Fix jsonschema validation that fires CustomKeyInConfigDeprecation (#11580)

* Fix detection of additional config property deprecation

Previously we were taking the first `key` on the `instance` property
of the jsonschema ValidationError. However, this validation error
is raised as an "anyOf" violation, which then has sub-errors in its
`context` property. To identify the key in violation, we have to
find the `additionalProperties` validation in the sub-errors. The key
that is an issue can then be parsed from that sub-error.

* Refactor key parsing from jsonschema ValidationError messages to single definition

* Update handling `additionalProperties` violations to handle multiple keys in violation

* Add changelog

* Remove guard logic in jsonschemas validation rule that is no longer needed

Author: QMalcolm

.changes/unreleased/Fixes-20250502-131822.yaml

@@ -0,0 +1,6 @@
+kind: Fixes
+body: Ensure the right key is associatd with the `CustomKeyInConfigDeprecation` deprecation
+time: 2025-05-02T13:18:22.940373-05:00
+custom:
+  Author: QMalcolm
+  Issue: "11576"

core/dbt/jsonschemas.py

@@ -3,7 +3,7 @@
 import re
 from datetime import date, datetime
 from pathlib import Path
-from typing import Any, Dict, Iterator
+from typing import Any, Dict, Iterator, List
 
 import jsonschema
 from jsonschema import ValidationError
@@ -57,6 +57,11 @@ def error_path_to_string(error: jsonschema.ValidationError) -> str:
         return path
 
 
+def _additional_properties_violation_keys(error: ValidationError) -> List[str]:
+    found_keys = re.findall(r"'\S+'", error.message)
+    return [key.strip("'") for key in found_keys]
+
+
 def jsonschema_validate(schema: Dict[str, Any], json: Dict[str, Any], file_path: str) -> None:
 
     if not os.environ.get("DBT_ENV_PRIVATE_RUN_JSONSCHEMA_VALIDATIONS"):
@@ -69,33 +74,38 @@ def jsonschema_validate(schema: Dict[str, Any], json: Dict[str, Any], file_path:
         # Listify the error path to make it easier to work with (it's a deque in the ValidationError object)
         error_path = list(error.path)
         if error.validator == "additionalProperties":
-            key = re.search(r"'\S+'", error.message)
+            keys = _additional_properties_violation_keys(error)
             if len(error.path) == 0:
-                deprecations.warn(
-                    "custom-top-level-key-deprecation",
-                    msg="Unexpected top-level key" + (" " + key.group() if key else ""),
-                    file=file_path,
-                )
+                for key in keys:
+                    deprecations.warn(
+                        "custom-top-level-key-deprecation",
+                        msg="Unexpected top-level key" + (" " + key if key else ""),
+                        file=file_path,
+                    )
             else:
-                deprecations.warn(
-                    "custom-key-in-object-deprecation",
-                    key=key.group() if key else "",
-                    file=file_path,
-                    key_path=error_path_to_string(error),
-                )
-        elif (
-            error.validator == "anyOf"
-            and len(error_path) > 0
-            and error_path[-1] == "config"
-            and isinstance(error.instance, dict)
-            and len(error.instance.keys()) > 0
-        ):
-            deprecations.warn(
-                "custom-key-in-config-deprecation",
-                key=(list(error.instance.keys()))[0],
-                file=file_path,
-                key_path=error_path_to_string(error),
-            )
+                key_path = error_path_to_string(error)
+                for key in keys:
+                    deprecations.warn(
+                        "custom-key-in-object-deprecation",
+                        key=key,
+                        file=file_path,
+                        key_path=key_path,
+                    )
+        elif error.validator == "anyOf" and len(error_path) > 0 and error_path[-1] == "config":
+            for sub_error in error.context or []:
+                if (
+                    isinstance(sub_error, ValidationError)
+                    and sub_error.validator == "additionalProperties"
+                ):
+                    keys = _additional_properties_violation_keys(sub_error)
+                    key_path = error_path_to_string(error)
+                    for key in keys:
+                        deprecations.warn(
+                            "custom-key-in-config-deprecation",
+                            key=key,
+                            file=file_path,
+                            key_path=key_path,
+                        )
         else:
             deprecations.warn(
                 "generic-json-schema-validation-deprecation",

tests/functional/deprecations/fixtures.py

@@ -168,6 +168,16 @@
       my_custom_key: "my_custom_value"
 """
 
+multiple_custom_keys_in_config_yaml = """
+models:
+  - name: models_trivial
+    description: "This is a test model"
+    deprecation_date: 1999-01-01 00:00:00.00+00:00
+    config:
+      my_custom_key: "my_custom_value"
+      my_custom_key2: "my_custom_value2"
+"""
+
 custom_key_in_object_yaml = """
 models:
   - name: models_trivial

tests/functional/deprecations/test_deprecations.py

@@ -26,6 +26,7 @@
     duplicate_keys_yaml,
     invalid_deprecation_date_yaml,
     models_trivial__model_sql,
+    multiple_custom_keys_in_config_yaml,
 )
 from tests.utils import EventCatcher
 
@@ -341,14 +342,43 @@ def models(self):
     @mock.patch.dict(os.environ, {"DBT_ENV_PRIVATE_RUN_JSONSCHEMA_VALIDATIONS": "True"})
     def test_duplicate_yaml_keys_in_schema_files(self, project):
         event_catcher = EventCatcher(CustomKeyInConfigDeprecation)
-        run_dbt(["parse", "--no-partial-parse"], callbacks=[event_catcher.catch])
+        run_dbt(
+            ["parse", "--no-partial-parse", "--show-all-deprecations"],
+            callbacks=[event_catcher.catch],
+        )
         assert len(event_catcher.caught_events) == 1
         assert (
             "Custom key `my_custom_key` found in `config` at path `models[0].config`"
             in event_catcher.caught_events[0].info.msg
         )
 
 
+class TestMultipleCustomKeysInConfigDeprecation:
+    @pytest.fixture(scope="class")
+    def models(self):
+        return {
+            "models_trivial.sql": models_trivial__model_sql,
+            "models.yml": multiple_custom_keys_in_config_yaml,
+        }
+
+    @mock.patch.dict(os.environ, {"DBT_ENV_PRIVATE_RUN_JSONSCHEMA_VALIDATIONS": "True"})
+    def test_duplicate_yaml_keys_in_schema_files(self, project):
+        event_catcher = EventCatcher(CustomKeyInConfigDeprecation)
+        run_dbt(
+            ["parse", "--no-partial-parse", "--show-all-deprecations"],
+            callbacks=[event_catcher.catch],
+        )
+        assert len(event_catcher.caught_events) == 2
+        assert (
+            "Custom key `my_custom_key` found in `config` at path `models[0].config`"
+            in event_catcher.caught_events[0].info.msg
+        )
+        assert (
+            "Custom key `my_custom_key2` found in `config` at path `models[0].config`"
+            in event_catcher.caught_events[1].info.msg
+        )
+
+
 class TestCustomKeyInObjectDeprecation:
     @pytest.fixture(scope="class")
     def models(self):
@@ -363,7 +393,7 @@ def test_custom_key_in_object_deprecation(self, project):
         run_dbt(["parse", "--no-partial-parse"], callbacks=[event_catcher.catch])
         assert len(event_catcher.caught_events) == 1
         assert (
-            "Custom key `'my_custom_property'` found at `models[0]` in file"
+            "Custom key `my_custom_property` found at `models[0]` in file"
             in event_catcher.caught_events[0].info.msg
         )