init

python/mrds_common/mrds/processors/xml_processor.py

@@ -0,0 +1,30 @@
+import logging
+
+from .base import TaskProcessor
+
+from mrds.utils import (
+    xml_utils,
+    csv_utils,
+)
+
+
+class XMLTaskProcessor(TaskProcessor):
+        
+    def _extract(self):
+        # Extract data from XML
+        csv_data = xml_utils.extract_data(
+            self.global_config.source_filepath,
+            self.xpath_entries,
+            self.xml_position_entries,
+            self.task_conf.namespaces,
+            self.workflow_context,
+            self.global_config.encoding_type,
+        )
+        logging.info(f"CSV data extracted for task '{self.task_conf.task_name}'")
+
+        # Generate CSV
+        logging.info(f"Writing core data to CSV file at '{self.output_filepath}'")
+        csv_utils.write_data_to_csv_file(
+            self.output_filepath, csv_data, self.global_config.encoding_type
+        )
+        logging.info(f"Core data written to CSV file at '{self.output_filepath}'")