From 39e52e8179972d92fe7cfa51da7f9dd5371b75d8 Mon Sep 17 00:00:00 2001
From: Alex Dowad <alexinbeijing@gmail.com>
Date: Mon, 14 Apr 2014 21:56:25 +0200
Subject: squark-filter: correctly identify URLs which use percent encoding

---
 src/blob.c         | 44 +++++++++++++++++++++++++++++++++++++++++++-
 src/sqdb-build.lua |  1 +
 2 files changed, 44 insertions(+), 1 deletion(-)

(limited to 'src')

diff --git a/src/blob.c b/src/blob.c
index 41e8d75..4ddc156 100644
--- a/src/blob.c
+++ b/src/blob.c
@@ -139,6 +139,24 @@ int blob_icmp(blob_t a, blob_t b)
 	return strncasecmp(a.ptr, b.ptr, a.len);
 }
 
+int blob_find_char(blob_t blob, char c)
+{
+	int i;
+	char *ptr = blob.ptr;
+	for (i = 0; i < blob.len; i++)
+		if (ptr[i] == c)
+			return i;
+	return -1;
+}
+
+unsigned char blob_read_hexbyte(blob_t *b, int i)
+{
+	if (b->len >= i+2)
+		return (dx(b->ptr[i]) << 4) + dx(b->ptr[i+1]);
+	else
+		return 0;
+}
+
 void blob_lowercase(blob_t blob)
 {
 	int i;
@@ -147,6 +165,30 @@ void blob_lowercase(blob_t blob)
 		ptr[i] = tolower(ptr[i]);
 }
 
+void blob_percent_decode(blob_t *blob)
+{
+	int i = blob_find_char(*blob, '%');
+
+	if (i >= 0) {
+		int dest  = i;
+		int len   = blob->len;
+		char* ptr = blob->ptr;
+
+		for ( ; i < blob->len; i++, dest++) {
+			if (ptr[i] == '%') {
+				ptr[dest] = blob_read_hexbyte(blob, i+1);
+				i   += 2;
+				len -= 2;
+			}
+			else {
+				ptr[dest] = ptr[i];
+			}
+		}
+
+		blob->len = len;
+	}
+}
+
 int blob_pull_inet_addr(blob_t *b, struct in_addr *saddr)
 {
 	unsigned long ip = 0;
@@ -280,7 +322,7 @@ void blob_push_urldecode(blob_t *to, blob_t url)
 			blob_expand_head(to, orig, '/');
 			blob_expand_head_bytes(to, 1); /* back up past the '/' separator */
 		} else {
-			/* copy decoded; FIXME decode percent encoding */
+			blob_percent_decode(&b);
 			blob_push_byte(to, '/');
 			blob_push(to, b);
 		}
diff --git a/src/sqdb-build.lua b/src/sqdb-build.lua
index cd039e2..2806bb2 100755
--- a/src/sqdb-build.lua
+++ b/src/sqdb-build.lua
@@ -117,6 +117,7 @@ local function read_urls(filename, category, locked)
 		url = url:gsub("#.*", "")
 		url = url:gsub(" *^", "")
 		url = url:lower()
+		url = url:gsub("%%(%x%x)", function(h) return string.char(tonumber(h,16)) end)
 		url = url:gsub("^(www%d*[.])([^.]*[.])", "%2")
 		domain, path = url:match("([^/]*)/?(.*)")
 		domain = domain:gsub(":.*", "")
-- 
cgit v1.2.3