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

collect parsing errors

Browse source
This commit is contained in:
Quentin 2023-06-16 17:24:51 +02:00
parent 326c903c43
commit 1527b41ab4
Signed by: quentin
GPG key ID: E9602264D639FF68
4 changed files with 257 additions and 206 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

11
src/datetime.rs
View file

@ -1,13 +1,10 @@
use chrono::DateTime; use chrono::{DateTime, FixedOffset};
use nom::IResult; use nom::IResult;
use crate::{model,misc_token}; use crate::misc_token;
pub fn section(input: &str) -> IResult<&str, model::HeaderDate> { pub fn section(input: &str) -> IResult<&str, DateTime<FixedOffset>> {
// @FIXME want to extract datetime our way in the future // @FIXME want to extract datetime our way in the future
// to better handle obsolete/bad cases instead of returning raw text. // to better handle obsolete/bad cases instead of returning raw text.
let (input, raw_date) = misc_token::unstructured(input)?; let (input, raw_date) = misc_token::unstructured(input)?;
match DateTime::parse_from_rfc2822(&raw_date) { Ok((input, DateTime::parse_from_rfc2822(&raw_date).unwrap()))
Ok(chronodt) => Ok((input, model::HeaderDate::Parsed(chronodt))),
Err(e) => Ok((input, model::HeaderDate::Unknown(raw_date, e))),
}
} }

390
src/header.rs
View file

@ -1,5 +1,7 @@
use chrono::{DateTime, FixedOffset};
use nom::{ use nom::{
IResult, IResult,
Parser,
branch::alt, branch::alt,
bytes::complete::{is_not, take_while1, take_while, tag, tag_no_case}, bytes::complete::{is_not, take_while1, take_while, tag, tag_no_case},
character::complete::space0, character::complete::space0,
@ -11,7 +13,7 @@ use nom::{
use crate::whitespace::{fws, perm_crlf}; use crate::whitespace::{fws, perm_crlf};
use crate::words::vchar_seq; use crate::words::vchar_seq;
use crate::misc_token::{phrase, unstructured}; use crate::misc_token::{phrase, unstructured};
use crate::model::{HeaderSection, HeaderDate, MailboxRef, AddressRef}; use crate::model::{HeaderSection, MailboxRef, AddressRef, Field, FieldBody};
use crate::mailbox::mailbox; use crate::mailbox::mailbox;
use crate::address::{mailbox_list, address_list, address_list_cfws}; use crate::address::{mailbox_list, address_list, address_list_cfws};
use crate::identification::msg_id; use crate::identification::msg_id;
@ -24,7 +26,7 @@ use crate::{datetime, trace, model};
/// See: https://www.rfc-editor.org/rfc/rfc5322.html#section-2.2 /// See: https://www.rfc-editor.org/rfc/rfc5322.html#section-2.2
pub fn section(input: &str) -> IResult<&str, HeaderSection> { pub fn section(input: &str) -> IResult<&str, HeaderSection> {
let (input, headers) = fold_many0( let (input, headers) = fold_many0(
alt((header_field, unknown_field, rescue)), alt((known_field, unknown_field, rescue_field)),
HeaderSection::default, HeaderSection::default,
|mut section, head| { |mut section, head| {
match head { match head {
@ -32,87 +34,91 @@ pub fn section(input: &str) -> IResult<&str, HeaderSection> {
// it may result in missing data or silently overriden data. // it may result in missing data or silently overriden data.
// 3.6.1. The Origination Date Field // 3.6.1. The Origination Date Field
HeaderField::Date(d) => {
// | orig-date | 1 | 1 | | // | orig-date | 1 | 1 | |
section.date = d; Field::Date(FieldBody::Correct(d)) => {
section.date = Some(d);
} }
// 3.6.2. Originator Fields // 3.6.2. Originator Fields
HeaderField::From(v) => { Field::From(FieldBody::Correct(v)) => {
// | from | 1 | 1 | See sender and 3.6.2 | // | from | 1 | 1 | See sender and 3.6.2 |
section.from = v; section.from = v;
} }
HeaderField::Sender(mbx) => { Field::Sender(FieldBody::Correct(mbx)) => {
// | sender | 0* | 1 | MUST occur with multi-address from - see 3.6.2 | // | sender | 0* | 1 | MUST occur with multi-address from - see 3.6.2 |
section.sender = Some(mbx); section.sender = Some(mbx);
} }
HeaderField::ReplyTo(addr_list) => { Field::ReplyTo(FieldBody::Correct(addr_list)) => {
// | reply-to | 0 | 1 | | // | reply-to | 0 | 1 | |
section.reply_to = addr_list; section.reply_to = addr_list;
} }
// 3.6.3. Destination Address Fields // 3.6.3. Destination Address Fields
HeaderField::To(addr_list) => { Field::To(FieldBody::Correct(addr_list)) => {
// | to | 0 | 1 | | // | to | 0 | 1 | |
section.to = addr_list; section.to = addr_list;
} }
HeaderField::Cc(addr_list) => { Field::Cc(FieldBody::Correct(addr_list)) => {
// | cc | 0 | 1 | | // | cc | 0 | 1 | |
section.cc = addr_list; section.cc = addr_list;
} }
HeaderField::Bcc(addr_list) => { Field::Bcc(FieldBody::Correct(addr_list)) => {
// | bcc | 0 | 1 | | // | bcc | 0 | 1 | |
section.bcc = addr_list; section.bcc = addr_list;
} }
// 3.6.4. Identification Fields // 3.6.4. Identification Fields
HeaderField::MessageID(msg_id) => { Field::MessageID(FieldBody::Correct(msg_id)) => {
// | message-id | 0* | 1 | SHOULD be present - see 3.6.4 | // | message-id | 0* | 1 | SHOULD be present - see 3.6.4 |
section.msg_id = Some(msg_id); section.msg_id = Some(msg_id);
} }
HeaderField::InReplyTo(id_list) => { Field::InReplyTo(FieldBody::Correct(id_list)) => {
// | in-reply-to | 0* | 1 | SHOULD occur in some replies - see 3.6.4 | // | in-reply-to | 0* | 1 | SHOULD occur in some replies - see 3.6.4 |
section.in_reply_to = id_list; section.in_reply_to = id_list;
} }
HeaderField::References(id_list) => { Field::References(FieldBody::Correct(id_list)) => {
// | in-reply-to | 0* | 1 | SHOULD occur in some replies - see 3.6.4 | // | in-reply-to | 0* | 1 | SHOULD occur in some replies - see 3.6.4 |
section.references = id_list; section.references = id_list;
} }
// 3.6.5. Informational Fields // 3.6.5. Informational Fields
HeaderField::Subject(title) => { Field::Subject(FieldBody::Correct(title)) => {
// | subject | 0 | 1 | | // | subject | 0 | 1 | |
section.subject = Some(title); section.subject = Some(title);
} }
HeaderField::Comments(coms) => { Field::Comments(FieldBody::Correct(coms)) => {
// | comments | 0 | unlimited | | // | comments | 0 | unlimited | |
section.comments.push(coms); section.comments.push(coms);
} }
HeaderField::Keywords(mut kws) => { Field::Keywords(FieldBody::Correct(mut kws)) => {
// | keywords | 0 | unlimited | | // | keywords | 0 | unlimited | |
section.keywords.append(&mut kws); section.keywords.append(&mut kws);
} }
// 3.6.6 Resent Fields are not implemented // 3.6.6 Resent Fields are not implemented
// 3.6.7 Trace Fields // 3.6.7 Trace Fields
HeaderField::ReturnPath(maybe_mbx) => { Field::ReturnPath(FieldBody::Correct(maybe_mbx)) => {
if let Some(mbx) = maybe_mbx { if let Some(mbx) = maybe_mbx {
section.return_path.push(mbx); section.return_path.push(mbx);
} }
} }
HeaderField::Received(log) => { Field::Received(FieldBody::Correct(log)) => {
section.received.push(log); section.received.push(log);
} }
// 3.6.8. Optional Fields // 3.6.8. Optional Fields
HeaderField::Optional(name, body) => { Field::Optional(name, body) => {
section.optional.insert(name, body); section.optional.insert(name, body);
} }
// Rescue // Rescue
HeaderField::Rescue(x) => { Field::Rescue(x) => {
section.unparsed.push(x); section.unparsed.push(x);
} }
bad_field => {
section.bad_fields.push(bad_field);
}
}; };
section section
} }
@ -122,48 +128,13 @@ pub fn section(input: &str) -> IResult<&str, HeaderSection> {
Ok((input, headers)) Ok((input, headers))
} }
#[derive(Debug, PartialEq)]
pub enum HeaderField<'a> {
// 3.6.1. The Origination Date Field
Date(HeaderDate),
// 3.6.2. Originator Fields
From(Vec<MailboxRef>),
Sender(MailboxRef),
ReplyTo(Vec<AddressRef>),
// 3.6.3. Destination Address Fields
To(Vec<AddressRef>),
Cc(Vec<AddressRef>),
Bcc(Vec<AddressRef>),
// 3.6.4. Identification Fields
MessageID(model::MessageId<'a>),
InReplyTo(Vec<model::MessageId<'a>>),
References(Vec<model::MessageId<'a>>),
// 3.6.5. Informational Fields
Subject(String),
Comments(String),
Keywords(Vec<String>),
// 3.6.6 Resent Fields (not implemented)
// 3.6.7 Trace Fields
Received(&'a str),
ReturnPath(Option<model::MailboxRef>),
// 3.6.8. Optional Fields
Optional(&'a str, String),
// None
Rescue(&'a str),
}
/// Parse one known header field /// Parse one known header field
/// ///
/// RFC5322 optional-field seems to be a generalization of the field terminology. /// RFC5322 optional-field seems to be a generalization of the field terminology.
/// We use it to parse all header names: /// We use it to parse all header names:
pub fn header_field(input: &str) -> IResult<&str, HeaderField> { pub fn known_field(input: &str) -> IResult<&str, Field> {
terminated( terminated(
alt(( alt((
// 3.6.1. The Origination Date Field // 3.6.1. The Origination Date Field
@ -183,96 +154,117 @@ pub fn header_field(input: &str) -> IResult<&str, HeaderField> {
)(input) )(input)
} }
// 3.6.1. The Origination Date Field /// A high-level function to match more easily a field name
fn date(input: &str) -> IResult<&str, HeaderField> {
let (input, body) = preceded(field_name_tag("Date"), datetime::section)(input)?;
Ok((input, HeaderField::Date(body)))
}
// 3.6.2. Originator Fields
fn from(input: &str) -> IResult<&str, HeaderField> {
let (input, body) = preceded(field_name_tag("From"), mailbox_list)(input)?;
Ok((input, HeaderField::From(body)))
}
fn sender(input: &str) -> IResult<&str, HeaderField> {
let (input, body) = preceded(field_name_tag("Sender"), mailbox)(input)?;
Ok((input, HeaderField::Sender(body)))
}
fn reply_to(input: &str) -> IResult<&str, HeaderField> {
let (input, body) = preceded(field_name_tag("Reply-To"), address_list)(input)?;
Ok((input, HeaderField::ReplyTo(body)))
}
// 3.6.3. Destination Address Fields
fn to(input: &str) -> IResult<&str, HeaderField> {
let (input, body) = preceded(field_name_tag("To"), address_list)(input)?;
Ok((input, HeaderField::To(body)))
}
fn cc(input: &str) -> IResult<&str, HeaderField> {
let (input, body) = preceded(field_name_tag("Cc"), address_list)(input)?;
Ok((input, HeaderField::Cc(body)))
}
fn bcc(input: &str) -> IResult<&str, HeaderField> {
let (input, body) = preceded(
field_name_tag("Bcc"),
opt(alt((address_list, address_list_cfws))),
)(input)?;
Ok((input, HeaderField::Bcc(body.unwrap_or(vec![]))))
}
// 3.6.4. Identification Fields
fn msg_id_field(input: &str) -> IResult<&str, HeaderField> {
let (input, body) = preceded(field_name_tag("Message-ID"), msg_id)(input)?;
Ok((input, HeaderField::MessageID(body)))
}
fn in_reply_to(input: &str) -> IResult<&str, HeaderField> {
let (input, body) = preceded(field_name_tag("In-Reply-To"), many1(msg_id))(input)?;
Ok((input, HeaderField::InReplyTo(body)))
}
fn references(input: &str) -> IResult<&str, HeaderField> {
let (input, body) = preceded(field_name_tag("References"), many1(msg_id))(input)?;
Ok((input, HeaderField::References(body)))
}
// 3.6.5. Informational Fields
fn subject(input: &str) -> IResult<&str, HeaderField> {
let (input, body) = preceded(field_name_tag("Subject"), unstructured)(input)?;
Ok((input, HeaderField::Subject(body)))
}
fn comments(input: &str) -> IResult<&str, HeaderField> {
let (input, body) = preceded(field_name_tag("Comments"), unstructured)(input)?;
Ok((input, HeaderField::Comments(body)))
}
fn keywords(input: &str) -> IResult<&str, HeaderField> {
let (input, body) = preceded(
field_name_tag("Keywords"),
separated_list1(tag(","), phrase),
)(input)?;
Ok((input, HeaderField::Keywords(body)))
}
fn field_name_tag(field_name: &str) -> impl FnMut(&str) -> IResult<&str, &str> + '_ { fn field_name_tag(field_name: &str) -> impl FnMut(&str) -> IResult<&str, &str> + '_ {
move |input: &str| { move |input: &str| {
recognize(tuple((tag_no_case(field_name), space0, tag(":"), space0)))(input) recognize(tuple((tag_no_case(field_name), space0, tag(":"), space0)))(input)
} }
} }
// 3.6.1. The Origination Date Field
fn date(input: &str) -> IResult<&str, Field> {
map(preceded(field_name_tag("Date"), alt((
map(datetime::section, |dt| FieldBody::Correct(dt)),
map(rescue, |r| FieldBody::Failed(r))))),
|b| Field::Date(b))(input)
}
// 3.6.2. Originator Fields
fn from(input: &str) -> IResult<&str, Field> {
map(preceded(field_name_tag("From"), alt((
map(mailbox_list, |dt| FieldBody::Correct(dt)),
map(rescue, |r| FieldBody::Failed(r))))),
|b| Field::From(b))(input)
}
fn sender(input: &str) -> IResult<&str, Field> {
map(preceded(field_name_tag("Sender"), alt((
map(mailbox, |dt| FieldBody::Correct(dt)),
map(rescue, |r| FieldBody::Failed(r))))),
|b| Field::Sender(b))(input)
}
fn reply_to(input: &str) -> IResult<&str, Field> {
map(preceded(field_name_tag("Reply-To"), alt((
map(address_list, |dt| FieldBody::Correct(dt)),
map(rescue, |r| FieldBody::Failed(r))))),
|b| Field::ReplyTo(b))(input)
}
// 3.6.3. Destination Address Fields
fn to(input: &str) -> IResult<&str, Field> {
map(preceded(field_name_tag("To"), alt((
map(address_list, |dt| FieldBody::Correct(dt)),
map(rescue, |r| FieldBody::Failed(r))))),
|b| Field::To(b))(input)
}
fn cc(input: &str) -> IResult<&str, Field> {
map(preceded(field_name_tag("Cc"), alt((
map(address_list, |dt| FieldBody::Correct(dt)),
map(rescue, |r| FieldBody::Failed(r))))),
|b| Field::Cc(b))(input)
}
fn bcc(input: &str) -> IResult<&str, Field> {
map(preceded(field_name_tag("Bcc"), alt((
map(opt(alt((address_list, address_list_cfws))), |dt| FieldBody::Correct(dt.unwrap_or(vec![]))),
map(rescue, |r| FieldBody::Failed(r))))),
|b| Field::Bcc(b))(input)
}
// 3.6.4. Identification Fields
fn msg_id_field(input: &str) -> IResult<&str, Field> {
map(preceded(field_name_tag("Message-ID"), alt((
map(msg_id, |dt| FieldBody::Correct(dt)),
map(rescue, |r| FieldBody::Failed(r))))),
|b| Field::MessageID(b))(input)
}
fn in_reply_to(input: &str) -> IResult<&str, Field> {
map(preceded(field_name_tag("In-Reply-To"), alt((
map(many1(msg_id), |dt| FieldBody::Correct(dt)),
map(rescue, |r| FieldBody::Failed(r))))),
|b| Field::InReplyTo(b))(input)
}
fn references(input: &str) -> IResult<&str, Field> {
map(preceded(field_name_tag("References"), alt((
map(many1(msg_id), |dt| FieldBody::Correct(dt)),
map(rescue, |r| FieldBody::Failed(r))))),
|b| Field::References(b))(input)
}
// 3.6.5. Informational Fields
fn subject(input: &str) -> IResult<&str, Field> {
map(preceded(field_name_tag("Subject"), alt((
map(unstructured, |dt| FieldBody::Correct(dt)),
map(rescue, |r| FieldBody::Failed(r))))),
|b| Field::Subject(b))(input)
}
fn comments(input: &str) -> IResult<&str, Field> {
map(preceded(field_name_tag("Comments"), alt((
map(unstructured, |dt| FieldBody::Correct(dt)),
map(rescue, |r| FieldBody::Failed(r))))),
|b| Field::Comments(b))(input)
}
fn keywords(input: &str) -> IResult<&str, Field> {
map(preceded(field_name_tag("Keywords"), alt((
map(separated_list1(tag(","), phrase), |dt| FieldBody::Correct(dt)),
map(rescue, |r| FieldBody::Failed(r))))),
|b| Field::Keywords(b))(input)
}
// 3.6.6 Resent fields // 3.6.6 Resent fields
// Not implemented // Not implemented
// 3.6.7 Trace fields // 3.6.7 Trace fields
fn return_path(input: &str) -> IResult<&str, HeaderField> { fn return_path(input: &str) -> IResult<&str, Field> {
map( map(preceded(field_name_tag("Return-Path"), alt((
preceded(pair(tag("Return-Path:"), space0), trace::return_path_body), map(trace::return_path_body, |dt| FieldBody::Correct(dt)),
|body| HeaderField::ReturnPath(body), map(rescue, |r| FieldBody::Failed(r))))),
)(input) |b| Field::ReturnPath(b))(input)
} }
fn received(input: &str) -> IResult<&str, HeaderField> { fn received(input: &str) -> IResult<&str, Field> {
map( map(preceded(field_name_tag("Received"), alt((
preceded(pair(tag("Received:"), space0), trace::received_body), map(trace::received_body, |dt| FieldBody::Correct(dt)),
|body| HeaderField::Received(body), map(rescue, |r| FieldBody::Failed(r))))),
)(input) |b| Field::Received(b))(input)
} }
/// Optional field /// Optional field
@ -284,12 +276,12 @@ fn received(input: &str) -> IResult<&str, HeaderField> {
/// %d59-126 ; characters not including /// %d59-126 ; characters not including
/// ; ":". /// ; ":".
/// ``` /// ```
fn unknown_field(input: &str) -> IResult<&str, HeaderField> { fn unknown_field(input: &str) -> IResult<&str, Field> {
// Extract field name // Extract field name
let (input, field_name) = field_name(input)?; let (input, field_name) = field_name(input)?;
let (input, body) = unstructured(input)?; let (input, body) = unstructured(input)?;
let (input, _) = perm_crlf(input)?; let (input, _) = perm_crlf(input)?;
Ok((input, HeaderField::Optional(field_name, body))) Ok((input, Field::Optional(field_name, body)))
} }
fn field_name(input: &str) -> IResult<&str, &str> { fn field_name(input: &str) -> IResult<&str, &str> {
terminated( terminated(
@ -306,68 +298,74 @@ fn field_name(input: &str) -> IResult<&str, &str> {
/// ///
/// ```abnf /// ```abnf
/// rescue = *(*any FWS) *any CRLF /// rescue = *(*any FWS) *any CRLF
fn rescue(input: &str) -> IResult<&str, HeaderField> { fn rescue(input: &str) -> IResult<&str, &str> {
map(recognize(pair( recognize(pair(
many0(pair(is_not("\r\n"), fws)), many0(pair(is_not("\r\n"), fws)),
pair(is_not("\r\n"), perm_crlf), is_not("\r\n"),
)), |x| HeaderField::Rescue(x))(input) ))(input)
} }
fn rescue_field(input: &str) -> IResult<&str, Field> {
map(terminated(rescue, perm_crlf), |r| Field::Rescue(r))(input)
}
#[cfg(test)] #[cfg(test)]
mod tests { mod tests {
use super::*; use super::*;
use crate::model::{GroupRef, AddrSpec}; use crate::model::{GroupRef, AddrSpec};
// 3.6.1. The Origination Date Field // 3.6.1. The Origination Date Field
#[test] /* #[test]
fn test_datetime() { fn test_datetime() {
let datefield = "Date: Thu,\r\n 13\r\n Feb\r\n 1969\r\n 23:32\r\n -0330 (Newfoundland Time)\r\n"; let datefield = "Date: Thu,\r\n 13\r\n Feb\r\n 1969\r\n 23:32\r\n -0330 (Newfoundland Time)\r\n";
let (input, v) = header_field(datefield).unwrap(); let (input, v) = known_field(datefield).unwrap();
assert_eq!(input, ""); assert_eq!(input, "");
match v { match v {
HeaderField::Date(HeaderDate::Parsed(_)) => (), Field::Date(HeaderDate::Parsed(_)) => (),
_ => panic!("Date has not been parsed"), _ => panic!("Date has not been parsed"),
}; };
} }*/
// 3.6.2. Originator Fields // 3.6.2. Originator Fields
#[test] #[test]
fn test_from() { fn test_from() {
assert_eq!( assert_eq!(
header_field("From: \"Joe Q. Public\" <john.q.public@example.com>\r\n"), known_field("From: \"Joe Q. Public\" <john.q.public@example.com>\r\n"),
Ok(("", HeaderField::From(vec![MailboxRef { Ok(("", Field::From(FieldBody::Correct(vec![MailboxRef {
name: Some("Joe Q. Public".into()), name: Some("Joe Q. Public".into()),
addrspec: AddrSpec { addrspec: AddrSpec {
local_part: "john.q.public".into(), local_part: "john.q.public".into(),
domain: "example.com".into(), domain: "example.com".into(),
} }
}]))), }])))),
); );
} }
#[test] #[test]
fn test_sender() { fn test_sender() {
assert_eq!( assert_eq!(
header_field("Sender: Michael Jones <mjones@machine.example>\r\n"), known_field("Sender: Michael Jones <mjones@machine.example>\r\n"),
Ok(("", HeaderField::Sender(MailboxRef { Ok(("", Field::Sender(FieldBody::Correct(MailboxRef {
name: Some("Michael Jones".into()), name: Some("Michael Jones".into()),
addrspec: AddrSpec { addrspec: AddrSpec {
local_part: "mjones".into(), local_part: "mjones".into(),
domain: "machine.example".into(), domain: "machine.example".into(),
}, },
}))), })))),
); );
} }
#[test] #[test]
fn test_reply_to() { fn test_reply_to() {
assert_eq!( assert_eq!(
header_field("Reply-To: \"Mary Smith: Personal Account\" <smith@home.example>\r\n"), known_field("Reply-To: \"Mary Smith: Personal Account\" <smith@home.example>\r\n"),
Ok(("", HeaderField::ReplyTo(vec![AddressRef::Single(MailboxRef { Ok(("", Field::ReplyTo(FieldBody::Correct(vec![AddressRef::Single(MailboxRef {
name: Some("Mary Smith: Personal Account".into()), name: Some("Mary Smith: Personal Account".into()),
addrspec: AddrSpec { addrspec: AddrSpec {
local_part: "smith".into(), local_part: "smith".into(),
domain: "home.example".into(), domain: "home.example".into(),
}, },
})]))) })]))))
); );
} }
@ -375,8 +373,8 @@ mod tests {
#[test] #[test]
fn test_to() { fn test_to() {
assert_eq!( assert_eq!(
header_field("To: A Group:Ed Jones <c@a.test>,joe@where.test,John <jdoe@one.test>;\r\n"), known_field("To: A Group:Ed Jones <c@a.test>,joe@where.test,John <jdoe@one.test>;\r\n"),
Ok(("", HeaderField::To(vec![AddressRef::Many(GroupRef { Ok(("", Field::To(FieldBody::Correct(vec![AddressRef::Many(GroupRef {
name: "A Group".into(), name: "A Group".into(),
participants: vec![ participants: vec![
MailboxRef { MailboxRef {
@ -392,28 +390,28 @@ mod tests {
addrspec: AddrSpec { local_part: "jdoe".into(), domain: "one.test".into() }, addrspec: AddrSpec { local_part: "jdoe".into(), domain: "one.test".into() },
}, },
] ]
})]))) })]))))
); );
} }
#[test] #[test]
fn test_cc() { fn test_cc() {
assert_eq!( assert_eq!(
header_field("Cc: Undisclosed recipients:;\r\n"), known_field("Cc: Undisclosed recipients:;\r\n"),
Ok(("", HeaderField::Cc(vec![AddressRef::Many(GroupRef { Ok(("", Field::Cc(FieldBody::Correct(vec![AddressRef::Many(GroupRef {
name: "Undisclosed recipients".into(), name: "Undisclosed recipients".into(),
participants: vec![], participants: vec![],
})]))) })]))))
); );
} }
#[test] #[test]
fn test_bcc() { fn test_bcc() {
assert_eq!( assert_eq!(
header_field("Bcc: (empty)\r\n"), known_field("Bcc: (empty)\r\n"),
Ok(("", HeaderField::Bcc(vec![]))) Ok(("", Field::Bcc(FieldBody::Correct(vec![]))))
); );
assert_eq!( assert_eq!(
header_field("Bcc: \r\n"), known_field("Bcc: \r\n"),
Ok(("", HeaderField::Bcc(vec![]))) Ok(("", Field::Bcc(FieldBody::Correct(vec![]))))
); );
} }
@ -422,28 +420,28 @@ mod tests {
#[test] #[test]
fn test_message_id() { fn test_message_id() {
assert_eq!( assert_eq!(
header_field("Message-ID: <310@[127.0.0.1]>\r\n"), known_field("Message-ID: <310@[127.0.0.1]>\r\n"),
Ok(("", HeaderField::MessageID(model::MessageId { left: "310", right: "127.0.0.1" }))) Ok(("", Field::MessageID(FieldBody::Correct(model::MessageId { left: "310", right: "127.0.0.1" }))))
); );
} }
#[test] #[test]
fn test_in_reply_to() { fn test_in_reply_to() {
assert_eq!( assert_eq!(
header_field("In-Reply-To: <a@b> <c@example.com>\r\n"), known_field("In-Reply-To: <a@b> <c@example.com>\r\n"),
Ok(("", HeaderField::InReplyTo(vec![ Ok(("", Field::InReplyTo(FieldBody::Correct(vec![
model::MessageId { left: "a", right: "b" }, model::MessageId { left: "a", right: "b" },
model::MessageId { left: "c", right: "example.com" }, model::MessageId { left: "c", right: "example.com" },
]))) ]))))
); );
} }
#[test] #[test]
fn test_references() { fn test_references() {
assert_eq!( assert_eq!(
header_field("References: <1234@local.machine.example> <3456@example.net>\r\n"), known_field("References: <1234@local.machine.example> <3456@example.net>\r\n"),
Ok(("", HeaderField::References(vec![ Ok(("", Field::References(FieldBody::Correct(vec![
model::MessageId { left: "1234", right: "local.machine.example" }, model::MessageId { left: "1234", right: "local.machine.example" },
model::MessageId { left: "3456", right: "example.net" }, model::MessageId { left: "3456", right: "example.net" },
]))) ]))))
); );
} }
@ -451,36 +449,54 @@ mod tests {
#[test] #[test]
fn test_subject() { fn test_subject() {
assert_eq!( assert_eq!(
header_field("Subject: Aérogramme\r\n"), known_field("Subject: Aérogramme\r\n"),
Ok(("", HeaderField::Subject("Aérogramme".into()))) Ok(("", Field::Subject(FieldBody::Correct("Aérogramme".into()))))
); );
} }
#[test] #[test]
fn test_comments() { fn test_comments() {
assert_eq!( assert_eq!(
header_field("Comments: 😛 easter egg!\r\n"), known_field("Comments: 😛 easter egg!\r\n"),
Ok(("", HeaderField::Comments("😛 easter egg!".into()))) Ok(("", Field::Comments(FieldBody::Correct("😛 easter egg!".into()))))
); );
} }
#[test] #[test]
fn test_keywords() { fn test_keywords() {
assert_eq!( assert_eq!(
header_field("Keywords: fantasque, farfelu, fanfreluche\r\n"), known_field("Keywords: fantasque, farfelu, fanfreluche\r\n"),
Ok(("", HeaderField::Keywords(vec!["fantasque".into(), "farfelu".into(), "fanfreluche".into()]))) Ok(("", Field::Keywords(FieldBody::Correct(vec!["fantasque".into(), "farfelu".into(), "fanfreluche".into()]))))
); );
} }
// Test invalid field name // Test invalid field name
#[test] #[test]
fn test_invalid_field_name() { fn test_invalid_field_name() {
assert!(header_field("Unknown: unknown\r\n").is_err()); assert!(known_field("Unknown: unknown\r\n").is_err());
} }
#[test] #[test]
fn test_rescue() { fn test_rescue_field() {
assert_eq!( assert_eq!(
rescue("Héron: élan\r\n\tnoël: test\r\n"), rescue_field("Héron: élan\r\n\tnoël: test\r\nFrom: ..."),
Ok(("", HeaderField::Rescue("Héron: élan\r\n\tnoël: test\r\n"))), Ok(("From: ...", Field::Rescue("Héron: élan\r\n\tnoël: test"))),
);
}
#[test]
fn test_wrong_fields() {
let fullmail = r#"Return-Path: xoxo
From: !!!!
Hello world"#;
assert_eq!(
section(fullmail),
Ok(("Hello world", HeaderSection {
bad_fields: vec![
Field::ReturnPath(FieldBody::Failed("xoxo")),
Field::From(FieldBody::Failed("!!!!")),
],
..Default::default()
}))
); );
} }
@ -524,7 +540,7 @@ This is a reply to your hello.
assert_eq!( assert_eq!(
section(fullmail), section(fullmail),
Ok(("This is a reply to your hello.\n", HeaderSection { Ok(("This is a reply to your hello.\n", HeaderSection {
date: model::HeaderDate::Parsed(FixedOffset::east_opt(2 * 3600).unwrap().with_ymd_and_hms(2023, 06, 13, 10, 01, 10).unwrap()), date: Some(FixedOffset::east_opt(2 * 3600).unwrap().with_ymd_and_hms(2023, 06, 13, 10, 01, 10).unwrap()),
from: vec![MailboxRef { from: vec![MailboxRef {
name: Some("Mary Smith".into()), name: Some("Mary Smith".into()),
@ -608,12 +624,12 @@ This is a reply to your hello.
]), ]),
unparsed: vec![ unparsed: vec![
"Héron: Raté\n Raté raté\n", "Héron: Raté\n Raté raté",
"Not a real header but should still recover\n", "Not a real header but should still recover",
], ],
bad_fields: vec![],
})) }))
); );
} }
} }

56
src/model.rs
View file

@ -1,14 +1,6 @@
use std::collections::HashMap; use std::collections::HashMap;
use chrono::{DateTime,FixedOffset,ParseError}; use chrono::{DateTime,FixedOffset,ParseError};
#[derive(Debug, PartialEq, Default)]
pub enum HeaderDate {
Parsed(DateTime<FixedOffset>),
Unknown(String, ParseError),
#[default]
None,
}
#[derive(Debug, PartialEq)] #[derive(Debug, PartialEq)]
pub struct AddrSpec { pub struct AddrSpec {
pub local_part: String, pub local_part: String,
@ -63,6 +55,49 @@ pub struct MessageId<'a> {
pub right: &'a str, pub right: &'a str,
} }
#[derive(Debug, PartialEq)]
pub enum FieldBody<'a, T> {
Correct(T),
Failed(&'a str),
}
#[derive(Debug, PartialEq)]
pub enum Field<'a> {
// 3.6.1. The Origination Date Field
Date(FieldBody<'a, DateTime<FixedOffset>>),
// 3.6.2. Originator Fields
From(FieldBody<'a, Vec<MailboxRef>>),
Sender(FieldBody<'a, MailboxRef>),
ReplyTo(FieldBody<'a, Vec<AddressRef>>),
// 3.6.3. Destination Address Fields
To(FieldBody<'a, Vec<AddressRef>>),
Cc(FieldBody<'a, Vec<AddressRef>>),
Bcc(FieldBody<'a, Vec<AddressRef>>),
// 3.6.4. Identification Fields
MessageID(FieldBody<'a, MessageId<'a>>),
InReplyTo(FieldBody<'a, Vec<MessageId<'a>>>),
References(FieldBody<'a, Vec<MessageId<'a>>>),
// 3.6.5. Informational Fields
Subject(FieldBody<'a, String>),
Comments(FieldBody<'a, String>),
Keywords(FieldBody<'a, Vec<String>>),
// 3.6.6 Resent Fields (not implemented)
// 3.6.7 Trace Fields
Received(FieldBody<'a, &'a str>),
ReturnPath(FieldBody<'a, Option<MailboxRef>>),
// 3.6.8. Optional Fields
Optional(&'a str, String),
// None
Rescue(&'a str),
}
/// Permissive Header Section /// Permissive Header Section
/// ///
/// This is a structure intended for parsing/decoding, /// This is a structure intended for parsing/decoding,
@ -72,7 +107,7 @@ pub struct MessageId<'a> {
#[derive(Debug, PartialEq, Default)] #[derive(Debug, PartialEq, Default)]
pub struct HeaderSection<'a> { pub struct HeaderSection<'a> {
// 3.6.1. The Origination Date Field // 3.6.1. The Origination Date Field
pub date: HeaderDate, pub date: Option<DateTime<FixedOffset>>,
// 3.6.2. Originator Fields // 3.6.2. Originator Fields
pub from: Vec<MailboxRef>, pub from: Vec<MailboxRef>,
@ -101,5 +136,8 @@ pub struct HeaderSection<'a> {
// 3.6.8. Optional Fields // 3.6.8. Optional Fields
pub optional: HashMap<&'a str, String>, pub optional: HashMap<&'a str, String>,
// Recovery
pub bad_fields: Vec<Field<'a>>,
pub unparsed: Vec<&'a str>, pub unparsed: Vec<&'a str>,
} }

4
src/parse.rs
View file

@ -1,4 +1,4 @@
use imf_codec::header; //use imf_codec::header;
fn main() { fn main() {
let hdr = r#"Return-Path: <gitlab@framasoft.org> let hdr = r#"Return-Path: <gitlab@framasoft.org>
@ -33,5 +33,5 @@ References: <1234@local.machine.example>
This is a reply to your hello. This is a reply to your hello.
"#; "#;
println!("{:?}", header::section(hdr)); //println!("{:?}", header::section(hdr));
} }