]>
Commit | Line | Data |
---|---|---|
558d9aab JB |
1 | <?php |
2 | ||
3 | namespace Wallabag\CoreBundle\Helper; | |
4 | ||
5 | use Graby\Graby; | |
45fd7e09 | 6 | use Psr\Log\LoggerInterface; |
558d9aab | 7 | use Wallabag\CoreBundle\Entity\Entry; |
da3d4998 | 8 | use Wallabag\CoreBundle\Tools\Utils; |
8d7b4f0e | 9 | use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser; |
558d9aab JB |
10 | |
11 | /** | |
12 | * This kind of proxy class take care of getting the content from an url | |
f1e29e69 | 13 | * and update the entry with what it found. |
558d9aab JB |
14 | */ |
15 | class ContentProxy | |
16 | { | |
17 | protected $graby; | |
c3510620 | 18 | protected $tagger; |
1c9cd2a7 | 19 | protected $logger; |
8d7b4f0e | 20 | protected $mimeGuesser; |
29dca432 | 21 | protected $fetchingErrorMessage; |
6bc6fb1f | 22 | protected $eventDispatcher; |
558d9aab | 23 | |
6bc6fb1f | 24 | public function __construct(Graby $graby, RuleBasedTagger $tagger, LoggerInterface $logger, $fetchingErrorMessage) |
558d9aab | 25 | { |
347fa6be | 26 | $this->graby = $graby; |
c3510620 | 27 | $this->tagger = $tagger; |
1c9cd2a7 | 28 | $this->logger = $logger; |
8d7b4f0e | 29 | $this->mimeGuesser = new MimeTypeExtensionGuesser(); |
29dca432 | 30 | $this->fetchingErrorMessage = $fetchingErrorMessage; |
558d9aab JB |
31 | } |
32 | ||
33 | /** | |
9e349f08 | 34 | * Fetch content using graby and hydrate given $entry with results information. |
f1e29e69 | 35 | * In case we couldn't find content, we'll try to use Open Graph data. |
558d9aab | 36 | * |
4d0ec0e7 JB |
37 | * We can also force the content, in case of an import from the v1 for example, so the function won't |
38 | * fetch the content from the website but rather use information given with the $content parameter. | |
39 | * | |
40 | * @param Entry $entry Entry to update | |
41 | * @param string $url Url to grab content for | |
9e349f08 | 42 | * @param array $content An array with AT LEAST keys title, html, url to skip the fetchContent from the url |
558d9aab JB |
43 | * |
44 | * @return Entry | |
45 | */ | |
4d0ec0e7 | 46 | public function updateEntry(Entry $entry, $url, array $content = []) |
558d9aab | 47 | { |
e668a812 JB |
48 | // ensure content is a bit cleaned up |
49 | if (!empty($content['html'])) { | |
0d6cfb88 | 50 | $content['html'] = $this->graby->cleanupHtml($content['html'], $url); |
e668a812 JB |
51 | } |
52 | ||
4d0ec0e7 JB |
53 | // do we have to fetch the content or the provided one is ok? |
54 | if (empty($content) || false === $this->validateContent($content)) { | |
29dca432 | 55 | $fetchedContent = $this->graby->fetchContent($url); |
106bdbcd JB |
56 | |
57 | // when content is imported, we have information in $content | |
58 | // in case fetching content goes bad, we'll keep the imported information instead of overriding them | |
29dca432 JC |
59 | if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) { |
60 | $content = $fetchedContent; | |
61 | } | |
4d0ec0e7 | 62 | } |
558d9aab JB |
63 | |
64 | $title = $content['title']; | |
e668a812 | 65 | if (!$title && !empty($content['open_graph']['og_title'])) { |
558d9aab JB |
66 | $title = $content['open_graph']['og_title']; |
67 | } | |
68 | ||
69 | $html = $content['html']; | |
70 | if (false === $html) { | |
36e6ef52 | 71 | $html = $this->fetchingErrorMessage; |
558d9aab | 72 | |
e668a812 | 73 | if (!empty($content['open_graph']['og_description'])) { |
558d9aab JB |
74 | $html .= '<p><i>But we found a short description: </i></p>'; |
75 | $html .= $content['open_graph']['og_description']; | |
76 | } | |
77 | } | |
78 | ||
79 | $entry->setUrl($content['url'] ?: $url); | |
80 | $entry->setTitle($title); | |
48656e0e | 81 | $entry->setContent($html); |
10b35097 | 82 | $entry->setHttpStatus(isset($content['status']) ? $content['status'] : ''); |
419214d7 | 83 | |
e668a812 | 84 | if (!empty($content['date'])) { |
f0378b4d JB |
85 | $date = $content['date']; |
86 | ||
87 | // is it a timestamp? | |
88 | if (filter_var($date, FILTER_VALIDATE_INT) !== false) { | |
89 | $date = '@'.$content['date']; | |
90 | } | |
91 | ||
e668a812 | 92 | try { |
f0378b4d | 93 | $entry->setPublishedAt(new \DateTime($date)); |
e668a812 | 94 | } catch (\Exception $e) { |
9e349f08 | 95 | $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $url, 'date' => $content['date']]); |
e668a812 | 96 | } |
5e9009ce NL |
97 | } |
98 | ||
7b0b3622 NL |
99 | if (!empty($content['authors'])) { |
100 | $entry->setPublishedBy($content['authors']); | |
101 | } | |
102 | ||
dda6a6ad NL |
103 | if (!empty($content['all_headers'])) { |
104 | $entry->setHeaders($content['all_headers']); | |
105 | } | |
106 | ||
e858018f JC |
107 | $entry->setLanguage(isset($content['language']) ? $content['language'] : ''); |
108 | $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : ''); | |
da3d4998 | 109 | $entry->setReadingTime(Utils::getReadingTime($html)); |
4d0ec0e7 JB |
110 | |
111 | $domainName = parse_url($entry->getUrl(), PHP_URL_HOST); | |
112 | if (false !== $domainName) { | |
113 | $entry->setDomainName($domainName); | |
114 | } | |
558d9aab | 115 | |
e668a812 | 116 | if (!empty($content['open_graph']['og_image'])) { |
558d9aab JB |
117 | $entry->setPreviewPicture($content['open_graph']['og_image']); |
118 | } | |
119 | ||
8d7b4f0e | 120 | // if content is an image define as a preview too |
e668a812 | 121 | if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) { |
8d7b4f0e JB |
122 | $entry->setPreviewPicture($content['url']); |
123 | } | |
124 | ||
1c9cd2a7 KG |
125 | try { |
126 | $this->tagger->tag($entry); | |
127 | } catch (\Exception $e) { | |
4094ea47 | 128 | $this->logger->error('Error while trying to automatically tag an entry.', [ |
1c9cd2a7 KG |
129 | 'entry_url' => $url, |
130 | 'error_msg' => $e->getMessage(), | |
4094ea47 | 131 | ]); |
1c9cd2a7 | 132 | } |
c3510620 | 133 | |
558d9aab JB |
134 | return $entry; |
135 | } | |
c2656f96 | 136 | |
4d0ec0e7 JB |
137 | /** |
138 | * Validate that the given content as enough value to be used | |
139 | * instead of fetch the content from the url. | |
140 | * | |
141 | * @param array $content | |
142 | * | |
143 | * @return bool true if valid otherwise false | |
144 | */ | |
145 | private function validateContent(array $content) | |
146 | { | |
e668a812 | 147 | return !empty($content['title']) && !empty($content['html']) && !empty($content['url']); |
4d0ec0e7 | 148 | } |
558d9aab | 149 | } |