natemoo-re / node-dom-parse-benchparks

Node.js DOM performance benchmarks

⚠️ This is a cheap test! My usecase was parsing a unique document once, and running a query selector once. This is very different from parsing a document once, and running many, many queries on it.

Methodology

1. Parse a long document (the GitHub.com homepage, © GitHub)
2. Query for a specific element, using whatever the parser has available
3. Repeat 999×

Results

Package	Result	Comparison
hypertag	1.48s	—
fast-html-parser	1.63s	10% slower
htmlparser2	3.69s	149% slower
node-html-parser	5.24s	254% slower
sax-wasm*	10.82s	631% slower
cheerio	22.48s	1419% slower
fast-html-dom-parser	94.48s	6284% slower
jsdom**	???

*I know what you’re thinking—“why is WASM slower?” Like I said, my methodology here was parsing a document 1,000×. WASM has a very expensive startup cost that JS doesn’t. I’m sure that once it’s loaded, it would run circles around everything else. So take these benchmarks with a grain of salt if you’re reusing the same document and not throwing it away each time like I was doing.

** jsdom ran out of memory after taking at least 30 seconds. I didn’t get to benchmark it, but that told me all I needed to know (dunno what run it was on either).