From 0386ca21ae1322b7e6c49858ab2525bb301467f7 Mon Sep 17 00:00:00 2001
From: Maxime Carbonneau-Leclerc <3360483+maxi297@users.noreply.github.com>
Date: Wed, 19 Jun 2024 09:29:45 -0400
Subject: [PATCH] Exclude airbyte-cdk modules from schema discovery (#39586)

---
 .../declarative/schema/json_file_schema_loader.py   |  2 +-
 .../schema/test_json_file_schema_loader.py          | 13 ++++++++++++-
 2 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/airbyte-cdk/python/airbyte_cdk/sources/declarative/schema/json_file_schema_loader.py b/airbyte-cdk/python/airbyte_cdk/sources/declarative/schema/json_file_schema_loader.py
index bfadbe8f9f080d..af51fe5db01e0a 100644
--- a/airbyte-cdk/python/airbyte_cdk/sources/declarative/schema/json_file_schema_loader.py
+++ b/airbyte-cdk/python/airbyte_cdk/sources/declarative/schema/json_file_schema_loader.py
@@ -18,7 +18,7 @@ def _default_file_path() -> str:
     # Schema files are always in "source_<connector_name>/schemas/<stream_name>.json
     # The connector's module name can be inferred by looking at the modules loaded and look for the one starting with source_
     source_modules = [
-        k for k, v in sys.modules.items() if "source_" in k
+        k for k, v in sys.modules.items() if "source_" in k and "airbyte_cdk" not in k
     ]  # example: ['source_exchange_rates', 'source_exchange_rates.source']
     if source_modules:
         module = source_modules[0].split(".")[0]
diff --git a/airbyte-cdk/python/unit_tests/sources/declarative/schema/test_json_file_schema_loader.py b/airbyte-cdk/python/unit_tests/sources/declarative/schema/test_json_file_schema_loader.py
index aa2f99b48901b2..8ef467618634ee 100644
--- a/airbyte-cdk/python/unit_tests/sources/declarative/schema/test_json_file_schema_loader.py
+++ b/airbyte-cdk/python/unit_tests/sources/declarative/schema/test_json_file_schema_loader.py
@@ -1,9 +1,10 @@
 #
 # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
 #
+from unittest.mock import patch
 
 import pytest
-from airbyte_cdk.sources.declarative.schema import JsonFileSchemaLoader
+from airbyte_cdk.sources.declarative.schema.json_file_schema_loader import JsonFileSchemaLoader, _default_file_path
 
 
 @pytest.mark.parametrize(
@@ -24,3 +25,13 @@ def test_extract_resource_and_schema_path(test_name, input_path, expected_resour
 
     assert actual_resource == expected_resource
     assert actual_path == expected_path
+
+
+@patch("airbyte_cdk.sources.declarative.schema.json_file_schema_loader.sys")
+def test_exclude_cdk_packages(mocked_sys):
+    keys = ["airbyte_cdk.sources.concurrent_source.concurrent_source_adapter", "source_gitlab.utils"]
+    mocked_sys.modules = {key: "" for key in keys}
+
+    default_file_path = _default_file_path()
+
+    assert "source_gitlab" in default_file_path