From 2e58fc0a7a69ecbe4a48b296bcf6313825fcfa7c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Timo=20Ter=C3=A4s?= <timo.teras@iki.fi>
Date: Wed, 18 Aug 2010 23:03:58 +0300
Subject: db, filter: fix db generation of ipv4 style addresses

Properly embed the ipv4 address in database now. Teach filter
to understand the two new reserved component id's.
---
 sqdb-build.lua  | 34 ++++++++++++++++++++++++++--------
 squark-filter.c | 27 +++++++++++++++++----------
 squarkdb.h      |  3 +++
 3 files changed, 46 insertions(+), 18 deletions(-)

diff --git a/sqdb-build.lua b/sqdb-build.lua
index 2b301fc..cd039e2 100755
--- a/sqdb-build.lua
+++ b/sqdb-build.lua
@@ -193,6 +193,14 @@ local function enum_tree(cb, category, dns, data)
 	end
 end
 
+function iptonumber(str)
+	local num = 0
+	for elem in str:gmatch("%d+") do
+		num = num * 256 + assert(tonumber(elem))
+	end
+	return num
+end
+
 local function enum_all(cb)
 	local ipaddr, data, category
 
@@ -204,8 +212,7 @@ local function enum_all(cb)
 		if data.paths ~= nil then
 			enum_paths(cb, data.category, ipaddr, data.paths)
 		end
-		-- fixme, calculate ip as 32-bit value
-		cb(ipaddr, nil, 0, data.category, nil, data.paths)
+		cb(ipaddr, nil, iptonumber(ipaddr), data.category, nil, data.paths)
 	end
 end
 
@@ -307,11 +314,22 @@ db:write_section("categories", all_categories_by_id)
 db:create_index(num_entries)
 enum_all(
 	function(uri, parent_uri, component, category, childs, paths)
-		db:assign_index(db:hash(uri),
-				category and category.id or 0,
-				childs and true or false,
-				paths and true or false,
-				all_strings[component] or 0,
-				parent_uri and db:hash(parent_uri) or 0)
+		if parent_uri == nil and type(component) == "number" then
+			-- Embedded IPv4 address
+			db:assign_index(db:hash(uri),
+					category and category.id or 0,
+					childs and true or false,
+					paths and true or false,
+					component,
+					-2)
+		else
+			-- Regular entry
+			db:assign_index(db:hash(uri),
+					category and category.id or 0,
+					childs and true or false,
+					paths and true or false,
+					all_strings[component] or 0,
+					parent_uri and db:hash(parent_uri) or -1)
+		end
 	end
 )
diff --git a/squark-filter.c b/squark-filter.c
index 097f420..fac9241 100644
--- a/squark-filter.c
+++ b/squark-filter.c
@@ -182,7 +182,7 @@ static int url_classify(struct url_info *url, struct sqdb *db)
 	blob_t b, key, got, tld, mkey;
 	void *cmph;
 	struct sqdb_index_entry *indx;
-	cmph_uint32 i = -1, previ;
+	cmph_uint32 i = SQDB_PARENT_ROOT, previ;
 	int dots_done = 1;
 
 	cmph = sqdb_section_get(db, SQDB_SECTION_INDEX_MPH, NULL);
@@ -201,26 +201,33 @@ static int url_classify(struct url_info *url, struct sqdb *db)
 
 		previ = i;
 		i = cmph_search_packed(cmph, key.ptr, key.len);
-		if (blob_cmp(got, sqdb_get_string_literal(db, indx[i].component)) != 0) {
-			/* the subdomain did no longer match, use 
-			 * parents classification */
-			i = previ;
-			goto parent_dns_match;
-		}
 		if (!blob_is_null(tld)) {
-			if (blob_cmp(tld, sqdb_get_string_literal(db, indx[indx[i].parent].component)) != 0) {
+			int p = indx[i].parent;
+
+			if (p == SQDB_PARENT_ROOT ||
+			    p == SQDB_PARENT_IPV4 ||
+			    indx[p].parent != SQDB_PARENT_ROOT ||
+			    blob_cmp(tld, sqdb_get_string_literal(db, indx[p].component)) != 0) {
 				/* top level domain did not match */
 				i = -1;
 				goto parent_dns_match;
 			}
 			tld = BLOB_NULL;
+			previ = p;
+		}
+		if (indx[i].parent != previ ||
+		    blob_cmp(got, sqdb_get_string_literal(db, indx[i].component)) != 0) {
+			/* the subdomain did no longer match, use
+			 * parents classification */
+			i = previ;
+			goto parent_dns_match;
 		}
 		mkey = key;
 		dots_done++;
 	} while (indx[i].has_subdomains);
 
 	/* No paths to match for */
-	if (!indx[i].has_paths)
+	if (i == SQDB_PARENT_ROOT || !indx[i].has_paths)
 		goto parent_dns_match;
 
 	if (key.ptr != url->host.ptr) {
@@ -268,7 +275,7 @@ static int url_classify(struct url_info *url, struct sqdb *db)
 	}
 
 parent_dns_match:
-	if (i == -1)
+	if (i == SQDB_PARENT_ROOT)
 		return 0; /* no category */
 
 	return indx[i].category;
diff --git a/squarkdb.h b/squarkdb.h
index 743325e..68c1a2a 100644
--- a/squarkdb.h
+++ b/squarkdb.h
@@ -34,6 +34,9 @@ struct sqdb_header {
 	struct sqdb_section	section[SQDB_SECTION_MAX];
 };
 
+#define SQDB_PARENT_ROOT	0xffffff
+#define SQDB_PARENT_IPV4	0xfffffe
+
 struct sqdb_index_entry {
 	uint32_t		has_subdomains : 1;
 	uint32_t		has_paths : 1;
-- 
cgit v1.2.3