Add more accurate JSON schema for Pydantic (#25)

* Add more accurate JSON schema for Pydantic

* chore: Python 3.9 doesn't allow union types with |

Author: mdomke

CHANGELOG.rst

@@ -5,9 +5,15 @@ Changelog
 
 Versions follow `Semantic Versioning <http://www.semver.org>`_
 
-`2.4.0`_ - 2024-04-02
+`2.5.0`_ - tbd
 ---------------------
+Changed
+~~~~~~~
+* Generate a more accurate JSON schema with Pydantic's ``BaseModel.model_json_schema()``. This
+  includes a specification for string and byte representations.
 
+`2.4.0`_ - 2024-04-02
+---------------------
 Added
 ~~~~~
 * :class:`.ULID` objects are now properly serialized when used as Pydantic types `@Avihais12344 <https://github.com/Avihais12344>`_.
@@ -151,6 +157,7 @@ Changed
 * The package now has no external dependencies.
 * The test-coverage has been raised to 100%.
 
+.. _2.5.0: https://github.com/mdomke/python-ulid/compare/2.4.0...2.5.0
 .. _2.4.0: https://github.com/mdomke/python-ulid/compare/2.3.0...2.4.0
 .. _2.3.0: https://github.com/mdomke/python-ulid/compare/2.2.0...2.3.0
 .. _2.2.0: https://github.com/mdomke/python-ulid/compare/2.1.0...2.2.0

tests/test_ulid.py

@@ -5,6 +5,7 @@
 from datetime import datetime
 from datetime import timedelta
 from datetime import timezone
+from typing import Optional
 from typing import Union
 
 import pytest
@@ -168,7 +169,7 @@ def test_pydantic_protocol() -> None:
     ulid = ULID()
 
     class Model(BaseModel):
-        ulid: ULID
+        ulid: Optional[ULID] = None
 
     for value in [ulid, str(ulid), int(ulid), bytes(ulid)]:
         model = Model(ulid=value)
@@ -188,3 +189,23 @@ class Model(BaseModel):
     model_json = model.model_dump_json()
     assert isinstance(json.loads(model_json)["ulid"], str)
     assert Model.model_validate_json(model_json) == model
+
+    model_json_schema = model.model_json_schema()
+    assert "properties" in model_json_schema
+    assert "ulid" in model_json_schema["properties"]
+    assert "anyOf" in model_json_schema["properties"]["ulid"]
+    assert {
+        "maxLength": 26,
+        "minLength": 26,
+        "pattern": "[A-Z0-9]{26}",
+        "type": "string",
+    } in model_json_schema["properties"]["ulid"]["anyOf"]
+    assert {
+        "maxLength": 16,
+        "minLength": 16,
+        "type": "string",
+        "format": "binary",
+    } in model_json_schema["properties"]["ulid"]["anyOf"]
+    assert {
+        "type": "null",
+    } in model_json_schema["properties"]["ulid"]["anyOf"]

ulid/__init__.py

@@ -259,6 +259,8 @@ def __get_pydantic_core_schema__(cls, source: Any, handler: GetCoreSchemaHandler
                 [
                     core_schema.is_instance_schema(ULID),
                     core_schema.no_info_plain_validator_function(ULID),
+                    core_schema.str_schema(pattern=r"[A-Z0-9]{26}", min_length=26, max_length=26),
+                    core_schema.bytes_schema(min_length=16, max_length=16),
                 ]
             ),
             serialization=core_schema.to_string_ser_schema(