Start with Garbled input text
Paste garbled text into Garbled input text. You can also use Upload to choose a .txt, .csv, .log, or .md file. Mojibake Fixer checks common UTF-8 decoding mistakes, lists the repair paths it can produce, and puts the selected result in Repaired output.
Repair garbled text encoding
- Paste text into Garbled input text, or use Upload for one of the four accepted file types.
- Read Suspected Repair Path & Candidates, including each candidate's corruption score.
- Keep the suggested option, select another candidate, or choose Keep original text (no repair).
- Check Repaired output. A repair candidate also opens Preview - What Changes.
- Use Copy or Download beside the output.
What the repair paths reverse
A decoding mistake can turn Café into text that appears as Café. One candidate can reverse that path:
Input: Café
Output: CaféThe page tests UTF-8 misread as Latin-1 (ISO-8859-1), UTF-8 misread as Windows-1252, and selected two-step paths for double-encoded text. Distinct results stay as separate candidates. Their corruption scores count byte-level mojibake patterns and replacement characters, so a lower score ranks a result without checking spelling, grammar, or meaning.
The lowest-scoring candidate gets the Suggested label and is selected only when its score is lower than the original. Selecting another radio option updates Repaired output immediately. The repair covers the full input, so inspect the output and preview before using Copy or Download.
When no mojibake repair candidate appears
The list can stay empty when the text is already clean or does not match a supported encoding-misinterpretation pattern. Unicode replacement characters can mean that the original bytes were lost before the text reached the page. Select Keep original text (no repair) to preserve the supplied input.
Preview - What Changes normally compares tokenized input and candidate text, then highlights changed candidate text. If the token arrays exceed 250000 pairs, the page shows an escaped preview instead, limited to 4000 characters, and adds [preview truncated] when it cuts the result.
Inspect common mojibake patterns
The input placeholder uses Café and ’ as examples of garbled text. The candidate label tells you if Mojibake Fixer reversed a Latin-1 or Windows-1252 interpretation, while the displayed path identifies the suspected sequence of repairs.
Double-encoded text can contain a longer sequence such as é. Compare the one-step and two-step candidates, then use Preview - What Changes to inspect the proposed character changes before you copy the result.
The tool applies a selected repair to the whole input. For a paragraph, a list, or one value, check the complete Repaired output before using Copy or Download.
Choose the right text transformation tool
Use Text Transformation Tools to choose another text operation. HTML Entity Encoder / Decoder handles named, decimal, or hexadecimal entity references. Unicode Normalizer handles NFC, NFD, NFKC, or NFKD. Find and Replace Text handles literal text and regular-expression matches.
The repair script contains no fetch, XMLHttpRequest, sendBeacon, or form submission for the input. Detection and repair use browser-side TextEncoder and TextDecoder logic. Repaired output is read-only, while Copy and Download read its current value.
Frequently asked questions
How do I fix mojibake in UTF-8 text?
Paste the text into Garbled input text. Review the suggested candidate and Preview - What Changes, then use Copy or Download beside Repaired output.
Can Mojibake Fixer repair double-encoded text?
It tests selected two-step paths for text misread as Latin-1 or Windows-1252 twice, plus mixed two-step paths shown in the candidate list.
Why does Mojibake Fixer show no repair candidate?
The text may already be clean, or it may contain no supported pattern. Replacement characters can also indicate that the original bytes were already lost.
Is my text sent to a server?
The repair script has no content upload request. Detection and repair use browser-side TextEncoder and TextDecoder logic.