Client-Side CSV — Privacy Without a Backend

Most “CSV tools in the browser” still upload your file to someone else’s server. That is convenient for the vendor. It is a privacy risk for anyone working with payroll, healthcare exports or customer data.

CSV View takes the opposite bet: the file never leaves the machine. No upload, no account, no backend for the core workflow — and still open, filter, edit and export multi-million-row sheets without freezing the UI.

Privacy is an architecture decision

If the product promise is “your data stays local”, you cannot bolt that on later. The architecture has to refuse the easy path:

• Parse on the client
• Persist session state locally (IndexedDB)
• Export from the client
• Keep domain logic free of server round-trips

Zero KB on the network is not a marketing line — it is the acceptance test.

Keep the main thread free

Million-row CSVs will punish a naive main-thread parse. The approach that worked:

1. Streaming parse in a Web Worker — chunks around 1 MB so the UI stays interactive while rows accumulate.
2. Virtualized table rendering — only mount what is visible; scrolling must not allocate the whole grid.
3. Pure services — parsing, stats, filters, diff and export live outside Vue so they are unit-testable and movable.

Session without a server

IndexedDB stores file content and UI session (search, filters, sort, column layout). Closing the tab should not mean losing an hour of exploration.

That same local-first model makes undo/redo coherent: one chronological stack for cell edits and column moves, without syncing a remote document model.

What you gain — and what you trade

You gain privacy by default, offline-friendly workflows and a deploy surface that is essentially static hosting. You trade away multi-device sync and collaborative editing — features that need a backend on purpose.

For a tool whose primary job is “open a sensitive file and understand it”, that trade is the product.