src/phinde/Elasticsearch.php

   1 <?php
   2 namespace phinde;
   3
   4 class Elasticsearch
   5 {
   6     protected $baseUrl;
   7
   8     public function __construct($baseUrl)
   9     {
  10         $this->baseUrl = $baseUrl;
  11     }
  12
  13     /**
  14      * @link https://www.elastic.co/guide/en/elasticsearch/guide/current/_finding_exact_values.html
  15      */
  16     public function isKnown($url)
  17     {
  18         $r = new Elasticsearch_Request(
  19             $this->baseUrl . 'document/_search/exists',
  20             \HTTP_Request2::METHOD_GET
  21         );
  22         $r->allow404 = true;
  23         $r->setBody(
  24             json_encode(
  25                 array(
  26                     'query' => array(
  27                         'filtered' => array(
  28                             'filter' => array(
  29                                 'term' => array(
  30                                     'url' => $url
  31                                 )
  32                             )
  33                         )
  34                     )
  35                 )
  36             )
  37         );
  38         $res = json_decode($r->send()->getBody());
  39         return $res->exists;
  40     }
  41
  42     public function get($url)
  43     {
  44         $r = new Elasticsearch_Request(
  45             $this->baseUrl . 'document/' . rawurlencode($url),
  46             \HTTP_Request2::METHOD_GET
  47         );
  48         $r->allow404 = true;
  49         $res = $r->send();
  50         if ($res->getStatus() != 200) {
  51             return null;
  52         }
  53         $d = json_decode($res->getBody());
  54         return $d->_source;
  55     }
  56
  57     public function markQueued($url)
  58     {
  59         $r = new Elasticsearch_Request(
  60             $this->baseUrl . 'document/' . rawurlencode($url),
  61             \HTTP_Request2::METHOD_PUT
  62         );
  63         $doc = array(
  64             'status' => 'queued',
  65             'url' => $url
  66         );
  67         $r->setBody(json_encode($doc));
  68         $r->send();
  69     }
  70
  71     public function search($query, $filters, $site, $page, $perPage, $sort)
  72     {
  73         if ($sort == 'date') {
  74             $sortCfg = array('modate' => array('order' => 'desc'));
  75         } else {
  76             $sortCfg = array();
  77         }
  78
  79         $r = new Elasticsearch_Request(
  80             $this->baseUrl . 'document/_search',
  81             \HTTP_Request2::METHOD_GET
  82         );
  83         $doc = array(
  84             '_source' => array(
  85                 'url',
  86                 'title',
  87                 'author',
  88                 'modate',
  89             ),
  90             'query' => array(
  91                 'bool' => array(
  92                     'must' => array(
  93                         array(
  94                             'query_string' => array(
  95                                 'default_field' => '_all',
  96                                 'default_operator' => 'AND',
  97                                 'query' => $query
  98                             )
  99                         ),
 100                         array(
 101                             'term' => array(
 102                                 'status' => 'indexed'
 103                             )
 104                         ),
 105                     )
 106                 )
 107             ),
 108             'highlight' => array(
 109                 'pre_tags' => array('<em class="hl">'),
 110                 'order' => 'score',
 111                 'encoder' => 'html',
 112                 'fields' => array(
 113                     'title' => array(
 114                         'require_field_match' => false,
 115                         'number_of_fragments' => 0,
 116                     ),
 117                     'url' => array(
 118                         'require_field_match' => false,
 119                         'number_of_fragments' => 0,
 120                     ),
 121                     'text' => array(
 122                         'require_field_match' => false,
 123                         'number_of_fragments' => 1,
 124                     ),
 125                 )
 126             ),
 127             'aggregations' => array(
 128                 'tags' => array(
 129                     'terms' => array(
 130                         'field' => 'tags'
 131                     )
 132                 ),
 133                 'language' => array(
 134                     'terms' => array(
 135                         'field' => 'language'
 136                     )
 137                 ),
 138                 'domain' => array(
 139                     'terms' => array(
 140                         'field' => 'domain'
 141                     )
 142                 ),
 143                 'type' => array(
 144                     'terms' => array(
 145                         'field' => 'type'
 146                     )
 147                 )
 148             ),
 149             'from' => $page * $perPage,
 150             'size' => $perPage,
 151             'sort' => $sortCfg,
 152         );
 153         foreach ($filters as $type => $value) {
 154             $doc['query']['bool']['must'][] = array(
 155                 'term' => array(
 156                     $type => $value
 157                 )
 158             );
 159         }
 160         if ($site != '') {
 161             $doc['query']['bool']['must'][] = array(
 162                 'prefix' => array(
 163                     'schemalessUrl' => array(
 164                         'value' => $site
 165                     )
 166                 )
 167             );
 168         }
 169
 170         //unset($doc['_source']);
 171
 172         //ini_set('xdebug.var_display_max_depth', 10);
 173         //echo json_encode($doc);die();
 174         $r->setBody(json_encode($doc));
 175         $res = $r->send();
 176         return json_decode($res->getBody());
 177     }
 178 }
 179 ?>