Aspecd

=============

Released 2021-11-05


New features
------------

* Reports

* Templates for reporting information contained in datasets come bundled with ASpecD.
* Context contains ``templates_dir`` allowing to include sub-templates.
* New class :class:`TxtReporter` for plain text reports

* Tasks/Recipe-driven data analysis

* YAML representation of recipe and tasks via :meth:`aspecd.tasks.Recipe.to_yaml` and :meth:`aspecd.tasks.Task.to_yaml`
* Figure labels can be set in plotters; otherwise a default label will be set and can be accessed from within reports.

* Utils

* :func:`change_working_dir` can be used as context manager to temporarily change the working directory.

* General

* :meth:`aspecd.utils.ToDictMixin.to_dict` can optionally remove keys with empty values.
* ``to_dict()`` method in :class:`aspecd.processing.ProcessingStep`, :class:`aspecd.analysis.AnalysisStep`, :class:`aspecd.annotation.Annotation`, :class:`aspecd.plotting.Plotter`, :class:`aspecd.table.Table`, :class:`aspecd.report.Reporter`, :class:`aspecd.model.Model`

* Models

* Axes quantities and units can be explicitly set on model creation.


Changes
-------

* Dataset labels do not contain source path.
* Recipe dataset_source and output directories are no longer converted to absolute paths.
* More complete recipe history for tasks, including more of their properties
* Recipe-driven data analysis: Figures get added to recipe with default label if no label is provided.
* :class:`aspecd.processing.Noise`: explicit noise amplitude can be given.
* Model can add label to created dataset.
* ModelTask adds result label as id to result.
* Plotter: Default figure size changed to (6., 4.) inch


Fixes
-----

* :meth:`aspecd.tasks.Task.to_yaml` serialises numpy arrays
* Datasets from foreign packages are correctly listed in recipe history
* :func:`aspecd.utils.copy_keys_between_dicts` properly traverses
* :class:`aspecd.utils.Yaml` handles :class:`numpy.double`
* Recipe-driven data analysis: automatically generated figure filenames get added to recipe figure record
* Models work now correctly when based on a dataset
* :class:`aspecd.model.FamilyOfCurves` sets correct values for additional axis
* :class:`aspecd.processing.Differentiation` works correctly for 2D datasets
* :class:`aspecd.processing.Noise`: normalisation works with >1D datasets
* :class:`aspecd.plotting.SinglePlotter2DStacked`: ylabel is set to third axis if offset = 0

=============

Released 2021-10-12

New features
------------

* Tasks/Recipe-driven data analysis

* YAML representation of tasks and recipes using :meth:`aspecd.tasks.Task.to_yaml` and :meth:`aspecd.tasks.Recipe.to_yaml` - convenience methods for later use in guided recipe generation
* :class:`aspecd.tasks.AggregatedanalysisTask` for performing a SingleAnalysisStep on a series of datasets, aggregating the result in a CalculatedDataset
* :class:`aspecd.tasks.TabulateTask` for tabular representation of data of a dataset

* Datasets

* New attribute :attr:`aspecd.dataset.Axis.index` (for individual labels for each data point, similar to pandas and for tabular data)
* :meth:`aspecd.dataset.Dataset.tabulate` to create tables from datasets

* Analysis steps

* New class :class:`aspecd.analysis.AggregatedAnalysisStep` for aggregating the results of a SingleAnalysisStep on multiple datasets in a CalculatedDataset

* Tabular representation of datasets

* New module :mod:`aspecd.table`
* Series of output formats for tables (including DokuWiki and LaTeX)
* Tables can have captions that are output as well

* Utils

* :func:`aspecd.utils.get_package_data` for obtaining package data (*i.e.*, non-code files contained in distribution)


Changes
-------

* :class:`aspecd.analysis.BasicCharacteristics` always returns scalars or lists in its results and writes index (for compatibility with :class:`aspecd.analysis.AggregatedAnalysisStep` and tabular output).

=============

Released 2021-10-08

**Note**: Starting with this release ASpecD requires **Python >= 3.7**.

New features
------------

* Tasks/Recipe-driven data analysis

* New attribute :attr:`aspecd.tasks.PlotTask.target` allows adding a plot to an already existing plot.
* :meth:`aspecd.tasks.Task.to_dict` adds (implicit) parameters of underlying task object
* Classes from the ASpecD framework can be used without prefixing them with "aspecd" in recipes with "default_package" set to a package based on the ASpecD framework.
* ``serve`` command outputs log messages for each task
* Command-line options for ``serve`` setting the log level/verbosity
* Catching of errors, excluding the stack trace and only showing the error message (but full stack trace in verbose mode)
* Switch in recipe to suppress writing history (for development/debugging, issuing warning on the command line via logging)
* New structure of recipes: Move ``default_package`` and ``autosave_plots`` to new dict ``settings``; ``output_directory`` and ``datasets_source_directory`` to new dict ``directories``
* Add ``format`` dict to recipe with fields ``type`` and ``version``
* Automatically convert old recipe formats within :class:`aspecd.io.RecipeYamlImporter`
* Processing steps writing parameters during execution and applied to multiple datasetes are unpacked in the recipe history if these parameters change for each dataset


* References in processing and analysis steps and models (using bibrecord package)

=============

Released 2021-09-21

The following bugs have been fixed:

* Handling of lists as properties in recipes
* Improved handling of axes labels with xkcd style
* Offset in SinglePlotter2DStacked can be set to zero

=============

Released 2021-09-02

**Note**: This is the last ASpecD release with explicit support for Python 3.5.

New features
------------

* Processing steps

* Adding (coloured) noise to datasets (:class:`aspecd.processing.Noise`)
* Provide a new range of axis values for a dataset for correction (:class:`aspecd.processing.ChangeAxesValues`)

* Analysis steps

* Power spectral density of 1D dataset (:class:`aspecd.analysis.PowerDensitySpectrum`), *e.g.*, for analysing noise
* Polynomial fit of 1D data (:class:`aspecd.analysis.PolynomialFit`)
* Linear regression of 1D data without fitting the intercept (:class:`aspecd.analysis.LinearRegressionWithFixedIntercept`)
* Additional methods in :class:`aspecd.analysis.BlindSNREstimation`

* Class :class:`aspecd.model.Model`

* New attribute :attr:`aspecd.model.Model.description`
* New non-public method ``_sanitise_parameters``

* New models

* :class:`aspecd.model.Polynomial` for evaluating polynomials (*e.g.*, as obtained using :class:`aspecd.analysis.PolynomialFit`)
* :class:`aspecd.model.Zeros`
* :class:`aspecd.model.Ones`
* :class:`aspecd.model.Gaussian`
* :class:`aspecd.model.NormalisedGaussian`
* :class:`aspecd.model.Lorentzian`
* :class:`aspecd.model.NormalisedLorentzian`
* :class:`aspecd.model.Sine`
* :class:`aspecd.model.Exponential`

* :class:`aspecd.model.CompositeModel` for models consisting of a (weighted) sum of individual models
* :class:`aspecd.model.FamilyOfCurves` for inspecting systematic variations of one parameter of a given model

* Tasks

* Comments can be added easily to processing and analysis steps using the top-level key ``comment`` of the respective task.


* Utils

* :func:`aspecd.utils.not_zero` ensuring a float not to cause DivisionByZero errors


Changes
-------

* :class:`aspecd.processing.Differentiation` uses :func:`numpy.gradient` instead of :func:`numpy.diff`
* :class:`aspecd.processing.BaselineCorrection` returns polynomial coefficients in unscaled data domain


Fixes
-----

* Axis labels without "/" if no unit is present
* :class:`aspecd.metadata.Measurement` handles dates imported from YAML (implicitly converted into datetime.date object)

=============

Released 2021-06-19

The following bugs have been fixed:

* Normalisation to minimum now divides by absolute value of minimum

* Normalisation raises ValueError in case of unknown kind

* Import with explicit importer when importer resides in (sub)package

* Recipe history shortens dataset source if dataset_source_directory has trailing slash