Improve mailbox parsing using chumsky

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
@@ -88,6 +89,10 @@ maud = "0.24"
 harness = false
 name = "transport_smtp"
 
+[[bench]]
+harness = false
+name = "mailbox_parsing"
+
 [features]
 default = ["smtp-transport", "pool", "native-tls", "hostname", "builder"]
 builder = ["httpdate", "mime", "fastrand", "quoted_printable", "email-encoding"]

README.md

@@ -89,7 +89,7 @@ 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);

examples/asyncstd1_smtp_starttls.rs

@@ -27,6 +27,6 @@ async fn main() {
     // Send the email
     match mailer.send(email).await {
         Ok(_) => println!("Email sent successfully!"),
-        Err(e) => panic!("Could not send email: {:?}", e),
+        Err(e) => panic!("Could not send email: {e:?}"),
     }
 }

examples/asyncstd1_smtp_tls.rs

@@ -27,6 +27,6 @@ async fn main() {
     // Send the email
     match mailer.send(email).await {
         Ok(_) => println!("Email sent successfully!"),
-        Err(e) => panic!("Could not send email: {:?}", e),
+        Err(e) => panic!("Could not send email: {e:?}"),
     }
 }

examples/smtp.rs

@@ -17,6 +17,6 @@ fn main() {
     // 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:?}"),
     }
 }

examples/smtp_selfsigned.rs

@@ -39,6 +39,6 @@ fn main() {
     // 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:?}"),
     }
 }

examples/smtp_starttls.rs

@@ -22,6 +22,6 @@ fn main() {
     // 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:?}"),
     }
 }

examples/smtp_tls.rs

@@ -22,6 +22,6 @@ fn main() {
     // 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:?}"),
     }
 }

examples/tokio1_smtp_starttls.rs

@@ -31,6 +31,6 @@ async fn main() {
     // Send the email
     match mailer.send(email).await {
         Ok(_) => println!("Email sent successfully!"),
-        Err(e) => panic!("Could not send email: {:?}", e),
+        Err(e) => panic!("Could not send email: {e:?}"),
     }
 }

examples/tokio1_smtp_tls.rs

@@ -31,6 +31,6 @@ async fn main() {
     // Send the email
     match mailer.send(email).await {
         Ok(_) => println!("Email sent successfully!"),
-        Err(e) => panic!("Could not send email: {:?}", e),
+        Err(e) => panic!("Could not send email: {e:?}"),
     }
 }

src/address/types.rs

@@ -184,7 +184,7 @@ where
         let domain = domain.as_ref();
         Address::check_domain(domain)?;
 
-        let serialized = format!("{}@{}", user, domain);
+        let serialized = format!("{user}@{domain}");
         Ok(Address {
             serialized,
             at_start: user.len(),
@@ -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/attachment.rs

@@ -96,7 +96,7 @@ impl Attachment {
                 builder.header(header::ContentDisposition::attachment(&filename))
             }
             Disposition::Inline(content_id) => builder
-                .header(header::ContentId::from(format!("<{}>", content_id)))
+                .header(header::ContentId::from(format!("<{content_id}>")))
                 .header(header::ContentDisposition::inline()),
         };
         builder = builder.header(content_type);

src/message/dkim.rs

@@ -588,7 +588,7 @@ cJ5Ku0OTwRtSMaseRPX+T4EfG1Caa/eunPPN4rh+CSup2BVVarOT
         );
         let signed = message.formatted();
         let signed = std::str::from_utf8(&signed).unwrap();
-        println!("{}", signed);
+        println!("{signed}");
         assert_eq!(
             signed,
             std::concat!(

src/message/header/content_disposition.rs

@@ -33,7 +33,7 @@ impl ContentDisposition {
     }
 
     fn with_name(kind: &str, file_name: &str) -> Self {
-        let raw_value = format!("{}; filename=\"{}\"", kind, file_name);
+        let raw_value = format!("{kind}; filename=\"{file_name}\"");
 
         let mut encoded_value = String::new();
         let line_len = "Content-Disposition: ".len();
@@ -90,12 +90,12 @@ mod test {
 
         headers.set(ContentDisposition::inline());
 
-        assert_eq!(format!("{}", headers), "Content-Disposition: inline\r\n");
+        assert_eq!(format!("{headers}"), "Content-Disposition: inline\r\n");
 
         headers.set(ContentDisposition::attachment("something.txt"));
 
         assert_eq!(
-            format!("{}", headers),
+            format!("{headers}"),
             "Content-Disposition: attachment; filename=\"something.txt\"\r\n"
         );
     }

src/message/header/content_type.rs

@@ -135,8 +135,7 @@ mod serde {
                     match ContentType::parse(mime) {
                         Ok(content_type) => Ok(content_type),
                         Err(_) => Err(E::custom(format!(
-                            "Couldn't parse the following MIME-Type: {}",
-                            mime
+                            "Couldn't parse the following MIME-Type: {mime}"
                         ))),
                     }
                 }

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(),
-            "Test2, test2 <2@example.com>".parse().unwrap(),
+            "\"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_string(),
+            "\"Test, test\" <1@example.com>, \"Test2, test2\" <2@example.com>".to_string(),
+        ));
+
+        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_string(),
         ));
 
         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_string(),
+            "\"Test, test\" <1@example.com>, \"Test2, test2\"".to_string(),
         ));
 
         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')))
+}