adding redaction of cause number

src/cleaner/__init__.py

@@ -1,12 +1,17 @@
 import json, argparse, os, datetime as dt, xxhash
-from azure.cosmos import CosmosClient, exceptions
 from dotenv import load_dotenv
 
 class Cleaner:
 
     def __init__(self, county):
         self.county = county.lower()
 
+    def redact_cause_number(self, input_dict: dict, out_file: dict) -> dict:
+        #This will hash and redact the cause number and then add it to the output file.
+        cause_number_hash = xxhash.xxh64(str(input_dict['code'])).hexdigest()
+        out_file["cause_number_redacted"] = cause_number_hash
+        return out_file
+
     def clean(self):
 
         case_json_folder_path = os.path.join(
@@ -26,7 +31,7 @@ def clean(self):
         list_case_json_files = os.listdir(case_json_folder_path)
         for case_json in list_case_json_files:
             print(case_json)
-            # List of motions identified as evidenciary
+            # List of motions identified as evidentiary. TODO: These should be moved to a separate JSON in resources
             good_motions = [
                 "Motion To Suppress",
                 "Motion to Reduce Bond",
@@ -61,7 +66,6 @@ def clean(self):
             charge_name_to_umich = charge_name_to_umich_dict
             # Cleaned Case Primary format
             out_file = {}
-            out_file["case_number"] = input_dict["code"] #Note: This may be closed to personally identifying information of the defendant.
             out_file["attorney_type"] = input_dict["party information"]["appointed or retained"]
             #Adding the county and hash values into the final version.
             out_file["county"] = input_dict["county"]
@@ -112,11 +116,12 @@ def contains_good_motion(motion, event):
             def_atty_hash = xxhash.xxh64(str(def_atty_unique_str)).hexdigest()
             out_file["defense attorney"] = def_atty_hash
 
+            out_file = self.redact_cause_number(input_dict, out_file)
+
             # Original Format
             out_filepath = os.path.join(
             os.path.dirname(__file__), "..", "..", "data", self.county, "case_json_cleaned",case_json
             )
 
             with open(out_filepath, "w") as f:
                 json.dump(out_file, f)
-