html parser: fix possible endless loop. Fixes matrix chat where parsing non-html as html happened

5 files changed, 60 insertions, 25 deletions

diff --git a/depends/html-parser b/depends/html-parser
-Subproject 684a3bc56d5c40ed3eb54ca263751fa4724e73f
+Subproject 66ec83b862ea2a8dbda1c1f3663af88a8d12d9b
diff --git a/depends/html-search b/depends/html-search
-Subproject 3dbbcc75199bd996163500beb39a6e902bc7ddc
+Subproject 0dd5817eecb3a8db9338394e2c164119a92d4db
diff --git a/plugins/Matrix.hpp b/plugins/Matrix.hpp
index 4afe28e..ac3ca3b 100644
--- a/plugins/Matrix.hpp
+++ b/plugins/Matrix.hpp
@@ -20,7 +20,6 @@ namespace QuickMedia {
 
     static const int AUTHOR_MAX_LENGTH = 48;
 
-    std::string remove_reply_formatting(const std::string &str);
     std::string extract_first_line_remove_newline_elipses(const std::string &str, size_t max_length);
     mgl::Color user_id_to_color(const std::string &user_id);
     std::string formatted_text_to_qm_text(const char *str, size_t size, bool allow_formatted_text);
@@ -103,6 +102,7 @@ namespace QuickMedia {
         RelatedEventType related_event_type = RelatedEventType::NONE;
         bool notification_mentions_me = false;
         bool cache = false;
+        bool body_is_formatted = false;
         std::string transaction_id;
         time_t timestamp = 0; // In milliseconds
         MessageType type;
diff --git a/src/QuickMedia.cpp b/src/QuickMedia.cpp
index 72df6eb..c9c7603 100644
--- a/src/QuickMedia.cpp
+++ b/src/QuickMedia.cpp
@@ -5319,10 +5319,17 @@ namespace QuickMedia {
         return load_cached_related_embedded_item(body_item, message, me.get(), current_room->get_user_display_name(me), me->user_id, message_body_items);
     }
 
+    static std::string message_to_qm_text(Message *message) {
+        if(message->body_is_formatted)
+            return formatted_text_to_qm_text(message->body.c_str(), message->body.size(), true);
+        else
+            return message->body;
+    }
+
     static std::shared_ptr<BodyItem> message_to_body_item(RoomData *room, Message *message, const std::string &my_display_name, const std::string &my_user_id) {
         auto body_item = BodyItem::create("");
         body_item->set_author(extract_first_line_remove_newline_elipses(room->get_user_display_name(message->user), AUTHOR_MAX_LENGTH));
-        body_item->set_description(strip(formatted_text_to_qm_text(message->body.c_str(), message->body.size(), true)));
+        body_item->set_description(strip(message_to_qm_text(message)));
         body_item->set_timestamp(message->timestamp);
         if(!message->thumbnail_url.empty()) {
             body_item->thumbnail_url = message->thumbnail_url;
@@ -5652,7 +5659,7 @@ namespace QuickMedia {
                             // TODO: Properly check reply message objects for mention of user instead of message data, but only when synapse fixes that notifications
                             // are not triggered by reply to a message with our display name/user id.
                             Message *edited_message_ref = static_cast<Message*>(body_item->userdata);
-                            std::string qm_formatted_text = formatted_text_to_qm_text(message->body.c_str(), message->body.size(), true);
+                            std::string qm_formatted_text = message_to_qm_text(message.get());
 
                             body_item->set_description(std::move(qm_formatted_text));
                             if(message->user != me && message_contains_user_mention(message.get(), my_display_name, me->user_id))
@@ -6116,6 +6123,7 @@ namespace QuickMedia {
                  }
 
                 auto message = std::make_shared<Message>();
+                message->body_is_formatted = true;
                 message->user = matrix->get_me(current_room);
                 if(msgtype == "m.emote")
                     message->body = "*" + current_room->get_user_display_name(me) + "* " + matrix->body_to_formatted_body(current_room, text);
diff --git a/src/plugins/Matrix.cpp b/src/plugins/Matrix.cpp
index d1a060d..31aab76 100644
--- a/src/plugins/Matrix.cpp
+++ b/src/plugins/Matrix.cpp
@@ -117,6 +117,34 @@ namespace QuickMedia {
         return colors[color_hash_code(user_id) % num_colors];
     }
 
+    static std::string remove_reply_formatting(const std::string &str) {
+        if(strncmp(str.c_str(), "> <@", 4) == 0) {
+            size_t index = str.find("> ", 4);
+            if(index != std::string::npos) {
+                size_t msg_begin = str.find("\n\n", index + 2);
+                if(msg_begin != std::string::npos)
+                    return str.substr(msg_begin + 2);
+            }
+        } else {
+            return formatted_text_to_qm_text(str.c_str(), str.size(), false);
+        }
+        return str;
+    }
+
+    static std::string remove_reply_formatting(const Message *message) {
+        if(!message->body_is_formatted && strncmp(message->body.c_str(), "> <@", 4) == 0) {
+            size_t index = message->body.find("> ", 4);
+            if(index != std::string::npos) {
+                size_t msg_begin = message->body.find("\n\n", index + 2);
+                if(msg_begin != std::string::npos)
+                    return message->body.substr(msg_begin + 2);
+            }
+            return message->body;
+        } else {
+            return formatted_text_to_qm_text(message->body.c_str(), message->body.size(), false);
+        }
+    }
+
     bool TimestampedDisplayData::set_data_if_newer(std::string new_data, time_t new_timestamp) {
         if(new_timestamp == 0) {
             data = std::move(new_data);
@@ -433,7 +461,7 @@ namespace QuickMedia {
         if(!sync_is_cache && message_dir == MessageDirection::AFTER) {
             for(auto &message : messages) {
                 if(message->notification_mentions_me) {
-                    std::string body = remove_reply_formatting(message->body);
+                    std::string body = remove_reply_formatting(message.get());
                     bool read = true;
                     // TODO: What if the message or username begins with "-"? also make the notification image be the avatar of the user
                     if((!is_window_focused || room != current_room) && message->related_event_type != RelatedEventType::EDIT && message->related_event_type != RelatedEventType::REDACTION) {
@@ -598,8 +626,15 @@ namespace QuickMedia {
         return nullptr;
     }
 
+    static std::string message_to_qm_text(const Message *message, bool allow_formatted_text = true) {
+        if(message->body_is_formatted)
+            return formatted_text_to_qm_text(message->body.c_str(), message->body.size(), allow_formatted_text);
+        else
+            return message->body;
+    }
+
     static std::string message_to_room_description_text(Message *message) {
-        std::string body = strip(formatted_text_to_qm_text(message->body.c_str(), message->body.size(), true));
+        std::string body = strip(message_to_qm_text(message));
         if(message->type == MessageType::REACTION)
             return "Reacted with: " + body;
         else if(message->related_event_type == RelatedEventType::REPLY)
@@ -2130,7 +2165,7 @@ namespace QuickMedia {
     }
 
     bool message_contains_user_mention(const Message *message, const std::string &username, const std::string &user_id) {
-        const std::string formatted_text = formatted_text_to_qm_text(message->body.c_str(), message->body.size(), false);
+        const std::string formatted_text = message_to_qm_text(message, false);
         return message_contains_user_mention(formatted_text, username) || message_contains_user_mention(formatted_text, user_id);
     }
 
@@ -2197,7 +2232,7 @@ namespace QuickMedia {
                 // TODO: Is @room ok? shouldn't we also check if the user has permission to do @room? (only when notifications are limited to @mentions)
                 // TODO: Is comparing against read marker timestamp ok enough?
                 if(me && message->timestamp > read_marker_message_timestamp) {
-                    std::string message_str = formatted_text_to_qm_text(message->body.c_str(), message->body.size(), false);
+                    std::string message_str = message_to_qm_text(message.get(), false);
                     message->notification_mentions_me = message_contains_user_mention(message_str, my_display_name) || message_contains_user_mention(message_str, me->user_id) || message_contains_user_mention(message_str, "@room");
                 }
             }
@@ -2370,8 +2405,10 @@ namespace QuickMedia {
                                 code_parse_userdata.allow_formatted_text = true;
                                 code_parse_userdata.inside_source_highlight = true;
                                 html_parser_parse(output.c_str(), output.size(), formattext_text_parser_callback, &code_parse_userdata);
-                                text_to_add = std::move(code_parse_userdata.result);
-                                formatted_text_flags = FORMATTED_TEXT_FLAG_NONE;
+                                if(!code_parse_userdata.result.empty()) {
+                                    text_to_add = std::move(code_parse_userdata.result);
+                                    formatted_text_flags = FORMATTED_TEXT_FLAG_NONE;
+                                }
                             }
                         }
                     }
@@ -2775,9 +2812,12 @@ namespace QuickMedia {
         if(!body_json.IsString())
             return nullptr;
 
+        bool body_is_formatted = true;
         const rapidjson::Value *formatted_body_json = &GetMember(*content_json, "formatted_body");
-        if(!formatted_body_json->IsString())
+        if(!formatted_body_json->IsString()) {
             formatted_body_json = &body_json;
+            body_is_formatted = false;
+        }
 
         auto message = std::make_shared<Message>();
         std::string prefix;
@@ -2846,6 +2886,7 @@ namespace QuickMedia {
         message->user = user;
         message->event_id = event_id_str;
         message->body = prefix + std::string(formatted_body_json->GetString(), formatted_body_json->GetStringLength());
+        message->body_is_formatted = body_is_formatted;
         message->related_event_id = std::move(related_event_id);
         message->related_event_type = related_event_type;
         message->timestamp = timestamp;
@@ -3567,20 +3608,6 @@ namespace QuickMedia {
         return PluginResult::OK;
     }
 
-    std::string remove_reply_formatting(const std::string &str) {
-        if(strncmp(str.c_str(), "> <@", 4) == 0) {
-            size_t index = str.find("> ", 4);
-            if(index != std::string::npos) {
-                size_t msg_begin = str.find("\n\n", index + 2);
-                if(msg_begin != std::string::npos)
-                    return str.substr(msg_begin + 2);
-            }
-        } else {
-            return formatted_text_to_qm_text(str.c_str(), str.size(), false);
-        }
-        return str;
-    }
-
     static std::string block_quote(const std::string &str) {
         std::string result;
         for(char c : str) {
@@ -3600,7 +3627,7 @@ namespace QuickMedia {
         switch(message->type) {
             case MessageType::TEXT: {
                 if(message->related_event_type != RelatedEventType::NONE)
-                    related_to_body = remove_reply_formatting(message->body);
+                    related_to_body = remove_reply_formatting(message);
                 else
                     related_to_body = message->body;
                 break;