maint: improve domain, email & hostname

yozachar · yozachar · commit 95e70ec2a47f · 2023-03-17T09:34:50.000+05:30
- in `domain`: decode value to `utf-8` instead of `ascii`; use `re.match` directly
- in `email`: split `ip_address` to `ipv4` and `ipv6` parameters
- in `hostname`: split `skip_ip_addr` to skip `ipv4` &amp; `ipv6` parameters
diff --git a/validators/domain.py b/validators/domain.py
@@ -1,59 +1,61 @@
-"""Domain."""
-# -*- coding: utf-8 -*-
-
-# standard
-import re
-
-# local
-from .utils import validator
-
-
-@validator
-def domain(value: str, /, *, rfc_1034: bool = False, rfc_2782: bool = False):
-    """Return whether or not given value is a valid domain.
-
-    Examples:
-        >>> domain('example.com')
-        # Output: True
-        >>> domain('example.com/')
-        # Output: ValidationFailure(func=domain, ...)
-        >>> # Supports IDN domains as well::
-        >>> domain('xn----gtbspbbmkef.xn--p1ai')
-        # Output: True
-
-    Args:
-        value:
-            Domain string to validate.
-        rfc_1034:
-            Allow trailing dot in domain name.
-            Ref: [RFC 1034](https://www.rfc-editor.org/rfc/rfc1034).
-        rfc_2782:
-            Domain name is of type service record.
-            Ref: [RFC 2782](https://www.rfc-editor.org/rfc/rfc2782).
-
-
-    Returns:
-        (Literal[True]):
-            If `value` is a valid domain name.
-        (ValidationFailure):
-            If `value` is an invalid domain name.
-
-    Note:
-        - *In version 0.10.0*:
-            - Added support for internationalized domain name (IDN) validation.
-
-    > *New in version 0.9.0*.
-    """
-    try:
-        return not re.search(r"\s", value) and re.compile(
-            # First character of the domain
-            rf"^(?:[a-zA-Z0-9{'_'if rfc_2782 else ''}]"
-            # Sub domain + hostname
-            + r"(?:[a-zA-Z0-9-_]{0,61}[A-Za-z0-9])?\.)"
-            # First 61 characters of the gTLD
-            + r"+[A-Za-z0-9][A-Za-z0-9-_]{0,61}"
-            # Last character of the gTLD
-            + rf"[A-Za-z]{r'.$' if rfc_1034 else r'$'}"
-        ).match(value.encode("idna").decode("ascii"))
-    except UnicodeError:
-        return False
+"""Domain."""
+# -*- coding: utf-8 -*-
+
+# standard
+import re
+
+# local
+from .utils import validator
+
+
+@validator
+def domain(value: str, /, *, rfc_1034: bool = False, rfc_2782: bool = False):
+    """Return whether or not given value is a valid domain.
+
+    Examples:
+        >>> domain('example.com')
+        # Output: True
+        >>> domain('example.com/')
+        # Output: ValidationFailure(func=domain, ...)
+        >>> # Supports IDN domains as well::
+        >>> domain('xn----gtbspbbmkef.xn--p1ai')
+        # Output: True
+
+    Args:
+        value:
+            Domain string to validate.
+        rfc_1034:
+            Allow trailing dot in domain name.
+            Ref: [RFC 1034](https://www.rfc-editor.org/rfc/rfc1034).
+        rfc_2782:
+            Domain name is of type service record.
+            Ref: [RFC 2782](https://www.rfc-editor.org/rfc/rfc2782).
+
+
+    Returns:
+        (Literal[True]):
+            If `value` is a valid domain name.
+        (ValidationFailure):
+            If `value` is an invalid domain name.
+
+    Note:
+        - *In version 0.10.0*:
+            - Added support for internationalized domain name (IDN) validation.
+
+    > *New in version 0.9.0*.
+    """
+    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
+            + r"(?:[a-zA-Z0-9-_]{0,61}[A-Za-z0-9])?\.)"
+            # First 61 characters of the gTLD
+            + r"+[A-Za-z0-9][A-Za-z0-9-_]{0,61}"
+            # Last character of the gTLD
+            + rf"[A-Za-z]{r'.$' if rfc_1034 else r'$'}",
+            value.encode("idna").decode("utf-8"),
+            re.IGNORECASE,
+        )
+    except UnicodeError:
+        return False
diff --git a/validators/email.py b/validators/email.py
@@ -1,96 +1,100 @@
-"""eMail."""
-# -*- coding: utf-8 -*-
-
-# standard
-import re
-
-# local
-from .hostname import hostname
-from .utils import validator
-
-
-@validator
-def email(
-    value: str,
-    /,
-    *,
-    simple_host: bool = False,
-    ip_address: bool = False,
-    rfc_1034: bool = False,
-    rfc_2782: bool = False,
-):
-    """Validate an email address.
-
-    This was inspired from [Django's email validator][1].
-    Also ref: [RFC 1034][2], [RFC 5321][3] and [RFC 5322][4].
-
-    [1]: https://github.com/django/django/blob/main/django/core/validators.py#L174
-    [2]: https://www.rfc-editor.org/rfc/rfc1034
-    [3]: https://www.rfc-editor.org/rfc/rfc5321
-    [4]: https://www.rfc-editor.org/rfc/rfc5322
-
-    Examples:
-        >>> email('someone@example.com')
-        # Output: True
-        >>> email('bogus@@')
-        # Output: ValidationFailure(email=email, args={'value': 'bogus@@'})
-
-    Args:
-        value:
-            eMail string to validate.
-        simple_host:
-            When the domain part is a simple hostname.
-        ip_address:
-            When the domain part is an IP address.
-        rfc_1034:
-            Allow trailing dot in domain name.
-            Ref: [RFC 1034](https://www.rfc-editor.org/rfc/rfc1034).
-        rfc_2782:
-            Domain name is of type service record.
-            Ref: [RFC 2782](https://www.rfc-editor.org/rfc/rfc2782).
-
-    Returns:
-        (Literal[True]):
-            If `value` is a valid eMail.
-        (ValidationFailure):
-            If `value` is an invalid eMail.
-
-    > *New in version 0.1.0*.
-    """
-    if not value or value.count("@") != 1:
-        return False
-
-    username_part, domain_part = value.rsplit("@", 1)
-
-    if len(username_part) > 64 or len(domain_part) > 253:
-        # ref: RFC 1034 and 5231
-        return False
-
-    if ip_address:
-        if domain_part.startswith("[") and domain_part.endswith("]"):
-            # ref: RFC 5321
-            domain_part = domain_part.lstrip("[").rstrip("]")
-        else:
-            return False
-
-    return (
-        bool(
-            hostname(
-                domain_part,
-                skip_ip_addr=not ip_address,
-                may_have_port=False,
-                maybe_simple=simple_host,
-                rfc_1034=rfc_1034,
-                rfc_2782=rfc_2782,
-            )
-        )
-        if re.match(
-            # dot-atom
-            r"(^[-!#$%&'*+/=?^_`{}|~0-9A-Z]+(\.[-!#$%&'*+/=?^_`{}|~0-9A-Z]+)*$"
-            # quoted-string
-            + r'|^"([\001-\010\013\014\016-\037!#-\[\]-\177]|\\[\001-\011\013\014\016-\177])*"$)',
-            username_part,
-            re.IGNORECASE,
-        )
-        else False
-    )
+"""eMail."""
+# -*- coding: utf-8 -*-
+
+# standard
+import re
+
+# local
+from .hostname import hostname
+from .utils import validator
+
+
+@validator
+def email(
+    value: str,
+    /,
+    *,
+    ipv6_address: bool = False,
+    ipv4_address: bool = False,
+    simple_host: bool = False,
+    rfc_1034: bool = False,
+    rfc_2782: bool = False,
+):
+    """Validate an email address.
+
+    This was inspired from [Django's email validator][1].
+    Also ref: [RFC 1034][2], [RFC 5321][3] and [RFC 5322][4].
+
+    [1]: https://github.com/django/django/blob/main/django/core/validators.py#L174
+    [2]: https://www.rfc-editor.org/rfc/rfc1034
+    [3]: https://www.rfc-editor.org/rfc/rfc5321
+    [4]: https://www.rfc-editor.org/rfc/rfc5322
+
+    Examples:
+        >>> email('someone@example.com')
+        # Output: True
+        >>> email('bogus@@')
+        # Output: ValidationFailure(email=email, args={'value': 'bogus@@'})
+
+    Args:
+        value:
+            eMail string to validate.
+        ipv6_address:
+            When the domain part is an IPv6 address.
+        ipv4_address:
+            When the domain part is an IPv4 address.
+        simple_host:
+            When the domain part is a simple hostname.
+        rfc_1034:
+            Allow trailing dot in domain name.
+            Ref: [RFC 1034](https://www.rfc-editor.org/rfc/rfc1034).
+        rfc_2782:
+            Domain name is of type service record.
+            Ref: [RFC 2782](https://www.rfc-editor.org/rfc/rfc2782).
+
+    Returns:
+        (Literal[True]):
+            If `value` is a valid eMail.
+        (ValidationFailure):
+            If `value` is an invalid eMail.
+
+    > *New in version 0.1.0*.
+    """
+    if not value or value.count("@") != 1:
+        return False
+
+    username_part, domain_part = value.rsplit("@", 1)
+
+    if len(username_part) > 64 or len(domain_part) > 253:
+        # ref: RFC 1034 and 5231
+        return False
+
+    if ipv6_address or ipv4_address:
+        if domain_part.startswith("[") and domain_part.endswith("]"):
+            # ref: RFC 5321
+            domain_part = domain_part.lstrip("[").rstrip("]")
+        else:
+            return False
+
+    return (
+        bool(
+            hostname(
+                domain_part,
+                skip_ipv6_addr=not ipv6_address,
+                skip_ipv4_addr=not ipv4_address,
+                may_have_port=False,
+                maybe_simple=simple_host,
+                rfc_1034=rfc_1034,
+                rfc_2782=rfc_2782,
+            )
+        )
+        if re.match(
+            # dot-atom
+            r"(^[-!#$%&'*+/=?^_`{}|~0-9A-Z]+(\.[-!#$%&'*+/=?^_`{}|~0-9A-Z]+)*$"
+            # quoted-string
+            + r'|^"([\001-\010\013\014\016-\037!#-\[\]-\177]|\\[\001-\011\013\014\016-\177])*"$)',
+            username_part,
+            re.IGNORECASE,
+        )
+        else False
+    )
diff --git a/validators/hostname.py b/validators/hostname.py
