wipelastic6

5 files changed, 48 insertions, 7 deletions

diff --git a/src/phinde/Elasticsearch.php b/src/phinde/Elasticsearch.php
index 5ca2180..8a6d8c7 100644
--- a/src/phinde/Elasticsearch.php
+++ b/src/phinde/Elasticsearch.php
@@ -167,6 +167,7 @@ class Elasticsearch
             ),
             'highlight' => array(
                 'pre_tags' => array('<em class="hl">'),
+                'post_tags' => array('</em>'),
                 'order' => 'score',
                 'encoder' => 'html',
                 'fields' => array(
@@ -231,8 +232,12 @@ class Elasticsearch
 
         //unset($doc['_source']);
 
-        //ini_set('xdebug.var_display_max_depth', 10);
-        //echo json_encode($doc);die();
+        if (false) {
+            ini_set('xdebug.var_display_max_depth', 10);
+            header('Content-type: application/json');
+            echo json_encode($doc, JSON_PRETTY_PRINT);die();
+        }
+
         $r->setBody(json_encode($doc));
         $res = $r->send();
         return json_decode($res->getBody());
diff --git a/src/phinde/Elasticsearch/Request.php b/src/phinde/Elasticsearch/Request.php
index 7bb6add..1f9cd99 100644
--- a/src/phinde/Elasticsearch/Request.php
+++ b/src/phinde/Elasticsearch/Request.php
@@ -30,6 +30,24 @@ class Elasticsearch_Request extends \HTTP_Request2
             . $error
         );
     }
-}
 
+    /**
+     * Sets the request body - inject content type
+     *
+     * @param mixed $body       Either a string with the body or filename
+     *                          containing body or pointer to an open file or
+     *                          object with multipart body data
+     * @param bool  $isFilename Whether first parameter is a filename
+     *
+     * @return HTTP_Request2
+     * @throws HTTP_Request2_LogicException
+     *
+     * @link https://www.elastic.co/blog/strict-content-type-checking-for-elasticsearch-rest-requests
+     */
+    public function setBody($body, $isFilename = false)
+    {
+        $this->setHeader('content-type', 'application/json');
+        return parent::setBody($body, $isFilename);
+    }
+}
 ?>
diff --git a/src/phinde/Fetcher.php b/src/phinde/Fetcher.php
index dccb118..63f5a43 100644
--- a/src/phinde/Fetcher.php
+++ b/src/phinde/Fetcher.php
@@ -60,6 +60,21 @@ class Fetcher
         }
         //FIXME: etag, hash on content
 
+        if ($esDoc === null) {
+            //not known yet
+            $esDoc = Helper::baseDoc($url);
+        }
+
+        $lm = $res->getHeader('last-modified');
+        if ($lm !== null) {
+            $esDoc->status->modate = gmdate('c', strtotime($lm));
+        } else {
+            $esDoc->status->modate = gmdate('c');
+        }
+        if ($esDoc->status->crdate == '') {
+            $esDoc->status->crdate = $esDoc->status->modate;
+        }
+
         $retrieved = new Retrieved();
         $retrieved->httpRes = $res;
         $retrieved->esDoc   = $esDoc;
diff --git a/src/phinde/Helper.php b/src/phinde/Helper.php
index aeb8ba5..55c8bbd 100644
--- a/src/phinde/Helper.php
+++ b/src/phinde/Helper.php
@@ -81,9 +81,15 @@ class Helper
     public static function baseDoc($url)
     {
         $esDoc = new \stdClass();
-        $esDoc->status = new \stdClass();
+        $esDoc->status = (object) array(
+            'findable'  => false,
+            'modate'    => '',
+            'crdate'    => '',
+            'processed' => '',
+        );
         $esDoc->url = $url;
         $esDoc->schemalessUrl = Helper::noSchema($url);
+        $esDoc->domain        = parse_url($url, PHP_URL_HOST);
         return $esDoc;
     }
 }
diff --git a/src/phinde/Indexer.php b/src/phinde/Indexer.php
index bdd5236..4efef42 100644
--- a/src/phinde/Indexer.php
+++ b/src/phinde/Indexer.php
@@ -77,12 +77,9 @@ class Indexer
             $xpContext = $doc->getElementById('content');
         }
 
-        $esDoc->url = $url;
-        $esDoc->schemalessUrl = Helper::noSchema($url);
         $esDoc->type = 'html';
         $esDoc->subtype = '';
         $esDoc->mimetype = $mimetype;
-        $esDoc->domain   = parse_url($url, PHP_URL_HOST);
 
         //$esDoc->source = 'FIXME';
         //$esDoc->sourcetitle = 'FIXME';