CSV Cleaner
Trims cells, normalises headers, fixes number and date formats and reports rows with the wrong column count.
Finds what is wrong with a file before changing anything: stray spaces, Hangul stored as separate jamo, invisible characters, numbers kept as text, duplicate headers. Then fixes the ones that have a safe fix and reports the ones that do not, because an ambiguous date column needs a person rather than a guess.
Drop the file that needs cleaning
Read in this tab. Nothing is uploaded. Files from Korean, Japanese, Chinese and Western Excel (CP949, Shift_JIS, GBK, Windows-1252) are recognised.
How to use it
Read the diagnosis
Each finding names the column, how many cells are affected and an example value. The example is the fastest way to see whether it is really a problem.
Choose the fixes
Nine of them, each independent. Normalising Hangul to NFC and turning text numbers into numbers are the two that fix the most downstream breakage.
Check the counts
Every number is cells actually changed. A fix that reports zero really did nothing, which is how you tell a fix that was not needed from one that did not work.
Download the cleaned file
UTF-8 with a byte-order mark, which is what makes Korean Excel open it without turning 한글 into 뷁.
Common questions
What is decomposed Hangul?
macOS stores 한 as ㅎ + ㅏ + ㄴ rather than as one character. It looks identical on screen and is a different string, so it never matches a value typed on Windows. Normalising to NFC joins them back up.
Why will it not fix my dates?
A column of 03/04/2026 could be day-first or month-first, and the file does not say. Picking one would silently corrupt half the rows in the other reading. It is reported so you can decide.
What are the invisible characters?
Zero-width spaces, soft hyphens and control characters, which arrive with text pasted from a website or a PDF. They take up no space, break every comparison, and are impossible to find by looking.
Does it change my original file?
No. The cleaned table is a new download; the file on your disk is never written to.
Feed it from
PDF Table Extractor
Finds ruled tables from the lines drawn on each page — merged cells included — and borderless ones from aligned columns, then exports each as CSV or TSV. Values are as printed; column types are not guessed.
PDF Form Data Extractor
Reads the filled field values out of one or many form PDFs into a single table.
Unicode Normaliser
Applies NFC, NFD, NFKC or NFKD — the fix for Korean text from macOS that looks right but never matches.
CSV Viewer
Opens a large delimited file with type detection, column stats and sorting, without loading it all into memory.
Next
These accept what this tool produces, so the result can go straight in.
CSV Deduplicator
Removes duplicate rows on all or selected key columns, keeping the first, last or most complete row.
Column Statistics
Profiles each column with count, nulls, unique values, min, max, mean, median and distribution.
CSV Viewer
Opens a large delimited file with type detection, column stats and sorting, without loading it all into memory.
Encoding Detector & Converter
Guesses a file’s character encoding — CP949, Shift_JIS, GB18030, Windows-1252 and more — with a side-by-side preview of each candidate, then re-saves it as UTF-8 (with or without BOM), UTF-16 or EUC-KR/CP949.
Unicode Normaliser
Applies NFC, NFD, NFKC or NFKD — the fix for Korean text from macOS that looks right but never matches.
CSV Join
Joins two tables on a key column with inner, left, right or outer semantics and shows unmatched rows.