]> git.immae.eu Git - github/wallabag/wallabag.git/blob - tests/Wallabag/CoreBundle/Helper/ContentProxyTest.php
Merge pull request #3398 from wallabag/issue-1735
[github/wallabag/wallabag.git] / tests / Wallabag / CoreBundle / Helper / ContentProxyTest.php
1 <?php
2
3 namespace Tests\Wallabag\CoreBundle\Helper;
4
5 use Graby\Graby;
6 use Monolog\Handler\TestHandler;
7 use Monolog\Logger;
8 use Psr\Log\NullLogger;
9 use Symfony\Component\Validator\ConstraintViolation;
10 use Symfony\Component\Validator\ConstraintViolationList;
11 use Symfony\Component\Validator\Validator\RecursiveValidator;
12 use Wallabag\CoreBundle\Entity\Entry;
13 use Wallabag\CoreBundle\Entity\Tag;
14 use Wallabag\CoreBundle\Helper\ContentProxy;
15 use Wallabag\CoreBundle\Helper\RuleBasedTagger;
16 use Wallabag\UserBundle\Entity\User;
17
18 class ContentProxyTest extends \PHPUnit_Framework_TestCase
19 {
20 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>.';
21
22 public function testWithBadUrl()
23 {
24 $tagger = $this->getTaggerMock();
25 $tagger->expects($this->once())
26 ->method('tag');
27
28 $graby = $this->getMockBuilder('Graby\Graby')
29 ->setMethods(['fetchContent'])
30 ->disableOriginalConstructor()
31 ->getMock();
32
33 $graby->expects($this->any())
34 ->method('fetchContent')
35 ->willReturn([
36 'html' => false,
37 'title' => '',
38 'url' => '',
39 'content_type' => '',
40 'language' => '',
41 ]);
42
43 $proxy = new ContentProxy($graby, $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
44 $entry = new Entry(new User());
45 $proxy->updateEntry($entry, 'http://user@:80');
46
47 $this->assertSame('http://user@:80', $entry->getUrl());
48 $this->assertEmpty($entry->getTitle());
49 $this->assertSame($this->fetchingErrorMessage, $entry->getContent());
50 $this->assertEmpty($entry->getPreviewPicture());
51 $this->assertEmpty($entry->getMimetype());
52 $this->assertEmpty($entry->getLanguage());
53 $this->assertSame(0.0, $entry->getReadingTime());
54 $this->assertNull($entry->getDomainName());
55 }
56
57 public function testWithEmptyContent()
58 {
59 $tagger = $this->getTaggerMock();
60 $tagger->expects($this->once())
61 ->method('tag');
62
63 $graby = $this->getMockBuilder('Graby\Graby')
64 ->setMethods(['fetchContent'])
65 ->disableOriginalConstructor()
66 ->getMock();
67
68 $graby->expects($this->any())
69 ->method('fetchContent')
70 ->willReturn([
71 'html' => false,
72 'title' => '',
73 'url' => '',
74 'content_type' => '',
75 'language' => '',
76 ]);
77
78 $proxy = new ContentProxy($graby, $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
79 $entry = new Entry(new User());
80 $proxy->updateEntry($entry, 'http://0.0.0.0');
81
82 $this->assertSame('http://0.0.0.0', $entry->getUrl());
83 $this->assertEmpty($entry->getTitle());
84 $this->assertSame($this->fetchingErrorMessage, $entry->getContent());
85 $this->assertEmpty($entry->getPreviewPicture());
86 $this->assertEmpty($entry->getMimetype());
87 $this->assertEmpty($entry->getLanguage());
88 $this->assertSame(0.0, $entry->getReadingTime());
89 $this->assertSame('0.0.0.0', $entry->getDomainName());
90 }
91
92 public function testWithEmptyContentButOG()
93 {
94 $tagger = $this->getTaggerMock();
95 $tagger->expects($this->once())
96 ->method('tag');
97
98 $graby = $this->getMockBuilder('Graby\Graby')
99 ->setMethods(['fetchContent'])
100 ->disableOriginalConstructor()
101 ->getMock();
102
103 $graby->expects($this->any())
104 ->method('fetchContent')
105 ->willReturn([
106 'html' => false,
107 'title' => '',
108 'url' => '',
109 'content_type' => '',
110 'language' => '',
111 'status' => '',
112 'open_graph' => [
113 'og_title' => 'my title',
114 'og_description' => 'desc',
115 ],
116 ]);
117
118 $proxy = new ContentProxy($graby, $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
119 $entry = new Entry(new User());
120 $proxy->updateEntry($entry, 'http://domain.io');
121
122 $this->assertSame('http://domain.io', $entry->getUrl());
123 $this->assertSame('my title', $entry->getTitle());
124 $this->assertSame($this->fetchingErrorMessage . '<p><i>But we found a short description: </i></p>desc', $entry->getContent());
125 $this->assertEmpty($entry->getPreviewPicture());
126 $this->assertEmpty($entry->getLanguage());
127 $this->assertEmpty($entry->getHttpStatus());
128 $this->assertEmpty($entry->getMimetype());
129 $this->assertSame(0.0, $entry->getReadingTime());
130 $this->assertSame('domain.io', $entry->getDomainName());
131 }
132
133 public function testWithContent()
134 {
135 $tagger = $this->getTaggerMock();
136 $tagger->expects($this->once())
137 ->method('tag');
138
139 $graby = $this->getMockBuilder('Graby\Graby')
140 ->setMethods(['fetchContent'])
141 ->disableOriginalConstructor()
142 ->getMock();
143
144 $graby->expects($this->any())
145 ->method('fetchContent')
146 ->willReturn([
147 'html' => str_repeat('this is my content', 325),
148 'title' => 'this is my title',
149 'url' => 'http://1.1.1.1',
150 'content_type' => 'text/html',
151 'language' => 'fr',
152 'status' => '200',
153 'open_graph' => [
154 'og_title' => 'my OG title',
155 'og_description' => 'OG desc',
156 'og_image' => 'http://3.3.3.3/cover.jpg',
157 ],
158 ]);
159
160 $proxy = new ContentProxy($graby, $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
161 $entry = new Entry(new User());
162 $proxy->updateEntry($entry, 'http://0.0.0.0');
163
164 $this->assertSame('http://1.1.1.1', $entry->getUrl());
165 $this->assertSame('this is my title', $entry->getTitle());
166 $this->assertContains('this is my content', $entry->getContent());
167 $this->assertSame('http://3.3.3.3/cover.jpg', $entry->getPreviewPicture());
168 $this->assertSame('text/html', $entry->getMimetype());
169 $this->assertSame('fr', $entry->getLanguage());
170 $this->assertSame('200', $entry->getHttpStatus());
171 $this->assertSame(4.0, $entry->getReadingTime());
172 $this->assertSame('1.1.1.1', $entry->getDomainName());
173 }
174
175 public function testWithContentAndNoOgImage()
176 {
177 $tagger = $this->getTaggerMock();
178 $tagger->expects($this->once())
179 ->method('tag');
180
181 $graby = $this->getMockBuilder('Graby\Graby')
182 ->setMethods(['fetchContent'])
183 ->disableOriginalConstructor()
184 ->getMock();
185
186 $graby->expects($this->any())
187 ->method('fetchContent')
188 ->willReturn([
189 'html' => str_repeat('this is my content', 325),
190 'title' => 'this is my title',
191 'url' => 'http://1.1.1.1',
192 'content_type' => 'text/html',
193 'language' => 'fr',
194 'status' => '200',
195 'open_graph' => [
196 'og_title' => 'my OG title',
197 'og_description' => 'OG desc',
198 'og_image' => null,
199 ],
200 ]);
201
202 $proxy = new ContentProxy($graby, $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
203 $entry = new Entry(new User());
204 $proxy->updateEntry($entry, 'http://0.0.0.0');
205
206 $this->assertSame('http://1.1.1.1', $entry->getUrl());
207 $this->assertSame('this is my title', $entry->getTitle());
208 $this->assertContains('this is my content', $entry->getContent());
209 $this->assertNull($entry->getPreviewPicture());
210 $this->assertSame('text/html', $entry->getMimetype());
211 $this->assertSame('fr', $entry->getLanguage());
212 $this->assertSame('200', $entry->getHttpStatus());
213 $this->assertSame(4.0, $entry->getReadingTime());
214 $this->assertSame('1.1.1.1', $entry->getDomainName());
215 }
216
217 public function testWithContentAndBadLanguage()
218 {
219 $tagger = $this->getTaggerMock();
220 $tagger->expects($this->once())
221 ->method('tag');
222
223 $validator = $this->getValidator();
224 $validator->expects($this->once())
225 ->method('validate')
226 ->willReturn(new ConstraintViolationList([new ConstraintViolation('oops', 'oops', [], 'oops', 'language', 'dontexist')]));
227
228 $graby = $this->getMockBuilder('Graby\Graby')
229 ->setMethods(['fetchContent'])
230 ->disableOriginalConstructor()
231 ->getMock();
232
233 $graby->expects($this->any())
234 ->method('fetchContent')
235 ->willReturn([
236 'html' => str_repeat('this is my content', 325),
237 'title' => 'this is my title',
238 'url' => 'http://1.1.1.1',
239 'content_type' => 'text/html',
240 'language' => 'dontexist',
241 'status' => '200',
242 ]);
243
244 $proxy = new ContentProxy($graby, $tagger, $validator, $this->getLogger(), $this->fetchingErrorMessage);
245 $entry = new Entry(new User());
246 $proxy->updateEntry($entry, 'http://0.0.0.0');
247
248 $this->assertSame('http://1.1.1.1', $entry->getUrl());
249 $this->assertSame('this is my title', $entry->getTitle());
250 $this->assertContains('this is my content', $entry->getContent());
251 $this->assertSame('text/html', $entry->getMimetype());
252 $this->assertNull($entry->getLanguage());
253 $this->assertSame('200', $entry->getHttpStatus());
254 $this->assertSame(4.0, $entry->getReadingTime());
255 $this->assertSame('1.1.1.1', $entry->getDomainName());
256 }
257
258 public function testWithContentAndBadOgImage()
259 {
260 $tagger = $this->getTaggerMock();
261 $tagger->expects($this->once())
262 ->method('tag');
263
264 $validator = $this->getValidator();
265 $validator->expects($this->exactly(2))
266 ->method('validate')
267 ->will($this->onConsecutiveCalls(
268 new ConstraintViolationList(),
269 new ConstraintViolationList([new ConstraintViolation('oops', 'oops', [], 'oops', 'url', 'https://')])
270 ));
271
272 $graby = $this->getMockBuilder('Graby\Graby')
273 ->setMethods(['fetchContent'])
274 ->disableOriginalConstructor()
275 ->getMock();
276
277 $graby->expects($this->any())
278 ->method('fetchContent')
279 ->willReturn([
280 'html' => str_repeat('this is my content', 325),
281 'title' => 'this is my title',
282 'url' => 'http://1.1.1.1',
283 'content_type' => 'text/html',
284 'language' => 'fr',
285 'status' => '200',
286 'open_graph' => [
287 'og_title' => 'my OG title',
288 'og_description' => 'OG desc',
289 'og_image' => 'https://',
290 ],
291 ]);
292
293 $proxy = new ContentProxy($graby, $tagger, $validator, $this->getLogger(), $this->fetchingErrorMessage);
294 $entry = new Entry(new User());
295 $proxy->updateEntry($entry, 'http://0.0.0.0');
296
297 $this->assertSame('http://1.1.1.1', $entry->getUrl());
298 $this->assertSame('this is my title', $entry->getTitle());
299 $this->assertContains('this is my content', $entry->getContent());
300 $this->assertNull($entry->getPreviewPicture());
301 $this->assertSame('text/html', $entry->getMimetype());
302 $this->assertSame('fr', $entry->getLanguage());
303 $this->assertSame('200', $entry->getHttpStatus());
304 $this->assertSame(4.0, $entry->getReadingTime());
305 $this->assertSame('1.1.1.1', $entry->getDomainName());
306 }
307
308 public function testWithForcedContent()
309 {
310 $tagger = $this->getTaggerMock();
311 $tagger->expects($this->once())
312 ->method('tag');
313
314 $proxy = new ContentProxy((new Graby()), $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage, true);
315 $entry = new Entry(new User());
316 $proxy->updateEntry(
317 $entry,
318 'http://0.0.0.0',
319 [
320 'html' => str_repeat('this is my content', 325),
321 'title' => 'this is my title',
322 'url' => 'http://1.1.1.1',
323 'content_type' => 'text/html',
324 'language' => 'fr',
325 'date' => '1395635872',
326 'authors' => ['Jeremy', 'Nico', 'Thomas'],
327 'all_headers' => [
328 'Cache-Control' => 'no-cache',
329 ],
330 ]
331 );
332
333 $this->assertSame('http://1.1.1.1', $entry->getUrl());
334 $this->assertSame('this is my title', $entry->getTitle());
335 $this->assertContains('this is my content', $entry->getContent());
336 $this->assertSame('text/html', $entry->getMimetype());
337 $this->assertSame('fr', $entry->getLanguage());
338 $this->assertSame(4.0, $entry->getReadingTime());
339 $this->assertSame('1.1.1.1', $entry->getDomainName());
340 $this->assertSame('24/03/2014', $entry->getPublishedAt()->format('d/m/Y'));
341 $this->assertContains('Jeremy', $entry->getPublishedBy());
342 $this->assertContains('Nico', $entry->getPublishedBy());
343 $this->assertContains('Thomas', $entry->getPublishedBy());
344 $this->assertNotNull($entry->getHeaders(), 'Headers are stored, so value is not null');
345 $this->assertContains('no-cache', $entry->getHeaders());
346 }
347
348 public function testWithForcedContentAndDatetime()
349 {
350 $tagger = $this->getTaggerMock();
351 $tagger->expects($this->once())
352 ->method('tag');
353
354 $logHandler = new TestHandler();
355 $logger = new Logger('test', [$logHandler]);
356
357 $proxy = new ContentProxy((new Graby()), $tagger, $this->getValidator(), $logger, $this->fetchingErrorMessage);
358 $entry = new Entry(new User());
359 $proxy->updateEntry(
360 $entry,
361 'http://1.1.1.1',
362 [
363 'html' => str_repeat('this is my content', 325),
364 'title' => 'this is my title',
365 'url' => 'http://1.1.1.1',
366 'content_type' => 'text/html',
367 'language' => 'fr',
368 'date' => '2016-09-08T11:55:58+0200',
369 ]
370 );
371
372 $this->assertSame('http://1.1.1.1', $entry->getUrl());
373 $this->assertSame('this is my title', $entry->getTitle());
374 $this->assertContains('this is my content', $entry->getContent());
375 $this->assertSame('text/html', $entry->getMimetype());
376 $this->assertSame('fr', $entry->getLanguage());
377 $this->assertSame(4.0, $entry->getReadingTime());
378 $this->assertSame('1.1.1.1', $entry->getDomainName());
379 $this->assertSame('08/09/2016', $entry->getPublishedAt()->format('d/m/Y'));
380 }
381
382 public function testWithForcedContentAndBadDate()
383 {
384 $tagger = $this->getTaggerMock();
385 $tagger->expects($this->once())
386 ->method('tag');
387
388 $logger = new Logger('foo');
389 $handler = new TestHandler();
390 $logger->pushHandler($handler);
391
392 $proxy = new ContentProxy((new Graby()), $tagger, $this->getValidator(), $logger, $this->fetchingErrorMessage);
393 $entry = new Entry(new User());
394 $proxy->updateEntry(
395 $entry,
396 'http://1.1.1.1',
397 [
398 'html' => str_repeat('this is my content', 325),
399 'title' => 'this is my title',
400 'url' => 'http://1.1.1.1',
401 'content_type' => 'text/html',
402 'language' => 'fr',
403 'date' => '01 02 2012',
404 ]
405 );
406
407 $this->assertSame('http://1.1.1.1', $entry->getUrl());
408 $this->assertSame('this is my title', $entry->getTitle());
409 $this->assertContains('this is my content', $entry->getContent());
410 $this->assertSame('text/html', $entry->getMimetype());
411 $this->assertSame('fr', $entry->getLanguage());
412 $this->assertSame(4.0, $entry->getReadingTime());
413 $this->assertSame('1.1.1.1', $entry->getDomainName());
414 $this->assertNull($entry->getPublishedAt());
415
416 $records = $handler->getRecords();
417
418 $this->assertCount(1, $records);
419 $this->assertContains('Error while defining date', $records[0]['message']);
420 }
421
422 public function testTaggerThrowException()
423 {
424 $tagger = $this->getTaggerMock();
425 $tagger->expects($this->once())
426 ->method('tag')
427 ->will($this->throwException(new \Exception()));
428
429 $proxy = new ContentProxy((new Graby()), $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
430 $entry = new Entry(new User());
431 $proxy->updateEntry(
432 $entry,
433 'http://1.1.1.1',
434 [
435 'html' => str_repeat('this is my content', 325),
436 'title' => 'this is my title',
437 'url' => 'http://1.1.1.1',
438 'content_type' => 'text/html',
439 'language' => 'fr',
440 ]
441 );
442
443 $this->assertCount(0, $entry->getTags());
444 }
445
446 public function dataForCrazyHtml()
447 {
448 return [
449 'script and comment' => [
450 '<strong>Script inside:</strong> <!--[if gte IE 4]><script>alert(\'lol\');</script><![endif]--><br />',
451 'lol',
452 ],
453 'script' => [
454 '<strong>Script inside:</strong><script>alert(\'lol\');</script>',
455 'script',
456 ],
457 ];
458 }
459
460 /**
461 * @dataProvider dataForCrazyHtml
462 */
463 public function testWithCrazyHtmlContent($html, $escapedString)
464 {
465 $tagger = $this->getTaggerMock();
466 $tagger->expects($this->once())
467 ->method('tag');
468
469 $proxy = new ContentProxy((new Graby()), $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
470 $entry = new Entry(new User());
471 $proxy->updateEntry(
472 $entry,
473 'http://1.1.1.1',
474 [
475 'html' => $html,
476 'title' => 'this is my title',
477 'url' => 'http://1.1.1.1',
478 'content_type' => 'text/html',
479 'language' => 'fr',
480 'status' => '200',
481 'open_graph' => [
482 'og_title' => 'my OG title',
483 'og_description' => 'OG desc',
484 'og_image' => 'http://3.3.3.3/cover.jpg',
485 ],
486 ]
487 );
488
489 $this->assertSame('http://1.1.1.1', $entry->getUrl());
490 $this->assertSame('this is my title', $entry->getTitle());
491 $this->assertNotContains($escapedString, $entry->getContent());
492 $this->assertSame('http://3.3.3.3/cover.jpg', $entry->getPreviewPicture());
493 $this->assertSame('text/html', $entry->getMimetype());
494 $this->assertSame('fr', $entry->getLanguage());
495 $this->assertSame('200', $entry->getHttpStatus());
496 $this->assertSame('1.1.1.1', $entry->getDomainName());
497 }
498
499 public function testWithImageAsContent()
500 {
501 $tagger = $this->getTaggerMock();
502 $tagger->expects($this->once())
503 ->method('tag');
504
505 $graby = $this->getMockBuilder('Graby\Graby')
506 ->setMethods(['fetchContent'])
507 ->disableOriginalConstructor()
508 ->getMock();
509
510 $graby->expects($this->any())
511 ->method('fetchContent')
512 ->willReturn([
513 'html' => '<p><img src="http://1.1.1.1/image.jpg" /></p>',
514 'title' => 'this is my title',
515 'url' => 'http://1.1.1.1/image.jpg',
516 'content_type' => 'image/jpeg',
517 'status' => '200',
518 'open_graph' => [],
519 ]);
520
521 $proxy = new ContentProxy($graby, $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
522 $entry = new Entry(new User());
523 $proxy->updateEntry($entry, 'http://0.0.0.0');
524
525 $this->assertSame('http://1.1.1.1/image.jpg', $entry->getUrl());
526 $this->assertSame('this is my title', $entry->getTitle());
527 $this->assertContains('http://1.1.1.1/image.jpg', $entry->getContent());
528 $this->assertSame('http://1.1.1.1/image.jpg', $entry->getPreviewPicture());
529 $this->assertSame('image/jpeg', $entry->getMimetype());
530 $this->assertSame('200', $entry->getHttpStatus());
531 $this->assertSame('1.1.1.1', $entry->getDomainName());
532 }
533
534 private function getTaggerMock()
535 {
536 return $this->getMockBuilder(RuleBasedTagger::class)
537 ->setMethods(['tag'])
538 ->disableOriginalConstructor()
539 ->getMock();
540 }
541
542 private function getLogger()
543 {
544 return new NullLogger();
545 }
546
547 private function getValidator()
548 {
549 return $this->getMockBuilder(RecursiveValidator::class)
550 ->setMethods(['validate'])
551 ->disableOriginalConstructor()
552 ->getMock();
553 }
554 }