wrap up JSON format completely and make it compliant with the spec

Signed-off-by: Tudor Plugaru <[email protected]>

Author: PlugaruT

src/cloudevents/core/formats/base.py

@@ -13,12 +13,12 @@
 #    under the License.
 
 
-from typing import Protocol, Union
+from typing import Callable, Optional, Protocol, Union
 
 from cloudevents.core.base import BaseCloudEvent
 
 
 class Format(Protocol):
-    def read(self, data: Union[str, bytes]) -> BaseCloudEvent: ...
+    def read(self, event_factory: Callable[[dict, Optional[Union[dict, str, bytes]]], BaseCloudEvent], data: Union[str, bytes]) -> BaseCloudEvent: ...
 
     def write(self, event: BaseCloudEvent) -> bytes: ...

src/cloudevents/core/formats/json.py

@@ -17,15 +17,13 @@
 import re
 from datetime import datetime
 from json import JSONEncoder, dumps, loads
-from typing import Any, Final, Pattern, Type, TypeVar, Union
+from typing import Any, Callable, Final, Optional, Pattern, Union
 
 from dateutil.parser import isoparse  # type: ignore[import-untyped]
 
 from cloudevents.core.base import BaseCloudEvent
 from cloudevents.core.formats.base import Format
 
-T = TypeVar("T", bound=BaseCloudEvent)
-
 
 class _JSONEncoderWithDatetime(JSONEncoder):
     """
@@ -47,13 +45,14 @@ def default(self, obj: Any) -> Any:
 class JSONFormat(Format):
     CONTENT_TYPE: Final[str] = "application/cloudevents+json"
     JSON_CONTENT_TYPE_PATTERN: Pattern[str] = re.compile(
-        r"^(application|text)\\/([a-zA-Z]+\\+)?json(;.*)*$"
+        r"^(application|text)/([a-zA-Z0-9\-\.]+\+)?json(;.*)?$"
     )
 
-    def read(self, event_klass: Type[T], data: Union[str, bytes]) -> T:
+    def read(self, event_factory: Callable[[dict, Optional[Union[dict, str, bytes]]], BaseCloudEvent], data: Union[str, bytes]) -> BaseCloudEvent:
         """
         Read a CloudEvent from a JSON formatted byte string.
 
+        :param event_factory: A factory function to create CloudEvent instances.
         :param data: The JSON formatted byte array.
         :return: The CloudEvent instance.
         """
@@ -67,16 +66,15 @@ def read(self, event_klass: Type[T], data: Union[str, bytes]) -> T:
         if "time" in event_attributes:
             event_attributes["time"] = isoparse(event_attributes["time"])
 
-        event_data: Union[str, bytes] = event_attributes.pop("data", None)
+        event_data: Union[dict, str, bytes, None] = event_attributes.pop("data", None)
         if event_data is None:
             event_data_base64 = event_attributes.pop("data_base64", None)
             if event_data_base64 is not None:
                 event_data = base64.b64decode(event_data_base64)
 
-        # disable mypy due to https://github.com/python/mypy/issues/9003
-        return event_klass(event_attributes, event_data)  # type: ignore
+        return event_factory(event_attributes, event_data)
 
-    def write(self, event: T) -> bytes:
+    def write(self, event: BaseCloudEvent) -> bytes:
         """
         Write a CloudEvent to a JSON formatted byte string.
 
@@ -90,12 +88,9 @@ def write(self, event: T) -> bytes:
             if isinstance(event_data, (bytes, bytearray)):
                 event_dict["data_base64"] = base64.b64encode(event_data).decode("utf-8")
             else:
-                datacontenttype = event_dict.get(
-                    "datacontenttype", JSONFormat.CONTENT_TYPE
-                )
-                # Should we fail if we can't serialize data to JSON?
+                datacontenttype = event_dict.get("datacontenttype", "application/json")
                 if re.match(JSONFormat.JSON_CONTENT_TYPE_PATTERN, datacontenttype):
-                    event_dict["data"] = dumps(event_data)
+                    event_dict["data"] = event_data
                 else:
                     event_dict["data"] = str(event_data)
 

tests/test_core/test_format/test_json.py

@@ -15,8 +15,6 @@
 
 from datetime import datetime, timezone
 
-import pytest
-
 from cloudevents.core.formats.json import JSONFormat
 from cloudevents.core.v1.event import CloudEvent
 
@@ -61,7 +59,7 @@ def test_write_cloud_event_to_json_with_data_as_json() -> None:
 
     assert (
         result
-        == '{"id": "123", "source": "source", "type": "type", "specversion": "1.0", "time": "2023-10-25T17:09:19.736166Z", "datacontenttype": "application/json", "dataschema": "http://example.com/schema", "subject": "test_subject", "data": "{\'key\': \'value\'}"}'.encode(
+        == '{"id": "123", "source": "source", "type": "type", "specversion": "1.0", "time": "2023-10-25T17:09:19.736166Z", "datacontenttype": "application/json", "dataschema": "http://example.com/schema", "subject": "test_subject", "data": {"key": "value"}}'.encode(
             "utf-8"
         )
     )
@@ -151,7 +149,7 @@ def test_write_cloud_event_to_json_with_no_content_type_set_and_data_as_json() -
 
     assert (
         result
-        == '{"id": "123", "source": "source", "type": "type", "specversion": "1.0", "time": "2023-10-25T17:09:19.736166Z", "dataschema": "http://example.com/schema", "subject": "test_subject", "data": "{\'key\': \'value\'}"}'.encode(
+        == '{"id": "123", "source": "source", "type": "type", "specversion": "1.0", "time": "2023-10-25T17:09:19.736166Z", "dataschema": "http://example.com/schema", "subject": "test_subject", "data": {"key": "value"}}'.encode(
             "utf-8"
         )
     )
@@ -215,3 +213,107 @@ def test_read_cloud_event_from_json_with_json_as_data() -> None:
     assert result.get_dataschema() == "http://example.com/schema"
     assert result.get_subject() == "test_subject"
     assert result.get_data() == {"key": "value"}
+
+
+def test_write_cloud_event_with_extension_attributes() -> None:
+    attributes = {
+        "id": "123",
+        "source": "source",
+        "type": "type",
+        "specversion": "1.0",
+        "customext1": "value1",
+        "customext2": 123,
+    }
+    event = CloudEvent(attributes=attributes, data=None)
+    formatter = JSONFormat()
+    result = formatter.write(event)
+
+    assert b'"customext1": "value1"' in result
+    assert b'"customext2": 123' in result
+
+
+def test_read_cloud_event_with_extension_attributes() -> None:
+    data = '{"id": "123", "source": "source", "type": "type", "specversion": "1.0", "customext1": "value1", "customext2": 123}'.encode("utf-8")
+    formatter = JSONFormat()
+    result = formatter.read(CloudEvent, data)
+
+    assert result.get_extension("customext1") == "value1"
+    assert result.get_extension("customext2") == 123
+
+
+def test_write_cloud_event_with_different_json_content_types() -> None:
+    test_cases = [
+        ("application/vnd.api+json", {"key": "value"}),
+        ("text/json", {"key": "value"}),
+        ("application/json; charset=utf-8", {"key": "value"}),
+    ]
+    
+    for content_type, data in test_cases:
+        attributes = {
+            "id": "123",
+            "source": "source",
+            "type": "type",
+            "specversion": "1.0",
+            "datacontenttype": content_type,
+        }
+        event = CloudEvent(attributes=attributes, data=data)
+        formatter = JSONFormat()
+        result = formatter.write(event)
+        
+        assert b'"data": {"key": "value"}' in result
+
+
+def test_read_cloud_event_with_string_data() -> None:
+    data = '{"id": "123", "source": "source", "type": "type", "specversion": "1.0", "data": "plain string data"}'.encode("utf-8")
+    formatter = JSONFormat()
+    result = formatter.read(CloudEvent, data)
+
+    assert result.get_data() == "plain string data"
+
+
+def test_write_cloud_event_with_utc_timezone_z_suffix() -> None:
+    attributes = {
+        "id": "123",
+        "source": "source",
+        "type": "type",
+        "specversion": "1.0",
+        "time": datetime(2023, 10, 25, 17, 9, 19, 736166, tzinfo=timezone.utc),
+    }
+    event = CloudEvent(attributes=attributes, data=None)
+    formatter = JSONFormat()
+    result = formatter.write(event)
+
+    assert b'"time": "2023-10-25T17:09:19.736166Z"' in result
+
+
+def test_write_cloud_event_with_unicode_data() -> None:
+    attributes = {
+        "id": "123",
+        "source": "source",
+        "type": "type",
+        "specversion": "1.0",
+    }
+    event = CloudEvent(attributes=attributes, data="Hello 世界 🌍")
+    formatter = JSONFormat()
+    result = formatter.write(event)
+    
+    decoded = result.decode("utf-8")
+    assert '"data": "Hello' in decoded
+    assert "Hello" in decoded
+
+
+def test_read_cloud_event_with_unicode_data() -> None:
+    data = '{"id": "123", "source": "source", "type": "type", "specversion": "1.0", "data": "Hello 世界 🌍"}'.encode("utf-8")
+    formatter = JSONFormat()
+    result = formatter.read(CloudEvent, data)
+
+    assert result.get_data() == "Hello 世界 🌍"
+
+
+def test_read_cloud_event_from_string_input() -> None:
+    data = '{"id": "123", "source": "source", "type": "type", "specversion": "1.0"}'
+    formatter = JSONFormat()
+    result = formatter.read(CloudEvent, data)
+
+    assert result.get_id() == "123"
+    assert result.get_source() == "source"