diff --git a/src/header.rs b/src/header.rs index 213dcbb..5c889a3 100644 --- a/src/header.rs +++ b/src/header.rs @@ -1,55 +1,62 @@ - -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::map, - multi::{fold_many0}, + combinator::{into, recognize}, + multi::many0, sequence::{pair, terminated, tuple}, IResult, }; +use std::fmt; -#[derive(Debug, PartialEq)] -pub enum CompField<'a, T> { - Known(T), - Unknown(Kv<'a>), - Bad(&'a [u8]), +use crate::text::misc_token::unstructured; +use crate::text::whitespace::{foldable_line, obs_crlf}; + +#[derive(PartialEq, Clone)] +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 struct Kv<'a>(pub &'a [u8], pub Unstructured<'a>); - - -pub fn header<'a, T>( - fx: impl Fn(&'a [u8]) -> IResult<&'a [u8], T> + Copy, -) -> impl Fn(&'a [u8]) -> IResult<&'a [u8], (Vec::, Vec::, Vec<&'a [u8]>)> { - move |input| { - terminated( - fold_many0( - alt(( - map(fx, CompField::Known), - map(opt_field, |(k, v)| CompField::Unknown(Kv(k, v))), - map(foldable_line, CompField::Bad), - )), - || (Vec::::new(), Vec::::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) +impl<'a> fmt::Debug for Kv2<'a> { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt.debug_tuple("header::Kv2") + .field(&String::from_utf8_lossy(self.0)) + .field(&String::from_utf8_lossy(self.1)) + .finish() } } -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) +#[derive(Debug, PartialEq, Clone)] +pub enum Field<'a> { + Good(Kv2<'a>), + Bad(&'a [u8]), +} +impl<'a> From> for Field<'a> { + fn from(kv: Kv2<'a>) -> Self { + Self::Good(kv) + } +} +impl<'a> From<&'a [u8]> for Field<'a> { + fn from(bad: &'a [u8]) -> Self { + Self::Bad(bad) + } +} + +/// Parse headers as key/values +pub fn header_kv(input: &[u8]) -> IResult<&[u8], Vec> { + terminated( + many0(alt((into(correct_field), into(foldable_line)))), + obs_crlf, + )(input) +} + +pub fn field_any(input: &[u8]) -> IResult<&[u8], &[u8]> { + terminated( + take_while1(|c| (0x21..=0x7E).contains(&c) && c != 0x3A), + tuple((space0, tag(b":"), space0)), + )(input) } /// Optional field @@ -61,15 +68,6 @@ pub fn field_name<'a>(name: &'static [u8]) -> impl Fn(&'a [u8]) -> IResult<&'a [ /// %d59-126 ; characters not including /// ; ":". /// ``` -pub fn opt_field(input: &[u8]) -> IResult<&[u8], (&[u8], Unstructured)> { - terminated( - pair( - terminated( - take_while1(|c| (0x21..=0x7E).contains(&c) && c != 0x3A), - tuple((space0, tag(b":"), space0)), - ), - unstructured, - ), - obs_crlf, - )(input) +pub fn correct_field(input: &[u8]) -> IResult<&[u8], Kv2> { + terminated(into(pair(field_any, recognize(unstructured))), obs_crlf)(input) } diff --git a/src/imf/field.rs b/src/imf/field.rs index c9ea2bd..0a7766d 100644 --- a/src/imf/field.rs +++ b/src/imf/field.rs @@ -1,21 +1,14 @@ use chrono::{DateTime, FixedOffset}; -use nom::{ - branch::alt, - combinator::map, - sequence::{preceded, terminated}, - IResult, -}; +use nom::combinator::map; -use crate::header::{field_name, header}; +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}; 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::imf::Imf; use crate::text::misc_token::{phrase_list, unstructured, PhraseList, Unstructured}; -use crate::text::whitespace::obs_crlf; #[derive(Debug, PartialEq)] pub enum Field<'a> { @@ -49,94 +42,34 @@ pub enum Field<'a> { MIMEVersion(Version), } - -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) -} - -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 - })(input) -} - -#[cfg(test)] -mod tests { - use super::*; - use crate::imf::address::*; - use crate::imf::mailbox::*; - use crate::text::misc_token::*; - use chrono::{FixedOffset, TimeZone}; - - #[test] - fn test_header() { - let fullmail = b"Date: 7 Mar 2023 08:00:00 +0200 -From: someone@example.com -To: someone_else@example.com -Subject: An RFC 822 formatted message - -This is the plain text body of the message. Note the blank line -between the header information and the body of the message."; - - assert_eq!( - imf(fullmail), - Ok(( - &b"This is the plain text body of the message. Note the blank line\nbetween the header information and the body of the message."[..], - Imf { - date: Some(FixedOffset::east_opt(2 * 3600).unwrap().with_ymd_and_hms(2023, 3, 7, 8, 0, 0).unwrap()), - from: vec![MailboxRef { - name: None, - addrspec: AddrSpec { - local_part: LocalPart(vec![LocalPartToken::Word(Word::Atom(&b"someone"[..]))]), - domain: Domain::Atoms(vec![&b"example"[..], &b"com"[..]]), - } - }], - to: vec![AddressRef::Single(MailboxRef { - name: None, - addrspec: AddrSpec { - local_part: LocalPart(vec![LocalPartToken::Word(Word::Atom(&b"someone_else"[..]))]), - domain: Domain::Atoms(vec![&b"example"[..], &b"com"[..]]), - } - })], - subject: Some(Unstructured(vec![ - UnstrToken::Plain(&b"An"[..]), - UnstrToken::Plain(&b"RFC"[..]), - UnstrToken::Plain(&b"822"[..]), - UnstrToken::Plain(&b"formatted"[..]), - UnstrToken::Plain(&b"message"[..]), - ])), - ..Imf::default() +impl<'a> TryFrom<&header::Field<'a>> for Field<'a> { + type Error = (); + fn try_from(f: &header::Field<'a>) -> Result { + 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(()), + }; + + content.map(|(_, content)| content).or(Err(())) } } diff --git a/src/imf/mod.rs b/src/imf/mod.rs index 7817ac7..b920c49 100644 --- a/src/imf/mod.rs +++ b/src/imf/mod.rs @@ -1,5 +1,4 @@ /// Parse and represent IMF (Internet Message Format) headers (RFC822, RFC5322) - pub mod address; pub mod datetime; pub mod field; @@ -8,13 +7,15 @@ pub mod mailbox; pub mod mime; pub mod trace; +use nom::{combinator::map, IResult}; + +use crate::header; use crate::imf::address::AddressRef; 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}; @@ -51,17 +52,13 @@ pub struct Imf<'a> { // MIME pub mime_version: Option, - // Junk - pub header_ext: Vec>, - pub header_bad: Vec<&'a [u8]>, + // Raw fields + pub kv: Vec>, } - impl<'a> Imf<'a> { - pub fn with_opt(mut self, opt: Vec>) -> 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_kv(mut self, v: Vec>) -> Self { + self.kv = v; + self } } @@ -92,3 +89,65 @@ impl<'a> FromIterator> for Imf<'a> { }) } } + +pub fn imf(input: &[u8]) -> IResult<&[u8], Imf> { + map(header::header_kv, |fields| { + fields + .iter() + .flat_map(Field::try_from) + .into_iter() + .collect::() + })(input) +} + +#[cfg(test)] +mod tests { + use super::*; + use crate::imf::address::*; + use crate::imf::mailbox::*; + use crate::text::misc_token::*; + use chrono::{FixedOffset, TimeZone}; + + #[test] + fn test_header() { + let fullmail = b"Date: 7 Mar 2023 08:00:00 +0200 +From: someone@example.com +To: someone_else@example.com +Subject: An RFC 822 formatted message + +This is the plain text body of the message. Note the blank line +between the header information and the body of the message."; + + assert_eq!( + imf(fullmail), + Ok(( + &b"This is the plain text body of the message. Note the blank line\nbetween the header information and the body of the message."[..], + Imf { + date: Some(FixedOffset::east_opt(2 * 3600).unwrap().with_ymd_and_hms(2023, 3, 7, 8, 0, 0).unwrap()), + from: vec![MailboxRef { + name: None, + addrspec: AddrSpec { + local_part: LocalPart(vec![LocalPartToken::Word(Word::Atom(&b"someone"[..]))]), + domain: Domain::Atoms(vec![&b"example"[..], &b"com"[..]]), + } + }], + to: vec![AddressRef::Single(MailboxRef { + name: None, + addrspec: AddrSpec { + local_part: LocalPart(vec![LocalPartToken::Word(Word::Atom(&b"someone_else"[..]))]), + domain: Domain::Atoms(vec![&b"example"[..], &b"com"[..]]), + } + })], + subject: Some(Unstructured(vec![ + UnstrToken::Plain(&b"An"[..]), + UnstrToken::Plain(&b"RFC"[..]), + UnstrToken::Plain(&b"822"[..]), + UnstrToken::Plain(&b"formatted"[..]), + UnstrToken::Plain(&b"message"[..]), + ])), + ..Imf::default() + } + )), + ) + } +} diff --git a/src/lib.rs b/src/lib.rs index b133b9e..bd79650 100644 --- a/src/lib.rs +++ b/src/lib.rs @@ -18,7 +18,7 @@ pub mod text; /// Manipulate buffer of bytes mod pointers; -use nom::{IResult, combinator::into}; +use nom::{combinator::into, IResult}; /// Parse a whole email including its (MIME) body /// @@ -37,7 +37,7 @@ use nom::{IResult, combinator::into}; /// * `msg` - The parsed message /// /// # Examples -/// +/// /// ``` /// let input = br#"Date: 7 Mar 2023 08:00:00 +0200 /// From: deuxfleurs@example.com @@ -57,7 +57,9 @@ use nom::{IResult, combinator::into}; /// ); /// ``` pub fn parse_message(input: &[u8]) -> IResult<&[u8], part::composite::Message> { - into(part::composite::message(mime::MIME::::default()))(input) + into(part::composite::message(mime::MIME::< + mime::r#type::DeductibleMessage, + >::default()))(input) } /// Only extract the headers of the email that are part of the Internet Message Format spec @@ -98,5 +100,5 @@ pub fn parse_message(input: &[u8]) -> IResult<&[u8], part::composite::Message> { /// ); /// ``` pub fn parse_imf(input: &[u8]) -> IResult<&[u8], imf::Imf> { - imf::field::imf(input) + imf::imf(input) } diff --git a/src/mime/field.rs b/src/mime/field.rs index b11210d..8736b47 100644 --- a/src/mime/field.rs +++ b/src/mime/field.rs @@ -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 +41,38 @@ impl<'a> Content<'a> { } } -/* -pub fn to_mime<'a, T: WithDefaultType>(list: Vec>) -> AnyMIMEWithDefault<'a, T> { - list.into_iter().collect::>() -}*/ +impl<'a> TryFrom<&header::Field<'a>> for Content<'a> { + type Error = (); + fn try_from(f: &header::Field<'a>) -> Result { + 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 +90,7 @@ mod tests { } else { panic!("Expected Content::Type, got {:?}", content); } - } + }*/ #[test] fn test_header() { @@ -116,7 +110,10 @@ 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![ diff --git a/src/mime/mod.rs b/src/mime/mod.rs index 2dd0c21..542d083 100644 --- a/src/mime/mod.rs +++ b/src/mime/mod.rs @@ -1,4 +1,4 @@ -/// Parsed and represent an email character set +/// Parsed and represent an email character set pub mod charset; /// MIME specific headers @@ -10,19 +10,20 @@ pub mod mechanism; /// Content-Type representation pub mod r#type; +use std::fmt; use std::marker::PhantomData; +use crate::header; use crate::imf::identification::MessageID; use crate::mime::field::Content; use crate::mime::mechanism::Mechanism; use crate::mime::r#type::{AnyType, NaiveType}; -use crate::header; use crate::text::misc_token::Unstructured; //Multipart, Message, Text, Binary}; #[derive(Debug, PartialEq, Clone)] pub struct MIME<'a, T> { - pub interpreted_type: T, - pub fields: NaiveMIME<'a> + pub interpreted_type: T, + pub fields: NaiveMIME<'a>, } impl<'a> Default for MIME<'a, r#type::DeductibleText> { fn default() -> Self { @@ -55,22 +56,32 @@ impl<'a, T: WithDefaultType> From> for AnyMIME<'a> { } } -#[derive(Debug, PartialEq, Default, Clone)] +#[derive(PartialEq, Default, Clone)] pub struct NaiveMIME<'a> { pub ctype: Option>, pub transfer_encoding: Mechanism<'a>, pub id: Option>, pub description: Option>, - pub header_ext: Vec>, - pub header_bad: Vec<&'a [u8]>, + pub kv: Vec>, pub raw: &'a [u8], } +impl<'a> fmt::Debug for NaiveMIME<'a> { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt.debug_struct("NaiveMime") + .field("ctype", &self.ctype) + .field("transfer_encoding", &self.transfer_encoding) + .field("id", &self.id) + .field("description", &self.description) + .field("kv", &self.kv) + .field("raw", &String::from_utf8_lossy(self.raw)) + .finish() + } +} impl<'a> FromIterator> for NaiveMIME<'a> { fn from_iter>>(it: I) -> Self { - it.into_iter().fold( - NaiveMIME::default(), - |mut section, field| { + it.into_iter() + .fold(NaiveMIME::default(), |mut section, field| { match field { Content::Type(v) => section.ctype = Some(v), Content::TransferEncoding(v) => section.transfer_encoding = v, @@ -78,28 +89,29 @@ impl<'a> FromIterator> for NaiveMIME<'a> { Content::Description(v) => section.description = Some(v), }; section - }, - ) + }) } } impl<'a> NaiveMIME<'a> { - pub fn with_opt(mut self, opt: Vec>) -> 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_kv(mut self, fields: Vec>) -> Self { + self.kv = fields; + self } pub fn with_raw(mut self, raw: &'a [u8]) -> Self { - self.raw = raw; self + self.raw = raw; + self } pub fn to_interpreted(self) -> AnyMIME<'a> { - self.ctype.as_ref().map(|c| c.to_type()).unwrap_or(T::default_type()).to_mime(self).into() + self.ctype + .as_ref() + .map(|c| c.to_type()) + .unwrap_or(T::default_type()) + .to_mime(self) + .into() } } - - pub trait WithDefaultType { fn default_type() -> AnyType; } diff --git a/src/mime/type.rs b/src/mime/type.rs index b83e643..7dafb29 100644 --- a/src/mime/type.rs +++ b/src/mime/type.rs @@ -5,19 +5,29 @@ use nom::{ sequence::{preceded, terminated, tuple}, IResult, }; +use std::fmt; use crate::mime::charset::EmailCharset; +use crate::mime::{AnyMIME, NaiveMIME, MIME}; use crate::text::misc_token::{mime_word, MIMEWord}; use crate::text::words::mime_atom; -use crate::mime::{AnyMIME, MIME, NaiveMIME}; // --------- NAIVE TYPE -#[derive(Debug, PartialEq, Clone)] +#[derive(PartialEq, Clone)] pub struct NaiveType<'a> { pub main: &'a [u8], pub sub: &'a [u8], pub params: Vec>, } +impl<'a> fmt::Debug for NaiveType<'a> { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt.debug_struct("mime::NaiveType") + .field("main", &String::from_utf8_lossy(self.main)) + .field("sub", &String::from_utf8_lossy(self.sub)) + .field("params", &self.params) + .finish() + } +} impl<'a> NaiveType<'a> { pub fn to_type(&self) -> AnyType { self.into() @@ -30,11 +40,20 @@ pub fn naive_type(input: &[u8]) -> IResult<&[u8], NaiveType> { )(input) } -#[derive(Debug, PartialEq, Clone)] +#[derive(PartialEq, Clone)] pub struct Parameter<'a> { pub name: &'a [u8], pub value: MIMEWord<'a>, } +impl<'a> fmt::Debug for Parameter<'a> { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt.debug_struct("mime::Parameter") + .field("name", &String::from_utf8_lossy(self.name)) + .field("value", &self.value) + .finish() + } +} + pub fn parameter(input: &[u8]) -> IResult<&[u8], Parameter> { map( tuple((mime_atom, tag(b"="), mime_word)), @@ -73,16 +92,27 @@ impl<'a> From<&'a NaiveType<'a>> for AnyType { impl<'a> AnyType { pub fn to_mime(self, fields: NaiveMIME<'a>) -> AnyMIME<'a> { - match self { - Self::Multipart(interpreted_type) => AnyMIME::Mult(MIME:: { interpreted_type, fields }), - Self::Message(interpreted_type) => AnyMIME::Msg(MIME:: { interpreted_type, fields }), - Self::Text(interpreted_type) => AnyMIME::Txt(MIME:: { interpreted_type, fields }), - Self::Binary(interpreted_type) => AnyMIME::Bin(MIME:: { interpreted_type, fields }), - } + match self { + Self::Multipart(interpreted_type) => AnyMIME::Mult(MIME:: { + interpreted_type, + fields, + }), + Self::Message(interpreted_type) => AnyMIME::Msg(MIME:: { + interpreted_type, + fields, + }), + Self::Text(interpreted_type) => AnyMIME::Txt(MIME:: { + interpreted_type, + fields, + }), + Self::Binary(interpreted_type) => AnyMIME::Bin(MIME:: { + interpreted_type, + fields, + }), + } } } - #[derive(Debug, PartialEq, Clone)] pub enum Deductible { Inferred(T), @@ -139,7 +169,8 @@ impl ToString for MultipartSubtype { Self::Parallel => "parallel", Self::Report => "report", Self::Unknown => "mixed", - }.into() + } + .into() } } impl<'a> From<&NaiveType<'a>> for MultipartSubtype { @@ -155,8 +186,6 @@ impl<'a> From<&NaiveType<'a>> for MultipartSubtype { } } - - #[derive(Debug, PartialEq, Default, Clone)] pub enum MessageSubtype { #[default] @@ -172,7 +201,8 @@ impl ToString for MessageSubtype { Self::Partial => "partial", Self::External => "external", Self::Unknown => "rfc822", - }.into() + } + .into() } } @@ -184,17 +214,25 @@ pub struct Message { impl<'a> From<&NaiveType<'a>> for Message { fn from(nt: &NaiveType<'a>) -> Self { match nt.sub.to_ascii_lowercase().as_slice() { - b"rfc822" => Self { subtype: MessageSubtype::RFC822 }, - b"partial" => Self { subtype: MessageSubtype::Partial }, - b"external" => Self { subtype: MessageSubtype::External }, - _ => Self { subtype: MessageSubtype::Unknown }, + b"rfc822" => Self { + subtype: MessageSubtype::RFC822, + }, + b"partial" => Self { + subtype: MessageSubtype::Partial, + }, + b"external" => Self { + subtype: MessageSubtype::External, + }, + _ => Self { + subtype: MessageSubtype::Unknown, + }, } } } impl From> for Message { fn from(d: Deductible) -> Self { match d { - Deductible::Inferred(t) | Deductible::Explicit(t) => t + Deductible::Inferred(t) | Deductible::Explicit(t) => t, } } } @@ -221,7 +259,7 @@ impl<'a> From<&NaiveType<'a>> for Text { impl From> for Text { fn from(d: Deductible) -> Self { match d { - Deductible::Inferred(t) | Deductible::Explicit(t) => t + Deductible::Inferred(t) | Deductible::Explicit(t) => t, } } } @@ -238,7 +276,8 @@ impl ToString for TextSubtype { match self { Self::Plain | Self::Unknown => "plain", Self::Html => "html", - }.into() + } + .into() } } impl<'a> From<&NaiveType<'a>> for TextSubtype { @@ -258,8 +297,8 @@ pub struct Binary {} mod tests { use super::*; use crate::mime::charset::EmailCharset; - use crate::text::quoted::QuotedString; use crate::mime::r#type::Deductible; + use crate::text::quoted::QuotedString; #[test] fn test_parameter() { @@ -317,7 +356,12 @@ mod tests { let (rest, nt) = naive_type(b"message/rfc822").unwrap(); assert_eq!(rest, &[]); - assert_eq!(nt.to_type(), AnyType::Message(Deductible::Explicit(Message { subtype: MessageSubtype::RFC822 }))); + assert_eq!( + nt.to_type(), + AnyType::Message(Deductible::Explicit(Message { + subtype: MessageSubtype::RFC822 + })) + ); } #[test] diff --git a/src/part/composite.rs b/src/part/composite.rs index dc2ac68..fabe5f5 100644 --- a/src/part/composite.rs +++ b/src/part/composite.rs @@ -1,20 +1,37 @@ use nom::IResult; +use std::fmt; -use crate::header::{header, self}; +use crate::header; use crate::imf; use crate::mime; -use crate::part::{self, AnyPart, field::MixedField}; -use crate::text::boundary::{boundary, Delimiter}; +use crate::part::{self, AnyPart}; use crate::pointers; +use crate::text::boundary::{boundary, Delimiter}; //--- Multipart -#[derive(Debug, PartialEq)] +#[derive(PartialEq)] pub struct Multipart<'a> { pub mime: mime::MIME<'a, mime::r#type::Multipart>, pub children: Vec>, pub raw_part_inner: &'a [u8], pub raw_part_outer: &'a [u8], } +impl<'a> fmt::Debug for Multipart<'a> { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt.debug_struct("part::Multipart") + .field("mime", &self.mime) + .field("children", &self.children) + .field( + "raw_part_inner", + &String::from_utf8_lossy(self.raw_part_inner), + ) + .field( + "raw_part_outer", + &String::from_utf8_lossy(self.raw_part_outer), + ) + .finish() + } +} impl<'a> Multipart<'a> { pub fn preamble(&self) -> &'a [u8] { pointers::parsed(self.raw_part_outer, self.raw_part_inner) @@ -65,7 +82,10 @@ pub fn multipart<'a>( mime: m.clone(), children: mparts, raw_part_inner: pointers::parsed(inner_orig, inp), - raw_part_outer: pointers::parsed(outer_orig, &outer_orig[outer_orig.len()..]), + raw_part_outer: pointers::parsed( + outer_orig, + &outer_orig[outer_orig.len()..], + ), }, )) } @@ -73,24 +93,30 @@ 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::() - .with_opt(unknown) - .with_bad(bad) - .with_raw(raw_hdrs); + .collect::(); + + let mime = mime.with_kv(fields).with_raw(raw_hdrs); + (input_eom, mime) - }, + } Err(_) => (input, mime::NaiveMIME::default()), }; // interpret mime according to context let mime = match m.interpreted_type.subtype { - mime::r#type::MultipartSubtype::Digest => naive_mime.to_interpreted::().into(), - _ => naive_mime.to_interpreted::().into(), + mime::r#type::MultipartSubtype::Digest => naive_mime + .to_interpreted::() + .into(), + _ => naive_mime + .to_interpreted::() + .into(), }; // parse raw part @@ -109,7 +135,7 @@ pub fn multipart<'a>( //--- Message -#[derive(Debug, PartialEq)] +#[derive(PartialEq)] pub struct Message<'a> { pub mime: mime::MIME<'a, mime::r#type::DeductibleMessage>, pub imf: imf::Imf<'a>, @@ -119,6 +145,18 @@ pub struct Message<'a> { pub raw_headers: &'a [u8], pub raw_body: &'a [u8], } +impl<'a> fmt::Debug for Message<'a> { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt.debug_struct("part::Message") + .field("mime", &self.mime) + .field("imf", &self.imf) + .field("child", &self.child) + .field("raw_part", &String::from_utf8_lossy(self.raw_part)) + .field("raw_headers", &String::from_utf8_lossy(self.raw_headers)) + .field("raw_body", &String::from_utf8_lossy(self.raw_body)) + .finish() + } +} pub fn message<'a>( m: mime::MIME<'a, mime::r#type::DeductibleMessage>, @@ -127,23 +165,27 @@ pub fn message<'a>( let orig = input; // parse header fields - let (input, (known, unknown, bad)): (_, (Vec::, Vec, 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); + let (naive_mime, imf) = part::field::split_and_build(&headers); - // attach bad headers to imf - let imf = imf.with_opt(unknown).with_bad(bad); + // Bind headers to the IMF object + let imf = imf.with_kv(headers); // interpret headers to choose a mime type - let in_mime = naive_mime.with_raw(raw_headers).to_interpreted::().into(); + let in_mime = naive_mime + .with_raw(raw_headers) + .to_interpreted::() + .into(); + //--------------- - // parse this mimetype + // parse a part following this mime specification let (input, part) = part::anypart(in_mime)(input)?; // extract raw parts 2/2 @@ -155,7 +197,9 @@ pub fn message<'a>( Message { mime: m.clone(), imf, - raw_part, raw_headers, raw_body, + raw_part, + raw_headers, + raw_body, child: Box::new(part), }, )) @@ -168,7 +212,7 @@ mod tests { use crate::part::discrete::Text; use crate::part::AnyPart; use crate::text::encoding::{Base64Word, EncodedWord, QuotedChunk, QuotedWord}; - use crate::text::misc_token::{Phrase, UnstrToken, Unstructured, Word, MIMEWord}; + use crate::text::misc_token::{MIMEWord, Phrase, UnstrToken, Unstructured, Word}; use crate::text::quoted::QuotedString; use chrono::{FixedOffset, TimeZone}; @@ -229,12 +273,15 @@ It DOES end with a linebreak. subtype: mime::r#type::TextSubtype::Plain, charset: mime::r#type::Deductible::Inferred(mime::charset::EmailCharset::US_ASCII), }), - fields: mime::NaiveMIME::default(), + fields: mime::NaiveMIME { + raw: &b"\n"[..], + ..mime::NaiveMIME::default() + }, }, body: &b"This is implicitly typed plain US-ASCII text.\nIt does NOT end with a linebreak."[..], }), AnyPart::Txt(Text { - mime: mime::MIME { + mime: mime::MIME { interpreted_type: mime::r#type::Deductible::Explicit(mime::r#type::Text { subtype: mime::r#type::TextSubtype::Plain, charset: mime::r#type::Deductible::Explicit(mime::charset::EmailCharset::US_ASCII), @@ -250,6 +297,10 @@ It DOES end with a linebreak. } ] }), + raw: &b"Content-type: text/plain; charset=us-ascii\n\n"[..], + kv: vec![ + header::Field::Good(header::Kv2(&b"Content-type"[..], &b"text/plain; charset=us-ascii"[..])) + ], ..mime::NaiveMIME::default() }, }, @@ -455,14 +506,18 @@ OoOoOoOoOoOoOoOoOoOoOoOoOoOoOoOoO
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"[..], + kv: vec![ + header::Field::Good(header::Kv2(&b"Date"[..], &b"Sat, 8 Jul 2023 07:14:29 +0200"[..])), + header::Field::Good(header::Kv2(&b"From"[..], &b"Grrrnd Zero "[..])), + header::Field::Good(header::Kv2(&b"To"[..], &b"John Doe "[..])), + header::Field::Good(header::Kv2(&b"CC"[..], &b"=?ISO-8859-1?Q?Andr=E9?= Pirard "[..])), + header::Field::Good(header::Kv2(&b"Subject"[..], &b"=?ISO-8859-1?B?SWYgeW91IGNhbiByZWFkIHRoaXMgeW8=?=\n =?ISO-8859-2?B?dSB1bmRlcnN0YW5kIHRoZSBleGFtcGxlLg==?="[..])), + header::Field::Good(header::Kv2(&b"X-Unknown"[..], &b"something something"[..])), + header::Field::Bad(&b"Bad entry\n on multiple lines\n"[..]), + header::Field::Good(header::Kv2(&b"Message-ID"[..], &b""[..])), + header::Field::Good(header::Kv2(&b"MIME-Version"[..], &b"1.0"[..])), + header::Field::Good(header::Kv2(&b"Content-Type"[..], &b"multipart/alternative;\n boundary=\"b1_e376dc71bafc953c0b0fdeb9983a9956\""[..])), + header::Field::Good(header::Kv2(&b"Content-Transfer-Encoding"[..], &b"7bit"[..])), ], ..imf::Imf::default() }, @@ -483,6 +538,7 @@ OoOoOoOoOoOoOoOoOoOoOoOoOoOoOoOoO
} ] }), + raw: hdrs, ..mime::NaiveMIME::default() }, }, @@ -507,6 +563,11 @@ OoOoOoOoOoOoOoOoOoOoOoOoOoOoOoOoO
] }), transfer_encoding: mime::mechanism::Mechanism::QuotedPrintable, + kv: vec![ + header::Field::Good(header::Kv2(&b"Content-Type"[..], &b"text/plain; charset=utf-8"[..])), + header::Field::Good(header::Kv2(&b"Content-Transfer-Encoding"[..], &b"quoted-printable"[..])), + ], + raw: &b"Content-Type: text/plain; charset=utf-8\nContent-Transfer-Encoding: quoted-printable\n\n"[..], ..mime::NaiveMIME::default() } }, @@ -529,7 +590,11 @@ OoOoOoOoOoOoOoOoOoOoOoOoOoOoOoOoO
value: MIMEWord::Atom(&b"us-ascii"[..]), } ] - }), + }), + kv: vec![ + header::Field::Good(header::Kv2(&b"Content-Type"[..], &b"text/html; charset=us-ascii"[..])), + ], + raw: &b"Content-Type: text/html; charset=us-ascii\n\n"[..], ..mime::NaiveMIME::default() }, }, diff --git a/src/part/discrete.rs b/src/part/discrete.rs index f92e032..8cfe5c3 100644 --- a/src/part/discrete.rs +++ b/src/part/discrete.rs @@ -12,10 +12,7 @@ impl<'a> fmt::Debug for Text<'a> { fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { fmt.debug_struct("part::Text") .field("mime", &self.mime) - .field( - "body", - &format_args!("\"{}\"", String::from_utf8_lossy(self.body)), - ) + .field("body", &String::from_utf8_lossy(self.body)) .finish() } } @@ -30,10 +27,7 @@ impl<'a> fmt::Debug for Binary<'a> { fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { fmt.debug_struct("part::Binary") .field("mime", &self.mime) - .field( - "body", - &format_args!("\"{}\"", String::from_utf8_lossy(self.body)), - ) + .field("body", &String::from_utf8_lossy(self.body)) .finish() } } diff --git a/src/part/field.rs b/src/part/field.rs index 3f4a6dc..e7fd32a 100644 --- a/src/part/field.rs +++ b/src/part/field.rs @@ -1,50 +1,24 @@ -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 to_mime(self) -> Option> { - 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> { - match self { - Self::IMF(v) => Some(v), - _ => None, - } - } -} +pub fn split_and_build<'a>(v: &Vec>) -> (mime::NaiveMIME<'a>, imf::Imf<'a>) { + let (mimev, imfv) = v.iter().fold( + ( + Vec::::new(), + Vec::::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 sections<'a>(list: Vec>) -> (mime::NaiveMIME<'a>, imf::Imf<'a>) { - let (v1, v2): (Vec, Vec<_>) = list.into_iter().partition(|v| v.mime().is_some()); - let mime = v1.into_iter().flat_map(MixedField::to_mime).collect::(); - let imf = v2.into_iter().flat_map(MixedField::to_imf).collect::(); - (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::(); + let fimf = imfv.into_iter().collect::(); + (fmime, fimf) } diff --git a/src/part/mod.rs b/src/part/mod.rs index 1d91f91..cc5c4b5 100644 --- a/src/part/mod.rs +++ b/src/part/mod.rs @@ -80,18 +80,19 @@ pub fn anypart<'a>(m: AnyMIME<'a>) -> impl FnOnce(&'a [u8]) -> IResult<&'a [u8], move |input| { let part = match m { AnyMIME::Mult(a) => multipart(a)(input) - .map(|(_, multi)| - multi.into()) - .unwrap_or(AnyPart::Txt(Text { - mime: mime::MIME::::default(), - body: input, - })), - AnyMIME::Msg(a) => message(a)(input) - .map(|(_, msg)| msg.into()) + .map(|(_, multi)| multi.into()) .unwrap_or(AnyPart::Txt(Text { mime: mime::MIME::::default(), body: input, })), + AnyMIME::Msg(a) => { + message(a)(input) + .map(|(_, msg)| msg.into()) + .unwrap_or(AnyPart::Txt(Text { + mime: mime::MIME::::default(), + body: input, + })) + } AnyMIME::Txt(a) => AnyPart::Txt(Text { mime: a, body: input, @@ -101,7 +102,7 @@ pub fn anypart<'a>(m: AnyMIME<'a>) -> impl FnOnce(&'a [u8]) -> IResult<&'a [u8], body: input, }), }; - + // This function always consumes the whole input Ok((&input[input.len()..], part)) } diff --git a/src/text/whitespace.rs b/src/text/whitespace.rs index c48e347..b7777c0 100644 --- a/src/text/whitespace.rs +++ b/src/text/whitespace.rs @@ -22,7 +22,12 @@ use nom::{ /// \r or \n is allowed nowhere else, so we also add this support. pub fn obs_crlf(input: &[u8]) -> IResult<&[u8], &[u8]> { - alt((tag(ascii::CRLF), tag(ascii::CRCRLF), tag(&[ascii::CR]), tag(&[ascii::LF])))(input) + alt(( + tag(ascii::CRLF), + tag(ascii::CRCRLF), + tag(&[ascii::CR]), + tag(&[ascii::LF]), + ))(input) } /// ```abnf