Skip to content

v0.17.0

Latest
Latest

Choose a tag to compare

View all tags
@github-actions github-actions released this 09 Aug 03:17
v0.17.0
13b7d96

Changelog

All notable changes to this project will be documented in this file.

Unreleased

0.17.0 - 2026-08-09

  • create_backend() (and the backend constructors it dispatches to) now accept a
    column_names argument: the labels the caller has for the data's columns, which are
    applied in place of the ones the data carries or lacks. create_backend(None, column_names=[...]) now builds an empty table with those columns instead of raising,
    and records use those names in place of f0, f1, ...
    (#165).

0.16.1 - 2026-08-07

  • textual_fastdatatable.backend can now be imported without importing Textual: the
    DataTable widget is now imported lazily by the package root (PEP 562), so
    from textual_fastdatatable.backend import create_backend no longer drags in the widget
    framework. from textual_fastdatatable import DataTable is unchanged, and still imports
    Textual at the moment the name is accessed.
  • pyarrow.parquet is now imported lazily, inside ArrowBackend.from_parquet(), instead
    of at module scope in backend.py.
  • Together these cut from textual_fastdatatable.backend import create_backend from 359
    to 186 imported modules on Python 3.10 (measured against this project's required
    dependencies, without the polars extra). There are no API changes.

0.16.0 - 2026-08-05

Features

  • DataTable(data=...) and create_backend() now accept a pandas DataFrame, which is
    converted with the ArrowBackend.
  • Copying a selection is now bound to super+c (cmd+c) as well as ctrl+c, matching the
    copy bindings Textual added in 7.3.0 for terminals that report the command key.

Dependency Changes (Breaking!)

  • Drops support for Python 3.9; the minimum supported version is now Python 3.10.
  • Requires Textual >= 7.3.0.
  • Drops the numpy and pandas dependencies.
  • pyarrow upper bounds have been removed.
  • Adds a test that resolves this project's dependencies for every supported Python (3.10 -
    3.14) and platform (Linux, macOS, and Windows, on x86_64 and arm64) with wheels only.
  • Declares typing-extensions as a dependency; it was imported but only installed as a
    transitive dependency of Textual.

0.15.0 - 2026-08-03

0.14.0 - 2025-10-25

  • Fixes column rendering bug to make compatible with recent versions of Textual.
  • Updates pyarrow and numpy dependencies to provide compatibility with Python 3.14

0.13.0 - 2025-10-16

  • Makes dependecies more exact to ensure installations use wheels for large dependencies (numpy, pyarrow, pandas, etc.).

0.12.0 - 2025-02-06

  • Catch overflow errors when casting arrow temporal types to python date and datetimes, and substitue date.max/min and datetime.max/min, instead of None.
  • Format date.max/min and datetime.max/min with an infinity symbol (∞) when rendering cells with those values.

0.11.0 - 2024-12-19

  • Drops support for Python 3.8
  • Adds support for Python 3.13

0.10.0 - 2024-10-31

  • Adds an optional parameter to DataTable to disable rendering of string data as Rich Markup.
  • Fixes a bug where None could be casted to a string and displayed as "None" (tconbeer/harlequin#658)

0.9.0 - 2024-07-23

  • Adds a PolarsBackend implementation of DataTableBackend. You must have polars installed to use the PolarsBackend. You can install it using the polars extra for this package.
  • Fixes a crash from the ArrowBackend when attempting to instantiate negative datetimes after a timezone conversion.

0.8.0 - 2024-07-10

  • Fixes a crash when cell contents contained bad Rich Markdown (tconbeer/harlequin#569).
  • Improves the appearance of data tooltips.

0.7.1 - 2024-02-09

  • Adds a backend.source_data property to exposue the underlying Arrow table, before slicing.

0.7.0 - 2024-02-07

Breaking Changes

  • Removes the NumpyBackend (#78).

Features

  • Values are now formatted based on their type. Numbers have separators based on the locale, and numbers, dates/times/etc., and bools are right-aligned (#70).

Bug Fixes

  • Fixes bug that caused either a crash or an empty table from initializing a table from_records or from_pydict with mixed (widening or narrowing) types in one column.

0.6.3 - 2024-01-09

Bug Fixes

  • Widens acceptable types for create_backend to accept a sequence of any iterable, not just iterables that are instances of typing.Iterable.

0.6.2 - 2024-01-08

Bug Fixes

  • Adds the tzdata package as a dependency for Windows installs, since Windows does not ship with a built-in tzdata database.

0.6.1 - 2024-01-05

Bug Fixes

  • Fixes the behavior of tab and shift+tab to cycle to the next/prev row if at the end/start of a row or table.
  • Fixes a crash from pressing ctrl+c when the cursor type is column.

0.6.0 - 2024-01-05

Features

  • Adds keybindings for navigating the cursor in the data table. ctrl+right/left/up/down/home/end (with shift variants), tab, shift+tab, ctrl+a now all do roughly what they do in Excel (if the cursor type is range).

0.5.1 - 2024-01-05

Bug Fixes

  • Adds a dependency on pytz for Python <3.9 for timezone support.
  • Fixes a bug where Arrow crashes while casting timestamptz to string (tconbeer/harlequin#382).

Performance

  • Vectorizes fallback string casting for datatypes unsupported by pc.cast (#8)

0.5.0 - 2023-12-21

Features

  • Adds a range cursor type that will highlight a range of selected cells, like Excel.
  • ctrl+c now posts a SelectionCopied message, with a values attribute that conttains a list of tuples of values from the data table.
  • Adds a max_column_content_width parameter to DataTable. If set, DataTable will truncate values longer than the width, but show the full value in a tooltip on hover.

0.4.1 - 2023-12-14

  • Fixes a crash caused by calling create_backend with an empty sequence.

0.4.0 - 2023-11-14

Breaking API Changes

  • When calling create_backend with a sequence of iterables, the default behavior now assumes the data does not contain headers. You can restore the old behavior with create_backend(has_headers=True).
  • When calling DataTable(data=...) with a sequence of iterables, the first row is treated as a header only if column_labels is not provided.

0.3.0 - 2023-11-11

Features

  • The DataTable now accepts a max_rows kwarg; if provided, backends will only store the first max_rows and the DataTable will only present max_rows. The original row count of the data source is available as DataTable().source_row_count (tconbeer/harlequin#281).

API Changes

  • Backends must now accept a max_rows kwarg on initialization.

0.2.1 - 2023-11-10

Bug Fixes

  • Tables with the ArrowBackend no longer display incorrect output when column labels are duplicated (#26).

0.2.0 - 2023-11-08

Features

  • Adds a null_rep: str argument when initializing the data table; this string will be used to replace missing data.
  • Adds a NumpyBackend that uses Numpy Record Arrays; this backend is marginally slower than the ArrowBackend in most scenarios (#23).

Bug Fixes

  • Fixes a crash when using ArrowBackend.from_records(has_header=False).

Performance

  • Drastically improves performance for tables that are much wider than the viewport (#12).

Benchmarks

  • Improves benchmarks to exclude data load times, disable garbage collection, and include more information about first paint and scroll performance.

0.1.4 - 2023-11-06

  • Fixes a crash when computing the widths of columns with no rows (#19).

0.1.3 - 2023-10-09

  • Fixes a crash when creating a column from a null or complex type.

0.1.2 - 2023-10-02

0.1.1 - 2023-09-29

  • Fixes a crash when rows were added to an empty table.

0.1.0 - 2023-09-29

  • Initial release. Adds DataTable and ArrowBackend, which is 1000x faster for datasets of 500k records or more.
Assets 5
Loading