OLD | NEW |
| (Empty) |
1 // Copyright (c) 2011, the Dart project authors. Please see the AUTHORS file | |
2 // for details. All rights reserved. Use of this source code is governed by a | |
3 // BSD-style license that can be found in the LICENSE file. | |
4 | |
5 #library("json"); | |
6 | |
7 // Pure Dart implementation of JSON protocol. | |
8 | |
9 /** | |
10 * Utility class to parse JSON and serialize objects to JSON. | |
11 */ | |
12 class JSON { | |
13 /** | |
14 * Parses [:json:] and build the corresponding object. | |
15 */ | |
16 static parse(String json) { | |
17 return JsonParser.parse(json); | |
18 } | |
19 | |
20 /** | |
21 * Serializes [:object:] into JSON string. | |
22 */ | |
23 static String stringify(Object object) { | |
24 return JsonStringifier.stringify(object); | |
25 } | |
26 } | |
27 | |
28 //// Implementation /////////////////////////////////////////////////////////// | |
29 | |
30 /** | |
31 * Union-like class for JSON tokens. | |
32 */ | |
33 class JsonToken { | |
34 static final int STRING = 0; | |
35 static final int NUMBER = 1; | |
36 static final int NULL = 2; | |
37 static final int FALSE = 3; | |
38 static final int TRUE = 4; | |
39 static final int RBRACKET = 5; | |
40 static final int LBRACKET = 6; | |
41 static final int RBRACE = 7; | |
42 static final int LBRACE = 8; | |
43 static final int COLON = 9; | |
44 static final int COMMA = 10; | |
45 | |
46 final int kind; | |
47 final String _s; | |
48 final num _n; | |
49 | |
50 String get str() { | |
51 assert(kind == STRING); | |
52 return _s; | |
53 } | |
54 | |
55 num get number() { | |
56 assert(kind == NUMBER); | |
57 return _n; | |
58 } | |
59 | |
60 const JsonToken._internal(this.kind, this._s, this._n); | |
61 | |
62 factory JsonToken.string(String s) { | |
63 return new JsonToken._internal(STRING, s, 0); | |
64 } | |
65 factory JsonToken.number(num n) { | |
66 return new JsonToken._internal(NUMBER, '', n); | |
67 } | |
68 factory JsonToken.atom(int kind) { | |
69 return new JsonToken._internal(kind, '', 0); | |
70 } | |
71 | |
72 String toString() { | |
73 switch (kind) { | |
74 case STRING: | |
75 return 'STRING(${str})'; | |
76 | |
77 case NUMBER: | |
78 return 'NUMBER(${number})'; | |
79 | |
80 case NULL: | |
81 return 'ATOM(null)'; | |
82 | |
83 case FALSE: | |
84 return 'ATOM(false)'; | |
85 | |
86 case TRUE: | |
87 return 'ATOM(true)'; | |
88 | |
89 case RBRACKET: | |
90 return 'ATOM(])'; | |
91 | |
92 case LBRACKET: | |
93 return 'ATOM([)'; | |
94 | |
95 case RBRACE: | |
96 return 'ATOM(})'; | |
97 | |
98 case LBRACE: | |
99 return 'ATOM({)'; | |
100 | |
101 case COLON: | |
102 return 'ATOM(:)'; | |
103 | |
104 case COMMA: | |
105 return 'ATOM(,)'; | |
106 } | |
107 } | |
108 } | |
109 | |
110 typedef bool Predicate(int c); | |
111 | |
112 class JsonTokenizer { | |
113 static final int BACKSPACE = 8; // '\b'.charCodeAt(0) | |
114 static final int TAB = 9; // '\t'.charCodeAt(0) | |
115 static final int NEW_LINE = 10; // '\n'.charCodeAt(0) | |
116 static final int FORM_FEED = 12; // '\f'.charCodeAt(0) | |
117 static final int LINE_FEED = 13; // '\r'.charCodeAt(0) | |
118 static final int SPACE = 32; // ' '.charCodeAt(0) | |
119 static final int QUOTE = 34; // '"'.charCodeAt(0) | |
120 static final int PLUS = 43; // '+'.charCodeAt(0) | |
121 static final int COMMA = 44; // ','.charCodeAt(0) | |
122 static final int MINUS = 45; // '-'.charCodeAt(0) | |
123 static final int DOT = 46; // '.'.charCodeAt(0) | |
124 static final int SLASH = 47; // '/'.charCodeAt(0) | |
125 static final int ZERO = 48; // '0'.charCodeAt(0) | |
126 static final int NINE = 57; // '9'.charCodeAt(0) | |
127 static final int COLON = 58; // ':'.charCodeAt(0) | |
128 static final int A_BIG = 65; // 'A'.charCodeAt(0) | |
129 static final int E_BIG = 69; // 'E'.charCodeAt(0) | |
130 static final int Z_BIG = 90; // 'Z'.charCodeAt(0) | |
131 static final int LBRACKET = 91; // '['.charCodeAt(0) | |
132 static final int BACKSLASH = 92; // '\\'.charCodeAt(0) | |
133 static final int RBRACKET = 93; // ']'.charCodeAt(0) | |
134 static final int A_SMALL = 97; // 'a'.charCodeAt(0) | |
135 static final int B_SMALL = 98; // 'b'.charCodeAt(0) | |
136 static final int E_SMALL = 101; // 'e'.charCodeAt(0) | |
137 static final int N_SMALL = 110; // 'n'.charCodeAt(0) | |
138 static final int R_SMALL = 114; // 'r'.charCodeAt(0) | |
139 static final int Z_SMALL = 122; // 'z'.charCodeAt(0) | |
140 static final int LBRACE = 123; // '{'.charCodeAt(0) | |
141 static final int RBRACE = 125; // '}'.charCodeAt(0) | |
142 | |
143 JsonTokenizer(String s) : _s = s + ' ', _pos = 0, _len = s.length + 1 {} | |
144 | |
145 /** | |
146 * Fetches next token or [:null:] if the stream has been exhausted. | |
147 */ | |
148 JsonToken next() { | |
149 while (_pos < _len && isWhitespace(_s.charCodeAt(_pos))) { | |
150 _pos++; | |
151 } | |
152 if (_pos == _len) { | |
153 return null; | |
154 } | |
155 | |
156 final int cur = _s.charCodeAt(_pos); | |
157 switch (true) { | |
158 case cur == QUOTE: | |
159 _pos++; | |
160 List<int> charCodes = new List<int>(); | |
161 while (_pos < _len) { | |
162 int c = _s.charCodeAt(_pos); | |
163 if (c == QUOTE) { | |
164 break; | |
165 } | |
166 if (c == BACKSLASH) { | |
167 _pos++; | |
168 if (_pos == _len) { | |
169 throw '\\ at the end'; | |
170 } | |
171 | |
172 switch (_s[_pos]) { | |
173 case '"': | |
174 c = QUOTE; | |
175 break; | |
176 case '\\': | |
177 c = BACKSLASH; | |
178 break; | |
179 case '/': | |
180 c = SLASH; | |
181 break; | |
182 case 'b': | |
183 c = BACKSPACE; | |
184 break; | |
185 case 'n': | |
186 c = NEW_LINE; | |
187 break; | |
188 case 'r': | |
189 c = LINE_FEED; | |
190 break; | |
191 case 'f': | |
192 c = FORM_FEED; | |
193 break; | |
194 case 't': | |
195 c = TAB; | |
196 break; | |
197 case 'u': | |
198 if (_pos + 5 > _len) { | |
199 throw 'Invalid unicode esacape sequence: \\' + | |
200 _s.substring(_pos, _len); | |
201 } | |
202 final codeString = _s.substring(_pos + 1, _pos + 5); | |
203 c = Math.parseInt('0x' + codeString); | |
204 if (c >= 128) { | |
205 // TODO(jmessery): the VM doesn't support 2-byte strings yet | |
206 // see runtime/lib/string.cc:49 | |
207 // So instead we replace these characters with '?' | |
208 c = '?'.charCodeAt(0); | |
209 } | |
210 _pos += 4; | |
211 break; | |
212 default: | |
213 throw 'Invalid esacape sequence: \\' + _s[_pos]; | |
214 } | |
215 } | |
216 charCodes.add(c); | |
217 _pos++; | |
218 } | |
219 if (_pos == _len) { | |
220 throw 'Unmatched quote'; | |
221 } | |
222 | |
223 final String body = new String.fromCharCodes(charCodes); | |
224 _pos++; | |
225 return new JsonToken.string(body); | |
226 | |
227 case cur == MINUS || isDigit(cur): | |
228 skipDigits() { | |
229 _scanWhile((int c) => isDigit(c), 'Invalid number'); | |
230 } | |
231 | |
232 int c = cur; | |
233 final int startPos = _pos; | |
234 int value = 0; | |
235 bool isNegative = false; | |
236 if (c == MINUS) { | |
237 isNegative = true; | |
238 _pos++; | |
239 c = _s.charCodeAt(_pos); | |
240 } | |
241 while (isDigit(c)) { | |
242 value = value * 10 + c - ZERO; | |
243 _pos++; | |
244 c = _s.charCodeAt(_pos); | |
245 } | |
246 | |
247 if (c != DOT) { | |
248 if (c != E_SMALL && cur != E_BIG) { | |
249 if (isNegative) value = -value; | |
250 return new JsonToken.number(value); | |
251 } | |
252 } else { | |
253 _pos++; | |
254 skipDigits(); | |
255 c = _s.charCodeAt(_pos); | |
256 } | |
257 | |
258 if (c == E_SMALL || c == E_BIG) { | |
259 // TODO: consider keeping E+ as an integer. | |
260 _pos++; | |
261 c = _s.charCodeAt(_pos); | |
262 if (c == PLUS || c == MINUS) { | |
263 _pos++; | |
264 } | |
265 skipDigits(); | |
266 } | |
267 | |
268 final String body = _s.substring(startPos, _pos); | |
269 return new JsonToken.number(Math.parseDouble(body)); | |
270 | |
271 case cur == LBRACE: | |
272 _pos++; | |
273 return new JsonToken.atom(JsonToken.LBRACE); | |
274 | |
275 case cur == RBRACE: | |
276 _pos++; | |
277 return new JsonToken.atom(JsonToken.RBRACE); | |
278 | |
279 case cur == LBRACKET: | |
280 _pos++; | |
281 return new JsonToken.atom(JsonToken.LBRACKET); | |
282 | |
283 case cur == RBRACKET: | |
284 _pos++; | |
285 return new JsonToken.atom(JsonToken.RBRACKET); | |
286 | |
287 case cur == COMMA: | |
288 _pos++; | |
289 return new JsonToken.atom(JsonToken.COMMA); | |
290 | |
291 case cur == COLON: | |
292 _pos++; | |
293 return new JsonToken.atom(JsonToken.COLON); | |
294 | |
295 case isLetter(cur): | |
296 final int startPos = _pos; | |
297 _pos++; | |
298 while (_pos < _len && isLetter(_s.charCodeAt(_pos))) { | |
299 _pos++; | |
300 } | |
301 final String body = _s.substring(startPos, _pos); | |
302 switch (body) { | |
303 case 'null': | |
304 return new JsonToken.atom(JsonToken.NULL); | |
305 | |
306 case 'false': | |
307 return new JsonToken.atom(JsonToken.FALSE); | |
308 | |
309 case 'true': | |
310 return new JsonToken.atom(JsonToken.TRUE); | |
311 | |
312 default: | |
313 throw 'Unexpected sequence ${body}'; | |
314 } | |
315 // TODO: Bogous, to please DartVM. | |
316 return null; | |
317 | |
318 default: | |
319 throw 'Invalid token'; | |
320 } | |
321 } | |
322 | |
323 final String _s; | |
324 int _pos; | |
325 final int _len; | |
326 | |
327 void _scanWhile(Predicate predicate, String errorMsg) { | |
328 while (_pos < _len && predicate(_s.charCodeAt(_pos))) { | |
329 _pos++; | |
330 } | |
331 if (_pos == _len) { | |
332 throw errorMsg; | |
333 } | |
334 } | |
335 | |
336 // TODO other kind of whitespace. | |
337 static bool isWhitespace(int c) { | |
338 return c == SPACE || c == TAB || c == NEW_LINE || c == LINE_FEED; | |
339 } | |
340 static bool isDigit(int c) { | |
341 return (ZERO <= c) && (c <= NINE); | |
342 } | |
343 static bool isLetter(int c) { | |
344 return ((A_SMALL <= c) && (c <= Z_SMALL)) || ((A_BIG <= c) && (c <= Z_BIG))
; | |
345 } | |
346 } | |
347 | |
348 class JsonParser { | |
349 static parse(String json) { | |
350 return new JsonParser._internal(json)._parseToplevel(); | |
351 } | |
352 | |
353 final JsonTokenizer _tokenizer; | |
354 | |
355 JsonParser._internal(String json) : _tokenizer = new JsonTokenizer(json) {} | |
356 | |
357 _parseToplevel() { | |
358 JsonToken token = _tokenizer.next(); | |
359 final result = _parseValue(token); | |
360 token = _tokenizer.next(); | |
361 if (token !== null) { | |
362 throw 'Junk at the end'; | |
363 } | |
364 return result; | |
365 } | |
366 | |
367 _parseValue(final JsonToken token) { | |
368 if (token === null) { | |
369 throw 'Nothing to parse'; | |
370 } | |
371 switch (token.kind) { | |
372 case JsonToken.STRING: | |
373 return token.str; | |
374 | |
375 case JsonToken.NUMBER: | |
376 return token.number; | |
377 | |
378 case JsonToken.NULL: | |
379 return null; | |
380 | |
381 case JsonToken.FALSE: | |
382 return false; | |
383 | |
384 case JsonToken.TRUE: | |
385 return true; | |
386 | |
387 case JsonToken.LBRACE: | |
388 return _parseObject(); | |
389 | |
390 case JsonToken.LBRACKET: | |
391 return _parseList(); | |
392 | |
393 default: | |
394 throw 'Unexpected token: ${token}'; | |
395 } | |
396 } | |
397 | |
398 _parseObject() { | |
399 Map<String, Object> object = new Map<String, Object>(); | |
400 | |
401 _parseSequence(JsonToken.RBRACE, (JsonToken token) { | |
402 _assertTokenKind(token, JsonToken.STRING); | |
403 final String key = token.str; | |
404 | |
405 token = _tokenizer.next(); | |
406 _assertTokenKind(token, JsonToken.COLON); | |
407 | |
408 token = _tokenizer.next(); | |
409 final value = _parseValue(token); | |
410 | |
411 object[key] = value; | |
412 }); | |
413 | |
414 return object; | |
415 } | |
416 | |
417 _parseList() { | |
418 List<Object> list = new List<Object>(); | |
419 | |
420 _parseSequence(JsonToken.RBRACKET, (JsonToken token) { | |
421 final value = _parseValue(token); | |
422 list.add(value); | |
423 }); | |
424 | |
425 return list; | |
426 } | |
427 | |
428 void _parseSequence(int endTokenKind, void parseElement(JsonToken token)) { | |
429 JsonToken token = _tokenizer.next(); | |
430 if (token === null) { | |
431 throw 'Unexpected end of stream'; | |
432 } | |
433 if (token.kind == endTokenKind) { | |
434 return; | |
435 } | |
436 | |
437 parseElement(token); | |
438 | |
439 token = _tokenizer.next(); | |
440 if (token === null) { | |
441 throw 'Expected either comma or terminator'; | |
442 } | |
443 while (token.kind != endTokenKind) { | |
444 _assertTokenKind(token, JsonToken.COMMA); | |
445 | |
446 token = _tokenizer.next(); | |
447 parseElement(token); | |
448 | |
449 token = _tokenizer.next(); | |
450 } | |
451 } | |
452 | |
453 void _assertTokenKind(JsonToken token, int kind) { | |
454 if (token === null || token.kind != kind) { | |
455 throw 'Unexpected token kind: token = ${token}, expected kind = ${kind}'; | |
456 } | |
457 } | |
458 | |
459 // TODO: consider factor out error throwing code and build more complicated | |
460 // data structure to provide more info for a caller. | |
461 } | |
462 | |
463 // TODO: proper base class. | |
464 class JsonUnsupportedObjectType { | |
465 const JsonUnsupportedObjectType(); | |
466 } | |
467 | |
468 class JsonStringifier { | |
469 static String stringify(final object) { | |
470 JsonStringifier stringifier = new JsonStringifier._internal(); | |
471 stringifier._stringify(object); | |
472 return stringifier._result; | |
473 } | |
474 | |
475 JsonStringifier._internal() | |
476 : _sb = new StringBuffer(), _seen = new List<Object>() {} | |
477 StringBuffer _sb; | |
478 List<Object> _seen; // TODO: that should be identity set. | |
479 String get _result() { return _sb.toString(); } | |
480 | |
481 static String _numberToString(num x) { | |
482 // TODO: need some more investigation what to do with precision | |
483 // of double values. | |
484 switch (true) { | |
485 case x is int: | |
486 return x.toString(); | |
487 | |
488 case x is double: | |
489 return x.toString(); | |
490 | |
491 default: | |
492 return x.toDouble().toString(); | |
493 } | |
494 } | |
495 | |
496 // TODO: add others. | |
497 static bool _needsEscape(int charCode) { | |
498 return JsonTokenizer.QUOTE == charCode || JsonTokenizer.BACKSLASH == charCod
e | |
499 || JsonTokenizer.NEW_LINE == charCode || JsonTokenizer.LINE_FEED == charCo
de; | |
500 } | |
501 | |
502 static void _escape(StringBuffer sb, String s) { | |
503 // TODO: support \u code points. | |
504 // TODO: use writeCodePoint when implemented. | |
505 // TODO: use for each if implemented. | |
506 final int length = s.length; | |
507 bool needsEscape = false; | |
508 final charCodes = new List<int>(); | |
509 for (int i = 0; i < length; i++) { | |
510 int charCode = s.charCodeAt(i); | |
511 if (_needsEscape(charCode)) { | |
512 charCodes.add(JsonTokenizer.BACKSLASH); | |
513 needsEscape = true; | |
514 | |
515 if (JsonTokenizer.NEW_LINE == charCode) { | |
516 charCode = JsonTokenizer.N_SMALL; | |
517 } else if (JsonTokenizer.LINE_FEED == charCode) { | |
518 charCode = JsonTokenizer.R_SMALL; | |
519 } | |
520 } | |
521 charCodes.add(charCode); | |
522 } | |
523 sb.add(needsEscape ? new String.fromCharCodes(charCodes) : s); | |
524 } | |
525 | |
526 void _checkCycle(final object) { | |
527 // TODO: use Iterables. | |
528 for (int i = 0; i < _seen.length; i++) { | |
529 if (_seen[i] === object) { | |
530 throw 'Cyclic structure'; | |
531 } | |
532 } | |
533 _seen.add(object); | |
534 } | |
535 | |
536 void _stringify(final object) { | |
537 switch (true) { | |
538 case object is num: | |
539 // TODO: use writeOn. | |
540 _sb.add(_numberToString(object)); | |
541 return; | |
542 | |
543 case object === true: | |
544 _sb.add('true'); | |
545 return; | |
546 | |
547 case object === false: | |
548 _sb.add('false'); | |
549 return; | |
550 | |
551 case object === null: | |
552 _sb.add('null'); | |
553 return; | |
554 | |
555 case object is String: | |
556 _sb.add('"'); | |
557 _escape(_sb, object); | |
558 _sb.add('"'); | |
559 return; | |
560 | |
561 case object is List: | |
562 _checkCycle(object); | |
563 List a = object; | |
564 _sb.add('['); | |
565 if (a.length > 0) { | |
566 _stringify(a[0]); | |
567 // TODO: switch to Iterables. | |
568 for (int i = 1; i < a.length; i++) { | |
569 _sb.add(','); | |
570 _stringify(a[i]); | |
571 } | |
572 } | |
573 _sb.add(']'); | |
574 _seen.removeLast(); | |
575 return; | |
576 | |
577 case object is Map: | |
578 _checkCycle(object); | |
579 Map<String, Object> m = object; | |
580 _sb.add('{'); | |
581 bool first = true; | |
582 m.forEach((String key, Object value) { | |
583 if (!first) { | |
584 _sb.add(',"'); | |
585 } else { | |
586 _sb.add('"'); | |
587 } | |
588 _escape(_sb, key); | |
589 _sb.add('":'); | |
590 _stringify(value); | |
591 first = false; | |
592 }); | |
593 _sb.add('}'); | |
594 _seen.removeLast(); | |
595 return; | |
596 | |
597 default: | |
598 throw const JsonUnsupportedObjectType(); | |
599 } | |
600 } | |
601 } | |
OLD | NEW |