perf: optimize reference link masking

[sarthakagrawal927]

Marked version: 681373c

Markdown flavor: n/a

Description

Reference-link masking rebuilt Object.keys(this.tokens.links) for every
queued inline source, then linearly searched it for every candidate. At 2,000
definitions and paragraphs, that is about four million temporary key entries
and two million string comparisons.

This checks the existing table with Object.hasOwn, removing both costs without
adding lexer state or changing output.

Performance

Thirty alternating fresh-process pairs on Node.js 26.7.0, with per-process
warmup. Values are medians; intervals are paired 95% bootstrap intervals.

References	Baseline	Current	Change (95% interval)
100	0.805 ms/op	0.489 ms/op	-39.2% (-40.3, -38.8)
500	11.512 ms/op	2.265 ms/op	-80.4% (-80.5, -80.2)
2,000	164.115 ms/op	10.095 ms/op	-93.8% (-93.9, -93.8)

A 2,000-paragraph no-definition control moved +1.3% (+0.3, +2.3). Guards that
avoid that scan also re-enumerated the link table and erased most of the target
gain, so this keeps the smaller implementation and scopes the claim.

Validation

• npm test
• 1,779 specification tests and 190 unit tests passed
• ESM/UMD/CJS, type, build, and lint checks passed

Contributor

• Existing reference-link unit and CommonMark/GFM specification tests cover
  the unchanged parsing behavior.
• No documentation is required because this does not add or change a
  feature.

Committer

• CI is green (no forced merge required).
• Squash and Merge PR following conventional commit guidelines.

[vercel]

@sarthakagrawal927 is attempting to deploy a commit to the MarkedJS Team on Vercel.

A member of the Team first needs to authorize it.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
marked-website	Ready	Preview	Aug 10, 2026 3:25pm

[UziTech]

We should still check if (this.tokens.links && Object.entries(this.tokens.links).length > 0) to prevent checking if we don't need to