Merge pull request #9 from tobyzerner/RegexpBuilder

Normalize codepoints for Twemoji

Author: Gummibeer

generate.php

@@ -8,7 +8,14 @@
 
 include __DIR__.'/vendor/autoload.php';
 
-$emoji = Spatie\Emoji\Emoji::all();
+// Normalize codepoints for Twemoji
+$emoji = array_map(function ($emoji) {
+    if (strpos($emoji, "\u{200D}") === false) {
+        $emoji = str_replace("\u{FE0F}", '', $emoji);
+    }
+
+    return $emoji;
+}, Spatie\Emoji\Emoji::all());
 
 // Work on bytes, output a PCRE regexp made of ASCII characters
 $builder = new s9e\RegexpBuilder\Builder([

src/Twemoji.php

@@ -16,7 +16,7 @@ class Twemoji implements JsonSerializable
     protected array $codepoints;
 
     /**
-     * @param string[] $codepoints
+     * @param  string[]  $codepoints
      */
     public function __construct(array $codepoints)
     {
@@ -32,9 +32,14 @@ public static function emoji(string $emoji): self
             $chars
         );
 
-        $normalized = array_diff($codepoints, ['fe0f']);
+        // Normalize codepoints for Twemoji
+        $codepoints[0] = ltrim($codepoints[0], '0');
 
-        return new static($normalized);
+        if (! in_array('200d', $codepoints)) {
+            $codepoints = array_diff($codepoints, ['fe0f']);
+        }
+
+        return new static($codepoints);
     }
 
     public static function text(string $text): EmojiText