1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace Remorhaz\UniLex\Tool\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 PhpParser\PrettyPrinterAbstract; |
14
|
|
|
use ReflectionClass; |
15
|
|
|
use Remorhaz\UniLex\RegExp\FSM\Range; |
16
|
|
|
use Remorhaz\UniLex\RegExp\FSM\RangeSet; |
17
|
|
|
use Remorhaz\UniLex\RegExp\FSM\RangeSetCalc; |
18
|
|
|
use SplFileObject; |
19
|
|
|
use Throwable; |
20
|
|
|
|
21
|
|
|
use function array_diff; |
22
|
|
|
use function array_keys; |
23
|
|
|
use function array_merge; |
24
|
|
|
use function count; |
25
|
|
|
use function var_export; |
26
|
|
|
|
27
|
|
|
final class PropertyBuilder |
28
|
|
|
{ |
29
|
|
|
|
30
|
|
|
private $rangeSetCalc; |
31
|
|
|
|
32
|
|
|
private $rangeSets = []; |
33
|
|
|
|
34
|
|
|
private $scripts = []; |
35
|
|
|
|
36
|
|
|
private $phpBuilder; |
37
|
|
|
|
38
|
|
|
private $printer; |
39
|
|
|
|
40
|
|
|
/** |
41
|
|
|
* @var Range[][] |
42
|
|
|
*/ |
43
|
|
|
private $rangeBuffer = []; |
44
|
|
|
|
45
|
|
|
public function __construct(RangeSetCalc $rangeSetCalc, PrettyPrinterAbstract $printer) |
46
|
|
|
{ |
47
|
|
|
$this->rangeSetCalc = $rangeSetCalc; |
48
|
|
|
$this->phpBuilder = new BuilderFactory(); |
49
|
|
|
$this->printer = $printer; |
50
|
|
|
} |
51
|
|
|
|
52
|
|
|
public function parseUnicodeData(SplFileObject $file, callable $onProgress): void |
53
|
|
|
{ |
54
|
|
|
foreach (new UnicodeDataRangeIterator($file, $onProgress) as $prop => $range) { |
55
|
|
|
$this->addRangeToBuffer($prop, $range); |
56
|
|
|
} |
57
|
|
|
} |
58
|
|
|
|
59
|
|
|
public function parseScripts(SplFileObject $file, callable $onProgress): void |
60
|
|
|
{ |
61
|
|
|
$otherProperties = array_keys($this->rangeBuffer); |
62
|
|
|
$this->parseProperties($file, $onProgress); |
63
|
|
|
$this->scripts = array_diff(array_keys($this->rangeBuffer), $otherProperties); |
64
|
|
|
} |
65
|
|
|
|
66
|
|
|
public function parseProperties(SplFileObject $file, callable $onProgress): void |
67
|
|
|
{ |
68
|
|
|
foreach (new PropertiesRangeIterator($file, $onProgress) as $prop => $range) { |
69
|
|
|
$this->addRangeToBuffer($prop, $range); |
70
|
|
|
} |
71
|
|
|
} |
72
|
|
|
|
73
|
|
|
private function addRangeToBuffer(string $prop, Range ...$ranges): void |
74
|
|
|
{ |
75
|
|
|
$this->rangeBuffer[$prop] = array_merge($this->rangeBuffer[$prop] ?? [], $ranges); |
76
|
|
|
} |
77
|
|
|
|
78
|
|
|
public function getRangeBufferSize(): int |
79
|
|
|
{ |
80
|
|
|
return count($this->rangeBuffer); |
81
|
|
|
} |
82
|
|
|
|
83
|
|
|
public function getFileCount(): int |
84
|
|
|
{ |
85
|
|
|
return count($this->rangeSets) + 1; |
86
|
|
|
} |
87
|
|
|
|
88
|
|
|
public function fetchBufferedRangeSets(callable $onProgress): void |
89
|
|
|
{ |
90
|
|
|
$count = 0; |
91
|
|
|
foreach ($this->rangeBuffer as $prop => $ranges) { |
92
|
|
|
$this->addRangeSet($prop, ...$ranges); |
93
|
|
|
$onProgress(++$count); |
94
|
|
|
} |
95
|
|
|
$this->rangeBuffer = []; |
96
|
|
|
} |
97
|
|
|
|
98
|
|
|
public function buildUnicodeDataDerivatives(callable $onProgress): void |
99
|
|
|
{ |
100
|
|
|
$map = [ |
101
|
|
|
'C' => ['Cc', 'Cf', 'Co', 'Cs'], |
102
|
|
|
'L' => ['Ll', 'Lm', 'Lo', 'Lt', 'Lu'], |
103
|
|
|
'L&' => ['Lu', 'Ll', 'Lt'], |
104
|
|
|
'M' => ['Mc', 'Me', 'Mn'], |
105
|
|
|
'N' => ['Nd', 'Nl', 'No'], |
106
|
|
|
'P' => ['Pc', 'Pd', 'Pe', 'Pf', 'Pi', 'Po', 'Ps'], |
107
|
|
|
'S' => ['Sc', 'Sk', 'Sm', 'So'], |
108
|
|
|
'Z' => ['Zl', 'Zp', 'Zs'], |
109
|
|
|
]; |
110
|
|
|
|
111
|
|
|
$notCnRanges = []; |
112
|
|
|
foreach ($map as $targetProp => $sourceProps) { |
113
|
|
|
foreach ($sourceProps as $prop) { |
114
|
|
|
$rangeSet = $this->getRangeSet($prop); |
115
|
|
|
$notCnRanges = array_merge($notCnRanges, $rangeSet->getRanges()); |
116
|
|
|
$this->addRangeToBuffer($targetProp, ...$rangeSet->getRanges()); |
117
|
|
|
} |
118
|
|
|
} |
119
|
|
|
|
120
|
|
|
try { |
121
|
|
|
$targetProp = 'Any'; |
122
|
|
|
$anyRangeSet = RangeSet::import([0x00, 0x10FFFF]); |
123
|
|
|
} catch (Throwable $e) { |
124
|
|
|
throw new Exception\RangeSetNotBuiltException($targetProp, $e); |
125
|
|
|
} |
126
|
|
|
$onProgress(); |
127
|
|
|
$this->addRangeToBuffer($targetProp, ...$anyRangeSet->getRanges()); |
128
|
|
|
|
129
|
|
|
try { |
130
|
|
|
$targetProp = 'Cn'; |
131
|
|
|
$notCnRangeSet = new RangeSet(...$notCnRanges); |
132
|
|
|
$onProgress(); |
133
|
|
|
$cnRanges = $this |
134
|
|
|
->rangeSetCalc |
135
|
|
|
->xor($notCnRangeSet, $anyRangeSet) |
136
|
|
|
->getRanges(); |
137
|
|
|
} catch (Throwable $e) { |
138
|
|
|
throw new Exception\RangeSetNotBuiltException($targetProp, $e); |
139
|
|
|
} |
140
|
|
|
$onProgress(); |
141
|
|
|
$this->addRangeToBuffer($targetProp, ...$cnRanges); |
142
|
|
|
} |
143
|
|
|
|
144
|
|
|
public function buildScriptsDerivatives(callable $onProgress): void |
145
|
|
|
{ |
146
|
|
|
$knownRanges = []; |
147
|
|
|
foreach ($this->scripts as $prop) { |
148
|
|
|
$knownRanges = array_merge($knownRanges, $this->getRangeSet($prop)->getRanges()); |
149
|
|
|
$onProgress(); |
150
|
|
|
} |
151
|
|
|
try { |
152
|
|
|
$targetProp = 'Unknown'; |
153
|
|
|
$knownRangeSet = new RangeSet(...$knownRanges); |
154
|
|
|
$onProgress(); |
155
|
|
|
$unknownRanges = $this |
156
|
|
|
->rangeSetCalc |
157
|
|
|
->xor($knownRangeSet, $this->getRangeSet('Any')) |
158
|
|
|
->getRanges(); |
159
|
|
|
} catch (Throwable $e) { |
160
|
|
|
throw new Exception\RangeSetNotBuiltException($targetProp, $e); |
161
|
|
|
} |
162
|
|
|
$onProgress(); |
163
|
|
|
$this->addRangeToBuffer($targetProp, ...$unknownRanges); |
164
|
|
|
} |
165
|
|
|
|
166
|
|
|
private function addRangeSet(string $prop, Range ...$ranges): void |
167
|
|
|
{ |
168
|
|
|
if (isset($this->rangeSets[$prop])) { |
169
|
|
|
throw new Exception\RangeSetAlreadyExistsException($prop); |
170
|
|
|
} |
171
|
|
|
try { |
172
|
|
|
$this->rangeSets[$prop] = new RangeSet(...$ranges); |
173
|
|
|
} catch (Throwable $e) { |
174
|
|
|
throw new Exception\RangeSetNotBuiltException($prop, $e); |
175
|
|
|
} |
176
|
|
|
} |
177
|
|
|
|
178
|
|
|
private function getRangeSet(string $prop): RangeSet |
179
|
|
|
{ |
180
|
|
|
if (isset($this->rangeSets[$prop])) { |
181
|
|
|
return $this->rangeSets[$prop]; |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
throw new Exception\RangeSetNotFoundException($prop); |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
public function writeFiles(string $targetRootDir, callable $onProgress): void |
188
|
|
|
{ |
189
|
|
|
$fileIndex = []; |
190
|
|
|
foreach ($this->rangeSets as $prop => $rangeSet) { |
191
|
|
|
$baseName = "/Properties/{$prop}.php"; |
192
|
|
|
$fileName = $targetRootDir . $baseName; |
193
|
|
|
try { |
194
|
|
|
$code = $this->buildPropertyFile($rangeSet); |
195
|
|
|
\Safe\file_put_contents($fileName, $code); |
196
|
|
|
} catch (Throwable $e) { |
197
|
|
|
throw new Exception\FileNotWrittenException($fileName); |
198
|
|
|
} |
199
|
|
|
$fileIndex[$prop] = $baseName; |
200
|
|
|
$onProgress(); |
201
|
|
|
} |
202
|
|
|
$fileName = $targetRootDir . "/PropertyIndex.php"; |
203
|
|
|
try { |
204
|
|
|
$code = $this->buildIndexFile($fileIndex); |
205
|
|
|
\Safe\file_put_contents($fileName, $code); |
206
|
|
|
} catch (Throwable $e) { |
207
|
|
|
throw new Exception\FileNotWrittenException($fileName); |
208
|
|
|
} |
209
|
|
|
$onProgress(); |
210
|
|
|
} |
211
|
|
|
|
212
|
|
|
private function buildIndexFile(array $index): string |
213
|
|
|
{ |
214
|
|
|
$array = var_export($index, true); |
215
|
|
|
|
216
|
|
|
return "<?php\n\nreturn {$array};\n"; |
217
|
|
|
} |
218
|
|
|
|
219
|
|
|
private function buildPropertyFile(RangeSet $rangeSet): string |
220
|
|
|
{ |
221
|
|
|
$rangeClass = new ReflectionClass(Range::class); |
222
|
|
|
$rangeSetClass = new ReflectionClass(RangeSet::class); |
223
|
|
|
|
224
|
|
|
$phpNodes = []; |
225
|
|
|
$declare = new Declare_([new DeclareDeclare('strict_types', $this->phpBuilder->val(1))]); |
226
|
|
|
$declare->setDocComment(new Doc('/** @noinspection PhpUnhandledExceptionInspection */')); |
227
|
|
|
$phpNodes[] = $declare; |
228
|
|
|
$phpNodes[] = $this->phpBuilder->namespace(__NAMESPACE__ . '\\Properties')->getNode(); |
229
|
|
|
$phpNodes[] = $this->phpBuilder->use($rangeClass->getName())->getNode(); |
230
|
|
|
$phpNodes[] = $this->phpBuilder->use($rangeSetClass->getName())->getNode(); |
231
|
|
|
$phpRanges = []; |
232
|
|
|
|
233
|
|
|
foreach ($rangeSet->getRanges() as $range) { |
234
|
|
|
$rangeStart = $range->getStart(); |
235
|
|
|
$rangeFinish = $range->getFinish(); |
236
|
|
|
$phpRangeStart = $this->phpBuilder->val($rangeStart); |
237
|
|
|
$phpRangeStart->setAttribute('kind', LNumber::KIND_HEX); |
238
|
|
|
$phpRangeArgs = [$phpRangeStart]; |
239
|
|
|
if ($rangeStart != $rangeFinish) { |
240
|
|
|
$phpRangeFinish = $this->phpBuilder->val($rangeFinish); |
241
|
|
|
$phpRangeFinish->setAttribute('kind', LNumber::KIND_HEX); |
242
|
|
|
$phpRangeArgs[] = $phpRangeFinish; |
243
|
|
|
} |
244
|
|
|
$phpRanges[] = $this->phpBuilder->new($rangeClass->getShortName(), $phpRangeArgs); |
245
|
|
|
} |
246
|
|
|
$phpReturn = new Return_( |
247
|
|
|
$this->phpBuilder->staticCall($rangeSetClass->getShortName(), 'loadUnsafe', $phpRanges) |
248
|
|
|
); |
249
|
|
|
$phpReturn->setDocComment(new Doc('/** phpcs:disable Generic.Files.LineLength.TooLong */')); |
250
|
|
|
$phpNodes[] = $phpReturn; |
251
|
|
|
|
252
|
|
|
return $this->printer->prettyPrintFile($phpNodes); |
253
|
|
|
} |
254
|
|
|
} |
255
|
|
|
|