A CSV file can look simple and still create import problems. A space around a product name, a repeated record or an incorrectly handled comma can change what another application receives. Cleaning should therefore mean checking the structure and meaning of the data, not just pressing every available button.
This guide uses the ToolsOnDuty CSV Viewer, Editor & Cleaner to work through a small inventory example. The screenshots show the actual browser workspace and the resulting rows. Processing happens in the browser, and you must download your result before closing or reloading the tab.
Choose the delimiter and header setting first
Open CSV Viewer, Editor & Cleaner. You can paste text or open a UTF-8 CSV or TSV file. Leave Import delimiter on Auto-detect for ordinary files, or choose comma, tab or semicolon when you know the source format. A CSV extension does not prove that the separator is a comma.
Keep First row contains headers selected for the example below. If your file starts with actual data, turn that setting off before loading. Import settings affect the next load; they do not reinterpret a table that is already open. Loading another file or pasted dataset replaces the current table, so save useful work first.
Start with a reproducible inventory example
Paste the following text into Or paste delimited text. There are five data rows: a product with surrounding spaces, two identical Notebook records, one empty record and a product name containing a comma. Quotation marks around Cable, USB-C keep that comma inside one cell.
Click Load pasted data. The workspace should report five data rows and three columns. Check the Item, Qty and UnitPrice headers before editing. If the product name containing a comma becomes two cells, stop and correct the delimiter or quoting rather than cleaning an incorrectly parsed table.
Item,Qty,UnitPrice
" Desk lamp ",2,18.50
Notebook,10,3.20
Notebook,10,3.20
,,
"Cable, USB-C",4,6.75Clean in a deliberate order
Trimming before deduplication matters when two otherwise identical records differ only in outside spaces. However, two purchases of the same item may be legitimate records even when their visible values match. Decide what a duplicate means for your source before removing it; this tool does not infer your business rules.
- Click Trim whitespace. Desk lamp loses its surrounding spaces. This also trims other cells and headers, so avoid it when leading or trailing spaces are meaningful.
- Click Remove empty rows. The row containing only separators is removed, leaving four data rows.
- Click Remove duplicate rows. One identical Notebook record is removed, leaving three data rows.
- Read the status messages and inspect the remaining cells. Deduplication compares complete rows, not just an ID, email address or product name.
Check the result, then sort or filter
The final table should contain Desk lamp with quantity 2 and price 18.50, Notebook with quantity 10 and price 3.20, and Cable, USB-C with quantity 4 and price 6.75. Values remain text; the workspace is not calculating totals, evaluating formulas or deciding whether prices are correct.
Choose Qty under Sort column and use Sort ascending if you want a natural order. Filter rows searches cell text and changes the visible view, not the underlying dataset. Searching Notebook should show one of three rows. Cleaning and sorting still apply to the complete dataset, even when a filter is active. Larger tables are shown in pages of 25 rows, so the visible page is not the complete export.
Download all rows or only the filtered result
Set a useful Download filename and leave Include column headers enabled when the receiving application expects them. Download all CSV exports the complete cleaned table. Download filtered CSV exports the current matching rows across pagination, not merely what fits on screen. In our check, the full export contained three records, while the Notebook-filtered export contained only Notebook.
Keep Protect spreadsheet formula-like text enabled for files that will be opened in a spreadsheet. It can prefix risky text with an apostrophe so a spreadsheet is less likely to interpret it as a formula. The workspace itself does not execute formulas. That prefix can matter to a downstream database importer, so check the receiving system's expectations rather than disabling protection automatically.
Downloads use UTF-8 with a byte-order mark, the imported delimiter and CRLF line endings. Quoting protects commas within cells. Reopen the downloaded file in the intended application and verify the row count and column alignment before replacing an important source file.
Troubleshooting and practical limits
Keep an untouched source copy for consequential data. Browser-side processing reduces the need to send the dataset to a server, but your browser, extensions, downloaded files and screen-sharing habits still matter. Do not describe a cleaned CSV as audited or anonymous just because it was edited locally.
- Everything appears in one column: choose the source delimiter explicitly and load the original data again.
- Names split across columns: check quoted fields and embedded quotation marks in the original CSV.
- Accents look wrong: convert the source to UTF-8. An older legacy encoding is not automatically repaired.
- The downloaded file has extra rows: check whether you selected Download all CSV instead of Download filtered CSV.
- A file is rejected: the workspace limits input to 5 MiB, 10,000 data rows, 100 columns and 200,000 cells.
Try the related tools
Apply the steps from this guide directly in your browser.
Frequently asked questions
Does removing duplicates merge rows with the same product name?
No. It removes exact duplicate rows across the complete dataset. Two rows with the same product name but different quantities or prices remain separate.
Does filtering delete the rows that do not match?
No. Filtering changes the view. Download all CSV still includes the complete dataset; Download filtered CSV includes only the current matches.
Will the workspace preserve my spreadsheet formulas?
CSV stores text, not a workbook's formulas and formatting model. No formulas are executed here. Formula-like export protection may prefix risky text, so review the exported file for your receiving application.
Are my edits saved if I close the tab?
No. Download your cleaned data before leaving or reloading. Keep the original file separately so you can restart if needed.
