Mojibake Detector and Fixer
Find garbled spans consistent with UTF-8 read as Windows-1252 or Latin-1. Review recovered bytes, select repairs and save the text as UTF-8.
Trace One and Two Passes of UTF-8 Repair
Mojibake Detector and Fixer: how to use it
- Paste garbled text or open a local Unicode text file. Choose the earlier decoder if it is known and set the maximum repair passes.
- Select Analyze encoding. Compare the original spans, proposed text and recovered bytes. Clear a candidate’s checkbox to keep that span as entered.
- Use optional normalization or cleanup where appropriate. Copy the repaired preview, save the UTF-8 text or download the JSON evidence.
Limits and notes
Up to 100,000 input UTF-16 units and 400,000 output units after normalization. Finds UTF-8 sequences that can be reconstructed from Windows-1252 or ISO-8859-1 display text, with up to two passes. Position and byte evidence describe candidates. Review the words when more than one interpretation is plausible.
How your privacy works →Mojibake Detector and Fixer: common questions
How is a candidate detected?
The tool reverses the selected legacy decoder for a span, then checks whether those bytes form a strict UTF-8 sequence. It shows the proposed character and the bytes used.
Can one span need two passes?
Yes. A saved copy can contain text that was misread twice. Two passes let adjacent candidates form a second valid UTF-8 sequence, such as the two steps from é to é.
Can it recover a replacement character?
U+FFFD signals that a previous decoder may already have discarded the original bytes. Its location is reported and the character stays in the preview.
How do I keep the original Unicode form and line endings?
Keep source form and leave optional cleanup unchecked. The UTF-8 output retains the source separators outside the selected candidate spans.