Emoji: added support for textual codepoint sequences

s9e · Jan 1, 2018 · efdcb8f · efdcb8f
1 parent f31b098
commit efdcb8f
Show file tree

Hide file tree

Showing 6 changed files with 8,938 additions and 3 deletions.
diff --git a/scripts/patchEmoji.php b/scripts/patchEmoji.php
@@ -83,8 +83,13 @@
 	{
 		echo bin2hex($utf8), " does not contain 0xE2, 0xEF or 0xF0. Parser.php would need to be updated.\n";
 	}
+	$seq = utf8ToSeq($utf8);
+
 	$allText .= $utf8 . "\n";
-	$allXml  .= '<EMOJI seq="' . utf8ToSeq($utf8) . '">' . $utf8 . "</EMOJI>\n";
+	$allXml  .= '<EMOJI seq="' . $seq . '">' . $utf8 . "</EMOJI>\n";
+
+	$allText .= ':' . $seq . ":\n";
+	$allXml  .= '<EMOJI seq="' . $seq . '">:' . $seq . ":</EMOJI>\n";
 }
 $allXml .= '</r>';
 

diff --git a/src/Plugins/Emoji/Parser.js b/src/Plugins/Emoji/Parser.js
diff --git a/src/Plugins/Emoji/Parser.php b/src/Plugins/Emoji/Parser.php
@@ -136,9 +136,14 @@ protected function parseAsciiEmoji($text)
 		foreach ($matches[0] as list($shortName, $tagPos))
 		{
 			$shortName = substr($shortName, 1);
+			$tagLen    = 2 + strlen($shortName);
 			if (isset(self::$map[$shortName]))
 			{
-				$this->addTag($tagPos, 2 + strlen($shortName), self::$map[$shortName]);
+				$this->addTag($tagPos, $tagLen, self::$map[$shortName]);
+			}
+			elseif (preg_match('/^[0-3][0-9a-f]{3,4}(?:-[0-9a-f]{4,5})*$/', $shortName))
+			{
+				$this->addTag($tagPos, $tagLen, preg_replace('/-(?:200d|fe0f)/', '', $shortName));
 			}
 		}
 	}

diff --git a/tests/Plugins/Emoji/ParserTest.php b/tests/Plugins/Emoji/ParserTest.php
@@ -129,6 +129,10 @@ function ($configurator, $plugin)
 				'👩‍❤️‍👨',
 				'<r><EMOJI seq="1f469-2764-1f468">&#128105;‍❤️‍&#128104;</EMOJI></r>'
 			],
+			[
+				':00a9: :1f1ef-1f1f5: :1f468-200d-2764-fe0f-200d-1f468:',
+				'<r><EMOJI seq="00a9">:00a9:</EMOJI> <EMOJI seq="1f1ef-1f1f5">:1f1ef-1f1f5:</EMOJI> <EMOJI seq="1f468-2764-1f468">:1f468-200d-2764-fe0f-200d-1f468:</EMOJI></r>'
+			],
 			[
 				file_get_contents(__DIR__ . '/all.txt'),
 				file_get_contents(__DIR__ . '/all.xml'),