[Emoji] More ordering and minor fixes

src/Symfony/Component/Emoji/EmojiTransliterator.php

@@ -119,7 +119,7 @@ public function transliterate(string $string, int $start = 0, int $end = -1): st
         }
 
         // Here we rely on intl to validate the $string, $start and $end arguments
-        // and to slice the string. Slicing is done by replacing the part if $string
+        // and to slice the string. Slicing is done by replacing the part of $string
         // between $start and $end by a unique cookie that can be reliably used to
         // identify which part of $string should be transliterated.
 

src/Symfony/Component/Emoji/Resources/bin/build.php

@@ -98,9 +98,6 @@ public static function buildRules(array $emojisCodePoints): Generator
                     ];
                     continue;
                 }
-                if (!self::testEmoji($emoji, $locale, $emojiCodePoints)) {
-                    continue;
-                }
                 $codePointsCount = mb_strlen($emoji);
                 $mapsByLocale[$locale][$codePointsCount][$emoji] = $name;
             }
@@ -131,105 +128,73 @@ public static function buildRules(array $emojisCodePoints): Generator
     public static function buildGitHubRules(array $emojisCodePoints): iterable
     {
         $emojis = json_decode((new Filesystem())->readFile(__DIR__.'/vendor/github-emojis.json'), true, flags: JSON_THROW_ON_ERROR);
-
-        $ignored = [];
         $maps = [];
 
         foreach ($emojis as $shortCode => $url) {
             $emojiCodePoints = str_replace('-', ' ', strtolower(basename(parse_url($url, \PHP_URL_PATH), '.png')));
+
             if (!array_key_exists($emojiCodePoints, $emojisCodePoints)) {
-                $ignored[] = [
-                    'emojiCodePoints' => $emojiCodePoints,
-                    'shortCode' => $shortCode,
-                ];
                 continue;
             }
             $emoji = $emojisCodePoints[$emojiCodePoints];
-            if (!self::testEmoji($emoji, 'github', $emojiCodePoints)) {
-                continue;
-            }
-            $codePointsCount = mb_strlen($emoji);
-            $maps[$codePointsCount][$emoji] = ":$shortCode:";
+            $maps[mb_strlen($emoji)][":$shortCode:"] = $emoji;
         }
 
-        $maps = self::createRules($maps);
+        [$map, $reverse] = self::createRules($maps, true);
 
-        return ['emoji-github' => $maps, 'github-emoji' => array_flip($maps)];
+        return ['emoji-github' => $map, 'github-emoji' => $reverse];
     }
 
     public static function buildGitlabRules(array $emojisCodePoints): iterable
     {
         $emojis = json_decode((new Filesystem())->readFile(__DIR__.'/vendor/gitlab-emojis.json'), true, flags: JSON_THROW_ON_ERROR);
-
-        $ignored = [];
         $maps = [];
 
         foreach ($emojis as $emojiItem) {
             $emojiCodePoints = strtolower($emojiItem['unicode']);
+
             if (!array_key_exists($emojiCodePoints, $emojisCodePoints)) {
-                $ignored[] = [
-                    'emojiCodePoints' => $emojiCodePoints,
-                    'name' => $emojiItem['name'],
-                ];
                 continue;
             }
             $emoji = $emojisCodePoints[$emojiCodePoints];
-            if (!self::testEmoji($emoji, 'gitlab', $emojiCodePoints)) {
-                continue;
-            }
-            $codePointsCount = mb_strlen($emoji);
-            $maps[$codePointsCount][$emoji] = $emojiItem['shortname'];
+            $maps[mb_strlen($emoji)][$emojiItem['shortname']] = $emoji;
         }
 
-        $maps = self::createRules($maps);
+        [$map, $reverse] = self::createRules($maps, true);
 
-        return ['emoji-gitlab' => $maps, 'gitlab-emoji' => array_flip($maps)];
+        return ['emoji-gitlab' => $map, 'gitlab-emoji' => $reverse];
     }
 
     public static function buildSlackRules(array $emojisCodePoints): iterable
     {
         $emojis = json_decode((new Filesystem())->readFile(__DIR__.'/vendor/slack-emojis.json'), true, flags: JSON_THROW_ON_ERROR);
-
-        $ignored = [];
-        $emojiSlackMaps = [];
-        $slackEmojiMaps = [];
+        $maps = [];
 
         foreach ($emojis as $data) {
             $emojiCodePoints = str_replace('-', ' ', strtolower($data['unified']));
-            $shortCode = $data['short_name'];
-            $shortCodes = $data['short_names'];
-            $shortCodes = array_map(fn ($v) => ":$v:", $shortCodes);
 
             if (!array_key_exists($emojiCodePoints, $emojisCodePoints)) {
-                $ignored[] = [
-                    'emojiCodePoints' => $emojiCodePoints,
-                    'shortCode' => $shortCode,
-                ];
                 continue;
             }
             $emoji = $emojisCodePoints[$emojiCodePoints];
-            if (!self::testEmoji($emoji, 'slack', $emojiCodePoints)) {
-                continue;
-            }
-            $codePointsCount = mb_strlen($emoji);
-            $emojiSlackMaps[$codePointsCount][$emoji] = ":$shortCode:";
-            foreach ($shortCodes as $short_name) {
-                $slackEmojiMaps[$codePointsCount][$short_name] = $emoji;
+            $codePointsCount = mb_strlen($emoji) << 1;
+            $maps[($codePointsCount << 1) + 1][":{$data['short_name']}:"] = $emoji;
+
+            foreach ($data['short_names'] as $shortName) {
+                $maps[$codePointsCount << 1][":$shortName:"] = $emoji;
             }
         }
 
-        return ['emoji-slack' => self::createRules($emojiSlackMaps), 'slack-emoji' => self::createRules($slackEmojiMaps)];
+        [$map, $reverse] = self::createRules($maps, true);
+
+        return ['emoji-slack' => $map, 'slack-emoji' => $reverse];
     }
 
     public static function buildStripRules(array $emojisCodePoints): iterable
     {
         $maps = [];
-        foreach ($emojisCodePoints as $codePoints => $emoji) {
-            if (!self::testEmoji($emoji, 'strip', $codePoints)) {
-                continue;
-            }
-            $codePointsCount = mb_strlen($emoji);
-            $maps[$codePointsCount][$emoji] = '';
+        foreach ($emojisCodePoints as $emoji) {
+            $maps[mb_strlen($emoji)][$emoji] = '';
         }
 
         return ['emoji-strip' => self::createRules($maps)];
@@ -269,24 +234,27 @@ public static function saveRules(iterable $rulesByLocale): void
         $fs->dumpFile($file, preg_replace('/QUICK_CHECK = .*;/m', "QUICK_CHECK = {$quickCheck};", $fs->readFile($file)));
     }
 
-    private static function testEmoji(string $emoji, string $locale, string $codePoints): bool
+    private static function createRules(array $maps, bool $reverse = false): array
     {
-        if (!Transliterator::createFromRules("\\$emoji > test ;")) {
-            printf('Could not create transliterator for "%s" in "%s" locale. Code Point: "%s". Error: "%s".'."\n", $emoji, $locale, $codePoints, intl_get_error_message());
+        // We must sort the maps by the number of code points, because the order really matters:
+        // 🫶🏼 must be before 🫶
+        krsort($maps);
 
-            return false;
+        if (!$reverse) {
+            return array_merge(...$maps);
         }
 
-        return true;
-    }
+        $emojiText = $textEmoji = [];
 
-    private static function createRules(array $maps): array
-    {
-        // We must sort the maps by the number of code points, because the order really matters:
-        // 🫶🏼 must be before 🫶
-        krsort($maps);
-        $maps = array_merge(...$maps);
+        foreach ($maps as $map) {
+            uksort($map, 'strnatcmp');
+            $textEmoji = $map + $textEmoji;
+
+            $map = array_flip($map);
+            ksort($map);
+            $emojiText += $map;
+        }
 
-        return $maps;
+        return [$emojiText, $textEmoji];
     }
 }