Full-width to Half-width: Fixing Numbers and Code

When you actually need it

Numbers are the classic problem. Full-width １２３ looks like 123 but sits at different code points, so a spreadsheet treats it as text, a script fails to parse it, and a search does not match. Phone numbers, amounts, and product codes copied from Japanese sources often arrive in this state.

Code and command lines have the same issue. A full-width space, bracket, or semicolon pasted into source is a syntax error that is very hard to spot. Tidying English and symbols copied from old documents, PDFs, or chat threads calls for the same fix.

• Numbers a spreadsheet stores as text
• IDs and amounts copied from Japanese documents
• Full-width spaces and brackets inside code

What the mapping covers

The mapping is a code point range: the full-width ASCII variants from U+FF01 to U+FF5E are shifted down by 0xFEE0, and the ideographic space U+3000 becomes an ordinary space. Chinese characters, kana, and half-width katakana are outside the range and stay as they are.

Full-width punctuation inside that range changes too: the full-width comma U+FF0C, full stop U+FF0E, exclamation and question marks, and parentheses all become ASCII. Japanese 、 and 。 sit at U+3001 and U+3002 and are unaffected. A full-width space turned into a normal space also removes first-line indentation.

• Full-width ASCII shifts down by 0xFEE0
• U+3000 becomes an ordinary space
• 、 and 。 are outside the range

Alternatives and when not to use it

If only the numbers matter, find-and-replace or VALUE() in a spreadsheet is enough; if the goal is consistent typing, switching the input method is faster. Japanese body text usually wants half-width digits and letters with full-width punctuation, so converting a whole passage overshoots.

For the opposite requirement, forms that insist on full-width input, use the half-width to full-width converter. Both directions are mechanical code point mappings that ignore context, so keep a backup before touching code, passwords, or URLs.