Struct Case 

pub struct Case {
    pub name: &'static str,
    pub section: &'static str,
    pub title: &'static str,
    pub expect: Expect,
    pub bytes: &'static [u8],
}

One message from the corpus.

Fields

name: &'static str

The RFC’s own name for the message, e.g. ipv6-good. RFC 5118 labels each message with this name (“Message Details: ipv6-good”), and the archive’s file names match, so the name is the link from a fixture back to the prose that describes it.

section: &'static str

The section of RFC 5118 that describes it, e.g. 4.1. Two sections describe two messages each, so this is not unique across cases.

title: &'static str

That section’s title, verbatim.

expect: Expect

Which layer must object, and how.

bytes: &'static [u8]

The message exactly as Appendix A’s archive holds it — LF-terminated, and for the §4.10 pair without a terminating blank line. Use Case::wire to get bytes a SIP parser is meant to see.

Implementations

impl Case

pub fn wire(&self) -> Bytes

The archive bytes turned into on-the-wire SIP.

Three transformations, all forced by the archive rather than chosen:

1. LF becomes CRLF. RFC 3261 §7 terminates every start line and header field with CRLF. The archive holds none, so without this every message is one unterminated header line and the corpus would measure nothing but that fact.

2. The header section is terminated if the archive left it open, which it does for §4.10’s two files. An unterminated header section is indistinguishable from a truncated message, so a parser is right to refuse it — see how RFC 4475’s baddn is classified in crate::rfc4475. Refusing it here would test the archive’s formatting, not sipx’s IPv6 handling.

3. Content-Length is set to the actual body length. The three messages that carry SDP declare a length that matches neither the LF-terminated body in the archive nor the CRLF-terminated one:

   case	declared	archive body (LF)	wire body (CRLF)
   §4.6 ipv6-in-sdp	268	242	251
   §4.8 mult-ip-in-sdp	181	180	189
   §4.9 ipv4-mapped-ipv6	236	236	245

   No single convention reconciles those, so the declared values are simply wrong — and RFC 5118 has one verified erratum, on §4.3’s wording, which does not mention them. A parser is right to refuse §4.6 as truncated and right to discard the tail of the other two, but doing so here would measure the RFC’s arithmetic instead of sipx’s IPv6 handling, and would cut §4.8’s and §4.9’s SDP off mid-body before sipx-sdp ever saw the c= lines the sections exist to exercise.

   Framing is covered thoroughly by RFC 4475, which has cases built for it (clerr, ncl, mcl01) and correct lengths everywhere else.

None of the three touches an octet inside an IPv6 reference, a URI, or a body — only line terminators and the digits of one header value. What the corpus is for survives intact, and wire_changes_only_terminators_and_content_length holds that claim.

pub fn lossy(&self) -> Cow<'_, str>

The message as a lossy string, for assertion messages.

pub fn is_classified(&self) -> bool

Whether this case is asserted on by the parser tests. Every RFC 5118 case is: unlike RFC 4475’s archive, this one carries no file that no section references.

pub fn has_sdp(&self) -> bool

Whether the message carries an SDP body, and so belongs to the sipx-sdp half of the harness as well as the sipx-sip half.

Trait Implementations

impl Clone for Case

fn clone(&self) -> Case

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Case

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Copy for Case