Skip to content

perf(estree/tokens): do not JSON-encode keyword, punctuator, etc tokens#19814

Merged
graphite-app[bot] merged 1 commit intomainfrom
om/02-22-perf_estree_tokens_do_not_json-encode_keyword_punctuator_etc_tokens
Feb 27, 2026
Merged

perf(estree/tokens): do not JSON-encode keyword, punctuator, etc tokens#19814
graphite-app[bot] merged 1 commit intomainfrom
om/02-22-perf_estree_tokens_do_not_json-encode_keyword_punctuator_etc_tokens

Conversation

@overlookmotel
Copy link
Member

@overlookmotel overlookmotel commented Feb 27, 2026

After #19744, #19813, #19812, and #19796, all tokens which require special handling or can contain values which are possibly not JSON-safe are handled separately. So everything remaining (keywords, punctuators etc) are JSON-safe and can be serialized via the optimized serializer introduced for identifiers in #19744.

To reflect that EstreeIdentToken now handles a lot more than identifiers, rename it to EstreeSafeToken, and rename EstreeToken to EstreeUnsafeToken.

Large positive perf impact. +24% on ESTree tokens benchmark.

@github-actions github-actions bot added the C-performance Category - Solution not expected to change functional behavior, only performance label Feb 27, 2026
Copy link
Member Author

overlookmotel commented Feb 27, 2026


How to use the Graphite Merge Queue

Add either label to this PR to merge it via the merge queue:

  • 0-merge - adds this PR to the back of the merge queue
  • hotfix - for urgent hot fixes, skip the queue and merge this PR next

You must have a Graphite account in order to use the merge queue. Sign up using this link.

An organization admin has enabled the Graphite Merge Queue in this repository.

Please do not merge from GitHub as this will restart CI on PRs being processed by the merge queue.

This stack of pull requests is managed by Graphite. Learn more about stacking.

@codspeed-hq
Copy link

codspeed-hq bot commented Feb 27, 2026

Merging this PR will improve performance by 24.01%

⚡ 1 improved benchmark
✅ 51 untouched benchmarks
⏩ 3 skipped benchmarks1

Performance Changes

Mode Benchmark BASE HEAD Efficiency
Simulation estree_tokens[checker.ts] 127.7 ms 102.9 ms +24.01%

Comparing om/02-22-perf_estree_tokens_do_not_json-encode_keyword_punctuator_etc_tokens (7369657) with om/02-22-perf_estree_tokens_do_not_json-encode_this_identifiers (efa2364)

Open in CodSpeed

Footnotes

  1. 3 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports.

@overlookmotel overlookmotel marked this pull request as ready for review February 27, 2026 18:24
Copilot AI review requested due to automatic review settings February 27, 2026 18:24
@overlookmotel overlookmotel added A-parser Area - Parser A-linter-plugins Area - Linter JS plugins labels Feb 27, 2026
Copy link
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

Optimizes ESTree token JSON serialization by routing JSON-safe tokens (keywords, punctuators, numbers, booleans, null, and most identifiers) through a fast manual serializer, while keeping escape-checked serialization for token values that may contain JSON-unsafe bytes.

Changes:

  • Introduces EstreeSafeToken (manual JSON assembly) and renames the previous general token type to EstreeUnsafeToken.
  • Refactors token emission to default to the safe fast-path and adds visitors to ensure string/template/JSXText tokens are emitted via the unsafe path.
  • Tightens invariants around which token kinds may reach get_token_type (debug assertion) after splitting safe vs unsafe handling.

@overlookmotel overlookmotel added the 0-merge Merge with Graphite Merge Queue label Feb 27, 2026
Copy link
Member Author

overlookmotel commented Feb 27, 2026

Merge activity

…ns (#19814)

After #19744, #19813, #19812, and #19796, all tokens which require special handling or can contain `value`s which are possibly not JSON-safe are handled separately. So everything remaining (keywords, punctuators etc) are JSON-safe and can be serialized via the optimized serializer introduced for identifiers in #19744.

To reflect that `EstreeIdentToken` now handles a lot more than identifiers, rename it to `EstreeSafeToken`, and rename `EstreeToken` to `EstreeUnsafeToken`.

Large positive perf impact. +24% on ESTree tokens benchmark.
graphite-app bot pushed a commit that referenced this pull request Feb 27, 2026
Remove the `regex` field from `EstreeToken` and serialize regex tokens via their own dedicated type `EstreeRegExpToken`. This removes a branch and logic from `emit_token` which is the path the processes all tokens except identifiers.

This PR is actually showing as a small perf degradation (-1% on ESTree tokens benchmark). But it enables a much bigger optimization later on in #19814.
graphite-app bot pushed a commit that referenced this pull request Feb 27, 2026
Same as #19744, but for `this` identifiers in TS types. Serialize them using the same optimized `EstreeIdentToken` serializer added in that PR.

This isn't a hot path so makes little difference, but it completes the preconditions for the next PR #19814.
@graphite-app graphite-app bot force-pushed the om/02-22-perf_estree_tokens_do_not_json-encode_this_identifiers branch from efa2364 to 6260ddd Compare February 27, 2026 18:40
@graphite-app graphite-app bot force-pushed the om/02-22-perf_estree_tokens_do_not_json-encode_keyword_punctuator_etc_tokens branch from 7369657 to 81bab90 Compare February 27, 2026 18:40
@graphite-app graphite-app bot removed the 0-merge Merge with Graphite Merge Queue label Feb 27, 2026
Base automatically changed from om/02-22-perf_estree_tokens_do_not_json-encode_this_identifiers to main February 27, 2026 18:48
@graphite-app graphite-app bot merged commit 81bab90 into main Feb 27, 2026
22 checks passed
@graphite-app graphite-app bot deleted the om/02-22-perf_estree_tokens_do_not_json-encode_keyword_punctuator_etc_tokens branch February 27, 2026 18:49
This was referenced Feb 28, 2026
camc314 pushed a commit that referenced this pull request Mar 2, 2026
### 🚀 Features

- 733d6dc parser: Report error on `infer` outside conditional type
(#19879) (camc314)
- c2a42f6 allocator: Add `Vec::into_bump_slice_mut` (#19895)
(overlookmotel)
- ee4982b parser: Add `VARIANTS` const to `Kind` via `fieldless_enum!`
macro (#19877) (overlookmotel)
- b3dceae data_structures: Add `fieldless_enum!` macro (#19876)
(overlookmotel)
- 12b841e parser: Make all `Kind::is_*` methods `const` (#19874)
(overlookmotel)
- 25c2e25 estree/tokens: Add function to update tokens in place (#19856)
(overlookmotel)
- f78e6df parser: Add `mutate_tokens` Cargo feature (#19853)
(overlookmotel)
- 5036bb6 parser: Report error on `for await` in static blocks (#19844)
(camc314)
- 42bd431 parser: Report error for missing initializer in using decl
(#19824) (camc314)
- a2f58e5 parser: Report error for `implements` clause in non-ts files
(#19820) (Cameron)
- b25228a estree: Add `IS_COMPACT` const to `Formatter` trait (#19787)
(overlookmotel)
- e2a1b79 estree: Expose buffer and formatter of serializers (#19773)
(overlookmotel)
- 4699498 data_structures: Add `CodeBuffer::print_strs_array` (#19760)
(overlookmotel)
- 233f947 estree: `oxc_estree` crate export config and formatter types
(#19724) (overlookmotel)
- 5937a32 semantic: Introduce `symbol_declarations` method (#19609)
(camc314)
- ea6b796 parser: Add `LexerConfig::TOKENS_METHOD_IS_STATIC` const
(#19683) (overlookmotel)
- 655c38f semantic: Add "did you mean?" suggestions to undefined name
errors (#19102) (copilot-swe-agent)
- 9e11dc6 parser,estree,coverage: Collect tokens in parser and convert
to ESTree format (#19497) (camc314)
- c4a3677 parser: Report error for initializer in ambient context
(#19187) (camc314)

### 🐛 Bug Fixes

- abc7e19 codegen: Improve parenthesised checks when printing types
(#19880) (camc314)
- 017de5d parser: Update error code for type annotation in `for...in`
statement (#19882) (camc314)
- 7682e5a linter/plugins: Decode escapes in identifier tokens (#19838)
(overlookmotel)
- 06767ed estree/tokens: Convert `this` tokens in `TSTypeName` (#19815)
(overlookmotel)
- ef798af parser: Use TS8037 for satisfies expression in JS files
diagnostic (#19819) (camc314)
- 98ea5c5 parser: Use TS8016 for type assertions in JS files diagnostic
(#19818) (camc314)
- 1710f56 codegen: Remove double indentation for enum inside namespace
(#19775) (Dunqing)
- 9e4995c codegen: Print type annotation on `CatchParameter` (#19790)
(camc314)
- 297b2bb codegen: Wrap `TSConditionalType` in parens when necessary
(#19788) (camc314)
- cec7878 codegen: Print `definite` property on AccessorProperty
(#19786) (camc314)
- 6f395cf codegen: Print `definite` property on PropertyDefinition
(#19785) (camc314)
- b749373 codegen: Correctly parenthesise TSArrayType (#19784) (camc314)
- 876dc1b codegen: Print object property `this` param (#19783) (camc314)
- 93bb861 formatter: Trim trailing whitespace before breaking line
(#19740) (leaysgur)
- ed17bbf codegen: Print `override` keyword for method and property
definitions (#19753) (Dunqing)
- 6a59a76 parser: Improve error recovery for private identifiers in
property names (#19710) (Boshen)
- 3b96f41 codegen: Print comments in JSX expression containers and
spread attributes (#19701) (Boshen)
- f5694ce estree/tokens: Reverse field order of `regex` object in tokens
(#19679) (overlookmotel)
- b2b7a55 estree/tokens: Generate tokens for files with BOM (#19535)
(overlookmotel)
- 50a7514 estree: Fix tokens for JSX (#19524) (overlookmotel)
- a35063e minifier: Preserve side effects for meta property url reads
(#19668) (Boshen)
- 8ad3430 semantic/jsdoc: Handle even-numbered backtick sequences in
JSDoc parsing (#19664) (Boshen)

### ⚡ Performance

- 05ccf9f linter/plugins: Transfer tokens via raw transfer (#19893)
(overlookmotel)
- c1bfdcf estree/tokens: Preallocate sufficient space for tokens JSON
(#19851) (overlookmotel)
- 4b0611a estree/tokens: Introduce `ESTreeTokenConfig` trait (#19842)
(overlookmotel)
- 81bab90 estree/tokens: Do not JSON-encode keyword, punctuator, etc
tokens (#19814) (overlookmotel)
- 6260ddd estree/tokens: Do not JSON-encode `this` identifiers (#19813)
(overlookmotel)
- b378f4a estree/tokens: Do not JSON-encode JSX identifiers (#19812)
(overlookmotel)
- 5016d92 estree/tokens: Handle regex tokens separately (#19796)
(overlookmotel)
- 780a68e estree/tokens: Use strings from AST for identifier tokens
(#19744) (overlookmotel)
- dc9c2e3 estree: Use `CodeBuffer::print_strs_array` to reduce bounds
checks (#19766) (overlookmotel)
- 845da35 estree: Use `CodeBuffer::print_indent` (#19727)
(overlookmotel)
- ec88f6a estree/tokens: Serialize tokens while visiting AST (#19726)
(overlookmotel)
- bc6507f estree/tokens: Serialize with `ESTree` not `serde` (#19725)
(overlookmotel)
- ec24859 estree/tokens: Do not branch on presence of override twice
(#19721) (overlookmotel)
- dac14be estree/tokens: Replace hash map with `Vec` (#19718)
(overlookmotel)
- b9d2443 estree/tokens: Replace multiple hash sets into a single hash
map (#19716) (overlookmotel)
- 7233548 parser: Remove branches from `finish_next_inner` (#19695)
(overlookmotel)
- b5d9845 parser: Remove const generic param from `finish_next_inner`
(#19684) (overlookmotel)
- 8940f66 estree/tokens: Serialize tokens to compact JSON (#19572)
(overlookmotel)
- 136e39b parser/tokens: Pre-allocate capacity for tokens (#19543)
(overlookmotel)
- 6a6513c linter/plugins: Use Oxc tokens in plugins (#19498) (camc314)
- b3b2d30 parser: Introduce `ParserConfig` (#19637) (overlookmotel)

### 📚 Documentation

- b2b7a64 estree/tokens: Correct comment (#19873) (overlookmotel)
- 0399311 estree/tokens: Improve comments (#19836) (overlookmotel)
- 1b392de minifier: Add `Function.prototype.toString` assumption
(#19758) (sapphi-red)
- 75c9cd8 parser: Improve doc comments for `ParserConfig` and
`LexerConfig` (#19682) (overlookmotel)
- 2fa936f README.md: Map npm package links to npmx.dev (#19666) (Boshen)

Co-authored-by: Boshen <1430279+Boshen@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

A-linter-plugins Area - Linter JS plugins A-parser Area - Parser C-performance Category - Solution not expected to change functional behavior, only performance

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants