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)
  72     {
  73         $r = new Elasticsearch_Request(
  74             $this->baseUrl . 'document/_search',
  75             \HTTP_Request2::METHOD_GET
  76         );
  77         $doc = array(
  78             '_source' => array(
  79                 'url',
  80                 'title',
  81                 'author',
  82                 'modate',
  83             ),
  84             'query' => array(
  85                 'bool' => array(
  86                     'must' => array(
  87                         array(
  88                             'query_string' => array(
  89                                 'default_field' => '_all',
  90                                 'default_operator' => 'AND',
  91                                 'query' => $query
  92                             )
  93                         ),
  94                         array(
  95                             'term' => array(
  96                                 'status' => 'indexed'
  97                             )
  98                         ),
  99                     )
 100                 )
 101             ),
 102             'highlight' => array(
 103                 'pre_tags' => array('<em class="hl">'),
 104                 'order' => 'score',
 105                 'encoder' => 'html',
 106                 'fields' => array(
 107                     'title' => array(
 108                         'require_field_match' => false,
 109                         'number_of_fragments' => 0,
 110                     ),
 111                     'url' => array(
 112                         'require_field_match' => false,
 113                         'number_of_fragments' => 0,
 114                     ),
 115                     'text' => array(
 116                         'require_field_match' => false,
 117                         'number_of_fragments' => 1,
 118                     ),
 119                 )
 120             ),
 121             'aggregations' => array(
 122                 'tags' => array(
 123                     'terms' => array(
 124                         'field' => 'tags'
 125                     )
 126                 ),
 127                 'language' => array(
 128                     'terms' => array(
 129                         'field' => 'language'
 130                     )
 131                 ),
 132                 'domain' => array(
 133                     'terms' => array(
 134                         'field' => 'domain'
 135                     )
 136                 ),
 137                 'type' => array(
 138                     'terms' => array(
 139                         'field' => 'type'
 140                     )
 141                 )
 142             ),
 143             'from' => $page * $perPage,
 144             'size' => $perPage,
 145             'sort' => array(
 146                 //array('modate' => array('order' => 'desc'))
 147             )
 148         );
 149         foreach ($filters as $type => $value) {
 150             $doc['query']['bool']['must'][] = array(
 151                 'term' => array(
 152                     $type => $value
 153                 )
 154             );
 155         }
 156         if ($site != '') {
 157             $doc['query']['bool']['must'][] = array(
 158                 'prefix' => array(
 159                     'schemalessUrl' => array(
 160                         'value' => $site
 161                     )
 162                 )
 163             );
 164         }
 165
 166         //unset($doc['_source']);
 167
 168         //ini_set('xdebug.var_display_max_depth', 10);
 169         //echo json_encode($doc);die();
 170         $r->setBody(json_encode($doc));
 171         $res = $r->send();
 172         return json_decode($res->getBody());
 173     }
 174 }
 175 ?>