Add one-off error messages for full-width-at and small-commercial-at which are not accepted for the @-sign

JoshData · JoshData · commit bc08faa2a74b · 2024-11-25T08:58:33.000-05:00
diff --git a/email_validator/syntax.py b/email_validator/syntax.py
@@ -83,7 +83,28 @@ def split_string_at_unquoted_special(text: str, specials: Tuple[str, ...]) -> Tu
             else:
                 left_part += c
 
+        # No special symbol found. The special symbols always
+        # include an at-sign, so this always indicates a missing
+        # at-sign. The other symbol is optional.
         if len(left_part) == len(text):
+            # The full-width at-sign might occur in CJK contexts.
+            # We can't accept it because we only accept addresess
+            # that are actually valid. But if this is common we
+            # may want to consider accepting and normalizing full-
+            # width characters for the other special symbols (and
+            # full-width dot is already accepted in internationalized
+            # domains) with a new option.
+            # See https://news.ycombinator.com/item?id=42235268.
+            if "＠" in text:
+                raise EmailSyntaxError("The email address has the \"full-width\" at-sign (@) character instead of a regular at-sign.")
+
+            # Check another near-homoglyph for good measure because
+            # homoglyphs in place of required characters could be
+            # very confusing. We may want to consider checking for
+            # homoglyphs anywhere we look for a special symbol.
+            if "﹫" in text:
+                raise EmailSyntaxError('The email address has the "small commercial at" character instead of a regular at-sign.')
+
             raise EmailSyntaxError("An email address must have an @-sign.")
 
         # The right part is whatever is left.
diff --git a/tests/test_syntax.py b/tests/test_syntax.py
@@ -304,6 +304,16 @@ def test_email_valid(email_input: str, output: ValidatedEmail) -> None:
                 normalized='\u1E69@nfc.tld',
             ),
         ),
+        (
+            '＠@fullwidth.at',
+            MakeValidatedEmail(
+                local_part='＠',
+                smtputf8=True,
+                ascii_domain='fullwidth.at',
+                domain='fullwidth.at',
+                normalized='＠@fullwidth.at',
+            ),
+        ),
     ],
 )
 def test_email_valid_intl_local_part(email_input: str, output: ValidatedEmail) -> None:
@@ -363,6 +373,8 @@ def test_domain_literal() -> None:
     'email_input,error_msg',
     [
         ('hello.world', 'An email address must have an @-sign.'),
+        ('hello＠world', 'The email address has the "full-width" at-sign (@) character instead of a regular at-sign.'),
+        ('hello﹫world', 'The email address has the "small commercial at" character instead of a regular at-sign.'),
         ('my@localhost', 'The part after the @-sign is not valid. It should have a period.'),
         ('my@.leadingdot.com', 'An email address cannot have a period immediately after the @-sign.'),
         ('my@．leadingfwdot.com', 'An email address cannot have a period immediately after the @-sign.'),
