|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* AnalyzerText package. |
|
4
|
|
|
* |
|
5
|
|
|
* @author Peter Gribanov <[email protected]> |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
namespace AnalyzerText\Filter; |
|
9
|
|
|
|
|
10
|
|
|
use AnalyzerText\Filter\WordList\Adverb; |
|
11
|
|
|
use AnalyzerText\Filter\WordList\Interjection; |
|
12
|
|
|
use AnalyzerText\Filter\WordList\Particle; |
|
13
|
|
|
use AnalyzerText\Filter\WordList\Preposition; |
|
14
|
|
|
use AnalyzerText\Filter\WordList\Pronoun; |
|
15
|
|
|
use AnalyzerText\Filter\WordList\Union; |
|
16
|
|
|
use AnalyzerText\Filter\WordList\WordList; |
|
17
|
|
|
use AnalyzerText\Text; |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* Оставляет только информационные слова. |
|
21
|
|
|
* |
|
22
|
|
|
* Сушествительные, глаголы, прилагательные, имена, фамилии |
|
23
|
|
|
* |
|
24
|
|
|
* @author Peter Gribanov <[email protected]> |
|
25
|
|
|
*/ |
|
26
|
|
|
class Informative extends Filter |
|
27
|
|
|
{ |
|
28
|
|
|
/** |
|
29
|
|
|
* Список фильтров. |
|
30
|
|
|
* |
|
31
|
|
|
* @var WordList[] |
|
32
|
|
|
*/ |
|
33
|
|
|
private $filters = array(); |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* @param Text $iterator Текст |
|
37
|
|
|
*/ |
|
38
|
8 |
|
public function __construct(Text $iterator) |
|
39
|
|
|
{ |
|
40
|
8 |
|
parent::__construct($iterator); |
|
41
|
8 |
|
$this->filters = array( |
|
42
|
8 |
|
new Interjection($this->getInnerIterator()), |
|
43
|
8 |
|
new Particle($this->getInnerIterator()), |
|
44
|
8 |
|
new Preposition($this->getInnerIterator()), |
|
45
|
8 |
|
new Pronoun($this->getInnerIterator()), |
|
46
|
8 |
|
new Union($this->getInnerIterator()), |
|
47
|
8 |
|
new Adverb($this->getInnerIterator()), |
|
48
|
|
|
); |
|
49
|
8 |
|
} |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* @see \FilterIterator::accept() |
|
53
|
|
|
*/ |
|
54
|
8 |
|
public function accept() |
|
55
|
|
|
{ |
|
56
|
8 |
|
$word = $this->current(); |
|
57
|
|
|
// сначала ищем последовательности |
|
58
|
8 |
|
foreach ($this->filters as $filter) { |
|
59
|
8 |
|
if ($filter->isSequence($word)) { |
|
60
|
3 |
|
return false; |
|
61
|
|
|
} |
|
62
|
|
|
} |
|
63
|
|
|
// ищем простые и сложные формы |
|
64
|
7 |
|
foreach ($this->filters as $filter) { |
|
65
|
7 |
|
if ($filter->isSimple($word) || $filter->isComposite($word)) { |
|
66
|
7 |
|
return false; |
|
67
|
|
|
} |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
6 |
|
return true; |
|
71
|
|
|
} |
|
72
|
|
|
} |
|
73
|
|
|
|
peter-gribanov /
AnalyzerText
