]>
Commit | Line | Data |
---|---|---|
19d9efab JB |
1 | <?php |
2 | ||
3 | namespace Wallabag\ImportBundle\Import; | |
4 | ||
f808b016 JB |
5 | use Doctrine\ORM\EntityManager; |
6 | use OldSound\RabbitMqBundle\RabbitMq\ProducerInterface; | |
19d9efab JB |
7 | use Psr\Log\LoggerInterface; |
8 | use Psr\Log\NullLogger; | |
f808b016 | 9 | use Symfony\Component\EventDispatcher\EventDispatcherInterface; |
19d9efab | 10 | use Wallabag\CoreBundle\Entity\Entry; |
8664069e | 11 | use Wallabag\CoreBundle\Entity\Tag; |
f808b016 JB |
12 | use Wallabag\CoreBundle\Event\EntrySavedEvent; |
13 | use Wallabag\CoreBundle\Helper\ContentProxy; | |
6bc6fb1f | 14 | use Wallabag\CoreBundle\Helper\TagsAssigner; |
3849a9f3 | 15 | use Wallabag\UserBundle\Entity\User; |
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; | |
d5c2cc54 | 27 | protected $disableContentUpdate = false; |
3aca0a9f JB |
28 | protected $skippedEntries = 0; |
29 | protected $importedEntries = 0; | |
c80cc01a | 30 | protected $queuedEntries = 0; |
19d9efab | 31 | |
6bc6fb1f | 32 | public function __construct(EntityManager $em, ContentProxy $contentProxy, TagsAssigner $tagsAssigner, EventDispatcherInterface $eventDispatcher) |
19d9efab JB |
33 | { |
34 | $this->em = $em; | |
35 | $this->logger = new NullLogger(); | |
36 | $this->contentProxy = $contentProxy; | |
6bc6fb1f | 37 | $this->tagsAssigner = $tagsAssigner; |
7816eb62 | 38 | $this->eventDispatcher = $eventDispatcher; |
19d9efab JB |
39 | } |
40 | ||
41 | public function setLogger(LoggerInterface $logger) | |
42 | { | |
43 | $this->logger = $logger; | |
44 | } | |
45 | ||
c98db1b6 | 46 | /** |
b3437d58 | 47 | * Set RabbitMQ/Redis Producer to send each entry to a queue. |
c98db1b6 | 48 | * This method should be called when user has enabled RabbitMQ. |
c98db1b6 | 49 | */ |
b3437d58 | 50 | public function setProducer(ProducerInterface $producer) |
c98db1b6 JB |
51 | { |
52 | $this->producer = $producer; | |
53 | } | |
54 | ||
55 | /** | |
56 | * Set current user. | |
57 | * Could the current *connected* user or one retrieve by the consumer. | |
c98db1b6 | 58 | */ |
3849a9f3 | 59 | public function setUser(User $user) |
c98db1b6 JB |
60 | { |
61 | $this->user = $user; | |
62 | } | |
63 | ||
64 | /** | |
65 | * Set whether articles must be all marked as read. | |
66 | * | |
67 | * @param bool $markAsRead | |
68 | */ | |
69 | public function setMarkAsRead($markAsRead) | |
70 | { | |
71 | $this->markAsRead = $markAsRead; | |
72 | ||
73 | return $this; | |
74 | } | |
75 | ||
76 | /** | |
77 | * Get whether articles must be all marked as read. | |
78 | */ | |
79 | public function getMarkAsRead() | |
80 | { | |
81 | return $this->markAsRead; | |
82 | } | |
83 | ||
d0e9b3d6 JC |
84 | /** |
85 | * Set whether articles should be fetched for updated content. | |
86 | * | |
432a24f5 | 87 | * @param bool $disableContentUpdate |
d0e9b3d6 JC |
88 | */ |
89 | public function setDisableContentUpdate($disableContentUpdate) | |
90 | { | |
91 | $this->disableContentUpdate = $disableContentUpdate; | |
92 | ||
93 | return $this; | |
94 | } | |
95 | ||
f808b016 JB |
96 | /** |
97 | * {@inheritdoc} | |
98 | */ | |
99 | public function getSummary() | |
100 | { | |
101 | return [ | |
102 | 'skipped' => $this->skippedEntries, | |
103 | 'imported' => $this->importedEntries, | |
104 | 'queued' => $this->queuedEntries, | |
105 | ]; | |
106 | } | |
107 | ||
108 | /** | |
109 | * Parse one entry. | |
110 | * | |
f808b016 JB |
111 | * @return Entry |
112 | */ | |
113 | abstract public function parseEntry(array $importedEntry); | |
114 | ||
9f8f188d JB |
115 | /** |
116 | * Validate that an entry is valid (like has some required keys, etc.). | |
117 | * | |
9f8f188d JB |
118 | * @return bool |
119 | */ | |
120 | abstract public function validateEntry(array $importedEntry); | |
121 | ||
19d9efab JB |
122 | /** |
123 | * Fetch content from the ContentProxy (using graby). | |
59b97fae | 124 | * If it fails return the given entry to be saved in all case (to avoid user to loose the content). |
19d9efab JB |
125 | * |
126 | * @param Entry $entry Entry to update | |
127 | * @param string $url Url to grab content for | |
128 | * @param array $content An array with AT LEAST keys title, html, url, language & content_type to skip the fetchContent from the url | |
19d9efab JB |
129 | */ |
130 | protected function fetchContent(Entry $entry, $url, array $content = []) | |
131 | { | |
132 | try { | |
6acadf8e | 133 | $this->contentProxy->updateEntry($entry, $url, $content, $this->disableContentUpdate); |
19d9efab | 134 | } catch (\Exception $e) { |
d0e9b3d6 | 135 | $this->logger->error('Error trying to import an entry.', [ |
6acadf8e | 136 | 'entry_url' => $url, |
d0e9b3d6 JC |
137 | 'error_msg' => $e->getMessage(), |
138 | ]); | |
19d9efab JB |
139 | } |
140 | } | |
c98db1b6 JB |
141 | |
142 | /** | |
143 | * Parse and insert all given entries. | |
c98db1b6 | 144 | */ |
9f8f188d | 145 | protected function parseEntries(array $entries) |
c98db1b6 JB |
146 | { |
147 | $i = 1; | |
7816eb62 | 148 | $entryToBeFlushed = []; |
c98db1b6 JB |
149 | |
150 | foreach ($entries as $importedEntry) { | |
ff1a5362 JB |
151 | if ($this->markAsRead) { |
152 | $importedEntry = $this->setEntryAsRead($importedEntry); | |
153 | } | |
154 | ||
9f8f188d JB |
155 | if (false === $this->validateEntry($importedEntry)) { |
156 | continue; | |
157 | } | |
158 | ||
c98db1b6 JB |
159 | $entry = $this->parseEntry($importedEntry); |
160 | ||
161 | if (null === $entry) { | |
162 | continue; | |
163 | } | |
164 | ||
7816eb62 JB |
165 | // store each entry to be flushed so we can trigger the entry.saved event for each of them |
166 | // entry.saved needs the entry to be persisted in db because it needs it id to generate | |
167 | // images (at least) | |
168 | $entryToBeFlushed[] = $entry; | |
169 | ||
c98db1b6 | 170 | // flush every 20 entries |
3ef055ce | 171 | if (0 === ($i % 20)) { |
c98db1b6 | 172 | $this->em->flush(); |
8664069e | 173 | |
7816eb62 JB |
174 | foreach ($entryToBeFlushed as $entry) { |
175 | $this->eventDispatcher->dispatch(EntrySavedEvent::NAME, new EntrySavedEvent($entry)); | |
176 | } | |
177 | ||
178 | $entryToBeFlushed = []; | |
179 | ||
8664069e JB |
180 | // clear only affected entities |
181 | $this->em->clear(Entry::class); | |
182 | $this->em->clear(Tag::class); | |
c98db1b6 JB |
183 | } |
184 | ++$i; | |
185 | } | |
186 | ||
187 | $this->em->flush(); | |
7816eb62 JB |
188 | |
189 | if (!empty($entryToBeFlushed)) { | |
190 | foreach ($entryToBeFlushed as $entry) { | |
191 | $this->eventDispatcher->dispatch(EntrySavedEvent::NAME, new EntrySavedEvent($entry)); | |
192 | } | |
193 | } | |
c98db1b6 JB |
194 | } |
195 | ||
3849a9f3 JB |
196 | /** |
197 | * Parse entries and send them to the queue. | |
198 | * It should just be a simple loop on all item, no call to the database should be done | |
199 | * to speedup queuing. | |
200 | * | |
201 | * Faster parse entries for Producer. | |
202 | * We don't care to make check at this time. They'll be done by the consumer. | |
3849a9f3 JB |
203 | */ |
204 | protected function parseEntriesForProducer(array $entries) | |
205 | { | |
206 | foreach ($entries as $importedEntry) { | |
207 | // set userId for the producer (it won't know which user is connected) | |
208 | $importedEntry['userId'] = $this->user->getId(); | |
209 | ||
210 | if ($this->markAsRead) { | |
211 | $importedEntry = $this->setEntryAsRead($importedEntry); | |
212 | } | |
213 | ||
c80cc01a | 214 | ++$this->queuedEntries; |
3849a9f3 JB |
215 | |
216 | $this->producer->publish(json_encode($importedEntry)); | |
217 | } | |
218 | } | |
219 | ||
3849a9f3 JB |
220 | /** |
221 | * Set current imported entry to archived / read. | |
222 | * Implementation is different accross all imports. | |
223 | * | |
3849a9f3 JB |
224 | * @return array |
225 | */ | |
226 | abstract protected function setEntryAsRead(array $importedEntry); | |
19d9efab | 227 | } |