3 namespace Wallabag\CoreBundle\Helper
;
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
;
15 * This kind of proxy class takes care of getting the content from an url
16 * and updates the entry with what it found.
24 protected $mimeGuesser;
25 protected $fetchingErrorMessage;
26 protected $eventDispatcher;
27 protected $storeArticleHeaders;
29 public function __construct(Graby
$graby, RuleBasedTagger
$tagger, ValidatorInterface
$validator, LoggerInterface
$logger, $fetchingErrorMessage, $storeArticleHeaders = false)
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;
41 * Update entry using either fetched or provided content.
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
48 public function updateEntry(Entry
$entry, $url, array $content = [], $disableContentUpdate = false)
50 $this->graby
->toggleImgNoReferrer(true);
51 if (!empty($content['html'])) {
52 $content['html'] = $this->graby
->cleanupHtml($content['html'], $url);
55 if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) {
56 $fetchedContent = $this->graby
->fetchContent($url);
57 $fetchedContent['title'] = $this->sanitizeContentTitle($fetchedContent['title'], $fetchedContent['content_type']);
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;
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;
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)) {
77 $this->stockEntry($entry, $content);
81 * Use a Symfony validator to ensure the language is well formatted.
84 * @param string $value Language to validate and save
86 public function updateLanguage(Entry
$entry, $value)
88 // some lang are defined as fr-FR, es-ES.
89 // replacing - by _ might increase language support
90 $value = str_replace('-', '_', $value);
92 $errors = $this->validator
->validate(
94 (new LocaleConstraint())
97 if (0 === \
count($errors)) {
98 $entry->setLanguage($value);
103 $this->logger
->warning('Language validation failed. ' . (string) $errors);
107 * Use a Symfony validator to ensure the preview picture is a real url.
109 * @param Entry $entry
110 * @param string $value URL to validate and save
112 public function updatePreviewPicture(Entry
$entry, $value)
114 $errors = $this->validator
->validate(
116 (new UrlConstraint())
119 if (0 === \
count($errors)) {
120 $entry->setPreviewPicture($value);
125 $this->logger
->warning('PreviewPicture validation failed. ' . (string) $errors);
131 * @param Entry $entry
132 * @param string $value Date to validate and save
134 public function updatePublishedAt(Entry
$entry, $value)
138 // is it a timestamp?
139 if (false !== filter_var($date, FILTER_VALIDATE_INT
)) {
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);
150 $entry->setPublishedAt($date);
151 } catch (\Exception
$e) {
152 $this->logger
->warning('Error while defining date', ['e' => $e, 'url' => $entry->getUrl(), 'date' => $value]);
157 * Helper to extract and save host from entry url.
159 * @param Entry $entry
161 public function setEntryDomainName(Entry
$entry)
163 $domainName = parse_url($entry->getUrl(), PHP_URL_HOST
);
164 if (false !== $domainName) {
165 $entry->setDomainName($domainName);
170 * Helper to set a default title using:
171 * - url basename, if applicable
174 * @param Entry $entry
176 public function setDefaultEntryTitle(Entry
$entry)
178 $url = parse_url($entry->getUrl());
179 $path = pathinfo($url['path'], PATHINFO_BASENAME
);
182 $path = $url['host'];
185 $entry->setTitle($path);
189 * Try to sanitize the title of the fetched content from wrong character encodings and invalid UTF-8 character.
192 * @param $contentType
196 private function sanitizeContentTitle($title, $contentType)
198 if ('application/pdf' === $contentType) {
199 $title = $this->convertPdfEncodingToUTF8($title);
202 return $this->sanitizeUTF8Text($title);
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.
211 * @return string (maybe contains invalid UTF-8 character)
213 private function convertPdfEncodingToUTF8($title)
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);
226 * Remove invalid UTF-8 characters from the given string.
228 * @param string $rawText
232 private function sanitizeUTF8Text($rawText)
234 if (mb_check_encoding($rawText, 'UTF-8')) {
238 return iconv('UTF-8', 'UTF-8//IGNORE', $rawText);
242 * Stock entry with fetched or imported content.
243 * Will fall back to OpenGraph data if available.
245 * @param Entry $entry Entry to stock
246 * @param array $content Array with at least title, url & html
248 private function stockEntry(Entry
$entry, array $content)
250 $this->updateOriginUrl($entry, $content['url']);
252 $this->setEntryDomainName($entry);
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']);
260 if (empty($content['html'])) {
261 $content['html'] = $this->fetchingErrorMessage
;
263 if (!empty($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'];
269 $entry->setContent($content['html']);
270 $entry->setReadingTime(Utils
::getReadingTime($content['html']));
272 if (!empty($content['status'])) {
273 $entry->setHttpStatus($content['status']);
276 if (!empty($content['authors']) && \
is_array($content['authors'])) {
277 $entry->setPublishedBy($content['authors']);
280 if (!empty($content['all_headers']) && $this->storeArticleHeaders
) {
281 $entry->setHeaders($content['all_headers']);
284 if (!empty($content['date'])) {
285 $this->updatePublishedAt($entry, $content['date']);
288 if (!empty($content['language'])) {
289 $this->updateLanguage($entry, $content['language']);
292 $previewPictureUrl = '';
293 if (!empty($content['open_graph']['og_image'])) {
294 $previewPictureUrl = $content['open_graph']['og_image'];
297 // if content is an image, define it as a preview too
298 if (!empty($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 = $imagesUrls[0];
309 if (!empty($previewPictureUrl)) {
310 $this->updatePreviewPicture($entry, $previewPictureUrl);
313 if (!empty($content['content_type'])) {
314 $entry->setMimetype($content['content_type']);
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(),
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.
331 * @param Entry $entry
334 private function updateOriginUrl(Entry
$entry, $url)
336 if (empty($url) || $entry->getUrl() === $url) {
340 $parsed_entry_url = parse_url($entry->getUrl());
341 $parsed_content_url = parse_url($url);
344 * The following part computes the list of part changes between two
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.
352 * The resulting array gives us all changing parts between the two
353 * urls: scheme, host, path, query and/or fragment.
355 $diff_ec = array_diff_assoc($parsed_entry_url, $parsed_content_url);
356 $diff_ce = array_diff_assoc($parsed_content_url, $parsed_entry_url);
358 $diff = array_merge($diff_ec, $diff_ce);
359 $diff_keys = array_keys($diff);
362 if ($this->ignoreUrl($entry->getUrl())) {
363 $entry->setUrl($url);
369 * This switch case lets us apply different behaviors according to
370 * changing parts of urls.
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'.
377 * Note that values in $diff_keys are sorted.
379 switch ($diff_keys) {
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);
387 $entry->setUrl($url);
393 if (empty($entry->getOriginUrl())) {
394 $entry->setOriginUrl($entry->getUrl());
396 $entry->setUrl($url);
402 * Check entry url against an ignore list to replace with content url.
404 * XXX: move the ignore list in the database to let users handle it
406 * @param string $url url to test
408 * @return bool true if url matches ignore list otherwise false
410 private function ignoreUrl($url)
412 $ignored_hosts = ['feedproxy.google.com', 'feeds.reuters.com'];
413 $ignored_patterns = ['https?://www\.lemonde\.fr/tiny.*'];
415 $parsed_url = parse_url($url);
417 $filtered = array_filter($ignored_hosts, function ($var) use ($parsed_url) {
418 return $var === $parsed_url['host'];
421 if ([] !== $filtered) {
425 $filtered = array_filter($ignored_patterns, function ($var) use ($url) {
426 return preg_match("`$var`i", $url);
429 if ([] !== $filtered) {
437 * Validate that the given content has at least a title, an html and a url.
439 * @param array $content
441 * @return bool true if valid otherwise false
443 private function validateContent(array $content)
445 return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);