OpenCSV

A fast, local-first CSV & Excel editor — a single self-contained binary.

Open 100MB+ files in a virtual grid, sort/filter/query them like a spreadsheet, run SQL against your data, and export anywhere. No cloud, no install, no telemetry. An open-source alternative to SmoothCSV.

---

Why OpenCSV?

• Local-first. Your data never leaves your machine. The whole app is one Go binary that embeds the web UI.
• Built for big files. A virtual grid renders only what's on screen, and huge files switch to a windowed mode so browser memory stays bounded (a 1M-row / 90MB file keeps the tab at ~9MB of heap).
• Spreadsheet ergonomics + SQL power. Excel-style per-column sort/filter dropdowns and a SQL console that queries your CSV as a table.

Features

• 📂 Open CSV / TSV / TXT / DAT (delimiter & encoding auto-detected: UTF-8/BOM, GBK, GB18030, UTF-16, Latin-1) and import/export Excel (.xlsx)
• ⚡ Virtual grid that stays smooth on millions of rows; inline cell editing, column rename, column resize
• 🔼 Sort — multi-key, type-aware (text / number / date / length); per-column buttons and a global dialog stay in sync
• 🔎 Filter — Excel-style per-column dropdown (filter by values with counts, or by condition) unified with a global builder and an editable WHERE bar. 20 operators incl. contains, starts/ends with, like, between, is one of, is empty, …
• 🔁 Find & Replace with regex, case sensitivity, and per-column scope
• 🧮 SQL console — query your data as a table named data (powered by an in-process SQLite engine, cached per session)
• ✏️ Editing — undo/redo, copy/paste (multi-cell, tab-separated), insert/delete rows & columns
• 🧰 Transforms — case conversion, trim, merge/split columns, deduplicate, transpose
• 📊 Aggregate stats on a selection (sum / avg / min / max / count / unique)
• 📤 Export — Excel, CSV, JSON, Markdown, HTML, SQL, LaTeX
• 🎨 Multi-tab UI, command palette (⌘P), light/dark/system theme, context menus, keyboard shortcuts

Tech stack

Layer	Tech
Backend	Go 1.25 · Gin · excelize · modernc.org/sqlite (pure Go, no CGO)
Frontend	Vue 3 · TypeScript · Pinia · Vite · lucide-vue-next
Packaging	Frontend builds into backend/dist/, embedded via //go:embed → one standalone executable

Quick start

Prerequisites

• Go 1.25+
• Node 20+

Run in development (two terminals)

# Terminal 1 — backend API on :7070 (does not serve the UI in dev)
cd backend
DEV=1 go run .

# Terminal 2 — Vite dev server on :5173 (proxies /api and /ws to :7070)
cd frontend
npm install
npm run dev

Open http://localhost:5173.

Build the production binary

cd frontend && npm run build      # outputs to ../backend/dist/
cd ../backend && go build -o opencsv .
./opencsv                         # serves UI + API on :7070, opens your browser

PORT=8080 ./opencsv overrides the port.

Type-check the frontend

cd frontend && npx vue-tsc --noEmit

Project structure

opencsv-cc/
├── backend/            # Go HTTP server + all business logic
│   ├── main.go         # Gin router, //go:embed dist, browser launch
│   ├── handlers/       # HTTP handlers (file, data, export, sql, upload)
│   ├── services/       # csv, excel, session, sql, transform, watcher
│   └── models/         # request/response + session models
├── frontend/           # Vue 3 SPA (builds into backend/dist/)
│   └── src/
│       ├── components/grid/VirtualGrid.vue   # core spreadsheet component
│       ├── components/dialogs|panels/        # sort/filter/find/sql UI
│       ├── stores/     # Pinia: tabs, settings, history
│       └── api/        # axios clients
├── scripts/gen_bench.py   # deterministic benchmark dataset generator
├── ARCHITECTURE.md        # detailed architecture notes (中文)
└── PRD.md                 # product requirements (中文)

The architecture and PRD docs are currently written in Chinese — translations are very welcome (see Contributing).

Performance

Measured on a 1,067,371-row, 90 MB real-world dataset (UCI Online Retail II):

Operation	Time
Open file (90 MB)	~280 ms
Filter (Country is one of …)	~17 ms
Filter (Description LIKE '%…%')	~240 ms
Sort (date / number, 1M rows)	~0.3 s
SQL query (repeat, cached table)	~80 ms
Browser tab heap (windowed mode)	~9 MB

Need test data? python3 scripts/gen_bench.py regenerates benchmark CSVs deterministically (they're intentionally not committed).

Contributing

Contributions are welcome — code, docs, translations, bug reports, and ideas.

1. Fork & clone, then get the dev servers running (see Quick start).
2. Create a branch off main: git checkout -b feat/my-change.
3. Make your change. Good first areas:
   • Adding a data operation: handler in backend/handlers/data.go → service in backend/services/ → route in backend/main.go → API call in frontend/src/api/data.ts → wire it into the relevant component / command palette.
   • Adding an export format: backend/handlers/export.go + a route, then call it from the frontend.
   • Translating ARCHITECTURE.md / PRD.md to English.
4. Check it builds: go build ./... (backend) and npx vue-tsc --noEmit (frontend).
5. Open a PR with a clear description and, where it helps, a screenshot or before/after numbers.

Conventions:

• Frontend uses Vue 3 <script setup lang="ts">. No CSS frameworks — theming is via CSS custom properties in frontend/src/style.css.
• Backend stays pure Go (no CGO) — keep SQLite on modernc.org/sqlite.
• Commit messages follow a type: summary style (feat:, fix:, perf:, chore:, docs:).

See CLAUDE.md for a deeper guide to conventions and the codebase layout.

License

Licensed under the Apache License 2.0. By contributing, you agree that your contributions will be licensed under the same terms.