Changelog

0.6.0 (2025-11-13)

Changed

• Upgrade to Python 3.12 to match Google Colab.

0.5.0 (2025-08-14)

Removed

• Remove support for Jupyter Notebook 6 or earlier.

0.4.0 (2025-05-13)

Changed

• Add spreadsheet_name as the first argument to save_dataframe_to_sheet().

Removed

• Remove ocdskingfishercolab.set_spreadsheet_name.

0.3.14 (2024-11-25)

Fixed

• Constrain the version of prettytable to be compatible with ipython-sql.

0.3.13 (2023-10-27)

Fixed

• Restore support for Jupyter Notebook 6.

0.3.12 (2023-10-23)

Changed

• Upgrade to Python 3.10 and other dependencies to match Google Colab.

0.3.11 (2023-07-04)

Added

• set_dark_mode(), if Tools > Settings > Site > Theme is set to dark in Google Colab.

• set_light_mode(), for the default Seaborn theme.

• format_thousands().

0.3.10 (2023-02-03)

Fixed

• Pin to SQLAlchemy 1.x.

0.3.9 (2022-06-30)

Changed

• save_dataframe_to_sheet() and save_dataframe_to_spreadsheet() do nothing if the data frame is empty. 9b83348

• calculate_coverage(): Rename keyword arguments from sql to print_sql and sql_only to return_sql. d706145

• calculate_coverage(): Simplify the query if "ALL " is prefixed to a field that is not within an array. 869a9d0

• calculate_coverage(): Raise an error if no fields are provided. 8896336

Fixed

• calculate_coverage(): Use the correct parent table if scope is not set. Previously, Kingfisher Colab would not use: a7c0936

  • award_documents

  • award_items

  • award_suppliers

  • contract_documents

  • contract_items

  • contract_milestones

  • contract_implementation_documents

  • contract_implementation_milestones

  • contract_implementation_transactions

• calculate_coverage(): Construct correct conditions and warnings if a field is within nested arrays. 3dced1a

• calculate_coverage(): Use the relatedprocesses_summary table for fields starting with relatedProcesses/, where appropriate. 9e6cdb7

• calculate_coverage(): Prefix all_ to the column if "ALL " is prefixed to the field, to avoid duplicate columns. e9427b2

• calculate_coverage(): No longer warn about address fields. e2b8d72

0.3.8 (2022-04-27)

Changed

• Removed dependency on libcoveocds (GPL).

0.3.7 (2022-03-11)

Added

• Add calculate_coverage() to calculate the co-occurence coverage of a group of fields.

0.3.6 (2021-09-15)

Changed

• list_collections(): source_id is now an optional argument. If omitted, all collections are returned.

0.3.5 (2021-08-09)

Added

• Add render_json() to render JSON into collapsible HTML.

0.3.4 (2021-04-16)

Fixed

• set_search_path() no longer outputs an error message.

0.3.3 (2021-04-10)

Added

• Add Python wheels distribution.

0.3.2 (2021-02-16)

Fixed

• Set the minimum version of ipython-sql.

0.3.1 (2021-02-16)

Fixed

• Fixed dependencies to install cleanly both locally and on Google Colab.

0.3.0 (2020-12-15)

Changed

• Refactor to build upon ipython-sql.

  You must run %load_ext sql first, and %config SqlMagic.autopandas = True to get a pandas DataFrame.

  • Remove functions that can be replaced by ipython-sql magics in the notebook.

    • create_connection, reset_connection: Use an ipython-sql connection string, e.g. %sql postgresql://user:pass@host/db

    • execute_statement, get_dataframe_from_cursor, get_dataframe_from_query: Use ipython-sql’s %sql magic, e.g. %sql SELECT a FROM b

  • Replace SQL calls with ipython-sql calls in remaining functions.

    • get_list_from_query: Use get_ipython_sql_resultset_from_query(). This returns an ipython-sql ResultSet, the type returned by the %sql% magic when autopandas is off. It behaves like a list, but with extra methods.

    • ocdskingfishercolab.download_package_from_query() no longer takes a params argument, and instead uses variables from the local scope, to be consistent with the ipython-sql’s %sql magic.

  There’s a shared (but not public) colab notebook of examples run against Kingfisher.

• download_package_from_ocid() orders packaged releases in reverse date order.

Fixed

• write_data_as_json() replaces path separators with underscores in filenames.

0.2.2 (2020-04-22)

Added

• set_search_path()

• get_list_from_query

Fixed

• execute_statement` no longer has a mutable default argument value.

0.2.1 (2020-04-21)

Fixed

• execute_statement no longer errors if given a psycopg2.sql.Composable.

0.2.0 (2020-04-21)

Upgrade instructions:

• Upgrade to 0.1.x if you have not already, and address any deprecation warnings. Then, upgrade to 0.2.x.

• Install alembic and SQLAlchemy separately, if used in the notebook.

Added

• list_source_ids()

• list_collections()

• execute_statement

• save_dataframe_to_spreadsheet()

• download_data_as_json()

• write_data_as_json()

• Add a comment to all SQL queries with a link to the notebook, for database administrators.

Changed

• Backwards-incompatible: The methods deprecated in 0.1.x are removed.

• Backwards-incompatible: alembic and SQLAlchemy are no longer installed.

Fixed

• download_package_from_ocid() returns structurally correct records. Previously, the ocid field was at the package-level instead of the record-level.

• download_package_from_ocid() and download_package_from_query() return structurally correct packages. Previously, required fields were omitted.

0.1.1 (2020-04-20)

Fixed

• The deprecated methods output_notebook and get_results return values again.

0.1.0 (2020-04-20)

Upgrade instructions:

• Use pip install 'ocdskingfishercolab<0.2' instead of any previous pip install command.

• Import ocdskingfishercolab instead of kingfishercolab.

Added

• download_package_from_query()

Changed

• Backwards-incompatible: Renamed package from kingfishercolab to ocdskingfishercolab.

• Renamed methods for consistent style. Old-style methods are deprecated:

  • saveToSheets is now save_dataframe_to_sheet()

  • saveStraightToSheets is now save_dataframe_to_sheet() with prompt=False

  • saveToCSV is now download_dataframe_as_csv()

  • downloadReleases is now download_package_from_ocid()

  • output_notebook is now get_dataframe_from_query

  • getResults is now get_dataframe_from_cursor

• get_dataframe_from_query raises an error instead of returning an error.

• download_package_from_ocid() raises an error instead of printing a message.

0.0.1 (2020-04-20)

Initial release.