1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Digia\GraphQL\Language; |
4
|
|
|
|
5
|
|
|
use Digia\GraphQL\Error\SyntaxErrorException; |
6
|
|
|
|
7
|
|
|
class TokenReader implements TokenReaderInterface |
8
|
|
|
{ |
9
|
|
|
|
10
|
|
|
/** |
11
|
|
|
* The lexer owning this token reader. |
12
|
|
|
* |
13
|
|
|
* @var LexerInterface |
14
|
|
|
*/ |
15
|
|
|
protected $lexer; |
16
|
|
|
|
17
|
|
|
/** |
18
|
|
|
* @inheritdoc |
19
|
|
|
*/ |
20
|
|
|
public function setLexer(LexerInterface $lexer) |
21
|
|
|
{ |
22
|
|
|
$this->lexer = $lexer; |
23
|
|
|
return $this; |
24
|
|
|
} |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* @inheritdoc |
28
|
|
|
*/ |
29
|
|
|
public function getLexer(): LexerInterface |
30
|
|
|
{ |
31
|
|
|
return $this->lexer; |
32
|
|
|
} |
33
|
|
|
|
34
|
|
|
/** |
35
|
|
|
* @inheritdoc |
36
|
|
|
* @throws SyntaxErrorException |
37
|
|
|
*/ |
38
|
|
|
public function read(string $body, int $bodyLength, int $code, int $pos, int $line, int $col, Token $prev): ?Token |
39
|
|
|
{ |
40
|
|
|
switch ($code) { |
41
|
|
|
case 33: // ! |
42
|
|
|
return $this->readBang($pos, $line, $col, $prev); |
43
|
|
|
case 35: // # |
44
|
|
|
return $this->readComment($pos, $line, $col, $prev); |
45
|
|
|
case 36: // $ |
46
|
|
|
return $this->readDollar($pos, $line, $col, $prev); |
47
|
|
|
case 38: // & |
48
|
|
|
return $this->readAmp($pos, $line, $col, $prev); |
49
|
|
|
case 58: // : |
50
|
|
|
return $this->readColon($pos, $line, $col, $prev); |
51
|
|
|
case 61: // = |
52
|
|
|
return $this->readEquals($pos, $line, $col, $prev); |
53
|
|
|
case 64: // @ |
54
|
|
|
return $this->readAt($pos, $line, $col, $prev); |
55
|
|
|
case 124: // | |
56
|
|
|
return $this->readPipe($pos, $line, $col, $prev); |
57
|
|
|
case 40: |
58
|
|
|
case 41: // ( or )~ |
59
|
|
|
return $this->readParenthesis($code, $pos, $line, $col, $prev); |
60
|
|
|
case 91: |
61
|
|
|
case 93: // [ or ] |
62
|
|
|
return $this->readBracket($code, $pos, $line, $col, $prev); |
63
|
|
|
case 123: |
64
|
|
|
case 125: // { or } |
65
|
|
|
return $this->readBrace($code, $pos, $line, $col, $prev); |
66
|
|
|
} |
67
|
|
|
|
68
|
|
|
// Int: -?(0|[1-9][0-9]*) |
69
|
|
|
// Float: -?(0|[1-9][0-9]*)(\.[0-9]+)?((E|e)(+|-)?[0-9]+)? |
70
|
|
|
if ($code === 45 || isNumber($code)) { |
71
|
|
|
return $this->readNumber($code, $pos, $line, $col, $prev); |
72
|
|
|
} |
73
|
|
|
|
74
|
|
|
// Name: [_A-Za-z][_0-9A-Za-z]* |
75
|
|
|
if (isAlphaNumeric($code)) { |
76
|
|
|
return $this->readName($body, $bodyLength, $pos, $line, $col, $prev); |
77
|
|
|
} |
78
|
|
|
|
79
|
|
|
// Spread: ... |
80
|
|
|
if ($bodyLength >= 3 && isSpread($body, $code, $pos)) { |
81
|
|
|
return $this->readSpread($pos, $line, $col, $prev); |
82
|
|
|
} |
83
|
|
|
|
84
|
|
|
// String: "([^"\\\u000A\u000D]|(\\(u[0-9a-fA-F]{4}|["\\/bfnrt])))*" |
85
|
|
|
if (isString($body, $code, $pos)) { |
86
|
|
|
return $this->readString($body, $bodyLength, $pos, $line, $col, $prev); |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
// Block String: """("?"?(\\"""|\\(?!=""")|[^"\\]))*""" |
90
|
|
|
if ($bodyLength >= 3 && isTripleQuote($body, $code, $pos)) { |
91
|
|
|
return $this->readBlockString($body, $bodyLength, $pos, $line, $col, $prev); |
92
|
|
|
} |
93
|
|
|
|
94
|
|
|
return null; |
95
|
|
|
} |
96
|
|
|
|
97
|
|
|
/** |
98
|
|
|
* @param string $body |
99
|
|
|
* @param int $bodyLength |
100
|
|
|
* @param int $pos |
101
|
|
|
* @param int $line |
102
|
|
|
* @param int $col |
103
|
|
|
* @param Token $prev |
104
|
|
|
* @return Token |
105
|
|
|
*/ |
106
|
|
|
protected function readName(string $body, int $bodyLength, int $pos, int $line, int $col, Token $prev): Token |
107
|
|
|
{ |
108
|
|
|
$start = $pos; |
109
|
|
|
$pos = $start + 1; |
110
|
|
|
|
111
|
|
|
while ($pos !== $bodyLength && ($code = charCodeAt($body, $pos)) !== null && isAlphaNumeric($code)) { |
112
|
|
|
++$pos; |
113
|
|
|
} |
114
|
|
|
|
115
|
|
|
return new Token(TokenKindEnum::NAME, $start, $pos, $line, $col, $prev, sliceString($body, $start, $pos)); |
116
|
|
|
} |
117
|
|
|
|
118
|
|
|
/** |
119
|
|
|
* @param string $body |
120
|
|
|
* @param int $bodyLength |
121
|
|
|
* @param int $pos |
122
|
|
|
* @param int $line |
123
|
|
|
* @param int $col |
124
|
|
|
* @param Token $prev |
125
|
|
|
* @return Token |
126
|
|
|
* @throws SyntaxErrorException |
127
|
|
|
*/ |
128
|
|
|
protected function readBlockString(string $body, int $bodyLength, int $pos, int $line, int $col, Token $prev): Token |
129
|
|
|
{ |
130
|
|
|
$start = $pos; |
131
|
|
|
$pos = $start + 3; |
132
|
|
|
$chunkStart = $pos; |
133
|
|
|
$rawValue = ''; |
134
|
|
|
|
135
|
|
|
while ($pos < $bodyLength && ($code = charCodeAt($body, $pos)) !== null) { |
136
|
|
|
// Closing Triple-Quote (""") |
137
|
|
|
if (isTripleQuote($body, $code, $pos)) { |
138
|
|
|
$rawValue .= sliceString($body, $chunkStart, $pos); |
139
|
|
|
return new Token( |
140
|
|
|
TokenKindEnum::BLOCK_STRING, |
141
|
|
|
$start, |
142
|
|
|
$pos + 3, |
143
|
|
|
$line, |
144
|
|
|
$col, |
145
|
|
|
$prev, |
146
|
|
|
blockStringValue($rawValue) |
147
|
|
|
); |
148
|
|
|
} |
149
|
|
|
|
150
|
|
|
if (isSourceCharacter($code) && !isLineTerminator($code)) { |
151
|
|
|
throw new SyntaxErrorException( |
152
|
|
|
$this->lexer->getSource(), |
153
|
|
|
$pos, |
154
|
|
|
\sprintf('Invalid character within String: %s.', printCharCode($code)) |
155
|
|
|
); |
156
|
|
|
} |
157
|
|
|
|
158
|
|
|
if (isEscapedTripleQuote($body, $code, $pos)) { |
159
|
|
|
$rawValue .= sliceString($body, $chunkStart, $pos) . '"""'; |
160
|
|
|
$pos += 4; |
161
|
|
|
$chunkStart = $pos; |
162
|
|
|
} else { |
163
|
|
|
++$pos; |
164
|
|
|
} |
165
|
|
|
} |
166
|
|
|
|
167
|
|
|
throw new SyntaxErrorException($this->lexer->getSource(), $pos, 'Unterminated string.'); |
168
|
|
|
} |
169
|
|
|
|
170
|
|
|
/** |
171
|
|
|
* @param int $code |
172
|
|
|
* @param int $pos |
173
|
|
|
* @param int $line |
174
|
|
|
* @param int $col |
175
|
|
|
* @param Token $prev |
176
|
|
|
* @return Token |
177
|
|
|
* @throws SyntaxErrorException |
178
|
|
|
*/ |
179
|
|
|
protected function readNumber(int $code, int $pos, int $line, int $col, Token $prev): Token |
180
|
|
|
{ |
181
|
|
|
$body = $this->lexer->getBody(); |
182
|
|
|
$start = $pos; |
183
|
|
|
$isFloat = false; |
184
|
|
|
|
185
|
|
|
if ($code === 45) { |
186
|
|
|
// - |
187
|
|
|
$code = charCodeAt($body, ++$pos); |
188
|
|
|
} |
189
|
|
|
|
190
|
|
|
if ($code === 48) { |
191
|
|
|
// 0 |
192
|
|
|
$code = charCodeAt($body, ++$pos); |
193
|
|
|
if (isNumber($code)) { |
194
|
|
|
throw new SyntaxErrorException( |
195
|
|
|
$this->lexer->getSource(), |
196
|
|
|
$pos, |
197
|
|
|
\sprintf('Invalid number, unexpected digit after 0: %s.', printCharCode($code)) |
198
|
|
|
); |
199
|
|
|
} |
200
|
|
|
} else { |
201
|
|
|
$pos = $this->readDigits($code, $pos); |
202
|
|
|
$code = charCodeAt($body, $pos); |
203
|
|
|
} |
204
|
|
|
|
205
|
|
|
if ($code === 46) { |
206
|
|
|
// . |
207
|
|
|
$isFloat = true; |
208
|
|
|
$code = charCodeAt($body, ++$pos); |
209
|
|
|
$pos = $this->readDigits($code, $pos); |
210
|
|
|
$code = charCodeAt($body, $pos); |
211
|
|
|
} |
212
|
|
|
|
213
|
|
|
if ($code === 69 || $code === 101) { |
214
|
|
|
// e or E |
215
|
|
|
$isFloat = true; |
216
|
|
|
$code = charCodeAt($body, ++$pos); |
217
|
|
|
|
218
|
|
|
if ($code === 43 || $code === 45) { |
219
|
|
|
// + or - |
220
|
|
|
$code = charCodeAt($body, ++$pos); |
221
|
|
|
} |
222
|
|
|
|
223
|
|
|
$pos = $this->readDigits($code, $pos); |
224
|
|
|
} |
225
|
|
|
|
226
|
|
|
return new Token( |
227
|
|
|
$isFloat ? TokenKindEnum::FLOAT : TokenKindEnum::INT, |
228
|
|
|
$start, |
229
|
|
|
$pos, |
230
|
|
|
$line, |
231
|
|
|
$col, |
232
|
|
|
$prev, |
233
|
|
|
sliceString($body, $start, $pos) |
234
|
|
|
); |
235
|
|
|
} |
236
|
|
|
|
237
|
|
|
/** |
238
|
|
|
* @param string $body |
239
|
|
|
* @param int $bodyLength |
240
|
|
|
* @param int $pos |
241
|
|
|
* @param int $line |
242
|
|
|
* @param int $col |
243
|
|
|
* @param Token $prev |
244
|
|
|
* @return Token |
245
|
|
|
* @throws SyntaxErrorException |
246
|
|
|
*/ |
247
|
|
|
protected function readString(string $body, int $bodyLength, int $pos, int $line, int $col, Token $prev): Token |
248
|
|
|
{ |
249
|
|
|
$start = $pos; |
250
|
|
|
$pos = $start + 1; |
251
|
|
|
$chunkStart = $pos; |
252
|
|
|
$value = ''; |
253
|
|
|
while ($pos < $bodyLength && ($code = charCodeAt($body, $pos)) !== null && !isLineTerminator($code)) { |
254
|
|
|
// Closing Quote (") |
255
|
|
|
if ($code === 34) { |
256
|
|
|
$value .= sliceString($body, $chunkStart, $pos); |
257
|
|
|
return new Token(TokenKindEnum::STRING, $start, $pos + 1, $line, $col, $prev, $value); |
258
|
|
|
} |
259
|
|
|
|
260
|
|
|
if (isSourceCharacter($code)) { |
261
|
|
|
throw new SyntaxErrorException( |
262
|
|
|
$this->lexer->getSource(), |
263
|
|
|
$pos, |
264
|
|
|
\sprintf('Invalid character within String: %s', printCharCode($code)) |
265
|
|
|
); |
266
|
|
|
} |
267
|
|
|
|
268
|
|
|
++$pos; |
269
|
|
|
|
270
|
|
|
if ($code === 92) { |
271
|
|
|
// \ |
272
|
|
|
$value .= sliceString($body, $chunkStart, $pos - 1); |
273
|
|
|
$code = charCodeAt($body, $pos); |
274
|
|
|
|
275
|
|
|
switch ($code) { |
276
|
|
|
case 34: |
277
|
|
|
$value .= '"'; |
278
|
|
|
break; |
279
|
|
|
case 47: |
280
|
|
|
$value .= '/'; |
281
|
|
|
break; |
282
|
|
|
case 92: |
283
|
|
|
$value .= '\\'; |
284
|
|
|
break; |
285
|
|
|
case 98: |
286
|
|
|
$value .= '\b'; |
287
|
|
|
break; |
288
|
|
|
case 102: |
289
|
|
|
$value .= '\f'; |
290
|
|
|
break; |
291
|
|
|
case 110: |
292
|
|
|
$value .= '\n'; |
293
|
|
|
break; |
294
|
|
|
case 114: |
295
|
|
|
$value .= '\r'; |
296
|
|
|
break; |
297
|
|
|
case 116: |
298
|
|
|
$value .= '\t'; |
299
|
|
|
break; |
300
|
|
|
case 117: |
301
|
|
|
// u |
302
|
|
|
$unicodeString = sliceString($body, $pos - 1, $pos + 5); |
303
|
|
|
$charCode = uniCharCode( |
304
|
|
|
charCodeAt($body, $pos + 1), |
305
|
|
|
charCodeAt($body, $pos + 2), |
306
|
|
|
charCodeAt($body, $pos + 3), |
307
|
|
|
charCodeAt($body, $pos + 4) |
308
|
|
|
); |
309
|
|
|
if ($charCode < 0) { |
310
|
|
|
throw new SyntaxErrorException( |
311
|
|
|
$this->lexer->getSource(), |
312
|
|
|
$pos, |
313
|
|
|
\sprintf('Invalid character escape sequence: %s.', $unicodeString) |
314
|
|
|
); |
315
|
|
|
} |
316
|
|
|
$value .= $unicodeString; |
317
|
|
|
$pos += 4; |
318
|
|
|
break; |
319
|
|
|
default: |
320
|
|
|
throw new SyntaxErrorException( |
321
|
|
|
$this->lexer->getSource(), |
322
|
|
|
$pos, |
323
|
|
|
\sprintf('Invalid character escape sequence: \\%s.', \chr($code)) |
324
|
|
|
); |
325
|
|
|
} |
326
|
|
|
|
327
|
|
|
++$pos; |
328
|
|
|
|
329
|
|
|
$chunkStart = $pos; |
330
|
|
|
} |
331
|
|
|
} |
332
|
|
|
|
333
|
|
|
throw new SyntaxErrorException($this->lexer->getSource(), $pos, 'Unterminated string.'); |
334
|
|
|
} |
335
|
|
|
|
336
|
|
|
/** |
337
|
|
|
* @param int $pos |
338
|
|
|
* @param int $line |
339
|
|
|
* @param int $col |
340
|
|
|
* @param Token $prev |
341
|
|
|
* @return Token |
342
|
|
|
*/ |
343
|
|
|
protected function readSpread(int $pos, int $line, int $col, Token $prev): Token |
344
|
|
|
{ |
345
|
|
|
return new Token(TokenKindEnum::SPREAD, $pos, $pos + 3, $line, $col, $prev); |
346
|
|
|
} |
347
|
|
|
|
348
|
|
|
/** |
349
|
|
|
* @param int $pos |
350
|
|
|
* @param int $line |
351
|
|
|
* @param int $col |
352
|
|
|
* @param Token $prev |
353
|
|
|
* @return Token |
354
|
|
|
*/ |
355
|
|
|
protected function readDollar(int $pos, int $line, int $col, Token $prev): Token |
356
|
|
|
{ |
357
|
|
|
return new Token(TokenKindEnum::DOLLAR, $pos, $pos + 1, $line, $col, $prev); |
358
|
|
|
} |
359
|
|
|
|
360
|
|
|
/** |
361
|
|
|
* @param int $pos |
362
|
|
|
* @param int $line |
363
|
|
|
* @param int $col |
364
|
|
|
* @param Token $prev |
365
|
|
|
* @return Token |
366
|
|
|
*/ |
367
|
|
|
protected function readPipe(int $pos, int $line, int $col, Token $prev): Token |
368
|
|
|
{ |
369
|
|
|
return new Token(TokenKindEnum::PIPE, $pos, $pos + 1, $line, $col, $prev); |
370
|
|
|
} |
371
|
|
|
|
372
|
|
|
/** |
373
|
|
|
* @param int $code |
374
|
|
|
* @param int $pos |
375
|
|
|
* @param int $line |
376
|
|
|
* @param int $col |
377
|
|
|
* @param Token $prev |
378
|
|
|
* @return Token |
379
|
|
|
*/ |
380
|
|
|
protected function readParenthesis(int $code, int $pos, int $line, int $col, Token $prev): Token |
381
|
|
|
{ |
382
|
|
|
return $code === 40 |
383
|
|
|
? new Token(TokenKindEnum::PAREN_L, $pos, $pos + 1, $line, $col, $prev) |
384
|
|
|
: new Token(TokenKindEnum::PAREN_R, $pos, $pos + 1, $line, $col, $prev); |
385
|
|
|
} |
386
|
|
|
|
387
|
|
|
/** |
388
|
|
|
* @param int $pos |
389
|
|
|
* @param int $line |
390
|
|
|
* @param int $col |
391
|
|
|
* @param Token $prev |
392
|
|
|
* @return Token |
393
|
|
|
*/ |
394
|
|
|
protected function readEquals(int $pos, int $line, int $col, Token $prev): Token |
395
|
|
|
{ |
396
|
|
|
return new Token(TokenKindEnum::EQUALS, $pos, $pos + 1, $line, $col, $prev); |
397
|
|
|
} |
398
|
|
|
|
399
|
|
|
/** |
400
|
|
|
* @param int $pos |
401
|
|
|
* @param int $line |
402
|
|
|
* @param int $col |
403
|
|
|
* @param Token $prev |
404
|
|
|
* @return Token |
405
|
|
|
*/ |
406
|
|
|
protected function readAt(int $pos, int $line, int $col, Token $prev): Token |
407
|
|
|
{ |
408
|
|
|
return new Token(TokenKindEnum::AT, $pos, $pos + 1, $line, $col, $prev); |
409
|
|
|
} |
410
|
|
|
|
411
|
|
|
/** |
412
|
|
|
* @param int $pos |
413
|
|
|
* @param int $line |
414
|
|
|
* @param int $col |
415
|
|
|
* @param Token $prev |
416
|
|
|
* @return Token |
417
|
|
|
*/ |
418
|
|
|
protected function readComment(int $pos, int $line, int $col, Token $prev): Token |
419
|
|
|
{ |
420
|
|
|
$body = $this->lexer->getBody(); |
421
|
|
|
$start = $pos; |
422
|
|
|
|
423
|
|
|
do { |
424
|
|
|
$code = charCodeAt($body, ++$pos); |
425
|
|
|
} while ($code !== null && ($code > 0x001f || $code === 0x0009)); // SourceCharacter but not LineTerminator |
426
|
|
|
|
427
|
|
|
return new Token( |
428
|
|
|
TokenKindEnum::COMMENT, |
429
|
|
|
$start, |
430
|
|
|
$pos, |
431
|
|
|
$line, |
432
|
|
|
$col, |
433
|
|
|
$prev, |
434
|
|
|
sliceString($body, $start + 1, $pos) |
435
|
|
|
); |
436
|
|
|
} |
437
|
|
|
|
438
|
|
|
/** |
439
|
|
|
* @param int $pos |
440
|
|
|
* @param int $line |
441
|
|
|
* @param int $col |
442
|
|
|
* @param Token $prev |
443
|
|
|
* @return Token |
444
|
|
|
*/ |
445
|
|
|
protected function readColon(int $pos, int $line, int $col, Token $prev): Token |
446
|
|
|
{ |
447
|
|
|
return new Token(TokenKindEnum::COLON, $pos, $pos + 1, $line, $col, $prev); |
448
|
|
|
} |
449
|
|
|
|
450
|
|
|
/** |
451
|
|
|
* @param int $pos |
452
|
|
|
* @param int $line |
453
|
|
|
* @param int $col |
454
|
|
|
* @param Token $prev |
455
|
|
|
* @return Token |
456
|
|
|
*/ |
457
|
|
|
protected function readAmp(int $pos, int $line, int $col, Token $prev): Token |
458
|
|
|
{ |
459
|
|
|
return new Token(TokenKindEnum::AMP, $pos, $pos + 1, $line, $col, $prev); |
460
|
|
|
} |
461
|
|
|
|
462
|
|
|
/** |
463
|
|
|
* @param int $pos |
464
|
|
|
* @param int $line |
465
|
|
|
* @param int $col |
466
|
|
|
* @param Token $prev |
467
|
|
|
* @return Token |
468
|
|
|
*/ |
469
|
|
|
protected function readBang(int $pos, int $line, int $col, Token $prev): Token |
470
|
|
|
{ |
471
|
|
|
return new Token(TokenKindEnum::BANG, $pos, $pos + 1, $line, $col, $prev); |
472
|
|
|
} |
473
|
|
|
|
474
|
|
|
/** |
475
|
|
|
* @param int $code |
476
|
|
|
* @param int $pos |
477
|
|
|
* @param int $line |
478
|
|
|
* @param int $col |
479
|
|
|
* @param Token $prev |
480
|
|
|
* @return Token |
481
|
|
|
*/ |
482
|
|
|
protected function readBrace(int $code, int $pos, int $line, int $col, Token $prev): Token |
483
|
|
|
{ |
484
|
|
|
return $code === 123 |
485
|
|
|
? new Token(TokenKindEnum::BRACE_L, $pos, $pos + 1, $line, $col, $prev) |
486
|
|
|
: new Token(TokenKindEnum::BRACE_R, $pos, $pos + 1, $line, $col, $prev); |
487
|
|
|
} |
488
|
|
|
|
489
|
|
|
/** |
490
|
|
|
* @param int $code |
491
|
|
|
* @param int $pos |
492
|
|
|
* @param int $line |
493
|
|
|
* @param int $col |
494
|
|
|
* @param Token $prev |
495
|
|
|
* @return Token |
496
|
|
|
*/ |
497
|
|
|
protected function readBracket(int $code, int $pos, int $line, int $col, Token $prev): Token |
498
|
|
|
{ |
499
|
|
|
return $code === 91 |
500
|
|
|
? new Token(TokenKindEnum::BRACKET_L, $pos, $pos + 1, $line, $col, $prev) |
501
|
|
|
: new Token(TokenKindEnum::BRACKET_R, $pos, $pos + 1, $line, $col, $prev); |
502
|
|
|
} |
503
|
|
|
|
504
|
|
|
/** |
505
|
|
|
* @param int $code |
506
|
|
|
* @param int $pos |
507
|
|
|
* @return int |
508
|
|
|
* @throws SyntaxErrorException |
509
|
|
|
*/ |
510
|
|
|
protected function readDigits(int $code, int $pos): int |
511
|
|
|
{ |
512
|
|
|
$body = $this->lexer->getBody(); |
513
|
|
|
|
514
|
|
|
if (isNumber($code)) { |
515
|
|
|
do { |
516
|
|
|
$code = charCodeAt($body, ++$pos); |
517
|
|
|
} while (isNumber($code)); |
518
|
|
|
|
519
|
|
|
return $pos; |
520
|
|
|
} |
521
|
|
|
|
522
|
|
|
throw new SyntaxErrorException( |
523
|
|
|
$this->lexer->getSource(), |
524
|
|
|
$pos, |
525
|
|
|
sprintf('Invalid number, expected digit but got: %s.', printCharCode($code)) |
526
|
|
|
); |
527
|
|
|
} |
528
|
|
|
} |
529
|
|
|
|