ContentProxy: replace ignoreUrl with new RuleBasedIgnoreOriginProcessor

[github/wallabag/wallabag.git] / src / Wallabag / CoreBundle / Helper / ContentProxy.php

diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
index c6fa0d983ee3321436520e6a198b7741068465be..7e93249ddd8dd8d708ea3a868065b0556a487c96 100644 (file)
--- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -19,6 +19,7 @@ class ContentProxy
 {
     protected $graby;
     protected $tagger;
+    protected $ignoreOriginProcessor;
     protected $validator;
     protected $logger;
     protected $mimeGuesser;
@@ -26,10 +27,11 @@ class ContentProxy
     protected $eventDispatcher;
     protected $storeArticleHeaders;
 
-    public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage, $storeArticleHeaders = false)
+    public function __construct(Graby $graby, RuleBasedTagger $tagger, RuleBasedIgnoreOriginProcessor $ignoreOriginProcessor, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage, $storeArticleHeaders = false)
     {
         $this->graby = $graby;
         $this->tagger = $tagger;
+        $this->ignoreOriginProcessor = $ignoreOriginProcessor;
         $this->validator = $validator;
         $this->logger = $logger;
         $this->mimeGuesser = new MimeTypeExtensionGuesser();
@@ -78,13 +80,14 @@ class ContentProxy
             $entry->setUrl($url);
         }
 
+        $entry->setGivenUrl($url);
+
         $this->stockEntry($entry, $content);
     }
 
     /**
      * Use a Symfony validator to ensure the language is well formatted.
      *
-     * @param Entry  $entry
      * @param string $value Language to validate and save
      */
     public function updateLanguage(Entry $entry, $value)
@@ -110,7 +113,6 @@ class ContentProxy
     /**
      * Use a Symfony validator to ensure the preview picture is a real url.
      *
-     * @param Entry  $entry
      * @param string $value URL to validate and save
      */
     public function updatePreviewPicture(Entry $entry, $value)
@@ -132,7 +134,6 @@ class ContentProxy
     /**
      * Update date.
      *
-     * @param Entry  $entry
      * @param string $value Date to validate and save
      */
     public function updatePublishedAt(Entry $entry, $value)
@@ -159,8 +160,6 @@ class ContentProxy
 
     /**
      * Helper to extract and save host from entry url.
-     *
-     * @param Entry $entry
      */
     public function setEntryDomainName(Entry $entry)
     {
@@ -174,8 +173,6 @@ class ContentProxy
      * Helper to set a default title using:
      * - url basename, if applicable
      * - hostname.
-     *
-     * @param Entry $entry
      */
     public function setDefaultEntryTitle(Entry $entry)
     {
@@ -331,7 +328,6 @@ class ContentProxy
      * Update the origin_url field when a redirection occurs
      * This field is set if it is empty and new url does not match ignore list.
      *
-     * @param Entry  $entry
      * @param string $url
      */
     private function updateOriginUrl(Entry $entry, $url)
@@ -362,7 +358,7 @@ class ContentProxy
         $diff_keys = array_keys($diff);
         sort($diff_keys);
 
-        if ($this->ignoreUrl($entry->getUrl())) {
+        if ($this->ignoreOriginProcessor->process($entry)) {
             $entry->setUrl($url);
 
             return false;
@@ -401,46 +397,9 @@ class ContentProxy
         }
     }
 
-    /**
-     * Check entry url against an ignore list to replace with content url.
-     *
-     * XXX: move the ignore list in the database to let users handle it
-     *
-     * @param string $url url to test
-     *
-     * @return bool true if url matches ignore list otherwise false
-     */
-    private function ignoreUrl($url)
-    {
-        $ignored_hosts = ['feedproxy.google.com', 'feeds.reuters.com'];
-        $ignored_patterns = ['https?://www\.lemonde\.fr/tiny.*'];
-
-        $parsed_url = parse_url($url);
-
-        $filtered = array_filter($ignored_hosts, function ($var) use ($parsed_url) {
-            return $var === $parsed_url['host'];
-        });
-
-        if ([] !== $filtered) {
-            return true;
-        }
-
-        $filtered = array_filter($ignored_patterns, function ($var) use ($url) {
-            return preg_match("`$var`i", $url);
-        });
-
-        if ([] !== $filtered) {
-            return true;
-        }
-
-        return false;
-    }
-
     /**
      * Validate that the given content has at least a title, an html and a url.
      *
-     * @param array $content
-     *
      * @return bool true if valid otherwise false
      */
     private function validateContent(array $content)