Slow performance decoding binary bytes

[wader]

Hi, I noticed while fuzzing fq that it sometimes times out and it turned out to be gomarkdown being quite slow when fed random binary data. Did a quick CPU profile and most of the time is spent in parser.codeSpan: https://github.com/gomarkdown/markdown/blob/90efaac57fb4d6c93a2f5dff442a6ae0774de269/parser/inline.go#L151-L161

A bit weird "use case" i admit but maybe it could be used for denial-of-service attacks etc? maybe the parser should fail somehow instead if hard to make faster?

Attaching file produced by go fuzzer. File takes around 15s to parser on my laptop.

test2.md

[kjk]

It does look like we could break out of the loop when we set hasCharsAfterDelimiter = true instead of looping till the end.

[kjk]

Doing the break speeds things up by 5000+x:

before:
BenchmarkIssue265-16                   1        18239097588 ns/op         787176 B/op        162 allocs/op

after:
BenchmarkIssue265-16                 448           3093183 ns/op          787170 B/op        161 allocs/op

[wader]

Pretty good :)

[kjk]

Thanks for the test case!

[wader]

No problem, thanks for quick fix and thanks for gomarkdown, works great! I use it in fq for being able to use markdown both for web and cli documentation. Is implemented as "format" you can decode, was the reason it got fuzzed.