735b64ff7ae6ad0a4fbe3e08d31f3fecab540ad1
[phinde.git] / 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                                 'query' => $query
91                             )
92                         ),
93                         array(
94                             'term' => array(
95                                 'status' => 'indexed'
96                             )
97                         ),
98                     )
99                 )
100             ),
101             'highlight' => array(
102                 'pre_tags' => array('<em class="hl">'),
103                 'order' => 'score',
104                 'fields' => array(
105                     'title' => array(
106                         'require_field_match' => false,
107                         'number_of_fragments' => 0,
108                     ),
109                     'url' => array(
110                         'require_field_match' => false,
111                         'number_of_fragments' => 0,
112                     ),
113                     'text' => array(
114                         'require_field_match' => false,
115                         'number_of_fragments' => 1,
116                     ),
117                 )
118             ),
119             'aggregations' => array(
120                 'tags' => array(
121                     'terms' => array(
122                         'field' => 'tags'
123                     )
124                 ),
125                 'language' => array(
126                     'terms' => array(
127                         'field' => 'language'
128                     )
129                 ),
130                 'domain' => array(
131                     'terms' => array(
132                         'field' => 'domain'
133                     )
134                 ),
135                 'type' => array(
136                     'terms' => array(
137                         'field' => 'type'
138                     )
139                 )
140             ),
141             'from' => $page * $perPage,
142             'size' => $perPage,
143             'sort' => array(
144                 //array('modate' => array('order' => 'desc'))
145             )
146         );
147         foreach ($filters as $type => $value) {
148             $doc['query']['bool']['must'][] = array(
149                 'term' => array(
150                     $type => $value
151                 )
152             );
153         }
154         if ($site != '') {
155             $doc['query']['bool']['must'][] = array(
156                 'prefix' => array(
157                     'schemalessUrl' => array(
158                         'value' => $site
159                     )
160                 )
161             );
162         }
163
164         //unset($doc['_source']);
165
166         //ini_set('xdebug.var_display_max_depth', 10);
167         //echo json_encode($doc);die();
168         $r->setBody(json_encode($doc));
169         $res = $r->send();
170         return json_decode($res->getBody());
171     }
172 }
173 ?>