|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace Remorhaz\UniLex\Lexer; |
|
6
|
|
|
|
|
7
|
|
|
use phpDocumentor\Reflection\DocBlock; |
|
8
|
|
|
use phpDocumentor\Reflection\DocBlockFactory; |
|
9
|
|
|
use phpDocumentor\Reflection\DocBlockFactoryInterface; |
|
10
|
|
|
use ReflectionException; |
|
11
|
|
|
use Remorhaz\UniLex\Exception; |
|
12
|
|
|
|
|
13
|
|
|
class TokenMatcherSpecParser |
|
14
|
|
|
{ |
|
15
|
|
|
private const TAG_LEX_TARGET_CLASS = 'lexTargetClass'; |
|
16
|
|
|
private const TAG_LEX_TEMPLATE_CLASS = 'lexTemplateClass'; |
|
17
|
|
|
private const TAG_LEX_HEADER = 'lexHeader'; |
|
18
|
|
|
private const TAG_LEX_BEFORE_MATCH = 'lexBeforeMatch'; |
|
19
|
|
|
private const TAG_LEX_ON_TRANSITION = 'lexOnTransition'; |
|
20
|
|
|
private const TAG_LEX_ON_ERROR = 'lexOnError'; |
|
21
|
|
|
private const TAG_LEX_TOKEN = 'lexToken'; |
|
22
|
|
|
private const TAG_LEX_MODE = 'lexMode'; |
|
23
|
|
|
private const LEX_NAMESPACE = 'namespace'; |
|
24
|
|
|
private const LEX_USE = 'use'; |
|
25
|
|
|
private const LEX_TOKEN_REGEXP = 'token_regexp'; |
|
26
|
|
|
private const LEX_TOKEN_CONTEXT = 'token_context'; |
|
27
|
|
|
|
|
28
|
|
|
private $source; |
|
29
|
|
|
|
|
30
|
|
|
private $matcherSpec; |
|
31
|
|
|
|
|
32
|
|
|
private $targetClassName; |
|
33
|
|
|
|
|
34
|
|
|
private $templateClassName; |
|
35
|
|
|
|
|
36
|
|
|
private $docBlockFactory; |
|
37
|
|
|
|
|
38
|
|
|
private $codeBlockList = []; |
|
39
|
|
|
|
|
40
|
|
|
private $codeBlockKey; |
|
41
|
|
|
|
|
42
|
|
|
private $codeBlockStack = []; |
|
43
|
|
|
|
|
44
|
|
|
private $skipToken = false; |
|
45
|
|
|
|
|
46
|
|
|
private $usedClassList = []; |
|
47
|
|
|
|
|
48
|
|
|
private $tokenSpecList = []; |
|
49
|
|
|
|
|
50
|
|
|
public function __construct(string $source) |
|
51
|
|
|
{ |
|
52
|
|
|
$this->source = $source; |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* @param string $fileName |
|
57
|
|
|
* @return TokenMatcherSpecParser |
|
58
|
|
|
* @throws Exception |
|
59
|
|
|
*/ |
|
60
|
|
|
public static function loadFromFile(string $fileName): self |
|
61
|
|
|
{ |
|
62
|
|
|
$source = @file_get_contents($fileName); |
|
63
|
|
|
if (false === $source) { |
|
64
|
|
|
throw new Exception("Failed to read lexer specification from file {$fileName}"); |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
return new self($source); |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
/** |
|
71
|
|
|
* @return TokenMatcherSpec |
|
72
|
|
|
* @throws Exception |
|
73
|
|
|
* @throws ReflectionException |
|
74
|
|
|
*/ |
|
75
|
|
|
public function getMatcherSpec(): TokenMatcherSpec |
|
76
|
|
|
{ |
|
77
|
|
|
if (!isset($this->matcherSpec)) { |
|
78
|
|
|
$this->matcherSpec = $this->buildMatcherSpec(); |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
|
|
return $this->matcherSpec; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* @return TokenMatcherSpec |
|
86
|
|
|
* @throws Exception |
|
87
|
|
|
* @throws ReflectionException |
|
88
|
|
|
*/ |
|
89
|
|
|
private function buildMatcherSpec(): TokenMatcherSpec |
|
90
|
|
|
{ |
|
91
|
|
|
$phpTokenList = token_get_all($this->source); |
|
92
|
|
|
foreach ($phpTokenList as $phpToken) { |
|
93
|
|
|
$argList = is_array($phpToken) ? $phpToken : [null, $phpToken]; |
|
94
|
|
|
$this->processPhpToken(...$argList); |
|
|
|
|
|
|
95
|
|
|
} |
|
96
|
|
|
$this->afterProcessingSource(); |
|
97
|
|
|
if (!isset($this->targetClassName)) { |
|
98
|
|
|
throw new Exception("Invalid lexer specification: target class is not defined"); |
|
99
|
|
|
} |
|
100
|
|
|
$targetClassName = $this->getCodeBlock(self::LEX_NAMESPACE) . $this->targetClassName; |
|
101
|
|
|
$templateClassName = $this->templateClassName ?? TokenMatcherTemplate::class; |
|
102
|
|
|
$matcherSpec = new TokenMatcherSpec($targetClassName, $templateClassName); |
|
103
|
|
|
$matcherSpec |
|
104
|
|
|
->setHeader($this->getCodeBlock(self::TAG_LEX_HEADER)) |
|
105
|
|
|
->setBeforeMatch($this->getCodeBlock(self::TAG_LEX_BEFORE_MATCH)) |
|
106
|
|
|
->setOnTransition($this->getCodeBlock(self::TAG_LEX_ON_TRANSITION)) |
|
107
|
|
|
->setOnError($this->getCodeBlock(self::TAG_LEX_ON_ERROR, "return false;")); |
|
108
|
|
|
foreach ($this->tokenSpecList as $mode => $tokenSpecList) { |
|
109
|
|
|
$matcherSpec->addTokenSpec($mode, ...array_values($tokenSpecList)); |
|
110
|
|
|
} |
|
111
|
|
|
foreach ($this->usedClassList as $usedClassAlias => $usedClassName) { |
|
112
|
|
|
$matcherSpec->addUsedClass($usedClassName, is_string($usedClassAlias) ? $usedClassAlias : null); |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
return $matcherSpec; |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
/** |
|
119
|
|
|
* @param int|null $tokenId |
|
120
|
|
|
* @param string $code |
|
121
|
|
|
* @throws Exception |
|
122
|
|
|
*/ |
|
123
|
|
|
private function processPhpToken(?int $tokenId, string $code): void |
|
124
|
|
|
{ |
|
125
|
|
|
$this->skipToken = false; |
|
126
|
|
|
if ($this->isCurrentCodeBlock(self::TAG_LEX_HEADER)) { |
|
127
|
|
|
if (T_NAMESPACE === $tokenId && !$this->codeBlockExists(self::LEX_NAMESPACE)) { |
|
128
|
|
|
$this->replaceCurrentCodeBlock(self::LEX_NAMESPACE); |
|
129
|
|
|
$this->skipToken = true; |
|
130
|
|
|
} |
|
131
|
|
|
if (T_USE === $tokenId) { |
|
132
|
|
|
$this->replaceCurrentCodeBlock(self::LEX_USE); |
|
133
|
|
|
$this->skipToken = true; |
|
134
|
|
|
} |
|
135
|
|
|
} |
|
136
|
|
|
if ($this->isCurrentCodeBlock(self::LEX_NAMESPACE) && null === $tokenId && ';' == $code) { |
|
137
|
|
|
$this->appendCodeBlock("\\"); |
|
138
|
|
|
$this->restoreCurrentCodeBlock(); |
|
139
|
|
|
$this->skipToken = true; |
|
140
|
|
|
} |
|
141
|
|
|
if ($this->isCurrentCodeBlock(self::LEX_USE) && null === $tokenId && ';' == $code) { |
|
142
|
|
|
[$usedClass, $alias] = $this->parseUsedClass($this->getCodeBlock(self::LEX_USE)); |
|
143
|
|
|
if (isset($alias)) { |
|
144
|
|
|
$this->usedClassList[$alias] = $usedClass; |
|
145
|
|
|
} else { |
|
146
|
|
|
$this->usedClassList[] = $usedClass; |
|
147
|
|
|
} |
|
148
|
|
|
$this->resetCodeBlock(); |
|
149
|
|
|
$this->restoreCurrentCodeBlock(); |
|
150
|
|
|
$this->skipToken = true; |
|
151
|
|
|
} |
|
152
|
|
|
if (T_DOC_COMMENT === $tokenId) { |
|
153
|
|
|
$docBlock = $this->getDocBlockFactory()->create($code); |
|
154
|
|
|
$this->detectTargetClassName($docBlock); |
|
155
|
|
|
$this->detectTemplateClassName($docBlock); |
|
156
|
|
|
$this->detectCodeBlock($docBlock); |
|
157
|
|
|
$this->detectTokenRegExp($docBlock); |
|
158
|
|
|
} |
|
159
|
|
|
if (!$this->skipToken) { |
|
160
|
|
|
$this->appendCodeBlock($code); |
|
161
|
|
|
} |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
/** |
|
165
|
|
|
* @throws Exception |
|
166
|
|
|
*/ |
|
167
|
|
|
private function afterProcessingSource(): void |
|
168
|
|
|
{ |
|
169
|
|
|
$this->detectUnprocessedTokenBlock(); |
|
170
|
|
|
} |
|
171
|
|
|
|
|
172
|
|
|
/** |
|
173
|
|
|
* @param string $usedClass |
|
174
|
|
|
* @return array |
|
175
|
|
|
* @throws Exception |
|
176
|
|
|
*/ |
|
177
|
|
|
private function parseUsedClass(string $usedClass): array |
|
178
|
|
|
{ |
|
179
|
|
|
/** @noinspection HtmlUnknownTag */ |
|
180
|
|
|
$pattern = '#^(?P<className>\S+)(?:\s+(?:as\s+)?(?P<alias>\S+))?$#i'; |
|
181
|
|
|
$pregResult = preg_match($pattern, $usedClass, $matches); |
|
182
|
|
|
if (1 !== $pregResult) { |
|
183
|
|
|
throw new Exception("Invalid lexer specification: invalid used class expression"); |
|
184
|
|
|
} |
|
185
|
|
|
|
|
186
|
|
|
return [$matches['className'], $matches['alias'] ?? null]; |
|
187
|
|
|
} |
|
188
|
|
|
|
|
189
|
|
|
/** |
|
190
|
|
|
* @param DocBlock $docBlock |
|
191
|
|
|
* @throws Exception |
|
192
|
|
|
*/ |
|
193
|
|
|
private function detectTargetClassName(DocBlock $docBlock): void |
|
194
|
|
|
{ |
|
195
|
|
|
$tagName = self::TAG_LEX_TARGET_CLASS; |
|
196
|
|
|
if (!$docBlock->hasTag($tagName)) { |
|
197
|
|
|
return; |
|
198
|
|
|
} |
|
199
|
|
|
$this->skipToken = true; |
|
200
|
|
|
$tagList = $docBlock->getTagsByName($tagName); |
|
201
|
|
|
if (isset($this->targetClassName) || count($tagList) != 1) { |
|
202
|
|
|
throw new Exception("Invalid lexer specification: duplicated @{$tagName} tag"); |
|
203
|
|
|
} |
|
204
|
|
|
$this->targetClassName = (string) array_pop($tagList); |
|
205
|
|
|
} |
|
206
|
|
|
|
|
207
|
|
|
/** |
|
208
|
|
|
* @param DocBlock $docBlock |
|
209
|
|
|
* @throws Exception |
|
210
|
|
|
*/ |
|
211
|
|
|
private function detectTemplateClassName(DocBlock $docBlock): void |
|
212
|
|
|
{ |
|
213
|
|
|
$tagName = self::TAG_LEX_TEMPLATE_CLASS; |
|
214
|
|
|
if (!$docBlock->hasTag($tagName)) { |
|
215
|
|
|
return; |
|
216
|
|
|
} |
|
217
|
|
|
$this->skipToken = true; |
|
218
|
|
|
$tagList = $docBlock->getTagsByName($tagName); |
|
219
|
|
|
if (isset($this->templateClassName) || count($tagList) != 1) { |
|
220
|
|
|
throw new Exception("Invalid lexer specification: duplicated @{$tagName} tag"); |
|
221
|
|
|
} |
|
222
|
|
|
$this->templateClassName = (string) array_pop($tagList); |
|
223
|
|
|
} |
|
224
|
|
|
|
|
225
|
|
|
/** |
|
226
|
|
|
* @param DocBlock $docBlock |
|
227
|
|
|
* @throws Exception |
|
228
|
|
|
*/ |
|
229
|
|
|
private function detectCodeBlock(DocBlock $docBlock): void |
|
230
|
|
|
{ |
|
231
|
|
|
$codeBlockKeyList = [ |
|
232
|
|
|
self::TAG_LEX_HEADER, |
|
233
|
|
|
self::TAG_LEX_BEFORE_MATCH, |
|
234
|
|
|
self::TAG_LEX_ON_TRANSITION, |
|
235
|
|
|
self::TAG_LEX_ON_ERROR, |
|
236
|
|
|
self::TAG_LEX_TOKEN, |
|
237
|
|
|
]; |
|
238
|
|
|
$codeBlockKey = null; |
|
239
|
|
|
foreach ($codeBlockKeyList as $tagName) { |
|
240
|
|
|
if ($docBlock->hasTag($tagName)) { |
|
241
|
|
|
if (isset($codeBlockKey)) { |
|
242
|
|
|
throw new Exception("Invalid lexer specification: @{$tagName} tag conflicts with @{$codeBlockKey}"); |
|
243
|
|
|
} |
|
244
|
|
|
$codeBlockKey = $tagName; |
|
245
|
|
|
} |
|
246
|
|
|
} |
|
247
|
|
|
if (isset($codeBlockKey)) { |
|
248
|
|
|
$this->detectUnprocessedTokenBlock(); |
|
249
|
|
|
if ($this->codeBlockExists($codeBlockKey)) { |
|
250
|
|
|
throw new Exception("Invalid lexer specification: duplicated @{$codeBlockKey} tag"); |
|
251
|
|
|
} |
|
252
|
|
|
$this->skipToken = true; |
|
253
|
|
|
$this->switchCurrentCodeBlock($codeBlockKey); |
|
254
|
|
|
} |
|
255
|
|
|
} |
|
256
|
|
|
|
|
257
|
|
|
/** |
|
258
|
|
|
* @throws Exception |
|
259
|
|
|
*/ |
|
260
|
|
|
private function detectUnprocessedTokenBlock(): void |
|
261
|
|
|
{ |
|
262
|
|
|
$codeBlockKey = self::TAG_LEX_TOKEN; |
|
263
|
|
|
if (!$this->codeBlockExists($codeBlockKey)) { |
|
264
|
|
|
return; |
|
265
|
|
|
} |
|
266
|
|
|
$context = $this->getCodeBlock(self::LEX_TOKEN_CONTEXT); |
|
267
|
|
|
$this->resetCodeBlock(self::LEX_TOKEN_CONTEXT); |
|
268
|
|
|
$tokenRegExp = $this->getCodeBlock(self::LEX_TOKEN_REGEXP); |
|
269
|
|
|
if (isset($this->tokenSpecList[$context][$tokenRegExp])) { |
|
270
|
|
|
throw new Exception("Invalid lexer specification: duplicated @{$codeBlockKey} /{$tokenRegExp}/ tag"); |
|
271
|
|
|
} |
|
272
|
|
|
$this->resetCodeBlock(self::LEX_TOKEN_REGEXP); |
|
273
|
|
|
$tokenCode = $this->getCodeBlock($codeBlockKey); |
|
274
|
|
|
$this->resetCodeBlock($codeBlockKey); |
|
275
|
|
|
$tokenSpec = new TokenSpec($tokenRegExp, $tokenCode); |
|
276
|
|
|
$this->tokenSpecList[$context][$tokenSpec->getRegExp()] = $tokenSpec; |
|
277
|
|
|
} |
|
278
|
|
|
|
|
279
|
|
|
/** |
|
280
|
|
|
* @param DocBlock $docBlock |
|
281
|
|
|
* @throws Exception |
|
282
|
|
|
*/ |
|
283
|
|
|
private function detectTokenRegExp(DocBlock $docBlock): void |
|
284
|
|
|
{ |
|
285
|
|
|
if (!$this->isCurrentCodeBlock(self::TAG_LEX_TOKEN)) { |
|
286
|
|
|
return; |
|
287
|
|
|
} |
|
288
|
|
|
$tagValue = (string) $docBlock->getTagsByName(self::TAG_LEX_TOKEN)[0]; |
|
289
|
|
|
$matchResult = preg_match('#^/(?P<regexp>.*)/$#', $tagValue, $matches); |
|
290
|
|
|
if (1 !== $matchResult) { |
|
291
|
|
|
throw new Exception("Invalid lexer specification: regular expression is not framed by \"/\""); |
|
292
|
|
|
} |
|
293
|
|
|
$regExp = $matches['regexp']; |
|
294
|
|
|
$this->replaceCurrentCodeBlock(self::LEX_TOKEN_REGEXP); |
|
295
|
|
|
$this->appendCodeBlock($regExp); |
|
296
|
|
|
$this->restoreCurrentCodeBlock(); |
|
297
|
|
|
$context = $docBlock->hasTag(self::TAG_LEX_MODE) |
|
298
|
|
|
? (string) $docBlock->getTagsByName(self::TAG_LEX_MODE)[0] |
|
299
|
|
|
: TokenMatcherInterface::DEFAULT_MODE; |
|
300
|
|
|
$matchResult = preg_match('#^[a-zA-Z][0-9a-zA-Z]*$#i', $context); |
|
301
|
|
|
if (1 !== $matchResult) { |
|
302
|
|
|
throw new Exception("Invalid lexer specification: invalid context name {$context}"); |
|
303
|
|
|
} |
|
304
|
|
|
$this->replaceCurrentCodeBlock(self::LEX_TOKEN_CONTEXT); |
|
305
|
|
|
$this->appendCodeBlock($context); |
|
306
|
|
|
$this->restoreCurrentCodeBlock(); |
|
307
|
|
|
} |
|
308
|
|
|
|
|
309
|
|
|
private function getDocBlockFactory(): DocBlockFactoryInterface |
|
310
|
|
|
{ |
|
311
|
|
|
if (!isset($this->docBlockFactory)) { |
|
312
|
|
|
$this->docBlockFactory = DocBlockFactory::createInstance(); |
|
313
|
|
|
} |
|
314
|
|
|
|
|
315
|
|
|
return $this->docBlockFactory; |
|
316
|
|
|
} |
|
317
|
|
|
|
|
318
|
|
|
private function switchCurrentCodeBlock(string $key): void |
|
319
|
|
|
{ |
|
320
|
|
|
$this->codeBlockKey = $key; |
|
321
|
|
|
$this->codeBlockList[$this->codeBlockKey] = ''; |
|
322
|
|
|
} |
|
323
|
|
|
|
|
324
|
|
|
private function replaceCurrentCodeBlock(string $key): void |
|
325
|
|
|
{ |
|
326
|
|
|
$this->codeBlockStack[] = $this->codeBlockKey; |
|
327
|
|
|
$this->switchCurrentCodeBlock($key); |
|
328
|
|
|
} |
|
329
|
|
|
|
|
330
|
|
|
private function restoreCurrentCodeBlock(): void |
|
331
|
|
|
{ |
|
332
|
|
|
$this->codeBlockKey = array_pop($this->codeBlockStack); |
|
333
|
|
|
} |
|
334
|
|
|
|
|
335
|
|
|
private function codeBlockExists(string $key): bool |
|
336
|
|
|
{ |
|
337
|
|
|
return isset($this->codeBlockList[$key]); |
|
338
|
|
|
} |
|
339
|
|
|
|
|
340
|
|
|
private function getCodeBlock(string $key = null, string $defaultValue = ''): string |
|
341
|
|
|
{ |
|
342
|
|
|
$effectiveKey = $key ?? $this->codeBlockKey; |
|
343
|
|
|
|
|
344
|
|
|
return isset($effectiveKey) |
|
345
|
|
|
? trim($this->codeBlockList[$effectiveKey] ?? $defaultValue) |
|
346
|
|
|
: trim($defaultValue); |
|
347
|
|
|
} |
|
348
|
|
|
|
|
349
|
|
|
private function resetCodeBlock(string $key = null): void |
|
350
|
|
|
{ |
|
351
|
|
|
$effectiveKey = $key ?? $this->codeBlockKey; |
|
352
|
|
|
if (isset($effectiveKey)) { |
|
353
|
|
|
unset($this->codeBlockList[$key]); |
|
354
|
|
|
} |
|
355
|
|
|
} |
|
356
|
|
|
|
|
357
|
|
|
private function appendCodeBlock(string $code, string $key = null): void |
|
358
|
|
|
{ |
|
359
|
|
|
$effectiveKey = $key ?? $this->codeBlockKey; |
|
360
|
|
|
if (isset($effectiveKey)) { |
|
361
|
|
|
$this->codeBlockList[$effectiveKey] .= $code; |
|
362
|
|
|
} |
|
363
|
|
|
} |
|
364
|
|
|
|
|
365
|
|
|
private function isCurrentCodeBlock(string $key): bool |
|
366
|
|
|
{ |
|
367
|
|
|
return $key === $this->codeBlockKey; |
|
368
|
|
|
} |
|
369
|
|
|
} |
|
370
|
|
|
|
remorhaz /
php-unilex
This check compares calls to functions or methods with their respective definitions. If the call has less arguments than are defined, it raises an issue.
If a function is defined several times with a different number of parameters, the check may pick up the wrong definition and report false positives. One codebase where this has been known to happen is Wordpress. Please note the @ignore annotation hint above.