tests/bookmark/LinkUtilsTest.php

   1 <?php
   2
   3 namespace Shaarli\Bookmark;
   4
   5 use PHPUnit\Framework\TestCase;
   6
   7 require_once 'tests/utils/CurlUtils.php';
   8
   9 /**
  10  * Class LinkUtilsTest.
  11  */
  12 class LinkUtilsTest extends TestCase
  13 {
  14     /**
  15      * Test html_extract_title() when the title is found.
  16      */
  17     public function testHtmlExtractExistentTitle()
  18     {
  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));
  24     }
  25
  26     /**
  27      * Test html_extract_title() when the title is not found.
  28      */
  29     public function testHtmlExtractNonExistentTitle()
  30     {
  31         $html = '<html><meta>stuff</meta></html>';
  32         $this->assertFalse(html_extract_title($html));
  33     }
  34
  35     /**
  36      * Test headers_extract_charset() when the charset is found.
  37      */
  38     public function testHeadersExtractExistentCharset()
  39     {
  40         $charset = 'x-MacCroatian';
  41         $headers = 'text/html; charset=' . $charset;
  42         $this->assertEquals(strtolower($charset), header_extract_charset($headers));
  43     }
  44
  45     /**
  46      * Test headers_extract_charset() when the charset is not found.
  47      */
  48     public function testHeadersExtractNonExistentCharset()
  49     {
  50         $headers = '';
  51         $this->assertFalse(header_extract_charset($headers));
  52
  53         $headers = 'text/html';
  54         $this->assertFalse(header_extract_charset($headers));
  55     }
  56
  57     /**
  58      * Test html_extract_charset() when the charset is found.
  59      */
  60     public function testHtmlExtractExistentCharset()
  61     {
  62         $charset = 'x-MacCroatian';
  63         $html = '<html><meta>stuff2</meta><meta charset="' . $charset . '"/></html>';
  64         $this->assertEquals(strtolower($charset), html_extract_charset($html));
  65     }
  66
  67     /**
  68      * Test html_extract_charset() when the charset is not found.
  69      */
  70     public function testHtmlExtractNonExistentCharset()
  71     {
  72         $html = '<html><meta>stuff</meta></html>';
  73         $this->assertFalse(html_extract_charset($html));
  74         $html = '<html><meta>stuff</meta><meta charset=""/></html>';
  75         $this->assertFalse(html_extract_charset($html));
  76     }
  77
  78     /**
  79      * Test html_extract_tag() when the tag <meta name= is found.
  80      */
  81     public function testHtmlExtractExistentNameTag()
  82     {
  83         $description = 'Bob and Alice share cookies.';
  84
  85         // Simple one line
  86         $html = '<html><meta>stuff2</meta><meta name="description" content="' . $description . '"/></html>';
  87         $this->assertEquals($description, html_extract_tag('description', $html));
  88
  89         // Simple OpenGraph
  90         $html = '<meta property="og:description" content="' . $description . '">';
  91         $this->assertEquals($description, html_extract_tag('description', $html));
  92
  93         // Simple reversed OpenGraph
  94         $html = '<meta content="' . $description . '" property="og:description">';
  95         $this->assertEquals($description, html_extract_tag('description', $html));
  96
  97         // ItemProp OpenGraph
  98         $html = '<meta itemprop="og:description" content="' . $description . '">';
  99         $this->assertEquals($description, html_extract_tag('description', $html));
 100
 101         // OpenGraph without quotes
 102         $html = '<meta property=og:description content="' . $description . '">';
 103         $this->assertEquals($description, html_extract_tag('description', $html));
 104
 105         // OpenGraph reversed without quotes
 106         $html = '<meta content="' . $description . '" property=og:description>';
 107         $this->assertEquals($description, html_extract_tag('description', $html));
 108
 109         // OpenGraph with noise
 110         $html = '<meta tag1="content1" property="og:description" tag2="content2" content="' .
 111             $description . '" tag3="content3">';
 112         $this->assertEquals($description, html_extract_tag('description', $html));
 113
 114         // OpenGraph reversed with noise
 115         $html = '<meta tag1="content1" content="' . $description . '" ' .
 116             'tag3="content3" tag2="content2" property="og:description">';
 117         $this->assertEquals($description, html_extract_tag('description', $html));
 118
 119         // OpenGraph multiple properties start
 120         $html = '<meta property="unrelated og:description" content="' . $description . '">';
 121         $this->assertEquals($description, html_extract_tag('description', $html));
 122
 123         // OpenGraph multiple properties end
 124         $html = '<meta property="og:description unrelated" content="' . $description . '">';
 125         $this->assertEquals($description, html_extract_tag('description', $html));
 126
 127         // OpenGraph multiple properties both end
 128         $html = '<meta property="og:unrelated1 og:description og:unrelated2" content="' . $description . '">';
 129         $this->assertEquals($description, html_extract_tag('description', $html));
 130
 131         // OpenGraph multiple properties both end with noise
 132         $html = '<meta tag1="content1" property="og:unrelated1 og:description og:unrelated2" '.
 133             'tag2="content2" content="' . $description . '" tag3="content3">';
 134         $this->assertEquals($description, html_extract_tag('description', $html));
 135
 136         // OpenGraph reversed multiple properties start
 137         $html = '<meta content="' . $description . '" property="unrelated og:description">';
 138         $this->assertEquals($description, html_extract_tag('description', $html));
 139
 140         // OpenGraph reversed multiple properties end
 141         $html = '<meta content="' . $description . '" property="og:description unrelated">';
 142         $this->assertEquals($description, html_extract_tag('description', $html));
 143
 144         // OpenGraph reversed multiple properties both end
 145         $html = '<meta content="' . $description . '" property="og:unrelated1 og:description og:unrelated2">';
 146         $this->assertEquals($description, html_extract_tag('description', $html));
 147
 148         // OpenGraph reversed multiple properties both end with noise
 149         $html = '<meta tag1="content1" content="' . $description . '" tag2="content2" '.
 150             'property="og:unrelated1 og:description og:unrelated2" tag3="content3">';
 151         $this->assertEquals($description, html_extract_tag('description', $html));
 152
 153         // Suggestion from #1375
 154         $html = '<meta property="og:description" name="description" content="' . $description . '">';
 155         $this->assertEquals($description, html_extract_tag('description', $html));
 156     }
 157
 158     /**
 159      * Test html_extract_tag() when the tag <meta name= is not found.
 160      */
 161     public function testHtmlExtractNonExistentNameTag()
 162     {
 163         $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>';
 164         $this->assertFalse(html_extract_tag('description', $html));
 165
 166         // Partial meta tag
 167         $html = '<meta content="Brief description">';
 168         $this->assertFalse(html_extract_tag('description', $html));
 169
 170         $html = '<meta property="og:description">';
 171         $this->assertFalse(html_extract_tag('description', $html));
 172
 173         $html = '<meta tag1="content1" property="og:description">';
 174         $this->assertFalse(html_extract_tag('description', $html));
 175
 176         $html = '<meta property="og:description" tag1="content1">';
 177         $this->assertFalse(html_extract_tag('description', $html));
 178
 179         $html = '<meta tag1="content1" content="Brief description">';
 180         $this->assertFalse(html_extract_tag('description', $html));
 181
 182         $html = '<meta content="Brief description" tag1="content1">';
 183         $this->assertFalse(html_extract_tag('description', $html));
 184     }
 185
 186     /**
 187      * Test html_extract_tag() when the tag <meta property="og: is found.
 188      */
 189     public function testHtmlExtractExistentOgTag()
 190     {
 191         $description = 'Bob and Alice share cookies.';
 192         $html = '<html><meta>stuff2</meta><meta property="og:description" content="' . $description . '"/></html>';
 193         $this->assertEquals($description, html_extract_tag('description', $html));
 194     }
 195
 196     /**
 197      * Test html_extract_tag() when the tag <meta property="og: is not found.
 198      */
 199     public function testHtmlExtractNonExistentOgTag()
 200     {
 201         $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>';
 202         $this->assertFalse(html_extract_tag('description', $html));
 203     }
 204
 205     /**
 206      * Test the download callback with valid value
 207      */
 208     public function testCurlDownloadCallbackOk()
 209     {
 210         $callback = get_curl_download_callback(
 211             $charset,
 212             $title,
 213             $desc,
 214             $keywords,
 215             false,
 216             'ut_curl_getinfo_ok'
 217         );
 218         $data = [
 219             'HTTP/1.1 200 OK',
 220             'Server: GitHub.com',
 221             'Date: Sat, 28 Oct 2017 12:01:33 GMT',
 222             'Content-Type: text/html; charset=utf-8',
 223             'Status: 200 OK',
 224             'end' => 'th=device-width">'
 225                 . '<title>Refactoring · GitHub</title>'
 226                 . '<link rel="search" type="application/opensea',
 227             '<title>ignored</title>'
 228                 . '<meta name="description" content="desc" />'
 229                 . '<meta name="keywords" content="key1,key2" />',
 230         ];
 231         foreach ($data as $key => $line) {
 232             $ignore = null;
 233             $expected = $key !== 'end' ? strlen($line) : false;
 234             $this->assertEquals($expected, $callback($ignore, $line));
 235             if ($expected === false) {
 236                 break;
 237             }
 238         }
 239         $this->assertEquals('utf-8', $charset);
 240         $this->assertEquals('Refactoring · GitHub', $title);
 241         $this->assertEmpty($desc);
 242         $this->assertEmpty($keywords);
 243     }
 244
 245     /**
 246      * Test the download callback with valid values and no charset
 247      */
 248     public function testCurlDownloadCallbackOkNoCharset()
 249     {
 250         $callback = get_curl_download_callback(
 251             $charset,
 252             $title,
 253             $desc,
 254             $keywords,
 255             false,
 256             'ut_curl_getinfo_no_charset'
 257         );
 258         $data = [
 259             'HTTP/1.1 200 OK',
 260             'end' => 'th=device-width">'
 261                 . '<title>Refactoring · GitHub</title>'
 262                 . '<link rel="search" type="application/opensea',
 263             '<title>ignored</title>'
 264             . '<meta name="description" content="desc" />'
 265             . '<meta name="keywords" content="key1,key2" />',
 266         ];
 267         foreach ($data as $key => $line) {
 268             $ignore = null;
 269             $this->assertEquals(strlen($line), $callback($ignore, $line));
 270         }
 271         $this->assertEmpty($charset);
 272         $this->assertEquals('Refactoring · GitHub', $title);
 273         $this->assertEmpty($desc);
 274         $this->assertEmpty($keywords);
 275     }
 276
 277     /**
 278      * Test the download callback with valid values and no charset
 279      */
 280     public function testCurlDownloadCallbackOkHtmlCharset()
 281     {
 282         $callback = get_curl_download_callback(
 283             $charset,
 284             $title,
 285             $desc,
 286             $keywords,
 287             false,
 288             'ut_curl_getinfo_no_charset'
 289         );
 290         $data = [
 291             'HTTP/1.1 200 OK',
 292             '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />',
 293             'end' => 'th=device-width">'
 294                 . '<title>Refactoring · GitHub</title>'
 295                 . '<link rel="search" type="application/opensea',
 296             '<title>ignored</title>'
 297             . '<meta name="description" content="desc" />'
 298             . '<meta name="keywords" content="key1,key2" />',
 299         ];
 300         foreach ($data as $key => $line) {
 301             $ignore = null;
 302             $expected = $key !== 'end' ? strlen($line) : false;
 303             $this->assertEquals($expected, $callback($ignore, $line));
 304             if ($expected === false) {
 305                 break;
 306             }
 307         }
 308         $this->assertEquals('utf-8', $charset);
 309         $this->assertEquals('Refactoring · GitHub', $title);
 310         $this->assertEmpty($desc);
 311         $this->assertEmpty($keywords);
 312     }
 313
 314     /**
 315      * Test the download callback with valid values and no title
 316      */
 317     public function testCurlDownloadCallbackOkNoTitle()
 318     {
 319         $callback = get_curl_download_callback(
 320             $charset,
 321             $title,
 322             $desc,
 323             $keywords,
 324             false,
 325             'ut_curl_getinfo_ok'
 326         );
 327         $data = [
 328             'HTTP/1.1 200 OK',
 329             'end' => 'th=device-width">Refactoring · GitHub<link rel="search" type="application/opensea',
 330             'ignored',
 331         ];
 332         foreach ($data as $key => $line) {
 333             $ignore = null;
 334             $this->assertEquals(strlen($line), $callback($ignore, $line));
 335         }
 336         $this->assertEquals('utf-8', $charset);
 337         $this->assertEmpty($title);
 338         $this->assertEmpty($desc);
 339         $this->assertEmpty($keywords);
 340     }
 341
 342     /**
 343      * Test the download callback with an invalid content type.
 344      */
 345     public function testCurlDownloadCallbackInvalidContentType()
 346     {
 347         $callback = get_curl_download_callback(
 348             $charset,
 349             $title,
 350             $desc,
 351             $keywords,
 352             false,
 353             'ut_curl_getinfo_ct_ko'
 354         );
 355         $ignore = null;
 356         $this->assertFalse($callback($ignore, ''));
 357         $this->assertEmpty($charset);
 358         $this->assertEmpty($title);
 359     }
 360
 361     /**
 362      * Test the download callback with an invalid response code.
 363      */
 364     public function testCurlDownloadCallbackInvalidResponseCode()
 365     {
 366         $callback = $callback = get_curl_download_callback(
 367             $charset,
 368             $title,
 369             $desc,
 370             $keywords,
 371             false,
 372             'ut_curl_getinfo_rc_ko'
 373         );
 374         $ignore = null;
 375         $this->assertFalse($callback($ignore, ''));
 376         $this->assertEmpty($charset);
 377         $this->assertEmpty($title);
 378     }
 379
 380     /**
 381      * Test the download callback with an invalid content type and response code.
 382      */
 383     public function testCurlDownloadCallbackInvalidContentTypeAndResponseCode()
 384     {
 385         $callback = $callback = get_curl_download_callback(
 386             $charset,
 387             $title,
 388             $desc,
 389             $keywords,
 390             false,
 391             'ut_curl_getinfo_rs_ct_ko'
 392         );
 393         $ignore = null;
 394         $this->assertFalse($callback($ignore, ''));
 395         $this->assertEmpty($charset);
 396         $this->assertEmpty($title);
 397     }
 398
 399     /**
 400      * Test the download callback with valid value, and retrieve_description option enabled.
 401      */
 402     public function testCurlDownloadCallbackOkWithDesc()
 403     {
 404         $callback = get_curl_download_callback(
 405             $charset,
 406             $title,
 407             $desc,
 408             $keywords,
 409             true,
 410             'ut_curl_getinfo_ok'
 411         );
 412         $data = [
 413             'HTTP/1.1 200 OK',
 414             'Server: GitHub.com',
 415             'Date: Sat, 28 Oct 2017 12:01:33 GMT',
 416             'Content-Type: text/html; charset=utf-8',
 417             'Status: 200 OK',
 418             'th=device-width">'
 419                 . '<title>Refactoring · GitHub</title>'
 420                 . '<link rel="search" type="application/opensea',
 421             'end' => '<title>ignored</title>'
 422             . '<meta name="description" content="link desc" />'
 423             . '<meta name="keywords" content="key1,key2" />',
 424         ];
 425         foreach ($data as $key => $line) {
 426             $ignore = null;
 427             $expected = $key !== 'end' ? strlen($line) : false;
 428             $this->assertEquals($expected, $callback($ignore, $line));
 429             if ($expected === false) {
 430                 break;
 431             }
 432         }
 433         $this->assertEquals('utf-8', $charset);
 434         $this->assertEquals('Refactoring · GitHub', $title);
 435         $this->assertEquals('link desc', $desc);
 436         $this->assertEquals('key1 key2', $keywords);
 437     }
 438
 439     /**
 440      * Test the download callback with valid value, and retrieve_description option enabled,
 441      * but no desc or keyword defined in the page.
 442      */
 443     public function testCurlDownloadCallbackOkWithDescNotFound()
 444     {
 445         $callback = get_curl_download_callback(
 446             $charset,
 447             $title,
 448             $desc,
 449             $keywords,
 450             true,
 451             'ut_curl_getinfo_ok'
 452         );
 453         $data = [
 454             'HTTP/1.1 200 OK',
 455             'Server: GitHub.com',
 456             'Date: Sat, 28 Oct 2017 12:01:33 GMT',
 457             'Content-Type: text/html; charset=utf-8',
 458             'Status: 200 OK',
 459             'th=device-width">'
 460                 . '<title>Refactoring · GitHub</title>'
 461                 . '<link rel="search" type="application/opensea',
 462             'end' => '<title>ignored</title>',
 463         ];
 464         foreach ($data as $key => $line) {
 465             $ignore = null;
 466             $expected = $key !== 'end' ? strlen($line) : false;
 467             $this->assertEquals($expected, $callback($ignore, $line));
 468             if ($expected === false) {
 469                 break;
 470             }
 471         }
 472         $this->assertEquals('utf-8', $charset);
 473         $this->assertEquals('Refactoring · GitHub', $title);
 474         $this->assertEmpty($desc);
 475         $this->assertEmpty($keywords);
 476     }
 477
 478     /**
 479      * Test text2clickable.
 480      */
 481     public function testText2clickable()
 482     {
 483         $text = 'stuff http://hello.there/is=someone#here otherstuff';
 484         $expectedText = 'stuff <a href="http://hello.there/is=someone#here">'
 485             . 'http://hello.there/is=someone#here</a> otherstuff';
 486         $processedText = text2clickable($text);
 487         $this->assertEquals($expectedText, $processedText);
 488
 489         $text = 'stuff http://hello.there/is=someone#here(please) otherstuff';
 490         $expectedText = 'stuff <a href="http://hello.there/is=someone#here(please)">'
 491             . 'http://hello.there/is=someone#here(please)</a> otherstuff';
 492         $processedText = text2clickable($text);
 493         $this->assertEquals($expectedText, $processedText);
 494
 495         $text = 'stuff http://hello.there/is=someone#here(please)&no otherstuff';
 496         $text = 'stuff http://hello.there/is=someone#here(please)&no otherstuff';
 497         $expectedText = 'stuff <a href="http://hello.there/is=someone#here(please)&no">'
 498             . 'http://hello.there/is=someone#here(please)&no</a> otherstuff';
 499         $processedText = text2clickable($text);
 500         $this->assertEquals($expectedText, $processedText);
 501     }
 502
 503     /**
 504      * Test testSpace2nbsp.
 505      */
 506     public function testSpace2nbsp()
 507     {
 508         $text = '  Are you   thrilled  by flags   ?' . PHP_EOL . ' Really?';
 509         $expectedText = '&nbsp; Are you &nbsp; thrilled &nbsp;by flags &nbsp; ?' . PHP_EOL . '&nbsp;Really?';
 510         $processedText = space2nbsp($text);
 511         $this->assertEquals($expectedText, $processedText);
 512     }
 513
 514     /**
 515      * Test hashtags auto-link.
 516      */
 517     public function testHashtagAutolink()
 518     {
 519         $index = 'http://domain.tld/';
 520         $rawDescription = '#hashtag\n
 521             # nothashtag\n
 522             test#nothashtag #hashtag \#nothashtag\n
 523             test #hashtag #hashtag test #hashtag.test\n
 524             #hashtag #hashtag-nothashtag #hashtag_hashtag\n
 525             What is #ашок anyway?\n
 526             カタカナ #カタカナ」カタカナ\n';
 527         $autolinkedDescription = hashtag_autolink($rawDescription, $index);
 528
 529         $this->assertContains($this->getHashtagLink('hashtag', $index), $autolinkedDescription);
 530         $this->assertNotContains(' #hashtag', $autolinkedDescription);
 531         $this->assertNotContains('>#nothashtag', $autolinkedDescription);
 532         $this->assertContains($this->getHashtagLink('ашок', $index), $autolinkedDescription);
 533         $this->assertContains($this->getHashtagLink('カタカナ', $index), $autolinkedDescription);
 534         $this->assertContains($this->getHashtagLink('hashtag_hashtag', $index), $autolinkedDescription);
 535         $this->assertNotContains($this->getHashtagLink('hashtag-nothashtag', $index), $autolinkedDescription);
 536     }
 537
 538     /**
 539      * Test hashtags auto-link without index URL.
 540      */
 541     public function testHashtagAutolinkNoIndex()
 542     {
 543         $rawDescription = 'blabla #hashtag x#nothashtag';
 544         $autolinkedDescription = hashtag_autolink($rawDescription);
 545
 546         $this->assertContains($this->getHashtagLink('hashtag'), $autolinkedDescription);
 547         $this->assertNotContains(' #hashtag', $autolinkedDescription);
 548         $this->assertNotContains('>#nothashtag', $autolinkedDescription);
 549     }
 550
 551     /**
 552      * Test is_note with note URLs.
 553      */
 554     public function testIsNote()
 555     {
 556         $this->assertTrue(is_note('?'));
 557         $this->assertTrue(is_note('?abcDEf'));
 558         $this->assertTrue(is_note('?_abcDEf#123'));
 559     }
 560
 561     /**
 562      * Test is_note with non note URLs.
 563      */
 564     public function testIsNotNote()
 565     {
 566         $this->assertFalse(is_note(''));
 567         $this->assertFalse(is_note('nope'));
 568         $this->assertFalse(is_note('https://github.com/shaarli/Shaarli/?hi'));
 569     }
 570
 571     /**
 572      * Util function to build an hashtag link.
 573      *
 574      * @param string $hashtag Hashtag name.
 575      * @param string $index Index URL.
 576      *
 577      * @return string HTML hashtag link.
 578      */
 579     private function getHashtagLink($hashtag, $index = '')
 580     {
 581         $hashtagLink = '<a href="' . $index . './add-tag/$1" title="Hashtag $1">#$1</a>';
 582         return str_replace('$1', $hashtag, $hashtagLink);
 583     }
 584 }