public api regex

README.md

@@ -102,8 +102,11 @@ console.log(sanitized);
  * @param {Function} [callbacks.cdata] Takes the content of the CDATA
  * @param {Function} [callbacks.xmlProlog] Takes no arguments
  * @param {Function} [callbacks.text] Takes the value of the text node
+ * @param {Object} [regex]
+ * @param {RegExp} [regex.name] Regex for element name. Default is [a-zA-Z_][\w:\-\.]*
+ * @param {RegExp} [regex.attribute] Regex for attribute name. Default is [a-zA-Z_][\w:\-\.]*
  */
-parse(htmlString, callbacks)
+parse(htmlString, callbacks, regex)
 
 /**
  * Parses the HTML contained in the given file asynchronously.

src/context.js

@@ -1,4 +1,4 @@
-exports.create = function(raw, options) {
+exports.create = function(raw, options, regex) {
 	var index = 0;
 	var context = {
 		text: '',
@@ -78,5 +78,16 @@ exports.create = function(raw, options) {
 		};
 	});
 
+	context.regex = {
+		name: /[a-zA-Z_][\w:\-\.]*/,
+		attribute: /[a-zA-Z_][\w:\-\.]*/
+	};
+	regex = regex || {};
+	for (var name in regex) {
+		if (regex.hasOwnProperty(name)) {
+			context.regex[name] = regex[name];
+		}
+	}
+
 	return context;
-}
+};

src/parser.js

@@ -1,8 +1,7 @@
 var parseContext = require('./context');
-var nameRegex = /[a-zA-Z_][\w:\-\.]*/;
 
 function readAttribute(context) {
-	var name = context.readRegex(nameRegex);
+	var name = context.readRegex(context.regex.attribute);
 	var value = null;
 	if (context.current === '=' || context.peekIgnoreWhitespace() === '=') {
 		context.readRegex(/\s*=\s*/);
@@ -30,7 +29,7 @@ function readAttributes(context, isXml) {
 
 	var next = context.current;
 	while (!context.isEof() && !isClosingToken()) {
-		if (nameRegex.test(next)) {
+		if (context.regex.attribute.test(next)) {
 			readAttribute(context);
 			next = context.current;
 		}
@@ -68,7 +67,7 @@ function readCloserForOpenedElement(context, name) {
 }
 
 function parseOpenElement(context) {
-	var name = context.readRegex(nameRegex);
+	var name = context.readRegex(context.regex.name);
 	context.callbacks.openElement(name);
 	readAttributes(context, false);
 	readCloserForOpenedElement(context, name);
@@ -90,7 +89,7 @@ function parseOpenElement(context) {
 }
 
 function parseEndElement(context) {
-	var name = context.readRegex(nameRegex);
+	var name = context.readRegex(context.regex.name);
 	context.callbacks.closeElement(name);
 	context.readRegex(/.*?(?:>|$)/);
 }
@@ -150,7 +149,7 @@ function parseNext(context) {
 		buffer += context.read();
 		if (context.current === '/') {
 			buffer += context.read();
-			if (nameRegex.test(context.current)) {
+			if (context.regex.name.test(context.current)) {
 				callbackText(context);
 				parseEndElement(context);
 			} else {
@@ -182,7 +181,7 @@ function parseNext(context) {
 				context.read();
 				appendText(buffer, context);
 			}
-		} else if (nameRegex.test(context.current)) {
+		} else if (context.regex.name.test(context.current)) {
 			callbackText(context);
 			parseOpenElement(context);
 		} else {
@@ -213,10 +212,13 @@ function parseNext(context) {
  * @param {Function} [callbacks.cdata] Takes the content of the CDATA
  * @param {Function} [callbacks.xmlProlog] Takes no arguments
  * @param {Function} [callbacks.text] Takes the value of the text node
+ * @param {Object} [regex]
+ * @param {RegExp} [regex.name] Regex for element name. Default is [a-zA-Z_][\w:\-\.]*
+ * @param {RegExp} [regex.attribute] Regex for attribute name. Default is [a-zA-Z_][\w:\-\.]*
  */
-exports.parse = function(htmlString, callbacks) {
+exports.parse = function(htmlString, callbacks, regex) {
 	htmlString = htmlString.replace(/\r\n/g, '\n').replace(/\r/g, '\n');
-	var context = parseContext.create(htmlString, callbacks);
+	var context = parseContext.create(htmlString, callbacks, regex);
 	do {
 		parseNext(context);
 	} while (!context.isEof());

tests/attribute-tests.js

@@ -172,4 +172,25 @@ describe('attributes', function() {
 		openCount.should.equal(1);
 		attrCount.should.equal(1);
 	});
+
+	it('with custom attribute regex', function() {
+		var openCount = 0, attrCount = 0;
+		helpers.parseString('<foo [bar]="baz">', {
+			openElement: function(name) {
+				name.should.equal('foo');
+				openCount++;
+			},
+
+			attribute: function(name, value) {
+				name.should.equal('[bar]');
+				value.should.equal('baz');
+				attrCount++;
+			}
+		}, {
+			attribute: /\[[\w\]]*/
+		});
+
+		openCount.should.equal(1);
+		attrCount.should.equal(1);
+	});
 });

tests/helpers.js

@@ -2,6 +2,6 @@ var htmlParser = require('../src/parser.js');
 
 exports.parser = htmlParser;
 
-exports.parseString = function(string, options) {
-	htmlParser.parse(string, options);
+exports.parseString = function(string, options, regex) {
+	htmlParser.parse(string, options, regex);
 };