qvd

High-performance Rust library for reading, writing, converting and merging Qlik QVD files.

Disclaimer: This project is not affiliated with Qlik Technologies or QlikTech. QVD is a proprietary format; all trademarks belong to Qlik. This library was built from scratch using publicly available information. See LEGAL.md for details.

Features

• Read/Write QVD with byte-identical roundtrip (tested on 399 files up to 2.8 GB)
• Streaming chunk-based reader for large files
• EXISTS() O(1) index + filtered reads (2.5x faster than Qlik Sense)
• Concatenate — pure append with strict/union schema modes
• PK Merge — upsert/dedup by primary key (replace/skip/error). First QVD library with this
• Parquet/Arrow — bidirectional conversion, write_arrow() for direct Arrow-to-QVD
• DuckDB / DataFusion — register QVD as SQL tables
• Python — PyArrow, pandas, Polars via zero-copy Arrow bridge
• Node.js / TypeScript — native bindings via napi-rs, async I/O
• CLI — inspect, convert, filter, head

Quick Start

# Cargo.toml
qvd = "0.7.1"

let table = qvd::read_qvd_file("data.qvd")?;
qvd::write_qvd_file(&table, "copy.qvd")?;

pip install qvdrs               # core + Arrow (pyarrow auto-installed)
pip install "qvdrs[pandas]"     # + pandas
pip install "qvdrs[polars]"     # + polars
pip install "qvdrs[duckdb]"     # + duckdb
pip install "qvdrs[all]"        # everything

import qvd
table = qvd.read_qvd("data.qvd")
table.save("copy.qvd")

npm install qvdrs

import { readQvd, saveQvd } from 'qvdrs'
const table = await readQvd('data.qvd')
await saveQvd(table, 'copy.qvd')

Documentation

License

MIT — Stanislav Chernov (@bintocher)