]>
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 | { | |
a05b6115 JB |
78 | $entry->setUrl($content['url']); |
79 | ||
80 | $domainName = parse_url($entry->getUrl(), PHP_URL_HOST); | |
81 | if (false !== $domainName) { | |
82 | $entry->setDomainName($domainName); | |
83 | } | |
84 | ||
85 | if (!empty($content['title'])) { | |
86 | $entry->setTitle($content['title']); | |
87 | } elseif (!empty($content['open_graph']['og_title'])) { | |
88 | $entry->setTitle($content['open_graph']['og_title']); | |
558d9aab JB |
89 | } |
90 | ||
91 | $html = $content['html']; | |
92 | if (false === $html) { | |
36e6ef52 | 93 | $html = $this->fetchingErrorMessage; |
558d9aab | 94 | |
e668a812 | 95 | if (!empty($content['open_graph']['og_description'])) { |
558d9aab JB |
96 | $html .= '<p><i>But we found a short description: </i></p>'; |
97 | $html .= $content['open_graph']['og_description']; | |
98 | } | |
99 | } | |
100 | ||
48656e0e | 101 | $entry->setContent($html); |
a05b6115 | 102 | $entry->setReadingTime(Utils::getReadingTime($html)); |
f0378b4d | 103 | |
a05b6115 JB |
104 | if (!empty($content['status'])) { |
105 | $entry->setHttpStatus($content['status']); | |
5e9009ce NL |
106 | } |
107 | ||
645291e8 | 108 | if (!empty($content['authors']) && is_array($content['authors'])) { |
7b0b3622 NL |
109 | $entry->setPublishedBy($content['authors']); |
110 | } | |
111 | ||
dda6a6ad NL |
112 | if (!empty($content['all_headers'])) { |
113 | $entry->setHeaders($content['all_headers']); | |
114 | } | |
115 | ||
a05b6115 JB |
116 | if (!empty($content['date'])) { |
117 | $this->updatePublishedAt($entry, $content['date']); | |
118 | } | |
0d349ea6 | 119 | |
a05b6115 JB |
120 | if (!empty($content['language'])) { |
121 | $this->updateLanguage($entry, $content['language']); | |
122 | } | |
123 | ||
124 | if (!empty($content['open_graph']['og_image'])) { | |
125 | $this->updatePreviewPicture($entry, $content['open_graph']['og_image']); | |
126 | } | |
0d349ea6 | 127 | |
be54dfe4 | 128 | // if content is an image, define it as a preview too |
0d349ea6 | 129 | if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) { |
d0ec2ddd | 130 | $this->updatePreviewPicture($entry, $content['url']); |
0d349ea6 JB |
131 | } |
132 | ||
a05b6115 JB |
133 | if (!empty($content['content_type'])) { |
134 | $entry->setMimetype($content['content_type']); | |
4d0ec0e7 | 135 | } |
558d9aab | 136 | |
1c9cd2a7 KG |
137 | try { |
138 | $this->tagger->tag($entry); | |
139 | } catch (\Exception $e) { | |
4094ea47 | 140 | $this->logger->error('Error while trying to automatically tag an entry.', [ |
d0e9b3d6 | 141 | 'entry_url' => $content['url'], |
1c9cd2a7 | 142 | 'error_msg' => $e->getMessage(), |
4094ea47 | 143 | ]); |
1c9cd2a7 | 144 | } |
558d9aab | 145 | } |
c2656f96 | 146 | |
4d0ec0e7 | 147 | /** |
d0e9b3d6 | 148 | * Validate that the given content has at least a title, an html and a url. |
4d0ec0e7 JB |
149 | * |
150 | * @param array $content | |
6acadf8e JB |
151 | * |
152 | * @return bool true if valid otherwise false | |
4d0ec0e7 JB |
153 | */ |
154 | private function validateContent(array $content) | |
155 | { | |
6acadf8e | 156 | return !empty($content['title']) && !empty($content['html']) && !empty($content['url']); |
4d0ec0e7 | 157 | } |
0d349ea6 JB |
158 | |
159 | /** | |
160 | * Use a Symfony validator to ensure the language is well formatted. | |
161 | * | |
162 | * @param Entry $entry | |
a05b6115 | 163 | * @param string $value Language to validate and save |
0d349ea6 | 164 | */ |
a05b6115 | 165 | public function updateLanguage(Entry $entry, $value) |
0d349ea6 | 166 | { |
80e49ba7 JB |
167 | // some lang are defined as fr-FR, es-ES. |
168 | // replacing - by _ might increase language support | |
169 | $value = str_replace('-', '_', $value); | |
170 | ||
0d349ea6 JB |
171 | $errors = $this->validator->validate( |
172 | $value, | |
42f3bb2c | 173 | (new LocaleConstraint()) |
0d349ea6 JB |
174 | ); |
175 | ||
176 | if (0 === count($errors)) { | |
177 | $entry->setLanguage($value); | |
178 | ||
179 | return; | |
180 | } | |
181 | ||
f808b016 | 182 | $this->logger->warning('Language validation failed. ' . (string) $errors); |
0d349ea6 JB |
183 | } |
184 | ||
185 | /** | |
186 | * Use a Symfony validator to ensure the preview picture is a real url. | |
187 | * | |
188 | * @param Entry $entry | |
a05b6115 | 189 | * @param string $value URL to validate and save |
0d349ea6 | 190 | */ |
a05b6115 | 191 | public function updatePreviewPicture(Entry $entry, $value) |
0d349ea6 JB |
192 | { |
193 | $errors = $this->validator->validate( | |
194 | $value, | |
195 | (new UrlConstraint()) | |
196 | ); | |
197 | ||
198 | if (0 === count($errors)) { | |
199 | $entry->setPreviewPicture($value); | |
200 | ||
201 | return; | |
202 | } | |
203 | ||
f808b016 | 204 | $this->logger->warning('PreviewPicture validation failed. ' . (string) $errors); |
0d349ea6 | 205 | } |
a05b6115 JB |
206 | |
207 | /** | |
208 | * Update date. | |
209 | * | |
210 | * @param Entry $entry | |
211 | * @param string $value Date to validate and save | |
212 | */ | |
213 | public function updatePublishedAt(Entry $entry, $value) | |
214 | { | |
215 | $date = $value; | |
216 | ||
217 | // is it a timestamp? | |
218 | if (filter_var($date, FILTER_VALIDATE_INT) !== false) { | |
219 | $date = '@'.$value; | |
220 | } | |
221 | ||
222 | try { | |
223 | $entry->setPublishedAt(new \DateTime($date)); | |
224 | } catch (\Exception $e) { | |
225 | $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $entry->getUrl(), 'date' => $value]); | |
226 | } | |
227 | } | |
558d9aab | 228 | } |