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