Exception flood caused by premature parsing

[Danielku15]

Hey there.

I am currently troubleshooting some performance problems in building my Docusaurus based website. While investigating I noticed that there was a flood of exceptions thrown in acorn while the MDX files are parsed. Looking up the callstack I noticed something suspicious: Micromark prematurely attempts to parse MDX expressions with acorn whenever it hits a } character:

https://github.com/micromark/micromark-extension-mdx-expression/blob/2891b75ff9e985c6df208a47348e76ced05dbfed/packages/micromark-factory-mdx-expression/dev/index.js#L156-L159

Assume following MDX being parsed: <Demo attr={[["1}","2}","3}","4}","5}","6}","7}","8}","9}","10}"]]} />
On every } character, it tries to parse the preliminary code with acorn leading to errors here. I added some logging there and this is what I see:

micromark-util-events-to-acorn:eventsToAcorn parseExpressionAt [["1
micromark-util-events-to-acorn:eventsToAcorn parse failed [["1 Unterminated string constant (1:2)
micromark-util-events-to-acorn:eventsToAcorn parseExpressionAt [["1}","2
micromark-util-events-to-acorn:eventsToAcorn parse failed [["1}","2 Unterminated string constant (1:7)
micromark-util-events-to-acorn:eventsToAcorn parseExpressionAt [["1}","2}","3
micromark-util-events-to-acorn:eventsToAcorn parse failed [["1}","2}","3 Unterminated string constant (1:12)
micromark-util-events-to-acorn:eventsToAcorn parseExpressionAt [["1}","2}","3}","4
micromark-util-events-to-acorn:eventsToAcorn parse failed [["1}","2}","3}","4 Unterminated string constant (1:17)
micromark-util-events-to-acorn:eventsToAcorn parseExpressionAt [["1}","2}","3}","4}","5
micromark-util-events-to-acorn:eventsToAcorn parse failed [["1}","2}","3}","4}","5 Unterminated string constant (1:22)
micromark-util-events-to-acorn:eventsToAcorn parseExpressionAt [["1}","2}","3}","4}","5}","6
micromark-util-events-to-acorn:eventsToAcorn parse failed [["1}","2}","3}","4}","5}","6 Unterminated string constant (1:27)
micromark-util-events-to-acorn:eventsToAcorn parseExpressionAt [["1}","2}","3}","4}","5}","6}","7
micromark-util-events-to-acorn:eventsToAcorn parse failed [["1}","2}","3}","4}","5}","6}","7 Unterminated string constant (1:32)
micromark-util-events-to-acorn:eventsToAcorn parseExpressionAt [["1}","2}","3}","4}","5}","6}","7}","8
micromark-util-events-to-acorn:eventsToAcorn parse failed [["1}","2}","3}","4}","5}","6}","7}","8 Unterminated string constant (1:37)
micromark-util-events-to-acorn:eventsToAcorn parseExpressionAt [["1}","2}","3}","4}","5}","6}","7}","8}","9
micromark-util-events-to-acorn:eventsToAcorn parse failed [["1}","2}","3}","4}","5}","6}","7}","8}","9 Unterminated string constant (1:42)
micromark-util-events-to-acorn:eventsToAcorn parseExpressionAt [["1}","2}","3}","4}","5}","6}","7}","8}","9}","10
micromark-util-events-to-acorn:eventsToAcorn parse failed [["1}","2}","3}","4}","5}","6}","7}","8}","9}","10 Unterminated string constant (1:47)
micromark-util-events-to-acorn:eventsToAcorn parseExpressionAt [["1}","2}","3}","4}","5}","6}","7}","8}","9}","10}"]]
micromark-util-events-to-acorn:eventsToAcorn parse successful [["1}","2}","3}","4}","5}","6}","7}","8}","9}","10}"]]

For my website with a lot generated reference docs (3228 files) in total, this sums up to 11304 exceptions.

I doubt that this is the intended behavior. The redundant parsing attempts and also the exceptions can lead to quite some performance hit.

A minimal example can be seen here: https://github.com/Danielku15/mdx-parsing This repo parses a MDX input file similar to how Docusaurus does it.

I wonder if there is something that could be improved around this parsing logic to avoid this trial&error parsing approach.

[wooorm]

This is indeed like that, I don’t think it can work some other way.