Merge pull request #1631 from ArthurHoaro/fix/html-extract-quote-fix

[github/shaarli/Shaarli.git] / application / bookmark / LinkUtils.php

diff --git a/application/bookmark/LinkUtils.php b/application/bookmark/LinkUtils.php
index e7af4d552b409b77595985cc635273a6a1d8e448..a74fda5795c09e9f0fd6b3ed1e8d98b81226bc5d 100644 (file)
--- a/application/bookmark/LinkUtils.php
+++ b/application/bookmark/LinkUtils.php
@@ -66,16 +66,18 @@ function html_extract_tag($tag, $html)
 {
     $propertiesKey = ['property', 'name', 'itemprop'];
     $properties = implode('|', $propertiesKey);
-    // Try to retrieve OpenGraph image.
-    $ogRegex = '#<meta[^>]+(?:'. $properties .')=["\']?(?:og:)?'. $tag .'["\'\s][^>]*content=["\']?(.*?)["\'/>]#';
+    // 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 tag.
+    $ogRegex = '#<meta[^>]+(?:'. $properties .')=(?:'. $orCondition .')[^>]*content=(["\'])([^\1]*?)\1.*?>#';
     // 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=(["\'])([^\1]*?)\1[^>]+(?:'. $properties .')=(?:'. $orCondition .').*?>#';
 
     if (preg_match($ogRegex, $html, $matches) > 0
         || preg_match($ogRegexReverse, $html, $matches) > 0
     ) {
-        return $matches[1];
+        return $matches[2];
     }
 
     return false;
@@ -136,12 +138,17 @@ function space2nbsp($text)
  *
  * @param string $description shaare's description.
  * @param string $indexUrl    URL to Shaarli's index.
-
+ * @param bool   $autolink    Turn on/off automatic linkifications of URLs and hashtags
+ *
  * @return string formatted description.
  */
-function format_description($description, $indexUrl = '')
+function format_description($description, $indexUrl = '', $autolink = true)
 {
-    return nl2br(space2nbsp(hashtag_autolink(text2clickable($description), $indexUrl)));
+    if ($autolink) {
+        $description = hashtag_autolink(text2clickable($description), $indexUrl);
+    }
+
+    return nl2br(space2nbsp($description));
 }
 
 /**