squarkdb: cmph based url database for squark filtering

Implement basics of squarkdb which will be used by squark-filter
to categorize URIs. Implementation is based on libcmph and uses
file format suitable to be mmap:ed from squark-filter.

Lua code is used to create the squark database from standard
domain / url blacklists.

1 files changed, 50 insertions, 0 deletions

diff --git a/squarkdb.h b/squarkdb.h
new file mode 100644
index 0000000..2f96155
--- /dev/null
+++ b/squarkdb.h
@@ -0,0 +1,50 @@
+#ifndef SQUARKDB_H
+#define SQUARKDB_H
+
+#include <stdint.h>
+
+#define SQDB_SECTION_STRINGS		0
+#define SQDB_SECTION_CATEGORIES		1
+#define SQDB_SECTION_INDEX		2
+#define SQDB_SECTION_INDEX_MPH		3
+#define SQDB_SECTION_KEYWORD		4
+#define SQDB_SECTION_KEYWORD_MPH	5
+#define SQDB_SECTION_MAX		8
+
+struct sqdb {
+	int			fd;
+	void *			mmap_base;
+	size_t			file_length;
+};
+
+struct sqdb_section {
+	u_int32_t		offset;
+	u_int32_t		length;
+};
+
+struct sqdb_header {
+	char			description[116];
+	u_int32_t		num_sections;
+	u_int32_t		version;
+	u_int32_t		magic;
+	struct sqdb_section	section[SQDB_SECTION_MAX];
+};
+
+struct sqdb_index_entry {
+	u_int32_t		has_subdomains : 1;
+	u_int32_t		has_paths : 1;
+	u_int32_t		category : 6;
+	u_int32_t		parent : 24;
+	u_int32_t		component;
+};
+
+const char *sqdb_section_names[SQDB_SECTION_MAX];
+
+int sqdb_create(struct sqdb *db, const char *fn);
+int sqdb_open(struct sqdb *db, const char *fn);
+void sqdb_close(struct sqdb *db);
+
+void *sqdb_section_create(struct sqdb *db, int id, u_int32_t size);
+void *sqdb_section_get(struct sqdb *db, int id, u_int32_t *size);
+
+#endif