Fix a CSV in the table, not the text
Hand-editing raw CSV goes wrong in predictable ways: a comma inside a quoted field, a quote that never closes, a column that shifts by one. Editing in a parsed grid avoids the whole class of problem, because quoting and escaping happen at import and export, not while you type.
Paste the data or import a file, check the detected delimiter, and load the table. Every cell is directly editable. Column controls sit in each header: move a column left or right, or delete it. Rows delete individually, and new rows and columns append with one click.
Cleaning operations
Three operations cover most cleanup jobs, and each reports what it changed:
- Trim cells removes leading and trailing whitespace from every cell. Exports from web forms and copy-pasted spreadsheet ranges are the usual source of padded values like
ada@example.com. - Drop empty rows removes rows where every cell is blank, which appear when a spreadsheet exports its trailing empty lines.
- Remove duplicates keeps the first occurrence of each fully identical row. When a header row is enabled, the header is never counted or removed.
Order matters: trim first, then dedupe, or rows that differ only by whitespace survive as “different”.
Delimiters in and out
The parser detects comma, semicolon, tab, and pipe automatically by scoring which character produces the most consistent column counts, and you can override it before loading. The export delimiter is chosen separately, so the editor doubles as a converter: load a semicolon-separated file exported by a spreadsheet with European locale settings and download a comma-separated version, or produce a .tsv for tools that expect tabs.
Example: cleaning a signup export
The built-in example is a small signup list with the usual defects: a duplicated record, whitespace around an email address, and a fully empty row. Trim cells, then Drop empty rows, then Remove duplicates leaves four clean records, and the stats row confirms each step’s effect.
Size limits
The grid renders the first 500 rows; anything beyond that stays in memory and is still covered by cleaning operations and export. Very large files (hundreds of thousands of rows) work for clean-and-convert jobs, but a file that size is better edited in a tool built for it, since every keystroke in a giant grid re-renders the table.