tests/Wallabag/CoreBundle/Helper/ContentProxyTest.php

   1 <?php
   2
   3 namespace Tests\Wallabag\CoreBundle\Helper;
   4
   5 use Psr\Log\NullLogger;
   6 use Wallabag\CoreBundle\Helper\ContentProxy;
   7 use Wallabag\CoreBundle\Entity\Entry;
   8 use Wallabag\CoreBundle\Entity\Tag;
   9 use Wallabag\UserBundle\Entity\User;
  10 use Wallabag\CoreBundle\Helper\RuleBasedTagger;
  11 use Graby\Graby;
  12 use Monolog\Handler\TestHandler;
  13 use Monolog\Logger;
  14
  15 class ContentProxyTest extends \PHPUnit_Framework_TestCase
  16 {
  17     private $fetchingErrorMessage = 'wallabag can\'t retrieve contents for this article. Please <a href="http://doc.wallabag.org/en/user/errors_during_fetching.html#how-can-i-help-to-fix-that">troubleshoot this issue</a>.';
  18
  19     public function testWithBadUrl()
  20     {
  21         $tagger = $this->getTaggerMock();
  22         $tagger->expects($this->once())
  23             ->method('tag');
  24
  25         $graby = $this->getMockBuilder('Graby\Graby')
  26             ->setMethods(['fetchContent'])
  27             ->disableOriginalConstructor()
  28             ->getMock();
  29
  30         $graby->expects($this->any())
  31             ->method('fetchContent')
  32             ->willReturn([
  33                 'html' => false,
  34                 'title' => '',
  35                 'url' => '',
  36                 'content_type' => '',
  37                 'language' => '',
  38             ]);
  39
  40         $proxy = new ContentProxy($graby, $tagger, $this->getLogger(), $this->fetchingErrorMessage);
  41         $entry = new Entry(new User());
  42         $proxy->updateEntry($entry, 'http://user@:80');
  43
  44         $this->assertEquals('http://user@:80', $entry->getUrl());
  45         $this->assertEmpty($entry->getTitle());
  46         $this->assertEquals($this->fetchingErrorMessage, $entry->getContent());
  47         $this->assertEmpty($entry->getPreviewPicture());
  48         $this->assertEmpty($entry->getMimetype());
  49         $this->assertEmpty($entry->getLanguage());
  50         $this->assertEquals(0.0, $entry->getReadingTime());
  51         $this->assertEquals(false, $entry->getDomainName());
  52     }
  53
  54     public function testWithEmptyContent()
  55     {
  56         $tagger = $this->getTaggerMock();
  57         $tagger->expects($this->once())
  58             ->method('tag');
  59
  60         $graby = $this->getMockBuilder('Graby\Graby')
  61             ->setMethods(['fetchContent'])
  62             ->disableOriginalConstructor()
  63             ->getMock();
  64
  65         $graby->expects($this->any())
  66             ->method('fetchContent')
  67             ->willReturn([
  68                 'html' => false,
  69                 'title' => '',
  70                 'url' => '',
  71                 'content_type' => '',
  72                 'language' => '',
  73             ]);
  74
  75         $proxy = new ContentProxy($graby, $tagger, $this->getLogger(), $this->fetchingErrorMessage);
  76         $entry = new Entry(new User());
  77         $proxy->updateEntry($entry, 'http://0.0.0.0');
  78
  79         $this->assertEquals('http://0.0.0.0', $entry->getUrl());
  80         $this->assertEmpty($entry->getTitle());
  81         $this->assertEquals($this->fetchingErrorMessage, $entry->getContent());
  82         $this->assertEmpty($entry->getPreviewPicture());
  83         $this->assertEmpty($entry->getMimetype());
  84         $this->assertEmpty($entry->getLanguage());
  85         $this->assertEquals(0.0, $entry->getReadingTime());
  86         $this->assertEquals('0.0.0.0', $entry->getDomainName());
  87     }
  88
  89     public function testWithEmptyContentButOG()
  90     {
  91         $tagger = $this->getTaggerMock();
  92         $tagger->expects($this->once())
  93             ->method('tag');
  94
  95         $graby = $this->getMockBuilder('Graby\Graby')
  96             ->setMethods(['fetchContent'])
  97             ->disableOriginalConstructor()
  98             ->getMock();
  99
 100         $graby->expects($this->any())
 101             ->method('fetchContent')
 102             ->willReturn([
 103                 'html' => false,
 104                 'title' => '',
 105                 'url' => '',
 106                 'content_type' => '',
 107                 'language' => '',
 108                 'status' => '',
 109                 'open_graph' => [
 110                     'og_title' => 'my title',
 111                     'og_description' => 'desc',
 112                 ],
 113             ]);
 114
 115         $proxy = new ContentProxy($graby, $tagger, $this->getLogger(), $this->fetchingErrorMessage);
 116         $entry = new Entry(new User());
 117         $proxy->updateEntry($entry, 'http://domain.io');
 118
 119         $this->assertEquals('http://domain.io', $entry->getUrl());
 120         $this->assertEquals('my title', $entry->getTitle());
 121         $this->assertEquals($this->fetchingErrorMessage.'<p><i>But we found a short description: </i></p>desc', $entry->getContent());
 122         $this->assertEmpty($entry->getPreviewPicture());
 123         $this->assertEmpty($entry->getLanguage());
 124         $this->assertEmpty($entry->getHttpStatus());
 125         $this->assertEmpty($entry->getMimetype());
 126         $this->assertEquals(0.0, $entry->getReadingTime());
 127         $this->assertEquals('domain.io', $entry->getDomainName());
 128     }
 129
 130     public function testWithContent()
 131     {
 132         $tagger = $this->getTaggerMock();
 133         $tagger->expects($this->once())
 134             ->method('tag');
 135
 136         $graby = $this->getMockBuilder('Graby\Graby')
 137             ->setMethods(['fetchContent'])
 138             ->disableOriginalConstructor()
 139             ->getMock();
 140
 141         $graby->expects($this->any())
 142             ->method('fetchContent')
 143             ->willReturn([
 144                 'html' => str_repeat('this is my content', 325),
 145                 'title' => 'this is my title',
 146                 'url' => 'http://1.1.1.1',
 147                 'content_type' => 'text/html',
 148                 'language' => 'fr',
 149                 'status' => '200',
 150                 'open_graph' => [
 151                     'og_title' => 'my OG title',
 152                     'og_description' => 'OG desc',
 153                     'og_image' => 'http://3.3.3.3/cover.jpg',
 154                 ],
 155             ]);
 156
 157         $proxy = new ContentProxy($graby, $tagger, $this->getLogger(), $this->fetchingErrorMessage);
 158         $entry = new Entry(new User());
 159         $proxy->updateEntry($entry, 'http://0.0.0.0');
 160
 161         $this->assertEquals('http://1.1.1.1', $entry->getUrl());
 162         $this->assertEquals('this is my title', $entry->getTitle());
 163         $this->assertContains('this is my content', $entry->getContent());
 164         $this->assertEquals('http://3.3.3.3/cover.jpg', $entry->getPreviewPicture());
 165         $this->assertEquals('text/html', $entry->getMimetype());
 166         $this->assertEquals('fr', $entry->getLanguage());
 167         $this->assertEquals('200', $entry->getHttpStatus());
 168         $this->assertEquals(4.0, $entry->getReadingTime());
 169         $this->assertEquals('1.1.1.1', $entry->getDomainName());
 170     }
 171
 172     public function testWithContentAndNoOgImage()
 173     {
 174         $tagger = $this->getTaggerMock();
 175         $tagger->expects($this->once())
 176             ->method('tag');
 177
 178         $graby = $this->getMockBuilder('Graby\Graby')
 179             ->setMethods(['fetchContent'])
 180             ->disableOriginalConstructor()
 181             ->getMock();
 182
 183         $graby->expects($this->any())
 184             ->method('fetchContent')
 185             ->willReturn([
 186                 'html' => str_repeat('this is my content', 325),
 187                 'title' => 'this is my title',
 188                 'url' => 'http://1.1.1.1',
 189                 'content_type' => 'text/html',
 190                 'language' => 'fr',
 191                 'status' => '200',
 192                 'open_graph' => [
 193                     'og_title' => 'my OG title',
 194                     'og_description' => 'OG desc',
 195                     'og_image' => false,
 196                 ],
 197             ]);
 198
 199         $proxy = new ContentProxy($graby, $tagger, $this->getLogger(), $this->fetchingErrorMessage);
 200         $entry = $proxy->updateEntry(new Entry(new User()), 'http://0.0.0.0');
 201
 202         $this->assertEquals('http://1.1.1.1', $entry->getUrl());
 203         $this->assertEquals('this is my title', $entry->getTitle());
 204         $this->assertContains('this is my content', $entry->getContent());
 205         $this->assertNull($entry->getPreviewPicture());
 206         $this->assertEquals('text/html', $entry->getMimetype());
 207         $this->assertEquals('fr', $entry->getLanguage());
 208         $this->assertEquals('200', $entry->getHttpStatus());
 209         $this->assertEquals(4.0, $entry->getReadingTime());
 210         $this->assertEquals('1.1.1.1', $entry->getDomainName());
 211     }
 212
 213     public function testWithForcedContent()
 214     {
 215         $tagger = $this->getTaggerMock();
 216         $tagger->expects($this->once())
 217             ->method('tag');
 218
 219         $proxy = new ContentProxy((new Graby()), $tagger, $this->getLogger(), $this->fetchingErrorMessage);
 220         $entry = new Entry(new User());
 221         $proxy->updateEntry(
 222             $entry,
 223             'http://0.0.0.0',
 224             [
 225                 'html' => str_repeat('this is my content', 325),
 226                 'title' => 'this is my title',
 227                 'url' => 'http://1.1.1.1',
 228                 'content_type' => 'text/html',
 229                 'language' => 'fr',
 230                 'date' => '1395635872',
 231                 'authors' => ['Jeremy', 'Nico', 'Thomas'],
 232                 'all_headers' => [
 233                     'Cache-Control' => 'no-cache',
 234                 ],
 235             ]
 236         );
 237
 238         $this->assertEquals('http://1.1.1.1', $entry->getUrl());
 239         $this->assertEquals('this is my title', $entry->getTitle());
 240         $this->assertContains('this is my content', $entry->getContent());
 241         $this->assertEquals('text/html', $entry->getMimetype());
 242         $this->assertEquals('fr', $entry->getLanguage());
 243         $this->assertEquals(4.0, $entry->getReadingTime());
 244         $this->assertEquals('1.1.1.1', $entry->getDomainName());
 245         $this->assertEquals('24/03/2014', $entry->getPublishedAt()->format('d/m/Y'));
 246         $this->assertContains('Jeremy', $entry->getPublishedBy());
 247         $this->assertContains('Nico', $entry->getPublishedBy());
 248         $this->assertContains('Thomas', $entry->getPublishedBy());
 249         $this->assertContains('no-cache', $entry->getHeaders());
 250     }
 251
 252     public function testWithForcedContentAndDatetime()
 253     {
 254         $tagger = $this->getTaggerMock();
 255         $tagger->expects($this->once())
 256             ->method('tag');
 257
 258         $proxy = new ContentProxy((new Graby()), $tagger, $this->getLogger(), $this->fetchingErrorMessage);
 259         $entry = new Entry(new User());
 260         $proxy->updateEntry(
 261             $entry,
 262             'http://0.0.0.0',
 263             [
 264                 'html' => str_repeat('this is my content', 325),
 265                 'title' => 'this is my title',
 266                 'url' => 'http://1.1.1.1',
 267                 'content_type' => 'text/html',
 268                 'language' => 'fr',
 269                 'date' => '2016-09-08T11:55:58+0200',
 270             ]
 271         );
 272
 273         $this->assertEquals('http://1.1.1.1', $entry->getUrl());
 274         $this->assertEquals('this is my title', $entry->getTitle());
 275         $this->assertContains('this is my content', $entry->getContent());
 276         $this->assertEquals('text/html', $entry->getMimetype());
 277         $this->assertEquals('fr', $entry->getLanguage());
 278         $this->assertEquals(4.0, $entry->getReadingTime());
 279         $this->assertEquals('1.1.1.1', $entry->getDomainName());
 280         $this->assertEquals('08/09/2016', $entry->getPublishedAt()->format('d/m/Y'));
 281     }
 282
 283     public function testWithForcedContentAndBadDate()
 284     {
 285         $tagger = $this->getTaggerMock();
 286         $tagger->expects($this->once())
 287             ->method('tag');
 288
 289         $logger = new Logger('foo');
 290         $handler = new TestHandler();
 291         $logger->pushHandler($handler);
 292
 293         $proxy = new ContentProxy((new Graby()), $tagger, $logger, $this->fetchingErrorMessage);
 294         $entry = new Entry(new User());
 295         $proxy->updateEntry(
 296             $entry,
 297             'http://0.0.0.0',
 298             [
 299                 'html' => str_repeat('this is my content', 325),
 300                 'title' => 'this is my title',
 301                 'url' => 'http://1.1.1.1',
 302                 'content_type' => 'text/html',
 303                 'language' => 'fr',
 304                 'date' => '01 02 2012',
 305             ]
 306         );
 307
 308         $this->assertEquals('http://1.1.1.1', $entry->getUrl());
 309         $this->assertEquals('this is my title', $entry->getTitle());
 310         $this->assertContains('this is my content', $entry->getContent());
 311         $this->assertEquals('text/html', $entry->getMimetype());
 312         $this->assertEquals('fr', $entry->getLanguage());
 313         $this->assertEquals(4.0, $entry->getReadingTime());
 314         $this->assertEquals('1.1.1.1', $entry->getDomainName());
 315         $this->assertNull($entry->getPublishedAt());
 316
 317         $records = $handler->getRecords();
 318
 319         $this->assertCount(1, $records);
 320         $this->assertContains('Error while defining date', $records[0]['message']);
 321     }
 322
 323     public function testTaggerThrowException()
 324     {
 325         $graby = $this->getMockBuilder('Graby\Graby')
 326             ->disableOriginalConstructor()
 327             ->getMock();
 328
 329         $tagger = $this->getTaggerMock();
 330         $tagger->expects($this->once())
 331             ->method('tag')
 332             ->will($this->throwException(new \Exception()));
 333
 334         $proxy = new ContentProxy($graby, $tagger, $this->getLogger(), $this->fetchingErrorMessage);
 335
 336         $entry = new Entry(new User());
 337         $proxy->updateEntry($entry, 'http://0.0.0.0', [
 338             'html' => str_repeat('this is my content', 325),
 339             'title' => 'this is my title',
 340             'url' => 'http://1.1.1.1',
 341             'content_type' => 'text/html',
 342             'language' => 'fr',
 343         ]);
 344
 345         $this->assertCount(0, $entry->getTags());
 346     }
 347
 348     public function dataForCrazyHtml()
 349     {
 350         return [
 351             'script and comment' => [
 352                 '<strong>Script inside:</strong> <!--[if gte IE 4]><script>alert(\'lol\');</script><![endif]--><br />',
 353                 'lol',
 354             ],
 355             'script' => [
 356                 '<strong>Script inside:</strong><script>alert(\'lol\');</script>',
 357                 'script',
 358             ],
 359         ];
 360     }
 361
 362     /**
 363      * @dataProvider dataForCrazyHtml
 364      */
 365     public function testWithCrazyHtmlContent($html, $escapedString)
 366     {
 367         $tagger = $this->getTaggerMock();
 368         $tagger->expects($this->once())
 369             ->method('tag');
 370
 371         $proxy = new ContentProxy((new Graby()), $tagger, $this->getLogger(), $this->fetchingErrorMessage);
 372         $entry = $proxy->updateEntry(
 373             new Entry(new User()),
 374             'http://1.1.1.1',
 375             [
 376                 'html' => $html,
 377                 'title' => 'this is my title',
 378                 'url' => 'http://1.1.1.1',
 379                 'content_type' => 'text/html',
 380                 'language' => 'fr',
 381                 'status' => '200',
 382                 'open_graph' => [
 383                     'og_title' => 'my OG title',
 384                     'og_description' => 'OG desc',
 385                     'og_image' => 'http://3.3.3.3/cover.jpg',
 386                 ],
 387             ]
 388         );
 389
 390         $this->assertEquals('http://1.1.1.1', $entry->getUrl());
 391         $this->assertEquals('this is my title', $entry->getTitle());
 392         $this->assertNotContains($escapedString, $entry->getContent());
 393         $this->assertEquals('http://3.3.3.3/cover.jpg', $entry->getPreviewPicture());
 394         $this->assertEquals('text/html', $entry->getMimetype());
 395         $this->assertEquals('fr', $entry->getLanguage());
 396         $this->assertEquals('200', $entry->getHttpStatus());
 397         $this->assertEquals('1.1.1.1', $entry->getDomainName());
 398     }
 399
 400     private function getTaggerMock()
 401     {
 402         return $this->getMockBuilder(RuleBasedTagger::class)
 403             ->setMethods(['tag'])
 404             ->disableOriginalConstructor()
 405             ->getMock();
 406     }
 407
 408     private function getLogger()
 409     {
 410         return new NullLogger();
 411     }
 412 }