If data has already been processed, return the data

src/doms_databasen/_utils.py

@@ -1,7 +1,7 @@
 """Utility function for the doms_databasen package."""
 
 import json
-from typing import Dict, List
+from typing import List
 
 import jsonlines
 
@@ -19,7 +19,7 @@ def save_dict_to_json(dict_, file_path) -> None:
         json.dump(dict_, f, indent=4)
 
 
-def read_json(file_path) -> Dict[str, str]:
+def read_json(file_path) -> dict:
     """Reads a json file.
 
     Args:

src/doms_databasen/processor.py

@@ -68,6 +68,8 @@ def process(self, case_id: str) -> Dict[str, Union[str, Dict[str, str]]]:
             processed_data (dict):
                 Processed data (only returned for testing purposes)
         """
+        processed_data: Dict[str, Union[str, Dict[str, str]]] = {}
+
         case_id = str(case_id)
         if case_id in self.blacklist:
             logger.info(f"{case_id} is blacklisted.")
@@ -88,7 +90,10 @@ def process(self, case_id: str) -> Dict[str, Union[str, Dict[str, str]]]:
             logger.info(
                 f"Case {case_id} has already been processed. Use --force to overwrite."
             )
-            return {}
+            processed_data = read_json(
+                file_path=case_dir_processed / self.config.file_names.processed_data
+            )
+            return processed_data
 
         # Process data for the case.
         logger.info(f"Processing case {case_id}...")
@@ -99,7 +104,6 @@ def process(self, case_id: str) -> Dict[str, Union[str, Dict[str, str]]]:
             case_dir_raw / self.config.file_names.tabular_data
         )
 
-        processed_data: Dict[str, Union[str, Dict[str, str]]] = {}
         processed_data["case_id"] = case_id
         processed_data["tabular_data"] = tabular_data