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 = $baseUrl;
  11     }
  12
  13     public function isKnown($url)
  14     {
  15         $r = new Elasticsearch_Request(
  16             $this->baseUrl . 'document/' . rawurlencode($url),
  17             \HTTP_Request2::METHOD_HEAD
  18         );
  19         $r->allow404 = true;
  20         $res = $r->send();
  21         return $res->getStatus() == 200;
  22     }
  23
  24     public function get($url)
  25     {
  26         $r = new Elasticsearch_Request(
  27             $this->baseUrl . 'document/' . rawurlencode($url),
  28             \HTTP_Request2::METHOD_GET
  29         );
  30         $r->allow404 = true;
  31         $res = $r->send();
  32         if ($res->getStatus() != 200) {
  33             return null;
  34         }
  35         $d = json_decode($res->getBody());
  36         return $d->_source;
  37     }
  38
  39     public function markQueued($url)
  40     {
  41         $r = new Elasticsearch_Request(
  42             $this->baseUrl . 'document/' . rawurlencode($url),
  43             \HTTP_Request2::METHOD_PUT
  44         );
  45         $doc = array(
  46             'status' => 'queued',
  47             'url' => $url
  48         );
  49         $r->setBody(json_encode($doc));
  50         $r->send();
  51     }
  52
  53     public function search($query, $filters, $site, $page, $perPage, $sort)
  54     {
  55         if (preg_match_all('#nick:([^ ]*)#', $query, $matches)) {
  56             foreach ($matches[1] as $authorName) {
  57                 $query = str_replace(
  58                     'nick:' . $authorName,
  59                     'author.name:' . $authorName,
  60                     $query
  61                 );
  62             }
  63         }
  64
  65         $qMust = array();//query parts for the MUST section
  66
  67         //modification date filters
  68         if (preg_match('#after:([^ ]+)#', $query, $matches)) {
  69             $dateAfter = $matches[1];
  70             $query      = trim(str_replace($matches[0], '', $query));
  71             $qMust[]    = array(
  72                 'range' => array(
  73                     'modate' => array(
  74                         'gt' => $dateAfter . '||/d',
  75                     )
  76                 )
  77             );
  78         }
  79         if (preg_match('#before:([^ ]+)#', $query, $matches)) {
  80             $dateBefore = $matches[1];
  81             $query      = trim(str_replace($matches[0], '', $query));
  82             $qMust[]    = array(
  83                 'range' => array(
  84                     'modate' => array(
  85                         'lt' => $dateBefore . '||/d',
  86                     )
  87                 )
  88             );
  89         }
  90         if (preg_match('#date:([^ ]+)#', $query, $matches)) {
  91             $dateExact = $matches[1];
  92             $query      = trim(str_replace($matches[0], '', $query));
  93             $qMust[]    = array(
  94                 'range' => array(
  95                     'modate' => array(
  96                         'gte' => $dateExact . '||/d',
  97                         'lte' => $dateExact . '||/d',
  98                     )
  99                 )
 100             );
 101         }
 102
 103         $qMust[] = array(
 104             'query_string' => array(
 105                 'default_field' => '_all',
 106                 'default_operator' => 'AND',
 107                 'query' => $query
 108             )
 109         );
 110         $qMust[] = array(
 111             'term' => array(
 112                 'status' => 'indexed'
 113             )
 114         );
 115
 116         if ($sort == 'date') {
 117             $sortCfg = array('modate' => array('order' => 'desc'));
 118         } else {
 119             $sortCfg = array();
 120         }
 121
 122         $contentMatchSize = 100;
 123         if ($GLOBALS['phinde']['showFullContent']) {
 124             $contentMatchSize = 999999;
 125         }
 126
 127         $r = new Elasticsearch_Request(
 128             $this->baseUrl . 'document/_search',
 129             \HTTP_Request2::METHOD_GET
 130         );
 131         $doc = array(
 132             '_source' => array(
 133                 'url',
 134                 'title',
 135                 'author',
 136                 'modate',
 137             ),
 138             'query' => array(
 139                 'bool' => array(
 140                     'must' => $qMust
 141                 )
 142             ),
 143             'highlight' => array(
 144                 'pre_tags' => array('<em class="hl">'),
 145                 'order' => 'score',
 146                 'encoder' => 'html',
 147                 'fields' => array(
 148                     'title' => array(
 149                         'require_field_match' => false,
 150                         'number_of_fragments' => 0,
 151                     ),
 152                     'url' => array(
 153                         'require_field_match' => false,
 154                         'number_of_fragments' => 0,
 155                     ),
 156                     'text' => array(
 157                         'require_field_match' => false,
 158                         'number_of_fragments' => 1,
 159                         'fragment_size' => $contentMatchSize,
 160                         'no_match_size' => $contentMatchSize,
 161                     ),
 162                 )
 163             ),
 164             'aggregations' => array(
 165                 'tags' => array(
 166                     'terms' => array(
 167                         'field' => 'tags'
 168                     )
 169                 ),
 170                 'language' => array(
 171                     'terms' => array(
 172                         'field' => 'language'
 173                     )
 174                 ),
 175                 'domain' => array(
 176                     'terms' => array(
 177                         'field' => 'domain'
 178                     )
 179                 ),
 180                 'type' => array(
 181                     'terms' => array(
 182                         'field' => 'type'
 183                     )
 184                 )
 185             ),
 186             'from' => $page * $perPage,
 187             'size' => $perPage,
 188             'sort' => $sortCfg,
 189         );
 190         foreach ($filters as $type => $value) {
 191             $doc['query']['bool']['must'][] = array(
 192                 'term' => array(
 193                     $type => $value
 194                 )
 195             );
 196         }
 197         if ($site != '') {
 198             $doc['query']['bool']['must'][] = array(
 199                 'prefix' => array(
 200                     'schemalessUrl' => array(
 201                         'value' => $site
 202                     )
 203                 )
 204             );
 205         }
 206
 207         //unset($doc['_source']);
 208
 209         //ini_set('xdebug.var_display_max_depth', 10);
 210         //echo json_encode($doc);die();
 211         $r->setBody(json_encode($doc));
 212         $res = $r->send();
 213         return json_decode($res->getBody());
 214     }
 215 }
 216 ?>