Home

qsv Wiki

qsv is a blazing-fast, composable data-wrangling toolkit for CSV, TSV, SSV, Excel, JSON/JSONL, Parquet and more. 78 commands across four binary variants, and a single binary that fits in your toolbox.

Latest release: 22.0.1 — the "Data Schematic" release — qsv's biggest release ever, with 560+ commits since 21.1.0. The headliner is viz, an entirely new command that turns a CSV into interactive plotly charts and maps, with viz smart auto-designing a Data Schematic — a single, self-contained, offline-capable HTML rendering of a dataset's schema and statistics in which every claim shown is checkable against the data it describes (format spec; gallery). Schematics are explorable, not just viewable: an embedded data viewer drawer puts the underlying rows beside the charts, cross-linked with map points in both directions, and the whole UI renders in the dataset's own language. Three more new commands land alongside it: denull (detect the null sentinels that silently degrade typing), fixedwidth (fixed-width text → CSV) and clean (remove qsv-generated cache files). ⚠️ Three breaking changes — minijinja 2.23 changes rendered template output, the cached 2 → 3 migration swaps the on-disk cache backend from sled to redb (invalidating existing caches and inverting the meaning of a TTL of 0), and describegpt's bundled prompt file bumps 8.0.0 → 9.0.0. See the full changelog.

Welcome! This wiki is the workflow + discoverability layer for qsv. Use it to learn how the commands fit together and find recipes for real-world tasks. For per-command flag references, follow the deep-links into /docs/help/ — that's the canonical reference.

Three ways to start

Path	When
Try it in your browser	You want to evaluate qsv before installing anything
Install qsv	You're ready to run it locally
Read the Whirlwind Tour	You want a guided 2.7M-row walkthrough

Then jump to Getting Started for a hands-on tutorial, or Why qsv? for the elevator pitch.

Where things live

qsv documentation lives on several surfaces. Each has a different job:

Surface	Job
README.md	Project overview, command catalog with legend, install options, feature flags
/docs/ (canonical reference)	The source of truth: PERFORMANCE.md, BENCHMARKS.md, ENVIRONMENT_VARIABLES.md, STATS_DEFINITIONS.md, Validate.md, Fetch.md, Describegpt.md, INTERPRETERS.md, plus the whirlwind tour
/docs/help/	Per-command help — flags, arguments, examples
This wiki	Workflow & discoverability: getting started, cookbook recipes, command-reference category guides, tuning, integrations, troubleshooting
100.dathere.com	Interactive lessons & exercises (WIP)
Discussions	Q&A, FAQ, announcements, ideas
	LLM-generated Wiki that crawls the codebase once a week, that you can chat with!

Browse by tier

Every wiki page has a Tier badge — Beginner, Intermediate, or Advanced. See the Tier Legend for what each tier assumes.

Beginner

Installation · Getting Started · Why qsv? · Command Reference (index) · Cookbook · Troubleshooting · FAQ · Glossary · External Resources · qsv pro

Intermediate

Binary Variants · Transform & Reshape · Aggregation & Statistics · Joins & Set Ops · Validation & Schema · Conversion & I/O · Geospatial · HTTP & Web · Indexing, Compression & Diff · AI & Documentation · Comparison vs others · Integrations · Claude Cowork Plugin · MCP Server

Advanced

SQL & Polars · Scripting (Luau / Python) · Performance Tuning · Environment Variables · Stats Cache & Caching · Lookup Tables · Contributing to the Wiki

Anchor datasets used across this wiki

Examples in this wiki reach for a small set of datasets repeatedly, so you build mental models:

• wcp.csv — World Cities Population, 2.7M rows / 124 MB. Used in Getting Started and many command-reference examples. Available in files/wcp.zip.
• NYC 311 Service Requests — public dataset, available as a 1M-row sample and a full 16 GB / 27M-row export. Used in Validation, SQL & Polars, Geographic Enrichment, Larger-than-RAM.
• Boston 311 — small (100 rows) and in many compression formats (.csv, .gz, .zst, .parquet). Used in Clean & Normalize and conversion demos.
• Allegheny County property sales — mid-size, already indexed. Used in Build a Data Pipeline and Stats → Insights.
• NOAA GHCN-Daily — weather time-series, fetched via qsv fetch. Used in Fetch & Cache and Multi-table Joins (asof joins).
• GitHub stargazers — qsv's own stars, fetched via qsv fetch. Used in HTTP & Web demos.

Try it in your browser

No install required: visit qsv.dathere.com to try qsv against a sample CSV right now.

Talks & blog series

• "Have we achieved ACI?" series: 1 · 2 · 3
• NYC SOD 2026 · csv,conf,v9 · PyCon US 2025 · csv,conf,v8 · NYC SOD 2022
• Full list: External Resources

Legacy pages (preserved during wiki rewrite)

These pages predate the wiki rewrite. Their content has been integrated into the new structure but the originals are kept for backwards-compatibility with external links:

• Cookbook (legacy) — original recipes; new Cookbook expands them
• qsv cheat sheet — pre-rewrite infographic & quick reference (WIP)
• Luau Development — folded into Scripting (Luau / Python)
• Luau Helper Functions Examples — referenced from Scripting (Luau / Python)
• describegpt model notes — folded into AI & Documentation
• xsv 0.13.0 stats vs qsv 1.0.0 stats — referenced from Comparison vs others

Contributing to this wiki

This wiki is itself a git repo at https://github.com/dathere/qsv.wiki.git. See Contributing to the Wiki for clone/edit/push instructions and the verification checklist.