4cee8e9d16
atob() does not handle Unicode well. The decoding routine creates a Unicode DOM string from the decoded bytes effectively treating the stream as ISO-8859-1. This breaks non-ASCII by decomposing characters into several meaningless "characters". Instead of decoding base64 in the browser, modify the server to return base64() BinaryResult as a JSON string if the only value of the Accept request header is application/json. RestApi client code currently always sets "Accept: application/json" in the request, as the browser JS strongly prefers JSON in the response body. Relying on the native JavaScript parser in the browser to read and decode the JSON string is smaller and more efficient than a pure JavaScript implementation of base64 decoding that supports UTF-8. Fix the Content-Type used during POST and PUT to document charset is UTF-8. The XMLHttpRequest standard requires browsers to always encode in UTF-8, however the server examines the request's Content-Type header to extract the charset attribute. Without setting the charset an arbitrary default encoding was chosen by the server, which may still mangle a commit message or file during save. Bug: issue 3093 Change-Id: I549037bfb82aff15bbffc8b4a01e8955ae0ce318 |
||
|---|---|---|
| .. | ||
| src | ||
| BUCK | ||