1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace Remorhaz\UniLex\RegExp; |
6
|
|
|
|
7
|
|
|
use PhpParser\BuilderFactory; |
8
|
|
|
use PhpParser\Comment\Doc; |
9
|
|
|
use PhpParser\Node\Scalar\LNumber; |
10
|
|
|
use PhpParser\Node\Stmt\Declare_; |
11
|
|
|
use PhpParser\Node\Stmt\DeclareDeclare; |
12
|
|
|
use PhpParser\Node\Stmt\Return_; |
13
|
|
|
use ReflectionClass; |
14
|
|
|
use ReflectionException; |
15
|
|
|
use Remorhaz\UniLex\Console\PrettyPrinter; |
16
|
|
|
use Remorhaz\UniLex\Exception as UniLexException; |
17
|
|
|
use Remorhaz\UniLex\RegExp\FSM\Range; |
18
|
|
|
use Remorhaz\UniLex\RegExp\FSM\RangeSet; |
19
|
|
|
use RuntimeException; |
20
|
|
|
use SplFileObject; |
21
|
|
|
|
22
|
|
|
use function explode; |
23
|
|
|
use function file_put_contents; |
24
|
|
|
use function hexdec; |
25
|
|
|
use function trim; |
26
|
|
|
|
27
|
|
|
final class PropertyBuilder |
28
|
|
|
{ |
29
|
|
|
|
30
|
|
|
private const PROP_DIR = '/Properties'; |
31
|
|
|
|
32
|
|
|
private $phpBuilder; |
33
|
|
|
|
34
|
|
|
private $printer; |
35
|
|
|
|
36
|
|
|
public function __construct() |
37
|
|
|
{ |
38
|
|
|
$this->phpBuilder = new BuilderFactory(); |
39
|
|
|
$this->printer = new PrettyPrinter(); |
40
|
|
|
} |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* @param array $index |
44
|
|
|
* @return array |
45
|
|
|
* @throws UniLexException |
46
|
|
|
* @throws ReflectionException |
47
|
|
|
*/ |
48
|
|
|
public function buildScripts(array $index): array |
49
|
|
|
{ |
50
|
|
|
/** @var RangeSet[] $ranges */ |
51
|
|
|
$ranges = []; |
52
|
|
|
$source = new SplFileObject(__DIR__ . '/../../data/Scripts.txt'); |
53
|
|
|
$lastKnownCode = null; |
54
|
|
|
$unknownRanges = []; |
55
|
|
|
|
56
|
|
|
echo "Parsing: "; |
57
|
|
|
$rangeCount = 0; |
58
|
|
|
|
59
|
|
|
while (!$source->eof()) { |
60
|
|
|
$line = $source->fgets(); |
61
|
|
|
if (false === $line) { |
62
|
|
|
throw new RuntimeException("Error reading line from scripts file"); |
63
|
|
|
} |
64
|
|
|
$dataWithComment = explode('#', $line, 2); |
65
|
|
|
$data = trim($dataWithComment[0] ?? ''); |
66
|
|
|
if ('' == $data) { |
67
|
|
|
continue; |
68
|
|
|
} |
69
|
|
|
$rangeWithProp = explode(';', $data); |
70
|
|
|
$unsplittedRange = trim($rangeWithProp[0] ?? null); |
71
|
|
|
$prop = trim($rangeWithProp[1] ?? null); |
72
|
|
|
if (!isset($unsplittedRange, $prop)) { |
73
|
|
|
throw new RuntimeException("Invalid range or property"); |
74
|
|
|
} |
75
|
|
|
$splittedRange = explode('..', $unsplittedRange); |
76
|
|
|
$start = hexdec($splittedRange[0]); |
77
|
|
|
$finish = isset($splittedRange[1]) |
78
|
|
|
? hexdec($splittedRange[1]) |
79
|
|
|
: $start; |
80
|
|
|
if (!isset($lastKnownCode)) { |
81
|
|
|
if ($start > 0) { |
82
|
|
|
$unknownRanges[] = new Range(0, $start - 1); |
|
|
|
|
83
|
|
|
} |
84
|
|
|
} elseif ($start - $lastKnownCode > 1) { |
85
|
|
|
$unknownRanges[] = new Range($lastKnownCode + 1, $start - 1); |
86
|
|
|
} |
87
|
|
|
$lastKnownCode = $finish; |
88
|
|
|
|
89
|
|
|
if (!isset($ranges[$prop])) { |
90
|
|
|
$ranges[$prop] = []; |
91
|
|
|
} |
92
|
|
|
$range = new Range($start, $finish); |
|
|
|
|
93
|
|
|
$ranges[$prop][] = $range; |
94
|
|
|
echo "."; |
95
|
|
|
$rangeCount++; |
96
|
|
|
} |
97
|
|
|
$ranges['Unknown'] = $unknownRanges; |
98
|
|
|
echo ". {$rangeCount} ranges\n"; |
99
|
|
|
$source = null; |
|
|
|
|
100
|
|
|
|
101
|
|
|
echo "Building range sets: "; |
102
|
|
|
$rangeSetCount = 0; |
103
|
|
|
$rangeSets = []; |
104
|
|
|
foreach ($ranges as $prop => $rangeList) { |
105
|
|
|
$rangeSets[$prop] = new RangeSet(...$rangeList); |
106
|
|
|
echo "."; |
107
|
|
|
$rangeSetCount++; |
108
|
|
|
} |
109
|
|
|
echo " {$rangeSetCount} range sets\n"; |
110
|
|
|
|
111
|
|
|
/*echo "Build Unknown property..."; |
112
|
|
|
$fullRangeSet = new RangeSet(new Range(0x0, 0x10FFFF)); |
113
|
|
|
$knownRangeSet = new RangeSet(...$knownRanges); |
114
|
|
|
$unknownRangeSet = $rangeSetCalc->xor($fullRangeSet, $knownRangeSet); |
115
|
|
|
$ranges['Unknown'] = $unknownRangeSet;*/ |
116
|
|
|
|
117
|
|
|
return $this->dumpProps($index, $rangeSets); |
118
|
|
|
} |
119
|
|
|
|
120
|
|
|
/** |
121
|
|
|
* @param array $index |
122
|
|
|
* @param array $rangeSets |
123
|
|
|
* @return array |
124
|
|
|
* @throws ReflectionException |
125
|
|
|
*/ |
126
|
|
|
private function dumpProps(array $index, array $rangeSets): array |
127
|
|
|
{ |
128
|
|
|
$rangeSetClass = new ReflectionClass(RangeSet::class); |
129
|
|
|
$rangeClass = new ReflectionClass(Range::class); |
130
|
|
|
foreach ($rangeSets as $prop => $rangeSet) { |
131
|
|
|
$targetFile = self::PROP_DIR . "/{$prop}.php"; |
132
|
|
|
|
133
|
|
|
$phpNodes = []; |
134
|
|
|
$declare = new Declare_([new DeclareDeclare('strict_types', $this->phpBuilder->val(1))]); |
135
|
|
|
$declare->setDocComment(new Doc('/** @noinspection PhpUnhandledExceptionInspection */')); |
136
|
|
|
$phpNodes[] = $declare; |
137
|
|
|
$phpNodes[] = $this->phpBuilder->namespace(__NAMESPACE__ . '\\Properties')->getNode(); |
138
|
|
|
$phpNodes[] = $this->phpBuilder->use($rangeClass->getName())->getNode(); |
139
|
|
|
$phpNodes[] = $this->phpBuilder->use($rangeSetClass->getName())->getNode(); |
140
|
|
|
$phpRanges = []; |
141
|
|
|
|
142
|
|
|
foreach ($rangeSet->getRanges() as $range) { |
143
|
|
|
$rangeStart = $range->getStart(); |
144
|
|
|
$rangeFinish = $range->getFinish(); |
145
|
|
|
$phpRangeStart = $this->phpBuilder->val($rangeStart); |
146
|
|
|
$phpRangeStart->setAttribute('kind', LNumber::KIND_HEX); |
147
|
|
|
$phpRangeArgs = [$phpRangeStart]; |
148
|
|
|
if ($rangeStart != $rangeFinish) { |
149
|
|
|
$phpRangeFinish = $this->phpBuilder->val($rangeFinish); |
150
|
|
|
$phpRangeFinish->setAttribute('kind', LNumber::KIND_HEX); |
151
|
|
|
$phpRangeArgs[] = $phpRangeFinish; |
152
|
|
|
} |
153
|
|
|
$phpRanges[] = $this->phpBuilder->new($rangeClass->getShortName(), $phpRangeArgs); |
154
|
|
|
} |
155
|
|
|
$phpReturn = new Return_( |
156
|
|
|
$this->phpBuilder->staticCall($rangeSetClass->getShortName(), 'loadUnsafe', $phpRanges) |
157
|
|
|
); |
158
|
|
|
$phpReturn->setDocComment(new Doc('/** phpcs:disable Generic.Files.LineLength.TooLong */')); |
159
|
|
|
$phpNodes[] = $phpReturn; |
160
|
|
|
file_put_contents(__DIR__ . $targetFile, $this->printer->prettyPrintFile($phpNodes)); |
161
|
|
|
$index[$prop] = $targetFile; |
162
|
|
|
} |
163
|
|
|
|
164
|
|
|
return $index; |
165
|
|
|
} |
166
|
|
|
|
167
|
|
|
public function dumpIndex(array $index): void |
168
|
|
|
{ |
169
|
|
|
$indexCode = "<?php\n\nreturn " . var_export($index, true) . ";\n"; |
170
|
|
|
file_put_contents(__DIR__ . '/PropertyIndex.php', $indexCode); |
171
|
|
|
} |
172
|
|
|
} |
173
|
|
|
|