Merge pull request #14 from DeputyApp/INTS-1482-invalid-json-data-type

INTS-1482 - Support for JSON data types

Author: RhysHutchison

tap_deputy/discover.py

@@ -67,7 +67,8 @@
     'VarChar': 'string',
     'Blob': 'string',
     'Bit': 'boolean',
-    'Time': 'string'
+    'Time': 'string',
+    'Json': 'string',
 }
 
 

tests/test_discover.py

@@ -0,0 +1,74 @@
+import pytest
+from unittest.mock import Mock
+
+from tap_deputy.discover import discover, RESOURCES
+
+
+@pytest.mark.parametrize(
+    "deputy_type, expected_jsonschema_type",
+    [
+        ("Integer", ["null", "integer"]),
+        ("Float", ["null", "number"]),
+        ("VarChar", ["null", "string"]),
+        ("Blob", ["null", "string"]),
+        ("Bit", ["null", "boolean"]),
+        ("Time", ["null", "string"]),
+        ("Json", ["null", "string"]),
+    ],
+)
+def test_discover_simple_type_mappings(deputy_type, expected_jsonschema_type):
+    """
+    Verify that Deputy API field types are correctly mapped to simple
+    JSON schema types in the generated catalog.
+    """
+    client = Mock()
+    resource_name = "Employee"
+    stream_name = RESOURCES[resource_name]
+
+    client.get.return_value = {
+        "fields": {"Id": "Integer", "SomeTestField": deputy_type}
+    }
+
+    catalog = discover(client)
+    employees_stream = None
+    for s in catalog.streams:
+        if s.tap_stream_id == stream_name:
+            employees_stream = s
+
+    assert employees_stream is not None
+    schema_dict = employees_stream.schema.to_dict()
+    schema_properties = schema_dict["properties"]
+
+    assert "SomeTestField" in schema_properties
+    field_schema = schema_properties["SomeTestField"]
+    assert field_schema["type"] == expected_jsonschema_type
+
+
+@pytest.mark.parametrize("deputy_type", ["Date", "DateTime"])
+def test_discover_datetime_type_mapping(deputy_type):
+    """
+    Verify that Deputy API 'Date' and 'DateTime' field types are correctly
+    mapped to a 'string' with 'date-time' format in the catalog schema.
+    """
+    client = Mock()
+    resource_name = "Employee"
+    stream_name = RESOURCES[resource_name]
+
+    client.get.return_value = {
+        "fields": {"Id": "Integer", "SomeDateField": deputy_type}
+    }
+
+    catalog = discover(client)
+    employees_stream = None
+    for s in catalog.streams:
+        if s.tap_stream_id == stream_name:
+            employees_stream = s
+
+    assert employees_stream is not None
+    schema_dict = employees_stream.schema.to_dict()
+    schema_properties = schema_dict["properties"]
+
+    assert "SomeDateField" in schema_properties
+    field_schema = schema_properties["SomeDateField"]
+    assert field_schema["type"] == ["null", "string"]
+    assert field_schema["format"] == "date-time"