fsspec-browser

Terminal and web browser for fsspec-backed filesystems.

Overview

fsspec-browser lets you inspect local files, object stores, and other fsspec-compatible filesystems from a terminal UI or a local web UI. It is built for browsing, previewing, and downloading files without writing one-off scripts.

Install

pip install fsspec-browser

Install any fsspec backend packages your URLs require, such as S3, GCS, SSH, or cloud vendor integrations.

Terminal Browser

fsspec-browser /tmp
fsspec-browser s3-rs://my-bucket/path -o endpoint_url=https://...

Use Ctrl-A, then Left or Right, to focus the browser or preview pane. Arrow keys operate on the focused pane: they navigate files in the browser and scroll vertically or horizontally in the preview. Enter opens directories, p previews remote files, and d downloads the selected file. Ctrl-U/Ctrl-D and H/L remain preview scrolling shortcuts. Arrow IPC and Parquet render as tables through fsspec-data; scrolling near the end loads another --preview-rows page.

Web Browser

fsspec-browser-web /tmp --host 127.0.0.1 --port 8765
fsspec-browser-web --host 127.0.0.1 --port 8765

When started without a path, the web UI opens a connection form. File previews are explicit and bounded by --preview-bytes. Database relations and columns, CSV, JSON/JSONL, and Parquet render as tables. Scrolling the preview loads the next --preview-rows page.

Browse a DuckDB database

Install the database extra and create the Superstore fixture:

pip install 'fsspec-browser[database]'
python examples/create_superstore_duckdb.py

Start the web browser with the DuckDB database as a storage option:

fsspec-browser-web db+duckdb:// \
  -o database=examples/superstore.duckdb \
  --preview-rows 20

Tables and stored views remain directory-like database relations, but are explicitly previewable as bounded row sets. Use SQL Preview for an arbitrary read query.

Documentation

• Preview local data: guided first use.
• Use the terminal or web browser: task-oriented guides.
• Understand browsing and previews: design and backend boundaries.
• Python API reference: embedding and server helpers.