Code refactoring

main.py

@@ -190,6 +190,31 @@ def delete_dataset_from_dest_index( the_cfg, the_uuid ):
     return
 
 
+def get_metadata_records_processor( the_root_url,
+                                    the_no_records_per_page,
+                                    the_uuid_to_get,
+                                    the_uuids_to_filter_out=None,
+                                    the_username=None,
+                                    the_password=None):
+
+    for record in get_metadata_records( root_url=the_root_url,
+                                        no_records_per_page=the_no_records_per_page,
+                                        uuid=the_uuid_to_get,
+                                        the_filter=the_uuids_to_filter_out,
+                                        username=the_username, password=the_password ):
+
+        the_uuid = record['geonet:info']['uuid']
+
+        if is_locked( cfg['session']['working_directory'], the_uuid ):
+            logging.info("Dataset with uuid = %s is already being indexed: skipping." % the_uuid)
+            continue
+        else:
+            logging.info("Setting lock for dataset with uuid = %s" % the_uuid)
+            lock(cfg['session']['working_directory'], the_uuid)
+
+        # delete_dataset_from_indices(cfg, record['geonet:info']['uuid'])
+        delete_dataset_from_dest_index(cfg, the_uuid)
+        send_record_to_the_metadata_processor(cfg, record)
 
 
 def main(cfg):
@@ -209,48 +234,63 @@ def main(cfg):
     password = cfg['geonetwork']['password']
     del cfg['geonetwork']['password'] # <- as this info is no longer needed
 
-    if 'all' not in uuids_to_get:
-
-        # get some datasets only
-        for uuid_to_get in uuids_to_get:
-            for record in get_metadata_records( cfg['geonetwork']['url'],
-                                                cfg['geonetwork']['records_per_page'],
-                                                uuid=uuid_to_get,
-                                                the_filter=uuids_to_filter_out,
-                                                username=username, password=password ):
-
-                the_uuid = record['geonet:info']['uuid']
-
-                if is_locked( cfg['session']['working_directory'], the_uuid ):
-                    logging.info("Dataset with uuid = %s is already being indexed: skipping." % the_uuid)
-                    continue
-                else:
-                    lock(cfg['session']['working_directory'], the_uuid)
-
-                # delete_dataset_from_indices(cfg, record['geonet:info']['uuid'])
-                delete_dataset_from_dest_index(cfg, the_uuid)
-                send_record_to_the_metadata_processor(cfg, record)
-
-    else:
-
-        # get all the datasets
-        for record in get_metadata_records( cfg['geonetwork']['url'],
-                                            cfg['geonetwork']['records_per_page'],
-                                            uuid=None,
-                                            the_filter=uuids_to_filter_out,
-                                            username=username, password=password ):
-
-            the_uuid = record['geonet:info']['uuid']
-
-            if is_locked( cfg['session']['working_directory'], the_uuid ):
-                logging.info("Dataset with uuid = %s is already being indexed: skipping." % the_uuid)
-                continue
-            else:
-                lock(cfg['session']['working_directory'], the_uuid)
-
-            # delete_dataset_from_indices(cfg, record['geonet:info']['uuid'])
-            delete_dataset_from_dest_index(cfg, record['geonet:info']['uuid'])
-            send_record_to_the_metadata_processor(cfg, record)
+    if 'all' in uuids_to_get:
+      uuids_to_get = [None]
+
+    for uuid_to_get in uuids_to_get:
+
+        get_metadata_records_processor(
+            the_root_url = cfg['geonetwork']['url'],
+            the_no_records_per_page = cfg['geonetwork']['records_per_page'],
+            the_uuid_to_get = uuid_to_get,
+            the_uuids_to_filter_out = uuids_to_filter_out,
+            the_username = username,
+            the_password = password)
+
+
+
+    # if 'all' not in uuids_to_get:
+    #
+    #     # get some datasets only
+    #     for uuid_to_get in uuids_to_get:
+    #         for record in get_metadata_records( cfg['geonetwork']['url'],
+    #                                             cfg['geonetwork']['records_per_page'],
+    #                                             uuid=uuid_to_get,
+    #                                             the_filter=uuids_to_filter_out,
+    #                                             username=username, password=password ):
+    #
+    #             the_uuid = record['geonet:info']['uuid']
+    #
+    #             if is_locked( cfg['session']['working_directory'], the_uuid ):
+    #                 logging.info("Dataset with uuid = %s is already being indexed: skipping." % the_uuid)
+    #                 continue
+    #             else:
+    #                 lock(cfg['session']['working_directory'], the_uuid)
+    #
+    #             # delete_dataset_from_indices(cfg, record['geonet:info']['uuid'])
+    #             delete_dataset_from_dest_index(cfg, the_uuid)
+    #             send_record_to_the_metadata_processor(cfg, record)
+    #
+    # else:
+    #
+    #     # get all the datasets
+    #     for record in get_metadata_records( cfg['geonetwork']['url'],
+    #                                         cfg['geonetwork']['records_per_page'],
+    #                                         uuid=None,
+    #                                         the_filter=uuids_to_filter_out,
+    #                                         username=username, password=password ):
+    #
+    #         the_uuid = record['geonet:info']['uuid']
+    #
+    #         if is_locked( cfg['session']['working_directory'], the_uuid ):
+    #             logging.info("Dataset with uuid = %s is already being indexed: skipping." % the_uuid)
+    #             continue
+    #         else:
+    #             lock(cfg['session']['working_directory'], the_uuid)
+    #
+    #         # delete_dataset_from_indices(cfg, record['geonet:info']['uuid'])
+    #         delete_dataset_from_dest_index(cfg, record['geonet:info']['uuid'])
+    #         send_record_to_the_metadata_processor(cfg, record)
 
 
     #connection.close()