victornpb / benchmark-html-parser-libraries Goto Github PK

this test setup does not support html streaming which is imho one of the biggest speed factors... as parsing can be already done when you have finished reading the file...(the beauty of a declarative language like html)

But yeah if you do get the html as a string and do not read it from the file system (which can be streamed) then it totally makes sense...

but it should be clarified that a comparison where you get the raw string and streaming is impossible is the goal here

in my experience a very fast parser. I currently process 800+ html files within 3s (I'm only reading a few attributes)

https://github.com/justinwilaby/sax-wasm

I think many parsers are built so they can parse multiple html files one after another while using the same instance.

I assume you will get fastly different numbers if you rewrite this https://github.com/victornpb/benchmark-html-parser-libraries/blob/master/tests/sax.js

const sax = require("sax");

module.exports = function (html) {
	return new Promise((resolve, reject) => {
		const parser = sax.parser(false);

		parser.onend = resolve;
		parser.onerror = reject;
		parser.write(html);
		parser.close();
	});
};

to something like this

const sax = require("sax");
const parser = sax.parser(false);

module.exports = function (html) {
	return new Promise((resolve, reject) => {
		parser.onend = resolve;
		parser.onerror = reject;
		parser.write(html);
		parser.close();
	});
};

Hey, Im currently developing an html parser and in my own benchmarks it was 2-3x times faster then the current top tiers, How Can I test my own html parser with this package?