Repository navigation
perf: ~16% faster parsing by skipping em/strong masking for runs without *, _ or ~ - #4142
Merged
Merged
Conversation
|
@deadstrobe5 is attempting to deploy a commit to the MarkedJS Team on Vercel. A member of the Team first needs to authorize it. |
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
UziTech
approved these changes
Oct 17, 2026
UziTech
left a comment
UziTech
left a comment
Member
There was a problem hiding this comment.
Thanks for making marked faster! 💯
1 of 5 tasks
github-actions Bot
pushed a commit
that referenced
this pull request
Oct 26, 2026
## [18.1.2](v18.1.1...v18.1.2) (2026-10-26) ### Bug Fixes * allow a tab or end of input after a type 6 html block tag name ([#4128](#4128)) ([e09c40d](e09c40d)) * apply emphasis flanking and rules 9-10 to the whole delimiter run ([#4145](#4145)) ([5d8e234](5d8e234)) * **cli:** await config functions before parsing ([#4138](#4138)) ([3a11ad1](3a11ad1)) * **cli:** create no-clobber output files atomically ([#4135](#4135)) ([e7c4f64](e7c4f64)) * do not continue a block quote paragraph from a whitespace-only line ([#4134](#4134)) ([42f34a8](42f34a8)) * expand a tab after the block quote marker to the next tab stop ([#4110](#4110)) ([2d3899b](2d3899b)) * honor per-call async options when composing hooks ([#4139](#4139)) ([379be9d](379be9d)) * keep a line starting with # but no space in the list item ([#4127](#4127)) ([6b63189](6b63189)) * keep duplicate link definitions in the tokens ([#4132](#4132)) ([2784acc](2784acc)) * keep lazy lines after a list in a nested blockquote ([#4137](#4137)) ([38c7e55](38c7e55)) * keep lazy setext underlines in list items as text ([#4126](#4126)) ([398234c](398234c)) * render nested image alt formatting as plain text ([#4131](#4131)) ([15f59bb](15f59bb)) * stop an unclosed inline raw tag from unescaping later blocks ([#4130](#4130)) ([4562b6f](4562b6f)) ### Performance Improvements * ~16% faster parsing by skipping em/strong masking for runs without *, _ or ~ ([#4142](#4142)) ([f0fd3e0](f0fd3e0))
This was referenced Oct 30, 2026
1 of 5 tasks
This branch was successfully deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Marked version: master
Markdown flavor: n/a
Description
Performance:
marked.parseis ~16% faster on real-world READMEs (and ~9% on the CommonMark spec examples), with byte-identical output.#inlineTokensInnerbuildsmaskedSrc(reflink mask,anyPunctuationOutsideCodereplace,blockSkipreplace) for every inline run, but onlyemStronganddelread it, and their left-delimiter regexes all begin with*,_or~. 89% of inline runs in real READMEs (23,912 of 26,843 in the corpus below) contain none of these, so that work is thrown away.This moves the masking into a private
#maskSrc()and skips it when the run has no*,_or~. It still runs whenever something else could readmaskedSrc: anemStrongMaskhook is set, oremStrong/delare overridden (viamarked.useor a subclass).No regex was added or changed, so ReDoS exposure is unchanged.
Benchmark
marked.parse, default options, A/B interleaved in one process (alternating order), Node 26, Apple M5:README.mdfiles fromnode_modules(2.4 MB), 40 rounds × 5 runsAn A/A run of the original against itself gave a ratio of 1.000–1.010.
Safety
gfm: false,pedanticandbreaksoptions (same SHA-256 before and after).npm test: spec 1911/1911, unit 200/200;test:typesandtest:lintpass.emStrongtokenizer still receives the masked source in a run without em delimiters (it fails if the custom-tokenizer check is removed).Contributor
emStrongtokenizers); or,Committer
In most cases, this should be a different person than the contributor.