mirror of
https://github.com/thangisme/notes.git
synced 2024-10-31 22:17:25 -04:00
402 lines
15 KiB
JavaScript
402 lines
15 KiB
JavaScript
/*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 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.NULL3 = 0x42;
|
|
var NULL3 = C.NULL2 = 0x43;
|
|
var NUMBER1 = C.NUMBER1 = 0x51;
|
|
var NUMBER2 = C.NUMBER2 = 0x52;
|
|
var NUMBER3 = C.NUMBER3 = 0x53;
|
|
var NUMBER4 = C.NUMBER4 = 0x54;
|
|
var NUMBER5 = C.NUMBER5 = 0x55;
|
|
var NUMBER6 = C.NUMBER6 = 0x56;
|
|
var NUMBER7 = C.NUMBER7 = 0x57;
|
|
var NUMBER8 = C.NUMBER8 = 0x58;
|
|
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;
|
|
|
|
// Slow code to string converter (only used when throwing syntax errors)
|
|
function toknam(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));
|
|
}
|
|
|
|
|
|
function Parser() {
|
|
this.tState = START;
|
|
this.value = undefined;
|
|
|
|
this.string = undefined; // string data
|
|
this.unicode = undefined; // unicode escapes
|
|
|
|
// For number parsing
|
|
this.negative = undefined;
|
|
this.magnatude = undefined;
|
|
this.position = undefined;
|
|
this.exponent = undefined;
|
|
this.negativeExponent = 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
|
|
}
|
|
var proto = Parser.prototype;
|
|
proto.charError = function (buffer, i) {
|
|
this.onError(new Error("Unexpected " + JSON.stringify(String.fromCharCode(buffer[i])) + " at position " + i + " in state " + toknam(this.tState)));
|
|
};
|
|
proto.onError = function (err) { throw err; };
|
|
proto.write = function (buffer) {
|
|
if (typeof buffer === "string") buffer = new Buffer(buffer);
|
|
//process.stdout.write("Input: ");
|
|
//console.dir(buffer.toString());
|
|
var n;
|
|
for (var i = 0, l = buffer.length; i < l; i++) {
|
|
if (this.tState === START){
|
|
n = buffer[i];
|
|
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.tState = STRING1; // "
|
|
}else if(n === 0x2d){ this.negative = true; this.tState = NUMBER1; // -
|
|
}else if(n === 0x30){ this.magnatude = 0; this.tState = NUMBER2; // 0
|
|
}else{
|
|
if (n > 0x30 && n < 0x40) { // 1-9
|
|
this.magnatude = n - 0x30; this.tState = NUMBER3;
|
|
} else if (n === 0x20 || n === 0x09 || n === 0x0a || n === 0x0d) {
|
|
// whitespace
|
|
} else { 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.string += this.temp_buffs[this.bytes_in_sequence].toString();
|
|
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 >= 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.string += buffer.slice(i, (i + this.bytes_in_sequence)).toString();
|
|
i = i + this.bytes_in_sequence - 1;
|
|
}
|
|
} else if (n === 0x22) { this.tState = START; this.onToken(STRING, this.string); this.string = undefined; }
|
|
else if (n === 0x5c) { this.tState = STRING2; }
|
|
else if (n >= 0x20) { this.string += String.fromCharCode(n); }
|
|
else { this.charError(buffer, i); }
|
|
}else if (this.tState === STRING2){ // After backslash
|
|
n = buffer[i];
|
|
if(n === 0x22){ this.string += "\""; this.tState = STRING1;
|
|
}else if(n === 0x5c){ this.string += "\\"; this.tState = STRING1;
|
|
}else if(n === 0x2f){ this.string += "\/"; this.tState = STRING1;
|
|
}else if(n === 0x62){ this.string += "\b"; this.tState = STRING1;
|
|
}else if(n === 0x66){ this.string += "\f"; this.tState = STRING1;
|
|
}else if(n === 0x6e){ this.string += "\n"; this.tState = STRING1;
|
|
}else if(n === 0x72){ this.string += "\r"; this.tState = STRING1;
|
|
}else if(n === 0x74){ this.string += "\t"; this.tState = STRING1;
|
|
}else if(n === 0x75){ this.unicode = ""; this.tState = STRING3;
|
|
}else{
|
|
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) {
|
|
this.string += String.fromCharCode(parseInt(this.unicode, 16));
|
|
this.unicode = undefined;
|
|
this.tState = STRING1;
|
|
}
|
|
} else {
|
|
this.charError(buffer, i);
|
|
}
|
|
}else if (this.tState === NUMBER1){ // after minus
|
|
n = buffer[i];
|
|
if (n === 0x30) { this.magnatude = 0; this.tState = NUMBER2; }
|
|
else if (n > 0x30 && n < 0x40) { this.magnatude = n - 0x30; this.tState = NUMBER3; }
|
|
else { this.charError(buffer, i); }
|
|
}else if (this.tState === NUMBER2){ // * After initial zero
|
|
n = buffer[i];
|
|
if(n === 0x2e){ // .
|
|
this.position = 0.1; this.tState = NUMBER4;
|
|
}else if(n === 0x65 || n === 0x45){ // e/E
|
|
this.exponent = 0; this.tState = NUMBER6;
|
|
}else{
|
|
this.tState = START;
|
|
this.onToken(NUMBER, 0);
|
|
this.magnatude = undefined;
|
|
this.negative = undefined;
|
|
i--;
|
|
}
|
|
}else if (this.tState === NUMBER3){ // * After digit (before period)
|
|
n = buffer[i];
|
|
if(n === 0x2e){ // .
|
|
this.position = 0.1; this.tState = NUMBER4;
|
|
}else if(n === 0x65 || n === 0x45){ // e/E
|
|
this.exponent = 0; this.tState = NUMBER6;
|
|
}else{
|
|
if (n >= 0x30 && n < 0x40) { this.magnatude = this.magnatude * 10 + n - 0x30; }
|
|
else {
|
|
this.tState = START;
|
|
if (this.negative) {
|
|
this.magnatude = -this.magnatude;
|
|
this.negative = undefined;
|
|
}
|
|
this.onToken(NUMBER, this.magnatude);
|
|
this.magnatude = undefined;
|
|
i--;
|
|
}
|
|
}
|
|
}else if (this.tState === NUMBER4){ // After period
|
|
n = buffer[i];
|
|
if (n >= 0x30 && n < 0x40) { // 0-9
|
|
this.magnatude += this.position * (n - 0x30);
|
|
this.position /= 10;
|
|
this.tState = NUMBER5;
|
|
} else { this.charError(buffer, i); }
|
|
}else if (this.tState === NUMBER5){ // * After digit (after period)
|
|
n = buffer[i];
|
|
if (n >= 0x30 && n < 0x40) { // 0-9
|
|
this.magnatude += this.position * (n - 0x30);
|
|
this.position /= 10;
|
|
}
|
|
else if (n === 0x65 || n === 0x45) { this.exponent = 0; this.tState = NUMBER6; } // E/e
|
|
else {
|
|
this.tState = START;
|
|
if (this.negative) {
|
|
this.magnatude = -this.magnatude;
|
|
this.negative = undefined;
|
|
}
|
|
this.onToken(NUMBER, this.negative ? -this.magnatude : this.magnatude);
|
|
this.magnatude = undefined;
|
|
this.position = undefined;
|
|
i--;
|
|
}
|
|
}else if (this.tState === NUMBER6){ // After E
|
|
n = buffer[i];
|
|
if (n === 0x2b || n === 0x2d) { // +/-
|
|
if (n === 0x2d) { this.negativeExponent = true; }
|
|
this.tState = NUMBER7;
|
|
}
|
|
else if (n >= 0x30 && n < 0x40) {
|
|
this.exponent = this.exponent * 10 + (n - 0x30);
|
|
this.tState = NUMBER8;
|
|
}
|
|
else { this.charError(buffer, i); }
|
|
}else if (this.tState === NUMBER7){ // After +/-
|
|
n = buffer[i];
|
|
if (n >= 0x30 && n < 0x40) { // 0-9
|
|
this.exponent = this.exponent * 10 + (n - 0x30);
|
|
this.tState = NUMBER8;
|
|
}
|
|
else { this.charError(buffer, i); }
|
|
}else if (this.tState === NUMBER8){ // * After digit (after +/-)
|
|
n = buffer[i];
|
|
if (n >= 0x30 && n < 0x40) { // 0-9
|
|
this.exponent = this.exponent * 10 + (n - 0x30);
|
|
}
|
|
else {
|
|
if (this.negativeExponent) {
|
|
this.exponent = -this.exponent;
|
|
this.negativeExponent = undefined;
|
|
}
|
|
this.magnatude *= Math.pow(10, this.exponent);
|
|
this.exponent = undefined;
|
|
if (this.negative) {
|
|
this.magnatude = -this.magnatude;
|
|
this.negative = undefined;
|
|
}
|
|
this.tState = START;
|
|
this.onToken(NUMBER, this.magnatude);
|
|
this.magnatude = undefined;
|
|
i--;
|
|
}
|
|
}else if (this.tState === TRUE1){ // r
|
|
if (buffer[i] === 0x72) { this.tState = TRUE2; }
|
|
else { this.charError(buffer, i); }
|
|
}else if (this.tState === TRUE2){ // u
|
|
if (buffer[i] === 0x75) { this.tState = TRUE3; }
|
|
else { this.charError(buffer, i); }
|
|
}else if (this.tState === TRUE3){ // e
|
|
if (buffer[i] === 0x65) { this.tState = START; this.onToken(TRUE, true); }
|
|
else { this.charError(buffer, i); }
|
|
}else if (this.tState === FALSE1){ // a
|
|
if (buffer[i] === 0x61) { this.tState = FALSE2; }
|
|
else { this.charError(buffer, i); }
|
|
}else if (this.tState === FALSE2){ // l
|
|
if (buffer[i] === 0x6c) { this.tState = FALSE3; }
|
|
else { this.charError(buffer, i); }
|
|
}else if (this.tState === FALSE3){ // s
|
|
if (buffer[i] === 0x73) { this.tState = FALSE4; }
|
|
else { this.charError(buffer, i); }
|
|
}else if (this.tState === FALSE4){ // e
|
|
if (buffer[i] === 0x65) { this.tState = START; this.onToken(FALSE, false); }
|
|
else { this.charError(buffer, i); }
|
|
}else if (this.tState === NULL1){ // u
|
|
if (buffer[i] === 0x75) { this.tState = NULL2; }
|
|
else { this.charError(buffer, i); }
|
|
}else if (this.tState === NULL2){ // l
|
|
if (buffer[i] === 0x6c) { this.tState = NULL3; }
|
|
else { this.charError(buffer, i); }
|
|
}else if (this.tState === NULL3){ // l
|
|
if (buffer[i] === 0x6c) { this.tState = START; this.onToken(NULL, null); }
|
|
else { this.charError(buffer, i); }
|
|
}
|
|
}
|
|
};
|
|
proto.onToken = function (token, value) {
|
|
// Override this to get events
|
|
};
|
|
|
|
proto.parseError = function (token, value) {
|
|
this.onError(new Error("Unexpected " + toknam(token) + (value ? ("(" + JSON.stringify(value) + ")") : "") + " in state " + toknam(this.state)));
|
|
};
|
|
proto.onError = function (err) { throw err; };
|
|
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) {
|
|
//console.log("OnToken: state=%s token=%s %s", toknam(this.state), toknam(token), value?JSON.stringify(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 {
|
|
this.parseError(token, value);
|
|
}
|
|
}else if(token === RIGHT_BRACKET){
|
|
if (this.mode === ARRAY) {
|
|
this.pop();
|
|
} else {
|
|
this.parseError(token, value);
|
|
}
|
|
}else{
|
|
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 {
|
|
this.parseError(token, value);
|
|
}
|
|
}else if(this.state === COLON){
|
|
if (token === COLON) { this.state = VALUE; }
|
|
else { 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 {
|
|
this.parseError(token, value);
|
|
}
|
|
}else{
|
|
this.parseError(token, value);
|
|
}
|
|
};
|
|
|
|
module.exports = Parser;
|