updateOriginUrl: add behavior when diff is fragment and query

[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 d4ea608f3c724d6ae9f896199fb23ede3fc96435..1a2a330ff085f5e922da295fbd5c908611936545 100644 (file)
--- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -66,6 +66,13 @@ class ContentProxy
         // so we'll be able to refetch it in the future
         $content['url'] = !empty($content['url']) ? $content['url'] : $url;
 
+        // In one case (at least in tests), url is empty here
+        // so we set it using $url provided in the updateEntry call.
+        // Not sure what are the other possible cases where this property is empty
+        if (empty($entry->getUrl()) && !empty($url)) {
+            $entry->setUrl($url);
+        }
+
         $this->stockEntry($entry, $content);
     }
 
@@ -239,7 +246,7 @@ class ContentProxy
      */
     private function stockEntry(Entry $entry, array $content)
     {
-        $entry->setUrl($content['url']);
+        $this->updateOriginUrl($entry, $content['url']);
 
         $this->setEntryDomainName($entry);
 
@@ -305,6 +312,51 @@ 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, $url)
+    {
+        if (!empty($url) && $entry->getUrl() !== $url) {
+            $parsed_entry_url = parse_url($entry->getUrl());
+            $parsed_content_url = parse_url($url);
+
+            $diff_ec = array_diff_assoc($parsed_entry_url, $parsed_content_url);
+            $diff_ce = array_diff_assoc($parsed_content_url, $parsed_entry_url);
+
+            $diff = array_merge($diff_ec, $diff_ce);
+            $diff_keys = array_keys($diff);
+            sort($diff_keys);
+
+            switch ($diff_keys) {
+                case ['path']:
+                    if (($parsed_entry_url['path'] . '/' === $parsed_content_url['path']) // diff is trailing slash, we only replace the url of the entry
+                        || ($url === urldecode($entry->getUrl()))) { // we update entry url if new url is a decoded version of it, see EntryRepository#findByUrlAndUserId
+                        $entry->setUrl($url);
+                    }
+                    break;
+                case ['scheme']:
+                    $entry->setUrl($url);
+                    break;
+                case ['fragment']:
+                case ['query']:
+                case ['fragment', 'query']:
+                    // noop
+                    break;
+                default:
+                    if (empty($entry->getOriginUrl())) {
+                        $entry->setOriginUrl($entry->getUrl());
+                    }
+                    $entry->setUrl($url);
+                    break;
+            }
+        }
+    }
+
     /**
      * Validate that the given content has at least a title, an html and a url.
      *