v0.3.1

• Dropped support for Python 3.7, added support for Python 3.12.
• Removed use of deprecated Py_UNICODE APIs.
• Also build wheels for Linux 64-bit ARM architecture (manylinux-aarch64)
• __main__: add options for indent level, pager and json output

v0.3.0.post1

• Build wheels for Python 3.11 and universal2 wheels for macOS with Apple Silicon. No code changes.

v0.3.0

Replace use of array.array with C++ vector to make it work on pypy3 (#12)
Added wheels for CPython 3.10 and for PyPy 3.7.

v0.2.2.post1

Add CPython 3.8 and 3.9 wheels, drop 2.7.

v0.2.2

Write strings quoted if they can be confused with an integer or a float, or if they contain any characters that is not in the following set:

$.0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz

v0.2.1

Reverted the default value of use_numbers to False in the parser's load and loads functions. So by default the parser will treat any plist string whether quoted or unquoted as a Python unicode string.

Set it to True to enable the feature whereby unquoted plist strings are attempted to be converted to Python int or float.

v0.2.0

Implemeted an OpenStep plist writer. Added two new public functions dump and dumps: one writes to a file object, the other returns a unicode string.

The parser now attempts to cast unquoted plist strings to Python int or float, unless use_numbers=False is disabled.

v0.1.0

Initial release. The only two public functions exported are loads and load: one reads from a string, the other from a file object.
They parse the plist into a Python dict, list, unicode string, bytes (for binary data) or any combination of them.
A custom dict_type can also be defined (e.g. OrderedDict).