[Kludex]

Summary

Add opt-in MAX_FIELDS and MAX_FIELD_SIZE configuration for URL-encoded forms in FormParser. Count fields before assembling them and check combined raw name/value bytes before buffering incoming data. Both limits default to infinity, raise FormParserError when exceeded, and leave multipart forms and file uploads unchanged.

This lets Starlette reuse upstream field assembly instead of maintaining its own message queue. The caller still calls finalize() after stream exhaustion.

Closes #332.

Validation

• Full suite: 230 tests passed with 100% coverage.
• New public-API tests cover exact boundaries, chunk splits, empty chunks, duplicate names, bare and empty fields, invalid limits, and rejection before large allocations.
• All 68 new tests also pass on Python 3.11 and 3.15 with warnings treated as errors.
• Ruff, mypy, strict documentation build, runnable documentation example, and sdist checks passed.

The full suite ignores pre-existing ResourceWarnings from unclosed files in three existing tests; these also reproduce on unchanged main.

Performance

Local interleaved microbenchmarks on CPython 3.13 measured approximately 10-14% overhead for small URL-encoded forms and 2% for a 1 MiB field, compared with the existing high-level FormParser. Whether replacing Starlette's queue improves end-to-end performance still needs an integration benchmark.

AI Disclaimer

This PR was developed with the assistance of either Claude or Codex. I've reviewed and verified the changes.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-04T10:01:50.890164Z	fc7d442	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[codspeed]

Merging this PR will not alter performance

✅ 5 untouched benchmarks
⏩ 5 skipped benchmarks¹

---

_{Comparing add-formparser-field-limits (fc7d442) with main (212819d)}

Footnotes

1. 5 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. ↩

[cubic-dev-ai]

No issues found across 4 files

_{Re-trigger cubic}

[MaciejZet]

I checked fc7d442 and its local integration with main@51ff1a4 on macOS arm64, using CPython 3.11–3.15.0rc1.

MAX_FIELDS and MAX_FIELD_SIZE matched expectations across bytewise input, empty chunks, and every one/two-cut layout of a 25-body corpus, including splits inside %26 and %3D. Exact boundaries, over-limit rejection and final fields behaved consistently. All 104 additional tests passed on both trees, including the unchanged multipart control.

The full PR suite had 230 passed; main + PR + the additional tests had 335 passed, with 100% coverage on each Python version. The combined 172 limit/review tests also passed with -W error.

On Python 3.11, Atheris 2.3.0 could not build because Apple Clang lacks libFuzzer. I used --no-install-package atheris and UV_NO_SYNC=true for that environment. No unit tests were skipped; Atheris fuzzing was not run. These are local results.