|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace rOpenDev\ExtractExpression; |
|
4
|
|
|
|
|
5
|
|
|
// curl: automatic set referrer |
|
6
|
|
|
// pomme de terre ne marche pas |
|
7
|
|
|
|
|
8
|
|
|
class ExtractExpression |
|
9
|
|
|
{ |
|
10
|
|
|
public $onlyInSentence = false; |
|
11
|
|
|
public $expressionMaxWords = 5; |
|
12
|
|
|
|
|
13
|
|
|
/** |
|
14
|
|
|
* @var int Minimum value for wich one we keep a trail. Set to 0 to disabled keeping trail |
|
15
|
|
|
**/ |
|
16
|
|
|
public $keepTrail = 5; |
|
17
|
|
|
|
|
18
|
|
|
protected $words; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* @var array contain the source texts |
|
22
|
|
|
**/ |
|
23
|
|
|
protected $text = []; |
|
24
|
|
|
/** |
|
25
|
|
|
* @var array textKey => [expression => number] |
|
26
|
|
|
**/ |
|
27
|
|
|
protected $expressions = []; |
|
28
|
|
|
|
|
29
|
|
|
protected $mergedExpressions; |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* @var array containing the number of word analyzed for a specific source text |
|
33
|
|
|
**/ |
|
34
|
|
|
protected $wordNumber = []; |
|
35
|
|
|
|
|
36
|
|
|
public function __construct() |
|
37
|
|
|
{ |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
public function addContent(string $text) |
|
41
|
|
|
{ |
|
42
|
|
|
$text = new ExtractExpressions($text); |
|
43
|
|
|
$text->onlyInSentence = $this->onlyInSentence; |
|
44
|
|
|
$text->expressionMaxWords = $this->expressionMaxWords; |
|
45
|
|
|
$text->keepTrail = $this->keepTrail; |
|
46
|
|
|
$text->extract(); |
|
47
|
|
|
|
|
48
|
|
|
$this->text[] = $text; |
|
49
|
|
|
|
|
50
|
|
|
return $this; |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
public function exec() |
|
54
|
|
|
{ |
|
55
|
|
|
$mergedExpressions = []; |
|
56
|
|
|
|
|
57
|
|
|
foreach ($this->text as $text) { |
|
58
|
|
|
$expressions = $text->getExpressionsByDensity(); |
|
59
|
|
|
foreach ($expressions as $expression => $density) { |
|
60
|
|
|
$mergedExpressions[$expression] = (isset($mergedExpressions[$expression]) ? $mergedExpressions[$expression] : 0) + $density; |
|
61
|
|
|
} |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
arsort($mergedExpressions); |
|
65
|
|
|
|
|
66
|
|
|
$this->mergedExpressions = $mergedExpressions; |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
public function getExpressions(?int $number = null) |
|
70
|
|
|
{ |
|
71
|
|
|
if (null === $this->mergedExpressions) { |
|
72
|
|
|
$this->exec(); |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
return !$number ? $this->mergedExpressions : array_slice($this->getExpressions(), 0, $number); |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
public function getWordNumber() |
|
79
|
|
|
{ |
|
80
|
|
|
if (null === $this->mergedExpressions) { |
|
81
|
|
|
$this->exec(); |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
$wn = 0; |
|
85
|
|
|
foreach ($this->text as $text) { |
|
86
|
|
|
$wn = $wn + $text->getWordNumber(); |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
return $wn; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
/** |
|
93
|
|
|
* @return array containing sentence where we can find expresion |
|
94
|
|
|
*/ |
|
95
|
|
|
public function getTrail(string $expression) |
|
96
|
|
|
{ |
|
97
|
|
|
if (null === $this->mergedExpressions) { |
|
98
|
|
|
$this->exec(); |
|
99
|
|
|
} |
|
100
|
|
|
$trail = []; |
|
101
|
|
|
|
|
102
|
|
|
foreach ($this->text as $text) { |
|
103
|
|
|
$trail = array_merge($trail, $tex->getTrail($expression)); |
|
|
|
|
|
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
return $trail; |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
public function getTrails() |
|
110
|
|
|
{ |
|
111
|
|
|
$trail = []; |
|
112
|
|
|
|
|
113
|
|
|
foreach ($this->text as $text) { |
|
114
|
|
|
$trail = array_merge($trail, $text->getTrails()); |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
return $trail; |
|
118
|
|
|
} |
|
119
|
|
|
} |
|
120
|
|
|
|
RobinDev /
ExtractExpression
Loading history...