Encoding Fixer

What it's for

Repair text that went through the wrong encoding, and save it as clean UTF-8. Garbled characters such as é for é or ’ for ’ (known as mojibake) appear when UTF-8 text is read as Windows-1252 or Latin-1. Files saved by older Windows programs, or exported as “ANSI” or UTF-16, are often turned down by imports that expect UTF-8.

For example, when a CSV exported from Excel shows Renée in another system, or a connector rejects a file because of its encoding, open the file here and save it as UTF-8, with a byte order mark (BOM) if the import wants one.

How to use it

Choose a file or drop it on the page, or paste text. For a file, File’s encoding is detected from its bytes: a BOM if it has one, UTF-8 if the bytes make valid UTF-8, UTF-16 if many alternate bytes are zero, and otherwise Windows-1252 (which also covers Latin-1). Choose it yourself if the guess is wrong. Pasted text is already decoded, so only the repairs and line endings apply to it.

• Repair garbled characters finds each run of characters that could be garbled UTF-8 and mends it in place, twice over if it was garbled twice. Runs that don’t turn back into valid UTF-8, such as real accented letters, are left alone.
• Line endings: keep them, or make them all LF or all CRLF. The message line says which kinds the text had.
• Start saved files with a UTF-8 BOM adds the three bytes some programs (Excel among them) look for.

The message line says what was found and done. Copy the result, or Save as a file, which adds -utf8 to the file’s name. Save keeps the line endings exactly; the Result box, and so Copy, shows them as plain line breaks. The settings, apart from the encoding, are remembered in this browser.

Example

Paste this:

Café, café and it’s, twice: café

The result is Café, café and it’s, twice: café, and the message says 3 garbled runs were repaired. The first “Café” was already right, so it’s left as it was.

Good to know

Detection is a guess from the bytes: a short Windows-1252 file with no accented characters is the same as UTF-8, which does no harm. Other legacy encodings, such as Shift JIS or KOI8-R, aren’t detected or offered. A garbled run directly next to a correct accented letter may not be repaired. The Character Inspector shows the code points and bytes, if you need to see what’s in the text.