From b6cfb432d4b36c57c7a0a5ff1989bb206c0101aa Mon Sep 17 00:00:00 2001
From: Jovial Joe Jayarson <jovial7joe@hotmail.com>
Date: Sun, 31 Mar 2024 08:09:47 +0530
Subject: [PATCH] fix: domain validation is now more consistent across rfcs

---
 src/validators/domain.py | 23 ++++++++++++++---------
 src/validators/utils.py  |  2 +-
 2 files changed, 15 insertions(+), 10 deletions(-)

diff --git a/src/validators/domain.py b/src/validators/domain.py
index c5eea7d4..540234e2 100644
--- a/src/validators/domain.py
+++ b/src/validators/domain.py
@@ -24,32 +24,37 @@ def domain(value: str, /, *, rfc_1034: bool = False, rfc_2782: bool = False):
         value:
             Domain string to validate.
         rfc_1034:
-            Allow trailing dot in domain name.
+            Allows optional trailing dot in the domain name.
             Ref: [RFC 1034](https://www.rfc-editor.org/rfc/rfc1034).
         rfc_2782:
             Domain name is of type service record.
+            Allows optional underscores in the domain name.
             Ref: [RFC 2782](https://www.rfc-editor.org/rfc/rfc2782).
 
 
     Returns:
         (Literal[True]): If `value` is a valid domain name.
         (ValidationError): If `value` is an invalid domain name.
+
+    Raises:
+        (UnicodeError): If `value` cannot be encoded into `idna` or decoded into `utf-8`.
     """
     if not value:
         return False
     try:
         return not re.search(r"\s", value) and re.match(
             # First character of the domain
-            rf"^(?:[a-zA-Z0-9{'_'if rfc_2782 else ''}]"
-            # Sub domain + hostname
-            + rf"(?:[a-zA-Z0-9-{'_'if rfc_2782 else ''}]{{0,61}}"
-            + rf"[A-Za-z0-9{'_'if rfc_2782 else ''}])?\.)"
+            rf"^(?:[a-z0-9{r'_?'if rfc_2782 else ''}]"
+            # Sub-domain
+            + rf"(?:[a-z0-9-{r'_?'if rfc_2782 else ''}]{{0,61}}"
+            # Hostname
+            + rf"[a-z0-9{r'_?'if rfc_2782 else ''}])?\.)"
             # First 61 characters of the gTLD
-            + r"+[A-Za-z0-9][A-Za-z0-9-_]{0,61}"
+            + r"+[a-z0-9][a-z0-9-_]{0,61}"
             # Last character of the gTLD
-            + rf"[A-Za-z]{r'.$' if rfc_1034 else r'$'}",
+            + rf"[a-z]{r'.?$' if rfc_1034 else r'$'}",
             value.encode("idna").decode("utf-8"),
             re.IGNORECASE,
         )
-    except UnicodeError:
-        return False
+    except UnicodeError as err:
+        raise UnicodeError(f"Unable to encode/decode {value}") from err
diff --git a/src/validators/utils.py b/src/validators/utils.py
index 80af7893..639de834 100644
--- a/src/validators/utils.py
+++ b/src/validators/utils.py
@@ -91,7 +91,7 @@ def wrapper(*args: Any, **kwargs: Any):
                     if func(*args, **kwargs)
                     else ValidationError(func, _func_args_as_dict(func, *args, **kwargs))
                 )
-        except (ValueError, TypeError) as exp:
+        except (ValueError, TypeError, UnicodeError) as exp:
             if raise_validation_error:
                 raise ValidationError(
                     func, _func_args_as_dict(func, *args, **kwargs), str(exp)