[mod_rewrite] rewrite_raw result gets decoded

doc/mod_rewrite.xml

@@ -76,13 +76,15 @@
 	</action>
 
 	<action name="rewrite_raw">
-		<short>modify request path and querystring, matching raw path</short>
+		<short>modify request path and querystring, matching and writing raw path</short>
 		<parameter name="rule">
 			<short>a simple target string or one rule, mapping a regular expression to a target string, or a list of rules.</short>
 		</parameter>
 		<description>
 			<textile>
-				Similar to "@rewrite@":mod_rewrite.html#mod_rewrite__action_rewrite, but matches the raw path (i.e. the path before URL decoding and sanitizing).
+				Similar to "@rewrite@":mod_rewrite.html#mod_rewrite__action_rewrite, but matches the raw path (i.e. the path before URL decoding and sanitizing) and the result is decoded again.
+
+				"@rewrite@":mod_rewrite.html#mod_rewrite__action_rewrite write the result to @request.path@ and uses URL encoding to generate @request.raw_path@; "@rewrite_raw@":mod_rewrite.html#mod_rewrite__action_rewrite_raw writes @request.raw_path@ and decodes it into @request.path@. In both cases @request.path@ gets simplified afterwards.
 			</textile>
 		</description>
 	</action>

src/modules/mod_rewrite.c

@@ -94,17 +94,10 @@ error:
 	return FALSE;
 }
 
-static gboolean rewrite_internal(liVRequest *vr, GString *dest_path, GString *dest_query, rewrite_rule *rule, gboolean raw) {
-	gchar *path;
+static gboolean rewrite_internal(liVRequest *vr, GString *dest_path, GString *dest_query, rewrite_rule *rule, gchar *path) {
 	GMatchInfo *match_info = NULL;
 	GMatchInfo *prev_match_info = NULL;
 
-	if (raw) {
-		path = vr->request.uri.raw_path->str;
-	} else {
-		path = vr->request.uri.path->str;
-	}
-
 	if (NULL != rule->regex && !g_regex_match(rule->regex, path, 0, &match_info)) {
 		if (NULL != match_info) {
 			g_match_info_free(match_info);
@@ -137,6 +130,7 @@ static liHandlerResult rewrite(liVRequest *vr, gpointer param, gpointer *context
 	rewrite_data *rd = param;
 	gboolean debug = _OPTION(vr, rd->p, 0).boolean;
 	GString *dest_query = g_string_sized_new(31);
+	gchar *path = rd->raw ? vr->request.uri.raw_path->str : vr->request.uri.path->str;
 	UNUSED(context);
 
 	for (i = 0; i < rd->rules->len; i++) {
@@ -144,18 +138,36 @@ static liHandlerResult rewrite(liVRequest *vr, gpointer param, gpointer *context
 
 		rule = &g_array_index(rd->rules, rewrite_rule, i);
 
-		if (rewrite_internal(vr, dest_path, dest_query, rule, rd->raw)) {
+		if (rewrite_internal(vr, dest_path, dest_query, rule, path)) {
 			/* regex matched */
 			if (debug) {
-				VR_DEBUG(vr, "rewrite: path \"%s\" => \"%s\", query \"%s\" => \"%s\"",
-					vr->request.uri.path->str, dest_path->str,
-					vr->request.uri.query->str, dest_query->str
-				);
+				if (NULL != rule->querystring) {
+					VR_DEBUG(vr, "rewrite%s: path \"%s\" => \"%s\", query \"%s\" => \"%s\"",
+						rd->raw ? " (raw)" : "",
+						path, dest_path->str,
+						vr->request.uri.query->str, dest_query->str
+					);
+				} else {
+					VR_DEBUG(vr, "rewrite%s: path \"%s\" => \"%s\"",
+						rd->raw ? " (raw)" : "",
+						path, dest_path->str
+					);
+				}
 			}
 
 			/* change request path */
-			g_string_truncate(vr->request.uri.path, 0);
-			g_string_append_len(vr->request.uri.path, GSTR_LEN(dest_path));
+			if (rd->raw) {
+				g_string_truncate(vr->request.uri.raw_path, 0);
+				g_string_append_len(vr->request.uri.raw_path, GSTR_LEN(dest_path));
+				g_string_truncate(vr->request.uri.path, 0);
+				g_string_append_len(vr->request.uri.path, GSTR_LEN(dest_path));
+				li_url_decode(vr->request.uri.path);
+			} else {
+				g_string_truncate(vr->request.uri.path, 0);
+				g_string_append_len(vr->request.uri.path, GSTR_LEN(dest_path));
+				li_string_encode(vr->request.uri.path->str, vr->request.uri.raw_path, LI_ENCODING_URI);
+			}
+			li_path_simplify(vr->request.uri.path);
 
 			/* change request query */
 			if (NULL != rule->querystring) {

tests/t-rewrite.py

@@ -22,6 +22,37 @@ class TestRewrite2(CurlRequest):
 rewrite "/somethingelse" => "/nothing", "^/somefile$" => "/test.txt";
 defaultaction;
 """
+
+# match decoded and simplified paths by default
+class TestRewrite3(CurlRequest):
+	URL = "/http://some%2Ffile"
+	EXPECT_RESPONSE_BODY = "/dest/file"
+	EXPECT_RESPONSE_CODE = 200
+	config = """
+rewrite "/http:/some(/.*)" => "/dest$1";
+respond 200 => "%{req.path}";
+"""
+
+# match raw paths and simplify path
+class TestRewrite4(CurlRequest):
+	URL = "/http://some%2Ffile"
+	EXPECT_RESPONSE_BODY = "/dest/http:/some/file"
+	EXPECT_RESPONSE_CODE = 200
+	config = """
+rewrite_raw "(/http://some%2F.*)" => "/dest$1";
+respond 200 => "%{req.path}";
+"""
+
+# match and write raw paths
+class TestRewrite5(CurlRequest):
+	URL = "/http://some%2Ffile"
+	EXPECT_RESPONSE_BODY = "/dest/http://some%2Ffile"
+	EXPECT_RESPONSE_CODE = 200
+	config = """
+rewrite_raw "(/http://some%2F.*)" => "/dest$1";
+respond 200 => "%{req.raw_path}";
+"""
+
 class Test(GroupTest):
 	plain_config = """
 setup { module_load "mod_rewrite"; }
@@ -29,5 +60,8 @@ setup { module_load "mod_rewrite"; }
 
 	group = [
 		TestRewrite1,
-		TestRewrite2
+		TestRewrite2,
+		TestRewrite3,
+		TestRewrite4,
+		TestRewrite5,
 	]