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() with double quoted content containing single quote, and the opposite.
 173      */
 174     public function testHtmlExtractExistentNameTagWithMixedQuotes(): void
 175     {
 176         $description = 'Bob and Alice share M&M\'s.';
 177
 178         $html = '<meta property="og:description" content="' . $description . '">';
 179         $this->assertEquals($description, html_extract_tag('description', $html));
 180
 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));
 184
 185         $html = '<meta property="og:description" name="description" content="' . $description . '">';
 186         $this->assertEquals($description, html_extract_tag('description', $html));
 187
 188         $description = 'Bob and Alice share "cookies".';
 189
 190         $html = '<meta property="og:description" content=\'' . $description . '\'>';
 191         $this->assertEquals($description, html_extract_tag('description', $html));
 192
 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));
 196
 197         $html = '<meta property="og:description" name="description" content=\'' . $description . '\'>';
 198         $this->assertEquals($description, html_extract_tag('description', $html));
 199     }
 200
 201     /**
 202      * Test html_extract_tag() when the tag <meta name= is not found.
 203      */
 204     public function testHtmlExtractNonExistentNameTag()
 205     {
 206         $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>';
 207         $this->assertFalse(html_extract_tag('description', $html));
 208
 209         // Partial meta tag
 210         $html = '<meta content="Brief description">';
 211         $this->assertFalse(html_extract_tag('description', $html));
 212
 213         $html = '<meta property="og:description">';
 214         $this->assertFalse(html_extract_tag('description', $html));
 215
 216         $html = '<meta tag1="content1" property="og:description">';
 217         $this->assertFalse(html_extract_tag('description', $html));
 218
 219         $html = '<meta property="og:description" tag1="content1">';
 220         $this->assertFalse(html_extract_tag('description', $html));
 221
 222         $html = '<meta tag1="content1" content="Brief description">';
 223         $this->assertFalse(html_extract_tag('description', $html));
 224
 225         $html = '<meta content="Brief description" tag1="content1">';
 226         $this->assertFalse(html_extract_tag('description', $html));
 227     }
 228
 229     /**
 230      * Test html_extract_tag() when the tag <meta property="og: is found.
 231      */
 232     public function testHtmlExtractExistentOgTag()
 233     {
 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));
 237     }
 238
 239     /**
 240      * Test html_extract_tag() when the tag <meta property="og: is not found.
 241      */
 242     public function testHtmlExtractNonExistentOgTag()
 243     {
 244         $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>';
 245         $this->assertFalse(html_extract_tag('description', $html));
 246     }
 247
 248     /**
 249      * Test the header callback with valid value
 250      */
 251     public function testCurlHeaderCallbackOk(): void
 252     {
 253         $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_ok');
 254         $data = [
 255             'HTTP/1.1 200 OK',
 256             'Server: GitHub.com',
 257             'Date: Sat, 28 Oct 2017 12:01:33 GMT',
 258             'Content-Type: text/html; charset=utf-8',
 259             'Status: 200 OK',
 260         ];
 261
 262         foreach ($data as $chunk) {
 263             static::assertIsInt($callback(null, $chunk));
 264         }
 265
 266         static::assertSame('utf-8', $charset);
 267     }
 268
 269     /**
 270      * Test the download callback with valid value
 271      */
 272     public function testCurlDownloadCallbackOk(): void
 273     {
 274         $charset = 'utf-8';
 275         $callback = get_curl_download_callback(
 276             $charset,
 277             $title,
 278             $desc,
 279             $keywords,
 280             false
 281         );
 282
 283         $data = [
 284             'th=device-width">'
 285                 . '<title>Refactoring · GitHub</title>'
 286                 . '<link rel="search" type="application/opensea',
 287             '<title>ignored</title>'
 288                 . '<meta name="description" content="desc" />'
 289                 . '<meta name="keywords" content="key1,key2" />',
 290         ];
 291
 292         foreach ($data as $chunk) {
 293             static::assertSame(strlen($chunk), $callback(null, $chunk));
 294         }
 295
 296         static::assertSame('utf-8', $charset);
 297         static::assertSame('Refactoring · GitHub', $title);
 298         static::assertEmpty($desc);
 299         static::assertEmpty($keywords);
 300     }
 301
 302     /**
 303      * Test the header callback with valid value
 304      */
 305     public function testCurlHeaderCallbackNoCharset(): void
 306     {
 307         $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_no_charset');
 308         $data = [
 309             'HTTP/1.1 200 OK',
 310         ];
 311
 312         foreach ($data as $chunk) {
 313             static::assertSame(strlen($chunk), $callback(null, $chunk));
 314         }
 315
 316         static::assertFalse($charset);
 317     }
 318
 319     /**
 320      * Test the download callback with valid values and no charset
 321      */
 322     public function testCurlDownloadCallbackOkNoCharset(): void
 323     {
 324         $charset = null;
 325         $callback = get_curl_download_callback(
 326             $charset,
 327             $title,
 328             $desc,
 329             $keywords,
 330             false
 331         );
 332
 333         $data = [
 334             'end' => 'th=device-width">'
 335                 . '<title>Refactoring · GitHub</title>'
 336                 . '<link rel="search" type="application/opensea',
 337             '<title>ignored</title>'
 338             . '<meta name="description" content="desc" />'
 339             . '<meta name="keywords" content="key1,key2" />',
 340         ];
 341
 342         foreach ($data as $chunk) {
 343             static::assertSame(strlen($chunk), $callback(null, $chunk));
 344         }
 345
 346         $this->assertEmpty($charset);
 347         $this->assertEquals('Refactoring · GitHub', $title);
 348         $this->assertEmpty($desc);
 349         $this->assertEmpty($keywords);
 350     }
 351
 352     /**
 353      * Test the download callback with valid values and no charset
 354      */
 355     public function testCurlDownloadCallbackOkHtmlCharset(): void
 356     {
 357         $charset = null;
 358         $callback = get_curl_download_callback(
 359             $charset,
 360             $title,
 361             $desc,
 362             $keywords,
 363             false
 364         );
 365
 366         $data = [
 367             '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />',
 368             'end' => 'th=device-width">'
 369                 . '<title>Refactoring · GitHub</title>'
 370                 . '<link rel="search" type="application/opensea',
 371             '<title>ignored</title>'
 372             . '<meta name="description" content="desc" />'
 373             . '<meta name="keywords" content="key1,key2" />',
 374         ];
 375         foreach ($data as $chunk) {
 376             static::assertSame(strlen($chunk), $callback(null, $chunk));
 377         }
 378
 379         $this->assertEquals('utf-8', $charset);
 380         $this->assertEquals('Refactoring · GitHub', $title);
 381         $this->assertEmpty($desc);
 382         $this->assertEmpty($keywords);
 383     }
 384
 385     /**
 386      * Test the download callback with valid values and no title
 387      */
 388     public function testCurlDownloadCallbackOkNoTitle(): void
 389     {
 390         $charset = 'utf-8';
 391         $callback = get_curl_download_callback(
 392             $charset,
 393             $title,
 394             $desc,
 395             $keywords,
 396             false
 397         );
 398
 399         $data = [
 400             'end' => 'th=device-width">Refactoring · GitHub<link rel="search" type="application/opensea',
 401             'ignored',
 402         ];
 403
 404         foreach ($data as $chunk) {
 405             static::assertSame(strlen($chunk), $callback(null, $chunk));
 406         }
 407
 408         $this->assertEquals('utf-8', $charset);
 409         $this->assertEmpty($title);
 410         $this->assertEmpty($desc);
 411         $this->assertEmpty($keywords);
 412     }
 413
 414     /**
 415      * Test the header callback with an invalid content type.
 416      */
 417     public function testCurlHeaderCallbackInvalidContentType(): void
 418     {
 419         $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_ct_ko');
 420         $data = [
 421             'HTTP/1.1 200 OK',
 422         ];
 423
 424         static::assertFalse($callback(null, $data[0]));
 425         static::assertNull($charset);
 426     }
 427
 428     /**
 429      * Test the header callback with an invalid response code.
 430      */
 431     public function testCurlHeaderCallbackInvalidResponseCode(): void
 432     {
 433         $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_rc_ko');
 434
 435         static::assertFalse($callback(null, ''));
 436         static::assertNull($charset);
 437     }
 438
 439     /**
 440      * Test the header callback with an invalid content type and response code.
 441      */
 442     public function testCurlHeaderCallbackInvalidContentTypeAndResponseCode(): void
 443     {
 444         $callback = get_curl_header_callback($charset, 'ut_curl_getinfo_rs_ct_ko');
 445
 446         static::assertFalse($callback(null, ''));
 447         static::assertNull($charset);
 448     }
 449
 450     /**
 451      * Test the download callback with valid value, and retrieve_description option enabled.
 452      */
 453     public function testCurlDownloadCallbackOkWithDesc(): void
 454     {
 455         $charset = 'utf-8';
 456         $callback = get_curl_download_callback(
 457             $charset,
 458             $title,
 459             $desc,
 460             $keywords,
 461             true
 462         );
 463         $data = [
 464             'th=device-width">'
 465                 . '<title>Refactoring · GitHub</title>'
 466                 . '<link rel="search" type="application/opensea',
 467             'end' => '<title>ignored</title>'
 468             . '<meta name="description" content="link desc" />'
 469             . '<meta name="keywords" content="key1,key2" />',
 470         ];
 471
 472         foreach ($data as $chunk) {
 473             static::assertSame(strlen($chunk), $callback(null, $chunk));
 474         }
 475
 476         $this->assertEquals('utf-8', $charset);
 477         $this->assertEquals('Refactoring · GitHub', $title);
 478         $this->assertEquals('link desc', $desc);
 479         $this->assertEquals('key1 key2', $keywords);
 480     }
 481
 482     /**
 483      * Test the download callback with valid value, and retrieve_description option enabled,
 484      * but no desc or keyword defined in the page.
 485      */
 486     public function testCurlDownloadCallbackOkWithDescNotFound(): void
 487     {
 488         $charset = 'utf-8';
 489         $callback = get_curl_download_callback(
 490             $charset,
 491             $title,
 492             $desc,
 493             $keywords,
 494             true,
 495             'ut_curl_getinfo_ok'
 496         );
 497         $data = [
 498             'th=device-width">'
 499                 . '<title>Refactoring · GitHub</title>'
 500                 . '<link rel="search" type="application/opensea',
 501             'end' => '<title>ignored</title>',
 502         ];
 503
 504         foreach ($data as $chunk) {
 505             static::assertSame(strlen($chunk), $callback(null, $chunk));
 506         }
 507
 508         $this->assertEquals('utf-8', $charset);
 509         $this->assertEquals('Refactoring · GitHub', $title);
 510         $this->assertEmpty($desc);
 511         $this->assertEmpty($keywords);
 512     }
 513
 514     /**
 515      * Test text2clickable.
 516      */
 517     public function testText2clickable()
 518     {
 519         $text = 'stuff http://hello.there/is=someone#here otherstuff';
 520         $expectedText = 'stuff <a href="http://hello.there/is=someone#here">'
 521             . 'http://hello.there/is=someone#here</a> otherstuff';
 522         $processedText = text2clickable($text);
 523         $this->assertEquals($expectedText, $processedText);
 524
 525         $text = 'stuff http://hello.there/is=someone#here(please) otherstuff';
 526         $expectedText = 'stuff <a href="http://hello.there/is=someone#here(please)">'
 527             . 'http://hello.there/is=someone#here(please)</a> otherstuff';
 528         $processedText = text2clickable($text);
 529         $this->assertEquals($expectedText, $processedText);
 530
 531         $text = 'stuff http://hello.there/is=someone#here(please)&no otherstuff';
 532         $text = 'stuff http://hello.there/is=someone#here(please)&no otherstuff';
 533         $expectedText = 'stuff <a href="http://hello.there/is=someone#here(please)&no">'
 534             . 'http://hello.there/is=someone#here(please)&no</a> otherstuff';
 535         $processedText = text2clickable($text);
 536         $this->assertEquals($expectedText, $processedText);
 537     }
 538
 539     /**
 540      * Test testSpace2nbsp.
 541      */
 542     public function testSpace2nbsp()
 543     {
 544         $text = '  Are you   thrilled  by flags   ?' . PHP_EOL . ' Really?';
 545         $expectedText = '&nbsp; Are you &nbsp; thrilled &nbsp;by flags &nbsp; ?' . PHP_EOL . '&nbsp;Really?';
 546         $processedText = space2nbsp($text);
 547         $this->assertEquals($expectedText, $processedText);
 548     }
 549
 550     /**
 551      * Test hashtags auto-link.
 552      */
 553     public function testHashtagAutolink()
 554     {
 555         $index = 'http://domain.tld/';
 556         $rawDescription = '#hashtag\n
 557             # nothashtag\n
 558             test#nothashtag #hashtag \#nothashtag\n
 559             test #hashtag #hashtag test #hashtag.test\n
 560             #hashtag #hashtag-nothashtag #hashtag_hashtag\n
 561             What is #ашок anyway?\n
 562             カタカナ #カタカナ」カタカナ\n';
 563         $autolinkedDescription = hashtag_autolink($rawDescription, $index);
 564
 565         $this->assertContainsPolyfill($this->getHashtagLink('hashtag', $index), $autolinkedDescription);
 566         $this->assertNotContainsPolyfill(' #hashtag', $autolinkedDescription);
 567         $this->assertNotContainsPolyfill('>#nothashtag', $autolinkedDescription);
 568         $this->assertContainsPolyfill($this->getHashtagLink('ашок', $index), $autolinkedDescription);
 569         $this->assertContainsPolyfill($this->getHashtagLink('カタカナ', $index), $autolinkedDescription);
 570         $this->assertContainsPolyfill($this->getHashtagLink('hashtag_hashtag', $index), $autolinkedDescription);
 571         $this->assertNotContainsPolyfill($this->getHashtagLink('hashtag-nothashtag', $index), $autolinkedDescription);
 572     }
 573
 574     /**
 575      * Test hashtags auto-link without index URL.
 576      */
 577     public function testHashtagAutolinkNoIndex()
 578     {
 579         $rawDescription = 'blabla #hashtag x#nothashtag';
 580         $autolinkedDescription = hashtag_autolink($rawDescription);
 581
 582         $this->assertContainsPolyfill($this->getHashtagLink('hashtag'), $autolinkedDescription);
 583         $this->assertNotContainsPolyfill(' #hashtag', $autolinkedDescription);
 584         $this->assertNotContainsPolyfill('>#nothashtag', $autolinkedDescription);
 585     }
 586
 587     /**
 588      * Test is_note with note URLs.
 589      */
 590     public function testIsNote()
 591     {
 592         $this->assertTrue(is_note('?'));
 593         $this->assertTrue(is_note('?abcDEf'));
 594         $this->assertTrue(is_note('?_abcDEf#123'));
 595     }
 596
 597     /**
 598      * Test is_note with non note URLs.
 599      */
 600     public function testIsNotNote()
 601     {
 602         $this->assertFalse(is_note(''));
 603         $this->assertFalse(is_note('nope'));
 604         $this->assertFalse(is_note('https://github.com/shaarli/Shaarli/?hi'));
 605     }
 606
 607     /**
 608      * Util function to build an hashtag link.
 609      *
 610      * @param string $hashtag Hashtag name.
 611      * @param string $index Index URL.
 612      *
 613      * @return string HTML hashtag link.
 614      */
 615     private function getHashtagLink($hashtag, $index = '')
 616     {
 617         $hashtagLink = '<a href="' . $index . './add-tag/$1" title="Hashtag $1">#$1</a>';
 618         return str_replace('$1', $hashtag, $hashtagLink);
 619     }
 620 }