CSV Cleaner + Change Log avatar

CSV Cleaner + Change Log

Pricing

Pay per usage

Go to Apify Store
CSV Cleaner + Change Log

CSV Cleaner + Change Log

Free CSV/TSV cleaner by TORIX (AI-assisted). Trims whitespace, converts date columns to ISO YYYY-MM-DD, normalizes number columns, optional Title Case, drops empty rows/columns and duplicates. Outputs cleaned CSV, a before/after change log and an issues list. Review results before use.

Pricing

Pay per usage

Rating

0.0

(0)

Developer

Torix

Torix

Maintained by Community

Actor stats

0

Bookmarked

2

Total users

1

Monthly active users

12 hours ago

Last modified

Categories

Features

Limited permissions

Share

CSV Cleaner: One Date Format, Numbers, Duplicates + Change Log

Turn a messy CSV into a clean one and see exactly what was changed. Paste the CSV or give a public link; you get:

  • Cleaned rows (Dataset - export as CSV, Excel or JSON with the Apify Export button),
  • CLEANED_CSV - the cleaned file with the same delimiter,
  • CHANGE_LOG_CSV / CHANGE_LOG - a before/after list: record number, column, rule, old value, new value,
  • ISSUES - values that could not be cleaned safely and were left unchanged (impossible dates, text in number columns),
  • SUMMARY - counts: input rows, output rows, cell changes, removed empty rows and duplicates.

Free test version. The Actor does not charge per result at the moment.

How to run it in 3 steps

  1. Add your CSV. Paste it into CSV text (the form is prefilled with made-up sample data) or put a direct public link in Public link to a CSV file.
  2. Press Start. The prefilled sample finishes in a few seconds.
  3. Open the Output tab. Check Change log and Issues first, then export the cleaned rows.

What it does

RuleExample (before -> after)
Trim and collapse whitespace" anna KOVAL " -> "anna KOVAL"
Dates to ISO YYYY-MM-DD03/11/2026 -> 2026-11-03 (day first by default, switch with dateOrder); 5 Nov 2026, 2026/11/3, 03.11.26 also work
Numbers stored as numbers1 200,50, 1,200.50, 1.200,50, $12, (12.50) -> 1200.5, 1200.5, 1200.5, 12, -12.5
Title Case for chosen columnsanna KOVAL -> Anna Koval
Remove empty rows and columns;; rows and columns with no values
Remove duplicates after cleaninganna KOVAL;03/11/2026;1 200,50 and Anna Koval;2026-11-03;1200.5 become the same row, the second is removed and logged
Header repairempty names become column_3, repeated names become name_2

Every removed row and every changed value is in the change log. Record numbers refer to the input file (header = record 1), so you can find the original line.

What it will NOT do (on purpose)

  • No guessing. A date that is impossible in the chosen order (31.02.2026, or 11/25/2026 when day-first) is reported in Issues and left as it was. Text in a number column is reported, not deleted.
  • No fuzzy matching. Only rows that are identical after cleaning are removed. Typos such as Kowal vs Kovalski are not merged.
  • Identifier columns are protected. Columns named like phone, zip, postal, id, sku, code, iban, nip, pesel are never turned into numbers automatically (leading zeros are kept). Values like 007 are never converted.
  • Ambiguous numbers: a lone 1,200 is read as 1200, a lone 1.200 as 1.2. If your file uses the other convention, set Decimal separator.

Input

{
"csvText": "name;date;amount\n anna KOVAL ;03/11/2026;\"1 200,50\"\nAnna Koval;2026-11-03;1200.5\nOleh P.;31.02.2026;abc\n;;\nJan Nowak;15.11.2026;99,9",
"titleCaseColumns": ["name"]
}
FieldMeaning
csvText / csvUrlPasted CSV or a public direct link (http/https). Not both.
delimiterauto (default), comma, semicolon, tab, pipe
dateOrderDMY (default) or MDY
decimalSeparatorauto (default), comma, dot
titleCaseColumnsColumns written in Title Case
dateColumns, numberColumnsOptional: force the type of a column (otherwise detected when at least 60 % of the values fit)
removeDuplicates, dedupeKeyRemove duplicate rows; optionally compare only some columns (e.g. email)
removeEmptyRows, removeEmptyColumnsBoth on by default
maxRowsSafety limit, default 100 000

Output

Result of the sample above:

name;date;amount
Anna Koval;2026-11-03;1200.5
Oleh P.;31.02.2026;abc
Jan Nowak;2026-11-15;99.9

CHANGE_LOG_CSV (excerpt; rows in the real file are ordered by cleaning step, and whitespace trims are listed too):

record,column,rule,before,after
2,date,normalize_date,03/11/2026,2026-11-03
2,amount,normalize_number,"1 200,50",1200.5
2,name,title_case,anna KOVAL,Anna Koval
3,,remove_duplicate,Anna Koval | 2026-11-03 | 1200.5,removed - duplicate of record 2
5,,remove_empty_row,,removed - empty row

ISSUES:

[
{ "record": 4, "column": "date", "value": "31.02.2026", "problem": "Not a valid date: not a real calendar date. Left unchanged." },
{ "record": 4, "column": "amount", "value": "abc", "problem": "Not a number (or an identifier with leading zeros). Left unchanged." }
]

Excel users

The Actor reads CSV/TSV text, not .xlsx files. In Excel use Save as -> CSV (UTF-8), then paste the text or upload the file to any place that gives a public direct link. After the run, use Export -> Excel on the Dataset to get an .xlsx back.

Limitations

  • CSV/TSV only (no .xlsx, no PDF, no scanned documents, no OCR).
  • Public links only: links that need a login (private OneDrive, Google Drive, SharePoint) do not work. No cookies, no accounts, no secrets are used or needed.
  • File size up to 20 MB. Encoding: UTF-8, or windows-1250 when the bytes are not valid UTF-8 (a note is shown).
  • Dates with a time part (2026-11-03 14:30) are not converted and are reported in Issues.
  • English month names only. Two-digit years: 00-69 = 2000s, 70-99 = 1900s.
  • Check your data before you rely on it: the Actor applies fixed rules and shows every change, it does not know what your data means.

Tips

  • Run once, read the Change log and Issues, then adjust dateOrder, decimalSeparator or titleCaseColumns and run again.
  • For a customer list, set dedupeKey to email to find the same person written in different ways.

FAQ

Is it free? Yes, this is a free test version. If a paid version appears later, the price will be shown on this page before you use it.

Who builds it? TORIX, a small project that builds and maintains Actors with AI assistance. Feedback and problems are welcome in the Issues tab.

Does it keep my data? The input and results live only in your Apify run storage under your account.