Validate built-in String formats

pyproject.toml

@@ -8,6 +8,8 @@ readme = "README.md"
 
 [tool.poetry.dependencies]
 python = "^3.8"
+rfc3987 = { version = "^1.3.8", optional = true }
+
 
 [tool.poetry.dev-dependencies]
 pytest = "^7.2.0"
@@ -17,6 +19,10 @@ vulture = "^2.6"
 flake8 = { version = "^6.0.0", python = ">=3.8.1" }
 pytest-cov = "^4.0.0"
 
+
+[tool.poetry.extras]
+strings = ["rfc3987"]
+
 [build-system]
 requires = ["poetry-core>=1.0.0"]
 build-backend = "poetry.core.masonry.api"

schemamodels/__init__.py

@@ -1,14 +1,19 @@
 import sys
 from dataclasses import make_dataclass, field, fields as fs
 from re import sub
+import re
 import importlib
+import ipaddress
 from operator import gt, ge, lt, le, mod, xor, not_
 from typing import Callable
+from urllib.parse import urlparse
 
 from functools import partial, reduce
 
 from schemamodels import exceptions as e, bases
 
+from datetime import datetime
+
 
 JSON_TYPE_MAP = {
     'string': lambda d: isinstance(d, str),
@@ -21,13 +26,46 @@
 
 PORCELINE_KEYWORDS = ['value', 'default', 'anyOf', 'allOf', 'oneOf', 'not']
 
+# Source: https://uibakery.io/regex-library/email-regex-python
+EMAIL_REGEX = re.compile(r"^[a-z0-9!#$%&'*+/=?^_`{|}~-]+(?:\.[a-z0-9!#$%&'*+/=?^_`{|}~-]+)*@(?:[a-z0-9](?:[a-z0-9-]*[a-z0-9])?\.)+[a-z0-9](?:[a-z0-9-]*[a-z0-9])?$")
+
+HOSTNAME_REGEX = re.compile(r'^(?P<scheme>(?:\w+(?:\+\w+)?:)//)?//(?P<reminder>.*)')
+
+
+def hostname_process(e):
+    results = HOSTNAME_REGEX.match(e)
+    if results:
+        if all(results.group('scheme'), results.group('remainder')):
+            return urlparse(e).hostname is not None
+    return False
+
+
+def ifraises(func, value, exception_class=ValueError):
+    try:
+        func(value)
+    except exception_class:
+        return False
+    return True
+
+
+STRING_FORMATS = {
+    'email': lambda e: EMAIL_REGEX.match(e) is not None,
+    'date-time': lambda e: sys.version_info >= (3, 11) and datetime.fromisoformat(e),
+    'date': lambda e: sys.version_info >= (3, 11) and datetime.fromisoformat(e),
+    'time': lambda e: sys.version_info >= (3, 11) and datetime.fromisoformat(e),
+    'ipv4': lambda e: ifraises(ipaddress.IPv4Address, e, exception_class=ipaddress.AddressValueError),
+    'ipv6': lambda e: ifraises(ipaddress.IPv6Address, e, exception_class=ipaddress.AddressValueError),
+    'hostname': lambda e: hostname_process(e),
+}
+
 COMPARISONS = {
     'type': lambda d: JSON_TYPE_MAP[d],
     'anyOf': lambda d: partial(lambda struct: generate_functors(struct), d),
     'allOf': lambda d: partial(lambda struct: generate_functors(struct), d),
     'oneOf': lambda d: partial(lambda struct: generate_functors(struct), d),
     'not': lambda d: not_(d),
     'string': lambda d: isinstance(d, str),
+    'format': lambda d: partial(lambda bound, v: STRING_FORMATS[bound](v), d),
     'integer': lambda d: isinstance(d, int),
     'number': lambda d: isinstance(d, (float, int)),
     'null': lambda d: d is None,
@@ -107,6 +145,11 @@ def constraints(dataclass_instance):
         raise e.SubSchemaFailureViolation("at least one subschema failed")
     if len([n for n in nodes if not n.get('type', True)]) > 0:
         raise e.ValueTypeViolation("incorrect type assigned to JSON property")
+    if len([n for n in nodes if not n.get('format', True)]) > 0:
+        if sys.version_info >= (3, 11):
+            raise e.StringFormatViolation("violates string format constraint")
+        else:
+            raise e.StringFormatViolation("Python versions < 3.11 does not support native ISO8601 datetime parsing")
     if len([n for n in nodes if not n.get('maximum', True)]) > 0:
         raise e.RangeConstraintViolation("violates range contraint")
     if len([n for n in nodes if not n.get('exclusiveMaximum', True)]) > 0:

schemamodels/exceptions.py

@@ -14,3 +14,6 @@ class ValueTypeViolation(SchemaViolation): pass
 
 
 class SubSchemaFailureViolation(SchemaViolation): pass
+
+
+class StringFormatViolation(SchemaViolation): pass

tests.py

@@ -1,3 +1,4 @@
+import sys
 from jsonschema import validators
 import json
 import importlib
@@ -523,3 +524,115 @@ def test_functor_generator():
     assert next(iter(fn.values()))(1)
     assert any(list(next(iter(fn.values()))(1.0)))
     assert all(list(next(iter(fn.values()))("e")))
+
+
+@pytest.mark.string
+@pytest.mark.format
+def test_string_email_format_support():
+    schemadoc = '''
+    {
+        "title": "email-format",
+        "description": "Blue Blah",
+        "type": "object",
+        "properties": {
+            "user_email": {
+              "type": "string",
+              "format": "email"
+            }
+        }
+    }
+    '''
+    stringformat = json.loads(schemadoc)
+    sm = SchemaModelFactory()
+    sm.register(stringformat)
+
+    from schemamodels.dynamic import EmailFormat
+
+    fs = EmailFormat(user_email="[email protected]")
+    with pytest.raises(exceptions.StringFormatViolation):
+        fs = EmailFormat(user_email="abcdefgh")
+
+
+@pytest.mark.string
+@pytest.mark.format
+def test_string_datetimes_format_support():
+    schemadoc = '''
+    {
+        "title": "date-format",
+        "description": "Blue Blah",
+        "type": "object",
+        "properties": {
+            "event_time": {
+              "type": "string",
+              "format": "date-time"
+            }
+        }
+    }
+    '''
+    stringformat = json.loads(schemadoc)
+    sm = SchemaModelFactory()
+    sm.register(stringformat)
+
+    from schemamodels.dynamic import DateFormat
+
+    if sys.version_info >= (3, 11):
+      fs = DateFormat(event_time="2018-11-13T20:20:39+00:00")
+      with pytest.raises(exceptions.StringFormatViolation):
+          fs = DateFormat(event_time="abcdefgh")
+    else:
+      with pytest.raises(exceptions.StringFormatViolation):
+          fs = DateFormat(event_time="2018-11-13T20:20:39+00:00")
+
+
+@pytest.mark.string
+@pytest.mark.format
+def test_string_ipaddress_format_support():
+    schemadoc = '''
+    {
+        "title": "internet-format",
+        "description": "Blue Blah",
+        "type": "object",
+        "properties": {
+            "event_location": {
+              "type": "string",
+              "format": "ipv4"
+            }
+        }
+    }
+    '''
+    stringformat = json.loads(schemadoc)
+    sm = SchemaModelFactory()
+    sm.register(stringformat)
+
+    from schemamodels.dynamic import InternetFormat
+
+    fs = InternetFormat(event_location='127.0.0.1')
+    with pytest.raises(exceptions.StringFormatViolation):
+        fs = InternetFormat(event_location='welpwelp')
+
+
+@pytest.mark.string
+@pytest.mark.format
+def test_string_hostname_format_support():
+    schemadoc = '''
+    {
+        "title": "internet-format",
+        "description": "Blue Blah",
+        "type": "object",
+        "properties": {
+            "webhost": {
+              "type": "string",
+              "format": "hostname"
+            }
+        }
+    }
+    '''
+    stringformat = json.loads(schemadoc)
+    sm = SchemaModelFactory()
+    sm.register(stringformat)
+
+    from schemamodels.dynamic import InternetFormat
+
+    fs = InternetFormat(webhost='127.0.0.1')
+    with pytest.raises(exceptions.StringFormatViolation):
+        fs = InternetFormat(webhost='qwdqwd//qwdqwdwd.com')