1
|
|
|
<?php declare(strict_types = 1); |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* Created by PhpStorm. |
5
|
|
|
* User: gordon |
6
|
|
|
* Date: 24/3/2561 |
7
|
|
|
* Time: 21:14 น. |
8
|
|
|
*/ |
9
|
|
|
|
10
|
|
|
namespace Suilven\ManticoreSearch\Service; |
11
|
|
|
|
12
|
|
|
use Suilven\FreeTextSearch\Indexes; |
13
|
|
|
use Suilven\FreeTextSearch\Types\FieldTypes; |
14
|
|
|
|
15
|
|
|
// @phpcs:disable Generic.Files.LineLength.TooLong |
16
|
|
|
// @phpcs:disable SlevomatCodingStandard.Files.LineLength.LineTooLong |
17
|
|
|
class IndexCreator extends \Suilven\FreeTextSearch\Base\IndexCreator implements \Suilven\FreeTextSearch\Interfaces\IndexCreator |
18
|
|
|
{ |
19
|
|
|
/** |
20
|
|
|
* Create an index |
21
|
|
|
* |
22
|
|
|
* @todo Refactor into Indexer base |
23
|
|
|
* @param string $indexName the name of the index |
24
|
|
|
*/ |
25
|
|
|
public function createIndex(string $indexName): void |
26
|
|
|
{ |
27
|
|
|
$fields = $this->getFields($indexName); |
28
|
|
|
$storedFields = $this->getStoredFields($indexName); |
29
|
|
|
$specs = $this->getFieldSpecs($indexName); |
30
|
|
|
|
31
|
|
|
\error_log('SPECS'); |
32
|
|
|
\print_r($specs); |
33
|
|
|
|
34
|
|
|
$columns = []; |
35
|
|
|
foreach ($fields as $field) { |
36
|
|
|
$fieldType = $specs[$field]; |
37
|
|
|
|
38
|
|
|
// this will be the most common |
39
|
|
|
$indexType = 'text'; |
40
|
|
|
|
41
|
|
|
// @todo configure index to strip HTML |
42
|
|
|
switch ($fieldType) { |
43
|
|
|
case FieldTypes::FOREIGN_KEY: |
44
|
|
|
// @todo this perhaps needs to be a token |
45
|
|
|
// See https://docs.manticoresearch.com/3.4.0/html/indexing/data_types.html |
46
|
|
|
|
47
|
|
|
// @todo also how to mark strings for tokenizing? |
48
|
|
|
$indexType = 'bigint'; |
49
|
|
|
|
50
|
|
|
break; |
51
|
|
|
case FieldTypes::INTEGER: |
52
|
|
|
$indexType = 'integer'; |
53
|
|
|
|
54
|
|
|
break; |
55
|
|
|
case FieldTypes::FLOAT: |
56
|
|
|
$indexType = 'float'; |
57
|
|
|
|
58
|
|
|
break; |
59
|
|
|
case FieldTypes::TIME: |
60
|
|
|
$indexType = 'timestamp'; |
61
|
|
|
|
62
|
|
|
break; |
63
|
|
|
case FieldTypes::BOOLEAN: |
64
|
|
|
// @todo is there a better type? |
65
|
|
|
$indexType = 'integer'; |
66
|
|
|
|
67
|
|
|
break; |
68
|
|
|
} |
69
|
|
|
|
70
|
|
|
$options = []; |
71
|
|
|
if ($indexType === 'text') { |
72
|
|
|
$options = ['indexed', 'stored']; |
73
|
|
|
} |
74
|
|
|
|
75
|
|
|
// override for Link, do not index it. The storing of the Link URL is to save on database hierarchy |
76
|
|
|
// traversal when rendering search results |
77
|
|
|
if ($field === 'Link' || \in_array($field, $storedFields, true)) { |
78
|
|
|
$options = ['stored']; |
79
|
|
|
} |
80
|
|
|
$columns[$field] = ['type' => $indexType, 'options' => $options]; |
81
|
|
|
} |
82
|
|
|
|
83
|
|
|
|
84
|
|
|
// @todo Add has one |
85
|
|
|
|
86
|
|
|
$indexes = new Indexes(); |
87
|
|
|
$index = $indexes->getIndex($indexName); |
88
|
|
|
$mvaFields = $index->getHasManyFields(); |
89
|
|
|
error_log(print_r($mvaFields, true)); |
90
|
|
|
|
91
|
|
|
foreach(array_keys($mvaFields) as $mvaColumnName) { |
92
|
|
|
$columns[$mvaColumnName] = ['type' => 'multi']; |
93
|
|
|
} |
94
|
|
|
|
95
|
|
|
$client = new Client(); |
96
|
|
|
$manticoreClient = $client->getConnection(); |
97
|
|
|
|
98
|
|
|
$settings = [ |
99
|
|
|
'rt_mem_limit' => '256M', |
100
|
|
|
'dict' => 'keywords', |
101
|
|
|
'min_infix_len' => 2, |
102
|
|
|
'html_strip' => 1, |
103
|
|
|
'bigram_index' => 'both_freq' |
104
|
|
|
]; |
105
|
|
|
|
106
|
|
|
|
107
|
|
|
// drop index, and updating an existing one does not effect change |
108
|
|
|
$manticoreClient->indices()->drop(['index' => $indexName, 'body'=>['silent'=>true]]); |
109
|
|
|
|
110
|
|
|
|
111
|
|
|
$manticoreIndex = new \Manticoresearch\Index($manticoreClient, $indexName); |
112
|
|
|
|
113
|
|
|
$manticoreIndex->create( |
114
|
|
|
$columns, |
115
|
|
|
$settings, |
116
|
|
|
true |
117
|
|
|
); |
118
|
|
|
} |
119
|
|
|
} |
120
|
|
|
|