Add Wallabag\CoreBundle\Helper\UrlHasher

Signed-off-by: Olivier Mehani <shtrom@ssji.net>

src/Wallabag/ApiBundle/Controller/EntryRestController.php

@@ -14,6 +14,7 @@ use Wallabag\CoreBundle\Entity\Entry;
 use Wallabag\CoreBundle\Entity\Tag;
 use Wallabag\CoreBundle\Event\EntryDeletedEvent;
 use Wallabag\CoreBundle\Event\EntrySavedEvent;
+use Wallabag\CoreBundle\Helper\UrlHasher;
 
 class EntryRestController extends WallabagRestController
 {
@@ -56,8 +57,8 @@ class EntryRestController extends WallabagRestController
         }
 
         $urlHashMap = [];
-        foreach($urls as $urlToHash) {
+        foreach ($urls as $urlToHash) {
-            $urlHash = hash('sha1', $urlToHash); // XXX: the hash logic would better be in a separate util to avoid duplication with GenerateUrlHashesCommand::generateHashedUrls
+            $urlHash = UrlHasher::hashUrl($urlToHash);
             $hashedUrls[] = $urlHash;
             $urlHashMap[$urlHash] = $urlToHash;
         }
@@ -77,25 +78,11 @@ class EntryRestController extends WallabagRestController
 
         if (!empty($url) || !empty($hashedUrl)) {
             $hu = array_keys($results)[0];
+
             return $this->sendResponse(['exists' => $results[$hu]]);
         }
-        return $this->sendResponse($results);
-    }
 
-    /**
+        return $this->sendResponse($results);
-     * Replace the hashedUrl keys in $results with the unhashed URL from the
-     * request, as recorded in $urlHashMap.
-     */
-    private function replaceUrlHashes(array $results, array $urlHashMap) {
-        $newResults = [];
-        foreach($results as $hash => $res) {
-            if (isset($urlHashMap[$hash])) {
-                $newResults[$urlHashMap[$hash]] = $res;
-            } else {
-                $newResults[$hash] = $res;
-            }
-        }
-        return $newResults;
     }
 
     /**
@@ -815,6 +802,24 @@ class EntryRestController extends WallabagRestController
         return $this->sendResponse($results);
     }
 
+    /**
+     * Replace the hashedUrl keys in $results with the unhashed URL from the
+     * request, as recorded in $urlHashMap.
+     */
+    private function replaceUrlHashes(array $results, array $urlHashMap)
+    {
+        $newResults = [];
+        foreach ($results as $hash => $res) {
+            if (isset($urlHashMap[$hash])) {
+                $newResults[$urlHashMap[$hash]] = $res;
+            } else {
+                $newResults[$hash] = $res;
+            }
+        }
+
+        return $newResults;
+    }
+
     /**
      * Retrieve value from the request.
      * Used for POST & PATCH on a an entry.

src/Wallabag/CoreBundle/Command/GenerateUrlHashesCommand.php

@@ -7,6 +7,7 @@ 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\CoreBundle\Helper\UrlHasher;
 use Wallabag\UserBundle\Entity\User;
 
 class GenerateUrlHashesCommand extends ContainerAwareCommand
@@ -65,7 +66,9 @@ class GenerateUrlHashesCommand extends ContainerAwareCommand
 
         $i = 1;
         foreach ($entries as $entry) {
-            $entry->setHashedUrl(hash('sha1', $entry->getUrl()));
+            $entry->setHashedUrl(
+                UrlHasher::hashUrl($entry->getUrl())
+            );
             $em->persist($entry);
 
             if (0 === ($i % 20)) {

src/Wallabag/CoreBundle/Entity/Entry.php

@@ -13,6 +13,7 @@ use JMS\Serializer\Annotation\XmlRoot;
 use Symfony\Component\Validator\Constraints as Assert;
 use Wallabag\AnnotationBundle\Entity\Annotation;
 use Wallabag\CoreBundle\Helper\EntityTimestampsTrait;
+use Wallabag\CoreBundle\Helper\UrlHasher;
 use Wallabag\UserBundle\Entity\User;
 
 /**
@@ -324,7 +325,7 @@ class Entry
     public function setUrl($url)
     {
         $this->url = $url;
-        $this->hashedUrl = hash('sha1', $url);
+        $this->hashedUrl = UrlHasher::hashUrl($url);
 
         return $this;
     }

src/Wallabag/CoreBundle/Helper/UrlHasher.php

@@ -0,0 +1,22 @@
+<?php
+
+namespace Wallabag\CoreBundle\Helper;
+
+/**
+ * Hash URLs for privacy and performance.
+ */
+class UrlHasher
+{
+    /** @var string */
+    const ALGORITHM = 'sha1';
+
+    /**
+     * @param string $url
+     *
+     * @return string hashed $url
+     */
+    public static function hashUrl(string $url)
+    {
+        return hash(static::ALGORITHM, $url);
+    }
+}

src/Wallabag/CoreBundle/Repository/EntryRepository.php

@@ -9,6 +9,7 @@ use Pagerfanta\Adapter\DoctrineORMAdapter;
 use Pagerfanta\Pagerfanta;
 use Wallabag\CoreBundle\Entity\Entry;
 use Wallabag\CoreBundle\Entity\Tag;
+use Wallabag\CoreBundle\Helper\UrlHasher;
 
 class EntryRepository extends EntityRepository
 {
@@ -349,7 +350,7 @@ class EntryRepository extends EntityRepository
     public function findByUrlAndUserId($url, $userId)
     {
         return $this->findByHashedUrlAndUserId(
-            hash('sha1', $url), // XXX: the hash logic would better be in a separate util to avoid duplication with GenerateUrlHashesCommand::generateHashedUrls
+            UrlHasher::hashUrl($url),
             $userId);
     }
 
@@ -506,6 +507,32 @@ class EntryRepository extends EntityRepository
         return $this->find($randomId);
     }
 
+    /**
+     * Inject a UrlHasher.
+     *
+     * @param UrlHasher $hasher
+     */
+    public function setUrlHasher(UrlHasher $hasher)
+    {
+        $this->urlHasher = $hasher;
+    }
+
+    /**
+     * Get the UrlHasher, or create a default one if not injected.
+     *
+     * XXX: the default uses the default hash algorithm
+     *
+     * @return UrlHasher
+     */
+    protected function getUrlHasher()
+    {
+        if (!isset($this->urlHasher)) {
+            $this->setUrlHasher(new UrlHasher());
+        }
+
+        return $this->urlHasher;
+    }
+
     /**
      * Return a query builder to be used by other getBuilderFor* method.
      *