CHANGELOG.md

TFS-Pandas Changelog

Version 3.8.2

• Changed:

  • The headers of a TfsDataFrame are now stored as a dict and no longer an OrderedDict. This is transparent to the user.

• Fixed:

  • Removed a workaround function which is no longer necessary due to the higher minimum pandas version.

Version 3.8.1

• Changed:

  • Migrated to standard pyproject.toml.
  • The minimum required numpy version is now numpy 1.24.

• Fixed:

  • The package is now compatible with numpy 2.x.
  • The package's HDF functionality is fully compatible with numpy 2.x on Python >= 3.10 thanks to a pytables compatibility release.
  • The package's HDF functionality limits to numpy < 2 on Python 3.9 due to the lack of compatibility from pytables on this versions.

Version 3.8.0

• Changed:

  • The minimum required pandas version is now pandas 2.1.
  • Support for Python 3.8 has been dropped. The minimum required Python version is now 3.9.

• Fixed:

  • Solved a DeprecationWarning appearing when writing a TfsDataFrame to disk due to the use of .applymap, byusing the now recommended .map method.
  • Solved a DeprecationWarning appearing when reading a file from disk due to the use of delim_whitespace in our reader, by using the now recommended sep option.
  • Solved a FutureWarning appearing when validating a TfsDataFrame due to the use of the pd.option_context('mode.use_inf_as_na', True) context manager during validation by explicitely casting infinite values to NaNs.
  • Solved a FutureWarning appearing when validating a TfsDataFrame due to object downcasting happening during validation by explicitely infering dtypes first.

Version 3.7.3

• Fixed:
  • Fixed a regression where the writing of a pd.Series-like object to disk was raising an error. It is now possible again.

Version 3.7.2

• Fixed:
  • fixing the issues with pandas >= v2.1.0 (see tfs-pandas v3.7.1) by overwriting the _constructor_from_mgr function.

Version 3.7.1

• Changed:
  • The dependency on pandas was restricted to avoid the latest version, 2.1.0 and above as a temporary workaround to an attribute access bug that arose with it.

Version 3.7.0

Minor API changes to the TFSCollections:

• the old write_to and get_filename are renamed to _write_to and _get_filename as they could only be accessed internally (due to the input parameters not available to the user). This also means, that - in case they are overwritten by a user's implementation - they need to be renamed there!!

• The column which is set as index can now also be defined manually, by overwriting the attribute INDEX, which defaults to "NAME".

• New Functions of TFSCollection Instances:

  • get_filename(name): Returns the associated filename to the property with name name.
  • get_path(name): Return the actual file path of the property name
  • flush(): Write the current state of the TFSDataFrames into their respective files.
  • write_tfs(filename, data_frame): Write the data_frame to self.directory with the given filename.

• New Special Properties of TFSCollection Instances:

  • defined_properties: Tuple of strings of the defined properties on this instance.
  • filenames is a convenience wrapper for get_filename():
    • When called (filenames(exist: bool)) returns a dictionary of the defined properties and their associated filenames. The exist boolean filters between existing files or filenames for all properties.
    • Can also be used either filenames.name or filenames[name] to call get_filename(name) on the instance.

• Moved the define-properties functions directly into the Tfs-attribute marker class.

• Return of None for the MaybeCall class in case of attribute not found (instead of empty function, which didn't make sense).

Version 3.6.0

• Removed:

  • The append and join methods of TfsDataFrame have been removed.

• Changed:

  • The dependency version on pandas has been restored to >=1.0.0 as the above removal restores compatibility with pandas 2.0.

Version 3.5.3

• Changed:
  • Fixed a wrong deprecation of the .merge method of TfsDataFrames.

Version 3.5.2

• Changed:
  • The dependency on pandas has been pinned to <2.0 to guarantee the proper functionning of the compability append and join methods in TfsDataFrames. These will be removed with the next release of tfs-pandas and users should use the tfs.frame.concat compatibility function instead.

Version 3.5.1

• Fixed:
  • Allow reading of empty lines in headers again.

Version 3.5.0

• Fixed:

  • Any empty strings ("") in a file's columns will now properly be read as such and not converted to NaN.

• Added:

  • It is now possible to only read the headers of a file by using a new function, read_headers. The function API is not exported at the top level of the package but is available to import from tfs.reader.

Version 3.4.0

• Added:
  • The read_tfs and write_tfs functions can now handle reading / writing compressed files, see documentation for details.

Version 3.3.1

• Changed:
  • Column types are now assigned at read time instead of later on, which should improve performance for large data frames.

Version 3.3.0

• Added:

  • The option is now given to the user to skip data frame validation after reading from file / before writing to file. Validation is left "on" by default, but can be turned off with a boolean argument.

• Changes:

  • The tfs.frame.validate function has seen its internal logic reworked to be more efficient and users performing validation on large data frames should notice a significant performance improvement.
  • The documentation has been expanded and improved, with notably the addition of example code snippets.

Version 3.2.1

• Changed:
  • Allow spaces in header names.

Version 3.2.0

• Added:

  • HDF5 read/write.

• Changed:

  • The minimum required Python version is now 3.7.

Version 3.1.0

• Fixed:

  • Removed dependency on depricated numpy.str

• Changed:

  • No logging of error messages internally for reading files and checking dataframes. Instead logging is either moved to debug-level or all info is now in the error message itself to be handled externally by the user.

Version 3.0.2

• Fixed:
  • String representation of empty headers is fixed (accidentally printed 'None' before).

Version 3.0.1

• Fixed:
  • Merging functionality from TfsDataFrame.append, TfsDataFrame.join, TfsDataFrame.merge and tfs.concat do not crash anymore when encountering a pandas.DataFrame (or more for tfs.concat) in their input. Signatures have been updated and tests were added for this behavior.

Version 3.0.0

A long-standing issue where merging functionality used on TfsDataFrame (through .merge or pandas.concat for instance) would cause them to be cast back to pandas.DataFrame and lose their headers has been patched.

• Breaking changes:

  • The internal API has been reworked for clarity and consistency. Note that anyone previously using the high-level exports tfs.read, tfs.write and tfs.TfsDataFrame will not be affected.

• Added:

  • The TfsDataFrame class now has new .append, .join and .merge methods wrapping the inherited methods of the same name and fixing the aforementioned issue.
  • A tfs.frame.concat function, exported as tfs.concat, has been added to wrap pandas.concat and fix the aforementioned issue.
  • A tfs.frame.merge_headers function has been added.
  • Top level exports are now: tfs.TfsDataFrame, tfs.read, tfs.write and tfs.concat.

• Changes:

  • The tfs.frame.validate function is now a public-facing documented API and may be used stably.
  • The write_tfs function now appends an EOL (\n) at the end of the file when writing out for visual clarity and readability. This is a purely cosmetic and does not change functionality / compatibility of the files.
  • Documentation and README have been updated and cleared up.

Please do refer to the documentation for the use of the new merging functionality to be aware of caveats, especially when merging headers.

Version 2.1.0

• Changes:
  • The parsing in read_tfs has been reworked to make use of pandas's C engine, resulting in drastic performance improvements when loading files. No functionality was lost or changed.

Version 2.0.3

• Fixed:

  • Took care of a numpy deprecation warning when using np.str, which should not appear anymore for users.

• Changes:

  • Prior to version 2.0.3, reading and writing would raise a TfsFormatError in case of non-unique indices or columns. From now on, this behavior is an option in read_tfs and write_tfscalled non_unique_bahvior which by default is set to log a warning. If explicitely asked by the user, the failed check will raise a TfsFormatError.

Version 2.0.2

• Fixed:
  • Proper error on non-string columns
  • Writing numeric-only mixed type dataframes bug

Version 2.0.1

• Fixed:
  • No longer warns on MAD-X styled string column types (%[num]s).
  • Documentation is up-to-date, and plays nicely with Sphinx's parsing.
  • Fix a wrong type hint.

Version 2.0.0

• Breaking Changes:

  • FixedColumn, FixedColumnCollection and FixedTfs have been removed from the package
  • Objects are not converted to strings upon read anymore, and will raise an error
  • Minimum pandas version is 1.0

• Fixed:

  • No longer writes an empty line to file in case of empty headers
  • "Planed" dataframes capitalize plane key attributes to be consistent with other pylhc packages, however they can be accessed with and without capitalizing your query.

• Changes:

  • Minimum required numpy version is now 1.19
  • TfsDataFrames now automatically cast themselves to pandas datatypes using .convert_dtypes()
  • Lighter dependency matrix
  • Full testing of supported Python versions across linux, macOS and windows systems through Github Actions

Version 1.0.5

• Fixed:

  • Bug with testing for headers, also in pandas DataFrames
  • Same testing method for all data-frame comparisons
  • Some minor fixes

• Added:

  • Testing of writing of pandas DataFrames

Version 1.0.4

• Added:

  • support for pathlib Paths
  • strings with spaces support (all strings in data are quoted)
  • more validation checks (no spaces in header/columns)
  • nicer string representation
  • left-align of index-column

• Removed:

  • .indx from class (use index="NAME" instead)

• Fixed:

  • Writing of empty dataframes
  • Doc imports
  • Minor bugfixes

Version 1.0.3

• Fixed:
  • From relative to absolute imports (IMPORTANT FIX!!)

Version 1.0.2

• Fixed:

  • Additional index column after writing is removed again
  • Renamded sigificant_numbers to significant_digits
  • significant_digits throws proper error if zero-error is given

• Added:

  • Fixed Dataframe Class
  • Type Annotations

Version 1.0.1

• Fixed:

  • Metaclass-Bug in Collections

• Added:

  • Additional Unit Tests
  • Versioning
  • Changelog

Version 1.0.0

• Initial Release