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

refactor headers

Browse source
This commit is contained in:
Quentin 2023-08-30 19:00:08 +02:00
parent d9cf6b225d
commit 18bb04340a
Signed by: quentin
GPG key ID: E9602264D639FF68
7 changed files with 107 additions and 221 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

66
src/header.rs
View file

@ -1,38 +1,31 @@
use crate::text::misc_token::{unstructured, Unstructured};
use crate::text::whitespace::{foldable_line, obs_crlf};
use nom::{
branch::alt,
bytes::complete::{tag, tag_no_case, take_while1},
bytes::complete::{tag, take_while1},
character::complete::space0,
combinator::{into, map},
multi::{fold_many0, many0},
combinator::{into, recognize},
multi::many0,
sequence::{pair, terminated, tuple},
IResult,
};
#[derive(Debug, PartialEq)]
pub enum CompField<'a, T> {
Known(T),
Unknown(Kv<'a>),
Bad(&'a [u8]),
}
use crate::text::whitespace::{foldable_line, obs_crlf};
use crate::text::misc_token::unstructured;
#[derive(Debug, PartialEq, Clone)]
pub struct Kv<'a>(pub &'a [u8], pub Unstructured<'a>);
impl<'a> From<(&'a [u8], Unstructured<'a>)> for Kv<'a> {
fn from(pair: (&'a [u8], Unstructured<'a>)) -> Self {
pub struct Kv2<'a>(pub &'a [u8], pub &'a [u8]);
impl<'a> From<(&'a [u8], &'a [u8])> for Kv2<'a> {
fn from(pair: (&'a [u8], &'a [u8])) -> Self {
Self(pair.0, pair.1)
}
}
#[derive(Debug, PartialEq, Clone)]
pub enum Field<'a> {
Good(Kv<'a>),
Good(Kv2<'a>),
Bad(&'a [u8]),
}
impl<'a> From<Kv<'a>> for Field<'a> {
fn from(kv: Kv<'a>) -> Self {
impl<'a> From<Kv2<'a>> for Field<'a> {
fn from(kv: Kv2<'a>) -> Self {
Self::Good(kv)
}
}
@ -47,7 +40,7 @@ pub fn header_kv(input: &[u8]) -> IResult<&[u8], Vec<Field>> {
terminated(
many0(
alt((
into(opt_field),
into(correct_field),
into(foldable_line),
))
),
@ -55,37 +48,6 @@ pub fn header_kv(input: &[u8]) -> IResult<&[u8], Vec<Field>> {
)(input)
}
pub fn header<'a, T>(
fx: impl Fn(&'a [u8]) -> IResult<&'a [u8], T> + Copy,
) -> impl Fn(&'a [u8]) -> IResult<&'a [u8], (Vec::<T>, Vec::<Kv>, Vec<&'a [u8]>)> {
move |input| {
terminated(
fold_many0(
alt((
map(fx, CompField::Known),
map(opt_field, CompField::Unknown),
map(foldable_line, CompField::Bad),
)),
|| (Vec::<T>::new(), Vec::<Kv>::new(), Vec::<&'a [u8]>::new()),
|(mut known, mut unknown, mut bad), item| {
match item {
CompField::Known(v) => known.push(v),
CompField::Unknown(v) => unknown.push(v),
CompField::Bad(v) => bad.push(v),
};
(known, unknown, bad)
}
),
obs_crlf,
)(input)
}
}
pub fn field_name<'a>(name: &'static [u8]) -> impl Fn(&'a [u8]) -> IResult<&'a [u8], &'a [u8]> {
move |input| terminated(tag_no_case(name), tuple((space0, tag(b":"), space0)))(input)
}
pub fn field_any(input: &[u8]) -> IResult<&[u8], &[u8]> {
terminated(
take_while1(|c| (0x21..=0x7E).contains(&c) && c != 0x3A),
@ -102,11 +64,11 @@ pub fn field_any(input: &[u8]) -> IResult<&[u8], &[u8]> {
/// %d59-126 ; characters not including
/// ; ":".
/// ```
pub fn opt_field(input: &[u8]) -> IResult<&[u8], Kv> {
pub fn correct_field(input: &[u8]) -> IResult<&[u8], Kv2> {
terminated(
into(pair(
field_any,
unstructured,
recognize(unstructured),
)),
obs_crlf,
)(input)

68
src/imf/field.rs
View file

@ -1,12 +1,7 @@
use chrono::{DateTime, FixedOffset};
use nom::{
branch::alt,
combinator::map,
sequence::{preceded, terminated},
IResult,
};
use nom::combinator::map;
use crate::header::{field_name};
use crate::header;
use crate::imf::address::{address_list, mailbox_list, nullable_address_list, AddressList};
use crate::imf::datetime::section as date;
use crate::imf::identification::{msg_id, msg_list, MessageID, MessageIDList};
@ -14,7 +9,6 @@ use crate::imf::mailbox::{mailbox, AddrSpec, MailboxList, MailboxRef};
use crate::imf::mime::{version, Version};
use crate::imf::trace::{received_log, return_path, ReceivedLog};
use crate::text::misc_token::{phrase_list, unstructured, PhraseList, Unstructured};
use crate::text::whitespace::obs_crlf;
#[derive(Debug, PartialEq)]
pub enum Field<'a> {
@ -48,36 +42,32 @@ pub enum Field<'a> {
MIMEVersion(Version),
}
/*impl<'a> From<header::Field<'a>> for Field<'a> {
fn from(raw: header::Field
}*/
impl<'a> TryFrom<&header::Field<'a>> for Field<'a> {
type Error = ();
fn try_from(f: &header::Field<'a>) -> Result<Self, Self::Error> {
let content = match f {
header::Field::Good(header::Kv2(key, value)) => match key.to_ascii_lowercase().as_slice() {
b"date" => map(date, Field::Date)(value),
b"from" => map(mailbox_list, Field::From)(value),
b"sender" => map(mailbox, Field::Sender)(value),
b"reply-to" => map(address_list, Field::ReplyTo)(value),
b"to" => map(address_list, Field::To)(value),
b"cc" => map(address_list, Field::Cc)(value),
b"bcc" => map(nullable_address_list, Field::Bcc)(value),
b"message-id" => map(msg_id, Field::MessageID)(value),
b"in-reply-to" => map(msg_list, Field::InReplyTo)(value),
b"references" => map(msg_list, Field::References)(value),
b"subject" => map(unstructured, Field::Subject)(value),
b"comments" => map(unstructured, Field::Comments)(value),
b"keywords" => map(phrase_list, Field::Keywords)(value),
b"return-path" => map(return_path, Field::ReturnPath)(value),
b"received" => map(received_log, Field::Received)(value),
b"mime-version" => map(version, Field::MIMEVersion)(value),
_ => return Err(()),
},
_ => return Err(()),
};
pub fn field(input: &[u8]) -> IResult<&[u8], Field> {
terminated(
alt((
preceded(field_name(b"date"), map(date, Field::Date)),
preceded(field_name(b"from"), map(mailbox_list, Field::From)),
preceded(field_name(b"sender"), map(mailbox, Field::Sender)),
preceded(field_name(b"reply-to"), map(address_list, Field::ReplyTo)),
preceded(field_name(b"to"), map(address_list, Field::To)),
preceded(field_name(b"cc"), map(address_list, Field::Cc)),
preceded(field_name(b"bcc"), map(nullable_address_list, Field::Bcc)),
preceded(field_name(b"message-id"), map(msg_id, Field::MessageID)),
preceded(field_name(b"in-reply-to"), map(msg_list, Field::InReplyTo)),
preceded(field_name(b"references"), map(msg_list, Field::References)),
preceded(field_name(b"subject"), map(unstructured, Field::Subject)),
preceded(field_name(b"comments"), map(unstructured, Field::Comments)),
preceded(field_name(b"keywords"), map(phrase_list, Field::Keywords)),
preceded(
field_name(b"return-path"),
map(return_path, Field::ReturnPath),
),
preceded(field_name(b"received"), map(received_log, Field::Received)),
preceded(
field_name(b"mime-version"),
map(version, Field::MIMEVersion),
),
)),
obs_crlf,
)(input)
content.map(|(_, content)| content).or(Err(()))
}
}

25
src/imf/mod.rs
View file

@ -13,14 +13,13 @@ use nom::{
IResult,
};
use crate::header::header;
use crate::header;
use crate::imf::address::AddressRef;
use crate::imf::field::{field, Field};
use crate::imf::field::Field;
use crate::imf::identification::MessageID;
use crate::imf::mailbox::{AddrSpec, MailboxRef};
use crate::imf::mime::Version;
use crate::imf::trace::ReceivedLog;
use crate::header;
use crate::text::misc_token::{PhraseList, Unstructured};
use chrono::{DateTime, FixedOffset};
@ -56,19 +55,6 @@ pub struct Imf<'a> {
// MIME
pub mime_version: Option<Version>,
// Junk
pub header_ext: Vec<header::Kv<'a>>,
pub header_bad: Vec<&'a [u8]>,
}
impl<'a> Imf<'a> {
pub fn with_opt(mut self, opt: Vec<header::Kv<'a>>) -> Self {
self.header_ext = opt; self
}
pub fn with_bad(mut self, bad: Vec<&'a [u8]>) -> Self {
self.header_bad = bad; self
}
}
//@FIXME min and max limits are not enforced,
@ -100,11 +86,8 @@ impl<'a> FromIterator<Field<'a>> for Imf<'a> {
}
pub fn imf(input: &[u8]) -> IResult<&[u8], Imf> {
map(header(field), |(known, unknown, bad)| {
let mut imf = Imf::from_iter(known);
imf.header_ext = unknown;
imf.header_bad = bad;
imf
map(header::header_kv, |fields| {
fields.iter().flat_map(Field::try_from).into_iter().collect::<Imf>()
})(input)
}

55
src/mime/field.rs
View file

@ -1,16 +1,10 @@
use nom::{
branch::alt,
combinator::map,
sequence::{preceded, terminated},
IResult,
};
use nom::combinator::map;
use crate::header::{field_name};
use crate::header;
use crate::imf::identification::{msg_id, MessageID};
use crate::mime::mechanism::{mechanism, Mechanism};
use crate::mime::r#type::{naive_type, NaiveType};
use crate::text::misc_token::{unstructured, Unstructured};
use crate::text::whitespace::obs_crlf;
#[derive(Debug, PartialEq)]
pub enum Content<'a> {
@ -47,38 +41,35 @@ impl<'a> Content<'a> {
}
}
/*
pub fn to_mime<'a, T: WithDefaultType>(list: Vec<Content<'a>>) -> AnyMIMEWithDefault<'a, T> {
list.into_iter().collect::<AnyMIMEWithDefault<T>>()
}*/
impl<'a> TryFrom<&header::Field<'a>> for Content<'a> {
type Error = ();
fn try_from(f: &header::Field<'a>) -> Result<Self, Self::Error> {
let content = match f {
header::Field::Good(header::Kv2(key, value)) => match key.to_ascii_lowercase().as_slice() {
b"content-type" => map(naive_type, Content::Type)(value),
b"content-transfer-encoding" => map(mechanism, Content::TransferEncoding)(value),
b"content-id" => map(msg_id, Content::ID)(value),
b"content-description" => map(unstructured, Content::Description)(value),
_ => return Err(()),
},
_ => return Err(()),
};
pub fn content(input: &[u8]) -> IResult<&[u8], Content> {
terminated(
alt((
preceded(field_name(b"content-type"), map(naive_type, Content::Type)),
preceded(
field_name(b"content-transfer-encoding"),
map(mechanism, Content::TransferEncoding),
),
preceded(field_name(b"content-id"), map(msg_id, Content::ID)),
preceded(
field_name(b"content-description"),
map(unstructured, Content::Description),
),
)),
obs_crlf,
)(input)
//@TODO check that the full value is parsed, otherwise maybe log an error ?!
content.map(|(_, content)| content).or(Err(()))
}
}
#[cfg(test)]
mod tests {
use super::*;
use crate::header::{header};
use crate::mime::charset::EmailCharset;
use crate::header;
//use crate::mime::charset::EmailCharset;
use crate::mime::r#type::*;
use crate::text::misc_token::MIMEWord;
use crate::text::quoted::QuotedString;
/*
#[test]
fn test_content_type() {
let (rest, content) =
@ -96,7 +87,7 @@ mod tests {
} else {
panic!("Expected Content::Type, got {:?}", content);
}
}
}*/
#[test]
fn test_header() {
@ -116,7 +107,7 @@ This is a multipart message.
.as_bytes();
assert_eq!(
map(header(content), |(k, _, _)| k)(fullmail),
map(header::header_kv, |k| k.iter().flat_map(Content::try_from).collect())(fullmail),
Ok((
&b"This is a multipart message.\n\n"[..],
vec![

10
src/mime/mod.rs
View file

@ -61,8 +61,7 @@ pub struct NaiveMIME<'a> {
pub transfer_encoding: Mechanism<'a>,
pub id: Option<MessageID<'a>>,
pub description: Option<Unstructured<'a>>,
pub header_ext: Vec<header::Kv<'a>>,
pub header_bad: Vec<&'a [u8]>,
pub fields: Vec<header::Field<'a>>,
pub raw: &'a [u8],
}
@ -84,11 +83,8 @@ impl<'a> FromIterator<Content<'a>> for NaiveMIME<'a> {
}
impl<'a> NaiveMIME<'a> {
pub fn with_opt(mut self, opt: Vec<header::Kv<'a>>) -> Self {
self.header_ext = opt; self
}
pub fn with_bad(mut self, bad: Vec<&'a [u8]>) -> Self {
self.header_bad = bad; self
pub fn with_fields(mut self, fields: Vec<header::Field<'a>>) -> Self {
self.fields = fields; self
}
pub fn with_raw(mut self, raw: &'a [u8]) -> Self {
self.raw = raw; self

43
src/part/composite.rs
View file

@ -1,9 +1,9 @@
use nom::IResult;
use crate::header::{header, self};
use crate::header;
use crate::imf;
use crate::mime;
use crate::part::{self, AnyPart, field::MixedField};
use crate::part::{self, AnyPart};
use crate::text::boundary::{boundary, Delimiter};
use crate::pointers;
@ -73,15 +73,19 @@ pub fn multipart<'a>(
};
// parse mime headers, otherwise pick default mime
let (input, naive_mime) = match header(mime::field::content)(input) {
Ok((input_eom, (known, unknown, bad))) => {
let (input, naive_mime) = match header::header_kv(input) {
Ok((input_eom, fields)) => {
let raw_hdrs = pointers::parsed(input, input_eom);
let mime = known
let mime = fields
.iter()
.flat_map(mime::field::Content::try_from)
.into_iter()
.collect::<mime::NaiveMIME>()
.with_opt(unknown)
.with_bad(bad)
.collect::<mime::NaiveMIME>();
let mime = mime
.with_fields(fields)
.with_raw(raw_hdrs);
(input_eom, mime)
},
Err(_) => (input, mime::NaiveMIME::default()),
@ -127,23 +131,21 @@ pub fn message<'a>(
let orig = input;
// parse header fields
let (input, (known, unknown, bad)): (_, (Vec::<MixedField>, Vec<header::Kv>, Vec<&[u8]>)) =
header(part::field::mixed_field)(input)?;
let (input, headers) = header::header_kv(input)?;
// extract raw parts 1/2
let raw_headers = pointers::parsed(orig, input);
let body_orig = input;
//---------------
// aggregate header fields
let (naive_mime, imf) = part::field::sections(known);
// attach bad headers to imf
let imf = imf.with_opt(unknown).with_bad(bad);
let (naive_mime, imf) = part::field::split_and_build(&headers);
// interpret headers to choose a mime type
let in_mime = naive_mime.with_raw(raw_headers).to_interpreted::<mime::WithGenericDefault>().into();
let in_mime = naive_mime.with_fields(headers).with_raw(raw_headers).to_interpreted::<mime::WithGenericDefault>().into();
//---------------
// parse this mimetype
// parse a part following this mime specification
let (input, part) = part::anypart(in_mime)(input)?;
// extract raw parts 2/2
@ -459,15 +461,6 @@ OoOoOoOoOoOoOoOoOoOoOoOoOoOoOoOoO<br />
right: &b"www.grrrndzero.org"[..],
}),
mime_version: Some(imf::mime::Version { major: 1, minor: 0}),
header_ext: vec![
header::Kv(&b"X-Unknown"[..], Unstructured(vec![
UnstrToken::Plain(&b"something"[..]),
UnstrToken::Plain(&b"something"[..]),
]))
],
header_bad: vec![
&b"Bad entry\n on multiple lines\n"[..],
],
..imf::Imf::default()
},
child: Box::new(AnyPart::Mult(Multipart {

61
src/part/field.rs
View file

@ -1,50 +1,21 @@
use nom::{branch::alt, combinator::map, IResult};
use crate::header;
use crate::imf;
use crate::mime;
pub enum MixedField<'a> {
MIME(mime::field::Content<'a>),
IMF(imf::field::Field<'a>),
}
#[allow(dead_code)]
impl<'a> MixedField<'a> {
pub fn mime(&self) -> Option<&mime::field::Content<'a>> {
match self {
Self::MIME(v) => Some(v),
_ => None,
pub fn split_and_build<'a>(v: &Vec<header::Field<'a>>) -> (mime::NaiveMIME<'a>, imf::Imf<'a>) {
let (mimev, imfv) = v.iter().fold(
(Vec::<mime::field::Content>::new(), Vec::<imf::field::Field>::new()),
|(mut mime, mut imf), f| {
if let Ok(m) = mime::field::Content::try_from(f) {
mime.push(m);
} else if let Ok(i) = imf::field::Field::try_from(f) {
imf.push(i);
}
(mime, imf)
}
}
pub fn to_mime(self) -> Option<mime::field::Content<'a>> {
match self {
Self::MIME(v) => Some(v),
_ => None,
}
}
pub fn imf(&self) -> Option<&imf::field::Field<'a>> {
match self {
Self::IMF(v) => Some(v),
_ => None,
}
}
pub fn to_imf(self) -> Option<imf::field::Field<'a>> {
match self {
Self::IMF(v) => Some(v),
_ => None,
}
}
}
pub fn sections<'a>(list: Vec<MixedField<'a>>) -> (mime::NaiveMIME<'a>, imf::Imf<'a>) {
let (v1, v2): (Vec<MixedField>, Vec<_>) = list.into_iter().partition(|v| v.mime().is_some());
let mime = v1.into_iter().flat_map(MixedField::to_mime).collect::<mime::NaiveMIME>();
let imf = v2.into_iter().flat_map(MixedField::to_imf).collect::<imf::Imf>();
(mime, imf)
}
pub fn mixed_field(input: &[u8]) -> IResult<&[u8], MixedField> {
alt((
map(mime::field::content, MixedField::MIME),
map(imf::field::field, MixedField::IMF),
))(input)
);
let fmime = mimev.into_iter().collect::<mime::NaiveMIME>();
let fimf = imfv.into_iter().collect::<imf::Imf>();
(fmime, fimf)
}