Mojibake Patterns

Python

When a UTF-8 string was incorrectly decoded as Latin-1:

# Re-encode as Latin-1, then decode as UTF-8
fixed = garbled.encode('latin-1').decode('utf-8')

# Or using ftfy (Fix Text For You)
import ftfy
fixed = ftfy.fix_text(garbled)

PHP

Latin-1 bytes that should be UTF-8:

// PHP 8.1+ (mb_convert_encoding)
$fixed = mb_convert_encoding($garbled, 'UTF-8', 'ISO-8859-1');

// Older PHP
$fixed = utf8_encode($garbled);

MySQL

Table stored as Latin-1 but contains UTF-8 bytes:

-- Step 1: change column charset without conversion
ALTER TABLE t MODIFY col BLOB;
-- Step 2: convert BLOB back to utf8mb4
ALTER TABLE t MODIFY col TEXT CHARACTER SET utf8mb4;

-- Or convert entire table
ALTER TABLE t CONVERT TO CHARACTER SET utf8mb4;

JavaScript / Node.js

Reading a buffer with the wrong encoding:

// Specify encoding when reading files
const text = fs.readFileSync('file.txt', 'utf8');

// Convert buffer manually
const decoder = new TextDecoder('utf-8');
const text = decoder.decode(buffer);

// If bytes are actually Windows-1252:
const decoder = new TextDecoder('windows-1252');
const text = decoder.decode(buffer);

Look at the bytes

Use a hex editor or xxd / hexdump to see the raw bytes. UTF-8 multi-byte sequences start with 0xC2–0xF4. Bytes 0x80–0xBF are continuation bytes. If you see standalone 0x80–0xFF, it's likely a single-byte encoding.

Identify the pattern

The signature é always means UTF-8 read as Latin-1. The signature ’ means UTF-8 read as Windows-1252. Lots of ? or � means bytes that don't fit the encoding at all.

Use detection libraries

chardet (Python) and uchardet (C/CLI) use statistical analysis to guess the encoding from byte frequency patterns. They work best on longer texts — short strings may produce wrong guesses. Use the Decode Bytes tool on this site to try different decodings interactively.