Merge pull request #42 from HideyoshiNakazone/improvement/better-string-validations

(improvement): Adds More Type Formats to String Parser

jambo/parser/string_type_parser.py

@@ -1,10 +1,12 @@
 from jambo.parser._type_parser import GenericTypeParser
 from jambo.types.type_parser_options import TypeParserOptions
 
-from pydantic import EmailStr, HttpUrl, IPvAnyAddress
+from pydantic import AnyUrl, EmailStr
 from typing_extensions import Unpack
 
 from datetime import date, datetime, time, timedelta
+from ipaddress import IPv4Address, IPv6Address
+from uuid import UUID
 
 
 class StringTypeParser(GenericTypeParser):
@@ -20,15 +22,22 @@ class StringTypeParser(GenericTypeParser):
     }
 
     format_type_mapping = {
-        "email": EmailStr,
+        # 7.3.1. Dates, Times, and Duration
-        "uri": HttpUrl,
-        "ipv4": IPvAnyAddress,
-        "ipv6": IPvAnyAddress,
-        "hostname": str,
         "date": date,
         "time": time,
         "date-time": datetime,
         "duration": timedelta,
+        # 7.3.2. Email Addresses
+        "email": EmailStr,
+        # 7.3.3. Hostnames
+        "hostname": str,
+        # 7.3.4. IP Addresses
+        "ipv4": IPv4Address,
+        "ipv6": IPv6Address,
+        # 7.3.5. Resource Identifiers
+        "uri": AnyUrl,
+        # "iri" # Not supported by pydantic and currently not supported by jambo
+        "uuid": UUID,
     }
 
     format_pattern_mapping = {

tests/parser/test_string_type_parser.py

@@ -1,9 +1,11 @@
 from jambo.parser import StringTypeParser
 
-from pydantic import EmailStr, HttpUrl, IPvAnyAddress
+from pydantic import AnyUrl, EmailStr
 
 from datetime import date, datetime, time, timedelta
+from ipaddress import IPv4Address, IPv6Address
 from unittest import TestCase
+from uuid import UUID
 
 
 class TestStringTypeParser(TestCase):
@@ -111,12 +113,14 @@ class TestStringTypeParser(TestCase):
 
         type_parsing, type_validator = parser.from_properties("placeholder", properties)
 
-        self.assertEqual(type_parsing, HttpUrl)
+        self.assertEqual(type_parsing, AnyUrl)
 
     def test_string_parser_with_ip_formats(self):
         parser = StringTypeParser()
 
-        for ip_format in ["ipv4", "ipv6"]:
+        formats = {"ipv4": IPv4Address, "ipv6": IPv6Address}
+
+        for ip_format, expected_type in formats.items():
             properties = {
                 "type": "string",
                 "format": ip_format,
@@ -126,7 +130,19 @@ class TestStringTypeParser(TestCase):
                 "placeholder", properties
             )
 
-            self.assertEqual(type_parsing, IPvAnyAddress)
+            self.assertEqual(type_parsing, expected_type)
+
+    def test_string_parser_with_uuid_format(self):
+        parser = StringTypeParser()
+
+        properties = {
+            "type": "string",
+            "format": "uuid",
+        }
+
+        type_parsing, type_validator = parser.from_properties("placeholder", properties)
+
+        self.assertEqual(type_parsing, UUID)
 
     def test_string_parser_with_time_format(self):
         parser = StringTypeParser()

tests/test_schema_converter.py

@@ -1,9 +1,10 @@
 from jambo import SchemaConverter
 
-from pydantic import BaseModel, HttpUrl
+from pydantic import AnyUrl, BaseModel
 
 from ipaddress import IPv4Address, IPv6Address
 from unittest import TestCase
+from uuid import UUID
 
 
 def is_pydantic_model(cls):
@@ -463,7 +464,7 @@ class TestSchemaConverter(TestCase):
         }
         model = SchemaConverter.build(schema)
         self.assertEqual(
-            model(website="https://example.com").website, HttpUrl("https://example.com")
+            model(website="https://example.com").website, AnyUrl("https://example.com")
         )
         with self.assertRaises(ValueError):
             model(website="invalid-uri")
@@ -493,6 +494,22 @@ class TestSchemaConverter(TestCase):
         with self.assertRaises(ValueError):
             model(ip="invalid-ipv6")
 
+    def test_string_format_uuid(self):
+        schema = {
+            "title": "UUIDTest",
+            "type": "object",
+            "properties": {"id": {"type": "string", "format": "uuid"}},
+        }
+        model = SchemaConverter.build(schema)
+
+        self.assertEqual(
+            model(id="123e4567-e89b-12d3-a456-426614174000").id,
+            UUID("123e4567-e89b-12d3-a456-426614174000"),
+        )
+
+        with self.assertRaises(ValueError):
+            model(id="invalid-uuid")
+
     def test_string_format_hostname(self):
         schema = {
             "title": "HostnameTest",