WHATWG HTML5 specification-compliant, fast and ready for production HTML parsing/serialization toolset for Node and io.js.
I needed fast and ready for production HTML parser, which will parse HTML as a modern browser's parser. Existing solutions were either too slow or their output was too inaccurate. So, this is how parse5 was born.
Included tools:
- Parser - HTML to DOM-tree parser.
- SimpleApiParser - SAX-style parser for HTML.
- Serializer - DOM-tree to HTML code serializer.
Install
$ npm install parse5
Usage
var Parser = require('parse5').Parser;
//Instantiate parser
var parser = new Parser();
//Then feed it with an HTML document
var document = parser.parse('<!DOCTYPE html><html><head></head><body>Hi there!</body></html>')
//Now let's parse HTML-snippet
var fragment = parser.parseFragment('<title>Parse5 is fucking awesome!</title><h1>42</h1>');
Is it fast?
Check out this benchmark.
Starting benchmark. Fasten your seatbelts...
html5 (https://github.com/aredridel/html5) x 0.18 ops/sec ±5.92% (5 runs sampled)
htmlparser (https://github.com/tautologistics/node-htmlparser/) x 3.83 ops/sec ±42.43% (14 runs sampled)
htmlparser2 (https://github.com/fb55/htmlparser2) x 4.05 ops/sec ±39.27% (15 runs sampled)
parse5 (https://github.com/inikulin/parse5) x 3.04 ops/sec ±51.81% (13 runs sampled)
Fastest is htmlparser2 (https://github.com/fb55/htmlparser2),parse5 (https://github.com/inikulin/parse5)
So, parse5 is as fast as simple specification incompatible parsers and ~15-times(!) faster than the current specification compatible parser available for the node.
API reference
Enum: TreeAdapters
Provides built-in tree adapters which can be passed as an optional argument to the Parser and Serializer constructors.
• TreeAdapters.default
Default tree format for parse5.
• TreeAdapters.htmlparser2
Quite popular htmlparser2 tree format (e.g. used in cheerio and jsdom).
Class: Parser
Provides HTML parsing functionality.
• Parser.ctor([treeAdapter, options])
Creates new reusable instance of the Parser. Optional treeAdapter argument specifies resulting tree format. If treeAdapter argument is not specified, default tree adapter will be used.
options object provides the parsing algorithm modifications:
options.decodeHtmlEntities
Decode HTML-entities like &, , etc. Default: true. Warning: disabling this option may cause output which is not conform HTML5 specification.
options.locationInfo
Enables source code location information for the nodes. Default: false. When enabled, each node (except root node) has __location property, which contains start and end indices of the node in the source code. If element was implicitly created by the parser it's __location property will be null. In case the node is not an empty element, __location has two addition properties startTag and endTag which contain location information for individual tags in a fashion similar to __location property.
Example:
var parse5 = require('parse5');
//Instantiate new parser with default tree adapter
var parser1 = new parse5.Parser();
//Instantiate new parser with htmlparser2 tree adapter
var parser2 = new parse5.Parser(parse5.TreeAdapters.htmlparser2);
• Parser.parse(html)
Parses specified html string. Returns document node.
Example:
var document = parser.parse('<!DOCTYPE html><html><head></head><body>Hi there!</body></html>');
• Parser.parseFragment(htmlFragment, [contextElement])
Parses given htmlFragment. Returns documentFragment node. Optional contextElement argument specifies context in which given htmlFragment will be parsed (consider it as setting contextElement.innerHTML property). If contextElement argument is not specified then <template> element will be used as a context and fragment will be parsed in 'forgiving' manner.
Example:
var documentFragment = parser.parseFragment('<table></table>');
//Parse html fragment in context of the parsed <table> element
var trFragment = parser.parseFragment('<tr><td>Shake it, baby</td></tr>', documentFragment.childNodes[0]);
Class: SimpleApiParser
Provides SAX-style HTML parsing functionality.
• SimpleApiParser.ctor(handlers, [options])
Creates new reusable instance of the SimpleApiParser. handlers argument specifies object that contains parser's event handlers. Possible events and their signatures are shown in the example.
options object provides the parsing algorithm modifications:
options.decodeHtmlEntities
Decode HTML-entities like &, , etc. Default: true. Warning: disabling this option may cause output which is not conform HTML5 specification.
options.locationInfo
Enables source code location information for the tokens. Default: false. When enabled, each node handler receives location object as it's last argument. location object contains start and end indices of the token in the source code.
Example:
var parse5 = require('parse5');
var parser = new parse5.SimpleApiParser({
doctype: function(name, publicId, systemId /*, [location] */) {
//Handle doctype here
},
startTag: function(tagName, attrs, selfClosing /*, [location] */) {
//Handle start tags here
},
endTag: function(tagName /*, [location] */) {
//Handle end tags here
},
text: function(text /*, [location] */) {
//Handle texts here
},
comment: function(text /*, [location] */) {
//Handle comments here
}
});
• SimpleApiParser.parse(html)
Raises parser events for the given html.
Example:
var parse5 = require('parse5');
var parser = new parse5.SimpleApiParser({
text: function(text) {
console.log(text);
}
});
parser.parse('<body>Yo!</body>');
Class: Serializer
Provides tree-to-HTML serialization functionality.
Note: prior to v1.2.0 this class was called TreeSerializer. However, it's still accessible as parse5.TreeSerializer for backward compatibility.
• Serializer.ctor([treeAdapter, options])
Creates new reusable instance of the Serializer. Optional treeAdapter argument specifies input tree format. If treeAdapter argument is not specified, default tree adapter will be used.
options object provides the serialization algorithm modifications:
options.encodeHtmlEntities
HTML-encode characters like <, >, &, etc. Default: true. Warning: disabling this option may cause output which is not conform HTML5 specification.
Example:
var parse5 = require('parse5');
//Instantiate new serializer with default tree adapter
var serializer1 = new parse5.Serializer();
//Instantiate new serializer with htmlparser2 tree adapter
var serializer2 = new parse5.Serializer(parse5.TreeAdapters.htmlparser2);
• Serializer.serialize(node)
Serializes the given node. Returns HTML string.
Example:
var document = parser.parse('<!DOCTYPE html><html><head></head><body>Hi there!</body></html>');
//Serialize document
var html = serializer.serialize(document);
//Serialize <body> element content
var bodyInnerHtml = serializer.serialize(document.childNodes[0].childNodes[1]);
Testing
Test data is adopted from html5lib project. Parser is covered by more than 8000 test cases. To run tests:
$ npm test
Custom tree adapter
You can create a custom tree adapter so parse5 can work with your own DOM-tree implementation. Just pass your adapter implementation to the parser's constructor as an argument:
var Parser = require('parse5').Parser;
var myTreeAdapter = {
//Adapter methods...
};
//Instantiate parser
var parser = new Parser(myTreeAdapter);
Sample implementation can be found here.
The custom tree adapter should implement all methods exposed via exports in the sample implementation.
Questions or suggestions?
If you have any questions, please feel free to create an issue here on github.
Author
Ivan Nikulin (ifaaan@gmail.com)