tests/bookmark/LinkUtilsTest.php

   1 <?php
   2
   3 namespace Shaarli\Bookmark;
   4
   5 use Shaarli\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 found with odd quotes.
  47      */
  48     public function testHeadersExtractExistentCharsetWithQuotes()
  49     {
  50         $charset = 'x-MacCroatian';
  51         $headers = 'text/html; charset="' . $charset . '"otherstuff="test"';
  52         $this->assertEquals(strtolower($charset), header_extract_charset($headers));
  53
  54         $headers = 'text/html; charset=\'' . $charset . '\'otherstuff="test"';
  55         $this->assertEquals(strtolower($charset), header_extract_charset($headers));
  56     }
  57
  58     /**
  59      * Test headers_extract_charset() when the charset is not found.
  60      */
  61     public function testHeadersExtractNonExistentCharset()
  62     {
  63         $headers = '';
  64         $this->assertFalse(header_extract_charset($headers));
  65
  66         $headers = 'text/html';
  67         $this->assertFalse(header_extract_charset($headers));
  68     }
  69
  70     /**
  71      * Test html_extract_charset() when the charset is found.
  72      */
  73     public function testHtmlExtractExistentCharset()
  74     {
  75         $charset = 'x-MacCroatian';
  76         $html = '<html><meta>stuff2</meta><meta charset="' . $charset . '"/></html>';
  77         $this->assertEquals(strtolower($charset), html_extract_charset($html));
  78     }
  79
  80     /**
  81      * Test html_extract_charset() when the charset is not found.
  82      */
  83     public function testHtmlExtractNonExistentCharset()
  84     {
  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));
  89     }
  90
  91     /**
  92      * Test html_extract_tag() when the tag <meta name= is found.
  93      */
  94     public function testHtmlExtractExistentNameTag()
  95     {
  96         $description = 'Bob and Alice share cookies.';
  97
  98         // Simple one line
  99         $html = '<html><meta>stuff2</meta><meta name="description" content="' . $description . '"/></html>';
 100         $this->assertEquals($description, html_extract_tag('description', $html));
 101
 102         // Simple OpenGraph
 103         $html = '<meta property="og:description" content="' . $description . '">';
 104         $this->assertEquals($description, html_extract_tag('description', $html));
 105
 106         // Simple reversed OpenGraph
 107         $html = '<meta content="' . $description . '" property="og:description">';
 108         $this->assertEquals($description, html_extract_tag('description', $html));
 109
 110         // ItemProp OpenGraph
 111         $html = '<meta itemprop="og:description" content="' . $description . '">';
 112         $this->assertEquals($description, html_extract_tag('description', $html));
 113
 114         // OpenGraph without quotes
 115         $html = '<meta property=og:description content="' . $description . '">';
 116         $this->assertEquals($description, html_extract_tag('description', $html));
 117
 118         // OpenGraph reversed without quotes
 119         $html = '<meta content="' . $description . '" property=og:description>';
 120         $this->assertEquals($description, html_extract_tag('description', $html));
 121
 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));
 126
 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));
 131
 132         // OpenGraph multiple properties start
 133         $html = '<meta property="unrelated og:description" content="' . $description . '">';
 134         $this->assertEquals($description, html_extract_tag('description', $html));
 135
 136         // OpenGraph multiple properties end
 137         $html = '<meta property="og:description unrelated" content="' . $description . '">';
 138         $this->assertEquals($description, html_extract_tag('description', $html));
 139
 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));
 143
 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));
 148
 149         // OpenGraph reversed multiple properties start
 150         $html = '<meta content="' . $description . '" property="unrelated og:description">';
 151         $this->assertEquals($description, html_extract_tag('description', $html));
 152
 153         // OpenGraph reversed multiple properties end
 154         $html = '<meta content="' . $description . '" property="og:description unrelated">';
 155         $this->assertEquals($description, html_extract_tag('description', $html));
 156
 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));
 160
 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));
 165
 166         // Suggestion from #1375
 167         $html = '<meta property="og:description" name="description" content="' . $description . '">';
 168         $this->assertEquals($description, html_extract_tag('description', $html));
 169     }
 170
 171     /**
 172      * Test html_extract_tag() when the tag <meta name= is not found.
 173      */
 174     public function testHtmlExtractNonExistentNameTag()
 175     {
 176         $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>';
 177         $this->assertFalse(html_extract_tag('description', $html));
 178
 179         // Partial meta tag
 180         $html = '<meta content="Brief description">';
 181         $this->assertFalse(html_extract_tag('description', $html));
 182
 183         $html = '<meta property="og:description">';
 184         $this->assertFalse(html_extract_tag('description', $html));
 185
 186         $html = '<meta tag1="content1" property="og:description">';
 187         $this->assertFalse(html_extract_tag('description', $html));
 188
 189         $html = '<meta property="og:description" tag1="content1">';
 190         $this->assertFalse(html_extract_tag('description', $html));
 191
 192         $html = '<meta tag1="content1" content="Brief description">';
 193         $this->assertFalse(html_extract_tag('description', $html));
 194
 195         $html = '<meta content="Brief description" tag1="content1">';
 196         $this->assertFalse(html_extract_tag('description', $html));
 197     }
 198
 199     /**
 200      * Test html_extract_tag() when the tag <meta property="og: is found.
 201      */
 202     public function testHtmlExtractExistentOgTag()
 203     {
 204         $description = 'Bob and Alice share cookies.';
 205         $html = '<html><meta>stuff2</meta><meta property="og:description" content="' . $description . '"/></html>';
 206         $this->assertEquals($description, html_extract_tag('description', $html));
 207     }
 208
 209     /**
 210      * Test html_extract_tag() when the tag <meta property="og: is not found.
 211      */
 212     public function testHtmlExtractNonExistentOgTag()
 213     {
 214         $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>';
 215         $this->assertFalse(html_extract_tag('description', $html));
 216     }
 217
 218     /**
 219      * Test the header callback with valid value
 220      */
 221     public function testCurlHeaderCallbackOk(): void
 222     {
 223         $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_ok');
 224         $data = [
 225             'HTTP/1.1 200 OK',
 226             'Server: GitHub.com',
 227             'Date: Sat, 28 Oct 2017 12:01:33 GMT',
 228             'Content-Type: text/html; charset=utf-8',
 229             'Status: 200 OK',
 230         ];
 231
 232         foreach ($data as $chunk) {
 233             static::assertIsInt($callback(null, $chunk));
 234         }
 235
 236         static::assertSame('utf-8', $charset);
 237     }
 238
 239     /**
 240      * Test the download callback with valid value
 241      */
 242     public function testCurlDownloadCallbackOk(): void
 243     {
 244         $charset = 'utf-8';
 245         $callback = get_curl_download_callback(
 246             $charset,
 247             $title,
 248             $desc,
 249             $keywords,
 250             false
 251         );
 252
 253         $data = [
 254             'th=device-width">'
 255                 . '<title>Refactoring · GitHub</title>'
 256                 . '<link rel="search" type="application/opensea',
 257             '<title>ignored</title>'
 258                 . '<meta name="description" content="desc" />'
 259                 . '<meta name="keywords" content="key1,key2" />',
 260         ];
 261
 262         foreach ($data as $chunk) {
 263             static::assertSame(strlen($chunk), $callback(null, $chunk));
 264         }
 265
 266         static::assertSame('utf-8', $charset);
 267         static::assertSame('Refactoring · GitHub', $title);
 268         static::assertEmpty($desc);
 269         static::assertEmpty($keywords);
 270     }
 271
 272     /**
 273      * Test the header callback with valid value
 274      */
 275     public function testCurlHeaderCallbackNoCharset(): void
 276     {
 277         $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_no_charset');
 278         $data = [
 279             'HTTP/1.1 200 OK',
 280         ];
 281
 282         foreach ($data as $chunk) {
 283             static::assertSame(strlen($chunk), $callback(null, $chunk));
 284         }
 285
 286         static::assertFalse($charset);
 287     }
 288
 289     /**
 290      * Test the download callback with valid values and no charset
 291      */
 292     public function testCurlDownloadCallbackOkNoCharset(): void
 293     {
 294         $charset = null;
 295         $callback = get_curl_download_callback(
 296             $charset,
 297             $title,
 298             $desc,
 299             $keywords,
 300             false
 301         );
 302
 303         $data = [
 304             'end' => 'th=device-width">'
 305                 . '<title>Refactoring · GitHub</title>'
 306                 . '<link rel="search" type="application/opensea',
 307             '<title>ignored</title>'
 308             . '<meta name="description" content="desc" />'
 309             . '<meta name="keywords" content="key1,key2" />',
 310         ];
 311
 312         foreach ($data as $chunk) {
 313             static::assertSame(strlen($chunk), $callback(null, $chunk));
 314         }
 315
 316         $this->assertEmpty($charset);
 317         $this->assertEquals('Refactoring · GitHub', $title);
 318         $this->assertEmpty($desc);
 319         $this->assertEmpty($keywords);
 320     }
 321
 322     /**
 323      * Test the download callback with valid values and no charset
 324      */
 325     public function testCurlDownloadCallbackOkHtmlCharset(): void
 326     {
 327         $charset = null;
 328         $callback = get_curl_download_callback(
 329             $charset,
 330             $title,
 331             $desc,
 332             $keywords,
 333             false
 334         );
 335
 336         $data = [
 337             '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />',
 338             'end' => 'th=device-width">'
 339                 . '<title>Refactoring · GitHub</title>'
 340                 . '<link rel="search" type="application/opensea',
 341             '<title>ignored</title>'
 342             . '<meta name="description" content="desc" />'
 343             . '<meta name="keywords" content="key1,key2" />',
 344         ];
 345         foreach ($data as $chunk) {
 346             static::assertSame(strlen($chunk), $callback(null, $chunk));
 347         }
 348
 349         $this->assertEquals('utf-8', $charset);
 350         $this->assertEquals('Refactoring · GitHub', $title);
 351         $this->assertEmpty($desc);
 352         $this->assertEmpty($keywords);
 353     }
 354
 355     /**
 356      * Test the download callback with valid values and no title
 357      */
 358     public function testCurlDownloadCallbackOkNoTitle(): void
 359     {
 360         $charset = 'utf-8';
 361         $callback = get_curl_download_callback(
 362             $charset,
 363             $title,
 364             $desc,
 365             $keywords,
 366             false
 367         );
 368
 369         $data = [
 370             'end' => 'th=device-width">Refactoring · GitHub<link rel="search" type="application/opensea',
 371             'ignored',
 372         ];
 373
 374         foreach ($data as $chunk) {
 375             static::assertSame(strlen($chunk), $callback(null, $chunk));
 376         }
 377
 378         $this->assertEquals('utf-8', $charset);
 379         $this->assertEmpty($title);
 380         $this->assertEmpty($desc);
 381         $this->assertEmpty($keywords);
 382     }
 383
 384     /**
 385      * Test the header callback with an invalid content type.
 386      */
 387     public function testCurlHeaderCallbackInvalidContentType(): void
 388     {
 389         $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_ct_ko');
 390         $data = [
 391             'HTTP/1.1 200 OK',
 392         ];
 393
 394         static::assertFalse($callback(null, $data[0]));
 395         static::assertNull($charset);
 396     }
 397
 398     /**
 399      * Test the header callback with an invalid response code.
 400      */
 401     public function testCurlHeaderCallbackInvalidResponseCode(): void
 402     {
 403         $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_rc_ko');
 404
 405         static::assertFalse($callback(null, ''));
 406         static::assertNull($charset);
 407     }
 408
 409     /**
 410      * Test the header callback with an invalid content type and response code.
 411      */
 412     public function testCurlHeaderCallbackInvalidContentTypeAndResponseCode(): void
 413     {
 414         $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_rs_ct_ko');
 415
 416         static::assertFalse($callback(null, ''));
 417         static::assertNull($charset);
 418     }
 419
 420     /**
 421      * Test the download callback with valid value, and retrieve_description option enabled.
 422      */
 423     public function testCurlDownloadCallbackOkWithDesc(): void
 424     {
 425         $charset = 'utf-8';
 426         $callback = get_curl_download_callback(
 427             $charset,
 428             $title,
 429             $desc,
 430             $keywords,
 431             true
 432         );
 433         $data = [
 434             'th=device-width">'
 435                 . '<title>Refactoring · GitHub</title>'
 436                 . '<link rel="search" type="application/opensea',
 437             'end' => '<title>ignored</title>'
 438             . '<meta name="description" content="link desc" />'
 439             . '<meta name="keywords" content="key1,key2" />',
 440         ];
 441
 442         foreach ($data as $chunk) {
 443             static::assertSame(strlen($chunk), $callback(null, $chunk));
 444         }
 445
 446         $this->assertEquals('utf-8', $charset);
 447         $this->assertEquals('Refactoring · GitHub', $title);
 448         $this->assertEquals('link desc', $desc);
 449         $this->assertEquals('key1 key2', $keywords);
 450     }
 451
 452     /**
 453      * Test the download callback with valid value, and retrieve_description option enabled,
 454      * but no desc or keyword defined in the page.
 455      */
 456     public function testCurlDownloadCallbackOkWithDescNotFound(): void
 457     {
 458         $charset = 'utf-8';
 459         $callback = get_curl_download_callback(
 460             $charset,
 461             $title,
 462             $desc,
 463             $keywords,
 464             true,
 465             'ut_curl_getinfo_ok'
 466         );
 467         $data = [
 468             'th=device-width">'
 469                 . '<title>Refactoring · GitHub</title>'
 470                 . '<link rel="search" type="application/opensea',
 471             'end' => '<title>ignored</title>',
 472         ];
 473
 474         foreach ($data as $chunk) {
 475             static::assertSame(strlen($chunk), $callback(null, $chunk));
 476         }
 477
 478         $this->assertEquals('utf-8', $charset);
 479         $this->assertEquals('Refactoring · GitHub', $title);
 480         $this->assertEmpty($desc);
 481         $this->assertEmpty($keywords);
 482     }
 483
 484     /**
 485      * Test text2clickable.
 486      */
 487     public function testText2clickable()
 488     {
 489         $text = 'stuff http://hello.there/is=someone#here otherstuff';
 490         $expectedText = 'stuff <a href="http://hello.there/is=someone#here">'
 491             . 'http://hello.there/is=someone#here</a> otherstuff';
 492         $processedText = text2clickable($text);
 493         $this->assertEquals($expectedText, $processedText);
 494
 495         $text = 'stuff http://hello.there/is=someone#here(please) otherstuff';
 496         $expectedText = 'stuff <a href="http://hello.there/is=someone#here(please)">'
 497             . 'http://hello.there/is=someone#here(please)</a> otherstuff';
 498         $processedText = text2clickable($text);
 499         $this->assertEquals($expectedText, $processedText);
 500
 501         $text = 'stuff http://hello.there/is=someone#here(please)&no otherstuff';
 502         $text = 'stuff http://hello.there/is=someone#here(please)&no otherstuff';
 503         $expectedText = 'stuff <a href="http://hello.there/is=someone#here(please)&no">'
 504             . 'http://hello.there/is=someone#here(please)&no</a> otherstuff';
 505         $processedText = text2clickable($text);
 506         $this->assertEquals($expectedText, $processedText);
 507     }
 508
 509     /**
 510      * Test testSpace2nbsp.
 511      */
 512     public function testSpace2nbsp()
 513     {
 514         $text = '  Are you   thrilled  by flags   ?' . PHP_EOL . ' Really?';
 515         $expectedText = '&nbsp; Are you &nbsp; thrilled &nbsp;by flags &nbsp; ?' . PHP_EOL . '&nbsp;Really?';
 516         $processedText = space2nbsp($text);
 517         $this->assertEquals($expectedText, $processedText);
 518     }
 519
 520     /**
 521      * Test hashtags auto-link.
 522      */
 523     public function testHashtagAutolink()
 524     {
 525         $index = 'http://domain.tld/';
 526         $rawDescription = '#hashtag\n
 527             # nothashtag\n
 528             test#nothashtag #hashtag \#nothashtag\n
 529             test #hashtag #hashtag test #hashtag.test\n
 530             #hashtag #hashtag-nothashtag #hashtag_hashtag\n
 531             What is #ашок anyway?\n
 532             カタカナ #カタカナ」カタカナ\n';
 533         $autolinkedDescription = hashtag_autolink($rawDescription, $index);
 534
 535         $this->assertContainsPolyfill($this->getHashtagLink('hashtag', $index), $autolinkedDescription);
 536         $this->assertNotContainsPolyfill(' #hashtag', $autolinkedDescription);
 537         $this->assertNotContainsPolyfill('>#nothashtag', $autolinkedDescription);
 538         $this->assertContainsPolyfill($this->getHashtagLink('ашок', $index), $autolinkedDescription);
 539         $this->assertContainsPolyfill($this->getHashtagLink('カタカナ', $index), $autolinkedDescription);
 540         $this->assertContainsPolyfill($this->getHashtagLink('hashtag_hashtag', $index), $autolinkedDescription);
 541         $this->assertNotContainsPolyfill($this->getHashtagLink('hashtag-nothashtag', $index), $autolinkedDescription);
 542     }
 543
 544     /**
 545      * Test hashtags auto-link without index URL.
 546      */
 547     public function testHashtagAutolinkNoIndex()
 548     {
 549         $rawDescription = 'blabla #hashtag x#nothashtag';
 550         $autolinkedDescription = hashtag_autolink($rawDescription);
 551
 552         $this->assertContainsPolyfill($this->getHashtagLink('hashtag'), $autolinkedDescription);
 553         $this->assertNotContainsPolyfill(' #hashtag', $autolinkedDescription);
 554         $this->assertNotContainsPolyfill('>#nothashtag', $autolinkedDescription);
 555     }
 556
 557     /**
 558      * Test is_note with note URLs.
 559      */
 560     public function testIsNote()
 561     {
 562         $this->assertTrue(is_note('?'));
 563         $this->assertTrue(is_note('?abcDEf'));
 564         $this->assertTrue(is_note('?_abcDEf#123'));
 565     }
 566
 567     /**
 568      * Test is_note with non note URLs.
 569      */
 570     public function testIsNotNote()
 571     {
 572         $this->assertFalse(is_note(''));
 573         $this->assertFalse(is_note('nope'));
 574         $this->assertFalse(is_note('https://github.com/shaarli/Shaarli/?hi'));
 575     }
 576
 577     /**
 578      * Util function to build an hashtag link.
 579      *
 580      * @param string $hashtag Hashtag name.
 581      * @param string $index Index URL.
 582      *
 583      * @return string HTML hashtag link.
 584      */
 585     private function getHashtagLink($hashtag, $index = '')
 586     {
 587         $hashtagLink = '<a href="' . $index . './add-tag/$1" title="Hashtag $1">#$1</a>';
 588         return str_replace('$1', $hashtag, $hashtagLink);
 589     }
 590 }