Modified to only read in keys needed from CSV file.

lib_afc_mosaic/helpers.py

@@ -5,7 +5,7 @@ Helper functions for the Mosaic library
 import csv
 import codecs
 
-def read_voterdata_csv_stream(input_stream) -> list:
+def read_voterdata_csv_stream(input_stream, keys_needed=None) -> list:
     """
     read_voterdata_csv_stream: Read csv stream to list
 
@@ -16,9 +16,13 @@ def read_voterdata_csv_stream(input_stream) -> list:
 
     rows = []
     # note: using 'utf-8-sig' removes the annoying \ufeff BOM signature
+    if keys_needed is None or len(keys_needed) > 0:
+        key_list = keys_needed
+
     reader = csv.DictReader(codecs.getreader("utf-8-sig")(input_stream))
     for row in reader:
-        key_list = row.keys()
+        if key_list is None:
+            key_list = row.keys()
         new_row = {}
         for k in key_list:
             new_row[k] = row[k]

lib_afc_mosaic/universe_map_file.py

@@ -202,18 +202,9 @@ class AFCMUniverseMapFile:
         :param voterset_filename: The VoterSet filename to process
         """
 
-        amplify_rows = []
-        removed_rows = []
-
-        try:
-            voterset_filename_s3_key = os.path.join(self.metadata.get("s3_key", ""), voterset_filename)
-            file_data_stream = self.voterset_s3_connection.get_object(self.voterset_s3_bucket,
-                                                                      voterset_filename_s3_key)
-        except Exception as ex:
-            raise Exception(f"universe_map_file.process_file: Failed to get {self.voterset_s3_bucket}/{voterset_filename_s3_key}: {ex}") from ex
-
-        rows = read_voterdata_csv_stream(file_data_stream)
-        logging.debug("rows = %s", len(rows))
+        if 'mapping' not in self.mosaic_file_list[voterset_filename]:
+            logging.debug("Missing 'mapping' key, skipping %s", voterset_filename)
+            return None
 
         mapping = self.mosaic_file_list[voterset_filename]['mapping']
 
@@ -225,6 +216,30 @@ class AFCMUniverseMapFile:
                           voterset_filename)
             return None
 
+        # Faster & less memory to only read the keys we need from CSV
+        csv_keys_needed = []
+        for k,v in mapping.items():
+            if isinstance(v, dict):
+                if 'field' in v:
+                    csv_keys_needed.append(v['field'])
+                elif 'combine' in v:
+                    for k1,v1 in v['combine'].items():
+                        if k1.startswith('field'):
+                            csv_keys_needed.append(v1)
+
+        amplify_rows = []
+        removed_rows = []
+
+        try:
+            voterset_filename_s3_key = os.path.join(self.metadata.get("s3_key", ""), voterset_filename)
+            file_data_stream = self.voterset_s3_connection.get_object(self.voterset_s3_bucket,
+                                                                      voterset_filename_s3_key)
+        except Exception as ex:
+            raise Exception(f"universe_map_file.process_file: Failed to get {self.voterset_s3_bucket}/{voterset_filename_s3_key}: {ex}") from ex
+
+        rows = read_voterdata_csv_stream(file_data_stream, csv_keys_needed)
+        logging.debug("rows = %s", len(rows))
+
         # Update incoming stats after reading this file map
         self.stats["incoming_count"][voterset_filename] = len(rows)
         self.stats["incoming_count"]["total"] += self.stats["incoming_count"][voterset_filename]