I’ve posted a rough beginnings of an implementation of xml5 for node.js. The core of this work is the tokenizer, for which I wrote a simple script to do the conversion of Anne van Kesteren’s implementation of the parse state methods to the style that Aria Stewart used for html5. Pretty much the remainder was “borrowed” from html5.
While this is not yet complete, you can see how it parses and dumps simple files with the following command:
node parse.js filename
Plenty still needs to be done. In particular:
When I run the above command, I get the following message which I will want to suppress:
###########################################################
# WARNING: No HTML parser could be found.
# Element.innerHTML setter support has been disabled
# Element.innerHTML getter support will still function
# Download: http://github.com/tautologistics/node-htmlparser
###########################################################
Accessing node names via DOM methods results in UPPERCASE values. I want to preserve case. For the moment, I do a case insensitive match on end tags with start tags. Thinking about it, that may be worth spec’ing and retaining.
As Anne indicated, it would be ideal to accept the full range of HTML5 entity names.
Namespaces in particular and the test suite in general need to be addressed
That message is coming from jsdom. You need to explicitly pass an instance of Aria’s parser to jsdom’s HtmlToDom method.
As the core dependency is jsdom which implements the W3C DOM standard, it should be possible to strip out the require statements and usage of Events, concatenate the rest, minimize the results and use the results in the browser of your choice.
get nodeName() {
var name = this._nodeName || this._tagName;
if (this.nodeType === this.ELEMENT_NODE &&
this.ownerDocument &&
this.ownerDocument.doctype &&
this.ownerDocument.doctype.name.indexOf("html") !== -1)
{
return name.toUpperCase();
}
return name;
},
Based on this, it looks like the workaround is to create a DOCTYPE for every XML document that doesn’t have one. Either that, or monkey-patch jsdom. Or switch to using localName, which brings up another potential issue: if I’m reading this code correctly, simple names without a colon are treated as being a prefix, and localName will not be set in such cases.
My quick performance testing has not shown any issues. In fact, due to the way events are pipelined, I’ve yet to see a case where the completion of parsing is more than 1 millisecond after the completion of the http request. That may not be the case for disk access.
var http = require('http'),
html5 = require('html5'),
jsdom = require('jsdom'),
window = jsdom.jsdom().createWindow(null, null, {parser: html5});
var rubix = http.createClient(80, 'planet.intertwingly.net');
var request = rubix.request('GET', '/', {'host': 'planet.intertwingly.net'});
request.end();
var start = new Date();
request.on('response', function (response) {
response.on('end', function () {
console.log('Response complete: ' + (new Date() - start));
});
var parser = new html5.Parser({document: window.document});
parser.parse(response);
parser.on('end', function() {
console.log('Parse complete: ' + (new Date() - start));
});
});
Web Standards Links: 10 January 2011 to 16 January 2011
Promoting inter opera bility is at the heart of Open The Web. If you had missed the topic of the week is Google planning to drop the support of H264 in Chrome. I’ll spare you the links, there were plenty all over twitter. Web standards...
John Cowan: I’ll openly admit at this point that I’m skeptical about the prospects of MicroXML. It doesn’t contain enough of XML to correctly parse feeds (be they RSS 1.0, RSS 2.0, or Atom). It doesn’t...
[more]
This is some nifty work you’re doing with Node.js here, Sam. The possibilities for expanding upon this are swirling around in my head right now. They seem limitless!