Why é turns into two characters
In UTF-8, every accented letter takes two bytes – é is C3 and A9. A program that reads the file as Windows-1252 turns each byte into a character of its own: C3 becomes Ã, A9 becomes ©. That’s why nearly all broken accents start with à – the second character tells you which letter was meant.
Before
Café crème at the résumé workshop – señor Müller’s naïve idea.
After
Café crème at the résumé workshop – señor Müller’s naïve idea.
The most common patterns
é→é,è→è,ê→êá→á,ó→ó,ú→ú,ñ→ñç→ç,ü→ü,ö→öÉ→É,Ç→Ç€→€,’→’,“→“– more on broken apostrophes and quotes
The full table is further down. If the garbage looks longer, like é, the text is double-encoded – Mojibuster fixes that too.
Better not to replace by hand
You could find-and-replace é with é. That only works as long as you know every error: à followed by a non-breaking space is a broken à, but a lone à can also be a real à in a Portuguese name. Mojibuster checks every spot byte by byte and only changes what is clearly broken.
Where the problem comes from
- Excel opens a UTF-8 CSV without a BOM as Windows-1252 – here’s how to open it correctly.
- Databases deliver UTF-8 over a
latin1connection – see MySQL and phpMyAdmin. - Scripts read files with the wrong encoding, for example PowerShell or Python with pandas.
- E-mails declare the wrong character set – help for Outlook.
- Editors open the file as ANSI – in Notepad++ one click usually fixes it.
Common broken characters
| Broken | Correct |
|---|---|
| é | é |
| è | è |
| Ã | à |
| ç | ç |
| ñ | ñ |
| ü | ü |
| ö | ö |
| ’ | ’ |
| “ | “ |
| – | – |
| — | — |
| … | … |
| € | € |
| ° | ° |
Frequently asked questions
Are my characters lost?
No. é still holds exactly the bytes of é, and Mojibuster turns them back. Characters are only lost once a program has replaced them with ? or � – more on question marks instead of characters.
Why do I see à followed by a space?
That’s usually a broken à: its second byte is a non-breaking space in Windows-1252. If copying turned it into a normal space, the spot is no longer unambiguous, and Mojibuster leaves it alone.
Can I repair a whole file?
Yes. Drop CSV, TXT, JSON or SQL files up to 100 MB into the box above. The file stays on your device.