Deuxfleurs/eml-codec
15
1
Fork 0
Code Issues 10 Pull requests Projects Releases 2 Packages Wiki Activity

fix mailbox tests

Browse source
This commit is contained in:
Quentin 2023-07-19 15:28:17 +02:00
parent 3c8d851ce6
commit fd0794fe6e
Signed by: quentin
GPG key ID: E9602264D639FF68
5 changed files with 163 additions and 167 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

228
src/rfc5322/mailbox.rs
View file

@ -113,7 +113,7 @@ pub fn addr_spec(input: &[u8]) -> IResult<&[u8], AddrSpec> {
obs_local_part, obs_local_part,
tag(&[ascii::AT]), tag(&[ascii::AT]),
obs_domain, obs_domain,
many0(pair(tag(&[ascii::AT]), obs_domain)), // for compatibility reasons many0(pair(tag(&[ascii::AT]), obs_domain)), // for compatibility reasons with ENRON
)), )),
|(local_part, _, domain, _)| AddrSpec { local_part, domain }, |(local_part, _, domain, _)| AddrSpec { local_part, domain },
)(input) )(input)
@ -175,7 +175,10 @@ impl<'a> Domain<'a> {
pub fn to_string(&self) -> String { pub fn to_string(&self) -> String {
match self { match self {
Domain::Atoms(v) => v.iter().map(|v| encoding_rs::UTF_8.decode_without_bom_handling(v).0.to_string()).collect::<Vec<String>>().join("."), Domain::Atoms(v) => v.iter().map(|v| encoding_rs::UTF_8.decode_without_bom_handling(v).0.to_string()).collect::<Vec<String>>().join("."),
Domain::Litteral(v) => v.iter().map(|v| encoding_rs::UTF_8.decode_without_bom_handling(v).0.to_string()).collect::<Vec<String>>().join(" "), Domain::Litteral(v) => {
let inner = v.iter().map(|v| encoding_rs::UTF_8.decode_without_bom_handling(v).0.to_string()).collect::<Vec<String>>().join(" ");
format!("[{}]", inner)
}
} }
} }
} }
@ -235,39 +238,28 @@ pub fn is_dtext(c: u8) -> bool {
#[cfg(test)] #[cfg(test)]
mod tests { mod tests {
use super::*; use super::*;
use crate::text::quoted::QuotedString;
#[test] #[test]
fn test_addr_spec() { fn test_addr_spec() {
assert_eq!( assert_eq!(
addr_spec("alice@example.com"), addr_spec(b"alice@example.com"),
Ok(( Ok((
"", &b""[..],
AddrSpec { AddrSpec {
local_part: "alice".into(), local_part: LocalPart(vec![LocalPartToken::Word(Word::Atom(&b"alice"[..]))]),
domain: "example.com".into() domain: Domain::Atoms(vec![&b"example"[..], &b"com"[..]]),
} }
)) ))
); );
assert_eq!( assert_eq!(
addr_spec("jsmith@[192.168.2.1]"), addr_spec(b"jsmith@[192.168.2.1]").unwrap().1.to_string(),
Ok(( "jsmith@[192.168.2.1]".to_string(),
"",
AddrSpec {
local_part: "jsmith".into(),
domain: "192.168.2.1".into()
}
))
); );
assert_eq!( assert_eq!(
addr_spec("jsmith@[IPv6:2001:db8::1]"), addr_spec(b"jsmith@[IPv6:2001:db8::1]").unwrap().1.to_string(),
Ok(( "jsmith@[IPv6:2001:db8::1]".to_string(),
"",
AddrSpec {
local_part: "jsmith".into(),
domain: "IPv6:2001:db8::1".into()
}
))
); );
// UTF-8 // UTF-8
@ -285,52 +277,42 @@ mod tests {
// ASCII Edge cases // ASCII Edge cases
assert_eq!( assert_eq!(
addr_spec("user+mailbox/department=shipping@example.com"), addr_spec(b"user+mailbox/department=shipping@example.com").unwrap().1.to_string(),
"user+mailbox/department=shipping@example.com".to_string(),
);
assert_eq!(
addr_spec(b"!#$%&'*+-/=?^_`.{|}~@example.com").unwrap().1.to_string(),
"!#$%&'*+-/=?^_`.{|}~@example.com".to_string(),
);
assert_eq!(
addr_spec(r#""Abc@def"@example.com"#.as_bytes()),
Ok(( Ok((
"", &b""[..],
AddrSpec { AddrSpec {
local_part: "user+mailbox/department=shipping".into(), local_part: LocalPart(vec![LocalPartToken::Word(Word::Quoted(QuotedString(vec![b"Abc@def"])))]),
domain: "example.com".into() domain: Domain::Atoms(vec![&b"example"[..], &b"com"[..]]),
} }
)) ))
); );
assert_eq!( assert_eq!(
addr_spec("!#$%&'*+-/=?^_`.{|}~@example.com"), addr_spec(r#""Fred\ Bloggs"@example.com"#.as_bytes()),
Ok(( Ok((
"", &b""[..],
AddrSpec { AddrSpec {
local_part: "!#$%&'*+-/=?^_`.{|}~".into(), local_part: LocalPart(vec![LocalPartToken::Word(Word::Quoted(QuotedString(vec![b"Fred", b" ", b"Bloggs"])))]),
domain: "example.com".into() domain: Domain::Atoms(vec![&b"example"[..], &b"com"[..]]),
} }
)) ))
); );
assert_eq!( assert_eq!(
addr_spec(r#""Abc@def"@example.com"#), addr_spec(r#""Joe.\\Blow"@example.com"#.as_bytes()),
Ok(( Ok((
"", &b""[..],
AddrSpec { AddrSpec {
local_part: "Abc@def".into(), local_part: LocalPart(vec![LocalPartToken::Word(Word::Quoted(QuotedString(vec![b"Joe.", &[ascii::BACKSLASH], b"Blow"])))]),
domain: "example.com".into() domain: Domain::Atoms(vec![&b"example"[..], &b"com"[..]]),
}
))
);
assert_eq!(
addr_spec(r#""Fred\ Bloggs"@example.com"#),
Ok((
"",
AddrSpec {
local_part: "Fred Bloggs".into(),
domain: "example.com".into()
}
))
);
assert_eq!(
addr_spec(r#""Joe.\\Blow"@example.com"#),
Ok((
"",
AddrSpec {
local_part: r#"Joe.\Blow"#.into(),
domain: "example.com".into()
} }
)) ))
); );
@ -339,84 +321,98 @@ mod tests {
#[test] #[test]
fn test_mailbox() { fn test_mailbox() {
assert_eq!( assert_eq!(
mailbox(r#""Joe Q. Public" <john.q.public@example.com>"#), mailbox(r#""Joe Q. Public" <john.q.public@example.com>"#.as_bytes()),
Ok(( Ok((
"", &b""[..],
MailboxRef { MailboxRef {
name: Some("Joe Q. Public".into()), name: Some(Phrase(vec![Word::Quoted(QuotedString(vec![&b"Joe"[..], &[ascii::SP], &b"Q."[..], &[ascii::SP], &b"Public"[..]]))])),
addrspec: AddrSpec { addrspec: AddrSpec {
local_part: "john.q.public".into(), local_part: LocalPart(vec![
domain: "example.com".into(), LocalPartToken::Word(Word::Atom(&b"john"[..])),
LocalPartToken::Dot,
LocalPartToken::Word(Word::Atom(&b"q"[..])),
LocalPartToken::Dot,
LocalPartToken::Word(Word::Atom(&b"public"[..])),
]),
domain: Domain::Atoms(vec![&b"example"[..], &b"com"[..]]),
} }
} }
)) ))
); );
assert_eq!( assert_eq!(
mailbox(r#"Mary Smith <mary@x.test>"#), mailbox(r#"Mary Smith <mary@x.test>"#.as_bytes()),
Ok(( Ok((
"", &b""[..],
MailboxRef { MailboxRef {
name: Some("Mary Smith".into()), name: Some(Phrase(vec![Word::Atom(&b"Mary"[..]), Word::Atom(&b"Smith"[..])])),
addrspec: AddrSpec { addrspec: AddrSpec {
local_part: "mary".into(), local_part: LocalPart(vec![LocalPartToken::Word(Word::Atom(&b"mary"[..]))]),
domain: "x.test".into(), domain: Domain::Atoms(vec![&b"x"[..], &b"test"[..]]),
} }
} }
)) ))
); );
assert_eq!( assert_eq!(
mailbox(r#"jdoe@example.org"#), mailbox(r#"jdoe@example.org"#.as_bytes()),
Ok(( Ok((
"", &b""[..],
MailboxRef { MailboxRef {
name: None, name: None,
addrspec: AddrSpec { addrspec: AddrSpec {
local_part: "jdoe".into(), local_part: LocalPart(vec![LocalPartToken::Word(Word::Atom(&b"jdoe"[..]))]),
domain: "example.org".into(), domain: Domain::Atoms(vec![&b"example"[..], &b"org"[..]]),
} }
} }
)) ))
); );
assert_eq!( assert_eq!(
mailbox(r#"Who? <one@y.test>"#), mailbox(r#"Who? <one@y.test>"#.as_bytes()),
Ok(( Ok((
"", &b""[..],
MailboxRef { MailboxRef {
name: Some("Who?".into()), name: Some(Phrase(vec![Word::Atom(&b"Who?"[..])])),
addrspec: AddrSpec { addrspec: AddrSpec {
local_part: "one".into(), local_part: LocalPart(vec![LocalPartToken::Word(Word::Atom(&b"one"[..]))]),
domain: "y.test".into(), domain: Domain::Atoms(vec![&b"y"[..], &b"test"[..]]),
} }
} }
)) ))
); );
assert_eq!( assert_eq!(
mailbox(r#"<boss@nil.test>"#), mailbox(r#"<boss@nil.test>"#.as_bytes()),
Ok(( Ok((
"", &b""[..],
MailboxRef { MailboxRef {
name: None, name: None,
addrspec: AddrSpec { addrspec: AddrSpec {
local_part: "boss".into(), local_part: LocalPart(vec![LocalPartToken::Word(Word::Atom(&b"boss"[..]))]),
domain: "nil.test".into(), domain: Domain::Atoms(vec![&b"nil"[..], &b"test"[..]]),
} }
} }
)) ))
); );
assert_eq!( assert_eq!(
mailbox(r#""Giant; \"Big\" Box" <sysservices@example.net>"#), mailbox(r#""Giant; \"Big\" Box" <sysservices@example.net>"#.as_bytes()),
Ok(( Ok((
"", &b""[..],
MailboxRef { MailboxRef {
name: Some(r#"Giant; "Big" Box"#.into()), name: Some(Phrase(vec![Word::Quoted(QuotedString(vec![
&b"Giant;"[..],
&[ascii::SP],
&[ascii::DQUOTE],
&b"Big"[..],
&[ascii::DQUOTE],
&[ascii::SP],
&b"Box"[..]
]))])),
addrspec: AddrSpec { addrspec: AddrSpec {
local_part: "sysservices".into(), local_part: LocalPart(vec![LocalPartToken::Word(Word::Atom(&b"sysservices"[..]))]),
domain: "example.net".into(), domain: Domain::Atoms(vec![&b"example"[..], &b"net"[..]]),
} }
} }
)) ))
@ -433,17 +429,20 @@ mod tests {
@33+4.com,,,, @33+4.com,,,,
,,,, ,,,,
(again) (again)
@example.com,@yep.com,@a,@b,,,@c"# @example.com,@yep.com,@a,@b,,,@c"#.as_bytes()
), ),
Ok(( Ok((
"", &b""[..],
vec![ vec![
"33+4.com".into(), None,
"example.com".into(), Some(Domain::Atoms(vec![&b"33+4"[..], &b"com"[..]])),
"yep.com".into(), None, None, None, None, None, None, None,
"a".into(), Some(Domain::Atoms(vec![&b"example"[..], &b"com"[..]])),
"b".into(), Some(Domain::Atoms(vec![&b"yep"[..], &b"com"[..]])),
"c".into() Some(Domain::Atoms(vec![&b"a"[..]])),
Some(Domain::Atoms(vec![&b"b"[..]])),
None, None,
Some(Domain::Atoms(vec![&b"c"[..]])),
] ]
)) ))
); );
@ -452,12 +451,17 @@ mod tests {
#[test] #[test]
fn test_enron1() { fn test_enron1() {
assert_eq!( assert_eq!(
addr_spec("a..howard@enron.com"), addr_spec("a..howard@enron.com".as_bytes()),
Ok(( Ok((
"", &b""[..],
AddrSpec { AddrSpec {
local_part: "a..howard".into(), local_part: LocalPart(vec![
domain: "enron.com".into(), LocalPartToken::Word(Word::Atom(&b"a"[..])),
LocalPartToken::Dot,
LocalPartToken::Dot,
LocalPartToken::Word(Word::Atom(&b"howard"[..])),
]),
domain: Domain::Atoms(vec![&b"enron"[..], &b"com"[..]]),
} }
)) ))
); );
@ -466,12 +470,15 @@ mod tests {
#[test] #[test]
fn test_enron2() { fn test_enron2() {
assert_eq!( assert_eq!(
addr_spec(".nelson@enron.com"), addr_spec(".nelson@enron.com".as_bytes()),
Ok(( Ok((
"", &b""[..],
AddrSpec { AddrSpec {
local_part: ".nelson".into(), local_part: LocalPart(vec![
domain: "enron.com".into(), LocalPartToken::Dot,
LocalPartToken::Word(Word::Atom(&b"nelson"[..])),
]),
domain: Domain::Atoms(vec![&b"enron"[..], &b"com"[..]]),
} }
)) ))
); );
@ -480,12 +487,17 @@ mod tests {
#[test] #[test]
fn test_enron3() { fn test_enron3() {
assert_eq!( assert_eq!(
addr_spec("ecn2760.conf.@enron.com"), addr_spec("ecn2760.conf.@enron.com".as_bytes()),
Ok(( Ok((
"", &b""[..],
AddrSpec { AddrSpec {
local_part: "ecn2760.conf.".into(), local_part: LocalPart(vec![
domain: "enron.com".into(), LocalPartToken::Word(Word::Atom(&b"ecn2760"[..])),
LocalPartToken::Dot,
LocalPartToken::Word(Word::Atom(&b"conf"[..])),
LocalPartToken::Dot,
]),
domain: Domain::Atoms(vec![&b"enron"[..], &b"com"[..]]),
} }
)) ))
); );
@ -494,14 +506,18 @@ mod tests {
#[test] #[test]
fn test_enron4() { fn test_enron4() {
assert_eq!( assert_eq!(
mailbox(r#"<"mark_kopinski/intl/acim/americancentury"@americancentury.com@enron.com>"#), mailbox(r#"<"mark_kopinski/intl/acim/americancentury"@americancentury.com@enron.com>"#.as_bytes()),
Ok(( Ok((
"", &b""[..],
MailboxRef { MailboxRef {
name: None, name: None,
addrspec: AddrSpec { addrspec: AddrSpec {
local_part: "mark_kopinski/intl/acim/americancentury".into(), local_part: LocalPart(vec![
domain: "americancentury.com".into(), LocalPartToken::Word(Word::Quoted(QuotedString(vec![
&b"mark_kopinski/intl/acim/americancentury"[..],
])))
]),
domain: Domain::Atoms(vec![&b"americancentury"[..], &b"com"[..]]),
} }
} }
)) ))

42
src/text/buffer.rs
View file

@ -1,42 +0,0 @@
#[derive(Debug, PartialEq, Default)]
pub struct Text<'a> {
parts: Vec<&'a [u8]>,
}
impl<'a> Text<'a> {
pub fn push(&mut self, e: &'a [u8]) {
self.parts.push(e)
}
pub fn to_string(&self) -> String {
let enc = encoding_rs::UTF_8;
let size = self.parts.iter().fold(0, |acc, v| acc + v.len());
self.parts.iter().fold(
String::with_capacity(size),
|mut acc, v| {
let (content, _) = enc.decode_without_bom_handling(v);
acc.push_str(content.as_ref());
acc
},
)
}
}
#[cfg(test)]
mod tests {
use super::*;
use crate::text::ascii;
#[test]
fn test_text() {
let mut text = Text::default();
text.push(b"hello");
text.push(&[ascii::SP]);
text.push(b"world");
assert_eq!(
text.to_string(),
"hello world".to_string(),
);
}
}

5
src/text/misc_token.rs
View file

@ -9,11 +9,10 @@ use nom::{
}; };
use crate::text::{ use crate::text::{
quoted::quoted_string, quoted::{QuotedString, quoted_string},
whitespace::{fws, is_obs_no_ws_ctl}, whitespace::{fws, is_obs_no_ws_ctl},
words::{atom, is_vchar}, words::{atom, is_vchar},
encoding::{self, encoded_word}, encoding::{self, encoded_word},
buffer,
ascii, ascii,
}; };
@ -43,7 +42,7 @@ impl<'a> TryFrom<&'a lazy::PhraseList<'a>> for PhraseList {
#[derive(Debug, PartialEq)] #[derive(Debug, PartialEq)]
pub enum Word<'a> { pub enum Word<'a> {
Quoted(buffer::Text<'a>), Quoted(QuotedString<'a>),
Encoded(encoding::EncodedWord<'a>), Encoded(encoding::EncodedWord<'a>),
Atom(&'a [u8]), Atom(&'a [u8]),
} }

1
src/text/mod.rs
View file

@ -4,4 +4,3 @@ pub mod misc_token;
pub mod quoted; pub mod quoted;
pub mod whitespace; pub mod whitespace;
pub mod words; pub mod words;
pub mod buffer;

54
src/text/quoted.rs
View file

@ -9,7 +9,29 @@ use nom::{
use crate::text::whitespace::{cfws, fws, is_obs_no_ws_ctl}; use crate::text::whitespace::{cfws, fws, is_obs_no_ws_ctl};
use crate::text::ascii; use crate::text::ascii;
use crate::text::buffer;
#[derive(Debug, PartialEq, Default)]
pub struct QuotedString<'a>(pub Vec<&'a [u8]>);
impl<'a> QuotedString<'a> {
pub fn push(&mut self, e: &'a [u8]) {
self.0.push(e)
}
pub fn to_string(&self) -> String {
let enc = encoding_rs::UTF_8;
let size = self.0.iter().fold(0, |acc, v| acc + v.len());
self.0.iter().fold(
String::with_capacity(size),
|mut acc, v| {
let (content, _) = enc.decode_without_bom_handling(v);
acc.push_str(content.as_ref());
acc
},
)
}
}
/// Quoted pair /// Quoted pair
/// ///
@ -55,7 +77,7 @@ fn qcontent(input: &[u8]) -> IResult<&[u8], &[u8]> {
/// DQUOTE *([FWS] qcontent) [FWS] DQUOTE /// DQUOTE *([FWS] qcontent) [FWS] DQUOTE
/// [CFWS] /// [CFWS]
/// ``` /// ```
pub fn quoted_string(input: &[u8]) -> IResult<&[u8], buffer::Text> { pub fn quoted_string(input: &[u8]) -> IResult<&[u8], QuotedString> {
let (input, _) = opt(cfws)(input)?; let (input, _) = opt(cfws)(input)?;
let (input, _) = tag("\"")(input)?; let (input, _) = tag("\"")(input)?;
let (input, content) = many0(pair(opt(fws), qcontent))(input)?; let (input, content) = many0(pair(opt(fws), qcontent))(input)?;
@ -63,7 +85,7 @@ pub fn quoted_string(input: &[u8]) -> IResult<&[u8], buffer::Text> {
// Rebuild string // Rebuild string
let mut qstring = content let mut qstring = content
.iter() .iter()
.fold(buffer::Text::default(), |mut acc, (maybe_wsp, c)| { .fold(QuotedString::default(), |mut acc, (maybe_wsp, c)| {
if let Some(_) = maybe_wsp { if let Some(_) = maybe_wsp {
acc.push(&[ascii::SP]); acc.push(&[ascii::SP]);
} }
@ -86,23 +108,25 @@ mod tests {
use super::*; use super::*;
#[test] #[test]
fn test_quoted_string() { fn test_quoted_string_parser() {
let mut text = buffer::Text::default();
text.push(b"hello");
text.push(&[ascii::DQUOTE]);
text.push(b"world");
assert_eq!( assert_eq!(
quoted_string(b" \"hello\\\"world\" "), quoted_string(b" \"hello\\\"world\" ").unwrap().1,
Ok((&b""[..], text)) QuotedString(vec![b"hello", &[ascii::DQUOTE], b"world"])
); );
let mut text = buffer::Text::default();
text.push(b"hello");
text.push(&[ascii::SP]);
text.push(b"world");
assert_eq!( assert_eq!(
quoted_string(b"\"hello\r\n world\""), quoted_string(b"\"hello\r\n world\""),
Ok((&b""[..], text)) Ok((&b""[..], QuotedString(vec![b"hello", &[ascii::SP], b"world"]))),
);
}
use crate::text::ascii;
#[test]
fn test_quoted_string_object() {
assert_eq!(
QuotedString(vec![b"hello", &[ascii::SP], b"world"]).to_string(),
"hello world".to_string(),
); );
} }
} }