htmlparser2 - npm Package Compare versions

Comparing version 3.1.5 to 3.1.6

lib/Parser.js

@@ -177,3 +177,8 @@ var Tokenizer = require("./Tokenizer.js");
 	if(this._cbs.onattribute) this._cbs.onattribute(this._attribname, value);
-	if(this._attribs) this._attribs[this._attribname] = value;
+	if(
+		this._attribs &&
+		!Object.prototype.hasOwnProperty.call(this._attribs, this._attribname)
+	){
+		this._attribs[this._attribname] = value;
+	}
 	this._attribname = "";
@@ -180,0 +185,0 @@ };

lib/Tokenizer.js

@@ -67,9 +67,40 @@ module.exports = Tokenizer;
     AFTER_STYLE_3 = i++, //L
-    AFTER_STYLE_4 = i++; //E
+    AFTER_STYLE_4 = i++, //E
 
+    SPECIAL_NONE = 0,
+    SPECIAL_SCRIPT = 1,
+    SPECIAL_STYLE = 2;
 
+
 function whitespace(c){
-	return c === " " || c === "\t" || c === "\r" || c === "\n";
+	return c === " " || c === "\n" || c === "\t" || c === "\f";
 }
 
+function ifElseState(upper, SUCCESS, FAILURE){
+	var lower = upper.toLowerCase();
+
+	if(upper === lower){
+		return function(c){
+			this._state = c === lower ? SUCCESS : FAILURE;
+		};
+	} else {
+		return function(c){
+			this._state = (c === lower || c === upper) ? SUCCESS : FAILURE;
+		};
+	}
+}
+
+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){
@@ -80,8 +111,331 @@ this._state = TEXT;
 	this._index = 0;
-	this._options = options;
-	this._special = 0; // 1 for script, 2 for style
+	this._special = SPECIAL_NONE;
 	this._cbs = cbs;
 	this._running = true;
+	this._xmlMode = !!(options && options.xmlMode);
 }
 
+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;
+	}
+};
+
+Tokenizer.prototype._stateBeforeTagName = function (c) {
+	if(c === "/"){
+		this._state = BEFORE_CLOSING_TAG_NAME;
+	} 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._special = SPECIAL_NONE;
+		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(whitespace(c)){
+		/* noop */
+	} else if(c === ">"){
+		this._state = TEXT;
+		this._cbs.onopentagend();
+		this._sectionStart = this._index + 1;
+	} else if(c === "/"){
+		this._cbs.onselfclosingtag();
+		this._state = AFTER_CLOSING_TAG_NAME;
+	} else {
+		this._state = IN_ATTRIBUTE_NAME;
+		this._sectionStart = this._index;
+	}
+};
+
+Tokenizer.prototype._stateInAttributeName = function (c) {
+	if(c === "=" || c === "/" || c === ">" || whitespace(c)){
+		if(this._index > this._sectionStart){
+			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._state = BEFORE_ATTRIBUTE_NAME;
+		this._index--;
+	} else if(!whitespace(c)){
+		this._state = IN_ATTRIBUTE_NAME;
+		this._sectionStart = this._index;
+	}
+};
+
+Tokenizer.prototype._stateBeforeAttributeValue = function (c) {
+	if(c === "\""){
+		this._state = IN_ATTRIBUTE_VALUE_DOUBLE_QUOTES;
+		this._sectionStart = this._index + 1;
+	} else if(c === "'"){
+		this._state = IN_ATTRIBUTE_VALUE_SINGLE_QUOTES;
+		this._sectionStart = this._index + 1;
+	} else if(!whitespace(c)){
+		this._state = IN_ATTRIBUTE_VALUE_NO_QUOTES;
+		this._sectionStart = this._index;
+	}
+};
+
+Tokenizer.prototype._stateInAttributeValueDoubleQuotes = function (c) {
+	if(c === "\""){
+		this._emitToken("onattribvalue");
+		this._state = BEFORE_ATTRIBUTE_NAME;
+	}
+};
+
+Tokenizer.prototype._stateInAttributeValueSingleQuotes = function (c) {
+	if(c === "'"){
+		this._emitToken("onattribvalue");
+		this._state = BEFORE_ATTRIBUTE_NAME;
+	}
+};
+
+Tokenizer.prototype._stateInAttributeValueNoQuotes = function (c) {
+	if(whitespace(c) || c === ">"){
+		this._emitToken("onattribvalue");
+		this._state = BEFORE_ATTRIBUTE_NAME;
+		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 = ifElseState("-", AFTER_COMMENT_2, 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;
+	}
+};
+
+Tokenizer.prototype._stateInCdata = function (c) {
+	if(c === "]") this._state = AFTER_CDATA_1;
+};
+
+Tokenizer.prototype._stateAfterCdata1 = ifElseState("]", AFTER_CDATA_2, IN_CDATA);
+
+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._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._state = IN_CLOSING_TAG_NAME;
+		this._sectionStart = this._index - 5;
+		this._index--; //reconsume the token
+	}
+	else this._state = TEXT;
+};
+
+Tokenizer.prototype._cleanup = function () {
+	if(this._sectionStart < 0){
+		this._buffer = "";
+		this._index = 0;
+	} else {
+		if(this._state === TEXT){
+			if(this._sectionStart !== this._index){
+				this._cbs.ontext(this._buffer.substr(this._sectionStart));
+			}
+			this._buffer = "";
+			this._index = 0;
+		} else if(this._sectionStart === this._index){
+			//the section just started
+			this._buffer = "";
+			this._index = 0;
+		} else {
+			//remove everything unnecessary
+			this._buffer = this._buffer.substr(this._sectionStart);
+			this._index -= this._sectionStart;
+		}
+
+		this._sectionStart = 0;
+	}
+};
+
 //TODO make events conditional
@@ -93,78 +447,14 @@ Tokenizer.prototype.write = function(chunk){
 		var c = this._buffer.charAt(this._index);
-		if(this._state === TEXT){
-			if(c === "<"){
-				this._emitIfToken("ontext");
-				this._state = BEFORE_TAG_NAME;
-				this._sectionStart = this._index;
-			}
+		if(this._state === TEXT) {
+			this._stateText(c);
 		} else if(this._state === BEFORE_TAG_NAME){
-			if(c === "/"){
-				this._state = BEFORE_CLOSING_TAG_NAME;
-			} else if(c === ">" || this._special > 0 || whitespace(c)) {
-				this._state = TEXT;
-			} else {
-				if(whitespace(c));
-				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 if(
-					!(this._options && this._options.xmlMode) &&
-					(c === "s" || c === "S")
-				){
-					this._state = BEFORE_SPECIAL;
-					this._sectionStart = this._index;
-				} else {
-					this._state = IN_TAG_NAME;
-					this._sectionStart = this._index;
-				}
-			}
-		} else if(this._state === IN_TAG_NAME){
-			if(c === "/"){
-				this._emitToken("onopentagname");
-				this._cbs.onselfclosingtag();
-				this._state = AFTER_CLOSING_TAG_NAME;
-			} else if(c === ">"){
-				this._emitToken("onopentagname");
-				this._cbs.onopentagend();
-				this._state = TEXT;
-				this._sectionStart = this._index + 1;
-			} else if(whitespace(c)){
-				this._emitToken("onopentagname");
-				this._state = BEFORE_ATTRIBUTE_NAME;
-			}
+			this._stateBeforeTagName(c);
+		} else if(this._state === IN_TAG_NAME) {
+			this._stateInTagName(c);
 		} else if(this._state === BEFORE_CLOSING_TAG_NAME){
-			if(whitespace(c));
-			else if(c === ">"){
-				this._state = TEXT;
-			} else if(this._special > 0){
-				if(c === "s" || c === "S"){
-					this._state = BEFORE_SPECIAL_END;
-				} else {
-					this._state = TEXT;
-					continue;
-				}
-			} else {
-				this._state = IN_CLOSING_TAG_NAME;
-				this._sectionStart = this._index;
-			}
+			this._stateBeforeCloseingTagName(c);
 		} else if(this._state === IN_CLOSING_TAG_NAME){
-			if(c === ">"){
-				this._emitToken("onclosetag");
-				this._state = TEXT;
-				this._sectionStart = this._index + 1;
-				this._special = 0;
-			} else if(whitespace(c)){
-				this._emitToken("onclosetag");
-				this._state = AFTER_CLOSING_TAG_NAME;
-				this._special = 0;
-			}
+			this._stateInCloseingTagName(c);
 		} else if(this._state === AFTER_CLOSING_TAG_NAME){
-			//skip everything until ">"
-			if(c === ">"){
-				this._state = TEXT;
-				this._sectionStart = this._index + 1;
-			}
+			this._stateAfterCloseingTagName(c);
 		}
@@ -176,66 +466,15 @@
 		else if(this._state === BEFORE_ATTRIBUTE_NAME){
-			if(c === ">"){
-				this._state = TEXT;
-				this._cbs.onopentagend();
-				this._sectionStart = this._index + 1;
-			} else if(c === "/"){
-				this._cbs.onselfclosingtag();
-				this._state = AFTER_CLOSING_TAG_NAME;
-			} else if(!whitespace(c)){
-				this._state = IN_ATTRIBUTE_NAME;
-				this._sectionStart = this._index;
-			}
+			this._stateBeforeAttributeName(c);
 		} else if(this._state === IN_ATTRIBUTE_NAME){
-			if(c === "="){
-				this._emitIfToken("onattribname");
-				this._state = BEFORE_ATTRIBUTE_VALUE;
-			} else if(whitespace(c)){
-				this._emitIfToken("onattribname");
-				this._state = AFTER_ATTRIBUTE_NAME;
-			} else if(c === "/" || c === ">"){
-				this._emitIfToken("onattribname");
-				this._state = BEFORE_ATTRIBUTE_NAME;
-				continue;
-			}
+			this._stateInAttributeName(c);
 		} else if(this._state === AFTER_ATTRIBUTE_NAME){
-			if(c === "="){
-				this._state = BEFORE_ATTRIBUTE_VALUE;
-			} else if(c === "/" || c === ">"){
-				this._state = BEFORE_ATTRIBUTE_NAME;
-				continue;
-			} else if(!whitespace(c)){
-				this._state = IN_ATTRIBUTE_NAME;
-				this._sectionStart = this._index;
-			}
+			this._stateAfterAttributeName(c);
 		} else if(this._state === BEFORE_ATTRIBUTE_VALUE){
-			if(c === "\""){
-				this._state = IN_ATTRIBUTE_VALUE_DOUBLE_QUOTES;
-				this._sectionStart = this._index + 1;
-			} else if(c === "'"){
-				this._state = IN_ATTRIBUTE_VALUE_SINGLE_QUOTES;
-				this._sectionStart = this._index + 1;
-			} else if(!whitespace(c)){
-				this._state = IN_ATTRIBUTE_VALUE_NO_QUOTES;
-				this._sectionStart = this._index;
-			}
+			this._stateBeforeAttributeValue(c);
 		} else if(this._state === IN_ATTRIBUTE_VALUE_DOUBLE_QUOTES){
-			if(c === "\""){
-				this._emitToken("onattribvalue");
-				this._state = BEFORE_ATTRIBUTE_NAME;
-			}
+			this._stateInAttributeValueDoubleQuotes(c);
 		} else if(this._state === IN_ATTRIBUTE_VALUE_SINGLE_QUOTES){
-			if(c === "'"){
-				this._state = BEFORE_ATTRIBUTE_NAME;	
-				this._emitToken("onattribvalue");
-			}
+			this._stateInAttributeValueSingleQuotes(c);
 		} else if(this._state === IN_ATTRIBUTE_VALUE_NO_QUOTES){
-			if(c === ">"){
-				this._emitToken("onattribvalue");
-				this._state = TEXT;
-				this._cbs.onopentagend();
-				this._sectionStart = this._index + 1;
-			} else if(whitespace(c)){
-				this._emitToken("onattribvalue");
-				this._state = BEFORE_ATTRIBUTE_NAME;
-			}
+			this._stateInAttributeValueNoQuotes(c);
 		}
@@ -247,11 +486,5 @@
 		else if(this._state === BEFORE_DECLARATION){
-			if(c === "[") this._state = BEFORE_CDATA_1;
-			else if(c === "-") this._state = BEFORE_COMMENT;
-			else this._state = IN_DECLARATION;
+			this._stateBeforeDeclaration(c);
 		} else if(this._state === IN_DECLARATION){
-			if(c === ">"){
-				this._emitToken("ondeclaration");
-				this._state = TEXT;
-				this._sectionStart = this._index + 1;
-			}
+			this._stateInDeclaration(c);
 		}
@@ -263,7 +496,3 @@
 		else if(this._state === IN_PROCESSING_INSTRUCTION){
-			if(c === ">"){
-				this._emitToken("onprocessinginstruction");
-				this._state = TEXT;
-				this._sectionStart = this._index + 1;
-			}
+			this._stateInProcessingInstruction(c);
 		}
@@ -275,23 +504,9 @@
 		else if(this._state === BEFORE_COMMENT){
-			if(c === "-"){
-				this._state = IN_COMMENT;
-				this._sectionStart = this._index + 1;
-			} else {
-				this._state = IN_DECLARATION;
-			}
+			this._stateBeforeComment(c);
 		} else if(this._state === IN_COMMENT){
-			if(c === "-") this._state = AFTER_COMMENT_1;
+			this._stateInComment(c);
 		} else if(this._state === AFTER_COMMENT_1){
-			if(c === "-") this._state = AFTER_COMMENT_2;
-			else this._state = IN_COMMENT;
+			this._stateAfterComment1(c);
 		} else if(this._state === AFTER_COMMENT_2){
-			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 (`--->`)
+			this._stateAfterComment2(c);
 		}
@@ -303,38 +518,19 @@
 		else if(this._state === BEFORE_CDATA_1){
-			if(c === "C") this._state = BEFORE_CDATA_2;
-			else this._state = IN_DECLARATION;
+			this._stateBeforeCdata1(c);
 		} else if(this._state === BEFORE_CDATA_2){
-			if(c === "D") this._state = BEFORE_CDATA_3;
-			else this._state = IN_DECLARATION;
+			this._stateBeforeCdata2(c);
 		} else if(this._state === BEFORE_CDATA_3){
-			if(c === "A") this._state = BEFORE_CDATA_4;
-			else this._state = IN_DECLARATION;
+			this._stateBeforeCdata3(c);
 		} else if(this._state === BEFORE_CDATA_4){
-			if(c === "T") this._state = BEFORE_CDATA_5;
-			else this._state = IN_DECLARATION;
+			this._stateBeforeCdata4(c);
 		} else if(this._state === BEFORE_CDATA_5){
-			if(c === "A") this._state = BEFORE_CDATA_6;
-			else this._state = IN_DECLARATION;
+			this._stateBeforeCdata5(c);
 		} else if(this._state === BEFORE_CDATA_6){
-			if(c === "["){
-				this._state = IN_CDATA;
-				this._sectionStart = this._index + 1;
-			} else {
-				this._state = IN_DECLARATION;
-			}
+			this._stateBeforeCdata6(c);
 		} else if(this._state === IN_CDATA){
-			if(c === "]") this._state = AFTER_CDATA_1;
+			this._stateInCdata(c);
 		} else if(this._state === AFTER_CDATA_1){
-			if(c === "]") this._state = AFTER_CDATA_2;
-			else this._state = IN_CDATA;
+			this._stateAfterCdata1(c);
 		} else if(this._state === AFTER_CDATA_2){
-			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 (`]]]>`)
+			this._stateAfterCdata2(c);
 		}
@@ -346,17 +542,5 @@
 		else if(this._state === BEFORE_SPECIAL){
-			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;
-				continue; //consume the token again
-			}
+			this._stateBeforeSpecial(c);
 		} else if(this._state === BEFORE_SPECIAL_END){
-			if(this._special === 1 && (c === "c" || c === "C")){
-				this._state = AFTER_SCRIPT_1;
-			} else if(this._special === 2 && (c === "t" || c === "T")){
-				this._state = AFTER_STYLE_1;
-			} 
-			else this._state = TEXT;
+			this._stateBeforeSpecialEnd(c);
 		}
@@ -368,64 +552,23 @@
 		else if(this._state === BEFORE_SCRIPT_1){
-			if(c === "r" || c === "R"){
-				this._state = BEFORE_SCRIPT_2;
-			} else {
-				this._state = IN_TAG_NAME;
-				continue; //consume the token again
-			}
+			this._stateBeforeScript1(c);
 		} else if(this._state === BEFORE_SCRIPT_2){
-			if(c === "i" || c === "I"){
-				this._state = BEFORE_SCRIPT_3;
-			} else {
-				this._state = IN_TAG_NAME;
-				continue; //consume the token again
-			}
+			this._stateBeforeScript2(c);
 		} else if(this._state === BEFORE_SCRIPT_3){
-			if(c === "p" || c === "P"){
-				this._state = BEFORE_SCRIPT_4;
-			} else {
-				this._state = IN_TAG_NAME;
-				continue; //consume the token again
-			}
+			this._stateBeforeScript3(c);
 		} else if(this._state === BEFORE_SCRIPT_4){
-			if(c === "t" || c === "T"){
-				this._state = BEFORE_SCRIPT_5;
-			} else {
-				this._state = IN_TAG_NAME;
-				continue; //consume the token again
-			}
+			this._stateBeforeScript4(c);
 		} else if(this._state === BEFORE_SCRIPT_5){
-			if(c === "/" || c === ">" || whitespace(c)){
-				this._special = 1;
-			}
-			this._state = IN_TAG_NAME;
-			continue; //consume the token again
+			this._stateBeforeScript5(c);
 		}
 
 		else if(this._state === AFTER_SCRIPT_1){
-			if(c === "r" || c === "R"){
-				this._state = AFTER_SCRIPT_2;
-			} 
-			else this._state = TEXT;
+			this._stateAfterScript1(c);
 		} else if(this._state === AFTER_SCRIPT_2){
-			if(c === "i" || c === "I"){
-				this._state = AFTER_SCRIPT_3;
-			} 
-			else this._state = TEXT;
+			this._stateAfterScript2(c);
 		} else if(this._state === AFTER_SCRIPT_3){
-			if(c === "p" || c === "P"){
-				this._state = AFTER_SCRIPT_4;
-			} 
-			else this._state = TEXT;
+			this._stateAfterScript3(c);
 		} else if(this._state === AFTER_SCRIPT_4){
-			if(c === "t" || c === "T"){
-				this._state = AFTER_SCRIPT_5;
-			} 
-			else this._state = TEXT;
+			this._stateAfterScript4(c);
 		} else if(this._state === AFTER_SCRIPT_5){
-			if(c === ">" || whitespace(c)){
-				this._state = IN_CLOSING_TAG_NAME;
-				this._sectionStart = this._index - 6;
-				continue; //reconsume the token
-			} 
-			else this._state = TEXT;
+			this._stateAfterScript5(c);
 		}
@@ -437,57 +580,23 @@
 		else if(this._state === BEFORE_STYLE_1){
-			if(c === "y" || c === "Y"){
-				this._state = BEFORE_STYLE_2;
-			} else {
-				this._state = IN_TAG_NAME;
-				continue; //consume the token again
-			}
+			this._stateBeforeStyle1(c);
 		} else if(this._state === BEFORE_STYLE_2){
-			if(c === "l" || c === "L"){
-				this._state = BEFORE_STYLE_3;
-			} else {
-				this._state = IN_TAG_NAME;
-				continue; //consume the token again
-			}
+			this._stateBeforeStyle2(c);
 		} else if(this._state === BEFORE_STYLE_3){
-			if(c === "e" || c === "E"){
-				this._state = BEFORE_STYLE_4;
-			} else {
-				this._state = IN_TAG_NAME;
-				continue; //consume the token again
-			}
+			this._stateBeforeStyle3(c);
 		} else if(this._state === BEFORE_STYLE_4){
-			if(c === "/" || c === ">" || whitespace(c)){
-				this._special = 2;
-			}
-			this._state = IN_TAG_NAME;
-			continue; //consume the token again
+			this._stateBeforeStyle4(c);
 		}
 
 		else if(this._state === AFTER_STYLE_1){
-			if(c === "y" || c === "Y"){
-				this._state = AFTER_STYLE_2;
-			} 
-			else this._state = TEXT;
+			this._stateAfterStyle1(c);
 		} else if(this._state === AFTER_STYLE_2){
-			if(c === "l" || c === "L"){
-				this._state = AFTER_STYLE_3;
-			} 
-			else this._state = TEXT;
+			this._stateAfterStyle2(c);
 		} else if(this._state === AFTER_STYLE_3){
-			if(c === "e" || c === "E"){
-				this._state = AFTER_STYLE_4;
-			} 
-			else this._state = TEXT;
+			this._stateAfterStyle3(c);
 		} else if(this._state === AFTER_STYLE_4){
-			if(c === ">" || whitespace(c)){
-				this._state = IN_CLOSING_TAG_NAME;
-				this._sectionStart = this._index - 5;
-				continue; //reconsume the token
-			} 
-			else this._state = TEXT;
+			this._stateAfterStyle4(c);
 		}
 
-
 		else {
-			this._cbs.onerror(Error("unknown state"), this._state);
+			this._cbs.onerror(Error("unknown _state"), this._state);
 		}
@@ -498,25 +607,3 @@
 
-	//cleanup
-	if(this._sectionStart === -1){
-		this._buffer = "";
-		this._index = 0;
-	} else {
-		if(this._state === TEXT){
-			if(this._sectionStart !== this._index){
-				this._cbs.ontext(this._buffer.substr(this._sectionStart));
-			}
-			this._buffer = "";
-			this._index = 0;
-		} else if(this._sectionStart === this._index){
-			//the section just started
-			this._buffer = "";
-			this._index = 0;
-		} else if(this._sectionStart > 0){
-			//remove everything unnecessary
-			this._buffer = this._buffer.substr(this._sectionStart);
-			this._index -= this._sectionStart;
-		}
-
-		this._sectionStart = 0;
-	}
+	this._cleanup();
 };
@@ -555,15 +642,12 @@
 Tokenizer.prototype.reset = function(){
-	Tokenizer.call(this, this._options, this._cbs);
+	Tokenizer.call(this, {xmlMode: this._xmlMode}, this._cbs);
 };
 
-Tokenizer.prototype._emitToken = function(name){
-	this._cbs[name](this._buffer.substring(this._sectionStart, this._index));
-	this._sectionStart = -1;
+Tokenizer.prototype._getSection = function(){
+	return this._buffer.substring(this._sectionStart, this._index);
 };
 
-Tokenizer.prototype._emitIfToken = function(name){
-	if(this._index > this._sectionStart){
-		this._cbs[name](this._buffer.substring(this._sectionStart, this._index));
-	}
+Tokenizer.prototype._emitToken = function(name){
+	this._cbs[name](this._getSection());
 	this._sectionStart = -1;
 };

package.json

 {
 	"name": "htmlparser2",
 	"description": "Fast & forgiving HTML/XML/RSS parser",
-	"version": "3.1.5",
+	"version": "3.1.6",
 	"author": "Felix Boehm <[e-mail]>",
@@ -6,0 +6,0 @@ "keywords": ["html", "parser", "streams", "xml", "dom", "rss", "feed", "atom"],

README.md

@@ -29,3 +29,3 @@ #htmlparser2 [![NPM version](https://badge.fury.io/js/htmlparser2.png)](https://npmjs.org/package/htmlparser2) [![Build Status](https://secure.travis-ci.org/fb55/htmlparser2.png)](http://travis-ci.org/fb55/htmlparser2) [![Dependency Status](https://david-dm.org/fb55/htmlparser2.png)](https://david-dm.org/fb55/htmlparser2)
 });
-parser.write("Xyz <script type='text/javascript'>var foo = '<<bar>>';< /  script>");
+parser.write("Xyz <script type='text/javascript'>var foo = '<<bar>>';</ script>");
 parser.end();
@@ -32,0 +32,0 @@ ```

tests/99-benchmark.js

@@ -0,0 +0,0 @@ var multiply = function(text){

No alert changes

Improved metrics

Total package byte prevSize

121970

increased by2.21%

Number of package files

49

increased by4.26%

Lines of code

5544

increased by0.89%

No dependency changes