1 files changed, 222 insertions, 129 deletions
diff --git a/tests/bookmark/LinkUtilsTest.php b/tests/bookmark/LinkUtilsTest.php
index 591976f2..3321242f 100644
--- a/tests/bookmark/LinkUtilsTest.php
+++ b/tests/bookmark/LinkUtilsTest.php
@@ -2,9 +2,7 @@
 namespace Shaarli\Bookmark;
-use PHPUnit\Framework\TestCase;
+use Shaarli\TestCase;
-use ReferenceLinkDB;
-use Shaarli\Config\ConfigManager;
 require_once 'tests/utils/CurlUtils.php';
@@ -45,6 +43,19 @@ class LinkUtilsTest extends TestCase
    }
    /**
+     * Test headers_extract_charset() when the charset is found with odd quotes.
+     */
+    public function testHeadersExtractExistentCharsetWithQuotes()
+    {
+        $charset = 'x-MacCroatian';
+        $headers = 'text/html; charset="' . $charset . '"otherstuff="test"';
+        $this->assertEquals(strtolower($charset), header_extract_charset($headers));
+        $headers = 'text/html; charset=\'' . $charset . '\'otherstuff="test"';
+        $this->assertEquals(strtolower($charset), header_extract_charset($headers));
+    }
+    /**
     * Test headers_extract_charset() when the charset is not found.
     */
    public function testHeadersExtractNonExistentCharset()
@@ -83,8 +94,78 @@ class LinkUtilsTest extends TestCase
    public function testHtmlExtractExistentNameTag()
    {
        $description = 'Bob and Alice share cookies.';
+        // Simple one line
        $html = '<html><meta>stuff2</meta><meta name="description" content="' . $description . '"/></html>';
        $this->assertEquals($description, html_extract_tag('description', $html));
+        // Simple OpenGraph
+        $html = '<meta property="og:description" content="' . $description . '">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // Simple reversed OpenGraph
+        $html = '<meta content="' . $description . '" property="og:description">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // ItemProp OpenGraph
+        $html = '<meta itemprop="og:description" content="' . $description . '">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // OpenGraph without quotes
+        $html = '<meta property=og:description content="' . $description . '">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // OpenGraph reversed without quotes
+        $html = '<meta content="' . $description . '" property=og:description>';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // OpenGraph with noise
+        $html = '<meta tag1="content1" property="og:description" tag2="content2" content="' .
+            $description . '" tag3="content3">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // OpenGraph reversed with noise
+        $html = '<meta tag1="content1" content="' . $description . '" ' .
+            'tag3="content3" tag2="content2" property="og:description">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // OpenGraph multiple properties start
+        $html = '<meta property="unrelated og:description" content="' . $description . '">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // OpenGraph multiple properties end
+        $html = '<meta property="og:description unrelated" content="' . $description . '">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // OpenGraph multiple properties both end
+        $html = '<meta property="og:unrelated1 og:description og:unrelated2" content="' . $description . '">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // OpenGraph multiple properties both end with noise
+        $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));
+        // OpenGraph reversed multiple properties start
+        $html = '<meta content="' . $description . '" property="unrelated og:description">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // OpenGraph reversed multiple properties end
+        $html = '<meta content="' . $description . '" property="og:description unrelated">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // OpenGraph reversed multiple properties both end
+        $html = '<meta content="' . $description . '" property="og:unrelated1 og:description og:unrelated2">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // OpenGraph reversed multiple properties both end with noise
+        $html = '<meta tag1="content1" content="' . $description . '" tag2="content2" '.
+            'property="og:unrelated1 og:description og:unrelated2" tag3="content3">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
+        // Suggestion from #1375
+        $html = '<meta property="og:description" name="description" content="' . $description . '">';
+        $this->assertEquals($description, html_extract_tag('description', $html));
    }
    /**
@@ -94,6 +175,25 @@ class LinkUtilsTest extends TestCase
    {
        $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>';
        $this->assertFalse(html_extract_tag('description', $html));
+        // Partial meta tag
+        $html = '<meta content="Brief description">';
+        $this->assertFalse(html_extract_tag('description', $html));
+        $html = '<meta property="og:description">';
+        $this->assertFalse(html_extract_tag('description', $html));
+        $html = '<meta tag1="content1" property="og:description">';
+        $this->assertFalse(html_extract_tag('description', $html));
+        $html = '<meta property="og:description" tag1="content1">';
+        $this->assertFalse(html_extract_tag('description', $html));
+        $html = '<meta tag1="content1" content="Brief description">';
+        $this->assertFalse(html_extract_tag('description', $html));
+        $html = '<meta content="Brief description" tag1="content1">';
+        $this->assertFalse(html_extract_tag('description', $html));
    }
    /**
@@ -116,60 +216,91 @@ class LinkUtilsTest extends TestCase
    }
    /**
+     * Test the header callback with valid value
+     */
+    public function testCurlHeaderCallbackOk(): void
+    {
+        $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_ok');
+        $data = [
+            'HTTP/1.1 200 OK',
+            'Server: GitHub.com',
+            'Date: Sat, 28 Oct 2017 12:01:33 GMT',
+            'Content-Type: text/html; charset=utf-8',
+            'Status: 200 OK',
+        ];
+        foreach ($data as $chunk) {
+            static::assertIsInt($callback(null, $chunk));
+        }
+        static::assertSame('utf-8', $charset);
+    }
+    /**
     * Test the download callback with valid value
     */
-    public function testCurlDownloadCallbackOk()
+    public function testCurlDownloadCallbackOk(): void
    {
+        $charset = 'utf-8';
        $callback = get_curl_download_callback(
            $charset,
            $title,
            $desc,
            $keywords,
-            false,
+            false
-            'ut_curl_getinfo_ok'
        );
        $data = [
-            'HTTP/1.1 200 OK',
+            'th=device-width">'
-            'Server: GitHub.com',
-            'Date: Sat, 28 Oct 2017 12:01:33 GMT',
-            'Content-Type: text/html; charset=utf-8',
-            'Status: 200 OK',
-            'end' => 'th=device-width">'
                . '<title>Refactoring · GitHub</title>'
                . '<link rel="search" type="application/opensea',
            '<title>ignored</title>'
                . '<meta name="description" content="desc" />'
                . '<meta name="keywords" content="key1,key2" />',
        ];
-        foreach ($data as $key => $line) {
-            $ignore = null;
+        foreach ($data as $chunk) {
-            $expected = $key !== 'end' ? strlen($line) : false;
+            static::assertSame(strlen($chunk), $callback(null, $chunk));
-            $this->assertEquals($expected, $callback($ignore, $line));
-            if ($expected === false) {
-                break;
-            }
        }
-        $this->assertEquals('utf-8', $charset);
-        $this->assertEquals('Refactoring · GitHub', $title);
+        static::assertSame('utf-8', $charset);
-        $this->assertEmpty($desc);
+        static::assertSame('Refactoring · GitHub', $title);
-        $this->assertEmpty($keywords);
+        static::assertEmpty($desc);
+        static::assertEmpty($keywords);
+    }
+    /**
+     * Test the header callback with valid value
+     */
+    public function testCurlHeaderCallbackNoCharset(): void
+    {
+        $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_no_charset');
+        $data = [
+            'HTTP/1.1 200 OK',
+        ];
+        foreach ($data as $chunk) {
+            static::assertSame(strlen($chunk), $callback(null, $chunk));
+        }
+        static::assertFalse($charset);
    }
    /**
     * Test the download callback with valid values and no charset
     */
-    public function testCurlDownloadCallbackOkNoCharset()
+    public function testCurlDownloadCallbackOkNoCharset(): void
    {
+        $charset = null;
        $callback = get_curl_download_callback(
            $charset,
            $title,
            $desc,
            $keywords,
-            false,
+            false
-            'ut_curl_getinfo_no_charset'
        );
        $data = [
-            'HTTP/1.1 200 OK',
            'end' => 'th=device-width">'
                . '<title>Refactoring · GitHub</title>'
                . '<link rel="search" type="application/opensea',
@@ -177,10 +308,11 @@ class LinkUtilsTest extends TestCase
            . '<meta name="description" content="desc" />'
            . '<meta name="keywords" content="key1,key2" />',
        ];
-        foreach ($data as $key => $line) {
-            $ignore = null;
+        foreach ($data as $chunk) {
-            $this->assertEquals(strlen($line), $callback($ignore, $line));
+            static::assertSame(strlen($chunk), $callback(null, $chunk));
        }
        $this->assertEmpty($charset);
        $this->assertEquals('Refactoring · GitHub', $title);
        $this->assertEmpty($desc);
@@ -190,18 +322,18 @@ class LinkUtilsTest extends TestCase
    /**
     * Test the download callback with valid values and no charset
     */
-    public function testCurlDownloadCallbackOkHtmlCharset()
+    public function testCurlDownloadCallbackOkHtmlCharset(): void
    {
+        $charset = null;
        $callback = get_curl_download_callback(
            $charset,
            $title,
            $desc,
            $keywords,
-            false,
+            false
-            'ut_curl_getinfo_no_charset'
        );
        $data = [
-            'HTTP/1.1 200 OK',
            '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />',
            'end' => 'th=device-width">'
                . '<title>Refactoring · GitHub</title>'
@@ -210,14 +342,10 @@ class LinkUtilsTest extends TestCase
            . '<meta name="description" content="desc" />'
            . '<meta name="keywords" content="key1,key2" />',
        ];
-        foreach ($data as $key => $line) {
+        foreach ($data as $chunk) {
-            $ignore = null;
+            static::assertSame(strlen($chunk), $callback(null, $chunk));
-            $expected = $key !== 'end' ? strlen($line) : false;
-            $this->assertEquals($expected, $callback($ignore, $line));
-            if ($expected === false) {
-                break;
-            }
        }
        $this->assertEquals('utf-8', $charset);
        $this->assertEquals('Refactoring · GitHub', $title);
        $this->assertEmpty($desc);
@@ -227,25 +355,26 @@ class LinkUtilsTest extends TestCase
    /**
     * Test the download callback with valid values and no title
     */
-    public function testCurlDownloadCallbackOkNoTitle()
+    public function testCurlDownloadCallbackOkNoTitle(): void
    {
+        $charset = 'utf-8';
        $callback = get_curl_download_callback(
            $charset,
            $title,
            $desc,
            $keywords,
-            false,
+            false
-            'ut_curl_getinfo_ok'
        );
        $data = [
-            'HTTP/1.1 200 OK',
            'end' => 'th=device-width">Refactoring · GitHub<link rel="search" type="application/opensea',
            'ignored',
        ];
-        foreach ($data as $key => $line) {
-            $ignore = null;
+        foreach ($data as $chunk) {
-            $this->assertEquals(strlen($line), $callback($ignore, $line));
+            static::assertSame(strlen($chunk), $callback(null, $chunk));
        }
        $this->assertEquals('utf-8', $charset);
        $this->assertEmpty($title);
        $this->assertEmpty($desc);
@@ -253,81 +382,55 @@ class LinkUtilsTest extends TestCase
    }
    /**
-     * Test the download callback with an invalid content type.
+     * Test the header callback with an invalid content type.
     */
-    public function testCurlDownloadCallbackInvalidContentType()
+    public function testCurlHeaderCallbackInvalidContentType(): void
    {
-        $callback = get_curl_download_callback(
+        $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_ct_ko');
-            $charset,
+        $data = [
-            $title,
+            'HTTP/1.1 200 OK',
-            $desc,
+        ];
-            $keywords,
-            false,
+        static::assertFalse($callback(null, $data[0]));
-            'ut_curl_getinfo_ct_ko'
+        static::assertNull($charset);
-        );
-        $ignore = null;
-        $this->assertFalse($callback($ignore, ''));
-        $this->assertEmpty($charset);
-        $this->assertEmpty($title);
    }
    /**
-     * Test the download callback with an invalid response code.
+     * Test the header callback with an invalid response code.
     */
-    public function testCurlDownloadCallbackInvalidResponseCode()
+    public function testCurlHeaderCallbackInvalidResponseCode(): void
    {
-        $callback = $callback = get_curl_download_callback(
+        $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_rc_ko');
-            $charset,
-            $title,
+        static::assertFalse($callback(null, ''));
-            $desc,
+        static::assertNull($charset);
-            $keywords,
-            false,
-            'ut_curl_getinfo_rc_ko'
-        );
-        $ignore = null;
-        $this->assertFalse($callback($ignore, ''));
-        $this->assertEmpty($charset);
-        $this->assertEmpty($title);
    }
    /**
-     * Test the download callback with an invalid content type and response code.
+     * Test the header callback with an invalid content type and response code.
     */
-    public function testCurlDownloadCallbackInvalidContentTypeAndResponseCode()
+    public function testCurlHeaderCallbackInvalidContentTypeAndResponseCode(): void
    {
-        $callback = $callback = get_curl_download_callback(
+        $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_rs_ct_ko');
-            $charset,
-            $title,
+        static::assertFalse($callback(null, ''));
-            $desc,
+        static::assertNull($charset);
-            $keywords,
-            false,
-            'ut_curl_getinfo_rs_ct_ko'
-        );
-        $ignore = null;
-        $this->assertFalse($callback($ignore, ''));
-        $this->assertEmpty($charset);
-        $this->assertEmpty($title);
    }
    /**
     * Test the download callback with valid value, and retrieve_description option enabled.
     */
-    public function testCurlDownloadCallbackOkWithDesc()
+    public function testCurlDownloadCallbackOkWithDesc(): void
    {
+        $charset = 'utf-8';
        $callback = get_curl_download_callback(
            $charset,
            $title,
            $desc,
            $keywords,
-            true,
+            true
-            'ut_curl_getinfo_ok'
        );
        $data = [
-            'HTTP/1.1 200 OK',
-            'Server: GitHub.com',
-            'Date: Sat, 28 Oct 2017 12:01:33 GMT',
-            'Content-Type: text/html; charset=utf-8',
-            'Status: 200 OK',
            'th=device-width">'
                . '<title>Refactoring · GitHub</title>'
                . '<link rel="search" type="application/opensea',
@@ -335,14 +438,11 @@ class LinkUtilsTest extends TestCase
            . '<meta name="description" content="link desc" />'
            . '<meta name="keywords" content="key1,key2" />',
        ];
-        foreach ($data as $key => $line) {
-            $ignore = null;
+        foreach ($data as $chunk) {
-            $expected = $key !== 'end' ? strlen($line) : false;
+            static::assertSame(strlen($chunk), $callback(null, $chunk));
-            $this->assertEquals($expected, $callback($ignore, $line));
-            if ($expected === false) {
-                break;
-            }
        }
        $this->assertEquals('utf-8', $charset);
        $this->assertEquals('Refactoring · GitHub', $title);
        $this->assertEquals('link desc', $desc);
@@ -353,8 +453,9 @@ class LinkUtilsTest extends TestCase
     * Test the download callback with valid value, and retrieve_description option enabled,
     * but no desc or keyword defined in the page.
     */
-    public function testCurlDownloadCallbackOkWithDescNotFound()
+    public function testCurlDownloadCallbackOkWithDescNotFound(): void
    {
+        $charset = 'utf-8';
        $callback = get_curl_download_callback(
            $charset,
            $title,
@@ -364,24 +465,16 @@ class LinkUtilsTest extends TestCase
            'ut_curl_getinfo_ok'
        );
        $data = [
-            'HTTP/1.1 200 OK',
-            'Server: GitHub.com',
-            'Date: Sat, 28 Oct 2017 12:01:33 GMT',
-            'Content-Type: text/html; charset=utf-8',
-            'Status: 200 OK',
            'th=device-width">'
                . '<title>Refactoring · GitHub</title>'
                . '<link rel="search" type="application/opensea',
            'end' => '<title>ignored</title>',
        ];
-        foreach ($data as $key => $line) {
-            $ignore = null;
+        foreach ($data as $chunk) {
-            $expected = $key !== 'end' ? strlen($line) : false;
+            static::assertSame(strlen($chunk), $callback(null, $chunk));
-            $this->assertEquals($expected, $callback($ignore, $line));
-            if ($expected === false) {
-                break;
-            }
        }
        $this->assertEquals('utf-8', $charset);
        $this->assertEquals('Refactoring · GitHub', $title);
        $this->assertEmpty($desc);
@@ -439,13 +532,13 @@ class LinkUtilsTest extends TestCase
            カタカナ #カタカナ」カタカナ\n';
        $autolinkedDescription = hashtag_autolink($rawDescription, $index);
-        $this->assertContains($this->getHashtagLink('hashtag', $index), $autolinkedDescription);
+        $this->assertContainsPolyfill($this->getHashtagLink('hashtag', $index), $autolinkedDescription);
-        $this->assertNotContains(' #hashtag', $autolinkedDescription);
+        $this->assertNotContainsPolyfill(' #hashtag', $autolinkedDescription);
-        $this->assertNotContains('>#nothashtag', $autolinkedDescription);
+        $this->assertNotContainsPolyfill('>#nothashtag', $autolinkedDescription);
-        $this->assertContains($this->getHashtagLink('ашок', $index), $autolinkedDescription);
+        $this->assertContainsPolyfill($this->getHashtagLink('ашок', $index), $autolinkedDescription);
-        $this->assertContains($this->getHashtagLink('カタカナ', $index), $autolinkedDescription);
+        $this->assertContainsPolyfill($this->getHashtagLink('カタカナ', $index), $autolinkedDescription);
-        $this->assertContains($this->getHashtagLink('hashtag_hashtag', $index), $autolinkedDescription);
+        $this->assertContainsPolyfill($this->getHashtagLink('hashtag_hashtag', $index), $autolinkedDescription);
-        $this->assertNotContains($this->getHashtagLink('hashtag-nothashtag', $index), $autolinkedDescription);
+        $this->assertNotContainsPolyfill($this->getHashtagLink('hashtag-nothashtag', $index), $autolinkedDescription);
    }
    /**
@@ -456,9 +549,9 @@ class LinkUtilsTest extends TestCase
        $rawDescription = 'blabla #hashtag x#nothashtag';
        $autolinkedDescription = hashtag_autolink($rawDescription);
-        $this->assertContains($this->getHashtagLink('hashtag'), $autolinkedDescription);
+        $this->assertContainsPolyfill($this->getHashtagLink('hashtag'), $autolinkedDescription);
-        $this->assertNotContains(' #hashtag', $autolinkedDescription);
+        $this->assertNotContainsPolyfill(' #hashtag', $autolinkedDescription);
-        $this->assertNotContains('>#nothashtag', $autolinkedDescription);
+        $this->assertNotContainsPolyfill('>#nothashtag', $autolinkedDescription);
    }
    /**
@@ -491,7 +584,7 @@ class LinkUtilsTest extends TestCase
     */
    private function getHashtagLink($hashtag, $index = '')
    {
-        $hashtagLink = '<a href="' . $index . '?addtag=$1" title="Hashtag $1">#$1</a>';
+        $hashtagLink = '<a href="' . $index . './add-tag/$1" title="Hashtag $1">#$1</a>';
        return str_replace('$1', $hashtag, $hashtagLink);
    }
 }