Implement emoji parsing

Fix Chatterino#60

src/emojis.hpp

@@ -13,7 +13,12 @@ namespace chatterino {
 
 struct EmojiData {
     QString value;
+
+    // what's used in the emoji-one url
     QString code;
+
+    // i.e. thinking
+    QString shortCode;
 };
 
 class Emojis

src/emotemanager.cpp

@@ -188,74 +188,102 @@ void EmoteManager::loadEmojis()
         EmojiData emojiData{
             QString::fromUcs4(unicodeBytes, numUnicodeBytes),  //
             code,                                              //
+            shortCode,                                         //
         };
 
-        shortCodeToEmoji.insert(shortCode, emojiData);
-        emojiToShortCode.insert(emojiData.value, shortCode);
-    }
-
-    /*
-    for (auto const &emoji : shortCodeToEmoji.toStdMap()) {
-        auto iter = firstEmojiChars.find(emoji.first.at(0));
-
-        if (iter != firstEmojiChars.end()) {
-            iter.value().insert(emoji.second.value, emoji.second.value);
-            continue;
-        }
+        this->emojiShortCodeToEmoji.insert(shortCode, emojiData);
 
-        firstEmojiChars.insert(emoji.first.at(0),
-                               QMap<QString, QString>{{emoji.second.value, emoji.second.code}});
+        this->emojiFirstByte[emojiData.value.at(0)].append(emojiData);
     }
-    */
 }
 
 void EmoteManager::parseEmojis(
-    std::vector<std::tuple<messages::LazyLoadedImage *, QString>> &vector, const QString &text)
+    std::vector<std::tuple<messages::LazyLoadedImage *, QString>> &parsedWords, const QString &text)
 {
-    // TODO(pajlada): Add this method to EmoteManager instead
-    long lastSlice = 0;
+    int lastParsedEmojiEndIndex = 0;
 
     for (auto i = 0; i < text.length() - 1; i++) {
-        if (!text.at(i).isLowSurrogate()) {
-            auto iter = firstEmojiChars.find(text.at(i));
-
-            if (iter != firstEmojiChars.end()) {
-                for (auto j = std::min(8, text.length() - i); j > 0; j--) {
-                    QString emojiString = text.mid(i, 2);
-                    auto emojiIter = iter.value().find(emojiString);
-
-                    if (emojiIter != iter.value().end()) {
-                        QString url = "https://cdnjs.cloudflare.com/ajax/libs/"
-                                      "emojione/2.2.6/assets/png/" +
-                                      emojiIter.value() + ".png";
-
-                        if (i - lastSlice != 0) {
-                            vector.push_back(std::tuple<messages::LazyLoadedImage *, QString>(
-                                nullptr, text.mid(lastSlice, i - lastSlice)));
-                        }
-
-                        vector.push_back(std::tuple<messages::LazyLoadedImage *, QString>(
-                            emojis.getOrAdd(url,
-                                            [this, &url] {
-                                                return new LazyLoadedImage(
-                                                    *this, this->windowManager, url, 0.35);  //
-                                            }),
-                            QString()));
-
-                        i += j - 1;
-
-                        lastSlice = i + 1;
-
-                        break;
-                    }
+        const QChar character = text.at(i);
+
+        if (character.isLowSurrogate()) {
+            continue;
+        }
+
+        auto it = this->emojiFirstByte.find(character);
+        if (it == this->emojiFirstByte.end()) {
+            // No emoji starts with this character
+            continue;
+        }
+
+        const QVector<EmojiData> possibleEmojis = it.value();
+
+        int remainingCharacters = text.length() - i;
+
+        EmojiData matchedEmoji;
+
+        int matchedEmojiLength = 0;
+
+        for (const EmojiData &emoji : possibleEmojis) {
+            if (remainingCharacters < emoji.value.length()) {
+                // It cannot be this emoji, there's not enough space for it
+                continue;
+            }
+
+            bool match = true;
+
+            for (int j = 1; j < emoji.value.length(); ++j) {
+                if (text.at(i + j) != emoji.value.at(j)) {
+                    match = false;
+
+                    break;
                 }
             }
+
+            if (match) {
+                matchedEmoji = emoji;
+                matchedEmojiLength = emoji.value.length();
+
+                break;
+            }
+        }
+
+        if (matchedEmojiLength == 0) {
+            continue;
+        }
+
+        int currentParsedEmojiFirstIndex = i;
+        int currentParsedEmojiEndIndex = i + (matchedEmojiLength);
+
+        int charactersFromLastParsedEmoji = currentParsedEmojiFirstIndex - lastParsedEmojiEndIndex;
+
+        if (charactersFromLastParsedEmoji > 0) {
+            // Add characters inbetween emojis
+            parsedWords.push_back(std::tuple<messages::LazyLoadedImage *, QString>(
+                nullptr, text.mid(lastParsedEmojiEndIndex, charactersFromLastParsedEmoji)));
         }
+
+        QString url = "https://cdnjs.cloudflare.com/ajax/libs/"
+                      "emojione/2.2.6/assets/png/" +
+                      matchedEmoji.code + ".png";
+
+        // Create or fetch cached emoji image
+        auto emojiImage = this->emojiCache.getOrAdd(url, [this, &url] {
+            return new LazyLoadedImage(*this, this->windowManager, url, 0.35);  //
+        });
+
+        // Push the emoji as a word to parsedWords
+        parsedWords.push_back(
+            std::tuple<messages::LazyLoadedImage *, QString>(emojiImage, QString()));
+
+        lastParsedEmojiEndIndex = currentParsedEmojiEndIndex;
+
+        i += matchedEmojiLength - 1;
     }
 
-    if (lastSlice < text.length()) {
-        vector.push_back(
-            std::tuple<messages::LazyLoadedImage *, QString>(nullptr, text.mid(lastSlice)));
+    if (lastParsedEmojiEndIndex < text.length()) {
+        // Add remaining characters
+        parsedWords.push_back(std::tuple<messages::LazyLoadedImage *, QString>(
+            nullptr, text.mid(lastParsedEmojiEndIndex)));
     }
 }
 

src/emotemanager.hpp

@@ -62,44 +62,42 @@ class EmoteManager
     WindowManager &windowManager;
     Resources &resources;
 
-    // Emojis
-    // shortCodeToEmoji maps strings like ":sunglasses:" to the unicode character
-    QMap<QString, EmojiData> shortCodeToEmoji;
+    /// Emojis
+    // shortCodeToEmoji maps strings like "sunglasses" to its emoji
+    QMap<QString, EmojiData> emojiShortCodeToEmoji;
 
-    // emojiToShortCode maps the unicode character to the shortcode like ":sunglasses:"
-    QMap<QString, QString> emojiToShortCode;
+    // Maps the first character of the emoji unicode string to a vector of possible emojis
+    QMap<QChar, QVector<EmojiData>> emojiFirstByte;
 
-    // TODO(pajlada): Figure out what this is for
-    QMap<QChar, QMap<QString, QString>> firstEmojiChars;
-
-    ConcurrentMap<QString, messages::LazyLoadedImage *> emojis;
+    //            url      Emoji-one image
+    ConcurrentMap<QString, messages::LazyLoadedImage *> emojiCache;
 
     void loadEmojis();
 
 public:
-    void parseEmojis(std::vector<std::tuple<messages::LazyLoadedImage *, QString>> &vector,
+    void parseEmojis(std::vector<std::tuple<messages::LazyLoadedImage *, QString>> &parsedWords,
                      const QString &text);
 
 private:
-    // Twitch emotes
+    /// Twitch emotes
     ConcurrentMap<QString, twitch::EmoteValue *> _twitchEmotes;
     ConcurrentMap<long, messages::LazyLoadedImage *> _twitchEmoteFromCache;
 
-    // BTTV emotes
+    /// BTTV emotes
     ConcurrentMap<QString, messages::LazyLoadedImage *> bttvChannelEmotes;
     ConcurrentMap<QString, messages::LazyLoadedImage *> _bttvEmotes;
     ConcurrentMap<QString, messages::LazyLoadedImage *> _bttvChannelEmoteFromCaches;
 
     void loadBTTVEmotes();
 
-    // FFZ emotes
+    /// FFZ emotes
     ConcurrentMap<QString, messages::LazyLoadedImage *> ffzChannelEmotes;
     ConcurrentMap<QString, messages::LazyLoadedImage *> _ffzEmotes;
     ConcurrentMap<int, messages::LazyLoadedImage *> _ffzChannelEmoteFromCaches;
 
     void loadFFZEmotes();
 
-    // Chatterino emotes
+    /// Chatterino emotes
     ConcurrentMap<QString, messages::LazyLoadedImage *> _chatterinoEmotes;
 
     // ???

src/twitch/twitchmessagebuilder.cpp

@@ -1,6 +1,5 @@
 #include "twitch/twitchmessagebuilder.hpp"
 #include "colorscheme.hpp"
-#include "emojis.hpp"
 #include "emotemanager.hpp"
 #include "ircmanager.hpp"
 #include "resources.hpp"
@@ -190,6 +189,7 @@ SharedMessage TwitchMessageBuilder::parse(const Communi::IrcPrivateMessage *ircM
         // split words
         std::vector<std::tuple<LazyLoadedImage *, QString>> parsed;
 
+        // Parse emojis and take all non-emojis and put them in parsed as full text-words
         emoteManager.parseEmojis(parsed, split);
 
         for (const std::tuple<LazyLoadedImage *, QString> &tuple : parsed) {