Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

Version 1.3.2 (2025-05-27)

• Added: support for Python 3.13

• Added: support for Python 3.12

• Fixed: audformat.Database.update() for misc tables

• Fixed: audformat.Database.get() if scheme is stored in a segmented table, and additional_schemes are stored in filewise tables. Before, values of the additional schemes were set to <NA>

• Fixed: if audformat.Column.set() assigns values to a column with scheme labels given by a misc table, the misc table is only converted once to a dictionary instead of once for each value

Version 1.3.1 (2024-09-16)

• Changed: replace unmaintained iso-639 dependency with iso639-lang

• Fixed: ensure poetry can manage audformat

Version 1.3.0 (2024-07-18)

• Added: strict argument to audformat.utils.hash(). If set to True, the order of the data, and its level/column names are taken into account when calculating the hash

• Changed: store tables per default as parquet files, by changing the default value of storage_format to "parquet" in audformat.Table.save() and audformat.Database.save()

• Fixed: load csv tables with pandas.read_csv(), if pyarrow.csv.read_csv() fails

Version 1.2.0 (2024-06-25)

• Added: expand format specifications to allow parquet files as table files

• Added: support for storing tables as parquet files by adding "parquet" (audformat.define.TableStorageFormat.PARQUET) as an option for the storage_format argument of audformat.Table.save() and audformat.Database.save()

• Added: support for numpy>=2.0

• Added: mention text files as potential media files in the documentation

• Added: mention in the documentation of audformat.utils.hash() that column/level names do not influence its hash value

• Added: warn in the documentation of audformat.utils.hash() that the hash of a dataframe or series, containing "Int64" as data type, changes with pandas>=2.2.0

• Fixed: ensure "boolean" data type is always used in indices of misc tables that store boolean values

Version 1.1.4 (2024-05-15)

• Fixed: audformat.Database.get(), if its argument additional_schemes contains a non-existent scheme

Version 1.1.3 (2024-04-26)

• Added: as_dataframe argument to audformat.utils.read_csv()

• Fixed: audformat.utils.read_csv() now treats float/integer values in start, end columns as seconds

Version 1.1.2 (2024-02-02)

• Fixed: audformat.Database.load() when loading databases with a misc table that has an assigned split

Version 1.1.1 (2024-01-25)

• Changed: depend on audeer>=2.0.0

• Fixed: pandas deprecation warnings

Version 1.1.0 (2023-11-30)

• Added: audformat.Database.get() method to retrieve labels based on their schemes and independent of the tables in which they are stored

• Added: aggregate_function and aggregate_strategy arguments to audformat.utils.concat() to support overlapping values in the objects that should be concatenated

• Changed: audformat.Column.get(map=...) now returns dtype of labels

• Changed: audformat.Column.get(map=...) does no longer raise an error if some of the mapped values are not available when stored in a dictionary as scheme labels

• Fixed: avoid deprecation warning by replacing pkg_resources internally with importlib.metadata

Version 1.0.3 (2023-10-11)

• Fixed: audformat.utils.hash() for pandas>=2.1.0

• Fixed: remove upper limit of pandas dependency

Version 1.0.2 (2023-10-09)

• Fixed: require pandas<2.1.0 as pandas>=2.1.0 introduced a bug in calculating the hash of an index

• Removed: deprecated root argument from audformat.testing.create_audio_files()

Version 1.0.1 (2023-05-08)

• Fixed: ensure audformat.utils.to_segmented_index() and audformat.Table.get() with as_segmented=True uses same precision for end values as audformat.segmented_index()

Version 1.0.0 (2023-04-27)

• Added: audformat.Scheme.labels_as_list property to list all scheme labels

• Added: example to the documentation of audformat.utils.to_filewise_index()

• Changed: convert dates to UTC timezone in audformat.Column.set() when using a scheme of type "date"

• Fixed: support pandas>=2.0.0

• Fixed: mention author, license, license_url, organization in the specification documentation of the database header

• Fixed: missing Raises section in the documentation of audformat.Database.load() and audformat.Database.attachments

• Fixed: when the root argument of audformat.utils.expand_file_path() is a relative path it is no longer expanded to an absolute path

Version 0.16.1 (2023-03-29)

• Added: copy_attachments argument to audformat.Database.update()

• Changed: preserve dtypes when audformat.Table.get() is called with an index

• Changed: speed up audformat.utils.union()

• Changed: allow to save a database with missing attachments

Version 0.16.0 (2023-01-12)

• Added: audformat.Attachment to store any kind of files/folders as part of the database

• Added: support for Python 3.10

• Added: support for Python 3.11

• Changed: require audeer>=1.19.0

• Changed: split API documentation into sub-pages for each function

• Fixed: support "meta" as key in meta dictionaries like the one passed as meta argument to audformat.Database

Version 0.15.4 (2022-11-01)

• Fixed: avoid FutureWarning when setting values in place for a series in audformat.Column.set()

• Fixed: improve sketches in the specifications section of the documentation

Version 0.15.3 (2022-09-19)

• Changed: audformat.Column.set() now lists values not matching the scheme of the column in the corresponding error message

• Fixed: audformat.Column.set() checking of values for a scheme with minimum and/or maximum when input values are given as np.array and contain NaN or None

• Fixed: audformat.Column.set() checking of values for a scheme with minimum and/or maximum when minimum or maximum is 0

Version 0.15.2 (2022-08-17)

• Added: audformat.Table.map_files()

• Fixed: audformat.Database.load() for databases that contain a scheme with labels stored in a misc table that is using schemes for its columns. Before it could fail if the schemes were not loaded in the correct order

• Fixed: audformat.Table.drop_index() and audformat.MiscTable.drop_index() when the provided index to drop contains entries not present in the index of the table. Before it was extending the table by those entries besides dropping overlapping indices

Version 0.15.1 (2022-08-11)

• Added: audformat.Scheme.uses_table to indicate if the scheme uses a misc table to store its labels

• Added: usage example to docstring of audfromat.utils.to_segmented_index()

• Changed: forbid nesting of misc tables as scheme labels

• Fixed: support for pd.Index and pd.Series in audformat.utils.to_filewise_index()

• Fixed: description of audformat.Schemes.labels in API documentation

Version 0.15.0 (2022-08-05)

• Added: audformat.MiscTable which can store data not associated with media files

• Added: store scheme labels in a misc table

• Added: dictionary audformat.Database.misc_tables holding misc tables of a database

• Added: audformat.utils.difference() for finding index entries that are only part of a single index for a given sequence of indices

• Added: audformat.utils.is_index_alike() for checking if a sequence of indices has the same number of levels, level names, and matching dtypes

• Added: audformat.define.DataType.OBJECT

• Added: audformat.utils.set_index_dtypes() to change dtypes of an index

• Added: audformat.testing.add_misc_table()

• Added: audformat.Database.__iter__ iterates through all (misc) tables, e.g. a user can do list(db) to get a list of all (misc) tables

• Changed: audformat.Database.update() can now join schemes with different labels

• Changed: audformat.utils.union(), audformat.utils.intersect(), and audformat.utils.concat() now support any kind of index

• Changed: audformat.utils.intersect() no longer removes segments from a segmented index that are contained in a filewise index

• Changed: require pandas>=1.4.1

• Changed: use pandas dtype "string" instead of "object" for storing audformat dtype "str" entries

• Changed: use a misc table to store the "speaker" scheme labels in the emodb example in the documentation

• Changed: audformat.utils.join_labels() raises ValueError if labels are of different dtype

• Fixed: ensure column IDs are different from index level names

• Fixed: make sure audformat.Column.set() converts data to dtype of scheme before checking if values are in min-max-range of scheme

• Fixed: links to pandas API in the documentation

• Fixed: include methods to_dict(), from_dict(), dump(), and attributes description, meta in the documentation for the classes audformat.Column, audformat.Database, audformat.Media, audformat.Rater, audformat.Scheme, audformat.Split, audformat.Table

• Fixed: type hint of argument dtype in the documentation of audformat.Scheme

• Removed: support for Python 3.7

Version 0.14.3 (2022-06-01)

• Added: audformat.utils.map_country()

• Changed: improve speed of audformat.Table.drop_files() for segmented tables

Version 0.14.2 (2022-04-29)

• Added: audformat.utils.index_has_overlap()

• Added: audformat.utils.iter_index_by_file()

• Changed: store categories with integers as int64 instead of Int64

• Changed: require audeer>=1.18.0

• Changed: support pandas>=1.4.0

Version 0.14.1 (2022-03-03)

• Added: audformat.utils.map_file_path()

Version 0.14.0 (2022-02-24)

• Changed: ensure audformat.testing.create_database() uses Unix path separators

• Changed: don’t allow \ path entries in a portable database

• Changed: mark deprecated root argument of audformat.testing.create_audio_files() to be removed in version 1.0.0

Version 0.13.3 (2022-02-07)

• Fixed: conversion of pickle protocol 5 files to pickle protocol 4 in cache

Version 0.13.2 (2022-01-27)

• Fixed: reintroduce sorting the output of audformat.Database.files and audformat.Database.segments

Version 0.13.1 (2022-01-26)

• Fixed: changelog for 0.13.0

Version 0.13.0 (2022-01-26)

• Changed: audformat.utils.union() no longer sorts levels

• Changed: audformat.Table.save() forces pickle format 4

• Changed: clean up test requirements

• Changed: require pandas < 1.4.0

Version 0.12.4 (2022-01-12)

• Changed: the API documentation on the language argument of audformat.Database is more verbose now

• Changed: the difference between audformat.define.DataType.TIME and audformat.define.DataType.DATE is now discussed in the API documentation

• Fixed: saving a not loaded table to CSV when a PKL file is present

• Fixed: pandas deprecation warnings

Version 0.12.3 (2022-01-03)

• Removed: Python 3.6 support

Version 0.12.2 (2021-11-18)

• Added: audformat.assert_no_duplicates()

• Changed: audformat.assert_index() no longer checks for duplicates

Version 0.12.1 (2021-11-17)

• Added: audformat.utils.hash()

• Added: audformat.utils.expand_file_path()

• Added: audformat.utils.replace_file_extension()

• Changed: use yaml.CLoader for faster header reading

Version 0.12.0 (2021-11-10)

• Added: as_segmented, allow_nat, root, num_workers arguments to audformat.Table.get()

• Added: as_segmented, allow_nat, root, num_workers arguments to audformat.Column.get()

• Added: files_duration argument to audformat.utils.to_segmented_index()

• Added: audformat.Database.files_duration()

• Changed: changed default value of load_data argument in audformat.Database.load() to False

• Changed: speed up audformat.Database.files and audformat.Database.segments

• Fixed: re-add support for pandas>=1.3

Version 0.11.6 (2021-08-20)

• Added: support for Python 3.9

• Fixed: speed up audformat.utils.union()

• Fixed: audformat.Column.set() with pd.Series and np.array for a scheme with fixed labels and containing NaN values

Version 0.11.5 (2021-08-09)

• Removed: duration scheme and column from conventions and emodb example

Version 0.11.4 (2021-08-05)

• Added: custom BadKeyError when key is not found

• Changed: limit to pandas <1.3 until it works again for newer pandas versions

• Changed: remove the <1.0.0 limit for audiofile as a stable release is available and the API has not changed

Version 0.11.3 (2021-06-10)

• Added: audformat.utils.duration

• Fixed: description of audformat.Database.is_portable in documentation

Version 0.11.2 (2021-05-12)

• Added: audformat.utils.join_schemes

Version 0.11.1 (2021-05-11)

• Added: Database.is_portable

• Added: copy_media argument to Database.update()

• Changed: remove root argument from testing.create_audio_files() and instead use Database.root

• Fixed: utils.concat() converts to nullable dtype

• Fixed: utils.concat() returns DataFrame if input contains at least one DataFrame

Version 0.11.0 (2021-05-06)

Note: tables stored from this version upwards cannot be loaded with older versions

• Added: Database.root

• Added: utils.join_labels()

• Added: Scheme.replace_labels()

• Changed: set dependency to pandas>=1.1.5

• Changed: do not compress pickled table files

Version 0.10.2 (2021-04-22)

• Changed: allow_nat argument to utils.to_segmented_index()

Version 0.10.1 (2021-03-31)

• Fixed: audformat.assert_index() checks for correct dtypes

Version 0.10.0 (2021-03-18)

• Added: audformat.Database.update()

• Added: audformat.Table.update()

• Added: overwrite argument to audformat.utils.concat()

• Changed: result of audformat.Table.__add__() is no longer assigned to a audformat.Database

Version 0.9.8 (2021-02-23)

• Added: audformat.Database.license

• Added: audformat.Database.license_url

• Added: audformat.Database.author

• Added: audformat.Database.organization

• Added: audformat.utils.intersect() for index objects

• Added: audformat.utils.union() for index objects

• Changed: Database.load() raises error if table file missing

• Changed: forbid duplicates in audformat conform indices

• Fixed: audformat.Table.__add__() returned wrong values for some index combinations

Version 0.9.7 (2021-02-01)

• Added: update_other_formats argument to audformat.Table.save() to make sure existing files in other formats are updated as well

• Changed: use round_trip argument when loading CSV files to ensure dataframes are equal after storing and loading again

Version 0.9.6 (2021-01-28)

• Fixed: implemented audformat.Database.__eq__ and return True for identical databases

Version 0.9.5 (2021-01-14)

• Changed: use nullable Pandas’ type "boolean" for bool schemes

• Fixed: Scheme.draw() generates boolean values if scheme is bool

Version 0.9.4 (2021-01-11)

• Changed: add arguments num_workers and verbose to audformat.Database.load()

Version 0.9.3 (2021-01-07)

• Fixed: avoid sphinx syntax in CHANGELOG

Version 0.9.2 (2021-01-07)

• Changed: add arguments num_workers and verbose to audformat.Database.drop_files(), audformat.Database.map_files(), audformat.Database.pick_files(), audformat.Database.save()

• Changed: audformat.segmented_index() support int and float, which will be interpreted as seconds

• Fixed: audformat.utils.to_segmented_index() returns correct index type for NaT

Version 0.9.1 (2020-12-21)

• Fixed: add column name to HTML Series output in docs

• Fixed: removed mentioning of NotConformToUnifiedFormat error and RedundantArgumentError error

• Fixed: add missing errors to docstring of audformat.Table.set() and audformat.Column.set()

Version 0.9.0 (2020-12-18)

• Added: initial release public release