Merge pull request #21 from deltachat/fix-email-ate-last-dot

fix: do not parse last dot for email addresses

CHANGELOG.md

@@ -2,9 +2,15 @@
 
 ## Unreleased
 
+## Changed
+
 - update rust toolchain to `1.60.0`
 - enable more clippy lints to prevent panics
 
+## Fixed
+
+- fix: do not parse last dot for email addresses #19
+
 ## 0.3.0 - Squashing Link Bugs
 
 ## Changed

src/parser/parse_from_text/text_elements.rs

@@ -34,7 +34,10 @@ fn hashtag(input: &str) -> IResult<&str, Element, CustomError<&str>> {
 }
 
 fn not_email_address_part_char(c: char) -> bool {
-    matches!(c, '@' | '\n' | '\r' | '\t' | ' ' | ':')
+    matches!(
+        c,
+        '@' | '\n' | '\r' | '\t' | ' ' | ':' | ';' | '!' | '?' | ','
+    )
 }
 
 fn email_address_part_char(c: char) -> bool {
@@ -50,7 +53,27 @@ fn email_intern(input: &str) -> IResult<&str, (), CustomError<&str>> {
 }
 
 fn email_address(input: &str) -> IResult<&str, Element, CustomError<&str>> {
-    let (input, content) = recognize(email_intern)(input)?;
+    // basically
+    // let (input, content) = recognize(email_intern)(input)?;
+    // but don't eat the last char if it is a dot.
+    let i = <&str>::clone(&input);
+    let i2 = <&str>::clone(&input);
+    let i3 = <&str>::clone(&input);
+    let (input, content) = match email_intern(i) {
+        Ok((remaining, _)) => {
+            let index = i2.offset(remaining);
+            let consumed = i2.slice(..index);
+            if let Some('.') = consumed.chars().last() {
+                let index = input.offset(remaining).saturating_sub(1);
+                let consumed = i3.slice(..index);
+                let remaining = input.slice(index..);
+                Ok((remaining, consumed))
+            } else {
+                Ok((remaining, consumed))
+            }
+        }
+        Err(e) => Err(e),
+    }?;
     // check if result is valid email
     if true {
         Ok((input, Element::EmailAddress(content)))

tests/text_to_ast/text_only.rs

@@ -172,6 +172,62 @@ fn email_address_example() {
     );
 }
 
+#[test]
+fn email_address_do_not_parse_last_dot() {
+    assert_eq!(
+        parse_only_text("you can reach me on [email protected]."),
+        vec![
+            Text("you can reach me on "),
+            EmailAddress("[email protected]"),
+            Text(".")
+        ]
+    );
+}
+
+#[test]
+fn email_address_do_not_parse_last_char_if_special() {
+    assert_eq!(
+        parse_only_text("you can reach me on [email protected]!"),
+        vec![
+            Text("you can reach me on "),
+            EmailAddress("[email protected]"),
+            Text("!")
+        ]
+    );
+    assert_eq!(
+        parse_only_text("you can reach me on [email protected]?"),
+        vec![
+            Text("you can reach me on "),
+            EmailAddress("[email protected]"),
+            Text("?")
+        ]
+    );
+    assert_eq!(
+        parse_only_text("you can reach me on [email protected],"),
+        vec![
+            Text("you can reach me on "),
+            EmailAddress("[email protected]"),
+            Text(",")
+        ]
+    );
+    assert_eq!(
+        parse_only_text("you can reach me on [email protected]:"),
+        vec![
+            Text("you can reach me on "),
+            EmailAddress("[email protected]"),
+            Text(":")
+        ]
+    );
+    assert_eq!(
+        parse_only_text("you can reach me on [email protected];"),
+        vec![
+            Text("you can reach me on "),
+            EmailAddress("[email protected]"),
+            Text(";")
+        ]
+    );
+}
+
 #[test]
 fn link() {
     let test_cases = vec![