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 take care of getting the content from an url
16 * and update 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 = [], $disableContentUpdate = false)
50 if (!empty($content['html'])) {
51 $content['html'] = $this->graby
->cleanupHtml($content['html'], $url);
54 if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) {
55 $fetchedContent = $this->graby
->fetchContent($url);
57 // when content is imported, we have information in $content
58 // in case fetching content goes bad, we'll keep the imported information instead of overriding them
59 if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage
) {
60 $content = $fetchedContent;
64 // be sure to keep the url in case of error
65 // so we'll be able to refetch it in the future
66 $content['url'] = !empty($content['url']) ? $content['url'] : $url;
68 $this->stockEntry($entry, $content);
72 * Use a Symfony validator to ensure the language is well formatted.
75 * @param string $value Language to validate and save
77 public function updateLanguage(Entry
$entry, $value)
79 // some lang are defined as fr-FR, es-ES.
80 // replacing - by _ might increase language support
81 $value = str_replace('-', '_', $value);
83 $errors = $this->validator
->validate(
85 (new LocaleConstraint())
88 if (0 === count($errors)) {
89 $entry->setLanguage($value);
94 $this->logger
->warning('Language validation failed. ' . (string) $errors);
98 * Use a Symfony validator to ensure the preview picture is a real url.
100 * @param Entry $entry
101 * @param string $value URL to validate and save
103 public function updatePreviewPicture(Entry
$entry, $value)
105 $errors = $this->validator
->validate(
107 (new UrlConstraint())
110 if (0 === count($errors)) {
111 $entry->setPreviewPicture($value);
116 $this->logger
->warning('PreviewPicture validation failed. ' . (string) $errors);
122 * @param Entry $entry
123 * @param string $value Date to validate and save
125 public function updatePublishedAt(Entry
$entry, $value)
129 // is it a timestamp?
130 if (false !== filter_var($date, FILTER_VALIDATE_INT
)) {
135 // is it already a DateTime?
136 // (it's inside the try/catch in case of fail to be parse time string)
137 if (!$date instanceof \DateTime
) {
138 $date = new \
DateTime($date);
141 $entry->setPublishedAt($date);
142 } catch (\Exception
$e) {
143 $this->logger
->warning('Error while defining date', ['e' => $e, 'url' => $entry->getUrl(), 'date' => $value]);
148 * Stock entry with fetched or imported content.
149 * Will fall back to OpenGraph data if available.
151 * @param Entry $entry Entry to stock
152 * @param array $content Array with at least title, url & html
154 private function stockEntry(Entry
$entry, array $content)
156 $entry->setUrl($content['url']);
158 $domainName = parse_url($entry->getUrl(), PHP_URL_HOST
);
159 if (false !== $domainName) {
160 $entry->setDomainName($domainName);
163 if (!empty($content['title'])) {
164 $entry->setTitle($content['title']);
165 } elseif (!empty($content['open_graph']['og_title'])) {
166 $entry->setTitle($content['open_graph']['og_title']);
169 $html = $content['html'];
170 if (false === $html) {
171 $html = $this->fetchingErrorMessage
;
173 if (!empty($content['open_graph']['og_description'])) {
174 $html .= '<p><i>But we found a short description: </i></p>';
175 $html .= $content['open_graph']['og_description'];
179 $entry->setContent($html);
180 $entry->setReadingTime(Utils
::getReadingTime($html));
182 if (!empty($content['status'])) {
183 $entry->setHttpStatus($content['status']);
186 if (!empty($content['authors']) && is_array($content['authors'])) {
187 $entry->setPublishedBy($content['authors']);
190 if (!empty($content['all_headers']) && $this->storeArticleHeaders
) {
191 $entry->setHeaders($content['all_headers']);
194 if (!empty($content['date'])) {
195 $this->updatePublishedAt($entry, $content['date']);
198 if (!empty($content['language'])) {
199 $this->updateLanguage($entry, $content['language']);
202 if (!empty($content['open_graph']['og_image'])) {
203 $this->updatePreviewPicture($entry, $content['open_graph']['og_image']);
206 // if content is an image, define it as a preview too
207 if (!empty($content['content_type']) && in_array($this->mimeGuesser
->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
208 $this->updatePreviewPicture($entry, $content['url']);
211 if (!empty($content['content_type'])) {
212 $entry->setMimetype($content['content_type']);
216 $this->tagger
->tag($entry);
217 } catch (\Exception
$e) {
218 $this->logger
->error('Error while trying to automatically tag an entry.', [
219 'entry_url' => $content['url'],
220 'error_msg' => $e->getMessage(),
226 * Validate that the given content has at least a title, an html and a url.
228 * @param array $content
230 * @return bool true if valid otherwise false
232 private function validateContent(array $content)
234 return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);