This repository was archived by the owner on Apr 29, 2026. It is now read-only.
Add typed exception hierarchy and regression-corpus scaffolding#17
Merged
Conversation
Typed exception hierarchy under llm_safe_pl.errors: - LlmSafeError (base, subclass of Exception) - MappingError, InputSizeError (also subclass ValueError so existing `except ValueError` code keeps catching) - DetectorError (also subclass RuntimeError; constructor signature is (detector_name) only — never input text or cause, to prevent PII from leaking into stack traces) Mapping.from_dict / from_json and the Shield input-size guard now raise the typed classes instead of bare ValueError. Constructor-time argument validation (Shield(max_input_bytes=-1), duplicate detector names) keeps raising plain ValueError. Regression-corpus scaffolding under tests/corpora/: - pl_pii_positive/ with three labeled samples (PESEL, email, IBAN) - pl_pii_negative/ with two safe-text samples - tests/test_corpus.py discovers .txt/.json pairs at collection time, asserts the default Shield matches every labeled span in positive samples and finds zero matches in negative samples - CONTRIBUTING.md gains an "Adding to the regression corpus" section with the JSON schema and offset rules 345 tests pass, coverage 96.50%, ruff/mypy clean.
Tatarinho
pushed a commit
that referenced
this pull request
Apr 29, 2026
…folding Resolves the CHANGELOG.md conflict between main's [Unreleased] entries (typed-exception hierarchy in llm_safe_pl.errors and tests/corpora/ regression scaffolding from PR #17) and the v0.2.1 deprecation tombstone that was branched directly from v0.2.0. CHANGELOG resolution: kept the v0.2.1 tombstone entry verbatim and replaced main's [Unreleased] block with an end-of-life note explaining that the typed-exception code is preserved in the source tree but will not ship to PyPI (v0.2.1 is final). Equivalent functionality lives in pii-veil.errors and pii-core. src/llm_safe_pl/__init__.py: auto-merged cleanly. The post-merge file imports the typed exception classes (LlmSafeError, MappingError, InputSizeError, DetectorError) AND fires the v0.2.1 DeprecationWarning on import. Anyone building from the archived repo source gets both; PyPI users on 0.2.1 only get the deprecation warning. Other files came in cleanly from main: src/llm_safe_pl/errors.py (new), docs/errors.md (new), tests/test_errors.py (new), tests/test_corpus.py (new), tests/corpora/ fixtures (new), and the typed-error refactors in models.py / shield.py / test_public_api.py / test_security_hardening.py. Local CI gates green on the merged tree: ruff check . -> All checks passed ruff format --check . -> 60 files already formatted mypy -> no issues found in 24 source files pytest -q -> 347 passed, 96.51% coverage
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Typed exception hierarchy under llm_safe_pl.errors:
except ValueErrorcode keeps catching)Mapping.from_dict / from_json and the Shield input-size guard now raise the typed classes instead of bare ValueError. Constructor-time argument validation (Shield(max_input_bytes=-1), duplicate detector names) keeps raising plain ValueError.
Regression-corpus scaffolding under tests/corpora/:
345 tests pass, coverage 96.50%, ruff/mypy clean.