Bugfix: wrong use of header "Encoding"

[laggardkernel]

Wrong use of experimental header "Encoding". Seems a bug.

class Request:
    # added by Reitz, 10/14/18
    # 36c57393180a03e19ca65424cf6d1bcb808b904e
    @property
    async def declared_encoding(self):
        if "Encoding" in self.headers:
            return self.headers["Encoding"]

class Response:
    @property
    async def body(self):
        if self._stream is not None:
            return (self._stream(), {})

        if self.content is not None:
            headers = {}
            content = self.content
            if self.mimetype is not None:
                headers["Content-Type"] = self.mimetype
            if self.mimetype == "text/plain" and self.encoding is not None:
                # added by taoufik07, 2/20/19
                # 0b261252e1b8865ecc4667dd35674199fce796eb
                headers["Encoding"] = self.encoding
                content = content.encode(self.encoding)
            return (content, headers)

Experimental header "Encoding" has a very different format compared what we thought, after I checked the RFCs,

• https://tools.ietf.org/html/rfc1505
• https://tools.ietf.org/html/rfc1154

Here's some examples of this experimental headers.

Encoding: 107 Text
Encoding: 458 uuencode LZW tar (Unix binary object)
Encoding: 496 Text, 8 Text Signature

---

Another possible reason of the intro of "Encoding" by Reitz, may be that he'd like users set encoding in the "endpoint function" or the "before_request" function like this, Request.headers["Encoding"]="utf-8". But the chance is slim for this assumption cause we could set encoding on Request.encoding directly.