Implement lua_html methods.

diff --git a/doc/Makefile b/doc/Makefile
index 729780fddfdacd93618e9b7d8566a0ec3df64a8b..91ada9117243f967345be5576b0f5b3afac4b24c 100644 (file)
--- a/doc/Makefile
+++ b/doc/Makefile
@@ -14,7 +14,7 @@ rspamc.1: rspamc.1.md
        
 lua-doc: lua_regexp lua_ip lua_config lua_task lua_ucl lua_http lua_trie \
        lua_dns lua_redis lua_upstream lua_expression lua_mimepart lua_logger lua_url \
-       lua_tcp lua_mempool
+       lua_tcp lua_mempool lua_html
 
 lua_regexp: ../src/lua/lua_regexp.c
        $(LUADOC) < ../src/lua/lua_regexp.c > markdown/lua/regexp.md
@@ -47,4 +47,6 @@ lua_url: ../src/lua/lua_url.c
 lua_tcp: ../src/lua/lua_tcp.c
        $(LUADOC) < ../src/lua/lua_tcp.c > markdown/lua/tcp.md
 lua_mempool: ../src/lua/lua_mempool.c
-       $(LUADOC) < ../src/lua/lua_mempool.c > markdown/lua/mempool.md  
\ No newline at end of file
+       $(LUADOC) < ../src/lua/lua_mempool.c > markdown/lua/mempool.md
+lua_html: ../src/lua/lua_html.c
+       $(LUADOC) < ../src/lua/lua_html.c > markdown/lua/html.md
\ No newline at end of file

diff --git a/src/lua/lua_html.c b/src/lua/lua_html.c
index 94bd7b9824d3e811ebdea76143ce9b5722756a9f..79abc0a3979bcbcf7ec49a55ec5fc8f2dd55e270 100644 (file)
--- a/src/lua/lua_html.c
+++ b/src/lua/lua_html.c
@@ -47,11 +47,158 @@ rspamd_config.R_HTML_IMAGE = function (task)
 end
  */
 
+/***
+ * @method html:has_tag(name)
+ * Checks if a specified tag `name` is presented in a part
+ * @param {string} name name of tag to check
+ * @return {boolean} `true` if the tag exists in HTML tree
+ */
+LUA_FUNCTION_DEF (html, has_tag);
+
+/***
+ * @method html:check_property(name)
+ * Checks if the HTML has a specific property. Here is the list of available properties:
+ *
+ * - `no_html` - no html tag presented
+ * - `bad_element` - part has some broken elements
+ * - `xml` - part is xhtml
+ * - `unknown_element` - part has some unknown elements
+ * - `duplicate_element` - part has some duplicate elements that should be unique (namely, `title` tag)
+ * - `unbalanced` - part has unbalanced tags
+ * @param {string} name name of property
+ * @return {boolean} true if the part has the specified property
+ */
+LUA_FUNCTION_DEF (html, has_property);
+
+/***
+ * @method html:get_images()
+ * Returns table of images found in html. Each image is, in turn, a table with the following fields:
+ *
+ * - `src` - link to the source
+ * - `height` - height in pixels
+ * - `width` - width in pixels
+ * - `embeded` - `true` if an image is embedded in a message
+ * @return {table} table of images in html part
+ */
+LUA_FUNCTION_DEF (html, get_images);
+
 static const struct luaL_reg htmllib_m[] = {
+       LUA_INTERFACE_DEF (html, has_tag),
+       LUA_INTERFACE_DEF (html, has_property),
+       LUA_INTERFACE_DEF (html, get_images),
        {"__tostring", rspamd_lua_class_tostring},
        {NULL, NULL}
 };
 
+static struct html_content *
+lua_check_html (lua_State * L, gint pos)
+{
+       void *ud = luaL_checkudata (L, pos, "rspamd{html}");
+       luaL_argcheck (L, ud != NULL, pos, "'html' expected");
+       return ud ? *((struct html_content **)ud) : NULL;
+}
+
+static gint
+lua_html_has_tag (lua_State *L)
+{
+       struct html_content *hc = lua_check_html (L, 1);
+       const gchar *tagname = luaL_checkstring (L, 2);
+       gboolean ret = FALSE;
+
+       if (hc && tagname) {
+               if (rspamd_html_tag_seen (hc, tagname)) {
+                       ret = TRUE;
+               }
+       }
+
+       lua_pushboolean (L, ret);
+
+       return 1;
+}
+
+static gint
+lua_html_has_property (lua_State *L)
+{
+       struct html_content *hc = lua_check_html (L, 1);
+       const gchar *propname = luaL_checkstring (L, 2);
+       gboolean ret = FALSE;
+
+       if (hc && propname) {
+               /*
+                * - `no_html`
+                * - `bad_element`
+                * - `xml`
+                * - `unknown_element`
+                * - `duplicate_element`
+                * - `unbalanced`
+                */
+               if (strcmp (propname, "no_html") == 0) {
+                       ret = hc->flags & RSPAMD_HTML_FLAG_BAD_START;
+               }
+               else if (strcmp (propname, "bad_element") == 0) {
+                       ret = hc->flags & RSPAMD_HTML_FLAG_BAD_ELEMENTS;
+               }
+               else if (strcmp (propname, "xml") == 0) {
+                       ret = hc->flags & RSPAMD_HTML_FLAG_XML;
+               }
+               else if (strcmp (propname, "unknown_element") == 0) {
+                       ret = hc->flags & RSPAMD_HTML_FLAG_UNKNOWN_ELEMENTS;
+               }
+               else if (strcmp (propname, "duplicate_element") == 0) {
+                       ret = hc->flags & RSPAMD_HTML_FLAG_DUPLICATE_ELEMENTS;
+               }
+               else if (strcmp (propname, "unbalanced") == 0) {
+                       ret = hc->flags & RSPAMD_HTML_FLAG_UNBALANCED;
+               }
+       }
+
+       lua_pushboolean (L, ret);
+
+       return 1;
+}
+
+static gint
+lua_html_get_images (lua_State *L)
+{
+       struct html_content *hc = lua_check_html (L, 1);
+       struct html_image *img;
+       guint i;
+
+       if (hc != NULL) {
+               lua_newtable (L);
+
+               if (hc->images) {
+                       for (i = 0; i < hc->images->len; i ++) {
+                               img = g_ptr_array_index (hc->images, i);
+
+                               lua_newtable (L);
+
+                               if (img->src) {
+                                       lua_pushstring (L, "src");
+                                       lua_pushstring (L, img->src);
+                                       lua_settable (L, -3);
+                               }
+
+                               lua_pushstring (L, "height");
+                               lua_pushnumber (L, img->height);
+                               lua_settable (L, -3);
+                               lua_pushstring (L, "width");
+                               lua_pushnumber (L, img->width);
+                               lua_settable (L, -3);
+                               lua_pushstring (L, "embedded");
+                               lua_pushboolean (L, img->flags & RSPAMD_HTML_FLAG_IMAGE_EMBEDDED);
+                               lua_settable (L, -3);
+
+                               lua_rawseti (L, -2, i + 1);
+                       }
+               }
+       }
+       else {
+               lua_pushnil (L);
+       }
+
+       return 1;
+}
 
 void
 luaopen_html (lua_State * L)