]>
git.immae.eu Git - github/shaarli/Shaarli.git/blob - tests/bookmark/LinkUtilsTest.php
3 namespace Shaarli\Bookmark
;
7 require_once 'tests/utils/CurlUtils.php';
10 * Class LinkUtilsTest.
12 class LinkUtilsTest
extends TestCase
15 * Test html_extract_title() when the title is found.
17 public function testHtmlExtractExistentTitle()
19 $title = 'Read me please.';
20 $html = '<html><meta>stuff</meta><title>' . $title . '</title></html>';
21 $this->assertEquals($title, html_extract_title($html));
22 $html = '<html><title>' . $title . '</title>blabla<title>another</title></html>';
23 $this->assertEquals($title, html_extract_title($html));
27 * Test html_extract_title() when the title is not found.
29 public function testHtmlExtractNonExistentTitle()
31 $html = '<html><meta>stuff</meta></html>';
32 $this->assertFalse(html_extract_title($html));
36 * Test headers_extract_charset() when the charset is found.
38 public function testHeadersExtractExistentCharset()
40 $charset = 'x-MacCroatian';
41 $headers = 'text/html; charset=' . $charset;
42 $this->assertEquals(strtolower($charset), header_extract_charset($headers));
46 * Test headers_extract_charset() when the charset is found with odd quotes.
48 public function testHeadersExtractExistentCharsetWithQuotes()
50 $charset = 'x-MacCroatian';
51 $headers = 'text/html; charset="' . $charset . '"otherstuff="test"';
52 $this->assertEquals(strtolower($charset), header_extract_charset($headers));
54 $headers = 'text/html; charset=\'' . $charset . '\'otherstuff="test"';
55 $this->assertEquals(strtolower($charset), header_extract_charset($headers));
59 * Test headers_extract_charset() when the charset is not found.
61 public function testHeadersExtractNonExistentCharset()
64 $this->assertFalse(header_extract_charset($headers));
66 $headers = 'text/html';
67 $this->assertFalse(header_extract_charset($headers));
71 * Test html_extract_charset() when the charset is found.
73 public function testHtmlExtractExistentCharset()
75 $charset = 'x-MacCroatian';
76 $html = '<html><meta>stuff2</meta><meta charset="' . $charset . '"/></html>';
77 $this->assertEquals(strtolower($charset), html_extract_charset($html));
81 * Test html_extract_charset() when the charset is not found.
83 public function testHtmlExtractNonExistentCharset()
85 $html = '<html><meta>stuff</meta></html>';
86 $this->assertFalse(html_extract_charset($html));
87 $html = '<html><meta>stuff</meta><meta charset=""/></html>';
88 $this->assertFalse(html_extract_charset($html));
92 * Test html_extract_tag() when the tag <meta name= is found.
94 public function testHtmlExtractExistentNameTag()
96 $description = 'Bob and Alice share cookies.';
99 $html = '<html><meta>stuff2</meta><meta name="description" content="' . $description . '"/></html>';
100 $this->assertEquals($description, html_extract_tag('description', $html));
103 $html = '<meta property="og:description" content="' . $description . '">';
104 $this->assertEquals($description, html_extract_tag('description', $html));
106 // Simple reversed OpenGraph
107 $html = '<meta content="' . $description . '" property="og:description">';
108 $this->assertEquals($description, html_extract_tag('description', $html));
110 // ItemProp OpenGraph
111 $html = '<meta itemprop="og:description" content="' . $description . '">';
112 $this->assertEquals($description, html_extract_tag('description', $html));
114 // OpenGraph without quotes
115 $html = '<meta property=og:description content="' . $description . '">';
116 $this->assertEquals($description, html_extract_tag('description', $html));
118 // OpenGraph reversed without quotes
119 $html = '<meta content="' . $description . '" property=og:description>';
120 $this->assertEquals($description, html_extract_tag('description', $html));
122 // OpenGraph with noise
123 $html = '<meta tag1="content1" property="og:description" tag2="content2" content="' .
124 $description . '" tag3="content3">';
125 $this->assertEquals($description, html_extract_tag('description', $html));
127 // OpenGraph reversed with noise
128 $html = '<meta tag1="content1" content="' . $description . '" ' .
129 'tag3="content3" tag2="content2" property="og:description">';
130 $this->assertEquals($description, html_extract_tag('description', $html));
132 // OpenGraph multiple properties start
133 $html = '<meta property="unrelated og:description" content="' . $description . '">';
134 $this->assertEquals($description, html_extract_tag('description', $html));
136 // OpenGraph multiple properties end
137 $html = '<meta property="og:description unrelated" content="' . $description . '">';
138 $this->assertEquals($description, html_extract_tag('description', $html));
140 // OpenGraph multiple properties both end
141 $html = '<meta property="og:unrelated1 og:description og:unrelated2" content="' . $description . '">';
142 $this->assertEquals($description, html_extract_tag('description', $html));
144 // OpenGraph multiple properties both end with noise
145 $html = '<meta tag1="content1" property="og:unrelated1 og:description og:unrelated2" '.
146 'tag2="content2" content="' . $description . '" tag3="content3">';
147 $this->assertEquals($description, html_extract_tag('description', $html));
149 // OpenGraph reversed multiple properties start
150 $html = '<meta content="' . $description . '" property="unrelated og:description">';
151 $this->assertEquals($description, html_extract_tag('description', $html));
153 // OpenGraph reversed multiple properties end
154 $html = '<meta content="' . $description . '" property="og:description unrelated">';
155 $this->assertEquals($description, html_extract_tag('description', $html));
157 // OpenGraph reversed multiple properties both end
158 $html = '<meta content="' . $description . '" property="og:unrelated1 og:description og:unrelated2">';
159 $this->assertEquals($description, html_extract_tag('description', $html));
161 // OpenGraph reversed multiple properties both end with noise
162 $html = '<meta tag1="content1" content="' . $description . '" tag2="content2" '.
163 'property="og:unrelated1 og:description og:unrelated2" tag3="content3">';
164 $this->assertEquals($description, html_extract_tag('description', $html));
166 // Suggestion from #1375
167 $html = '<meta property="og:description" name="description" content="' . $description . '">';
168 $this->assertEquals($description, html_extract_tag('description', $html));
172 * Test html_extract_tag() with double quoted content containing single quote, and the opposite.
174 public function testHtmlExtractExistentNameTagWithMixedQuotes(): void
176 $description = 'Bob and Alice share M&M\'s.';
178 $html = '<meta property="og:description" content="' . $description . '">';
179 $this->assertEquals($description, html_extract_tag('description', $html));
181 $html = '<meta tag1="content1" property="og:unrelated1 og:description og:unrelated2" '.
182 'tag2="content2" content="' . $description . '" tag3="content3">';
183 $this->assertEquals($description, html_extract_tag('description', $html));
185 $html = '<meta property="og:description" name="description" content="' . $description . '">';
186 $this->assertEquals($description, html_extract_tag('description', $html));
188 $description = 'Bob and Alice share "cookies".';
190 $html = '<meta property="og:description" content=\'' . $description . '\'>';
191 $this->assertEquals($description, html_extract_tag('description', $html));
193 $html = '<meta tag1="content1" property="og:unrelated1 og:description og:unrelated2" '.
194 'tag2="content2" content=\'' . $description . '\' tag3="content3">';
195 $this->assertEquals($description, html_extract_tag('description', $html));
197 $html = '<meta property="og:description" name="description" content=\'' . $description . '\'>';
198 $this->assertEquals($description, html_extract_tag('description', $html));
202 * Test html_extract_tag() when the tag <meta name= is not found.
204 public function testHtmlExtractNonExistentNameTag()
206 $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>';
207 $this->assertFalse(html_extract_tag('description', $html));
210 $html = '<meta content="Brief description">';
211 $this->assertFalse(html_extract_tag('description', $html));
213 $html = '<meta property="og:description">';
214 $this->assertFalse(html_extract_tag('description', $html));
216 $html = '<meta tag1="content1" property="og:description">';
217 $this->assertFalse(html_extract_tag('description', $html));
219 $html = '<meta property="og:description" tag1="content1">';
220 $this->assertFalse(html_extract_tag('description', $html));
222 $html = '<meta tag1="content1" content="Brief description">';
223 $this->assertFalse(html_extract_tag('description', $html));
225 $html = '<meta content="Brief description" tag1="content1">';
226 $this->assertFalse(html_extract_tag('description', $html));
230 * Test html_extract_tag() when the tag <meta property="og: is found.
232 public function testHtmlExtractExistentOgTag()
234 $description = 'Bob and Alice share cookies.';
235 $html = '<html><meta>stuff2</meta><meta property="og:description" content="' . $description . '"/></html>';
236 $this->assertEquals($description, html_extract_tag('description', $html));
240 * Test html_extract_tag() when the tag <meta property="og: is not found.
242 public function testHtmlExtractNonExistentOgTag()
244 $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>';
245 $this->assertFalse(html_extract_tag('description', $html));
249 * Test the header callback with valid value
251 public function testCurlHeaderCallbackOk(): void
253 $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_ok');
256 'Server: GitHub.com',
257 'Date: Sat, 28 Oct 2017 12:01:33 GMT',
258 'Content-Type: text/html; charset=utf-8',
262 foreach ($data as $chunk) {
263 static::assertIsInt($callback(null, $chunk));
266 static::assertSame('utf-8', $charset);
270 * Test the download callback with valid value
272 public function testCurlDownloadCallbackOk(): void
275 $callback = get_curl_download_callback(
286 . '<title>Refactoring · GitHub</title>'
287 . '<link rel="search" type="application/opensea',
288 '<title>ignored</title>'
289 . '<meta name="description" content="desc" />'
290 . '<meta name="keywords" content="key1,key2" />',
293 foreach ($data as $chunk) {
294 static::assertSame(strlen($chunk), $callback(null, $chunk));
297 static::assertSame('utf-8', $charset);
298 static::assertSame('Refactoring · GitHub', $title);
299 static::assertEmpty($desc);
300 static::assertEmpty($keywords);
304 * Test the header callback with valid value
306 public function testCurlHeaderCallbackNoCharset(): void
308 $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_no_charset');
313 foreach ($data as $chunk) {
314 static::assertSame(strlen($chunk), $callback(null, $chunk));
317 static::assertFalse($charset);
321 * Test the download callback with valid values and no charset
323 public function testCurlDownloadCallbackOkNoCharset(): void
326 $callback = get_curl_download_callback(
336 'end' => 'th=device-width">'
337 . '<title>Refactoring · GitHub</title>'
338 . '<link rel="search" type="application/opensea',
339 '<title>ignored</title>'
340 . '<meta name="description" content="desc" />'
341 . '<meta name="keywords" content="key1,key2" />',
344 foreach ($data as $chunk) {
345 static::assertSame(strlen($chunk), $callback(null, $chunk));
348 $this->assertEmpty($charset);
349 $this->assertEquals('Refactoring · GitHub', $title);
350 $this->assertEmpty($desc);
351 $this->assertEmpty($keywords);
355 * Test the download callback with valid values and no charset
357 public function testCurlDownloadCallbackOkHtmlCharset(): void
360 $callback = get_curl_download_callback(
370 '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />',
371 'end' => 'th=device-width">'
372 . '<title>Refactoring · GitHub</title>'
373 . '<link rel="search" type="application/opensea',
374 '<title>ignored</title>'
375 . '<meta name="description" content="desc" />'
376 . '<meta name="keywords" content="key1,key2" />',
378 foreach ($data as $chunk) {
379 static::assertSame(strlen($chunk), $callback(null, $chunk));
382 $this->assertEquals('utf-8', $charset);
383 $this->assertEquals('Refactoring · GitHub', $title);
384 $this->assertEmpty($desc);
385 $this->assertEmpty($keywords);
389 * Test the download callback with valid values and no title
391 public function testCurlDownloadCallbackOkNoTitle(): void
394 $callback = get_curl_download_callback(
404 'end' => 'th=device-width">Refactoring · GitHub<link rel="search" type="application/opensea',
408 foreach ($data as $chunk) {
409 static::assertSame(strlen($chunk), $callback(null, $chunk));
412 $this->assertEquals('utf-8', $charset);
413 $this->assertEmpty($title);
414 $this->assertEmpty($desc);
415 $this->assertEmpty($keywords);
419 * Test the header callback with an invalid content type.
421 public function testCurlHeaderCallbackInvalidContentType(): void
423 $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_ct_ko');
428 static::assertFalse($callback(null, $data[0]));
429 static::assertNull($charset);
433 * Test the header callback with an invalid response code.
435 public function testCurlHeaderCallbackInvalidResponseCode(): void
437 $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_rc_ko');
439 static::assertFalse($callback(null, ''));
440 static::assertNull($charset);
444 * Test the header callback with an invalid content type and response code.
446 public function testCurlHeaderCallbackInvalidContentTypeAndResponseCode(): void
448 $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_rs_ct_ko');
450 static::assertFalse($callback(null, ''));
451 static::assertNull($charset);
455 * Test the download callback with valid value, and retrieve_description option enabled.
457 public function testCurlDownloadCallbackOkWithDesc(): void
460 $callback = get_curl_download_callback(
470 . '<title>Refactoring · GitHub</title>'
471 . '<link rel="search" type="application/opensea',
472 'end' => '<title>ignored</title>'
473 . '<meta name="description" content="link desc" />'
474 . '<meta name="keywords" content="key1,key2" />',
477 foreach ($data as $chunk) {
478 static::assertSame(strlen($chunk), $callback(null, $chunk));
481 $this->assertEquals('utf-8', $charset);
482 $this->assertEquals('Refactoring · GitHub', $title);
483 $this->assertEquals('link desc', $desc);
484 $this->assertEquals('key1 key2', $keywords);
488 * Test the download callback with valid value, and retrieve_description option enabled,
489 * but no desc or keyword defined in the page.
491 public function testCurlDownloadCallbackOkWithDescNotFound(): void
494 $callback = get_curl_download_callback(
504 . '<title>Refactoring · GitHub</title>'
505 . '<link rel="search" type="application/opensea',
506 'end' => '<title>ignored</title>',
509 foreach ($data as $chunk) {
510 static::assertSame(strlen($chunk), $callback(null, $chunk));
513 $this->assertEquals('utf-8', $charset);
514 $this->assertEquals('Refactoring · GitHub', $title);
515 $this->assertEmpty($desc);
516 $this->assertEmpty($keywords);
520 * Test text2clickable.
522 public function testText2clickable()
524 $text = 'stuff http://hello.there/is=someone#here otherstuff';
525 $expectedText = 'stuff <a href="http://hello.there/is=someone#here">'
526 . 'http://hello.there/is=someone#here</a> otherstuff';
527 $processedText = text2clickable($text);
528 $this->assertEquals($expectedText, $processedText);
530 $text = 'stuff http://hello.there/is=someone#here(please) otherstuff';
531 $expectedText = 'stuff <a href="http://hello.there/is=someone#here(please)">'
532 . 'http://hello.there/is=someone#here(please)</a> otherstuff';
533 $processedText = text2clickable($text);
534 $this->assertEquals($expectedText, $processedText);
536 $text = 'stuff http://hello.there/is=someone#here(please)&no otherstuff';
537 $text = 'stuff http://hello.there/is=someone#here(please)&no otherstuff';
538 $expectedText = 'stuff <a href="http://hello.there/is=someone#here(please)&no">'
539 . 'http://hello.there/is=someone#here(please)&no</a> otherstuff';
540 $processedText = text2clickable($text);
541 $this->assertEquals($expectedText, $processedText);
545 * Test testSpace2nbsp.
547 public function testSpace2nbsp()
549 $text = ' Are you thrilled by flags ?' . PHP_EOL
. ' Really?';
550 $expectedText = ' Are you thrilled by flags ?' . PHP_EOL
. ' Really?';
551 $processedText = space2nbsp($text);
552 $this->assertEquals($expectedText, $processedText);
556 * Test hashtags auto-link.
558 public function testHashtagAutolink()
560 $index = 'http://domain.tld/';
561 $rawDescription = '#hashtag\n
563 test#nothashtag #hashtag \#nothashtag\n
564 test #hashtag #hashtag test #hashtag.test\n
565 #hashtag #hashtag-nothashtag #hashtag_hashtag\n
566 What is #ашок anyway?\n
568 $autolinkedDescription = hashtag_autolink($rawDescription, $index);
570 $this->assertContainsPolyfill($this->getHashtagLink('hashtag', $index), $autolinkedDescription);
571 $this->assertNotContainsPolyfill(' #hashtag', $autolinkedDescription);
572 $this->assertNotContainsPolyfill('>#nothashtag', $autolinkedDescription);
573 $this->assertContainsPolyfill($this->getHashtagLink('ашок', $index), $autolinkedDescription);
574 $this->assertContainsPolyfill($this->getHashtagLink('カタカナ', $index), $autolinkedDescription);
575 $this->assertContainsPolyfill($this->getHashtagLink('hashtag_hashtag', $index), $autolinkedDescription);
576 $this->assertNotContainsPolyfill($this->getHashtagLink('hashtag-nothashtag', $index), $autolinkedDescription);
580 * Test hashtags auto-link without index URL.
582 public function testHashtagAutolinkNoIndex()
584 $rawDescription = 'blabla #hashtag x#nothashtag';
585 $autolinkedDescription = hashtag_autolink($rawDescription);
587 $this->assertContainsPolyfill($this->getHashtagLink('hashtag'), $autolinkedDescription);
588 $this->assertNotContainsPolyfill(' #hashtag', $autolinkedDescription);
589 $this->assertNotContainsPolyfill('>#nothashtag', $autolinkedDescription);
593 * Test is_note with note URLs.
595 public function testIsNote()
597 $this->assertTrue(is_note('?'));
598 $this->assertTrue(is_note('?abcDEf'));
599 $this->assertTrue(is_note('?_abcDEf#123'));
603 * Test is_note with non note URLs.
605 public function testIsNotNote()
607 $this->assertFalse(is_note(''));
608 $this->assertFalse(is_note('nope'));
609 $this->assertFalse(is_note('https://github.com/shaarli/Shaarli/?hi'));
613 * Test tags_str2array with whitespace separator.
615 public function testTagsStr2ArrayWithSpaceSeparator(): void
619 static::assertSame(['tag1', 'tag2', 'tag3'], tags_str2array('tag1 tag2 tag3', $separator));
620 static::assertSame(['tag1', 'tag2', 'tag3'], tags_str2array('tag1 tag2 tag3', $separator));
621 static::assertSame(['tag1', 'tag2', 'tag3'], tags_str2array(' tag1 tag2 tag3 ', $separator));
622 static::assertSame(['tag1@', 'tag2,', '.tag3'], tags_str2array(' tag1@ tag2, .tag3 ', $separator));
623 static::assertSame([], tags_str2array('', $separator));
624 static::assertSame([], tags_str2array(' ', $separator));
625 static::assertSame([], tags_str2array(null, $separator));
629 * Test tags_str2array with @ separator.
631 public function testTagsStr2ArrayWithCharSeparator(): void
635 static::assertSame(['tag1', 'tag2', 'tag3'], tags_str2array('tag1@tag2@tag3', $separator));
636 static::assertSame(['tag1', 'tag2', 'tag3'], tags_str2array('tag1@@@@tag2@@@@tag3', $separator));
637 static::assertSame(['tag1', 'tag2', 'tag3'], tags_str2array('@@@tag1@@@tag2@@@@tag3@@', $separator));
639 ['tag1#', 'tag2, and other', '.tag3'],
640 tags_str2array('@@@ tag1# @@@ tag2, and other @@@@.tag3@@', $separator)
642 static::assertSame([], tags_str2array('', $separator));
643 static::assertSame([], tags_str2array(' ', $separator));
644 static::assertSame([], tags_str2array(null, $separator));
648 * Test tags_array2str with ' ' separator.
650 public function testTagsArray2StrWithSpaceSeparator(): void
654 static::assertSame('tag1 tag2 tag3', tags_array2str(['tag1', 'tag2', 'tag3'], $separator));
655 static::assertSame('tag1, tag2@ tag3', tags_array2str(['tag1,', 'tag2@', 'tag3'], $separator));
656 static::assertSame('tag1 tag2 tag3', tags_array2str([' tag1 ', 'tag2', 'tag3 '], $separator));
657 static::assertSame('tag1 tag2 tag3', tags_array2str([' tag1 ', ' ', 'tag2', ' ', 'tag3 '], $separator));
658 static::assertSame('tag1', tags_array2str([' tag1 '], $separator));
659 static::assertSame('', tags_array2str([' '], $separator));
660 static::assertSame('', tags_array2str([], $separator));
661 static::assertSame('', tags_array2str(null, $separator));
665 * Test tags_array2str with @ separator.
667 public function testTagsArray2StrWithCharSeparator(): void
671 static::assertSame('tag1@tag2@tag3', tags_array2str(['tag1', 'tag2', 'tag3'], $separator));
672 static::assertSame('tag1,@tag2@tag3', tags_array2str(['tag1,', 'tag2@', 'tag3'], $separator));
674 'tag1@tag2, and other@tag3',
675 tags_array2str(['@@@@ tag1@@@', ' @tag2, and other @', 'tag3@@@@'], $separator)
677 static::assertSame('tag1@tag2@tag3', tags_array2str(['@@@tag1@@@', '@', 'tag2', '@@@', 'tag3@@@'], $separator));
678 static::assertSame('tag1', tags_array2str(['@@@@tag1@@@@'], $separator));
679 static::assertSame('', tags_array2str(['@@@'], $separator));
680 static::assertSame('', tags_array2str([], $separator));
681 static::assertSame('', tags_array2str(null, $separator));
685 * Test tags_array2str with @ separator.
687 public function testTagsFilterWithSpaceSeparator(): void
691 static::assertSame(['tag1', 'tag2', 'tag3'], tags_filter(['tag1', 'tag2', 'tag3'], $separator));
692 static::assertSame(['tag1,', 'tag2@', 'tag3'], tags_filter(['tag1,', 'tag2@', 'tag3'], $separator));
693 static::assertSame(['tag1', 'tag2', 'tag3'], tags_filter([' tag1 ', 'tag2', 'tag3 '], $separator));
694 static::assertSame(['tag1', 'tag2', 'tag3'], tags_filter([' tag1 ', ' ', 'tag2', ' ', 'tag3 '], $separator));
695 static::assertSame(['tag1'], tags_filter([' tag1 '], $separator));
696 static::assertSame([], tags_filter([' '], $separator));
697 static::assertSame([], tags_filter([], $separator));
698 static::assertSame([], tags_filter(null, $separator));
702 * Test tags_array2str with @ separator.
704 public function testTagsArrayFilterWithSpaceSeparator(): void
708 static::assertSame(['tag1', 'tag2', 'tag3'], tags_filter(['tag1', 'tag2', 'tag3'], $separator));
709 static::assertSame(['tag1,', 'tag2#', 'tag3'], tags_filter(['tag1,', 'tag2#', 'tag3'], $separator));
711 ['tag1', 'tag2, and other', 'tag3'],
712 tags_filter(['@@@@ tag1@@@', ' @tag2, and other @', 'tag3@@@@'], $separator)
714 static::assertSame(['tag1', 'tag2', 'tag3'], tags_filter(['@@@tag1@@@', '@', 'tag2', '@@@', 'tag3@@@'], $separator));
715 static::assertSame(['tag1'], tags_filter(['@@@@tag1@@@@'], $separator));
716 static::assertSame([], tags_filter(['@@@'], $separator));
717 static::assertSame([], tags_filter([], $separator));
718 static::assertSame([], tags_filter(null, $separator));
722 * Util function to build an hashtag link.
724 * @param string $hashtag Hashtag name.
725 * @param string $index Index URL.
727 * @return string HTML hashtag link.
729 private function getHashtagLink($hashtag, $index = '')
731 $hashtagLink = '<a href="' . $index . './add-tag/$1" title="Hashtag $1">#$1</a>';
732 return str_replace('$1', $hashtag, $hashtagLink);