Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
1 out of 2 committers have signed the CLA.

✅ luka-mimi
❌ luka

Atob will produce garbled characters when processing Base64 encoding of Chinese or other non-ASCII characters because it was originally designed to support only Latin1 (ISO-8859-1) encoding, that is, each character only occupies one byte (0~255).

Chinese characters are multi-byte encodings (such as UTF-8, which usually occupies 2~4 bytes). When you directly decode the Base64 encoded string with atob, a single-byte decoded string is returned. If these bytes are directly interpreted as UTF-16 strings (JavaScript's internal string encoding), garbled characters will result.

'data:' + mimeType + ';base64,' + binaryData

`data:${mimeType};charset=utf-8;base64,${binaryData}`;

This happens because when decoding Base64 data, the browser needs to know which character encoding (like UTF-8) to use to interpret the text content—especially for non-ASCII characters like Chinese. If the character set isn't explicitly specified (charset=utf-8), the default behavior may lead to garbled text (mojibake).

✅ Why does the second version not cause garbled text?

`data:${mimeType};charset=utf-8;base64,${binaryData}`

This includes charset=utf-8, which tells the browser or parser:

After decoding the Base64 data, interpret it as UTF-8 encoded text.

Chinese characters like "你好" are multi-byte in UTF-8. If you don’t explicitly specify the charset, some environments might default to another encoding (like ISO-8859-1), which causes incorrect interpretation.

❌ Why does the first version cause garbled text?

'data:' + mimeType + ';base64,' + binaryData

This omits the charset, so the browser guesses the encoding based on the mimeType:

• For text/plain, it might default to ISO-8859-1 or something else.
• Since Chinese characters are multi-byte in UTF-8, but ISO-8859-1 interprets bytes as single characters, you get garbled output.

Hey @luka-mimi,

Thanks for the PR, We have created "GHC-2405" as the internal reference to get this reviewed.

One of us will be in touch if there are any changes needed, in most cases this is normally within a couple of weeks but it depends on the current workload of the team.