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 (preg_match('#nick:([^ ]*)#', $query, $matches)) {
  74             $authorName = $matches[1];
  75             $query = str_replace(
  76                 'nick:' . $authorName,
  77                 'author.name:' . $authorName,
  78                 $query
  79             );
  80         }
  81
  82         if ($sort == 'date') {
  83             $sortCfg = array('modate' => array('order' => 'desc'));
  84         } else {
  85             $sortCfg = array();
  86         }
  87
  88         $r = new Elasticsearch_Request(
  89             $this->baseUrl . 'document/_search',
  90             \HTTP_Request2::METHOD_GET
  91         );
  92         $doc = array(
  93             '_source' => array(
  94                 'url',
  95                 'title',
  96                 'author',
  97                 'modate',
  98             ),
  99             'query' => array(
 100                 'bool' => array(
 101                     'must' => array(
 102                         array(
 103                             'query_string' => array(
 104                                 'default_field' => '_all',
 105                                 'default_operator' => 'AND',
 106                                 'query' => $query
 107                             )
 108                         ),
 109                         array(
 110                             'term' => array(
 111                                 'status' => 'indexed'
 112                             )
 113                         ),
 114                     )
 115                 )
 116             ),
 117             'highlight' => array(
 118                 'pre_tags' => array('<em class="hl">'),
 119                 'order' => 'score',
 120                 'encoder' => 'html',
 121                 'fields' => array(
 122                     'title' => array(
 123                         'require_field_match' => false,
 124                         'number_of_fragments' => 0,
 125                     ),
 126                     'url' => array(
 127                         'require_field_match' => false,
 128                         'number_of_fragments' => 0,
 129                     ),
 130                     'text' => array(
 131                         'require_field_match' => false,
 132                         'number_of_fragments' => 1,
 133                     ),
 134                 )
 135             ),
 136             'aggregations' => array(
 137                 'tags' => array(
 138                     'terms' => array(
 139                         'field' => 'tags'
 140                     )
 141                 ),
 142                 'language' => array(
 143                     'terms' => array(
 144                         'field' => 'language'
 145                     )
 146                 ),
 147                 'domain' => array(
 148                     'terms' => array(
 149                         'field' => 'domain'
 150                     )
 151                 ),
 152                 'type' => array(
 153                     'terms' => array(
 154                         'field' => 'type'
 155                     )
 156                 )
 157             ),
 158             'from' => $page * $perPage,
 159             'size' => $perPage,
 160             'sort' => $sortCfg,
 161         );
 162         foreach ($filters as $type => $value) {
 163             $doc['query']['bool']['must'][] = array(
 164                 'term' => array(
 165                     $type => $value
 166                 )
 167             );
 168         }
 169         if ($site != '') {
 170             $doc['query']['bool']['must'][] = array(
 171                 'prefix' => array(
 172                     'schemalessUrl' => array(
 173                         'value' => $site
 174                     )
 175                 )
 176             );
 177         }
 178
 179         //unset($doc['_source']);
 180
 181         //ini_set('xdebug.var_display_max_depth', 10);
 182         //echo json_encode($doc);die();
 183         $r->setBody(json_encode($doc));
 184         $res = $r->send();
 185         return json_decode($res->getBody());
 186     }
 187 }
 188 ?>