]>
Commit | Line | Data |
---|---|---|
558d9aab JB |
1 | <?php |
2 | ||
3 | namespace Wallabag\CoreBundle\Helper; | |
4 | ||
5 | use Graby\Graby; | |
1c9cd2a7 | 6 | use Psr\Log\LoggerInterface as Logger; |
558d9aab | 7 | use Wallabag\CoreBundle\Entity\Entry; |
c2656f96 | 8 | use Wallabag\CoreBundle\Entity\Tag; |
da3d4998 | 9 | use Wallabag\CoreBundle\Tools\Utils; |
c2656f96 | 10 | use Wallabag\CoreBundle\Repository\TagRepository; |
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; |
c2656f96 | 21 | protected $tagRepository; |
558d9aab | 22 | |
c2656f96 | 23 | public function __construct(Graby $graby, RuleBasedTagger $tagger, TagRepository $tagRepository, Logger $logger) |
558d9aab | 24 | { |
347fa6be | 25 | $this->graby = $graby; |
c3510620 | 26 | $this->tagger = $tagger; |
1c9cd2a7 | 27 | $this->logger = $logger; |
c2656f96 | 28 | $this->tagRepository = $tagRepository; |
558d9aab JB |
29 | } |
30 | ||
31 | /** | |
32 | * Fetch content using graby and hydrate given entry with results information. | |
f1e29e69 | 33 | * In case we couldn't find content, we'll try to use Open Graph data. |
558d9aab | 34 | * |
4d0ec0e7 JB |
35 | * We can also force the content, in case of an import from the v1 for example, so the function won't |
36 | * fetch the content from the website but rather use information given with the $content parameter. | |
37 | * | |
38 | * @param Entry $entry Entry to update | |
39 | * @param string $url Url to grab content for | |
40 | * @param array $content An array with AT LEAST keys title, html, url, language & content_type to skip the fetchContent from the url | |
558d9aab JB |
41 | * |
42 | * @return Entry | |
43 | */ | |
4d0ec0e7 | 44 | public function updateEntry(Entry $entry, $url, array $content = []) |
558d9aab | 45 | { |
4d0ec0e7 JB |
46 | // do we have to fetch the content or the provided one is ok? |
47 | if (empty($content) || false === $this->validateContent($content)) { | |
48 | $content = $this->graby->fetchContent($url); | |
49 | } | |
558d9aab JB |
50 | |
51 | $title = $content['title']; | |
52 | if (!$title && isset($content['open_graph']['og_title'])) { | |
53 | $title = $content['open_graph']['og_title']; | |
54 | } | |
55 | ||
56 | $html = $content['html']; | |
57 | if (false === $html) { | |
58 | $html = '<p>Unable to retrieve readable content.</p>'; | |
59 | ||
60 | if (isset($content['open_graph']['og_description'])) { | |
61 | $html .= '<p><i>But we found a short description: </i></p>'; | |
62 | $html .= $content['open_graph']['og_description']; | |
63 | } | |
64 | } | |
65 | ||
66 | $entry->setUrl($content['url'] ?: $url); | |
67 | $entry->setTitle($title); | |
419214d7 | 68 | |
98f0929f | 69 | $entry->setLanguage($content['language']); |
558d9aab | 70 | $entry->setMimetype($content['content_type']); |
da3d4998 | 71 | $entry->setReadingTime(Utils::getReadingTime($html)); |
4d0ec0e7 JB |
72 | |
73 | $domainName = parse_url($entry->getUrl(), PHP_URL_HOST); | |
74 | if (false !== $domainName) { | |
75 | $entry->setDomainName($domainName); | |
76 | } | |
558d9aab | 77 | |
419214d7 TC |
78 | if (true) { |
79 | $this->logger->log('debug','Starting to download images'); | |
80 | $downloadImages = new DownloadImages($html, $url, $this->logger); | |
81 | $html = $downloadImages->process(); | |
82 | } | |
83 | ||
84 | $entry->setContent($html); | |
85 | ||
558d9aab JB |
86 | if (isset($content['open_graph']['og_image'])) { |
87 | $entry->setPreviewPicture($content['open_graph']['og_image']); | |
88 | } | |
89 | ||
1c9cd2a7 KG |
90 | try { |
91 | $this->tagger->tag($entry); | |
92 | } catch (\Exception $e) { | |
4094ea47 | 93 | $this->logger->error('Error while trying to automatically tag an entry.', [ |
1c9cd2a7 KG |
94 | 'entry_url' => $url, |
95 | 'error_msg' => $e->getMessage(), | |
4094ea47 | 96 | ]); |
1c9cd2a7 | 97 | } |
c3510620 | 98 | |
558d9aab JB |
99 | return $entry; |
100 | } | |
c2656f96 JB |
101 | |
102 | /** | |
103 | * Assign some tags to an entry. | |
104 | * | |
105 | * @param Entry $entry | |
82fc3290 JB |
106 | * @param array|string $tags An array of tag or a string coma separated of tag |
107 | * @param array $entitiesReady Entities from the EntityManager which are persisted but not yet flushed | |
108 | * It is mostly to fix duplicate tag on import @see http://stackoverflow.com/a/7879164/569101 | |
c2656f96 | 109 | */ |
40113585 | 110 | public function assignTagsToEntry(Entry $entry, $tags, array $entitiesReady = []) |
c2656f96 JB |
111 | { |
112 | if (!is_array($tags)) { | |
113 | $tags = explode(',', $tags); | |
114 | } | |
115 | ||
40113585 JB |
116 | // keeps only Tag entity from the "not yet flushed entities" |
117 | $tagsNotYetFlushed = []; | |
118 | foreach ($entitiesReady as $entity) { | |
119 | if ($entity instanceof Tag) { | |
120 | $tagsNotYetFlushed[$entity->getLabel()] = $entity; | |
121 | } | |
122 | } | |
123 | ||
c2656f96 JB |
124 | foreach ($tags as $label) { |
125 | $label = trim($label); | |
126 | ||
127 | // avoid empty tag | |
128 | if (0 === strlen($label)) { | |
129 | continue; | |
130 | } | |
131 | ||
40113585 JB |
132 | if (isset($tagsNotYetFlushed[$label])) { |
133 | $tagEntity = $tagsNotYetFlushed[$label]; | |
134 | } else { | |
135 | $tagEntity = $this->tagRepository->findOneByLabel($label); | |
c2656f96 | 136 | |
40113585 JB |
137 | if (is_null($tagEntity)) { |
138 | $tagEntity = new Tag(); | |
139 | $tagEntity->setLabel($label); | |
140 | } | |
c2656f96 JB |
141 | } |
142 | ||
143 | // only add the tag on the entry if the relation doesn't exist | |
144 | if (false === $entry->getTags()->contains($tagEntity)) { | |
145 | $entry->addTag($tagEntity); | |
146 | } | |
147 | } | |
148 | } | |
4d0ec0e7 JB |
149 | |
150 | /** | |
151 | * Validate that the given content as enough value to be used | |
152 | * instead of fetch the content from the url. | |
153 | * | |
154 | * @param array $content | |
155 | * | |
156 | * @return bool true if valid otherwise false | |
157 | */ | |
158 | private function validateContent(array $content) | |
159 | { | |
160 | return isset($content['title']) && isset($content['html']) && isset($content['url']) && isset($content['language']) && isset($content['content_type']); | |
161 | } | |
558d9aab | 162 | } |