|
1
|
|
|
<?php |
|
2
|
|
|
namespace Mystem; |
|
3
|
|
|
|
|
4
|
|
|
/** |
|
5
|
|
|
* Class Mystem |
|
6
|
|
|
* Helper for execute mystem |
|
7
|
|
|
*/ |
|
8
|
|
|
class Mystem |
|
9
|
|
|
{ |
|
10
|
|
|
private static $handle; |
|
11
|
|
|
protected static $pipes; |
|
12
|
|
|
|
|
13
|
|
|
/* @var string $mystemPath path to mystem binary */ |
|
14
|
|
|
public static $mystemPath = null; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* Runs mystem binary and returns raw morphological data for each word |
|
18
|
|
|
* Ex. for 'хрюкотали' returns: |
|
19
|
|
|
* array(2) { |
|
20
|
|
|
* ["text"]=> string(18) "хрюкотали" |
|
21
|
|
|
* ["analysis"]=> array(1) { |
|
22
|
|
|
* [0]=> array(3) { |
|
23
|
|
|
* ["lex"] =>string(18) "хрюкотать" |
|
24
|
|
|
* ["gr"] =>string(42) "V,несов,нп=прош,мн,изъяв" |
|
25
|
|
|
* ["qual"]=>string(7) "bastard" |
|
26
|
|
|
* } |
|
27
|
|
|
* } |
|
28
|
|
|
* } |
|
29
|
|
|
* @param string $text |
|
30
|
|
|
* @throws \Exception |
|
31
|
|
|
* @return array[] lexical strings associative array |
|
32
|
|
|
*/ |
|
33
|
40 |
|
public static function stemm($text) |
|
34
|
|
|
{ |
|
35
|
40 |
|
self::procOpen(); |
|
36
|
|
|
do { |
|
37
|
40 |
|
$endMark = 'end' . rand(99999, PHP_INT_MAX); |
|
38
|
40 |
|
} while (mb_strpos($text, $endMark) !== false); |
|
39
|
40 |
|
fwrite(self::$pipes[0], $text . ".$endMark\n"); |
|
40
|
40 |
|
$raw = self::readUntil(self::$pipes[1], $endMark); |
|
41
|
40 |
|
$possibleError = stream_get_contents(self::$pipes[2], 1024); |
|
42
|
40 |
|
if (!empty($possibleError)) { |
|
43
|
1 |
|
throw new \Exception("Error: ".$possibleError); |
|
44
|
|
|
} |
|
45
|
39 |
|
$lines = explode("\n", $raw); |
|
46
|
39 |
|
foreach ($lines as &$line) { |
|
47
|
39 |
|
$line = json_decode($line, true); |
|
48
|
|
|
} |
|
49
|
39 |
|
$lines = array_filter($lines, function ($value) { |
|
50
|
39 |
|
return !empty($value['analysis']); |
|
51
|
39 |
|
}); |
|
52
|
39 |
|
return $lines; |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* @param $pipe |
|
57
|
|
|
* @param string $endMark |
|
58
|
|
|
* @return string |
|
59
|
|
|
*/ |
|
60
|
40 |
|
private static function readUntil($pipe, $endMark) |
|
61
|
|
|
{ |
|
62
|
40 |
|
$w = null; |
|
63
|
40 |
|
$read = array($pipe); |
|
64
|
40 |
|
if (stream_select($read, $w, $e, 4, 1000) == 0) { |
|
|
|
|
|
|
65
|
|
|
return ''; |
|
66
|
|
|
} |
|
67
|
40 |
|
$raw = ''; |
|
68
|
40 |
|
$newOffset = 0; |
|
69
|
40 |
|
$counter = 0; |
|
70
|
|
|
do { |
|
71
|
40 |
|
$offset = $newOffset; |
|
72
|
40 |
|
usleep(500); |
|
73
|
40 |
|
$raw .= stream_get_contents($pipe); |
|
74
|
40 |
|
$newOffset = mb_strlen($raw); |
|
75
|
40 |
|
} while (mb_strpos($raw, $endMark, $offset) == false && $counter++<20); |
|
76
|
40 |
|
return $raw; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
40 |
|
private static function procOpen() |
|
80
|
|
|
{ |
|
81
|
40 |
|
if (self::$handle !== null) { |
|
82
|
7 |
|
return array(); |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
40 |
|
self::$handle = proc_open(self::getMystem() . ' -incs --format=json', array( |
|
86
|
40 |
|
0 => array("pipe", "r"), 1 => array("pipe", "w"), 2 => array("pipe", "w") |
|
87
|
40 |
|
), self::$pipes); |
|
88
|
|
|
|
|
89
|
40 |
|
if (!is_resource(self::$handle)) { |
|
90
|
|
|
throw new \Exception("Can't proc_open mystem"); |
|
91
|
|
|
} |
|
92
|
40 |
|
stream_set_blocking(self::$pipes[1], 0); |
|
93
|
40 |
|
stream_set_blocking(self::$pipes[2], 0); |
|
94
|
|
|
|
|
95
|
40 |
|
register_shutdown_function(array('\Mystem\Mystem', 'destruct')); |
|
96
|
40 |
|
} |
|
97
|
|
|
|
|
98
|
|
|
public static function destruct() |
|
99
|
|
|
{ |
|
100
|
|
|
if (self::$handle === null) { |
|
101
|
|
|
return false; |
|
102
|
|
|
} |
|
103
|
|
|
if (is_array(self::$pipes)) { |
|
104
|
|
|
foreach (self::$pipes as $pipe) { |
|
105
|
|
|
fflush($pipe); |
|
106
|
|
|
fclose($pipe); |
|
107
|
|
|
} |
|
108
|
|
|
} |
|
109
|
|
|
proc_terminate(self::$handle); |
|
110
|
|
|
proc_close(self::$handle); |
|
111
|
|
|
self::$handle = null; |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
/** |
|
115
|
|
|
* Returns mystem executable depends bit depth of operating system and OS type |
|
116
|
|
|
* @return string |
|
117
|
|
|
*/ |
|
118
|
40 |
|
private static function getMystem() |
|
119
|
|
|
{ |
|
120
|
40 |
|
if (self::$mystemPath === null) { |
|
|
|
|
|
|
121
|
39 |
|
if (is_dir(__DIR__ . '/../../vendor/bin/')) { |
|
122
|
39 |
|
self::$mystemPath = __DIR__ . '/../../vendor/bin/'; |
|
123
|
|
|
} else { |
|
124
|
|
|
self::$mystemPath = __DIR__ . '/../../../../bin/'; |
|
125
|
|
|
} |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
40 |
|
return self::$mystemPath . ( |
|
129
|
40 |
|
strtoupper(substr(PHP_OS, 0, 3)) === 'WIN' ? 'mystem.exe' : 'mystem' |
|
130
|
|
|
); |
|
131
|
|
|
} |
|
132
|
|
|
} |
|
133
|
|
|
|
aotd1 /
mystem
Loading history...