Mojibake Fixer

Encoding hypothesis comparison
Your pasted text and every candidate stay in this browser. Funnel mode keeps one validated record in this tab for up to 30 minutes and places only an opaque job ID in the URL. Nothing is sent to our servers or a repair API.

Paste the damaged-looking text

Paste text such as café or a Japanese string that became unreadable after an encoding mismatch. The original remains unchanged in the editor.

Maximum 50,000 characters. This tool compares pasted Unicode text only; it does not inspect or repair files, byte streams, databases, or HTTP headers.

Mojibake is readable data displayed as the wrong characters because bytes were decoded with an incompatible character set. A familiar example is café where café was encoded as UTF-8 but its bytes were interpreted as Latin-1 or Windows-1252. Paste the visible text to compare strictly reversible repair hypotheses. The tool keeps the original unchanged, labels every candidate as a hypothesis, and performs the comparison locally in your browser.

How to compare a mojibake repair hypothesis

  1. 1

    Paste the visible text

    Use the text exactly as received. The tool does not upload or inspect a source file.

  2. 2

    Opt in to the comparison

    Ask the browser to test Latin-1 and Windows-1252 byte interpretations against a strict UTF-8 decoder.

  3. 3

    Compare, do not assume

    Read the unchanged original beside every reversible candidate and choose only when context supports it.

  4. 4

    Copy plain text

    Copy, download, or print a selected hypothesis without overwriting the original.

What a mojibake repair can and cannot establish

An encoding maps characters to bytes and back. UTF-8 represents é with the two bytes C3 A9. If software decodes those bytes as Windows-1252 or ISO-8859-1, the visible result can become é. A useful repair hypothesis reverses that particular mistake: treat the visible legacy characters as byte values, decode those bytes strictly as UTF-8, and verify that encoding the candidate as UTF-8 produces the same bytes again.

That round-trip check rejects incomplete and malformed UTF-8 sequences. It also rejects any candidate containing the replacement character , because information represented by that character has already been lost. A valid round-trip is evidence for a hypothesis, not proof of which encoding the source system used or what the author meant.

Visible text Hypothesis tested Possible candidate What to verify
café UTF-8 bytes read as Latin-1 café Compare spelling with the source
It’s UTF-8 bytes read as Windows-1252 It’s Check punctuation and house style
文字化け UTF-8 bytes read as Latin-1 文字化け Check the Japanese text with a trusted copy
café Two encoding mistakes in sequence café Inspect both guarded passes

Why Japanese text needs context

The Japanese term 文字化け literally describes garbled characters. Japanese UTF-8 text can turn into a long mixture of Latin letters, symbols, and control-like characters when its bytes are decoded through the wrong legacy mapping. Reversibility is useful here, but a short name or isolated character can still be ambiguous. Compare the candidate with the original document, database export, sender, or another authoritative copy.

Limits and safer recovery

This tool receives Unicode text that the browser has already decoded. It cannot recover bytes discarded earlier, infer an encoding from a binary file, fix a database column, or change an HTTP Content-Type header. If no candidate appears, keep the original. Obtain the actual source bytes when possible, make a backup, identify the declared and real encodings, then decode once with a tool designed for files or byte streams. Never repeatedly save damaged text over the only source copy.

Frequently Asked Questions

No. It means that a specific legacy-byte interpretation decoded as valid UTF-8 and passed an exact byte round-trip. More than one interpretation may produce the same or different readable text. Context and an authoritative source are still required.

The visible characters may not represent UTF-8 bytes misread as ISO-8859-1 or Windows-1252, the sequence may be incomplete, or a replacement character may already have removed information. Keep the original and inspect the actual bytes and encoding metadata.

No. It compares pasted Unicode text only. It does not read files, detect a file encoding, connect to a database, rewrite stored values, or repair server headers. Back up the source before using a byte-aware conversion workflow.

No. The comparison, candidate selection, copying, TXT creation, and print preparation happen in your browser. Funnel mode can keep a validated record in this browser tab for up to 30 minutes and puts only an opaque job ID in the URL.

Related Tools