\-escaping in quoted strings

[mnot@…]

quoted-strings support \-escaping, so that <"> and "\" can appear in the final string. This process is not explicitly defined, however (although it is defined in RFC 5322, it isn't referenced from HTTP).

From testing Content-Disposition, it appears that some implementations will do as RFC 5322 suggests, while others will replace the character with _ or -, or stop processing the string.

This needs to be explicitly defined for HTTP.

Additionally, the range of characters that can be \-escaped should be considered; it's only useful for the two characters above, and we already discourage senders from using others.

[mnot@…]

Prague editors' meeting: need to test both clients and servers to see if / how they do escaping.

[julian.reschke@…]

From testing Content-Disposition, it appears that some implementations will do as RFC 5322 suggests, while others will replace the character with _ or -, or stop processing the string.

This is somewhat misleading. What happens is that those that do not unescape will handle the "\" as character not allowed in filenames, and substitute it.

[mnot@…]

That might make things more straightforward.

Currently, the relevant text is:

The backslash octet ("\") can be used as a single-octet quoting mechanism within quoted-string constructs:

quoted-pair = "\" ( WSP / VCHAR / obs-text )

Senders SHOULD NOT escape octets that do not require escaping (i.e., other than DQUOTE and the backslash octet).

I'd propose:

The quoted-string construct uses the backslash octet ("\") as a signle-octet quoting mechanism:

quoted-pair = "\" ( WSP / VCHAR / obs-text )

Receivers that process the value of the quoted-string MUST handle a quoted-pair as if it were replaced by the octet following the backslash.

Senders SHOULD NOT escape octets that do not require escaping (i.e., other than DQUOTE and the backslash octet).

[julian.reschke@…]

5322 link: ​http://tools.ietf.org/html/rfc5322#section-3.2.1

[julian.reschke@…]

From [1361]:

define \-escaping in quoted strings for characters other than \ and DQUOTE (see #270)