#htmlparser2 [](https://npmjs.org/package/htmlparser2) [](http://travis-ci.org/fb55/htmlparser2) [](https://david-dm.org/fb55/htmlparser2) A forgiving HTML/XML/RSS parser written in JS for NodeJS. The parser can handle streams (chunked data) and supports custom handlers for writing custom DOMs/output. ##Installing npm install htmlparser2 A live demo of htmlparser2 is available at http://demos.forbeslindesay.co.uk/htmlparser2/ ##Usage ```javascript var htmlparser = require("htmlparser2"); var parser = new htmlparser.Parser({ onopentag: function(name, attribs){ if(name === "script" && attribs.type === "text/javascript"){ console.log("JS! Hooray!"); } }, ontext: function(text){ console.log("-->", text); }, onclosetag: function(tagname){ if(tagname === "script"){ console.log("That's it?!"); } } }); parser.write("Xyz <script type='text/javascript'>var foo = '<<bar>>';</ script>"); parser.end(); ``` Output (simplified): ```javascript --> Xyz JS! Hooray! --> var foo = '<<bar>>'; That's it?! ``` Read more about the parser in the [wiki](https://github.com/fb55/htmlparser2/wiki/Parser-options). ##Get a DOM The `DomHandler` (known as `DefaultHandler` in the original `htmlparser` module) produces a DOM (document object model) that can be manipulated using the [`DomUtils`](https://github.com/fb55/DomUtils) helper. The `DomHandler`, while still bundled with this module, was moved to its [own module](https://github.com/fb55/domhandler). Have a look at it for further information. ##Parsing RSS/RDF/Atom Feeds ```javascript new htmlparser.FeedHandler(function(<error> error, <object> feed){ ... }); ``` ##Performance After having some artificial benchmarks for some time, __@AndreasMadsen__ published his [`htmlparser-benchmark`](https://github.com/AndreasMadsen/htmlparser-benchmark), which benchmarks HTML parses based on real-world websites. At the time of writing, the latest versions of all supported parsers show the following performance characteristics on [Travis CI](https://travis-ci.org/AndreasMadsen/htmlparser-benchmark/builds/10805007) (please note that Travis doesn't guarantee equal conditions for all tests): ``` gumbo-parser : 34.9208 ms/file ± 21.4238 html-parser : 24.8224 ms/file ± 15.8703 html5 : 419.597 ms/file ± 264.265 htmlparser : 60.0722 ms/file ± 384.844 htmlparser2-dom: 12.0749 ms/file ± 6.49474 htmlparser2 : 7.49130 ms/file ± 5.74368 hubbub : 30.4980 ms/file ± 16.4682 libxmljs : 14.1338 ms/file ± 18.6541 parse5 : 22.0439 ms/file ± 15.3743 sax : 49.6513 ms/file ± 26.6032 ``` ##How is this different from [node-htmlparser](https://github.com/tautologistics/node-htmlparser)? This is a fork of the `htmlparser` module. The main difference is that this is intended to be used only with node (it runs on other platforms using [browserify](https://github.com/substack/node-browserify)). `htmlparser2` was rewritten multiple times and, while it maintains an API that's compatible with `htmlparser` in most cases, the projects don't share any code anymore. The parser now provides a callback interface close to [sax.js](https://github.com/isaacs/sax-js) (originally targeted at [readabilitySAX](https://github.com/fb55/readabilitysax)). As a result, old handlers won't work anymore. The `DefaultHandler` and the `RssHandler` were renamed to clarify their purpose (to `DomHandler` and `FeedHandler`). The old names are still available when requiring `htmlparser2`, so your code should work as expected.