databrowser 1.1.1

Quick tool to browse and view data files (json,csv,parquet,more) on local disk and s3

databrowser

A terminal file browser for quickly viewing data files, from local disk or S3.

Supports csv, tsv, parquet, feather, orc, json, xlsx, xls, xml and html — read into pandas DataFrames and shown in a scrollable table (with a dtype view). Built on Textual.

Originally based on the example code_browser from Textual.

install

Run without installing (uvx)

With uv, run it in a throwaway environment — no install needed:

uvx databrowser [path]
uvx databrowser s3://bucket/dir/
uvx databrowser hf://datasets/org/name

For Google Cloud Storage or Azure, pull in the matching extra:

uvx --from 'databrowser[gcs]' databrowser gs://bucket/dir/
uvx --from 'databrowser[azure]' databrowser az://container/dir/

Install from PyPI

The package is published on pypi https://pypi.org/project/databrowser/

run pip install databrowser (pip install databrowser --upgrade to get latest version). Then just execute databrowser. The gcs / azure / all extras add those cloud backends.

Build

This project uses uv (Python >= 3.13).

Execute uv sync to create the virtual env and install the package and its dependencies.

Run uv run databrowser to execute in the virtual env (or uv run python src/databrowser/data_browser.py).

Usage

databrowser [optional path] (defaults to the current directory)

Remote filesystems

Any fsspec URL works, not just local paths:

• databrowser s3://bucket/path/ — S3 (uses the default AWS credentials in the environment)
• databrowser hf://datasets/org/name — Hugging Face datasets (set HF_TOKEN for private ones)
• databrowser gs://bucket/path/ — Google Cloud Storage (needs gcsfs installed)

The protocol is resolved automatically, so the same browsing works for every backend.

Select a data file to view

• press F to hide the filebrowser

• press D to show the dtypes

• press S to save a screenshot in svg

• press Q to quit

By default the preview shows the first 100 rows. Set DATABROWSER_ROWS to change it, e.g. DATABROWSER_ROWS=500 databrowser data/ (or pass --rows 500).

Run databrowser --help for all options.

examples