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 = $baseUrl;
  11     }
  12
  13     public static function getDocId($url)
  14     {
  15         return hash('sha256', $url);
  16     }
  17
  18     public function isKnown($url)
  19     {
  20         $r = new Elasticsearch_Request(
  21             $this->baseUrl . 'document/' . static::getDocId($url),
  22             \HTTP_Request2::METHOD_HEAD
  23         );
  24         $r->allow404 = true;
  25         $res = $r->send();
  26         return $res->getStatus() == 200;
  27     }
  28
  29     public function get($url)
  30     {
  31         $r = new Elasticsearch_Request(
  32             $this->baseUrl . 'document/' . static::getDocId($url),
  33             \HTTP_Request2::METHOD_GET
  34         );
  35         $r->allow404 = true;
  36         $res = $r->send();
  37         if ($res->getStatus() != 200) {
  38             return null;
  39         }
  40         $d = json_decode($res->getBody());
  41         return $d->_source;
  42     }
  43
  44     public function markQueued($url)
  45     {
  46         $r = new Elasticsearch_Request(
  47             $this->baseUrl . 'document/' . static::getDocId($url),
  48             \HTTP_Request2::METHOD_PUT
  49         );
  50         $doc = (object) array(
  51             'url' => $url,
  52             'status' => (object) array(
  53                 'processed' => null,
  54                 'findable'  => false,
  55             )
  56         );
  57         $r->setBody(json_encode($doc));
  58         $r->send();
  59     }
  60
  61     public function getIndexStatus()
  62     {
  63         $r = new Elasticsearch_Request(
  64             $this->baseUrl . '_stats/docs,store',
  65             \HTTP_Request2::METHOD_GET
  66         );
  67         $res = $r->send();
  68         $data = json_decode($res->getBody());
  69         return array(
  70             'documents' => $data->_all->total->docs->count,
  71             'size'      => $data->_all->total->store->size_in_bytes,
  72         );
  73     }
  74
  75     public function search($query, $filters, $site, $page, $perPage, $sort)
  76     {
  77         if (preg_match_all('#nick:([^ ]*)#', $query, $matches)) {
  78             foreach ($matches[1] as $authorName) {
  79                 $query = str_replace(
  80                     'nick:' . $authorName,
  81                     'author.name:' . $authorName,
  82                     $query
  83                 );
  84             }
  85         }
  86
  87         $qMust = array();//query parts for the MUST section
  88
  89         //modification date filters
  90         if (preg_match('#after:([^ ]+)#', $query, $matches)) {
  91             $dateAfter = $matches[1];
  92             $query      = trim(str_replace($matches[0], '', $query));
  93             $qMust[]    = array(
  94                 'range' => array(
  95                     'modate' => array(
  96                         'gt' => $dateAfter . '||/d',
  97                     )
  98                 )
  99             );
 100         }
 101         if (preg_match('#before:([^ ]+)#', $query, $matches)) {
 102             $dateBefore = $matches[1];
 103             $query      = trim(str_replace($matches[0], '', $query));
 104             $qMust[]    = array(
 105                 'range' => array(
 106                     'modate' => array(
 107                         'lt' => $dateBefore . '||/d',
 108                     )
 109                 )
 110             );
 111         }
 112         if (preg_match('#date:([^ ]+)#', $query, $matches)) {
 113             $dateExact = $matches[1];
 114             $query      = trim(str_replace($matches[0], '', $query));
 115             $qMust[]    = array(
 116                 'range' => array(
 117                     'modate' => array(
 118                         'gte' => $dateExact . '||/d',
 119                         'lte' => $dateExact . '||/d',
 120                     )
 121                 )
 122             );
 123         }
 124
 125         $qMust[] = array(
 126             'query_string' => array(
 127                 'default_field' => '_all',
 128                 'default_operator' => 'AND',
 129                 'query' => $query
 130             )
 131         );
 132         $qMust[] = array(
 133             'term' => array(
 134                 'status.findable' => true
 135             )
 136         );
 137
 138         if ($sort == '' && $GLOBALS['phinde']['defaultSort'] == 'date') {
 139             $sort = 'date';
 140         }
 141         if ($sort == 'date') {
 142             $sortCfg = array('status.modate' => array('order' => 'desc'));
 143         } else {
 144             $sortCfg = array();
 145         }
 146
 147         $contentMatchSize = 100;
 148         if ($GLOBALS['phinde']['showFullContent']) {
 149             $contentMatchSize = 999999;
 150         }
 151
 152         $r = new Elasticsearch_Request(
 153             $this->baseUrl . 'document/_search',
 154             \HTTP_Request2::METHOD_GET
 155         );
 156         $doc = array(
 157             '_source' => array(
 158                 'url',
 159                 'title',
 160                 'author',
 161                 'status.modate',
 162             ),
 163             'query' => array(
 164                 'bool' => array(
 165                     'must' => $qMust
 166                 )
 167             ),
 168             'highlight' => array(
 169                 'pre_tags' => array('<em class="hl">'),
 170                 'order' => 'score',
 171                 'encoder' => 'html',
 172                 'fields' => array(
 173                     'title' => array(
 174                         'require_field_match' => false,
 175                         'number_of_fragments' => 0,
 176                     ),
 177                     'url' => array(
 178                         'require_field_match' => false,
 179                         'number_of_fragments' => 0,
 180                     ),
 181                     'text' => array(
 182                         'require_field_match' => false,
 183                         'number_of_fragments' => 1,
 184                         'fragment_size' => $contentMatchSize,
 185                         'no_match_size' => $contentMatchSize,
 186                     ),
 187                 )
 188             ),
 189             'aggregations' => array(
 190                 'tags' => array(
 191                     'terms' => array(
 192                         'field' => 'tags'
 193                     )
 194                 ),
 195                 'language' => array(
 196                     'terms' => array(
 197                         'field' => 'language'
 198                     )
 199                 ),
 200                 'domain' => array(
 201                     'terms' => array(
 202                         'field' => 'domain'
 203                     )
 204                 ),
 205                 'type' => array(
 206                     'terms' => array(
 207                         'field' => 'type'
 208                     )
 209                 )
 210             ),
 211             'from' => $page * $perPage,
 212             'size' => $perPage,
 213             'sort' => $sortCfg,
 214         );
 215         foreach ($filters as $type => $value) {
 216             $doc['query']['bool']['must'][] = array(
 217                 'term' => array(
 218                     $type => $value
 219                 )
 220             );
 221         }
 222         if ($site != '') {
 223             $doc['query']['bool']['must'][] = array(
 224                 'prefix' => array(
 225                     'schemalessUrl' => array(
 226                         'value' => $site
 227                     )
 228                 )
 229             );
 230         }
 231
 232         //unset($doc['_source']);
 233
 234         //ini_set('xdebug.var_display_max_depth', 10);
 235         //echo json_encode($doc);die();
 236         $r->setBody(json_encode($doc));
 237         $res = $r->send();
 238         return json_decode($res->getBody());
 239     }
 240 }
 241 ?>