Batch Text Format Cleaner Remove full/half-width mixing, invisible Unicode characters and AI copy-paste artifacts in one click
Step 1: Paste the text you want to clean
Step 2: Detection results
These numbers come from an exact, character-by-character scan — not a semantic guess. This is precisely what AI chat tools struggle with: they're great at understanding meaning, but notoriously unreliable at precise character-level counting.
Step 3: Choose which cleanups to apply
Detects fullwidth letters/digits/punctuation mixed into halfwidth text (or vice versa), and lets you pick a direction to normalize.
Removes zero-width spaces, zero-width joiners, BOM and other characters that are invisible to the eye but break search and layout.
Fixes formatting fingerprints commonly left behind when copying from AI chat tools.
Collapses repeated spaces and blank lines, and trims stray whitespace at the start/end of every line.
Why does pasted text sometimes look "broken" or behave strangely?
When you copy text from ChatGPT, Word, PDFs, or translated documents, it often carries characters you cannot see with your eyes but that computers still process — zero-width spaces, leftover byte-order-mark (BOM) bytes, or bidirectional control marks. These invisible characters can break search matching, mess up word counts, or even trigger false "duplicate content" flags once pasted into a CMS, spreadsheet, or code file. On top of that, AI chat tools tend to leave stylistic fingerprints behind: curly "smart" quotes, unified em dashes, and Markdown symbols (**bold**, # heading) that were meant for chat rendering, not your document.
This tool runs 100% locally in your browser. Nothing you paste is ever uploaded anywhere.
What this tool detects and cleans
- Full/half-width normalization: detects fullwidth Latin letters, digits and punctuation mixed into half-width text (or vice versa), and lets you convert everything to one consistent width.
- Invisible Unicode characters: finds and removes zero-width space (U+200B), zero-width joiner/non-joiner (U+200C/U+200D), BOM (U+FEFF), word joiner, soft hyphen, and directional override marks — and shows exactly how many of each were found.
- AI copy-paste artifacts: converts curly quotes to straight quotes, unifies em/en dashes, and strips leftover Markdown symbols.
- Whitespace cleanup: collapses repeated spaces and blank lines, and trims stray spaces at the start/end of every line.
Why not just ask an AI to clean it?
Large language models read text semantically — they are notoriously unreliable at precise, character-by-character counting and byte-level Unicode detection, a well-documented weakness of LLMs. This tool instead scans every character deterministically, so the counts you see are exact, not an AI's best guess.
FAQ
Q1: Does this tool upload my text anywhere?
No. All detection and cleaning happen locally in your browser — nothing is ever sent to a server, which makes it safe for text containing personal or confidential information.
Q2: Will full/half-width normalization mess up my Chinese/Japanese/Korean text?
No. Full/half-width normalization only affects the ASCII-range Latin letters, digits, and punctuation (U+FF01-U+FF5E). Characters that are naturally full-width, like Chinese, Japanese, or Korean text, are left untouched.