A free, fully offline, cross-platform Apache Parquet viewer. Double-click a file bigger than your RAM — it opens in under a second, in an interface that belongs on your desktop.
A Parquet file is made for machines: columnar, compressed, binary. Superb to compute on, mute to look at — you can't open it the way you open text, so between you and your own data there is always a layer of code.
The questions haven't changed and won't: what's the schema? how many rows? what do they look like? where are the nulls? None of them should cost an environment, a query, or somebody else's server. Viewda makes the glance free: double-click, look, close — whether Parquet is your daily currency or a file someone just handed you.
- Speed comes first. A glance that waits is a failure; when a feature would cost responsiveness, the feature loses.
- Your data never leaves your machine. Viewda has no telemetry and never uploads data. Its only network request is an identifier-free update check, attempted automatically at most once per 24 hours and available on demand; automatic checks can be disabled.
- The basics are free forever. Viewing, SQL, export, metadata — never behind a paywall. Apache License 2.0, in the open.
- Opens local Parquet files, from the app or straight from your file manager, and keeps recent ones a click away.
- Shows the data: typed values across every row, with visual filters, multi-column sorting, resizable columns and selectable ranges.
- Copies and exports data: copy grid selections or export a selection or the current filtered and sorted view to CSV.
- Shows the structure: size, row counts, row groups, the nested schema and on-demand column statistics.
- Matches your desktop: follows the system theme or forces Light or Dark, native window chrome included.
- Open and read everything: any size, local or over SSH; the data and the file's anatomy — schema, statistics, metadata down to row groups.
- Search and query: full-text search and full SQL with history when visual filters and sorting are not enough.
- Take the data anywhere: add JSON, Markdown and Parquet export alongside CSV.
- Withstand broken files and help debug them: everything readable opens, the damage gets named.
Every build below comes from GitHub Releases.
Install with Homebrew — it picks the right architecture, and the app keeps updating itself:
brew tap meisquietude/viewda https://github.com/MeisQuietude/Viewda
brew trust --cask meisquietude/viewda/viewda
brew install --cask viewdaOr download a DMG: Apple silicon for an M-series Mac, Intel for an older Intel Mac, or Universal if unsure. Open it and drag Viewda to Applications.
Either way, builds are not notarized yet; if macOS blocks the first launch, choose System Settings → Privacy & Security → Open Anyway.
Download and run the .exe. It is not code-signed yet, so SmartScreen may
require More info → Run anyway.
Download the AppImage, run chmod +x Viewda_*.AppImage, then open it. The
AppImage is recommended because it supports in-app updates. The .deb
package is a manual-update fallback and is not recommended for most users.
AppImage file associations require integration through appimaged or
AppImageLauncher.
A Rust core owns the data — windowed reads, metadata, SQL (DuckDB, arrow-rs) — and streams Arrow columns to a virtualized React DOM grid in a Tauri shell. Between them sits one transport-agnostic engine protocol: shell, engine and UI can each be swapped without rewriting the rest.
Bug reports and ideas are welcome at any time; code once the change is agreed in an issue. CONTRIBUTING.md has the rest: the toolchain, the one command that gates every change, and what a pull request is expected to carry. Everyone taking part follows the Code of Conduct.
Report a vulnerability privately, the way SECURITY.md describes, never in a public issue.