v0.6.0

[v0.6.0] - 2026-03-10

Added

• Added the public Schema base class for class-based schema definitions while keeping validation on the existing dict engine.
• Added zodify/schema.py as the extracted runtime module for class-based schemas while keeping ValidatedDict as an internal runtime carrier.
• Added comprehensive Schema-vs-dict parity coverage, typing-contract checks, package-artifact checks, and release-gate validation for the public class-based schema surface.
• Added README Class-Based Schemas documentation with side-by-side dict/class examples, nested composition, supported boundaries, and intentional non-goals.

Changed

• Aligned package metadata, module version, and tests on 0.6.0.
• Updated package/README positioning from "one file" to "one engine" to match the extracted runtime layout.

v0.5.0

[v0.5.0] - 2026-03-06

Added

• Added Validator class with reusable default configuration (coerce, max_depth, unknown_keys, error_mode) and per-call overrides.
• Added comprehensive tests/test_validator_class.py coverage for defaults, override non-mutation, multi-instance isolation, and parity with bare validate().
• Added README Configuration section documenting Validator usage and call-level override patterns while retaining bare validate() as the recommended starting point.

Changed

• Refactored mode-option validation into shared internal helper logic to keep validate() and Validator option contracts aligned.
• Updated static typing configuration for pyright package-targeted checks ([tool.pyright] include = ["zodify"]).
• Updated benchmark comparison harness to use zodify's dict-schema API directly.

v0.4.1

[v0.4.1] - 2026-03-04

Added

• Added six runnable example scripts in examples/: basic_validation.py, nested_schemas.py, custom_validators.py, union_types.py, env_config.py, and structured_errors.py.
• Added tests/test_examples.py smoke coverage for required example presence, runnable entrypoints, expected-output verification, and runtime API usage checks.
• Added public API export parity assertion to keep __version__ included in zodify.__all__.
• Added package-boundary verification policy for examples (wheel excludes runtime examples; source distribution may include them for documentation visibility).
• Added README Schema Composition section showing plain-dict schema reuse, nested composition, nested union member usage, and a runnable reference to examples/nested_schemas.py.
• Added README consistency updates: Schema composition comparison-table row, roadmap status aligned to v0.4.1, and release tagging example updated to git tag v0.4.1.

v0.4.0

[v0.4.0] - 2026-03-02

Added

• ValidationError exception class: subclasses ValueError with .issues attribute containing structured list[dict] of validation failures. Each issue dict has path, message, expected, and got keys.
• error_mode parameter on validate(): error_mode="structured" raises ValidationError with machine-readable .issues; default error_mode="text" preserves existing ValueError string behavior.
• Structured error output covers all error types including union mismatch (str | int), coercion failure, custom validator failure, depth exceeded, and unknown key errors.
• copy/deepcopy/pickle support for ValidationError via __reduce__.

v0.3.0

[v0.3.0] - 2026-03-02

Added

• Union type support: str | int runtime schema syntax for validating values against multiple types.
• Union coercion with left-to-right priority: when coerce=True, union members are tried in declaration order — first exact match (non-str types), then coercion attempts. First success wins.

Known Limitations

• str catch-all in union coercion: when str is a union member and coerce=True, any value that fails earlier union members will fall through to str() coercion (e.g., int | str with True produces "True"). Place str last in unions to use it as a fallback.

v0.2.1

[v0.2.1] - 2026-03-01

Added

• PEP 561 py.typed marker file for type checker recognition.
• Inline type annotations on all public and internal symbols.
• @overload signatures for env() enabling IDE type inference (e.g., env("PORT", int) infers int).
• Google-style docstrings with usage examples on all public API symbols (validate, env, Optional).
• mypy (strict mode) and pyright CI gates as merge-blocking checks.

Changed

• Internal function docstrings (_coerce_value, _check_value, _validate) trimmed to one-line per architecture pattern.
• flake8 CI step changed to report-only (--exit-zero) — type checkers now handle correctness gating.

Known Limitations

• Optional default values are not type-checked. Optional(int, "text") will use the string default without validation. Verify your defaults match the expected type.
• When coerce=True with target type str, any value is accepted via Python's str() builtin (e.g., str(None) becomes "None"). Non-string targets require string input for coercion.

v0.2.0

[v0.2.0] - 2026-02-27

Added

• max_depth parameter on validate() with default of 32 to prevent stack overflow from deeply nested data.
• Custom validator functions: use any callable (lambda or function) as a schema value for value-level validation beyond type checking.
• unknown_keys parameter on validate() with default "reject" to control handling of extra keys in data. Use "strip" for previous silent-ignore behavior.
• Performance benchmark scripts (benchmarks/) for compliance verification: import time, validate speed, env overhead.

Changed

• Internal error representation refactored from format strings to structured 4-tuples (path, message, expected, got) for future structured error support.
• validate() parameters after data are now keyword-only (enforced via *).
• test_env.py extracted from test_zodify.py for cleaner test organization.
• Unknown keys in data are now rejected by default. Use unknown_keys="strip" to restore v0.1.0 behavior.
• CI pipeline now includes benchmark reporting on Python 3.12 (non-gating).

v0.1.0

[v0.1.0] - 2026-02-25

Added

• Nested dict schema validation with recursive validation and dot-path errors (for example, db.port).
• Optional schema marker with optional defaults for missing keys.
• List element typing via single-element schema lists (for example, {"tags": [str]}), including nested list-of-dict support.
• Broader automated test coverage for nested schemas, optionals, list validation, and coercion edge cases.

Changed

• Updated README with v0.1.0 examples for nested dicts, optional keys, and list element validation.
• Updated package metadata and versioning to 0.1.0.

Fixed

• CI/publish workflow robustness for install/test/build flow and release handling.
• Packaging metadata cleanup in pyproject.toml (license format/classifier cleanup).

v0.0.1

Full Changelog: https://github.com/junyoung2015/zodify/commits/v0.0.1