V3 Coverage Strategy¶

Goal¶

Maintain full line/branch coverage for v3 behavior while keeping tests readable and representative of user workflows.

Testing split¶

Common usage paths (user-facing)¶

Prefer doctests in documentation/examples.
Cover stable API entry points and common data flows.
Keep examples readable and copy/paste friendly.

Edge and failure paths (engineering-facing)¶

Prefer pytest in tests/test_000_detextive.
Cover boundary and error behaviors.
Use dependency injection surfaces (detector registries, behavior arguments) instead of monkey-patching internals.

Component focus areas¶

Detection and Inference¶

Detector ordering and fallback behavior.
Header/location context handling.
Default-vs-error policy behavior with confidence expectations.

Decoding¶

Trial decode sequencing and validator interactions.
HTTP Content-Type charset behavior, including non-textual rejection.
decode_inform textual MIME guarantees and metadata shape.

Validation¶

Profile behavior for reasonable vs unreasonable text.
Confidence-threshold behavior and validation gating.

Coverage workflow¶

Run hatch --env develop run testers for the combined coverage view.
Use hatch --env develop run coverage report --show-missing to identify exact uncovered lines/branches.
Add/adjust doctests for common user-facing paths.
Add/adjust pytest cases for edge/error branches.
Re-run testers and verify coverage remains at target.