Mojibake Fixer (Repair Garbled Text)

Mojibake is the jumble you get when text is decoded with the wrong character set: “é” instead of “é”, “’” instead of an apostrophe, “한글” or “Çѱ۔ instead of 한글, or “縺薙ｓ縺ｫ縺｡縺ｯ” instead of こんにちは. The original bytes are usually still there, just shown through the wrong table. This tool turns the characters back into those bytes with the wrong table and decodes them with the right one, for every common pair of encodings, and shows the reading that looks most like real text.

How to use this tool

1. Paste the garbled text.
2. Leave “Automatically” selected, or pick the pair if you know what happened (for example a UTF-8 file opened as Windows-1252).
3. For short Korean, Japanese or Chinese text, choose the original language to break ties.
4. Copy the repaired text, and check the table if the top reading is not right.

Formula

Repaired text = decode(right charset, encode(wrong charset, garbled text))

The encode tables are built in your browser by decoding every byte, or every lead and trail byte pair, with the browser’s own decoder and inverting the result, so they match exactly what browsers do. Each candidate is scored for how natural it looks: rare symbols, control characters, the replacement character �, a switch of script between two neighbouring letters (Ã followed by ©) and a capital letter in the middle of a lower-case word all add to the score. When UTF-8 is the target, only byte runs that form valid UTF-8 are changed, so a correct é next to a garbled dash survives.

Example

The sample Café menu – crème brûlée, it’s €4.50 is UTF-8 shown as Windows-1252, the most common case on the web. The repair gives Café menu – crème brûlée, it’s €4.50. Double damage such as ’ is undone in two rounds and becomes ’.