MIME type parsing

[annevk]

See http://www.rfc-editor.org/errata_search.php?rfc=7231&eid=4031 (rejected due to scope). Clients typically (or all?) allow text/html; despite it being invalid.

[annevk]

One of the things that I couldn't find addressed anywhere is something like text/html;charset=gbk;charset=utf-8. The introduction of https://tools.ietf.org/html/rfc2046 mentions parameters as a set, thereby potentially outlawing duplicate names, but there's nothing normative to that effect anywhere. (Browsers are evenly split on how they handle this case by the way.)

[mnot]

See also whatwg/mimesniff#30 e.g., discussion of multiple content-types. This has potential security impact.

[annevk]

For MIME types I ended up (re)writing https://mimesniff.spec.whatwg.org/#understanding-mime-types btw and also have a bunch of tests at https://github.com/w3c/web-platform-tests/tree/master/mimesniff/mime-types.

[mnot]

Discussed in Basel; inclination is to change the ABNF, and survey similar places to align if necessary.

[snuggs]

@mnot pardon my ignorance but what is ABNF? Am I wrong to assume it is this ?

Thanks in advance!

[mnot]

It is.

[mnot]

Note that this is about the trailing semicolon, at least initially.

[mnot]

Discussed on editors' call; we should create a new ABNF rule parameters in 5.4.1.4. Parameters that allows semicolons without following parameter, and use that where appropriate.

[reschke]

Just to check (grep in httpbis.abnf):

expectation = token [ "=" ( token / quoted-string ) *parameter ]

media-range = ( "*/*" / ( type "/*" ) / ( type "/" subtype ) ) *( OWS ";" OWS parameter )
media-type = type "/" subtype *( OWS ";" OWS parameter )

parameter = parameter-name "=" parameter-value
parameter-name = token
parameter-value = ( token / quoted-string )

transfer-coding = token *( OWS ";" OWS transfer-parameter )
transfer-parameter = token BWS "=" BWS ( token / quoted-string )

So "expectation" needs a fix (see #420). So "Expect" would be the only affected field in addition to "Content-Type". (Because "TE" is special). Are we ok with that?

That said, do we want:

parameters = *( OWS ";" OWS [ parameter ] )

or

parameters = *( OWS ";" OWS parameter ) [ OWS ";" OWS ]

?