CSV Cleaner Online

What this tool does

This tool applies a small set of well-defined cleaning operations — removing duplicate or empty rows, trimming whitespace, renaming or selecting columns, filtering rows and replacing values — as an ordered, visible pipeline. Every step you add stays listed, can be disabled without losing its settings, and can be removed. The “Cleaned” tab always shows the exact result of the current pipeline; the “Original” tab always shows your file untouched, so you can compare the two at any time.

Nothing happens automatically

There is no “auto-clean” button and no step is ever added for you. If you also use the Data Quality Analyzer on the same file, its findings are not read or applied by this tool — you decide which, if any, of its observations are worth acting on, and you build the corresponding step yourself. This is a deliberate limit: a tool that silently reinterprets a heuristic finding as a required fix could make a wrong assumption about your data.

Available transformations

• Remove duplicate rows (identical in every column)
• Remove completely empty rows
• Trim leading/trailing whitespace in chosen text columns
• Normalize empty values: turn a blank or whitespace-only text value into empty
• Rename a column
• Select or remove columns, and reorder the ones you keep
• Filter rows by one column and condition
• Replace an exact, whole-cell text value

Add as many steps as you need; each one runs on the result of the steps before it, in the order you added them. A step that would no longer make sense — for example trimming a column that an earlier step already removed — is rejected with an explanation instead of being silently skipped or silently corrupting the result.

How it works

Each step becomes part of one SQL query, run by DuckDB as WebAssembly in your browser, in a background Web Worker. The “Cleaned” view is a live query over your original file, not a saved copy, so disabling or removing a step takes effect immediately. Exporting runs that same query once more and writes the result as a CSV file entirely inside the engine.

Browser-local processing

Your file is not uploaded. The WebAssembly engine and this page's own code are served from this site, and the browser is told to refuse requests to any other site while the tool runs. See the privacy page for what that does and does not guarantee.

Exporting

“Export cleaned CSV” downloads the full result of the current pipeline, with any search and sort applied in the Cleaned tab — not just the current page. As with every export on this site, a text value that would be read as a formula by a spreadsheet program (starting with =, +, -, @, a tab or a carriage return) is prefixed with a quote character so it opens as plain text instead.

Large files

This tool warns you above roughly 100 MB, warns more strongly above roughly 500 MB, and declines to open a file above roughly 2 GB rather than risk crashing the tab. Every transformation still runs on the whole file, not just the preview page, so a heavier pipeline on a large file can take a noticeable moment to update.