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 not found.
  47      */
  48     public function testHeadersExtractNonExistentCharset()
  49     {
  50         $headers = '';
  51         $this->assertFalse(header_extract_charset($headers));
  52
  53         $headers = 'text/html';
  54         $this->assertFalse(header_extract_charset($headers));
  55     }
  56
  57     /**
  58      * Test html_extract_charset() when the charset is found.
  59      */
  60     public function testHtmlExtractExistentCharset()
  61     {
  62         $charset = 'x-MacCroatian';
  63         $html = '<html><meta>stuff2</meta><meta charset="' . $charset . '"/></html>';
  64         $this->assertEquals(strtolower($charset), html_extract_charset($html));
  65     }
  66
  67     /**
  68      * Test html_extract_charset() when the charset is not found.
  69      */
  70     public function testHtmlExtractNonExistentCharset()
  71     {
  72         $html = '<html><meta>stuff</meta></html>';
  73         $this->assertFalse(html_extract_charset($html));
  74         $html = '<html><meta>stuff</meta><meta charset=""/></html>';
  75         $this->assertFalse(html_extract_charset($html));
  76     }
  77
  78     /**
  79      * Test html_extract_tag() when the tag <meta name= is found.
  80      */
  81     public function testHtmlExtractExistentNameTag()
  82     {
  83         $description = 'Bob and Alice share cookies.';
  84         $html = '<html><meta>stuff2</meta><meta name="description" content="' . $description . '"/></html>';
  85         $this->assertEquals($description, html_extract_tag('description', $html));
  86     }
  87
  88     /**
  89      * Test html_extract_tag() when the tag <meta name= is not found.
  90      */
  91     public function testHtmlExtractNonExistentNameTag()
  92     {
  93         $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>';
  94         $this->assertFalse(html_extract_tag('description', $html));
  95     }
  96
  97     /**
  98      * Test html_extract_tag() when the tag <meta property="og: is found.
  99      */
 100     public function testHtmlExtractExistentOgTag()
 101     {
 102         $description = 'Bob and Alice share cookies.';
 103         $html = '<html><meta>stuff2</meta><meta property="og:description" content="' . $description . '"/></html>';
 104         $this->assertEquals($description, html_extract_tag('description', $html));
 105     }
 106
 107     /**
 108      * Test html_extract_tag() when the tag <meta property="og: is not found.
 109      */
 110     public function testHtmlExtractNonExistentOgTag()
 111     {
 112         $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>';
 113         $this->assertFalse(html_extract_tag('description', $html));
 114     }
 115
 116     /**
 117      * Test the download callback with valid value
 118      */
 119     public function testCurlDownloadCallbackOk()
 120     {
 121         $callback = get_curl_download_callback(
 122             $charset,
 123             $title,
 124             $desc,
 125             $keywords,
 126             false,
 127             'ut_curl_getinfo_ok'
 128         );
 129         $data = [
 130             'HTTP/1.1 200 OK',
 131             'Server: GitHub.com',
 132             'Date: Sat, 28 Oct 2017 12:01:33 GMT',
 133             'Content-Type: text/html; charset=utf-8',
 134             'Status: 200 OK',
 135             'end' => 'th=device-width">'
 136                 . '<title>Refactoring · GitHub</title>'
 137                 . '<link rel="search" type="application/opensea',
 138             '<title>ignored</title>'
 139                 . '<meta name="description" content="desc" />'
 140                 . '<meta name="keywords" content="key1,key2" />',
 141         ];
 142         foreach ($data as $key => $line) {
 143             $ignore = null;
 144             $expected = $key !== 'end' ? strlen($line) : false;
 145             $this->assertEquals($expected, $callback($ignore, $line));
 146             if ($expected === false) {
 147                 break;
 148             }
 149         }
 150         $this->assertEquals('utf-8', $charset);
 151         $this->assertEquals('Refactoring · GitHub', $title);
 152         $this->assertEmpty($desc);
 153         $this->assertEmpty($keywords);
 154     }
 155
 156     /**
 157      * Test the download callback with valid values and no charset
 158      */
 159     public function testCurlDownloadCallbackOkNoCharset()
 160     {
 161         $callback = get_curl_download_callback(
 162             $charset,
 163             $title,
 164             $desc,
 165             $keywords,
 166             false,
 167             'ut_curl_getinfo_no_charset'
 168         );
 169         $data = [
 170             'HTTP/1.1 200 OK',
 171             'end' => 'th=device-width">'
 172                 . '<title>Refactoring · GitHub</title>'
 173                 . '<link rel="search" type="application/opensea',
 174             '<title>ignored</title>'
 175             . '<meta name="description" content="desc" />'
 176             . '<meta name="keywords" content="key1,key2" />',
 177         ];
 178         foreach ($data as $key => $line) {
 179             $ignore = null;
 180             $this->assertEquals(strlen($line), $callback($ignore, $line));
 181         }
 182         $this->assertEmpty($charset);
 183         $this->assertEquals('Refactoring · GitHub', $title);
 184         $this->assertEmpty($desc);
 185         $this->assertEmpty($keywords);
 186     }
 187
 188     /**
 189      * Test the download callback with valid values and no charset
 190      */
 191     public function testCurlDownloadCallbackOkHtmlCharset()
 192     {
 193         $callback = get_curl_download_callback(
 194             $charset,
 195             $title,
 196             $desc,
 197             $keywords,
 198             false,
 199             'ut_curl_getinfo_no_charset'
 200         );
 201         $data = [
 202             'HTTP/1.1 200 OK',
 203             '<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />',
 204             'end' => 'th=device-width">'
 205                 . '<title>Refactoring · GitHub</title>'
 206                 . '<link rel="search" type="application/opensea',
 207             '<title>ignored</title>'
 208             . '<meta name="description" content="desc" />'
 209             . '<meta name="keywords" content="key1,key2" />',
 210         ];
 211         foreach ($data as $key => $line) {
 212             $ignore = null;
 213             $expected = $key !== 'end' ? strlen($line) : false;
 214             $this->assertEquals($expected, $callback($ignore, $line));
 215             if ($expected === false) {
 216                 break;
 217             }
 218         }
 219         $this->assertEquals('utf-8', $charset);
 220         $this->assertEquals('Refactoring · GitHub', $title);
 221         $this->assertEmpty($desc);
 222         $this->assertEmpty($keywords);
 223     }
 224
 225     /**
 226      * Test the download callback with valid values and no title
 227      */
 228     public function testCurlDownloadCallbackOkNoTitle()
 229     {
 230         $callback = get_curl_download_callback(
 231             $charset,
 232             $title,
 233             $desc,
 234             $keywords,
 235             false,
 236             'ut_curl_getinfo_ok'
 237         );
 238         $data = [
 239             'HTTP/1.1 200 OK',
 240             'end' => 'th=device-width">Refactoring · GitHub<link rel="search" type="application/opensea',
 241             'ignored',
 242         ];
 243         foreach ($data as $key => $line) {
 244             $ignore = null;
 245             $this->assertEquals(strlen($line), $callback($ignore, $line));
 246         }
 247         $this->assertEquals('utf-8', $charset);
 248         $this->assertEmpty($title);
 249         $this->assertEmpty($desc);
 250         $this->assertEmpty($keywords);
 251     }
 252
 253     /**
 254      * Test the download callback with an invalid content type.
 255      */
 256     public function testCurlDownloadCallbackInvalidContentType()
 257     {
 258         $callback = get_curl_download_callback(
 259             $charset,
 260             $title,
 261             $desc,
 262             $keywords,
 263             false,
 264             'ut_curl_getinfo_ct_ko'
 265         );
 266         $ignore = null;
 267         $this->assertFalse($callback($ignore, ''));
 268         $this->assertEmpty($charset);
 269         $this->assertEmpty($title);
 270     }
 271
 272     /**
 273      * Test the download callback with an invalid response code.
 274      */
 275     public function testCurlDownloadCallbackInvalidResponseCode()
 276     {
 277         $callback = $callback = get_curl_download_callback(
 278             $charset,
 279             $title,
 280             $desc,
 281             $keywords,
 282             false,
 283             'ut_curl_getinfo_rc_ko'
 284         );
 285         $ignore = null;
 286         $this->assertFalse($callback($ignore, ''));
 287         $this->assertEmpty($charset);
 288         $this->assertEmpty($title);
 289     }
 290
 291     /**
 292      * Test the download callback with an invalid content type and response code.
 293      */
 294     public function testCurlDownloadCallbackInvalidContentTypeAndResponseCode()
 295     {
 296         $callback = $callback = get_curl_download_callback(
 297             $charset,
 298             $title,
 299             $desc,
 300             $keywords,
 301             false,
 302             'ut_curl_getinfo_rs_ct_ko'
 303         );
 304         $ignore = null;
 305         $this->assertFalse($callback($ignore, ''));
 306         $this->assertEmpty($charset);
 307         $this->assertEmpty($title);
 308     }
 309
 310     /**
 311      * Test the download callback with valid value, and retrieve_description option enabled.
 312      */
 313     public function testCurlDownloadCallbackOkWithDesc()
 314     {
 315         $callback = get_curl_download_callback(
 316             $charset,
 317             $title,
 318             $desc,
 319             $keywords,
 320             true,
 321             'ut_curl_getinfo_ok'
 322         );
 323         $data = [
 324             'HTTP/1.1 200 OK',
 325             'Server: GitHub.com',
 326             'Date: Sat, 28 Oct 2017 12:01:33 GMT',
 327             'Content-Type: text/html; charset=utf-8',
 328             'Status: 200 OK',
 329             'th=device-width">'
 330                 . '<title>Refactoring · GitHub</title>'
 331                 . '<link rel="search" type="application/opensea',
 332             'end' => '<title>ignored</title>'
 333             . '<meta name="description" content="link desc" />'
 334             . '<meta name="keywords" content="key1,key2" />',
 335         ];
 336         foreach ($data as $key => $line) {
 337             $ignore = null;
 338             $expected = $key !== 'end' ? strlen($line) : false;
 339             $this->assertEquals($expected, $callback($ignore, $line));
 340             if ($expected === false) {
 341                 break;
 342             }
 343         }
 344         $this->assertEquals('utf-8', $charset);
 345         $this->assertEquals('Refactoring · GitHub', $title);
 346         $this->assertEquals('link desc', $desc);
 347         $this->assertEquals('key1 key2', $keywords);
 348     }
 349
 350     /**
 351      * Test the download callback with valid value, and retrieve_description option enabled,
 352      * but no desc or keyword defined in the page.
 353      */
 354     public function testCurlDownloadCallbackOkWithDescNotFound()
 355     {
 356         $callback = get_curl_download_callback(
 357             $charset,
 358             $title,
 359             $desc,
 360             $keywords,
 361             true,
 362             'ut_curl_getinfo_ok'
 363         );
 364         $data = [
 365             'HTTP/1.1 200 OK',
 366             'Server: GitHub.com',
 367             'Date: Sat, 28 Oct 2017 12:01:33 GMT',
 368             'Content-Type: text/html; charset=utf-8',
 369             'Status: 200 OK',
 370             'th=device-width">'
 371                 . '<title>Refactoring · GitHub</title>'
 372                 . '<link rel="search" type="application/opensea',
 373             'end' => '<title>ignored</title>',
 374         ];
 375         foreach ($data as $key => $line) {
 376             $ignore = null;
 377             $expected = $key !== 'end' ? strlen($line) : false;
 378             $this->assertEquals($expected, $callback($ignore, $line));
 379             if ($expected === false) {
 380                 break;
 381             }
 382         }
 383         $this->assertEquals('utf-8', $charset);
 384         $this->assertEquals('Refactoring · GitHub', $title);
 385         $this->assertEmpty($desc);
 386         $this->assertEmpty($keywords);
 387     }
 388
 389     /**
 390      * Test text2clickable.
 391      */
 392     public function testText2clickable()
 393     {
 394         $text = 'stuff http://hello.there/is=someone#here otherstuff';
 395         $expectedText = 'stuff <a href="http://hello.there/is=someone#here">'
 396             . 'http://hello.there/is=someone#here</a> otherstuff';
 397         $processedText = text2clickable($text);
 398         $this->assertEquals($expectedText, $processedText);
 399
 400         $text = 'stuff http://hello.there/is=someone#here(please) otherstuff';
 401         $expectedText = 'stuff <a href="http://hello.there/is=someone#here(please)">'
 402             . 'http://hello.there/is=someone#here(please)</a> otherstuff';
 403         $processedText = text2clickable($text);
 404         $this->assertEquals($expectedText, $processedText);
 405
 406         $text = 'stuff http://hello.there/is=someone#here(please)&no otherstuff';
 407         $text = 'stuff http://hello.there/is=someone#here(please)&no otherstuff';
 408         $expectedText = 'stuff <a href="http://hello.there/is=someone#here(please)&no">'
 409             . 'http://hello.there/is=someone#here(please)&no</a> otherstuff';
 410         $processedText = text2clickable($text);
 411         $this->assertEquals($expectedText, $processedText);
 412     }
 413
 414     /**
 415      * Test testSpace2nbsp.
 416      */
 417     public function testSpace2nbsp()
 418     {
 419         $text = '  Are you   thrilled  by flags   ?' . PHP_EOL . ' Really?';
 420         $expectedText = '&nbsp; Are you &nbsp; thrilled &nbsp;by flags &nbsp; ?' . PHP_EOL . '&nbsp;Really?';
 421         $processedText = space2nbsp($text);
 422         $this->assertEquals($expectedText, $processedText);
 423     }
 424
 425     /**
 426      * Test hashtags auto-link.
 427      */
 428     public function testHashtagAutolink()
 429     {
 430         $index = 'http://domain.tld/';
 431         $rawDescription = '#hashtag\n
 432             # nothashtag\n
 433             test#nothashtag #hashtag \#nothashtag\n
 434             test #hashtag #hashtag test #hashtag.test\n
 435             #hashtag #hashtag-nothashtag #hashtag_hashtag\n
 436             What is #ашок anyway?\n
 437             カタカナ #カタカナ」カタカナ\n';
 438         $autolinkedDescription = hashtag_autolink($rawDescription, $index);
 439
 440         $this->assertContainsPolyfill($this->getHashtagLink('hashtag', $index), $autolinkedDescription);
 441         $this->assertNotContainsPolyfill(' #hashtag', $autolinkedDescription);
 442         $this->assertNotContainsPolyfill('>#nothashtag', $autolinkedDescription);
 443         $this->assertContainsPolyfill($this->getHashtagLink('ашок', $index), $autolinkedDescription);
 444         $this->assertContainsPolyfill($this->getHashtagLink('カタカナ', $index), $autolinkedDescription);
 445         $this->assertContainsPolyfill($this->getHashtagLink('hashtag_hashtag', $index), $autolinkedDescription);
 446         $this->assertNotContainsPolyfill($this->getHashtagLink('hashtag-nothashtag', $index), $autolinkedDescription);
 447     }
 448
 449     /**
 450      * Test hashtags auto-link without index URL.
 451      */
 452     public function testHashtagAutolinkNoIndex()
 453     {
 454         $rawDescription = 'blabla #hashtag x#nothashtag';
 455         $autolinkedDescription = hashtag_autolink($rawDescription);
 456
 457         $this->assertContainsPolyfill($this->getHashtagLink('hashtag'), $autolinkedDescription);
 458         $this->assertNotContainsPolyfill(' #hashtag', $autolinkedDescription);
 459         $this->assertNotContainsPolyfill('>#nothashtag', $autolinkedDescription);
 460     }
 461
 462     /**
 463      * Test is_note with note URLs.
 464      */
 465     public function testIsNote()
 466     {
 467         $this->assertTrue(is_note('?'));
 468         $this->assertTrue(is_note('?abcDEf'));
 469         $this->assertTrue(is_note('?_abcDEf#123'));
 470     }
 471
 472     /**
 473      * Test is_note with non note URLs.
 474      */
 475     public function testIsNotNote()
 476     {
 477         $this->assertFalse(is_note(''));
 478         $this->assertFalse(is_note('nope'));
 479         $this->assertFalse(is_note('https://github.com/shaarli/Shaarli/?hi'));
 480     }
 481
 482     /**
 483      * Util function to build an hashtag link.
 484      *
 485      * @param string $hashtag Hashtag name.
 486      * @param string $index Index URL.
 487      *
 488      * @return string HTML hashtag link.
 489      */
 490     private function getHashtagLink($hashtag, $index = '')
 491     {
 492         $hashtagLink = '<a href="' . $index . './add-tag/$1" title="Hashtag $1">#$1</a>';
 493         return str_replace('$1', $hashtag, $hashtagLink);
 494     }
 495 }