|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
/** |
|
6
|
|
|
* This file is part of Scout Extended. |
|
7
|
|
|
* |
|
8
|
|
|
* (c) Algolia Team <[email protected]> |
|
9
|
|
|
* |
|
10
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
11
|
|
|
* file that was distributed with this source code. |
|
12
|
|
|
*/ |
|
13
|
|
|
|
|
14
|
|
|
namespace Algolia\ScoutExtended\Splitters; |
|
15
|
|
|
|
|
16
|
|
|
use DOMXPath; |
|
17
|
|
|
use DOMDocument; |
|
18
|
|
|
use Algolia\ScoutExtended\Contracts\SplitterContract; |
|
19
|
|
|
|
|
20
|
|
|
class HtmlSplitter implements SplitterContract |
|
21
|
|
|
{ |
|
22
|
|
|
/** |
|
23
|
|
|
* The list of html tags. |
|
24
|
|
|
* |
|
25
|
|
|
* @var string[] |
|
26
|
|
|
*/ |
|
27
|
|
|
protected $nodes = [ |
|
28
|
|
|
'h1', |
|
29
|
|
|
'h2', |
|
30
|
|
|
'h3', |
|
31
|
|
|
'h4', |
|
32
|
|
|
'h5', |
|
33
|
|
|
'h6', |
|
34
|
|
|
'p', |
|
35
|
|
|
]; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* String for key check purpose. |
|
39
|
|
|
* |
|
40
|
|
|
* @const string IMPORTANCE |
|
41
|
|
|
*/ |
|
42
|
|
|
const IMPORTANCE = 'importance'; |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* Creates a new instance of the class. |
|
46
|
|
|
* |
|
47
|
|
|
* @param array $nodes |
|
48
|
|
|
* |
|
49
|
|
|
* @return void |
|
50
|
|
|
*/ |
|
51
|
3 |
|
public function __construct(array $nodes = null) |
|
52
|
|
|
{ |
|
53
|
3 |
|
if ($nodes !== null) { |
|
54
|
2 |
|
$this->nodes = $nodes; |
|
55
|
|
|
} |
|
56
|
3 |
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* Find weight of current nodes. |
|
60
|
|
|
* |
|
61
|
|
|
* @param array $object |
|
62
|
|
|
* |
|
63
|
|
|
* @return int |
|
64
|
|
|
*/ |
|
65
|
3 |
|
public function findWeight(array $object): int |
|
66
|
|
|
{ |
|
67
|
3 |
|
return (int) array_search((key($object)), $this->nodes); |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
/** |
|
71
|
|
|
* Add object to queue. |
|
72
|
|
|
* |
|
73
|
|
|
* @param array $object |
|
74
|
|
|
* @param array $queue |
|
75
|
|
|
* |
|
76
|
|
|
* @return array |
|
77
|
|
|
*/ |
|
78
|
3 |
|
public function addObjectToQueue(array $object, array $queue): array |
|
79
|
|
|
{ |
|
80
|
3 |
|
if (count($queue) == 0) { |
|
81
|
3 |
|
$queue[] = $object; |
|
82
|
|
|
|
|
83
|
3 |
|
return $queue; |
|
84
|
|
|
} else { |
|
85
|
3 |
|
if ($this->findWeight($object) > $this->findWeight(end($queue))) { |
|
86
|
1 |
|
$queue[] = $object; |
|
87
|
|
|
|
|
88
|
1 |
|
return $queue; |
|
89
|
|
|
} else { |
|
90
|
3 |
|
array_pop($queue); |
|
91
|
|
|
|
|
92
|
3 |
|
return $this->addObjectToQueue($object, $queue); |
|
93
|
|
|
} |
|
94
|
|
|
} |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
/** |
|
98
|
|
|
* Importance formula. |
|
99
|
|
|
* Give integer from tags ranking. |
|
100
|
|
|
* |
|
101
|
|
|
* @param \DOMElement $node |
|
102
|
|
|
* @param array $queue |
|
103
|
|
|
* |
|
104
|
|
|
* @return int |
|
105
|
|
|
*/ |
|
106
|
3 |
|
public function importanceWeight(\DOMElement $node, array $queue): int |
|
107
|
|
|
{ |
|
108
|
3 |
|
if ($node->nodeName === 'p') { |
|
109
|
1 |
|
if (empty(end($queue))) { |
|
110
|
|
|
return 0; |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
1 |
|
return (int) (count($this->nodes) - 1) + (int) (array_search(key(end($queue)), $this->nodes)); |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
3 |
|
return (int) array_search($node->nodeName, $this->nodes); |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
/** |
|
120
|
|
|
* Clean Records to have a correct format. |
|
121
|
|
|
* |
|
122
|
|
|
* |
|
123
|
|
|
* @param array $objects |
|
124
|
|
|
* |
|
125
|
|
|
* @return array |
|
126
|
|
|
*/ |
|
127
|
3 |
|
public function cleanRecords(array $objects): array |
|
128
|
|
|
{ |
|
129
|
3 |
|
$records = []; |
|
130
|
3 |
|
foreach ($objects as $object) { |
|
131
|
3 |
|
foreach ($object as $data) { |
|
132
|
3 |
|
foreach ($data as $key => $value) { |
|
133
|
3 |
|
$record[$key] = $value; |
|
134
|
3 |
|
if ($key === self::IMPORTANCE) { |
|
135
|
3 |
|
$records[] = $record; |
|
136
|
3 |
|
$record = []; |
|
137
|
|
|
} |
|
138
|
|
|
} |
|
139
|
|
|
} |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
3 |
|
return $records; |
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
|
|
/** |
|
146
|
|
|
* Acts a static factory. |
|
147
|
|
|
* |
|
148
|
|
|
* @param string|array $tags |
|
149
|
|
|
* |
|
150
|
|
|
* @return static |
|
151
|
|
|
*/ |
|
152
|
2 |
|
public static function by($tags) |
|
153
|
|
|
{ |
|
154
|
2 |
|
return new static((array) $tags); |
|
155
|
|
|
} |
|
156
|
|
|
|
|
157
|
|
|
/** |
|
158
|
|
|
* Splits the given value. |
|
159
|
|
|
* |
|
160
|
|
|
* @param object $searchable |
|
161
|
|
|
* @param string $value |
|
162
|
|
|
* |
|
163
|
|
|
* @return array |
|
164
|
|
|
*/ |
|
165
|
3 |
|
public function split($searchable, $value): array |
|
166
|
|
|
{ |
|
167
|
3 |
|
$dom = new DOMDocument(); |
|
168
|
3 |
|
$dom->loadHTML($value); |
|
169
|
3 |
|
$xpath = new DOMXpath($dom); |
|
170
|
3 |
|
$queue = []; |
|
171
|
3 |
|
$objects = []; |
|
172
|
3 |
|
$xpathQuery = '//'.implode(' | //', $this->nodes); |
|
173
|
3 |
|
$nodes = $xpath->query($xpathQuery); |
|
174
|
|
|
|
|
175
|
3 |
|
foreach ($nodes as $node) { |
|
176
|
3 |
|
$object = [$node->nodeName => $node->textContent]; |
|
177
|
3 |
|
$importance = $this->importanceWeight($node, $queue); |
|
178
|
3 |
|
$queue = $this->addObjectToQueue($object, $queue); |
|
179
|
3 |
|
$cloneQueue = $queue; |
|
180
|
3 |
|
$cloneQueue[] = [self::IMPORTANCE => $importance]; |
|
181
|
3 |
|
$objects[] = $cloneQueue; |
|
182
|
|
|
} |
|
183
|
|
|
|
|
184
|
3 |
|
$records = $this->cleanRecords($objects); |
|
185
|
|
|
|
|
186
|
3 |
|
return $records; |
|
187
|
|
|
} |
|
188
|
|
|
} |
|
189
|
|
|
|
algolia /
scout-extended
