Fix: Resolve issue where PyAirbyte would fail if property names contain the dot character ('.'), e.g. with source-google-ads

.gitignore

@@ -130,3 +130,6 @@ dmypy.json
 
 # Cython debug symbols
 cython_debug/
+
+# Pycharm
+.idea

airbyte/shared/catalog_providers.py

@@ -16,6 +16,7 @@
 )
 
 from airbyte import exceptions as exc
+from airbyte._util.name_normalizers import LowerCaseNormalizer
 from airbyte.strategies import WriteMethod, WriteStrategy
 
 
@@ -149,7 +150,9 @@ def get_primary_keys(
  if not pks:
  return []
 
- joined_pks = [".".join(pk) for pk in pks]
+ normalized_pks = [[LowerCaseNormalizer.normalize(c) for c in pk] for pk in pks]
+ joined_pks = [".".join(pk) for pk in normalized_pks]
+
  for pk in joined_pks:
  if "." in pk:
  msg = f"Nested primary keys are not yet supported. Found: {pk}"

tests/integration_tests/fixtures/source-test/source_test/run.py

@@ -65,6 +65,23 @@
  },
  },
  },
+ {
+ "name": "primary-key-with-dot",
+ "description": "This stream has a primary key with dot similar what is there in GAds.",
+ "source_defined_primary_key": [["table1.Column1"]],
+ "source_defined_cursor": False,
+ "supported_sync_modes": ["full_refresh"],
+ "json_schema": {
+ "$schema": "http://json-schema.org/draft-07/schema#",
+ "type": "object",
+ "properties": {
+ "table1.Column1": {"type": "string"},
+ "table1.Column2": {"type": "number"},
+ "table1.empty_column": {"type": "string"},
+ "table1.big_number": {"type": "number"},
+ },
+ },
+ },
  ]
  },
 }
@@ -137,6 +154,19 @@
  "emitted_at": 1704067200,
  },
 }
+sample_record_primary_key_with_dot = {
+ "type": "RECORD",
+ "record": {
+ "data": {
+ "table1.Column1": "value1",
+ "table1.Column2": 1,
+ "table1.empty_column": None,
+ "table1.big_number": 1234567890123456,
+ },
+ "stream": "primary-key-with-dot",
+ "emitted_at": 1704067200,
+ },
+}
 
 
 def parse_args():
@@ -184,3 +214,5 @@ def run():
  print(json.dumps(sample_record2_stream1))
  elif stream["stream"]["name"] == "stream2":
  print(json.dumps(sample_record_stream2))
+ elif stream["stream"]["name"] == "primary-key-with-dot":
+ print(json.dumps(sample_record_primary_key_with_dot))