Incorrect header encoding conversion

[821938089]

        private static String fixHeaderEncoding(String val) {
            byte[] bytes = val.getBytes(ISO_8859_1);
            if (!looksLikeUtf8(bytes))
                return val;
            return new String(bytes, UTF_8);
        }

This encoding conversion is wrong, you cannot restore the original binary content from a string without knowing its encoding.
Such conversion leads to loss of some characters.

Related references: https://stackoverflow.com/a/39308860

By the way: when will the next version be released?

[jhy]

Can you give me the code for this vs screenshots so that I can review?

[821938089]

Is this it? search.php?search=我的

[jhy]

OK, I've moved the re-encoding fix-up to only response headers. That's in place to fix #706 where the header was encoded as 8559 but held UTF bytes instead. Browsers seem to do this fix up too so the solution seems necessary. We do need better tests for this - I wasn't able to get Jetty to emit the header incorrectly so can't directly add a test case.

For request headers, the value set by the user is now retained directly. When making the request, Java will encode the header as UTF-8. Servers will probably expect 8559 and so this may or not work. Per spec, the content should either be limited to 8559 content or encoded with RFC 2047. We don't attempt to automatically do that (and some servers will be OK). A bit of a grey area here. Happy for other suggestions.

[821938089]

This header comes from the server's response. Is there a way to fix it?

[jhy]

Can you give me a sample URL or code so that I can actually review the server's response properly?

[821938089]

https://www.zhenshezw.com/

Enter "我的" and click on the search icon.

[821938089]

Hi, if you can't reproduce the issue could you add a configuration option to skip the fixHeaderEncoding?

[jhy]

I get caught in bot detections when I try this. Can you provide sample code so that I can try and repro?

I won't add a configuration option unless I can validate it. You could always fork the code yourself, of course.