cloudevents · cumason123 · Aug 19, 2020 · Aug 19, 2020 · Aug 19, 2020 · Aug 19, 2020
diff --git a/cloudevents/__init__.py b/cloudevents/__init__.py
@@ -1 +1 @@
-__version__ = "1.1.0"
+__version__ = "1.2.0"
diff --git a/cloudevents/exceptions.py b/cloudevents/exceptions.py
@@ -11,17 +11,29 @@
 #    WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
 #    License for the specific language governing permissions and limitations
 #    under the License.
-class MissingRequiredFields(Exception):
+class GenericException(Exception):
     pass
 
 
-class InvalidRequiredFields(Exception):
+class MissingRequiredFields(GenericException):
     pass
 
 
-class InvalidStructuredJSON(Exception):
+class InvalidRequiredFields(GenericException):
     pass
 
 
-class InvalidHeadersFormat(Exception):
+class InvalidStructuredJSON(GenericException):
+    pass
+
+
+class InvalidHeadersFormat(GenericException):
+    pass
+
+
+class DataMarshallerError(GenericException):
+    pass
+
+
+class DataUnmarshallerError(GenericException):
     pass
diff --git a/cloudevents/http/http_methods.py b/cloudevents/http/http_methods.py
@@ -20,13 +20,15 @@ def from_http(
     Unwrap a CloudEvent (binary or structured) from an HTTP request.
     :param headers: the HTTP headers
     :type headers: typing.Dict[str, str]
-    :param data: the HTTP request body
+    :param data: the HTTP request body. If set to None, "" or b'', the returned
+        event's data field will be set to None
     :type data: typing.IO
     :param data_unmarshaller: Callable function to map data to a python object
         e.g. lambda x: x or lambda x: json.loads(x)
     :type data_unmarshaller: types.UnmarshallerType
     """
-    if data is None:
+    if data is None or data == b"":
+        # Empty string will cause data to be marshalled into None
         data = ""
 
     if not isinstance(data, (str, bytes, bytearray)):
@@ -79,7 +81,13 @@ def from_http(
     attrs.pop("extensions", None)
     attrs.update(**event.extensions)
 
-    return CloudEvent(attrs, event.data)
+    if event.data == "" or event.data == b"":
+        # TODO: Check binary unmarshallers to debug why setting data to ""
+        # returns an event with data set to None, but structured will return ""
+        data = None
+    else:
+        data = event.data
+    return CloudEvent(attrs, data)
 
 
 def _to_http(

diff --git a/cloudevents/http/util.py b/cloudevents/http/util.py
@@ -3,7 +3,7 @@
 
 
 def default_marshaller(content: any):
-    if content is None or len(content) == 0:
+    if content is None:
         return None
     try:
         return json.dumps(content)
@@ -12,7 +12,7 @@ def default_marshaller(content: any):
 
 
 def _json_or_string(content: typing.Union[str, bytes]):
-    if content is None or len(content) == 0:
+    if content is None:
         return None
     try:
         return json.loads(content)

diff --git a/cloudevents/sdk/event/base.py b/cloudevents/sdk/event/base.py
@@ -201,7 +201,14 @@ def MarshalJSON(self, data_marshaller: types.MarshallerType) -> str:
             data_marshaller = lambda x: x  # noqa: E731
         props = self.Properties()
         if "data" in props:
-            data = data_marshaller(props.pop("data"))
+            data = props.pop("data")
+            try:
+                data = data_marshaller(data)
+            except Exception as e:
+                raise cloud_exceptions.DataMarshallerError(
+                    "Failed to marshall data with error: "
+                    f"{type(e).__name__}('{e}')"
+                )
             if isinstance(data, (bytes, bytes, memoryview)):
                 props["data_base64"] = base64.b64encode(data).decode("ascii")
             else:
@@ -225,14 +232,23 @@ def UnmarshalJSON(
             )
 
         for name, value in raw_ce.items():
+            decoder = lambda x: x
             if name == "data":
                 # Use the user-provided serializer, which may have customized
                 # JSON decoding
-                value = data_unmarshaller(json.dumps(value))
+                decoder = lambda v: data_unmarshaller(json.dumps(v))
             if name == "data_base64":
-                value = data_unmarshaller(base64.b64decode(value))
+                decoder = lambda v: data_unmarshaller(base64.b64decode(v))
                 name = "data"
-            self.Set(name, value)
+
+            try:
+                set_value = decoder(value)
+            except Exception as e:
+                raise cloud_exceptions.DataUnmarshallerError(
+                    "Failed to unmarshall data with error: "
+                    f"{type(e).__name__}('{e}')"
+                )
+            self.Set(name, set_value)
 
     def UnmarshalBinary(
         self,
@@ -256,7 +272,15 @@ def UnmarshalBinary(
                 self.SetContentType(value)
             elif header.startswith("ce-"):
                 self.Set(header[3:], value)
-        self.Set("data", data_unmarshaller(body))
+
+        try:
+            raw_ce = data_unmarshaller(body)
+        except Exception as e:
+            raise cloud_exceptions.DataUnmarshallerError(
+                "Failed to unmarshall data with error: "
+                f"{type(e).__name__}('{e}')"
+            )
+        self.Set("data", raw_ce)
 
     def MarshalBinary(
         self, data_marshaller: types.MarshallerType
@@ -276,7 +300,13 @@ def MarshalBinary(
             headers["ce-{0}".format(key)] = value
 
         data, _ = self.Get("data")
-        data = data_marshaller(data)
+        try:
+            data = data_marshaller(data)
+        except Exception as e:
+            raise cloud_exceptions.DataMarshallerError(
+                "Failed to marshall data with error: "
+                f"{type(e).__name__}('{e}')"
+            )
         if isinstance(data, str):  # Convenience method for json.dumps
             data = data.encode("utf-8")
         return headers, data
diff --git a/cloudevents/tests/test_http_cloudevent.py b/cloudevents/tests/test_http_cloudevent.py
@@ -2,6 +2,7 @@
 
 import cloudevents.exceptions as cloud_exceptions
 from cloudevents.http import CloudEvent
+from cloudevents.http.util import _json_or_string
 
 
 @pytest.mark.parametrize("specversion", ["0.3", "1.0"])
@@ -114,3 +115,7 @@ def test_cloudevent_general_overrides():
         assert attribute in event
         del event[attribute]
     assert len(event) == 0
+
+
+def test_none_json_or_string():
+    assert _json_or_string(None) is None
diff --git a/cloudevents/tests/test_http_events.py b/cloudevents/tests/test_http_events.py
@@ -286,9 +286,17 @@ def test_empty_data_structured_event(specversion):
         "source": "<source-url>",
     }
 
-    _ = from_http(
+    event = from_http(
+        {"content-type": "application/cloudevents+json"}, json.dumps(attributes)
+    )
+    assert event.data == None
+
+    attributes["data"] = ""
+    # Data of empty string will be marshalled into None
+    event = from_http(
         {"content-type": "application/cloudevents+json"}, json.dumps(attributes)
     )
+    assert event.data == None
 
 
 @pytest.mark.parametrize("specversion", ["1.0", "0.3"])
@@ -302,7 +310,13 @@ def test_empty_data_binary_event(specversion):
         "ce-time": "2018-10-23T12:28:22.4579346Z",
         "ce-source": "<source-url>",
     }
-    _ = from_http(headers, "")
+    event = from_http(headers, None)
+    assert event.data == None
+
+    data = ""
+    # Data of empty string will be marshalled into None
+    event = from_http(headers, data)
+    assert event.data == None
 
 
 @pytest.mark.parametrize("specversion", ["1.0", "0.3"])
@@ -476,6 +490,35 @@ def test_uppercase_headers_with_none_data_binary():
     assert new_data == None
 
 
+def test_generic_exception():
+    headers = {"Content-Type": "application/cloudevents+json"}
+    data = json.dumps(
+        {
+            "specversion": "1.0",
+            "source": "s",
+            "type": "t",
+            "id": "1234-1234-1234",
+            "data": "",
+        }
+    )
+    with pytest.raises(cloud_exceptions.GenericException) as e:
+        from_http({}, None)
+    e.errisinstance(cloud_exceptions.MissingRequiredFields)
+
+    with pytest.raises(cloud_exceptions.GenericException) as e:
+        from_http({}, 123)
+    e.errisinstance(cloud_exceptions.InvalidStructuredJSON)
+
+    with pytest.raises(cloud_exceptions.GenericException) as e:
+        from_http(headers, data, data_unmarshaller=lambda x: 1 / 0)
+    e.errisinstance(cloud_exceptions.DataUnmarshallerError)
+
+    with pytest.raises(cloud_exceptions.GenericException) as e:
+        event = from_http(headers, data)
+        to_binary(event, data_marshaller=lambda x: 1 / 0)
+    e.errisinstance(cloud_exceptions.DataMarshallerError)
+
+
 def test_non_dict_data_no_headers_bug():
     # Test for issue #116
     headers = {"Content-Type": "application/cloudevents+json"}

diff --git a/cloudevents/tests/test_marshaller.py b/cloudevents/tests/test_marshaller.py
@@ -12,15 +12,19 @@
 #    License for the specific language governing permissions and limitations
 #    under the License.
 
+import json
+
 import pytest
 
+import cloudevents.exceptions as cloud_exceptions
+from cloudevents.http import CloudEvent, from_http, to_binary, to_structured
 from cloudevents.sdk import converters, exceptions, marshaller
 from cloudevents.sdk.converters import binary, structured
 from cloudevents.sdk.event import v1
 
 
 @pytest.fixture
-def headers():
+def binary_headers():
     return {
         "ce-specversion": "1.0",
         "ce-source": "1.0",
@@ -29,6 +33,19 @@ def headers():
     }
 
 
+@pytest.fixture
+def structured_data():
+    return json.dumps(
+        {
+            "specversion": "1.0",
+            "source": "pytest",
+            "type": "com.pytest.test",
+            "id": "1234-1234-1234",
+            "data": "test",
+        }
+    )
+
+
 def test_from_request_wrong_unmarshaller():
     with pytest.raises(exceptions.InvalidDataUnmarshaller):
         m = marshaller.NewDefaultHTTPMarshaller()
@@ -41,7 +58,7 @@ def test_to_request_wrong_marshaller():
         _ = m.ToRequest(v1.Event(), data_marshaller="")
 
 
-def test_from_request_cannot_read(headers):
+def test_from_request_cannot_read(binary_headers):
     with pytest.raises(exceptions.UnsupportedEventConverter):
         m = marshaller.HTTPMarshaller(
             [binary.NewBinaryHTTPCloudEventConverter(),]
@@ -52,7 +69,7 @@ def test_from_request_cannot_read(headers):
         m = marshaller.HTTPMarshaller(
             [structured.NewJSONHTTPCloudEventConverter()]
         )
-        m.FromRequest(v1.Event(), headers, "")
+        m.FromRequest(v1.Event(), binary_headers, "")
 
 
 def test_to_request_invalid_converter():
@@ -61,3 +78,65 @@ def test_to_request_invalid_converter():
             [structured.NewJSONHTTPCloudEventConverter()]
         )
         m.ToRequest(v1.Event(), "")
+
+
+def test_http_data_unmarshaller_exceptions(binary_headers, structured_data):
+    # binary
+    with pytest.raises(cloud_exceptions.DataUnmarshallerError) as e:
+        from_http(binary_headers, None, data_unmarshaller=lambda x: 1 / 0)
+    assert (
+        "Failed to unmarshall data with error: "
+        "ZeroDivisionError('division by zero')" in str(e.value)
+    )
+
+    # structured
+    headers = {"Content-Type": "application/cloudevents+json"}
+    with pytest.raises(cloud_exceptions.DataUnmarshallerError) as e:
+        from_http(headers, structured_data, data_unmarshaller=lambda x: 1 / 0)
+    assert (
+        "Failed to unmarshall data with error: "
+        "ZeroDivisionError('division by zero')" in str(e.value)
+    )
+
+
+def test_http_data_marshaller_exception(binary_headers, structured_data):
+    # binary
+    event = from_http(binary_headers, None)
+    with pytest.raises(cloud_exceptions.DataMarshallerError) as e:
+        to_binary(event, data_marshaller=lambda x: 1 / 0)
+    assert (
+        "Failed to marshall data with error: "
+        "ZeroDivisionError('division by zero')" in str(e.value)
+    )
+
+    # structured
+    headers = {"Content-Type": "application/cloudevents+json"}
+
+    event = from_http(headers, structured_data)
+    with pytest.raises(cloud_exceptions.DataMarshallerError) as e:
+        to_structured(event, data_marshaller=lambda x: 1 / 0)
+    assert (
+        "Failed to marshall data with error: "
+        "ZeroDivisionError('division by zero')" in str(e.value)
+    )
+
+
+@pytest.mark.parametrize("test_data", [[], {}, (), "", b"", None])
+def test_known_empty_edge_cases(binary_headers, test_data):
+    expect_data = test_data
+    if test_data in ["", b""]:
+        expect_data = None
+    elif test_data == ():
+        # json.dumps(()) outputs '[]' hence list not tuple check
+        expect_data = []
+
+    # Remove ce- prefix
+    headers = {key[3:]: value for key, value in binary_headers.items()}
+
+    # binary
+    event = from_http(*to_binary(CloudEvent(headers, test_data)))
+    assert event.data == expect_data
+
+    # structured
+    event = from_http(*to_structured(CloudEvent(headers, test_data)))
+    assert event.data == expect_data