Installation¶

tabsight needs Python 3.11 or newer (3.11, 3.12, 3.13 and 3.14 are tested) and pandas 2.2 or newer (pandas 2 and 3 are both supported).

Note

tabsight is at version 0.1. It is tested on Windows, macOS and Linux; some names may still change before 1.0, and the changelog says when they do.

From PyPI¶

python -m pip install tabsight

For the newest code, or to work on tabsight itself:

git clone https://github.com/yesh069/tabsight.git
cd tabsight
python -m pip install -e .

The core package only needs pandas. Extra file formats need extra packages:

Extra	Installs	Needed for
`excel`	`openpyxl`, `defusedxml`	Reading `.xlsx` / `.xlsm` files. `defusedxml` protects against malicious XML ("XML bombs"); tabsight refuses to read Excel without it.
`parquet`	`pyarrow` (15 or newer)	Reading `.parquet` files. Older pyarrow has a known code-execution flaw (CVE-2023-47248), so tabsight refuses it.
`polars`	`polars`, `pyarrow`	Passing polars DataFrames.
`static`	`dukpy`	Still SVG pictures of charts (`fig.to_svg()`), and the picture a saved notebook shows when it is reopened or viewed on GitHub. dukpy is a small JavaScript engine for Python (MIT licence) with ready-made wheels for every supported Python; no Node.js or browser is needed.
`png`	`dukpy`, `resvg-py`	PNG pictures of charts (`fig.to_png()`, `tabsight chart ... -o chart.png`). resvg-py (MIT) turns the SVG picture into pixels; no browser is needed.
`web`	`fastapi`, `flask`	Serving dashboards in server mode from a web app. Install only the framework you use; `tabsight serve` needs neither.
`dev`	test and lint tools	Working on tabsight itself.

python -m pip install "tabsight[excel,parquet]"

import tabsight as ts

print(ts.__version__)
print(ts.ECHARTS_VERSION)  # the bundled, offline chart library