From bbf8fd9b04169a94ddba002ca8877910df4e0929 Mon Sep 17 00:00:00 2001
From: Christian Weiske <cweiske@cweiske.de>
Date: Fri, 2 Sep 2016 11:01:28 +0200
Subject: massively improve crawl speed by ditching "exists" queries

---
 src/phinde/Elasticsearch.php | 26 ++++----------------------
 1 file changed, 4 insertions(+), 22 deletions(-)

(limited to 'src/phinde')

diff --git a/src/phinde/Elasticsearch.php b/src/phinde/Elasticsearch.php
index 43ef4f9..1732bbb 100644
--- a/src/phinde/Elasticsearch.php
+++ b/src/phinde/Elasticsearch.php
@@ -10,33 +10,15 @@ class Elasticsearch
         $this->baseUrl = $baseUrl;
     }
 
-    /**
-     * @link https://www.elastic.co/guide/en/elasticsearch/guide/current/_finding_exact_values.html
-     */
     public function isKnown($url)
     {
         $r = new Elasticsearch_Request(
-            $this->baseUrl . 'document/_search/exists',
-            \HTTP_Request2::METHOD_GET
+            $this->baseUrl . 'document/' . rawurlencode($url),
+            \HTTP_Request2::METHOD_HEAD
         );
         $r->allow404 = true;
-        $r->setBody(
-            json_encode(
-                array(
-                    'query' => array(
-                        'filtered' => array(
-                            'filter' => array(
-                                'term' => array(
-                                    'url' => $url
-                                )
-                            )
-                        )
-                    )
-                )
-            )
-        );
-        $status = $r->send()->getStatus();
-        return $status !== 404;
+        $res = $r->send();
+        return $res->getStatus() == 200;
     }
 
     public function get($url)
-- 
cgit v1.2.3