I have an application with a browser page that allows upload of CSV to Google Drive via V2 of the api. It is only handling ASCII (I get "failed to execute btoa / string to be encoded contains characters outside the Latin1 range").

I've updated my code so that the content-type is now "text/csv;charset=UTF-8;". I am still using content-transfer-encoding of base64.

In multiple places on the web, I found recommendations of using btoa(unescape(encodeURIComponent(data))). Previously, the code was only doing btoa(data). It seems to work, though I have not tested it exhaustively.

I understand that unescape is considered in a precarious light relative to ECMA-262. Here are the questions:

• a) does the above work for all of UTF-8 (including 4-byte chars)?
• b) if not, will it just fail to encode properly, or will I get a browser error?
• c) are there any other risks of using this approach (even if it handles 4-byte chars)?
• d) is there a better / newer way of doing this? I probably have the luxury of not supporting older browsers.