|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace spec\drupol\phpngrams; |
|
4
|
|
|
|
|
5
|
|
|
use drupol\phpngrams\NGrams; |
|
6
|
|
|
use PhpSpec\ObjectBehavior; |
|
7
|
|
|
|
|
8
|
|
|
class NGramsSpec extends ObjectBehavior |
|
9
|
|
|
{ |
|
10
|
|
|
public function it_is_initializable() |
|
11
|
|
|
{ |
|
12
|
|
|
$this->shouldHaveType(NGrams::class); |
|
13
|
|
|
} |
|
14
|
|
|
|
|
15
|
|
|
public function it_can_get_ngram_from_a_string() |
|
16
|
|
|
{ |
|
17
|
|
|
$result = [ |
|
18
|
|
|
'he', |
|
19
|
|
|
'el', |
|
20
|
|
|
'll', |
|
21
|
|
|
'lo', |
|
22
|
|
|
]; |
|
23
|
|
|
|
|
24
|
|
|
$this->ngrams('hello', 2)->shouldIterateAs(new \ArrayIterator($result)); |
|
25
|
|
|
} |
|
26
|
|
|
|
|
27
|
|
|
public function it_can_get_ngram_from_a_string_with_big_n() |
|
28
|
|
|
{ |
|
29
|
|
|
$result = [ |
|
30
|
|
|
'hello', |
|
31
|
|
|
]; |
|
32
|
|
|
|
|
33
|
|
|
$this->ngrams('hello', 10)->shouldIterateAs(new \ArrayIterator($result)); |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
public function it_can_get_ngram_from_an_array() |
|
37
|
|
|
{ |
|
38
|
|
|
$result = [ |
|
39
|
|
|
['h', 'e'], |
|
40
|
|
|
['e', 'l'], |
|
41
|
|
|
['l', 'l'], |
|
42
|
|
|
['l', 'o'], |
|
43
|
|
|
]; |
|
44
|
|
|
|
|
45
|
|
|
$this->ngrams(['h', 'e', 'l', 'l', 'o'], 2)->shouldIterateAs(new \ArrayIterator($result)); |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
public function it_can_get_ngram_from_an_array_without_cycling() |
|
49
|
|
|
{ |
|
50
|
|
|
$result = [ |
|
51
|
|
|
['h', 'e', 'l'], |
|
52
|
|
|
['e', 'l', 'l'], |
|
53
|
|
|
['l', 'l', 'o'], |
|
54
|
|
|
['l', 'o', ' '], |
|
55
|
|
|
['o', ' ', 'w'], |
|
56
|
|
|
[' ', 'w', 'o'], |
|
57
|
|
|
['w', 'o', 'r'], |
|
58
|
|
|
['o', 'r', 'l'], |
|
59
|
|
|
['r', 'l', 'd'], |
|
60
|
|
|
]; |
|
61
|
|
|
|
|
62
|
|
|
$this->ngrams(['h', 'e', 'l', 'l', 'o', ' ', 'w', 'o', 'r', 'l', 'd'], 3)->shouldIterateAs(new \ArrayIterator($result)); |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
public function it_can_get_ngram_from_an_array_with_big_n() |
|
66
|
|
|
{ |
|
67
|
|
|
$result = [ |
|
68
|
|
|
['h', 'e', 'l', 'l', 'o'], |
|
69
|
|
|
]; |
|
70
|
|
|
|
|
71
|
|
|
$this->ngrams(['h', 'e', 'l', 'l', 'o'], 10)->shouldIterateAs(new \ArrayIterator($result)); |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
public function it_can_calculate_the_frequency() |
|
75
|
|
|
{ |
|
76
|
|
|
$input= 'Hold my beer'; |
|
77
|
|
|
$ngrams = $this->getWrappedObject()->ngrams($input, 2); |
|
78
|
|
|
$this->frequency($ngrams, 'my')->shouldBe(1/11); |
|
79
|
|
|
|
|
80
|
|
|
$input= ['h', 'e', 'l', 'l', 'o']; |
|
81
|
|
|
$ngrams = $this->getWrappedObject()->ngrams($input, 2); |
|
82
|
|
|
$this->frequency($ngrams, ['l', 'l'])->shouldBe(1/4); |
|
83
|
|
|
} |
|
84
|
|
|
} |
|
85
|
|
|
|
drupol /
phpngrams
