]>
Commit | Line | Data |
---|---|---|
ae669126 TC |
1 | <?php |
2 | ||
3 | namespace Wallabag\ImportBundle\Import; | |
4 | ||
ae669126 TC |
5 | use Wallabag\CoreBundle\Entity\Entry; |
6 | use Wallabag\UserBundle\Entity\User; | |
7 | use Wallabag\CoreBundle\Helper\ContentProxy; | |
8 | ||
59201088 | 9 | abstract class BrowserImport extends AbstractImport |
ae669126 | 10 | { |
ae669126 | 11 | protected $filepath; |
ae669126 TC |
12 | |
13 | /** | |
14 | * {@inheritdoc} | |
15 | */ | |
59201088 | 16 | abstract public function getName(); |
ae669126 TC |
17 | |
18 | /** | |
19 | * {@inheritdoc} | |
20 | */ | |
59201088 | 21 | abstract public function getUrl(); |
ae669126 TC |
22 | |
23 | /** | |
24 | * {@inheritdoc} | |
25 | */ | |
59201088 | 26 | abstract public function getDescription(); |
ae669126 TC |
27 | |
28 | /** | |
29 | * {@inheritdoc} | |
30 | */ | |
31 | public function import() | |
32 | { | |
33 | if (!$this->user) { | |
64b1229b | 34 | $this->logger->error('Wallabag Browser Import: user is not defined'); |
ae669126 TC |
35 | |
36 | return false; | |
37 | } | |
38 | ||
39 | if (!file_exists($this->filepath) || !is_readable($this->filepath)) { | |
64b1229b | 40 | $this->logger->error('Wallabag Browser Import: unable to read file', ['filepath' => $this->filepath]); |
ae669126 TC |
41 | |
42 | return false; | |
43 | } | |
44 | ||
45 | $data = json_decode(file_get_contents($this->filepath), true); | |
46 | ||
47 | if (empty($data)) { | |
48 | return false; | |
49 | } | |
50 | ||
59201088 TC |
51 | if ($this->producer) { |
52 | $this->parseEntriesForProducer($data); | |
53 | ||
54 | return true; | |
55 | } | |
56 | ||
ae669126 | 57 | $this->parseEntries($data); |
ae669126 TC |
58 | |
59 | return true; | |
60 | } | |
61 | ||
59201088 TC |
62 | /** |
63 | * Set file path to the json file. | |
64 | * | |
65 | * @param string $filepath | |
66 | */ | |
67 | public function setFilepath($filepath) | |
68 | { | |
69 | $this->filepath = $filepath; | |
70 | ||
71 | return $this; | |
72 | } | |
73 | ||
74 | /** | |
75 | * Parse and insert all given entries. | |
76 | * | |
77 | * @param $entries | |
78 | */ | |
79 | protected function parseEntries($entries) | |
ae669126 | 80 | { |
59201088 TC |
81 | $i = 1; |
82 | ||
83 | foreach ($entries as $importedEntry) { | |
84 | if ((array) $importedEntry !== $importedEntry) { | |
85 | continue; | |
86 | } | |
87 | ||
88 | $entry = $this->parseEntry($importedEntry); | |
89 | ||
90 | if (null === $entry) { | |
91 | continue; | |
92 | } | |
93 | ||
94 | // flush every 20 entries | |
95 | if (($i % 20) === 0) { | |
96 | $this->em->flush(); | |
59201088 TC |
97 | } |
98 | ++$i; | |
ae669126 | 99 | } |
59201088 TC |
100 | |
101 | $this->em->flush(); | |
ae669126 TC |
102 | } |
103 | ||
59201088 TC |
104 | /** |
105 | * Parse entries and send them to the queue. | |
106 | * It should just be a simple loop on all item, no call to the database should be done | |
107 | * to speedup queuing. | |
108 | * | |
109 | * Faster parse entries for Producer. | |
110 | * We don't care to make check at this time. They'll be done by the consumer. | |
111 | * | |
112 | * @param array $entries | |
113 | */ | |
114 | protected function parseEntriesForProducer(array $entries) | |
ae669126 | 115 | { |
59201088 | 116 | foreach ($entries as $importedEntry) { |
59201088 TC |
117 | if ((array) $importedEntry !== $importedEntry) { |
118 | continue; | |
119 | } | |
120 | ||
121 | // set userId for the producer (it won't know which user is connected) | |
122 | $importedEntry['userId'] = $this->user->getId(); | |
123 | ||
124 | if ($this->markAsRead) { | |
125 | $importedEntry = $this->setEntryAsRead($importedEntry); | |
126 | } | |
127 | ||
128 | ++$this->queuedEntries; | |
129 | ||
130 | $this->producer->publish(json_encode($importedEntry)); | |
ae669126 | 131 | } |
59201088 | 132 | } |
ae669126 | 133 | |
59201088 TC |
134 | /** |
135 | * {@inheritdoc} | |
136 | */ | |
137 | public function parseEntry(array $importedEntry) | |
138 | { | |
12d93e68 | 139 | if ((!array_key_exists('guid', $importedEntry) || (!array_key_exists('id', $importedEntry))) && is_array(reset($importedEntry))) { |
ae669126 | 140 | $this->parseEntries($importedEntry); |
2c61db30 | 141 | |
ae669126 TC |
142 | return; |
143 | } | |
59201088 | 144 | |
12d93e68 | 145 | if (array_key_exists('children', $importedEntry)) { |
ae669126 | 146 | $this->parseEntries($importedEntry['children']); |
2c61db30 | 147 | |
ae669126 TC |
148 | return; |
149 | } | |
ae669126 | 150 | |
12d93e68 | 151 | if (!array_key_exists('uri', $importedEntry) && !array_key_exists('url', $importedEntry)) { |
59201088 TC |
152 | return; |
153 | } | |
ae669126 | 154 | |
12d93e68 | 155 | $url = array_key_exists('uri', $importedEntry) ? $importedEntry['uri'] : $importedEntry['url']; |
ae669126 | 156 | |
59201088 TC |
157 | $existingEntry = $this->em |
158 | ->getRepository('WallabagCoreBundle:Entry') | |
12d93e68 | 159 | ->findByUrlAndUserId($url, $this->user->getId()); |
ae669126 | 160 | |
59201088 TC |
161 | if (false !== $existingEntry) { |
162 | ++$this->skippedEntries; | |
ae669126 | 163 | |
59201088 TC |
164 | return; |
165 | } | |
ae669126 | 166 | |
59201088 | 167 | $data = $this->prepareEntry($importedEntry); |
ae669126 | 168 | |
59201088 TC |
169 | $entry = new Entry($this->user); |
170 | $entry->setUrl($data['url']); | |
171 | $entry->setTitle($data['title']); | |
ae669126 | 172 | |
59201088 TC |
173 | // update entry with content (in case fetching failed, the given entry will be return) |
174 | $entry = $this->fetchContent($entry, $data['url'], $data); | |
ae669126 | 175 | |
59201088 TC |
176 | if (array_key_exists('tags', $data)) { |
177 | $this->contentProxy->assignTagsToEntry( | |
178 | $entry, | |
179 | $data['tags'] | |
180 | ); | |
ae669126 | 181 | } |
ae669126 | 182 | |
59201088 | 183 | $entry->setArchived($data['is_archived']); |
ae669126 | 184 | |
59201088 TC |
185 | if (!empty($data['created_at'])) { |
186 | $dt = new \DateTime(); | |
12d93e68 | 187 | $entry->setCreatedAt($dt->setTimestamp($data['created_at'])); |
59201088 | 188 | } |
ae669126 | 189 | |
59201088 TC |
190 | $this->em->persist($entry); |
191 | ++$this->importedEntries; | |
ae669126 | 192 | |
59201088 | 193 | return $entry; |
ae669126 TC |
194 | } |
195 | ||
27acc6dd JB |
196 | /** |
197 | * {@inheritdoc} | |
198 | */ | |
12d93e68 | 199 | protected function prepareEntry(array $entry = []) |
27acc6dd | 200 | { |
12d93e68 JB |
201 | $url = array_key_exists('uri', $entry) ? $entry['uri'] : $entry['url']; |
202 | $date = array_key_exists('date_added', $entry) ? $entry['date_added'] : $entry['dateAdded']; | |
203 | $title = array_key_exists('name', $entry) ? $entry['name'] : $entry['title']; | |
204 | ||
205 | if (16 === strlen($date)) { | |
206 | // firefox ... | |
207 | $date = (int) ceil($date / 1000000); | |
208 | } else if (17 === strlen($date)) { | |
209 | // chrome ... | |
210 | $date = (int) ceil($date / 10000000); | |
211 | } else { | |
212 | $date = ''; | |
213 | } | |
214 | ||
27acc6dd | 215 | $data = [ |
12d93e68 | 216 | 'title' => $title, |
27acc6dd | 217 | 'html' => '', |
12d93e68 | 218 | 'url' => $url, |
27acc6dd JB |
219 | 'is_archived' => $this->markAsRead, |
220 | 'tags' => '', | |
12d93e68 | 221 | 'created_at' => $date, |
27acc6dd JB |
222 | ]; |
223 | ||
224 | if (array_key_exists('tags', $entry) && $entry['tags'] != '') { | |
225 | $data['tags'] = $entry['tags']; | |
226 | } | |
227 | ||
228 | return $data; | |
229 | } | |
230 | ||
ae669126 TC |
231 | /** |
232 | * {@inheritdoc} | |
233 | */ | |
59201088 | 234 | protected function setEntryAsRead(array $importedEntry) |
ae669126 | 235 | { |
59201088 TC |
236 | $importedEntry['is_archived'] = 1; |
237 | ||
238 | return $importedEntry; | |
ae669126 TC |
239 | } |
240 | } |