Fix accept header in crawler

[phinde.git] / src / phinde / Fetcher.php

diff --git a/src/phinde/Fetcher.php b/src/phinde/Fetcher.php
index b5644af1310d16f2dbe107b36e75d2b059d57868..666d7eda2449a6240a14d657bca6844922ecf3da 100644 (file)
--- a/src/phinde/Fetcher.php
+++ b/src/phinde/Fetcher.php
@@ -15,18 +15,21 @@ class Fetcher
      */
     public function fetch($url, $actions, $force = false)
     {
+        $url = Helper::rewriteUrl($url);
+
         $esDoc = $this->es->get($url);
         if (isset($esDoc->status->location)
             && $esDoc->status->location != ''
         ) {
             //TODO: what if location redirects change?
             $url = $esDoc->status->location;
+            $url = Helper::rewriteUrl($url);
             $esDoc = $this->es->get($url);
         }
 
         $types = array();
         foreach ($actions as $action) {
-            $types = array_merge($action::$supportedTypes);
+            $types = array_merge($types, array_keys($action::$supportedTypes));
         }
         $types = array_unique($types);
 
@@ -43,7 +46,7 @@ class Fetcher
         $res = $req->send();
         if ($res->getStatus() === 304) {
             //not modified since last time, so don't crawl again
-            echo "Not modified since last fetch\n";
+            Log::info("Not modified since last fetch");
             return false;
         } else if ($res->getStatus() !== 200) {
             throw new \Exception(
@@ -52,7 +55,8 @@ class Fetcher
             );
         }
 
-        $effUrl = $res->getEffectiveUrl();
+        $effUrl = Helper::removeAnchor($res->getEffectiveUrl());
+        $effUrl = Helper::rewriteUrl($effUrl);
         if ($effUrl != $url) {
             $this->storeRedirect($url, $effUrl);
             $url = $effUrl;
@@ -69,17 +73,17 @@ class Fetcher
 
     protected function storeRedirect($url, $target)
     {
-        $esDoc = new \stdClass();
+        $esDoc = Helper::baseDoc($url);
         $esDoc->status = (object) array(
-            'location' => $target
+            'location' => $target,
+            'findable' => false,
         );
-        $esDoc->url = $url;
         $this->storeDoc($url, $esDoc);
     }
 
     public function storeDoc($url, $esDoc)
     {
-        echo "Store $url\n";
+        Log::info("Store $url");
         $esDoc->status->processed = gmdate('c');
         $r = new Elasticsearch_Request(
             $GLOBALS['phinde']['elasticsearch'] . 'document/'