aboutsummaryrefslogtreecommitdiffhomepage
diff options
context:
space:
mode:
authorThomas Citharel <tcit@tcit.fr>2017-02-24 11:27:03 +0100
committerThomas Citharel <tcit@tcit.fr>2017-05-01 13:04:31 +0200
commite2f3800ccb884682547769d9e4b5d6b7cafe4e07 (patch)
treee0aec759b37b66ef795e28c652a94e4fa116e85f
parent3cbb0cc3ef89873a06bda6583747a2660b989fb8 (diff)
downloadwallabag-e2f3800ccb884682547769d9e4b5d6b7cafe4e07.tar.gz
wallabag-e2f3800ccb884682547769d9e4b5d6b7cafe4e07.tar.zst
wallabag-e2f3800ccb884682547769d9e4b5d6b7cafe4e07.zip
Add Clean Duplicates Command
-rw-r--r--src/Wallabag/CoreBundle/Command/CleanDuplicatesCommand.php119
-rw-r--r--src/Wallabag/CoreBundle/Repository/EntryRepository.php13
2 files changed, 132 insertions, 0 deletions
diff --git a/src/Wallabag/CoreBundle/Command/CleanDuplicatesCommand.php b/src/Wallabag/CoreBundle/Command/CleanDuplicatesCommand.php
new file mode 100644
index 00000000..65f35d8e
--- /dev/null
+++ b/src/Wallabag/CoreBundle/Command/CleanDuplicatesCommand.php
@@ -0,0 +1,119 @@
1<?php
2
3namespace Wallabag\CoreBundle\Command;
4
5use Doctrine\ORM\NoResultException;
6use Symfony\Bundle\FrameworkBundle\Command\ContainerAwareCommand;
7use Symfony\Component\Console\Input\InputArgument;
8use Symfony\Component\Console\Input\InputInterface;
9use Symfony\Component\Console\Output\OutputInterface;
10use Wallabag\CoreBundle\Entity\Entry;
11use Wallabag\UserBundle\Entity\User;
12
13class CleanDuplicatesCommand extends ContainerAwareCommand
14{
15 /** @var OutputInterface */
16 protected $output;
17
18 protected $duplicates = 0;
19
20 protected function configure()
21 {
22 $this
23 ->setName('wallabag:clean-duplicates')
24 ->setDescription('Cleans the database for duplicates')
25 ->setHelp('This command helps you to clean your articles list in case of duplicates')
26 ->addArgument(
27 'username',
28 InputArgument::OPTIONAL,
29 'User to clean'
30 );
31 }
32
33 protected function execute(InputInterface $input, OutputInterface $output)
34 {
35 $this->output = $output;
36
37 $username = $input->getArgument('username');
38
39 if ($username) {
40 try {
41 $user = $this->getUser($username);
42 $this->cleanDuplicates($user);
43 } catch (NoResultException $e) {
44 $output->writeln(sprintf('<error>User "%s" not found.</error>', $username));
45
46 return 1;
47 }
48 } else {
49 $users = $this->getDoctrine()->getRepository('WallabagUserBundle:User')->findAll();
50
51 $output->writeln(sprintf('Cleaning through %d user accounts', count($users)));
52
53 foreach ($users as $user) {
54 $output->writeln(sprintf('Processing user %s', $user->getUsername()));
55 $this->cleanDuplicates($user);
56 }
57 $output->writeln(sprintf('Finished cleaning. %d duplicates found in total', $this->duplicates));
58 }
59
60 return 0;
61 }
62
63 /**
64 * @param User $user
65 */
66 private function cleanDuplicates(User $user)
67 {
68 $em = $this->getContainer()->get('doctrine.orm.entity_manager');
69 $repo = $this->getDoctrine()->getRepository('WallabagCoreBundle:Entry');
70
71 $entries = $repo->getAllEntriesIdAndUrl($user->getId());
72
73 $duplicatesCount = 0;
74 $urls = [];
75 foreach ($entries as $entry) {
76 $url = $this->similarUrl($entry['url']);
77
78 /* @var $entry Entry */
79 if (in_array($url, $urls)) {
80 ++$duplicatesCount;
81
82 $em->remove($repo->find($entry['id']));
83 $em->flush(); // Flushing at the end of the loop would require the instance not being online
84 } else {
85 $urls[] = $entry['url'];
86 }
87 }
88
89 $this->duplicates += $duplicatesCount;
90
91 $this->output->writeln(sprintf('Cleaned %d duplicates for user %s', $duplicatesCount, $user->getUserName()));
92 }
93
94 private function similarUrl($url)
95 {
96 if (in_array(substr($url, -1), ['/', '#'])) { // get rid of "/" and "#" and the end of urls
97 return substr($url, 0, strlen($url));
98 }
99
100 return $url;
101 }
102
103 /**
104 * Fetches a user from its username.
105 *
106 * @param string $username
107 *
108 * @return \Wallabag\UserBundle\Entity\User
109 */
110 private function getUser($username)
111 {
112 return $this->getDoctrine()->getRepository('WallabagUserBundle:User')->findOneByUserName($username);
113 }
114
115 private function getDoctrine()
116 {
117 return $this->getContainer()->get('doctrine');
118 }
119}
diff --git a/src/Wallabag/CoreBundle/Repository/EntryRepository.php b/src/Wallabag/CoreBundle/Repository/EntryRepository.php
index 1f22e901..5e7b0d3a 100644
--- a/src/Wallabag/CoreBundle/Repository/EntryRepository.php
+++ b/src/Wallabag/CoreBundle/Repository/EntryRepository.php
@@ -379,4 +379,17 @@ class EntryRepository extends EntityRepository
379 ->setParameter('userId', $userId) 379 ->setParameter('userId', $userId)
380 ->execute(); 380 ->execute();
381 } 381 }
382
383 /**
384 * Get id and url from all entries
385 * Used for the clean-duplicates command.
386 */
387 public function getAllEntriesIdAndUrl($userId)
388 {
389 $qb = $this->createQueryBuilder('e')
390 ->select('e.id, e.url')
391 ->where('e.user = :userid')->setParameter(':userid', $userId);
392
393 return $qb->getQuery()->getArrayResult();
394 }
382} 395}