substance/vendor/htmlparser2.js

Substance is a JavaScript library for web-based content editing. It provides building blocks for realizing custom text editors and web-based publishing system. It is developed to power our online editing platform [Substance](http://substance.io).

import { decode_codepoint } from './entities';
import { entities } from './entities';
import { legacy } from './entities';
import { xml } from './entities';

var Tokenizer_1 = Tokenizer;

var i = 0,

    TEXT                      = i++,
    BEFORE_TAG_NAME           = i++, //after <
    IN_TAG_NAME               = i++,
    IN_SELF_CLOSING_TAG       = i++,
    BEFORE_CLOSING_TAG_NAME   = i++,
    IN_CLOSING_TAG_NAME       = i++,
    AFTER_CLOSING_TAG_NAME    = i++,

    //attributes
    BEFORE_ATTRIBUTE_NAME     = i++,
    IN_ATTRIBUTE_NAME         = i++,
    AFTER_ATTRIBUTE_NAME      = i++,
    BEFORE_ATTRIBUTE_VALUE    = i++,
    IN_ATTRIBUTE_VALUE_DQ     = i++, // "
    IN_ATTRIBUTE_VALUE_SQ     = i++, // '
    IN_ATTRIBUTE_VALUE_NQ     = i++,

    //declarations
    BEFORE_DECLARATION        = i++, // !
    IN_DECLARATION            = i++,

    //processing instructions
    IN_PROCESSING_INSTRUCTION = i++, // ?

    //comments
    BEFORE_COMMENT            = i++,
    IN_COMMENT                = i++,
    AFTER_COMMENT_1           = i++,
    AFTER_COMMENT_2           = i++,

    //cdata
    BEFORE_CDATA_1            = i++, // [
    BEFORE_CDATA_2            = i++, // C
    BEFORE_CDATA_3            = i++, // D
    BEFORE_CDATA_4            = i++, // A
    BEFORE_CDATA_5            = i++, // T
    BEFORE_CDATA_6            = i++, // A
    IN_CDATA                  = i++, // [
    AFTER_CDATA_1             = i++, // ]
    AFTER_CDATA_2             = i++, // ]

    //special tags
    BEFORE_SPECIAL            = i++, //S
    BEFORE_SPECIAL_END        = i++,   //S

    BEFORE_SCRIPT_1           = i++, //C
    BEFORE_SCRIPT_2           = i++, //R
    BEFORE_SCRIPT_3           = i++, //I
    BEFORE_SCRIPT_4           = i++, //P
    BEFORE_SCRIPT_5           = i++, //T
    AFTER_SCRIPT_1            = i++, //C
    AFTER_SCRIPT_2            = i++, //R
    AFTER_SCRIPT_3            = i++, //I
    AFTER_SCRIPT_4            = i++, //P
    AFTER_SCRIPT_5            = i++, //T

    BEFORE_STYLE_1            = i++, //T
    BEFORE_STYLE_2            = i++, //Y
    BEFORE_STYLE_3            = i++, //L
    BEFORE_STYLE_4            = i++, //E
    AFTER_STYLE_1             = i++, //T
    AFTER_STYLE_2             = i++, //Y
    AFTER_STYLE_3             = i++, //L
    AFTER_STYLE_4             = i++, //E

    BEFORE_ENTITY             = i++, //&
    BEFORE_NUMERIC_ENTITY     = i++, //#
    IN_NAMED_ENTITY           = i++,
    IN_NUMERIC_ENTITY         = i++,
    IN_HEX_ENTITY             = i++, //X

    j = 0,

    SPECIAL_NONE              = j++,
    SPECIAL_SCRIPT            = j++,
    SPECIAL_STYLE             = j++;

function whitespace(c){
	return c === " " || c === "\n" || c === "\t" || c === "\f" || c === "\r";
}

function characterState(char, SUCCESS){
	return function(c){
		if(c === char) this._state = SUCCESS;
	};
}

function ifElseState(upper, SUCCESS, FAILURE){
	var lower = upper.toLowerCase();

	if(upper === lower){
		return function(c){
			if(c === lower){
				this._state = SUCCESS;
			} else {
				this._state = FAILURE;
				this._index--;
			}
		};
	} else {
		return function(c){
			if(c === lower || c === upper){
				this._state = SUCCESS;
			} else {
				this._state = FAILURE;
				this._index--;
			}
		};
	}
}

function consumeSpecialNameChar(upper, NEXT_STATE){
	var lower = upper.toLowerCase();

	return function(c){
		if(c === lower || c === upper){
			this._state = NEXT_STATE;
		} else {
			this._state = IN_TAG_NAME;
			this._index--; //consume the token again
		}
	};
}

function Tokenizer(options, cbs){
	this._state = TEXT;
	this._buffer = "";
	this._sectionStart = 0;
	this._index = 0;
	this._bufferOffset = 0; //chars removed from _buffer
	this._baseState = TEXT;
	this._special = SPECIAL_NONE;
	this._cbs = cbs;
	this._running = true;
	this._ended = false;
	this._xmlMode = !!(options && options.xmlMode);
	this._decodeEntities = !!(options && options.decodeEntities);
}

Tokenizer.prototype._stateText = function(c){
	if(c === "<"){
		if(this._index > this._sectionStart){
			this._cbs.ontext(this._getSection());
		}
		this._state = BEFORE_TAG_NAME;
		this._sectionStart = this._index;
	} else if(this._decodeEntities && this._special === SPECIAL_NONE && c === "&"){
		if(this._index > this._sectionStart){
			this._cbs.ontext(this._getSection());
		}
		this._baseState = TEXT;
		this._state = BEFORE_ENTITY;
		this._sectionStart = this._index;
	}
};

Tokenizer.prototype._stateBeforeTagName = function(c){
	if(c === "/"){
		this._state = BEFORE_CLOSING_TAG_NAME;
	} else if(c === "<"){
		this._cbs.ontext(this._getSection());
		this._sectionStart = this._index;
	} else if(c === ">" || this._special !== SPECIAL_NONE || whitespace(c)) {
		this._state = TEXT;
	} else if(c === "!"){
		this._state = BEFORE_DECLARATION;
		this._sectionStart = this._index + 1;
	} else if(c === "?"){
		this._state = IN_PROCESSING_INSTRUCTION;
		this._sectionStart = this._index + 1;
	} else {
		this._state = (!this._xmlMode && (c === "s" || c === "S")) ?
						BEFORE_SPECIAL : IN_TAG_NAME;
		this._sectionStart = this._index;
	}
};

Tokenizer.prototype._stateInTagName = function(c){
	if(c === "/" || c === ">" || whitespace(c)){
		this._emitToken("onopentagname");
		this._state = BEFORE_ATTRIBUTE_NAME;
		this._index--;
	}
};

Tokenizer.prototype._stateBeforeCloseingTagName = function(c){
	if(whitespace(c));
	else if(c === ">"){
		this._state = TEXT;
	} else if(this._special !== SPECIAL_NONE){
		if(c === "s" || c === "S"){
			this._state = BEFORE_SPECIAL_END;
		} else {
			this._state = TEXT;
			this._index--;
		}
	} else {
		this._state = IN_CLOSING_TAG_NAME;
		this._sectionStart = this._index;
	}
};

Tokenizer.prototype._stateInCloseingTagName = function(c){
	if(c === ">" || whitespace(c)){
		this._emitToken("onclosetag");
		this._state = AFTER_CLOSING_TAG_NAME;
		this._index--;
	}
};

Tokenizer.prototype._stateAfterCloseingTagName = function(c){
	//skip everything until ">"
	if(c === ">"){
		this._state = TEXT;
		this._sectionStart = this._index + 1;
	}
};

Tokenizer.prototype._stateBeforeAttributeName = function(c){
	if(c === ">"){
		this._cbs.onopentagend();
		this._state = TEXT;
		this._sectionStart = this._index + 1;
	} else if(c === "/"){
		this._state = IN_SELF_CLOSING_TAG;
	} else if(!whitespace(c)){
		this._state = IN_ATTRIBUTE_NAME;
		this._sectionStart = this._index;
	}
};

Tokenizer.prototype._stateInSelfClosingTag = function(c){
	if(c === ">"){
		this._cbs.onselfclosingtag();
		this._state = TEXT;
		this._sectionStart = this._index + 1;
	} else if(!whitespace(c)){
		this._state = BEFORE_ATTRIBUTE_NAME;
		this._index--;
	}
};

Tokenizer.prototype._stateInAttributeName = function(c){
	if(c === "=" || c === "/" || c === ">" || whitespace(c)){
		this._cbs.onattribname(this._getSection());
		this._sectionStart = -1;
		this._state = AFTER_ATTRIBUTE_NAME;
		this._index--;
	}
};

Tokenizer.prototype._stateAfterAttributeName = function(c){
	if(c === "="){
		this._state = BEFORE_ATTRIBUTE_VALUE;
	} else if(c === "/" || c === ">"){
		this._cbs.onattribend();
		this._state = BEFORE_ATTRIBUTE_NAME;
		this._index--;
	} else if(!whitespace(c)){
		this._cbs.onattribend();
		this._state = IN_ATTRIBUTE_NAME;
		this._sectionStart = this._index;
	}
};

Tokenizer.prototype._stateBeforeAttributeValue = function(c){
	if(c === "\""){
		this._state = IN_ATTRIBUTE_VALUE_DQ;
		this._sectionStart = this._index + 1;
	} else if(c === "'"){
		this._state = IN_ATTRIBUTE_VALUE_SQ;
		this._sectionStart = this._index + 1;
	} else if(!whitespace(c)){
		this._state = IN_ATTRIBUTE_VALUE_NQ;
		this._sectionStart = this._index;
		this._index--; //reconsume token
	}
};

Tokenizer.prototype._stateInAttributeValueDoubleQuotes = function(c){
	if(c === "\""){
		this._emitToken("onattribdata");
		this._cbs.onattribend();
		this._state = BEFORE_ATTRIBUTE_NAME;
	} else if(this._decodeEntities && c === "&"){
		this._emitToken("onattribdata");
		this._baseState = this._state;
		this._state = BEFORE_ENTITY;
		this._sectionStart = this._index;
	}
};

Tokenizer.prototype._stateInAttributeValueSingleQuotes = function(c){
	if(c === "'"){
		this._emitToken("onattribdata");
		this._cbs.onattribend();
		this._state = BEFORE_ATTRIBUTE_NAME;
	} else if(this._decodeEntities && c === "&"){
		this._emitToken("onattribdata");
		this._baseState = this._state;
		this._state = BEFORE_ENTITY;
		this._sectionStart = this._index;
	}
};

Tokenizer.prototype._stateInAttributeValueNoQuotes = function(c){
	if(whitespace(c) || c === ">"){
		this._emitToken("onattribdata");
		this._cbs.onattribend();
		this._state = BEFORE_ATTRIBUTE_NAME;
		this._index--;
	} else if(this._decodeEntities && c === "&"){
		this._emitToken("onattribdata");
		this._baseState = this._state;
		this._state = BEFORE_ENTITY;
		this._sectionStart = this._index;
	}
};

Tokenizer.prototype._stateBeforeDeclaration = function(c){
	this._state = c === "[" ? BEFORE_CDATA_1 :
					c === "-" ? BEFORE_COMMENT :
						IN_DECLARATION;
};

Tokenizer.prototype._stateInDeclaration = function(c){
	if(c === ">"){
		this._cbs.ondeclaration(this._getSection());
		this._state = TEXT;
		this._sectionStart = this._index + 1;
	}
};

Tokenizer.prototype._stateInProcessingInstruction = function(c){
	if(c === ">"){
		this._cbs.onprocessinginstruction(this._getSection());
		this._state = TEXT;
		this._sectionStart = this._index + 1;
	}
};

Tokenizer.prototype._stateBeforeComment = function(c){
	if(c === "-"){
		this._state = IN_COMMENT;
		this._sectionStart = this._index + 1;
	} else {
		this._state = IN_DECLARATION;
	}
};

Tokenizer.prototype._stateInComment = function(c){
	if(c === "-") this._state = AFTER_COMMENT_1;
};

Tokenizer.prototype._stateAfterComment1 = function(c){
	if(c === "-"){
		this._state = AFTER_COMMENT_2;
	} else {
		this._state = IN_COMMENT;
	}
};

Tokenizer.prototype._stateAfterComment2 = function(c){
	if(c === ">"){
		//remove 2 trailing chars
		this._cbs.oncomment(this._buffer.substring(this._sectionStart, this._index - 2));
		this._state = TEXT;
		this._sectionStart = this._index + 1;
	} else if(c !== "-"){
		this._state = IN_COMMENT;
	}
	// else: stay in AFTER_COMMENT_2 (`--->`)
};

Tokenizer.prototype._stateBeforeCdata1 = ifElseState("C", BEFORE_CDATA_2, IN_DECLARATION);
Tokenizer.prototype._stateBeforeCdata2 = ifElseState("D", BEFORE_CDATA_3, IN_DECLARATION);
Tokenizer.prototype._stateBeforeCdata3 = ifElseState("A", BEFORE_CDATA_4, IN_DECLARATION);
Tokenizer.prototype._stateBeforeCdata4 = ifElseState("T", BEFORE_CDATA_5, IN_DECLARATION);
Tokenizer.prototype._stateBeforeCdata5 = ifElseState("A", BEFORE_CDATA_6, IN_DECLARATION);

Tokenizer.prototype._stateBeforeCdata6 = function(c){
	if(c === "["){
		this._state = IN_CDATA;
		this._sectionStart = this._index + 1;
	} else {
		this._state = IN_DECLARATION;
		this._index--;
	}
};

Tokenizer.prototype._stateInCdata = function(c){
	if(c === "]") this._state = AFTER_CDATA_1;
};

Tokenizer.prototype._stateAfterCdata1 = characterState("]", AFTER_CDATA_2);

Tokenizer.prototype._stateAfterCdata2 = function(c){
	if(c === ">"){
		//remove 2 trailing chars
		this._cbs.oncdata(this._buffer.substring(this._sectionStart, this._index - 2));
		this._state = TEXT;
		this._sectionStart = this._index + 1;
	} else if(c !== "]") {
		this._state = IN_CDATA;
	}
	//else: stay in AFTER_CDATA_2 (`]]]>`)
};

Tokenizer.prototype._stateBeforeSpecial = function(c){
	if(c === "c" || c === "C"){
		this._state = BEFORE_SCRIPT_1;
	} else if(c === "t" || c === "T"){
		this._state = BEFORE_STYLE_1;
	} else {
		this._state = IN_TAG_NAME;
		this._index--; //consume the token again
	}
};

Tokenizer.prototype._stateBeforeSpecialEnd = function(c){
	if(this._special === SPECIAL_SCRIPT && (c === "c" || c === "C")){
		this._state = AFTER_SCRIPT_1;
	} else if(this._special === SPECIAL_STYLE && (c === "t" || c === "T")){
		this._state = AFTER_STYLE_1;
	}
	else this._state = TEXT;
};

Tokenizer.prototype._stateBeforeScript1 = consumeSpecialNameChar("R", BEFORE_SCRIPT_2);
Tokenizer.prototype._stateBeforeScript2 = consumeSpecialNameChar("I", BEFORE_SCRIPT_3);
Tokenizer.prototype._stateBeforeScript3 = consumeSpecialNameChar("P", BEFORE_SCRIPT_4);
Tokenizer.prototype._stateBeforeScript4 = consumeSpecialNameChar("T", BEFORE_SCRIPT_5);

Tokenizer.prototype._stateBeforeScript5 = function(c){
	if(c === "/" || c === ">" || whitespace(c)){
		this._special = SPECIAL_SCRIPT;
	}
	this._state = IN_TAG_NAME;
	this._index--; //consume the token again
};

Tokenizer.prototype._stateAfterScript1 = ifElseState("R", AFTER_SCRIPT_2, TEXT);
Tokenizer.prototype._stateAfterScript2 = ifElseState("I", AFTER_SCRIPT_3, TEXT);
Tokenizer.prototype._stateAfterScript3 = ifElseState("P", AFTER_SCRIPT_4, TEXT);
Tokenizer.prototype._stateAfterScript4 = ifElseState("T", AFTER_SCRIPT_5, TEXT);

Tokenizer.prototype._stateAfterScript5 = function(c){
	if(c === ">" || whitespace(c)){
		this._special = SPECIAL_NONE;
		this._state = IN_CLOSING_TAG_NAME;
		this._sectionStart = this._index - 6;
		this._index--; //reconsume the token
	}
	else this._state = TEXT;
};

Tokenizer.prototype._stateBeforeStyle1 = consumeSpecialNameChar("Y", BEFORE_STYLE_2);
Tokenizer.prototype._stateBeforeStyle2 = consumeSpecialNameChar("L", BEFORE_STYLE_3);
Tokenizer.prototype._stateBeforeStyle3 = consumeSpecialNameChar("E", BEFORE_STYLE_4);

Tokenizer.prototype._stateBeforeStyle4 = function(c){
	if(c === "/" || c === ">" || whitespace(c)){
		this._special = SPECIAL_STYLE;
	}
	this._state = IN_TAG_NAME;
	this._index--; //consume the token again
};

Tokenizer.prototype._stateAfterStyle1 = ifElseState("Y", AFTER_STYLE_2, TEXT);
Tokenizer.prototype._stateAfterStyle2 = ifElseState("L", AFTER_STYLE_3, TEXT);
Tokenizer.prototype._stateAfterStyle3 = ifElseState("E", AFTER_STYLE_4, TEXT);

Tokenizer.prototype._stateAfterStyle4 = function(c){
	if(c === ">" || whitespace(c)){
		this._special = SPECIAL_NONE;
		this._state = IN_CLOSING_TAG_NAME;
		this._sectionStart = this._index - 5;
		this._index--; //reconsume the token
	}
	else this._state = TEXT;
};

Tokenizer.prototype._stateBeforeEntity = ifElseState("#", BEFORE_NUMERIC_ENTITY, IN_NAMED_ENTITY);
Tokenizer.prototype._stateBeforeNumericEntity = ifElseState("X", IN_HEX_ENTITY, IN_NUMERIC_ENTITY);

//for entities terminated with a semicolon
Tokenizer.prototype._parseNamedEntityStrict = function(){
	//offset = 1
	if(this._sectionStart + 1 < this._index){
		var entity = this._buffer.substring(this._sectionStart + 1, this._index),
		    map = this._xmlMode ? xml : entities;

		if(map.hasOwnProperty(entity)){
			this._emitPartial(map[entity]);
			this._sectionStart = this._index + 1;
		}
	}
};


//parses legacy entities (without trailing semicolon)
Tokenizer.prototype._parseLegacyEntity = function(){
	var start = this._sectionStart + 1,
	    limit = this._index - start;

	if(limit > 6) limit = 6; //the max length of legacy entities is 6

	while(limit >= 2){ //the min length of legacy entities is 2
		var entity = this._buffer.substr(start, limit);

		if(legacy.hasOwnProperty(entity)){
			this._emitPartial(legacy[entity]);
			this._sectionStart += limit + 1;
			return;
		} else {
			limit--;
		}
	}
};

Tokenizer.prototype._stateInNamedEntity = function(c){
	if(c === ";"){
		this._parseNamedEntityStrict();
		if(this._sectionStart + 1 < this._index && !this._xmlMode){
			this._parseLegacyEntity();
		}
		this._state = this._baseState;
	} else if((c < "a" || c > "z") && (c < "A" || c > "Z") && (c < "0" || c > "9")){
		if(this._xmlMode);
		else if(this._sectionStart + 1 === this._index);
		else if(this._baseState !== TEXT){
			if(c !== "="){
				this._parseNamedEntityStrict();
			}
		} else {
			this._parseLegacyEntity();
		}

		this._state = this._baseState;
		this._index--;
	}
};

Tokenizer.prototype._decodeNumericEntity = function(offset, base){
	var sectionStart = this._sectionStart + offset;

	if(sectionStart !== this._index){
		//parse entity
		var entity = this._buffer.substring(sectionStart, this._index);
		var parsed = parseInt(entity, base);

		this._emitPartial(decode_codepoint(parsed));
		this._sectionStart = this._index;
	} else {
		this._sectionStart--;
	}

	this._state = this._baseState;
};

Tokenizer.prototype._stateInNumericEntity = function(c){
	if(c === ";"){
		this._decodeNumericEntity(2, 10);
		this._sectionStart++;
	} else if(c < "0" || c > "9"){
		if(!this._xmlMode){
			this._decodeNumericEntity(2, 10);
		} else {
			this._state = this._baseState;
		}
		this._index--;
	}
};

Tokenizer.prototype._stateInHexEntity = function(c){
	if(c === ";"){
		this._decodeNumericEntity(3, 16);
		this._sectionStart++;
	} else if((c < "a" || c > "f") && (c < "A" || c > "F") && (c < "0" || c > "9")){
		if(!this._xmlMode){
			this._decodeNumericEntity(3, 16);
		} else {
			this._state = this._baseState;
		}
		this._index--;
	}
};

Tokenizer.prototype._cleanup = function (){
	if(this._sectionStart < 0){
		this._buffer = "";
		this._bufferOffset += this._index;
		this._index = 0;
	} else if(this._running){
		if(this._state === TEXT){
			if(this._sectionStart !== this._index){
				this._cbs.ontext(this._buffer.substr(this._sectionStart));
			}
			this._buffer = "";
			this._bufferOffset += this._index;
			this._index = 0;
		} else if(this._sectionStart === this._index){
			//the section just started
			this._buffer = "";
			this._bufferOffset += this._index;
			this._index = 0;
		} else {
			//remove everything unnecessary
			this._buffer = this._buffer.substr(this._sectionStart);
			this._index -= this._sectionStart;
			this._bufferOffset += this._sectionStart;
		}

		this._sectionStart = 0;
	}
};

//TODO make events conditional
Tokenizer.prototype.write = function(chunk){
	if(this._ended) this._cbs.onerror(Error(".write() after done!"));

	this._buffer += chunk;
	this._parse();
};

Tokenizer.prototype._parse = function(){
	while(this._index < this._buffer.length && this._running){
		var c = this._buffer.charAt(this._index);
		if(this._state === TEXT) {
			this._stateText(c);
		} else if(this._state === BEFORE_TAG_NAME){
			this._stateBeforeTagName(c);
		} else if(this._state === IN_TAG_NAME) {
			this._stateInTagName(c);
		} else if(this._state === BEFORE_CLOSING_TAG_NAME){
			this._stateBeforeCloseingTagName(c);
		} else if(this._state === IN_CLOSING_TAG_NAME){
			this._stateInCloseingTagName(c);
		} else if(this._state === AFTER_CLOSING_TAG_NAME){
			this._stateAfterCloseingTagName(c);
		} else if(this._state === IN_SELF_CLOSING_TAG){
			this._stateInSelfClosingTag(c);
		}

		/*
		*	attributes
		*/
		else if(this._state === BEFORE_ATTRIBUTE_NAME){
			this._stateBeforeAttributeName(c);
		} else if(this._state === IN_ATTRIBUTE_NAME){
			this._stateInAttributeName(c);
		} else if(this._state === AFTER_ATTRIBUTE_NAME){
			this._stateAfterAttributeName(c);
		} else if(this._state === BEFORE_ATTRIBUTE_VALUE){
			this._stateBeforeAttributeValue(c);
		} else if(this._state === IN_ATTRIBUTE_VALUE_DQ){
			this._stateInAttributeValueDoubleQuotes(c);
		} else if(this._state === IN_ATTRIBUTE_VALUE_SQ){
			this._stateInAttributeValueSingleQuotes(c);
		} else if(this._state === IN_ATTRIBUTE_VALUE_NQ){
			this._stateInAttributeValueNoQuotes(c);
		}

		/*
		*	declarations
		*/
		else if(this._state === BEFORE_DECLARATION){
			this._stateBeforeDeclaration(c);
		} else if(this._state === IN_DECLARATION){
			this._stateInDeclaration(c);
		}

		/*
		*	processing instructions
		*/
		else if(this._state === IN_PROCESSING_INSTRUCTION){
			this._stateInProcessingInstruction(c);
		}

		/*
		*	comments
		*/
		else if(this._state === BEFORE_COMMENT){
			this._stateBeforeComment(c);
		} else if(this._state === IN_COMMENT){
			this._stateInComment(c);
		} else if(this._state === AFTER_COMMENT_1){
			this._stateAfterComment1(c);
		} else if(this._state === AFTER_COMMENT_2){
			this._stateAfterComment2(c);
		}

		/*
		*	cdata
		*/
		else if(this._state === BEFORE_CDATA_1){
			this._stateBeforeCdata1(c);
		} else if(this._state === BEFORE_CDATA_2){
			this._stateBeforeCdata2(c);
		} else if(this._state === BEFORE_CDATA_3){
			this._stateBeforeCdata3(c);
		} else if(this._state === BEFORE_CDATA_4){
			this._stateBeforeCdata4(c);
		} else if(this._state === BEFORE_CDATA_5){
			this._stateBeforeCdata5(c);
		} else if(this._state === BEFORE_CDATA_6){
			this._stateBeforeCdata6(c);
		} else if(this._state === IN_CDATA){
			this._stateInCdata(c);
		} else if(this._state === AFTER_CDATA_1){
			this._stateAfterCdata1(c);
		} else if(this._state === AFTER_CDATA_2){
			this._stateAfterCdata2(c);
		}

		/*
		* special tags
		*/
		else if(this._state === BEFORE_SPECIAL){
			this._stateBeforeSpecial(c);
		} else if(this._state === BEFORE_SPECIAL_END){
			this._stateBeforeSpecialEnd(c);
		}

		/*
		* script
		*/
		else if(this._state === BEFORE_SCRIPT_1){
			this._stateBeforeScript1(c);
		} else if(this._state === BEFORE_SCRIPT_2){
			this._stateBeforeScript2(c);
		} else if(this._state === BEFORE_SCRIPT_3){
			this._stateBeforeScript3(c);
		} else if(this._state === BEFORE_SCRIPT_4){
			this._stateBeforeScript4(c);
		} else if(this._state === BEFORE_SCRIPT_5){
			this._stateBeforeScript5(c);
		}

		else if(this._state === AFTER_SCRIPT_1){
			this._stateAfterScript1(c);
		} else if(this._state === AFTER_SCRIPT_2){
			this._stateAfterScript2(c);
		} else if(this._state === AFTER_SCRIPT_3){
			this._stateAfterScript3(c);
		} else if(this._state === AFTER_SCRIPT_4){
			this._stateAfterScript4(c);
		} else if(this._state === AFTER_SCRIPT_5){
			this._stateAfterScript5(c);
		}

		/*
		* style
		*/
		else if(this._state === BEFORE_STYLE_1){
			this._stateBeforeStyle1(c);
		} else if(this._state === BEFORE_STYLE_2){
			this._stateBeforeStyle2(c);
		} else if(this._state === BEFORE_STYLE_3){
			this._stateBeforeStyle3(c);
		} else if(this._state === BEFORE_STYLE_4){
			this._stateBeforeStyle4(c);
		}

		else if(this._state === AFTER_STYLE_1){
			this._stateAfterStyle1(c);
		} else if(this._state === AFTER_STYLE_2){
			this._stateAfterStyle2(c);
		} else if(this._state === AFTER_STYLE_3){
			this._stateAfterStyle3(c);
		} else if(this._state === AFTER_STYLE_4){
			this._stateAfterStyle4(c);
		}

		/*
		* entities
		*/
		else if(this._state === BEFORE_ENTITY){
			this._stateBeforeEntity(c);
		} else if(this._state === BEFORE_NUMERIC_ENTITY){
			this._stateBeforeNumericEntity(c);
		} else if(this._state === IN_NAMED_ENTITY){
			this._stateInNamedEntity(c);
		} else if(this._state === IN_NUMERIC_ENTITY){
			this._stateInNumericEntity(c);
		} else if(this._state === IN_HEX_ENTITY){
			this._stateInHexEntity(c);
		}

		else {
			this._cbs.onerror(Error("unknown _state"), this._state);
		}

		this._index++;
	}

	this._cleanup();
};

Tokenizer.prototype.pause = function(){
	this._running = false;
};
Tokenizer.prototype.resume = function(){
	this._running = true;

	if(this._index < this._buffer.length){
		this._parse();
	}
	if(this._ended){
		this._finish();
	}
};

Tokenizer.prototype.end = function(chunk){
	if(this._ended) this._cbs.onerror(Error(".end() after done!"));
	if(chunk) this.write(chunk);

	this._ended = true;

	if(this._running) this._finish();
};

Tokenizer.prototype._finish = function(){
	//if there is remaining data, emit it in a reasonable way
	if(this._sectionStart < this._index){
		this._handleTrailingData();
	}

	this._cbs.onend();
};

Tokenizer.prototype._handleTrailingData = function(){
	var data = this._buffer.substr(this._sectionStart);

	if(this._state === IN_CDATA || this._state === AFTER_CDATA_1 || this._state === AFTER_CDATA_2){
		this._cbs.oncdata(data);
	} else if(this._state === IN_COMMENT || this._state === AFTER_COMMENT_1 || this._state === AFTER_COMMENT_2){
		this._cbs.oncomment(data);
	} else if(this._state === IN_NAMED_ENTITY && !this._xmlMode){
		this._parseLegacyEntity();
		if(this._sectionStart < this._index){
			this._state = this._baseState;
			this._handleTrailingData();
		}
	} else if(this._state === IN_NUMERIC_ENTITY && !this._xmlMode){
		this._decodeNumericEntity(2, 10);
		if(this._sectionStart < this._index){
			this._state = this._baseState;
			this._handleTrailingData();
		}
	} else if(this._state === IN_HEX_ENTITY && !this._xmlMode){
		this._decodeNumericEntity(3, 16);
		if(this._sectionStart < this._index){
			this._state = this._baseState;
			this._handleTrailingData();
		}
	} else if(
		this._state !== IN_TAG_NAME &&
		this._state !== BEFORE_ATTRIBUTE_NAME &&
		this._state !== BEFORE_ATTRIBUTE_VALUE &&
		this._state !== AFTER_ATTRIBUTE_NAME &&
		this._state !== IN_ATTRIBUTE_NAME &&
		this._state !== IN_ATTRIBUTE_VALUE_SQ &&
		this._state !== IN_ATTRIBUTE_VALUE_DQ &&
		this._state !== IN_ATTRIBUTE_VALUE_NQ &&
		this._state !== IN_CLOSING_TAG_NAME
	){
		this._cbs.ontext(data);
	}
	//else, ignore remaining data
	//TODO add a way to remove current tag
};

Tokenizer.prototype.reset = function(){
	Tokenizer.call(this, {xmlMode: this._xmlMode, decodeEntities: this._decodeEntities}, this._cbs);
};

Tokenizer.prototype.getAbsoluteIndex = function(){
	return this._bufferOffset + this._index;
};

Tokenizer.prototype._getSection = function(){
	return this._buffer.substring(this._sectionStart, this._index);
};

Tokenizer.prototype._emitToken = function(name){
	let section = this._getSection();
	switch (name) {
		case 'onattribute':
			this._cbs.onattribute(section);
			break
		case 'onattribdata':
			this._cbs.onattribdata(section);
			break
		case 'oncdatastart':
			this._cbs.oncdatastart(section);
			break
		case 'oncdataend':
			this._cbs.oncdataend(section);
			break
		case 'ontext':
			this._cbs.ontext(section);
			break
		case 'onprocessinginstruction':
			this._cbs.onprocessinginstruction(section);
			break
		case 'oncomment':
			this._cbs.oncomment(section);
			break
		case 'oncommentend':
			this._cbs.oncommentend(section);
			break
		case 'onclosetag':
			this._cbs.onclosetag(section);
			break
		case 'onopentag':
			this._cbs.onopentag(section);
			break
		case 'onopentagname':
			this._cbs.onopentagname(section);
			break
		case 'onerror':
			this._cbs.onerror(section);
			break
		case 'onend':
			this._cbs.onend(section);
			break
		default:
			throw new Error('Unsupported event: ' + name)
	}
	this._sectionStart = -1;
};

Tokenizer.prototype._emitPartial = function(value){
	if(this._baseState !== TEXT){
		this._cbs.onattribdata(value); //TODO implement the new event
	} else {
		this._cbs.ontext(value);
	}
};

var Tokenizer$1 = Tokenizer_1;

/*
	Options:

	xmlMode: Disables the special behavior for script/style tags (false by default)
	lowerCaseAttributeNames: call .toLowerCase for each attribute name (true if xmlMode is `false`)
	lowerCaseTags: call .toLowerCase for each tag name (true if xmlMode is `false`)
*/

/*
	Callbacks:

	oncdataend,
	oncdatastart,
	onclosetag,
	oncomment,
	oncommentend,
	onerror,
	onopentag,
	onprocessinginstruction,
	onreset,
	ontext
*/

var formTags = new Set([
	'input',
	'option',
	'optgroup',
	'select',
	'button',
	'datalist',
	'textarea'
]);

var openImpliesClose = new Map([
	['tr'      , new Set('tr','th','td')],
	['th'      , new Set('th')],
	['td'      , new Set('thead','th','td')],
	['body'    , new Set('head','link','script')],
	['li'      , new Set('li')],
	['p'       , new Set('p')],
	['h1'      , new Set('p')],
	['h2'      , new Set('p')],
	['h3'      , new Set('p')],
	['h4'      , new Set('p')],
	['h5'      , new Set('p')],
	['h6'      , new Set('p')],
	['select'  , formTags],
	['input'   , formTags],
	['output'  , formTags],
	['button'  , formTags],
	['datalist', formTags],
	['textarea', formTags],
	['option'  , new Set('option')],
	['optgroup', new Set('optgroup')]
]);

var voidElements = new Set([
	'area',
	'base',
	'basefont',
	'br',
	'col',
	'command',
	'embed',
	'frame',
	'hr',
	'img',
	'input',
	'isindex',
	'keygen',
	'link',
	'meta',
	'param',
	'source',
	'track',
	'wbr',
	//common self closing svg elements
	'path',
	'circle',
	'ellipse',
	'line',
	'rect',
	'use',
	'stop',
	'polyline',
	'polygon'
]);

var re_nameEnd = /\s|\//;

function Parser(cbs, options){
	this._options = options || {};
	this._cbs = cbs || {};

	this._tagname = "";
	this._attribname = "";
	this._attribvalue = "";
	this._attribs = null;
	this._stack = [];

	this.startIndex = 0;
	this.endIndex = null;

	this._lowerCaseTagNames = this._options.lowerCaseTags || !this._options.xmlMode;
	this._lowerCaseAttributeNames = this._options.lowerCaseAttributeNames || !this._options.xmlMode;

	if(this._options.Tokenizer) {
		Tokenizer$1 = this._options.Tokenizer;
	}
	this._tokenizer = new Tokenizer$1(this._options, this);

	if(this._cbs.onparserinit) this._cbs.onparserinit(this);
}

Parser.prototype._updatePosition = function(initialOffset){
	if(this.endIndex === null){
		if(this._tokenizer._sectionStart <= initialOffset){
			this.startIndex = 0;
		} else {
			this.startIndex = this._tokenizer._sectionStart - initialOffset;
		}
	}
	else this.startIndex = this.endIndex + 1;
	this.endIndex = this._tokenizer.getAbsoluteIndex();
};

//Tokenizer event handlers
Parser.prototype.ontext = function(data){
	this._updatePosition(1);
	this.endIndex--;

	if(this._cbs.ontext) this._cbs.ontext(data);
};

Parser.prototype.onopentagname = function(name){
	if(this._lowerCaseTagNames){
		name = name.toLowerCase();
	}

	this._tagname = name;

	if(!this._options.xmlMode && openImpliesClose.has(name)) {
		while (this._stack.length > 0) {
			var el = this._stack[this._stack.length - 1];
			if (openImpliesClose.get(name).has(el)) {
				this.onclosetag(el);
			} else {
				break
			}
		}
	}

	if(this._options.xmlMode || !(voidElements.has(name))){
		this._stack.push(name);
	}

	if(this._cbs.onopentagname) this._cbs.onopentagname(name);
	if(this._cbs.onopentag) this._attribs = {};
};

Parser.prototype.onopentagend = function(){
	this._updatePosition(1);

	if(this._attribs){
		if(this._cbs.onopentag) this._cbs.onopentag(this._tagname, this._attribs);
		this._attribs = null;
	}

	if(!this._options.xmlMode && this._cbs.onclosetag && voidElements.has(this._tagname)){
		this._cbs.onclosetag(this._tagname);
	}

	this._tagname = "";
};

Parser.prototype.onclosetag = function(name){
	this._updatePosition(1);
	if(this._lowerCaseTagNames){
		name = name.toLowerCase();
	}
	// NOTE: the original implementation allows 'voidElements' in XML, such as in `<input>`, which
	// IMO is not valid. Only self-closing tags are allowd, such as in `<foo />`, but actually only
	// if it is declared as EMPTY in the XSD
	if(this._options.xmlMode) {
		const stack = this._stack;
		let last = stack.pop();
		while(last !== name) {
			if(this._cbs.onerror) {
				this._cbs.onerror("Unclosed tag <"+last+">");
			}
			last = stack.pop();
		}
		this.onopentagend();
		if(this._cbs.onclosetag) {
			this._cbs.onclosetag(last);
		}
	} else {
		if(this._stack.length && (!(voidElements.has(name)))) {
			let pos = this._stack.lastIndexOf(name);
			if(pos !== -1){
				if(this._cbs.onclosetag){
					pos = this._stack.length - pos;
					while(pos--) this._cbs.onclosetag(this._stack.pop());
				}
				else this._stack.length = pos;
			} else if(name === "p"){
				this.onopentagname(name);
				this._closeCurrentTag();
			}
		} else if(name === "br" || name === "p"){
			this.onopentagname(name);
			this._closeCurrentTag();
		}
	}
};

Parser.prototype.onselfclosingtag = function(){
	if(this._options.xmlMode || this._options.recognizeSelfClosing){
		this._closeCurrentTag();
	} else {
		this.onopentagend();
	}
};

Parser.prototype._closeCurrentTag = function(){
	var name = this._tagname;

	this.onopentagend();

	//self-closing tags will be on the top of the stack
	//(cheaper check than in onclosetag)
	if(this._stack[this._stack.length - 1] === name){
		if(this._cbs.onclosetag){
			this._cbs.onclosetag(name);
		}
		this._stack.pop();
	}
};

Parser.prototype.onattribname = function(name){
	if(this._lowerCaseAttributeNames){
		name = name.toLowerCase();
	}
	this._attribname = name;
};

Parser.prototype.onattribdata = function(value){
	this._attribvalue += value;
};

Parser.prototype.onattribend = function(){
	if(this._cbs.onattribute) this._cbs.onattribute(this._attribname, this._attribvalue);
	if(
		this._attribs &&
		!Object.prototype.hasOwnProperty.call(this._attribs, this._attribname)
	){
		this._attribs[this._attribname] = this._attribvalue;
	}
	this._attribname = "";
	this._attribvalue = "";
};

Parser.prototype._getInstructionName = function(value){
	var idx = value.search(re_nameEnd),
	    name = idx < 0 ? value : value.substr(0, idx);

	if(this._lowerCaseTagNames){
		name = name.toLowerCase();
	}

	return name;
};

Parser.prototype.ondeclaration = function(value){
	if(this._cbs.onprocessinginstruction){
		var name = this._getInstructionName(value);
		this._cbs.onprocessinginstruction("!" + name, "!" + value);
	}
};

Parser.prototype.onprocessinginstruction = function(value){
	if(this._cbs.onprocessinginstruction){
		var name = this._getInstructionName(value);
		this._cbs.onprocessinginstruction("?" + name, "?" + value);
	}
};

Parser.prototype.oncomment = function(value){
	this._updatePosition(4);

	if(this._cbs.oncomment) this._cbs.oncomment(value);
	if(this._cbs.oncommentend) this._cbs.oncommentend();
};

Parser.prototype.oncdata = function(value){
	this._updatePosition(1);

	if(this._options.xmlMode || this._options.recognizeCDATA){
		if(this._cbs.oncdatastart) this._cbs.oncdatastart();
		if(this._cbs.ontext) this._cbs.ontext(value);
		if(this._cbs.oncdataend) this._cbs.oncdataend();
	} else {
		this.oncomment("[CDATA[" + value + "]]");
	}
};

Parser.prototype.onerror = function(err){
	if(this._cbs.onerror) this._cbs.onerror(err);
};

Parser.prototype.onend = function(){
	if(this._cbs.onclosetag){
		for(
			var i = this._stack.length;
			i > 0;
			this._cbs.onclosetag(this._stack[--i])
		);
	}
	if(this._cbs.onend) this._cbs.onend();
};


//Resets the parser to a blank state, ready to parse a new HTML document
Parser.prototype.reset = function(){
	if(this._cbs.onreset) this._cbs.onreset();
	this._tokenizer.reset();

	this._tagname = "";
	this._attribname = "";
	this._attribs = null;
	this._stack = [];

	if(this._cbs.onparserinit) this._cbs.onparserinit(this);
};

//Parses a complete HTML document and pushes it to the handler
Parser.prototype.parseComplete = function(data){
	this.reset();
	this.end(data);
};

Parser.prototype.write = function(chunk){
	this._tokenizer.write(chunk);
};

Parser.prototype.end = function(chunk){
	this._tokenizer.end(chunk);
};

Parser.prototype.pause = function(){
	this._tokenizer.pause();
};

Parser.prototype.resume = function(){
	this._tokenizer.resume();
};

//alias for backwards compat
Parser.prototype.parseChunk = Parser.prototype.write;
Parser.prototype.done = Parser.prototype.end;

var Parser_1 = Parser;

// monkey patching

Parser_1.prototype.ondeclaration = function(value){
  if(this._cbs.ondeclaration){
    this._cbs.ondeclaration(value);
  } else if(this._cbs.onprocessinginstruction){
    var name = this._getInstructionName(value);
    this._cbs.onprocessinginstruction("!" + name, "!" + value);
  }
};

Parser_1.prototype.oncdata = function(value){
  this._updatePosition(1);

  if(this._options.xmlMode || this._options.recognizeCDATA){
    if(this._cbs.oncdatastart) this._cbs.oncdatastart(value);
    // we don't want `ontext` getting called here
    if(this._cbs.oncdataend) this._cbs.oncdataend();
  } else {
    this.oncomment("[CDATA[" + value + "]]");
  }
};

export default Parser_1;