Pyorc

--------------------

Added
~~~~~

- Module level variables for the ORC library version: orc_version string and
orc_version_info namedtuple.
- New parameter for Writer: row_index_stride.
- New read-only properties for Reader: row_index_stride and software_version.
- Trino and Scritchley writer ids.
- Type annotations support for ORC types.
- Support for `timestamp with local time zone` type.
- New parameter for Reader and Writer: timezone.
- The backported zoneinfo module dependency pior to Python 3.9.
- Predicate (SearchArgument) support for filtering row groups during ORC file
reads. New classes: Predicate and PredicateColumn.
- New parameter for Reader: predicate.
- Build for aarch64 wheels. (PR 43, contribution of odidev)

Changed
~~~~~~~

- ORC C++ Core updated to 1.7.0, and because many of the new features are not
backported to the 1.6 branch, currently this is the minimum required lib
version.
- TimestampConverter's to_orc and from_orc methods got an extra timezone
parameter, that will be bound to the same ZoneInfo object passed to the
Reader or Writer via their timezone parameters during type convert.
- Renamed Reader.metadata property and Writer.set_metadata method to
user_metadata and set_user_metadata respectively to avoid confusion.

--------------------

Added
~~~~~

- Experimental Windows support.
- tzdata package dependency on Windows. Automatically setting TZDIR
to the path of the tzdata package's data dir after importing PyORC.

Changed
~~~~~~~
- Create ORC Type from TypeDescription directly (instead of string parsing)
for Writer. (PR 26, contribution of blkerby)
- Dotted column names are allowed to use in TypeDescription.find_column_id
method with escaping them backticks.
- ORC C++ Core updated to 1.6.6.

Fixed
~~~~~

- Handling large negative seconds on Windows for TimestampConverter.from_orc.

--------------------

Added
~~~~~

- Metadata property for Reader and set_metadata for Writer to
handle ORC file's metadata.
- Meta info attributes like writer_id, writer_version, bytes_length,
compression and compression_block_size for Reader.
- New TypeDescription subclasses to represent ORC types.

Changed
~~~~~~~

- Reimplemented TypeDescription in Python.
- ORC C++ Core updated to 1.6.3.

Fixed
~~~~~

- Converting date from ORC on systems where the system's timezone
has a negative UTC offset (Issues 5)

--------------------

Added
~~~~~

- Converters for date, decimal and timestamp ORC types in Python and
option to change them via Reader's and Writer's converters parameter.
- Column object for accessing statistics about ORC columns.
- An attribute to Reader for selected schema.

Changed
~~~~~~~

- Use timezone-aware datetime objects (in UTC) for ORC timestamps by default.
- Wrapped C++ stripe object to Python Stripe.

Fixed
~~~~~

- Decrementing reference for bytes object after reading from file stream.

--------------------

Added
~~~~~

- A Reader object to read ORC files.
- A stripe object to read only a stripe in an ORC file.
- A Writer object to write ORC files.
- A typedescription object to represent the ORC schema.
- Support to represent a struct type either a Python tuple or a dictionary.