Merge pull request #13384 from annando/smilies

Posts without text or only with emojis are now always accepted in the language check

src/Content/Smilies.php

@@ -285,4 +285,33 @@ class Smilies
 
 		return str_replace($matches[0], $t, $matches[0]);
 	}
+
+	/**
+	 * Checks if the body doesn't contain any alphanumeric characters
+	 *
+	 * @param string $body Possibly-HTML post body
+	 * @return boolean
+	 */
+	public static function isEmojiPost(string $body): bool
+	{
+		// Strips all whitespace
+		$conv = preg_replace('#\s#u', '', html_entity_decode($body));
+		if (empty($conv)) {
+			return false;
+		}
+
+		if (!class_exists('IntlChar')) {
+			// Most Emojis are 4 byte Unicode characters, so this is a good workaround, when IntlChar does not exist on the system
+			return strlen($conv) / mb_strlen($conv) == 4;
+		}
+
+		for ($i = 0; $i < mb_strlen($conv); $i++) {
+			$character = mb_substr($conv, $i, 1);
+
+			if (\IntlChar::isalnum($character) || \IntlChar::ispunct($character) || \IntlChar::isgraph($character) && (strlen($character) <= 2)) {
+				return false;
+			}
+		}
+		return true;
+	}
 }

src/Content/Text/BBCode.php

@@ -1791,13 +1791,9 @@ class BBCode
 					$text = preg_replace("/\[event\-id\](.*?)\[\/event\-id\]/ism", '', $text);
 				}
 
-				if (!$for_plaintext && DI::config()->get('system', 'big_emojis') && ($simple_html != self::DIASPORA)) {
-					$conv = html_entity_decode(str_replace([' ', "\n", "\r"], '', $text));
-					// Emojis are always 4 byte Unicode characters
-					if (!empty($conv) && (strlen($conv) / mb_strlen($conv) == 4)) {
+				if (!$for_plaintext && DI::config()->get('system', 'big_emojis') && ($simple_html != self::DIASPORA) && Smilies::isEmojiPost($text)) {
 					$text = '<span style="font-size: xx-large; line-height: normal;">' . $text . '</span>';
 				}
-				}
 
 				// Handle mentions and hashtag links
 				if ($simple_html == self::DIASPORA) {

src/Protocol/Relay.php

@@ -21,6 +21,7 @@
 
 namespace Friendica\Protocol;
 
+use Friendica\Content\Smilies;
 use Friendica\Content\Text\BBCode;
 use Friendica\Core\Logger;
 use Friendica\Core\Protocol;
@@ -172,6 +173,11 @@ class Relay
 	 */
 	public static function isWantedLanguage(string $body, int $uri_id = 0, int $author_id = 0)
 	{
+		if (empty($body) || Smilies::isEmojiPost($body)) {
+			Logger::debug('Empty body or only emojis', ['body' => $body]);
+			return true;
+		}
+
 		$languages = [];
 		foreach (Item::getLanguageArray($body, 10, $uri_id, $author_id) as $language => $reliability) {
 			if ($reliability > 0) {

tests/src/Content/SmiliesTest.php

@@ -72,4 +72,75 @@ class SmiliesTest extends FixtureTest
 		$output = Smilies::replaceFromArray($text, $smilies);
 		self::assertEquals($expected, $output);
 	}
+
+	public function dataIsEmojiPost(): array
+	{
+		return [
+			'emoji' => [
+				'expected' => true,
+				'body' => '👀',
+			],
+			'emojis' => [
+				'expected' => true,
+				'body' => '👀🤷',
+			],
+			'emoji+whitespace' => [
+				'expected' => true,
+				'body' => ' 👀 ',
+			],
+			'empty' => [
+				'expected' => false,
+				'body' => '',
+			],
+			'whitespace' => [
+				'expected' => false,
+				'body' => '
+				',
+			],
+			'emoji+ASCII' => [
+				'expected' => false,
+				'body' => '🤷a',
+			],
+			'HTML entity whitespace' => [
+				'expected' => false,
+				'body' => ' ',
+			],
+			'HTML entity else' => [
+				'expected' => false,
+				'body' => '°',
+			],
+			'emojis+HTML whitespace' => [
+				'expected' => true,
+				'body' => '👀 🤷',
+			],
+			'emojis+HTML else' => [
+				'expected' => false,
+				'body' => '👀<🤷',
+			],
+			'zwj' => [
+				'expected' => true,
+				'body' => '👨‍👨‍👧‍',
+			],
+			'zwj+whitespace' => [
+				'expected' => true,
+				'body' => ' 👨‍👨‍👧‍ ',
+			],
+			'zwj+HTML whitespace' => [
+				'expected' => true,
+				'body' => ' 👨‍👨‍👧‍ ',
+			],
+		];
+	}
+
+	/**
+	 * @dataProvider dataIsEmojiPost
+	 *
+	 * @param bool   $expected
+	 * @param string $body
+	 * @return void
+	 */
+	public function testIsEmojiPost(bool $expected, string $body)
+	{
+		$this->assertEquals($expected, Smilies::isEmojiPost($body));
+	}
 }