Hash the urls to check if they exist

Signed-off-by: Thomas Citharel <tcit@tcit.fr>

src/Wallabag/ApiBundle/Controller/EntryRestController.php

@@ -29,6 +29,8 @@ class EntryRestController extends WallabagRestController
      *          {"name"="return_id", "dataType"="string", "required"=false, "format"="1 or 0", "description"="Set 1 if you want to retrieve ID in case entry(ies) exists, 0 by default"},
      *          {"name"="url", "dataType"="string", "required"=true, "format"="An url", "description"="Url to check if it exists"},
      *          {"name"="urls", "dataType"="string", "required"=false, "format"="An array of urls (?urls[]=http...&urls[]=http...)", "description"="Urls (as an array) to check if it exists"}
+     *          {"name"="hashedurl", "dataType"="string", "required"=true, "format"="An url", "description"="Md5 url to check if it exists"},
+     *          {"name"="hashedurls", "dataType"="string", "required"=false, "format"="An array of urls (?urls[]=http...&urls[]=http...)", "description"="Md5 urls (as an array) to check if it exists"}
      *       }
      * )
      *
@@ -41,34 +43,46 @@ class EntryRestController extends WallabagRestController
         $returnId = (null === $request->query->get('return_id')) ? false : (bool) $request->query->get('return_id');
         $urls = $request->query->get('urls', []);
 
+        $hashedUrls = $request->query->get('hashedurls', []);
+
         // handle multiple urls first
-        if (!empty($urls)) {
+        if (!empty($hashedUrls)) {
             $results = [];
-            foreach ($urls as $url) {
+            foreach ($hashedUrls as $hashedUrl) {
                 $res = $this->getDoctrine()
                     ->getRepository('WallabagCoreBundle:Entry')
-                    ->findByUrlAndUserId($url, $this->getUser()->getId());
+                    ->findOneBy([
+                        'hashedUrl' => $hashedUrl,
+                        'user' => $this->getUser()->getId(),
+                    ]);
 
-                $results[$url] = $this->returnExistInformation($res, $returnId);
+                // $results[$url] = $this->returnExistInformation($res, $returnId);
+                $results[$hashedUrl] = $this->returnExistInformation($res, $returnId);
             }
 
             return $this->sendResponse($results);
         }
 
         // let's see if it is a simple url?
-        $url = $request->query->get('url', '');
+        $hashedUrl = $request->query->get('hashedurl', '');
 
-        if (empty($url)) {
+        // if (empty($url)) {
-            throw $this->createAccessDeniedException('URL is empty?, logged user id: ' . $this->getUser()->getId());
+        //     throw $this->createAccessDeniedException('URL is empty?, logged user id: ' . $this->getUser()->getId());
+        // }
+
+        if (empty($hashedUrl)) {
+            throw $this->createAccessDeniedException('URL is empty?, logged user id: '.$this->getUser()->getId());
         }
 
         $res = $this->getDoctrine()
             ->getRepository('WallabagCoreBundle:Entry')
-            ->findByUrlAndUserId($url, $this->getUser()->getId());
+            // ->findByUrlAndUserId($url, $this->getUser()->getId());
+            ->findOneBy([
+                'hashedUrl' => $hashedUrl,
+                'user' => $this->getUser()->getId(),
+            ]);
 
-        $exists = $this->returnExistInformation($res, $returnId);
+        return $this->sendResponse(['exists' => $this->returnExistInformation($res, $returnId)]);
-
-        return $this->sendResponse(['exists' => $exists]);
     }
 
     /**

src/Wallabag/CoreBundle/Command/GenerateUrlHashesCommand.php

@@ -0,0 +1,95 @@
+<?php
+
+namespace Wallabag\CoreBundle\Command;
+
+use Doctrine\ORM\NoResultException;
+use Symfony\Bundle\FrameworkBundle\Command\ContainerAwareCommand;
+use Symfony\Component\Console\Input\InputArgument;
+use Symfony\Component\Console\Input\InputInterface;
+use Symfony\Component\Console\Output\OutputInterface;
+use Wallabag\UserBundle\Entity\User;
+
+class GenerateUrlHashesCommand extends ContainerAwareCommand
+{
+    /** @var OutputInterface */
+    protected $output;
+
+    protected function configure()
+    {
+        $this
+            ->setName('wallabag:generate-hashed-urls')
+            ->setDescription('Generates hashed urls for each entry')
+            ->setHelp('This command helps you to generates hashes of the url of each entry, to check through API if an URL is already saved')
+            ->addArgument(
+                'username',
+                InputArgument::OPTIONAL,
+                'User to process entries'
+            );
+    }
+
+    protected function execute(InputInterface $input, OutputInterface $output)
+    {
+        $this->output = $output;
+
+        $username = $input->getArgument('username');
+
+        if ($username) {
+            try {
+                $user = $this->getUser($username);
+                $this->generateHashedUrls($user);
+            } catch (NoResultException $e) {
+                $output->writeln(sprintf('<error>User "%s" not found.</error>', $username));
+
+                return 1;
+            }
+        } else {
+            $users = $this->getDoctrine()->getRepository('WallabagUserBundle:User')->findAll();
+
+            $output->writeln(sprintf('Generating hashed urls for the %d user account entries', count($users)));
+
+            foreach ($users as $user) {
+                $output->writeln(sprintf('Processing user %s', $user->getUsername()));
+                $this->generateHashedUrls($user);
+            }
+            $output->writeln(sprintf('Finished generated hashed urls'));
+        }
+
+        return 0;
+    }
+
+    /**
+     * @param User $user
+     */
+    private function generateHashedUrls(User $user)
+    {
+        $em = $this->getContainer()->get('doctrine.orm.entity_manager');
+        $repo = $this->getDoctrine()->getRepository('WallabagCoreBundle:Entry');
+
+        $entries = $repo->findByUser($user->getId());
+
+        foreach ($entries as $entry) {
+            $entry->setHashedUrl(hash('sha512', $entry->getUrl()));
+            $em->persist($entry);
+            $em->flush();
+        }
+
+        $this->output->writeln(sprintf('Generated hashed urls for user %s', $user->getUserName()));
+    }
+
+    /**
+     * Fetches a user from its username.
+     *
+     * @param string $username
+     *
+     * @return \Wallabag\UserBundle\Entity\User
+     */
+    private function getUser($username)
+    {
+        return $this->getDoctrine()->getRepository('WallabagUserBundle:User')->findOneByUserName($username);
+    }
+
+    private function getDoctrine()
+    {
+        return $this->getContainer()->get('doctrine');
+    }
+}

src/Wallabag/CoreBundle/DataFixtures/EntryFixtures.php

@@ -30,6 +30,7 @@ class EntryFixtures extends Fixture implements DependentFixtureInterface
             'entry2' => [
                 'user' => 'admin-user',
                 'url' => 'http://0.0.0.0/entry2',
+                'hashed_url' => hash('md5', 'http://0.0.0.0/entry2'),
                 'reading_time' => 1,
                 'domain' => 'domain.io',
                 'mime' => 'text/html',

src/Wallabag/CoreBundle/Entity/Entry.php

@@ -25,7 +25,8 @@ use Wallabag\UserBundle\Entity\User;
  *     options={"collate"="utf8mb4_unicode_ci", "charset"="utf8mb4"},
  *     indexes={
  *         @ORM\Index(name="created_at", columns={"created_at"}),
- *         @ORM\Index(name="uid", columns={"uid"})
+ *         @ORM\Index(name="uid", columns={"uid"}),
+ *         @ORM\Index(name="hashedurl", columns={"hashedurl"})
  *     }
  * )
  * @ORM\HasLifecycleCallbacks()
@@ -75,6 +76,13 @@ class Entry
      */
     private $url;
 
+    /**
+     * @var string
+     *
+     * @ORM\Column(name="hashedurl", type="text", nullable=true)
+     */
+    private $hashedUrl;
+
     /**
      * @var bool
      *
@@ -911,4 +919,24 @@ class Entry
     {
         return $this->originUrl;
     }
+
+    /**
+     * @return string
+     */
+    public function getHashedUrl()
+    {
+        return $this->hashedUrl;
+    }
+
+    /**
+     * @param mixed $hashedUrl
+     *
+     * @return Entry
+     */
+    public function setHashedUrl($hashedUrl)
+    {
+        $this->hashedUrl = $hashedUrl;
+
+        return $this;
+    }
 }

src/Wallabag/CoreBundle/Helper/ContentProxy.php

@@ -248,6 +248,8 @@ class ContentProxy
     {
         $this->updateOriginUrl($entry, $content['url']);
 
+        $entry->setHashedUrl(hash('md5', $entry->getUrl()));
+
         $this->setEntryDomainName($entry);
 
         if (!empty($content['title'])) {

tests/Wallabag/ApiBundle/Controller/EntryRestControllerTest.php

@@ -987,6 +987,8 @@ class EntryRestControllerTest extends WallabagApiTestCase
     {
         $this->client->request('GET', '/api/entries/exists?url=http://0.0.0.0/entry2');
 
+        $this->client->request('GET', '/api/entries/exists?hashedurl=' . hash('md5', 'http://0.0.0.0/entry2'));
+
         $this->assertSame(200, $this->client->getResponse()->getStatusCode());
 
         $content = json_decode($this->client->getResponse()->getContent(), true);
@@ -994,10 +996,22 @@ class EntryRestControllerTest extends WallabagApiTestCase
         $this->assertTrue($content['exists']);
     }
 
+    public function testGetEntriesExistsWithHash()
+    {
+        $this->client->request('GET', '/api/entries/exists?hashedurl=' . hash('md5', 'http://0.0.0.0/entry2'));
+
+        $this->assertSame(200, $this->client->getResponse()->getStatusCode());
+
+        $content = json_decode($this->client->getResponse()->getContent(), true);
+
+        $this->assertSame(2, $content['exists']);
+    }
+
     public function testGetEntriesExistsWithManyUrls()
     {
         $url1 = 'http://0.0.0.0/entry2';
         $url2 = 'http://0.0.0.0/entry10';
+
         $this->client->request('GET', '/api/entries/exists?urls[]=' . $url1 . '&urls[]=' . $url2 . '&return_id=1');
 
         $this->assertSame(200, $this->client->getResponse()->getStatusCode());
@@ -1027,9 +1041,46 @@ class EntryRestControllerTest extends WallabagApiTestCase
         $this->assertFalse($content[$url2]);
     }
 
+    public function testGetEntriesExistsWithManyUrlsHashed()
+    {
+        $url1 = 'http://0.0.0.0/entry2';
+        $url2 = 'http://0.0.0.0/entry10';
+        $this->client->request('GET', '/api/entries/exists?hashedurls[]='.hash('md5',$url1).'&hashedurls[]='.hash('md5',$url2) . '&return_id=1');
+
+        $this->assertSame(200, $this->client->getResponse()->getStatusCode());
+
+        $content = json_decode($this->client->getResponse()->getContent(), true);
+
+        $this->assertArrayHasKey($url1, $content);
+        $this->assertArrayHasKey($url2, $content);
+        $this->assertSame(2, $content[$url1]);
+        $this->assertNull($content[$url2]);
+
+        $this->assertArrayHasKey(hash('md5', $url1), $content);
+        $this->assertArrayHasKey(hash('md5', $url2), $content);
+        $this->assertEquals(2, $content[hash('md5', $url1)]);
+        $this->assertEquals(false, $content[hash('md5', $url2)]);
+    }
+
+    public function testGetEntriesExistsWithManyUrlsHashedReturnBool()
+    {
+        $url1 = 'http://0.0.0.0/entry2';
+        $url2 = 'http://0.0.0.0/entry10';
+        $this->client->request('GET', '/api/entries/exists?hashedurls[]='.hash('md5',$url1).'&hashedurls[]='.hash('md5',$url2));
+
+        $this->assertSame(200, $this->client->getResponse()->getStatusCode());
+
+        $content = json_decode($this->client->getResponse()->getContent(), true);
+
+        $this->assertArrayHasKey($url1, $content);
+        $this->assertArrayHasKey($url2, $content);
+        $this->assertTrue($content[$url1]);
+        $this->assertFalse($content[$url2]);
+    }
+
     public function testGetEntriesExistsWhichDoesNotExists()
     {
-        $this->client->request('GET', '/api/entries/exists?url=http://google.com/entry2');
+        $this->client->request('GET', '/api/entries/exists?hashedurl='.hash('md5','http://google.com/entry2'));
 
         $this->assertSame(200, $this->client->getResponse()->getStatusCode());
 
@@ -1040,7 +1091,7 @@ class EntryRestControllerTest extends WallabagApiTestCase
 
     public function testGetEntriesExistsWithNoUrl()
     {
-        $this->client->request('GET', '/api/entries/exists?url=');
+        $this->client->request('GET', '/api/entries/exists?hashedurl=');
 
         $this->assertSame(403, $this->client->getResponse()->getStatusCode());
     }

tests/Wallabag/CoreBundle/Command/GenerateUrlHashesCommandTest.php

@@ -0,0 +1,101 @@
+<?php
+
+namespace Tests\Wallabag\CoreBundle\Command;
+
+use Symfony\Bundle\FrameworkBundle\Console\Application;
+use Symfony\Component\Console\Tester\CommandTester;
+use Tests\Wallabag\CoreBundle\WallabagCoreTestCase;
+use Wallabag\CoreBundle\Command\GenerateUrlHashesCommand;
+use Wallabag\CoreBundle\Entity\Entry;
+
+class GenerateUrlHashesCommandTest extends WallabagCoreTestCase
+{
+    public function testRunGenerateUrlHashesCommand()
+    {
+        $application = new Application($this->getClient()->getKernel());
+        $application->add(new GenerateUrlHashesCommand());
+
+        $command = $application->find('wallabag:generate-hashed-urls');
+
+        $tester = new CommandTester($command);
+        $tester->execute([
+            'command' => $command->getName(),
+        ]);
+
+        $this->assertContains('Generating hashed urls for the 3 user account entries', $tester->getDisplay());
+        $this->assertContains('Finished generated hashed urls', $tester->getDisplay());
+    }
+
+    public function testRunGenerateUrlHashesCommandWithBadUsername()
+    {
+        $application = new Application($this->getClient()->getKernel());
+        $application->add(new GenerateUrlHashesCommand());
+
+        $command = $application->find('wallabag:generate-hashed-urls');
+
+        $tester = new CommandTester($command);
+        $tester->execute([
+            'command' => $command->getName(),
+            'username' => 'unknown',
+        ]);
+
+        $this->assertContains('User "unknown" not found', $tester->getDisplay());
+    }
+
+    public function testRunGenerateUrlHashesCommandForUser()
+    {
+        $application = new Application($this->getClient()->getKernel());
+        $application->add(new GenerateUrlHashesCommand());
+
+        $command = $application->find('wallabag:generate-hashed-urls');
+
+        $tester = new CommandTester($command);
+        $tester->execute([
+            'command' => $command->getName(),
+            'username' => 'admin',
+        ]);
+
+        $this->assertContains('Generated hashed urls for user admin', $tester->getDisplay());
+    }
+
+    public function testGenerateUrls()
+    {
+        $url = 'http://www.lemonde.fr/sport/visuel/2017/05/05/rondelle-prison-blanchissage-comprendre-le-hockey-sur-glace_5122587_3242.html';
+        $client = $this->getClient();
+        $em = $client->getContainer()->get('doctrine.orm.entity_manager');
+
+        $this->logInAs('admin');
+
+        $user = $em->getRepository('WallabagUserBundle:User')->findOneById($this->getLoggedInUserId());
+
+        $entry1 = new Entry($user);
+        $entry1->setUrl($url);
+
+        $em->persist($entry1);
+
+        $em->flush();
+
+        $this->assertNull($entry1->getHashedUrl());
+
+        $application = new Application($this->getClient()->getKernel());
+        $application->add(new GenerateUrlHashesCommand());
+
+        $command = $application->find('wallabag:generate-hashed-urls');
+
+        $tester = new CommandTester($command);
+        $tester->execute([
+            'command' => $command->getName(),
+            'username' => 'admin',
+        ]);
+
+        $this->assertContains('Generated hashed urls for user admin', $tester->getDisplay());
+
+        $entry = $em->getRepository('WallabagCoreBundle:Entry')->findOneByUrl($url);
+
+        $this->assertEquals($entry->getHashedUrl(), hash('sha512', $url));
+
+        $query = $em->createQuery('DELETE FROM Wallabag\CoreBundle\Entity\Entry e WHERE e.url = :url');
+        $query->setParameter('url', $url);
+        $query->execute();
+    }
+}