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