From f3bfb875e94021a93e24a41fbc0f8d86d4dee378 Mon Sep 17 00:00:00 2001
From: Jeremy Benoist <jeremy.benoist@gmail.com>
Date: Wed, 29 May 2019 14:18:04 +0200
Subject: Use hash given url to avoid duplicate

Using hashed url we can ensure an index on them to ensure it's fast.
---
 src/Wallabag/CoreBundle/Entity/Entry.php           | 99 ++++++++++++----------
 src/Wallabag/CoreBundle/Helper/ContentProxy.php    |  3 +-
 .../CoreBundle/Repository/EntryRepository.php      |  2 +-
 3 files changed, 59 insertions(+), 45 deletions(-)

(limited to 'src/Wallabag/CoreBundle')

diff --git a/src/Wallabag/CoreBundle/Entity/Entry.php b/src/Wallabag/CoreBundle/Entity/Entry.php
index 62274136..304dd1b3 100644
--- a/src/Wallabag/CoreBundle/Entity/Entry.php
+++ b/src/Wallabag/CoreBundle/Entity/Entry.php
@@ -27,9 +27,8 @@ use Wallabag\UserBundle\Entity\User;
  *     indexes={
  *         @ORM\Index(name="created_at", columns={"created_at"}),
  *         @ORM\Index(name="uid", columns={"uid"}),
- *         @ORM\Index(name="hashed_url_user_id", columns={"user_id", "hashed_url"}, options={"lengths"={null, 40}})
- *     },
- *     uniqueConstraints={@ORM\UniqueConstraint(name="IDX_entry_given_url",columns={"url", "given_url", "user_id"})}
+ *         @ORM\Index(name="hashed_urls_user_id", columns={"user_id", "hashed_url", "hashed_given_url"}, options={"lengths"={null, 40, 40}})
+ *     }
  * )
  * @ORM\HasLifecycleCallbacks()
  * @Hateoas\Relation("self", href = "expr('/api/entries/' ~ object.getId())")
@@ -69,30 +68,52 @@ class Entry
     private $title;
 
     /**
+     * Define the url fetched by wallabag (the final url after potential redirections).
+     *
      * @var string
      *
-     * @ORM\Column(name="given_url", type="text", nullable=true)
+     * @Assert\NotBlank()
+     * @ORM\Column(name="url", type="text", nullable=true)
      *
      * @Groups({"entries_for_user", "export_all"})
      */
-    private $givenUrl;
+    private $url;
 
     /**
      * @var string
      *
-     * @Assert\NotBlank()
-     * @ORM\Column(name="url", type="text", nullable=true)
+     * @ORM\Column(name="hashed_url", type="string", length=40, nullable=true)
+     */
+    private $hashedUrl;
+
+    /**
+     * From where user retrieved/found the url (an other article, a twitter, or the given_url if non are provided).
+     *
+     * @var string
+     *
+     * @ORM\Column(name="origin_url", type="text", nullable=true)
      *
      * @Groups({"entries_for_user", "export_all"})
      */
-    private $url;
+    private $originUrl;
 
     /**
+     * Define the url entered by the user (without redirections).
+     *
      * @var string
      *
-     * @ORM\Column(name="hashed_url", type="string", length=40, nullable=true)
+     * @ORM\Column(name="given_url", type="text", nullable=true)
+     *
+     * @Groups({"entries_for_user", "export_all"})
      */
-    private $hashedUrl;
+    private $givenUrl;
+
+    /**
+     * @var string
+     *
+     * @ORM\Column(name="hashed_given_url", type="string", length=40, nullable=true)
+     */
+    private $hashedGivenUrl;
 
     /**
      * @var bool
@@ -273,15 +294,6 @@ class Entry
      */
     private $tags;
 
-    /**
-     * @var string
-     *
-     * @ORM\Column(name="origin_url", type="text", nullable=true)
-     *
-     * @Groups({"entries_for_user", "export_all"})
-     */
-    private $originUrl;
-
     /*
      * @param User     $user
      */
@@ -325,30 +337,6 @@ class Entry
         return $this->title;
     }
 
-    /**
-     * Set given url.
-     *
-     * @param string $givenUrl
-     *
-     * @return Entry
-     */
-    public function setGivenUrl($givenUrl)
-    {
-        $this->givenUrl = $givenUrl;
-
-        return $this;
-    }
-
-    /**
-     * Get given Url.
-     *
-     * @return string
-     */
-    public function getGivenUrl()
-    {
-        return $this->givenUrl;
-    }
-
     /**
      * Set url.
      *
@@ -956,6 +944,31 @@ class Entry
         return $this->originUrl;
     }
 
+    /**
+     * Set origin url.
+     *
+     * @param string $givenUrl
+     *
+     * @return Entry
+     */
+    public function setGivenUrl($givenUrl)
+    {
+        $this->givenUrl = $givenUrl;
+        $this->hashedGivenUrl = UrlHasher::hashUrl($givenUrl);
+
+        return $this;
+    }
+
+    /**
+     * Get origin url.
+     *
+     * @return string
+     */
+    public function getGivenUrl()
+    {
+        return $this->givenUrl;
+    }
+
     /**
      * @return string
      */
diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
index 0d6a412d..5901df8b 100644
--- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -76,9 +76,10 @@ class ContentProxy
         // Not sure what are the other possible cases where this property is empty
         if (empty($entry->getUrl()) && !empty($url)) {
             $entry->setUrl($url);
-            $entry->setGivenUrl($url);
         }
 
+        $entry->setGivenUrl($url);
+
         $this->stockEntry($entry, $content);
     }
 
diff --git a/src/Wallabag/CoreBundle/Repository/EntryRepository.php b/src/Wallabag/CoreBundle/Repository/EntryRepository.php
index 299b0b27..8b29aad2 100644
--- a/src/Wallabag/CoreBundle/Repository/EntryRepository.php
+++ b/src/Wallabag/CoreBundle/Repository/EntryRepository.php
@@ -368,7 +368,7 @@ class EntryRepository extends EntityRepository
     {
         $res = $this->createQueryBuilder('e')
             ->where('e.hashedUrl = :hashed_url')->setParameter('hashed_url', $hashedUrl)
-            // ->orWhere('e.givenUrl = :url')->setParameter('url', $url)
+            ->orWhere('e.hashedGivenUrl = :hashed_given_url')->setParameter('hashed_given_url', $hashedUrl)
             ->andWhere('e.user = :user_id')->setParameter('user_id', $userId)
             ->getQuery()
             ->getResult();
-- 
cgit v1.2.3