]>
Commit | Line | Data |
---|---|---|
1 | <?php | |
2 | ||
3 | namespace Wallabag\CoreBundle\Helper; | |
4 | ||
5 | use Graby\Graby; | |
6 | use Psr\Log\LoggerInterface; | |
7 | use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser; | |
8 | use Symfony\Component\Validator\Constraints\Locale as LocaleConstraint; | |
9 | use Symfony\Component\Validator\Constraints\Url as UrlConstraint; | |
10 | use Symfony\Component\Validator\Validator\ValidatorInterface; | |
11 | use Wallabag\CoreBundle\Entity\Entry; | |
12 | use Wallabag\CoreBundle\Tools\Utils; | |
13 | ||
14 | /** | |
15 | * This kind of proxy class takes care of getting the content from an url | |
16 | * and updates the entry with what it found. | |
17 | */ | |
18 | class ContentProxy | |
19 | { | |
20 | protected $graby; | |
21 | protected $tagger; | |
22 | protected $validator; | |
23 | protected $logger; | |
24 | protected $mimeGuesser; | |
25 | protected $fetchingErrorMessage; | |
26 | protected $eventDispatcher; | |
27 | protected $storeArticleHeaders; | |
28 | ||
29 | public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage, $storeArticleHeaders = false) | |
30 | { | |
31 | $this->graby = $graby; | |
32 | $this->tagger = $tagger; | |
33 | $this->validator = $validator; | |
34 | $this->logger = $logger; | |
35 | $this->mimeGuesser = new MimeTypeExtensionGuesser(); | |
36 | $this->fetchingErrorMessage = $fetchingErrorMessage; | |
37 | $this->storeArticleHeaders = $storeArticleHeaders; | |
38 | } | |
39 | ||
40 | /** | |
41 | * Update entry using either fetched or provided content. | |
42 | * | |
43 | * @param Entry $entry Entry to update | |
44 | * @param string $url Url of the content | |
45 | * @param array $content Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url | |
46 | * @param bool $disableContentUpdate Whether to skip trying to fetch content using Graby | |
47 | */ | |
48 | public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false) | |
49 | { | |
50 | $this->graby->toggleImgNoReferrer(true); | |
51 | if (!empty($content['html'])) { | |
52 | $content['html'] = $this->graby->cleanupHtml($content['html'], $url); | |
53 | } | |
54 | ||
55 | if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) { | |
56 | $fetchedContent = $this->graby->fetchContent($url); | |
57 | $fetchedContent['title'] = $this->sanitizeContentTitle($fetchedContent['title'], $fetchedContent['content_type']); | |
58 | ||
59 | // when content is imported, we have information in $content | |
60 | // in case fetching content goes bad, we'll keep the imported information instead of overriding them | |
61 | if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) { | |
62 | $content = $fetchedContent; | |
63 | } | |
64 | } | |
65 | ||
66 | // be sure to keep the url in case of error | |
67 | // so we'll be able to refetch it in the future | |
68 | $content['url'] = !empty($content['url']) ? $content['url'] : $url; | |
69 | ||
70 | // In one case (at least in tests), url is empty here | |
71 | // so we set it using $url provided in the updateEntry call. | |
72 | // Not sure what are the other possible cases where this property is empty | |
73 | if (empty($entry->getUrl()) && !empty($url)) { | |
74 | $entry->setUrl($url); | |
75 | } | |
76 | ||
77 | $this->stockEntry($entry, $content); | |
78 | } | |
79 | ||
80 | /** | |
81 | * Use a Symfony validator to ensure the language is well formatted. | |
82 | * | |
83 | * @param Entry $entry | |
84 | * @param string $value Language to validate and save | |
85 | */ | |
86 | public function updateLanguage(Entry $entry, $value) | |
87 | { | |
88 | // some lang are defined as fr-FR, es-ES. | |
89 | // replacing - by _ might increase language support | |
90 | $value = str_replace('-', '_', $value); | |
91 | ||
92 | $errors = $this->validator->validate( | |
93 | $value, | |
94 | (new LocaleConstraint()) | |
95 | ); | |
96 | ||
97 | if (0 === \count($errors)) { | |
98 | $entry->setLanguage($value); | |
99 | ||
100 | return; | |
101 | } | |
102 | ||
103 | $this->logger->warning('Language validation failed. ' . (string) $errors); | |
104 | } | |
105 | ||
106 | /** | |
107 | * Use a Symfony validator to ensure the preview picture is a real url. | |
108 | * | |
109 | * @param Entry $entry | |
110 | * @param string $value URL to validate and save | |
111 | */ | |
112 | public function updatePreviewPicture(Entry $entry, $value) | |
113 | { | |
114 | $errors = $this->validator->validate( | |
115 | $value, | |
116 | (new UrlConstraint()) | |
117 | ); | |
118 | ||
119 | if (0 === \count($errors)) { | |
120 | $entry->setPreviewPicture($value); | |
121 | ||
122 | return; | |
123 | } | |
124 | ||
125 | $this->logger->warning('PreviewPicture validation failed. ' . (string) $errors); | |
126 | } | |
127 | ||
128 | /** | |
129 | * Update date. | |
130 | * | |
131 | * @param Entry $entry | |
132 | * @param string $value Date to validate and save | |
133 | */ | |
134 | public function updatePublishedAt(Entry $entry, $value) | |
135 | { | |
136 | $date = $value; | |
137 | ||
138 | // is it a timestamp? | |
139 | if (false !== filter_var($date, FILTER_VALIDATE_INT)) { | |
140 | $date = '@' . $date; | |
141 | } | |
142 | ||
143 | try { | |
144 | // is it already a DateTime? | |
145 | // (it's inside the try/catch in case of fail to be parse time string) | |
146 | if (!$date instanceof \DateTime) { | |
147 | $date = new \DateTime($date); | |
148 | } | |
149 | ||
150 | $entry->setPublishedAt($date); | |
151 | } catch (\Exception $e) { | |
152 | $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $entry->getUrl(), 'date' => $value]); | |
153 | } | |
154 | } | |
155 | ||
156 | /** | |
157 | * Helper to extract and save host from entry url. | |
158 | * | |
159 | * @param Entry $entry | |
160 | */ | |
161 | public function setEntryDomainName(Entry $entry) | |
162 | { | |
163 | $domainName = parse_url($entry->getUrl(), PHP_URL_HOST); | |
164 | if (false !== $domainName) { | |
165 | $entry->setDomainName($domainName); | |
166 | } | |
167 | } | |
168 | ||
169 | /** | |
170 | * Helper to set a default title using: | |
171 | * - url basename, if applicable | |
172 | * - hostname. | |
173 | * | |
174 | * @param Entry $entry | |
175 | */ | |
176 | public function setDefaultEntryTitle(Entry $entry) | |
177 | { | |
178 | $url = parse_url($entry->getUrl()); | |
179 | $path = pathinfo($url['path'], PATHINFO_BASENAME); | |
180 | ||
181 | if (empty($path)) { | |
182 | $path = $url['host']; | |
183 | } | |
184 | ||
185 | $entry->setTitle($path); | |
186 | } | |
187 | ||
188 | /** | |
189 | * Try to sanitize the title of the fetched content from wrong character encodings and invalid UTF-8 character. | |
190 | * | |
191 | * @param $title | |
192 | * @param $contentType | |
193 | * | |
194 | * @return string | |
195 | */ | |
196 | private function sanitizeContentTitle($title, $contentType) | |
197 | { | |
198 | if ('application/pdf' === $contentType) { | |
199 | $title = $this->convertPdfEncodingToUTF8($title); | |
200 | } | |
201 | ||
202 | return $this->sanitizeUTF8Text($title); | |
203 | } | |
204 | ||
205 | /** | |
206 | * If the title from the fetched content comes from a PDF, then its very possible that the character encoding is not | |
207 | * UTF-8. This methods tries to identify the character encoding and translate the title to UTF-8. | |
208 | * | |
209 | * @param $title | |
210 | * | |
211 | * @return string (maybe contains invalid UTF-8 character) | |
212 | */ | |
213 | private function convertPdfEncodingToUTF8($title) | |
214 | { | |
215 | // first try UTF-8 because its easier to detect its present/absence | |
216 | foreach (['UTF-8', 'UTF-16BE', 'WINDOWS-1252'] as $encoding) { | |
217 | if (mb_check_encoding($title, $encoding)) { | |
218 | return mb_convert_encoding($title, 'UTF-8', $encoding); | |
219 | } | |
220 | } | |
221 | ||
222 | return $title; | |
223 | } | |
224 | ||
225 | /** | |
226 | * Remove invalid UTF-8 characters from the given string. | |
227 | * | |
228 | * @param string $rawText | |
229 | * | |
230 | * @return string | |
231 | */ | |
232 | private function sanitizeUTF8Text($rawText) | |
233 | { | |
234 | if (mb_check_encoding($rawText, 'UTF-8')) { | |
235 | return $rawText; | |
236 | } | |
237 | ||
238 | return iconv('UTF-8', 'UTF-8//IGNORE', $rawText); | |
239 | } | |
240 | ||
241 | /** | |
242 | * Stock entry with fetched or imported content. | |
243 | * Will fall back to OpenGraph data if available. | |
244 | * | |
245 | * @param Entry $entry Entry to stock | |
246 | * @param array $content Array with at least title, url & html | |
247 | */ | |
248 | private function stockEntry(Entry $entry, array $content) | |
249 | { | |
250 | $this->updateOriginUrl($entry, $content['url']); | |
251 | ||
252 | $this->setEntryDomainName($entry); | |
253 | ||
254 | if (!empty($content['title'])) { | |
255 | $entry->setTitle($content['title']); | |
256 | } elseif (!empty($content['open_graph']['og_title'])) { | |
257 | $entry->setTitle($content['open_graph']['og_title']); | |
258 | } | |
259 | ||
260 | if (empty($content['html'])) { | |
261 | $content['html'] = $this->fetchingErrorMessage; | |
262 | ||
263 | if (!empty($content['open_graph']['og_description'])) { | |
264 | $content['html'] .= '<p><i>But we found a short description: </i></p>'; | |
265 | $content['html'] .= $content['open_graph']['og_description']; | |
266 | } | |
267 | } | |
268 | ||
269 | $entry->setContent($content['html']); | |
270 | $entry->setReadingTime(Utils::getReadingTime($content['html'])); | |
271 | ||
272 | if (!empty($content['status'])) { | |
273 | $entry->setHttpStatus($content['status']); | |
274 | } | |
275 | ||
276 | if (!empty($content['authors']) && \is_array($content['authors'])) { | |
277 | $entry->setPublishedBy($content['authors']); | |
278 | } | |
279 | ||
280 | if (!empty($content['all_headers']) && $this->storeArticleHeaders) { | |
281 | $entry->setHeaders($content['all_headers']); | |
282 | } | |
283 | ||
284 | if (!empty($content['date'])) { | |
285 | $this->updatePublishedAt($entry, $content['date']); | |
286 | } | |
287 | ||
288 | if (!empty($content['language'])) { | |
289 | $this->updateLanguage($entry, $content['language']); | |
290 | } | |
291 | ||
292 | $previewPictureUrl = ''; | |
293 | if (!empty($content['open_graph']['og_image'])) { | |
294 | $previewPictureUrl = $content['open_graph']['og_image']; | |
295 | } | |
296 | ||
297 | // if content is an image, define it as a preview too | |
298 | if (!empty($content['content_type']) && \in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) { | |
299 | $previewPictureUrl = $content['url']; | |
300 | } elseif (empty($previewPictureUrl)) { | |
301 | $this->logger->debug('Extracting images from content to provide a default preview picture'); | |
302 | $imagesUrls = DownloadImages::extractImagesUrlsFromHtml($content['html']); | |
303 | $this->logger->debug(\count($imagesUrls) . ' pictures found'); | |
304 | if (!empty($imagesUrls)) { | |
305 | $previewPictureUrl = $imagesUrls[0]; | |
306 | } | |
307 | } | |
308 | ||
309 | if (!empty($previewPictureUrl)) { | |
310 | $this->updatePreviewPicture($entry, $previewPictureUrl); | |
311 | } | |
312 | ||
313 | if (!empty($content['content_type'])) { | |
314 | $entry->setMimetype($content['content_type']); | |
315 | } | |
316 | ||
317 | try { | |
318 | $this->tagger->tag($entry); | |
319 | } catch (\Exception $e) { | |
320 | $this->logger->error('Error while trying to automatically tag an entry.', [ | |
321 | 'entry_url' => $content['url'], | |
322 | 'error_msg' => $e->getMessage(), | |
323 | ]); | |
324 | } | |
325 | } | |
326 | ||
327 | /** | |
328 | * Update the origin_url field when a redirection occurs | |
329 | * This field is set if it is empty and new url does not match ignore list. | |
330 | * | |
331 | * @param Entry $entry | |
332 | * @param string $url | |
333 | */ | |
334 | private function updateOriginUrl(Entry $entry, $url) | |
335 | { | |
336 | if (empty($url) || $entry->getUrl() === $url) { | |
337 | return false; | |
338 | } | |
339 | ||
340 | $parsed_entry_url = parse_url($entry->getUrl()); | |
341 | $parsed_content_url = parse_url($url); | |
342 | ||
343 | /** | |
344 | * The following part computes the list of part changes between two | |
345 | * parse_url arrays. | |
346 | * | |
347 | * As array_diff_assoc only computes changes to go from the left array | |
348 | * to the right one, we make two differents arrays to have both | |
349 | * directions. We merge these two arrays and sort keys before passing | |
350 | * the result to the switch. | |
351 | * | |
352 | * The resulting array gives us all changing parts between the two | |
353 | * urls: scheme, host, path, query and/or fragment. | |
354 | */ | |
355 | $diff_ec = array_diff_assoc($parsed_entry_url, $parsed_content_url); | |
356 | $diff_ce = array_diff_assoc($parsed_content_url, $parsed_entry_url); | |
357 | ||
358 | $diff = array_merge($diff_ec, $diff_ce); | |
359 | $diff_keys = array_keys($diff); | |
360 | sort($diff_keys); | |
361 | ||
362 | if ($this->ignoreUrl($entry->getUrl())) { | |
363 | $entry->setUrl($url); | |
364 | ||
365 | return false; | |
366 | } | |
367 | ||
368 | /** | |
369 | * This switch case lets us apply different behaviors according to | |
370 | * changing parts of urls. | |
371 | * | |
372 | * As $diff_keys is an array, we provide arrays as cases. ['path'] means | |
373 | * 'only the path is different between the two urls' whereas | |
374 | * ['fragment', 'query'] means 'only fragment and query string parts are | |
375 | * different between the two urls'. | |
376 | * | |
377 | * Note that values in $diff_keys are sorted. | |
378 | */ | |
379 | switch ($diff_keys) { | |
380 | case ['path']: | |
381 | if (($parsed_entry_url['path'] . '/' === $parsed_content_url['path']) // diff is trailing slash, we only replace the url of the entry | |
382 | || ($url === urldecode($entry->getUrl()))) { // we update entry url if new url is a decoded version of it, see EntryRepository#findByUrlAndUserId | |
383 | $entry->setUrl($url); | |
384 | } | |
385 | break; | |
386 | case ['scheme']: | |
387 | $entry->setUrl($url); | |
388 | break; | |
389 | case ['fragment']: | |
390 | // noop | |
391 | break; | |
392 | default: | |
393 | if (empty($entry->getOriginUrl())) { | |
394 | $entry->setOriginUrl($entry->getUrl()); | |
395 | } | |
396 | $entry->setUrl($url); | |
397 | break; | |
398 | } | |
399 | } | |
400 | ||
401 | /** | |
402 | * Check entry url against an ignore list to replace with content url. | |
403 | * | |
404 | * XXX: move the ignore list in the database to let users handle it | |
405 | * | |
406 | * @param string $url url to test | |
407 | * | |
408 | * @return bool true if url matches ignore list otherwise false | |
409 | */ | |
410 | private function ignoreUrl($url) | |
411 | { | |
412 | $ignored_hosts = ['feedproxy.google.com', 'feeds.reuters.com']; | |
413 | $ignored_patterns = ['https?://www\.lemonde\.fr/tiny.*']; | |
414 | ||
415 | $parsed_url = parse_url($url); | |
416 | ||
417 | $filtered = array_filter($ignored_hosts, function ($var) use ($parsed_url) { | |
418 | return $var === $parsed_url['host']; | |
419 | }); | |
420 | ||
421 | if ([] !== $filtered) { | |
422 | return true; | |
423 | } | |
424 | ||
425 | $filtered = array_filter($ignored_patterns, function ($var) use ($url) { | |
426 | return preg_match("`$var`i", $url); | |
427 | }); | |
428 | ||
429 | if ([] !== $filtered) { | |
430 | return true; | |
431 | } | |
432 | ||
433 | return false; | |
434 | } | |
435 | ||
436 | /** | |
437 | * Validate that the given content has at least a title, an html and a url. | |
438 | * | |
439 | * @param array $content | |
440 | * | |
441 | * @return bool true if valid otherwise false | |
442 | */ | |
443 | private function validateContent(array $content) | |
444 | { | |
445 | return !empty($content['title']) && !empty($content['html']) && !empty($content['url']); | |
446 | } | |
447 | } |