02 Jan 11:07

v3.0.0 Latest

Latest

Add Pandas integration and unify result APIs

Added

Add pandas export helpers (to_series(), to_dataframe()) behind an optional pandas extra (cobjectric[pandas]).
Add new result types/collections to support unified metric results and aggregation across models/lists.

Fixed

Validate metric function outputs more strictly by raising dedicated errors when similarity/fill-rate-accuracy values are invalid or out of range.

Updated

Move built-in similarity functions/decorators to cobjectric.similarity (replace cobjectric.similarities) and update public exports accordingly.
Refactor list comparison typing to use the unified ModelResult API instead of fill-rate-specific result types.

Removed

Remove reliance on fill-rate-specific result classes as the primary public result API in favor of unified metric result classes.

Full Changelog: v2.1.1...v3.0.0

Assets 4

01 Jan 22:19

v2.1.1

Mark library as production-ready and enhance documentation

Updated

Update development status classifier from "Alpha" to "Production/Stable" in package metadata
Enhance documentation section in README with comprehensive overview of available resources
Improve formatting and clarity in specifications documentation
Add documentation badge to README for better visibility

Full Changelog: v2.1.0...v2.1.1

Assets 4

01 Jan 21:46

v2.1.0

Pre-defined Specs and Contextual Normalizers

Added

Add pre-defined Specs for common field types:
- KeywordSpec for identifiers, codes, and enums with exact matching, optional whitespace stripping, and int-to-string conversion
- TextSpec for free-form text with fuzzy matching and comprehensive normalization (lowercase, strip, collapse spaces, remove accents)
- NumericSpec for numeric fields with intelligent type coercion (JSON Number → Python int/float) and optional tolerance-based similarity
- BooleanSpec for boolean fields with exact matching and value conversion
- DatetimeSpec for datetime strings with ISO format normalization, auto-detection of common formats, and time difference tolerance for similarity
Add contextual normalizers that receive FieldContext (field name, type, and spec) for type-aware normalization
Add FieldContext dataclass to pass context information to normalizers
Add datetime_similarity_factory() for datetime comparison with optional time difference tolerance
Add automatic detection of normalizer signature (simple 1-param vs contextual 2-param) in BaseModel

Updated

Update FieldSpec to support both simple and contextual normalizers
Update BaseModel._build_combined_normalizer() to handle contextual normalizers and pass FieldContext when needed
Update field processing to preserve all FieldSpec attributes when creating combined normalizers

Full Changelog: v2.0.0...v2.1.0

Assets 4

01 Jan 18:13

v2.0.0

Similarity, List Comparison Strategies, Path Access, and Enhanced Metrics

Added

Add similarity computation to compare field values between models with fuzzy text matching support
Add fill rate accuracy computation to compare data completeness between two models
Add path access notation to access nested fields using dot notation (e.g., result["address.city"] or result["items[0].name"])
Add list comparison strategies for comparing list fields with three alignment algorithms:
- Pairwise comparison by index (default)
- Levenshtein-based alignment for order-preserving matching
- Optimal assignment using Hungarian algorithm for best one-to-one matching (requires scipy)
Add built-in similarity functions:
- Exact similarity for strict equality matching
- Fuzzy similarity factory with multiple rapidfuzz scorers (ratio, partial_ratio, token_sort_ratio, token_set_ratio, WRatio, QRatio)
- Numeric similarity factory with configurable difference thresholds
Add statistical aggregation methods (mean, std, var, min, max, quantile) on all result objects
Add support for custom similarity functions per field via Spec(similarity_func=...) or @similarity_func decorator
Add support for custom fill rate accuracy functions per field via Spec(fill_rate_accuracy_func=...) or @fill_rate_accuracy_func decorator
Add aggregated field access for list results to compute statistics across list items via items.aggregated_fields
Add improved model representation with __repr__ showing field values and missing fields
Add scipy dependency for optimal assignment list comparison strategy

Updated

Update fill rate computation to support nested models and list aggregation
Update field specifications to accept ListCompareStrategy enum directly in Spec()
Update error handling with new exceptions for invalid aggregated fields and list comparison strategy usage
Update weight validation to support both fill_rate_weight and fill_rate_accuracy_weight

Fixed

Fix path resolution for nested list structures and complex nested models

Full Changelog: v1.1.0...v2.0.0

Assets 4

31 Dec 17:30

v1.1.0

Core Model System with Fill Rate Computation

Added

Add BaseModel class for defining typed models with automatic type validation
Add Field class to represent individual model fields with type, value, and specification
Add FieldCollection class providing attribute-based access to model fields
Add FieldSpec and Spec() function for field configuration with metadata, normalizers, fill rate functions, and weights
Add MissingValue sentinel for representing missing or invalid field values
Add field_normalizer decorator for transforming field values before validation
Add fill rate computation system with compute_fill_rate() method and statistical aggregation (mean, weighted mean)
Add support for nested models with recursive fill rate computation
Add support for list fields with typed elements (list[str], list[int], list[BaseModel])
Add support for optional fields using | None type annotations
Add support for union types in field definitions
Add comprehensive exception hierarchy (CobjectricError, UnsupportedTypeError, UnsupportedListTypeError, MissingListTypeArgError, DuplicateFillRateFuncError, InvalidFillRateValueError, InvalidWeightError)
Add from_dict() class method for creating model instances from dictionaries
Add readonly model instances (immutable after creation)
Add documentation with quick start guide, API reference, and usage examples

Removed

Remove status() function from public API

Full Changelog: v1.0.1...v1.1.0

Assets 4

07 Dec 14:30

v1.0.1

Enhance readme & CI metadata

Added

Add Codecov integration for coverage reporting
Add project classifiers, keywords, and URLs to package metadata
Add README badges for CI status, coverage, PyPI version, downloads, and code style
Add automatic README version update in release workflow

Updated

Update Python version requirement from >=3.13.9 to >=3.13
Update release workflow to publish on official PyPI instead of TestPyPI
Update coverage report format for Codecov compatibility
Improve README documentation formatting and installation instructions

Full Changelog: v1.0.0...v1.0.1

Assets 4

07 Dec 13:56

v1.0.0

Hello World

Updated

Update release workflow to publish packages to official PyPI instead of TestPyPI

Full Changelog: v0.2.2...v1.0.0

Assets 4

07 Dec 13:38

v0.2.2

Improve release workflow with lockfile sync and changelog links

Added

Add automatic uv.lock update step during version bump in release workflow
Add dynamic changelog comparison link between previous and current tag in GitHub releases

Updated

Update version bump commit to include both pyproject.toml and uv.lock
Update GitHub release notes to show proper changelog comparison URL (fallback to commits list for first release)

Full Changelog: v0.2.1...v0.2.2

Assets 4

07 Dec 13:22

v0.2.1

Fix GitHub release description

Fixed

Modified the release workflow to generate a release notes file instead of using inline notes, addressing shell escaping issues.

Full Changelog: ...v0.2.1

Assets 4