src/Wallabag/ImportBundle/Import/BrowserImport.php

   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\Entity\Entry;
   9 use Wallabag\UserBundle\Entity\User;
  10 use Wallabag\CoreBundle\Helper\ContentProxy;
  11
  12 abstract class BrowserImport extends AbstractImport
  13 {
  14     protected $filepath;
  15
  16     /**
  17      * {@inheritdoc}
  18      */
  19     abstract public function getName();
  20
  21     /**
  22      * {@inheritdoc}
  23      */
  24     abstract public function getUrl();
  25
  26     /**
  27      * {@inheritdoc}
  28      */
  29     abstract public function getDescription();
  30
  31     /**
  32      * {@inheritdoc}
  33      */
  34     public function import()
  35     {
  36         if (!$this->user) {
  37             $this->logger->error('WallabagImport: user is not defined');
  38
  39             return false;
  40         }
  41
  42         if (!file_exists($this->filepath) || !is_readable($this->filepath)) {
  43             $this->logger->error('WallabagImport: unable to read file', ['filepath' => $this->filepath]);
  44
  45             return false;
  46         }
  47
  48         $data = json_decode(file_get_contents($this->filepath), true);
  49
  50         if (empty($data)) {
  51             return false;
  52         }
  53
  54         if ($this->producer) {
  55             $this->parseEntriesForProducer($data);
  56
  57             return true;
  58         }
  59
  60         $this->parseEntries($data);
  61
  62         return true;
  63     }
  64
  65     /**
  66      * Set file path to the json file.
  67      *
  68      * @param string $filepath
  69      */
  70     public function setFilepath($filepath)
  71     {
  72         $this->filepath = $filepath;
  73
  74         return $this;
  75     }
  76
  77     /**
  78      * Parse and insert all given entries.
  79      *
  80      * @param $entries
  81      */
  82     protected function parseEntries($entries)
  83     {
  84         $i = 1;
  85
  86         foreach ($entries as $importedEntry) {
  87             if ((array) $importedEntry !== $importedEntry) {
  88                 continue;
  89             }
  90
  91             $entry = $this->parseEntry($importedEntry);
  92
  93             if (null === $entry) {
  94                 continue;
  95             }
  96
  97             // flush every 20 entries
  98             if (($i % 20) === 0) {
  99                 $this->em->flush();
 100
 101                 // clear only affected entities
 102                 $this->em->clear(Entry::class);
 103                 $this->em->clear(Tag::class);
 104             }
 105             ++$i;
 106         }
 107
 108         $this->em->flush();
 109     }
 110
 111     /**
 112      * Parse entries and send them to the queue.
 113      * It should just be a simple loop on all item, no call to the database should be done
 114      * to speedup queuing.
 115      *
 116      * Faster parse entries for Producer.
 117      * We don't care to make check at this time. They'll be done by the consumer.
 118      *
 119      * @param array $entries
 120      */
 121     protected function parseEntriesForProducer(array $entries)
 122     {
 123         foreach ($entries as $importedEntry) {
 124
 125             if ((array) $importedEntry !== $importedEntry) {
 126                 continue;
 127             }
 128
 129             // set userId for the producer (it won't know which user is connected)
 130             $importedEntry['userId'] = $this->user->getId();
 131
 132             if ($this->markAsRead) {
 133                 $importedEntry = $this->setEntryAsRead($importedEntry);
 134             }
 135
 136             ++$this->queuedEntries;
 137
 138             $this->producer->publish(json_encode($importedEntry));
 139         }
 140     }
 141
 142     /**
 143      * {@inheritdoc}
 144      */
 145     public function parseEntry(array $importedEntry)
 146     {
 147
 148         if ((!key_exists('guid', $importedEntry) || (!key_exists('id', $importedEntry))) && is_array(reset($importedEntry))) {
 149             $this->parseEntries($importedEntry);
 150             return;
 151         }
 152
 153         if (key_exists('children', $importedEntry)) {
 154             $this->parseEntries($importedEntry['children']);
 155             return;
 156         }
 157
 158         if (!key_exists('uri', $importedEntry) && !key_exists('url', $importedEntry)) {
 159             return;
 160         }
 161
 162         $firefox = key_exists('uri', $importedEntry);
 163
 164         $existingEntry = $this->em
 165             ->getRepository('WallabagCoreBundle:Entry')
 166             ->findByUrlAndUserId(($firefox) ? $importedEntry['uri'] : $importedEntry['url'], $this->user->getId());
 167
 168         if (false !== $existingEntry) {
 169             ++$this->skippedEntries;
 170
 171             return;
 172         }
 173
 174         $data = $this->prepareEntry($importedEntry);
 175
 176         $entry = new Entry($this->user);
 177         $entry->setUrl($data['url']);
 178         $entry->setTitle($data['title']);
 179
 180         // update entry with content (in case fetching failed, the given entry will be return)
 181         $entry = $this->fetchContent($entry, $data['url'], $data);
 182
 183         if (array_key_exists('tags', $data)) {
 184             $this->contentProxy->assignTagsToEntry(
 185                 $entry,
 186                 $data['tags']
 187             );
 188         }
 189
 190         $entry->setArchived($data['is_archived']);
 191
 192         if (!empty($data['created_at'])) {
 193             $dt = new \DateTime();
 194             $entry->setCreatedAt($dt->setTimestamp($data['created_at']/1000));
 195         }
 196
 197         $this->em->persist($entry);
 198         ++$this->importedEntries;
 199
 200         return $entry;
 201     }
 202
 203     /**
 204      * {@inheritdoc}
 205      */
 206     protected function setEntryAsRead(array $importedEntry)
 207     {
 208         $importedEntry['is_archived'] = 1;
 209
 210         return $importedEntry;
 211     }
 212 }