diff options
| -rw-r--r-- | Makefile | 6 | ||||
| -rw-r--r-- | include/arpa/ftp.h | 109 | ||||
| -rw-r--r-- | include/arpa/inet.h | 101 | ||||
| -rw-r--r-- | include/arpa/nameser.h | 64 | ||||
| -rw-r--r-- | include/arpa/telnet.h | 11 | ||||
| -rw-r--r-- | include/arpa/tftp.h | 4 | ||||
| -rw-r--r-- | include/netinet/ether.h | 54 | ||||
| -rw-r--r-- | include/netinet/icmp6.h | 232 | ||||
| -rw-r--r-- | include/netinet/if_ether.h | 109 | ||||
| -rw-r--r-- | include/netinet/if_fddi.h | 37 | ||||
| -rw-r--r-- | include/netinet/if_tr.h | 41 | ||||
| -rw-r--r-- | include/netinet/igmp.h | 95 | ||||
| -rw-r--r-- | include/netinet/in.h | 289 | ||||
| -rw-r--r-- | include/netinet/in_systm.h | 42 | ||||
| -rw-r--r-- | include/netinet/ip.h | 291 | ||||
| -rw-r--r-- | include/netinet/ip6.h | 106 | ||||
| -rw-r--r-- | include/netinet/ip_icmp.h | 288 | ||||
| -rw-r--r-- | include/netinet/tcp.h | 167 | ||||
| -rw-r--r-- | include/netinet/udp.h | 56 | ||||
| -rw-r--r-- | include/regex.h | 3754 | ||||
| -rw-r--r-- | include/regexp.h | 241 | ||||
| -rw-r--r-- | include/stdlib.h | 25 | ||||
| -rw-r--r-- | include/sys/kd.h | 29 | ||||
| -rw-r--r-- | include/sys/klog.h | 34 | ||||
| -rw-r--r-- | include/sys/mtio.h | 277 | ||||
| -rw-r--r-- | libc/misc/regex/Makefile | 17 | ||||
| -rw-r--r-- | libc/misc/regex/rx.c | 7522 | ||||
| -rw-r--r-- | libc/stdlib/Makefile | 2 | ||||
| -rw-r--r-- | libc/stdlib/realpath.c | 168 |
29 files changed, 14004 insertions, 167 deletions
@@ -1,4 +1,4 @@ -DIRS = headers error getent malloc-simple misc regexp stdio2 \ +DIRS = headers error getent malloc-simple misc regex stdio2 \ string sysdeps termios time #rpc all: libc.a @@ -29,8 +29,8 @@ misc: dummy net: dummy make -C net -regexp: dummy - make -C regexp +regex: dummy + make -C regex rpc: dummy make -C rpc diff --git a/include/arpa/ftp.h b/include/arpa/ftp.h new file mode 100644 index 000000000..ac864aa70 --- /dev/null +++ b/include/arpa/ftp.h @@ -0,0 +1,109 @@ +/* + * Copyright (c) 1983, 1989, 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)ftp.h 8.1 (Berkeley) 6/2/93 + */ + +#ifndef _ARPA_FTP_H +#define _ARPA_FTP_H 1 + +/* Definitions for FTP; see RFC-765. */ + +/* + * Reply codes. + */ +#define PRELIM 1 /* positive preliminary */ +#define COMPLETE 2 /* positive completion */ +#define CONTINUE 3 /* positive intermediate */ +#define TRANSIENT 4 /* transient negative completion */ +#define ERROR 5 /* permanent negative completion */ + +/* + * Type codes + */ +#define TYPE_A 1 /* ASCII */ +#define TYPE_E 2 /* EBCDIC */ +#define TYPE_I 3 /* image */ +#define TYPE_L 4 /* local byte size */ + +#ifdef FTP_NAMES +char *typenames[] = {"0", "ASCII", "EBCDIC", "Image", "Local" }; +#endif + +/* + * Form codes + */ +#define FORM_N 1 /* non-print */ +#define FORM_T 2 /* telnet format effectors */ +#define FORM_C 3 /* carriage control (ASA) */ +#ifdef FTP_NAMES +char *formnames[] = {"0", "Nonprint", "Telnet", "Carriage-control" }; +#endif + +/* + * Structure codes + */ +#define STRU_F 1 /* file (no record structure) */ +#define STRU_R 2 /* record structure */ +#define STRU_P 3 /* page structure */ +#ifdef FTP_NAMES +char *strunames[] = {"0", "File", "Record", "Page" }; +#endif + +/* + * Mode types + */ +#define MODE_S 1 /* stream */ +#define MODE_B 2 /* block */ +#define MODE_C 3 /* compressed */ +#ifdef FTP_NAMES +char *modenames[] = {"0", "Stream", "Block", "Compressed" }; +#endif + +/* + * Record Tokens + */ +#define REC_ESC '\377' /* Record-mode Escape */ +#define REC_EOR '\001' /* Record-mode End-of-Record */ +#define REC_EOF '\002' /* Record-mode End-of-File */ + +/* + * Block Header + */ +#define BLK_EOR 0x80 /* Block is End-of-Record */ +#define BLK_EOF 0x40 /* Block is End-of-File */ +#define BLK_ERRORS 0x20 /* Block is suspected of containing errors */ +#define BLK_RESTART 0x10 /* Block is Restart Marker */ + +#define BLK_BYTECOUNT 2 /* Bytes in this block */ + +#endif /* arpa/ftp.h */ diff --git a/include/arpa/inet.h b/include/arpa/inet.h index cf4cd3275..e34c4726a 100644 --- a/include/arpa/inet.h +++ b/include/arpa/inet.h @@ -1,12 +1,95 @@ -#ifndef __ARPA_INET_H -#define __ARPA_INET_H +/* Copyright (C) 1997 Free Software Foundation, Inc. + This file is part of the GNU C Library. -#include <netinet/in.h> + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. -int inet_aton(const char *cp, struct in_addr *inp); - -unsigned long int inet_addr(const char *cp); + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. -char *inet_ntoa(struct in_addr in); - -#endif + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +#ifndef _ARPA_INET_H +#define _ARPA_INET_H 1 + +#include <features.h> + +#include <sys/types.h> +#include <netinet/in.h> /* To define `struct in_addr'. */ + +__BEGIN_DECLS + +/* Convert Internet host address from numbers-and-dots notation in CP + into binary data in network byte order. */ +extern u_int32_t inet_addr __P ((__const char *__cp)); + +/* Convert Internet host address from numbers-and-dots notation in CP + into binary data and store the result in the structure INP. */ +extern int inet_aton __P ((__const char *__cp, struct in_addr *__inp)); + +/* Return the local host address part of the Internet address in IN. */ +extern u_int32_t inet_lnaof __P ((struct in_addr __in)); + +/* Make Internet host address in network byte order by combining the + network number NET with the local address HOST. */ +extern struct in_addr inet_makeaddr __P ((u_int32_t __net, u_int32_t __host)); + +/* Format a network number NET into presentation format and place result + in buffer starting at BUF with length of LEN bytes. */ +extern char *inet_neta __P ((u_int32_t __net, char *__buf, size_t __len)); + +/* Return network number part of the Internet address IN. */ +extern u_int32_t inet_netof __P ((struct in_addr __in)); + +/* Extract the network number in network byte order from the address + in numbers-and-dots natation starting at CP. */ +extern u_int32_t inet_network __P ((__const char *__cp)); + +/* Convert network number for interface type AF in buffer starting at + CP to presentation format. The result will specifiy BITS bits of + the number. */ +extern char *inet_net_ntop __P ((int __af, __const void *__cp, int __bits, + char *__buf, size_t __len)); + +/* Convert network number for interface type AF from presentation in + buffer starting at CP to network format and store result int + buffer starting at BUF of size LEN. */ +extern int inet_net_pton __P ((int __af, __const char *__cp, + void *__buf, size_t __len)); + +/* Convert Internet number in IN to ASCII representation. The return value + is a pointer to an internal array containing the string. */ +extern char *inet_ntoa __P ((struct in_addr __in)); + +/* Convert from presentation format of an Internet number in buffer + starting at CP to the binary network format and store result for + interface type AF in buffer starting at BUF. */ +extern int inet_pton __P ((int __af, __const char *__cp, void *__buf)); + +/* Convert a Internet address in binary network format for interface + type AF in buffer starting at CP to presentation form and place + result in buffer of length LEN astarting at BUF. */ +extern __const char *inet_ntop __P ((int __af, __const void *__cp, + char *__buf, size_t __len)); + +/* Convert ASCII representation in hexadecimal form of the Internet + address to binary form and place result in buffer of length LEN + starting at BUF. */ +extern unsigned int inet_nsap_addr __P ((__const char *__cp, + unsigned char *__buf, int __len)); + +/* Convert internet address in binary form in LEN bytes starting at CP + a presentation form and place result in BUF. */ +extern char *inet_nsap_ntoa __P ((int __len, __const unsigned char *__cp, + char *__buf)); + +__END_DECLS + +#endif /* arpa/inet.h */ diff --git a/include/arpa/nameser.h b/include/arpa/nameser.h index e88ad8226..4a8ef67c7 100644 --- a/include/arpa/nameser.h +++ b/include/arpa/nameser.h @@ -77,23 +77,15 @@ /* * @(#)nameser.h 8.1 (Berkeley) 6/2/93 - * $Id: nameser.h,v 1.2 2000/05/14 06:07:30 erik Exp $ + * $Id: nameser.h,v 1.3 2000/07/09 06:39:14 andersen Exp $ */ -#ifndef _NAMESER_H_ -#define _NAMESER_H_ +#ifndef _ARPA_NAMESER_H +#define _ARPA_NAMESER_H 1 +#include <features.h> #include <sys/param.h> -#if (!defined(BSD)) || (BSD < 199306) -# include <sys/bitypes.h> -#else -# include <sys/types.h> -#endif -#include <sys/cdefs.h> - -#ifdef _AUX_SOURCE -# include <sys/types.h> -#endif +#include <sys/types.h> /* * revision information. this is the release date in YYYYMMDD format. @@ -254,46 +246,10 @@ #define CONV_BADCKSUM (-3) #define CONV_BADBUFLEN (-4) -#ifndef BYTE_ORDER -#if (BSD >= 199103) -# include <machine/endian.h> -#else -#ifdef __linux__ -# include <endian.h> -#else -#define LITTLE_ENDIAN 1234 /* least-significant byte first (vax, pc) */ -#define BIG_ENDIAN 4321 /* most-significant byte first (IBM, net) */ -#define PDP_ENDIAN 3412 /* LSB first in word, MSW first in long (pdp)*/ +/* glibc always has byte order info in <endian.h> */ +#include <endian.h> -#if defined(vax) || defined(ns32000) || defined(sun386) || defined(i386) || \ - defined(MIPSEL) || defined(_MIPSEL) || defined(BIT_ZERO_ON_RIGHT) || \ - defined(__alpha__) || defined(__alpha) -#define BYTE_ORDER LITTLE_ENDIAN -#endif - -#if defined(sel) || defined(pyr) || defined(mc68000) || defined(sparc) || \ - defined(is68k) || defined(tahoe) || defined(ibm032) || defined(ibm370) || \ - defined(MIPSEB) || defined(_MIPSEB) || defined(_IBMR2) || defined(DGUX) ||\ - defined(apollo) || defined(__convex__) || defined(_CRAY) || \ - defined(__hppa) || defined(__hp9000) || \ - defined(__hp9000s300) || defined(__hp9000s700) || \ - defined (BIT_ZERO_ON_LEFT) || defined(m68k) -#define BYTE_ORDER BIG_ENDIAN -#endif -#endif /* __linux__ */ -#endif /* BSD */ -#endif /* BYTE_ORDER */ - -#if !defined(BYTE_ORDER) || \ - (BYTE_ORDER != BIG_ENDIAN && BYTE_ORDER != LITTLE_ENDIAN && \ - BYTE_ORDER != PDP_ENDIAN) - /* you must determine what the correct bit order is for - * your compiler - the next line is an intentional error - * which will force your compiles to bomb until you fix - * the above macros. - */ - error "Undefined or invalid BYTE_ORDER"; -#endif +__BEGIN_DECLS /* * Structure for query header. The order of the fields is machine- and @@ -389,4 +345,6 @@ extern u_int32_t _getlong __P((const u_char *)); (cp) += INT32SZ; \ } -#endif /* !_NAMESER_H_ */ +__END_DECLS + +#endif /* arpa/nameser.h */ diff --git a/include/arpa/telnet.h b/include/arpa/telnet.h index 25085b89a..3309e5d72 100644 --- a/include/arpa/telnet.h +++ b/include/arpa/telnet.h @@ -34,7 +34,7 @@ */ #ifndef _ARPA_TELNET_H -#define _ARPA_TELNET_H +#define _ARPA_TELNET_H 1 /* * Definitions for the TELNET protocol. @@ -96,7 +96,7 @@ extern char *telcmds[]; #define TELOPT_NAOVTS 14 /* negotiate about vertical tab stops */ #define TELOPT_NAOVTD 15 /* negotiate about vertical tab disposition */ #define TELOPT_NAOLFD 16 /* negotiate about output LF disposition */ -#define TELOPT_XASCII 17 /* extended ascic character set */ +#define TELOPT_XASCII 17 /* extended ascii character set */ #define TELOPT_LOGOUT 18 /* force logout */ #define TELOPT_BM 19 /* byte macro */ #define TELOPT_DET 20 /* data entry terminal */ @@ -200,7 +200,7 @@ char *telopts[NTELOPTS+1] = { #define NSLC 18 /* - * For backwards compatability, we define SLC_NAMES to be the + * For backwards compatibility, we define SLC_NAMES to be the * list of names if SLC_NAMES is not defined. */ #define SLC_NAMELIST "0", "SYNCH", "BRK", "IP", "AO", "AYT", "EOR", \ @@ -237,7 +237,7 @@ extern char *slc_names[]; #define NEW_ENV_VAR 0 #define NEW_ENV_VALUE 1 #define ENV_ESC 2 -#define ENV_USERVAR 3 +#define ENV_USERVAR 3 /* * AUTHENTICATION suboptions @@ -316,4 +316,5 @@ extern char *enctype_names[]; #define ENCTYPE_NAME_OK(x) ((unsigned int)(x) < ENCTYPE_CNT) #define ENCTYPE_NAME(x) enctype_names[x] -#endif /* _ARPA_TELNET_H */ + +#endif /* arpa/telnet.h */ diff --git a/include/arpa/tftp.h b/include/arpa/tftp.h index 0904407c7..69187da08 100644 --- a/include/arpa/tftp.h +++ b/include/arpa/tftp.h @@ -56,9 +56,9 @@ struct tftphdr { unsigned short tu_block; /* block # */ short tu_code; /* error code */ char tu_stuff[1]; /* request packet stuff */ - } th_u; + } __attribute__ ((__packed__)) th_u; char th_data[1]; /* data or error string */ -}; +} __attribute__ ((__packed__)); #define th_block th_u.tu_block #define th_code th_u.tu_code diff --git a/include/netinet/ether.h b/include/netinet/ether.h new file mode 100644 index 000000000..c7985d7ac --- /dev/null +++ b/include/netinet/ether.h @@ -0,0 +1,54 @@ +/* Functions for storing Ethernet addresses in ASCII and mapping to hostnames. + Copyright (C) 1996, 1997 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +#ifndef _NETINET_ETHER_H +#define _NETINET_ETHER_H 1 + +#include <features.h> + +/* Get definition of `struct ether_addr'. */ +#include <netinet/if_ether.h> + +__BEGIN_DECLS + +/* Convert 48 bit Ethernet ADDRess to ASCII. */ +extern char *ether_ntoa __P ((__const struct ether_addr *__addr)); +extern char *ether_ntoa_r __P ((__const struct ether_addr *__addr, + char *__buf)); + +/* Convert ASCII string S to 48 bit Ethernet address. */ +extern struct ether_addr *ether_aton __P ((__const char *__asc)); +extern struct ether_addr *ether_aton_r __P ((__const char *__asc, + struct ether_addr *__addr)); + +/* Map 48 bit Ethernet number ADDR to HOSTNAME. */ +extern int ether_ntohost __P ((char *__hostname, + __const struct ether_addr *__addr)); + +/* Map HOSTNAME to 48 bit Ethernet address. */ +extern int ether_hostton __P ((__const char *__hostname, + struct ether_addr *__addr)); + +/* Scan LINE and set ADDR and HOSTNAME. */ +extern int ether_line __P ((__const char *__line, struct ether_addr *__addr, + char *__hostname)); + +__END_DECLS + +#endif /* netinet/ether.h */ diff --git a/include/netinet/icmp6.h b/include/netinet/icmp6.h new file mode 100644 index 000000000..5a3863953 --- /dev/null +++ b/include/netinet/icmp6.h @@ -0,0 +1,232 @@ +/* Copyright (C) 1991, 92, 93, 94, 95, 96, 97 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +#ifndef _NETINET_ICMP6_H +#define _NETINET_ICMP6_H 1 + +#include <inttypes.h> +#include <string.h> +#include <sys/types.h> +#include <netinet/in.h> + +#define ICMP6_FILTER 1 + +#define ICMP6_FILTER_BLOCK 1 +#define ICMP6_FILTER_PASS 2 +#define ICMP6_FILTER_BLOCKOTHERS 3 +#define ICMP6_FILTER_PASSONLY 4 + +struct icmp6_filter + { + uint32_t data[8]; + }; + +struct icmp6_hdr + { + uint8_t icmp6_type; /* type field */ + uint8_t icmp6_code; /* code field */ + uint16_t icmp6_cksum; /* checksum field */ + union + { + uint32_t icmp6_un_data32[1]; /* type-specific field */ + uint16_t icmp6_un_data16[2]; /* type-specific field */ + uint8_t icmp6_un_data8[4]; /* type-specific field */ + } icmp6_dataun; + }; + +#define icmp6_data32 icmp6_dataun.icmp6_un_data32 +#define icmp6_data16 icmp6_dataun.icmp6_un_data16 +#define icmp6_data8 icmp6_dataun.icmp6_un_data8 +#define icmp6_pptr icmp6_data32[0] /* parameter prob */ +#define icmp6_mtu icmp6_data32[0] /* packet too big */ +#define icmp6_id icmp6_data16[0] /* echo request/reply */ +#define icmp6_seq icmp6_data16[1] /* echo request/reply */ +#define icmp6_maxdelay icmp6_data16[0] /* mcast group membership */ + +#define ICMP6_DST_UNREACH 1 +#define ICMP6_PACKET_TOO_BIG 2 +#define ICMP6_TIME_EXCEEDED 3 +#define ICMP6_PARAM_PROB 4 + +#define ICMP6_INFOMSG_MASK 0x80 /* all informational messages */ + +#define ICMP6_ECHO_REQUEST 128 +#define ICMP6_ECHO_REPLY 129 +#define ICMP6_MEMBERSHIP_QUERY 130 +#define ICMP6_MEMBERSHIP_REPORT 131 +#define ICMP6_MEMBERSHIP_REDUCTION 132 + +#define ICMP6_DST_UNREACH_NOROUTE 0 /* no route to destination */ +#define ICMP6_DST_UNREACH_ADMIN 1 /* communication with destination */ + /* administratively prohibited */ +#define ICMP6_DST_UNREACH_NOTNEIGHBOR 2 /* not a neighbor */ +#define ICMP6_DST_UNREACH_ADDR 3 /* address unreachable */ +#define ICMP6_DST_UNREACH_NOPORT 4 /* bad port */ + +#define ICMP6_TIME_EXCEED_TRANSIT 0 /* Hop Limit == 0 in transit */ +#define ICMP6_TIME_EXCEED_REASSEMBLY 1 /* Reassembly time out */ + +#define ICMP6_PARAMPROB_HEADER 0 /* erroneous header field */ +#define ICMP6_PARAMPROB_NEXTHEADER 1 /* unrecognized Next Header */ +#define ICMP6_PARAMPROB_OPTION 2 /* unrecognized IPv6 option */ + +#define ICMP6_FILTER_WILLPASS(type, filterp) \ + ((((filterp)->data[(type) >> 5]) & (1 << ((type) & 31))) == 0) + +#define ICMP6_FILTER_WILLBLOCK(type, filterp) \ + ((((filterp)->data[(type) >> 5]) & (1 << ((type) & 31))) != 0) + +#define ICMP6_FILTER_SETPASS(type, filterp) \ + ((((filterp)->data[(type) >> 5]) &= ~(1 << ((type) & 31)))) + +#define ICMP6_FILTER_SETBLOCK(type, filterp) \ + ((((filterp)->data[(type) >> 5]) |= (1 << ((type) & 31)))) + +#define ICMP6_FILTER_SETPASSALL(filterp) \ + memset (filterp, 0, sizeof (struct icmp6_filter)); + +#define ICMP6_FILTER_SETBLOCKALL(filterp) \ + memset (filterp, 0xFF, sizeof (struct icmp6_filter)); + +#define ND_ROUTER_SOLICIT 133 +#define ND_ROUTER_ADVERT 134 +#define ND_NEIGHBOR_SOLICIT 135 +#define ND_NEIGHBOR_ADVERT 136 +#define ND_REDIRECT 137 + +struct nd_router_solicit /* router solicitation */ + { + struct icmp6_hdr nd_rs_hdr; + /* could be followed by options */ + }; + +#define nd_rs_type nd_rs_hdr.icmp6_type +#define nd_rs_code nd_rs_hdr.icmp6_code +#define nd_rs_cksum nd_rs_hdr.icmp6_cksum +#define nd_rs_reserved nd_rs_hdr.icmp6_data32[0] + +struct nd_router_advert /* router advertisement */ + { + struct icmp6_hdr nd_ra_hdr; + uint32_t nd_ra_reachable; /* reachable time */ + uint32_t nd_ra_retransmit; /* retransmit timer */ + /* could be followed by options */ + }; + +#define nd_ra_type nd_ra_hdr.icmp6_type +#define nd_ra_code nd_ra_hdr.icmp6_code +#define nd_ra_cksum nd_ra_hdr.icmp6_cksum +#define nd_ra_curhoplimit nd_ra_hdr.icmp6_data8[0] +#define nd_ra_flags_reserved nd_ra_hdr.icmp6_data8[1] +#define ND_RA_FLAG_MANAGED 0x80 +#define ND_RA_FLAG_OTHER 0x40 +#define nd_ra_router_lifetime nd_ra_hdr.icmp6_data16[1] + +struct nd_neighbor_solicit /* neighbor solicitation */ + { + struct icmp6_hdr nd_ns_hdr; + struct in6_addr nd_ns_target; /* target address */ + /* could be followed by options */ + }; + +#define nd_ns_type nd_ns_hdr.icmp6_type +#define nd_ns_code nd_ns_hdr.icmp6_code +#define nd_ns_cksum nd_ns_hdr.icmp6_cksum +#define nd_ns_reserved nd_ns_hdr.icmp6_data32[0] + +struct nd_neighbor_advert /* neighbor advertisement */ + { + struct icmp6_hdr nd_na_hdr; + struct in6_addr nd_na_target; /* target address */ + /* could be followed by options */ + }; + +#define nd_na_type nd_na_hdr.icmp6_type +#define nd_na_code nd_na_hdr.icmp6_code +#define nd_na_cksum nd_na_hdr.icmp6_cksum +#define nd_na_flags_reserved nd_na_hdr.icmp6_data32[0] +#if BYTE_ORDER == BIG_ENDIAN +#define ND_NA_FLAG_ROUTER 0x80000000 +#define ND_NA_FLAG_SOLICITED 0x40000000 +#define ND_NA_FLAG_OVERRIDE 0x20000000 +#else /* BYTE_ORDER == LITTLE_ENDIAN */ +#define ND_NA_FLAG_ROUTER 0x00000080 +#define ND_NA_FLAG_SOLICITED 0x00000040 +#define ND_NA_FLAG_OVERRIDE 0x00000020 +#endif + +struct nd_redirect /* redirect */ + { + struct icmp6_hdr nd_rd_hdr; + struct in6_addr nd_rd_target; /* target address */ + struct in6_addr nd_rd_dst; /* destination address */ + /* could be followed by options */ + }; + +#define nd_rd_type nd_rd_hdr.icmp6_type +#define nd_rd_code nd_rd_hdr.icmp6_code +#define nd_rd_cksum nd_rd_hdr.icmp6_cksum +#define nd_rd_reserved nd_rd_hdr.icmp6_data32[0] + +struct nd_opt_hdr /* Neighbor discovery option header */ + { + uint8_t nd_opt_type; + uint8_t nd_opt_len; /* in units of 8 octets */ + /* followed by option specific data */ + }; + +#define ND_OPT_SOURCE_LINKADDR 1 +#define ND_OPT_TARGET_LINKADDR 2 +#define ND_OPT_PREFIX_INFORMATION 3 +#define ND_OPT_REDIRECTED_HEADER 4 +#define ND_OPT_MTU 5 + +struct nd_opt_prefix_info /* prefix information */ + { + uint8_t nd_opt_pi_type; + uint8_t nd_opt_pi_len; + uint8_t nd_opt_pi_prefix_len; + uint8_t nd_opt_pi_flags_reserved; + uint32_t nd_opt_pi_valid_time; + uint32_t nd_opt_pi_preferred_time; + uint32_t nd_opt_pi_reserved2; + struct in6_addr nd_opt_pi_prefix; + }; + +#define ND_OPT_PI_FLAG_ONLINK 0x80 +#define ND_OPT_PI_FLAG_AUTO 0x40 + +struct nd_opt_rd_hdr /* redirected header */ + { + uint8_t nd_opt_rh_type; + uint8_t nd_opt_rh_len; + uint16_t nd_opt_rh_reserved1; + uint32_t nd_opt_rh_reserved2; + /* followed by IP header and data */ + }; + +struct nd_opt_mtu /* MTU option */ + { + uint8_t nd_opt_mtu_type; + uint8_t nd_opt_mtu_len; + uint16_t nd_opt_mtu_reserved; + uint32_t nd_opt_mtu_mtu; + }; + + +#endif /* netinet/icmpv6.h */ diff --git a/include/netinet/if_ether.h b/include/netinet/if_ether.h new file mode 100644 index 000000000..7194490f5 --- /dev/null +++ b/include/netinet/if_ether.h @@ -0,0 +1,109 @@ +/* Copyright (C) 1996, 1997 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +#ifndef __NETINET_IF_ETHER_H + +#define __NETINET_IF_ETHER_H 1 +#include <features.h> +#include <sys/types.h> + +/* Get definitions from kernel header file. */ +#include <linux/if_ether.h> + +#ifdef __USE_BSD +/* + * Copyright (c) 1982, 1986, 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)if_ether.h 8.3 (Berkeley) 5/2/95 + * $FreeBSD$ + */ + +#include <net/ethernet.h> +#include <net/if_arp.h> + +__BEGIN_DECLS +/* + * Ethernet Address Resolution Protocol. + * + * See RFC 826 for protocol description. Structure below is adapted + * to resolving internet addresses. Field names used correspond to + * RFC 826. + */ +struct ether_arp { + struct arphdr ea_hdr; /* fixed-size header */ + u_int8_t arp_sha[ETH_ALEN]; /* sender hardware address */ + u_int8_t arp_spa[4]; /* sender protocol address */ + u_int8_t arp_tha[ETH_ALEN]; /* target hardware address */ + u_int8_t arp_tpa[4]; /* target protocol address */ +}; +#define arp_hrd ea_hdr.ar_hrd +#define arp_pro ea_hdr.ar_pro +#define arp_hln ea_hdr.ar_hln +#define arp_pln ea_hdr.ar_pln +#define arp_op ea_hdr.ar_op + +/* + * Macro to map an IP multicast address to an Ethernet multicast address. + * The high-order 25 bits of the Ethernet address are statically assigned, + * and the low-order 23 bits are taken from the low end of the IP address. + */ +#define ETHER_MAP_IP_MULTICAST(ipaddr, enaddr) \ + /* struct in_addr *ipaddr; */ \ + /* u_char enaddr[ETH_ALEN]; */ \ +{ \ + (enaddr)[0] = 0x01; \ + (enaddr)[1] = 0x00; \ + (enaddr)[2] = 0x5e; \ + (enaddr)[3] = ((u_int8_t *)ipaddr)[1] & 0x7f; \ + (enaddr)[4] = ((u_int8_t *)ipaddr)[2]; \ + (enaddr)[5] = ((u_int8_t *)ipaddr)[3]; \ +} + +__END_DECLS +#endif /* __USE_BSD */ + +#endif /* netinet/if_ether.h */ diff --git a/include/netinet/if_fddi.h b/include/netinet/if_fddi.h new file mode 100644 index 000000000..d5d6dbdbe --- /dev/null +++ b/include/netinet/if_fddi.h @@ -0,0 +1,37 @@ +/* Copyright (C) 1997 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +#ifndef _NETINET_IF_FDDI_H +#define _NETINET_IF_FDDI_H 1 + +#include <sys/cdefs.h> +#include <sys/types.h> +#include <asm/types.h> + +#include <linux/if_fddi.h> + +#ifdef __USE_BSD + +struct fddi_header { + u_int8_t fddi_fc; /* Frame Control (FC) value */ + u_int8_t fddi_dhost[FDDI_K_ALEN]; /* Destination host */ + u_int8_t fddi_shost[FDDI_K_ALEN]; /* Source host */ +}; +#endif + +#endif /* netinet/if_fddi.h */ diff --git a/include/netinet/if_tr.h b/include/netinet/if_tr.h new file mode 100644 index 000000000..3c1be2102 --- /dev/null +++ b/include/netinet/if_tr.h @@ -0,0 +1,41 @@ +/* Copyright (C) 1997 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +#ifndef _NETINET_IF_TR_H +#define _NETINET_IF_TR_H 1 + +#include <sys/cdefs.h> +#include <sys/types.h> +#include <asm/types.h> + +#include <linux/if_tr.h> + +#ifdef __USE_BSD + +struct trn_hdr { + u_int8_t trn_ac; /* access control field */ + u_int8_t trn_fc; /* field control field */ + u_int8_t trn_dhost[TR_ALEN]; /* destination host */ + u_int8_t trn_shost[TR_ALEN]; /* source host */ + u_int16_t trn_rcf; /* route control field */ + u_int16_t trn_rseg[8]; /* routing registers */ +}; + +#endif + +#endif /* netinet/if_tr.h */ diff --git a/include/netinet/igmp.h b/include/netinet/igmp.h index 4525630e1..7a6ed6e8c 100644 --- a/include/netinet/igmp.h +++ b/include/netinet/igmp.h @@ -1,24 +1,93 @@ +/* Copyright (C) 1997 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + #ifndef _NETINET_IGMP_H -#define _NETINET_IGMP_H +#define _NETINET_IGMP_H 1 +#include <sys/cdefs.h> +#include <sys/types.h> + +#include <asm/types.h> #include <linux/igmp.h> -#ifdef __BSD_SOURCE +#ifdef __USE_BSD + +#include <netinet/in.h> + +__BEGIN_DECLS + +/* + * Copyright (c) 1988 Stephen Deering. + * Copyright (c) 1992, 1993 + * The Regents of the University of California. All rights reserved. + * + * This code is derived from software contributed to Berkeley by + * Stephen Deering of Stanford University. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)igmp.h 8.1 (Berkeley) 6/10/93 + * $FreeBSD$ + */ -struct igmp -{ - __u8 igmp_type; - __u8 igmp_code; - __u16 igmp_cksum; - struct in_addr igmp_group; +struct igmp { + u_int8_t igmp_type; /* IGMP type */ + u_int8_t igmp_code; /* routing code */ + u_int16_t igmp_cksum; /* checksum */ + struct in_addr igmp_group; /* group address */ }; -#define IGMP_MINLEN 8 -#define IGMP_MAX_HOST_REPORT_DELAY 10 -#define IGMP_TIMER_SCALE 10 +/* + * Message types, including version number. + */ +#define IGMP_MEMBERSHIP_QUERY 0x11 /* membership query */ +#define IGMP_V1_MEMBERSHIP_REPORT 0x12 /* Ver. 1 membership report */ +#define IGMP_V2_MEMBERSHIP_REPORT 0x16 /* Ver. 2 membership report */ +#define IGMP_V2_LEAVE_GROUP 0x17 /* Leave-group message */ -#define IGMP_AGE_THRESHOLD 540 +__END_DECLS #endif -#endif /* _NETINET_IGMP_H */ +#endif /* netinet/igmp.h */ diff --git a/include/netinet/in.h b/include/netinet/in.h index d666c67ac..3dea92632 100644 --- a/include/netinet/in.h +++ b/include/netinet/in.h @@ -1,29 +1,66 @@ -/* Copyright (C) 1991 Free Software Foundation, Inc. -This file is part of the GNU C Library. +/* Copyright (C) 1991,92,93,94,95,96,97,98,99 Free Software Foundation, Inc. + This file is part of the GNU C Library. -The GNU C Library is free software; you can redistribute it and/or modify -it under the terms of the GNU General Public License as published by -the Free Software Foundation; either version 1, or (at your option) -any later version. + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. -The GNU C Library is distributed in the hope that it will be useful, -but WITHOUT ANY WARRANTY; without even the implied warranty of -MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the -GNU General Public License for more details. + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. -You should have received a copy of the GNU General Public License -along with the GNU C Library; see the file COPYING. If not, write to -the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */ + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ #ifndef _NETINET_IN_H - #define _NETINET_IN_H 1 + #include <features.h> +#include <limits.h> +#include <stdint.h> + +#include <sys/types.h> +#include <bits/socket.h> -#include <sys/socket.h> __BEGIN_DECLS +/* Standard well-defined IP protocols. */ +enum + { + IPPROTO_IP = 0, /* Dummy protocol for TCP. */ + IPPROTO_HOPOPTS = 0, /* IPv6 Hop-by-Hop options. */ + IPPROTO_ICMP = 1, /* Internet Control Message Protocol. */ + IPPROTO_IGMP = 2, /* Internet Group Management Protocol. */ + IPPROTO_IPIP = 4, /* IPIP tunnels (older KA9Q tunnels use 94). */ + IPPROTO_TCP = 6, /* Transmission Control Protocol. */ + IPPROTO_EGP = 8, /* Exterior Gateway Protocol. */ + IPPROTO_PUP = 12, /* PUP protocol. */ + IPPROTO_UDP = 17, /* User Datagram Protocol. */ + IPPROTO_IDP = 22, /* XNS IDP protocol. */ + IPPROTO_TP = 29, /* SO Transport Protocol Class 4. */ + IPPROTO_IPV6 = 41, /* IPv6 header. */ + IPPROTO_ROUTING = 43, /* IPv6 routing header. */ + IPPROTO_FRAGMENT = 44, /* IPv6 fragmentation header. */ + IPPROTO_RSVP = 46, /* Reservation Protocol. */ + IPPROTO_GRE = 47, /* General Routing Encapsulation. */ + IPPROTO_ESP = 50, /* encapsulating security payload. */ + IPPROTO_AH = 51, /* authentication header. */ + IPPROTO_ICMPV6 = 58, /* ICMPv6. */ + IPPROTO_NONE = 59, /* IPv6 no next header. */ + IPPROTO_DSTOPTS = 60, /* IPv6 destination options. */ + IPPROTO_MTP = 92, /* Multicast Transport Protocol. */ + IPPROTO_ENCAP = 98, /* Encapsulation Header. */ + IPPROTO_PIM = 103, /* Protocol Independent Multicast. */ + IPPROTO_COMP = 108, /* Compression Header Protocol. */ + IPPROTO_RAW = 255, /* Raw IP packets. */ + IPPROTO_MAX + }; + /* Standard well-known ports. */ enum { @@ -65,23 +102,217 @@ enum }; -/* Link numbers. */ -#define IMPLINK_IP 155 -#define IMPLINK_LOWEXPER 156 -#define IMPLINK_HIGHEXPER 158 +/* Internet address. */ +struct in_addr + { + uint32_t s_addr; + }; + + +/* Definitions of the bits in an Internet address integer. + + On subnets, host and network parts are found according to + the subnet mask, not these masks. */ + +#define IN_CLASSA(a) ((((uint32_t) (a)) & 0x80000000) == 0) +#define IN_CLASSA_NET 0xff000000 +#define IN_CLASSA_NSHIFT 24 +#define IN_CLASSA_HOST (0xffffffff & ~IN_CLASSA_NET) +#define IN_CLASSA_MAX 128 + +#define IN_CLASSB(a) ((((uint32_t) (a)) & 0xc0000000) == 0x80000000) +#define IN_CLASSB_NET 0xffff0000 +#define IN_CLASSB_NSHIFT 16 +#define IN_CLASSB_HOST (0xffffffff & ~IN_CLASSB_NET) +#define IN_CLASSB_MAX 65536 + +#define IN_CLASSC(a) ((((uint32_t) (a)) & 0xe0000000) == 0xc0000000) +#define IN_CLASSC_NET 0xffffff00 +#define IN_CLASSC_NSHIFT 8 +#define IN_CLASSC_HOST (0xffffffff & ~IN_CLASSC_NET) + +#define IN_CLASSD(a) ((((uint32_t) (a)) & 0xf0000000) == 0xe0000000) +#define IN_MULTICAST(a) IN_CLASSD(a) + +#define IN_EXPERIMENTAL(a) ((((uint32_t) (a)) & 0xe0000000) == 0xe0000000) +#define IN_BADCLASS(a) ((((uint32_t) (a)) & 0xf0000000) == 0xf0000000) + +/* Address to accept any incoming messages. */ +#define INADDR_ANY ((uint32_t) 0x00000000) +/* Address to send to all hosts. */ +#define INADDR_BROADCAST ((uint32_t) 0xffffffff) +/* Address indicating an error return. */ +#define INADDR_NONE ((uint32_t) 0xffffffff) + +/* Network number for local host loopback. */ +#define IN_LOOPBACKNET 127 +/* Address to loopback in software to local host. */ +#ifndef INADDR_LOOPBACK +# define INADDR_LOOPBACK ((uint32_t) 0x7f000001) /* Inet 127.0.0.1. */ +#endif + +/* Defines for Multicast INADDR. */ +#define INADDR_UNSPEC_GROUP ((uint32_t) 0xe0000000) /* 224.0.0.0 */ +#define INADDR_ALLHOSTS_GROUP ((uint32_t) 0xe0000001) /* 224.0.0.1 */ +#define INADDR_ALLRTRS_GROUP ((uint32_t) 0xe0000002) /* 224.0.0.2 */ +#define INADDR_MAX_LOCAL_GROUP ((uint32_t) 0xe00000ff) /* 224.0.0.255 */ + + +/* IPv6 address */ +struct in6_addr + { + union + { + uint8_t u6_addr8[16]; + uint16_t u6_addr16[8]; + uint32_t u6_addr32[4]; +#if ULONG_MAX > 0xffffffff + uint64_t u6_addr64[2]; +#endif + } in6_u; +#define s6_addr in6_u.u6_addr8 +#define s6_addr16 in6_u.u6_addr16 +#define s6_addr32 in6_u.u6_addr32 +#define s6_addr64 in6_u.u6_addr64 + }; + +extern const struct in6_addr in6addr_any; /* :: */ +extern const struct in6_addr in6addr_loopback; /* ::1 */ +#define IN6ADDR_ANY_INIT { { { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } } } +#define IN6ADDR_LOOPBACK_INIT { { { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1 } } } + +#define INET_ADDRSTRLEN 16 +#define INET6_ADDRSTRLEN 46 + +/* Get the definition of the macro to define the common sockaddr members. */ +#include <bits/sockaddr.h> + + +/* Structure describing an Internet socket address. */ +struct sockaddr_in + { + __SOCKADDR_COMMON (sin_); + uint16_t sin_port; /* Port number. */ + struct in_addr sin_addr; /* Internet address. */ + + /* Pad to size of `struct sockaddr'. */ + unsigned char sin_zero[sizeof (struct sockaddr) - + __SOCKADDR_COMMON_SIZE - + sizeof (uint16_t) - + sizeof (struct in_addr)]; + }; + +/* Ditto, for IPv6. */ +struct sockaddr_in6 + { + __SOCKADDR_COMMON (sin6_); + uint16_t sin6_port; /* Transport layer port # */ + uint32_t sin6_flowinfo; /* IPv6 flow information */ + struct in6_addr sin6_addr; /* IPv6 address */ + }; + +/* IPv6 multicast request. */ +struct ipv6_mreq + { + /* IPv6 multicast address of group */ + struct in6_addr ipv6mr_multiaddr; + + /* local interface */ + unsigned int ipv6mr_interface; + }; + +/* Get system-specific definitions. */ +#include <bits/in.h> + +/* Functions to convert between host and network byte order. + + Please note that these functions normally take `unsigned long int' or + `unsigned short int' values as arguments and also return them. But + this was a short-sighted decision since on different systems the types + may have different representations but the values are always the same. */ + +extern uint32_t ntohl __P ((uint32_t __netlong)); +extern uint16_t ntohs __P ((uint16_t __netshort)); +extern uint32_t htonl __P ((uint32_t __hostlong)); +extern uint16_t htons __P ((uint16_t __hostshort)); + +#include <endian.h> + +/* Get machine dependent optimized versions of byte swapping functions. */ +#include <bits/byteswap.h> + +#if __BYTE_ORDER == __BIG_ENDIAN && defined __OPTIMIZE__ +/* The host byte order is the same as network byte order, + so these functions are all just identity. */ +# define ntohl(x) (x) +# define ntohs(x) (x) +# define htonl(x) (x) +# define htons(x) (x) +#else +# if __BYTE_ORDER == __LITTLE_ENDIAN && defined __OPTIMIZE__ +# define ntohl(x) __bswap_32 (x) +# define ntohs(x) __bswap_16 (x) +# define htonl(x) __bswap_32 (x) +# define htons(x) __bswap_16 (x) +# endif +#endif + +#define IN6_IS_ADDR_UNSPECIFIED(a) \ + (((uint32_t *) (a))[0] == 0 && ((uint32_t *) (a))[1] == 0 && \ + ((uint32_t *) (a))[2] == 0 && ((uint32_t *) (a))[3] == 0) +#define IN6_IS_ADDR_LOOPBACK(a) \ + (((uint32_t *) (a))[0] == 0 && ((uint32_t *) (a))[1] == 0 && \ + ((uint32_t *) (a))[2] == 0 && ((uint32_t *) (a))[3] == htonl (1)) -/* - * Many other definitions have been moved to <linux/in.h>, - * because several parts of the kernel need them. -FvK - */ -#include <linux/in.h> +#define IN6_IS_ADDR_MULTICAST(a) (((u_int8_t *) (a))[0] == 0xff) -/* - * Bind a socket to a privileged IP port - */ -extern int bindresvport __P ((int __sockfd, - struct sockaddr_in * __sin)); +#define IN6_IS_ADDR_LINKLOCAL(a) \ + ((((uint32_t *) (a))[0] & htonl (0xffc00000)) == htonl (0xfe800000)) + +#define IN6_IS_ADDR_SITELOCAL(a) \ + ((((uint32_t *) (a))[0] & htonl (0xffc00000)) == htonl (0xfec00000)) + +#define IN6_IS_ADDR_V4MAPPED(a) \ + ((((uint32_t *) (a))[0] == 0) && (((uint32_t *) (a))[1] == 0) && \ + (((uint32_t *) (a))[2] == htonl (0xffff))) + +#define IN6_IS_ADDR_V4COMPAT(a) \ + ((((uint32_t *) (a))[0] == 0) && (((uint32_t *) (a))[1] == 0) && \ + (((uint32_t *) (a))[2] == 0) && (ntohl (((uint32_t *) (a))[3]) > 1)) + +#define IN6_ARE_ADDR_EQUAL(a,b) \ + ((((uint32_t *) (a))[0] == ((uint32_t *) (b))[0]) && \ + (((uint32_t *) (a))[1] == ((uint32_t *) (b))[2]) && \ + (((uint32_t *) (a))[2] == ((uint32_t *) (b))[1]) && \ + (((uint32_t *) (a))[3] == ((uint32_t *) (b))[3])) + +/* Bind socket to a privileged IP port. */ +extern int bindresvport __P ((int __sockfd, struct sockaddr_in *__sock_in)); + + + +#define IN6_IS_ADDR_MC_NODELOCAL(a) \ + (IN6_IS_ADDR_MULTICAST(a) && ((((u_int8_t *) (a))[1] & 0xf) == 0x1)) + +#define IN6_IS_ADDR_MC_LINKLOCAL(a) \ + (IN6_IS_ADDR_MULTICAST(a) && ((((u_int8_t *) (a))[1] & 0xf) == 0x2)) + +#define IN6_IS_ADDR_MC_SITELOCAL(a) \ + (IN6_IS_ADDR_MULTICAST(a) && ((((u_int8_t *) (a))[1] & 0xf) == 0x5)) + +#define IN6_IS_ADDR_MC_ORGLOCAL(a) \ + (IN6_IS_ADDR_MULTICAST(a) && ((((u_int8_t *) (a))[1] & 0xf) == 0x8)) + +#define IN6_IS_ADDR_MC_GLOBAL(a) \ + (IN6_IS_ADDR_MULTICAST(a) && ((((u_int8_t *) (a))[1] & 0xf) == 0xe)) + +/* IPv6 packet information. */ +struct in6_pktinfo + { + struct in6_addr ipi6_addr; /* src/dst IPv6 address */ + unsigned int ipi6_ifindex; /* send/recv interface index */ + }; __END_DECLS diff --git a/include/netinet/in_systm.h b/include/netinet/in_systm.h index f481c5511..902fe6ea0 100644 --- a/include/netinet/in_systm.h +++ b/include/netinet/in_systm.h @@ -1 +1,41 @@ -#include <linux/in_systm.h> +/* System specific type definitions for networking code. + Copyright (C) 1997 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +#ifndef _NETINET_IN_SYSTM_H +#define _NETINET_IN_SYSTM_H 1 + +#include <sys/cdefs.h> +#include <sys/types.h> + +__BEGIN_DECLS + +/* + * Network order versions of various data types. Unfortunately, BSD + * assumes specific sizes for shorts (16 bit) and longs (32 bit) which + * don't hold in general. As a consequence, the network order versions + * may not reflect the actual size of the native data types. + */ + +typedef u_int16_t n_short; /* short as received from the net */ +typedef u_int32_t n_long; /* long as received from the net */ +typedef u_int32_t n_time; /* ms since 00:00 GMT, byte rev */ + +__END_DECLS + +#endif /* netinet/in_systm.h */ diff --git a/include/netinet/ip.h b/include/netinet/ip.h index a6c49b6af..246a56a5d 100644 --- a/include/netinet/ip.h +++ b/include/netinet/ip.h @@ -1,39 +1,276 @@ -#ifndef _NETINET_IP_H -#define _NETINET_IP_H +/* Copyright (C) 1991, 92, 93, 95, 96, 97, 98 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +#ifndef __NETINET_IP_H +#define __NETINET_IP_H 1 #include <features.h> +#include <sys/types.h> + #include <netinet/in.h> -#include <linux/ip.h> -#ifdef _BSD_SOURCE +__BEGIN_DECLS + +struct timestamp + { + u_int8_t len; + u_int8_t ptr; +#if __BYTE_ORDER == __LITTLE_ENDIAN + unsigned int flags:4; + unsigned int overflow:4; +#elif __BYTE_ORDER == __BIG_ENDIAN + unsigned int overflow:4; + unsigned int flags:4; +#else +# error "Please fix <bits/endian.h>" +#endif + u_int32_t data[9]; + }; + +struct ip_options + { + u_int32_t faddr; /* Saved first hop address */ + u_int8_t optlen; + u_int8_t srr; + u_int8_t rr; + u_int8_t ts; + unsigned int is_setbyuser:1; /* Set by setsockopt? */ + unsigned int is_data:1; /* Options in __data, rather than skb */ + unsigned int is_strictroute:1; /* Strict source route */ + unsigned int srr_is_hit:1; /* Packet destination addr was our one */ + unsigned int is_changed:1; /* IP checksum more not valid */ + unsigned int rr_needaddr:1; /* Need to record addr of outgoing dev */ + unsigned int ts_needtime:1; /* Need to record timestamp */ + unsigned int ts_needaddr:1; /* Need to record addr of outgoing dev */ + u_int8_t router_alert; + u_int8_t __pad1; + u_int8_t __pad2; +#ifdef __GNUC__ + u_int8_t __data[0]; +#endif + }; + +struct iphdr + { +#if __BYTE_ORDER == __LITTLE_ENDIAN + unsigned int ihl:4; + unsigned int version:4; +#elif __BYTE_ORDER == __BIG_ENDIAN + unsigned int version:4; + unsigned int ihl:4; +#else +# error "Please fix <bits/endian.h>" +#endif + u_int8_t tos; + u_int16_t tot_len; + u_int16_t id; + u_int16_t frag_off; + u_int8_t ttl; + u_int8_t protocol; + u_int16_t check; + u_int32_t saddr; + u_int32_t daddr; + /*The options start here. */ + }; +#ifdef __USE_BSD /* - * BSD has the following structure + * Copyright (c) 1982, 1986, 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)ip.h 8.1 (Berkeley) 6/10/93 + */ + +/* + * Definitions for internet protocol version 4. + * Per RFC 791, September 1981. + */ + +/* + * Structure of an internet header, naked of options. */ - struct ip -{ -#if defined(__LITTLE_ENDIAN_BITFIELD) - __u8 ip_hl:4, - ip_v:4; -#else - __u8 ip_v:4, - ip_hl:4; + { +#if __BYTE_ORDER == __LITTLE_ENDIAN + unsigned int ip_hl:4; /* header length */ + unsigned int ip_v:4; /* version */ +#endif +#if __BYTE_ORDER == __BIG_ENDIAN + unsigned int ip_v:4; /* version */ + unsigned int ip_hl:4; /* header length */ #endif -#define IPVERSION 4 - __u8 ip_tos; - __u16 ip_len; - __u16 ip_id; - __u16 ip_off; - __u8 ip_ttl; - __u8 ip_p; - __u16 ip_csum; - struct in_addr ip_src,ip_dst; -}; - -#define IP_DF 0x4000 /* dont fragment flag */ -#define IP_MF 0x2000 /* more fragments flag */ + u_int8_t ip_tos; /* type of service */ + u_short ip_len; /* total length */ + u_short ip_id; /* identification */ + u_short ip_off; /* fragment offset field */ +#define IP_RF 0x8000 /* reserved fragment flag */ +#define IP_DF 0x4000 /* dont fragment flag */ +#define IP_MF 0x2000 /* more fragments flag */ +#define IP_OFFMASK 0x1fff /* mask for fragmenting bits */ + u_int8_t ip_ttl; /* time to live */ + u_int8_t ip_p; /* protocol */ + u_short ip_sum; /* checksum */ + struct in_addr ip_src, ip_dst; /* source and dest address */ + }; +/* + * Time stamp option structure. + */ +struct ip_timestamp + { + u_int8_t ipt_code; /* IPOPT_TS */ + u_int8_t ipt_len; /* size of structure (variable) */ + u_int8_t ipt_ptr; /* index of current entry */ +#if __BYTE_ORDER == __LITTLE_ENDIAN + unsigned int ipt_flg:4; /* flags, see below */ + unsigned int ipt_oflw:4; /* overflow counter */ +#endif +#if __BYTE_ORDER == __BIG_ENDIAN + unsigned int ipt_oflw:4; /* overflow counter */ + unsigned int ipt_flg:4; /* flags, see below */ #endif + u_int32_t data[9]; + }; +#endif /* __USE_BSD */ + +#define IPVERSION 4 /* IP version number */ +#define IP_MAXPACKET 65535 /* maximum packet size */ + +/* + * Definitions for IP type of service (ip_tos) + */ +#define IPTOS_TOS_MASK 0x1E +#define IPTOS_TOS(tos) ((tos) & IPTOS_TOS_MASK) +#define IPTOS_LOWDELAY 0x10 +#define IPTOS_THROUGHPUT 0x08 +#define IPTOS_RELIABILITY 0x04 +#define IPTOS_LOWCOST 0x02 +#define IPTOS_MINCOST IPTOS_LOWCOST + +/* + * Definitions for IP precedence (also in ip_tos) (hopefully unused) + */ +#define IPTOS_PREC_MASK 0xe0 +#define IPTOS_PREC(tos) ((tos) & IPTOS_PREC_MASK) +#define IPTOS_PREC_NETCONTROL 0xe0 +#define IPTOS_PREC_INTERNETCONTROL 0xc0 +#define IPTOS_PREC_CRITIC_ECP 0xa0 +#define IPTOS_PREC_FLASHOVERRIDE 0x80 +#define IPTOS_PREC_FLASH 0x60 +#define IPTOS_PREC_IMMEDIATE 0x40 +#define IPTOS_PREC_PRIORITY 0x20 +#define IPTOS_PREC_ROUTINE 0x00 + +/* + * Definitions for options. + */ +#define IPOPT_COPY 0x80 +#define IPOPT_CLASS_MASK 0x60 +#define IPOPT_NUMBER_MASK 0x1f + +#define IPOPT_COPIED(o) ((o) & IPOPT_COPY) +#define IPOPT_CLASS(o) ((o) & IPOPT_CLASS_MASK) +#define IPOPT_NUMBER(o) ((o) & IPOPT_NUMBER_MASK) + +#define IPOPT_CONTROL 0x00 +#define IPOPT_RESERVED1 0x20 +#define IPOPT_DEBMEAS 0x40 +#define IPOPT_MEASUREMENT IPOPT_DEBMEAS +#define IPOPT_RESERVED2 0x60 + +#define IPOPT_EOL 0 /* end of option list */ +#define IPOPT_END IPOPT_EOL +#define IPOPT_NOP 1 /* no operation */ +#define IPOPT_NOOP IPOPT_NOP + +#define IPOPT_RR 7 /* record packet route */ +#define IPOPT_TS 68 /* timestamp */ +#define IPOPT_TIMESTAMP IPOPT_TS +#define IPOPT_SECURITY 130 /* provide s,c,h,tcc */ +#define IPOPT_SEC IPOPT_SECURITY +#define IPOPT_LSRR 131 /* loose source route */ +#define IPOPT_SATID 136 /* satnet id */ +#define IPOPT_SID IPOPT_SATID +#define IPOPT_SSRR 137 /* strict source route */ +#define IPOPT_RA 148 /* router alert */ + +/* + * Offsets to fields in options other than EOL and NOP. + */ +#define IPOPT_OPTVAL 0 /* option ID */ +#define IPOPT_OLEN 1 /* option length */ +#define IPOPT_OFFSET 2 /* offset within option */ +#define IPOPT_MINOFF 4 /* min value of above */ + +#define MAX_IPOPTLEN 40 + +/* flag bits for ipt_flg */ +#define IPOPT_TS_TSONLY 0 /* timestamps only */ +#define IPOPT_TS_TSANDADDR 1 /* timestamps and addresses */ +#define IPOPT_TS_PRESPEC 3 /* specified modules only */ + +/* bits for security (not byte swapped) */ +#define IPOPT_SECUR_UNCLASS 0x0000 +#define IPOPT_SECUR_CONFID 0xf135 +#define IPOPT_SECUR_EFTO 0x789a +#define IPOPT_SECUR_MMMM 0xbc4d +#define IPOPT_SECUR_RESTR 0xaf13 +#define IPOPT_SECUR_SECRET 0xd788 +#define IPOPT_SECUR_TOPSECRET 0x6bc5 + +/* + * Internet implementation parameters. + */ +#define MAXTTL 255 /* maximum time to live (seconds) */ +#define IPDEFTTL 64 /* default ttl, from RFC 1340 */ +#define IPFRAGTTL 60 /* time to live for frags, slowhz */ +#define IPTTLDEC 1 /* subtracted when forwarding */ + +#define IP_MSS 576 /* default maximum segment size */ + +__END_DECLS -#endif /* _NETINET_IP_H */ +#endif /* netinet/ip.h */ diff --git a/include/netinet/ip6.h b/include/netinet/ip6.h new file mode 100644 index 000000000..cd42ef64b --- /dev/null +++ b/include/netinet/ip6.h @@ -0,0 +1,106 @@ +/* Copyright (C) 1991, 92, 93, 94, 95, 96, 97 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +#ifndef _NETINET_IP6_H +#define _NETINET_IP6_H 1 + +#include <inttypes.h> +#include <netinet/in.h> + +struct ip6_hdr + { + union + { + struct ip6_hdrctl + { + uint32_t ip6_un1_flow; /* 24 bits of flow-ID */ + uint16_t ip6_un1_plen; /* payload length */ + uint8_t ip6_un1_nxt; /* next header */ + uint8_t ip6_un1_hlim; /* hop limit */ + } ip6_un1; + uint8_t ip6_un2_vfc; /* 4 bits version, 4 bits priority */ + } ip6_ctlun; + struct in6_addr ip6_src; /* source address */ + struct in6_addr ip6_dst; /* destination address */ + }; + +#define ip6_vfc ip6_ctlun.ip6_un2_vfc +#define ip6_flow ip6_ctlun.ip6_un1.ip6_un1_flow +#define ip6_plen ip6_ctlun.ip6_un1.ip6_un1_plen +#define ip6_nxt ip6_ctlun.ip6_un1.ip6_un1_nxt +#define ip6_hlim ip6_ctlun.ip6_un1.ip6_un1_hlim +#define ip6_hops ip6_ctlun.ip6_un1.ip6_un1_hlim + +/* Hop-by-Hop options header. */ +struct ip6_hbh + { + uint8_t ip6h_nxt; /* next hesder. */ + uint8_t ip6h_len; /* length in units of 8 octets. */ + /* followed by options */ + }; + +/* Destination options header */ +struct ip6_dest + { + uint8_t ip6d_nxt; /* next header */ + uint8_t ip6d_len; /* length in units of 8 octets */ + /* followed by options */ + }; + +/* Routing header */ +struct ip6_rthdr + { + uint8_t ip6r_nxt; /* next header */ + uint8_t ip6r_len; /* length in units of 8 octets */ + uint8_t ip6r_type; /* routing type */ + uint8_t ip6r_segleft; /* segments left */ + /* followed by routing type specific data */ + }; + +/* Type 0 Routing header */ +struct ip6_rthdr0 + { + uint8_t ip6r0_nxt; /* next header */ + uint8_t ip6r0_len; /* length in units of 8 octets */ + uint8_t ip6r0_type; /* always zero */ + uint8_t ip6r0_segleft; /* segments left */ + uint8_t ip6r0_reserved; /* reserved field */ + uint8_t ip6r0_slmap[3]; /* strict/loose bit map */ + struct in6_addr ip6r0_addr[1]; /* up to 23 addresses */ + }; + +/* Fragment header */ +struct ip6_frag + { + uint8_t ip6f_nxt; /* next header */ + uint8_t ip6f_reserved; /* reserved field */ + uint16_t ip6f_offlg; /* offset, reserved, and flag */ + uint32_t ip6f_ident; /* identification */ + }; + +#if BYTE_ORDER == BIG_ENDIAN +#define IP6F_OFF_MASK 0xfff8 /* mask out offset from _offlg */ +#define IP6F_RESERVED_MASK 0x0006 /* reserved bits in ip6f_offlg */ +#define IP6F_MORE_FRAG 0x0001 /* more-fragments flag */ +#else /* BYTE_ORDER == LITTLE_ENDIAN */ +#define IP6F_OFF_MASK 0xf8ff /* mask out offset from _offlg */ +#define IP6F_RESERVED_MASK 0x0600 /* reserved bits in ip6f_offlg */ +#define IP6F_MORE_FRAG 0x0100 /* more-fragments flag */ +#endif + +#endif /* netinet/ip6.h */ diff --git a/include/netinet/ip_icmp.h b/include/netinet/ip_icmp.h index 1662e3f7f..be7959ee7 100644 --- a/include/netinet/ip_icmp.h +++ b/include/netinet/ip_icmp.h @@ -1 +1,287 @@ -#include <linux/icmp.h> +/* Copyright (C) 1991, 92, 93, 95, 96, 97 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +#ifndef __NETINET_IP_ICMP_H +#define __NETINET_IP_ICMP_H 1 + +#include <sys/cdefs.h> +#include <sys/types.h> + +__BEGIN_DECLS + +struct icmphdr +{ + u_int8_t type; /* message type */ + u_int8_t code; /* type sub-code */ + u_int16_t checksum; + union + { + struct + { + u_int16_t id; + u_int16_t sequence; + } echo; /* echo datagram */ + u_int32_t gateway; /* gateway address */ + struct + { + u_int16_t __unused; + u_int16_t mtu; + } frag; /* path mtu discovery */ + } un; +}; + +#define ICMP_ECHOREPLY 0 /* Echo Reply */ +#define ICMP_DEST_UNREACH 3 /* Destination Unreachable */ +#define ICMP_SOURCE_QUENCH 4 /* Source Quench */ +#define ICMP_REDIRECT 5 /* Redirect (change route) */ +#define ICMP_ECHO 8 /* Echo Request */ +#define ICMP_TIME_EXCEEDED 11 /* Time Exceeded */ +#define ICMP_PARAMETERPROB 12 /* Parameter Problem */ +#define ICMP_TIMESTAMP 13 /* Timestamp Request */ +#define ICMP_TIMESTAMPREPLY 14 /* Timestamp Reply */ +#define ICMP_INFO_REQUEST 15 /* Information Request */ +#define ICMP_INFO_REPLY 16 /* Information Reply */ +#define ICMP_ADDRESS 17 /* Address Mask Request */ +#define ICMP_ADDRESSREPLY 18 /* Address Mask Reply */ +#define NR_ICMP_TYPES 18 + + +/* Codes for UNREACH. */ +#define ICMP_NET_UNREACH 0 /* Network Unreachable */ +#define ICMP_HOST_UNREACH 1 /* Host Unreachable */ +#define ICMP_PROT_UNREACH 2 /* Protocol Unreachable */ +#define ICMP_PORT_UNREACH 3 /* Port Unreachable */ +#define ICMP_FRAG_NEEDED 4 /* Fragmentation Needed/DF set */ +#define ICMP_SR_FAILED 5 /* Source Route failed */ +#define ICMP_NET_UNKNOWN 6 +#define ICMP_HOST_UNKNOWN 7 +#define ICMP_HOST_ISOLATED 8 +#define ICMP_NET_ANO 9 +#define ICMP_HOST_ANO 10 +#define ICMP_NET_UNR_TOS 11 +#define ICMP_HOST_UNR_TOS 12 +#define ICMP_PKT_FILTERED 13 /* Packet filtered */ +#define ICMP_PREC_VIOLATION 14 /* Precedence violation */ +#define ICMP_PREC_CUTOFF 15 /* Precedence cut off */ +#define NR_ICMP_UNREACH 15 /* instead of hardcoding immediate value */ + +/* Codes for REDIRECT. */ +#define ICMP_REDIR_NET 0 /* Redirect Net */ +#define ICMP_REDIR_HOST 1 /* Redirect Host */ +#define ICMP_REDIR_NETTOS 2 /* Redirect Net for TOS */ +#define ICMP_REDIR_HOSTTOS 3 /* Redirect Host for TOS */ + +/* Codes for TIME_EXCEEDED. */ +#define ICMP_EXC_TTL 0 /* TTL count exceeded */ +#define ICMP_EXC_FRAGTIME 1 /* Fragment Reass time exceeded */ + + +#ifdef __USE_BSD +/* + * Copyright (c) 1982, 1986, 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)ip_icmp.h 8.1 (Berkeley) 6/10/93 + */ + +#include <netinet/in.h> +#include <netinet/ip.h> + +/* + * Internal of an ICMP Router Advertisement + */ +struct icmp_ra_addr +{ + u_int32_t ira_addr; + u_int32_t ira_preference; +}; + +struct icmp +{ + u_int8_t icmp_type; /* type of message, see below */ + u_int8_t icmp_code; /* type sub code */ + u_int16_t icmp_cksum; /* ones complement checksum of struct */ + union + { + u_char ih_pptr; /* ICMP_PARAMPROB */ + struct in_addr ih_gwaddr; /* gateway address */ + struct ih_idseq /* echo datagram */ + { + u_int16_t icd_id; + u_int16_t icd_seq; + } ih_idseq; + u_int32_t ih_void; + + /* ICMP_UNREACH_NEEDFRAG -- Path MTU Discovery (RFC1191) */ + struct ih_pmtu + { + u_int16_t ipm_void; + u_int16_t ipm_nextmtu; + } ih_pmtu; + + struct ih_rtradv + { + u_int8_t irt_num_addrs; + u_int8_t irt_wpa; + u_int16_t irt_lifetime; + } ih_rtradv; + } icmp_hun; +#define icmp_pptr icmp_hun.ih_pptr +#define icmp_gwaddr icmp_hun.ih_gwaddr +#define icmp_id icmp_hun.ih_idseq.icd_id +#define icmp_seq icmp_hun.ih_idseq.icd_seq +#define icmp_void icmp_hun.ih_void +#define icmp_pmvoid icmp_hun.ih_pmtu.ipm_void +#define icmp_nextmtu icmp_hun.ih_pmtu.ipm_nextmtu +#define icmp_num_addrs icmp_hun.ih_rtradv.irt_num_addrs +#define icmp_wpa icmp_hun.ih_rtradv.irt_wpa +#define icmp_lifetime icmp_hun.ih_rtradv.irt_lifetime + union + { + struct + { + u_int32_t its_otime; + u_int32_t its_rtime; + u_int32_t its_ttime; + } id_ts; + struct + { + struct ip idi_ip; + /* options and then 64 bits of data */ + } id_ip; + struct icmp_ra_addr id_radv; + u_int32_t id_mask; + u_int8_t id_data[1]; + } icmp_dun; +#define icmp_otime icmp_dun.id_ts.its_otime +#define icmp_rtime icmp_dun.id_ts.its_rtime +#define icmp_ttime icmp_dun.id_ts.its_ttime +#define icmp_ip icmp_dun.id_ip.idi_ip +#define icmp_radv icmp_dun.id_radv +#define icmp_mask icmp_dun.id_mask +#define icmp_data icmp_dun.id_data +}; + +/* + * Lower bounds on packet lengths for various types. + * For the error advice packets must first insure that the + * packet is large enough to contain the returned ip header. + * Only then can we do the check to see if 64 bits of packet + * data have been returned, since we need to check the returned + * ip header length. + */ +#define ICMP_MINLEN 8 /* abs minimum */ +#define ICMP_TSLEN (8 + 3 * sizeof (n_time)) /* timestamp */ +#define ICMP_MASKLEN 12 /* address mask */ +#define ICMP_ADVLENMIN (8 + sizeof (struct ip) + 8) /* min */ +#ifndef _IP_VHL +#define ICMP_ADVLEN(p) (8 + ((p)->icmp_ip.ip_hl << 2) + 8) + /* N.B.: must separately check that ip_hl >= 5 */ +#else +#define ICMP_ADVLEN(p) (8 + (IP_VHL_HL((p)->icmp_ip.ip_vhl) << 2) + 8) + /* N.B.: must separately check that header length >= 5 */ +#endif + +/* Definition of type and code fields. */ +/* defined above: ICMP_ECHOREPLY, ICMP_REDIRECT, ICMP_ECHO */ +#define ICMP_UNREACH 3 /* dest unreachable, codes: */ +#define ICMP_SOURCEQUENCH 4 /* packet lost, slow down */ +#define ICMP_ROUTERADVERT 9 /* router advertisement */ +#define ICMP_ROUTERSOLICIT 10 /* router solicitation */ +#define ICMP_TIMXCEED 11 /* time exceeded, code: */ +#define ICMP_PARAMPROB 12 /* ip header bad */ +#define ICMP_TSTAMP 13 /* timestamp request */ +#define ICMP_TSTAMPREPLY 14 /* timestamp reply */ +#define ICMP_IREQ 15 /* information request */ +#define ICMP_IREQREPLY 16 /* information reply */ +#define ICMP_MASKREQ 17 /* address mask request */ +#define ICMP_MASKREPLY 18 /* address mask reply */ + +#define ICMP_MAXTYPE 18 + +/* UNREACH codes */ +#define ICMP_UNREACH_NET 0 /* bad net */ +#define ICMP_UNREACH_HOST 1 /* bad host */ +#define ICMP_UNREACH_PROTOCOL 2 /* bad protocol */ +#define ICMP_UNREACH_PORT 3 /* bad port */ +#define ICMP_UNREACH_NEEDFRAG 4 /* IP_DF caused drop */ +#define ICMP_UNREACH_SRCFAIL 5 /* src route failed */ +#define ICMP_UNREACH_NET_UNKNOWN 6 /* unknown net */ +#define ICMP_UNREACH_HOST_UNKNOWN 7 /* unknown host */ +#define ICMP_UNREACH_ISOLATED 8 /* src host isolated */ +#define ICMP_UNREACH_NET_PROHIB 9 /* net denied */ +#define ICMP_UNREACH_HOST_PROHIB 10 /* host denied */ +#define ICMP_UNREACH_TOSNET 11 /* bad tos for net */ +#define ICMP_UNREACH_TOSHOST 12 /* bad tos for host */ +#define ICMP_UNREACH_FILTER_PROHIB 13 /* admin prohib */ +#define ICMP_UNREACH_HOST_PRECEDENCE 14 /* host prec vio. */ +#define ICMP_UNREACH_PRECEDENCE_CUTOFF 15 /* prec cutoff */ + +/* REDIRECT codes */ +#define ICMP_REDIRECT_NET 0 /* for network */ +#define ICMP_REDIRECT_HOST 1 /* for host */ +#define ICMP_REDIRECT_TOSNET 2 /* for tos and net */ +#define ICMP_REDIRECT_TOSHOST 3 /* for tos and host */ + +/* TIMEXCEED codes */ +#define ICMP_TIMXCEED_INTRANS 0 /* ttl==0 in transit */ +#define ICMP_TIMXCEED_REASS 1 /* ttl==0 in reass */ + +/* PARAMPROB code */ +#define ICMP_PARAMPROB_OPTABSENT 1 /* req. opt. absent */ + +#define ICMP_INFOTYPE(type) \ + ((type) == ICMP_ECHOREPLY || (type) == ICMP_ECHO || \ + (type) == ICMP_ROUTERADVERT || (type) == ICMP_ROUTERSOLICIT || \ + (type) == ICMP_TSTAMP || (type) == ICMP_TSTAMPREPLY || \ + (type) == ICMP_IREQ || (type) == ICMP_IREQREPLY || \ + (type) == ICMP_MASKREQ || (type) == ICMP_MASKREPLY) + +#endif /* __USE_BSD */ + +__END_DECLS + +#endif /* netinet/ip_icmp.h */ diff --git a/include/netinet/tcp.h b/include/netinet/tcp.h index 8636ec7d0..36976369e 100644 --- a/include/netinet/tcp.h +++ b/include/netinet/tcp.h @@ -1 +1,166 @@ -#include <netinet/ip_tcp.h> +/* + * Copyright (c) 1982, 1986, 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)tcp.h 8.1 (Berkeley) 6/10/93 + */ + +#ifndef _NETINET_TCP_H +#define _NETINET_TCP_H 1 + +#include <features.h> +#include <sys/types.h> + +__BEGIN_DECLS + +#ifdef __FAVOR_BSD +typedef u_int32_t tcp_seq; +/* + * TCP header. + * Per RFC 793, September, 1981. + */ +struct tcphdr + { + u_int16_t th_sport; /* source port */ + u_int16_t th_dport; /* destination port */ + tcp_seq th_seq; /* sequence number */ + tcp_seq th_ack; /* acknowledgement number */ +#if __BYTE_ORDER == __LITTLE_ENDIAN + u_int8_t th_x2:4; /* (unused) */ + u_int8_t th_off:4; /* data offset */ +#endif +#if __BYTE_ORDER == __BIG_ENDIAN + u_int8_t th_off:4; /* data offset */ + u_int8_t th_x2:4; /* (unused) */ +#endif + u_int8_t th_flags; +#define TH_FIN 0x01 +#define TH_SYN 0x02 +#define TH_RST 0x04 +#define TH_PUSH 0x08 +#define TH_ACK 0x10 +#define TH_URG 0x20 + u_int16_t th_win; /* window */ + u_int16_t th_sum; /* checksum */ + u_int16_t th_urp; /* urgent pointer */ +}; + +#else /* !__FAVOR_BSD */ +struct tcphdr + { + u_int16_t source; + u_int16_t dest; + u_int32_t seq; + u_int32_t ack_seq; +#if __BYTE_ORDER == __LITTLE_ENDIAN + u_int16_t res1:4; + u_int16_t doff:4; + u_int16_t fin:1; + u_int16_t syn:1; + u_int16_t rst:1; + u_int16_t psh:1; + u_int16_t ack:1; + u_int16_t urg:1; + u_int16_t res2:2; +#elif __BYTE_ORDER == __BIG_ENDIAN + u_int16_t doff:4; + u_int16_t res1:4; + u_int16_t res2:2; + u_int16_t urg:1; + u_int16_t ack:1; + u_int16_t psh:1; + u_int16_t rst:1; + u_int16_t syn:1; + u_int16_t fin:1; +#else +#error "Adjust your <bits/endian.h> defines" +#endif + u_int16_t window; + u_int16_t check; + u_int16_t urg_ptr; +}; +#endif /* __FAVOR_BSD */ + +enum +{ + TCP_ESTABLISHED = 1, + TCP_SYN_SENT, + TCP_SYN_RECV, + TCP_FIN_WAIT1, + TCP_FIN_WAIT2, + TCP_TIME_WAIT, + TCP_CLOSE, + TCP_CLOSE_WAIT, + TCP_LAST_ACK, + TCP_LISTEN, + TCP_CLOSING /* now a valid state */ +}; + +#define TCPOPT_EOL 0 +#define TCPOPT_NOP 1 +#define TCPOPT_MAXSEG 2 +#define TCPOLEN_MAXSEG 4 +#define TCPOPT_WINDOW 3 +#define TCPOLEN_WINDOW 3 +#define TCPOPT_SACK_PERMITTED 4 /* Experimental */ +#define TCPOLEN_SACK_PERMITTED 2 +#define TCPOPT_SACK 5 /* Experimental */ +#define TCPOPT_TIMESTAMP 8 +#define TCPOLEN_TIMESTAMP 10 +#define TCPOLEN_TSTAMP_APPA (TCPOLEN_TIMESTAMP+2) /* appendix A */ + +#define TCPOPT_TSTAMP_HDR \ + (TCPOPT_NOP<<24|TCPOPT_NOP<<16|TCPOPT_TIMESTAMP<<8|TCPOLEN_TIMESTAMP) + +/* + * Default maximum segment size for TCP. + * With an IP MSS of 576, this is 536, + * but 512 is probably more convenient. + * This should be defined as MIN(512, IP_MSS - sizeof (struct tcpiphdr)). + */ +#define TCP_MSS 512 + +#define TCP_MAXWIN 65535 /* largest value for (unscaled) window */ + +#define TCP_MAX_WINSHIFT 14 /* maximum window shift */ + +/* + * User-settable options (used with setsockopt). + */ +#define TCP_NODELAY 0x01 /* don't delay send to coalesce packets */ +#define TCP_MAXSEG 0x02 /* set maximum segment size */ +#define TCP_CORK 0x03 /* control sending of partial frames */ + +#define SOL_TCP 6 /* TCP level */ + +__END_DECLS + +#endif /* netinet/tcp.h */ diff --git a/include/netinet/udp.h b/include/netinet/udp.h index 39a57a3c7..9ee66f0fc 100644 --- a/include/netinet/udp.h +++ b/include/netinet/udp.h @@ -1 +1,55 @@ -#include <netinet/ip_udp.h> +/* Copyright (C) 1991, 92, 93, 95, 96, 97 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +/* + * Copyright (c) 1982, 1986 Regents of the University of California. + * All rights reserved. The Berkeley software License Agreement + * specifies the terms and conditions for redistribution. + */ + +#ifndef __NETINET_UDP_H +#define __NETINET_UDP_H 1 + +#include <sys/cdefs.h> +#include <sys/types.h> + +__BEGIN_DECLS + +/* UDP header as specified by RFC 768, August 1980. */ +#ifdef __FAVOR_BSD +struct udphdr { + u_int16_t uh_sport; /* source port */ + u_int16_t uh_dport; /* destination port */ + u_int16_t uh_ulen; /* udp length */ + u_int16_t uh_sum; /* udp checksum */ +}; +#else + +struct udphdr { + u_int16_t source; + u_int16_t dest; + u_int16_t len; + u_int16_t check; +}; +#endif + +#define SOL_UDP 17 /* sockopt level for UDP */ + +__END_DECLS + +#endif /* netinet/udp.h */ diff --git a/include/regex.h b/include/regex.h new file mode 100644 index 000000000..813882c42 --- /dev/null +++ b/include/regex.h @@ -0,0 +1,3754 @@ +#if !defined(_RX_H) || defined(RX_WANT_SE_DEFS) +#define _RX_H + +/* Copyright (C) 1992, 1993 Free Software Foundation, Inc. + +This file is part of the librx library. + +Librx is free software; you can redistribute it and/or modify it under +the terms of the GNU Library General Public License as published by +the Free Software Foundation; either version 2, or (at your option) +any later version. + +Librx is distributed in the hope that it will be useful, but WITHOUT +ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or +FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +for more details. + +You should have received a copy of the GNU Library General Public +License along with this software; see the file COPYING.LIB. If not, +write to the Free Software Foundation, 675 Mass Ave, Cambridge, MA +02139, USA. */ +/* t. lord Wed Sep 23 18:20:57 1992 */ + + + + +#include <features.h> + +#define __need_size_t +#include <stddef.h> + +#include <string.h> + +#if RX_WANT_SE_DEFS != 1 +__BEGIN_DECLS +#endif + +#ifndef RX_WANT_SE_DEFS + +/* This page: Bitsets */ + +#ifndef RX_subset +typedef unsigned int RX_subset; +#define RX_subset_bits (32) +#define RX_subset_mask (RX_subset_bits - 1) +#endif + +typedef RX_subset * rx_Bitset; + +#ifdef __STDC__ +typedef void (*rx_bitset_iterator) (void *, int member_index); +#else +typedef void (*rx_bitset_iterator) (); +#endif + +#define rx_bitset_subset(N) ((N) / RX_subset_bits) +#define rx_bitset_subset_val(B,N) ((B)[rx_bitset_subset(N)]) +#define RX_bitset_access(B,N,OP) \ + ((B)[rx_bitset_subset(N)] OP rx_subset_singletons[(N) & RX_subset_mask]) +#define RX_bitset_member(B,N) RX_bitset_access(B, N, &) +#define RX_bitset_enjoin(B,N) RX_bitset_access(B, N, |=) +#define RX_bitset_remove(B,N) RX_bitset_access(B, N, &= ~) +#define RX_bitset_toggle(B,N) RX_bitset_access(B, N, ^= ) +#define rx_bitset_numb_subsets(N) (((N) + RX_subset_bits - 1) / RX_subset_bits) +#define rx_sizeof_bitset(N) (rx_bitset_numb_subsets(N) * sizeof(RX_subset)) + + + +/* This page: Splay trees. */ + +#ifdef __STDC__ +typedef int (*rx_sp_comparer) (void * a, void * b); +#else +typedef int (*rx_sp_comparer) (); +#endif + +struct rx_sp_node +{ + void * key; + void * data; + struct rx_sp_node * kids[2]; +}; + +#ifdef __STDC__ +typedef void (*rx_sp_key_data_freer) (struct rx_sp_node *); +#else +typedef void (*rx_sp_key_data_freer) (); +#endif + + +/* giant inflatable hash trees */ + +struct rx_hash_item +{ + struct rx_hash_item * next_same_hash; + struct rx_hash * table; + unsigned long hash; + void * data; + void * binding; +}; + +struct rx_hash +{ + struct rx_hash * parent; + int refs; + struct rx_hash * children[13]; + struct rx_hash_item * buckets [13]; + int bucket_size [13]; +}; + +struct rx_hash_rules; + +#ifdef __STDC__ +/* should return like == */ +typedef int (*rx_hash_eq)(void *, void *); +typedef struct rx_hash * (*rx_alloc_hash)(struct rx_hash_rules *); +typedef void (*rx_free_hash)(struct rx_hash *, + struct rx_hash_rules *); +typedef struct rx_hash_item * (*rx_alloc_hash_item)(struct rx_hash_rules *, + void *); +typedef void (*rx_free_hash_item)(struct rx_hash_item *, + struct rx_hash_rules *); +#else +typedef int (*rx_hash_eq)(); +typedef struct rx_hash * (*rx_alloc_hash)(); +typedef void (*rx_free_hash)(); +typedef struct rx_hash_item * (*rx_alloc_hash_item)(); +typedef void (*rx_free_hash_item)(); +#endif + +struct rx_hash_rules +{ + rx_hash_eq eq; + rx_alloc_hash hash_alloc; + rx_free_hash free_hash; + rx_alloc_hash_item hash_item_alloc; + rx_free_hash_item free_hash_item; +}; + + +/* Forward declarations */ + +struct rx_cache; +struct rx_superset; +struct rx; +struct rx_se_list; + + + +/* + * GLOSSARY + * + * regexp + * regular expression + * expression + * pattern - a `regular' expression. The expression + * need not be formally regular -- it can contain + * constructs that don't correspond to purely regular + * expressions. + * + * buffer + * string - the string (or strings) being searched or matched. + * + * pattern buffer - a structure of type `struct re_pattern_buffer' + * This in turn contains a `struct rx', which holds the + * NFA compiled from a pattern, as well as some of the state + * of a matcher using the pattern. + * + * NFA - nondeterministic finite automata. Some people + * use this term to a member of the class of + * regular automata (those corresponding to a regular + * language). However, in this code, the meaning is + * more general. The automata used by Rx are comperable + * in power to what are usually called `push down automata'. + * + * Two NFA are built by rx for every pattern. One is built + * by the compiler. The other is built from the first, on + * the fly, by the matcher. The latter is called the `superstate + * NFA' because its states correspond to sets of states from + * the first NFA. (Joe Keane gets credit for the name + * `superstate NFA'). + * + * NFA edges + * epsilon edges + * side-effect edges - The NFA compiled from a pattern can have three + * kinds of edges. Epsilon edges can be taken freely anytime + * their source state is reached. Character set edges can be + * taken when their source state is reached and when the next + * character in the buffer is a member of the set. Side effect + * edges imply a transition that can only be taken after the + * indicated side effect has been successfully accomplished. + * Some examples of side effects are: + * + * Storing the current match position to record the + * location of a parentesized subexpression. + * + * Advancing the matcher over N characters if they + * match the N characters previously matched by a + * parentesized subexpression. + * + * Both of those kinds of edges occur in the NFA generated + * by the pattern: \(.\)\1 + * + * Epsilon and side effect edges are similar. Unfortunately, + * some of the code uses the name `epsilon edge' to mean + * both epsilon and side effect edges. For example, the + * function has_non_idempotent_epsilon_path computes the existance + * of a non-trivial path containing only a mix of epsilon and + * side effect edges. In that case `nonidempotent epsilon' is being + * used to mean `side effect'. + */ + + + + + +/* LOW LEVEL PATTERN BUFFERS */ + +/* Suppose that from some NFA state, more than one path through + * side-effect edges is possible. In what order should the paths + * be tried? A function of type rx_se_list_order answers that + * question. It compares two lists of side effects, and says + * which list comes first. + */ + +#ifdef __STDC__ +typedef int (*rx_se_list_order) (struct rx *, + struct rx_se_list *, + struct rx_se_list *); +#else +typedef int (*rx_se_list_order) (); +#endif + + + +/* Struct RX holds a compiled regular expression - that is, an nfa + * ready to be converted on demand to a more efficient superstate nfa. + * This is for the low level interface. The high-level interfaces enclose + * this in a `struct re_pattern_buffer'. + */ +struct rx +{ + /* The compiler assigns a unique id to every pattern. + * Like sequence numbers in X, there is a subtle bug here + * if you use Rx in a system that runs for a long time. + * But, because of the way the caches work out, it is almost + * impossible to trigger the Rx version of this bug. + * + * The id is used to validate superstates found in a cache + * of superstates. It isn't sufficient to let a superstate + * point back to the rx for which it was compiled -- the caller + * may be re-using a `struct rx' in which case the superstate + * is not really valid. So instead, superstates are validated + * by checking the sequence number of the pattern for which + * they were built. + */ + int rx_id; + + /* This is memory mgt. state for superstates. This may be + * shared by more than one struct rx. + */ + struct rx_cache * cache; + + /* Every regex defines the size of its own character set. + * A superstate has an array of this size, with each element + * a `struct rx_inx'. So, don't make this number too large. + * In particular, don't make it 2^16. + */ + int local_cset_size; + + /* After the NFA is built, it is copied into a contiguous region + * of memory (mostly for compatability with GNU regex). + * Here is that region, and it's size: + */ + void * buffer; + unsigned long allocated; + + /* Clients of RX can ask for some extra storage in the space pointed + * to by BUFFER. The field RESERVED is an input parameter to the + * compiler. After compilation, this much space will be available + * at (buffer + allocated - reserved) + */ + unsigned long reserved; + + /* --------- The remaining fields are for internal use only. --------- */ + /* --------- But! they must be initialized to 0. --------- */ + + /* NODEC is the number of nodes in the NFA with non-epsilon + * transitions. + */ + int nodec; + + /* EPSNODEC is the number of nodes with only epsilon transitions. */ + int epsnodec; + + /* The sum (NODEC + EPSNODEC) is the total number of states in the + * compiled NFA. + */ + + /* Lists of side effects as stored in the NFA are `hash consed'..meaning + * that lists with the same elements are ==. During compilation, + * this table facilitates hash-consing. + */ + struct rx_hash se_list_memo; + + /* Lists of NFA states are also hashed. + */ + struct rx_hash set_list_memo; + + + + + /* The compiler and matcher must build a number of instruction frames. + * The format of these frames is fixed (c.f. struct rx_inx). The values + * of the instructions is not fixed. + * + * An enumerated type (enum rx_opcode) defines the set of instructions + * that the compiler or matcher might generate. When filling an instruction + * frame, the INX field is found by indexing this instruction table + * with an opcode: + */ + void ** instruction_table; + + /* The list of all states in an NFA. + * During compilation, the NEXT field of NFA states links this list. + * After compilation, all the states are compacted into an array, + * ordered by state id numbers. At that time, this points to the base + * of that array. + */ + struct rx_nfa_state *nfa_states; + + /* Every nfa begins with one distinguished starting state: + */ + struct rx_nfa_state *start; + + /* This orders the search through super-nfa paths. + * See the comment near the typedef of rx_se_list_order. + */ + rx_se_list_order se_list_cmp; + + struct rx_superset * start_set; +}; + + + + +/* SYNTAX TREES */ + +/* Compilation is in stages. + * + * In the first stage, a pattern specified by a string is + * translated into a syntax tree. Later stages will convert + * the syntax tree into an NFA optimized for conversion to a + * superstate-NFA. + * + * This page is about syntax trees. + */ + +enum rexp_node_type +{ + r_cset, /* Match from a character set. `a' or `[a-z]'*/ + r_concat, /* Concat two subexpressions. `ab' */ + r_alternate, /* Choose one of two subexpressions. `a\|b' */ + r_opt, /* Optional subexpression. `a?' */ + r_star, /* Repeated subexpression. `a*' */ + + + /* A 2phase-star is a variation on a repeated subexpression. + * In this case, there are two subexpressions. The first, if matched, + * begins a repitition (otherwise, the whole expression is matches the + * empth string). + * + * After matching the first subexpression, a 2phase star either finishes, + * or matches the second subexpression. If the second subexpression is + * matched, then the whole construct repeats. + * + * 2phase stars are used in two circumstances. First, they + * are used as part of the implementation of POSIX intervals (counted + * repititions). Second, they are used to implement proper star + * semantics when the repeated subexpression contains paths of + * only side effects. See rx_compile for more information. + */ + r_2phase_star, + + + /* c.f. "typedef void * rx_side_effect" */ + r_side_effect, + + /* This is an extension type: It is for transient use in source->source + * transformations (implemented over syntax trees). + */ + r_data +}; + +/* A side effect is a matcher-specific action associated with + * transitions in the NFA. The details of side effects are up + * to the matcher. To the compiler and superstate constructors + * side effects are opaque: + */ + +typedef void * rx_side_effect; + +/* Nodes in a syntax tree are of this type: + */ +struct rexp_node +{ + enum rexp_node_type type; + union + { + rx_Bitset cset; + rx_side_effect side_effect; + struct + { + struct rexp_node *left; + struct rexp_node *right; + } pair; + void * data; + } params; +}; + + + +/* NFA + * + * A syntax tree is compiled into an NFA. This page defines the structure + * of that NFA. + */ + +struct rx_nfa_state +{ + /* These are kept in a list as the NFA is being built. */ + struct rx_nfa_state *next; + + /* After the NFA is built, states are given integer id's. + * States whose outgoing transitions are all either epsilon or + * side effect edges are given ids less than 0. Other states + * are given successive non-negative ids starting from 0. + */ + int id; + + /* The list of NFA edges that go from this state to some other. */ + struct rx_nfa_edge *edges; + + /* If you land in this state, then you implicitly land + * in all other states reachable by only epsilon translations. + * Call the set of maximal paths to such states the epsilon closure + * of this state. + * + * There may be other states that are reachable by a mixture of + * epsilon and side effect edges. Consider the set of maximal paths + * of that sort from this state. Call it the epsilon-side-effect + * closure of the state. + * + * The epsilon closure of the state is a subset of the epsilon-side- + * effect closure. It consists of all the paths that contain + * no side effects -- only epsilon edges. + * + * The paths in the epsilon-side-effect closure can be partitioned + * into equivalance sets. Two paths are equivalant if they have the + * same set of side effects, in the same order. The epsilon-closure + * is one of these equivalance sets. Let's call these equivalance + * sets: observably equivalant path sets. That name is chosen + * because equivalance of two paths means they cause the same side + * effects -- so they lead to the same subsequent observations other + * than that they may wind up in different target states. + * + * The superstate nfa, which is derived from this nfa, is based on + * the observation that all of the paths in an observably equivalant + * path set can be explored at the same time, provided that the + * matcher keeps track not of a single nfa state, but of a set of + * states. In particular, after following all the paths in an + * observably equivalant set, you wind up at a set of target states. + * That set of target states corresponds to one state in the + * superstate NFA. + * + * Staticly, before matching begins, it is convenient to analyze the + * nfa. Each state is labeled with a list of the observably + * equivalant path sets who's union covers all the + * epsilon-side-effect paths beginning in this state. This list is + * called the possible futures of the state. + * + * A trivial example is this NFA: + * s1 + * A ---> B + * + * s2 + * ---> C + * + * epsilon s1 + * ---------> D ------> E + * + * + * In this example, A has two possible futures. + * One invokes the side effect `s1' and contains two paths, + * one ending in state B, the other in state E. + * The other invokes the side effect `s2' and contains only + * one path, landing in state C. + */ + struct rx_possible_future *futures; + + + /* There are exactly two distinguished states in every NFA: */ + unsigned int is_final:1; + unsigned int is_start:1; + + /* These are used during NFA construction... */ + unsigned int eclosure_needed:1; + unsigned int mark:1; +}; + + +/* An edge in an NFA is typed: */ +enum rx_nfa_etype +{ + /* A cset edge is labled with a set of characters one of which + * must be matched for the edge to be taken. + */ + ne_cset, + + /* An epsilon edge is taken whenever its starting state is + * reached. + */ + ne_epsilon, + + /* A side effect edge is taken whenever its starting state is + * reached. Side effects may cause the match to fail or the + * position of the matcher to advance. + */ + ne_side_effect /* A special kind of epsilon. */ +}; + +struct rx_nfa_edge +{ + struct rx_nfa_edge *next; + enum rx_nfa_etype type; + struct rx_nfa_state *dest; + union + { + rx_Bitset cset; + rx_side_effect side_effect; + } params; +}; + + + +/* A possible future consists of a list of side effects + * and a set of destination states. Below are their + * representations. These structures are hash-consed which + * means that lists with the same elements share a representation + * (their addresses are ==). + */ + +struct rx_nfa_state_set +{ + struct rx_nfa_state * car; + struct rx_nfa_state_set * cdr; +}; + +struct rx_se_list +{ + rx_side_effect car; + struct rx_se_list * cdr; +}; + +struct rx_possible_future +{ + struct rx_possible_future *next; + struct rx_se_list * effects; + struct rx_nfa_state_set * destset; +}; + + + +/* This begins the description of the superstate NFA. + * + * The superstate NFA corresponds to the NFA in these ways: + * + * Every superstate NFA states SUPER correspond to sets of NFA states, + * nfa_states(SUPER). + * + * Superstate edges correspond to NFA paths. + * + * The superstate has no epsilon transitions; + * every edge has a character label, and a (possibly empty) side + * effect label. The side effect label corresponds to a list of + * side effects that occur in the NFA. These parts are referred + * to as: superedge_character(EDGE) and superedge_sides(EDGE). + * + * For a superstate edge EDGE starting in some superstate SUPER, + * the following is true (in pseudo-notation :-): + * + * exists DEST in nfa_states s.t. + * exists nfaEDGE in nfa_edges s.t. + * origin (nfaEDGE) == DEST + * && origin (nfaEDGE) is a member of nfa_states(SUPER) + * && exists PF in possible_futures(dest(nfaEDGE)) s.t. + * sides_of_possible_future (PF) == superedge_sides (EDGE) + * + * also: + * + * let SUPER2 := superedge_destination(EDGE) + * nfa_states(SUPER2) + * == union of all nfa state sets S s.t. + * exists PF in possible_futures(dest(nfaEDGE)) s.t. + * sides_of_possible_future (PF) == superedge_sides (EDGE) + * && S == dests_of_possible_future (PF) } + * + * Or in english, every superstate is a set of nfa states. A given + * character and a superstate implies many transitions in the NFA -- + * those that begin with an edge labeled with that character from a + * state in the set corresponding to the superstate. + * + * The destinations of those transitions each have a set of possible + * futures. A possible future is a list of side effects and a set of + * destination NFA states. Two sets of possible futures can be + * `merged' by combining all pairs of possible futures that have the + * same side effects. A pair is combined by creating a new future + * with the same side effect but the union of the two destination sets. + * In this way, all the possible futures suggested by a superstate + * and a character can be merged into a set of possible futures where + * no two elements of the set have the same set of side effects. + * + * The destination of a possible future, being a set of NFA states, + * corresponds to a supernfa state. So, the merged set of possible + * futures we just created can serve as a set of edges in the + * supernfa. + * + * The representation of the superstate nfa and the nfa is critical. + * The nfa has to be compact, but has to facilitate the rapid + * computation of missing superstates. The superstate nfa has to + * be fast to interpret, lazilly constructed, and bounded in space. + * + * To facilitate interpretation, the superstate data structures are + * peppered with `instruction frames'. There is an instruction set + * defined below which matchers using the supernfa must be able to + * interpret. + * + * We'd like to make it possible but not mandatory to use code + * addresses to represent instructions (c.f. gcc's computed goto). + * Therefore, we define an enumerated type of opcodes, and when + * writing one of these instructions into a data structure, use + * the opcode as an index into a table of instruction values. + * + * Here are the opcodes that occur in the superstate nfa: + */ + + +/* Every superstate contains a table of instruction frames indexed + * by characters. A normal `move' in a matcher is to fetch the next + * character and use it as an index into a superstates transition + * table. + * + * In the fasted case, only one edge follows from that character. + * In other cases there is more work to do. + * + * The descriptions of the opcodes refer to data structures that are + * described further below. + */ + +enum rx_opcode +{ + /* + * BACKTRACK_POINT is invoked when a character transition in + * a superstate leads to more than one edge. In that case, + * the edges have to be explored independently using a backtracking + * strategy. + * + * A BACKTRACK_POINT instruction is stored in a superstate's + * transition table for some character when it is known that that + * character crosses more than one edge. On encountering this + * instruction, the matcher saves enough state to backtrack to this + * point in the match later. + */ + rx_backtrack_point = 0, /* data is (struct transition_class *) */ + + /* + * RX_DO_SIDE_EFFECTS evaluates the side effects of an epsilon path. + * There is one occurence of this instruction per rx_distinct_future. + * This instruction is skipped if a rx_distinct_future has no side effects. + */ + rx_do_side_effects = rx_backtrack_point + 1, + + /* data is (struct rx_distinct_future *) */ + + /* + * RX_CACHE_MISS instructions are stored in rx_distinct_futures whose + * destination superstate has been reclaimed (or was never built). + * It recomputes the destination superstate. + * RX_CACHE_MISS is also stored in a superstate transition table before + * any of its edges have been built. + */ + rx_cache_miss = rx_do_side_effects + 1, + /* data is (struct rx_distinct_future *) */ + + /* + * RX_NEXT_CHAR is called to consume the next character and take the + * corresponding transition. This is the only instruction that uses + * the DATA field of the instruction frame instead of DATA_2. + * (see EXPLORE_FUTURE in regex.c). + */ + rx_next_char = rx_cache_miss + 1, /* data is (struct superstate *) */ + + /* RX_BACKTRACK indicates that a transition fails. + */ + rx_backtrack = rx_next_char + 1, /* no data */ + + /* + * RX_ERROR_INX is stored only in places that should never be executed. + */ + rx_error_inx = rx_backtrack + 1, /* Not supposed to occur. */ + + rx_num_instructions = rx_error_inx + 1 +}; + +/* An id_instruction_table holds the values stored in instruction + * frames. The table is indexed by the enums declared above. + */ +extern void * rx_id_instruction_table[rx_num_instructions]; + +/* The heart of the matcher is a `word-code-interpreter' + * (like a byte-code interpreter, except that instructions + * are a full word wide). + * + * Instructions are not stored in a vector of code, instead, + * they are scattered throughout the data structures built + * by the regexp compiler and the matcher. One word-code instruction, + * together with the arguments to that instruction, constitute + * an instruction frame (struct rx_inx). + * + * This structure type is padded by hand to a power of 2 because + * in one of the dominant cases, we dispatch by indexing a table + * of instruction frames. If that indexing can be accomplished + * by just a shift of the index, we're happy. + * + * Instructions take at most one argument, but there are two + * slots in an instruction frame that might hold that argument. + * These are called data and data_2. The data slot is only + * used for one instruction (RX_NEXT_CHAR). For all other + * instructions, data should be set to 0. + * + * RX_NEXT_CHAR is the most important instruction by far. + * By reserving the data field for its exclusive use, + * instruction dispatch is sped up in that case. There is + * no need to fetch both the instruction and the data, + * only the data is needed. In other words, a `cycle' begins + * by fetching the field data. If that is non-0, then it must + * be the destination state of a next_char transition, so + * make that value the current state, advance the match position + * by one character, and start a new cycle. On the other hand, + * if data is 0, fetch the instruction and do a more complicated + * dispatch on that. + */ + +struct rx_inx +{ + void * data; + void * data_2; + void * inx; + void * fnord; +}; + +#ifndef RX_TAIL_ARRAY +#define RX_TAIL_ARRAY 1 +#endif + +/* A superstate corresponds to a set of nfa states. Those sets are + * represented by STRUCT RX_SUPERSET. The constructors + * guarantee that only one (shared) structure is created for a given set. + */ +struct rx_superset +{ + int refs; /* This is a reference counted structure. */ + + /* We keep these sets in a cache because (in an unpredictable way), + * the same set is often created again and again. But that is also + * problematic -- compatibility with POSIX and GNU regex requires + * that we not be able to tell when a program discards a particular + * NFA (thus invalidating the supersets created from it). + * + * But when a cache hit appears to occur, we will have in hand the + * nfa for which it may have happened. That is why every nfa is given + * its own sequence number. On a cache hit, the cache is validated + * by comparing the nfa sequence number to this field: + */ + int id; + + struct rx_nfa_state * car; /* May or may not be a valid addr. */ + struct rx_superset * cdr; + + /* If the corresponding superstate exists: */ + struct rx_superstate * superstate; + + + /* There is another bookkeeping problem. It is expensive to + * compute the starting nfa state set for an nfa. So, once computed, + * it is cached in the `struct rx'. + * + * But, the state set can be flushed from the superstate cache. + * When that happens, we can't know if the corresponding `struct rx' + * is still alive or if it has been freed or re-used by the program. + * So, the cached pointer to this set in a struct rx might be invalid + * and we need a way to validate it. + * + * Fortunately, even if this set is flushed from the cache, it is + * not freed. It just goes on the free-list of supersets. + * So we can still examine it. + * + * So to validate a starting set memo, check to see if the + * starts_for field still points back to the struct rx in question, + * and if the ID matches the rx sequence number. + */ + struct rx * starts_for; + + /* This is used to link into a hash bucket so these objects can + * be `hash-consed'. + */ + struct rx_hash_item hash_item; +}; + +#define rx_protect_superset(RX,CON) (++(CON)->refs) + +/* The terminology may be confusing (rename this structure?). + * Every character occurs in at most one rx_super_edge per super-state. + * But, that structure might have more than one option, indicating a point + * of non-determinism. + * + * In other words, this structure holds a list of superstate edges + * sharing a common starting state and character label. The edges + * are in the field OPTIONS. All superstate edges sharing the same + * starting state and character are in this list. + */ +struct rx_super_edge +{ + struct rx_super_edge *next; + struct rx_inx rx_backtrack_frame; + int cset_size; + rx_Bitset cset; + struct rx_distinct_future *options; +}; + +/* A superstate is a set of nfa states (RX_SUPERSET) along + * with a transition table. Superstates are built on demand and reclaimed + * without warning. To protect a superstate from this ghastly fate, + * use LOCK_SUPERSTATE. + */ +struct rx_superstate +{ + int rx_id; /* c.f. the id field of rx_superset */ + int locks; /* protection from reclamation */ + + /* Within a superstate cache, all the superstates are kept in a big + * queue. The tail of the queue is the state most likely to be + * reclaimed. The *recyclable fields hold the queue position of + * this state. + */ + struct rx_superstate * next_recyclable; + struct rx_superstate * prev_recyclable; + + /* The supernfa edges that exist in the cache and that have + * this state as their destination are kept in this list: + */ + struct rx_distinct_future * transition_refs; + + /* The list of nfa states corresponding to this superstate: */ + struct rx_superset * contents; + + /* The list of edges in the cache beginning from this state. */ + struct rx_super_edge * edges; + + /* A tail of the recyclable queue is marked as semifree. A semifree + * state has no incoming next_char transitions -- any transition + * into a semifree state causes a complex dispatch with the side + * effect of rescuing the state from its semifree state. + * + * An alternative to this might be to make next_char more expensive, + * and to move a state to the head of the recyclable queue whenever + * it is entered. That way, popular states would never be recycled. + * + * But unilaterally making next_char more expensive actually loses. + * So, incoming transitions are only made expensive for states near + * the tail of the recyclable queue. The more cache contention + * there is, the more frequently a state will have to prove itself + * and be moved back to the front of the queue. If there is less + * contention, then popular states just aggregate in the front of + * the queue and stay there. + */ + int is_semifree; + + + /* This keeps track of the size of the transition table for this + * state. There is a half-hearted attempt to support variable sized + * superstates. + */ + int trans_size; + + /* Indexed by characters... */ + struct rx_inx transitions[RX_TAIL_ARRAY]; +}; + + +/* A list of distinct futures define the edges that leave from a + * given superstate on a given character. c.f. rx_super_edge. + */ + +struct rx_distinct_future +{ + struct rx_distinct_future * next_same_super_edge[2]; + struct rx_distinct_future * next_same_dest; + struct rx_distinct_future * prev_same_dest; + struct rx_superstate * present; /* source state */ + struct rx_superstate * future; /* destination state */ + struct rx_super_edge * edge; + + + /* The future_frame holds the instruction that should be executed + * after all the side effects are done, when it is time to complete + * the transition to the next state. + * + * Normally this is a next_char instruction, but it may be a + * cache_miss instruction as well, depending on whether or not + * the superstate is in the cache and semifree. + * + * If this is the only future for a given superstate/char, and + * if there are no side effects to be performed, this frame is + * not used (directly) at all. Instead, its contents are copied + * into the transition table of the starting state of this dist. future. + */ + struct rx_inx future_frame; + + struct rx_inx side_effects_frame; + struct rx_se_list * effects; +}; + +#define rx_lock_superstate(R,S) ((S)->locks++) +#define rx_unlock_superstate(R,S) (--(S)->locks) + + +/* This page destined for rx.h */ + +struct rx_blocklist +{ + struct rx_blocklist * next; + int bytes; +}; + +struct rx_freelist +{ + struct rx_freelist * next; +}; + +struct rx_cache; + +#ifdef __STDC__ +typedef void (*rx_morecore_fn)(struct rx_cache *); +#else +typedef void (*rx_morecore_fn)(); +#endif + +/* You use this to control the allocation of superstate data + * during matching. Most of it should be initialized to 0. + * + * A MORECORE function is necessary. It should allocate + * a new block of memory or return 0. + * A default that uses malloc is called `rx_morecore'. + * + * The number of SUPERSTATES_ALLOWED indirectly limits how much memory + * the system will try to allocate. The default is 128. Batch style + * applications that are very regexp intensive should use as high a number + * as possible without thrashing. + * + * The LOCAL_CSET_SIZE is the number of characters in a character set. + * It is therefore the number of entries in a superstate transition table. + * Generally, it should be 256. If your character set has 16 bits, + * it is better to translate your regexps into equivalent 8 bit patterns. + */ + +struct rx_cache +{ + struct rx_hash_rules superset_hash_rules; + + /* Objects are allocated by incrementing a pointer that + * scans across rx_blocklists. + */ + struct rx_blocklist * memory; + struct rx_blocklist * memory_pos; + int bytes_left; + char * memory_addr; + rx_morecore_fn morecore; + + /* Freelists. */ + struct rx_freelist * free_superstates; + struct rx_freelist * free_transition_classes; + struct rx_freelist * free_discernable_futures; + struct rx_freelist * free_supersets; + struct rx_freelist * free_hash; + + /* Two sets of superstates -- those that are semifreed, and those + * that are being used. + */ + struct rx_superstate * lru_superstate; + struct rx_superstate * semifree_superstate; + + struct rx_superset * empty_superset; + + int superstates; + int semifree_superstates; + int hits; + int misses; + int superstates_allowed; + + int local_cset_size; + void ** instruction_table; + + struct rx_hash superset_table; +}; + + + +/* The lowest-level search function supports arbitrarily fragmented + * strings and (optionally) suspendable/resumable searches. + * + * Callers have to provide a few hooks. + */ + +#ifndef __GNUC__ +#ifdef __STDC__ +#define __const__ const +#else +#define __const__ +#endif +#endif + +/* This holds a matcher position */ +struct rx_string_position +{ + __const__ unsigned char * pos; /* The current pos. */ + __const__ unsigned char * string; /* The current string burst. */ + __const__ unsigned char * end; /* First invalid position >= POS. */ + int offset; /* Integer address of the current burst. */ + int size; /* Current string's size. */ + int search_direction; /* 1 or -1 */ + int search_end; /* First position to not try. */ +}; + + +enum rx_get_burst_return +{ + rx_get_burst_continuation, + rx_get_burst_error, + rx_get_burst_ok, + rx_get_burst_no_more +}; + + +/* A call to get burst should make POS valid. It might be invalid + * if the STRING field doesn't point to a burst that actually + * contains POS. + * + * GET_BURST should take a clue from SEARCH_DIRECTION (1 or -1) as to + * whether or not to pad to the left. Padding to the right is always + * appropriate, but need not go past the point indicated by STOP. + * + * If a continuation is returned, then the reentering call to + * a search function will retry the get_burst. + */ + +#ifdef __STDC__ +typedef enum rx_get_burst_return + (*rx_get_burst_fn) (struct rx_string_position * pos, + void * app_closure, + int stop); + +#else +typedef enum rx_get_burst_return (*rx_get_burst_fn) (); +#endif + + +enum rx_back_check_return +{ + rx_back_check_continuation, + rx_back_check_error, + rx_back_check_pass, + rx_back_check_fail +}; + +/* Back_check should advance the position it is passed + * over rparen - lparen characters and return pass iff + * the characters starting at POS match those indexed + * by [LPAREN..RPAREN]. + * + * If a continuation is returned, then the reentering call to + * a search function will retry the back_check. + */ + +#ifdef __STDC__ +typedef enum rx_back_check_return + (*rx_back_check_fn) (struct rx_string_position * pos, + int lparen, + int rparen, + unsigned char * translate, + void * app_closure, + int stop); + +#else +typedef enum rx_back_check_return (*rx_back_check_fn) (); +#endif + + + + +/* A call to fetch_char should return the character at POS or POS + 1. + * Returning continuations here isn't supported. OFFSET is either 0 or 1 + * and indicates which characters is desired. + */ + +#ifdef __STDC__ +typedef int (*rx_fetch_char_fn) (struct rx_string_position * pos, + int offset, + void * app_closure, + int stop); +#else +typedef int (*rx_fetch_char_fn) (); +#endif + + +enum rx_search_return +{ + rx_search_continuation = -4, + rx_search_error = -3, + rx_search_soft_fail = -2, /* failed by running out of string */ + rx_search_fail = -1 /* failed only by reaching failure states */ + /* return values >= 0 indicate the position of a successful match */ +}; + + + + + + +/* regex.h + * + * The remaining declarations replace regex.h. + */ + +/* This is an array of error messages corresponding to the error codes. + */ +extern __const__ char *re_error_msg[]; + +/* If any error codes are removed, changed, or added, update the + `re_error_msg' table in regex.c. */ +typedef enum +{ + REG_NOERROR = 0, /* Success. */ + REG_NOMATCH, /* Didn't find a match (for regexec). */ + + /* POSIX regcomp return error codes. (In the order listed in the + standard.) */ + REG_BADPAT, /* Invalid pattern. */ + REG_ECOLLATE, /* Not implemented. */ + REG_ECTYPE, /* Invalid character class name. */ + REG_EESCAPE, /* Trailing backslash. */ + REG_ESUBREG, /* Invalid back reference. */ + REG_EBRACK, /* Unmatched left bracket. */ + REG_EPAREN, /* Parenthesis imbalance. */ + REG_EBRACE, /* Unmatched \{. */ + REG_BADBR, /* Invalid contents of \{\}. */ + REG_ERANGE, /* Invalid range end. */ + REG_ESPACE, /* Ran out of memory. */ + REG_BADRPT, /* No preceding re for repetition op. */ + + /* Error codes we've added. */ + REG_EEND, /* Premature end. */ + REG_ESIZE, /* Compiled pattern bigger than 2^16 bytes. */ + REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */ +} reg_errcode_t; + +/* The regex.c support, as a client of rx, defines a set of possible + * side effects that can be added to the edge lables of nfa edges. + * Here is the list of sidef effects in use. + */ + +enum re_side_effects +{ +#define RX_WANT_SE_DEFS 1 +#undef RX_DEF_SE +#undef RX_DEF_CPLX_SE +#define RX_DEF_SE(IDEM, NAME, VALUE) NAME VALUE, +#define RX_DEF_CPLX_SE(IDEM, NAME, VALUE) NAME VALUE, +#include <regex.h> +#undef RX_DEF_SE +#undef RX_DEF_CPLX_SE +#undef RX_WANT_SE_DEFS + re_floogle_flap = 65533 +}; + +/* These hold paramaters for the kinds of side effects that are possible + * in the supported pattern languages. These include things like the + * numeric bounds of {} operators and the index of paren registers for + * subexpression measurement or backreferencing. + */ +struct re_se_params +{ + enum re_side_effects se; + int op1; + int op2; +}; + +typedef unsigned reg_syntax_t; + +struct re_pattern_buffer +{ + struct rx rx; + reg_syntax_t syntax; /* See below for syntax bit definitions. */ + + unsigned int no_sub:1; /* If set, don't return register offsets. */ + unsigned int not_bol:1; /* If set, the anchors ('^' and '$') don't */ + unsigned int not_eol:1; /* match at the ends of the string. */ + unsigned int newline_anchor:1;/* If true, an anchor at a newline matches.*/ + unsigned int least_subs:1; /* If set, and returning registers, return + * as few values as possible. Only + * backreferenced groups and group 0 (the whole + * match) will be returned. + */ + + /* If true, this says that the matcher should keep registers on its + * backtracking stack. For many patterns, we can easily determine that + * this isn't necessary. + */ + unsigned int match_regs_on_stack:1; + unsigned int search_regs_on_stack:1; + + /* is_anchored and begbuf_only are filled in by rx_compile. */ + unsigned int is_anchored:1; /* Anchorded by ^? */ + unsigned int begbuf_only:1; /* Anchored to char position 0? */ + + + /* If REGS_UNALLOCATED, allocate space in the `regs' structure + * for `max (RE_NREGS, re_nsub + 1)' groups. + * If REGS_REALLOCATE, reallocate space if necessary. + * If REGS_FIXED, use what's there. + */ +#define REGS_UNALLOCATED 0 +#define REGS_REALLOCATE 1 +#define REGS_FIXED 2 + unsigned int regs_allocated:2; + + + /* Either a translate table to apply to all characters before + * comparing them, or zero for no translation. The translation + * is applied to a pattern when it is compiled and to a string + * when it is matched. + */ + unsigned char * translate; + + /* If this is a valid pointer, it tells rx not to store the extents of + * certain subexpressions (those corresponding to non-zero entries). + * Passing 0x1 is the same as passing an array of all ones. Passing 0x0 + * is the same as passing an array of all zeros. + * The array should contain as many entries as their are subexps in the + * regexp. + * + * For POSIX compatability, when using regcomp and regexec this field + * is zeroed and ignored. + */ + char * syntax_parens; + + /* Number of subexpressions found by the compiler. */ + size_t re_nsub; + + void * buffer; /* Malloced memory for the nfa. */ + unsigned long allocated; /* Size of that memory. */ + + /* Pointer to a fastmap, if any, otherwise zero. re_search uses + * the fastmap, if there is one, to skip over impossible + * starting points for matches. */ + char *fastmap; + + unsigned int fastmap_accurate:1; /* These three are internal. */ + unsigned int can_match_empty:1; + struct rx_nfa_state * start; /* The nfa starting state. */ + + /* This is the list of iterator bounds for {lo,hi} constructs. + * The memory pointed to is part of the rx->buffer. + */ + struct re_se_params *se_params; + + /* This is a bitset representation of the fastmap. + * This is a true fastmap that already takes the translate + * table into account. + */ + rx_Bitset fastset; +}; + +/* Type for byte offsets within the string. POSIX mandates this. */ +typedef int regoff_t; + +/* This is the structure we store register match data in. See + regex.texinfo for a full description of what registers match. */ +struct re_registers +{ + unsigned num_regs; + regoff_t *start; + regoff_t *end; +}; + +typedef struct re_pattern_buffer regex_t; + +/* POSIX specification for registers. Aside from the different names than + `re_registers', POSIX uses an array of structures, instead of a + structure of arrays. */ +typedef struct +{ + regoff_t rm_so; /* Byte offset from string's start to substring's start. */ + regoff_t rm_eo; /* Byte offset from string's start to substring's end. */ +} regmatch_t; + + +/* The following bits are used to determine the regexp syntax we + recognize. The set/not-set meanings are chosen so that Emacs syntax + remains the value 0. The bits are given in alphabetical order, and + the definitions shifted by one from the previous bit; thus, when we + add or remove a bit, only one other definition need change. */ + +/* If this bit is not set, then \ inside a bracket expression is literal. + If set, then such a \ quotes the following character. */ +#define RE_BACKSLASH_ESCAPE_IN_LISTS (1) + +/* If this bit is not set, then + and ? are operators, and \+ and \? are + literals. + If set, then \+ and \? are operators and + and ? are literals. */ +#define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1) + +/* If this bit is set, then character classes are supported. They are: + [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:], + [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:]. + If not set, then character classes are not supported. */ +#define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1) + +/* If this bit is set, then ^ and $ are always anchors (outside bracket + expressions, of course). + If this bit is not set, then it depends: + ^ is an anchor if it is at the beginning of a regular + expression or after an open-group or an alternation operator; + $ is an anchor if it is at the end of a regular expression, or + before a close-group or an alternation operator. + + This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because + POSIX draft 11.2 says that * etc. in leading positions is undefined. + We already implemented a previous draft which made those constructs + invalid, though, so we haven't changed the code back. */ +#define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1) + +/* If this bit is set, then special characters are always special + regardless of where they are in the pattern. + If this bit is not set, then special characters are special only in + some contexts; otherwise they are ordinary. Specifically, + * + ? and intervals are only special when not after the beginning, + open-group, or alternation operator. */ +#define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1) + +/* If this bit is set, then *, +, ?, and { cannot be first in an re or + immediately after an alternation or begin-group operator. */ +#define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1) + +/* If this bit is set, then . matches newline. + If not set, then it doesn't. */ +#define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1) + +/* If this bit is set, then . doesn't match NUL. + If not set, then it does. */ +#define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1) + +/* If this bit is set, nonmatching lists [^...] do not match newline. + If not set, they do. */ +#define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1) + +/* If this bit is set, either \{...\} or {...} defines an + interval, depending on RE_NO_BK_BRACES. + If not set, \{, \}, {, and } are literals. */ +#define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1) + +/* If this bit is set, +, ? and | aren't recognized as operators. + If not set, they are. */ +#define RE_LIMITED_OPS (RE_INTERVALS << 1) + +/* If this bit is set, newline is an alternation operator. + If not set, newline is literal. */ +#define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1) + +/* If this bit is set, then `{...}' defines an interval, and \{ and \} + are literals. + If not set, then `\{...\}' defines an interval. */ +#define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1) + +/* If this bit is set, (...) defines a group, and \( and \) are literals. + If not set, \(...\) defines a group, and ( and ) are literals. */ +#define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1) + +/* If this bit is set, then \<digit> matches <digit>. + If not set, then \<digit> is a back-reference. */ +#define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1) + +/* If this bit is set, then | is an alternation operator, and \| is literal. + If not set, then \| is an alternation operator, and | is literal. */ +#define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1) + +/* If this bit is set, then an ending range point collating higher + than the starting range point, as in [z-a], is invalid. + If not set, then when ending range point collates higher than the + starting range point, the range is ignored. */ +#define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1) + +/* If this bit is set, then an unmatched ) is ordinary. + If not set, then an unmatched ) is invalid. */ +#define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1) + +/* If this bit is set, do not process the GNU regex operators. + IF not set, then the GNU regex operators are recognized. */ +#define RE_NO_GNU_OPS (RE_UNMATCHED_RIGHT_PAREN_ORD << 1) + +/* This global variable defines the particular regexp syntax to use (for + some interfaces). When a regexp is compiled, the syntax used is + stored in the pattern buffer, so changing this does not affect + already-compiled regexps. */ +extern reg_syntax_t re_syntax_options; + +/* Define combinations of the above bits for the standard possibilities. + (The [[[ comments delimit what gets put into the Texinfo file, so + don't delete them!) */ +/* [[[begin syntaxes]]] */ +#define RE_SYNTAX_EMACS 0 + +#define RE_SYNTAX_AWK \ + (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \ + | RE_NO_BK_PARENS | RE_NO_BK_REFS \ + | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \ + | RE_DOT_NEWLINE \ + | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS) + +#define RE_SYNTAX_GNU_AWK \ + ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS) \ + & ~(RE_DOT_NOT_NULL|RE_INTERVALS)) + +#define RE_SYNTAX_POSIX_AWK \ + (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS | RE_NO_GNU_OPS) + +#define RE_SYNTAX_GREP \ + (RE_BK_PLUS_QM | RE_CHAR_CLASSES \ + | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \ + | RE_NEWLINE_ALT) + +#define RE_SYNTAX_EGREP \ + (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \ + | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \ + | RE_NEWLINE_ALT | RE_NO_BK_PARENS \ + | RE_NO_BK_VBAR) + +#define RE_SYNTAX_POSIX_EGREP \ + (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES) + +/* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */ +#define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC + +#define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC + +/* Syntax bits common to both basic and extended POSIX regex syntax. */ +#define _RE_SYNTAX_POSIX_COMMON \ + (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \ + | RE_INTERVALS | RE_NO_EMPTY_RANGES) + +#define RE_SYNTAX_POSIX_BASIC \ + (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM) + +/* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes + RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this + isn't minimal, since other operators, such as \`, aren't disabled. */ +#define RE_SYNTAX_POSIX_MINIMAL_BASIC \ + (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS) + +#define RE_SYNTAX_POSIX_EXTENDED \ + (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ + | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \ + | RE_NO_BK_PARENS | RE_NO_BK_VBAR \ + | RE_UNMATCHED_RIGHT_PAREN_ORD) + +/* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INVALID_OPS + replaces RE_CONTEXT_INDEP_OPS and RE_NO_BK_REFS is added. */ +#define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \ + (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ + | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \ + | RE_NO_BK_PARENS | RE_NO_BK_REFS \ + | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD) +/* [[[end syntaxes]]] */ + +/* Maximum number of duplicates an interval can allow. Some systems + (erroneously) define this in other header files, but we want our + value, so remove any previous define. */ +#ifdef RE_DUP_MAX +#undef RE_DUP_MAX +#endif +/* if sizeof(int) == 2, then ((1 << 15) - 1) overflows */ +#define RE_DUP_MAX (0x7fff) + + +/* POSIX `cflags' bits (i.e., information for `regcomp'). */ + +/* If this bit is set, then use extended regular expression syntax. + If not set, then use basic regular expression syntax. */ +#define REG_EXTENDED 1 + +/* If this bit is set, then ignore case when matching. + If not set, then case is significant. */ +#define REG_ICASE (REG_EXTENDED << 1) + +/* If this bit is set, then anchors do not match at newline + characters in the string. + If not set, then anchors do match at newlines. */ +#define REG_NEWLINE (REG_ICASE << 1) + +/* If this bit is set, then report only success or fail in regexec. + If not set, then returns differ between not matching and errors. */ +#define REG_NOSUB (REG_NEWLINE << 1) + + +/* POSIX `eflags' bits (i.e., information for regexec). */ + +/* If this bit is set, then the beginning-of-line operator doesn't match + the beginning of the string (presumably because it's not the + beginning of a line). + If not set, then the beginning-of-line operator does match the + beginning of the string. */ +#define REG_NOTBOL 1 + +/* Like REG_NOTBOL, except for the end-of-line. */ +#define REG_NOTEOL (1 << 1) + +/* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer, + * `re_match_2' returns information about at least this many registers + * the first time a `regs' structure is passed. + * + * Also, this is the greatest number of backreferenced subexpressions + * allowed in a pattern being matched without caller-supplied registers. + */ +#ifndef RE_NREGS +#define RE_NREGS 30 +#endif + +extern int rx_cache_bound; +extern char rx_version_string[]; + + + +#ifdef RX_WANT_RX_DEFS + +/* This is decls to the interesting subsystems and lower layers + * of rx. Everything which doesn't have a public counterpart in + * regex.c is declared here. + */ + + +#ifdef __STDC__ +typedef void (*rx_hash_freefn) (struct rx_hash_item * it); +#else /* ndef __STDC__ */ +typedef void (*rx_hash_freefn) (); +#endif /* ndef __STDC__ */ + + + + +#ifdef __STDC__ +RX_DECL int rx_bitset_is_equal (int size, rx_Bitset a, rx_Bitset b); +RX_DECL int rx_bitset_is_subset (int size, rx_Bitset a, rx_Bitset b); +RX_DECL int rx_bitset_empty (int size, rx_Bitset set); +RX_DECL void rx_bitset_null (int size, rx_Bitset b); +RX_DECL void rx_bitset_universe (int size, rx_Bitset b); +RX_DECL void rx_bitset_complement (int size, rx_Bitset b); +RX_DECL void rx_bitset_assign (int size, rx_Bitset a, rx_Bitset b); +RX_DECL void rx_bitset_union (int size, rx_Bitset a, rx_Bitset b); +RX_DECL void rx_bitset_intersection (int size, + rx_Bitset a, rx_Bitset b); +RX_DECL void rx_bitset_difference (int size, rx_Bitset a, rx_Bitset b); +RX_DECL void rx_bitset_revdifference (int size, + rx_Bitset a, rx_Bitset b); +RX_DECL void rx_bitset_xor (int size, rx_Bitset a, rx_Bitset b); +RX_DECL unsigned long rx_bitset_hash (int size, rx_Bitset b); +RX_DECL struct rx_hash_item * rx_hash_find (struct rx_hash * table, + unsigned long hash, + void * value, + struct rx_hash_rules * rules); +RX_DECL struct rx_hash_item * rx_hash_store (struct rx_hash * table, + unsigned long hash, + void * value, + struct rx_hash_rules * rules); +RX_DECL void rx_hash_free (struct rx_hash_item * it, struct rx_hash_rules * rules); +RX_DECL void rx_free_hash_table (struct rx_hash * tab, rx_hash_freefn freefn, + struct rx_hash_rules * rules); +RX_DECL rx_Bitset rx_cset (struct rx *rx); +RX_DECL rx_Bitset rx_copy_cset (struct rx *rx, rx_Bitset a); +RX_DECL void rx_free_cset (struct rx * rx, rx_Bitset c); +RX_DECL struct rexp_node * rexp_node (struct rx *rx, + enum rexp_node_type type); +RX_DECL struct rexp_node * rx_mk_r_cset (struct rx * rx, + rx_Bitset b); +RX_DECL struct rexp_node * rx_mk_r_concat (struct rx * rx, + struct rexp_node * a, + struct rexp_node * b); +RX_DECL struct rexp_node * rx_mk_r_alternate (struct rx * rx, + struct rexp_node * a, + struct rexp_node * b); +RX_DECL struct rexp_node * rx_mk_r_opt (struct rx * rx, + struct rexp_node * a); +RX_DECL struct rexp_node * rx_mk_r_star (struct rx * rx, + struct rexp_node * a); +RX_DECL struct rexp_node * rx_mk_r_2phase_star (struct rx * rx, + struct rexp_node * a, + struct rexp_node * b); +RX_DECL struct rexp_node * rx_mk_r_side_effect (struct rx * rx, + rx_side_effect a); +RX_DECL struct rexp_node * rx_mk_r_data (struct rx * rx, + void * a); +RX_DECL void rx_free_rexp (struct rx * rx, struct rexp_node * node); +RX_DECL struct rexp_node * rx_copy_rexp (struct rx *rx, + struct rexp_node *node); +RX_DECL struct rx_nfa_state * rx_nfa_state (struct rx *rx); +RX_DECL void rx_free_nfa_state (struct rx_nfa_state * n); +RX_DECL struct rx_nfa_state * rx_id_to_nfa_state (struct rx * rx, + int id); +RX_DECL struct rx_nfa_edge * rx_nfa_edge (struct rx *rx, + enum rx_nfa_etype type, + struct rx_nfa_state *start, + struct rx_nfa_state *dest); +RX_DECL void rx_free_nfa_edge (struct rx_nfa_edge * e); +RX_DECL void rx_free_nfa (struct rx *rx); +RX_DECL int rx_build_nfa (struct rx *rx, + struct rexp_node *rexp, + struct rx_nfa_state **start, + struct rx_nfa_state **end); +RX_DECL void rx_name_nfa_states (struct rx *rx); +RX_DECL int rx_eclose_nfa (struct rx *rx); +RX_DECL void rx_delete_epsilon_transitions (struct rx *rx); +RX_DECL int rx_compactify_nfa (struct rx *rx, + void **mem, unsigned long *size); +RX_DECL void rx_release_superset (struct rx *rx, + struct rx_superset *set); +RX_DECL struct rx_superset * rx_superset_cons (struct rx * rx, + struct rx_nfa_state *car, struct rx_superset *cdr); +RX_DECL struct rx_superset * rx_superstate_eclosure_union + (struct rx * rx, struct rx_superset *set, struct rx_nfa_state_set *ecl); +RX_DECL struct rx_superstate * rx_superstate (struct rx *rx, + struct rx_superset *set); +RX_DECL struct rx_inx * rx_handle_cache_miss + (struct rx *rx, struct rx_superstate *super, unsigned char chr, void *data); +RX_DECL reg_errcode_t rx_compile (__const__ char *pattern, int size, + reg_syntax_t syntax, + struct re_pattern_buffer * rxb); +RX_DECL void rx_blow_up_fastmap (struct re_pattern_buffer * rxb); +#else /* STDC */ +RX_DECL int rx_bitset_is_equal (); +RX_DECL int rx_bitset_is_subset (); +RX_DECL int rx_bitset_empty (); +RX_DECL void rx_bitset_null (); +RX_DECL void rx_bitset_universe (); +RX_DECL void rx_bitset_complement (); +RX_DECL void rx_bitset_assign (); +RX_DECL void rx_bitset_union (); +RX_DECL void rx_bitset_intersection (); +RX_DECL void rx_bitset_difference (); +RX_DECL void rx_bitset_revdifference (); +RX_DECL void rx_bitset_xor (); +RX_DECL unsigned long rx_bitset_hash (); +RX_DECL struct rx_hash_item * rx_hash_find (); +RX_DECL struct rx_hash_item * rx_hash_store (); +RX_DECL void rx_hash_free (); +RX_DECL void rx_free_hash_table (); +RX_DECL rx_Bitset rx_cset (); +RX_DECL rx_Bitset rx_copy_cset (); +RX_DECL void rx_free_cset (); +RX_DECL struct rexp_node * rexp_node (); +RX_DECL struct rexp_node * rx_mk_r_cset (); +RX_DECL struct rexp_node * rx_mk_r_concat (); +RX_DECL struct rexp_node * rx_mk_r_alternate (); +RX_DECL struct rexp_node * rx_mk_r_opt (); +RX_DECL struct rexp_node * rx_mk_r_star (); +RX_DECL struct rexp_node * rx_mk_r_2phase_star (); +RX_DECL struct rexp_node * rx_mk_r_side_effect (); +RX_DECL struct rexp_node * rx_mk_r_data (); +RX_DECL void rx_free_rexp (); +RX_DECL struct rexp_node * rx_copy_rexp (); +RX_DECL struct rx_nfa_state * rx_nfa_state (); +RX_DECL void rx_free_nfa_state (); +RX_DECL struct rx_nfa_state * rx_id_to_nfa_state (); +RX_DECL struct rx_nfa_edge * rx_nfa_edge (); +RX_DECL void rx_free_nfa_edge (); +RX_DECL void rx_free_nfa (); +RX_DECL int rx_build_nfa (); +RX_DECL void rx_name_nfa_states (); +RX_DECL int rx_eclose_nfa (); +RX_DECL void rx_delete_epsilon_transitions (); +RX_DECL int rx_compactify_nfa (); +RX_DECL void rx_release_superset (); +RX_DECL struct rx_superset * rx_superset_cons (); +RX_DECL struct rx_superset * rx_superstate_eclosure_union (); +RX_DECL struct rx_superstate * rx_superstate (); +RX_DECL struct rx_inx * rx_handle_cache_miss (); +RX_DECL reg_errcode_t rx_compile (); +RX_DECL void rx_blow_up_fastmap (); +#endif /* STDC */ + + +#endif /* RX_WANT_RX_DEFS */ + + + +#ifdef __STDC__ +extern int re_search_2 (struct re_pattern_buffer *rxb, + __const__ char * string1, int size1, + __const__ char * string2, int size2, + int startpos, int range, + struct re_registers *regs, + int stop); +extern int re_search (struct re_pattern_buffer * rxb, __const__ char *string, + int size, int startpos, int range, + struct re_registers *regs); +extern int re_match_2 (struct re_pattern_buffer * rxb, + __const__ char * string1, int size1, + __const__ char * string2, int size2, + int pos, struct re_registers *regs, int stop); +extern int re_match (struct re_pattern_buffer * rxb, + __const__ char * string, + int size, int pos, + struct re_registers *regs); +extern reg_syntax_t re_set_syntax (reg_syntax_t syntax); +extern void re_set_registers (struct re_pattern_buffer *bufp, + struct re_registers *regs, + unsigned num_regs, + regoff_t * starts, regoff_t * ends); +extern __const__ char * re_compile_pattern (__const__ char *pattern, + int length, + struct re_pattern_buffer * rxb); +extern int re_compile_fastmap (struct re_pattern_buffer * rxb); +extern char * re_comp (__const__ char *s); +extern int re_exec (__const__ char *s); +extern int regcomp (regex_t * preg, __const__ char * pattern, int cflags); +extern int regexec (__const__ regex_t *preg, __const__ char *string, + size_t nmatch, regmatch_t pmatch[], + int eflags); +extern size_t regerror (int errcode, __const__ regex_t *preg, + char *errbuf, size_t errbuf_size); +extern void regfree (regex_t *preg); + +#else /* STDC */ +extern int re_search_2 (); +extern int re_search (); +extern int re_match_2 (); +extern int re_match (); +extern reg_syntax_t re_set_syntax (); +extern void re_set_registers (); +extern __const__ char * re_compile_pattern (); +extern int re_compile_fastmap (); +extern char * re_comp (); +extern int re_exec (); +extern int regcomp (); +extern int regexec (); +extern size_t regerror (); +extern void regfree (); + +#endif /* STDC */ + + + +#ifdef RX_WANT_RX_DEFS + +struct rx_counter_frame +{ + int tag; + int val; + struct rx_counter_frame * inherited_from; /* If this is a copy. */ + struct rx_counter_frame * cdr; +}; + +struct rx_backtrack_frame +{ + char * counter_stack_sp; + + /* A frame is used to save the matchers state when it crosses a + * backtracking point. The `stk_' fields correspond to variables + * in re_search_2 (just strip off thes `stk_'). They are documented + * tere. + */ + struct rx_superstate * stk_super; + unsigned int stk_c; + struct rx_string_position stk_test_pos; + int stk_last_l; + int stk_last_r; + int stk_test_ret; + + /* This is the list of options left to explore at the backtrack + * point for which this frame was created. + */ + struct rx_distinct_future * df; + struct rx_distinct_future * first_df; + +#ifdef RX_DEBUG + int stk_line_no; +#endif +}; + +struct rx_stack_chunk +{ + struct rx_stack_chunk * next_chunk; + int bytes_left; + char * sp; +}; + +enum rx_outer_entry +{ + rx_outer_start, + rx_outer_fastmap, + rx_outer_test, + rx_outer_restore_pos +}; + +enum rx_fastmap_return +{ + rx_fastmap_continuation, + rx_fastmap_error, + rx_fastmap_ok, + rx_fastmap_fail +}; + +enum rx_fastmap_entry +{ + rx_fastmap_start, + rx_fastmap_string_break +}; + +enum rx_test_return +{ + rx_test_continuation, + rx_test_error, + rx_test_fail, + rx_test_ok +}; + +enum rx_test_internal_return +{ + rx_test_internal_error, + rx_test_found_first, + rx_test_line_finished +}; + +enum rx_test_match_entry +{ + rx_test_start, + rx_test_cache_hit_loop, + rx_test_backreference_check, + rx_test_backtrack_return +}; + +struct rx_search_state +{ + /* Two groups of registers are kept. The group with the register state + * of the current test match, and the group that holds the state at the end + * of the best known match, if any. + * + * For some patterns, there may also be registers saved on the stack. + */ + unsigned num_regs; /* Includes an element for register zero. */ + regoff_t * lparen; /* scratch space for register returns */ + regoff_t * rparen; + regoff_t * best_lpspace; /* in case the user doesn't want these */ + regoff_t * best_rpspace; /* values, we still need space to store + * them. Normally, this memoryis unused + * and the space pointed to by REGS is + * used instead. + */ + + int last_l; /* Highest index of a valid lparen. */ + int last_r; /* It's dual. */ + + int * best_lparen; /* This contains the best known register */ + int * best_rparen; /* assignments. + * This may point to the same mem as + * best_lpspace, or it might point to memory + * passed by the caller. + */ + int best_last_l; /* best_last_l:best_lparen::last_l:lparen */ + int best_last_r; + + + unsigned char * translate; + + struct rx_string_position outer_pos; + + struct rx_superstate * start_super; + int nfa_choice; + int first_found; /* If true, return after finding any match. */ + int ret_val; + + /* For continuations... */ + enum rx_outer_entry outer_search_resume_pt; + struct re_pattern_buffer * saved_rxb; + int saved_startpos; + int saved_range; + int saved_stop; + int saved_total_size; + rx_get_burst_fn saved_get_burst; + rx_back_check_fn saved_back_check; + struct re_registers * saved_regs; + + /** + ** state for fastmap + **/ + char * fastmap; + int fastmap_chr; + int fastmap_val; + + /* for continuations in the fastmap procedure: */ + enum rx_fastmap_entry fastmap_resume_pt; + + /** + ** state for test_match + **/ + + /* The current superNFA position of the matcher. */ + struct rx_superstate * super; + + /* The matcher interprets a series of instruction frames. + * This is the `instruction counter' for the interpretation. + */ + struct rx_inx * ifr; + + /* We insert a ghost character in the string to prime + * the nfa. test_pos.pos, test_pos.str_half, and test_pos.end_half + * keep track of the test-match position and string-half. + */ + unsigned char c; + + /* Position within the string. */ + struct rx_string_position test_pos; + + struct rx_stack_chunk * counter_stack; + struct rx_stack_chunk * backtrack_stack; + int backtrack_frame_bytes; + int chunk_bytes; + struct rx_stack_chunk * free_chunks; + + /* To return from this function, set test_ret and + * `goto test_do_return'. + * + * Possible return values are: + * 1 --- end of string while the superNFA is still going + * 0 --- internal error (out of memory) + * -1 --- search completed by reaching the superNFA fail state + * -2 --- a match was found, maybe not the longest. + * + * When the search is complete (-1), best_last_r indicates whether + * a match was found. + * + * -2 is return only if search_state.first_found is non-zero. + * + * if search_state.first_found is non-zero, a return of -1 indicates no match, + * otherwise, best_last_r has to be checked. + */ + int test_ret; + + int could_have_continued; + +#ifdef RX_DEBUG + int backtrack_depth; + /* There is a search tree with every node as set of deterministic + * transitions in the super nfa. For every branch of a + * backtrack point is an edge in the tree. + * This counts up a pre-order of nodes in that tree. + * It's saved on the search stack and printed when debugging. + */ + int line_no; + int lines_found; +#endif + + + /* For continuations within the match tester */ + enum rx_test_match_entry test_match_resume_pt; + struct rx_inx * saved_next_tr_table; + struct rx_inx * saved_this_tr_table; + int saved_reg; + struct rx_backtrack_frame * saved_bf; + +}; +static __inline__ void init_fastmap( struct re_pattern_buffer *, + struct rx_search_state * ); + + +extern char rx_slowmap[]; +extern unsigned char rx_id_translation[]; + +static __inline__ void +init_fastmap( struct re_pattern_buffer * rxb, + struct rx_search_state * search_state ) +{ + search_state->fastmap = (rxb->fastmap + ? (char *)rxb->fastmap + : (char *)rx_slowmap); + /* Update the fastmap now if not correct already. + * When the regexp was compiled, the fastmap was computed + * and stored in a bitset. This expands the bitset into a + * character array containing 1s and 0s. + */ + if ((search_state->fastmap == rxb->fastmap) && !rxb->fastmap_accurate) + rx_blow_up_fastmap (rxb); + search_state->fastmap_chr = -1; + search_state->fastmap_val = 0; + search_state->fastmap_resume_pt = rx_fastmap_start; +} + +static __inline__ void +uninit_fastmap ( struct re_pattern_buffer * rxb, + struct rx_search_state * search_state ) +{ + /* Unset the fastmap sentinel */ + if (search_state->fastmap_chr >= 0) + search_state->fastmap[search_state->fastmap_chr] + = search_state->fastmap_val; +} + +static __inline__ int +fastmap_search ( struct re_pattern_buffer * rxb, int stop, + rx_get_burst_fn get_burst, void * app_closure, + struct rx_search_state * search_state ) +{ + enum rx_fastmap_entry pc; + + if (0) + { + return_continuation: + search_state->fastmap_resume_pt = pc; + return rx_fastmap_continuation; + } + + pc = search_state->fastmap_resume_pt; + + switch (pc) + { + default: + return rx_fastmap_error; + case rx_fastmap_start: + init_fastmap_sentinal: + /* For the sake of fast fastmapping, set a sentinal in the fastmap. + * This sentinal will trap the fastmap loop when it reaches the last + * valid character in a string half. + * + * This must be reset when the fastmap/search loop crosses a string + * boundry, and before returning to the caller. So sometimes, + * the fastmap loop is restarted with `continue', othertimes by + * `goto init_fastmap_sentinal'. + */ + if (search_state->outer_pos.size) + { + search_state->fastmap_chr = ((search_state->outer_pos.search_direction == 1) + ? *(search_state->outer_pos.end - 1) + : *search_state->outer_pos.string); + search_state->fastmap_val + = search_state->fastmap[search_state->fastmap_chr]; + search_state->fastmap[search_state->fastmap_chr] = 1; + } + else + { + search_state->fastmap_chr = -1; + search_state->fastmap_val = 0; + } + + if (search_state->outer_pos.pos >= search_state->outer_pos.end) + goto fastmap_hit_bound; + else + { + if (search_state->outer_pos.search_direction == 1) + { + if (search_state->fastmap_val) + { + for (;;) + { + while (!search_state->fastmap[*search_state->outer_pos.pos]) + ++search_state->outer_pos.pos; + return rx_fastmap_ok; + } + } + else + { + for (;;) + { + while (!search_state->fastmap[*search_state->outer_pos.pos]) + ++search_state->outer_pos.pos; + if (*search_state->outer_pos.pos != search_state->fastmap_chr) + return rx_fastmap_ok; + else + { + ++search_state->outer_pos.pos; + if (search_state->outer_pos.pos == search_state->outer_pos.end) + goto fastmap_hit_bound; + } + } + } + } + else + { + __const__ unsigned char * bound; + bound = search_state->outer_pos.string - 1; + if (search_state->fastmap_val) + { + for (;;) + { + while (!search_state->fastmap[*search_state->outer_pos.pos]) + --search_state->outer_pos.pos; + return rx_fastmap_ok; + } + } + else + { + for (;;) + { + while (!search_state->fastmap[*search_state->outer_pos.pos]) + --search_state->outer_pos.pos; + if ((*search_state->outer_pos.pos != search_state->fastmap_chr) || search_state->fastmap_val) + return rx_fastmap_ok; + else + { + --search_state->outer_pos.pos; + if (search_state->outer_pos.pos == bound) + goto fastmap_hit_bound; + } + } + } + } + } + + case rx_fastmap_string_break: + fastmap_hit_bound: + { + /* If we hit a bound, it may be time to fetch another burst + * of string, or it may be time to return a continuation to + * the caller, or it might be time to fail. + */ + + int burst_state; + burst_state = get_burst (&search_state->outer_pos, app_closure, stop); + switch (burst_state) + { + default: + case rx_get_burst_error: + return rx_fastmap_error; + case rx_get_burst_continuation: + { + pc = rx_fastmap_string_break; + goto return_continuation; + } + case rx_get_burst_ok: + goto init_fastmap_sentinal; + case rx_get_burst_no_more: + /* ...not a string split, simply no more string. + * + * When searching backward, running out of string + * is reason to quit. + * + * When searching forward, we allow the possibility + * of an (empty) match after the last character in the + * virtual string. So, fall through to the matcher + */ + return ( (search_state->outer_pos.search_direction == 1) + ? rx_fastmap_ok + : rx_fastmap_fail); + } + } + } + +} + + + +#ifdef emacs +/* The `emacs' switch turns on certain matching commands + * that make sense only in Emacs. + */ +#include "config.h" +#include "lisp.h" +#include "buffer.h" +#include "syntax.h" +#endif /* emacs */ + +/* Setting RX_MEMDBUG is useful if you have dbmalloc. Maybe with similar + * packages too. + */ +#ifdef RX_MEMDBUG +#include <malloc.h> +#endif /* RX_RX_MEMDBUG */ + +/* We used to test for `BSTRING' here, but only GCC and Emacs define + * `BSTRING', as far as I know, and neither of them use this code. + */ +#if HAVE_STRING_H || __STDC__ +#include <string.h> + +#ifndef bcmp +#define bcmp(s1, s2, n) memcmp ((s1), (s2), (n)) +#endif + +#ifndef bcopy +#define bcopy(s, d, n) memcpy ((d), (s), (n)) +#endif + +#ifndef bzero +#define bzero(s, n) memset ((s), 0, (n)) +#endif + +#else /* HAVE_STRING_H || __STDC__ */ +#include <strings.h> +#endif /* not (HAVE_STRING_H || __STDC__) */ + +#ifdef __STDC__ +#include <stdlib.h> +#else /* not __STDC__ */ +char *malloc (); +char *realloc (); +#endif /* not __STDC__ */ + + + + +/* How many characters in the character set. */ +#define CHAR_SET_SIZE (1 << CHARBITS) + +#ifndef emacs +/* Define the syntax basics for \<, \>, etc. + * This must be nonzero for the wordchar and notwordchar pattern + * commands in re_match_2. + */ +#ifndef Sword +#define Sword 1 +#endif +#define SYNTAX(c) re_syntax_table[c] +RX_DECL char re_syntax_table[CHAR_SET_SIZE]; +#endif /* not emacs */ + + +/* Test if at very beginning or at very end of the virtual concatenation + * of `string1' and `string2'. If only one string, it's `string2'. + */ + +#define AT_STRINGS_BEG() \ + ( -1 \ + == ((search_state.test_pos.pos - search_state.test_pos.string) \ + + search_state.test_pos.offset)) + +#define AT_STRINGS_END() \ + ( (total_size - 1) \ + == ((search_state.test_pos.pos - search_state.test_pos.string) \ + + search_state.test_pos.offset)) + + +/* Test if POS + 1 points to a character which is word-constituent. We have + * two special cases to check for: if past the end of string1, look at + * the first character in string2; and if before the beginning of + * string2, look at the last character in string1. + * + * Assumes `string1' exists, so use in conjunction with AT_STRINGS_BEG (). + */ +#define LETTER_P(POS,OFF) \ + ( SYNTAX (fetch_char(POS, OFF, app_closure, stop)) \ + == Sword) + +/* Test if the character at D and the one after D differ with respect + * to being word-constituent. + */ +#define AT_WORD_BOUNDARY(d) \ + (AT_STRINGS_BEG () || AT_STRINGS_END () || LETTER_P (d,0) != LETTER_P (d, 1)) + + +#ifdef RX_SUPPORT_CONTINUATIONS +#define RX_STACK_ALLOC(BYTES) malloc(BYTES) +#define RX_STACK_FREE(MEM) free(MEM) +#else +#define RX_STACK_ALLOC(BYTES) alloca(BYTES) +#define RX_STACK_FREE(MEM) \ + ((struct rx_stack_chunk *)MEM)->next_chunk = search_state.free_chunks; \ + search_state.free_chunks = ((struct rx_stack_chunk *)MEM); + +#endif + +#define PUSH(CHUNK_VAR,BYTES) \ + if (!CHUNK_VAR || (CHUNK_VAR->bytes_left < (BYTES))) \ + { \ + struct rx_stack_chunk * new_chunk; \ + if (search_state.free_chunks) \ + { \ + new_chunk = search_state.free_chunks; \ + search_state.free_chunks = search_state.free_chunks->next_chunk; \ + } \ + else \ + { \ + new_chunk = (struct rx_stack_chunk *)RX_STACK_ALLOC(search_state.chunk_bytes); \ + if (!new_chunk) \ + { \ + search_state.ret_val = 0; \ + goto test_do_return; \ + } \ + } \ + new_chunk->sp = (char *)new_chunk + sizeof (struct rx_stack_chunk); \ + new_chunk->bytes_left = (search_state.chunk_bytes \ + - (BYTES) \ + - sizeof (struct rx_stack_chunk)); \ + new_chunk->next_chunk = CHUNK_VAR; \ + CHUNK_VAR = new_chunk; \ + } \ + else \ + (CHUNK_VAR->sp += (BYTES)), (CHUNK_VAR->bytes_left -= (BYTES)) + +#define POP(CHUNK_VAR,BYTES) \ + if (CHUNK_VAR->sp == ((char *)CHUNK_VAR + sizeof(*CHUNK_VAR))) \ + { \ + struct rx_stack_chunk * new_chunk = CHUNK_VAR->next_chunk; \ + RX_STACK_FREE(CHUNK_VAR); \ + CHUNK_VAR = new_chunk; \ + } \ + else \ + (CHUNK_VAR->sp -= BYTES), (CHUNK_VAR->bytes_left += BYTES) + + + +#define SRCH_TRANSLATE(C) search_state.translate[(unsigned char) (C)] + + + + +#ifdef __STDC__ +RX_DECL __inline__ int +rx_search (struct re_pattern_buffer * rxb, + int startpos, + int range, + int stop, + int total_size, + rx_get_burst_fn get_burst, + rx_back_check_fn back_check, + rx_fetch_char_fn fetch_char, + void * app_closure, + struct re_registers * regs, + struct rx_search_state * resume_state, + struct rx_search_state * save_state) +#else +RX_DECL __inline__ int +rx_search (rxb, startpos, range, stop, total_size, + get_burst, back_check, fetch_char, + app_closure, regs, resume_state, save_state) + struct re_pattern_buffer * rxb; + int startpos; + int range; + int stop; + int total_size; + rx_get_burst_fn get_burst; + rx_back_check_fn back_check; + rx_fetch_char_fn fetch_char; + void * app_closure; + struct re_registers * regs; + struct rx_search_state * resume_state; + struct rx_search_state * save_state; +#endif +{ + int pc; + int test_state; + struct rx_search_state search_state; + + search_state.free_chunks = 0; + if (!resume_state) + pc = rx_outer_start; + else + { + search_state = *resume_state; + regs = search_state.saved_regs; + rxb = search_state.saved_rxb; + startpos = search_state.saved_startpos; + range = search_state.saved_range; + stop = search_state.saved_stop; + total_size = search_state.saved_total_size; + get_burst = search_state.saved_get_burst; + back_check = search_state.saved_back_check; + pc = search_state.outer_search_resume_pt; + if (0) + { + return_continuation: + if (save_state) + { + *save_state = search_state; + save_state->saved_regs = regs; + save_state->saved_rxb = rxb; + save_state->saved_startpos = startpos; + save_state->saved_range = range; + save_state->saved_stop = stop; + save_state->saved_total_size = total_size; + save_state->saved_get_burst = get_burst; + save_state->saved_back_check = back_check; + save_state->outer_search_resume_pt = pc; + } + return rx_search_continuation; + } + } + + switch (pc) + { + case rx_outer_start: + search_state.ret_val = rx_search_fail; + ( search_state.lparen + = search_state.rparen + = search_state.best_lpspace + = search_state.best_rpspace + = 0); + + /* figure the number of registers we may need for use in backreferences. + * the number here includes an element for register zero. + */ + search_state.num_regs = rxb->re_nsub + 1; + + + /* check for out-of-range startpos. */ + if ((startpos < 0) || (startpos > total_size)) + return rx_search_fail; + + /* fix up range if it might eventually take us outside the string. */ + { + int endpos; + endpos = startpos + range; + if (endpos < -1) + range = (-1 - startpos); + else if (endpos > (total_size + 1)) + range = total_size - startpos; + } + + /* if the search isn't to be a backwards one, don't waste time in a + * long search for a pattern that says it is anchored. + */ + if (rxb->begbuf_only && (range > 0)) + { + if (startpos > 0) + return rx_search_fail; + else + range = 1; + } + + /* decide whether to use internal or user-provided reg buffers. */ + if (!regs || rxb->no_sub) + { + search_state.best_lpspace = + (regoff_t *)REGEX_ALLOCATE (search_state.num_regs * sizeof(regoff_t)); + search_state.best_rpspace = + (regoff_t *)REGEX_ALLOCATE (search_state.num_regs * sizeof(regoff_t)); + search_state.best_lparen = search_state.best_lpspace; + search_state.best_rparen = search_state.best_rpspace; + } + else + { + /* have the register data arrays been allocated? */ + if (rxb->regs_allocated == REGS_UNALLOCATED) + { /* no. so allocate them with malloc. we need one + extra element beyond `search_state.num_regs' for the `-1' marker + gnu code uses. */ + regs->num_regs = MAX (RE_NREGS, rxb->re_nsub + 1); + regs->start = ((regoff_t *) + malloc (regs->num_regs * sizeof ( regoff_t))); + regs->end = ((regoff_t *) + malloc (regs->num_regs * sizeof ( regoff_t))); + if (regs->start == 0 || regs->end == 0) + return rx_search_error; + rxb->regs_allocated = REGS_REALLOCATE; + } + else if (rxb->regs_allocated == REGS_REALLOCATE) + { /* yes. if we need more elements than were already + allocated, reallocate them. if we need fewer, just + leave it alone. */ + if (regs->num_regs < search_state.num_regs + 1) + { + regs->num_regs = search_state.num_regs + 1; + regs->start = ((regoff_t *) + realloc (regs->start, + regs->num_regs * sizeof (regoff_t))); + regs->end = ((regoff_t *) + realloc (regs->end, + regs->num_regs * sizeof ( regoff_t))); + if (regs->start == 0 || regs->end == 0) + return rx_search_error; + } + } + else if (rxb->regs_allocated != REGS_FIXED) + return rx_search_error; + + if (regs->num_regs < search_state.num_regs + 1) + { + search_state.best_lpspace = + ((regoff_t *) + REGEX_ALLOCATE (search_state.num_regs * sizeof(regoff_t))); + search_state.best_rpspace = + ((regoff_t *) + REGEX_ALLOCATE (search_state.num_regs * sizeof(regoff_t))); + search_state.best_lparen = search_state.best_lpspace; + search_state.best_rparen = search_state.best_rpspace; + } + else + { + search_state.best_lparen = regs->start; + search_state.best_rparen = regs->end; + } + } + + search_state.lparen = + (regoff_t *) REGEX_ALLOCATE (search_state.num_regs * sizeof(regoff_t)); + search_state.rparen = + (regoff_t *) REGEX_ALLOCATE (search_state.num_regs * sizeof(regoff_t)); + + if (! ( search_state.best_rparen + && search_state.best_lparen + && search_state.lparen && search_state.rparen)) + return rx_search_error; + + search_state.best_last_l = search_state.best_last_r = -1; + + search_state.translate = (rxb->translate + ? rxb->translate + : rx_id_translation); + + + + /* + * two nfa's were compiled. + * `0' is complete. + * `1' faster but gets registers wrong and ends too soon. + */ + search_state.nfa_choice = (regs && !rxb->least_subs) ? '\0' : '\1'; + + /* we have the option to look for the best match or the first + * one we can find. if the user isn't asking for register information, + * we don't need to find the best match. + */ + search_state.first_found = !regs; + + if (range >= 0) + { + search_state.outer_pos.search_end = startpos + range; + search_state.outer_pos.search_direction = 1; + } + else + { + search_state.outer_pos.search_end = startpos + range; + search_state.outer_pos.search_direction = -1; + } + + /* the vacuous search always turns up nothing. */ + if ((search_state.outer_pos.search_direction == 1) + ? (startpos > search_state.outer_pos.search_end) + : (startpos < search_state.outer_pos.search_end)) + return rx_search_fail; + + /* now we build the starting state of the supernfa. */ + { + struct rx_superset * start_contents; + struct rx_nfa_state_set * start_nfa_set; + + /* we presume here that the nfa start state has only one + * possible future with no side effects. + */ + start_nfa_set = rxb->start->futures->destset; + if ( rxb->rx.start_set + && (rxb->rx.start_set->starts_for == &rxb->rx)) + start_contents = rxb->rx.start_set; + else + { + start_contents = + rx_superstate_eclosure_union (&rxb->rx, + rx_superset_cons (&rxb->rx, 0, 0), + start_nfa_set); + + if (!start_contents) + return rx_search_fail; + + start_contents->starts_for = &rxb->rx; + rxb->rx.start_set = start_contents; + } + if ( start_contents->superstate + && (start_contents->superstate->rx_id == rxb->rx.rx_id)) + { + search_state.start_super = start_contents->superstate; + rx_lock_superstate (&rxb->rx, search_state.start_super); + } + else + { + rx_protect_superset (&rxb->rx, start_contents); + + search_state.start_super = rx_superstate (&rxb->rx, start_contents); + if (!search_state.start_super) + return rx_search_fail; + rx_lock_superstate (&rxb->rx, search_state.start_super); + rx_release_superset (&rxb->rx, start_contents); + } + } + + + /* The outer_pos tracks the position within the strings + * as seen by loop that calls fastmap_search. + * + * The caller supplied get_burst function actually + * gives us pointers to chars. + * + * Communication with the get_burst function is through an + * rx_string_position structure. Here, the structure for + * outer_pos is initialized. It is set to point to the + * NULL string, at an offset of STARTPOS. STARTPOS is out + * of range of the NULL string, so the first call to + * getburst will patch up the rx_string_position to point + * to valid characters. + */ + + ( search_state.outer_pos.string + = search_state.outer_pos.end + = 0); + + search_state.outer_pos.offset = 0; + search_state.outer_pos.size = 0; + search_state.outer_pos.pos = (unsigned char *)startpos; + init_fastmap (rxb, &search_state); + + search_state.fastmap_resume_pt = rx_fastmap_start; + case rx_outer_fastmap: + /* do { */ + pseudo_do: + { + { + int fastmap_state; + fastmap_state = fastmap_search (rxb, stop, get_burst, app_closure, + &search_state); + switch (fastmap_state) + { + case rx_fastmap_continuation: + pc = rx_outer_fastmap; + goto return_continuation; + case rx_fastmap_fail: + goto finish; + case rx_fastmap_ok: + break; + } + } + + /* now the fastmap loop has brought us to a plausible + * starting point for a match. so, it's time to run the + * nfa and see if a match occured. + */ + startpos = ( search_state.outer_pos.pos + - search_state.outer_pos.string + + search_state.outer_pos.offset); +#if 0 +/*|*/ if ((range > 0) && (startpos == search_state.outer_pos.search_end)) +/*|*/ goto finish; +#endif + } + + search_state.test_match_resume_pt = rx_test_start; + /* do interrupted for entry point... */ + case rx_outer_test: + /* ...do continued */ + { + goto test_match; + test_returns_to_search: + switch (test_state) + { + case rx_test_continuation: + pc = rx_outer_test; + goto return_continuation; + case rx_test_error: + search_state.ret_val = rx_search_error; + goto finish; + case rx_test_fail: + break; + case rx_test_ok: + goto finish; + } + search_state.outer_pos.pos += search_state.outer_pos.search_direction; + startpos += search_state.outer_pos.search_direction; +#if 0 +/*|*/ if (search_state.test_pos.pos < search_state.test_pos.end) +/*|*/ break; +#endif + } + /* do interrupted for entry point... */ + case rx_outer_restore_pos: + { + int x; + x = get_burst (&search_state.outer_pos, app_closure, stop); + switch (x) + { + case rx_get_burst_continuation: + pc = rx_outer_restore_pos; + goto return_continuation; + case rx_get_burst_error: + search_state.ret_val = rx_search_error; + goto finish; + case rx_get_burst_no_more: + if (rxb->can_match_empty) + break; + goto finish; + case rx_get_burst_ok: + break; + } + } /* } while (...see below...) */ + + if ((search_state.outer_pos.search_direction == 1) + ? (startpos <= search_state.outer_pos.search_end) + : (startpos > search_state.outer_pos.search_end)) + goto pseudo_do; + + + finish: + uninit_fastmap (rxb, &search_state); + if (search_state.start_super) + rx_unlock_superstate (&rxb->rx, search_state.start_super); + +#ifdef regex_malloc + if (search_state.lparen) free (search_state.lparen); + if (search_state.rparen) free (search_state.rparen); + if (search_state.best_lpspace) free (search_state.best_lpspace); + if (search_state.best_rpspace) free (search_state.best_rpspace); +#endif + return search_state.ret_val; + } + + + test_match: + { + enum rx_test_match_entry test_pc; + int inx; + test_pc = search_state.test_match_resume_pt; + if (test_pc == rx_test_start) + { +#ifdef RX_DEBUG + search_state.backtrack_depth = 0; +#endif + search_state.last_l = search_state.last_r = 0; + search_state.lparen[0] = startpos; + search_state.super = search_state.start_super; + search_state.c = search_state.nfa_choice; + search_state.test_pos.pos = search_state.outer_pos.pos - 1; + search_state.test_pos.string = search_state.outer_pos.string; + search_state.test_pos.end = search_state.outer_pos.end; + search_state.test_pos.offset = search_state.outer_pos.offset; + search_state.test_pos.size = search_state.outer_pos.size; + search_state.test_pos.search_direction = 1; + search_state.counter_stack = 0; + search_state.backtrack_stack = 0; + search_state.backtrack_frame_bytes = + (sizeof (struct rx_backtrack_frame) + + (rxb->match_regs_on_stack + ? sizeof (regoff_t) * (search_state.num_regs + 1) * 2 + : 0)); + search_state.chunk_bytes = search_state.backtrack_frame_bytes * 64; + search_state.test_ret = rx_test_line_finished; + search_state.could_have_continued = 0; + } + /* This is while (1)...except that the body of the loop is interrupted + * by some alternative entry points. + */ + pseudo_while_1: + switch (test_pc) + { + case rx_test_cache_hit_loop: + goto resume_continuation_1; + case rx_test_backreference_check: + goto resume_continuation_2; + case rx_test_backtrack_return: + goto resume_continuation_3; + case rx_test_start: +#ifdef RX_DEBUG + /* There is a search tree with every node as set of deterministic + * transitions in the super nfa. For every branch of a + * backtrack point is an edge in the tree. + * This counts up a pre-order of nodes in that tree. + * It's saved on the search stack and printed when debugging. + */ + search_state.line_no = 0; + search_state.lines_found = 0; +#endif + + top_of_cycle: + /* A superstate is basicly a transition table, indexed by + * characters from the string being tested, and containing + * RX_INX (`instruction frame') structures. + */ + search_state.ifr = &search_state.super->transitions [search_state.c]; + + recurse_test_match: + /* This is the point to which control is sent when the + * test matcher `recurses'. Before jumping here, some variables + * need to be saved on the stack and the next instruction frame + * has to be computed. + */ + + restart: + /* Some instructions don't advance the matcher, but just + * carry out some side effects and fetch a new instruction. + * To dispatch that new instruction, `goto restart'. + */ + + { + struct rx_inx * next_tr_table; + struct rx_inx * this_tr_table; + + /* The fastest route through the loop is when the instruction + * is RX_NEXT_CHAR. This case is detected when SEARCH_STATE.IFR->DATA + * is non-zero. In that case, it points to the next + * superstate. + * + * This allows us to not bother fetching the bytecode. + */ + next_tr_table = (struct rx_inx *)search_state.ifr->data; + this_tr_table = search_state.super->transitions; + while (next_tr_table) + { +#ifdef RX_DEBUG_0 + if (rx_debug_trace) + { + struct rx_superset * setp; + + fprintf (stderr, "%d %d>> re_next_char @ %d (%d)", + search_state.line_no, + search_state.backtrack_depth, + (search_state.test_pos.pos - search_state.test_pos.string + + search_state.test_pos.offset), search_state.c); + + search_state.super = + ((struct rx_superstate *) + ((char *)this_tr_table + - ((unsigned long) + ((struct rx_superstate *)0)->transitions))); + + setp = search_state.super->contents; + fprintf (stderr, " superstet (rx=%d, &=%x: ", + rxb->rx.rx_id, setp); + while (setp) + { + fprintf (stderr, "%d ", setp->id); + setp = setp->cdr; + } + fprintf (stderr, "\n"); + } +#endif + this_tr_table = next_tr_table; + ++search_state.test_pos.pos; + if (search_state.test_pos.pos == search_state.test_pos.end) + { + int burst_state; + try_burst_1: + burst_state = get_burst (&search_state.test_pos, + app_closure, stop); + switch (burst_state) + { + case rx_get_burst_continuation: + search_state.saved_this_tr_table = this_tr_table; + search_state.saved_next_tr_table = next_tr_table; + test_pc = rx_test_cache_hit_loop; + goto test_return_continuation; + + resume_continuation_1: + /* Continuation one jumps here to do its work: */ + search_state.saved_this_tr_table = this_tr_table; + search_state.saved_next_tr_table = next_tr_table; + goto try_burst_1; + + case rx_get_burst_ok: + /* get_burst succeeded...keep going */ + break; + + case rx_get_burst_no_more: + search_state.test_ret = rx_test_line_finished; + search_state.could_have_continued = 1; + goto test_do_return; + + case rx_get_burst_error: + /* An error... */ + search_state.test_ret = rx_test_internal_error; + goto test_do_return; + } + } + search_state.c = *search_state.test_pos.pos; + search_state.ifr = this_tr_table + search_state.c; + next_tr_table = (struct rx_inx *)search_state.ifr->data; + } /* Fast loop through cached transition tables */ + + /* Here when we ran out of cached next-char transitions. + * So, it will be necessary to do a more expensive + * dispatch on the current instruction. The superstate + * pointer is allowed to become invalid during next-char + * transitions -- now we must bring it up to date. + */ + search_state.super = + ((struct rx_superstate *) + ((char *)this_tr_table + - ((unsigned long) + ((struct rx_superstate *)0)->transitions))); + } + + /* We've encountered an instruction other than next-char. + * Dispatch that instruction: + */ + inx = (int)search_state.ifr->inx; +#ifdef RX_DEBUG_0 + if (rx_debug_trace) + { + struct rx_superset * setp = search_state.super->contents; + + fprintf (stderr, "%d %d>> %s @ %d (%d)", search_state.line_no, + search_state.backtrack_depth, + inx_names[inx], + (search_state.test_pos.pos - search_state.test_pos.string + + (test_pos.half == 0 ? 0 : size1)), search_state.c); + + fprintf (stderr, " superstet (rx=%d, &=%x: ", + rxb->rx.rx_id, setp); + while (setp) + { + fprintf (stderr, "%d ", setp->id); + setp = setp->cdr; + } + fprintf (stderr, "\n"); + } +#endif + switch ((enum rx_opcode)inx) + { + case rx_do_side_effects: + + /* RX_DO_SIDE_EFFECTS occurs when we cross epsilon + * edges associated with parentheses, backreferencing, etc. + */ + { + struct rx_distinct_future * df = + (struct rx_distinct_future *)search_state.ifr->data_2; + struct rx_se_list * el = df->effects; + /* Side effects come in lists. This walks down + * a list, dispatching. + */ + while (el) + { + long effect; + effect = (long)el->car; + if (effect < 0) + { +#ifdef RX_DEBUG_0 + if (rx_debug_trace) + { + struct rx_superset * setp = search_state.super->contents; + + fprintf (stderr, "....%d %d>> %s\n", search_state.line_no, + search_state.backtrack_depth, + efnames[-effect]); + } +#endif + switch ((enum re_side_effects) effect) + + { + case re_se_pushback: + search_state.ifr = &df->future_frame; + if (!search_state.ifr->data) + { + struct rx_superstate * sup; + sup = search_state.super; + rx_lock_superstate (rx, sup); + if (!rx_handle_cache_miss (&rxb->rx, + search_state.super, + search_state.c, + (search_state.ifr + ->data_2))) + { + rx_unlock_superstate (rx, sup); + search_state.test_ret = rx_test_internal_error; + goto test_do_return; + } + rx_unlock_superstate (rx, sup); + } + /* --search_state.test_pos.pos; */ + search_state.c = 't'; + search_state.super + = ((struct rx_superstate *) + ((char *)search_state.ifr->data + - (long)(((struct rx_superstate *)0) + ->transitions))); + goto top_of_cycle; + break; + case re_se_push0: + { + struct rx_counter_frame * old_cf + = (search_state.counter_stack + ? ((struct rx_counter_frame *) + search_state.counter_stack->sp) + : 0); + struct rx_counter_frame * cf; + PUSH (search_state.counter_stack, + sizeof (struct rx_counter_frame)); + cf = ((struct rx_counter_frame *) + search_state.counter_stack->sp); + cf->tag = re_se_iter; + cf->val = 0; + cf->inherited_from = 0; + cf->cdr = old_cf; + break; + } + case re_se_fail: + goto test_do_return; + case re_se_begbuf: + if (!AT_STRINGS_BEG ()) + goto test_do_return; + break; + case re_se_endbuf: + if (!AT_STRINGS_END ()) + goto test_do_return; + break; + case re_se_wordbeg: + if ( LETTER_P (&search_state.test_pos, 1) + && ( AT_STRINGS_BEG() + || !LETTER_P (&search_state.test_pos, 0))) + break; + else + goto test_do_return; + case re_se_wordend: + if ( !AT_STRINGS_BEG () + && LETTER_P (&search_state.test_pos, 0) + && (AT_STRINGS_END () + || !LETTER_P (&search_state.test_pos, 1))) + break; + else + goto test_do_return; + case re_se_wordbound: + if (AT_WORD_BOUNDARY (&search_state.test_pos)) + break; + else + goto test_do_return; + case re_se_notwordbound: + if (!AT_WORD_BOUNDARY (&search_state.test_pos)) + break; + else + goto test_do_return; + case re_se_hat: + if (AT_STRINGS_BEG ()) + { + if (rxb->not_bol) + goto test_do_return; + else + break; + } + else + { + char pos_c = *search_state.test_pos.pos; + if ( (SRCH_TRANSLATE (pos_c) + == SRCH_TRANSLATE('\n')) + && rxb->newline_anchor) + break; + else + goto test_do_return; + } + case re_se_dollar: + if (AT_STRINGS_END ()) + { + if (rxb->not_eol) + goto test_do_return; + else + break; + } + else + { + if ( ( SRCH_TRANSLATE (fetch_char + (&search_state.test_pos, 1, + app_closure, stop)) + == SRCH_TRANSLATE ('\n')) + && rxb->newline_anchor) + break; + else + goto test_do_return; + } + + case re_se_try: + /* This is the first side effect in every + * expression. + * + * FOR NO GOOD REASON...get rid of it... + */ + break; + + case re_se_pushpos: + { + int urhere = + ((int)(search_state.test_pos.pos + - search_state.test_pos.string) + + search_state.test_pos.offset); + struct rx_counter_frame * old_cf + = (search_state.counter_stack + ? ((struct rx_counter_frame *) + search_state.counter_stack->sp) + : 0); + struct rx_counter_frame * cf; + PUSH(search_state.counter_stack, + sizeof (struct rx_counter_frame)); + cf = ((struct rx_counter_frame *) + search_state.counter_stack->sp); + cf->tag = re_se_pushpos; + cf->val = urhere; + cf->inherited_from = 0; + cf->cdr = old_cf; + break; + } + + case re_se_chkpos: + { + int urhere = + ((int)(search_state.test_pos.pos + - search_state.test_pos.string) + + search_state.test_pos.offset); + struct rx_counter_frame * cf + = ((struct rx_counter_frame *) + search_state.counter_stack->sp); + if (cf->val == urhere) + goto test_do_return; + cf->val = urhere; + break; + } + break; + + case re_se_poppos: + POP(search_state.counter_stack, + sizeof (struct rx_counter_frame)); + break; + + + case re_se_at_dot: + case re_se_syntax: + case re_se_not_syntax: +#ifdef emacs + /* + * this release lacks emacs support + */ +#endif + break; + case re_se_win: + case re_se_lparen: + case re_se_rparen: + case re_se_backref: + case re_se_iter: + case re_se_end_iter: + case re_se_tv: + case re_floogle_flap: + search_state.ret_val = 0; + goto test_do_return; + } + } + else + { +#ifdef RX_DEBUG_0 + if (rx_debug_trace) + fprintf (stderr, "....%d %d>> %s %d %d\n", search_state.line_no, + search_state.backtrack_depth, + efnames2[rxb->se_params [effect].se], + rxb->se_params [effect].op1, + rxb->se_params [effect].op2); +#endif + switch (rxb->se_params [effect].se) + { + case re_se_win: + /* This side effect indicates that we've + * found a match, though not necessarily the + * best match. This is a fancy assignment to + * register 0 unless the caller didn't + * care about registers. In which case, + * this stops the match. + */ + { + int urhere = + ((int)(search_state.test_pos.pos + - search_state.test_pos.string) + + search_state.test_pos.offset); + + if ( (search_state.best_last_r < 0) + || (urhere + 1 > search_state.best_rparen[0])) + { + /* Record the best known and keep + * looking. + */ + int x; + for (x = 0; x <= search_state.last_l; ++x) + search_state.best_lparen[x] = search_state.lparen[x]; + search_state.best_last_l = search_state.last_l; + for (x = 0; x <= search_state.last_r; ++x) + search_state.best_rparen[x] = search_state.rparen[x]; + search_state.best_rparen[0] = urhere + 1; + search_state.best_last_r = search_state.last_r; + } + /* If we're not reporting the match-length + * or other register info, we need look no + * further. + */ + if (search_state.first_found) + { + search_state.test_ret = rx_test_found_first; + goto test_do_return; + } + } + break; + case re_se_lparen: + { + int urhere = + ((int)(search_state.test_pos.pos + - search_state.test_pos.string) + + search_state.test_pos.offset); + + int reg = rxb->se_params [effect].op1; +#if 0 + if (reg > search_state.last_l) +#endif + { + search_state.lparen[reg] = urhere + 1; + /* In addition to making this assignment, + * we now know that lower numbered regs + * that haven't already been assigned, + * won't be. We make sure they're + * filled with -1, so they can be + * recognized as unassigned. + */ + if (search_state.last_l < reg) + while (++search_state.last_l < reg) + search_state.lparen[search_state.last_l] = -1; + } + break; + } + + case re_se_rparen: + { + int urhere = + ((int)(search_state.test_pos.pos + - search_state.test_pos.string) + + search_state.test_pos.offset); + int reg = rxb->se_params [effect].op1; + search_state.rparen[reg] = urhere + 1; + if (search_state.last_r < reg) + { + while (++search_state.last_r < reg) + search_state.rparen[search_state.last_r] + = -1; + } + break; + } + + case re_se_backref: + { + int reg = rxb->se_params [effect].op1; + if ( reg > search_state.last_r + || search_state.rparen[reg] < 0) + goto test_do_return; + + { + int backref_status; + check_backreference: + backref_status + = back_check (&search_state.test_pos, + search_state.lparen[reg], + search_state.rparen[reg], + search_state.translate, + app_closure, + stop); + switch (backref_status) + { + case rx_back_check_continuation: + search_state.saved_reg = reg; + test_pc = rx_test_backreference_check; + goto test_return_continuation; + resume_continuation_2: + reg = search_state.saved_reg; + goto check_backreference; + case rx_back_check_fail: + /* Fail */ + goto test_do_return; + case rx_back_check_pass: + /* pass -- + * test_pos now advanced to last + * char matched by backref + */ + break; + } + } + break; + } + case re_se_iter: + { + struct rx_counter_frame * csp + = ((struct rx_counter_frame *) + search_state.counter_stack->sp); + if (csp->val == rxb->se_params[effect].op2) + goto test_do_return; + else + ++csp->val; + break; + } + case re_se_end_iter: + { + struct rx_counter_frame * csp + = ((struct rx_counter_frame *) + search_state.counter_stack->sp); + if (csp->val < rxb->se_params[effect].op1) + goto test_do_return; + else + { + struct rx_counter_frame * source = csp; + while (source->inherited_from) + source = source->inherited_from; + if (!source || !source->cdr) + { + POP(search_state.counter_stack, + sizeof(struct rx_counter_frame)); + } + else + { + source = source->cdr; + csp->val = source->val; + csp->tag = source->tag; + csp->cdr = 0; + csp->inherited_from = source; + } + } + break; + } + case re_se_tv: + /* is a noop */ + break; + case re_se_try: + case re_se_pushback: + case re_se_push0: + case re_se_pushpos: + case re_se_chkpos: + case re_se_poppos: + case re_se_at_dot: + case re_se_syntax: + case re_se_not_syntax: + case re_se_begbuf: + case re_se_hat: + case re_se_wordbeg: + case re_se_wordbound: + case re_se_notwordbound: + case re_se_wordend: + case re_se_endbuf: + case re_se_dollar: + case re_se_fail: + case re_floogle_flap: + search_state.ret_val = 0; + goto test_do_return; + } + } + el = el->cdr; + } + /* Now the side effects are done, + * so get the next instruction. + * and move on. + */ + search_state.ifr = &df->future_frame; + goto restart; + } + + case rx_backtrack_point: + { + /* A backtrack point indicates that we've reached a + * non-determinism in the superstate NFA. This is a + * loop that exhaustively searches the possibilities. + * + * A backtracking strategy is used. We keep track of what + * registers are valid so we can erase side effects. + * + * First, make sure there is some stack space to hold + * our state. + */ + + struct rx_backtrack_frame * bf; + + PUSH(search_state.backtrack_stack, + search_state.backtrack_frame_bytes); +#ifdef RX_DEBUG_0 + ++search_state.backtrack_depth; +#endif + + bf = ((struct rx_backtrack_frame *) + search_state.backtrack_stack->sp); + { + bf->stk_super = search_state.super; + /* We prevent the current superstate from being + * deleted from the superstate cache. + */ + rx_lock_superstate (&rxb->rx, search_state.super); +#ifdef RX_DEBUG_0 + bf->stk_search_state.line_no = search_state.line_no; +#endif + bf->stk_c = search_state.c; + bf->stk_test_pos = search_state.test_pos; + bf->stk_last_l = search_state.last_l; + bf->stk_last_r = search_state.last_r; + bf->df = ((struct rx_super_edge *) + search_state.ifr->data_2)->options; + bf->first_df = bf->df; + bf->counter_stack_sp = (search_state.counter_stack + ? search_state.counter_stack->sp + : 0); + bf->stk_test_ret = search_state.test_ret; + if (rxb->match_regs_on_stack) + { + int x; + regoff_t * stk = + (regoff_t *)((char *)bf + sizeof (*bf)); + for (x = 0; x <= search_state.last_l; ++x) + stk[x] = search_state.lparen[x]; + stk += x; + for (x = 0; x <= search_state.last_r; ++x) + stk[x] = search_state.rparen[x]; + } + } + + /* Here is a while loop whose body is mainly a function + * call and some code to handle a return from that + * function. + * + * From here on for the rest of `case backtrack_point' it + * is unsafe to assume that the search_state copies of + * variables saved on the backtracking stack are valid + * -- so read their values from the backtracking stack. + * + * This lets us use one generation fewer stack saves in + * the call-graph of a search. + */ + + while_non_det_options: +#ifdef RX_DEBUG_0 + ++search_state.lines_found; + if (rx_debug_trace) + fprintf (stderr, "@@@ %d calls %d @@@\n", + search_state.line_no, search_state.lines_found); + + search_state.line_no = search_state.lines_found; +#endif + + if (bf->df->next_same_super_edge[0] == bf->first_df) + { + /* This is a tail-call optimization -- we don't recurse + * for the last of the possible futures. + */ + search_state.ifr = (bf->df->effects + ? &bf->df->side_effects_frame + : &bf->df->future_frame); + + rx_unlock_superstate (&rxb->rx, search_state.super); + POP(search_state.backtrack_stack, + search_state.backtrack_frame_bytes); +#ifdef RX_DEBUG + --search_state.backtrack_depth; +#endif + goto restart; + } + else + { + if (search_state.counter_stack) + { + struct rx_counter_frame * old_cf + = ((struct rx_counter_frame *)search_state.counter_stack->sp); + struct rx_counter_frame * cf; + PUSH(search_state.counter_stack, sizeof (struct rx_counter_frame)); + cf = ((struct rx_counter_frame *)search_state.counter_stack->sp); + cf->tag = old_cf->tag; + cf->val = old_cf->val; + cf->inherited_from = old_cf; + cf->cdr = 0; + } + /* `Call' this test-match block */ + search_state.ifr = (bf->df->effects + ? &bf->df->side_effects_frame + : &bf->df->future_frame); + goto recurse_test_match; + } + + /* Returns in this block are accomplished by + * goto test_do_return. There are two cases. + * If there is some search-stack left, + * then it is a return from a `recursive' call. + * If there is no search-stack left, then + * we should return to the fastmap/search loop. + */ + + test_do_return: + + if (!search_state.backtrack_stack) + { +#ifdef RX_DEBUG_0 + if (rx_debug_trace) + fprintf (stderr, "!!! %d bails returning %d !!!\n", + search_state.line_no, search_state.test_ret); +#endif + + /* No more search-stack -- this test is done. */ + if (search_state.test_ret != rx_test_internal_error) + goto return_from_test_match; + else + goto error_in_testing_match; + } + + /* Returning from a recursive call to + * the test match block: + */ + + bf = ((struct rx_backtrack_frame *) + search_state.backtrack_stack->sp); +#ifdef RX_DEBUG_0 + if (rx_debug_trace) + fprintf (stderr, "+++ %d returns %d (to %d)+++\n", + search_state.line_no, + search_state.test_ret, + bf->stk_search_state.line_no); +#endif + + while (search_state.counter_stack + && (!bf->counter_stack_sp + || (bf->counter_stack_sp + != search_state.counter_stack->sp))) + { + POP(search_state.counter_stack, + sizeof (struct rx_counter_frame)); + } + + if (search_state.test_ret == rx_test_internal_error) + { + POP (search_state.backtrack_stack, + search_state.backtrack_frame_bytes); + goto test_do_return; + } + + /* If a non-longest match was found and that is good + * enough, return immediately. + */ + if ( (search_state.test_ret == rx_test_found_first) + && search_state.first_found) + { + rx_unlock_superstate (&rxb->rx, bf->stk_super); + POP (search_state.backtrack_stack, + search_state.backtrack_frame_bytes); + goto test_do_return; + } + + search_state.test_ret = bf->stk_test_ret; + search_state.last_l = bf->stk_last_l; + search_state.last_r = bf->stk_last_r; + bf->df = bf->df->next_same_super_edge[0]; + search_state.super = bf->stk_super; + search_state.c = bf->stk_c; +#ifdef RX_DEBUG_0 + search_state.line_no = bf->stk_search_state.line_no; +#endif + + if (rxb->match_regs_on_stack) + { + int x; + regoff_t * stk = + (regoff_t *)((char *)bf + sizeof (*bf)); + for (x = 0; x <= search_state.last_l; ++x) + search_state.lparen[x] = stk[x]; + stk += x; + for (x = 0; x <= search_state.last_r; ++x) + search_state.rparen[x] = stk[x]; + } + + if ((search_state.test_ret != rx_test_line_finished) && + (search_state.test_ret != rx_test_internal_error)) + { + int x; + try_burst_2: + x = get_burst (&bf->stk_test_pos, app_closure, stop); + switch (x) + { + case rx_get_burst_continuation: + search_state.saved_bf = bf; + test_pc = rx_test_backtrack_return; + goto test_return_continuation; + resume_continuation_3: + bf = search_state.saved_bf; + goto try_burst_2; + case rx_get_burst_no_more: + /* Since we've been here before, it is some kind of + * error that we can't return. + */ + case rx_get_burst_error: + search_state.test_ret = rx_test_internal_error; + goto test_do_return; + case rx_get_burst_ok: + break; + } + } + search_state.test_pos = bf->stk_test_pos; + goto while_non_det_options; + } + + + case rx_cache_miss: + /* Because the superstate NFA is lazily constructed, + * and in fact may erode from underneath us, we sometimes + * have to construct the next instruction from the hard way. + * This invokes one step in the lazy-conversion. + */ + search_state.ifr = rx_handle_cache_miss (&rxb->rx, + search_state.super, + search_state.c, + search_state.ifr->data_2); + if (!search_state.ifr) + { + search_state.test_ret = rx_test_internal_error; + goto test_do_return; + } + goto restart; + + case rx_backtrack: + /* RX_BACKTRACK means that we've reached the empty + * superstate, indicating that match can't succeed + * from this point. + */ + goto test_do_return; + + case rx_next_char: + case rx_error_inx: + case rx_num_instructions: + search_state.ret_val = 0; + goto test_do_return; + } + goto pseudo_while_1; + } + + /* Healthy exits from the test-match loop do a + * `goto return_from_test_match' On the other hand, + * we might end up here. + */ + error_in_testing_match: + test_state = rx_test_error; + goto test_returns_to_search; + + /***** fastmap/search loop body + * considering the results testing for a match + */ + + return_from_test_match: + + if (search_state.best_last_l >= 0) + { + if (regs && (regs->start != search_state.best_lparen)) + { + bcopy (search_state.best_lparen, regs->start, + regs->num_regs * sizeof (int)); + bcopy (search_state.best_rparen, regs->end, + regs->num_regs * sizeof (int)); + } + if (regs && !rxb->no_sub) + { + int q; + int bound = (regs->num_regs < search_state.num_regs + ? regs->num_regs + : search_state.num_regs); + regoff_t * s = regs->start; + regoff_t * e = regs->end; + for (q = search_state.best_last_l + 1; q < bound; ++q) + s[q] = e[q] = -1; + } + search_state.ret_val = search_state.best_lparen[0]; + test_state = rx_test_ok; + goto test_returns_to_search; + } + else + { + test_state = rx_test_fail; + goto test_returns_to_search; + } + + test_return_continuation: + search_state.test_match_resume_pt = test_pc; + test_state = rx_test_continuation; + goto test_returns_to_search; + } +} + + + +#endif /* RX_WANT_RX_DEFS */ + + + +#else /* RX_WANT_SE_DEFS */ + /* Integers are used to represent side effects. + * + * Simple side effects are given negative integer names by these enums. + * + * Non-negative names are reserved for complex effects. + * + * Complex effects are those that take arguments. For example, + * a register assignment associated with a group is complex because + * it requires an argument to tell which group is being matched. + * + * The integer name of a complex effect is an index into rxb->se_params. + */ + + RX_DEF_SE(1, re_se_try, = -1) /* Epsilon from start state */ + + RX_DEF_SE(0, re_se_pushback, = re_se_try - 1) + RX_DEF_SE(0, re_se_push0, = re_se_pushback -1) + RX_DEF_SE(0, re_se_pushpos, = re_se_push0 - 1) + RX_DEF_SE(0, re_se_chkpos, = re_se_pushpos -1) + RX_DEF_SE(0, re_se_poppos, = re_se_chkpos - 1) + + RX_DEF_SE(1, re_se_at_dot, = re_se_poppos - 1) /* Emacs only */ + RX_DEF_SE(0, re_se_syntax, = re_se_at_dot - 1) /* Emacs only */ + RX_DEF_SE(0, re_se_not_syntax, = re_se_syntax - 1) /* Emacs only */ + + RX_DEF_SE(1, re_se_begbuf, = re_se_not_syntax - 1) /* match beginning of buffer */ + RX_DEF_SE(1, re_se_hat, = re_se_begbuf - 1) /* match beginning of line */ + + RX_DEF_SE(1, re_se_wordbeg, = re_se_hat - 1) + RX_DEF_SE(1, re_se_wordbound, = re_se_wordbeg - 1) + RX_DEF_SE(1, re_se_notwordbound, = re_se_wordbound - 1) + + RX_DEF_SE(1, re_se_wordend, = re_se_notwordbound - 1) + RX_DEF_SE(1, re_se_endbuf, = re_se_wordend - 1) + + /* This fails except at the end of a line. + * It deserves to go here since it is typicly one of the last steps + * in a match. + */ + RX_DEF_SE(1, re_se_dollar, = re_se_endbuf - 1) + + /* Simple effects: */ + RX_DEF_SE(1, re_se_fail, = re_se_dollar - 1) + + /* Complex effects. These are used in the 'se' field of + * a struct re_se_params. Indexes into the se array + * are stored as instructions on nfa edges. + */ + RX_DEF_CPLX_SE(1, re_se_win, = 0) + RX_DEF_CPLX_SE(1, re_se_lparen, = re_se_win + 1) + RX_DEF_CPLX_SE(1, re_se_rparen, = re_se_lparen + 1) + RX_DEF_CPLX_SE(0, re_se_backref, = re_se_rparen + 1) + RX_DEF_CPLX_SE(0, re_se_iter, = re_se_backref + 1) + RX_DEF_CPLX_SE(0, re_se_end_iter, = re_se_iter + 1) + RX_DEF_CPLX_SE(0, re_se_tv, = re_se_end_iter + 1) + +#endif + +#if RX_WANT_SE_DEFS != 1 +__END_DECLS +#endif + +#endif diff --git a/include/regexp.h b/include/regexp.h index 73d6bf412..174e10b75 100644 --- a/include/regexp.h +++ b/include/regexp.h @@ -1,21 +1,224 @@ /* - * Definitions etc. for regexp(3) routines. + * regexp.h -- old-style regexp compile and step (emulated with POSIX regex) + * Copyright (C) 1993 Rick Sladkey <jrs@world.std.com> * - * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof], - * not the System V one. - */ -#define NSUBEXP 10 -typedef struct regexp { - char *startp[NSUBEXP]; - char *endp[NSUBEXP]; - char regstart; /* Internal use only. */ - char reganch; /* Internal use only. */ - char *regmust; /* Internal use only. */ - int regmlen; /* Internal use only. */ - char program[1]; /* Unwarranted chumminess with compiler. */ -} regexp; - -extern regexp *regcomp(); -extern int regexec(); -extern void regsub(); -extern void regerror(); + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU Library Public License as published by + * the Free Software Foundation; either version 2, or (at your option) + * any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU Library Public License for more details. + */ + +/* + * Think really hard before you intentionally include this file. + * You should really be using the POSIX regex interface instead. + * This emulation file is intended solely for compiling old code. + * + * A program that uses this file must define six macros: INIT, + * GETC, PEEKC, UNGETC, RETURN, and ERROR. This interface is + * so arcane that VMS hackers point at it in ridicule. + */ + +#ifndef _REGEXP_H +#define _REGEXP_H + +#include <sys/types.h> /* regex.h needs size_t */ +#include <regex.h> /* POSIX.2 regexp routines */ +#include <stdlib.h> /* for malloc, realloc and free */ + +/* + * These three advertised external variables record state information + * for compile and step. They are so gross, I'm choking as I write this. + */ +char *loc1; /* the beginning of a match */ +char *loc2; /* the end of a match */ +int circf; /* current pattern begins with '^' */ + +/* + * These are the other variables mentioned in the regexp.h manpage. + * Since we don't emulate them (whatever they do), we want errors if + * they are referenced. Therefore they are commented out here. + */ +#if 0 +char *locs; +int sed; +int nbra; +#endif + +/* + * We need to stuff a regex_t into an arbitrary buffer so align it. + * GCC make this easy. For the others we have to guess. + */ +#ifdef __GNUC__ +#define __REGEX_T_ALIGN (__alignof__(regex_t)) +#else /* !__GNUC__ */ +#define __REGEX_T_ALIGN 8 +#endif /* !__GNUC__ */ + +#define __regex_t_align(p) \ + ((regex_t *) ((((unsigned long) p) + __REGEX_T_ALIGN - 1) \ + / __REGEX_T_ALIGN * __REGEX_T_ALIGN)) + +/* + * We just slurp the whole pattern into a string and then compile + * it `normally'. With this implementation we never use the PEEKC + * macro. Please feel free to die laughing when we translate + * error symbols into hard-coded numbers. + */ +char * +compile(char *instring, char *expbuf, char *endbuf, int eof) +{ + int __c; + int __len; + char *__buf; + int __buflen; + int __error; + regex_t *__preg; + INIT; + + __buflen = 128; + __buf = malloc(__buflen); + if (!__buf) { + ERROR(50); + return 0; + } + __len = 0; + circf = 0; + for (;;) { + __c = GETC(); + if (__c == eof) + break; + if (__c == '\0' || __c == '\n') { + UNGETC(__c); + break; + } + if (__len + 2 > __buflen) { + __buflen *= 2; + __buf = realloc(__buf, __buflen); + if (!__buf) { + ERROR(50); + return 0; + } + } + if (__len == 0 && !circf && __c == '^') + circf = 1; + else + __buf[__len++] = __c; + } + if (__len == 0 && !circf) { + free(__buf); + ERROR(41); + return 0; + } + __buf[__len] = '\0'; + if (endbuf <= expbuf + sizeof(regex_t)) { + free(__buf); + ERROR(50); + return 0; + } + __preg = __regex_t_align(expbuf); + __preg->buffer = (char *) (__preg + 1); + __preg->allocated = endbuf - (char *) __preg->buffer; + __error = regcomp(__preg, __buf, REG_NEWLINE); + free(__buf); + switch (__error) { + case 0: + break; + case REG_BADRPT: + __error = 36; /* poor fit */ + break; + case REG_BADBR: + __error = 16; + break; + case REG_EBRACE: + __error = 44; /* poor fit */ + break; + case REG_EBRACK: + __error = 49; + break; + case REG_ERANGE: + __error = 36; /* poor fit */ + break; + case REG_ECTYPE: + __error = 36; /* poor fit */ + break; + case REG_EPAREN: + __error = 42; + break; + case REG_ESUBREG: + __error = 36; /* poor fit */ + break; + case REG_EEND: + __error = 36; /* poor fit */ + break; + case REG_EESCAPE: + __error = 36; + break; + case REG_BADPAT: + __error = 36; /* poor fit */ + break; + case REG_ESIZE: + __error = 50; + break; + case REG_ESPACE: + __error = 50; + break; + default: + __error = 36; /* as good as any */ + break; + } + if (__error) { + ERROR(__error); + return 0; + } +#ifdef _RX_H + RETURN((__preg->buffer + __preg->rx.allocated - __preg->rx.reserved)); +#else + RETURN((__preg->buffer + __preg->used)); +#endif +} + +/* + * Note how we carefully emulate the gross `circf' hack. Otherwise, + * this just looks like an ordinary matching call that records the + * starting and ending match positions. + */ +int +step(char *string, char *expbuf) +{ + int __result; + regmatch_t __pmatch[1]; + + __result = regexec(__regex_t_align(expbuf), string, 1, __pmatch, 0); + if (circf && __pmatch[0].rm_so != 0) + __result = REG_NOMATCH; + if (__result == 0) { + loc1 = string + __pmatch[0].rm_so; + loc2 = string + __pmatch[0].rm_eo; + } + return __result == 0; +} + +/* + * For advance we are only supposed to match at the beginning of the + * string. You have to read the man page really carefully to find this + * one. We'll match them kludge-for-kludge. + */ +int +advance(char *string, char *expbuf) +{ + int __old_circf; + int __result; + + __old_circf = circf; + circf = 1; + __result = step(string, expbuf); + circf = __old_circf; + return __result; +} + +#endif /* _REGEXP_H */ diff --git a/include/stdlib.h b/include/stdlib.h index daff04b07..51ce80ba2 100644 --- a/include/stdlib.h +++ b/include/stdlib.h @@ -61,10 +61,29 @@ extern int setenv __P ((__const char *__name, __const char *__value, extern int system __P ((__const char *__command)); -extern int qsort __P ((char *base, int num, int size, int (*cmp)())); - extern char * gcvt __P ((float number, size_t ndigit, char * buf)); +#if defined __USE_BSD || defined __USE_XOPEN_EXTENDED +/* Return the canonical absolute name of file NAME. The last file name + * component need not exist, and may be a symlink to a nonexistent file. + * If RESOLVED is null, the result is malloc'd; otherwise, if the canonical + * name is PATH_MAX chars or more, returns null with `errno' set to + * ENAMETOOLONG; if the name fits in fewer than PATH_MAX chars, returns the + * name in RESOLVED. */ +extern char *realpath __P ((__const char *__restrict __name, + char *__restrict __resolved)); +#endif + + +/* Shorthand for type of comparison functions. */ +typedef int (*__compar_fn_t) __P ((__const __ptr_t, __const __ptr_t)); +typedef __compar_fn_t comparison_fn_t; +/* Sort NMEMB elements of BASE, of SIZE bytes each, + using COMPAR to perform the comparisons. */ +extern void qsort __P ((__ptr_t __base, size_t __nmemb, size_t __size, + __compar_fn_t __compar)); + + #define atof(x) strtod((x),(char**)0) #define atoi(x) (int)strtol((x),(char**)0,10) #define atol(x) strtol((x),(char**)0,10) @@ -84,4 +103,6 @@ typedef struct long int rem; /* Remainder. */ } ldiv_t; + + #endif /* __STDLIB_H */ diff --git a/include/sys/kd.h b/include/sys/kd.h new file mode 100644 index 000000000..b37586abf --- /dev/null +++ b/include/sys/kd.h @@ -0,0 +1,29 @@ +/* Copyright (C) 1996, 1997 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +#ifndef _SYS_KD_H +#define _SYS_KD_H 1 + +/* Make sure the <linux/types.h> header is not loaded. */ +#ifndef _LINUX_TYPES_H +# define _LINUX_TYPES_H 1 +#endif + +#include <linux/kd.h> + +#endif /* sys/kd.h */ diff --git a/include/sys/klog.h b/include/sys/klog.h new file mode 100644 index 000000000..62243e9ec --- /dev/null +++ b/include/sys/klog.h @@ -0,0 +1,34 @@ +/* Copyright (C) 1996 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +#ifndef _SYS_KLOG_H + +#define _SYS_KLOG_H 1 +#include <features.h> + +__BEGIN_DECLS + +/* Control the kernel's logging facility. This corresponds exactly to + the kernel's syslog system call, but that name is easily confused + with the user-level syslog facility, which is something completely + different. */ +extern int klogctl __P ((int __type, char *__bufp, int __len)); + +__END_DECLS + +#endif /* _SYS_KLOG_H */ diff --git a/include/sys/mtio.h b/include/sys/mtio.h new file mode 100644 index 000000000..7498dfa78 --- /dev/null +++ b/include/sys/mtio.h @@ -0,0 +1,277 @@ +/* Structures and definitions for magnetic tape I/O control commands. + Copyright (C) 1996, 1997 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +/* Written by H. Bergman <hennus@cybercomm.nl>. */ + +#ifndef _SYS_MTIO_H +#define _SYS_MTIO_H 1 + +/* Get necessary definitions from system and kernel headers. */ +#include <sys/types.h> +#include <sys/ioctl.h> + + +/* Structure for MTIOCTOP - magnetic tape operation command. */ +struct mtop + { + short int mt_op; /* Operations defined below. */ + int mt_count; /* How many of them. */ + }; +#define _IOT_mtop /* Hurd ioctl type field. */ \ + _IOT (_IOTS (short), 1, _IOTS (int), 1, 0, 0) + +/* Magnetic Tape operations [Not all operations supported by all drivers]. */ +#define MTRESET 0 /* +reset drive in case of problems. */ +#define MTFSF 1 /* Forward space over FileMark, + * position at first record of next file. */ +#define MTBSF 2 /* Backward space FileMark (position before FM). */ +#define MTFSR 3 /* Forward space record. */ +#define MTBSR 4 /* Backward space record. */ +#define MTWEOF 5 /* Write an end-of-file record (mark). */ +#define MTREW 6 /* Rewind. */ +#define MTOFFL 7 /* Rewind and put the drive offline (eject?). */ +#define MTNOP 8 /* No op, set status only (read with MTIOCGET). */ +#define MTRETEN 9 /* Retension tape. */ +#define MTBSFM 10 /* +backward space FileMark, position at FM. */ +#define MTFSFM 11 /* +forward space FileMark, position at FM. */ +#define MTEOM 12 /* Goto end of recorded media (for appending files). + MTEOM positions after the last FM, ready for + appending another file. */ +#define MTERASE 13 /* Erase tape -- be careful! */ + +#define MTRAS1 14 /* Run self test 1 (nondestructive). */ +#define MTRAS2 15 /* Run self test 2 (destructive). */ +#define MTRAS3 16 /* Reserved for self test 3. */ + +#define MTSETBLK 20 /* Set block length (SCSI). */ +#define MTSETDENSITY 21 /* Set tape density (SCSI). */ +#define MTSEEK 22 /* Seek to block (Tandberg, etc.). */ +#define MTTELL 23 /* Tell block (Tandberg, etc.). */ +#define MTSETDRVBUFFER 24 /* Set the drive buffering according to SCSI-2. + Ordinary buffered operation with code 1. */ +#define MTFSS 25 /* Space forward over setmarks. */ +#define MTBSS 26 /* Space backward over setmarks. */ +#define MTWSM 27 /* Write setmarks. */ + +#define MTLOCK 28 /* Lock the drive door. */ +#define MTUNLOCK 29 /* Unlock the drive door. */ +#define MTLOAD 30 /* Execute the SCSI load command. */ +#define MTUNLOAD 31 /* Execute the SCSI unload command. */ +#define MTCOMPRESSION 32/* Control compression with SCSI mode page 15. */ +#define MTSETPART 33 /* Change the active tape partition. */ +#define MTMKPART 34 /* Format the tape with one or two partitions. */ + +/* structure for MTIOCGET - mag tape get status command */ + +struct mtget + { + long int mt_type; /* Type of magtape device. */ + long int mt_resid; /* Residual count: (not sure) + number of bytes ignored, or + number of files not skipped, or + number of records not skipped. */ + /* The following registers are device dependent. */ + long int mt_dsreg; /* Status register. */ + long int mt_gstat; /* Generic (device independent) status. */ + long int mt_erreg; /* Error register. */ + /* The next two fields are not always used. */ + __daddr_t mt_fileno; /* Number of current file on tape. */ + __daddr_t mt_blkno; /* Current block number. */ + }; +#define _IOT_mtget /* Hurd ioctl type field. */ \ + _IOT (_IOTS (long), 7, 0, 0, 0, 0) + + +/* Constants for mt_type. Not all of these are supported, and + these are not all of the ones that are supported. */ +#define MT_ISUNKNOWN 0x01 +#define MT_ISQIC02 0x02 /* Generic QIC-02 tape streamer. */ +#define MT_ISWT5150 0x03 /* Wangtek 5150EQ, QIC-150, QIC-02. */ +#define MT_ISARCHIVE_5945L2 0x04 /* Archive 5945L-2, QIC-24, QIC-02?. */ +#define MT_ISCMSJ500 0x05 /* CMS Jumbo 500 (QIC-02?). */ +#define MT_ISTDC3610 0x06 /* Tandberg 6310, QIC-24. */ +#define MT_ISARCHIVE_VP60I 0x07 /* Archive VP60i, QIC-02. */ +#define MT_ISARCHIVE_2150L 0x08 /* Archive Viper 2150L. */ +#define MT_ISARCHIVE_2060L 0x09 /* Archive Viper 2060L. */ +#define MT_ISARCHIVESC499 0x0A /* Archive SC-499 QIC-36 controller. */ +#define MT_ISQIC02_ALL_FEATURES 0x0F /* Generic QIC-02 with all features. */ +#define MT_ISWT5099EEN24 0x11 /* Wangtek 5099-een24, 60MB, QIC-24. */ +#define MT_ISTEAC_MT2ST 0x12 /* Teac MT-2ST 155mb drive, + Teac DC-1 card (Wangtek type). */ +#define MT_ISEVEREX_FT40A 0x32 /* Everex FT40A (QIC-40). */ +#define MT_ISDDS1 0x51 /* DDS device without partitions. */ +#define MT_ISDDS2 0x52 /* DDS device with partitions. */ +#define MT_ISSCSI1 0x71 /* Generic ANSI SCSI-1 tape unit. */ +#define MT_ISSCSI2 0x72 /* Generic ANSI SCSI-2 tape unit. */ + +/* QIC-40/80/3010/3020 ftape supported drives. + 20bit vendor ID + 0x800000 (see vendors.h in ftape distribution). */ +#define MT_ISFTAPE_UNKNOWN 0x800000 /* obsolete */ +#define MT_ISFTAPE_FLAG 0x800000 + +struct mt_tape_info + { + long int t_type; /* Device type id (mt_type). */ + char *t_name; /* Descriptive name. */ + }; + +#define MT_TAPE_INFO \ + { \ + {MT_ISUNKNOWN, "Unknown type of tape device"}, \ + {MT_ISQIC02, "Generic QIC-02 tape streamer"}, \ + {MT_ISWT5150, "Wangtek 5150, QIC-150"}, \ + {MT_ISARCHIVE_5945L2, "Archive 5945L-2"}, \ + {MT_ISCMSJ500, "CMS Jumbo 500"}, \ + {MT_ISTDC3610, "Tandberg TDC 3610, QIC-24"}, \ + {MT_ISARCHIVE_VP60I, "Archive VP60i, QIC-02"}, \ + {MT_ISARCHIVE_2150L, "Archive Viper 2150L"}, \ + {MT_ISARCHIVE_2060L, "Archive Viper 2060L"}, \ + {MT_ISARCHIVESC499, "Archive SC-499 QIC-36 controller"}, \ + {MT_ISQIC02_ALL_FEATURES, "Generic QIC-02 tape, all features"}, \ + {MT_ISWT5099EEN24, "Wangtek 5099-een24, 60MB"}, \ + {MT_ISTEAC_MT2ST, "Teac MT-2ST 155mb data cassette drive"}, \ + {MT_ISEVEREX_FT40A, "Everex FT40A, QIC-40"}, \ + {MT_ISSCSI1, "Generic SCSI-1 tape"}, \ + {MT_ISSCSI2, "Generic SCSI-2 tape"}, \ + {0, NULL} \ + } + + +/* Structure for MTIOCPOS - mag tape get position command. */ + +struct mtpos + { + long int mt_blkno; /* Current block number. */ + }; +#define _IOT_mtpos /* Hurd ioctl type field. */ \ + _IOT_SIMPLE (long) + + +/* Structure for MTIOCGETCONFIG/MTIOCSETCONFIG primarily intended + as an interim solution for QIC-02 until DDI is fully implemented. */ +struct mtconfiginfo + { + long int mt_type; /* Drive type. */ + long int ifc_type; /* Interface card type. */ + unsigned short int irqnr; /* IRQ number to use. */ + unsigned short int dmanr; /* DMA channel to use. */ + unsigned short int port; /* IO port base address. */ + + unsigned long int debug; /* Debugging flags. */ + + unsigned have_dens:1; + unsigned have_bsf:1; + unsigned have_fsr:1; + unsigned have_bsr:1; + unsigned have_eod:1; + unsigned have_seek:1; + unsigned have_tell:1; + unsigned have_ras1:1; + unsigned have_ras2:1; + unsigned have_ras3:1; + unsigned have_qfa:1; + + unsigned pad1:5; + char reserved[10]; + }; +#define _IOT_mtconfiginfo /* Hurd ioctl type field. */ \ + _IOT (_IOTS (long), 2, _IOTS (short), 3, _IOTS (long), 1) /* XXX wrong */ + + +/* Magnetic tape I/O control commands. */ +#define MTIOCTOP _IOW('m', 1, struct mtop) /* Do a mag tape op. */ +#define MTIOCGET _IOR('m', 2, struct mtget) /* Get tape status. */ +#define MTIOCPOS _IOR('m', 3, struct mtpos) /* Get tape position.*/ + +/* The next two are used by the QIC-02 driver for runtime reconfiguration. + See tpqic02.h for struct mtconfiginfo. */ +#define MTIOCGETCONFIG _IOR('m', 4, struct mtconfiginfo) /* Get tape config.*/ +#define MTIOCSETCONFIG _IOW('m', 5, struct mtconfiginfo) /* Set tape config.*/ + +/* Generic Mag Tape (device independent) status macros for examining + mt_gstat -- HP-UX compatible. + There is room for more generic status bits here, but I don't + know which of them are reserved. At least three or so should + be added to make this really useful. */ +#define GMT_EOF(x) ((x) & 0x80000000) +#define GMT_BOT(x) ((x) & 0x40000000) +#define GMT_EOT(x) ((x) & 0x20000000) +#define GMT_SM(x) ((x) & 0x10000000) /* DDS setmark */ +#define GMT_EOD(x) ((x) & 0x08000000) /* DDS EOD */ +#define GMT_WR_PROT(x) ((x) & 0x04000000) +/* #define GMT_ ? ((x) & 0x02000000) */ +#define GMT_ONLINE(x) ((x) & 0x01000000) +#define GMT_D_6250(x) ((x) & 0x00800000) +#define GMT_D_1600(x) ((x) & 0x00400000) +#define GMT_D_800(x) ((x) & 0x00200000) +/* #define GMT_ ? ((x) & 0x00100000) */ +/* #define GMT_ ? ((x) & 0x00080000) */ +#define GMT_DR_OPEN(x) ((x) & 0x00040000) /* Door open (no tape). */ +/* #define GMT_ ? ((x) & 0x00020000) */ +#define GMT_IM_REP_EN(x) ((x) & 0x00010000) /* Immediate report mode.*/ +/* 16 generic status bits unused. */ + + +/* SCSI-tape specific definitions. Bitfield shifts in the status */ +#define MT_ST_BLKSIZE_SHIFT 0 +#define MT_ST_BLKSIZE_MASK 0xffffff +#define MT_ST_DENSITY_SHIFT 24 +#define MT_ST_DENSITY_MASK 0xff000000 + +#define MT_ST_SOFTERR_SHIFT 0 +#define MT_ST_SOFTERR_MASK 0xffff + +/* Bitfields for the MTSETDRVBUFFER ioctl. */ +#define MT_ST_OPTIONS 0xf0000000 +#define MT_ST_BOOLEANS 0x10000000 +#define MT_ST_SETBOOLEANS 0x30000000 +#define MT_ST_CLEARBOOLEANS 0x40000000 +#define MT_ST_WRITE_THRESHOLD 0x20000000 +#define MT_ST_DEF_BLKSIZE 0x50000000 +#define MT_ST_DEF_OPTIONS 0x60000000 + +#define MT_ST_BUFFER_WRITES 0x1 +#define MT_ST_ASYNC_WRITES 0x2 +#define MT_ST_READ_AHEAD 0x4 +#define MT_ST_DEBUGGING 0x8 +#define MT_ST_TWO_FM 0x10 +#define MT_ST_FAST_MTEOM 0x20 +#define MT_ST_AUTO_LOCK 0x40 +#define MT_ST_DEF_WRITES 0x80 +#define MT_ST_CAN_BSR 0x100 +#define MT_ST_NO_BLKLIMS 0x200 +#define MT_ST_CAN_PARTITIONS 0x400 +#define MT_ST_SCSI2LOGICAL 0x800 + +/* The mode parameters to be controlled. Parameter chosen with bits 20-28. */ +#define MT_ST_CLEAR_DEFAULT 0xfffff +#define MT_ST_DEF_DENSITY (MT_ST_DEF_OPTIONS | 0x100000) +#define MT_ST_DEF_COMPRESSION (MT_ST_DEF_OPTIONS | 0x200000) +#define MT_ST_DEF_DRVBUFFER (MT_ST_DEF_OPTIONS | 0x300000) + +/* The offset for the arguments for the special HP changer load command. */ +#define MT_ST_HPLOADER_OFFSET 10000 + + +/* Specify default tape device. */ +#ifndef DEFTAPE +# define DEFTAPE "/dev/tape" +#endif + +#endif /* mtio.h */ diff --git a/libc/misc/regex/Makefile b/libc/misc/regex/Makefile new file mode 100644 index 000000000..c6c8d8e52 --- /dev/null +++ b/libc/misc/regex/Makefile @@ -0,0 +1,17 @@ +TOPDIR=../ +include $(TOPDIR)Rules.make + +LIBC=../libc.a + +OBJ=rx.o + +all: $(LIBC) + +$(LIBC): $(LIBC)($(OBJ)) + +$(LIBC)(rx.o): rx.c + $(CC) $(CFLAGS) -DL_$* $< -c -o $*.o + $(AR) $(ARFLAGS) $@ $*.o + +clean: + rm -f libc.a *.o core mon.out timer.t.h dMakefile dtr try timer diff --git a/libc/misc/regex/rx.c b/libc/misc/regex/rx.c new file mode 100644 index 000000000..8e85782f2 --- /dev/null +++ b/libc/misc/regex/rx.c @@ -0,0 +1,7522 @@ +/* Copyright (C) 1992, 1993, 1994, 1995 Free Software Foundation, Inc. + +This file is part of the librx library. + +Librx is free software; you can redistribute it and/or modify it under +the terms of the GNU Library General Public License as published by +the Free Software Foundation; either version 2, or (at your option) +any later version. + +Librx is distributed in the hope that it will be useful, but WITHOUT +ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or +FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +for more details. + +You should have received a copy of the GNU Library General Public +License along with this software; see the file COPYING.LIB. If not, +write to the Free Software Foundation, 675 Mass Ave, Cambridge, MA +02139, USA. */ + +/* NOTE!!! AIX is so losing it requires this to be the first thing in the + * file. + * Do not put ANYTHING before it! + */ +#if !defined (__GNUC__) && defined (_AIX) + #pragma alloca +#endif + +/* To make linux happy? */ +#ifndef _GNU_SOURCE +#define _GNU_SOURCE +#endif + + +#include <stdlib.h> +#include <stdio.h> +#include <string.h> +#include <ctype.h> +#ifndef isgraph +#define isgraph(c) (isprint (c) && !isspace (c)) +#endif +#ifndef isblank +#define isblank(c) ((c) == ' ' || (c) == '\t') +#endif + +#include <sys/types.h> + +#undef MAX +#undef MIN +#define MAX(a, b) ((a) > (b) ? (a) : (b)) +#define MIN(a, b) ((a) < (b) ? (a) : (b)) + +typedef char boolean; +#define false 0 +#define true 1 + +#ifndef __GCC__ +#undef __inline__ +#define __inline__ +#endif + +/* Emacs already defines alloca, sometimes. */ +#ifndef alloca + +/* Make alloca work the best possible way. */ +#ifdef __GNUC__ +#define alloca __builtin_alloca +#else /* not __GNUC__ */ +#if HAVE_ALLOCA_H +#include <alloca.h> +#else /* not __GNUC__ or HAVE_ALLOCA_H */ +#ifndef _AIX /* Already did AIX, up at the top. */ +char *alloca (); +#endif /* not _AIX */ +#endif /* not HAVE_ALLOCA_H */ +#endif /* not __GNUC__ */ + +#endif /* not alloca */ + +/* Memory management and stuff for emacs. */ + +#define CHARBITS 8 +#define remalloc(M, S) (M ? realloc (M, S) : malloc (S)) + + +/* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we + * use `alloca' instead of `malloc' for the backtracking stack. + * + * Emacs will die miserably if we don't do this. + */ + +#ifdef REGEX_MALLOC +#define REGEX_ALLOCATE malloc +#else /* not REGEX_MALLOC */ +#define REGEX_ALLOCATE alloca +#endif /* not REGEX_MALLOC */ + + +#ifdef RX_WANT_RX_DEFS +#define RX_DECL extern +#define RX_DEF_QUAL +#else +#define RX_WANT_RX_DEFS +#define RX_DECL static +#define RX_DEF_QUAL static +#endif + +#include <regex.h> +#undef RX_DECL +#define RX_DECL RX_DEF_QUAL + + +/* + * Prototypes. + */ +#ifdef __STDC__ +RX_DECL struct rx_hash_item + *rx_hash_find (struct rx_hash *, unsigned long, + void *, struct rx_hash_rules *); +RX_DECL struct rx_hash_item + *rx_hash_find (struct rx_hash *, unsigned long, + void *, struct rx_hash_rules *); +RX_DECL struct rx_hash_item + *rx_hash_store (struct rx_hash *, unsigned long, + void *, struct rx_hash_rules *); +RX_DECL void rx_hash_free (struct rx_hash_item *, + struct rx_hash_rules *); +RX_DECL void rx_free_hash_table (struct rx_hash *, rx_hash_freefn, + struct rx_hash_rules *); +RX_DECL rx_Bitset + rx_cset (struct rx *); +RX_DECL rx_Bitset + rx_copy_cset (struct rx *, rx_Bitset); +RX_DECL void rx_free_cset (struct rx *, rx_Bitset); +static struct rx_hash_item + *compiler_hash_item_alloc (struct rx_hash_rules *, void *); +static struct rx_hash + *compiler_hash_alloc (struct rx_hash_rules *); +static void compiler_free_hash (struct rx_hash *, + struct rx_hash_rules *); +static void compiler_free_hash_item (struct rx_hash_item *, + struct rx_hash_rules *); +RX_DECL struct rexp_node + *rexp_node (struct rx *, enum rexp_node_type); +RX_DECL struct rexp_node + *rx_mk_r_cset (struct rx *, rx_Bitset); +RX_DECL struct rexp_node + *rx_mk_r_concat (struct rx *, struct rexp_node *, + struct rexp_node *); +RX_DECL struct rexp_node + *rx_mk_r_alternate (struct rx *, struct rexp_node *, + struct rexp_node *); +RX_DECL struct rexp_node + *rx_mk_r_alternate (struct rx *, struct rexp_node *, + struct rexp_node *); +RX_DECL struct rexp_node + *rx_mk_r_opt (struct rx *, struct rexp_node *); +RX_DECL struct rexp_node + *rx_mk_r_star (struct rx *, struct rexp_node *); +RX_DECL struct rexp_node + *rx_mk_r_2phase_star (struct rx *, struct rexp_node *, + struct rexp_node *); +RX_DECL struct rexp_node + *rx_mk_r_side_effect (struct rx *, rx_side_effect); +RX_DECL struct rexp_node + *rx_mk_r_data (struct rx *, void *); +RX_DECL void rx_free_rexp (struct rx *, struct rexp_node *); +RX_DECL struct rexp_node + *rx_copy_rexp (struct rx *, struct rexp_node *); +RX_DECL struct rx_nfa_state + *rx_nfa_state (struct rx *); +RX_DECL void rx_free_nfa_state (struct rx_nfa_state *); +RX_DECL struct rx_nfa_state + *rx_id_to_nfa_state (struct rx *, int); +RX_DECL struct rx_nfa_edge + *rx_nfa_edge (struct rx *, enum rx_nfa_etype, + struct rx_nfa_state *, + struct rx_nfa_state *); +RX_DECL void rx_free_nfa_edge (struct rx_nfa_edge *); +static struct rx_possible_future + *rx_possible_future (struct rx *, struct rx_se_list *); +static void rx_free_possible_future (struct rx_possible_future *); +RX_DECL void rx_free_nfa (struct rx *); +RX_DECL int rx_build_nfa (struct rx *, struct rexp_node *, + struct rx_nfa_state **, + struct rx_nfa_state **); +RX_DECL void rx_name_nfa_states (struct rx *); +static int se_list_cmp (void *, void *); +static int se_list_equal (void *, void *); +static struct rx_se_list + *hash_cons_se_prog (struct rx *, struct rx_hash *, + void *, struct rx_se_list *); +static struct rx_se_list + *hash_se_prog (struct rx *, struct rx_hash *, + struct rx_se_list *); +static int nfa_set_cmp (void *, void *); +static int nfa_set_equal (void *, void *); +static struct rx_nfa_state_set + *nfa_set_cons (struct rx *, struct rx_hash *, + struct rx_nfa_state *, + struct rx_nfa_state_set *); +static struct rx_nfa_state_set + *nfa_set_enjoin (struct rx *, struct rx_hash *, + struct rx_nfa_state *, + struct rx_nfa_state_set *); +#endif + +#ifndef emacs + +#ifdef SYNTAX_TABLE +extern char *re_syntax_table; +#else /* not SYNTAX_TABLE */ + +#ifndef RX_WANT_RX_DEFS +RX_DECL char re_syntax_table[CHAR_SET_SIZE]; +#endif + +#ifdef __STDC__ +static void +init_syntax_once (void) +#else +static void +init_syntax_once () +#endif +{ + register int c; + static int done = 0; + + if (done) + return; + + bzero (re_syntax_table, sizeof re_syntax_table); + + for (c = 'a'; c <= 'z'; c++) + re_syntax_table[c] = Sword; + + for (c = 'A'; c <= 'Z'; c++) + re_syntax_table[c] = Sword; + + for (c = '0'; c <= '9'; c++) + re_syntax_table[c] = Sword; + + re_syntax_table['_'] = Sword; + + done = 1; +} +#endif /* not SYNTAX_TABLE */ +#endif /* not emacs */ + +/* Compile with `-DRX_DEBUG' and use the following flags. + * + * Debugging flags: + * rx_debug - print information as a regexp is compiled + * rx_debug_trace - print information as a regexp is executed + */ + +#ifdef RX_DEBUG + +int rx_debug_compile = 0; +int rx_debug_trace = 0; +static struct re_pattern_buffer * dbug_rxb = 0; + + +/* + * More Prototypes + */ +#ifdef __STDC__ +typedef void (*side_effect_printer) (struct rx *, void *, FILE *); +static void print_cset (struct rx *, rx_Bitset, FILE *); +static void print_rexp (struct rx *, struct rexp_node *, int, + side_effect_printer, FILE *); +static void print_nfa (struct rx *, struct rx_nfa_state *, + side_effect_printer, FILE *); +static void re_seprint (struct rx *, void *, FILE *); +void print_compiled_pattern (struct re_pattern_buffer *); +void print_fastmap (char *); +#else +typedef void (*side_effect_printer) (); +static void print_cset (); +#endif + +#ifdef __STDC__ +static void +print_rexp (struct rx *rx, + struct rexp_node *node, int depth, + side_effect_printer seprint, FILE * fp) +#else +static void +print_rexp (rx, node, depth, seprint, fp) + struct rx *rx; + struct rexp_node *node; + int depth; + side_effect_printer seprint; + FILE * fp; +#endif +{ + if (!node) + return; + else + { + switch (node->type) + { + case r_cset: + { + fprintf (fp, "%*s", depth, ""); + print_cset (rx, node->params.cset, fp); + fputc ('\n', fp); + break; + } + + case r_opt: + case r_star: + fprintf (fp, "%*s%s\n", depth, "", + node->type == r_opt ? "opt" : "star"); + print_rexp (rx, node->params.pair.left, depth + 3, seprint, fp); + break; + + case r_2phase_star: + fprintf (fp, "%*s2phase star\n", depth, ""); + print_rexp (rx, node->params.pair.right, depth + 3, seprint, fp); + print_rexp (rx, node->params.pair.left, depth + 3, seprint, fp); + break; + + + case r_alternate: + case r_concat: + fprintf (fp, "%*s%s\n", depth, "", + node->type == r_alternate ? "alt" : "concat"); + print_rexp (rx, node->params.pair.left, depth + 3, seprint, fp); + print_rexp (rx, node->params.pair.right, depth + 3, seprint, fp); + break; + case r_side_effect: + fprintf (fp, "%*sSide effect: ", depth, ""); + seprint (rx, node->params.side_effect, fp); + fputc ('\n', fp); + } + } +} + +#ifdef __STDC__ +static void +print_nfa (struct rx * rx, + struct rx_nfa_state * n, + side_effect_printer seprint, FILE * fp) +#else +static void +print_nfa (rx, n, seprint, fp) + struct rx * rx; + struct rx_nfa_state * n; + side_effect_printer seprint; + FILE * fp; +#endif +{ + while (n) + { + struct rx_nfa_edge *e = n->edges; + struct rx_possible_future *ec = n->futures; + fprintf (fp, "node %d %s\n", n->id, + n->is_final ? "final" : (n->is_start ? "start" : "")); + while (e) + { + fprintf (fp, " edge to %d, ", e->dest->id); + switch (e->type) + { + case ne_epsilon: + fprintf (fp, "epsilon\n"); + break; + case ne_side_effect: + fprintf (fp, "side effect "); + seprint (rx, e->params.side_effect, fp); + fputc ('\n', fp); + break; + case ne_cset: + fprintf (fp, "cset "); + print_cset (rx, e->params.cset, fp); + fputc ('\n', fp); + break; + } + e = e->next; + } + + while (ec) + { + int x; + struct rx_nfa_state_set * s; + struct rx_se_list * l; + fprintf (fp, " eclosure to {"); + for (s = ec->destset; s; s = s->cdr) + fprintf (fp, "%d ", s->car->id); + fprintf (fp, "} ("); + for (l = ec->effects; l; l = l->cdr) + { + seprint (rx, l->car, fp); + fputc (' ', fp); + } + fprintf (fp, ")\n"); + ec = ec->next; + } + n = n->next; + } +} + +static char * efnames [] = +{ + "bogon", + "re_se_try", + "re_se_pushback", + "re_se_push0", + "re_se_pushpos", + "re_se_chkpos", + "re_se_poppos", + "re_se_at_dot", + "re_se_syntax", + "re_se_not_syntax", + "re_se_begbuf", + "re_se_hat", + "re_se_wordbeg", + "re_se_wordbound", + "re_se_notwordbound", + "re_se_wordend", + "re_se_endbuf", + "re_se_dollar", + "re_se_fail", +}; + +static char * efnames2[] = +{ + "re_se_win", + "re_se_lparen", + "re_se_rparen", + "re_se_backref", + "re_se_iter", + "re_se_end_iter", + "re_se_tv" +}; + +static char * inx_names[] = +{ + "rx_backtrack_point", + "rx_do_side_effects", + "rx_cache_miss", + "rx_next_char", + "rx_backtrack", + "rx_error_inx", + "rx_num_instructions" +}; + + +#ifdef __STDC__ +static void +re_seprint (struct rx * rx, void * effect, FILE * fp) +#else +static void +re_seprint (rx, effect, fp) + struct rx * rx; + void * effect; + FILE * fp; +#endif +{ + if ((int)effect < 0) + fputs (efnames[-(int)effect], fp); + else if (dbug_rxb) + { + struct re_se_params * p = &dbug_rxb->se_params[(int)effect]; + fprintf (fp, "%s(%d,%d)", efnames2[p->se], p->op1, p->op2); + } + else + fprintf (fp, "[complex op # %d]", (int)effect); +} + +/* These are so the regex.c regression tests will compile. */ +void +print_compiled_pattern (rxb) + struct re_pattern_buffer * rxb; +{ +} + +void +print_fastmap (fm) + char * fm; +{ +} + +#endif /* RX_DEBUG */ + + + +/* This page: Bitsets. Completely unintersting. */ + +RX_DECL int rx_bitset_is_equal (int, rx_Bitset, rx_Bitset); +RX_DECL int rx_bitset_is_subset (int, rx_Bitset, rx_Bitset); +RX_DECL int rx_bitset_empty (int, rx_Bitset); +RX_DECL void rx_bitset_null (int, rx_Bitset); +RX_DECL void rx_bitset_complement (int, rx_Bitset); +RX_DECL void rx_bitset_complement (int, rx_Bitset); +RX_DECL void rx_bitset_assign (int, rx_Bitset, rx_Bitset); +RX_DECL void rx_bitset_union (int, rx_Bitset, rx_Bitset); +RX_DECL void rx_bitset_intersection (int, rx_Bitset, rx_Bitset); +RX_DECL void rx_bitset_difference (int, rx_Bitset, rx_Bitset); +RX_DECL void rx_bitset_revdifference (int, rx_Bitset, rx_Bitset); +RX_DECL void rx_bitset_xor (int, rx_Bitset, rx_Bitset); +RX_DECL unsigned long + rx_bitset_hash (int, rx_Bitset); + +#ifdef __STDC__ +RX_DECL int +rx_bitset_is_equal (int size, rx_Bitset a, rx_Bitset b) +#else +RX_DECL int +rx_bitset_is_equal (size, a, b) + int size; + rx_Bitset a; + rx_Bitset b; +#endif +{ + int x; + RX_subset s = b[0]; + b[0] = ~a[0]; + + for (x = rx_bitset_numb_subsets(size) - 1; a[x] == b[x]; --x) + ; + + b[0] = s; + return !x && s == a[0]; +} + +#ifdef __STDC__ +RX_DECL int +rx_bitset_is_subset (int size, rx_Bitset a, rx_Bitset b) +#else +RX_DECL int +rx_bitset_is_subset (size, a, b) + int size; + rx_Bitset a; + rx_Bitset b; +#endif +{ + int x = rx_bitset_numb_subsets(size) - 1; + while (x-- && (a[x] & b[x]) == a[x]); + return x == -1; +} + + +#ifdef __STDC__ +RX_DECL int +rx_bitset_empty (int size, rx_Bitset set) +#else +RX_DECL int +rx_bitset_empty (size, set) + int size; + rx_Bitset set; +#endif +{ + int x; + RX_subset s = set[0]; + set[0] = 1; + for (x = rx_bitset_numb_subsets(size) - 1; !set[x]; --x) + ; + set[0] = s; + return !s; +} + +#ifdef __STDC__ +RX_DECL void +rx_bitset_null (int size, rx_Bitset b) +#else +RX_DECL void +rx_bitset_null (size, b) + int size; + rx_Bitset b; +#endif +{ + bzero (b, rx_sizeof_bitset(size)); +} + + +#ifdef __STDC__ +RX_DECL void +rx_bitset_universe (int size, rx_Bitset b) +#else +RX_DECL void +rx_bitset_universe (size, b) + int size; + rx_Bitset b; +#endif +{ + int x = rx_bitset_numb_subsets (size); + while (x--) + *b++ = ~(RX_subset)0; +} + + +#ifdef __STDC__ +RX_DECL void +rx_bitset_complement (int size, rx_Bitset b) +#else +RX_DECL void +rx_bitset_complement (size, b) + int size; + rx_Bitset b; +#endif +{ + int x = rx_bitset_numb_subsets (size); + while (x--) + { + *b = ~*b; + ++b; + } +} + + +#ifdef __STDC__ +RX_DECL void +rx_bitset_assign (int size, rx_Bitset a, rx_Bitset b) +#else +RX_DECL void +rx_bitset_assign (size, a, b) + int size; + rx_Bitset a; + rx_Bitset b; +#endif +{ + int x; + for (x = rx_bitset_numb_subsets(size) - 1; x >=0; --x) + a[x] = b[x]; +} + +#ifdef __STDC__ +RX_DECL void +rx_bitset_union (int size, rx_Bitset a, rx_Bitset b) +#else +RX_DECL void +rx_bitset_union (size, a, b) + int size; + rx_Bitset a; + rx_Bitset b; +#endif +{ + int x; + for (x = rx_bitset_numb_subsets(size) - 1; x >=0; --x) + a[x] |= b[x]; +} + + +#ifdef __STDC__ +RX_DECL void +rx_bitset_intersection (int size, + rx_Bitset a, rx_Bitset b) +#else +RX_DECL void +rx_bitset_intersection (size, a, b) + int size; + rx_Bitset a; + rx_Bitset b; +#endif +{ + int x; + for (x = rx_bitset_numb_subsets(size) - 1; x >=0; --x) + a[x] &= b[x]; +} + + +#ifdef __STDC__ +RX_DECL void +rx_bitset_difference (int size, rx_Bitset a, rx_Bitset b) +#else +RX_DECL void +rx_bitset_difference (size, a, b) + int size; + rx_Bitset a; + rx_Bitset b; +#endif +{ + int x; + for (x = rx_bitset_numb_subsets(size) - 1; x >=0; --x) + a[x] &= ~ b[x]; +} + + +#ifdef __STDC__ +RX_DECL void +rx_bitset_revdifference (int size, + rx_Bitset a, rx_Bitset b) +#else +RX_DECL void +rx_bitset_revdifference (size, a, b) + int size; + rx_Bitset a; + rx_Bitset b; +#endif +{ + int x; + for (x = rx_bitset_numb_subsets(size) - 1; x >=0; --x) + a[x] = ~a[x] & b[x]; +} + +#ifdef __STDC__ +RX_DECL void +rx_bitset_xor (int size, rx_Bitset a, rx_Bitset b) +#else +RX_DECL void +rx_bitset_xor (size, a, b) + int size; + rx_Bitset a; + rx_Bitset b; +#endif +{ + int x; + for (x = rx_bitset_numb_subsets(size) - 1; x >=0; --x) + a[x] ^= b[x]; +} + + +#ifdef __STDC__ +RX_DECL unsigned long +rx_bitset_hash (int size, rx_Bitset b) +#else +RX_DECL unsigned long +rx_bitset_hash (size, b) + int size; + rx_Bitset b; +#endif +{ + int x; + unsigned long hash = (unsigned long)rx_bitset_hash; + + for (x = rx_bitset_numb_subsets(size) - 1; x >= 0; --x) + hash ^= rx_bitset_subset_val(b, x); + + return hash; +} + +RX_DECL RX_subset rx_subset_singletons [RX_subset_bits] = +{ + 0x1, + 0x2, + 0x4, + 0x8, + 0x10, + 0x20, + 0x40, + 0x80, + 0x100, + 0x200, + 0x400, + 0x800, + 0x1000, + 0x2000, + 0x4000, + 0x8000, + 0x10000, + 0x20000, + 0x40000, + 0x80000, + 0x100000, + 0x200000, + 0x400000, + 0x800000, + 0x1000000, + 0x2000000, + 0x4000000, + 0x8000000, + 0x10000000, + 0x20000000, + 0x40000000, + 0x80000000 +}; + +#ifdef RX_DEBUG + +#ifdef __STDC__ +static void +print_cset (struct rx *rx, rx_Bitset cset, FILE * fp) +#else +static void +print_cset (rx, cset, fp) + struct rx *rx; + rx_Bitset cset; + FILE * fp; +#endif +{ + int x; + fputc ('[', fp); + for (x = 0; x < rx->local_cset_size; ++x) + if (RX_bitset_member (cset, x)) + { + if (isprint(x)) + fputc (x, fp); + else + fprintf (fp, "\\0%o ", x); + } + fputc (']', fp); +} + +#endif /* RX_DEBUG */ + + + +static unsigned long rx_hash_masks[4] = +{ + 0x12488421, + 0x96699669, + 0xbe7dd7eb, + 0xffffffff +}; + + +/* Hash tables */ +#ifdef __STDC__ +RX_DECL struct rx_hash_item * +rx_hash_find (struct rx_hash * table, + unsigned long hash, + void * value, + struct rx_hash_rules * rules) +#else +RX_DECL struct rx_hash_item * +rx_hash_find (table, hash, value, rules) + struct rx_hash * table; + unsigned long hash; + void * value; + struct rx_hash_rules * rules; +#endif +{ + rx_hash_eq eq = rules->eq; + int maskc = 0; + long mask = rx_hash_masks [0]; + int bucket = (hash & mask) % 13; + + while (table->children [bucket]) + { + table = table->children [bucket]; + ++maskc; + mask = rx_hash_masks[maskc]; + bucket = (hash & mask) % 13; + } + + { + struct rx_hash_item * it = table->buckets[bucket]; + while (it) + if (eq (it->data, value)) + return it; + else + it = it->next_same_hash; + } + + return 0; +} + +#ifdef __STDC__ +RX_DECL struct rx_hash_item * +rx_hash_store (struct rx_hash * table, + unsigned long hash, + void * value, + struct rx_hash_rules * rules) +#else +RX_DECL struct rx_hash_item * +rx_hash_store (table, hash, value, rules) + struct rx_hash * table; + unsigned long hash; + void * value; + struct rx_hash_rules * rules; +#endif +{ + rx_hash_eq eq = rules->eq; + int maskc = 0; + long mask = rx_hash_masks[0]; + int bucket = (hash & mask) % 13; + int depth = 0; + + while (table->children [bucket]) + { + table = table->children [bucket]; + ++maskc; + mask = rx_hash_masks[maskc]; + bucket = (hash & mask) % 13; + ++depth; + } + + { + struct rx_hash_item * it = table->buckets[bucket]; + while (it) + if (eq (it->data, value)) + return it; + else + it = it->next_same_hash; + } + + { + if ( (depth < 3) + && (table->bucket_size [bucket] >= 4)) + { + struct rx_hash * newtab = ((struct rx_hash *) + rules->hash_alloc (rules)); + if (!newtab) + goto add_to_bucket; + bzero (newtab, sizeof (*newtab)); + newtab->parent = table; + { + struct rx_hash_item * them = table->buckets[bucket]; + unsigned long newmask = rx_hash_masks[maskc + 1]; + while (them) + { + struct rx_hash_item * save = them->next_same_hash; + int new_buck = (them->hash & newmask) % 13; + them->next_same_hash = newtab->buckets[new_buck]; + newtab->buckets[new_buck] = them; + them->table = newtab; + them = save; + ++newtab->bucket_size[new_buck]; + ++newtab->refs; + } + table->refs = (table->refs - table->bucket_size[bucket] + 1); + table->bucket_size[bucket] = 0; + table->buckets[bucket] = 0; + table->children[bucket] = newtab; + table = newtab; + bucket = (hash & newmask) % 13; + } + } + } + add_to_bucket: + { + struct rx_hash_item * it = ((struct rx_hash_item *) + rules->hash_item_alloc (rules, value)); + if (!it) + return 0; + it->hash = hash; + it->table = table; + /* DATA and BINDING are to be set in hash_item_alloc */ + it->next_same_hash = table->buckets [bucket]; + table->buckets[bucket] = it; + ++table->bucket_size [bucket]; + ++table->refs; + return it; + } +} + + +#ifdef __STDC__ +RX_DECL void +rx_hash_free (struct rx_hash_item * it, struct rx_hash_rules * rules) +#else +RX_DECL void +rx_hash_free (it, rules) + struct rx_hash_item * it; + struct rx_hash_rules * rules; +#endif +{ + if (it) + { + struct rx_hash * table = it->table; + unsigned long hash = it->hash; + int depth = (table->parent + ? (table->parent->parent + ? (table->parent->parent->parent + ? 3 + : 2) + : 1) + : 0); + int bucket = (hash & rx_hash_masks [depth]) % 13; + struct rx_hash_item ** pos = &table->buckets [bucket]; + + while (*pos != it) + pos = &(*pos)->next_same_hash; + *pos = it->next_same_hash; + rules->free_hash_item (it, rules); + --table->bucket_size[bucket]; + --table->refs; + while (!table->refs && depth) + { + struct rx_hash * save = table; + table = table->parent; + --depth; + bucket = (hash & rx_hash_masks [depth]) % 13; + --table->refs; + table->children[bucket] = 0; + rules->free_hash (save, rules); + } + } +} + +#ifdef __STDC__ +RX_DECL void +rx_free_hash_table (struct rx_hash * tab, rx_hash_freefn freefn, + struct rx_hash_rules * rules) +#else +RX_DECL void +rx_free_hash_table (tab, freefn, rules) + struct rx_hash * tab; + rx_hash_freefn freefn; + struct rx_hash_rules * rules; +#endif +{ + int x; + + for (x = 0; x < 13; ++x) + if (tab->children[x]) + { + rx_free_hash_table (tab->children[x], freefn, rules); + rules->free_hash (tab->children[x], rules); + } + else + { + struct rx_hash_item * them = tab->buckets[x]; + while (them) + { + struct rx_hash_item * that = them; + them = that->next_same_hash; + freefn (that); + rules->free_hash_item (that, rules); + } + } +} + + + +/* Utilities for manipulating bitset represntations of characters sets. */ + +#ifdef __STDC__ +RX_DECL rx_Bitset +rx_cset (struct rx *rx) +#else +RX_DECL rx_Bitset +rx_cset (rx) + struct rx *rx; +#endif +{ + rx_Bitset b = (rx_Bitset) malloc (rx_sizeof_bitset (rx->local_cset_size)); + if (b) + rx_bitset_null (rx->local_cset_size, b); + return b; +} + + +#ifdef __STDC__ +RX_DECL rx_Bitset +rx_copy_cset (struct rx *rx, rx_Bitset a) +#else +RX_DECL rx_Bitset +rx_copy_cset (rx, a) + struct rx *rx; + rx_Bitset a; +#endif +{ + rx_Bitset cs = rx_cset (rx); + + if (cs) + rx_bitset_union (rx->local_cset_size, cs, a); + + return cs; +} + + +#ifdef __STDC__ +RX_DECL void +rx_free_cset (struct rx * rx, rx_Bitset c) +#else +RX_DECL void +rx_free_cset (rx, c) + struct rx * rx; + rx_Bitset c; +#endif +{ + if (c) + free ((char *)c); +} + + +/* Hash table memory allocation policy for the regexp compiler */ + +#ifdef __STDC__ +static struct rx_hash * +compiler_hash_alloc (struct rx_hash_rules * rules) +#else +static struct rx_hash * +compiler_hash_alloc (rules) + struct rx_hash_rules * rules; +#endif +{ + return (struct rx_hash *)malloc (sizeof (struct rx_hash)); +} + + +#ifdef __STDC__ +static struct rx_hash_item * +compiler_hash_item_alloc (struct rx_hash_rules * rules, void * value) +#else +static struct rx_hash_item * +compiler_hash_item_alloc (rules, value) + struct rx_hash_rules * rules; + void * value; +#endif +{ + struct rx_hash_item * it; + it = (struct rx_hash_item *)malloc (sizeof (*it)); + if (it) + { + it->data = value; + it->binding = 0; + } + return it; +} + +#ifdef __STDC__ +static void +compiler_free_hash (struct rx_hash * tab, + struct rx_hash_rules * rules) +#else +static void +compiler_free_hash (tab, rules) + struct rx_hash * tab; + struct rx_hash_rules * rules; +#endif +{ + free ((char *)tab); +} + + +#ifdef __STDC__ +static void +compiler_free_hash_item (struct rx_hash_item * item, + struct rx_hash_rules * rules) +#else +static void +compiler_free_hash_item (item, rules) + struct rx_hash_item * item; + struct rx_hash_rules * rules; +#endif +{ + free ((char *)item); +} + + +/* This page: REXP_NODE (expression tree) structures. */ + +#ifdef __STDC__ +RX_DECL struct rexp_node * +rexp_node (struct rx *rx, + enum rexp_node_type type) +#else +RX_DECL struct rexp_node * +rexp_node (rx, type) + struct rx *rx; + enum rexp_node_type type; +#endif +{ + struct rexp_node *n; + + n = (struct rexp_node *)malloc (sizeof (*n)); + if (n) + { + bzero (n, sizeof (*n)); + n->type = type; + } + return n; +} + + +/* free_rexp_node assumes that the bitset passed to rx_mk_r_cset + * can be freed using rx_free_cset. + */ +#ifdef __STDC__ +RX_DECL struct rexp_node * +rx_mk_r_cset (struct rx * rx, + rx_Bitset b) +#else +RX_DECL struct rexp_node * +rx_mk_r_cset (rx, b) + struct rx * rx; + rx_Bitset b; +#endif +{ + struct rexp_node * n = rexp_node (rx, r_cset); + if (n) + n->params.cset = b; + return n; +} + +#ifdef __STDC__ +RX_DECL struct rexp_node * +rx_mk_r_concat (struct rx * rx, + struct rexp_node * a, + struct rexp_node * b) +#else +RX_DECL struct rexp_node * +rx_mk_r_concat (rx, a, b) + struct rx * rx; + struct rexp_node * a; + struct rexp_node * b; +#endif +{ + struct rexp_node * n = rexp_node (rx, r_concat); + if (n) + { + n->params.pair.left = a; + n->params.pair.right = b; + } + return n; +} + + +#ifdef __STDC__ +RX_DECL struct rexp_node * +rx_mk_r_alternate (struct rx * rx, + struct rexp_node * a, + struct rexp_node * b) +#else +RX_DECL struct rexp_node * +rx_mk_r_alternate (rx, a, b) + struct rx * rx; + struct rexp_node * a; + struct rexp_node * b; +#endif +{ + struct rexp_node * n = rexp_node (rx, r_alternate); + if (n) + { + n->params.pair.left = a; + n->params.pair.right = b; + } + return n; +} + + +#ifdef __STDC__ +RX_DECL struct rexp_node * +rx_mk_r_opt (struct rx * rx, + struct rexp_node * a) +#else +RX_DECL struct rexp_node * +rx_mk_r_opt (rx, a) + struct rx * rx; + struct rexp_node * a; +#endif +{ + struct rexp_node * n = rexp_node (rx, r_opt); + if (n) + { + n->params.pair.left = a; + n->params.pair.right = 0; + } + return n; +} + +#ifdef __STDC__ +RX_DECL struct rexp_node * +rx_mk_r_star (struct rx * rx, + struct rexp_node * a) +#else +RX_DECL struct rexp_node * +rx_mk_r_star (rx, a) + struct rx * rx; + struct rexp_node * a; +#endif +{ + struct rexp_node * n = rexp_node (rx, r_star); + if (n) + { + n->params.pair.left = a; + n->params.pair.right = 0; + } + return n; +} + + +#ifdef __STDC__ +RX_DECL struct rexp_node * +rx_mk_r_2phase_star (struct rx * rx, + struct rexp_node * a, + struct rexp_node * b) +#else +RX_DECL struct rexp_node * +rx_mk_r_2phase_star (rx, a, b) + struct rx * rx; + struct rexp_node * a; + struct rexp_node * b; +#endif +{ + struct rexp_node * n = rexp_node (rx, r_2phase_star); + if (n) + { + n->params.pair.left = a; + n->params.pair.right = b; + } + return n; +} + +#ifdef __STDC__ +RX_DECL struct rexp_node * +rx_mk_r_side_effect (struct rx * rx, + rx_side_effect a) +#else +RX_DECL struct rexp_node * +rx_mk_r_side_effect (rx, a) + struct rx * rx; + rx_side_effect a; +#endif +{ + struct rexp_node * n = rexp_node (rx, r_side_effect); + if (n) + { + n->params.side_effect = a; + n->params.pair.right = 0; + } + return n; +} + + +#ifdef __STDC__ +RX_DECL struct rexp_node * +rx_mk_r_data (struct rx * rx, + void * a) +#else +RX_DECL struct rexp_node * +rx_mk_r_data (rx, a) + struct rx * rx; + void * a; +#endif +{ + struct rexp_node * n = rexp_node (rx, r_data); + if (n) + { + n->params.pair.left = a; + n->params.pair.right = 0; + } + return n; +} + +#ifdef __STDC__ +RX_DECL void +rx_free_rexp (struct rx * rx, struct rexp_node * node) +#else +RX_DECL void +rx_free_rexp (rx, node) + struct rx * rx; + struct rexp_node * node; +#endif +{ + if (node) + { + switch (node->type) + { + case r_cset: + if (node->params.cset) + rx_free_cset (rx, node->params.cset); + + case r_side_effect: + break; + + case r_concat: + case r_alternate: + case r_2phase_star: + case r_opt: + case r_star: + rx_free_rexp (rx, node->params.pair.left); + rx_free_rexp (rx, node->params.pair.right); + break; + + case r_data: + /* This shouldn't occur. */ + break; + } + free ((char *)node); + } +} + +#ifdef __STDC__ +RX_DECL struct rexp_node * +rx_copy_rexp (struct rx *rx, + struct rexp_node *node) +#else +RX_DECL struct rexp_node * +rx_copy_rexp (rx, node) + struct rx *rx; + struct rexp_node *node; +#endif +{ + if (!node) + return 0; + else + { + struct rexp_node *n = rexp_node (rx, node->type); + if (!n) + return 0; + switch (node->type) + { + case r_cset: + n->params.cset = rx_copy_cset (rx, node->params.cset); + if (!n->params.cset) + { + rx_free_rexp (rx, n); + return 0; + } + break; + + case r_side_effect: + n->params.side_effect = node->params.side_effect; + break; + + case r_concat: + case r_alternate: + case r_opt: + case r_2phase_star: + case r_star: + n->params.pair.left = + rx_copy_rexp (rx, node->params.pair.left); + n->params.pair.right = + rx_copy_rexp (rx, node->params.pair.right); + if ( (node->params.pair.left && !n->params.pair.left) + || (node->params.pair.right && !n->params.pair.right)) + { + rx_free_rexp (rx, n); + return 0; + } + break; + case r_data: + /* shouldn't happen */ + break; + } + return n; + } +} + + + +/* This page: functions to build and destroy graphs that describe nfa's */ + +/* Constructs a new nfa node. */ +#ifdef __STDC__ +RX_DECL struct rx_nfa_state * +rx_nfa_state (struct rx *rx) +#else +RX_DECL struct rx_nfa_state * +rx_nfa_state (rx) + struct rx *rx; +#endif +{ + struct rx_nfa_state * n = (struct rx_nfa_state *)malloc (sizeof (*n)); + if (!n) + return 0; + bzero (n, sizeof (*n)); + n->next = rx->nfa_states; + rx->nfa_states = n; + return n; +} + + +#ifdef __STDC__ +RX_DECL void +rx_free_nfa_state (struct rx_nfa_state * n) +#else +RX_DECL void +rx_free_nfa_state (n) + struct rx_nfa_state * n; +#endif +{ + free ((char *)n); +} + + +/* This looks up an nfa node, given a numeric id. Numeric id's are + * assigned after the nfa has been built. + */ +#ifdef __STDC__ +RX_DECL struct rx_nfa_state * +rx_id_to_nfa_state (struct rx * rx, + int id) +#else +RX_DECL struct rx_nfa_state * +rx_id_to_nfa_state (rx, id) + struct rx * rx; + int id; +#endif +{ + struct rx_nfa_state * n; + for (n = rx->nfa_states; n; n = n->next) + if (n->id == id) + return n; + return 0; +} + + +/* This adds an edge between two nodes, but doesn't initialize the + * edge label. + */ + +#ifdef __STDC__ +RX_DECL struct rx_nfa_edge * +rx_nfa_edge (struct rx *rx, + enum rx_nfa_etype type, + struct rx_nfa_state *start, + struct rx_nfa_state *dest) +#else +RX_DECL struct rx_nfa_edge * +rx_nfa_edge (rx, type, start, dest) + struct rx *rx; + enum rx_nfa_etype type; + struct rx_nfa_state *start; + struct rx_nfa_state *dest; +#endif +{ + struct rx_nfa_edge *e; + e = (struct rx_nfa_edge *)malloc (sizeof (*e)); + if (!e) + return 0; + e->next = start->edges; + start->edges = e; + e->type = type; + e->dest = dest; + return e; +} + + +#ifdef __STDC__ +RX_DECL void +rx_free_nfa_edge (struct rx_nfa_edge * e) +#else +RX_DECL void +rx_free_nfa_edge (e) + struct rx_nfa_edge * e; +#endif +{ + free ((char *)e); +} + + +/* This constructs a POSSIBLE_FUTURE, which is a kind epsilon-closure + * of an NFA. These are added to an nfa automaticly by eclose_nfa. + */ + +#ifdef __STDC__ +static struct rx_possible_future * +rx_possible_future (struct rx * rx, + struct rx_se_list * effects) +#else +static struct rx_possible_future * +rx_possible_future (rx, effects) + struct rx * rx; + struct rx_se_list * effects; +#endif +{ + struct rx_possible_future *ec; + ec = (struct rx_possible_future *) malloc (sizeof (*ec)); + if (!ec) + return 0; + ec->destset = 0; + ec->next = 0; + ec->effects = effects; + return ec; +} + + +#ifdef __STDC__ +static void +rx_free_possible_future (struct rx_possible_future * pf) +#else +static void +rx_free_possible_future (pf) + struct rx_possible_future * pf; +#endif +{ + free ((char *)pf); +} + + +#ifdef __STDC__ +RX_DECL void +rx_free_nfa (struct rx *rx) +#else +RX_DECL void +rx_free_nfa (rx) + struct rx *rx; +#endif +{ + while (rx->nfa_states) + { + while (rx->nfa_states->edges) + { + switch (rx->nfa_states->edges->type) + { + case ne_cset: + rx_free_cset (rx, rx->nfa_states->edges->params.cset); + break; + default: + break; + } + { + struct rx_nfa_edge * e; + e = rx->nfa_states->edges; + rx->nfa_states->edges = rx->nfa_states->edges->next; + rx_free_nfa_edge (e); + } + } /* while (rx->nfa_states->edges) */ + { + /* Iterate over the partial epsilon closures of rx->nfa_states */ + struct rx_possible_future * pf = rx->nfa_states->futures; + while (pf) + { + struct rx_possible_future * pft = pf; + pf = pf->next; + rx_free_possible_future (pft); + } + } + { + struct rx_nfa_state *n; + n = rx->nfa_states; + rx->nfa_states = rx->nfa_states->next; + rx_free_nfa_state (n); + } + } +} + + + +/* This page: translating a pattern expression into an nfa and doing the + * static part of the nfa->super-nfa translation. + */ + +/* This is the thompson regexp->nfa algorithm. + * It is modified to allow for `side-effect epsilons.' Those are + * edges that are taken whenever a similar epsilon edge would be, + * but which imply that some side effect occurs when the edge + * is taken. + * + * Side effects are used to model parts of the pattern langauge + * that are not regular (in the formal sense). + */ + +#ifdef __STDC__ +RX_DECL int +rx_build_nfa (struct rx *rx, + struct rexp_node *rexp, + struct rx_nfa_state **start, + struct rx_nfa_state **end) +#else +RX_DECL int +rx_build_nfa (rx, rexp, start, end) + struct rx *rx; + struct rexp_node *rexp; + struct rx_nfa_state **start; + struct rx_nfa_state **end; +#endif +{ + struct rx_nfa_edge *edge; + + /* Start & end nodes may have been allocated by the caller. */ + *start = *start ? *start : rx_nfa_state (rx); + + if (!*start) + return 0; + + if (!rexp) + { + *end = *start; + return 1; + } + + *end = *end ? *end : rx_nfa_state (rx); + + if (!*end) + { + rx_free_nfa_state (*start); + return 0; + } + + switch (rexp->type) + { + case r_data: + return 0; + + case r_cset: + edge = rx_nfa_edge (rx, ne_cset, *start, *end); + if (!edge) + return 0; + edge->params.cset = rx_copy_cset (rx, rexp->params.cset); + if (!edge->params.cset) + { + rx_free_nfa_edge (edge); + return 0; + } + return 1; + + case r_opt: + return (rx_build_nfa (rx, rexp->params.pair.left, start, end) + && rx_nfa_edge (rx, ne_epsilon, *start, *end)); + + case r_star: + { + struct rx_nfa_state * star_start = 0; + struct rx_nfa_state * star_end = 0; + return (rx_build_nfa (rx, rexp->params.pair.left, + &star_start, &star_end) + && star_start + && star_end + && rx_nfa_edge (rx, ne_epsilon, star_start, star_end) + && rx_nfa_edge (rx, ne_epsilon, *start, star_start) + && rx_nfa_edge (rx, ne_epsilon, star_end, *end) + + && rx_nfa_edge (rx, ne_epsilon, star_end, star_start)); + } + + case r_2phase_star: + { + struct rx_nfa_state * star_start = 0; + struct rx_nfa_state * star_end = 0; + struct rx_nfa_state * loop_exp_start = 0; + struct rx_nfa_state * loop_exp_end = 0; + + return (rx_build_nfa (rx, rexp->params.pair.left, + &star_start, &star_end) + && rx_build_nfa (rx, rexp->params.pair.right, + &loop_exp_start, &loop_exp_end) + && star_start + && star_end + && loop_exp_end + && loop_exp_start + && rx_nfa_edge (rx, ne_epsilon, star_start, *end) + && rx_nfa_edge (rx, ne_epsilon, *start, star_start) + && rx_nfa_edge (rx, ne_epsilon, star_end, *end) + + && rx_nfa_edge (rx, ne_epsilon, star_end, loop_exp_start) + && rx_nfa_edge (rx, ne_epsilon, loop_exp_end, star_start)); + } + + + case r_concat: + { + struct rx_nfa_state *shared = 0; + return + (rx_build_nfa (rx, rexp->params.pair.left, start, &shared) + && rx_build_nfa (rx, rexp->params.pair.right, &shared, end)); + } + + case r_alternate: + { + struct rx_nfa_state *ls = 0; + struct rx_nfa_state *le = 0; + struct rx_nfa_state *rs = 0; + struct rx_nfa_state *re = 0; + return (rx_build_nfa (rx, rexp->params.pair.left, &ls, &le) + && rx_build_nfa (rx, rexp->params.pair.right, &rs, &re) + && rx_nfa_edge (rx, ne_epsilon, *start, ls) + && rx_nfa_edge (rx, ne_epsilon, *start, rs) + && rx_nfa_edge (rx, ne_epsilon, le, *end) + && rx_nfa_edge (rx, ne_epsilon, re, *end)); + } + + case r_side_effect: + edge = rx_nfa_edge (rx, ne_side_effect, *start, *end); + if (!edge) + return 0; + edge->params.side_effect = rexp->params.side_effect; + return 1; + } + + /* this should never happen */ + return 0; +} + + +/* RX_NAME_NFA_STATES identifies all nodes with outgoing non-epsilon + * transitions. Only these nodes can occur in super-states. + * All nodes are given an integer id. + * The id is non-negative if the node has non-epsilon out-transitions, negative + * otherwise (this is because we want the non-negative ids to be used as + * array indexes in a few places). + */ + +#ifdef __STDC__ +RX_DECL void +rx_name_nfa_states (struct rx *rx) +#else +RX_DECL void +rx_name_nfa_states (rx) + struct rx *rx; +#endif +{ + struct rx_nfa_state *n = rx->nfa_states; + + rx->nodec = 0; + rx->epsnodec = -1; + + while (n) + { + struct rx_nfa_edge *e = n->edges; + + if (n->is_start) + n->eclosure_needed = 1; + + while (e) + { + switch (e->type) + { + case ne_epsilon: + case ne_side_effect: + break; + + case ne_cset: + n->id = rx->nodec++; + { + struct rx_nfa_edge *from_n = n->edges; + while (from_n) + { + from_n->dest->eclosure_needed = 1; + from_n = from_n->next; + } + } + goto cont; + } + e = e->next; + } + n->id = rx->epsnodec--; + cont: + n = n->next; + } + rx->epsnodec = -rx->epsnodec; +} + + +/* This page: data structures for the static part of the nfa->supernfa + * translation. + * + * There are side effect lists -- lists of side effects occuring + * along an uninterrupted, acyclic path of side-effect epsilon edges. + * Such paths are collapsed to single edges in the course of computing + * epsilon closures. Such single edges are labled with a list of all + * the side effects entailed in crossing them. Like lists of side + * effects are made == by the constructors below. + * + * There are also nfa state sets. These are used to hold a list of all + * states reachable from a starting state for a given type of transition + * and side effect list. These are also hash-consed. + */ + +/* The next several functions compare, construct, etc. lists of side + * effects. See ECLOSE_NFA (below) for details. + */ + +/* Ordering of rx_se_list + * (-1, 0, 1 return value convention). + */ + +#ifdef __STDC__ +static int +se_list_cmp (void * va, void * vb) +#else +static int +se_list_cmp (va, vb) + void * va; + void * vb; +#endif +{ + struct rx_se_list * a = (struct rx_se_list *)va; + struct rx_se_list * b = (struct rx_se_list *)vb; + + return ((va == vb) + ? 0 + : (!va + ? -1 + : (!vb + ? 1 + : ((long)a->car < (long)b->car + ? 1 + : ((long)a->car > (long)b->car + ? -1 + : se_list_cmp ((void *)a->cdr, (void *)b->cdr)))))); +} + + +#ifdef __STDC__ +static int +se_list_equal (void * va, void * vb) +#else +static int +se_list_equal (va, vb) + void * va; + void * vb; +#endif +{ + return !(se_list_cmp (va, vb)); +} + +static struct rx_hash_rules se_list_hash_rules = +{ + se_list_equal, + compiler_hash_alloc, + compiler_free_hash, + compiler_hash_item_alloc, + compiler_free_hash_item +}; + + +#ifdef __STDC__ +static struct rx_se_list * +side_effect_cons (struct rx * rx, + void * se, struct rx_se_list * list) +#else +static struct rx_se_list * +side_effect_cons (rx, se, list) + struct rx * rx; + void * se; + struct rx_se_list * list; +#endif +{ + struct rx_se_list * l; + l = ((struct rx_se_list *) malloc (sizeof (*l))); + if (!l) + return 0; + l->car = se; + l->cdr = list; + return l; +} + + +#ifdef __STDC__ +static struct rx_se_list * +hash_cons_se_prog (struct rx * rx, + struct rx_hash * memo, + void * car, struct rx_se_list * cdr) +#else +static struct rx_se_list * +hash_cons_se_prog (rx, memo, car, cdr) + struct rx * rx; + struct rx_hash * memo; + void * car; + struct rx_se_list * cdr; +#endif +{ + long hash = (long)car ^ (long)cdr; + struct rx_se_list template; + + template.car = car; + template.cdr = cdr; + { + struct rx_hash_item * it = rx_hash_store (memo, hash, + (void *)&template, + &se_list_hash_rules); + if (!it) + return 0; + if (it->data == (void *)&template) + { + struct rx_se_list * consed; + consed = (struct rx_se_list *) malloc (sizeof (*consed)); + if (! consed) + { + free ((char *)it); + return 0; + } + *consed = template; + it->data = (void *)consed; + } + return (struct rx_se_list *)it->data; + } +} + + +#ifdef __STDC__ +static struct rx_se_list * +hash_se_prog (struct rx * rx, struct rx_hash * memo, struct rx_se_list * prog) +#else +static struct rx_se_list * +hash_se_prog (rx, memo, prog) + struct rx * rx; + struct rx_hash * memo; + struct rx_se_list * prog; +#endif +{ + struct rx_se_list * answer = 0; + while (prog) + { + answer = hash_cons_se_prog (rx, memo, prog->car, answer); + if (!answer) + return 0; + prog = prog->cdr; + } + return answer; +} + +#ifdef __STDC__ +static int +nfa_set_cmp (void * va, void * vb) +#else +static int +nfa_set_cmp (va, vb) + void * va; + void * vb; +#endif +{ + struct rx_nfa_state_set * a = (struct rx_nfa_state_set *)va; + struct rx_nfa_state_set * b = (struct rx_nfa_state_set *)vb; + + return ((va == vb) + ? 0 + : (!va + ? -1 + : (!vb + ? 1 + : (a->car->id < b->car->id + ? 1 + : (a->car->id > b->car->id + ? -1 + : nfa_set_cmp ((void *)a->cdr, (void *)b->cdr)))))); +} + +#ifdef __STDC__ +static int +nfa_set_equal (void * va, void * vb) +#else +static int +nfa_set_equal (va, vb) + void * va; + void * vb; +#endif +{ + return !nfa_set_cmp (va, vb); +} + +static struct rx_hash_rules nfa_set_hash_rules = +{ + nfa_set_equal, + compiler_hash_alloc, + compiler_free_hash, + compiler_hash_item_alloc, + compiler_free_hash_item +}; + + +#ifdef __STDC__ +static struct rx_nfa_state_set * +nfa_set_cons (struct rx * rx, + struct rx_hash * memo, struct rx_nfa_state * state, + struct rx_nfa_state_set * set) +#else +static struct rx_nfa_state_set * +nfa_set_cons (rx, memo, state, set) + struct rx * rx; + struct rx_hash * memo; + struct rx_nfa_state * state; + struct rx_nfa_state_set * set; +#endif +{ + struct rx_nfa_state_set template; + struct rx_hash_item * node; + template.car = state; + template.cdr = set; + node = rx_hash_store (memo, + (((long)state) >> 8) ^ (long)set, + &template, &nfa_set_hash_rules); + if (!node) + return 0; + if (node->data == &template) + { + struct rx_nfa_state_set * l; + l = (struct rx_nfa_state_set *) malloc (sizeof (*l)); + node->data = (void *) l; + if (!l) + return 0; + *l = template; + } + return (struct rx_nfa_state_set *)node->data; +} + +#ifdef __STDC__ +static struct rx_nfa_state_set * +nfa_set_enjoin (struct rx * rx, + struct rx_hash * memo, struct rx_nfa_state * state, + struct rx_nfa_state_set * set) +#else +static struct rx_nfa_state_set * +nfa_set_enjoin (rx, memo, state, set) + struct rx * rx; + struct rx_hash * memo; + struct rx_nfa_state * state; + struct rx_nfa_state_set * set; +#endif +{ + if (!set || state->id < set->car->id) + return nfa_set_cons (rx, memo, state, set); + if (state->id == set->car->id) + return set; + else + { + struct rx_nfa_state_set * newcdr + = nfa_set_enjoin (rx, memo, state, set->cdr); + if (newcdr != set->cdr) + set = nfa_set_cons (rx, memo, set->car, newcdr); + return set; + } +} + + + +/* This page: computing epsilon closures. The closures aren't total. + * Each node's closures are partitioned according to the side effects entailed + * along the epsilon edges. Return true on success. + */ + +struct eclose_frame +{ + struct rx_se_list *prog_backwards; +}; +static int eclose_node (struct rx *, struct rx_nfa_state *, + struct rx_nfa_state *, + struct eclose_frame *); +RX_DECL int rx_eclose_nfa (struct rx *); +RX_DECL void rx_delete_epsilon_transitions + (struct rx *); +static int nfacmp (void *, void *); +static int count_hash_nodes (struct rx_hash *); +static void nfa_set_freer (struct rx_hash_item *); +RX_DECL int rx_compactify_nfa (struct rx *, void **, + unsigned long *); +static char *rx_cache_malloc (struct rx_cache *, int); +static void rx_cache_free (struct rx_cache *, + struct rx_freelist **, char *); +static void install_transition (struct rx_superstate *, + struct rx_inx *, rx_Bitset); +static int qlen (struct rx_superstate *); +static void check_cache (struct rx_cache *); +static void semifree_superstate (struct rx_cache *); +static void refresh_semifree_superstate + (struct rx_cache *, + struct rx_superstate *); +static void rx_refresh_this_superstate + (struct rx_cache *, + struct rx_superstate *); +static void release_superset_low (struct rx_cache *, + struct rx_superset *); +RX_DECL void rx_release_superset (struct rx *, struct rx_superset *); +static int rx_really_free_superstate (struct rx_cache *); +static char *rx_cache_get (struct rx_cache *, + struct rx_freelist **); +static char *rx_cache_malloc_or_get (struct rx_cache *, + struct rx_freelist **, int); +static char *rx_cache_get_superstate (struct rx_cache *); +static int supersetcmp (void *, void *); +static struct rx_hash_item + *superset_allocator (struct rx_hash_rules *, void *); +static struct rx_hash + *super_hash_allocator (struct rx_hash_rules *); +static void super_hash_liberator (struct rx_hash *, + struct rx_hash_rules *); +static void superset_hash_item_liberator + (struct rx_hash_item *, + struct rx_hash_rules *); +static int bytes_for_cache_size (int, int); +static void rx_morecore (struct rx_cache *); +RX_DECL struct rx_superset + *rx_superset_cons (struct rx *, struct rx_nfa_state *, + struct rx_superset *); +RX_DECL struct rx_superset + *rx_superstate_eclosure_union + (struct rx *, struct rx_superset *, + struct rx_nfa_state_set *); +static struct rx_distinct_future + *include_futures (struct rx *, + struct rx_distinct_future *, + struct rx_nfa_state *, + struct rx_superstate *); +RX_DECL struct rx_superstate + *rx_superstate (struct rx *, struct rx_superset *); +static int solve_destination (struct rx *, + struct rx_distinct_future *); +static int compute_super_edge (struct rx *, + struct rx_distinct_future **, + rx_Bitset, struct rx_superstate *, + unsigned char); +static struct rx_super_edge + *rx_super_edge (struct rx *, struct rx_superstate *, + rx_Bitset, + struct rx_distinct_future *); +static void install_partial_transition + (struct rx_superstate *, + struct rx_inx *, RX_subset, int); +RX_DECL struct rx_inx + *rx_handle_cache_miss (struct rx *, struct rx_superstate *, + unsigned char, void *); +static boolean + at_begline_loc_p (__const__ char *, __const__ char *, + reg_syntax_t); +static boolean + at_endline_loc_p (__const__ char *, __const__ char *, + int); +static rx_Bitset + inverse_translation (struct re_pattern_buffer *, char *, + rx_Bitset, unsigned char *, int); + + +#ifdef __STDC__ +static int +eclose_node (struct rx *rx, struct rx_nfa_state *outnode, + struct rx_nfa_state *node, struct eclose_frame *frame) +#else +static int +eclose_node (rx, outnode, node, frame) + struct rx *rx; + struct rx_nfa_state *outnode; + struct rx_nfa_state *node; + struct eclose_frame *frame; +#endif +{ + struct rx_nfa_edge *e = node->edges; + + /* For each node, we follow all epsilon paths to build the closure. + * The closure omits nodes that have only epsilon edges. + * The closure is split into partial closures -- all the states in + * a partial closure are reached by crossing the same list of + * of side effects (though not necessarily the same path). + */ + if (node->mark) + return 1; + node->mark = 1; + + if (node->id >= 0 || node->is_final) + { + struct rx_possible_future **ec; + struct rx_se_list * prog_in_order + = ((struct rx_se_list *)hash_se_prog (rx, + &rx->se_list_memo, + frame->prog_backwards)); + int cmp; + + ec = &outnode->futures; + + while (*ec) + { + cmp = se_list_cmp ((void *)(*ec)->effects, (void *)prog_in_order); + if (cmp <= 0) + break; + ec = &(*ec)->next; + } + if (!*ec || (cmp < 0)) + { + struct rx_possible_future * saved = *ec; + *ec = rx_possible_future (rx, prog_in_order); + (*ec)->next = saved; + if (!*ec) + return 0; + } + if (node->id >= 0) + { + (*ec)->destset = nfa_set_enjoin (rx, &rx->set_list_memo, + node, (*ec)->destset); + if (!(*ec)->destset) + return 0; + } + } + + while (e) + { + switch (e->type) + { + case ne_epsilon: + if (!eclose_node (rx, outnode, e->dest, frame)) + return 0; + break; + case ne_side_effect: + { + frame->prog_backwards = side_effect_cons (rx, + e->params.side_effect, + frame->prog_backwards); + if (!frame->prog_backwards) + return 0; + if (!eclose_node (rx, outnode, e->dest, frame)) + return 0; + { + struct rx_se_list * dying = frame->prog_backwards; + frame->prog_backwards = frame->prog_backwards->cdr; + free ((char *)dying); + } + break; + } + default: + break; + } + e = e->next; + } + node->mark = 0; + return 1; +} + +#ifdef __STDC__ +RX_DECL int +rx_eclose_nfa (struct rx *rx) +#else +RX_DECL int +rx_eclose_nfa (rx) + struct rx *rx; +#endif +{ + struct rx_nfa_state *n = rx->nfa_states; + struct eclose_frame frame; + static int rx_id = 0; + + frame.prog_backwards = 0; + rx->rx_id = rx_id++; + bzero (&rx->se_list_memo, sizeof (rx->se_list_memo)); + bzero (&rx->set_list_memo, sizeof (rx->set_list_memo)); + while (n) + { + n->futures = 0; + if (n->eclosure_needed && !eclose_node (rx, n, n, &frame)) + return 0; + /* clear_marks (rx); */ + n = n->next; + } + return 1; +} + + +/* This deletes epsilon edges from an NFA. After running eclose_node, + * we have no more need for these edges. They are removed to simplify + * further operations on the NFA. + */ + +#ifdef __STDC__ +RX_DECL void +rx_delete_epsilon_transitions (struct rx *rx) +#else +RX_DECL void +rx_delete_epsilon_transitions (rx) + struct rx *rx; +#endif +{ + struct rx_nfa_state *n = rx->nfa_states; + struct rx_nfa_edge **e; + + while (n) + { + e = &n->edges; + while (*e) + { + struct rx_nfa_edge *t; + switch ((*e)->type) + { + case ne_epsilon: + case ne_side_effect: + t = *e; + *e = t->next; + rx_free_nfa_edge (t); + break; + + default: + e = &(*e)->next; + break; + } + } + n = n->next; + } +} + + +/* This page: storing the nfa in a contiguous region of memory for + * subsequent conversion to a super-nfa. + */ + +/* This is for qsort on an array of nfa_states. The order + * is based on state ids and goes + * [0...MAX][MIN..-1] where (MAX>=0) and (MIN<0) + * This way, positive ids double as array indices. + */ + +#ifdef __STDC__ +static int +nfacmp (void * va, void * vb) +#else +static int +nfacmp (va, vb) + void * va; + void * vb; +#endif +{ + struct rx_nfa_state **a = (struct rx_nfa_state **)va; + struct rx_nfa_state **b = (struct rx_nfa_state **)vb; + return (*a == *b /* &&&& 3.18 */ + ? 0 + : (((*a)->id < 0) == ((*b)->id < 0) + ? (((*a)->id < (*b)->id) ? -1 : 1) + : (((*a)->id < 0) + ? 1 : -1))); +} + +#ifdef __STDC__ +static int +count_hash_nodes (struct rx_hash * st) +#else +static int +count_hash_nodes (st) + struct rx_hash * st; +#endif +{ + int x; + int count = 0; + for (x = 0; x < 13; ++x) + count += ((st->children[x]) + ? count_hash_nodes (st->children[x]) + : st->bucket_size[x]); + + return count; +} + + +#ifdef __STDC__ +static void +se_memo_freer (struct rx_hash_item * node) +#else +static void +se_memo_freer (node) + struct rx_hash_item * node; +#endif +{ + free ((char *)node->data); +} + + +#ifdef __STDC__ +static void +nfa_set_freer (struct rx_hash_item * node) +#else +static void +nfa_set_freer (node) + struct rx_hash_item * node; +#endif +{ + free ((char *)node->data); +} + + +/* This copies an entire NFA into a single malloced block of memory. + * Mostly this is for compatability with regex.c, though it is convenient + * to have the nfa nodes in an array. + */ + +#ifdef __STDC__ +RX_DECL int +rx_compactify_nfa (struct rx *rx, + void **mem, unsigned long *size) +#else +RX_DECL int +rx_compactify_nfa (rx, mem, size) + struct rx *rx; + void **mem; + unsigned long *size; +#endif +{ + int total_nodec; + struct rx_nfa_state *n; + int edgec = 0; + int eclosec = 0; + int se_list_consc = count_hash_nodes (&rx->se_list_memo); + int nfa_setc = count_hash_nodes (&rx->set_list_memo); + unsigned long total_size; + + /* This takes place in two stages. First, the total size of the + * nfa is computed, then structures are copied. + */ + n = rx->nfa_states; + total_nodec = 0; + while (n) + { + struct rx_nfa_edge *e = n->edges; + struct rx_possible_future *ec = n->futures; + ++total_nodec; + while (e) + { + ++edgec; + e = e->next; + } + while (ec) + { + ++eclosec; + ec = ec->next; + } + n = n->next; + } + + total_size = (total_nodec * sizeof (struct rx_nfa_state) + + edgec * rx_sizeof_bitset (rx->local_cset_size) + + edgec * sizeof (struct rx_nfa_edge) + + nfa_setc * sizeof (struct rx_nfa_state_set) + + eclosec * sizeof (struct rx_possible_future) + + se_list_consc * sizeof (struct rx_se_list) + + rx->reserved); + + if (total_size > *size) + { + *mem = remalloc (*mem, total_size); + if (*mem) + *size = total_size; + else + return 0; + } + /* Now we've allocated the memory; this copies the NFA. */ + { + static struct rx_nfa_state **scratch = 0; + static int scratch_alloc = 0; + struct rx_nfa_state *state_base = (struct rx_nfa_state *) * mem; + struct rx_nfa_state *new_state = state_base; + struct rx_nfa_edge *new_edge = + (struct rx_nfa_edge *) + ((char *) state_base + total_nodec * sizeof (struct rx_nfa_state)); + struct rx_se_list * new_se_list = + (struct rx_se_list *) + ((char *)new_edge + edgec * sizeof (struct rx_nfa_edge)); + struct rx_possible_future *new_close = + ((struct rx_possible_future *) + ((char *) new_se_list + + se_list_consc * sizeof (struct rx_se_list))); + struct rx_nfa_state_set * new_nfa_set = + ((struct rx_nfa_state_set *) + ((char *)new_close + eclosec * sizeof (struct rx_possible_future))); + char *new_bitset = + ((char *) new_nfa_set + nfa_setc * sizeof (struct rx_nfa_state_set)); + int x; + struct rx_nfa_state *n; + + if (scratch_alloc < total_nodec) + { + scratch = ((struct rx_nfa_state **) + remalloc (scratch, total_nodec * sizeof (*scratch))); + if (scratch) + scratch_alloc = total_nodec; + else + { + scratch_alloc = 0; + return 0; + } + } + + for (x = 0, n = rx->nfa_states; n; n = n->next) + scratch[x++] = n; + + qsort (scratch, total_nodec, sizeof (struct rx_nfa_state *), + (__compar_fn_t)nfacmp); + + for (x = 0; x < total_nodec; ++x) + { + struct rx_possible_future *eclose = scratch[x]->futures; + struct rx_nfa_edge *edge = scratch[x]->edges; + struct rx_nfa_state *cn = new_state++; + cn->futures = 0; + cn->edges = 0; + cn->next = (x == total_nodec - 1) ? 0 : (cn + 1); + cn->id = scratch[x]->id; + cn->is_final = scratch[x]->is_final; + cn->is_start = scratch[x]->is_start; + cn->mark = 0; + while (edge) + { + int indx = (edge->dest->id < 0 + ? (total_nodec + edge->dest->id) + : edge->dest->id); + struct rx_nfa_edge *e = new_edge++; + rx_Bitset cset = (rx_Bitset) new_bitset; + new_bitset += rx_sizeof_bitset (rx->local_cset_size); + rx_bitset_null (rx->local_cset_size, cset); + rx_bitset_union (rx->local_cset_size, cset, edge->params.cset); + e->next = cn->edges; + cn->edges = e; + e->type = edge->type; + e->dest = state_base + indx; + e->params.cset = cset; + edge = edge->next; + } + while (eclose) + { + struct rx_possible_future *ec = new_close++; + struct rx_hash_item * sp; + struct rx_se_list ** sepos; + struct rx_se_list * sesrc; + struct rx_nfa_state_set * destlst; + struct rx_nfa_state_set ** destpos; + ec->next = cn->futures; + cn->futures = ec; + for (sepos = &ec->effects, sesrc = eclose->effects; + sesrc; + sesrc = sesrc->cdr, sepos = &(*sepos)->cdr) + { + sp = rx_hash_find (&rx->se_list_memo, + (long)sesrc->car ^ (long)sesrc->cdr, + sesrc, &se_list_hash_rules); + if (sp->binding) + { + sesrc = (struct rx_se_list *)sp->binding; + break; + } + *new_se_list = *sesrc; + sp->binding = (void *)new_se_list; + *sepos = new_se_list; + ++new_se_list; + } + *sepos = sesrc; + for (destpos = &ec->destset, destlst = eclose->destset; + destlst; + destpos = &(*destpos)->cdr, destlst = destlst->cdr) + { + sp = rx_hash_find (&rx->set_list_memo, + ((((long)destlst->car) >> 8) + ^ (long)destlst->cdr), + destlst, &nfa_set_hash_rules); + if (sp->binding) + { + destlst = (struct rx_nfa_state_set *)sp->binding; + break; + } + *new_nfa_set = *destlst; + new_nfa_set->car = state_base + destlst->car->id; + sp->binding = (void *)new_nfa_set; + *destpos = new_nfa_set; + ++new_nfa_set; + } + *destpos = destlst; + eclose = eclose->next; + } + } + } + rx_free_hash_table (&rx->se_list_memo, se_memo_freer, &se_list_hash_rules); + bzero (&rx->se_list_memo, sizeof (rx->se_list_memo)); + rx_free_hash_table (&rx->set_list_memo, nfa_set_freer, &nfa_set_hash_rules); + bzero (&rx->set_list_memo, sizeof (rx->set_list_memo)); + + rx_free_nfa (rx); + rx->nfa_states = (struct rx_nfa_state *)*mem; + return 1; +} + + +/* The functions in the next several pages define the lazy-NFA-conversion used + * by matchers. The input to this construction is an NFA such as + * is built by compactify_nfa (rx.c). The output is the superNFA. + */ + +/* Match engines can use arbitrary values for opcodes. So, the parse tree + * is built using instructions names (enum rx_opcode), but the superstate + * nfa is populated with mystery opcodes (void *). + * + * For convenience, here is an id table. The opcodes are == to their inxs + * + * The lables in re_search_2 would make good values for instructions. + */ + +void * rx_id_instruction_table[rx_num_instructions] = +{ + (void *) rx_backtrack_point, + (void *) rx_do_side_effects, + (void *) rx_cache_miss, + (void *) rx_next_char, + (void *) rx_backtrack, + (void *) rx_error_inx +}; + + + +/* Memory mgt. for superstate graphs. */ + +#ifdef __STDC__ +static char * +rx_cache_malloc (struct rx_cache * cache, int bytes) +#else +static char * +rx_cache_malloc (cache, bytes) + struct rx_cache * cache; + int bytes; +#endif +{ + while (cache->bytes_left < bytes) + { + if (cache->memory_pos) + cache->memory_pos = cache->memory_pos->next; + if (!cache->memory_pos) + { + cache->morecore (cache); + if (!cache->memory_pos) + return 0; + } + cache->bytes_left = cache->memory_pos->bytes; + cache->memory_addr = ((char *)cache->memory_pos + + sizeof (struct rx_blocklist)); + } + cache->bytes_left -= bytes; + { + char * addr = cache->memory_addr; + cache->memory_addr += bytes; + return addr; + } +} + +#ifdef __STDC__ +static void +rx_cache_free (struct rx_cache * cache, + struct rx_freelist ** freelist, char * mem) +#else +static void +rx_cache_free (cache, freelist, mem) + struct rx_cache * cache; + struct rx_freelist ** freelist; + char * mem; +#endif +{ + struct rx_freelist * it = (struct rx_freelist *)mem; + it->next = *freelist; + *freelist = it; +} + +/* The partially instantiated superstate graph has a transition + * table at every node. There is one entry for every character. + * This fills in the transition for a set. + */ +#ifdef __STDC__ +static void +install_transition (struct rx_superstate *super, + struct rx_inx *answer, rx_Bitset trcset) +#else +static void +install_transition (super, answer, trcset) + struct rx_superstate *super; + struct rx_inx *answer; + rx_Bitset trcset; +#endif +{ + struct rx_inx * transitions = super->transitions; + int chr; + for (chr = 0; chr < 256; ) + if (!*trcset) + { + ++trcset; + chr += 32; + } + else + { + RX_subset sub = *trcset; + RX_subset mask = 1; + int bound = chr + 32; + while (chr < bound) + { + if (sub & mask) + transitions [chr] = *answer; + ++chr; + mask <<= 1; + } + ++trcset; + } +} + +#ifdef __STDC__ +static int +qlen (struct rx_superstate * q) +#else +static int +qlen (q) + struct rx_superstate * q; +#endif +{ + int count = 1; + struct rx_superstate * it; + if (!q) + return 0; + for (it = q->next_recyclable; it != q; it = it->next_recyclable) + ++count; + return count; +} + +#ifdef __STDC__ +static void +check_cache (struct rx_cache * cache) +#else +static void +check_cache (cache) + struct rx_cache * cache; +#endif +{ + struct rx_cache * you_fucked_up = 0; + int total = cache->superstates; + int semi = cache->semifree_superstates; + if (semi != qlen (cache->semifree_superstate)) + check_cache (you_fucked_up); + if ((total - semi) != qlen (cache->lru_superstate)) + check_cache (you_fucked_up); +} + +/* When a superstate is old and neglected, it can enter a + * semi-free state. A semi-free state is slated to die. + * Incoming transitions to a semi-free state are re-written + * to cause an (interpreted) fault when they are taken. + * The fault handler revives the semi-free state, patches + * incoming transitions back to normal, and continues. + * + * The idea is basicly to free in two stages, aborting + * between the two if the state turns out to be useful again. + * When a free is aborted, the rescued superstate is placed + * in the most-favored slot to maximize the time until it + * is next semi-freed. + */ + +#ifdef __STDC__ +static void +semifree_superstate (struct rx_cache * cache) +#else +static void +semifree_superstate (cache) + struct rx_cache * cache; +#endif +{ + int disqualified = cache->semifree_superstates; + if (disqualified == cache->superstates) + return; + while (cache->lru_superstate->locks) + { + cache->lru_superstate = cache->lru_superstate->next_recyclable; + ++disqualified; + if (disqualified == cache->superstates) + return; + } + { + struct rx_superstate * it = cache->lru_superstate; + it->next_recyclable->prev_recyclable = it->prev_recyclable; + it->prev_recyclable->next_recyclable = it->next_recyclable; + cache->lru_superstate = (it == it->next_recyclable + ? 0 + : it->next_recyclable); + if (!cache->semifree_superstate) + { + cache->semifree_superstate = it; + it->next_recyclable = it; + it->prev_recyclable = it; + } + else + { + it->prev_recyclable = cache->semifree_superstate->prev_recyclable; + it->next_recyclable = cache->semifree_superstate; + it->prev_recyclable->next_recyclable = it; + it->next_recyclable->prev_recyclable = it; + } + { + struct rx_distinct_future *df; + it->is_semifree = 1; + ++cache->semifree_superstates; + df = it->transition_refs; + if (df) + { + df->prev_same_dest->next_same_dest = 0; + for (df = it->transition_refs; df; df = df->next_same_dest) + { + df->future_frame.inx = cache->instruction_table[rx_cache_miss]; + df->future_frame.data = 0; + df->future_frame.data_2 = (void *) df; + /* If there are any NEXT-CHAR instruction frames that + * refer to this state, we convert them to CACHE-MISS frames. + */ + if (!df->effects + && (df->edge->options->next_same_super_edge[0] + == df->edge->options)) + install_transition (df->present, &df->future_frame, + df->edge->cset); + } + df = it->transition_refs; + df->prev_same_dest->next_same_dest = df; + } + } + } +} + +#ifdef __STDC__ +static void +refresh_semifree_superstate (struct rx_cache * cache, + struct rx_superstate * super) +#else +static void +refresh_semifree_superstate (cache, super) + struct rx_cache * cache; + struct rx_superstate * super; +#endif +{ + struct rx_distinct_future *df; + + if (super->transition_refs) + { + super->transition_refs->prev_same_dest->next_same_dest = 0; + for (df = super->transition_refs; df; df = df->next_same_dest) + { + df->future_frame.inx = cache->instruction_table[rx_next_char]; + df->future_frame.data = (void *) super->transitions; + /* CACHE-MISS instruction frames that refer to this state, + * must be converted to NEXT-CHAR frames. + */ + if (!df->effects + && (df->edge->options->next_same_super_edge[0] + == df->edge->options)) + install_transition (df->present, &df->future_frame, + df->edge->cset); + } + super->transition_refs->prev_same_dest->next_same_dest + = super->transition_refs; + } + if (cache->semifree_superstate == super) + cache->semifree_superstate = (super->prev_recyclable == super + ? 0 + : super->prev_recyclable); + super->next_recyclable->prev_recyclable = super->prev_recyclable; + super->prev_recyclable->next_recyclable = super->next_recyclable; + + if (!cache->lru_superstate) + (cache->lru_superstate + = super->next_recyclable + = super->prev_recyclable + = super); + else + { + super->next_recyclable = cache->lru_superstate; + super->prev_recyclable = cache->lru_superstate->prev_recyclable; + super->next_recyclable->prev_recyclable = super; + super->prev_recyclable->next_recyclable = super; + } + super->is_semifree = 0; + --cache->semifree_superstates; +} + +#ifdef __STDC__ +static void +rx_refresh_this_superstate (struct rx_cache * cache, struct rx_superstate * superstate) +#else +static void +rx_refresh_this_superstate (cache, superstate) + struct rx_cache * cache; + struct rx_superstate * superstate; +#endif +{ + if (superstate->is_semifree) + refresh_semifree_superstate (cache, superstate); + else if (cache->lru_superstate == superstate) + cache->lru_superstate = superstate->next_recyclable; + else if (superstate != cache->lru_superstate->prev_recyclable) + { + superstate->next_recyclable->prev_recyclable + = superstate->prev_recyclable; + superstate->prev_recyclable->next_recyclable + = superstate->next_recyclable; + superstate->next_recyclable = cache->lru_superstate; + superstate->prev_recyclable = cache->lru_superstate->prev_recyclable; + superstate->next_recyclable->prev_recyclable = superstate; + superstate->prev_recyclable->next_recyclable = superstate; + } +} + +#ifdef __STDC__ +static void +release_superset_low (struct rx_cache * cache, + struct rx_superset *set) +#else +static void +release_superset_low (cache, set) + struct rx_cache * cache; + struct rx_superset *set; +#endif +{ + if (!--set->refs) + { + if (set->cdr) + release_superset_low (cache, set->cdr); + + set->starts_for = 0; + + rx_hash_free + (rx_hash_find + (&cache->superset_table, + (unsigned long)set->car ^ set->id ^ (unsigned long)set->cdr, + (void *)set, + &cache->superset_hash_rules), + &cache->superset_hash_rules); + rx_cache_free (cache, &cache->free_supersets, (char *)set); + } +} + +#ifdef __STDC__ +RX_DECL void +rx_release_superset (struct rx *rx, + struct rx_superset *set) +#else +RX_DECL void +rx_release_superset (rx, set) + struct rx *rx; + struct rx_superset *set; +#endif +{ + release_superset_low (rx->cache, set); +} + +/* This tries to add a new superstate to the superstate freelist. + * It might, as a result, free some edge pieces or hash tables. + * If nothing can be freed because too many locks are being held, fail. + */ + +#ifdef __STDC__ +static int +rx_really_free_superstate (struct rx_cache * cache) +#else +static int +rx_really_free_superstate (cache) + struct rx_cache * cache; +#endif +{ + int locked_superstates = 0; + struct rx_superstate * it; + + if (!cache->superstates) + return 0; + + { + /* This is a total guess. The idea is that we should expect as + * many misses as we've recently experienced. I.e., cache->misses + * should be the same as cache->semifree_superstates. + */ + while ((cache->hits + cache->misses) > cache->superstates_allowed) + { + cache->hits >>= 1; + cache->misses >>= 1; + } + if ( ((cache->hits + cache->misses) * cache->semifree_superstates) + < (cache->superstates * cache->misses)) + { + semifree_superstate (cache); + semifree_superstate (cache); + } + } + + while (cache->semifree_superstate && cache->semifree_superstate->locks) + { + refresh_semifree_superstate (cache, cache->semifree_superstate); + ++locked_superstates; + if (locked_superstates == cache->superstates) + return 0; + } + + if (cache->semifree_superstate) + { + it = cache->semifree_superstate; + it->next_recyclable->prev_recyclable = it->prev_recyclable; + it->prev_recyclable->next_recyclable = it->next_recyclable; + cache->semifree_superstate = ((it == it->next_recyclable) + ? 0 + : it->next_recyclable); + --cache->semifree_superstates; + } + else + { + while (cache->lru_superstate->locks) + { + cache->lru_superstate = cache->lru_superstate->next_recyclable; + ++locked_superstates; + if (locked_superstates == cache->superstates) + return 0; + } + it = cache->lru_superstate; + it->next_recyclable->prev_recyclable = it->prev_recyclable; + it->prev_recyclable->next_recyclable = it->next_recyclable; + cache->lru_superstate = ((it == it->next_recyclable) + ? 0 + : it->next_recyclable); + } + + if (it->transition_refs) + { + struct rx_distinct_future *df; + for (df = it->transition_refs, + df->prev_same_dest->next_same_dest = 0; + df; + df = df->next_same_dest) + { + df->future_frame.inx = cache->instruction_table[rx_cache_miss]; + df->future_frame.data = 0; + df->future_frame.data_2 = (void *) df; + df->future = 0; + } + it->transition_refs->prev_same_dest->next_same_dest = + it->transition_refs; + } + { + struct rx_super_edge *tc = it->edges; + while (tc) + { + struct rx_distinct_future * df; + struct rx_super_edge *tct = tc->next; + df = tc->options; + df->next_same_super_edge[1]->next_same_super_edge[0] = 0; + while (df) + { + struct rx_distinct_future *dft = df; + df = df->next_same_super_edge[0]; + + + if (dft->future && dft->future->transition_refs == dft) + { + dft->future->transition_refs = dft->next_same_dest; + if (dft->future->transition_refs == dft) + dft->future->transition_refs = 0; + } + dft->next_same_dest->prev_same_dest = dft->prev_same_dest; + dft->prev_same_dest->next_same_dest = dft->next_same_dest; + rx_cache_free (cache, &cache->free_discernable_futures, + (char *)dft); + } + rx_cache_free (cache, &cache->free_transition_classes, (char *)tc); + tc = tct; + } + } + + if (it->contents->superstate == it) + it->contents->superstate = 0; + release_superset_low (cache, it->contents); + rx_cache_free (cache, &cache->free_superstates, (char *)it); + --cache->superstates; + return 1; +} + +#ifdef __STDC__ +static char * +rx_cache_get (struct rx_cache * cache, + struct rx_freelist ** freelist) +#else +static char * +rx_cache_get (cache, freelist) + struct rx_cache * cache; + struct rx_freelist ** freelist; +#endif +{ + while (!*freelist && rx_really_free_superstate (cache)) + ; + if (!*freelist) + return 0; + { + struct rx_freelist * it = *freelist; + *freelist = it->next; + return (char *)it; + } +} + +#ifdef __STDC__ +static char * +rx_cache_malloc_or_get (struct rx_cache * cache, + struct rx_freelist ** freelist, int bytes) +#else +static char * +rx_cache_malloc_or_get (cache, freelist, bytes) + struct rx_cache * cache; + struct rx_freelist ** freelist; + int bytes; +#endif +{ + if (!*freelist) + { + char * answer = rx_cache_malloc (cache, bytes); + if (answer) + return answer; + } + + return rx_cache_get (cache, freelist); +} + +#ifdef __STDC__ +static char * +rx_cache_get_superstate (struct rx_cache * cache) +#else +static char * +rx_cache_get_superstate (cache) + struct rx_cache * cache; +#endif +{ + char * answer; + int bytes = ( sizeof (struct rx_superstate) + + cache->local_cset_size * sizeof (struct rx_inx)); + if (!cache->free_superstates + && (cache->superstates < cache->superstates_allowed)) + { + answer = rx_cache_malloc (cache, bytes); + if (answer) + { + ++cache->superstates; + return answer; + } + } + answer = rx_cache_get (cache, &cache->free_superstates); + if (!answer) + { + answer = rx_cache_malloc (cache, bytes); + if (answer) + ++cache->superstates_allowed; + } + ++cache->superstates; + return answer; +} + + + +#ifdef __STDC__ +static int +supersetcmp (void * va, void * vb) +#else +static int +supersetcmp (va, vb) + void * va; + void * vb; +#endif +{ + struct rx_superset * a = (struct rx_superset *)va; + struct rx_superset * b = (struct rx_superset *)vb; + return ( (a == b) + || (a && b && (a->car == b->car) && (a->cdr == b->cdr))); +} + +#ifdef __STDC__ +static struct rx_hash_item * +superset_allocator (struct rx_hash_rules * rules, void * val) +#else +static struct rx_hash_item * +superset_allocator (rules, val) + struct rx_hash_rules * rules; + void * val; +#endif +{ + struct rx_cache * cache + = ((struct rx_cache *) + ((char *)rules + - (unsigned long)(&((struct rx_cache *)0)->superset_hash_rules))); + struct rx_superset * template = (struct rx_superset *)val; + struct rx_superset * newset + = ((struct rx_superset *) + rx_cache_malloc_or_get (cache, + &cache->free_supersets, + sizeof (*template))); + if (!newset) + return 0; + newset->refs = 0; + newset->car = template->car; + newset->id = template->car->id; + newset->cdr = template->cdr; + newset->superstate = 0; + rx_protect_superset (rx, template->cdr); + newset->hash_item.data = (void *)newset; + newset->hash_item.binding = 0; + return &newset->hash_item; +} + +#ifdef __STDC__ +static struct rx_hash * +super_hash_allocator (struct rx_hash_rules * rules) +#else +static struct rx_hash * +super_hash_allocator (rules) + struct rx_hash_rules * rules; +#endif +{ + struct rx_cache * cache + = ((struct rx_cache *) + ((char *)rules + - (unsigned long)(&((struct rx_cache *)0)->superset_hash_rules))); + return ((struct rx_hash *) + rx_cache_malloc_or_get (cache, + &cache->free_hash, sizeof (struct rx_hash))); +} + + +#ifdef __STDC__ +static void +super_hash_liberator (struct rx_hash * hash, struct rx_hash_rules * rules) +#else +static void +super_hash_liberator (hash, rules) + struct rx_hash * hash; + struct rx_hash_rules * rules; +#endif +{ + struct rx_cache * cache + = ((struct rx_cache *) + (char *)rules - (long)(&((struct rx_cache *)0)->superset_hash_rules)); + rx_cache_free (cache, &cache->free_hash, (char *)hash); +} + +#ifdef __STDC__ +static void +superset_hash_item_liberator (struct rx_hash_item * it, + struct rx_hash_rules * rules) +#else +static void +superset_hash_item_liberator (it, rules) /* Well, it does ya know. */ + struct rx_hash_item * it; + struct rx_hash_rules * rules; +#endif +{ +} + +int rx_cache_bound = 128; +static int rx_default_cache_got = 0; + +#ifdef __STDC__ +static int +bytes_for_cache_size (int supers, int cset_size) +#else +static int +bytes_for_cache_size (supers, cset_size) + int supers; + int cset_size; +#endif +{ + /* What the hell is this? !!!*/ + return (int) + ((float)supers * + ( (1.03 * (float) ( rx_sizeof_bitset (cset_size) + + sizeof (struct rx_super_edge))) + + (1.80 * (float) sizeof (struct rx_possible_future)) + + (float) ( sizeof (struct rx_superstate) + + cset_size * sizeof (struct rx_inx)))); +} + +#ifdef __STDC__ +static void +rx_morecore (struct rx_cache * cache) +#else +static void +rx_morecore (cache) + struct rx_cache * cache; +#endif +{ + if (rx_default_cache_got >= rx_cache_bound) + return; + + rx_default_cache_got += 16; + cache->superstates_allowed = rx_cache_bound; + { + struct rx_blocklist ** pos = &cache->memory; + int size = bytes_for_cache_size (16, cache->local_cset_size); + while (*pos) + pos = &(*pos)->next; + *pos = ((struct rx_blocklist *) + malloc (size + sizeof (struct rx_blocklist))); + if (!*pos) + return; + + (*pos)->next = 0; + (*pos)->bytes = size; + cache->memory_pos = *pos; + cache->memory_addr = (char *)*pos + sizeof (**pos); + cache->bytes_left = size; + } +} + +static struct rx_cache default_cache = +{ + { + supersetcmp, + super_hash_allocator, + super_hash_liberator, + superset_allocator, + superset_hash_item_liberator, + }, + 0, + 0, + 0, + 0, + rx_morecore, + + 0, + 0, + 0, + 0, + 0, + + 0, + 0, + + 0, + + 0, + 0, + 0, + 0, + 128, + + 256, + rx_id_instruction_table, + + { + 0, + 0, + {0}, + {0}, + {0} + } +}; + +/* This adds an element to a superstate set. These sets are lists, such + * that lists with == elements are ==. The empty set is returned by + * superset_cons (rx, 0, 0) and is NOT equivelent to + * (struct rx_superset)0. + */ + +#ifdef __STDC__ +RX_DECL struct rx_superset * +rx_superset_cons (struct rx * rx, + struct rx_nfa_state *car, struct rx_superset *cdr) +#else +RX_DECL struct rx_superset * +rx_superset_cons (rx, car, cdr) + struct rx * rx; + struct rx_nfa_state *car; + struct rx_superset *cdr; +#endif +{ + struct rx_cache * cache = rx->cache; + if (!car && !cdr) + { + if (!cache->empty_superset) + { + cache->empty_superset + = ((struct rx_superset *) + rx_cache_malloc_or_get (cache, &cache->free_supersets, + sizeof (struct rx_superset))); + if (!cache->empty_superset) + return 0; + bzero (cache->empty_superset, sizeof (struct rx_superset)); + cache->empty_superset->refs = 1000; + } + return cache->empty_superset; + } + { + struct rx_superset template; + struct rx_hash_item * hit; + template.car = car; + template.cdr = cdr; + template.id = car->id; + /* While hash_store will protect cdr itself it might first allocate hash + tables and stuff which might cause it to be garbage collected before + it's protected -- [gsstark:19961026.2155EST] */ + rx_protect_superset (rx, cdr); + hit = rx_hash_store (&cache->superset_table, + (unsigned long)car ^ car->id ^ (unsigned long)cdr, + (void *)&template, + &cache->superset_hash_rules); + rx_release_superset (rx, cdr); + return (hit + ? (struct rx_superset *)hit->data + : 0); + } +} + +/* This computes a union of two NFA state sets. The sets do not have the + * same representation though. One is a RX_SUPERSET structure (part + * of the superstate NFA) and the other is an NFA_STATE_SET (part of the NFA). + */ + +#ifdef __STDC__ +RX_DECL struct rx_superset * +rx_superstate_eclosure_union + (struct rx * rx, struct rx_superset *set, struct rx_nfa_state_set *ecl) +#else +RX_DECL struct rx_superset * +rx_superstate_eclosure_union (rx, set, ecl) + struct rx * rx; + struct rx_superset *set; + struct rx_nfa_state_set *ecl; +#endif +{ + if (!ecl) + return set; + + if (!set->car) + return rx_superset_cons (rx, ecl->car, + rx_superstate_eclosure_union (rx, set, ecl->cdr)); + if (set->car == ecl->car) + return rx_superstate_eclosure_union (rx, set, ecl->cdr); + + { + struct rx_superset * tail; + struct rx_nfa_state * first; + + if (set->car > ecl->car) + { + tail = rx_superstate_eclosure_union (rx, set->cdr, ecl); + first = set->car; + } + else + { + tail = rx_superstate_eclosure_union (rx, set, ecl->cdr); + first = ecl->car; + } + if (!tail) + return 0; + else + { + struct rx_superset * answer; + answer = rx_superset_cons (rx, first, tail); + if (!answer) + { + rx_protect_superset (rx, tail); + rx_release_superset (rx, tail); + return 0; + } + else + return answer; + } + } +} + + + + +/* + * This makes sure that a list of rx_distinct_futures contains + * a future for each possible set of side effects in the eclosure + * of a given state. This is some of the work of filling in a + * superstate transition. + */ + +#ifdef __STDC__ +static struct rx_distinct_future * +include_futures (struct rx *rx, + struct rx_distinct_future *df, struct rx_nfa_state + *state, struct rx_superstate *superstate) +#else +static struct rx_distinct_future * +include_futures (rx, df, state, superstate) + struct rx *rx; + struct rx_distinct_future *df; + struct rx_nfa_state *state; + struct rx_superstate *superstate; +#endif +{ + struct rx_possible_future *future; + struct rx_cache * cache = rx->cache; + for (future = state->futures; future; future = future->next) + { + struct rx_distinct_future *dfp; + struct rx_distinct_future *insert_before = 0; + if (df) + df->next_same_super_edge[1]->next_same_super_edge[0] = 0; + for (dfp = df; dfp; dfp = dfp->next_same_super_edge[0]) + if (dfp->effects == future->effects) + break; + else + { + int order = rx->se_list_cmp (rx, dfp->effects, future->effects); + if (order > 0) + { + insert_before = dfp; + dfp = 0; + break; + } + } + if (df) + df->next_same_super_edge[1]->next_same_super_edge[0] = df; + if (!dfp) + { + dfp + = ((struct rx_distinct_future *) + rx_cache_malloc_or_get (cache, &cache->free_discernable_futures, + sizeof (struct rx_distinct_future))); + if (!dfp) + return 0; + if (!df) + { + df = insert_before = dfp; + df->next_same_super_edge[0] = df->next_same_super_edge[1] = df; + } + else if (!insert_before) + insert_before = df; + else if (insert_before == df) + df = dfp; + + dfp->next_same_super_edge[0] = insert_before; + dfp->next_same_super_edge[1] + = insert_before->next_same_super_edge[1]; + dfp->next_same_super_edge[1]->next_same_super_edge[0] = dfp; + dfp->next_same_super_edge[0]->next_same_super_edge[1] = dfp; + dfp->next_same_dest = dfp->prev_same_dest = dfp; + dfp->future = 0; + dfp->present = superstate; + dfp->future_frame.inx = rx->instruction_table[rx_cache_miss]; + dfp->future_frame.data = 0; + dfp->future_frame.data_2 = (void *) dfp; + dfp->side_effects_frame.inx + = rx->instruction_table[rx_do_side_effects]; + dfp->side_effects_frame.data = 0; + dfp->side_effects_frame.data_2 = (void *) dfp; + dfp->effects = future->effects; + } + } + return df; +} + + +/* This constructs a new superstate from its state set. The only + * complexity here is memory management. + */ +#ifdef __STDC__ +RX_DECL struct rx_superstate * +rx_superstate (struct rx *rx, + struct rx_superset *set) +#else +RX_DECL struct rx_superstate * +rx_superstate (rx, set) + struct rx *rx; + struct rx_superset *set; +#endif +{ + struct rx_cache * cache = rx->cache; + struct rx_superstate * superstate = 0; + + /* Does the superstate already exist in the cache? */ + if (set->superstate) + { + if (set->superstate->rx_id != rx->rx_id) + { + /* Aha. It is in the cache, but belongs to a superstate + * that refers to an NFA that no longer exists. + * (We know it no longer exists because it was evidently + * stored in the same region of memory as the current nfa + * yet it has a different id.) + */ + superstate = set->superstate; + if (!superstate->is_semifree) + { + if (cache->lru_superstate == superstate) + { + cache->lru_superstate = superstate->next_recyclable; + if (cache->lru_superstate == superstate) + cache->lru_superstate = 0; + } + { + superstate->next_recyclable->prev_recyclable + = superstate->prev_recyclable; + superstate->prev_recyclable->next_recyclable + = superstate->next_recyclable; + if (!cache->semifree_superstate) + { + (cache->semifree_superstate + = superstate->next_recyclable + = superstate->prev_recyclable + = superstate); + } + else + { + superstate->next_recyclable = cache->semifree_superstate; + superstate->prev_recyclable + = cache->semifree_superstate->prev_recyclable; + superstate->next_recyclable->prev_recyclable + = superstate; + superstate->prev_recyclable->next_recyclable + = superstate; + cache->semifree_superstate = superstate; + } + ++cache->semifree_superstates; + } + } + set->superstate = 0; + goto handle_cache_miss; + } + ++cache->hits; + superstate = set->superstate; + + rx_refresh_this_superstate (cache, superstate); + return superstate; + } + + handle_cache_miss: + + /* This point reached only for cache misses. */ + ++cache->misses; +#if RX_DEBUG + if (rx_debug_trace > 1) + { + struct rx_superset * setp = set; + fprintf (stderr, "Building a superstet %d(%d): ", rx->rx_id, set); + while (setp) + { + fprintf (stderr, "%d ", setp->id); + setp = setp->cdr; + } + fprintf (stderr, "(%d)\n", set); + } +#endif + superstate = (struct rx_superstate *)rx_cache_get_superstate (cache); + if (!superstate) + return 0; + + if (!cache->lru_superstate) + (cache->lru_superstate + = superstate->next_recyclable + = superstate->prev_recyclable + = superstate); + else + { + superstate->next_recyclable = cache->lru_superstate; + superstate->prev_recyclable = cache->lru_superstate->prev_recyclable; + ( superstate->prev_recyclable->next_recyclable + = superstate->next_recyclable->prev_recyclable + = superstate); + } + superstate->rx_id = rx->rx_id; + superstate->transition_refs = 0; + superstate->locks = 0; + superstate->is_semifree = 0; + set->superstate = superstate; + superstate->contents = set; + rx_protect_superset (rx, set); + superstate->edges = 0; + { + int x; + /* None of the transitions from this superstate are known yet. */ + for (x = 0; x < rx->local_cset_size; ++x) /* &&&&& 3.8 % */ + { + struct rx_inx * ifr = &superstate->transitions[x]; + ifr->inx = rx->instruction_table [rx_cache_miss]; + ifr->data = ifr->data_2 = 0; + } + } + return superstate; +} + + +/* This computes the destination set of one edge of the superstate NFA. + * Note that a RX_DISTINCT_FUTURE is a superstate edge. + * Returns 0 on an allocation failure. + */ + +#ifdef __STDC__ +static int +solve_destination (struct rx *rx, struct rx_distinct_future *df) +#else +static int +solve_destination (rx, df) + struct rx *rx; + struct rx_distinct_future *df; +#endif +{ + struct rx_super_edge *tc = df->edge; + struct rx_superset *nfa_state; + struct rx_superset *nil_set = rx_superset_cons (rx, 0, 0); + struct rx_superset *solution = nil_set; + struct rx_superstate *dest; + + rx_protect_superset (rx, solution); + /* Iterate over all NFA states in the state set of this superstate. */ + for (nfa_state = df->present->contents; + nfa_state->car; + nfa_state = nfa_state->cdr) + { + struct rx_nfa_edge *e; + /* Iterate over all edges of each NFA state. */ + for (e = nfa_state->car->edges; e; e = e->next) + /* If we find an edge that is labeled with + * the characters we are solving for..... + */ + if (rx_bitset_is_subset (rx->local_cset_size, + tc->cset, e->params.cset)) + { + struct rx_nfa_state *n = e->dest; + struct rx_possible_future *pf; + /* ....search the partial epsilon closures of the destination + * of that edge for a path that involves the same set of + * side effects we are solving for. + * If we find such a RX_POSSIBLE_FUTURE, we add members to the + * stateset we are computing. + */ + for (pf = n->futures; pf; pf = pf->next) + if (pf->effects == df->effects) + { + struct rx_superset * old_sol; + old_sol = solution; + solution = rx_superstate_eclosure_union (rx, solution, + pf->destset); + if (!solution) + return 0; + rx_protect_superset (rx, solution); + rx_release_superset (rx, old_sol); + } + } + } + /* It is possible that the RX_DISTINCT_FUTURE we are working on has + * the empty set of NFA states as its definition. In that case, this + * is a failure point. + */ + if (solution == nil_set) + { + df->future_frame.inx = (void *) rx_backtrack; + df->future_frame.data = 0; + df->future_frame.data_2 = 0; + return 1; + } + dest = rx_superstate (rx, solution); + rx_release_superset (rx, solution); + if (!dest) + return 0; + + { + struct rx_distinct_future *dft; + dft = df; + df->prev_same_dest->next_same_dest = 0; + while (dft) + { + dft->future = dest; + dft->future_frame.inx = rx->instruction_table[rx_next_char]; + dft->future_frame.data = (void *) dest->transitions; + dft = dft->next_same_dest; + } + df->prev_same_dest->next_same_dest = df; + } + if (!dest->transition_refs) + dest->transition_refs = df; + else + { + struct rx_distinct_future *dft = dest->transition_refs->next_same_dest; + dest->transition_refs->next_same_dest = df->next_same_dest; + df->next_same_dest->prev_same_dest = dest->transition_refs; + df->next_same_dest = dft; + dft->prev_same_dest = df; + } + return 1; +} + + +/* This takes a superstate and a character, and computes some edges + * from the superstate NFA. In particular, this computes all edges + * that lead from SUPERSTATE given CHR. This function also + * computes the set of characters that share this edge set. + * This returns 0 on allocation error. + * The character set and list of edges are returned through + * the paramters CSETOUT and DFOUT. +} */ + +#ifdef __STDC__ +static int +compute_super_edge (struct rx *rx, struct rx_distinct_future **dfout, + rx_Bitset csetout, struct rx_superstate *superstate, + unsigned char chr) +#else +static int +compute_super_edge (rx, dfout, csetout, superstate, chr) + struct rx *rx; + struct rx_distinct_future **dfout; + rx_Bitset csetout; + struct rx_superstate *superstate; + unsigned char chr; +#endif +{ + struct rx_superset *stateset = superstate->contents; + + /* To compute the set of characters that share edges with CHR, + * we start with the full character set, and subtract. + */ + rx_bitset_universe (rx->local_cset_size, csetout); + *dfout = 0; + + /* Iterate over the NFA states in the superstate state-set. */ + while (stateset->car) + { + struct rx_nfa_edge *e; + for (e = stateset->car->edges; e; e = e->next) + if (RX_bitset_member (e->params.cset, chr)) + { + /* If we find an NFA edge that applies, we make sure there + * are corresponding edges in the superstate NFA. + */ + { + struct rx_distinct_future * saved; + saved = *dfout; + *dfout = include_futures (rx, *dfout, e->dest, superstate); + if (!*dfout) + { + struct rx_distinct_future * df; + df = saved; + if (df) + df->next_same_super_edge[1]->next_same_super_edge[0] = 0; + while (df) + { + struct rx_distinct_future *dft; + dft = df; + df = df->next_same_super_edge[0]; + + if (dft->future && dft->future->transition_refs == dft) + { + dft->future->transition_refs = dft->next_same_dest; + if (dft->future->transition_refs == dft) + dft->future->transition_refs = 0; + } + dft->next_same_dest->prev_same_dest = dft->prev_same_dest; + dft->prev_same_dest->next_same_dest = dft->next_same_dest; + rx_cache_free (rx->cache, + &rx->cache->free_discernable_futures, + (char *)dft); + } + return 0; + } + } + /* We also trim the character set a bit. */ + rx_bitset_intersection (rx->local_cset_size, + csetout, e->params.cset); + } + else + /* An edge that doesn't apply at least tells us some characters + * that don't share the same edge set as CHR. + */ + rx_bitset_difference (rx->local_cset_size, csetout, e->params.cset); + stateset = stateset->cdr; + } + return 1; +} + + +/* This is a constructor for RX_SUPER_EDGE structures. These are + * wrappers for lists of superstate NFA edges that share character sets labels. + * If a transition class contains more than one rx_distinct_future (superstate + * edge), then it represents a non-determinism in the superstate NFA. + */ + + +#ifdef __STDC__ +static struct rx_super_edge * +rx_super_edge (struct rx *rx, + struct rx_superstate *super, rx_Bitset cset, + struct rx_distinct_future *df) +#else +static struct rx_super_edge * +rx_super_edge (rx, super, cset, df) + struct rx *rx; + struct rx_superstate *super; + rx_Bitset cset; + struct rx_distinct_future *df; +#endif +{ + struct rx_super_edge *tc = + (struct rx_super_edge *)rx_cache_malloc_or_get + (rx->cache, &rx->cache->free_transition_classes, + sizeof (struct rx_super_edge) + rx_sizeof_bitset (rx->local_cset_size)); + + if (!tc) + return 0; + tc->next = super->edges; + super->edges = tc; + tc->rx_backtrack_frame.inx = rx->instruction_table[rx_backtrack_point]; + tc->rx_backtrack_frame.data = 0; + tc->rx_backtrack_frame.data_2 = (void *) tc; + tc->options = df; + tc->cset = (rx_Bitset) ((char *) tc + sizeof (*tc)); + rx_bitset_assign (rx->local_cset_size, tc->cset, cset); + if (df) + { + struct rx_distinct_future * dfp = df; + df->next_same_super_edge[1]->next_same_super_edge[0] = 0; + while (dfp) + { + dfp->edge = tc; + dfp = dfp->next_same_super_edge[0]; + } + df->next_same_super_edge[1]->next_same_super_edge[0] = df; + } + return tc; +} + + +/* There are three kinds of cache miss. The first occurs when a + * transition is taken that has never been computed during the + * lifetime of the source superstate. That cache miss is handled by + * calling COMPUTE_SUPER_EDGE. The second kind of cache miss + * occurs when the destination superstate of a transition doesn't + * exist. SOLVE_DESTINATION is used to construct the destination superstate. + * Finally, the third kind of cache miss occurs when the destination + * superstate of a transition is in a `semi-free state'. That case is + * handled by UNFREE_SUPERSTATE. + * + * The function of HANDLE_CACHE_MISS is to figure out which of these + * cases applies. + */ + + +#ifdef __STDC__ +static void +install_partial_transition (struct rx_superstate *super, + struct rx_inx *answer, + RX_subset set, int offset) +#else +static void +install_partial_transition (super, answer, set, offset) + struct rx_superstate *super; + struct rx_inx *answer; + RX_subset set; + int offset; +#endif +{ + int start = offset; + int end = start + 32; + RX_subset pos = 1; + struct rx_inx * transitions = super->transitions; + + while (start < end) + { + if (set & pos) + transitions[start] = *answer; + pos <<= 1; + ++start; + } +} + +#ifdef __STDC__ +RX_DECL struct rx_inx * +rx_handle_cache_miss + (struct rx *rx, struct rx_superstate *super, unsigned char chr, void *data) +#else +RX_DECL struct rx_inx * +rx_handle_cache_miss (rx, super, chr, data) + struct rx *rx; + struct rx_superstate *super; + unsigned char chr; + void *data; +#endif +{ + int offset = chr / RX_subset_bits; + struct rx_distinct_future *df = data; + + if (!df) /* must be the shared_cache_miss_frame */ + { + /* Perhaps this is just a transition waiting to be filled. */ + struct rx_super_edge *tc; + RX_subset mask = rx_subset_singletons [chr % RX_subset_bits]; + + for (tc = super->edges; tc; tc = tc->next) + if (tc->cset[offset] & mask) + { + struct rx_inx * answer; + df = tc->options; + answer = ((tc->options->next_same_super_edge[0] != tc->options) + ? &tc->rx_backtrack_frame + : (df->effects + ? &df->side_effects_frame + : &df->future_frame)); + install_partial_transition (super, answer, + tc->cset [offset], offset * 32); + return answer; + } + /* Otherwise, it's a flushed or newly encountered edge. */ + { + char cset_space[1024]; /* this limit is far from unreasonable */ + rx_Bitset trcset; + struct rx_inx *answer; + + if (rx_sizeof_bitset (rx->local_cset_size) > sizeof (cset_space)) + return 0; /* If the arbitrary limit is hit, always fail */ + /* cleanly. */ + trcset = (rx_Bitset)cset_space; + rx_lock_superstate (rx, super); + if (!compute_super_edge (rx, &df, trcset, super, chr)) + { + rx_unlock_superstate (rx, super); + return 0; + } + if (!df) /* We just computed the fail transition. */ + { + static struct rx_inx + shared_fail_frame = { 0, 0, (void *)rx_backtrack, 0 }; + answer = &shared_fail_frame; + } + else + { + tc = rx_super_edge (rx, super, trcset, df); + if (!tc) + { + rx_unlock_superstate (rx, super); + return 0; + } + answer = ((tc->options->next_same_super_edge[0] != tc->options) + ? &tc->rx_backtrack_frame + : (df->effects + ? &df->side_effects_frame + : &df->future_frame)); + } + install_partial_transition (super, answer, + trcset[offset], offset * 32); + rx_unlock_superstate (rx, super); + return answer; + } + } + else if (df->future) /* A cache miss on an edge with a future? Must be + * a semi-free destination. */ + { + if (df->future->is_semifree) + refresh_semifree_superstate (rx->cache, df->future); + return &df->future_frame; + } + else + /* no future superstate on an existing edge */ + { + rx_lock_superstate (rx, super); + if (!solve_destination (rx, df)) + { + rx_unlock_superstate (rx, super); + return 0; + } + if (!df->effects + && (df->edge->options->next_same_super_edge[0] == df->edge->options)) + install_partial_transition (super, &df->future_frame, + df->edge->cset[offset], offset * 32); + rx_unlock_superstate (rx, super); + return &df->future_frame; + } +} + + + + +/* The rest of the code provides a regex.c compatable interface. */ + + +__const__ char *re_error_msg[] = +{ + 0, /* REG_NOUT */ + "No match", /* REG_NOMATCH */ + "Invalid regular expression", /* REG_BADPAT */ + "Invalid collation character", /* REG_ECOLLATE */ + "Invalid character class name", /* REG_ECTYPE */ + "Trailing backslash", /* REG_EESCAPE */ + "Invalid back reference", /* REG_ESUBREG */ + "Unmatched [ or [^", /* REG_EBRACK */ + "Unmatched ( or \\(", /* REG_EPAREN */ + "Unmatched \\{", /* REG_EBRACE */ + "Invalid content of \\{\\}", /* REG_BADBR */ + "Invalid range end", /* REG_ERANGE */ + "Memory exhausted", /* REG_ESPACE */ + "Invalid preceding regular expression", /* REG_BADRPT */ + "Premature end of regular expression", /* REG_EEND */ + "Regular expression too big", /* REG_ESIZE */ + "Unmatched ) or \\)", /* REG_ERPAREN */ +}; + + + +/* + * Macros used while compiling patterns. + * + * By convention, PEND points just past the end of the uncompiled pattern, + * P points to the read position in the pattern. `translate' is the name + * of the translation table (`TRANSLATE' is the name of a macro that looks + * things up in `translate'). + */ + + +/* + * Fetch the next character in the uncompiled pattern---translating it + * if necessary. *Also cast from a signed character in the constant + * string passed to us by the user to an unsigned char that we can use + * as an array index (in, e.g., `translate'). + */ +#define PATFETCH(c) \ + do {if (p == pend) return REG_EEND; \ + c = (unsigned char) *p++; \ + c = translate[c]; \ + } while (0) + +/* + * Fetch the next character in the uncompiled pattern, with no + * translation. + */ +#define PATFETCH_RAW(c) \ + do {if (p == pend) return REG_EEND; \ + c = (unsigned char) *p++; \ + } while (0) + +/* Go backwards one character in the pattern. */ +#define PATUNFETCH p-- + + +#define TRANSLATE(d) translate[(unsigned char) (d)] + +typedef unsigned regnum_t; + +/* Since offsets can go either forwards or backwards, this type needs to + * be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. + */ +typedef int pattern_offset_t; + +typedef struct +{ + struct rexp_node ** top_expression; /* was begalt */ + struct rexp_node ** last_expression; /* was laststart */ + pattern_offset_t inner_group_offset; + regnum_t regnum; +} compile_stack_elt_t; +typedef struct +{ + compile_stack_elt_t *stack; + unsigned size; + unsigned avail; /* Offset of next open position. */ +} compile_stack_type; + +static boolean + group_in_compile_stack (compile_stack_type, regnum_t); +static reg_errcode_t + compile_range (struct re_pattern_buffer *, rx_Bitset, + __const__ char **, __const__ char *, + unsigned char *, reg_syntax_t, + rx_Bitset, char *); +static void find_backrefs (char *, struct rexp_node *, + struct re_se_params *); +static int compute_fastset (struct re_pattern_buffer *, + struct rexp_node *); +static int is_anchored (struct rexp_node *, rx_side_effect); +static struct rexp_node + *remove_unecessary_side_effects + (struct rx *, char *, + struct rexp_node *, + struct re_se_params *); +static int pointless_if_repeated (struct rexp_node *, + struct re_se_params *); +static int registers_on_stack (struct re_pattern_buffer *, + struct rexp_node *, + int, struct re_se_params *); +static int has_any_se (struct rx *, struct rexp_node *); +static int has_non_idempotent_epsilon_path + (struct rx *, struct rexp_node *, + struct re_se_params *); +static int begins_with_complex_se (struct rx *, struct rexp_node *); +static void speed_up_alt (struct rx *, struct rexp_node *, int); +RX_DECL reg_errcode_t + rx_compile (__const__ char *, int, reg_syntax_t, + struct re_pattern_buffer *); +RX_DECL void rx_blow_up_fastmap (struct re_pattern_buffer *); +static __inline__ enum rx_get_burst_return + re_search_2_get_burst (struct rx_string_position *, + void *, int); +static __inline__ enum rx_back_check_return + re_search_2_back_check (struct rx_string_position *, int, + int, unsigned char *, void *, int); +static __inline__ int + re_search_2_fetch_char (struct rx_string_position *, + int, void *, int); + + +#define INIT_COMPILE_STACK_SIZE 32 + +#define COMPILE_STACK_EMPTY (compile_stack.avail == 0) +#define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size) + +/* The next available element. */ +#define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail]) + + +/* Set the bit for character C in a list. */ +#define SET_LIST_BIT(c) \ + (b[((unsigned char) (c)) / CHARBITS] \ + |= 1 << (((unsigned char) c) % CHARBITS)) + +/* Get the next unsigned number in the uncompiled pattern. */ +#define GET_UNSIGNED_NUMBER(num) \ + { if (p != pend) \ + { \ + PATFETCH (c); \ + while (isdigit (c)) \ + { \ + if (num < 0) \ + num = 0; \ + num = num * 10 + c - '0'; \ + if (p == pend) \ + break; \ + PATFETCH (c); \ + } \ + } \ + } + +#define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */ + +#define IS_CHAR_CLASS(string) \ + (!strcmp (string, "alpha") || !strcmp (string, "upper") \ + || !strcmp (string, "lower") || !strcmp (string, "digit") \ + || !strcmp (string, "alnum") || !strcmp (string, "xdigit") \ + || !strcmp (string, "space") || !strcmp (string, "print") \ + || !strcmp (string, "punct") || !strcmp (string, "graph") \ + || !strcmp (string, "cntrl") || !strcmp (string, "blank")) + + +/* These predicates are used in regex_compile. */ + +/* P points to just after a ^ in PATTERN. Return true if that ^ comes + * after an alternative or a begin-subexpression. We assume there is at + * least one character before the ^. + */ + +#ifdef __STDC__ +static boolean +at_begline_loc_p (__const__ char *pattern, __const__ char * p, reg_syntax_t syntax) +#else +static boolean +at_begline_loc_p (pattern, p, syntax) + __const__ char *pattern; + __const__ char * p; + reg_syntax_t syntax; +#endif +{ + __const__ char *prev = p - 2; + boolean prev_prev_backslash = ((prev > pattern) && (prev[-1] == '\\')); + + return + + (/* After a subexpression? */ + ((*prev == '(') && ((syntax & RE_NO_BK_PARENS) || prev_prev_backslash)) + || + /* After an alternative? */ + ((*prev == '|') && ((syntax & RE_NO_BK_VBAR) || prev_prev_backslash)) + ); +} + +/* The dual of at_begline_loc_p. This one is for $. We assume there is + * at least one character after the $, i.e., `P < PEND'. + */ + +#ifdef __STDC__ +static boolean +at_endline_loc_p (__const__ char *p, __const__ char *pend, int syntax) +#else +static boolean +at_endline_loc_p (p, pend, syntax) + __const__ char *p; + __const__ char *pend; + int syntax; +#endif +{ + __const__ char *next = p; + boolean next_backslash = (*next == '\\'); + __const__ char *next_next = (p + 1 < pend) ? (p + 1) : 0; + + return + ( + /* Before a subexpression? */ + ((syntax & RE_NO_BK_PARENS) + ? (*next == ')') + : (next_backslash && next_next && (*next_next == ')'))) + || + /* Before an alternative? */ + ((syntax & RE_NO_BK_VBAR) + ? (*next == '|') + : (next_backslash && next_next && (*next_next == '|'))) + ); +} + + +unsigned char rx_id_translation[256] = +{ + 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, + 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, + 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, + 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, + 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, + 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, + 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, + 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, + 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, + 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, + + 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, + 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, + 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, + 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, + 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, + 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, + 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, + 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, + 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, + 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, + + 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, + 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, + 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, + 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, + 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, + 250, 251, 252, 253, 254, 255 +}; + +/* The compiler keeps an inverted translation table. + * This looks up/inititalize elements. + * VALID is an array of booleans that validate CACHE. + */ + +#ifdef __STDC__ +static rx_Bitset +inverse_translation (struct re_pattern_buffer * rxb, + char * valid, rx_Bitset cache, + unsigned char * translate, int c) +#else +static rx_Bitset +inverse_translation (rxb, valid, cache, translate, c) + struct re_pattern_buffer * rxb; + char * valid; + rx_Bitset cache; + unsigned char * translate; + int c; +#endif +{ + rx_Bitset cs + = cache + c * rx_bitset_numb_subsets (rxb->rx.local_cset_size); + + if (!valid[c]) + { + int x; + int c_tr = TRANSLATE(c); + rx_bitset_null (rxb->rx.local_cset_size, cs); + for (x = 0; x < 256; ++x) /* &&&& 13.37 */ + if (TRANSLATE(x) == c_tr) + RX_bitset_enjoin (cs, x); + valid[c] = 1; + } + return cs; +} + + + + +/* More subroutine declarations and macros for regex_compile. */ + +/* Returns true if REGNUM is in one of COMPILE_STACK's elements and + false if it's not. */ + +#ifdef __STDC__ +static boolean +group_in_compile_stack (compile_stack_type compile_stack, regnum_t regnum) +#else +static boolean +group_in_compile_stack (compile_stack, regnum) + compile_stack_type compile_stack; + regnum_t regnum; +#endif +{ + int this_element; + + for (this_element = compile_stack.avail - 1; + this_element >= 0; + this_element--) + if (compile_stack.stack[this_element].regnum == regnum) + return true; + + return false; +} + + +/* + * Read the ending character of a range (in a bracket expression) from the + * uncompiled pattern *P_PTR (which ends at PEND). We assume the + * starting character is in `P[-2]'. (`P[-1]' is the character `-'.) + * Then we set the translation of all bits between the starting and + * ending characters (inclusive) in the compiled pattern B. + * + * Return an error code. + * + * We use these short variable names so we can use the same macros as + * `regex_compile' itself. + */ + +#ifdef __STDC__ +static reg_errcode_t +compile_range (struct re_pattern_buffer * rxb, rx_Bitset cs, + __const__ char ** p_ptr, __const__ char * pend, + unsigned char * translate, reg_syntax_t syntax, + rx_Bitset inv_tr, char * valid_inv_tr) +#else +static reg_errcode_t +compile_range (rxb, cs, p_ptr, pend, translate, syntax, inv_tr, valid_inv_tr) + struct re_pattern_buffer * rxb; + rx_Bitset cs; + __const__ char ** p_ptr; + __const__ char * pend; + unsigned char * translate; + reg_syntax_t syntax; + rx_Bitset inv_tr; + char * valid_inv_tr; +#endif +{ + unsigned this_char; + + __const__ char *p = *p_ptr; + + unsigned char range_end; + unsigned char range_start = TRANSLATE(p[-2]); + + if (p == pend) + return REG_ERANGE; + + PATFETCH (range_end); + + (*p_ptr)++; + + if (range_start > range_end) + return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR; + + for (this_char = range_start; this_char <= range_end; this_char++) + { + rx_Bitset it = + inverse_translation (rxb, valid_inv_tr, inv_tr, translate, this_char); + rx_bitset_union (rxb->rx.local_cset_size, cs, it); + } + + return REG_NOERROR; +} + + +/* This searches a regexp for backreference side effects. + * It fills in the array OUT with 1 at the index of every register pair + * referenced by a backreference. + * + * This is used to help optimize patterns for searching. The information is + * useful because, if the caller doesn't want register values, backreferenced + * registers are the only registers for which we need rx_backtrack. + */ + +#ifdef __STDC__ +static void +find_backrefs (char * out, struct rexp_node * rexp, + struct re_se_params * params) +#else +static void +find_backrefs (out, rexp, params) + char * out; + struct rexp_node * rexp; + struct re_se_params * params; +#endif +{ + if (rexp) + switch (rexp->type) + { + case r_cset: + case r_data: + return; + case r_alternate: + case r_concat: + case r_opt: + case r_star: + case r_2phase_star: + find_backrefs (out, rexp->params.pair.left, params); + find_backrefs (out, rexp->params.pair.right, params); + return; + case r_side_effect: + if ( ((long)rexp->params.side_effect >= 0) + && (params [(long)rexp->params.side_effect].se == re_se_backref)) + out[ params [(long)rexp->params.side_effect].op1] = 1; + return; + } +} + + + +/* Returns 0 unless the pattern can match the empty string. */ + +#ifdef __STDC__ +static int +compute_fastset (struct re_pattern_buffer * rxb, struct rexp_node * rexp) +#else +static int +compute_fastset (rxb, rexp) + struct re_pattern_buffer * rxb; + struct rexp_node * rexp; +#endif +{ + if (!rexp) + return 1; + switch (rexp->type) + { + case r_data: + return 1; + case r_cset: + { + rx_bitset_union (rxb->rx.local_cset_size, + rxb->fastset, rexp->params.cset); + } + return 0; + case r_concat: + return (compute_fastset (rxb, rexp->params.pair.left) + && compute_fastset (rxb, rexp->params.pair.right)); + case r_2phase_star: + compute_fastset (rxb, rexp->params.pair.left); + /* compute_fastset (rxb, rexp->params.pair.right); nope... */ + return 1; + case r_alternate: + return !!(compute_fastset (rxb, rexp->params.pair.left) + + compute_fastset (rxb, rexp->params.pair.right)); + case r_opt: + case r_star: + compute_fastset (rxb, rexp->params.pair.left); + return 1; + case r_side_effect: + return 1; + } + + /* this should never happen */ + return 0; +} + + +/* returns + * 1 -- yes, definately anchored by the given side effect. + * 2 -- maybe anchored, maybe the empty string. + * 0 -- definately not anchored + * There is simply no other possibility. + */ + +#ifdef __STDC__ +static int +is_anchored (struct rexp_node * rexp, rx_side_effect se) +#else +static int +is_anchored (rexp, se) + struct rexp_node * rexp; + rx_side_effect se; +#endif +{ + if (!rexp) + return 2; + switch (rexp->type) + { + case r_cset: + case r_data: + return 0; + case r_concat: + case r_2phase_star: + { + int l = is_anchored (rexp->params.pair.left, se); + return (l == 2 ? is_anchored (rexp->params.pair.right, se) : l); + } + case r_alternate: + { + int l = is_anchored (rexp->params.pair.left, se); + int r = l ? is_anchored (rexp->params.pair.right, se) : 0; + + if (l == r) + return l; + else if ((l == 0) || (r == 0)) + return 0; + else + return 2; + } + case r_opt: + case r_star: + return is_anchored (rexp->params.pair.left, se) ? 2 : 0; + + case r_side_effect: + return ((rexp->params.side_effect == se) + ? 1 : 2); + } + + /* this should never happen */ + return 0; +} + + +/* This removes register assignments that aren't required by backreferencing. + * This can speed up explore_future, especially if it eliminates + * non-determinism in the superstate NFA. + * + * NEEDED is an array of characters, presumably filled in by FIND_BACKREFS. + * The non-zero elements of the array indicate which register assignments + * can NOT be removed from the expression. + */ + +#ifdef __STDC__ +static struct rexp_node * +remove_unecessary_side_effects (struct rx * rx, char * needed, + struct rexp_node * rexp, + struct re_se_params * params) +#else +static struct rexp_node * +remove_unecessary_side_effects (rx, needed, rexp, params) + struct rx * rx; + char * needed; + struct rexp_node * rexp; + struct re_se_params * params; +#endif +{ + struct rexp_node * l; + struct rexp_node * r; + if (!rexp) + return 0; + else + switch (rexp->type) + { + case r_cset: + case r_data: + return rexp; + case r_alternate: + case r_concat: + case r_2phase_star: + l = remove_unecessary_side_effects (rx, needed, + rexp->params.pair.left, params); + r = remove_unecessary_side_effects (rx, needed, + rexp->params.pair.right, params); + if ((l && r) || (rexp->type != r_concat)) + { + rexp->params.pair.left = l; + rexp->params.pair.right = r; + return rexp; + } + else + { + rexp->params.pair.left = rexp->params.pair.right = 0; + rx_free_rexp (rx, rexp); + return l ? l : r; + } + case r_opt: + case r_star: + l = remove_unecessary_side_effects (rx, needed, + rexp->params.pair.left, params); + if (l) + { + rexp->params.pair.left = l; + return rexp; + } + else + { + rexp->params.pair.left = 0; + rx_free_rexp (rx, rexp); + return 0; + } + case r_side_effect: + { + int se = (long)rexp->params.side_effect; + if ( (se >= 0) + && ( ((enum re_side_effects)params[se].se == re_se_lparen) + || ((enum re_side_effects)params[se].se == re_se_rparen)) + && (params [se].op1 > 0) + && (!needed [params [se].op1])) + { + rx_free_rexp (rx, rexp); + return 0; + } + else + return rexp; + } + } + + /* this should never happen */ + return 0; +} + + + +#ifdef __STDC__ +static int +pointless_if_repeated (struct rexp_node * node, struct re_se_params * params) +#else +static int +pointless_if_repeated (node, params) + struct rexp_node * node; + struct re_se_params * params; +#endif +{ + if (!node) + return 1; + switch (node->type) + { + case r_cset: + return 0; + case r_alternate: + case r_concat: + case r_2phase_star: + return (pointless_if_repeated (node->params.pair.left, params) + && pointless_if_repeated (node->params.pair.right, params)); + case r_opt: + case r_star: + return pointless_if_repeated (node->params.pair.left, params); + case r_side_effect: + switch (((long)node->params.side_effect < 0) + ? (enum re_side_effects)node->params.side_effect + : (enum re_side_effects)params[(long)node->params.side_effect].se) + { + case re_se_try: + case re_se_at_dot: + case re_se_begbuf: + case re_se_hat: + case re_se_wordbeg: + case re_se_wordbound: + case re_se_notwordbound: + case re_se_wordend: + case re_se_endbuf: + case re_se_dollar: + case re_se_fail: + case re_se_win: + return 1; + case re_se_lparen: + case re_se_rparen: + case re_se_iter: + case re_se_end_iter: + case re_se_syntax: + case re_se_not_syntax: + case re_se_backref: + return 0; + } + case r_data: + default: + return 0; + } +} + + + +#ifdef __STDC__ +static int +registers_on_stack (struct re_pattern_buffer * rxb, + struct rexp_node * rexp, int in_danger, + struct re_se_params * params) +#else +static int +registers_on_stack (rxb, rexp, in_danger, params) + struct re_pattern_buffer * rxb; + struct rexp_node * rexp; + int in_danger; + struct re_se_params * params; +#endif +{ + if (!rexp) + return 0; + else + switch (rexp->type) + { + case r_cset: + case r_data: + return 0; + case r_alternate: + case r_concat: + return ( registers_on_stack (rxb, rexp->params.pair.left, + in_danger, params) + || (registers_on_stack + (rxb, rexp->params.pair.right, + in_danger, params))); + case r_opt: + return registers_on_stack (rxb, rexp->params.pair.left, 0, params); + case r_star: + return registers_on_stack (rxb, rexp->params.pair.left, 1, params); + case r_2phase_star: + return + ( registers_on_stack (rxb, rexp->params.pair.left, 1, params) + || registers_on_stack (rxb, rexp->params.pair.right, 1, params)); + case r_side_effect: + { + int se = (long)rexp->params.side_effect; + if ( in_danger + && (se >= 0) + && (params [se].op1 > 0) + && ( ((enum re_side_effects)params[se].se == re_se_lparen) + || ((enum re_side_effects)params[se].se == re_se_rparen))) + return 1; + else + return 0; + } + } + + /* this should never happen */ + return 0; +} + + + +static char idempotent_complex_se[] = +{ +#define RX_WANT_SE_DEFS 1 +#undef RX_DEF_SE +#undef RX_DEF_CPLX_SE +#define RX_DEF_SE(IDEM, NAME, VALUE) +#define RX_DEF_CPLX_SE(IDEM, NAME, VALUE) IDEM, +#include <regex.h> +#undef RX_DEF_SE +#undef RX_DEF_CPLX_SE +#undef RX_WANT_SE_DEFS + 23 +}; + +static char idempotent_se[] = +{ + 13, +#define RX_WANT_SE_DEFS 1 +#undef RX_DEF_SE +#undef RX_DEF_CPLX_SE +#define RX_DEF_SE(IDEM, NAME, VALUE) IDEM, +#define RX_DEF_CPLX_SE(IDEM, NAME, VALUE) +#include <regex.h> +#undef RX_DEF_SE +#undef RX_DEF_CPLX_SE +#undef RX_WANT_SE_DEFS + 42 +}; + + + +#ifdef __STDC__ +static int +has_any_se (struct rx * rx, + struct rexp_node * rexp) +#else +static int +has_any_se (rx, rexp) + struct rx * rx; + struct rexp_node * rexp; +#endif +{ + if (!rexp) + return 0; + + switch (rexp->type) + { + case r_cset: + case r_data: + return 0; + + case r_side_effect: + return 1; + + case r_2phase_star: + case r_concat: + case r_alternate: + return + ( has_any_se (rx, rexp->params.pair.left) + || has_any_se (rx, rexp->params.pair.right)); + + case r_opt: + case r_star: + return has_any_se (rx, rexp->params.pair.left); + } + + /* this should never happen */ + return 0; +} + + + +/* This must be called AFTER `convert_hard_loops' for a given REXP. */ +#ifdef __STDC__ +static int +has_non_idempotent_epsilon_path (struct rx * rx, + struct rexp_node * rexp, + struct re_se_params * params) +#else +static int +has_non_idempotent_epsilon_path (rx, rexp, params) + struct rx * rx; + struct rexp_node * rexp; + struct re_se_params * params; +#endif +{ + if (!rexp) + return 0; + + switch (rexp->type) + { + case r_cset: + case r_data: + case r_star: + return 0; + + case r_side_effect: + return + !((long)rexp->params.side_effect > 0 + ? idempotent_complex_se [ params [(long)rexp->params.side_effect].se ] + : idempotent_se [-(long)rexp->params.side_effect]); + + case r_alternate: + return + ( has_non_idempotent_epsilon_path (rx, + rexp->params.pair.left, params) + || has_non_idempotent_epsilon_path (rx, + rexp->params.pair.right, params)); + + case r_2phase_star: + case r_concat: + return + ( has_non_idempotent_epsilon_path (rx, + rexp->params.pair.left, params) + && has_non_idempotent_epsilon_path (rx, + rexp->params.pair.right, params)); + + case r_opt: + return has_non_idempotent_epsilon_path (rx, + rexp->params.pair.left, params); + } + + /* this should never happen */ + return 0; +} + + + +/* This computes rougly what it's name suggests. It can (and does) go wrong + * in the direction of returning spurious 0 without causing disasters. + */ +#ifdef __STDC__ +static int +begins_with_complex_se (struct rx * rx, struct rexp_node * rexp) +#else +static int +begins_with_complex_se (rx, rexp) + struct rx * rx; + struct rexp_node * rexp; +#endif +{ + if (!rexp) + return 0; + + switch (rexp->type) + { + case r_cset: + case r_data: + return 0; + + case r_side_effect: + return ((long)rexp->params.side_effect >= 0); + + case r_alternate: + return + ( begins_with_complex_se (rx, rexp->params.pair.left) + && begins_with_complex_se (rx, rexp->params.pair.right)); + + + case r_concat: + return has_any_se (rx, rexp->params.pair.left); + case r_opt: + case r_star: + case r_2phase_star: + return 0; + } + + /* this should never happen */ + return 0; +} + + +/* This destructively removes some of the re_se_tv side effects from + * a rexp tree. In particular, during parsing re_se_tv was inserted on the + * right half of every | to guarantee that posix path preference could be + * honored. This function removes some which it can be determined aren't + * needed. + */ + +#ifdef __STDC__ +static void +speed_up_alt (struct rx * rx, + struct rexp_node * rexp, + int unposix) +#else +static void +speed_up_alt (rx, rexp, unposix) + struct rx * rx; + struct rexp_node * rexp; + int unposix; +#endif +{ + if (!rexp) + return; + + switch (rexp->type) + { + case r_cset: + case r_data: + case r_side_effect: + return; + + case r_opt: + case r_star: + speed_up_alt (rx, rexp->params.pair.left, unposix); + return; + + case r_2phase_star: + case r_concat: + speed_up_alt (rx, rexp->params.pair.left, unposix); + speed_up_alt (rx, rexp->params.pair.right, unposix); + return; + + case r_alternate: + /* the right child is guaranteed to be (concat re_se_tv <subexp>) */ + + speed_up_alt (rx, rexp->params.pair.left, unposix); + speed_up_alt (rx, rexp->params.pair.right->params.pair.right, unposix); + + if ( unposix + || (begins_with_complex_se + (rx, rexp->params.pair.right->params.pair.right)) + || !( has_any_se (rx, rexp->params.pair.right->params.pair.right) + || has_any_se (rx, rexp->params.pair.left))) + { + struct rexp_node * conc = rexp->params.pair.right; + rexp->params.pair.right = conc->params.pair.right; + conc->params.pair.right = 0; + rx_free_rexp (rx, conc); + } + } +} + + + + + +/* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX. + Returns one of error codes defined in `regex.h', or zero for success. + + Assumes the `allocated' (and perhaps `buffer') and `translate' + fields are set in BUFP on entry. + + If it succeeds, results are put in BUFP (if it returns an error, the + contents of BUFP are undefined): + `buffer' is the compiled pattern; + `syntax' is set to SYNTAX; + `used' is set to the length of the compiled pattern; + `fastmap_accurate' is set to zero; + `re_nsub' is set to the number of groups in PATTERN; + `not_bol' and `not_eol' are set to zero. + + The `fastmap' and `newline_anchor' fields are neither + examined nor set. */ + + +#ifdef __STDC__ +RX_DECL reg_errcode_t +rx_compile (__const__ char *pattern, int size, + reg_syntax_t syntax, + struct re_pattern_buffer * rxb) +#else +RX_DECL reg_errcode_t +rx_compile (pattern, size, syntax, rxb) + __const__ char *pattern; + int size; + reg_syntax_t syntax; + struct re_pattern_buffer * rxb; +#endif +{ + RX_subset + inverse_translate [CHAR_SET_SIZE * rx_bitset_numb_subsets(CHAR_SET_SIZE)]; + char + validate_inv_tr [CHAR_SET_SIZE * rx_bitset_numb_subsets(CHAR_SET_SIZE)]; + + /* We fetch characters from PATTERN here. Even though PATTERN is + `char *' (i.e., signed), we declare these variables as unsigned, so + they can be reliably used as array indices. */ + register unsigned char c, c1; + + /* A random tempory spot in PATTERN. */ + __const__ char *p1; + + /* Keeps track of unclosed groups. */ + compile_stack_type compile_stack; + + /* Points to the current (ending) position in the pattern. */ + __const__ char *p = pattern; + __const__ char *pend = pattern + size; + + /* How to translate the characters in the pattern. */ + unsigned char *translate = (rxb->translate + ? rxb->translate + : rx_id_translation); + + /* When parsing is done, this will hold the expression tree. */ + struct rexp_node * rexp = 0; + + /* In the midst of compilation, this holds onto the regexp + * first parst while rexp goes on to aquire additional constructs. + */ + struct rexp_node * orig_rexp = 0; + struct rexp_node * fewer_side_effects = 0; + + /* This and top_expression are saved on the compile stack. */ + struct rexp_node ** top_expression = &rexp; + struct rexp_node ** last_expression = top_expression; + + /* Parameter to `goto append_node' */ + struct rexp_node * append; + + /* Counts open-groups as they are encountered. This is the index of the + * innermost group being compiled. + */ + regnum_t regnum = 0; + + /* Place in the uncompiled pattern (i.e., the {) to + * which to go back if the interval is invalid. + */ + __const__ char *beg_interval; + + struct re_se_params * params = 0; + int paramc = 0; /* How many complex side effects so far? */ + + rx_side_effect side; /* param to `goto add_side_effect' */ + + bzero (validate_inv_tr, sizeof (validate_inv_tr)); + + rxb->rx.instruction_table = rx_id_instruction_table; + + + /* Initialize the compile stack. */ + compile_stack.stack = (( compile_stack_elt_t *) malloc ((INIT_COMPILE_STACK_SIZE) * sizeof ( compile_stack_elt_t))); + if (compile_stack.stack == 0) + return REG_ESPACE; + + compile_stack.size = INIT_COMPILE_STACK_SIZE; + compile_stack.avail = 0; + + /* Initialize the pattern buffer. */ + rxb->rx.cache = &default_cache; + rxb->syntax = syntax; + rxb->fastmap_accurate = 0; + rxb->not_bol = rxb->not_eol = 0; + rxb->least_subs = 0; + + /* Always count groups, whether or not rxb->no_sub is set. + * The whole pattern is implicitly group 0, so counting begins + * with 1. + */ + rxb->re_nsub = 0; + +#if !defined (emacs) && !defined (SYNTAX_TABLE) + /* Initialize the syntax table. */ + init_syntax_once (); +#endif + + /* Loop through the uncompiled pattern until we're at the end. */ + while (p != pend) + { + PATFETCH (c); + + switch (c) + { + case '^': + { + if ( /* If at start of pattern, it's an operator. */ + p == pattern + 1 + /* If context independent, it's an operator. */ + || syntax & RE_CONTEXT_INDEP_ANCHORS + /* Otherwise, depends on what's come before. */ + || at_begline_loc_p (pattern, p, syntax)) + { + struct rexp_node * n + = rx_mk_r_side_effect (&rxb->rx, (rx_side_effect)re_se_hat); + if (!n) + return REG_ESPACE; + append = n; + goto append_node; + } + else + goto normal_char; + } + break; + + + case '$': + { + if ( /* If at end of pattern, it's an operator. */ + p == pend + /* If context independent, it's an operator. */ + || syntax & RE_CONTEXT_INDEP_ANCHORS + /* Otherwise, depends on what's next. */ + || at_endline_loc_p (p, pend, syntax)) + { + struct rexp_node * n + = rx_mk_r_side_effect (&rxb->rx, (rx_side_effect)re_se_dollar); + if (!n) + return REG_ESPACE; + append = n; + goto append_node; + } + else + goto normal_char; + } + break; + + + case '+': + case '?': + if ((syntax & RE_BK_PLUS_QM) + || (syntax & RE_LIMITED_OPS)) + goto normal_char; + + handle_plus: + case '*': + /* If there is no previous pattern... */ + if (pointless_if_repeated (*last_expression, params)) + { + if (syntax & RE_CONTEXT_INVALID_OPS) + return REG_BADRPT; + else if (!(syntax & RE_CONTEXT_INDEP_OPS)) + goto normal_char; + } + + { + /* 1 means zero (many) matches is allowed. */ + char zero_times_ok = 0, many_times_ok = 0; + + /* If there is a sequence of repetition chars, collapse it + down to just one (the right one). We can't combine + interval operators with these because of, e.g., `a{2}*', + which should only match an even number of `a's. */ + + for (;;) + { + zero_times_ok |= c != '+'; + many_times_ok |= c != '?'; + + if (p == pend) + break; + + PATFETCH (c); + + if (c == '*' + || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?'))) + ; + + else if (syntax & RE_BK_PLUS_QM && c == '\\') + { + if (p == pend) return REG_EESCAPE; + + PATFETCH (c1); + if (!(c1 == '+' || c1 == '?')) + { + PATUNFETCH; + PATUNFETCH; + break; + } + + c = c1; + } + else + { + PATUNFETCH; + break; + } + + /* If we get here, we found another repeat character. */ + } + + /* Star, etc. applied to an empty pattern is equivalent + to an empty pattern. */ + if (!last_expression) + break; + + /* Now we know whether or not zero matches is allowed + * and also whether or not two or more matches is allowed. + */ + + { + struct rexp_node * inner_exp = *last_expression; + int need_sync = 0; + + if (many_times_ok + && has_non_idempotent_epsilon_path (&rxb->rx, + inner_exp, params)) + { + struct rexp_node * pusher + = rx_mk_r_side_effect (&rxb->rx, + (rx_side_effect)re_se_pushpos); + struct rexp_node * checker + = rx_mk_r_side_effect (&rxb->rx, + (rx_side_effect)re_se_chkpos); + struct rexp_node * pushback + = rx_mk_r_side_effect (&rxb->rx, + (rx_side_effect)re_se_pushback); + rx_Bitset cs = rx_cset (&rxb->rx); + struct rexp_node * lit_t; + struct rexp_node * fake_state; + struct rexp_node * phase2; + struct rexp_node * popper; + struct rexp_node * star; + struct rexp_node * a; + struct rexp_node * whole_thing; + + if (! cs) + return REG_ESPACE; + lit_t = rx_mk_r_cset (&rxb->rx, cs); + fake_state = rx_mk_r_concat (&rxb->rx, pushback, lit_t); + phase2 = rx_mk_r_concat (&rxb->rx, checker, fake_state); + popper = rx_mk_r_side_effect (&rxb->rx, + (rx_side_effect)re_se_poppos); + star = rx_mk_r_2phase_star (&rxb->rx, inner_exp, phase2); + a = rx_mk_r_concat (&rxb->rx, pusher, star); + whole_thing = rx_mk_r_concat (&rxb->rx, a, popper); + + if (!(pusher && star && pushback && lit_t && fake_state + && lit_t && phase2 && checker && popper + && a && whole_thing)) + return REG_ESPACE; + RX_bitset_enjoin (cs, 't'); + *last_expression = whole_thing; + } + else + { + struct rexp_node * star = + (many_times_ok ? rx_mk_r_star : rx_mk_r_opt) + (&rxb->rx, *last_expression); + if (!star) + return REG_ESPACE; + *last_expression = star; + need_sync = has_any_se (&rxb->rx, *last_expression); + } + if (!zero_times_ok) + { + struct rexp_node * concat + = rx_mk_r_concat (&rxb->rx, inner_exp, + rx_copy_rexp (&rxb->rx, + *last_expression)); + if (!concat) + return REG_ESPACE; + *last_expression = concat; + } + if (need_sync) + { + int sync_se = paramc; + params = (params + ? ((struct re_se_params *) + realloc (params, + sizeof (*params) * (1 + paramc))) + : ((struct re_se_params *) + malloc (sizeof (*params)))); + if (!params) + return REG_ESPACE; + ++paramc; + params [sync_se].se = re_se_tv; + side = (rx_side_effect)sync_se; + goto add_side_effect; + } + } + /* The old regex.c used to optimize `.*\n'. + * Maybe rx should too? + */ + } + break; + + + case '.': + { + rx_Bitset cs = rx_cset (&rxb->rx); + struct rexp_node * n = rx_mk_r_cset (&rxb->rx, cs); + if (!(cs && n)) + return REG_ESPACE; + + rx_bitset_universe (rxb->rx.local_cset_size, cs); + if (!(rxb->syntax & RE_DOT_NEWLINE)) + RX_bitset_remove (cs, '\n'); + if (!(rxb->syntax & RE_DOT_NOT_NULL)) + RX_bitset_remove (cs, 0); + + append = n; + goto append_node; + break; + } + + + case '[': + if (p == pend) return REG_EBRACK; + { + boolean had_char_class = false; + rx_Bitset cs = rx_cset (&rxb->rx); + struct rexp_node * node = rx_mk_r_cset (&rxb->rx, cs); + int is_inverted = *p == '^'; + + if (!(node && cs)) + return REG_ESPACE; + + /* This branch of the switch is normally exited with + *`goto append_node' + */ + append = node; + + if (is_inverted) + p++; + + /* Remember the first position in the bracket expression. */ + p1 = p; + + /* Read in characters and ranges, setting map bits. */ + for (;;) + { + if (p == pend) return REG_EBRACK; + + PATFETCH (c); + + /* \ might escape characters inside [...] and [^...]. */ + if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\') + { + if (p == pend) return REG_EESCAPE; + + PATFETCH (c1); + { + rx_Bitset it = inverse_translation (rxb, + validate_inv_tr, + inverse_translate, + translate, + c1); + rx_bitset_union (rxb->rx.local_cset_size, cs, it); + } + continue; + } + + /* Could be the end of the bracket expression. If it's + not (i.e., when the bracket expression is `[]' so + far), the ']' character bit gets set way below. */ + if (c == ']' && p != p1 + 1) + goto finalize_class_and_append; + + /* Look ahead to see if it's a range when the last thing + was a character class. */ + if (had_char_class && c == '-' && *p != ']') + return REG_ERANGE; + + /* Look ahead to see if it's a range when the last thing + was a character: if this is a hyphen not at the + beginning or the end of a list, then it's the range + operator. */ + if (c == '-' + && !(p - 2 >= pattern && p[-2] == '[') + && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^') + && *p != ']') + { + reg_errcode_t ret + = compile_range (rxb, cs, &p, pend, translate, syntax, + inverse_translate, validate_inv_tr); + if (ret != REG_NOERROR) return ret; + } + + else if (p[0] == '-' && p[1] != ']') + { /* This handles ranges made up of characters only. */ + reg_errcode_t ret; + + /* Move past the `-'. */ + PATFETCH (c1); + + ret = compile_range (rxb, cs, &p, pend, translate, syntax, + inverse_translate, validate_inv_tr); + if (ret != REG_NOERROR) return ret; + } + + /* See if we're at the beginning of a possible character + class. */ + + else if ((syntax & RE_CHAR_CLASSES) + && (c == '[') && (*p == ':')) + { + char str[CHAR_CLASS_MAX_LENGTH + 1]; + + PATFETCH (c); + c1 = 0; + + /* If pattern is `[[:'. */ + if (p == pend) return REG_EBRACK; + + for (;;) + { + PATFETCH (c); + if (c == ':' || c == ']' || p == pend + || c1 == CHAR_CLASS_MAX_LENGTH) + break; + str[c1++] = c; + } + str[c1] = '\0'; + + /* If isn't a word bracketed by `[:' and:`]': + undo the ending character, the letters, and leave + the leading `:' and `[' (but set bits for them). */ + if (c == ':' && *p == ']') + { + int ch; + boolean is_alnum = !strcmp (str, "alnum"); + boolean is_alpha = !strcmp (str, "alpha"); + boolean is_blank = !strcmp (str, "blank"); + boolean is_cntrl = !strcmp (str, "cntrl"); + boolean is_digit = !strcmp (str, "digit"); + boolean is_graph = !strcmp (str, "graph"); + boolean is_lower = !strcmp (str, "lower"); + boolean is_print = !strcmp (str, "print"); + boolean is_punct = !strcmp (str, "punct"); + boolean is_space = !strcmp (str, "space"); + boolean is_upper = !strcmp (str, "upper"); + boolean is_xdigit = !strcmp (str, "xdigit"); + + if (!IS_CHAR_CLASS (str)) return REG_ECTYPE; + + /* Throw away the ] at the end of the character + class. */ + PATFETCH (c); + + if (p == pend) return REG_EBRACK; + + for (ch = 0; ch < 1 << CHARBITS; ch++) + { + if ( (is_alnum && isalnum (ch)) + || (is_alpha && isalpha (ch)) + || (is_blank && isblank (ch)) + || (is_cntrl && iscntrl (ch)) + || (is_digit && isdigit (ch)) + || (is_graph && isgraph (ch)) + || (is_lower && islower (ch)) + || (is_print && isprint (ch)) + || (is_punct && ispunct (ch)) + || (is_space && isspace (ch)) + || (is_upper && isupper (ch)) + || (is_xdigit && isxdigit (ch))) + { + rx_Bitset it = + inverse_translation (rxb, + validate_inv_tr, + inverse_translate, + translate, + ch); + rx_bitset_union (rxb->rx.local_cset_size, + cs, it); + } + } + had_char_class = true; + } + else + { + c1++; + while (c1--) + PATUNFETCH; + { + rx_Bitset it = + inverse_translation (rxb, + validate_inv_tr, + inverse_translate, + translate, + '['); + rx_bitset_union (rxb->rx.local_cset_size, + cs, it); + } + { + rx_Bitset it = + inverse_translation (rxb, + validate_inv_tr, + inverse_translate, + translate, + ':'); + rx_bitset_union (rxb->rx.local_cset_size, + cs, it); + } + had_char_class = false; + } + } + else + { + had_char_class = false; + { + rx_Bitset it = inverse_translation (rxb, + validate_inv_tr, + inverse_translate, + translate, + c); + rx_bitset_union (rxb->rx.local_cset_size, cs, it); + } + } + } + + finalize_class_and_append: + if (is_inverted) + { + rx_bitset_complement (rxb->rx.local_cset_size, cs); + if (syntax & RE_HAT_LISTS_NOT_NEWLINE) + RX_bitset_remove (cs, '\n'); + } + goto append_node; + } + break; + + + case '(': + if (syntax & RE_NO_BK_PARENS) + goto handle_open; + else + goto normal_char; + + + case ')': + if (syntax & RE_NO_BK_PARENS) + goto handle_close; + else + goto normal_char; + + + case '\n': + if (syntax & RE_NEWLINE_ALT) + goto handle_alt; + else + goto normal_char; + + + case '|': + if (syntax & RE_NO_BK_VBAR) + goto handle_alt; + else + goto normal_char; + + + case '{': + if ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) + goto handle_interval; + else + goto normal_char; + + + case '\\': + if (p == pend) return REG_EESCAPE; + + /* Do not translate the character after the \, so that we can + distinguish, e.g., \B from \b, even if we normally would + translate, e.g., B to b. */ + PATFETCH_RAW (c); + + switch (c) + { + case '(': + if (syntax & RE_NO_BK_PARENS) + goto normal_backslash; + + handle_open: + rxb->re_nsub++; + regnum++; + if (COMPILE_STACK_FULL) + { + ((compile_stack.stack) = + (compile_stack_elt_t *) realloc (compile_stack.stack, ( compile_stack.size << 1) * sizeof ( + compile_stack_elt_t))); + if (compile_stack.stack == 0) return REG_ESPACE; + + compile_stack.size <<= 1; + } + + if (*last_expression) + { + struct rexp_node * concat + = rx_mk_r_concat (&rxb->rx, *last_expression, 0); + if (!concat) + return REG_ESPACE; + *last_expression = concat; + last_expression = &concat->params.pair.right; + } + + /* + * These are the values to restore when we hit end of this + * group. + */ + COMPILE_STACK_TOP.top_expression = top_expression; + COMPILE_STACK_TOP.last_expression = last_expression; + COMPILE_STACK_TOP.regnum = regnum; + + compile_stack.avail++; + + top_expression = last_expression; + break; + + + case ')': + if (syntax & RE_NO_BK_PARENS) goto normal_backslash; + + handle_close: + /* See similar code for backslashed left paren above. */ + if (COMPILE_STACK_EMPTY) + if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) + goto normal_char; + else + return REG_ERPAREN; + + /* Since we just checked for an empty stack above, this + ``can't happen''. */ + + { + /* We don't just want to restore into `regnum', because + later groups should continue to be numbered higher, + as in `(ab)c(de)' -- the second group is #2. */ + regnum_t this_group_regnum; + struct rexp_node ** inner = top_expression; + + compile_stack.avail--; + top_expression = COMPILE_STACK_TOP.top_expression; + last_expression = COMPILE_STACK_TOP.last_expression; + this_group_regnum = COMPILE_STACK_TOP.regnum; + { + int left_se = paramc; + int right_se = paramc + 1; + + params = (params + ? ((struct re_se_params *) + realloc (params, + (paramc + 2) * sizeof (params[0]))) + : ((struct re_se_params *) + malloc (2 * sizeof (params[0])))); + if (!params) + return REG_ESPACE; + paramc += 2; + + params[left_se].se = re_se_lparen; + params[left_se].op1 = this_group_regnum; + params[right_se].se = re_se_rparen; + params[right_se].op1 = this_group_regnum; + { + struct rexp_node * left + = rx_mk_r_side_effect (&rxb->rx, + (rx_side_effect)left_se); + struct rexp_node * right + = rx_mk_r_side_effect (&rxb->rx, + (rx_side_effect)right_se); + struct rexp_node * c1 + = (*inner + ? rx_mk_r_concat (&rxb->rx, left, *inner) : left); + struct rexp_node * c2 + = rx_mk_r_concat (&rxb->rx, c1, right); + if (!(left && right && c1 && c2)) + return REG_ESPACE; + *inner = c2; + } + } + break; + } + + case '|': /* `\|'. */ + if ((syntax & RE_LIMITED_OPS) || (syntax & RE_NO_BK_VBAR)) + goto normal_backslash; + handle_alt: + if (syntax & RE_LIMITED_OPS) + goto normal_char; + + { + struct rexp_node * alt + = rx_mk_r_alternate (&rxb->rx, *top_expression, 0); + if (!alt) + return REG_ESPACE; + *top_expression = alt; + last_expression = &alt->params.pair.right; + { + int sync_se = paramc; + + params = (params + ? ((struct re_se_params *) + realloc (params, + (paramc + 1) * sizeof (params[0]))) + : ((struct re_se_params *) + malloc (sizeof (params[0])))); + if (!params) + return REG_ESPACE; + ++paramc; + + params[sync_se].se = re_se_tv; + { + struct rexp_node * sync + = rx_mk_r_side_effect (&rxb->rx, + (rx_side_effect)sync_se); + struct rexp_node * conc + = rx_mk_r_concat (&rxb->rx, sync, 0); + + if (!sync || !conc) + return REG_ESPACE; + + *last_expression = conc; + last_expression = &conc->params.pair.right; + } + } + } + break; + + + case '{': + /* If \{ is a literal. */ + if (!(syntax & RE_INTERVALS) + /* If we're at `\{' and it's not the open-interval + operator. */ + || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) + || (p - 2 == pattern && p == pend)) + goto normal_backslash; + + handle_interval: + { + /* If got here, then the syntax allows intervals. */ + + /* At least (most) this many matches must be made. */ + int lower_bound = -1, upper_bound = -1; + + beg_interval = p - 1; + + if (p == pend) + { + if (syntax & RE_NO_BK_BRACES) + goto unfetch_interval; + else + return REG_EBRACE; + } + + GET_UNSIGNED_NUMBER (lower_bound); + + if (c == ',') + { + GET_UNSIGNED_NUMBER (upper_bound); + if (upper_bound < 0) upper_bound = RE_DUP_MAX; + } + else + /* Interval such as `{1}' => match exactly once. */ + upper_bound = lower_bound; + + if (lower_bound < 0 || upper_bound > RE_DUP_MAX + || lower_bound > upper_bound) + { + if (syntax & RE_NO_BK_BRACES) + goto unfetch_interval; + else + return REG_BADBR; + } + + if (!(syntax & RE_NO_BK_BRACES)) + { + if (c != '\\') return REG_EBRACE; + PATFETCH (c); + } + + if (c != '}') + { + if (syntax & RE_NO_BK_BRACES) + goto unfetch_interval; + else + return REG_BADBR; + } + + /* We just parsed a valid interval. */ + + /* If it's invalid to have no preceding re. */ + if (pointless_if_repeated (*last_expression, params)) + { + if (syntax & RE_CONTEXT_INVALID_OPS) + return REG_BADRPT; + else if (!(syntax & RE_CONTEXT_INDEP_OPS)) + goto unfetch_interval; + /* was: else laststart = b; */ + } + + /* If the upper bound is zero, don't want to iterate + * at all. + */ + if (upper_bound == 0) + { + if (*last_expression) + { + rx_free_rexp (&rxb->rx, *last_expression); + *last_expression = 0; + } + } + else + /* Otherwise, we have a nontrivial interval. */ + { + int iter_se = paramc; + int end_se = paramc + 1; + params = (params + ? ((struct re_se_params *) + realloc (params, + sizeof (*params) * (2 + paramc))) + : ((struct re_se_params *) + malloc (2 * sizeof (*params)))); + if (!params) + return REG_ESPACE; + paramc += 2; + params [iter_se].se = re_se_iter; + params [iter_se].op1 = lower_bound; + params[iter_se].op2 = upper_bound; + + params[end_se].se = re_se_end_iter; + params[end_se].op1 = lower_bound; + params[end_se].op2 = upper_bound; + { + struct rexp_node * push0 + = rx_mk_r_side_effect (&rxb->rx, + (rx_side_effect)re_se_push0); + struct rexp_node * start_one_iter + = rx_mk_r_side_effect (&rxb->rx, + (rx_side_effect)iter_se); + struct rexp_node * phase1 + = rx_mk_r_concat (&rxb->rx, start_one_iter, + *last_expression); + struct rexp_node * pushback + = rx_mk_r_side_effect (&rxb->rx, + (rx_side_effect)re_se_pushback); + rx_Bitset cs = rx_cset (&rxb->rx); + struct rexp_node * lit_t; + struct rexp_node * phase2; + struct rexp_node * loop; + struct rexp_node * push_n_loop; + struct rexp_node * final_test; + struct rexp_node * full_exp; + + if (! cs) + return REG_ESPACE; + lit_t = rx_mk_r_cset (&rxb->rx, cs); + phase2 = rx_mk_r_concat (&rxb->rx, pushback, lit_t); + loop = rx_mk_r_2phase_star (&rxb->rx, phase1, phase2); + push_n_loop = rx_mk_r_concat (&rxb->rx, push0, loop); + final_test = rx_mk_r_side_effect (&rxb->rx, + (rx_side_effect)end_se); + full_exp = rx_mk_r_concat (&rxb->rx, push_n_loop, final_test); + + if (!(push0 && start_one_iter && phase1 + && pushback && lit_t && phase2 + && loop && push_n_loop && final_test && full_exp)) + return REG_ESPACE; + + RX_bitset_enjoin(cs, 't'); + + *last_expression = full_exp; + } + } + beg_interval = 0; + } + break; + + unfetch_interval: + /* If an invalid interval, match the characters as literals. */ + p = beg_interval; + beg_interval = 0; + + /* normal_char and normal_backslash need `c'. */ + PATFETCH (c); + + if (!(syntax & RE_NO_BK_BRACES)) + { + if (p > pattern && p[-1] == '\\') + goto normal_backslash; + } + goto normal_char; + +#ifdef emacs + /* There is no way to specify the before_dot and after_dot + operators. rms says this is ok. --karl */ + case '=': + side = (rx_side_effect)rx_se_at_dot; + goto add_side_effect; + break; + + case 's': + case 'S': + { + rx_Bitset cs = rx_cset (&rxb->rx); + struct rexp_node * set = rx_mk_r_cset (&rxb->rx, cs); + if (!(cs && set)) + return REG_ESPACE; + if (c == 'S') + rx_bitset_universe (rxb->rx.local_cset_size, cs); + + PATFETCH (c); + { + int x; + enum syntaxcode code = syntax_spec_code [c]; + for (x = 0; x < 256; ++x) + { + + if (SYNTAX (x) == code) + { + rx_Bitset it = + inverse_translation (rxb, validate_inv_tr, + inverse_translate, + translate, x); + rx_bitset_xor (rxb->rx.local_cset_size, cs, it); + } + } + } + append = set; + goto append_node; + } + break; +#endif /* emacs */ + + + case 'w': + case 'W': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + { + rx_Bitset cs = rx_cset (&rxb->rx); + struct rexp_node * n = (cs ? rx_mk_r_cset (&rxb->rx, cs) : 0); + if (!(cs && n)) + return REG_ESPACE; + if (c == 'W') + rx_bitset_universe (rxb->rx.local_cset_size ,cs); + { + int x; + for (x = rxb->rx.local_cset_size - 1; x > 0; --x) + if (SYNTAX(x) & Sword) + RX_bitset_toggle (cs, x); + } + append = n; + goto append_node; + } + break; + +/* With a little extra work, some of these side effects could be optimized + * away (basicly by looking at what we already know about the surrounding + * chars). + */ + case '<': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + side = (rx_side_effect)re_se_wordbeg; + goto add_side_effect; + break; + + case '>': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + side = (rx_side_effect)re_se_wordend; + goto add_side_effect; + break; + + case 'b': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + side = (rx_side_effect)re_se_wordbound; + goto add_side_effect; + break; + + case 'B': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + side = (rx_side_effect)re_se_notwordbound; + goto add_side_effect; + break; + + case '`': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + side = (rx_side_effect)re_se_begbuf; + goto add_side_effect; + break; + + case '\'': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + side = (rx_side_effect)re_se_endbuf; + goto add_side_effect; + break; + + add_side_effect: + { + struct rexp_node * se + = rx_mk_r_side_effect (&rxb->rx, side); + if (!se) + return REG_ESPACE; + append = se; + goto append_node; + } + break; + + case '1': case '2': case '3': case '4': case '5': + case '6': case '7': case '8': case '9': + if (syntax & RE_NO_BK_REFS) + goto normal_char; + + c1 = c - '0'; + + if (c1 > regnum) + return REG_ESUBREG; + + /* Can't back reference to a subexpression if inside of it. */ + if (group_in_compile_stack (compile_stack, c1)) + return REG_ESUBREG; + + { + int backref_se = paramc; + params = (params + ? ((struct re_se_params *) + realloc (params, + sizeof (*params) * (1 + paramc))) + : ((struct re_se_params *) + malloc (sizeof (*params)))); + if (!params) + return REG_ESPACE; + ++paramc; + params[backref_se].se = re_se_backref; + params[backref_se].op1 = c1; + side = (rx_side_effect)backref_se; + goto add_side_effect; + } + break; + + case '+': + case '?': + if (syntax & RE_BK_PLUS_QM) + goto handle_plus; + else + goto normal_backslash; + + default: + normal_backslash: + /* You might think it would be useful for \ to mean + not to translate; but if we don't translate it + it will never match anything. */ + c = TRANSLATE (c); + goto normal_char; + } + break; + + + default: + /* Expects the character in `c'. */ + normal_char: + { + rx_Bitset cs = rx_cset(&rxb->rx); + struct rexp_node * match = rx_mk_r_cset (&rxb->rx, cs); + rx_Bitset it; + if (!(cs && match)) + return REG_ESPACE; + it = inverse_translation (rxb, validate_inv_tr, + inverse_translate, translate, c); + rx_bitset_union (CHAR_SET_SIZE, cs, it); + append = match; + + append_node: + /* This genericly appends the rexp APPEND to *LAST_EXPRESSION + * and then parses the next character normally. + */ + if (*last_expression) + { + struct rexp_node * concat + = rx_mk_r_concat (&rxb->rx, *last_expression, append); + if (!concat) + return REG_ESPACE; + *last_expression = concat; + last_expression = &concat->params.pair.right; + } + else + *last_expression = append; + } + } /* switch (c) */ + } /* while p != pend */ + + + { + int win_se = paramc; + params = (params + ? ((struct re_se_params *) + realloc (params, + sizeof (*params) * (1 + paramc))) + : ((struct re_se_params *) + malloc (sizeof (*params)))); + if (!params) + return REG_ESPACE; + ++paramc; + params[win_se].se = re_se_win; + { + struct rexp_node * se + = rx_mk_r_side_effect (&rxb->rx, (rx_side_effect)win_se); + struct rexp_node * concat + = rx_mk_r_concat (&rxb->rx, rexp, se); + if (!(se && concat)) + return REG_ESPACE; + rexp = concat; + } + } + + + /* Through the pattern now. */ + + if (!COMPILE_STACK_EMPTY) + return REG_EPAREN; + + free (compile_stack.stack); + + orig_rexp = rexp; +#ifdef RX_DEBUG + if (rx_debug_compile) + { + dbug_rxb = rxb; + fputs ("\n\nCompiling ", stdout); + fwrite (pattern, 1, size, stdout); + fputs (":\n", stdout); + rxb->se_params = params; + print_rexp (&rxb->rx, orig_rexp, 2, re_seprint, stdout); + } +#endif + { + rx_Bitset cs = rx_cset(&rxb->rx); + rx_Bitset cs2 = rx_cset(&rxb->rx); + char * se_map = (char *) alloca (paramc); + struct rexp_node * new_rexp = 0; + + + bzero (se_map, paramc); + find_backrefs (se_map, rexp, params); + fewer_side_effects = + remove_unecessary_side_effects (&rxb->rx, se_map, + rx_copy_rexp (&rxb->rx, rexp), params); + + speed_up_alt (&rxb->rx, rexp, 0); + speed_up_alt (&rxb->rx, fewer_side_effects, 1); + + { + char * syntax_parens = rxb->syntax_parens; + if (syntax_parens == (char *)0x1) + rexp = remove_unecessary_side_effects + (&rxb->rx, se_map, rexp, params); + else if (syntax_parens) + { + int x; + for (x = 0; x < paramc; ++x) + if (( (params[x].se == re_se_lparen) + || (params[x].se == re_se_rparen)) + && (!syntax_parens [params[x].op1])) + se_map [x] = 1; + rexp = remove_unecessary_side_effects + (&rxb->rx, se_map, rexp, params); + } + } + + /* At least one more optimization would be nice to have here but i ran out + * of time. The idea would be to delay side effects. + * For examle, `(abc)' is the same thing as `abc()' except that the + * left paren is offset by 3 (which we know at compile time). + * (In this comment, write that second pattern `abc(:3:)' + * where `(:3:' is a syntactic unit.) + * + * Trickier: `(abc|defg)' is the same as `(abc(:3:|defg(:4:))' + * (The paren nesting may be hard to follow -- that's an alternation + * of `abc(:3:' and `defg(:4:' inside (purely syntactic) parens + * followed by the closing paren from the original expression.) + * + * Neither the expression tree representation nor the the nfa make + * this very easy to write. :( + */ + + /* What we compile is different than what the parser returns. + * Suppose the parser returns expression R. + * Let R' be R with unnecessary register assignments removed + * (see REMOVE_UNECESSARY_SIDE_EFFECTS, above). + * + * What we will compile is the expression: + * + * m{try}R{win}\|s{try}R'{win} + * + * {try} and {win} denote side effect epsilons (see EXPLORE_FUTURE). + * + * When trying a match, we insert an `m' at the beginning of the + * string if the user wants registers to be filled, `s' if not. + */ + new_rexp = + rx_mk_r_alternate + (&rxb->rx, + rx_mk_r_concat (&rxb->rx, rx_mk_r_cset (&rxb->rx, cs2), rexp), + rx_mk_r_concat (&rxb->rx, + rx_mk_r_cset (&rxb->rx, cs), fewer_side_effects)); + + if (!(new_rexp && cs && cs2)) + return REG_ESPACE; + RX_bitset_enjoin (cs2, '\0'); /* prefixed to the rexp used for matching. */ + RX_bitset_enjoin (cs, '\1'); /* prefixed to the rexp used for searching. */ + rexp = new_rexp; + } + +#ifdef RX_DEBUG + if (rx_debug_compile) + { + fputs ("\n...which is compiled as:\n", stdout); + print_rexp (&rxb->rx, rexp, 2, re_seprint, stdout); + } +#endif + { + struct rx_nfa_state *start = 0; + struct rx_nfa_state *end = 0; + + if (!rx_build_nfa (&rxb->rx, rexp, &start, &end)) + return REG_ESPACE; /* */ + else + { + void * mem = (void *)rxb->buffer; + unsigned long size = rxb->allocated; + int start_id; + char * perm_mem; + int iterator_size = paramc * sizeof (params[0]); + + end->is_final = 1; + start->is_start = 1; + rx_name_nfa_states (&rxb->rx); + start_id = start->id; +#ifdef RX_DEBUG + if (rx_debug_compile) + { + fputs ("...giving the NFA: \n", stdout); + dbug_rxb = rxb; + print_nfa (&rxb->rx, rxb->rx.nfa_states, re_seprint, stdout); + } +#endif + if (!rx_eclose_nfa (&rxb->rx)) + return REG_ESPACE; + else + { + rx_delete_epsilon_transitions (&rxb->rx); + + /* For compatability reasons, we need to shove the + * compiled nfa into one chunk of malloced memory. + */ + rxb->rx.reserved = ( sizeof (params[0]) * paramc + + rx_sizeof_bitset (rxb->rx.local_cset_size)); +#ifdef RX_DEBUG + if (rx_debug_compile) + { + dbug_rxb = rxb; + fputs ("...which cooks down (uncompactified) to: \n", stdout); + print_nfa (&rxb->rx, rxb->rx.nfa_states, re_seprint, stdout); + } +#endif + if (!rx_compactify_nfa (&rxb->rx, &mem, &size)) + return REG_ESPACE; + rxb->buffer = mem; + rxb->allocated = size; + rxb->rx.buffer = mem; + rxb->rx.allocated = size; + perm_mem = ((char *)rxb->rx.buffer + + rxb->rx.allocated - rxb->rx.reserved); + rxb->se_params = ((struct re_se_params *)perm_mem); + bcopy (params, rxb->se_params, iterator_size); + perm_mem += iterator_size; + rxb->fastset = (rx_Bitset) perm_mem; + rxb->start = rx_id_to_nfa_state (&rxb->rx, start_id); + } + rx_bitset_null (rxb->rx.local_cset_size, rxb->fastset); + rxb->can_match_empty = compute_fastset (rxb, orig_rexp); + rxb->match_regs_on_stack = + registers_on_stack (rxb, orig_rexp, 0, params); + rxb->search_regs_on_stack = + registers_on_stack (rxb, fewer_side_effects, 0, params); + if (rxb->can_match_empty) + rx_bitset_universe (rxb->rx.local_cset_size, rxb->fastset); + rxb->is_anchored = is_anchored (orig_rexp, (rx_side_effect) re_se_hat); + rxb->begbuf_only = is_anchored (orig_rexp, + (rx_side_effect) re_se_begbuf); + } + rx_free_rexp (&rxb->rx, rexp); + if (params) + free (params); +#ifdef RX_DEBUG + if (rx_debug_compile) + { + dbug_rxb = rxb; + fputs ("...which cooks down to: \n", stdout); + print_nfa (&rxb->rx, rxb->rx.nfa_states, re_seprint, stdout); + } +#endif + } + return REG_NOERROR; +} + + + +/* This table gives an error message for each of the error codes listed + in regex.h. Obviously the order here has to be same as there. */ + +__const__ char * rx_error_msg[] = +{ 0, /* REG_NOERROR */ + "No match", /* REG_NOMATCH */ + "Invalid regular expression", /* REG_BADPAT */ + "Invalid collation character", /* REG_ECOLLATE */ + "Invalid character class name", /* REG_ECTYPE */ + "Trailing backslash", /* REG_EESCAPE */ + "Invalid back reference", /* REG_ESUBREG */ + "Unmatched [ or [^", /* REG_EBRACK */ + "Unmatched ( or \\(", /* REG_EPAREN */ + "Unmatched \\{", /* REG_EBRACE */ + "Invalid content of \\{\\}", /* REG_BADBR */ + "Invalid range end", /* REG_ERANGE */ + "Memory exhausted", /* REG_ESPACE */ + "Invalid preceding regular expression", /* REG_BADRPT */ + "Premature end of regular expression", /* REG_EEND */ + "Regular expression too big", /* REG_ESIZE */ + "Unmatched ) or \\)", /* REG_ERPAREN */ +}; + + + + +char rx_slowmap [256] = +{ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, +}; + +#ifdef __STDC__ +RX_DECL void +rx_blow_up_fastmap (struct re_pattern_buffer * rxb) +#else +RX_DECL void +rx_blow_up_fastmap (rxb) + struct re_pattern_buffer * rxb; +#endif +{ + int x; + for (x = 0; x < 256; ++x) /* &&&& 3.6 % */ + rxb->fastmap [x] = !!RX_bitset_member (rxb->fastset, x); + rxb->fastmap_accurate = 1; +} + + + + +#if !defined(REGEX_MALLOC) && !defined(__GNUC__) +#define RE_SEARCH_2_FN inner_re_search_2 +#define RE_S2_QUAL static +#else +#define RE_SEARCH_2_FN re_search_2 +#define RE_S2_QUAL +#endif + +struct re_search_2_closure +{ + __const__ char * string1; + int size1; + __const__ char * string2; + int size2; +}; + +RE_S2_QUAL int + RE_SEARCH_2_FN (struct re_pattern_buffer *, + __const__ char *, + int, __const__ char *, int, int, + int, struct re_registers *, int); +int re_rx_search (struct re_pattern_buffer *, int, + int, int, int, rx_get_burst_fn, + rx_back_check_fn, rx_fetch_char_fn, + void *, struct re_registers *, + struct rx_search_state *, + struct rx_search_state *); +#if !defined(REGEX_MALLOC) && !defined(__GNUC__) +int re_search_2 (struct re_pattern_buffer *, + __const__ char *, int, + __const__ char *, int, + int, int, struct re_registers *, + int); +#endif +int re_search (struct re_pattern_buffer *, + __const__ char *, int, int, int, + struct re_registers *); +int re_match_2 (struct re_pattern_buffer *, + __const__ char *, int, + __const__ char *, int, + int, struct re_registers *, int); +int re_match (struct re_pattern_buffer *, + __const__ char *, int, int, + struct re_registers *); +reg_syntax_t re_set_syntax (reg_syntax_t); +void re_set_registers (struct re_pattern_buffer *, + struct re_registers *, unsigned, + regoff_t *, regoff_t *); +static int cplx_se_sublist_len (struct rx_se_list *); +static int posix_se_list_order (struct rx *, struct rx_se_list *, + struct rx_se_list *); +__const__ char + *re_compile_pattern (__const__ char *, int, + struct re_pattern_buffer *); +int re_compile_fastmap (struct re_pattern_buffer *); +char *re_comp (__const__ char *); +int re_exec (__const__ char *); +int regcomp (regex_t *, __const__ char *, int); +int regexec (__const__ regex_t *, + __const__ char *, size_t, + regmatch_t pmatch[], int); +size_t regerror (int, __const__ regex_t *, + char *, size_t); + +#ifdef __STDC__ +static __inline__ enum rx_get_burst_return +re_search_2_get_burst ( struct rx_string_position * pos, + void * vclosure, int stop ) +#else +static __inline__ enum rx_get_burst_return +re_search_2_get_burst (pos, vclosure, stop) + struct rx_string_position * pos; + void * vclosure; + int stop; +#endif +{ + struct re_search_2_closure * closure; + closure = (struct re_search_2_closure *)vclosure; + if (!closure->string2) + { + int inset; + + inset = pos->pos - pos->string; + if ((inset < -1) || (inset > closure->size1)) + return rx_get_burst_no_more; + else + { + pos->pos = (__const__ unsigned char *) closure->string1 + inset; + pos->string = (__const__ unsigned char *) closure->string1; + pos->size = closure->size1; + pos->end = ((__const__ unsigned char *) + MIN(closure->string1 + closure->size1, + closure->string1 + stop)); + pos->offset = 0; + return ((pos->pos < pos->end) + ? rx_get_burst_ok + : rx_get_burst_no_more); + } + } + else if (!closure->string1) + { + int inset; + + inset = pos->pos - pos->string; + pos->pos = (__const__ unsigned char *) closure->string2 + inset; + pos->string = (__const__ unsigned char *) closure->string2; + pos->size = closure->size2; + pos->end = ((__const__ unsigned char *) + MIN(closure->string2 + closure->size2, + closure->string2 + stop)); + pos->offset = 0; + return ((pos->pos < pos->end) + ? rx_get_burst_ok + : rx_get_burst_no_more); + } + else + { + int inset; + + inset = pos->pos - pos->string + pos->offset; + if (inset < closure->size1) + { + pos->pos = (__const__ unsigned char *) closure->string1 + inset; + pos->string = (__const__ unsigned char *) closure->string1; + pos->size = closure->size1; + pos->end = ((__const__ unsigned char *) + MIN(closure->string1 + closure->size1, + closure->string1 + stop)); + pos->offset = 0; + return rx_get_burst_ok; + } + else + { + pos->pos = ((__const__ unsigned char *) + closure->string2 + inset - closure->size1); + pos->string = (__const__ unsigned char *) closure->string2; + pos->size = closure->size2; + pos->end = ((__const__ unsigned char *) + MIN(closure->string2 + closure->size2, + closure->string2 + stop - closure->size1)); + pos->offset = closure->size1; + return ((pos->pos < pos->end) + ? rx_get_burst_ok + : rx_get_burst_no_more); + } + } +} + + +#ifdef __STDC__ +static __inline__ enum rx_back_check_return +re_search_2_back_check ( struct rx_string_position * pos, + int lparen, int rparen, unsigned char * translate, + void * vclosure, int stop ) +#else +static __inline__ enum rx_back_check_return +re_search_2_back_check (pos, lparen, rparen, translate, vclosure, stop) + struct rx_string_position * pos; + int lparen; + int rparen; + unsigned char * translate; + void * vclosure; + int stop; +#endif +{ + struct rx_string_position there; + struct rx_string_position past; + + there = *pos; + there.pos = there.string + lparen - there.offset; + re_search_2_get_burst (&there, vclosure, stop); + + past = *pos; + past.pos = past.string + rparen - there.offset; + re_search_2_get_burst (&past, vclosure, stop); + + ++pos->pos; + re_search_2_get_burst (pos, vclosure, stop); + + while ( (there.pos != past.pos) + && (pos->pos != pos->end)) + if (TRANSLATE(*there.pos) != TRANSLATE(*pos->pos)) + return rx_back_check_fail; + else + { + ++there.pos; + ++pos->pos; + if (there.pos == there.end) + re_search_2_get_burst (&there, vclosure, stop); + if (pos->pos == pos->end) + re_search_2_get_burst (pos, vclosure, stop); + } + + if (there.pos != past.pos) + return rx_back_check_fail; + --pos->pos; + re_search_2_get_burst (pos, vclosure, stop); + return rx_back_check_pass; +} + +#ifdef __STDC__ +static __inline__ int +re_search_2_fetch_char ( struct rx_string_position * pos, int offset, + void * app_closure, int stop ) +#else +static __inline__ int +re_search_2_fetch_char (pos, offset, app_closure, stop) + struct rx_string_position * pos; + int offset; + void * app_closure; + int stop; +#endif +{ + struct re_search_2_closure * closure; + closure = (struct re_search_2_closure *)app_closure; + if (offset == 0) + { + if (pos->pos >= pos->string) + return *pos->pos; + else + { + if ( (pos->string == (__const__ unsigned char *) closure->string2) + && (closure->string1) + && (closure->size1)) + return closure->string1[closure->size1 - 1]; + else + return 0; /* sure, why not. */ + } + } + if (pos->pos == pos->end) + return *closure->string2; + else +#if 0 + return pos->pos[1]; +#else + return pos->pos[offset]; /* FIXME */ +#endif +} + +#ifdef __STDC__ +RE_S2_QUAL int +RE_SEARCH_2_FN (struct re_pattern_buffer *rxb, + __const__ char * string1, int size1, + __const__ char * string2, int size2, + int startpos, int range, + struct re_registers *regs, + int stop) +#else +RE_S2_QUAL int +RE_SEARCH_2_FN (rxb, + string1, size1, string2, size2, startpos, range, regs, stop) + struct re_pattern_buffer *rxb; + __const__ char * string1; + int size1; + __const__ char * string2; + int size2; + int startpos; + int range; + struct re_registers *regs; + int stop; +#endif +{ + int answer; + struct re_search_2_closure closure; + closure.string1 = string1; + closure.size1 = size1; + closure.string2 = string2; + closure.size2 = size2; + answer = rx_search (rxb, startpos, range, stop, size1 + size2, + re_search_2_get_burst, + re_search_2_back_check, + re_search_2_fetch_char, + (void *)&closure, + regs, + 0, + 0); + switch (answer) + { + case rx_search_continuation: + abort (); + case rx_search_error: + return -2; + case rx_search_soft_fail: + case rx_search_fail: + return -1; + default: + return answer; + } +} + +/* Export rx_search to callers outside this file. */ + +#ifdef __STDC__ +int +re_rx_search ( struct re_pattern_buffer * rxb, int startpos, int range, + int stop, int total_size, rx_get_burst_fn get_burst, + rx_back_check_fn back_check, rx_fetch_char_fn fetch_char, + void * app_closure, struct re_registers * regs, + struct rx_search_state * resume_state, + struct rx_search_state * save_state ) +#else +int +re_rx_search (rxb, startpos, range, stop, total_size, + get_burst, back_check, fetch_char, + app_closure, regs, resume_state, save_state) + struct re_pattern_buffer * rxb; + int startpos; + int range; + int stop; + int total_size; + rx_get_burst_fn get_burst; + rx_back_check_fn back_check; + rx_fetch_char_fn fetch_char; + void * app_closure; + struct re_registers * regs; + struct rx_search_state * resume_state; + struct rx_search_state * save_state; +#endif +{ + return rx_search (rxb, startpos, range, stop, total_size, + get_burst, back_check, fetch_char, app_closure, + regs, resume_state, save_state); +} + +#if !defined(REGEX_MALLOC) && !defined(__GNUC__) +#ifdef __STDC__ +int +re_search_2 (struct re_pattern_buffer *rxb, + __const__ char * string1, int size1, + __const__ char * string2, int size2, + int startpos, int range, + struct re_registers *regs, + int stop) +#else +int +re_search_2 (rxb, string1, size1, string2, size2, startpos, range, regs, stop) + struct re_pattern_buffer *rxb; + __const__ char * string1; + int size1; + __const__ char * string2; + int size2; + int startpos; + int range; + struct re_registers *regs; + int stop; +#endif +{ + int ret; + ret = inner_re_search_2 (rxb, string1, size1, string2, size2, startpos, + range, regs, stop); + alloca (0); + return ret; +} +#endif + + +/* Like re_search_2, above, but only one string is specified, and + * doesn't let you say where to stop matching. + */ + +#ifdef __STDC__ +int +re_search (struct re_pattern_buffer * rxb, __const__ char *string, + int size, int startpos, int range, + struct re_registers *regs) +#else +int +re_search (rxb, string, size, startpos, range, regs) + struct re_pattern_buffer * rxb; + __const__ char * string; + int size; + int startpos; + int range; + struct re_registers *regs; +#endif +{ + return re_search_2 (rxb, 0, 0, string, size, startpos, range, regs, size); +} + +#ifdef __STDC__ +int +re_match_2 (struct re_pattern_buffer * rxb, + __const__ char * string1, int size1, + __const__ char * string2, int size2, + int pos, struct re_registers *regs, int stop) +#else +int +re_match_2 (rxb, string1, size1, string2, size2, pos, regs, stop) + struct re_pattern_buffer * rxb; + __const__ char * string1; + int size1; + __const__ char * string2; + int size2; + int pos; + struct re_registers *regs; + int stop; +#endif +{ + struct re_registers some_regs; + regoff_t start; + regoff_t end; + int srch; + int save = rxb->regs_allocated; + struct re_registers * regs_to_pass = regs; + char *old_fastmap = rxb->fastmap; + + if (!regs) + { + some_regs.start = &start; + some_regs.end = &end; + some_regs.num_regs = 1; + regs_to_pass = &some_regs; + rxb->regs_allocated = REGS_FIXED; + } + + rxb->fastmap = NULL; + srch = re_search_2 (rxb, string1, size1, string2, size2, + pos, 1, regs_to_pass, stop); + rxb->fastmap = old_fastmap; + if (regs_to_pass != regs) + rxb->regs_allocated = save; + if (srch < 0) + return srch; + return regs_to_pass->end[0] - regs_to_pass->start[0]; +} + +/* re_match is like re_match_2 except it takes only a single string. */ + +#ifdef __STDC__ +int +re_match (struct re_pattern_buffer * rxb, + __const__ char * string, + int size, int pos, + struct re_registers *regs) +#else +int +re_match (rxb, string, size, pos, regs) + struct re_pattern_buffer * rxb; + __const__ char *string; + int size; + int pos; + struct re_registers *regs; +#endif +{ + return re_match_2 (rxb, string, size, 0, 0, pos, regs, size); +} + + + +/* Set by `re_set_syntax' to the current regexp syntax to recognize. Can + also be assigned to arbitrarily: each pattern buffer stores its own + syntax, so it can be changed between regex compilations. */ +reg_syntax_t re_syntax_options = RE_SYNTAX_EMACS; + + +/* Specify the precise syntax of regexps for compilation. This provides + for compatibility for various utilities which historically have + different, incompatible syntaxes. + + The argument SYNTAX is a bit mask comprised of the various bits + defined in regex.h. We return the old syntax. */ + +#ifdef __STDC__ +reg_syntax_t +re_set_syntax (reg_syntax_t syntax) +#else +reg_syntax_t +re_set_syntax (syntax) + reg_syntax_t syntax; +#endif +{ + reg_syntax_t ret = re_syntax_options; + + re_syntax_options = syntax; + return ret; +} + + +/* Set REGS to hold NUM_REGS registers, storing them in STARTS and + ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use + this memory for recording register information. STARTS and ENDS + must be allocated using the malloc library routine, and must each + be at least NUM_REGS * sizeof (regoff_t) bytes long. + + If NUM_REGS == 0, then subsequent matches should allocate their own + register data. + + Unless this function is called, the first search or match using + PATTERN_BUFFER will allocate its own register data, without + freeing the old data. */ + +#ifdef __STDC__ +void +re_set_registers (struct re_pattern_buffer *bufp, + struct re_registers *regs, + unsigned num_regs, + regoff_t * starts, regoff_t * ends) +#else +void +re_set_registers (bufp, regs, num_regs, starts, ends) + struct re_pattern_buffer *bufp; + struct re_registers *regs; + unsigned num_regs; + regoff_t * starts; + regoff_t * ends; +#endif +{ + if (num_regs) + { + bufp->regs_allocated = REGS_REALLOCATE; + regs->num_regs = num_regs; + regs->start = starts; + regs->end = ends; + } + else + { + bufp->regs_allocated = REGS_UNALLOCATED; + regs->num_regs = 0; + regs->start = regs->end = (regoff_t) 0; + } +} + + + + +#ifdef __STDC__ +static int +cplx_se_sublist_len (struct rx_se_list * list) +#else +static int +cplx_se_sublist_len (list) + struct rx_se_list * list; +#endif +{ + int x = 0; + while (list) + { + if ((long)list->car >= 0) + ++x; + list = list->cdr; + } + return x; +} + + +/* For rx->se_list_cmp */ + +#ifdef __STDC__ +static int +posix_se_list_order (struct rx * rx, + struct rx_se_list * a, struct rx_se_list * b) +#else +static int +posix_se_list_order (rx, a, b) + struct rx * rx; + struct rx_se_list * a; + struct rx_se_list * b; +#endif +{ + int al = cplx_se_sublist_len (a); + int bl = cplx_se_sublist_len (b); + + if (!al && !bl) + return ((a == b) + ? 0 + : ((a < b) ? -1 : 1)); + + else if (!al) + return -1; + + else if (!bl) + return 1; + + else + { + rx_side_effect * av = ((rx_side_effect *) + alloca (sizeof (rx_side_effect) * (al + 1))); + rx_side_effect * bv = ((rx_side_effect *) + alloca (sizeof (rx_side_effect) * (bl + 1))); + struct rx_se_list * ap = a; + struct rx_se_list * bp = b; + int ai, bi; + + for (ai = al - 1; ai >= 0; --ai) + { + while ((long)ap->car < 0) + ap = ap->cdr; + av[ai] = ap->car; + ap = ap->cdr; + } + av[al] = (rx_side_effect)-2; + for (bi = bl - 1; bi >= 0; --bi) + { + while ((long)bp->car < 0) + bp = bp->cdr; + bv[bi] = bp->car; + bp = bp->cdr; + } + bv[bl] = (rx_side_effect)-1; + + { + int ret; + int x = 0; + while (av[x] == bv[x]) + ++x; + ret = (((unsigned *)(av[x]) < (unsigned *)(bv[x])) ? -1 : 1); + return ret; + } + } +} + + + + +/* re_compile_pattern is the GNU regular expression compiler: it + compiles PATTERN (of length SIZE) and puts the result in RXB. + Returns 0 if the pattern was valid, otherwise an error string. + + Assumes the `allocated' (and perhaps `buffer') and `translate' fields + are set in RXB on entry. + + We call rx_compile to do the actual compilation. */ + +#ifdef __STDC__ +__const__ char * +re_compile_pattern (__const__ char *pattern, + int length, + struct re_pattern_buffer * rxb) +#else +__const__ char * +re_compile_pattern (pattern, length, rxb) + __const__ char *pattern; + int length; + struct re_pattern_buffer * rxb; +#endif +{ + reg_errcode_t ret; + + /* GNU code is written to assume at least RE_NREGS registers will be set + (and at least one extra will be -1). */ + rxb->regs_allocated = REGS_UNALLOCATED; + + /* And GNU code determines whether or not to get register information + by passing null for the REGS argument to re_match, etc., not by + setting no_sub. */ + rxb->no_sub = 0; + + rxb->rx.local_cset_size = 256; + + /* Match anchors at newline. */ + rxb->newline_anchor = 1; + + rxb->re_nsub = 0; + rxb->start = 0; + rxb->se_params = 0; + rxb->rx.nodec = 0; + rxb->rx.epsnodec = 0; + rxb->rx.instruction_table = 0; + rxb->rx.nfa_states = 0; + rxb->rx.se_list_cmp = posix_se_list_order; + rxb->rx.start_set = 0; + + ret = rx_compile (pattern, length, re_syntax_options, rxb); + alloca (0); + return rx_error_msg[(int) ret]; +} + + +#ifdef __STDC__ +int +re_compile_fastmap (struct re_pattern_buffer * rxb) +#else +int +re_compile_fastmap (rxb) + struct re_pattern_buffer * rxb; +#endif +{ + rx_blow_up_fastmap (rxb); + return 0; +} + + + + +/* Entry points compatible with 4.2 BSD regex library. We don't define + them if this is an Emacs or POSIX compilation. */ + +#if (!defined (emacs) && !defined (_POSIX_SOURCE)) || defined(USE_BSD_REGEX) + +/* BSD has one and only one pattern buffer. */ +static struct re_pattern_buffer rx_comp_buf; + +#ifdef __STDC__ +char * +re_comp (__const__ char *s) +#else +char * +re_comp (s) + __const__ char *s; +#endif +{ + reg_errcode_t ret; + + if (!s || (*s == '\0')) + { + if (!rx_comp_buf.buffer) + return "No previous regular expression"; + return 0; + } + + if (!rx_comp_buf.fastmap) + { + rx_comp_buf.fastmap = (char *) malloc (1 << CHARBITS); + if (!rx_comp_buf.fastmap) + return "Memory exhausted"; + } + + /* Since `rx_exec' always passes NULL for the `regs' argument, we + don't need to initialize the pattern buffer fields which affect it. */ + + /* Match anchors at newlines. */ + rx_comp_buf.newline_anchor = 1; + + rx_comp_buf.re_nsub = 0; + rx_comp_buf.start = 0; + rx_comp_buf.se_params = 0; + rx_comp_buf.rx.nodec = 0; + rx_comp_buf.rx.epsnodec = 0; + rx_comp_buf.rx.instruction_table = 0; + rx_comp_buf.rx.nfa_states = 0; + rx_comp_buf.rx.start = 0; + rx_comp_buf.rx.se_list_cmp = posix_se_list_order; + rx_comp_buf.rx.start_set = 0; + rx_comp_buf.rx.local_cset_size = 256; + + ret = rx_compile (s, strlen (s), re_syntax_options, &rx_comp_buf); + alloca (0); + + /* Yes, we're discarding `__const__' here. */ + return (char *) rx_error_msg[(int) ret]; +} + + +#ifdef __STDC__ +int +re_exec (__const__ char *s) +#else +int +re_exec (s) + __const__ char *s; +#endif +{ + __const__ int len = strlen (s); + return + 0 <= re_search (&rx_comp_buf, s, len, 0, len, (struct re_registers *) 0); +} +#endif /* not emacs and not _POSIX_SOURCE */ + + + +/* POSIX.2 functions. Don't define these for Emacs. */ + +#if !defined(emacs) + +/* regcomp takes a regular expression as a string and compiles it. + + PREG is a regex_t *. We do not expect any fields to be initialized, + since POSIX says we shouldn't. Thus, we set + + `buffer' to the compiled pattern; + `used' to the length of the compiled pattern; + `syntax' to RE_SYNTAX_POSIX_EXTENDED if the + REG_EXTENDED bit in CFLAGS is set; otherwise, to + RE_SYNTAX_POSIX_BASIC; + `newline_anchor' to REG_NEWLINE being set in CFLAGS; + `fastmap' and `fastmap_accurate' to zero; + `re_nsub' to the number of subexpressions in PATTERN. + + PATTERN is the address of the pattern string. + + CFLAGS is a series of bits which affect compilation. + + If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we + use POSIX basic syntax. + + If REG_NEWLINE is set, then . and [^...] don't match newline. + Also, regexec will try a match beginning after every newline. + + If REG_ICASE is set, then we considers upper- and lowercase + versions of letters to be equivalent when matching. + + If REG_NOSUB is set, then when PREG is passed to regexec, that + routine will report only success or failure, and nothing about the + registers. + + It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for + the return codes and their meanings.) */ + + +#ifdef __STDC__ +int +regcomp (regex_t * preg, __const__ char * pattern, int cflags) +#else +int +regcomp (preg, pattern, cflags) + regex_t * preg; + __const__ char * pattern; + int cflags; +#endif +{ + reg_errcode_t ret; + unsigned syntax + = cflags & REG_EXTENDED ? RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC; + + /* regex_compile will allocate the space for the compiled pattern. */ + preg->buffer = 0; + preg->allocated = 0; + preg->fastmap = malloc (256); + if (!preg->fastmap) + return REG_ESPACE; + preg->fastmap_accurate = 0; + + if (cflags & REG_ICASE) + { + unsigned i; + + preg->translate = (unsigned char *) malloc (256); + if (!preg->translate) + return (int) REG_ESPACE; + + /* Map uppercase characters to corresponding lowercase ones. */ + for (i = 0; i < CHAR_SET_SIZE; i++) + preg->translate[i] = isupper (i) ? tolower (i) : i; + } + else + preg->translate = 0; + + /* If REG_NEWLINE is set, newlines are treated differently. */ + if (cflags & REG_NEWLINE) + { /* REG_NEWLINE implies neither . nor [^...] match newline. */ + syntax &= ~RE_DOT_NEWLINE; + syntax |= RE_HAT_LISTS_NOT_NEWLINE; + /* It also changes the matching behavior. */ + preg->newline_anchor = 1; + } + else + preg->newline_anchor = 0; + + preg->no_sub = !!(cflags & REG_NOSUB); + + /* POSIX says a null character in the pattern terminates it, so we + can use strlen here in compiling the pattern. */ + preg->re_nsub = 0; + preg->start = 0; + preg->se_params = 0; + preg->syntax_parens = 0; + preg->rx.nodec = 0; + preg->rx.epsnodec = 0; + preg->rx.instruction_table = 0; + preg->rx.nfa_states = 0; + preg->rx.local_cset_size = 256; + preg->rx.start = 0; + preg->rx.se_list_cmp = posix_se_list_order; + preg->rx.start_set = 0; + ret = rx_compile (pattern, strlen (pattern), syntax, preg); + alloca (0); + + /* POSIX doesn't distinguish between an unmatched open-group and an + unmatched close-group: both are REG_EPAREN. */ + if (ret == REG_ERPAREN) ret = REG_EPAREN; + + return (int) ret; +} + + +/* regexec searches for a given pattern, specified by PREG, in the + string STRING. + + If NMATCH is zero or REG_NOSUB was set in the cflags argument to + `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at + least NMATCH elements, and we set them to the offsets of the + corresponding matched substrings. + + EFLAGS specifies `execution flags' which affect matching: if + REG_NOTBOL is set, then ^ does not match at the beginning of the + string; if REG_NOTEOL is set, then $ does not match at the end. + + We return 0 if we find a match and REG_NOMATCH if not. */ + +#ifdef __STDC__ +int +regexec (__const__ regex_t *preg, __const__ char *string, + size_t nmatch, regmatch_t pmatch[], + int eflags) +#else +int +regexec (preg, string, nmatch, pmatch, eflags) + __const__ regex_t *preg; + __const__ char *string; + size_t nmatch; + regmatch_t pmatch[]; + int eflags; +#endif +{ + int ret; + struct re_registers regs; + regex_t private_preg; + int len = strlen (string); + boolean want_reg_info = !preg->no_sub && nmatch > 0; + + private_preg = *preg; + + private_preg.not_bol = !!(eflags & REG_NOTBOL); + private_preg.not_eol = !!(eflags & REG_NOTEOL); + + /* The user has told us exactly how many registers to return + * information about, via `nmatch'. We have to pass that on to the + * matching routines. + */ + private_preg.regs_allocated = REGS_FIXED; + + if (want_reg_info) + { + regs.num_regs = nmatch; + regs.start = (( regoff_t *) malloc ((nmatch) * sizeof ( regoff_t))); + regs.end = (( regoff_t *) malloc ((nmatch) * sizeof ( regoff_t))); + if (regs.start == 0 || regs.end == 0) + return (int) REG_NOMATCH; + } + + /* Perform the searching operation. */ + ret = re_search (&private_preg, + string, len, + /* start: */ 0, + /* range: */ len, + want_reg_info ? ®s : (struct re_registers *) 0); + + /* Copy the register information to the POSIX structure. */ + if (want_reg_info) + { + if (ret >= 0) + { + unsigned r; + + for (r = 0; r < nmatch; r++) + { + pmatch[r].rm_so = regs.start[r]; + pmatch[r].rm_eo = regs.end[r]; + } + } + + /* If we needed the temporary register info, free the space now. */ + free (regs.start); + free (regs.end); + } + + /* We want zero return to mean success, unlike `re_search'. */ + return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH; +} + + +/* Returns a message corresponding to an error code, ERRCODE, returned + from either regcomp or regexec. */ + +#ifdef __STDC__ +size_t +regerror (int errcode, __const__ regex_t *preg, + char *errbuf, size_t errbuf_size) +#else +size_t +regerror (errcode, preg, errbuf, errbuf_size) + int errcode; + __const__ regex_t *preg; + char *errbuf; + size_t errbuf_size; +#endif +{ + __const__ char *msg + = rx_error_msg[errcode] == 0 ? "Success" : rx_error_msg[errcode]; + size_t msg_size = strlen (msg) + 1; /* Includes the 0. */ + + if (errbuf_size != 0) + { + if (msg_size > errbuf_size) + { + strncpy (errbuf, msg, errbuf_size - 1); + errbuf[errbuf_size - 1] = 0; + } + else + strcpy (errbuf, msg); + } + + return msg_size; +} + + +/* Free dynamically allocated space used by PREG. */ + +#ifdef __STDC__ +void +regfree (regex_t *preg) +#else +void +regfree (preg) + regex_t *preg; +#endif +{ + if (preg->buffer != 0) + free (preg->buffer); + preg->buffer = 0; + preg->allocated = 0; + + if (preg->fastmap != 0) + free (preg->fastmap); + preg->fastmap = 0; + preg->fastmap_accurate = 0; + + if (preg->translate != 0) + free (preg->translate); + preg->translate = 0; +} + +#endif /* not emacs */ diff --git a/libc/stdlib/Makefile b/libc/stdlib/Makefile index 2c2d0f5fb..6f33f8afc 100644 --- a/libc/stdlib/Makefile +++ b/libc/stdlib/Makefile @@ -18,7 +18,7 @@ EOBJ=on_exit.o atexit.o __do_exit.o exit.o GOBJ=atoi.o atol.o ltoa.o ltostr.o \ ctype.o qsort.o bsearch.o rand.o lsearch.o getopt.o \ glob.o fnmatch.o itoa.o strtol.o crypt.o sleep.o mkstemp.o \ - mktemp.o + mktemp.o realpath.o UOBJ=getenv.o putenv.o popen.o system.o getcwd.o setenv.o \ execl.o execv.o execlp.o execvp.o execvep.o diff --git a/libc/stdlib/realpath.c b/libc/stdlib/realpath.c new file mode 100644 index 000000000..d053cfcaf --- /dev/null +++ b/libc/stdlib/realpath.c @@ -0,0 +1,168 @@ +/* + * realpath.c -- canonicalize pathname by removing symlinks + * Copyright (C) 1993 Rick Sladkey <jrs@world.std.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU Library Public License as published by + * the Free Software Foundation; either version 2, or (at your option) + * any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU Library Public License for more details. + */ + +#ifdef HAVE_CONFIG_H +#include <config.h> +#endif + +#include <sys/types.h> +#include <unistd.h> +#include <stdio.h> +#include <string.h> +#include <strings.h> +#include <limits.h> /* for PATH_MAX */ +#include <sys/param.h> /* for MAXPATHLEN */ +#include <errno.h> + +#include <sys/stat.h> /* for S_IFLNK */ + +#ifndef PATH_MAX +#ifdef _POSIX_VERSION +#define PATH_MAX _POSIX_PATH_MAX +#else +#ifdef MAXPATHLEN +#define PATH_MAX MAXPATHLEN +#else +#define PATH_MAX 1024 +#endif +#endif +#endif + +#define MAX_READLINKS 32 + +#ifdef __STDC__ +char *realpath(const char *path, char resolved_path []) +#else +char *realpath(path, resolved_path) +const char *path; +char resolved_path []; +#endif +{ + char copy_path[PATH_MAX]; + char link_path[PATH_MAX]; + char got_path [PATH_MAX]; + char *new_path = got_path; + char *max_path; + int readlinks = 0; + int n; + + /* Make a copy of the source path since we may need to modify it. */ + if (strlen(path)>=PATH_MAX-2) { + errno = ENAMETOOLONG; + return NULL; + } + strcpy(copy_path, path); + path = copy_path; + max_path = copy_path + PATH_MAX - 2; + /* If it's a relative pathname use getwd for starters. */ + if (*path != '/') { + /* Ohoo... */ +#define HAVE_GETCWD +#ifdef HAVE_GETCWD + getcwd(new_path, PATH_MAX - 1); +#else + getwd(new_path); +#endif + new_path += strlen(new_path); + if (new_path[-1] != '/') + *new_path++ = '/'; + } + else { + *new_path++ = '/'; + path++; + } + /* Expand each slash-separated pathname component. */ + while (*path != '\0') { + /* Ignore stray "/". */ + if (*path == '/') { + path++; + continue; + } + if (*path == '.') { + /* Ignore ".". */ + if (path[1] == '\0' || path[1] == '/') { + path++; + continue; + } + if (path[1] == '.') { + if (path[2] == '\0' || path[2] == '/') { + path += 2; + /* Ignore ".." at root. */ + if (new_path == got_path + 1) + continue; + /* Handle ".." by backing up. */ + while ((--new_path)[-1] != '/') + ; + continue; + } + } + } + /* Safely copy the next pathname component. */ + while (*path != '\0' && *path != '/') { + if (path > max_path) { + errno = ENAMETOOLONG; + return NULL; + } + *new_path++ = *path++; + } +#ifdef S_IFLNK + /* Protect against infinite loops. */ + if (readlinks++ > MAX_READLINKS) { + errno = ELOOP; + return NULL; + } + /* See if latest pathname component is a symlink. */ + *new_path = '\0'; + n = readlink(got_path, link_path, PATH_MAX - 1); + if (n < 0) { + /* EINVAL means the file exists but isn't a symlink. */ + if (errno != EINVAL) { + /* Make sure it's null terminated. */ + *new_path = '\0'; + strcpy (resolved_path, got_path); + return NULL; + } + } + else { + /* Note: readlink doesn't add the null byte. */ + link_path[n] = '\0'; + if (*link_path == '/') + /* Start over for an absolute symlink. */ + new_path = got_path; + else + /* Otherwise back up over this component. */ + while (*(--new_path) != '/') + ; + /* Safe sex check. */ + if (strlen(path) + n >= PATH_MAX-2) { + errno = ENAMETOOLONG; + return NULL; + } + /* Insert symlink contents into path. */ + strcat(link_path, path); + strcpy(copy_path, link_path); + path = copy_path; + } +#endif /* S_IFLNK */ + *new_path++ = '/'; + } + /* Delete trailing slash but don't whomp a lone slash. */ + if (new_path != got_path + 1 && new_path[-1] == '/') + new_path--; + /* Make sure it's null terminated. */ + *new_path = '\0'; + strcpy (resolved_path, got_path); + return resolved_path; +} |