]>
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 | 43 | */ |
4d0ec0e7 | 44 | public function updateEntry(Entry $entry, $url, array $content = []) |
558d9aab | 45 | { |
e668a812 JB |
46 | // ensure content is a bit cleaned up |
47 | if (!empty($content['html'])) { | |
0d6cfb88 | 48 | $content['html'] = $this->graby->cleanupHtml($content['html'], $url); |
e668a812 JB |
49 | } |
50 | ||
4d0ec0e7 JB |
51 | // do we have to fetch the content or the provided one is ok? |
52 | if (empty($content) || false === $this->validateContent($content)) { | |
29dca432 | 53 | $fetchedContent = $this->graby->fetchContent($url); |
106bdbcd JB |
54 | |
55 | // when content is imported, we have information in $content | |
56 | // in case fetching content goes bad, we'll keep the imported information instead of overriding them | |
29dca432 JC |
57 | if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) { |
58 | $content = $fetchedContent; | |
59 | } | |
4d0ec0e7 | 60 | } |
558d9aab JB |
61 | |
62 | $title = $content['title']; | |
e668a812 | 63 | if (!$title && !empty($content['open_graph']['og_title'])) { |
558d9aab JB |
64 | $title = $content['open_graph']['og_title']; |
65 | } | |
66 | ||
67 | $html = $content['html']; | |
68 | if (false === $html) { | |
36e6ef52 | 69 | $html = $this->fetchingErrorMessage; |
558d9aab | 70 | |
e668a812 | 71 | if (!empty($content['open_graph']['og_description'])) { |
558d9aab JB |
72 | $html .= '<p><i>But we found a short description: </i></p>'; |
73 | $html .= $content['open_graph']['og_description']; | |
74 | } | |
75 | } | |
76 | ||
77 | $entry->setUrl($content['url'] ?: $url); | |
78 | $entry->setTitle($title); | |
48656e0e | 79 | $entry->setContent($html); |
10b35097 | 80 | $entry->setHttpStatus(isset($content['status']) ? $content['status'] : ''); |
419214d7 | 81 | |
e668a812 | 82 | if (!empty($content['date'])) { |
f0378b4d JB |
83 | $date = $content['date']; |
84 | ||
85 | // is it a timestamp? | |
86 | if (filter_var($date, FILTER_VALIDATE_INT) !== false) { | |
87 | $date = '@'.$content['date']; | |
88 | } | |
89 | ||
e668a812 | 90 | try { |
f0378b4d | 91 | $entry->setPublishedAt(new \DateTime($date)); |
e668a812 | 92 | } catch (\Exception $e) { |
9e349f08 | 93 | $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $url, 'date' => $content['date']]); |
e668a812 | 94 | } |
5e9009ce NL |
95 | } |
96 | ||
7b0b3622 NL |
97 | if (!empty($content['authors'])) { |
98 | $entry->setPublishedBy($content['authors']); | |
99 | } | |
100 | ||
dda6a6ad NL |
101 | if (!empty($content['all_headers'])) { |
102 | $entry->setHeaders($content['all_headers']); | |
103 | } | |
104 | ||
e858018f JC |
105 | $entry->setLanguage(isset($content['language']) ? $content['language'] : ''); |
106 | $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : ''); | |
da3d4998 | 107 | $entry->setReadingTime(Utils::getReadingTime($html)); |
4d0ec0e7 JB |
108 | |
109 | $domainName = parse_url($entry->getUrl(), PHP_URL_HOST); | |
110 | if (false !== $domainName) { | |
111 | $entry->setDomainName($domainName); | |
112 | } | |
558d9aab | 113 | |
e668a812 | 114 | if (!empty($content['open_graph']['og_image'])) { |
558d9aab JB |
115 | $entry->setPreviewPicture($content['open_graph']['og_image']); |
116 | } | |
117 | ||
8d7b4f0e | 118 | // if content is an image define as a preview too |
e668a812 | 119 | if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) { |
8d7b4f0e JB |
120 | $entry->setPreviewPicture($content['url']); |
121 | } | |
122 | ||
1c9cd2a7 KG |
123 | try { |
124 | $this->tagger->tag($entry); | |
125 | } catch (\Exception $e) { | |
4094ea47 | 126 | $this->logger->error('Error while trying to automatically tag an entry.', [ |
1c9cd2a7 KG |
127 | 'entry_url' => $url, |
128 | 'error_msg' => $e->getMessage(), | |
4094ea47 | 129 | ]); |
1c9cd2a7 | 130 | } |
558d9aab | 131 | } |
c2656f96 | 132 | |
4d0ec0e7 JB |
133 | /** |
134 | * Validate that the given content as enough value to be used | |
135 | * instead of fetch the content from the url. | |
136 | * | |
137 | * @param array $content | |
138 | * | |
139 | * @return bool true if valid otherwise false | |
140 | */ | |
141 | private function validateContent(array $content) | |
142 | { | |
e668a812 | 143 | return !empty($content['title']) && !empty($content['html']) && !empty($content['url']); |
4d0ec0e7 | 144 | } |
558d9aab | 145 | } |