How to Read %E3%81%82, \u3042, and \343: Why Japanese Turns into Codes in URLs, JSON, and Git

Encoded text isn’t mojibake; it’s another way of writing Japanese

Mojibake is what you get when text is read as different characters because the encoding was mistaken. By contrast, %E3%81%82 and \u3042 are deliberate ASCII replacements, used to write Japanese where only ASCII is allowed. Once you know the notation, you can return to the Japanese without losing a single character.

Broadly, there are two approaches: writing a character’s number (its Unicode code point), or writing its bytes (the numbers it’s stored as in UTF-8 or another encoding). The hiragana “a” has the Unicode number U+3042 (12354 in decimal) and the UTF-8 bytes E3 81 82. Remember these two numbers and you can quickly tell which notation you’re looking at.

Chart of nine notations

We checked these examples by entering them into “Read escapes” in the Mojibake Fixer & Decoder. The Base64 example decodes to the Japanese word for “mojibake”, and the others to 文字 (Punycode to 日本語.jp). If text mixes several notations, the tool finds and reads each one.

%E3%81%82 in URLs: what you get when copying a Japanese URL

URLs allow only ASCII letters, digits, and some symbols, so Japanese is written as UTF-8 bytes, each as “%” plus two hex digits. A Japanese character is 3 bytes in UTF-8, so three %XX make one character. The 4-character Japanese word for “mojibake” becomes “%E6%96%87%E5%AD%97%E5%8C%96%E3%81%91”, which is 36 characters.

• Even if the browser’s address bar shows Japanese, copying and pasting into a chat or email gives the %XX form
• Sequences starting with %E3 are usually hiragana or katakana, and those starting with %E4–%E9 are usually kanji
• If you see many %82 or %83 and two %XX seem to make one character, it may be an old URL written in Shift_JIS
• “+” can represent a space (in URLs submitted from search forms, for example)

When sending a URL to someone, the %XX form is the safer choice. A URL turned back into Japanese can have its link cut off partway in some apps.

\u3042 in JSON, \343 in Git, \xe3 in Python

In programming, Japanese is often written in escaped form instead of being output directly. Each case below has a setting or method for displaying Japanese.

• \u3042 in JSON: a Unicode number written in 4 hex digits. For JSON output in Python, adding ensure_ascii=False keeps the Japanese as is. Emoji are written as a pair, as in “🍣” (🍣)
• \343\201\202 in Git: a Japanese filename shown in octal by git status and similar commands. Setting git config --global core.quotepath false displays it in Japanese
• \xe3\x81\x82 in Python: a bytes object displayed as is. Convert it to a string with .decode('utf-8') and you get Japanese
• あ and あ in HTML: a Unicode number written in decimal or hex. Browsers show the hiragana “a”, but in source or RSS viewed as text you see the code

Domains starting with xn-- are internationalized domains

Japanese domains like 日本語.jp are converted to ASCII because of how DNS works. The conversion method is Punycode, and converted names carry the prefix “xn--”. “xn--wgv71a119e.jp” is 日本語.jp.

To prevent impersonation using lookalike characters, browsers sometimes deliberately display Japanese domains in the xn-- form. For email or links from an unfamiliar xn-- domain, it’s safest to convert it back and confirm that it really is the domain you expect.

If real mojibake like “縺” or “ã” is mixed in rather than codes, check the pattern in the Mojibake Patterns Chart. The =?UTF-8?B?…?= in email subjects belongs to the same family, and it’s explained in Garbled Email Subjects.

FAQ

How do I turn %E3%81%82 in a URL back into Japanese?

Collect the two hex digits after each % as one byte and read them as UTF-8. %E3%81%82 is the hiragana “a”. In Japanese, three %XX make one character. Paste it into “Read escapes” in the Mojibake Fixer & Decoder to convert everything at once.

What is \u3042 in JSON?

It’s a Unicode escape, a Unicode number written in 4 hex digits; \u3042 is the hiragana “a”. JSON output is often configured to write Japanese this way. Emoji use a pair to represent one character, as in 🍣.

Git shows Japanese filenames like \343\201\202. Why?

Git is displaying the UTF-8 bytes of the Japanese filename in octal. Set git config --global core.quotepath false, and git status and similar commands will display them in Japanese.

Are URLs starting with xn-- dangerous?

xn-- simply marks an internationalized domain, such as a Japanese one, converted to ASCII. It isn’t dangerous by itself. But it’s also used for impersonation with lookalike characters, so converting it back and checking the domain name is the safe move.