123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414 |
- /*global Buffer*/
- // Named constants with unique integer values
- var C = {};
- // Tokens
- var LEFT_BRACE = C.LEFT_BRACE = 0x1;
- var RIGHT_BRACE = C.RIGHT_BRACE = 0x2;
- var LEFT_BRACKET = C.LEFT_BRACKET = 0x3;
- var RIGHT_BRACKET = C.RIGHT_BRACKET = 0x4;
- var COLON = C.COLON = 0x5;
- var COMMA = C.COMMA = 0x6;
- var TRUE = C.TRUE = 0x7;
- var FALSE = C.FALSE = 0x8;
- var NULL = C.NULL = 0x9;
- var STRING = C.STRING = 0xa;
- var NUMBER = C.NUMBER = 0xb;
- // Tokenizer States
- var START = C.START = 0x11;
- var STOP = C.STOP = 0x12;
- var TRUE1 = C.TRUE1 = 0x21;
- var TRUE2 = C.TRUE2 = 0x22;
- var TRUE3 = C.TRUE3 = 0x23;
- var FALSE1 = C.FALSE1 = 0x31;
- var FALSE2 = C.FALSE2 = 0x32;
- var FALSE3 = C.FALSE3 = 0x33;
- var FALSE4 = C.FALSE4 = 0x34;
- var NULL1 = C.NULL1 = 0x41;
- var NULL2 = C.NULL2 = 0x42;
- var NULL3 = C.NULL3 = 0x43;
- var NUMBER1 = C.NUMBER1 = 0x51;
- var NUMBER3 = C.NUMBER3 = 0x53;
- var STRING1 = C.STRING1 = 0x61;
- var STRING2 = C.STRING2 = 0x62;
- var STRING3 = C.STRING3 = 0x63;
- var STRING4 = C.STRING4 = 0x64;
- var STRING5 = C.STRING5 = 0x65;
- var STRING6 = C.STRING6 = 0x66;
- // Parser States
- var VALUE = C.VALUE = 0x71;
- var KEY = C.KEY = 0x72;
- // Parser Modes
- var OBJECT = C.OBJECT = 0x81;
- var ARRAY = C.ARRAY = 0x82;
- // Character constants
- var BACK_SLASH = "\\".charCodeAt(0);
- var FORWARD_SLASH = "\/".charCodeAt(0);
- var BACKSPACE = "\b".charCodeAt(0);
- var FORM_FEED = "\f".charCodeAt(0);
- var NEWLINE = "\n".charCodeAt(0);
- var CARRIAGE_RETURN = "\r".charCodeAt(0);
- var TAB = "\t".charCodeAt(0);
- var STRING_BUFFER_SIZE = 64 * 1024;
- function Parser() {
- this.tState = START;
- this.value = undefined;
- this.string = undefined; // string data
- this.stringBuffer = Buffer.alloc ? Buffer.alloc(STRING_BUFFER_SIZE) : new Buffer(STRING_BUFFER_SIZE);
- this.stringBufferOffset = 0;
- this.unicode = undefined; // unicode escapes
- this.highSurrogate = undefined;
- this.key = undefined;
- this.mode = undefined;
- this.stack = [];
- this.state = VALUE;
- this.bytes_remaining = 0; // number of bytes remaining in multi byte utf8 char to read after split boundary
- this.bytes_in_sequence = 0; // bytes in multi byte utf8 char to read
- this.temp_buffs = { "2": new Buffer(2), "3": new Buffer(3), "4": new Buffer(4) }; // for rebuilding chars split before boundary is reached
- // Stream offset
- this.offset = -1;
- }
- // Slow code to string converter (only used when throwing syntax errors)
- Parser.toknam = function (code) {
- var keys = Object.keys(C);
- for (var i = 0, l = keys.length; i < l; i++) {
- var key = keys[i];
- if (C[key] === code) { return key; }
- }
- return code && ("0x" + code.toString(16));
- };
- var proto = Parser.prototype;
- proto.onError = function (err) { throw err; };
- proto.charError = function (buffer, i) {
- this.tState = STOP;
- this.onError(new Error("Unexpected " + JSON.stringify(String.fromCharCode(buffer[i])) + " at position " + i + " in state " + Parser.toknam(this.tState)));
- };
- proto.appendStringChar = function (char) {
- if (this.stringBufferOffset >= STRING_BUFFER_SIZE) {
- this.string += this.stringBuffer.toString('utf8');
- this.stringBufferOffset = 0;
- }
- this.stringBuffer[this.stringBufferOffset++] = char;
- };
- proto.appendStringBuf = function (buf, start, end) {
- var size = buf.length;
- if (typeof start === 'number') {
- if (typeof end === 'number') {
- if (end < 0) {
- // adding a negative end decreeses the size
- size = buf.length - start + end;
- } else {
- size = end - start;
- }
- } else {
- size = buf.length - start;
- }
- }
- if (size < 0) {
- size = 0;
- }
- if (this.stringBufferOffset + size > STRING_BUFFER_SIZE) {
- this.string += this.stringBuffer.toString('utf8', 0, this.stringBufferOffset);
- this.stringBufferOffset = 0;
- }
- buf.copy(this.stringBuffer, this.stringBufferOffset, start, end);
- this.stringBufferOffset += size;
- };
- proto.write = function (buffer) {
- if (typeof buffer === "string") buffer = new Buffer(buffer);
- var n;
- for (var i = 0, l = buffer.length; i < l; i++) {
- if (this.tState === START){
- n = buffer[i];
- this.offset++;
- if(n === 0x7b){ this.onToken(LEFT_BRACE, "{"); // {
- }else if(n === 0x7d){ this.onToken(RIGHT_BRACE, "}"); // }
- }else if(n === 0x5b){ this.onToken(LEFT_BRACKET, "["); // [
- }else if(n === 0x5d){ this.onToken(RIGHT_BRACKET, "]"); // ]
- }else if(n === 0x3a){ this.onToken(COLON, ":"); // :
- }else if(n === 0x2c){ this.onToken(COMMA, ","); // ,
- }else if(n === 0x74){ this.tState = TRUE1; // t
- }else if(n === 0x66){ this.tState = FALSE1; // f
- }else if(n === 0x6e){ this.tState = NULL1; // n
- }else if(n === 0x22){ // "
- this.string = "";
- this.stringBufferOffset = 0;
- this.tState = STRING1;
- }else if(n === 0x2d){ this.string = "-"; this.tState = NUMBER1; // -
- }else{
- if (n >= 0x30 && n < 0x40) { // 1-9
- this.string = String.fromCharCode(n); this.tState = NUMBER3;
- } else if (n === 0x20 || n === 0x09 || n === 0x0a || n === 0x0d) {
- // whitespace
- } else {
- return this.charError(buffer, i);
- }
- }
- }else if (this.tState === STRING1){ // After open quote
- n = buffer[i]; // get current byte from buffer
- // check for carry over of a multi byte char split between data chunks
- // & fill temp buffer it with start of this data chunk up to the boundary limit set in the last iteration
- if (this.bytes_remaining > 0) {
- for (var j = 0; j < this.bytes_remaining; j++) {
- this.temp_buffs[this.bytes_in_sequence][this.bytes_in_sequence - this.bytes_remaining + j] = buffer[j];
- }
- this.appendStringBuf(this.temp_buffs[this.bytes_in_sequence]);
- this.bytes_in_sequence = this.bytes_remaining = 0;
- i = i + j - 1;
- } else if (this.bytes_remaining === 0 && n >= 128) { // else if no remainder bytes carried over, parse multi byte (>=128) chars one at a time
- if (n <= 193 || n > 244) {
- return this.onError(new Error("Invalid UTF-8 character at position " + i + " in state " + Parser.toknam(this.tState)));
- }
- if ((n >= 194) && (n <= 223)) this.bytes_in_sequence = 2;
- if ((n >= 224) && (n <= 239)) this.bytes_in_sequence = 3;
- if ((n >= 240) && (n <= 244)) this.bytes_in_sequence = 4;
- if ((this.bytes_in_sequence + i) > buffer.length) { // if bytes needed to complete char fall outside buffer length, we have a boundary split
- for (var k = 0; k <= (buffer.length - 1 - i); k++) {
- this.temp_buffs[this.bytes_in_sequence][k] = buffer[i + k]; // fill temp buffer of correct size with bytes available in this chunk
- }
- this.bytes_remaining = (i + this.bytes_in_sequence) - buffer.length;
- i = buffer.length - 1;
- } else {
- this.appendStringBuf(buffer, i, i + this.bytes_in_sequence);
- i = i + this.bytes_in_sequence - 1;
- }
- } else if (n === 0x22) {
- this.tState = START;
- this.string += this.stringBuffer.toString('utf8', 0, this.stringBufferOffset);
- this.stringBufferOffset = 0;
- this.onToken(STRING, this.string);
- this.offset += Buffer.byteLength(this.string, 'utf8') + 1;
- this.string = undefined;
- }
- else if (n === 0x5c) {
- this.tState = STRING2;
- }
- else if (n >= 0x20) { this.appendStringChar(n); }
- else {
- return this.charError(buffer, i);
- }
- }else if (this.tState === STRING2){ // After backslash
- n = buffer[i];
- if(n === 0x22){ this.appendStringChar(n); this.tState = STRING1;
- }else if(n === 0x5c){ this.appendStringChar(BACK_SLASH); this.tState = STRING1;
- }else if(n === 0x2f){ this.appendStringChar(FORWARD_SLASH); this.tState = STRING1;
- }else if(n === 0x62){ this.appendStringChar(BACKSPACE); this.tState = STRING1;
- }else if(n === 0x66){ this.appendStringChar(FORM_FEED); this.tState = STRING1;
- }else if(n === 0x6e){ this.appendStringChar(NEWLINE); this.tState = STRING1;
- }else if(n === 0x72){ this.appendStringChar(CARRIAGE_RETURN); this.tState = STRING1;
- }else if(n === 0x74){ this.appendStringChar(TAB); this.tState = STRING1;
- }else if(n === 0x75){ this.unicode = ""; this.tState = STRING3;
- }else{
- return this.charError(buffer, i);
- }
- }else if (this.tState === STRING3 || this.tState === STRING4 || this.tState === STRING5 || this.tState === STRING6){ // unicode hex codes
- n = buffer[i];
- // 0-9 A-F a-f
- if ((n >= 0x30 && n < 0x40) || (n > 0x40 && n <= 0x46) || (n > 0x60 && n <= 0x66)) {
- this.unicode += String.fromCharCode(n);
- if (this.tState++ === STRING6) {
- var intVal = parseInt(this.unicode, 16);
- this.unicode = undefined;
- if (this.highSurrogate !== undefined && intVal >= 0xDC00 && intVal < (0xDFFF + 1)) { //<56320,57343> - lowSurrogate
- this.appendStringBuf(new Buffer(String.fromCharCode(this.highSurrogate, intVal)));
- this.highSurrogate = undefined;
- } else if (this.highSurrogate === undefined && intVal >= 0xD800 && intVal < (0xDBFF + 1)) { //<55296,56319> - highSurrogate
- this.highSurrogate = intVal;
- } else {
- if (this.highSurrogate !== undefined) {
- this.appendStringBuf(new Buffer(String.fromCharCode(this.highSurrogate)));
- this.highSurrogate = undefined;
- }
- this.appendStringBuf(new Buffer(String.fromCharCode(intVal)));
- }
- this.tState = STRING1;
- }
- } else {
- return this.charError(buffer, i);
- }
- } else if (this.tState === NUMBER1 || this.tState === NUMBER3) {
- n = buffer[i];
- switch (n) {
- case 0x30: // 0
- case 0x31: // 1
- case 0x32: // 2
- case 0x33: // 3
- case 0x34: // 4
- case 0x35: // 5
- case 0x36: // 6
- case 0x37: // 7
- case 0x38: // 8
- case 0x39: // 9
- case 0x2e: // .
- case 0x65: // e
- case 0x45: // E
- case 0x2b: // +
- case 0x2d: // -
- this.string += String.fromCharCode(n);
- this.tState = NUMBER3;
- break;
- default:
- this.tState = START;
- var result = Number(this.string);
- if (isNaN(result)){
- return this.charError(buffer, i);
- }
- if ((this.string.match(/[0-9]+/) == this.string) && (result.toString() != this.string)) {
- // Long string of digits which is an ID string and not valid and/or safe JavaScript integer Number
- this.onToken(STRING, this.string);
- } else {
- this.onToken(NUMBER, result);
- }
- this.offset += this.string.length - 1;
- this.string = undefined;
- i--;
- break;
- }
- }else if (this.tState === TRUE1){ // r
- if (buffer[i] === 0x72) { this.tState = TRUE2; }
- else { return this.charError(buffer, i); }
- }else if (this.tState === TRUE2){ // u
- if (buffer[i] === 0x75) { this.tState = TRUE3; }
- else { return this.charError(buffer, i); }
- }else if (this.tState === TRUE3){ // e
- if (buffer[i] === 0x65) { this.tState = START; this.onToken(TRUE, true); this.offset+= 3; }
- else { return this.charError(buffer, i); }
- }else if (this.tState === FALSE1){ // a
- if (buffer[i] === 0x61) { this.tState = FALSE2; }
- else { return this.charError(buffer, i); }
- }else if (this.tState === FALSE2){ // l
- if (buffer[i] === 0x6c) { this.tState = FALSE3; }
- else { return this.charError(buffer, i); }
- }else if (this.tState === FALSE3){ // s
- if (buffer[i] === 0x73) { this.tState = FALSE4; }
- else { return this.charError(buffer, i); }
- }else if (this.tState === FALSE4){ // e
- if (buffer[i] === 0x65) { this.tState = START; this.onToken(FALSE, false); this.offset+= 4; }
- else { return this.charError(buffer, i); }
- }else if (this.tState === NULL1){ // u
- if (buffer[i] === 0x75) { this.tState = NULL2; }
- else { return this.charError(buffer, i); }
- }else if (this.tState === NULL2){ // l
- if (buffer[i] === 0x6c) { this.tState = NULL3; }
- else { return this.charError(buffer, i); }
- }else if (this.tState === NULL3){ // l
- if (buffer[i] === 0x6c) { this.tState = START; this.onToken(NULL, null); this.offset += 3; }
- else { return this.charError(buffer, i); }
- }
- }
- };
- proto.onToken = function (token, value) {
- // Override this to get events
- };
- proto.parseError = function (token, value) {
- this.tState = STOP;
- this.onError(new Error("Unexpected " + Parser.toknam(token) + (value ? ("(" + JSON.stringify(value) + ")") : "") + " in state " + Parser.toknam(this.state)));
- };
- proto.push = function () {
- this.stack.push({value: this.value, key: this.key, mode: this.mode});
- };
- proto.pop = function () {
- var value = this.value;
- var parent = this.stack.pop();
- this.value = parent.value;
- this.key = parent.key;
- this.mode = parent.mode;
- this.emit(value);
- if (!this.mode) { this.state = VALUE; }
- };
- proto.emit = function (value) {
- if (this.mode) { this.state = COMMA; }
- this.onValue(value);
- };
- proto.onValue = function (value) {
- // Override me
- };
- proto.onToken = function (token, value) {
- if(this.state === VALUE){
- if(token === STRING || token === NUMBER || token === TRUE || token === FALSE || token === NULL){
- if (this.value) {
- this.value[this.key] = value;
- }
- this.emit(value);
- }else if(token === LEFT_BRACE){
- this.push();
- if (this.value) {
- this.value = this.value[this.key] = {};
- } else {
- this.value = {};
- }
- this.key = undefined;
- this.state = KEY;
- this.mode = OBJECT;
- }else if(token === LEFT_BRACKET){
- this.push();
- if (this.value) {
- this.value = this.value[this.key] = [];
- } else {
- this.value = [];
- }
- this.key = 0;
- this.mode = ARRAY;
- this.state = VALUE;
- }else if(token === RIGHT_BRACE){
- if (this.mode === OBJECT) {
- this.pop();
- } else {
- return this.parseError(token, value);
- }
- }else if(token === RIGHT_BRACKET){
- if (this.mode === ARRAY) {
- this.pop();
- } else {
- return this.parseError(token, value);
- }
- }else{
- return this.parseError(token, value);
- }
- }else if(this.state === KEY){
- if (token === STRING) {
- this.key = value;
- this.state = COLON;
- } else if (token === RIGHT_BRACE) {
- this.pop();
- } else {
- return this.parseError(token, value);
- }
- }else if(this.state === COLON){
- if (token === COLON) { this.state = VALUE; }
- else { return this.parseError(token, value); }
- }else if(this.state === COMMA){
- if (token === COMMA) {
- if (this.mode === ARRAY) { this.key++; this.state = VALUE; }
- else if (this.mode === OBJECT) { this.state = KEY; }
- } else if (token === RIGHT_BRACKET && this.mode === ARRAY || token === RIGHT_BRACE && this.mode === OBJECT) {
- this.pop();
- } else {
- return this.parseError(token, value);
- }
- }else{
- return this.parseError(token, value);
- }
- };
- Parser.C = C;
- module.exports = Parser;
|