]>
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; |
d0e9b3d6 | 10 | use Symfony\Component\Config\Definition\Exception\Exception; |
558d9aab JB |
11 | |
12 | /** | |
13 | * This kind of proxy class take care of getting the content from an url | |
f1e29e69 | 14 | * and update the entry with what it found. |
558d9aab JB |
15 | */ |
16 | class ContentProxy | |
17 | { | |
18 | protected $graby; | |
c3510620 | 19 | protected $tagger; |
1c9cd2a7 | 20 | protected $logger; |
8d7b4f0e | 21 | protected $mimeGuesser; |
29dca432 | 22 | protected $fetchingErrorMessage; |
6bc6fb1f | 23 | protected $eventDispatcher; |
558d9aab | 24 | |
6bc6fb1f | 25 | public function __construct(Graby $graby, RuleBasedTagger $tagger, LoggerInterface $logger, $fetchingErrorMessage) |
558d9aab | 26 | { |
347fa6be | 27 | $this->graby = $graby; |
c3510620 | 28 | $this->tagger = $tagger; |
1c9cd2a7 | 29 | $this->logger = $logger; |
8d7b4f0e | 30 | $this->mimeGuesser = new MimeTypeExtensionGuesser(); |
29dca432 | 31 | $this->fetchingErrorMessage = $fetchingErrorMessage; |
558d9aab JB |
32 | } |
33 | ||
34 | /** | |
d0e9b3d6 | 35 | * Update existing entry by fetching from URL using Graby. |
4d0ec0e7 JB |
36 | * |
37 | * @param Entry $entry Entry to update | |
38 | * @param string $url Url to grab content for | |
558d9aab | 39 | */ |
d0e9b3d6 | 40 | public function updateEntry(Entry $entry, $url) |
558d9aab | 41 | { |
d0e9b3d6 JC |
42 | $content = $this->graby->fetchContent($url); |
43 | ||
44 | $this->stockEntry($entry, $content); | |
45 | } | |
46 | ||
47 | /** | |
48 | * Import entry using either fetched or provided content. | |
49 | * | |
50 | * @param Entry $entry Entry to update | |
51 | * @param array $content Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url | |
52 | * @param bool $disableContentUpdate Whether to skip trying to fetch content using Graby | |
53 | */ | |
54 | public function importEntry(Entry $entry, array $content, $disableContentUpdate = false) | |
55 | { | |
56 | $this->validateContent($content); | |
e668a812 | 57 | |
d0e9b3d6 JC |
58 | if (false === $disableContentUpdate) { |
59 | try { | |
60 | $fetchedContent = $this->graby->fetchContent($content['url']); | |
61 | } catch (\Exception $e) { | |
62 | $this->logger->error('Error while trying to fetch content from URL.', [ | |
63 | 'entry_url' => $content['url'], | |
64 | 'error_msg' => $e->getMessage(), | |
65 | ]); | |
66 | } | |
106bdbcd JB |
67 | |
68 | // when content is imported, we have information in $content | |
69 | // in case fetching content goes bad, we'll keep the imported information instead of overriding them | |
d0e9b3d6 | 70 | if ($fetchedContent['html'] !== $this->fetchingErrorMessage) { |
29dca432 JC |
71 | $content = $fetchedContent; |
72 | } | |
4d0ec0e7 | 73 | } |
558d9aab | 74 | |
d0e9b3d6 JC |
75 | $this->stockEntry($entry, $content); |
76 | } | |
77 | ||
78 | /** | |
79 | * Stock entry with fetched or imported content. | |
80 | * Will fall back to OpenGraph data if available. | |
81 | * | |
82 | * @param Entry $entry Entry to stock | |
83 | * @param array $content Array with at least title and URL | |
84 | */ | |
85 | private function stockEntry(Entry $entry, array $content) | |
86 | { | |
558d9aab | 87 | $title = $content['title']; |
e668a812 | 88 | if (!$title && !empty($content['open_graph']['og_title'])) { |
558d9aab JB |
89 | $title = $content['open_graph']['og_title']; |
90 | } | |
91 | ||
92 | $html = $content['html']; | |
93 | if (false === $html) { | |
36e6ef52 | 94 | $html = $this->fetchingErrorMessage; |
558d9aab | 95 | |
e668a812 | 96 | if (!empty($content['open_graph']['og_description'])) { |
558d9aab JB |
97 | $html .= '<p><i>But we found a short description: </i></p>'; |
98 | $html .= $content['open_graph']['og_description']; | |
99 | } | |
100 | } | |
101 | ||
d0e9b3d6 | 102 | $entry->setUrl($content['url']); |
558d9aab | 103 | $entry->setTitle($title); |
48656e0e | 104 | $entry->setContent($html); |
10b35097 | 105 | $entry->setHttpStatus(isset($content['status']) ? $content['status'] : ''); |
419214d7 | 106 | |
e668a812 | 107 | if (!empty($content['date'])) { |
f0378b4d JB |
108 | $date = $content['date']; |
109 | ||
110 | // is it a timestamp? | |
111 | if (filter_var($date, FILTER_VALIDATE_INT) !== false) { | |
112 | $date = '@'.$content['date']; | |
113 | } | |
114 | ||
e668a812 | 115 | try { |
f0378b4d | 116 | $entry->setPublishedAt(new \DateTime($date)); |
e668a812 | 117 | } catch (\Exception $e) { |
9e349f08 | 118 | $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $url, 'date' => $content['date']]); |
e668a812 | 119 | } |
5e9009ce NL |
120 | } |
121 | ||
7b0b3622 NL |
122 | if (!empty($content['authors'])) { |
123 | $entry->setPublishedBy($content['authors']); | |
124 | } | |
125 | ||
dda6a6ad NL |
126 | if (!empty($content['all_headers'])) { |
127 | $entry->setHeaders($content['all_headers']); | |
128 | } | |
129 | ||
e858018f JC |
130 | $entry->setLanguage(isset($content['language']) ? $content['language'] : ''); |
131 | $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : ''); | |
da3d4998 | 132 | $entry->setReadingTime(Utils::getReadingTime($html)); |
4d0ec0e7 JB |
133 | |
134 | $domainName = parse_url($entry->getUrl(), PHP_URL_HOST); | |
135 | if (false !== $domainName) { | |
136 | $entry->setDomainName($domainName); | |
137 | } | |
558d9aab | 138 | |
e668a812 | 139 | if (!empty($content['open_graph']['og_image'])) { |
558d9aab JB |
140 | $entry->setPreviewPicture($content['open_graph']['og_image']); |
141 | } | |
142 | ||
8d7b4f0e | 143 | // if content is an image define as a preview too |
e668a812 | 144 | if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) { |
8d7b4f0e JB |
145 | $entry->setPreviewPicture($content['url']); |
146 | } | |
147 | ||
1c9cd2a7 KG |
148 | try { |
149 | $this->tagger->tag($entry); | |
150 | } catch (\Exception $e) { | |
4094ea47 | 151 | $this->logger->error('Error while trying to automatically tag an entry.', [ |
d0e9b3d6 | 152 | 'entry_url' => $content['url'], |
1c9cd2a7 | 153 | 'error_msg' => $e->getMessage(), |
4094ea47 | 154 | ]); |
1c9cd2a7 | 155 | } |
558d9aab | 156 | } |
c2656f96 | 157 | |
4d0ec0e7 | 158 | /** |
d0e9b3d6 | 159 | * Validate that the given content has at least a title, an html and a url. |
4d0ec0e7 JB |
160 | * |
161 | * @param array $content | |
4d0ec0e7 JB |
162 | */ |
163 | private function validateContent(array $content) | |
164 | { | |
d0e9b3d6 JC |
165 | if (!empty($content['title']))) { |
166 | throw new Exception('Missing title from imported entry!'); | |
167 | } | |
168 | ||
169 | if (!empty($content['url']))) { | |
170 | throw new Exception('Missing URL from imported entry!'); | |
171 | } | |
172 | ||
173 | if (!empty($content['html']))) { | |
174 | throw new Exception('Missing html from imported entry!'); | |
175 | } | |
4d0ec0e7 | 176 | } |
558d9aab | 177 | } |