|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace PiedWeb\TextAnalyzer; |
|
4
|
|
|
|
|
5
|
|
|
class MultiAnalyzer |
|
6
|
|
|
{ |
|
7
|
|
|
protected $onlyInSentence; |
|
8
|
|
|
protected $expressionMaxWords; |
|
9
|
|
|
protected $keepTrail; |
|
10
|
|
|
|
|
11
|
|
|
protected $words; |
|
12
|
|
|
|
|
13
|
|
|
/** |
|
14
|
|
|
* @var array contain the source texts |
|
15
|
|
|
*/ |
|
16
|
|
|
protected $text = []; |
|
17
|
|
|
|
|
18
|
3 |
|
public function __construct( |
|
19
|
|
|
bool $onlySentence = false, |
|
20
|
|
|
int $expressionMaxWords = 5, |
|
21
|
|
|
int $keepTrail = 3 |
|
22
|
|
|
) { |
|
23
|
3 |
|
$this->onlyInSentence = $onlySentence; |
|
24
|
3 |
|
$this->expressionMaxWords = $expressionMaxWords; |
|
25
|
3 |
|
$this->keepTrail = $keepTrail; |
|
26
|
3 |
|
} |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* @return Analysis |
|
30
|
|
|
*/ |
|
31
|
3 |
|
public function addContent(string $text) |
|
32
|
|
|
{ |
|
33
|
3 |
|
$text = Analyzer::get( |
|
34
|
3 |
|
$text, |
|
35
|
3 |
|
$this->onlyInSentence, |
|
36
|
3 |
|
$this->expressionMaxWords, |
|
37
|
3 |
|
$this->keepTrail |
|
38
|
|
|
); |
|
39
|
|
|
|
|
40
|
3 |
|
$this->text[] = $text; |
|
41
|
|
|
|
|
42
|
3 |
|
return $text; |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
3 |
|
public function exec() |
|
46
|
|
|
{ |
|
47
|
3 |
|
$mergedExpressions = []; |
|
48
|
|
|
|
|
49
|
3 |
|
foreach ($this->text as $text) { |
|
50
|
3 |
|
$expressions = $text->getExpressionsByDensity(); |
|
51
|
3 |
|
foreach ($expressions as $expression => $density) { |
|
52
|
3 |
|
$mergedExpressions[$expression] = |
|
53
|
3 |
|
(isset($mergedExpressions[$expression]) ? $mergedExpressions[$expression] : 0) |
|
54
|
3 |
|
+ $density |
|
55
|
|
|
; |
|
56
|
|
|
} |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
3 |
|
arsort($mergedExpressions); |
|
60
|
|
|
|
|
61
|
3 |
|
return new Analysis($mergedExpressions, $this->getWordNumber(), $this->getTrails()); |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
3 |
|
protected function getWordNumber() |
|
65
|
|
|
{ |
|
66
|
3 |
|
$wn = 0; |
|
67
|
3 |
|
foreach ($this->text as $text) { |
|
68
|
3 |
|
$wn = $wn + $text->getWordNumber(); |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
3 |
|
return $wn; |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
3 |
|
protected function getTrails() |
|
75
|
|
|
{ |
|
76
|
3 |
|
$trails = []; |
|
77
|
|
|
|
|
78
|
3 |
|
foreach ($this->text as $text) { |
|
79
|
3 |
|
$trails = array_merge($trails, $text->getTrails()); |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
3 |
|
return $trails; |
|
83
|
|
|
} |
|
84
|
|
|
} |
|
85
|
|
|
|
PiedWeb /
TextAnalyzer
