From: Aleš Mrázek <ales.mrazek@nic.cz>
Date: Mon, 11 Apr 2022 20:52:01 +0000 (+0200)
Subject: manager: datamodel: types: punycode for DomainName
X-Git-Tag: v6.0.0a1~37^2~3
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=7b8a69ca73afad796077ad5bf1d92cb11f9e87f2;p=thirdparty%2Fknot-resolver.git

manager: datamodel: types: punycode for DomainName
---

diff --git a/manager/knot_resolver_manager/datamodel/types/types.py b/manager/knot_resolver_manager/datamodel/types/types.py
index 1d2806c4c..73603f2c9 100644
--- a/manager/knot_resolver_manager/datamodel/types/types.py
+++ b/manager/knot_resolver_manager/datamodel/types/types.py
@@ -55,17 +55,57 @@ class TimeUnit(UnitBase):
         return self._value
 
 
-class DomainName(PatternBase):
-    _spec_chars = "ßàÁâãóôþüúðæåïçèõöÿýòäœêëìíøùîûñé"
+class DomainName(StrBase):
+    """
+    Fully or partially qualified domain name.
+    """
+
     _re = re.compile(
-        # max 253 chars
-        r"(?=^.{,253}$)"
-        # do not start/end with dash; 1-63 chars in name; allow special chars; max 126 levels+TLD
-        rf"^((?!-)([{_spec_chars}]|[a-zA-Z0-9-]){{1,62}}[a-zA-Z0-9]\.){{0,126}}"
-        # TLD
-        r"[a-zA-Z]{2,6}($|.$)"
+        r"(?=^.{,253}$)"  # max 253 chars
+        r"^([a-zA-Z0-9]"  # do not start with hyphen
+        r"([a-zA-Z0-9-]){1,61}"  # max 63 chars in label
+        r"[a-zA-Z0-9]\.)"  # do not end with hyphen
+        r"{0,126}"  # max 126 levels+TLD
+        r"([a-zA-Z]){2,6}($|.$)"  # TLD; end with or without '.'
     )
 
+    def __init__(self, source_value: Any, object_path: str = "/") -> None:
+        super().__init__(source_value)
+        if isinstance(source_value, str):
+            try:
+                punycode = source_value.encode("idna").decode("utf-8")
+            except ValueError:
+                raise SchemaException(
+                    f"conversion of '{source_value}' to IDN punycode representation failed",
+                    object_path,
+                )
+
+            if type(self)._re.match(punycode):
+                self._value = source_value
+            else:
+                raise SchemaException(
+                    f"'{source_value}' represented in punycode '{punycode}' does not match '{self._re.pattern}' pattern",
+                    object_path,
+                )
+        else:
+            raise SchemaException(
+                "Unexpected value for '<domain-name>'."
+                f" Expected string, got '{source_value}' with type '{type(source_value)}'",
+                object_path,
+            )
+
+    def __hash__(self) -> int:
+        if self._value.endswith("."):
+            return hash(self._value)
+        return hash(f"{self._value}.")
+
+    def punycode(self) -> bytes:
+        return self._value.encode("idna")
+
+    @classmethod
+    def json_schema(cls: Type["DomainName"]) -> Dict[Any, Any]:
+        return {"type": "string", "pattern": rf"{cls._re.pattern}"}
+
 
 class InterfaceName(PatternBase):
     _re = re.compile(r"^[a-zA-Z0-9]+(?:[-_][a-zA-Z0-9]+)*$")
diff --git a/manager/tests/unit/datamodel/types/test_custom_types.py b/manager/tests/unit/datamodel/types/test_custom_types.py
index 2a0ec429c..2130d8acf 100644
--- a/manager/tests/unit/datamodel/types/test_custom_types.py
+++ b/manager/tests/unit/datamodel/types/test_custom_types.py
@@ -83,11 +83,12 @@ def test_checked_path():
     assert str(TestSchema({"p": "/tmp"}).p) == "/tmp"
 
 
-@pytest.mark.parametrize("val", ["example.com.", "test.example.com", "test-example.com"])
+@pytest.mark.parametrize("val", ["example.com.", "test.example.com", "test-example.com", "bücher.com.", "příklad.cz"])
 def test_domain_name_valid(val: str):
     o = DomainName(val)
     assert str(o) == val
     assert o == DomainName(val)
+    assert o.punycode() == val.encode("idna")
 
 
 @pytest.mark.parametrize("val", ["test.example.com..", "-example.com", "test-.example.net"])