]>
Commit | Line | Data |
---|---|---|
19d9efab JB |
1 | <?php |
2 | ||
3 | namespace Wallabag\ImportBundle\Import; | |
4 | ||
5 | use Psr\Log\LoggerInterface; | |
6 | use Psr\Log\NullLogger; | |
7 | use Doctrine\ORM\EntityManager; | |
8 | use Wallabag\CoreBundle\Helper\ContentProxy; | |
9 | use Wallabag\CoreBundle\Entity\Entry; | |
8664069e | 10 | use Wallabag\CoreBundle\Entity\Tag; |
6bc6fb1f | 11 | use Wallabag\CoreBundle\Helper\TagsAssigner; |
3849a9f3 | 12 | use Wallabag\UserBundle\Entity\User; |
b3437d58 | 13 | use OldSound\RabbitMqBundle\RabbitMq\ProducerInterface; |
7816eb62 JB |
14 | use Symfony\Component\EventDispatcher\EventDispatcherInterface; |
15 | use Wallabag\CoreBundle\Event\EntrySavedEvent; | |
19d9efab JB |
16 | |
17 | abstract class AbstractImport implements ImportInterface | |
18 | { | |
19 | protected $em; | |
20 | protected $logger; | |
21 | protected $contentProxy; | |
6bc6fb1f | 22 | protected $tagsAssigner; |
7816eb62 | 23 | protected $eventDispatcher; |
c98db1b6 JB |
24 | protected $producer; |
25 | protected $user; | |
26 | protected $markAsRead; | |
3aca0a9f JB |
27 | protected $skippedEntries = 0; |
28 | protected $importedEntries = 0; | |
c80cc01a | 29 | protected $queuedEntries = 0; |
19d9efab | 30 | |
6bc6fb1f | 31 | public function __construct(EntityManager $em, ContentProxy $contentProxy, TagsAssigner $tagsAssigner, EventDispatcherInterface $eventDispatcher) |
19d9efab JB |
32 | { |
33 | $this->em = $em; | |
34 | $this->logger = new NullLogger(); | |
35 | $this->contentProxy = $contentProxy; | |
6bc6fb1f | 36 | $this->tagsAssigner = $tagsAssigner; |
7816eb62 | 37 | $this->eventDispatcher = $eventDispatcher; |
19d9efab JB |
38 | } |
39 | ||
40 | public function setLogger(LoggerInterface $logger) | |
41 | { | |
42 | $this->logger = $logger; | |
43 | } | |
44 | ||
c98db1b6 | 45 | /** |
b3437d58 | 46 | * Set RabbitMQ/Redis Producer to send each entry to a queue. |
c98db1b6 JB |
47 | * This method should be called when user has enabled RabbitMQ. |
48 | * | |
b3437d58 | 49 | * @param ProducerInterface $producer |
c98db1b6 | 50 | */ |
b3437d58 | 51 | public function setProducer(ProducerInterface $producer) |
c98db1b6 JB |
52 | { |
53 | $this->producer = $producer; | |
54 | } | |
55 | ||
56 | /** | |
57 | * Set current user. | |
58 | * Could the current *connected* user or one retrieve by the consumer. | |
59 | * | |
3849a9f3 | 60 | * @param User $user |
c98db1b6 | 61 | */ |
3849a9f3 | 62 | public function setUser(User $user) |
c98db1b6 JB |
63 | { |
64 | $this->user = $user; | |
65 | } | |
66 | ||
67 | /** | |
68 | * Set whether articles must be all marked as read. | |
69 | * | |
70 | * @param bool $markAsRead | |
71 | */ | |
72 | public function setMarkAsRead($markAsRead) | |
73 | { | |
74 | $this->markAsRead = $markAsRead; | |
75 | ||
76 | return $this; | |
77 | } | |
78 | ||
79 | /** | |
80 | * Get whether articles must be all marked as read. | |
81 | */ | |
82 | public function getMarkAsRead() | |
83 | { | |
84 | return $this->markAsRead; | |
85 | } | |
86 | ||
19d9efab JB |
87 | /** |
88 | * Fetch content from the ContentProxy (using graby). | |
59b97fae | 89 | * If it fails return the given entry to be saved in all case (to avoid user to loose the content). |
19d9efab JB |
90 | * |
91 | * @param Entry $entry Entry to update | |
92 | * @param string $url Url to grab content for | |
93 | * @param array $content An array with AT LEAST keys title, html, url, language & content_type to skip the fetchContent from the url | |
94 | * | |
59b97fae | 95 | * @return Entry |
19d9efab JB |
96 | */ |
97 | protected function fetchContent(Entry $entry, $url, array $content = []) | |
98 | { | |
99 | try { | |
100 | return $this->contentProxy->updateEntry($entry, $url, $content); | |
101 | } catch (\Exception $e) { | |
59b97fae | 102 | return $entry; |
19d9efab JB |
103 | } |
104 | } | |
c98db1b6 JB |
105 | |
106 | /** | |
107 | * Parse and insert all given entries. | |
108 | * | |
109 | * @param $entries | |
110 | */ | |
111 | protected function parseEntries($entries) | |
112 | { | |
113 | $i = 1; | |
7816eb62 | 114 | $entryToBeFlushed = []; |
c98db1b6 JB |
115 | |
116 | foreach ($entries as $importedEntry) { | |
ff1a5362 JB |
117 | if ($this->markAsRead) { |
118 | $importedEntry = $this->setEntryAsRead($importedEntry); | |
119 | } | |
120 | ||
c98db1b6 JB |
121 | $entry = $this->parseEntry($importedEntry); |
122 | ||
123 | if (null === $entry) { | |
124 | continue; | |
125 | } | |
126 | ||
7816eb62 JB |
127 | // store each entry to be flushed so we can trigger the entry.saved event for each of them |
128 | // entry.saved needs the entry to be persisted in db because it needs it id to generate | |
129 | // images (at least) | |
130 | $entryToBeFlushed[] = $entry; | |
131 | ||
c98db1b6 JB |
132 | // flush every 20 entries |
133 | if (($i % 20) === 0) { | |
134 | $this->em->flush(); | |
8664069e | 135 | |
7816eb62 JB |
136 | foreach ($entryToBeFlushed as $entry) { |
137 | $this->eventDispatcher->dispatch(EntrySavedEvent::NAME, new EntrySavedEvent($entry)); | |
138 | } | |
139 | ||
140 | $entryToBeFlushed = []; | |
141 | ||
8664069e JB |
142 | // clear only affected entities |
143 | $this->em->clear(Entry::class); | |
144 | $this->em->clear(Tag::class); | |
c98db1b6 JB |
145 | } |
146 | ++$i; | |
147 | } | |
148 | ||
149 | $this->em->flush(); | |
7816eb62 JB |
150 | |
151 | if (!empty($entryToBeFlushed)) { | |
152 | foreach ($entryToBeFlushed as $entry) { | |
153 | $this->eventDispatcher->dispatch(EntrySavedEvent::NAME, new EntrySavedEvent($entry)); | |
154 | } | |
155 | } | |
c98db1b6 JB |
156 | } |
157 | ||
3849a9f3 JB |
158 | /** |
159 | * Parse entries and send them to the queue. | |
160 | * It should just be a simple loop on all item, no call to the database should be done | |
161 | * to speedup queuing. | |
162 | * | |
163 | * Faster parse entries for Producer. | |
164 | * We don't care to make check at this time. They'll be done by the consumer. | |
165 | * | |
166 | * @param array $entries | |
167 | */ | |
168 | protected function parseEntriesForProducer(array $entries) | |
169 | { | |
170 | foreach ($entries as $importedEntry) { | |
171 | // set userId for the producer (it won't know which user is connected) | |
172 | $importedEntry['userId'] = $this->user->getId(); | |
173 | ||
174 | if ($this->markAsRead) { | |
175 | $importedEntry = $this->setEntryAsRead($importedEntry); | |
176 | } | |
177 | ||
c80cc01a | 178 | ++$this->queuedEntries; |
3849a9f3 JB |
179 | |
180 | $this->producer->publish(json_encode($importedEntry)); | |
181 | } | |
182 | } | |
183 | ||
c80cc01a JB |
184 | /** |
185 | * {@inheritdoc} | |
186 | */ | |
187 | public function getSummary() | |
188 | { | |
189 | return [ | |
190 | 'skipped' => $this->skippedEntries, | |
191 | 'imported' => $this->importedEntries, | |
192 | 'queued' => $this->queuedEntries, | |
193 | ]; | |
194 | } | |
195 | ||
c98db1b6 JB |
196 | /** |
197 | * Parse one entry. | |
198 | * | |
199 | * @param array $importedEntry | |
200 | * | |
201 | * @return Entry | |
202 | */ | |
203 | abstract public function parseEntry(array $importedEntry); | |
3849a9f3 JB |
204 | |
205 | /** | |
206 | * Set current imported entry to archived / read. | |
207 | * Implementation is different accross all imports. | |
208 | * | |
209 | * @param array $importedEntry | |
210 | * | |
211 | * @return array | |
212 | */ | |
213 | abstract protected function setEntryAsRead(array $importedEntry); | |
19d9efab | 214 | } |