feat(oxfmt): Support trailing ignore comments

[alubbe]

Summary

Prettier supports end-of-line // prettier-ignore (undocumented but widely used), especially for long logging/debug statements.
oxfmt previously only respected suppression comments before the node, which forced extra lines and reduced readability in these cases, see #14669 (comment).

In the discussion on #16915 (comment), it was stated that the performance could be negatively impacted.

This PR is an attempt to show how it could be supported without sacrificing performance.

So you can now write lines like this, which won't be transformed:

logger({ level: 'debug', message: `Really long debugging message about how ${user} called a certain part of our application with a certain ${payload}` }); // prettier-ignore

I'm very happy to receive feedback/criticism and iterate on this PR.

Performance Notes

• No extra AST pass.
• No backtracking.
• No new allocations or hash maps.
• Reuses existing on-demand comment queries and parser metadata (is_line, is_trailing, newline flags).

Benchmarking

cargo bench -p oxc_benchmark --bench formatter --no-default-features --features compiler -- --save-baseline pr


   Compiling oxc_formatter v0.31.0 (/Users/andreaslubbe/code/oxc/crates/oxc_formatter)
   Compiling oxc_benchmark v0.0.0 (/Users/andreaslubbe/code/oxc/tasks/benchmark)
    Finished `bench` profile [optimized] target(s) in 23.62s
     Running benches/formatter.rs (target/release/deps/formatter-a2b283f388af781c)
Benchmarking formatter/RadixUIAdoptionSection.jsx: Warming up fo
Benchmarking formatter/RadixUIAdoptionSection.jsx: Collecting 10
formatter/RadixUIAdoptionSection.jsx
                        time:   [34.680 µs 34.752 µs 34.824 µs]
                        change: [−0.1831% +0.7658% +1.4018%] (p = 0.05 < 0.05)
                        Change within noise threshold.
Found 6 outliers among 100 measurements (6.00%)
  2 (2.00%) low severe
  2 (2.00%) low mild
  1 (1.00%) high mild
  1 (1.00%) high severe
Benchmarking formatter/errors.ts: Collecting 100 samples in esti
formatter/errors.ts     time:   [61.615 µs 61.727 µs 61.840 µs]
                        change: [−0.4673% −0.1798% +0.0776%] (p = 0.20 > 0.05)
                        No change in performance detected.
Found 5 outliers among 100 measurements (5.00%)
  2 (2.00%) low mild
  3 (3.00%) high mild
Benchmarking formatter/Search.tsx: Collecting 100 samples in est
formatter/Search.tsx    time:   [193.20 µs 193.46 µs 193.71 µs]
                        change: [+0.1333% +0.3976% +0.6556%] (p = 0.00 < 0.05)
                        Change within noise threshold.
Found 4 outliers among 100 measurements (4.00%)
  1 (1.00%) low severe
  2 (2.00%) low mild
  1 (1.00%) high severe
Benchmarking formatter/core.js: Collecting 100 samples in estima
formatter/core.js       time:   [202.53 µs 202.83 µs 203.15 µs]
                        change: [−0.8484% −0.4336% −0.0516%] (p = 0.03 < 0.05)
                        Change within noise threshold.
Found 4 outliers among 100 measurements (4.00%)
  2 (2.00%) low severe
  2 (2.00%) low mild
Benchmarking formatter/next.ts: Collecting 100 samples in estima
formatter/next.ts       time:   [288.66 µs 289.09 µs 289.54 µs]
                        change: [−0.4780% +0.0460% +0.5260%] (p = 0.86 > 0.05)
                        No change in performance detected.
Found 3 outliers among 100 measurements (3.00%)
  1 (1.00%) high mild
  2 (2.00%) high severe
Benchmarking formatter/index.tsx: Collecting 100 samples in esti
formatter/index.tsx     time:   [519.27 µs 519.96 µs 520.70 µs]
                        change: [−1.6068% −0.3630% +0.5398%] (p = 0.58 > 0.05)
                        No change in performance detected.
Found 7 outliers among 100 measurements (7.00%)
  1 (1.00%) low mild
  3 (3.00%) high mild
  3 (3.00%) high severe
Benchmarking formatter/handle-comments.js: Warming up for 3.0000
Benchmarking formatter/handle-comments.js: Collecting 100 sample
formatter/handle-comments.js
                        time:   [333.26 µs 333.70 µs 334.17 µs]
                        change: [+0.1965% +0.4268% +0.6624%] (p = 0.00 < 0.05)
                        Change within noise threshold.
Found 2 outliers among 100 measurements (2.00%)
  2 (2.00%) high severe
Benchmarking formatter/types.ts: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase target time to 7.3s, enable flat sampling, or reduce sample count to 50.
Benchmarking formatter/types.ts: Collecting 100 samples in estim
formatter/types.ts      time:   [1.1793 ms 1.1808 ms 1.1825 ms]
                        change: [+0.1914% +0.5340% +0.8613%] (p = 0.00 < 0.05)
                        Change within noise threshold.
Found 5 outliers among 100 measurements (5.00%)
  1 (1.00%) low mild
  1 (1.00%) high mild
  3 (3.00%) high severe
Benchmarking formatter/App.tsx: Collecting 100 samples in estima
formatter/App.tsx       time:   [5.8312 ms 5.8385 ms 5.8462 ms]
                        change: [+0.7680% +0.9831% +1.1990%] (p = 0.00 < 0.05)
                        Change within noise threshold.
Found 1 outliers among 100 measurements (1.00%)
  1 (1.00%) high severe

AI Disclosure

This PR was created over 2 days of trying different 5 different approaches, which both Claude Opus 4.6 and OpenAI GPT 5.3 helped with, until settling for this version of trying to combine the best parts.

[Copilot]

Pull request overview

This PR adds support for end-of-line suppression comments (e.g. // prettier-ignore and // oxfmt-ignore) so the formatter can preserve original source text for a statement even when the suppression marker appears on the same line (matching common Prettier usage and reducing diff noise).

Changes:

• Add a comment query helper to detect suppression comments that appear as trailing end-of-line line comments.
• Preserve original source text for suppressed statements when suppression is at end-of-line (with special handling for ExpressionStatement to remain ASI-safe).
• Extend fixture-based snapshot coverage to include same-line suppression scenarios.

Reviewed changes

Copilot reviewed 10 out of 11 changed files in this pull request and generated 9 comments.

Show a summary per file

File	Description
tasks/ast_tools/src/generators/formatter/format.rs	Updates generator to emit inline trailing-suppression handling for Statement formatting (excluding ExpressionStatement).
crates/oxc_formatter/src/formatter/comments.rs	Adds has_line_suppression_comment_at_end_of_line to detect same-line trailing suppression.
crates/oxc_formatter/src/print/mod.rs	Implements ASI-safe preservation for ExpressionStatement when a same-line suppression is present.
crates/oxc_formatter/src/utils/statement_body.rs	Ensures same-line suppression works for if consequents where trailing comments are handled specially.
crates/oxc_formatter/src/ast_nodes/generated/format.rs	Generated output reflecting the new Statement suppression logic and required imports.
crates/oxc_formatter/tests/fixtures/js/...	Adds/updates fixtures & snapshots to validate same-line suppression behavior.

[codspeed-hq]

Merging this PR will not alter performance

✅ 49 untouched benchmarks
⏩ 7 skipped benchmarks¹

---

_{Comparing alubbe:support-trailing-comments (de164f6) with main (abc7e19)}

Footnotes

1. 7 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. ↩

[Dunqing]

Hi @fisker, can I know why the trailing ignore comment is undocumented?

[fisker]

Hi @fisker, can I know why the trailing ignore comment is undocumented?

I don't know, I think it's bug (but I don't think we are going to fix), I didn't know about it util few month ago.

[fisker]

Looks like intentionally prettier/prettier#671 (comment)

[Dunqing]

Looks like intentionally prettier/prettier#671 (comment)

Thank you! It looks like it just forgot to add documentation. 😅

[alubbe]

In that case, would you like to review this PR? I could also add documentation to it, if you think this is could be mergeable. I could also open a PR on prettier adding documentation for it.

[Dunqing]

In that case, would you like to review this PR? I could also add documentation to it, if you think this is could be mergeable. I could also open a PR on prettier adding documentation for it.

I am on holiday right now. I will review this PR when I come back. It would be better to add documentation in Prettier first!

[alubbe]

Sounds great, enjoy your holiday!

[Dunqing]

Works well! I pushed a commit to support trailing block ignore comment

[graphite-app]

Merge activity

• Mar 2, 3:42 AM UTC: @alubbe we removed the merge queue label because we could not find a Graphite account associated with your GitHub profile.

You must have a Graphite account in order to use the merge queue. Create an account and try again using this link