See the characters that look like nothing
Two values can look identical and still fail a lookup or duplicate check. A non-breaking space, a zero-width space or an initial byte-order mark can explain the difference. Inspect the original text by character name, code point, count and location, then review a highlighted copy with labeled markers. The output previews your chosen cleanup; the input stays unchanged until you choose Replace input with output. Ordinary letters, accents and emoji are kept. Processing runs locally in your browser.
Hidden characters in everyday data
Two values that look alike
Input
North Star North Star
Result
North Star North Star
The second value contains a non-breaking space, U+00A0, between the words. The default cleanup replaces it with an ordinary space so the values can match. Nothing else is changed.
Punctuation pasted from Word
Input
“Ready” — don’t wait…
Result
"Ready" - don't wait...
These are visible typographic look-alikes, not errors. Enable the ASCII punctuation option to replace curly quotes, the em dash and the ellipsis. Leave it off when you want the original typography.
A CSV with a byte-order mark and mixed endings
Input
name,code Avery,00123 Sam,04567
Result
name,code Avery,00123 Sam,04567
The initial BOM is removed by default. Enable line-ending normalization and choose LF to make the CRLF, LF and CR endings consistent. Postal codes keep their leading zeros.
Inspect first, then choose cleanup
- Paste text or open a UTF-8 file. To check a file's line endings, open the file. Pasted text shows the line endings your clipboard provides.
- Review the summary table. Counts cover the full input, with the first five locations per type. Show all expands beyond twelve types. Locations count Unicode code points from line 1, column 1.
- Use Show ordinary spaces and tabs to reveal them. Neutral markers are kept; colored markers show what your cleanup options will change.
- Choose cleanup options and inspect the cleaned output. Copy, download, send it to a next step, or explicitly replace the input after reviewing the changes.
Hidden character questions
Does cleanup remove accents, emoji or other normal text?
Cleanup keeps accents, emoji and normal text. Joiners and single attached variation selectors are always kept. Repeated or unattached selectors and misplaced emoji tags are reported. Replacing look-alike letters in mixed words requires its own option, off by default.
Are all findings mistakes?
No. Non-breaking spaces prevent wrapping, direction marks support right-to-left writing, and curly quotes are intentional typography. Inspect the context before removing them. Direction removal and ASCII punctuation conversion are off by default.
How are line endings handled?
Line endings are counted by type, with mixed endings flagged. Control cleanup replaces next line, vertical tab and form feed with the main existing line ending, or LF when none exists. Optional normalization converts all line endings to LF or CRLF.
Is the full input inspected?
Yes, within the 10 MB input limit. Summary counts and cleanup cover everything. Only the highlighted copy and stored example locations are capped. A single emoji sequence or a letter with combining accents can contain several code points, so visible glyph counts can differ from reported columns.
Targeted inspection, not a spelling checker
Only Cyrillic, Greek and Armenian letters mapped to a single ASCII letter in Unicode 18.0.0 confusables are checked, and only in words containing Latin letters. A whole word typed entirely in look-alike letters is not detected. Greek lowercase α, γ, ν, ρ, σ, ι, υ and ϱ in mixed words may be intentional scientific symbols: they are reported but never replaced. Greek capitals and omicron remain eligible for optional replacement. Replacement and object-replacement characters are report-only. The preview is capped at 16,000 code points, 1,200 lines or 2,000 markers; counts cover the full input. Typing or undo can normalize editor line endings.