X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=application%2Fbookmark%2FLinkUtils.php;h=faf5dbfd4fe24906bf980d8f4cc72e0472b7e008;hb=47d15818502b11315b484131aaf67a3e3f441c5a;hp=68914fcab749a19b1ba15193decd99247158375a;hpb=af41d5ab5d2bd3ba64d052c997bc6afa6966a63c;p=github%2Fshaarli%2FShaarli.git

diff --git a/application/bookmark/LinkUtils.php b/application/bookmark/LinkUtils.php
index 68914fca..faf5dbfd 100644
--- a/application/bookmark/LinkUtils.php
+++ b/application/bookmark/LinkUtils.php
@@ -26,7 +26,7 @@ function html_extract_title($html)
  */
 function header_extract_charset($header)
 {
-    preg_match('/charset="?([^; ]+)/i', $header, $match);
+    preg_match('/charset=["\']?([^; "\']+)/i', $header, $match);
     if (! empty($match[1])) {
         return strtolower(trim($match[1]));
     }
@@ -66,11 +66,13 @@ function html_extract_tag($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 . '[^\'"]*?[\'"]';
     // Try to retrieve OpenGraph image.
-    $ogRegex = '#<meta[^>]+(?:'. $properties .')=["\']?(?:og:)?'. $tag .'["\'\s][^>]*content=["\']?(.*?)["\'/>]#';
+    $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=["\']([^"\']+)[^>]+(?:'. $properties .')=["\']?(?:og)?:'. $tag .'["\'\s/>]#';
+    $ogRegexReverse = '#<meta[^>]+content=["\'](.*?)["\'][^>]+(?:'. $properties .')=(?:'. $orCondition .').*?>#';
 
     if (preg_match($ogRegex, $html, $matches) > 0
         || preg_match($ogRegexReverse, $html, $matches) > 0