Tools

CSV Cleaner

Clean quote-aware CSV in your browser. Remove blank or exact duplicate records, trim selected cells, check uneven rows and preview every change before download.

This CSV cleaner parses the file as a table before applying only the operations you choose. For example, it keeps “Hong Kong, China” in one quoted cell, can remove a fully repeated contact row and reports a row with six cells when the header has five. Cleaning always starts again from the original table, so changing an option does not stack edits on an earlier result.

How to use this tool

Choose a CSV or delimited-text file up to 10,000,000 bytes, or paste up to 1,000,000 UTF-8 bytes. BOM-marked UTF-8, UTF-16LE and UTF-16BE are detected; Windows-1252, Big5 and GB18030 require your explicit selection.
Confirm the comma, tab, semicolon or pipe delimiter and whether the first record is a header. Ambiguous delimiter detection, malformed quotes, NUL data and uneven records are shown instead of being silently guessed.
Review the default removal of fully empty records, then choose any trimming, empty-column removal, header repair or exact deduplication rules. The preview lists affected record numbers and before-and-after counts.
Choose UTF-8 BOM and CRLF or LF output, review any formula-like cells, then download the reparsed CSV or TSV. Keep the source until the new file has been checked in its destination system.

Ways to use it

Clean a CRM or form export: Remove blank records and exact repeated contacts before re-importing a list, while keeping quoted names, addresses and notes together.
Check a supplier or inventory file: Find records with fewer or more cells than the header before a database or ecommerce import rejects the file.
Deduplicate with chosen key columns: Compare a complete row or selected fields, keep the first or last match, and decide whether comparison ignores case or surrounding whitespace.
Prepare a consistent UTF-8 copy: Read a supported source encoding and save CSV or TSV with an explicit delimiter, line ending and UTF-8 BOM choice.

Supported files and limits

The file limit is 10,000,000 bytes and pasted text is limited to 1,000,000 UTF-8 bytes. A table may contain at most 100,000 records including its header, 250 columns, 1,000,000 cells and 1,000,000 decoded characters in one field.
The preview shows at most 100 rows and 30 columns while reporting the full counts. Generated output is limited to 50,000,000 UTF-8 bytes; preview and Copy stop at 1,000,000 bytes, although a larger valid result can still be downloaded.
Spaces are part of CSV data. Trimming data or headers is optional and changes the values; removing empty columns, repairing headers and rectangularising uneven rows also require an explicit choice.
Exact deduplication is not identity matching. It cannot determine that differently written names, addresses, phone numbers or companies refer to the same record. Empty selected keys are ignored by default.
Strict rectangular shape is the default. Rectangularising pads short rows and extends the table to its widest row without truncating cells; it does not infer missing values.
Formula-like cells are preserved by default and may be interpreted when the file opens in spreadsheet software. Prefixing a single quote is optional, changes the cell value and is not a universal spreadsheet-safety guarantee.
A valid cleaned table can still contain incorrect, outdated or sensitive information. The tool does not validate email addresses, business rules, dates, totals or whether a record should be kept.
Processing runs in this browser without sending the selected file, pasted content, filename, headers or output to a cleaning server. Browser extensions, clipboard history, screen sharing and downloaded copies remain outside the tool.

Common questions

Does the cleaner support commas and line breaks inside a field?

Yes, when the selected delimiter and quoting are valid. Double-quoted fields can contain delimiters, doubled quotes and CRLF, LF or CR line breaks. Malformed quotes produce an error instead of a guessed table.

How are duplicate CSV records detected?

You can compare complete rows or selected key columns, keep the first or last match, and optionally ignore surrounding whitespace or letter case for comparison only. The tool reports exact matches under those rules; it does not perform fuzzy identity matching.

Will trimming change my CSV data?

Yes. CSV spaces are data, so cell and header trimming are optional and recomputed from the original table. Review the affected records before downloading.

What happens when rows have different column counts?

Strict shape is selected by default, so uneven records are reported with their logical record numbers. If you choose rectangularise, short rows are padded and the table extends to the widest row without deleting any extra cells.

Can the cleaner repair blank or duplicate headers?

Only when you select header repair. It can trim header text, fill blanks as collision-safe column_N names and make repeated names unique with suffixes. The preview shows the resulting header before download.

Can a cleaned CSV still contain spreadsheet formulas?

Yes. Formula-like cells are counted and warned about, but their exact text is kept by default. An optional single-quote prefix changes those values and may reduce risk in some spreadsheet workflows, but it is not a guarantee for every application.

Does the tool upload or store my CSV?

No. Parsing, cleaning and serialization run in the page worker in this browser. The workflow does not send or store the file, filename, cell values, headers, error excerpts or output; other browser software and copies you make are outside its control.

Why can I download a result that is too large to preview or copy?

Preview and Copy stop at 1,000,000 UTF-8 bytes to keep the page responsive. A valid result up to 50,000,000 bytes remains downloadable after it has been serialized and reparsed successfully.

Book a ConsultationConsult WhatsApp