From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Tue, 12 May 2015 11:46:10 +0000 (+0100)
Subject: Use raw regexp instead of utf ones by default.
X-Git-Tag: 0.9.0~32
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=3a201043a8eca6be7e7d9a8a06bf4ec18c4047ef;p=thirdparty%2Frspamd.git

Use raw regexp instead of utf ones by default.

Now to enable utf8 match, one should specify 'u' modifier, for example,
by adding '/u' to the slashed patterm.
---

diff --git a/src/libutil/regexp.c b/src/libutil/regexp.c
index 6f7d07f6fc..deaca80a5b 100644
--- a/src/libutil/regexp.c
+++ b/src/libutil/regexp.c
@@ -176,7 +176,8 @@ rspamd_regexp_new (const gchar *pattern, const gchar *flags,
 		flags_str = flags;
 	}
 
-	regexp_flags |= PCRE_UTF8 ;
+	rspamd_flags |= RSPAMD_REGEXP_FLAG_RAW;
+	regexp_flags &= ~PCRE_UTF8;
 
 	if (flags_str != NULL) {
 		while (*flags_str) {
@@ -194,7 +195,8 @@ rspamd_regexp_new (const gchar *pattern, const gchar *flags,
 				regexp_flags |= PCRE_EXTENDED;
 				break;
 			case 'u':
-				regexp_flags |= PCRE_UNGREEDY;
+				rspamd_flags &= ~RSPAMD_REGEXP_FLAG_RAW;
+				regexp_flags |= PCRE_UTF8;
 				break;
 			case 'O':
 				/* We optimize all regexps by default */
diff --git a/test/lua/unit/regxep.lua b/test/lua/unit/regxep.lua
index 04db54aa9b..8cc2db2bdc 100644
--- a/test/lua/unit/regxep.lua
+++ b/test/lua/unit/regxep.lua
@@ -15,7 +15,7 @@ context("Regexp unit tests", function()
       {'m,test,', 'test123', false},
       {'m{https?://[^/?\\s]+?:\\d+(?<!:80)(?<!:443)(?<!:8080)(?:/|\\s|$)}', '', false},
       {'/test/i', 'TeSt123', true},
-      {'/тест/i', 'ТесТ', true},
+      {'/ТесТ/iu', 'тест', true},
       -- Raw regexp
       {'/\\S<[-\\w\\.]+\\@[-\\w\\.]+>/r', 'some<example@example.com>', true},
       -- Cyrillic utf8 letter