Fix UTF-8 support in splitCodeIntoArray() (#76)

allejo · web-flow · commit efb6e445494a · 2020-08-26T20:24:44.000-07:00
diff --git a/HighlightUtilities/functions.php b/HighlightUtilities/functions.php
@@ -175,7 +175,8 @@ function splitCodeIntoArray($html)
 
     $dom = new \DOMDocument();
 
-    if (!$dom->loadHTML($html)) {
+    // https://stackoverflow.com/a/8218649
+    if (!$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'))) {
         throw new \UnexpectedValueException("The given HTML could not be parsed correctly.");
     }
 
@@ -186,15 +187,15 @@ function splitCodeIntoArray($html)
         $classes = $span->getAttribute("class");
         $renderedSpan = $dom->saveHTML($span);
 
-        if (preg_match('/\R/', $renderedSpan)) {
+        if (preg_match('/\R/u', $renderedSpan)) {
             $finished = preg_replace(
-                '/\R/',
+                '/\R/u',
                 sprintf('</span>%s<span class="%s">', PHP_EOL, $classes),
                 $renderedSpan
             );
             $html = str_replace($renderedSpan, $finished, $html);
         }
     }
 
-    return preg_split('/\R/', $html);
+    return preg_split('/\R/u', $html);
 }
diff --git a/test/HighlightUtilitiesTest.php b/test/HighlightUtilitiesTest.php
@@ -116,6 +116,26 @@ public function testSplitCodeIntoArray_MultilineComment()
         }
     }
 
+    public function testSplitCodeIntoArray_Emojis()
+    {
+        $raw = <<<'PHP'
+// ✅ ...
+$user = new \stdClass();
+$isUserPending = $user->isStatus('pending');
+PHP;
+        $highlighted = $this->hl->highlight('php', $raw);
+        $split = \HighlightUtilities\splitCodeIntoArray($highlighted->value);
+
+        $this->assertEquals(
+            $split,
+            array(
+                '<span class="hljs-comment">// ✅ ...</span>',
+                '$user = <span class="hljs-keyword">new</span> \stdClass();',
+                '$isUserPending = $user-&gt;isStatus(<span class="hljs-string">\'pending\'</span>);',
+            )
+        );
+    }
+
     public function testGetThemeBackgroundColorSingleColor()
     {
         $theme = 'atom-one-dark';

Original file line number	Diff line number	Diff line change
`@@ -175,7 +175,8 @@ function splitCodeIntoArray($html)`
`175`	`175`
`176`	`176`	`$dom = new \DOMDocument();`
`177`	`177`
`178`		`- if (!$dom->loadHTML($html)) {`
	`178`	`+ // https://stackoverflow.com/a/8218649`
	`179`	`+ if (!$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'))) {`
`179`	`180`	`throw new \UnexpectedValueException("The given HTML could not be parsed correctly.");`
`180`	`181`	`}`
`181`	`182`
`@@ -186,15 +187,15 @@ function splitCodeIntoArray($html)`
`186`	`187`	`$classes = $span->getAttribute("class");`
`187`	`188`	`$renderedSpan = $dom->saveHTML($span);`
`188`	`189`
`189`		`- if (preg_match('/\R/', $renderedSpan)) {`
	`190`	`+ if (preg_match('/\R/u', $renderedSpan)) {`
`190`	`191`	`$finished = preg_replace(`
`191`		`- '/\R/',`
	`192`	`+ '/\R/u',`
`192`	`193`	`sprintf('</span>%s<span class="%s">', PHP_EOL, $classes),`
`193`	`194`	`$renderedSpan`
`194`	`195`	`);`
`195`	`196`	`$html = str_replace($renderedSpan, $finished, $html);`
`196`	`197`	`}`
`197`	`198`	`}`
`198`	`199`
`199`		`- return preg_split('/\R/', $html);`
	`200`	`+ return preg_split('/\R/u', $html);`
`200`	`201`	`}`