Fix metadata extract regex (2)

[github/shaarli/Shaarli.git] / tests / bookmark / LinkUtilsTest.php

diff --git a/tests/bookmark/LinkUtilsTest.php b/tests/bookmark/LinkUtilsTest.php
index c2f9f9305716829b91ba6ffe36cfc9a8a29cb80e..46a7f1fe7a4fd42b749bee30bd455d41ccdc107a 100644 (file)
--- 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.
      */
@@ -215,6 +245,16 @@ class LinkUtilsTest extends TestCase
         $this->assertFalse(html_extract_tag('description', $html));
     }
 
+    public function testHtmlExtractDescriptionFromGoogleRealCase(): void
+    {
+        $html = 'id="gsr"><meta content="Fêtes de fin d\'année" property="twitter:title"><meta '.
+                'content="Bonnes fêtes de fin d\'année ! #GoogleDoodle" property="twitter:description">'.
+                '<meta content="Bonnes fêtes de fin d\'année ! #GoogleDoodle" property="og:description">'.
+                '<meta content="summary_large_image" property="twitter:card"><meta co'
+        ;
+        $this->assertSame('Bonnes fêtes de fin d\'année ! #GoogleDoodle', html_extract_tag('description', $html));
+    }
+
     /**
      * Test the header callback with valid value
      */