From 00d3dd91ef42df13eeafbcc54dcebe3238e322c6 Mon Sep 17 00:00:00 2001
From: ArthurHoaro <arthur@hoa.ro>
Date: Sun, 8 Nov 2020 13:54:39 +0100
Subject: Fix an issue truncating extracted metadata content

Previous regex forced the selection to stop at either the first single or double quote found, regardless of the opening quote. Using '\1', we're sure to wait for the proper quote before stopping the capture.
---
 tests/bookmark/LinkUtilsTest.php | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)

(limited to 'tests')

diff --git a/tests/bookmark/LinkUtilsTest.php b/tests/bookmark/LinkUtilsTest.php
index 3321242f..9bddf84b 100644
--- a/tests/bookmark/LinkUtilsTest.php
+++ b/tests/bookmark/LinkUtilsTest.php
@@ -168,6 +168,36 @@ class LinkUtilsTest extends TestCase
         $this->assertEquals($description, html_extract_tag('description', $html));
     }
 
+    /**
+     * Test html_extract_tag() with double quoted content containing single quote, and the opposite.
+     */
+    public function testHtmlExtractExistentNameTagWithMixedQuotes(): void
+    {
+        $description = 'Bob and Alice share M&M\'s.';
+
+        $html = '<meta property="og:description" content="' . $description . '">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+
+        $html = '<meta tag1="content1" property="og:unrelated1 og:description og:unrelated2" '.
+            'tag2="content2" content="' . $description . '" tag3="content3">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+
+        $html = '<meta property="og:description" name="description" content="' . $description . '">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+
+        $description = 'Bob and Alice share "cookies".';
+
+        $html = '<meta property="og:description" content=\'' . $description . '\'>';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+
+        $html = '<meta tag1="content1" property="og:unrelated1 og:description og:unrelated2" '.
+            'tag2="content2" content=\'' . $description . '\' tag3="content3">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+
+        $html = '<meta property="og:description" name="description" content=\'' . $description . '\'>';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+    }
+
     /**
      * Test html_extract_tag() when the tag <meta name= is not found.
      */
-- 
cgit v1.2.3