Improve mailbox parsing using chumsky (#839)

Cargo.toml

@@ -19,6 +19,7 @@ is-it-maintained-open-issues = { repository = "lettre/lettre" }
 maintenance = { status = "actively-developed" }
 
 [dependencies]
+chumsky = "0.8.0"
 idna = "0.3"
 once_cell = { version = "1", optional = true }
 tracing = { version = "0.1.16", default-features = false, features = ["std"], optional = true } # feature
@@ -89,6 +90,10 @@ maud = "0.24"
 harness = false
 name = "transport_smtp"
 
+[[bench]]
+harness = false
+name = "mailbox_parsing"
+
 [features]
 default = ["smtp-transport", "pool", "native-tls", "hostname", "builder"]
 builder = ["dep:httpdate", "dep:mime", "dep:fastrand", "dep:quoted_printable", "dep:email-encoding"]

README.md

@@ -91,7 +91,7 @@ let mailer = SmtpTransport::relay("smtp.gmail.com")
 // Send the email
 match mailer.send(&email) {
     Ok(_) => println!("Email sent successfully!"),
-    Err(e) => panic!("Could not send email: {:?}", e),
+    Err(e) => panic!("Could not send email: {e:?}"),
 }
 ```
 

benches/mailbox_parsing.rs

@@ -0,0 +1,27 @@
+use criterion::{black_box, criterion_group, criterion_main, Criterion};
+use lettre::message::{Mailbox, Mailboxes};
+
+fn bench_parse_single(mailbox: &str) {
+    assert!(mailbox.parse::<Mailbox>().is_ok());
+}
+
+fn bench_parse_multiple(mailboxes: &str) {
+    assert!(mailboxes.parse::<Mailboxes>().is_ok());
+}
+
+fn criterion_benchmark(c: &mut Criterion) {
+    c.bench_function("parse single mailbox", |b| {
+        b.iter(|| bench_parse_single(black_box("\"Benchmark test\" <test@mail.local>")))
+    });
+
+    c.bench_function("parse multiple mailboxes", |b| {
+        b.iter(|| {
+            bench_parse_multiple(black_box(
+                "\"Benchmark test\" <test@mail.local>, Test <test@mail.local>, <test@mail.local>, test@mail.local",
+            ))
+        })
+    });
+}
+
+criterion_group!(benches, criterion_benchmark);
+criterion_main!(benches);

src/address/types.rs

@@ -227,6 +227,7 @@ fn check_address(val: &str) -> Result<usize, AddressError> {
 }
 
 #[derive(Debug, PartialEq, Eq, Clone, Copy)]
+#[non_exhaustive]
 /// Errors in email addresses parsing
 pub enum AddressError {
     /// Missing domain or user
@@ -237,6 +238,8 @@ pub enum AddressError {
     InvalidUser,
     /// Invalid email domain
     InvalidDomain,
+    /// Invalid input found
+    InvalidInput,
 }
 
 impl Error for AddressError {}
@@ -248,6 +251,7 @@ impl Display for AddressError {
             AddressError::Unbalanced => f.write_str("Unbalanced angle bracket"),
             AddressError::InvalidUser => f.write_str("Invalid email user"),
             AddressError::InvalidDomain => f.write_str("Invalid email domain"),
+            AddressError::InvalidInput => f.write_str("Invalid input"),
         }
     }
 }

src/message/header/mailbox.rs

@@ -306,14 +306,30 @@ mod test {
     #[test]
     fn parse_multi_with_name_containing_comma() {
         let from: Vec<Mailbox> = vec![
-            "Test, test <1@example.com>".parse().unwrap(),
+            "\"Test, test\" <1@example.com>".parse().unwrap(),
-            "Test2, test2 <2@example.com>".parse().unwrap(),
+            "\"Test2, test2\" <2@example.com>".parse().unwrap(),
         ];
 
         let mut headers = Headers::new();
         headers.insert_raw(HeaderValue::new(
             HeaderName::new_from_ascii_str("From"),
-            "Test, test <1@example.com>, Test2, test2 <2@example.com>".to_owned(),
+            "\"Test, test\" <1@example.com>, \"Test2, test2\" <2@example.com>".to_owned(),
+        ));
+
+        assert_eq!(headers.get::<From>(), Some(From(from.into())));
+    }
+
+    #[test]
+    fn parse_multi_with_name_containing_double_quotes() {
+        let from: Vec<Mailbox> = vec![
+            "\"Test, test\" <1@example.com>".parse().unwrap(),
+            "\"Test2, \"test2\"\" <2@example.com>".parse().unwrap(),
+        ];
+
+        let mut headers = Headers::new();
+        headers.insert_raw(HeaderValue::new(
+            HeaderName::new_from_ascii_str("From"),
+            "\"Test, test\" <1@example.com>, \"Test2, \"test2\"\" <2@example.com>".to_owned(),
         ));
 
         assert_eq!(headers.get::<From>(), Some(From(from.into())));
@@ -324,9 +340,20 @@ mod test {
         let mut headers = Headers::new();
         headers.insert_raw(HeaderValue::new(
             HeaderName::new_from_ascii_str("From"),
-            "Test, test <1@example.com>, Test2, test2".to_owned(),
+            "\"Test, test\" <1@example.com>, \"Test2, test2\"".to_owned(),
         ));
 
         assert_eq!(headers.get::<From>(), None);
     }
+
+    #[test]
+    fn mailbox_format_address_with_angle_bracket() {
+        assert_eq!(
+            format!(
+                "{}",
+                Mailbox::new(Some("<3".into()), "i@love.example".parse().unwrap())
+            ),
+            r#""<3" <i@love.example>"#
+        );
+    }
 }

src/message/mailbox/mod.rs

@@ -1,3 +1,4 @@
+mod parsers;
 #[cfg(feature = "serde")]
 mod serde;
 mod types;

src/message/mailbox/parsers/mod.rs

@@ -0,0 +1,5 @@
+mod rfc2234;
+mod rfc2822;
+mod rfc5336;
+
+pub(crate) use rfc2822::{mailbox, mailbox_list};

src/message/mailbox/parsers/rfc2234.rs

@@ -0,0 +1,32 @@
+//! Partial parsers implementation of [RFC2234]: Augmented BNF for
+//! Syntax Specifications: ABNF.
+//!
+//! [RFC2234]: https://datatracker.ietf.org/doc/html/rfc2234
+
+use chumsky::{error::Cheap, prelude::*};
+
+// 6.1  Core Rules
+// https://datatracker.ietf.org/doc/html/rfc2234#section-6.1
+
+// ALPHA          =  %x41-5A / %x61-7A   ; A-Z / a-z
+pub(super) fn alpha() -> impl Parser<char, char, Error = Cheap<char>> {
+    filter(|c: &char| c.is_ascii_alphabetic())
+}
+
+// DIGIT          =  %x30-39
+//                        ; 0-9
+pub(super) fn digit() -> impl Parser<char, char, Error = Cheap<char>> {
+    filter(|c: &char| c.is_ascii_digit())
+}
+
+// DQUOTE         =  %x22
+//                        ; " (Double Quote)
+pub(super) fn dquote() -> impl Parser<char, char, Error = Cheap<char>> {
+    just('"')
+}
+
+// WSP            =  SP / HTAB
+//                        ; white space
+pub(super) fn wsp() -> impl Parser<char, char, Error = Cheap<char>> {
+    choice((just(' '), just('\t')))
+}

src/message/mailbox/parsers/rfc2822.rs

@@ -0,0 +1,248 @@
+//! Partial parsers implementation of [RFC2822]: Internet Message
+//! Format.
+//!
+//! [RFC2822]: https://datatracker.ietf.org/doc/html/rfc2822
+
+use chumsky::{error::Cheap, prelude::*};
+
+use super::{rfc2234, rfc5336};
+
+// 3.2.1. Primitive Tokens
+// https://datatracker.ietf.org/doc/html/rfc2822#section-3.2.1
+
+// NO-WS-CTL       =       %d1-8 /         ; US-ASCII control characters
+//                         %d11 /          ;  that do not include the
+//                         %d12 /          ;  carriage return, line feed,
+//                         %d14-31 /       ;  and white space characters
+//                         %d127
+fn no_ws_ctl() -> impl Parser<char, char, Error = Cheap<char>> {
+    filter(|c| matches!(u32::from(*c), 1..=8 | 11 | 12 | 14..=31 | 127))
+}
+
+// text            =       %d1-9 /         ; Characters excluding CR and LF
+//                         %d11 /
+//                         %d12 /
+//                         %d14-127 /
+//                         obs-text
+fn text() -> impl Parser<char, char, Error = Cheap<char>> {
+    filter(|c| matches!(u32::from(*c), 1..=9 | 11 | 12 | 14..=127))
+}
+
+// 3.2.2. Quoted characters
+// https://datatracker.ietf.org/doc/html/rfc2822#section-3.2.2
+
+// quoted-pair     =       ("\" text) / obs-qp
+fn quoted_pair() -> impl Parser<char, char, Error = Cheap<char>> {
+    just('\\').ignore_then(text())
+}
+
+// 3.2.3. Folding white space and comments
+// https://datatracker.ietf.org/doc/html/rfc2822#section-3.2.3
+
+// FWS             =       ([*WSP CRLF] 1*WSP) /   ; Folding white space
+//                         obs-FWS
+pub fn fws() -> impl Parser<char, Option<char>, Error = Cheap<char>> {
+    rfc2234::wsp()
+        .or_not()
+        .then_ignore(rfc2234::wsp().ignored().repeated())
+}
+
+// CFWS            =       *([FWS] comment) (([FWS] comment) / FWS)
+pub fn cfws() -> impl Parser<char, Option<char>, Error = Cheap<char>> {
+    // TODO: comment are not currently supported, so for now a cfws is
+    // the same as a fws.
+    fws()
+}
+
+// 3.2.4. Atom
+// https://datatracker.ietf.org/doc/html/rfc2822#section-3.2.4
+
+// atext           =       ALPHA / DIGIT / ; Any character except controls,
+//                         "!" / "#" /     ;  SP, and specials.
+//                         "$" / "%" /     ;  Used for atoms
+//                         "&" / "'" /
+//                         "*" / "+" /
+//                         "-" / "/" /
+//                         "=" / "?" /
+//                         "^" / "_" /
+//                         "`" / "{" /
+//                         "|" / "}" /
+//                         "~"
+pub(super) fn atext() -> impl Parser<char, char, Error = Cheap<char>> {
+    choice((
+        rfc2234::alpha(),
+        rfc2234::digit(),
+        filter(|c| {
+            matches!(
+                *c,
+                '!' | '#'
+                    | '$'
+                    | '%'
+                    | '&'
+                    | '\''
+                    | '*'
+                    | '+'
+                    | '-'
+                    | '/'
+                    | '='
+                    | '?'
+                    | '^'
+                    | '_'
+                    | '`'
+                    | '{'
+                    | '|'
+                    | '}'
+                    | '~'
+            )
+        }),
+        // also allow non ASCII UTF8 chars
+        rfc5336::utf8_non_ascii(),
+    ))
+}
+
+// atom            =       [CFWS] 1*atext [CFWS]
+pub(super) fn atom() -> impl Parser<char, Vec<char>, Error = Cheap<char>> {
+    cfws().chain(atext().repeated().at_least(1))
+}
+
+// dot-atom        =       [CFWS] dot-atom-text [CFWS]
+pub fn dot_atom() -> impl Parser<char, Vec<char>, Error = Cheap<char>> {
+    cfws().chain(dot_atom_text())
+}
+
+// dot-atom-text   =       1*atext *("." 1*atext)
+pub fn dot_atom_text() -> impl Parser<char, Vec<char>, Error = Cheap<char>> {
+    atext().repeated().at_least(1).chain(
+        just('.')
+            .chain(atext().repeated().at_least(1))
+            .repeated()
+            .at_least(1)
+            .flatten(),
+    )
+}
+
+// 3.2.5. Quoted strings
+// https://datatracker.ietf.org/doc/html/rfc2822#section-3.2.5
+
+// qtext           =       NO-WS-CTL /     ; Non white space controls
+//
+//                         %d33 /          ; The rest of the US-ASCII
+//                         %d35-91 /       ;  characters not including "\"
+//                         %d93-126        ;  or the quote character
+fn qtext() -> impl Parser<char, char, Error = Cheap<char>> {
+    choice((
+        filter(|c| matches!(u32::from(*c), 33 | 35..=91 | 93..=126)),
+        no_ws_ctl(),
+    ))
+}
+
+// qcontent        =       qtext / quoted-pair
+pub(super) fn qcontent() -> impl Parser<char, char, Error = Cheap<char>> {
+    choice((qtext(), quoted_pair(), rfc5336::utf8_non_ascii()))
+}
+
+// quoted-string   =       [CFWS]
+//                         DQUOTE *([FWS] qcontent) [FWS] DQUOTE
+//                         [CFWS]
+fn quoted_string() -> impl Parser<char, Vec<char>, Error = Cheap<char>> {
+    rfc2234::dquote()
+        .ignore_then(fws().chain(qcontent()).repeated().flatten())
+        .then_ignore(text::whitespace())
+        .then_ignore(rfc2234::dquote())
+}
+
+// 3.2.6. Miscellaneous tokens
+// https://datatracker.ietf.org/doc/html/rfc2822#section-3.2.6
+
+// word            =       atom / quoted-string
+fn word() -> impl Parser<char, Vec<char>, Error = Cheap<char>> {
+    choice((quoted_string(), atom()))
+}
+
+// phrase          =       1*word / obs-phrase
+fn phrase() -> impl Parser<char, Vec<char>, Error = Cheap<char>> {
+    choice((obs_phrase(), word().repeated().at_least(1).flatten()))
+}
+
+// 3.4. Address Specification
+// https://datatracker.ietf.org/doc/html/rfc2822#section-3.4
+
+// mailbox         =       name-addr / addr-spec
+pub(crate) fn mailbox() -> impl Parser<char, (Option<String>, (String, String)), Error = Cheap<char>>
+{
+    choice((name_addr(), addr_spec().map(|addr| (None, addr)))).then_ignore(end())
+}
+
+// name-addr       =       [display-name] angle-addr
+fn name_addr() -> impl Parser<char, (Option<String>, (String, String)), Error = Cheap<char>> {
+    display_name().collect().or_not().then(angle_addr())
+}
+
+// angle-addr      =       [CFWS] "<" addr-spec ">" [CFWS] / obs-angle-addr
+fn angle_addr() -> impl Parser<char, (String, String), Error = Cheap<char>> {
+    addr_spec()
+        .delimited_by(just('<').ignored(), just('>').ignored())
+        .padded()
+}
+
+// display-name    =       phrase
+fn display_name() -> impl Parser<char, Vec<char>, Error = Cheap<char>> {
+    phrase()
+}
+
+// mailbox-list    =       (mailbox *("," mailbox)) / obs-mbox-list
+pub(crate) fn mailbox_list(
+) -> impl Parser<char, Vec<(Option<String>, (String, String))>, Error = Cheap<char>> {
+    choice((name_addr(), addr_spec().map(|addr| (None, addr))))
+        .separated_by(just(',').padded())
+        .then_ignore(end())
+}
+
+// 3.4.1. Addr-spec specification
+// https://datatracker.ietf.org/doc/html/rfc2822#section-3.4.1
+
+// addr-spec       =       local-part "@" domain
+pub fn addr_spec() -> impl Parser<char, (String, String), Error = Cheap<char>> {
+    local_part()
+        .collect()
+        .then_ignore(just('@'))
+        .then(domain().collect())
+}
+
+// local-part      =       dot-atom / quoted-string / obs-local-part
+pub fn local_part() -> impl Parser<char, Vec<char>, Error = Cheap<char>> {
+    choice((dot_atom(), quoted_string(), obs_local_part()))
+}
+
+// domain          =       dot-atom / domain-literal / obs-domain
+pub fn domain() -> impl Parser<char, Vec<char>, Error = Cheap<char>> {
+    // NOTE: omitting domain-literal since it may never be used
+    choice((dot_atom(), obs_domain()))
+}
+
+// 4.1. Miscellaneous obsolete tokens
+// https://datatracker.ietf.org/doc/html/rfc2822#section-4.1
+
+// obs-phrase      =       word *(word / "." / CFWS)
+fn obs_phrase() -> impl Parser<char, Vec<char>, Error = Cheap<char>> {
+    // NOTE: the CFWS is already captured by the word, no need to add
+    // it there.
+    word().chain(
+        choice((word(), just('.').repeated().exactly(1)))
+            .repeated()
+            .flatten(),
+    )
+}
+
+// 4.4. Obsolete Addressing
+// https://datatracker.ietf.org/doc/html/rfc2822#section-4.4
+
+// obs-local-part  =       word *("." word)
+pub fn obs_local_part() -> impl Parser<char, Vec<char>, Error = Cheap<char>> {
+    word().chain(just('.').chain(word()).repeated().flatten())
+}
+
+// obs-domain      =       atom *("." atom)
+pub fn obs_domain() -> impl Parser<char, Vec<char>, Error = Cheap<char>> {
+    atom().chain(just('.').chain(atom()).repeated().flatten())
+}

src/message/mailbox/parsers/rfc5336.rs

@@ -0,0 +1,17 @@
+//! Partial parsers implementation of [RFC5336]: SMTP Extension for
+//! Internationalized Email Addresses.
+//!
+//! [RFC5336]: https://datatracker.ietf.org/doc/html/rfc5336
+
+use chumsky::{error::Cheap, prelude::*};
+
+// 3.3.  Extended Mailbox Address Syntax
+// https://datatracker.ietf.org/doc/html/rfc5336#section-3.3
+
+// UTF8-non-ascii = UTF8-2 / UTF8-3 / UTF8-4
+// UTF8-2 =  <See Section 4 of RFC 3629>
+// UTF8-3 =  <See Section 4 of RFC 3629>
+// UTF8-4 =  <See Section 4 of RFC 3629>
+pub(super) fn utf8_non_ascii() -> impl Parser<char, char, Error = Cheap<char>> {
+    filter(|c: &char| c.len_utf8() > 1)
+}

src/message/mailbox/serde.rs

@@ -198,7 +198,7 @@ mod test {
             from_str(r#""yin@dtb.com, Hei <hei@dtb.com>, Kai <kayo@example.com>""#).unwrap();
         assert_eq!(
             m,
-            "<yin@dtb.com>, Hei <hei@dtb.com>, Kai <kayo@example.com>"
+            "yin@dtb.com, Hei <hei@dtb.com>, Kai <kayo@example.com>"
                 .parse()
                 .unwrap()
         );
@@ -211,7 +211,7 @@ mod test {
                 .unwrap();
         assert_eq!(
             m,
-            "<yin@dtb.com>, Hei <hei@dtb.com>, Kai <kayo@example.com>"
+            "yin@dtb.com, Hei <hei@dtb.com>, Kai <kayo@example.com>"
                 .parse()
                 .unwrap()
         );

src/message/mailbox/types.rs

@@ -5,8 +5,10 @@ use std::{
     str::FromStr,
 };
 
+use chumsky::prelude::*;
 use email_encoding::headers::EmailWriter;
 
+use super::parsers;
 use crate::address::{Address, AddressError};
 
 /// Represents an email address with an optional name for the sender/recipient.
@@ -108,40 +110,18 @@ impl<S: Into<String>, T: Into<String>> TryFrom<(S, T)> for Mailbox {
     }
 }
 
-/*
-impl<S: AsRef<&str>, T: AsRef<&str>> TryFrom<(S, T)> for Mailbox {
-    type Error = AddressError;
-
-    fn try_from(header: (S, T)) -> Result<Self, Self::Error> {
-        let (name, address) = header;
-        Ok(Mailbox::new(Some(name.as_ref()), address.as_ref().parse()?))
-    }
-}*/
-
 impl FromStr for Mailbox {
     type Err = AddressError;
 
     fn from_str(src: &str) -> Result<Mailbox, Self::Err> {
-        match (src.find('<'), src.find('>')) {
+        let (name, (user, domain)) = parsers::mailbox().parse(src).map_err(|_errs| {
-            (Some(addr_open), Some(addr_close)) if addr_open < addr_close => {
+            // TODO: improve error management
-                let name = src.split_at(addr_open).0;
+            AddressError::InvalidInput
-                let addr_open = addr_open + 1;
+        })?;
-                let addr = src.split_at(addr_open).1.split_at(addr_close - addr_open).0;
+
-                let addr = addr.parse()?;
+        let mailbox = Mailbox::new(name, Address::new(user, domain)?);
-                let name = name.trim();
+
-                let name = if name.is_empty() {
+        Ok(mailbox)
-                    None
-                } else {
-                    Some(name.into())
-                };
-                Ok(Mailbox::new(name, addr))
-            }
-            (Some(_), _) => Err(AddressError::Unbalanced),
-            _ => {
-                let addr = src.parse()?;
-                Ok(Mailbox::new(None, addr))
-            }
-        }
     }
 }
 
@@ -356,34 +336,16 @@ impl Display for Mailboxes {
 impl FromStr for Mailboxes {
     type Err = AddressError;
 
-    fn from_str(mut src: &str) -> Result<Self, Self::Err> {
+    fn from_str(src: &str) -> Result<Self, Self::Err> {
         let mut mailboxes = Vec::new();
 
-        if !src.is_empty() {
+        let parsed_mailboxes = parsers::mailbox_list().parse(src).map_err(|_errs| {
-            // n-1 elements
+            // TODO: improve error management
-            let mut skip = 0;
+            AddressError::InvalidInput
-            while let Some(i) = src[skip..].find(',') {
+        })?;
-                let left = &src[..skip + i];
 
-                match left.trim().parse() {
+        for (name, (user, domain)) in parsed_mailboxes {
-                    Ok(mailbox) => {
+            mailboxes.push(Mailbox::new(name, Address::new(user, domain)?))
-                        mailboxes.push(mailbox);
-
-                        src = &src[left.len() + ",".len()..];
-                        skip = 0;
-                    }
-                    Err(AddressError::MissingParts) => {
-                        skip = left.len() + ",".len();
-                    }
-                    Err(err) => {
-                        return Err(err);
-                    }
-                }
-            }
-
-            // last element
-            let mailbox = src.trim().parse()?;
-            mailboxes.push(mailbox);
         }
 
         Ok(Mailboxes(mailboxes))

src/transport/mod.rs

@@ -75,7 +75,7 @@
 //! // Send the email
 //! match mailer.send(&email) {
 //!     Ok(_) => println!("Email sent successfully!"),
-//!     Err(e) => panic!("Could not send email: {:?}", e),
+//!     Err(e) => panic!("Could not send email: {e:?}"),
 //! }
 //! # Ok(())
 //! # }

src/transport/smtp/extension.rs

@@ -281,7 +281,7 @@ impl Display for RcptParameter {
             RcptParameter::Other {
                 ref keyword,
                 value: Some(ref value),
-            } => write!(f, "{}={}", keyword, XText(value)),
+            } => write!(f, "{keyword}={}", XText(value)),
             RcptParameter::Other {
                 ref keyword,
                 value: None,