|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace NirjharLo\Cgss\Lib\Analysis\Lib; |
|
4
|
|
|
|
|
5
|
|
|
if ( ! defined( 'ABSPATH' ) ) exit; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* An object to extract most used pharases in an array of phrases, containing one or more words. |
|
9
|
|
|
* But, for now we are using one word elements only. |
|
10
|
|
|
* |
|
11
|
|
|
* 2 properties. |
|
12
|
|
|
* @property string $words Input words array |
|
13
|
|
|
* @property string $text Input words array |
|
14
|
|
|
*/ |
|
15
|
|
|
class Keywords { |
|
16
|
|
|
|
|
17
|
|
|
|
|
18
|
|
|
public $words; |
|
19
|
|
|
public $text; |
|
20
|
|
|
|
|
21
|
|
|
|
|
22
|
|
|
//generate individual words from text content |
|
23
|
|
|
public function output() { |
|
24
|
|
|
|
|
25
|
|
|
$one = $this->output_one(); |
|
26
|
|
|
$two = $this->fetch(2); |
|
27
|
|
|
$three = $this->fetch(3); |
|
28
|
|
|
$four = $this->fetch(4); |
|
29
|
|
|
$five = $this->fetch(5); |
|
30
|
|
|
$six = $this->fetch(6); |
|
31
|
|
|
$keys = array_merge( $six, $five, $four, $three, $two, $one ); |
|
32
|
|
|
return $keys; |
|
33
|
|
|
} |
|
34
|
|
|
|
|
35
|
|
|
//generate individual words from text content |
|
36
|
|
|
public function output_one() { |
|
37
|
|
|
|
|
38
|
|
|
$one_arr = $this->fetch(1); |
|
39
|
|
|
$stop = $this->stop(); |
|
40
|
|
|
$output = array(); |
|
41
|
|
|
foreach ( $one_arr as $val => $key ) { |
|
42
|
|
|
if ( strlen( $val ) > 3 and ! in_array( $val, $stop ) ) { |
|
43
|
|
|
$output[$val] = $key; |
|
44
|
|
|
} |
|
45
|
|
|
} |
|
46
|
|
|
return array_slice( $output, 0, 10 ); |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
//generate individual words from text content |
|
50
|
|
|
public function fetch( $n ) { |
|
51
|
|
|
|
|
52
|
|
|
//compare text and words to find keyword |
|
53
|
|
|
$keys = array(); |
|
54
|
|
|
$phrs = $this->prepare( $n ); |
|
55
|
|
|
$phrs_count = count($phrs); |
|
56
|
|
|
foreach( $phrs as $val ) { |
|
57
|
|
|
$key_num = substr_count( $this->text, $val ); |
|
58
|
|
|
$key_per = round( ( ( ( $key_num * $n ) / $phrs_count ) * 100 ), 3 ); |
|
59
|
|
|
if ( $key_num > 1 and $key_per > 0.2 and ! isset( $keys[$val] ) ) { |
|
60
|
|
|
$keys[$val] = $key_per; |
|
61
|
|
|
} |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
//arrange them to select top final 3 keyword |
|
65
|
|
|
if ( is_array( $keys ) ) { |
|
66
|
|
|
arsort( $keys ); |
|
67
|
|
|
if ( $n != 1 ) { |
|
68
|
|
|
|
|
69
|
|
|
//remove keys without much significance |
|
70
|
|
|
$del_keys = array(); |
|
71
|
|
|
foreach ( $keys as $val => $key ) { |
|
72
|
|
|
$key_size = strlen( implode( '', explode( ' ', $val ) ) ) / 2; |
|
73
|
|
|
if ( $key_size <= $n ) { |
|
74
|
|
|
$del_keys[] = $val; |
|
75
|
|
|
} |
|
76
|
|
|
} |
|
77
|
|
|
$new_keys = array_diff( $keys, $del_keys ); |
|
78
|
|
|
|
|
79
|
|
|
$key_top = array_slice( $new_keys, 0, 3 ); |
|
80
|
|
|
} else { |
|
81
|
|
|
$key_top = $keys; |
|
82
|
|
|
} |
|
83
|
|
|
} |
|
84
|
|
|
return $key_top; |
|
|
|
|
|
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
//generate individual words from text content |
|
88
|
|
|
public function prepare( $n ) { |
|
89
|
|
|
$i = 0; |
|
|
|
|
|
|
90
|
|
|
$part = array(); |
|
91
|
|
|
$w_input = $this->words; |
|
92
|
|
|
for( $i = 0; $i < $n; $i++ ) { |
|
93
|
|
|
|
|
94
|
|
|
//create an array of words of pharses as sub-arrays |
|
95
|
|
|
$pieces = array_chunk( $w_input, $n ); |
|
|
|
|
|
|
96
|
|
|
$last_one = array_pop( $pieces ); |
|
97
|
|
|
if ( is_array( $last_one ) and count( $last_one ) != $i ) { |
|
98
|
|
|
unset($pieces[count($pieces) - 1]); |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
//create phrase out of those sub array |
|
102
|
|
|
foreach( $pieces as $val ) { |
|
103
|
|
|
$part[] = implode( ' ', $val ); |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
//Offset input words first value to create more unique words |
|
107
|
|
|
$del_w_input = array_diff( $w_input, array( $w_input[0] ) ); |
|
|
|
|
|
|
108
|
|
|
$w_input = array_values($del_w_input); |
|
109
|
|
|
} |
|
110
|
|
|
return $part; |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
//array of stop words |
|
114
|
|
|
public function stop() { |
|
115
|
|
|
return array( 'from', 'that', 'this', 'what', 'when', 'where', 'have', 'they', 'just', 'your', 'most', 'their', 'some', 'then', 'there', 'them', 'make', 'ever', 'never', 'enough', 'should', 'would', 'could', 'also', 'such', 'shall', 'will', 'with' ); |
|
116
|
|
|
} |
|
117
|
|
|
} |
|
118
|
|
|
?> |
|
|
|
|
|
|
119
|
|
|
|