Release v1.0.1

• 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()

Release v1.0.0

• 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

Release v0.16.1

• 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

Release v0.16.0

• 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

Release v0.15.4

• 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

Release v0.15.3

• 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

Release v0.15.2

• 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

Release v0.15.1

• 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

Release v0.15.0

• 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

Release v0.14.3

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