Clean dates and text with an ordered, reviewable pipeline

Remove repetitive text noise while keeping every original value available.

Open toolFiles stay in your browser. No account needed.

What the cleaner does

Apply explicit text and calendar-date rules to one selected column. The result is a source-row audit report of original and final values, not a silent rewrite of the workbook. Changed rows and row errors have separate audit sheets; source rows are never dropped.

Build the pipeline

Select a local file, sheet and header; choose a target column. Add, remove and reorder up to 16 rules: trim, whitespace normalization, remove prefix/suffix/version suffix, literal split/join, extract first date, normalize date, keep/remove by length, or literal replace. Arguments are literal text, not code. Validate before running, inspect a source row step by step, then export or clear.

Precise text rules

Whitespace normalization replaces runs with a space; add trim separately to remove outer spaces. Version removal recognizes a terminal -v1 or _v1.2, up to four numeric parts. Split creates at most 256 parts, join combines those parts, and an unjoined final list exports as JSON text. Length counts Unicode code points, not bytes or grapheme clusters. Length filters mark the value blank; literal replace treats dollar signs and backslashes literally.

Dates require evidence

ISO YYYY-MM-DD is supported. Day/month dates using slash, dot or hyphen require MDY or DMY when ambiguous: 03/04/2026 blocks until you choose. Extract first date uses the first recognized token, even if it is invalid; it does not skip a bad date to find a later valid one. Excel serial candidates retain the 1900/1904 epoch. Fictitious serial 60 in 1900 and fractional-day serials are rejected; there is no timezone conversion or automatic text-to-number step.

Synthetic example

The downloadable fixture contains " SYN-00123-v1 " and " SYN-中🙂-v2 ". Apply trim, remove prefix SYN-, then remove version suffix. Expected final values are 00123 and 中🙂. Changing the order can change the result; the source-row step preview shows before, after, raw date substring, parsed date and status. No real enterprise identifiers are used.

Privacy and safe audit export

The Worker retains data only for this session, with no upload, analytics or browser database. Clear/Cancel terminates it and releases download URLs. XLSX includes Data, Changed, Errors, Summary and Run_Settings. Formula-like strings remain inert; source formulas are never evaluated. CSV is Data only and loses audit sheets and reliable types. Non-ambiguity row errors retain the original value instead of a partially transformed value.

Limitations and frequently asked questions

Can I use a regular expression? No: advanced regex is backlog, not a disguised literal feature. Can I clean several columns at once? This version audits one selected column per run. Does an error remove a record? No. Blank, empty and zero remain distinct where the input format preserves them. Shared approximately 10k default / 100k experimental advanced limits and output budgets still apply; long expansion is blocked. Physical Safari and 16 GiB performance validation are pending.