Guidata

In this release, test coverage is 74%.

💥 New features:

* New Sphinx autodoc extension:
* Allows to document dataset classes and functions using Sphinx directives, thus generating a comprehensive documentation for datasets with labels, descriptions, default values, etc.
* The extension is available in the `guidata.dataset.autodoc` module
* Directives:
* `autodataset`: document a dataset class
* `autodataset_create`: document a dataset creation function
* `datasetnote`: add a note explaining how to use a dataset
* `BoolItem`/`TextItem`: add support for callbacks when the item value changes

🛠️ Bug fixes:

* Documentation generation: automatic requirement table generation feature was failing
when using version conditions in the `pyproject.toml` file (e.g. `pyqt5 >= 5.15`).
* [Issue 72](https://github.com/PlotPyStack/guidata/issues/72) - unit test leave files during the build usr/lib/python3/dist-packages/test.json
* [Issue 73](https://github.com/PlotPyStack/guidata/issues/73) - `ChoiceItem` radio buttons are duplicated when using callbacks

In this release, test coverage is 76%.

🛠️ Bug fixes:

* [Issue 71](https://github.com/PlotPyStack/guidata/issues/71) - Random segmentation faults with applications embedding `CodeEditor`
* [Issue 70](https://github.com/PlotPyStack/guidata/issues/70) - PermissionError: [Errno 13] Permission denied: '/usr/lib/python3/dist-packages/guidata/tests/data/genreqs/requirements.rst'

In this release, test coverage is 76%.

💥 New features:
* `dataset.io.h5fmt.HDF5Reader.read` method: added new `default` argument to set
default value for missing data in the HDF5 file (backward compatible). The default
value of `default` is `NoDefault` (a special value to indicate that no default value
should be used, and that an exception should be raised if the data is missing).
* `widgets.codeeditor.CodeEditor`: added new `inactivity_timeout` argument to set
the time (in milliseconds) to wait after the user has stopped typing before
emitting the `CodeEditor.SIG_EDIT_STOPPED` signal.
* Added `execenv.accept_dialogs` attribute to control whether dialogs should be
automatically accepted or not (default is `None`, meaning no automatic acceptance):
this allows more coverage of the test suite. For now, this attribute has only been
proven useful in `tests/dataset/test_all_features.py`.
* Added unit tests for HDF5 and JSON serialization/deserialization:
* Testing an arbitrary data model saved/loaded to/from HDF5 and JSON files,
with various data sets and other data types.
* Testing for backward compatibility with previous versions of the data model
(e.g. new attributes, removed attributes, etc.)

⚠️ API breaking changes:
* `guidata.dataset.io` module is now deprecated and will be removed in a future
release. Please use `guidata.io` instead. This change is backward compatible
(the old module is still available and will be removed in a future release).
The motivation for this change is to simplify the module structure and to help
understand that the scope of the `io` module is not limited to `dataset.DataSet`
objects, but may be used for any kind of data serialization/deserialization.

📖 Documentation:

* Added missing `DataSetEditDialog` and `DataSetEditLayout` classes
* Added missing inheritance/member details on some classes
* Reduced table of contents depth in left sidebar for better readability

In this release, test coverage is 72%.

💥 New features:
* Array editor now supports row/column insertion/deletion:
* Added `variable_size` argument to `setup_and_check` method
* The feature is disabled by default (backward compatible)
* It supports standard arrays, masked arrays, record arrays and N-dimensional arrays
* New dataset read-only mode:
* Added `readonly` argument to `DataSet` constructor
* This is useful to create a dataset that will be displayed in read-only mode
(e.g. string editing widgets will be in read-only mode: text will be selectable
but not editable)
* The items remain modifiable programmatically (e.g. `dataset.item = 42`)
* New dataset group edit mode:
* Added `mode` argument to `DataSetGroup.edit` method, with the following options:
* `mode='tabs'` (default): each dataset is displayed in a separate tab
* `mode='table'`: all datasets are displayed in a single table
* In the new table mode, the datasets are displayed in a single table with
one row per dataset and one column per item
* Clicking on a row will display the corresponding dataset in a modal dialog box

🛠️ Bug fixes:

* Qt console:
* Fixed `RuntimeError: wrapped C/C++ object of type DockableConsole has been deleted`
when closing the console widget (parent widget, e.g. a `QMainWindow`, was deleted)
while an output stream is still writing to the console (e.g. a `logging` handler
which will flush the output stream when closing the application)
* This concerns all console-related widgets: `DockableConsole`, `Console`,
`InternalShell`, `PythonShellWidget` and `ShellBaseWidget`
* Code editor: fixed compatibility issue with PySide6
(`AttributeError: 'QFont' object has no attribute 'Bold'`)

🛠️ Bug fixes:

* Fixed translation support (`gettext`):
* Locale detection has been fixed in 3.1.1 (deprecation of `locale.getdefaultlocale`)
* However, on frozen distributions on Windows (e.g. with `pyinstaller`), function
`locale.getlocale` is returning `(None, None)` instead of proper locale infos
* Added a workaround: on Windows, if locale can't be detected, we now use the
Windows API to retrieve it (using the `GetUserDefaultLocaleName` function)
* [Issue 68](https://github.com/PlotPyStack/guidata/issues/68) - Windows: gettext translation is not working on frozen applications
* Embedded Qt console:
* Fixed default encoding detection on frozen applications on Windows
* [Issue 69](https://github.com/PlotPyStack/guidata/issues/69) - Windows/Qt console: output encoding is not detected on frozen applications

🛠️ Bug fixes:

* Tests only: `qthelpers.close_widgets_and_quit` now ignores deleted widgets

💥 Changes:

* `dataset.ImageChoiceItem` and `dataset.ButtonItem`: added `size` argument to set the icon size
* `dataset.io` reader and writer classes: removed deprecated `write_unicode` method