]>
Commit | Line | Data |
---|---|---|
558d9aab JB |
1 | <?php |
2 | ||
3 | namespace Wallabag\CoreBundle\Helper; | |
4 | ||
5 | use Graby\Graby; | |
45fd7e09 | 6 | use Psr\Log\LoggerInterface; |
8d7b4f0e | 7 | use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser; |
42f3bb2c | 8 | use Symfony\Component\Validator\Constraints\Locale as LocaleConstraint; |
0d349ea6 JB |
9 | use Symfony\Component\Validator\Constraints\Url as UrlConstraint; |
10 | use Symfony\Component\Validator\Validator\ValidatorInterface; | |
f808b016 JB |
11 | use Wallabag\CoreBundle\Entity\Entry; |
12 | use Wallabag\CoreBundle\Tools\Utils; | |
558d9aab JB |
13 | |
14 | /** | |
15 | * This kind of proxy class take care of getting the content from an url | |
f1e29e69 | 16 | * and update the entry with what it found. |
558d9aab JB |
17 | */ |
18 | class ContentProxy | |
19 | { | |
20 | protected $graby; | |
c3510620 | 21 | protected $tagger; |
be54dfe4 | 22 | protected $validator; |
1c9cd2a7 | 23 | protected $logger; |
8d7b4f0e | 24 | protected $mimeGuesser; |
29dca432 | 25 | protected $fetchingErrorMessage; |
6bc6fb1f | 26 | protected $eventDispatcher; |
558d9aab | 27 | |
0d349ea6 | 28 | public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage) |
558d9aab | 29 | { |
347fa6be | 30 | $this->graby = $graby; |
c3510620 | 31 | $this->tagger = $tagger; |
0d349ea6 | 32 | $this->validator = $validator; |
1c9cd2a7 | 33 | $this->logger = $logger; |
8d7b4f0e | 34 | $this->mimeGuesser = new MimeTypeExtensionGuesser(); |
29dca432 | 35 | $this->fetchingErrorMessage = $fetchingErrorMessage; |
558d9aab JB |
36 | } |
37 | ||
38 | /** | |
6acadf8e | 39 | * Update entry using either fetched or provided content. |
4d0ec0e7 | 40 | * |
6acadf8e JB |
41 | * @param Entry $entry Entry to update |
42 | * @param string $url Url of the content | |
43 | * @param array $content Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url | |
44 | * @param bool $disableContentUpdate Whether to skip trying to fetch content using Graby | |
558d9aab | 45 | */ |
6acadf8e | 46 | public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false) |
558d9aab | 47 | { |
6acadf8e JB |
48 | if (!empty($content['html'])) { |
49 | $content['html'] = $this->graby->cleanupHtml($content['html'], $url); | |
d5c2cc54 | 50 | } |
e668a812 | 51 | |
6acadf8e | 52 | if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) { |
ec970721 | 53 | $fetchedContent = $this->graby->fetchContent($url); |
106bdbcd JB |
54 | |
55 | // when content is imported, we have information in $content | |
56 | // in case fetching content goes bad, we'll keep the imported information instead of overriding them | |
6acadf8e | 57 | if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) { |
29dca432 JC |
58 | $content = $fetchedContent; |
59 | } | |
4d0ec0e7 | 60 | } |
558d9aab | 61 | |
6acadf8e JB |
62 | // be sure to keep the url in case of error |
63 | // so we'll be able to refetch it in the future | |
64 | $content['url'] = !empty($content['url']) ? $content['url'] : $url; | |
65 | ||
d0e9b3d6 JC |
66 | $this->stockEntry($entry, $content); |
67 | } | |
68 | ||
69 | /** | |
70 | * Stock entry with fetched or imported content. | |
71 | * Will fall back to OpenGraph data if available. | |
72 | * | |
d5c2cc54 | 73 | * @param Entry $entry Entry to stock |
ec970721 | 74 | * @param array $content Array with at least title, url & html |
d0e9b3d6 JC |
75 | */ |
76 | private function stockEntry(Entry $entry, array $content) | |
77 | { | |
558d9aab | 78 | $title = $content['title']; |
e668a812 | 79 | if (!$title && !empty($content['open_graph']['og_title'])) { |
558d9aab JB |
80 | $title = $content['open_graph']['og_title']; |
81 | } | |
82 | ||
83 | $html = $content['html']; | |
84 | if (false === $html) { | |
36e6ef52 | 85 | $html = $this->fetchingErrorMessage; |
558d9aab | 86 | |
e668a812 | 87 | if (!empty($content['open_graph']['og_description'])) { |
558d9aab JB |
88 | $html .= '<p><i>But we found a short description: </i></p>'; |
89 | $html .= $content['open_graph']['og_description']; | |
90 | } | |
91 | } | |
92 | ||
d0e9b3d6 | 93 | $entry->setUrl($content['url']); |
558d9aab | 94 | $entry->setTitle($title); |
48656e0e | 95 | $entry->setContent($html); |
10b35097 | 96 | $entry->setHttpStatus(isset($content['status']) ? $content['status'] : ''); |
419214d7 | 97 | |
e668a812 | 98 | if (!empty($content['date'])) { |
f0378b4d JB |
99 | $date = $content['date']; |
100 | ||
101 | // is it a timestamp? | |
102 | if (filter_var($date, FILTER_VALIDATE_INT) !== false) { | |
f808b016 | 103 | $date = '@' . $content['date']; |
f0378b4d JB |
104 | } |
105 | ||
e668a812 | 106 | try { |
f0378b4d | 107 | $entry->setPublishedAt(new \DateTime($date)); |
e668a812 | 108 | } catch (\Exception $e) { |
6acadf8e | 109 | $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $content['url'], 'date' => $content['date']]); |
e668a812 | 110 | } |
5e9009ce NL |
111 | } |
112 | ||
645291e8 | 113 | if (!empty($content['authors']) && is_array($content['authors'])) { |
7b0b3622 NL |
114 | $entry->setPublishedBy($content['authors']); |
115 | } | |
116 | ||
dda6a6ad NL |
117 | if (!empty($content['all_headers'])) { |
118 | $entry->setHeaders($content['all_headers']); | |
119 | } | |
120 | ||
0d349ea6 JB |
121 | $this->validateAndSetLanguage( |
122 | $entry, | |
41d45c61 | 123 | isset($content['language']) ? $content['language'] : null |
0d349ea6 JB |
124 | ); |
125 | ||
126 | $this->validateAndSetPreviewPicture( | |
127 | $entry, | |
41d45c61 | 128 | isset($content['open_graph']['og_image']) ? $content['open_graph']['og_image'] : null |
0d349ea6 JB |
129 | ); |
130 | ||
be54dfe4 | 131 | // if content is an image, define it as a preview too |
0d349ea6 JB |
132 | if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) { |
133 | $this->validateAndSetPreviewPicture( | |
134 | $entry, | |
135 | $content['url'] | |
136 | ); | |
137 | } | |
138 | ||
e858018f | 139 | $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : ''); |
da3d4998 | 140 | $entry->setReadingTime(Utils::getReadingTime($html)); |
4d0ec0e7 JB |
141 | |
142 | $domainName = parse_url($entry->getUrl(), PHP_URL_HOST); | |
143 | if (false !== $domainName) { | |
144 | $entry->setDomainName($domainName); | |
145 | } | |
558d9aab | 146 | |
1c9cd2a7 KG |
147 | try { |
148 | $this->tagger->tag($entry); | |
149 | } catch (\Exception $e) { | |
4094ea47 | 150 | $this->logger->error('Error while trying to automatically tag an entry.', [ |
d0e9b3d6 | 151 | 'entry_url' => $content['url'], |
1c9cd2a7 | 152 | 'error_msg' => $e->getMessage(), |
4094ea47 | 153 | ]); |
1c9cd2a7 | 154 | } |
558d9aab | 155 | } |
c2656f96 | 156 | |
4d0ec0e7 | 157 | /** |
d0e9b3d6 | 158 | * Validate that the given content has at least a title, an html and a url. |
4d0ec0e7 JB |
159 | * |
160 | * @param array $content | |
6acadf8e JB |
161 | * |
162 | * @return bool true if valid otherwise false | |
4d0ec0e7 JB |
163 | */ |
164 | private function validateContent(array $content) | |
165 | { | |
6acadf8e | 166 | return !empty($content['title']) && !empty($content['html']) && !empty($content['url']); |
4d0ec0e7 | 167 | } |
0d349ea6 JB |
168 | |
169 | /** | |
170 | * Use a Symfony validator to ensure the language is well formatted. | |
171 | * | |
172 | * @param Entry $entry | |
173 | * @param string $value Language to validate | |
174 | */ | |
175 | private function validateAndSetLanguage($entry, $value) | |
176 | { | |
80e49ba7 JB |
177 | // some lang are defined as fr-FR, es-ES. |
178 | // replacing - by _ might increase language support | |
179 | $value = str_replace('-', '_', $value); | |
180 | ||
0d349ea6 JB |
181 | $errors = $this->validator->validate( |
182 | $value, | |
42f3bb2c | 183 | (new LocaleConstraint()) |
0d349ea6 JB |
184 | ); |
185 | ||
186 | if (0 === count($errors)) { | |
187 | $entry->setLanguage($value); | |
188 | ||
189 | return; | |
190 | } | |
191 | ||
f808b016 | 192 | $this->logger->warning('Language validation failed. ' . (string) $errors); |
0d349ea6 JB |
193 | } |
194 | ||
195 | /** | |
196 | * Use a Symfony validator to ensure the preview picture is a real url. | |
197 | * | |
198 | * @param Entry $entry | |
199 | * @param string $value URL to validate | |
200 | */ | |
201 | private function validateAndSetPreviewPicture($entry, $value) | |
202 | { | |
203 | $errors = $this->validator->validate( | |
204 | $value, | |
205 | (new UrlConstraint()) | |
206 | ); | |
207 | ||
208 | if (0 === count($errors)) { | |
209 | $entry->setPreviewPicture($value); | |
210 | ||
211 | return; | |
212 | } | |
213 | ||
f808b016 | 214 | $this->logger->warning('PreviewPicture validation failed. ' . (string) $errors); |
0d349ea6 | 215 | } |
558d9aab | 216 | } |