api: mark Format as #[non_exhaustive]

MA
2mo ago
+11-0

Scope

Marks the public Format enum #[non_exhaustive] and adds defensive wildcard arms to the in-repo Node and WASM bindings. No conversion behavior changes.

Motivation

#147 and #149 add Html and Mhtml variants to Format. Adding a variant to a public enum breaks downstream users with exhaustive match expressions (flagged by Cubic on #149, src/lib.rs:50). #[non_exhaustive] makes future variant additions forward-compatible: crates outside this workspace must include a wildcard arm, while references like Format::Pdf keep working unchanged.

Changes

  • src/lib.rs: #[non_exhaustive] on Format + doc note.
  • node/src/lib.rs and wasm/src/lib.rs: the bindings match anydoc::Format exhaustively in From<anydoc::Format>, so each gains an explicit other => unreachable!(...) arm. The bindings are released in lockstep with the core crate; the arm is defensive only.
  • Python bindings are untouched: they resolve format names through a lookup table, not an exhaustive match.

Validation (full gate on the exact candidate SHA 73f851b66c8f1926221bb4e4028e833e88685ef2, Windows, cargo 1.98.0)

  • Rust: cargo fmt --all --check, cargo clippy --workspace --all-targets --all-features -- -D warnings, cargo test --locked — all clean, all suites pass.
  • Node: npm ci / npm run build / npm test — 19 passed, 0 failed; committed node/index.js / node/index.d.ts are deterministic (zero diff after rebuild).
  • WASM: wasm32 Clippy clean; wasm-pack build wasm --release --target web --scope firecrawl; node --test wasm/test.mjs — 8 passed, 0 failed.
  • Python: locked Maturin release wheel, installed into a clean venv from site-packages, python -m unittest discover -s python/tests from the repo root — 11 tests OK.

Fixture note: this candidate is based on main, which intentionally does not include HTML/MHTML support, so the private real-fixture conversion matrix from the repository policy does not apply here (this PR changes no conversion logic).

Summary by cubic

Marks the Format enum #[non_exhaustive] so future variant additions (like the in-flight Html and Mhtml) won't break downstream matches. Downstream callers must now include a wildcard arm when matching. Adds defensive unreachable!() arms to the Node and WASM bindings' From<anydoc::Format> implementations; the Python bindings are unchanged since they resolve format names via a lookup table. No conversion behavior changes.