X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=application%2Fbookmark%2FLinkUtils.php;h=0ab2d2138c5a01f620744dfac40e091ddd3aadaa;hb=refs%2Fpull%2F1665%2Fhead;hp=cf97e3b0a904a08d72a5ee7d3467efb1536f5c96;hpb=8affa2243159052675da7e23b737c76f6b43aeed;p=github%2Fshaarli%2FShaarli.git

diff --git a/application/bookmark/LinkUtils.php b/application/bookmark/LinkUtils.php
index cf97e3b0..0ab2d213 100644
--- a/application/bookmark/LinkUtils.php
+++ b/application/bookmark/LinkUtils.php
@@ -67,14 +67,17 @@ function html_extract_tag($tag, $html)
     $propertiesKey = ['property', 'name', 'itemprop'];
     $properties = implode('|', $propertiesKey);
     // We need a OR here to accept either 'property=og:noquote' or 'property="og:unrelated og:my-tag"'
-    $orCondition  = '["\']?(?:og:)?'. $tag .'["\']?|["\'][^\'"]*?(?:og:)?' . $tag . '[^\'"]*?[\'"]';
+    $orCondition  = '["\']?(?:og:)?' . $tag . '["\']?|["\'][^\'"]*?(?:og:)?' . $tag . '[^\'"]*?[\'"]';
+    // Support quotes in double quoted content, and the other way around
+    $content = 'content=(["\'])((?:(?!\1).)*)\1';
     // Try to retrieve OpenGraph tag.
-    $ogRegex = '#<meta[^>]+(?:'. $properties .')=(?:'. $orCondition .')[^>]*content=(["\'])([^\1]*?)\1.*?>#';
+    $ogRegex = '#<meta[^>]+(?:' . $properties . ')=(?:' . $orCondition . ')[^>]*' . $content . '.*?>#';
     // If the attributes are not in the order property => content (e.g. Github)
     // New regex to keep this readable... more or less.
-    $ogRegexReverse = '#<meta[^>]+content=(["\'])([^\1]*?)\1[^>]+(?:'. $properties .')=(?:'. $orCondition .').*?>#';
+    $ogRegexReverse = '#<meta[^>]+' . $content . '[^>]+(?:' . $properties . ')=(?:' . $orCondition . ').*?>#';
 
-    if (preg_match($ogRegex, $html, $matches) > 0
+    if (
+        preg_match($ogRegex, $html, $matches) > 0
         || preg_match($ogRegexReverse, $html, $matches) > 0
     ) {
         return $matches[2];
@@ -116,7 +119,7 @@ function hashtag_autolink($description, $indexUrl = '')
      * \p{Mn} - any non marking space (accents, umlauts, etc)
      */
     $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}]+)/mui';
-    $replacement = '$1<a href="'. $indexUrl .'./add-tag/$2" title="Hashtag $2">#$2</a>';
+    $replacement = '$1<a href="' . $indexUrl . './add-tag/$2" title="Hashtag $2">#$2</a>';
     return preg_replace($regex, $replacement, $description);
 }