X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=src%2FWallabag%2FCoreBundle%2FHelper%2FContentProxy.php;h=c7ccfe4a0e2f871c9daa1528b6394663e7ddfc55;hb=refs%2Fheads%2Furl-3529;hp=fd97e0ff8a92473b42ac97abe06f00a4656ad439;hpb=b236d3f627a21bc9b02e7726bbb72d632937a45e;p=github%2Fwallabag%2Fwallabag.git

diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
index fd97e0ff..c7ccfe4a 100644
--- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -24,8 +24,9 @@ class ContentProxy
     protected $mimeGuesser;
     protected $fetchingErrorMessage;
     protected $eventDispatcher;
+    protected $storeArticleHeaders;
 
-    public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage)
+    public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage, $storeArticleHeaders = false)
     {
         $this->graby = $graby;
         $this->tagger = $tagger;
@@ -33,6 +34,7 @@ class ContentProxy
         $this->logger = $logger;
         $this->mimeGuesser = new MimeTypeExtensionGuesser();
         $this->fetchingErrorMessage = $fetchingErrorMessage;
+        $this->storeArticleHeaders = $storeArticleHeaders;
     }
 
     /**
@@ -63,6 +65,13 @@ 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);
     }
 
@@ -122,20 +131,58 @@ class ContentProxy
      */
     public function updatePublishedAt(Entry $entry, $value)
     {
-        $date = $value instanceof \DateTime ? $value->date : $value;
+        $date = $value;
 
         // is it a timestamp?
-        if (filter_var($date, FILTER_VALIDATE_INT) !== false) {
-            $date = '@' . $value;
+        if (false !== filter_var($date, FILTER_VALIDATE_INT)) {
+            $date = '@' . $date;
         }
 
         try {
-            $entry->setPublishedAt(new \DateTime($date));
+            // is it already a DateTime?
+            // (it's inside the try/catch in case of fail to be parse time string)
+            if (!$date instanceof \DateTime) {
+                $date = new \DateTime($date);
+            }
+
+            $entry->setPublishedAt($date);
         } catch (\Exception $e) {
             $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $entry->getUrl(), 'date' => $value]);
         }
     }
 
+    /**
+     * Helper to extract and save host from entry url.
+     *
+     * @param Entry $entry
+     */
+    public function setEntryDomainName(Entry $entry)
+    {
+        $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
+        if (false !== $domainName) {
+            $entry->setDomainName($domainName);
+        }
+    }
+
+    /**
+     * Helper to set a default title using:
+     * - url basename, if applicable
+     * - hostname.
+     *
+     * @param Entry $entry
+     */
+    public function setDefaultEntryTitle(Entry $entry)
+    {
+        $url = parse_url($entry->getUrl());
+        $path = pathinfo($url['path'], PATHINFO_BASENAME);
+
+        if (empty($path)) {
+            $path = $url['host'];
+        }
+
+        $entry->setTitle($path);
+    }
+
     /**
      * Stock entry with fetched or imported content.
      * Will fall back to OpenGraph data if available.
@@ -145,13 +192,18 @@ class ContentProxy
      */
     private function stockEntry(Entry $entry, array $content)
     {
-        $entry->setUrl($content['url']);
-
-        $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
-        if (false !== $domainName) {
-            $entry->setDomainName($domainName);
+        // When a redirection occurs while fetching an entry
+        // we move the original url in origin_url property if empty
+        // and set the entry url with the final value
+        if (!empty($content['url']) && $entry->getUrl() !== $content['url']) {
+            if (empty($entry->getOriginUrl())) {
+                $entry->setOriginUrl($entry->getUrl());
+            }
+            $entry->setUrl($content['url']);
         }
 
+        $this->setEntryDomainName($entry);
+
         if (!empty($content['title'])) {
             $entry->setTitle($content['title']);
         } elseif (!empty($content['open_graph']['og_title'])) {
@@ -179,7 +231,7 @@ class ContentProxy
             $entry->setPublishedBy($content['authors']);
         }
 
-        if (!empty($content['all_headers'])) {
+        if (!empty($content['all_headers']) && $this->storeArticleHeaders) {
             $entry->setHeaders($content['all_headers']);
         }