Add SAT test for no duplicates in enum properties (#19243)

* Make pytest aware of backward compatibility mark * Unit tests for duplicate values in enum properties * Add docs about airbyte enforcements to enum usage * Add docs info to test

Author: erohmensing

airbyte-integrations/bases/source-acceptance-test/pytest.ini

@@ -7,3 +7,4 @@ testpaths =
 markers =
  default_timeout
  slow: marks tests as slow (deselect with '-m "not slow"')
+ backward_compatibility

airbyte-integrations/bases/source-acceptance-test/source_acceptance_test/tests/test_core.py

@@ -95,6 +95,18 @@ def test_docker_env(self, actual_connector_spec: ConnectorSpecification, docker_
  docker_runner.entry_point
  ), "env should be equal to space-joined entrypoint"
 
+ def test_enum_usage(self, actual_connector_spec: ConnectorSpecification):
+ """Check that enum lists in specs contain distinct values."""
+ docs_url = "https://docs.airbyte.io/connector-development/connector-specification-reference"
+ docs_msg = f"See specification reference at {docs_url}."
+
+ schema_helper = JsonSchemaHelper(actual_connector_spec.connectionSpecification)
+ enum_paths = schema_helper.find_nodes(keys=["enum"])
+
+ for path in enum_paths:
+ enum_list = schema_helper.get_node(path)
+ assert len(set(enum_list)) == len(enum_list), f"Enum lists should not contain duplicate values. Misconfigured enum array: {enum_list}. {docs_msg}"
+
  def test_oneof_usage(self, actual_connector_spec: ConnectorSpecification):
  """Check that if spec contains oneOf it follows the rules according to reference
  https://docs.airbyte.io/connector-development/connector-specification-reference

airbyte-integrations/bases/source-acceptance-test/unit_tests/test_spec.py

@@ -350,6 +350,49 @@ def test_oneof_usage(connector_spec, should_fail):
  else:
  t.test_oneof_usage(actual_connector_spec=ConnectorSpecification(connectionSpecification=connector_spec))
 
+@parametrize_test_case(
+ {
+ "test_id": "successful",
+ "connector_spec": {
+ "type": "object",
+ "properties": {
+ "property_with_options": {
+ "title": "Property with options",
+ "description": "A property in the form of an enumerated list",
+ "type": "string",
+ "default": "Option 1",
+ "enum": ["Option 1", "Option 2", "Option 3"],
+ }
+ },
+ },
+ "should_fail": False,
+ },
+ {
+ "test_id": "duplicate_values",
+ "connector_spec": {
+ "type": "object",
+ "properties": {
+ "property_with_options": {
+ "title": "Property with options",
+ "description": "A property in the form of an enumerated list",
+ "type": "string",
+ "default": "Option 1",
+ "enum": ["Option 1", "Option 2", "Option 3", "Option 2"],
+ }
+ },
+ },
+ "should_fail": True,
+ },
+)
+def test_enum_usage(connector_spec, should_fail):
+ t = _TestSpec()
+ if should_fail is True:
+ with pytest.raises(AssertionError):
+ t.test_enum_usage(actual_connector_spec=ConnectorSpecification(connectionSpecification=connector_spec))
+ else:
+ t.test_enum_usage(actual_connector_spec=ConnectorSpecification(connectionSpecification=connector_spec))
+
+
 
 @pytest.mark.parametrize(
  "connector_spec, expected_error",

docs/connector-development/connector-specification-reference.md

@@ -101,12 +101,14 @@ In this case, use the `"airbyte_hidden": true` keyword to hide that field from t
 }
 ```
 
+Results in the following form:
+
 ![hidden fields](../.gitbook/assets/spec_reference_hidden_field_screenshot.png)
 
-Results in the following form: 
 
+## Airbyte Modifications to `jsonschema`
 
-### Using `oneOf`s
+### Using `oneOf`
 
 In some cases, a connector needs to accept one out of many options. For example, a connector might need to know the compression codec of the file it will read, which will render in the Airbyte UI as a list of the available codecs. In JSONSchema, this can be expressed using the [oneOf](https://json-schema.org/understanding-json-schema/reference/combining.html#oneof) keyword.
 
@@ -183,3 +185,28 @@ In each item in the `oneOf` array, the `option_title` string field exists with t
 }
 ```
 
+### Using `enum`
+
+In regular `jsonschema`, some drafts enforce that `enum` lists must contain distinct values, while others do not. For consistency, Airbyte enforces this restriction.
+
+For example, this spec is invalid, since `a_format` is listed twice under the enumerated property `format`:
+
+```javascript
+{
+ "connection_specification": {
+ "$schema": "http://json-schema.org/draft-07/schema#",
+ "title": "File Source Spec",
+ "type": "object",
+ "required": ["format"],
+ "properties": {
+ "dataset_name": {
+ ...
+ },
+ "format": {
+ type: "string",
+ enum: ["a_format", "another_format", "a_format"]
+ },
+ }
+ }
+}
+```