src/Wallabag/ImportBundle/Import/BrowserImport.php

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