WIP

mongo_analyser/program.py

@@ -0,0 +1,28 @@
+import argparse
+
+import mongo_analyser.extractor as data_extractor
+
+
+def main():
+    parser = argparse.ArgumentParser(description='Usage: mongo_analyser <command> [<args>]'
+                                                 'Commands:'
+                                                 '    analyse_schema  Analyse MongoDB schema and generate metadata'
+                                                 '    extract_data    Extract data from MongoDB and save to JSON file'
+                                     )
+
+    parser.add_argument('command', help='Subcommand to run')
+
+    # Parse the arguments
+    args = parser.parse_args()
+
+    # Dispatch based on the subcommand
+    if args.command == 'analyse_schema':
+        data_extractor.main()
+    elif args.command == 'extract_data':
+        data_extractor.main()
+    else:
+        parser.print_help()
+
+
+if __name__ == "__main__":
+    main()

tests/test_extractor.py

@@ -64,71 +64,3 @@ def test_convert_to_json_compatible():
     assert result["int_field"] == 123
     assert result["date_field"] == "2024-01-01T00:00:00+00:00"
     assert result["str_field"] == "test"
-
-# def test_extract_data(mocker):
-#     # Arrange
-#     mock_client = mocker.patch("pymongo.MongoClient", autospec=True)
-#     mock_collection = mock_client.return_value.__getitem__.return_value.__getitem__.return_value
-#     mock_cursor = MagicMock()
-#     mock_cursor.limit.return_value = iter([{"field": "value"}])
-#     mock_collection.find.return_value = mock_cursor
-#
-#     mock_gzip_open = mocker.patch("gzip.open", mocker.mock_open())
-#     mock_json_dump = mocker.patch("json.dump")
-#
-#     timezone = pytz.timezone("UTC")
-#     schema = {"field": {"type": "str"}}
-#     output_file = "output.json.gz"
-#
-#     # Act
-#     DataExtractor.extract_data("mongodb://localhost:27017", "test_db",
-#                                "test_collection",
-#                                schema, output_file, timezone, 1000, 100)
-#
-#     # Assert
-#     mock_client.assert_called_once_with("mongodb://localhost:27017")
-#     mock_cursor.limit.assert_called_once_with(1)
-#     mock_gzip_open.assert_called_once_with(output_file, 'wt', encoding="utf-8")
-#     mock_json_dump.assert_called_once()
-
-
-# def test_extract_data_no_limit(mocker):
-#     # Arrange
-#     mock_client = mocker.patch("pymongo.MongoClient", autospec=True)
-#     mock_collection = mock_client.return_value.__getitem__.return_value.__getitem__.return_value
-#     mock_cursor = MagicMock()
-#     mock_cursor.return_value = iter([{"field": "value"}])
-#     mock_collection.find.return_value = mock_cursor
-#
-#     mock_gzip_open = mocker.patch("gzip.open", mocker.mock_open())
-#     mock_json_dump = mocker.patch("json.dump")
-#     mock_timezone = mocker.MagicMock()
-#
-#     schema = {"field": {"type": "str"}}
-#     output_file = "output.json.gz"
-#
-#     # Act
-#     DataExtractor.extract_data("mongodb://localhost:27017", "test_db",
-#                                "test_collection",
-#                                schema, output_file, mock_timezone, 1000, -1)
-#
-#     # Assert
-#     mock_client.assert_called_once_with("mongodb://localhost:27017")
-#     mock_cursor.assert_called_once()
-#     mock_gzip_open.assert_called_once_with(output_file, 'wt', encoding="utf-8")
-#     mock_json_dump.assert_called_once()
-
-
-# def test_extract_data_connection_failure(mocker):
-#
-#     # Arrange
-#     # mock_client = mocker.patch("pymongo.MongoClient", side_effect=ConnectionFailure)
-#     timezone = pytz.timezone("UTC")
-#     schema = {"field": {"type": "str"}}
-#     output_file = "output.json.gz"
-#
-#     # Act & Assert
-#     with pytest.raises(ConnectionFailure):
-#         DataExtractor.extract_data("mongodb://localhost:27017", "test_db",
-#                                    "test_collection",
-#                                    schema, output_file, timezone, 1000, 100)

tests/test_shared.py

@@ -70,6 +70,7 @@ def test_handle_binary_with_known_subtype(mocker):
 
 
 def test_handle_binary_with_unknown_subtype(mocker):
+    """Test for handling binary data with an unknown subtype."""
     # Arrange
     mock_value = mocker.Mock()
     mock_value.subtype = 99  # Unknown subtype
@@ -84,6 +85,7 @@ def test_handle_binary_with_unknown_subtype(mocker):
 
 
 def test_handle_binary_in_array(mocker):
+    """Test for handling binary data in an array."""
     # Arrange
     mock_value = mocker.Mock()
     mock_value.subtype = 4  # Known subtype 'binary<UUID>'