V0 Changelog#

0.49.0 (2026-05-26)#

Added#

  • DataProviderToolkit.drop_discrepant_processed_endpoint_tables_rows for dropping the discrepant rows

Changed#

  • FMP and LSEG fundamentals: on CommonDataDiscrepancyError, drop the discrepant rows entirely instead of nulling their non-key columns and retrying.

  • DataProviderToolkit.format_consolidated_discrepancy_table_for_output is now a class method

  • Added visual separators before and after discrepancy table logs for better visual separation and consistency

Fixed#

  • Handle FMP “Column ‘FundamentalDataRow.filing_date’ not found in table.” errors

  • DataProviderToolkit.consolidate_processed_endpoint_tables crashed with pyarrow.lib.ArrowInvalid: Data type null is not supported in join non-key field __indicator_for_validity when an endpoint table had zero rows; the validity indicator array is now explicitly typed as bool.

Removed#

  • DataProviderToolkit.clear_discrepant_processed_endpoint_tables_rows

  • DataProviderToolkit._clear_table_rows_by_primary_key

0.48.1 (2026-05-04)#

Fixed#

  • MarketDataDailyRow non-negative value validation was leaking to fields added to extended entities

  • Entity validation error messages now mention the correct entity name for extended entities

0.48.0 (2026-05-01)#

Changed#

  • Moved __all__ declarations to the top of the files

Fixed#

  • FMP fundamentals: handle all-null columns that would crash with pyarrow.lib.ArrowInvalid: Data type null is not supported in join non-key field

  • FMP fundamentals: handle duplicate statement periods crashing with DataProviderToolkitRuntimeError: Primary key merge table contains duplicate rows.

  • FMP fundamentals: handle some but not all endpoints having no rows, crashing with KeyError: 'Field "FundamentalDataRow.filing_date$filingDate" does not exist in schema'

  • FMP fundamentals: handle duplicate statement primary keys crashing with pyarrow.lib.ArrowInvalid: Filter inputs must all be the same length

  • FMP fundamentals: handle mismatched statement key columns crashing with TypeError: '<' not supported between instances of 'NoneType' and 'datetime.date'

0.47.0 (2026-04-14)#

Added#

  • Official support for Python 3.14

Changed#

  • Breaking: parameters_datacurator.xlsx configuration file renamed to data_curator_parameters.xlsx

  • Breaking: FundamentalDataRowCashFlow.net_cash_from_investing_activites renamed to FundamentalDataRowCashFlow.net_cash_from_investing_activities to fix typo

  • custom_calculations.py template c_test calculation now uses split adjusted prices for better provider compatibility

  • Simplified Excel configuration entry script template

  • Include Yahoo Finance data provider in the production docker image

0.46.1 (2026-03-17)#

Fixed#

  • Handle error when FMP data provider returns empty market data

0.46.0 (2026-03-02)#

Added#

  • LSEG Workspace data provider

  • More data provider error exceptions

0.45.1 (2026-02-09)#

Fixed#

  • Handle error pyarrow.lib.ArrowInvalid: Filter inputs must all be the same length when there’s duplicate filing dates for different statements in FinancialModelingPrep fundamental data

0.45.0 (2025-12-16)#

Changed#

  • Load ReadTheDocs dependencies from pyproject.toml instead of requirements.txt

Fixed#

  • DataProviderToolkit entity field mapping methods fail on subclassed entities

  • BaseDataBlock entity packing methods fail on subclassed entities

0.44.0 (2025-12-11)#

Added#

  • BaseDataEntity class as new parent class for all data entities

  • First partial implementation of data blocks, for generalizing data entity assembly from consolidated data tables

  • BaseDataBlock class as data block parent class, containing common entity assembling logic

  • Data block classes for all current data entities

  • DataProviderToolkit for generalizing the logic for constructing the consolidated data tables passed to the data blocks, including validation and better error handling/debugging

  • Dependency on the networkx library for topological sorting and related uses

  • Devcontainer configuration

Changed#

  • Moved dividend and split date and factor field declarations out of ColumnBuilder and into the respective entity modules

  • Standardized the way in which data providers declare endpoints and their respective entity field to tag mappings, including tag preprocessors

  • Refactored FinancialModelingPrep to use the new generalized entity creation and validation APIs

Deprecated#

  • services/entity_helper.py module will be removed in a future version, with all functionality being moved into the BaseDataBlock class.`

0.43.1 (2025-09-09)#

Removed#

  • ta and pandas-ta unused dependencies

0.43.0 (2025-08-26)#

Changed#

  • Improved README file

Fixed#

  • DataCulumn division between int columns should return float column

0.42.0 (2025-08-04)#

Added#

  • DataProviderPaymentError and DataProviderConnectionError exceptions

  • FinancialModelingPrep now has class methods for setting and getting whether the user’s account plan is paid

  • pdm run docs script as shortcut to the corresponding docs maker for the current OS. Use with pdm run docs html, etc.

Changed#

  • DataProviderInterface._request_data() now has special handling for 402 Payment Required errors

  • FinancialModelingPrep fundamental data methods now handle “The values for ‘limit’ must be between 0 and 5 based on your current subscription” 402 Payment Required errors

  • data_curator.main() now handles uncaught DataProviderPaymentError exceptions

0.41.0 (2025-07-02)#

Added#

  • DataColumn.__hash__ method for hashing DataColumns

  • InMemoryOutput output handler for saving data to memory

  • CI: CODEOWNERS file for GitHub code review automation

  • Docs: Inserted hidden toctree entries in each category index.rst to ensure all functions are registered in the Sphinx documentation build.

  • Docs: Added Use Cases section, with links to the Data-Curator-Use-Cases repo

Changed#

  • Docs: Refactored features_extension.py to group calculation functions by category for documentation generation.

  • Docs: Calculation functions are now organized in per-category folders under api/, improving maintainability.

  • Docs: Each category generates its own index.rst file with a clean table layout listing all functions.

  • Docs: The Section Navigation panel now displays categories as expandable subsections instead of a flat list.

  • Docs: Added :ref:-based linking for function references without affecting Excel configuration behavior.

0.40.2 (2025-06-05)#

Fixed#

  • Cli init excel command not creating .env file in the Config folder

  • c_market_cap calculation now uses split-adjusted close price, to reduce the jumps right after a split

  • ReadTheDocs integration basic setup

0.40.1 (2025-06-04)#

Fixed#

  • OS Error on cli init script because of entry script template missing from wheel data

0.40.0 (2025-06-04)#

Changed#

  • First public release, now on PyPI

0.39 (2025-06-03)#

Changed#

  • ticker has been renamed to main_identifier in Excel configuration and all related code

  • Ticker entity is now a MainIdentifier entity, with Ticker.symbol becoming MainIdentifier.identifier

  • MainIdentifier.identifier now only validates that no whitespace is present

  • TickerNotFoundError is now IdentifierNotFoundError

  • Configuration entity no longer validates identifiers format, as different data providers will have different required formats

  • fi_ prefix for income statement columns now becomes fis_

  • Versioning changed to remove 1.0b prefix, as version 1.0 still requires quite a few features.

  • FundamentalDataRowIncome entity is now FundamentalDataRowIncomeStatement

  • DataProviderInterface.init_config() is now DataProviderInterface.initialize()

  • DividendDataRow.adjusted_dividend renamed to DividendDataRow.split_adjusted_dividend

  • MarketDataDailyRow entity fields added for all split-adjusted and dividend-and-split-adjusted data sets

  • FMP updated all endpoints to the new “stable” version

  • FMP class and all related code usages now renamed to FinancialModelingPrep;

  • Parameters file now uses financial_modeling_prep instead of fmp as the provider name

  • Renamed most calculation feature functions for consistency

  • Updated Excel parameters file template to the new public API

Fixed#

  • CLI update excel command not working on dev

Removed#

  • MarketDataDailyRow.adjusted_close field, as it is now redundant

0.38 (2025-05-08)#

Added#

  • DataColumn comparison operators ==, !=, <, <=, >, >= for use in binary logic, filters, etc.

  • DataColumn.boolean_and and DataColumn.boolean_or methods for boolean comparisons between multiple BooleanArray DataColumns

  • Missing DataColumn reverse arithmetic operation tests

  • DataColumn._replace_array_mask_with_nones private method to replace the mask of a DataColumn with None values

  • DataColumn related exceptions

Changed#

  • DataColumn divisions involving decimal inputs now output as float64, as any decimal precision was lost in the division anyway and becomes innecessary

0.37 (2025-04-11)#

Added#

  • Created conftest.py with pytest fixtures to load example data from CSV files used in calculation tests.

  • Helper methods in helpers.py to compute technical indicators, including Exponential Moving Average (EMA).

  • Pytest coverage for calculations.py, reaching 96%.

Changed#

  • Prefixed all functions in calculations.py with c_.

  • Alphabetically ordered all functions in calculations.py and their corresponding tests.

  • Replaced usage of ta library for technical indicators in almost all features, except for c_chaikin_money_flow and c_relative_strength_index_14d, which are still under development.

  • Removed CSV-based test data fixtures previously defined directly in calculations_test.py.

0.36 (2025-03-24)#

Changed#

  • Simplify dev installation procedure in README.md and Dockerfile

Fixed#

  • ExcelConfigurator: Data providers returning validate_api_key() as None should not raise an error

  • ColumnBuilder: Extended Fundamental entity properties fail when full data period contains no fundamental data

0.35 (2025-03-13)#

Fixed#

  • DataColumn extended entity subclasses should now work with empty data on the first rows

0.34 (2025-03-04)#

Added#

  • DataProviderInterface tests (incomplete)

Changed#

  • DataProviderInterface._find_first_date_before_start_date_in_descending_dates renamed to DataProviderInterface._find_first_date_before_start_date, added descending_order parameter

  • DataProviderInterface._find_unordered_dates_in_descending_dates renamed to DataProviderInterface._find_unordered_dates, added descending_order parameter

0.33 (2025-02-20)#

Changed#

  • Feature and custom calculation function names should now always start with a c_ prefix.

  • Improved some error texts’ legibility

Deprecated#

  • Feature and custom calculation function names without the c_ prefix will stop working in the public release version.

0.32 (2025-02-17)#

Changed#

  • Breaking: entity_helper.fill_fields renamed to entity_helper.convert_data_row_into_entity_fields

  • entity_helper.convert_data_row_into_entity_fields now skips the type conversion logic if the value is already the expected type.

0.31 (2025-02-13)#

Changed#

  • Can now subclass all fundamental statement row entities to add custom data columns

Fixed#

  • Regression: Debugger initialization was not being called

0.30 (2025-01-31)#

Changed#

  • Python 3.13 now officially supported

  • Can now subclass MarketDataDailyRow to add custom market data columns

Fixed#

  • CLI update should load the templates from the templates dir instead of the Python data dir when installed in editable mode

  • templates dir not found when installed in editable mode on Windows

0.29 (2025-01-25)#

Added#

  • kaxanuk.data_curator.modules.extension_handler aliased to kaxanuk.data_curator.extension_handler for loading external extension modules

Changed#

Removed#

  • YahooFinance from this main package

0.28 (2024-12-19)#

Added#

  • Sphinx documentation generator with Readthedocs support

  • Creation of a .py extension for parsing the calculations file and classifying its content using the ..category:: directive found in the docstring

  • Implementation of a structure to support .md files located outside the docs folder

  • Development of a Sphinx custom template for the documentation.

  • CLI --version command

  • kaxanuk.datacurator.__package_name__

  • kaxanuk.datacurator.__package_title__

Changed#

  • __version__ and __parameters_format_version__ moved from __version__.py to __init__.py

  • Migration of the documentation structure to the docs folder, replacing the deprecated Read_the_Docs folder.

  • Parquet and CSV output handlers no longer take data_features_subdir parameter, everything is saved to the Output folder

Fixed#

  • Regression: YahooFinance not correctly loaded by Excel configuration entry script

  • Editable install CLI init and update now work in any directory

  • CLI catch OSError when there’s a file permissions issue

Removed#

  • __version__.py file

  • kaxanuk.datacurator.version

  • Data_and_Features subfolder

0.27 (2024-11-17)#

Added#

  • CLI update format ‘entry_script’ for updating just the entry script

  • validate_api_key abstract (required) method to DataProviderInterface

Changed#

  • FinancialDataProviderInterface is now DataProviderInterface once more

  • ExcelConfigurator.__init__ API changed, data_providers now receive a typed dict with class and api_key params

  • Data provider API keys now are per provider

  • Excel entry script now gets api keys from environment (loading from Config/.env if available)

  • FMP changed _endpoints MappingProxyType for StrEnum

  • FMP.validate_api_key makes a request to get AAPL company information

  • Updated parameters_datacurator file version

0.26 (2024-11-13)#

Added#

  • Datacolumn.__neg__

  • Left a not implemented placeholder for Datacolumn.__pos__ for completeness

Changed#

  • DataColumn.all_equal() renamed to DataColumn.fully_equal()

  • Updated usage instructions in README.md

Fixed#

  • DataColumn reflected operators for +, -, *, /, //, %

0.25 (2024-11-08)#

Added#

  • Tests to replicate reflected arithmetic operation errors.

  • Missing typing on data_column methods.

  • CLI interface through services.cli

  • click library for the CLI implementation

  • .env file to templates (unused at the moment)

  • __main__.py file to templates

Changed#

  • DataColumn arithmetic operation tests now grouped into classes.

  • Docker images now install the library and use the CLI interface

  • templates dir files moved into data_curator subdir so that installed data files are also installed in a subdir of the data dir

  • Config templates are now in the templates/data_curator/Config subdir

  • The PyCharm debugger now only depends on the KNDC_DEBUG_PORT env variable for activation

Fixed#

  • GitHub workflow broke on pull request merges

Removed#

  • Config dir (gets created by the CLI interface now)

  • Output dir (gets created by the CLI interface now)

  • __main__.py (gets created by the CLI interface now)

0.24 (2024-10-12)#

Changed#

  • FMP: use unadjustedVolume for MarketDataDailyRow.volume

  • Standardized .gitignore based on the official GitHub one

Fixed#

  • Pin Dockerfile base image to python:3.12-slim, as Python 3.13 is now a thing and Pyarrow breaks on it

0.23 (2024-09-21)#

Added#

  • The GitHub Actions pipeline now builds and pushes the container image to GitHub Packages when pushing the dev branch or a main branch tag.

0.22 (2024-09-21)#

Changed#

  • DataColumn.__getitem__ - simplified the logic

  • First steps to build the full GitHub Packages publishing pipeline

Fixed#

  • FinancialDataProviderInterface._request_data - “AttributeError: ‘URLError’ object has no attribute ‘read’” when _load_ssl_context throws “ConnectionResetError: [WinError 10054] An existing connection was forcibly closed by the remote host”

  • FinancialDataProviderInterface._request_data - connection errors were still throwing up error trace after all connection attempts

0.21 (2024-08-04)#

Added#

  • tests: ColumnBuilder._process_columns_with_available_dependencies now fully tested

Changed#

  • prepended KNDC_ to all env vars

  • increased parameters_datacurator.xlsx version to 0.21

  • ExcelConfigurator now receives the names of the api key env vars, and loads them itself, as otherwise the warnings can’t really know if the keys were loaded from env vars or not

  • added optional logger_file parameter to data_curator.main, to enable logging to a file

  • Fundamental data entity error logs now include the date, where available

Fixed#

  • logger_level parameter in data_curator.main was being ignored if ExcelConfigurator had been used beforehand, as it was not closing its own logger

  • ExcelConfigurator now changes its own level to the one in the Excel file, as soon as it can

0.20 (2024-07-28)#

Changed#

  • parameters.xlsx file renamed to parameters_datacurator.xlsx as a workaround for Excel’s mind-numbingly idiotic inability to have 2 open files with the same name as of 2024

  • MarketDataDailyRow: validate low <= high

  • MarketDataDailyRow: validate volume and vwap are not negative

  • FundamentalDataRowIncome: validate weighted_average_shares_outstanding and weighted_average_shares_outstanding_diluted are not negative

  • ColumnBuilder._process_columns_with_available_dependencies is now class method, for easier unit testing

  • pytest: added data_column_debugger.dump_data_columns_to_csv() fixture to conftest.py for dumping DataColumns to a csv file for debugging tests

  • more unit tests

0.19 (2024-06-30)#

Added#

  • FinancialDataProviderInterface._build_url_with_ticker_path_and_query_params method for building standard URLs

  • calculations.annualized_volatility_5d() test with csv fixture

Changed#

  • features.helpers.annualized_volatility and features.helpers.indexed_rolling_window_operation now require kwargs as they are meant as user-facing functions

  • improved column_builder typing and docstrings

  • FinancialDataProviderInterface._request_data now accepts a url_builder callable argument for building the URLs, _build_url_with_ticker_path_and_query_params by default

  • made MarketDataDailyRow.adjusted_close nullable

  • refactored DataColumn.equal() to use both absolute and relative thresholds when comparing with approximate_floats=true

  • better documentation and linting

## Fixed * Fundamental Data Provider ‘none’ failed regression * DataColumn.equal failed when presented 2 NullArray arguments

0.18 (2024-05-26)#

Added#

  • logger_format param to ExcelConfigurator.__init__

  • Ticker value object for entities

  • services.validator module, with a validator for date pattern strings

  • date pattern tests for dividend and split rows

  • package descriptions for each __init__.py

Changed#

  • Started removing redundant docstring types from typehinted functions

  • providers package is now data-providers

  • internals package is now services

  • entities now use Ticker instead of str for field ticker

  • Moved all interface modules to their respective implementation packages

  • test coverage report now skips fully covered files

Fixed#

  • ExcelConfigurator missing checks

  • Handle FMP throwing error 404 when symbol data not found

  • entity_helper.detect_field_type_errors bug that prevented validation of all fields of an entity

  • entity_helper and validator services now fully tested, full library test coverage @ 58%

0.17 (2024-05-12)#

Added#

  • pdm.lock to .gitignore (don’t forget to remove on main branch!)

  • FinancialDataProviderInterface.init_config() method for running code before looping through each ticker

  • DataProviderMissingKeyError

  • YahooFinance missing interface implementation methods

  • __all__ to all __init__.py files, to be able to remove the F401 linter ignore

  • ConfiguratorInterface

Changed#

  • Data providers now use object instances instead of classes

  • Inject api keys into data providers at object initialization, removed them from individual method params

  • data_curator.main() now receives individual data provider objects instead of dict with possible choices

  • data_curator.main() now receives output_handlers list, runs all

  • data_curator.main() now receives an int logger_level

  • Complete refactor of ExcelConfigurator, is now an implementation of the new ConfiguratorInterface and has methods for outputting each dependency type

  • PassedArgumentError Exception

Fixed#

  • SplitData error with Fundamental data provider set as None

  • data_curator.py linter G004 error

Removed#

  • Configuration entity fields related to providers, handlers and loggers

0.16 (2024-05-02)#

Added#

  • yahoo_finance market data - yahoo!

Changed#

  • increased parameters file version

0.15 (2024-04-28)#

Added#

  • split data

  • initial entity_helper tests

Changed#

  • MarketDataDailyRow.vwap now nullable

  • use entity_helper.detect_field_type_errors() for all downloaded data entity validation

  • entity_helper.fill_fields() now accepts null field correspondences

  • updated parameters.xlsx template to add new fields, column width adjustments

0.14 (2024-04-23)#

Added#

  • modules/data_column 100% effective test coverage!

  • dividend data - had to rework some internal machinery

  • entity_helper.detect_field_type_errors() for validating entity field types

Changed#

  • pdm run test now includes coverage term-missing report

  • attribute_filler.py renamed to entity_helper.py

0.13 (2024-04-06)#

Added#

  • DataColumn.__mod__() actual implementation

  • DataColumn private members tests, class coverage at 90%

Changed#

  • Replaced generic exceptions by custom ones

Fixed#

  • Decimal precision out of range errors on DataColumn.__add__(), .__subtract__()

  • Decimal precision out of range error on DataColumn._mask_zeroes()

  • Multiple DataColumn private methods minor bugs

0.12 (2024-03-24)#

Changed#

  • CHANGELOG.md updated Keep a Changelog link to version 1.1.0

  • Fundamental data: missing cash flow or balance sheet now returns just the income statement data if available

  • Improved Api server error retries handler

  • Api server errors (after retries) now fully stop execution

  • Publicly exposed entity classes from kaxanuk.data_curator.entities namespace

  • Improved error handler for circular dependencies

  • Improved error handler for missing custom calculation functions

  • Improved error handler for Excel parameters file configuration errors

  • Removed redundant src/kaxanuk/py.typed file

Fixed#

  • “Decimal precision out of range” error when dividing decimal columns with too many decimals

0.11 (2024-03-17)#

Added#

  • pdm run install_dev

  • parquet output handler

  • py.typed files to declare the whole library is typehinted

Changed#

  • now loading api_keys from env vars, only using templates/parameters.xlsx as fallback

  • increased templates/parameters.xlsx version

  • fixed templates/parameters.xlsx start date validation

  • added all default output columns to templates/parameters.xlsx

  • improved internal documentation in templates/parameters.xlsx

  • Docker: don’t uninstall pdm on dev environment

  • renamed pdm run test_with_coverage to pdm run test

Fixed#

  • support index tickers with ^ character

  • “Decimal precision out of range” error when multiplying decimal columns with too many decimals

  • ExcelConfigurator typecasting None values to ‘None’ string

  • mypy configuration and some typehint errors

0.10 (2024-03-10)#

Added#

  • coverage tests under pytest-cov

Changed#

  • modules/security_calculations.py is now features/calculations.py

  • moved security_calculations helper functions into features/helpers.py

  • moved modules/attribute_filler.py to internals/attribute_filler.py

  • moved modules/column_builder.py to internals/column_builder.py

  • exposed FinancialDataProviderInterface as kaxanuk.data_curator.interfaces.FinancialDataProviderInterface

  • exposed OutputHandlerInterface as kaxanuk.data_curator.interfaces.OutputHandlerInterface

  • refactored DataColumn tests

Fixed#

  • Circular references error when custom calculation function parameter columns are missing from the output column list

  • __main__.py now only injects src to sys.path if not loading as installed library

0.9 (2024-03-03)#

Added#

  • An actual public API for the library by means of __all__ and imports in __init__.py files

  • DataColumn methods, fully unit tested:: all_equal, concatenate, equal

  • DataColumn property: type

  • Basic url request retry functionality

Changed#

  • CsvOutputter is now CsvOutput

  • __main__.py now uses the public library API in its imports

Fixed#

  • security_calculations._indexed_rolling_window_operation was broken, so last_twelve_month… functions returned wrong data

  • Improved internal documentation

  • Improved/simplified some type hints

  • Regression: Empty market data no longer terminates the whole process

Removed#

  • Numpy no longer a dependency

0.8 (2024-02-26)#

Added#

  • pytest pyarrow_helper fixture, for helper functions for testing PyArrow arrays

Changed#

  • No more Numpy! DataColumn, and thus security_calculations, now work on top of PyArrow!

  • DataColumn public API changed, but operator overloading works the same

  • security_calculations refactor as PyArrow simplifies many operations, allows easier use of pandas

  • security_calculations output columns are automatically wrapped in DataColumn

0.7 (2024-02-11)#

Added#

  • Add main() optional parameter logger_format, for configuring the logger

Changed#

  • Make entity attribute typing/casting errors more explicit

  • Remove revenue>=0 checks

  • Simplify entity attribute type validations

  • Change the default logger format for more readable logs

  • Allow floats as entity attribute type

  • Moved all general helper methods from FMP to FinancialDataProviderInterface, so any provider can use them

Fixed#

  • Require Configurator.start_date and end_date to be explicitly datetime.date

0.6 (2024-02-05)#

Fixed#

  • Add blank FundamentalData rows for ommited data (in case of ammendments, missing fundamentals, &c.)

0.5 (2024-02-04)#

Added#

  • Custom exceptions

  • New library dependency for semver version comparisons: packaging

Changed#

  • New parameters file template

  • Now versioning the parameters file formats, and checking them in ExcelConfigurator

  • Fundamental data provider can be now set independently of market data one, or even as disabled

  • Separate input and config handlers into their own folders

  • Rename the “quarter” period to “quarterly” in config

Fixed#

  • Missing fundamental data for a ticker will only omit that data, but keep the market data

0.4 (2024-02-03)#

Changed#

  • Replace numpy.array with custom DataColumn to remove “where” kwarg boilerplate code.

  • Inject custom calculations from entry script

  • Move templates outside src

0.3 (2024-01-07)#

  • Restructure src to implement under organization/project_name namespace.

On this page
Show Source