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         $html = '<html><meta>stuff2</meta><meta name="description" content="' . $description . '"/></html>';
  98         $this->assertEquals($description, html_extract_tag('description', $html));
  99     }
 100
 101     /**
 102      * Test html_extract_tag() when the tag <meta name= is not found.
 103      */
 104     public function testHtmlExtractNonExistentNameTag()
 105     {
 106         $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>';
 107         $this->assertFalse(html_extract_tag('description', $html));
 108     }
 109
 110     /**
 111      * Test html_extract_tag() when the tag <meta property="og: is found.
 112      */
 113     public function testHtmlExtractExistentOgTag()
 114     {
 115         $description = 'Bob and Alice share cookies.';
 116         $html = '<html><meta>stuff2</meta><meta property="og:description" content="' . $description . '"/></html>';
 117         $this->assertEquals($description, html_extract_tag('description', $html));
 118     }
 119
 120     /**
 121      * Test html_extract_tag() when the tag <meta property="og: is not found.
 122      */
 123     public function testHtmlExtractNonExistentOgTag()
 124     {
 125         $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>';
 126         $this->assertFalse(html_extract_tag('description', $html));
 127     }
 128
 129     /**
 130      * Test the download callback with valid value
 131      */
 132     public function testCurlDownloadCallbackOk()
 133     {
 134         $callback = get_curl_download_callback(
 135             $charset,
 136             $title,
 137             $desc,
 138             $keywords,
 139             false,
 140             'ut_curl_getinfo_ok'
 141         );
 142         $data = [
 143             'HTTP/1.1 200 OK',
 144             'Server: GitHub.com',
 145             'Date: Sat, 28 Oct 2017 12:01:33 GMT',
 146             'Content-Type: text/html; charset=utf-8',
 147             'Status: 200 OK',
 148             'end' => 'th=device-width">'
 149                 . '<title>Refactoring · GitHub</title>'
 150                 . '<link rel="search" type="application/opensea',
 151             '<title>ignored</title>'
 152                 . '<meta name="description" content="desc" />'
 153                 . '<meta name="keywords" content="key1,key2" />',
 154         ];
 155         foreach ($data as $key => $line) {
 156             $ignore = null;
 157             $expected = $key !== 'end' ? strlen($line) : false;
 158             $this->assertEquals($expected, $callback($ignore, $line));
 159             if ($expected === false) {
 160                 break;
 161             }
 162         }
 163         $this->assertEquals('utf-8', $charset);
 164         $this->assertEquals('Refactoring · GitHub', $title);
 165         $this->assertEmpty($desc);
 166         $this->assertEmpty($keywords);
 167     }
 168
 169     /**
 170      * Test the download callback with valid values and no charset
 171      */
 172     public function testCurlDownloadCallbackOkNoCharset()
 173     {
 174         $callback = get_curl_download_callback(
 175             $charset,
 176             $title,
 177             $desc,
 178             $keywords,
 179             false,
 180             'ut_curl_getinfo_no_charset'
 181         );
 182         $data = [
 183             'HTTP/1.1 200 OK',
 184             'end' => 'th=device-width">'
 185                 . '<title>Refactoring · GitHub</title>'
 186                 . '<link rel="search" type="application/opensea',
 187             '<title>ignored</title>'
 188             . '<meta name="description" content="desc" />'
 189             . '<meta name="keywords" content="key1,key2" />',
 190         ];
 191         foreach ($data as $key => $line) {
 192             $ignore = null;
 193             $this->assertEquals(strlen($line), $callback($ignore, $line));
 194         }
 195         $this->assertEmpty($charset);
 196         $this->assertEquals('Refactoring · GitHub', $title);
 197         $this->assertEmpty($desc);
 198         $this->assertEmpty($keywords);
 199     }
 200
 201     /**
 202      * Test the download callback with valid values and no charset
 203      */
 204     public function testCurlDownloadCallbackOkHtmlCharset()
 205     {
 206         $callback = get_curl_download_callback(
 207             $charset,
 208             $title,
 209             $desc,
 210             $keywords,
 211             false,
 212             'ut_curl_getinfo_no_charset'
 213         );
 214         $data = [
 215             'HTTP/1.1 200 OK',
 216             '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />',
 217             'end' => 'th=device-width">'
 218                 . '<title>Refactoring · GitHub</title>'
 219                 . '<link rel="search" type="application/opensea',
 220             '<title>ignored</title>'
 221             . '<meta name="description" content="desc" />'
 222             . '<meta name="keywords" content="key1,key2" />',
 223         ];
 224         foreach ($data as $key => $line) {
 225             $ignore = null;
 226             $expected = $key !== 'end' ? strlen($line) : false;
 227             $this->assertEquals($expected, $callback($ignore, $line));
 228             if ($expected === false) {
 229                 break;
 230             }
 231         }
 232         $this->assertEquals('utf-8', $charset);
 233         $this->assertEquals('Refactoring · GitHub', $title);
 234         $this->assertEmpty($desc);
 235         $this->assertEmpty($keywords);
 236     }
 237
 238     /**
 239      * Test the download callback with valid values and no title
 240      */
 241     public function testCurlDownloadCallbackOkNoTitle()
 242     {
 243         $callback = get_curl_download_callback(
 244             $charset,
 245             $title,
 246             $desc,
 247             $keywords,
 248             false,
 249             'ut_curl_getinfo_ok'
 250         );
 251         $data = [
 252             'HTTP/1.1 200 OK',
 253             'end' => 'th=device-width">Refactoring · GitHub<link rel="search" type="application/opensea',
 254             'ignored',
 255         ];
 256         foreach ($data as $key => $line) {
 257             $ignore = null;
 258             $this->assertEquals(strlen($line), $callback($ignore, $line));
 259         }
 260         $this->assertEquals('utf-8', $charset);
 261         $this->assertEmpty($title);
 262         $this->assertEmpty($desc);
 263         $this->assertEmpty($keywords);
 264     }
 265
 266     /**
 267      * Test the download callback with an invalid content type.
 268      */
 269     public function testCurlDownloadCallbackInvalidContentType()
 270     {
 271         $callback = get_curl_download_callback(
 272             $charset,
 273             $title,
 274             $desc,
 275             $keywords,
 276             false,
 277             'ut_curl_getinfo_ct_ko'
 278         );
 279         $ignore = null;
 280         $this->assertFalse($callback($ignore, ''));
 281         $this->assertEmpty($charset);
 282         $this->assertEmpty($title);
 283     }
 284
 285     /**
 286      * Test the download callback with an invalid response code.
 287      */
 288     public function testCurlDownloadCallbackInvalidResponseCode()
 289     {
 290         $callback = $callback = get_curl_download_callback(
 291             $charset,
 292             $title,
 293             $desc,
 294             $keywords,
 295             false,
 296             'ut_curl_getinfo_rc_ko'
 297         );
 298         $ignore = null;
 299         $this->assertFalse($callback($ignore, ''));
 300         $this->assertEmpty($charset);
 301         $this->assertEmpty($title);
 302     }
 303
 304     /**
 305      * Test the download callback with an invalid content type and response code.
 306      */
 307     public function testCurlDownloadCallbackInvalidContentTypeAndResponseCode()
 308     {
 309         $callback = $callback = get_curl_download_callback(
 310             $charset,
 311             $title,
 312             $desc,
 313             $keywords,
 314             false,
 315             'ut_curl_getinfo_rs_ct_ko'
 316         );
 317         $ignore = null;
 318         $this->assertFalse($callback($ignore, ''));
 319         $this->assertEmpty($charset);
 320         $this->assertEmpty($title);
 321     }
 322
 323     /**
 324      * Test the download callback with valid value, and retrieve_description option enabled.
 325      */
 326     public function testCurlDownloadCallbackOkWithDesc()
 327     {
 328         $callback = get_curl_download_callback(
 329             $charset,
 330             $title,
 331             $desc,
 332             $keywords,
 333             true,
 334             'ut_curl_getinfo_ok'
 335         );
 336         $data = [
 337             'HTTP/1.1 200 OK',
 338             'Server: GitHub.com',
 339             'Date: Sat, 28 Oct 2017 12:01:33 GMT',
 340             'Content-Type: text/html; charset=utf-8',
 341             'Status: 200 OK',
 342             'th=device-width">'
 343                 . '<title>Refactoring · GitHub</title>'
 344                 . '<link rel="search" type="application/opensea',
 345             'end' => '<title>ignored</title>'
 346             . '<meta name="description" content="link desc" />'
 347             . '<meta name="keywords" content="key1,key2" />',
 348         ];
 349         foreach ($data as $key => $line) {
 350             $ignore = null;
 351             $expected = $key !== 'end' ? strlen($line) : false;
 352             $this->assertEquals($expected, $callback($ignore, $line));
 353             if ($expected === false) {
 354                 break;
 355             }
 356         }
 357         $this->assertEquals('utf-8', $charset);
 358         $this->assertEquals('Refactoring · GitHub', $title);
 359         $this->assertEquals('link desc', $desc);
 360         $this->assertEquals('key1 key2', $keywords);
 361     }
 362
 363     /**
 364      * Test the download callback with valid value, and retrieve_description option enabled,
 365      * but no desc or keyword defined in the page.
 366      */
 367     public function testCurlDownloadCallbackOkWithDescNotFound()
 368     {
 369         $callback = get_curl_download_callback(
 370             $charset,
 371             $title,
 372             $desc,
 373             $keywords,
 374             true,
 375             'ut_curl_getinfo_ok'
 376         );
 377         $data = [
 378             'HTTP/1.1 200 OK',
 379             'Server: GitHub.com',
 380             'Date: Sat, 28 Oct 2017 12:01:33 GMT',
 381             'Content-Type: text/html; charset=utf-8',
 382             'Status: 200 OK',
 383             'th=device-width">'
 384                 . '<title>Refactoring · GitHub</title>'
 385                 . '<link rel="search" type="application/opensea',
 386             'end' => '<title>ignored</title>',
 387         ];
 388         foreach ($data as $key => $line) {
 389             $ignore = null;
 390             $expected = $key !== 'end' ? strlen($line) : false;
 391             $this->assertEquals($expected, $callback($ignore, $line));
 392             if ($expected === false) {
 393                 break;
 394             }
 395         }
 396         $this->assertEquals('utf-8', $charset);
 397         $this->assertEquals('Refactoring · GitHub', $title);
 398         $this->assertEmpty($desc);
 399         $this->assertEmpty($keywords);
 400     }
 401
 402     /**
 403      * Test text2clickable.
 404      */
 405     public function testText2clickable()
 406     {
 407         $text = 'stuff http://hello.there/is=someone#here otherstuff';
 408         $expectedText = 'stuff <a href="http://hello.there/is=someone#here">'
 409             . 'http://hello.there/is=someone#here</a> otherstuff';
 410         $processedText = text2clickable($text);
 411         $this->assertEquals($expectedText, $processedText);
 412
 413         $text = 'stuff http://hello.there/is=someone#here(please) otherstuff';
 414         $expectedText = 'stuff <a href="http://hello.there/is=someone#here(please)">'
 415             . 'http://hello.there/is=someone#here(please)</a> otherstuff';
 416         $processedText = text2clickable($text);
 417         $this->assertEquals($expectedText, $processedText);
 418
 419         $text = 'stuff http://hello.there/is=someone#here(please)&no otherstuff';
 420         $text = 'stuff http://hello.there/is=someone#here(please)&no otherstuff';
 421         $expectedText = 'stuff <a href="http://hello.there/is=someone#here(please)&no">'
 422             . 'http://hello.there/is=someone#here(please)&no</a> otherstuff';
 423         $processedText = text2clickable($text);
 424         $this->assertEquals($expectedText, $processedText);
 425     }
 426
 427     /**
 428      * Test testSpace2nbsp.
 429      */
 430     public function testSpace2nbsp()
 431     {
 432         $text = '  Are you   thrilled  by flags   ?' . PHP_EOL . ' Really?';
 433         $expectedText = '&nbsp; Are you &nbsp; thrilled &nbsp;by flags &nbsp; ?' . PHP_EOL . '&nbsp;Really?';
 434         $processedText = space2nbsp($text);
 435         $this->assertEquals($expectedText, $processedText);
 436     }
 437
 438     /**
 439      * Test hashtags auto-link.
 440      */
 441     public function testHashtagAutolink()
 442     {
 443         $index = 'http://domain.tld/';
 444         $rawDescription = '#hashtag\n
 445             # nothashtag\n
 446             test#nothashtag #hashtag \#nothashtag\n
 447             test #hashtag #hashtag test #hashtag.test\n
 448             #hashtag #hashtag-nothashtag #hashtag_hashtag\n
 449             What is #ашок anyway?\n
 450             カタカナ #カタカナ」カタカナ\n';
 451         $autolinkedDescription = hashtag_autolink($rawDescription, $index);
 452
 453         $this->assertContainsPolyfill($this->getHashtagLink('hashtag', $index), $autolinkedDescription);
 454         $this->assertNotContainsPolyfill(' #hashtag', $autolinkedDescription);
 455         $this->assertNotContainsPolyfill('>#nothashtag', $autolinkedDescription);
 456         $this->assertContainsPolyfill($this->getHashtagLink('ашок', $index), $autolinkedDescription);
 457         $this->assertContainsPolyfill($this->getHashtagLink('カタカナ', $index), $autolinkedDescription);
 458         $this->assertContainsPolyfill($this->getHashtagLink('hashtag_hashtag', $index), $autolinkedDescription);
 459         $this->assertNotContainsPolyfill($this->getHashtagLink('hashtag-nothashtag', $index), $autolinkedDescription);
 460     }
 461
 462     /**
 463      * Test hashtags auto-link without index URL.
 464      */
 465     public function testHashtagAutolinkNoIndex()
 466     {
 467         $rawDescription = 'blabla #hashtag x#nothashtag';
 468         $autolinkedDescription = hashtag_autolink($rawDescription);
 469
 470         $this->assertContainsPolyfill($this->getHashtagLink('hashtag'), $autolinkedDescription);
 471         $this->assertNotContainsPolyfill(' #hashtag', $autolinkedDescription);
 472         $this->assertNotContainsPolyfill('>#nothashtag', $autolinkedDescription);
 473     }
 474
 475     /**
 476      * Test is_note with note URLs.
 477      */
 478     public function testIsNote()
 479     {
 480         $this->assertTrue(is_note('?'));
 481         $this->assertTrue(is_note('?abcDEf'));
 482         $this->assertTrue(is_note('?_abcDEf#123'));
 483     }
 484
 485     /**
 486      * Test is_note with non note URLs.
 487      */
 488     public function testIsNotNote()
 489     {
 490         $this->assertFalse(is_note(''));
 491         $this->assertFalse(is_note('nope'));
 492         $this->assertFalse(is_note('https://github.com/shaarli/Shaarli/?hi'));
 493     }
 494
 495     /**
 496      * Util function to build an hashtag link.
 497      *
 498      * @param string $hashtag Hashtag name.
 499      * @param string $index Index URL.
 500      *
 501      * @return string HTML hashtag link.
 502      */
 503     private function getHashtagLink($hashtag, $index = '')
 504     {
 505         $hashtagLink = '<a href="' . $index . './add-tag/$1" title="Hashtag $1">#$1</a>';
 506         return str_replace('$1', $hashtag, $hashtagLink);
 507     }
 508 }