perf(lexer): reduce bounds checks in Lexer::get_string
#16317
Conversation
How to use the Graphite Merge QueueAdd either label to this PR to merge it via the merge queue:
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. |
github-actions
bot
added
the
C-performance
CodSpeed Performance ReportMerging #16317 will not alter performanceComparing Summary
Footnotes
|
There was a problem hiding this comment.
Pull request overview
This PR optimizes the Lexer::get_string method by reducing redundant string slicing operations. The original implementation sliced the source text twice - first to create a raw slice spanning the token, then again to trim quotes or the # symbol. The refactored code adjusts start/end indices directly before performing a single slice, eliminating 2 bounds checks and 2 UTF-8 character boundary checks.
Key Changes:
- Refactored
Lexer::get_stringto perform a single slice operation instead of two sequential slices - Adjusted start/end indices in-place based on token kind before slicing
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
overlookmotel
requested review from
camc314 and
camchenry
and removed request for
camc314
|
Very slight improvement in parser benchmarks. @camchenry I suggest we merge this and then rebase your PR on top, and see how much incremental difference the unsafe makes. |
overlookmotel
deleted the
12-01-perf_lexer_reduce_bounds_checks_in_lexer_get_string_
branch
### 💥 BREAKING CHANGES - 74cf572 ast: [**BREAKING**] Make `source` field of `TSImportType` a `StringLiteral` (#16114) (copilot-swe-agent) - 43156ae ast: [**BREAKING**] Rename `TSImportType` `argument` field to `source` (#16110) (overlookmotel) - 934d873 napi: [**BREAKING**] Drop `armv7-unknown-linux-musleabihf` support (#16105) (Boshen) ### 🚀 Features - 669afe0 ast: Add `Expression::is_jsx` method (#16154) (Dunqing) - 17a8caa parser: Add diagnostic for JSX identifiers with hyphens (#16133) (camchenry) - 0549ae5 parser: Add diagnostic for expected ident after optional chain (#16132) (camchenry) - db839ae parser: Improve diagnostic for unexpected optional declarations (#16131) (camchenry) - bab4bc8 napi/parser: Add type annotations to parse-raw-worker test (#15998) (camc314) ### 🐛 Bug Fixes - 35ed36c traverse: Fix panic when truncating non-ASCII variable names (#16265) (peter) - 9149a26 linter/plugins, napi/parser: Deep freeze visitor keys (#16293) (overlookmotel) - 6b54dab minifier: Incorrect non-null object condition simplification with `&&` and `||` (#16161) (sapphi-red) - 9cc20a1 minifier: Avoid merging side effectful expressions to next assignment statement if the side effect may change the left hand side reference (#16165) (sapphi-red) - 91eb3f2 ast/estree: Convert `TSImportType` `argument` field to `Literal` (#16109) (overlookmotel) - 1199cee parser: Reject invalid modifiers on parameter properties with binding patterns (#16083) (camc314) - f376325 traverse: Remove `console.log` from build script (#16049) (overlookmotel) ### ⚡ Performance - 82d784f lexer: Reduce bounds checks in `Lexer::get_string` (#16317) (overlookmotel) - cc2f352 span: Add `#[inline]` to `Atom` methods (#16311) (overlookmotel) - ffca070 span: Add `#[repr(transparent)]` to `Atom` (#16310) (overlookmotel) - 02bdf90 linter/plugins, napi/parser: Reuse arrays in visitor keys (#16294) (overlookmotel) ### 📚 Documentation - 891e0b4 parser: Add note about falling back to parse TSType in TSImportType (#16119) (camc314) Co-authored-by: Boshen <1430279+Boshen@users.noreply.github.com>
Lexer::get_stringwas slicing source text twice - once to obtainrawand then again to trim off a byte from start/end in the case of strings.Instead, slice only once. This removes 2 bounds checks and 2 x UTF-8 character boundary checks for strings.