dataview-tui 0.4.0

Keyboard-driven TUI for browsing columnar data files.

dataview-tui

dataview-tui is a keyboard-driven terminal UI for inspecting columnar data files. The command line entry point is dview.

The project supports Parquet, ORC, Avro, and optional Lance datasets. It is built for quickly moving between schema, data preview, metadata, and SQL query views without leaving the terminal.

Features

• Textual-based terminal application with a project-specific dark theme.
• Automatic format detection by extension and magic bytes.
• S3-compatible object storage reads for Parquet, ORC, and Avro objects.
• PyArrow-backed Parquet and ORC readers.
• fastavro-backed Avro reader.
• Optional Lance reader via the lance extra.
• Schema, Data, Metadata, and SQL sidebar views.
• Schema tree entries include field type and nullable state.
• Metadata view includes file info, row groups/fragments/blocks, and per-column details.
• DuckDB SQL engine exposing the opened file as data and t.
• CLI entry points: dview and dataview.

Installation

pipx install dataview-tui
dview --version

For Lance datasets:

pipx install "dataview-tui[lance]"

To update an existing install:

pipx upgrade dataview-tui

To install the current main branch from source:

pipx install git+https://github.com/lyne7-sc/dataview-tui.git

Quick Start

Open supported files or datasets:

dview path/to/file.parquet
dview path/to/file.orc
dview path/to/file.avro
dview path/to/dataset.lance

Open a Parquet, ORC, or Avro object from S3-compatible storage:

dview s3://bucket/path/to/file.parquet

For MinIO or another S3-compatible endpoint:

dview s3://bucket/path/to/file.parquet \
  --s3-endpoint-url http://localhost:9000 \
  --s3-access-key minioadmin \
  --s3-secret-key minioadmin \
  --s3-region us-east-1

S3-compatible reads use PyArrow's S3 filesystem. If explicit credentials are not provided, PyArrow uses the available AWS credential chain. SQL is disabled for remote S3-compatible objects in this version. Remote Lance datasets are not supported yet.

Choose an initial tab:

dview path/to/file.parquet --tab data

Shell Completion

dview can print static shell completion scripts. Static completion keeps Tab completion fast because the shell does not need to start the Python TUI app for each completion request.

For zsh:

mkdir -p ~/.zfunc
dview --print-completion zsh > ~/.zfunc/_dview

Make sure ~/.zfunc is on your fpath, then restart your shell or run compinit.

For bash or fish, replace zsh with bash or fish.

Keyboard

Key	Action
q	Quit
1 / 2 / 3	Show Schema / Data / Metadata tab
] / [	Next page / previous page
Ctrl+F	Filter columns by regex
/	Search cell contents
n / Shift+N	Next / previous match
Enter	Show full current cell value
y	Copy current cell value
s	Toggle SQL sidebar
Ctrl+Enter / Ctrl+J / F5	Run SQL

SQL

The SQL engine exposes the opened file as DuckDB views named data and t:

SELECT * FROM data LIMIT 20;
SELECT count(*) FROM t;

Query results are capped by the engine to keep the terminal responsive.

Terminal Setup

The app cannot force a terminal font. For a clearer table and SQL experience, use a monospaced font with readable punctuation and digits, such as JetBrains Mono or Maple Mono.

Development

Install dependencies and run checks:

uv sync --extra dev
uv run pytest -q
uv run ruff check .
uv build

Project layout:

src/dataview/
  app.py
  cli.py
  readers/
  widgets/
tests/

Publishing

Releases are built by GitHub Actions when a GitHub Release is published. PyPI publishing uses PyPI Trusted Publishing, so the PyPI project must be configured with this repository and .github/workflows/release.yml before publishing a release.

Dependencies

Runtime:

• pyarrow>=12
• textual>=0.40
• fastavro>=1.9
• duckdb>=0.9
• tree-sitter>=0.25
• tree-sitter-sql>=0.3.11

Optional:

• pylance>=0.20 via the lance extra

Development:

• pytest
• pytest-asyncio
• ruff

License

MIT