Implement lua regexp split, write test case.

4 files changed, 65 insertions, 15 deletions

diff --git a/src/libutil/regexp.c b/src/libutil/regexp.c
index 444496f45..6da0a663e 100644
--- a/src/libutil/regexp.c
+++ b/src/libutil/regexp.c
@@ -488,12 +488,27 @@ rspamd_regexp_cache_create (struct rspamd_regexp_cache *cache,
 	res = rspamd_regexp_new (pattern, flags, err);
 
 	if (res) {
+		REF_RETAIN (res);
 		g_hash_table_insert (cache->tbl, res->id, res);
 	}
 
 	return res;
 }
 
+gboolean
+rspamd_regexp_cache_remove (struct rspamd_regexp_cache *cache,
+		rspamd_regexp_t *re)
+{
+	if (cache == NULL) {
+		cache = global_re_cache;
+	}
+
+	g_assert (cache != NULL);
+	g_assert (re != NULL);
+
+	return g_hash_table_remove (cache->tbl, re->id);
+}
+
 void
 rspamd_regexp_cache_destroy (struct rspamd_regexp_cache *cache)
 {
diff --git a/src/libutil/regexp.h b/src/libutil/regexp.h
index 6e8aa7295..41835af84 100644
--- a/src/libutil/regexp.h
+++ b/src/libutil/regexp.h
@@ -112,6 +112,15 @@ rspamd_regexp_t* rspamd_regexp_cache_create (struct rspamd_regexp_cache *cache,
 		const gchar *flags, GError **err);
 
 /**
+ * Remove regexp from the cache
+ * @param cache regexp cache. if NULL, the superglobal cache is used (*not* thread-safe)
+ * @param re re to remove
+ * @return TRUE if a regexp has been removed
+ */
+gboolean rspamd_regexp_cache_remove (struct rspamd_regexp_cache *cache,
+		rspamd_regexp_t *re);
+
+/**
  * Destroy regexp cache and unref all elements inside it
  * @param cache
  */
diff --git a/src/lua/lua_regexp.c b/src/lua/lua_regexp.c
index a67bed334..6fa0de772 100644
--- a/src/lua/lua_regexp.c
+++ b/src/lua/lua_regexp.c
@@ -322,26 +322,29 @@ lua_regexp_split (lua_State *L)
 {
 	struct rspamd_lua_regexp *re = lua_check_regexp (L);
 	const gchar *data;
-	gchar **parts;
+	gboolean matched = FALSE;
+	gsize len;
+	const gchar *start = NULL, *end = NULL, *old_start;
 	gint i;
 
 	if (re) {
-		data = luaL_checkstring (L, 2);
+		data = luaL_checklstring (L, 2, &len);
 		if (data) {
-			if ((re->re_flags & G_REGEX_RAW) == 0) {
-				/* Validate input */
-				if (!g_utf8_validate (data, -1, NULL)) {
-					lua_pushnil (L);
-					return 1;
+			lua_newtable (L);
+			i = 0;
+			old_start = data;
+			while (rspamd_regexp_search (re->re, data, len, &start, &end, FALSE)) {
+				if (start - old_start > 0) {
+					lua_pushlstring (L, old_start, start - old_start);
+					lua_rawseti (L, -2, ++i);
+					matched = TRUE;
 				}
+				old_start = end;
 			}
-			parts = g_regex_split (re->re, data, 0);
-			lua_newtable (L);
-			for (i = 1; parts[i - 1] != NULL; i++) {
-				lua_pushstring (L, parts[i - 1]);
-				lua_rawseti (L, -2, i);
+			if (!matched) {
+				lua_pop (L, 1);
+				lua_pushnil (L);
 			}
-			g_strfreev (parts);
 			return 1;
 		}
 	}
@@ -362,7 +365,7 @@ lua_regexp_destroy (lua_State *L)
 	struct rspamd_lua_regexp *to_del = lua_check_regexp (L);
 
 	if (to_del) {
-		re_cache_del (to_del->re_pattern, regexp_static_pool);
+		rspamd_regexp_cache_remove (NULL, to_del->re);
 		rspamd_regexp_unref (to_del->re);
 		g_slice_free1 (sizeof (struct rspamd_lua_regexp), to_del);
 	}
diff --git a/test/lua/unit/regxep.lua b/test/lua/unit/regxep.lua
index 81234ddb6..1277cce11 100644
--- a/test/lua/unit/regxep.lua
+++ b/test/lua/unit/regxep.lua
@@ -19,12 +19,35 @@ context("Regexp unit tests", function()
     
     for _,c in ipairs(cases) do
       local r = re.create_cached(c[1])
-      assert_not_nil(r)
+      assert_not_nil(r, "cannot parse " .. c[1])
       local res = r:match(c[2])
       
       assert_equal(res, c[3], string.format("'%s' doesn't match with '%s'",
         c[2], c[1]))
     end
   end)
+  
+  test("Regexp split", function()
+    local cases = {
+      {'\\s', 'one two', {'one', 'two'}}, -- trivial
+      {'\\s', 'one   two', {'one', 'two'}}, -- multiple delimiters
+      {'\\s', '  one   two  ', {'one', 'two'}}, -- multiple delimiters
+      {'\\s', '  one   ', {'one', 'two'}}, -- multiple delimiters
+      {'[:,]', ',,,:::one,two,,', {'one', 'two'}}, -- multiple delimiters
+    }
+  
+    for _,c in ipairs(cases) do
+      local r = re.create_cached(c[1])
+      assert_not_nil(r, "cannot parse " .. c[1])
+      
+      local res = r:split(c[2])
+      assert_not_nil(res, "cannot split " .. c[2])
+      
+      for i,r in ipairs(res) do
+        assert_equal(r, c[3][i])
+      end
+    end
+  end)
+  
   end
 )
\ No newline at end of file