1 | /*global Buffer*/
|
---|
2 | // Named constants with unique integer values
|
---|
3 | var C = {};
|
---|
4 | // Tokens
|
---|
5 | var LEFT_BRACE = C.LEFT_BRACE = 0x1;
|
---|
6 | var RIGHT_BRACE = C.RIGHT_BRACE = 0x2;
|
---|
7 | var LEFT_BRACKET = C.LEFT_BRACKET = 0x3;
|
---|
8 | var RIGHT_BRACKET = C.RIGHT_BRACKET = 0x4;
|
---|
9 | var COLON = C.COLON = 0x5;
|
---|
10 | var COMMA = C.COMMA = 0x6;
|
---|
11 | var TRUE = C.TRUE = 0x7;
|
---|
12 | var FALSE = C.FALSE = 0x8;
|
---|
13 | var NULL = C.NULL = 0x9;
|
---|
14 | var STRING = C.STRING = 0xa;
|
---|
15 | var NUMBER = C.NUMBER = 0xb;
|
---|
16 | // Tokenizer States
|
---|
17 | var START = C.START = 0x11;
|
---|
18 | var STOP = C.STOP = 0x12;
|
---|
19 | var TRUE1 = C.TRUE1 = 0x21;
|
---|
20 | var TRUE2 = C.TRUE2 = 0x22;
|
---|
21 | var TRUE3 = C.TRUE3 = 0x23;
|
---|
22 | var FALSE1 = C.FALSE1 = 0x31;
|
---|
23 | var FALSE2 = C.FALSE2 = 0x32;
|
---|
24 | var FALSE3 = C.FALSE3 = 0x33;
|
---|
25 | var FALSE4 = C.FALSE4 = 0x34;
|
---|
26 | var NULL1 = C.NULL1 = 0x41;
|
---|
27 | var NULL2 = C.NULL2 = 0x42;
|
---|
28 | var NULL3 = C.NULL3 = 0x43;
|
---|
29 | var NUMBER1 = C.NUMBER1 = 0x51;
|
---|
30 | var NUMBER3 = C.NUMBER3 = 0x53;
|
---|
31 | var STRING1 = C.STRING1 = 0x61;
|
---|
32 | var STRING2 = C.STRING2 = 0x62;
|
---|
33 | var STRING3 = C.STRING3 = 0x63;
|
---|
34 | var STRING4 = C.STRING4 = 0x64;
|
---|
35 | var STRING5 = C.STRING5 = 0x65;
|
---|
36 | var STRING6 = C.STRING6 = 0x66;
|
---|
37 | // Parser States
|
---|
38 | var VALUE = C.VALUE = 0x71;
|
---|
39 | var KEY = C.KEY = 0x72;
|
---|
40 | // Parser Modes
|
---|
41 | var OBJECT = C.OBJECT = 0x81;
|
---|
42 | var ARRAY = C.ARRAY = 0x82;
|
---|
43 | // Character constants
|
---|
44 | var BACK_SLASH = "\\".charCodeAt(0);
|
---|
45 | var FORWARD_SLASH = "\/".charCodeAt(0);
|
---|
46 | var BACKSPACE = "\b".charCodeAt(0);
|
---|
47 | var FORM_FEED = "\f".charCodeAt(0);
|
---|
48 | var NEWLINE = "\n".charCodeAt(0);
|
---|
49 | var CARRIAGE_RETURN = "\r".charCodeAt(0);
|
---|
50 | var TAB = "\t".charCodeAt(0);
|
---|
51 |
|
---|
52 | var STRING_BUFFER_SIZE = 64 * 1024;
|
---|
53 |
|
---|
54 | function Parser() {
|
---|
55 | this.tState = START;
|
---|
56 | this.value = undefined;
|
---|
57 |
|
---|
58 | this.string = undefined; // string data
|
---|
59 | this.stringBuffer = Buffer.alloc ? Buffer.alloc(STRING_BUFFER_SIZE) : new Buffer(STRING_BUFFER_SIZE);
|
---|
60 | this.stringBufferOffset = 0;
|
---|
61 | this.unicode = undefined; // unicode escapes
|
---|
62 | this.highSurrogate = undefined;
|
---|
63 |
|
---|
64 | this.key = undefined;
|
---|
65 | this.mode = undefined;
|
---|
66 | this.stack = [];
|
---|
67 | this.state = VALUE;
|
---|
68 | this.bytes_remaining = 0; // number of bytes remaining in multi byte utf8 char to read after split boundary
|
---|
69 | this.bytes_in_sequence = 0; // bytes in multi byte utf8 char to read
|
---|
70 | this.temp_buffs = { "2": new Buffer(2), "3": new Buffer(3), "4": new Buffer(4) }; // for rebuilding chars split before boundary is reached
|
---|
71 |
|
---|
72 | // Stream offset
|
---|
73 | this.offset = -1;
|
---|
74 | }
|
---|
75 |
|
---|
76 | // Slow code to string converter (only used when throwing syntax errors)
|
---|
77 | Parser.toknam = function (code) {
|
---|
78 | var keys = Object.keys(C);
|
---|
79 | for (var i = 0, l = keys.length; i < l; i++) {
|
---|
80 | var key = keys[i];
|
---|
81 | if (C[key] === code) { return key; }
|
---|
82 | }
|
---|
83 | return code && ("0x" + code.toString(16));
|
---|
84 | };
|
---|
85 |
|
---|
86 | var proto = Parser.prototype;
|
---|
87 | proto.onError = function (err) { throw err; };
|
---|
88 | proto.charError = function (buffer, i) {
|
---|
89 | this.tState = STOP;
|
---|
90 | this.onError(new Error("Unexpected " + JSON.stringify(String.fromCharCode(buffer[i])) + " at position " + i + " in state " + Parser.toknam(this.tState)));
|
---|
91 | };
|
---|
92 | proto.appendStringChar = function (char) {
|
---|
93 | if (this.stringBufferOffset >= STRING_BUFFER_SIZE) {
|
---|
94 | this.string += this.stringBuffer.toString('utf8');
|
---|
95 | this.stringBufferOffset = 0;
|
---|
96 | }
|
---|
97 |
|
---|
98 | this.stringBuffer[this.stringBufferOffset++] = char;
|
---|
99 | };
|
---|
100 | proto.appendStringBuf = function (buf, start, end) {
|
---|
101 | var size = buf.length;
|
---|
102 | if (typeof start === 'number') {
|
---|
103 | if (typeof end === 'number') {
|
---|
104 | if (end < 0) {
|
---|
105 | // adding a negative end decreeses the size
|
---|
106 | size = buf.length - start + end;
|
---|
107 | } else {
|
---|
108 | size = end - start;
|
---|
109 | }
|
---|
110 | } else {
|
---|
111 | size = buf.length - start;
|
---|
112 | }
|
---|
113 | }
|
---|
114 |
|
---|
115 | if (size < 0) {
|
---|
116 | size = 0;
|
---|
117 | }
|
---|
118 |
|
---|
119 | if (this.stringBufferOffset + size > STRING_BUFFER_SIZE) {
|
---|
120 | this.string += this.stringBuffer.toString('utf8', 0, this.stringBufferOffset);
|
---|
121 | this.stringBufferOffset = 0;
|
---|
122 | }
|
---|
123 |
|
---|
124 | buf.copy(this.stringBuffer, this.stringBufferOffset, start, end);
|
---|
125 | this.stringBufferOffset += size;
|
---|
126 | };
|
---|
127 | proto.write = function (buffer) {
|
---|
128 | if (typeof buffer === "string") buffer = new Buffer(buffer);
|
---|
129 | var n;
|
---|
130 | for (var i = 0, l = buffer.length; i < l; i++) {
|
---|
131 | if (this.tState === START){
|
---|
132 | n = buffer[i];
|
---|
133 | this.offset++;
|
---|
134 | if(n === 0x7b){ this.onToken(LEFT_BRACE, "{"); // {
|
---|
135 | }else if(n === 0x7d){ this.onToken(RIGHT_BRACE, "}"); // }
|
---|
136 | }else if(n === 0x5b){ this.onToken(LEFT_BRACKET, "["); // [
|
---|
137 | }else if(n === 0x5d){ this.onToken(RIGHT_BRACKET, "]"); // ]
|
---|
138 | }else if(n === 0x3a){ this.onToken(COLON, ":"); // :
|
---|
139 | }else if(n === 0x2c){ this.onToken(COMMA, ","); // ,
|
---|
140 | }else if(n === 0x74){ this.tState = TRUE1; // t
|
---|
141 | }else if(n === 0x66){ this.tState = FALSE1; // f
|
---|
142 | }else if(n === 0x6e){ this.tState = NULL1; // n
|
---|
143 | }else if(n === 0x22){ // "
|
---|
144 | this.string = "";
|
---|
145 | this.stringBufferOffset = 0;
|
---|
146 | this.tState = STRING1;
|
---|
147 | }else if(n === 0x2d){ this.string = "-"; this.tState = NUMBER1; // -
|
---|
148 | }else{
|
---|
149 | if (n >= 0x30 && n < 0x40) { // 1-9
|
---|
150 | this.string = String.fromCharCode(n); this.tState = NUMBER3;
|
---|
151 | } else if (n === 0x20 || n === 0x09 || n === 0x0a || n === 0x0d) {
|
---|
152 | // whitespace
|
---|
153 | } else {
|
---|
154 | return this.charError(buffer, i);
|
---|
155 | }
|
---|
156 | }
|
---|
157 | }else if (this.tState === STRING1){ // After open quote
|
---|
158 | n = buffer[i]; // get current byte from buffer
|
---|
159 | // check for carry over of a multi byte char split between data chunks
|
---|
160 | // & fill temp buffer it with start of this data chunk up to the boundary limit set in the last iteration
|
---|
161 | if (this.bytes_remaining > 0) {
|
---|
162 | for (var j = 0; j < this.bytes_remaining; j++) {
|
---|
163 | this.temp_buffs[this.bytes_in_sequence][this.bytes_in_sequence - this.bytes_remaining + j] = buffer[j];
|
---|
164 | }
|
---|
165 |
|
---|
166 | this.appendStringBuf(this.temp_buffs[this.bytes_in_sequence]);
|
---|
167 | this.bytes_in_sequence = this.bytes_remaining = 0;
|
---|
168 | i = i + j - 1;
|
---|
169 | } else if (this.bytes_remaining === 0 && n >= 128) { // else if no remainder bytes carried over, parse multi byte (>=128) chars one at a time
|
---|
170 | if (n <= 193 || n > 244) {
|
---|
171 | return this.onError(new Error("Invalid UTF-8 character at position " + i + " in state " + Parser.toknam(this.tState)));
|
---|
172 | }
|
---|
173 | if ((n >= 194) && (n <= 223)) this.bytes_in_sequence = 2;
|
---|
174 | if ((n >= 224) && (n <= 239)) this.bytes_in_sequence = 3;
|
---|
175 | if ((n >= 240) && (n <= 244)) this.bytes_in_sequence = 4;
|
---|
176 | if ((this.bytes_in_sequence + i) > buffer.length) { // if bytes needed to complete char fall outside buffer length, we have a boundary split
|
---|
177 | for (var k = 0; k <= (buffer.length - 1 - i); k++) {
|
---|
178 | this.temp_buffs[this.bytes_in_sequence][k] = buffer[i + k]; // fill temp buffer of correct size with bytes available in this chunk
|
---|
179 | }
|
---|
180 | this.bytes_remaining = (i + this.bytes_in_sequence) - buffer.length;
|
---|
181 | i = buffer.length - 1;
|
---|
182 | } else {
|
---|
183 | this.appendStringBuf(buffer, i, i + this.bytes_in_sequence);
|
---|
184 | i = i + this.bytes_in_sequence - 1;
|
---|
185 | }
|
---|
186 | } else if (n === 0x22) {
|
---|
187 | this.tState = START;
|
---|
188 | this.string += this.stringBuffer.toString('utf8', 0, this.stringBufferOffset);
|
---|
189 | this.stringBufferOffset = 0;
|
---|
190 | this.onToken(STRING, this.string);
|
---|
191 | this.offset += Buffer.byteLength(this.string, 'utf8') + 1;
|
---|
192 | this.string = undefined;
|
---|
193 | }
|
---|
194 | else if (n === 0x5c) {
|
---|
195 | this.tState = STRING2;
|
---|
196 | }
|
---|
197 | else if (n >= 0x20) { this.appendStringChar(n); }
|
---|
198 | else {
|
---|
199 | return this.charError(buffer, i);
|
---|
200 | }
|
---|
201 | }else if (this.tState === STRING2){ // After backslash
|
---|
202 | n = buffer[i];
|
---|
203 | if(n === 0x22){ this.appendStringChar(n); this.tState = STRING1;
|
---|
204 | }else if(n === 0x5c){ this.appendStringChar(BACK_SLASH); this.tState = STRING1;
|
---|
205 | }else if(n === 0x2f){ this.appendStringChar(FORWARD_SLASH); this.tState = STRING1;
|
---|
206 | }else if(n === 0x62){ this.appendStringChar(BACKSPACE); this.tState = STRING1;
|
---|
207 | }else if(n === 0x66){ this.appendStringChar(FORM_FEED); this.tState = STRING1;
|
---|
208 | }else if(n === 0x6e){ this.appendStringChar(NEWLINE); this.tState = STRING1;
|
---|
209 | }else if(n === 0x72){ this.appendStringChar(CARRIAGE_RETURN); this.tState = STRING1;
|
---|
210 | }else if(n === 0x74){ this.appendStringChar(TAB); this.tState = STRING1;
|
---|
211 | }else if(n === 0x75){ this.unicode = ""; this.tState = STRING3;
|
---|
212 | }else{
|
---|
213 | return this.charError(buffer, i);
|
---|
214 | }
|
---|
215 | }else if (this.tState === STRING3 || this.tState === STRING4 || this.tState === STRING5 || this.tState === STRING6){ // unicode hex codes
|
---|
216 | n = buffer[i];
|
---|
217 | // 0-9 A-F a-f
|
---|
218 | if ((n >= 0x30 && n < 0x40) || (n > 0x40 && n <= 0x46) || (n > 0x60 && n <= 0x66)) {
|
---|
219 | this.unicode += String.fromCharCode(n);
|
---|
220 | if (this.tState++ === STRING6) {
|
---|
221 | var intVal = parseInt(this.unicode, 16);
|
---|
222 | this.unicode = undefined;
|
---|
223 | if (this.highSurrogate !== undefined && intVal >= 0xDC00 && intVal < (0xDFFF + 1)) { //<56320,57343> - lowSurrogate
|
---|
224 | this.appendStringBuf(new Buffer(String.fromCharCode(this.highSurrogate, intVal)));
|
---|
225 | this.highSurrogate = undefined;
|
---|
226 | } else if (this.highSurrogate === undefined && intVal >= 0xD800 && intVal < (0xDBFF + 1)) { //<55296,56319> - highSurrogate
|
---|
227 | this.highSurrogate = intVal;
|
---|
228 | } else {
|
---|
229 | if (this.highSurrogate !== undefined) {
|
---|
230 | this.appendStringBuf(new Buffer(String.fromCharCode(this.highSurrogate)));
|
---|
231 | this.highSurrogate = undefined;
|
---|
232 | }
|
---|
233 | this.appendStringBuf(new Buffer(String.fromCharCode(intVal)));
|
---|
234 | }
|
---|
235 | this.tState = STRING1;
|
---|
236 | }
|
---|
237 | } else {
|
---|
238 | return this.charError(buffer, i);
|
---|
239 | }
|
---|
240 | } else if (this.tState === NUMBER1 || this.tState === NUMBER3) {
|
---|
241 | n = buffer[i];
|
---|
242 |
|
---|
243 | switch (n) {
|
---|
244 | case 0x30: // 0
|
---|
245 | case 0x31: // 1
|
---|
246 | case 0x32: // 2
|
---|
247 | case 0x33: // 3
|
---|
248 | case 0x34: // 4
|
---|
249 | case 0x35: // 5
|
---|
250 | case 0x36: // 6
|
---|
251 | case 0x37: // 7
|
---|
252 | case 0x38: // 8
|
---|
253 | case 0x39: // 9
|
---|
254 | case 0x2e: // .
|
---|
255 | case 0x65: // e
|
---|
256 | case 0x45: // E
|
---|
257 | case 0x2b: // +
|
---|
258 | case 0x2d: // -
|
---|
259 | this.string += String.fromCharCode(n);
|
---|
260 | this.tState = NUMBER3;
|
---|
261 | break;
|
---|
262 | default:
|
---|
263 | this.tState = START;
|
---|
264 | var result = Number(this.string);
|
---|
265 |
|
---|
266 | if (isNaN(result)){
|
---|
267 | return this.charError(buffer, i);
|
---|
268 | }
|
---|
269 |
|
---|
270 | if ((this.string.match(/[0-9]+/) == this.string) && (result.toString() != this.string)) {
|
---|
271 | // Long string of digits which is an ID string and not valid and/or safe JavaScript integer Number
|
---|
272 | this.onToken(STRING, this.string);
|
---|
273 | } else {
|
---|
274 | this.onToken(NUMBER, result);
|
---|
275 | }
|
---|
276 |
|
---|
277 | this.offset += this.string.length - 1;
|
---|
278 | this.string = undefined;
|
---|
279 | i--;
|
---|
280 | break;
|
---|
281 | }
|
---|
282 | }else if (this.tState === TRUE1){ // r
|
---|
283 | if (buffer[i] === 0x72) { this.tState = TRUE2; }
|
---|
284 | else { return this.charError(buffer, i); }
|
---|
285 | }else if (this.tState === TRUE2){ // u
|
---|
286 | if (buffer[i] === 0x75) { this.tState = TRUE3; }
|
---|
287 | else { return this.charError(buffer, i); }
|
---|
288 | }else if (this.tState === TRUE3){ // e
|
---|
289 | if (buffer[i] === 0x65) { this.tState = START; this.onToken(TRUE, true); this.offset+= 3; }
|
---|
290 | else { return this.charError(buffer, i); }
|
---|
291 | }else if (this.tState === FALSE1){ // a
|
---|
292 | if (buffer[i] === 0x61) { this.tState = FALSE2; }
|
---|
293 | else { return this.charError(buffer, i); }
|
---|
294 | }else if (this.tState === FALSE2){ // l
|
---|
295 | if (buffer[i] === 0x6c) { this.tState = FALSE3; }
|
---|
296 | else { return this.charError(buffer, i); }
|
---|
297 | }else if (this.tState === FALSE3){ // s
|
---|
298 | if (buffer[i] === 0x73) { this.tState = FALSE4; }
|
---|
299 | else { return this.charError(buffer, i); }
|
---|
300 | }else if (this.tState === FALSE4){ // e
|
---|
301 | if (buffer[i] === 0x65) { this.tState = START; this.onToken(FALSE, false); this.offset+= 4; }
|
---|
302 | else { return this.charError(buffer, i); }
|
---|
303 | }else if (this.tState === NULL1){ // u
|
---|
304 | if (buffer[i] === 0x75) { this.tState = NULL2; }
|
---|
305 | else { return this.charError(buffer, i); }
|
---|
306 | }else if (this.tState === NULL2){ // l
|
---|
307 | if (buffer[i] === 0x6c) { this.tState = NULL3; }
|
---|
308 | else { return this.charError(buffer, i); }
|
---|
309 | }else if (this.tState === NULL3){ // l
|
---|
310 | if (buffer[i] === 0x6c) { this.tState = START; this.onToken(NULL, null); this.offset += 3; }
|
---|
311 | else { return this.charError(buffer, i); }
|
---|
312 | }
|
---|
313 | }
|
---|
314 | };
|
---|
315 | proto.onToken = function (token, value) {
|
---|
316 | // Override this to get events
|
---|
317 | };
|
---|
318 |
|
---|
319 | proto.parseError = function (token, value) {
|
---|
320 | this.tState = STOP;
|
---|
321 | this.onError(new Error("Unexpected " + Parser.toknam(token) + (value ? ("(" + JSON.stringify(value) + ")") : "") + " in state " + Parser.toknam(this.state)));
|
---|
322 | };
|
---|
323 | proto.push = function () {
|
---|
324 | this.stack.push({value: this.value, key: this.key, mode: this.mode});
|
---|
325 | };
|
---|
326 | proto.pop = function () {
|
---|
327 | var value = this.value;
|
---|
328 | var parent = this.stack.pop();
|
---|
329 | this.value = parent.value;
|
---|
330 | this.key = parent.key;
|
---|
331 | this.mode = parent.mode;
|
---|
332 | this.emit(value);
|
---|
333 | if (!this.mode) { this.state = VALUE; }
|
---|
334 | };
|
---|
335 | proto.emit = function (value) {
|
---|
336 | if (this.mode) { this.state = COMMA; }
|
---|
337 | this.onValue(value);
|
---|
338 | };
|
---|
339 | proto.onValue = function (value) {
|
---|
340 | // Override me
|
---|
341 | };
|
---|
342 | proto.onToken = function (token, value) {
|
---|
343 | if(this.state === VALUE){
|
---|
344 | if(token === STRING || token === NUMBER || token === TRUE || token === FALSE || token === NULL){
|
---|
345 | if (this.value) {
|
---|
346 | this.value[this.key] = value;
|
---|
347 | }
|
---|
348 | this.emit(value);
|
---|
349 | }else if(token === LEFT_BRACE){
|
---|
350 | this.push();
|
---|
351 | if (this.value) {
|
---|
352 | this.value = this.value[this.key] = {};
|
---|
353 | } else {
|
---|
354 | this.value = {};
|
---|
355 | }
|
---|
356 | this.key = undefined;
|
---|
357 | this.state = KEY;
|
---|
358 | this.mode = OBJECT;
|
---|
359 | }else if(token === LEFT_BRACKET){
|
---|
360 | this.push();
|
---|
361 | if (this.value) {
|
---|
362 | this.value = this.value[this.key] = [];
|
---|
363 | } else {
|
---|
364 | this.value = [];
|
---|
365 | }
|
---|
366 | this.key = 0;
|
---|
367 | this.mode = ARRAY;
|
---|
368 | this.state = VALUE;
|
---|
369 | }else if(token === RIGHT_BRACE){
|
---|
370 | if (this.mode === OBJECT) {
|
---|
371 | this.pop();
|
---|
372 | } else {
|
---|
373 | return this.parseError(token, value);
|
---|
374 | }
|
---|
375 | }else if(token === RIGHT_BRACKET){
|
---|
376 | if (this.mode === ARRAY) {
|
---|
377 | this.pop();
|
---|
378 | } else {
|
---|
379 | return this.parseError(token, value);
|
---|
380 | }
|
---|
381 | }else{
|
---|
382 | return this.parseError(token, value);
|
---|
383 | }
|
---|
384 | }else if(this.state === KEY){
|
---|
385 | if (token === STRING) {
|
---|
386 | this.key = value;
|
---|
387 | this.state = COLON;
|
---|
388 | } else if (token === RIGHT_BRACE) {
|
---|
389 | this.pop();
|
---|
390 | } else {
|
---|
391 | return this.parseError(token, value);
|
---|
392 | }
|
---|
393 | }else if(this.state === COLON){
|
---|
394 | if (token === COLON) { this.state = VALUE; }
|
---|
395 | else { return this.parseError(token, value); }
|
---|
396 | }else if(this.state === COMMA){
|
---|
397 | if (token === COMMA) {
|
---|
398 | if (this.mode === ARRAY) { this.key++; this.state = VALUE; }
|
---|
399 | else if (this.mode === OBJECT) { this.state = KEY; }
|
---|
400 |
|
---|
401 | } else if (token === RIGHT_BRACKET && this.mode === ARRAY || token === RIGHT_BRACE && this.mode === OBJECT) {
|
---|
402 | this.pop();
|
---|
403 | } else {
|
---|
404 | return this.parseError(token, value);
|
---|
405 | }
|
---|
406 | }else{
|
---|
407 | return this.parseError(token, value);
|
---|
408 | }
|
---|
409 | };
|
---|
410 |
|
---|
411 | Parser.C = C;
|
---|
412 |
|
---|
413 | module.exports = Parser;
|
---|