From 3ec03548e2c72b69448bdadcc67bc008dddd43d5 Mon Sep 17 00:00:00 2001
From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Sat, 12 Aug 2017 18:04:13 +0100
Subject: [PATCH] [Minor] Store raw urls

---
 src/libserver/url.c |  5 ++++-
 src/libserver/url.h |  2 ++
 src/libutil/http.c  |  1 +
 src/lua/lua_http.c  |  3 +++
 src/lua/lua_url.c   | 22 ++++++++++++++++++++++
 5 files changed, 32 insertions(+), 1 deletion(-)

diff --git a/src/libserver/url.c b/src/libserver/url.c
index 0e28cf2d8..c908685bb 100644
--- a/src/libserver/url.c
+++ b/src/libserver/url.c
@@ -1592,7 +1592,10 @@ rspamd_url_parse (struct rspamd_url *uri, gchar *uristring, gsize len,
 		uri->flags |= RSPAMD_URL_FLAG_OBSCURED;
 	}
 
-	uri->string = p;
+	uri->raw = p;
+	uri->rawlen = len;
+	uri->string = rspamd_mempool_alloc (pool, len + 1);
+	rspamd_strlcpy (uri->string, p, len + 1);
 	uri->urllen = len;
 
 	/* Now decode url symbols */
diff --git a/src/libserver/url.h b/src/libserver/url.h
index e4834d9bc..48f5acc12 100644
--- a/src/libserver/url.h
+++ b/src/libserver/url.h
@@ -25,6 +25,7 @@ struct rspamd_url_tag {
 };
 
 struct rspamd_url {
+	gchar *raw;
 	gchar *string;
 	gint protocol;
 	guint port;
@@ -48,6 +49,7 @@ struct rspamd_url {
 	guint surbllen;
 	guint tldlen;
 	guint urllen;
+	guint rawlen;
 
 	enum rspamd_url_flags flags;
 	guint count;
diff --git a/src/libutil/http.c b/src/libutil/http.c
index 80eff63d3..7f69843ea 100644
--- a/src/libutil/http.c
+++ b/src/libutil/http.c
@@ -2385,6 +2385,7 @@ rspamd_http_message_from_url (const gchar *url)
 
 	urllen = strlen (url);
 	memset (&pu, 0, sizeof (pu));
+
 	if (http_parser_parse_url (url, urllen, FALSE, &pu) != 0) {
 		msg_warn ("cannot parse URL: %s", url);
 		return NULL;
diff --git a/src/lua/lua_http.c b/src/lua/lua_http.c
index 99a967db7..ad2a7d981 100644
--- a/src/lua/lua_http.c
+++ b/src/lua/lua_http.c
@@ -371,13 +371,16 @@ lua_http_request (lua_State *L)
 	if (lua_gettop (L) >= 2) {
 		/* url, callback and event_base format */
 		url = luaL_checkstring (L, 1);
+
 		if (url == NULL || lua_type (L, 2) != LUA_TFUNCTION) {
 			msg_err ("http request has bad params");
 			lua_pushboolean (L, FALSE);
 			return 1;
 		}
+
 		lua_pushvalue (L, 2);
 		cbref = luaL_ref (L, LUA_REGISTRYINDEX);
+
 		if (lua_gettop (L) >= 3 && rspamd_lua_check_udata_maybe (L, 3, "rspamd{ev_base}")) {
 			ev_base = *(struct event_base **)lua_touserdata (L, 3);
 		}
diff --git a/src/lua/lua_url.c b/src/lua/lua_url.c
index 15966bfa3..cbe9148a3 100644
--- a/src/lua/lua_url.c
+++ b/src/lua/lua_url.c
@@ -47,6 +47,7 @@ LUA_FUNCTION_DEF (url, get_path);
 LUA_FUNCTION_DEF (url, get_query);
 LUA_FUNCTION_DEF (url, get_fragment);
 LUA_FUNCTION_DEF (url, get_text);
+LUA_FUNCTION_DEF (url, get_raw);
 LUA_FUNCTION_DEF (url, get_tld);
 LUA_FUNCTION_DEF (url, to_table);
 LUA_FUNCTION_DEF (url, is_phished);
@@ -73,6 +74,7 @@ static const struct luaL_reg urllib_m[] = {
 	LUA_INTERFACE_DEF (url, get_fragment),
 	LUA_INTERFACE_DEF (url, get_text),
 	LUA_INTERFACE_DEF (url, get_tld),
+	LUA_INTERFACE_DEF (url, get_raw),
 	LUA_INTERFACE_DEF (url, to_table),
 	LUA_INTERFACE_DEF (url, is_phished),
 	LUA_INTERFACE_DEF (url, is_redirected),
@@ -262,6 +264,26 @@ lua_url_get_text (lua_State *L)
 	return 1;
 }
 
+/***
+ * @method url:get_raw()
+ * Get full content of the url as it was parsed (e.g. with urldecode)
+ * @return {string} url string
+ */
+static gint
+lua_url_get_raw (lua_State *L)
+{
+	struct rspamd_lua_url *url = lua_check_url (L, 1);
+
+	if (url != NULL) {
+		lua_pushlstring (L, url->url->raw, url->url->rawlen);
+	}
+	else {
+		lua_pushnil (L);
+	}
+
+	return 1;
+}
+
 /***
  * @method url:is_phished()
  * Check whether URL is treated as phished
-- 
2.39.5