|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* User: jensk |
|
4
|
|
|
* Date: 21-2-2017 |
|
5
|
|
|
* Time: 10:38 |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
namespace library\search; |
|
9
|
|
|
|
|
10
|
|
|
|
|
11
|
|
|
use library\storage\Document; |
|
12
|
|
|
use library\storage\JsonStorage; |
|
13
|
|
|
|
|
14
|
|
|
class DocumentTokenizer |
|
15
|
|
|
{ |
|
16
|
|
|
/** |
|
17
|
|
|
* @var Document |
|
18
|
|
|
*/ |
|
19
|
|
|
protected $document; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* @var array |
|
23
|
|
|
*/ |
|
24
|
|
|
protected $tokenVector = array(); |
|
25
|
|
|
protected $storage; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Tokenizer constructor. |
|
29
|
|
|
* |
|
30
|
|
|
* @param \library\storage\Document $document |
|
31
|
|
|
* @param JsonStorage $storage |
|
32
|
|
|
*/ |
|
33
|
|
|
public function __construct(Document $document, JsonStorage $storage) |
|
34
|
|
|
{ |
|
35
|
|
|
$this->document = $document; |
|
36
|
|
|
$this->storage = $storage; |
|
37
|
|
|
$this->tokenize(); |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* Execute tokenization of all document fields |
|
42
|
|
|
*/ |
|
43
|
|
|
private function tokenize() |
|
44
|
|
|
{ |
|
45
|
|
|
$this->tokenizeTitle(); |
|
46
|
|
|
$this->tokenizeFields(); |
|
47
|
|
|
$this->tokenizeBricks(); |
|
48
|
|
|
$this->tokenizeDynamicBricks(); |
|
49
|
|
|
$this->tokenVector = array_filter($this->tokenVector); |
|
50
|
|
|
arsort($this->tokenVector); |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
private function tokenizeTitle() |
|
54
|
|
|
{ |
|
55
|
|
|
$filteredString = new CharacterFilter($this->document->title); |
|
56
|
|
|
$tokenizer = new Tokenizer($filteredString); |
|
57
|
|
|
$this->addTokenVectorToVector($tokenizer->getTokenVector(), 'title'); |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
private function tokenizeFields() |
|
61
|
|
|
{ |
|
62
|
|
|
$fields = $this->document->fields; |
|
63
|
|
|
$documentDefinition = $this->storage->getDocumentTypeBySlug($this->document->documentTypeSlug); |
|
64
|
|
|
foreach ($fields as $fieldName => $field) { |
|
65
|
|
|
$fieldType = $this->getFieldType($fieldName, $documentDefinition); |
|
66
|
|
|
$this->tokenizeField($field, $fieldName, $fieldType); |
|
67
|
|
|
} |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
private function tokenizeField($field, $fieldName, $fieldType) |
|
71
|
|
|
{ |
|
72
|
|
|
foreach ($field as $value) { |
|
73
|
|
|
// Only index fields that contain text |
|
74
|
|
|
if (in_array($fieldType, array('String', 'Text', 'Rich Text'))) { |
|
75
|
|
|
$filteredString = new CharacterFilter($value); |
|
76
|
|
|
$tokenizer = new Tokenizer($filteredString); |
|
77
|
|
|
$this->addTokenVectorToVector($tokenizer->getTokenVector(), $fieldName); |
|
78
|
|
|
} |
|
79
|
|
|
} |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
private function tokenizeBricks() |
|
83
|
|
|
{ |
|
84
|
|
|
$bricks = $this->document->bricks; |
|
85
|
|
|
foreach ($bricks as $brickSlug => $brick) { |
|
86
|
|
|
$this->tokenizeBrick($brick, $brickSlug); |
|
87
|
|
|
} |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
private function tokenizeBrick($brick, $brickSlug) |
|
91
|
|
|
{ |
|
92
|
|
|
$fields = $brick->fields; |
|
93
|
|
|
$brickDefinition = $this->storage->getBrickBySlug($brick->type); |
|
94
|
|
|
foreach ($fields as $fieldName => $field) { |
|
95
|
|
|
$fieldType = $this->getFieldType($fieldName, $brickDefinition); |
|
96
|
|
|
$this->tokenizeField($field, $brickSlug . '__' . $fieldName, $fieldType); |
|
97
|
|
|
} |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
private function tokenizeDynamicBricks() |
|
101
|
|
|
{ |
|
102
|
|
|
$dynamicBricks = $this->document->dynamicBricks; |
|
103
|
|
|
foreach ($dynamicBricks as $key => $brick) { |
|
104
|
|
|
$this->tokenizeBrick($brick, 'dynamicBricks__' . $brick->type . $key); |
|
105
|
|
|
} |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
public function getTokens() |
|
109
|
|
|
{ |
|
110
|
|
|
return $this->tokenVector; |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
/** |
|
114
|
|
|
* Add a token to the existing tokenvector |
|
115
|
|
|
* @param $token |
|
116
|
|
|
* @param string $field |
|
117
|
|
|
* @param int $count |
|
118
|
|
|
*/ |
|
119
|
|
|
private function addTokenToVector($token, $field, $count = 1) |
|
120
|
|
|
{ |
|
121
|
|
|
if (!empty($token)) { |
|
122
|
|
|
if (isset($this->tokenVector[$field][$token])) { |
|
123
|
|
|
$this->tokenVector[$field][$token] += $count; |
|
124
|
|
|
} else { |
|
125
|
|
|
$this->tokenVector[$field][$token] = $count; |
|
126
|
|
|
} |
|
127
|
|
|
} |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
/** |
|
131
|
|
|
* Add a complete token vector to the existing one. |
|
132
|
|
|
* @param $tokenVector |
|
133
|
|
|
* @param $field |
|
134
|
|
|
*/ |
|
135
|
|
|
private function addTokenVectorToVector($tokenVector, $field) |
|
136
|
|
|
{ |
|
137
|
|
|
foreach ($tokenVector as $token => $count) { |
|
138
|
|
|
$this->addTokenToVector($token, $field, $count); |
|
139
|
|
|
} |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
|
|
/** |
|
143
|
|
|
* Get the type for a field |
|
144
|
|
|
* @param $fieldName |
|
145
|
|
|
* @param $documentDefinition |
|
146
|
|
|
* @return mixed |
|
147
|
|
|
* @throws \Exception |
|
148
|
|
|
*/ |
|
149
|
|
|
private function getFieldType($fieldName, $documentDefinition) |
|
150
|
|
|
{ |
|
151
|
|
|
foreach ($documentDefinition->fields as $fieldTypeDefinition) { |
|
152
|
|
|
if ($fieldTypeDefinition->slug === $fieldName) { |
|
153
|
|
|
return $fieldTypeDefinition->type; |
|
154
|
|
|
} |
|
155
|
|
|
} |
|
156
|
|
|
|
|
157
|
|
|
throw new \Exception('Unknown field type for field' . $fieldName . ' in document ' . $this->document->path); |
|
158
|
|
|
} |
|
159
|
|
|
} |
jenskooij /
cloudcontrol
