json-sax-parser.ts

// inspired by https://gist.github.com/creationix/1821394
// from Tim Caswell

import { EventEmitter } from 'events';

interface ParserEvents {
	StartObject: () => void;
	EndObject: () => void;
	StartArray: () => void;
	EndArray: () => void;
	Colon: () => void;
	Comma: () => void;
	String: (str: string) => void;
	Boolean: (bool: boolean) => void;
	Null: () => void;
	Number: (num: number) => void;
	Error: (err: Error) => void;
}

declare interface SaxParser {
	on<U extends keyof ParserEvents>(event: U, listener: ParserEvents[U]): this;

	emit<U extends keyof ParserEvents>(
		event: U,
		...args: Parameters<ParserEvents[U]>
	): boolean;
}

// Named constants with unique integer values
const C = {
	START: 0x11,
	TRUE1: 0x21,
	TRUE2: 0x22,
	TRUE3: 0x23,
	FALSE1: 0x31,
	FALSE2: 0x32,
	FALSE3: 0x33,
	FALSE4: 0x34,
	NULL1: 0x41,
	NULL3: 0x42,
	NULL2: 0x43,
	NUMBER1: 0x51,
	NUMBER2: 0x52,
	NUMBER3: 0x53,
	NUMBER4: 0x54,
	NUMBER5: 0x55,
	NUMBER6: 0x56,
	NUMBER7: 0x57,
	NUMBER8: 0x58,
	STRING1: 0x61,
	STRING2: 0x62,
	STRING3: 0x63,
	STRING4: 0x64,
	STRING5: 0x65,
	STRING6: 0x66
};

// slow to string converter (only used when throwing syntax errors)
function toknam(code: number) {
	let k: keyof typeof C;
	for (k in C) {
		if (C[k] === code) {
			return k;
		}
	}

	return code && '0x' + code.toString(16);
}

class SaxParser extends EventEmitter {
	state = C.START;

	// for string parsing
	string?: string; // string data
	unicode?: string; // unicode escapes

	// For number parsing
	negative?: boolean;
	magnatude?: number;
	position?: number;
	exponent?: number;
	negativeExponent?: boolean;

	charError(buffer: Buffer, i: number) {
		this.emit(
			'Error',
			new Error(
				'Unexpected ' +
					JSON.stringify(String.fromCharCode(buffer[i])) +
					' at position ' +
					i +
					' in state ' +
					toknam(this.state)
			)
		);
	}

	parse(buffer: Buffer) {
		if (typeof buffer === 'string') buffer = new Buffer(buffer);
		var n;
		for (var i = 0, l = buffer.length; i < l; i++) {
			switch (this.state) {
				case C.START:
					n = buffer[i];
					switch (n) {
						case 0x7b: // `{`
							this.emit('StartObject');
							continue;
						case 0x7d: // `}`
							this.emit('EndObject');
							continue;
						case 0x5b: // `[`
							this.emit('StartArray');
							continue;
						case 0x5d: // `]`
							this.emit('EndArray');
							continue;
						case 0x3a: // `:`
							this.emit('Colon');
							continue;
						case 0x2c: // `,`
							this.emit('Comma');
							continue;
						case 0x74: // `t`
							this.state = C.TRUE1;
							continue;
						case 0x66: // `f`
							this.state = C.FALSE1;
							continue;
						case 0x6e: // `n`
							this.state = C.NULL1;
							continue;
						case 0x22: // `"`
							this.string = '';
							this.state = C.STRING1;
							continue;
						case 0x2d: // `-`
							this.negative = true;
							this.state = C.NUMBER1;
							continue;
						case 0x30: // `0`
							this.magnatude = 0;
							this.state = C.NUMBER2;
							continue;
					}
					if (n > 0x30 && n < 0x40) {
						// 1-9
						this.magnatude = n - 0x30;
						this.state = C.NUMBER3;
						continue;
					}
					if (n === 0x20 || n === 0x09 || n === 0x0a || n === 0x0d) {
						continue; // whitespace
					}
					this.charError(buffer, i);
				case C.STRING1: // After open quote
					n = buffer[i];
					switch (n) {
						case 0x22: // `"`
							this.emit('String', this.string!);
							this.string = undefined;
							this.state = C.START;
							continue;
						case 0x5c: // `\`
							this.state = C.STRING2;
							continue;
					}
					if (n >= 0x20) {
						this.string += String.fromCharCode(n);
						continue;
					}
					this.charError(buffer, i);
				case C.STRING2: // After backslash
					n = buffer[i];
					switch (n) {
						case 0x22:
							this.string += '"';
							this.state = C.STRING1;
							continue;
						case 0x5c:
							this.string += '\\';
							this.state = C.STRING1;
							continue;
						case 0x2f:
							this.string += '/';
							this.state = C.STRING1;
							continue;
						case 0x62:
							this.string += '\b';
							this.state = C.STRING1;
							continue;
						case 0x66:
							this.string += '\f';
							this.state = C.STRING1;
							continue;
						case 0x6e:
							this.string += '\n';
							this.state = C.STRING1;
							continue;
						case 0x72:
							this.string += '\r';
							this.state = C.STRING1;
							continue;
						case 0x74:
							this.string += '\t';
							this.state = C.STRING1;
							continue;
						case 0x75:
							this.unicode = '';
							this.state = C.STRING3;
							continue;
					}
					this.charError(buffer, i);
				case C.STRING3:
				case C.STRING4:
				case C.STRING5:
				case C.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.state++ === C.STRING6) {
							this.string += String.fromCharCode(parseInt(this.unicode!, 16));
							this.unicode = undefined;
							this.state = C.STRING1;
						}
						continue;
					}
					this.charError(buffer, i);
				case C.NUMBER1: // after minus
					n = buffer[i];
					if (n === 0x30) {
						// `0`
						this.magnatude = 0;
						this.state = C.NUMBER2;
						continue;
					}
					if (n > 0x30 && n < 0x40) {
						// `1`-`9`
						this.magnatude = n - 0x30;
						this.state = C.NUMBER3;
						continue;
					}
					this.charError(buffer, i);
				case C.NUMBER2: // * After initial zero
					switch (buffer[i]) {
						case 0x2e: // .
							this.position = 0.1;
							this.state = C.NUMBER4;
							continue;
						case 0x65:
						case 0x45: // e/E
							this.exponent = 0;
							this.state = C.NUMBER6;
							continue;
					}
					this.finish();
					i--; // rewind to re-check this char
					continue;
				case C.NUMBER3: // * After digit (before period)
					n = buffer[i];
					switch (n) {
						case 0x2e: // .
							this.position = 0.1;
							this.state = C.NUMBER4;
							continue;
						case 0x65:
						case 0x45: // e/E
							this.exponent = 0;
							this.state = C.NUMBER6;
							continue;
					}
					if (n >= 0x30 && n < 0x40) {
						// 0-9
						this.magnatude = this.magnatude! * 10 + (n - 0x30);
						continue;
					}
					this.finish();
					i--; // rewind to re-check
					continue;
				case C.NUMBER4: // After period
					n = buffer[i];
					if (n >= 0x30 && n < 0x40) {
						// 0-9
						this.magnatude! += this.position! * (n - 0x30);
						this.position! /= 10;
						this.state = C.NUMBER5;
						continue;
					}
					this.charError(buffer, i);
				case C.NUMBER5: // * After digit (after period)
					n = buffer[i];
					if (n >= 0x30 && n < 0x40) {
						// 0-9
						this.magnatude! += this.position! * (n - 0x30);
						this.position! /= 10;
						continue;
					}
					if (n === 0x65 || n === 0x45) {
						// E/e
						this.exponent = 0;
						this.state = C.NUMBER6;
						continue;
					}
					this.finish();
					i--; // rewind
					continue;
				case C.NUMBER6: // After E
					n = buffer[i];
					if (n === 0x2b || n === 0x2d) {
						// +/-
						if (n === 0x2d) {
							this.negativeExponent = true;
						}
						this.state = C.NUMBER7;
						continue;
					}
					if (n >= 0x30 && n < 0x40) {
						this.exponent = this.exponent! * 10 + (n - 0x30);
						this.state = C.NUMBER8;
						continue;
					}
					this.charError(buffer, i);
				case C.NUMBER7: // After +/-
					n = buffer[i];
					if (n >= 0x30 && n < 0x40) {
						// 0-9
						this.exponent = this.exponent! * 10 + (n - 0x30);
						this.state = C.NUMBER8;
						continue;
					}
					this.charError(buffer, i);
				case C.NUMBER8: // * After digit (after +/-)
					n = buffer[i];
					if (n >= 0x30 && n < 0x40) {
						// 0-9
						this.exponent = this.exponent! * 10 + (n - 0x30);
						continue;
					}
					this.finish();
					i--;
					continue;
				case C.TRUE1: // r
					if (buffer[i] === 0x72) {
						this.state = C.TRUE2;
						continue;
					}
					this.charError(buffer, i);
				case C.TRUE2: // u
					if (buffer[i] === 0x75) {
						this.state = C.TRUE3;
						continue;
					}
					this.charError(buffer, i);
				case C.TRUE3: // e
					if (buffer[i] === 0x65) {
						this.state = C.START;
						this.emit('Boolean', true);
						continue;
					}
					this.charError(buffer, i);
				case C.FALSE1: // a
					if (buffer[i] === 0x61) {
						this.state = C.FALSE2;
						continue;
					}
					this.charError(buffer, i);
				case C.FALSE2: // l
					if (buffer[i] === 0x6c) {
						this.state = C.FALSE3;
						continue;
					}
					this.charError(buffer, i);
				case C.FALSE3: // s
					if (buffer[i] === 0x73) {
						this.state = C.FALSE4;
						continue;
					}
					this.charError(buffer, i);
				case C.FALSE4: // e
					if (buffer[i] === 0x65) {
						this.state = C.START;
						this.emit('Boolean', false);
						continue;
					}
					this.charError(buffer, i);
				case C.NULL1: // u
					if (buffer[i] === 0x75) {
						this.state = C.NULL2;
						continue;
					}
					this.charError(buffer, i);
				case C.NULL2: // l
					if (buffer[i] === 0x6c) {
						this.state = C.NULL3;
						continue;
					}
					this.charError(buffer, i);
				case C.NULL3: // l
					if (buffer[i] === 0x6c) {
						this.state = C.START;
						this.emit('Null');
						continue;
					}
					this.charError(buffer, i);
			}
		}
	}

	finish() {
		switch (this.state) {
			case C.NUMBER2: // * After initial zero
				this.emit('Number', 0);
				this.state = C.START;
				this.magnatude = undefined;
				this.negative = undefined;
				break;
			case C.NUMBER3: // * After digit (before period)
				this.state = C.START;
				if (this.negative) {
					this.magnatude = -this.magnatude!;
					this.negative = undefined;
				}
				this.emit('Number', this.magnatude!);
				this.magnatude = undefined;
				break;
			case C.NUMBER5: // * After digit (after period)
				this.state = C.START;
				if (this.negative) {
					this.magnatude = -this.magnatude!;
					this.negative = undefined;
				}
				this.emit('Number', this.negative ? -this.magnatude! : this.magnatude!);
				this.magnatude = undefined;
				this.position = undefined;
				break;
			case C.NUMBER8: // * After digit (after +/-)
				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.state = C.START;
				this.emit('Number', this.magnatude!);
				this.magnatude = undefined;
				break;
		}
		if (this.state !== C.START) {
			this.emit('Error', new Error('Unexpected end of input stream'));
		}
	}
}

export default SaxParser;