|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace Remorhaz\UniLex\Tool\RegExp; |
|
6
|
|
|
|
|
7
|
|
|
use Iterator; |
|
8
|
|
|
use IteratorAggregate; |
|
9
|
|
|
use Remorhaz\UniLex\RegExp\FSM\Range; |
|
10
|
|
|
use SplFileObject; |
|
11
|
|
|
use Throwable; |
|
12
|
|
|
|
|
13
|
|
|
use function strlen; |
|
14
|
|
|
|
|
15
|
|
|
final class UnicodeDataRangeIterator implements IteratorAggregate |
|
16
|
|
|
{ |
|
17
|
|
|
|
|
18
|
|
|
private $file; |
|
19
|
|
|
|
|
20
|
|
|
private $onProgress; |
|
21
|
|
|
|
|
22
|
|
|
private $code; |
|
23
|
|
|
|
|
24
|
|
|
private $name; |
|
25
|
|
|
|
|
26
|
|
|
private $prop; |
|
27
|
|
|
|
|
28
|
|
|
private $lastCode; |
|
29
|
|
|
|
|
30
|
|
|
private $lastProp; |
|
31
|
|
|
|
|
32
|
|
|
private $rangeStart; |
|
33
|
|
|
|
|
34
|
|
|
private $namedStarts = []; |
|
35
|
|
|
|
|
36
|
|
|
public function __construct(SplFileObject $file, callable $onProgress) |
|
37
|
|
|
{ |
|
38
|
|
|
$this->file = $file; |
|
39
|
|
|
$this->onProgress = $onProgress; |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
|
|
public function getIterator(): Iterator |
|
43
|
|
|
{ |
|
44
|
|
|
while (!$this->file->eof()) { |
|
45
|
|
|
$line = $this->fetchNextLine($this->file); |
|
46
|
|
|
if (!isset($line)) { |
|
47
|
|
|
continue; |
|
48
|
|
|
} |
|
49
|
|
|
$range = $this->fetchUnicodeDataRange($line); |
|
50
|
|
|
if (isset($range)) { |
|
51
|
|
|
yield $this->lastProp => $range; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
$this->lastCode = $this->code; |
|
55
|
|
|
$this->lastProp = $this->prop; |
|
56
|
|
|
|
|
57
|
|
|
($this->onProgress)(strlen($line)); |
|
58
|
|
|
} |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
private function fetchNextLine(SplFileObject $file): ?string |
|
62
|
|
|
{ |
|
63
|
|
|
$line = $file->fgets(); |
|
64
|
|
|
if (false === $line) { |
|
65
|
|
|
throw new Exception\LineNotReadException($file->getFilename()); |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
return '' == $line ? null : $line; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
private function parseUnicodeDataLineLine(string $line): void |
|
72
|
|
|
{ |
|
73
|
|
|
$splitLine = explode(';', $line); |
|
74
|
|
|
$codeHex = $splitLine[0] ?? null; |
|
75
|
|
|
$name = $splitLine[1] ?? null; |
|
76
|
|
|
$prop = $splitLine[2] ?? null; |
|
77
|
|
|
if (!isset($codeHex, $name, $prop)) { |
|
78
|
|
|
throw new Exception\InvalidLineException($line); |
|
79
|
|
|
} |
|
80
|
|
|
$this->code = hexdec($codeHex); |
|
81
|
|
|
$this->name = $name; |
|
82
|
|
|
$this->prop = $prop; |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
private function fetchUnicodeDataRange(string $line): ?Range |
|
86
|
|
|
{ |
|
87
|
|
|
$this->parseUnicodeDataLineLine($line); |
|
88
|
|
|
|
|
89
|
|
|
[$firstName, $lastName] = $this->parseRangeBoundary($this->name); |
|
90
|
|
|
if (isset($firstName)) { |
|
91
|
|
|
$this->namedStarts[$firstName] = $this->code; |
|
92
|
|
|
$this->rangeStart = null; |
|
93
|
|
|
|
|
94
|
|
|
return null; |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
if (isset($lastName)) { |
|
98
|
|
|
if ( |
|
99
|
|
|
!isset($this->namedStarts[$lastName]) || |
|
100
|
|
|
isset($this->rangeStart) || |
|
101
|
|
|
$this->lastCode !== $this->namedStarts[$lastName] |
|
102
|
|
|
) { |
|
103
|
|
|
throw new Exception\InvalidLineException($line); |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
return $this->createRange($this->lastCode, $this->code); |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
if ($this->prop === $this->lastProp && $this->code - 1 === $this->lastCode) { |
|
110
|
|
|
return null; |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
$range = isset($this->rangeStart, $this->lastCode) |
|
114
|
|
|
? $this->createRange($this->rangeStart, $this->lastCode) |
|
115
|
|
|
: null; |
|
116
|
|
|
|
|
117
|
|
|
$this->rangeStart = $this->code; |
|
118
|
|
|
|
|
119
|
|
|
return $range; |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
private function parseRangeBoundary(string $name): array |
|
123
|
|
|
{ |
|
124
|
|
|
try { |
|
125
|
|
|
$isFirst = 1 === \Safe\preg_match('#^<(.+), First>$#', $name, $matches); |
|
126
|
|
|
if ($isFirst) { |
|
127
|
|
|
return [$matches[1] ?? null, null]; |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
$isLast = 1 === \Safe\preg_match('#^<(.+), Last>$#', $name, $matches); |
|
131
|
|
|
|
|
132
|
|
|
return $isLast |
|
133
|
|
|
? [null, $matches[1] ?? null] |
|
134
|
|
|
: [null, null]; |
|
135
|
|
|
} catch (Throwable $e) { |
|
136
|
|
|
throw new Exception\CodePointNameNotParsedException($name, $e); |
|
137
|
|
|
} |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
private function createRange(int $start, ?int $finish): Range |
|
141
|
|
|
{ |
|
142
|
|
|
try { |
|
143
|
|
|
return new Range($start, $finish); |
|
144
|
|
|
} catch (Throwable $e) { |
|
145
|
|
|
throw new Exception\RangeNotCreatedException($e); |
|
146
|
|
|
} |
|
147
|
|
|
} |
|
148
|
|
|
} |
|
149
|
|
|
|
remorhaz /
php-unilex
