add Origin field to bag-info file, fixes #378

RockefellerArchiveCenter · Jan 19, 2020 · 3dd0673 · 3dd0673
1 parent f912b9d
commit 3dd0673
Show file tree

Hide file tree

Showing 3 changed files with 25 additions and 0 deletions.
diff --git a/aurora/bag_transfer/lib/cron.py b/aurora/bag_transfer/lib/cron.py
@@ -121,6 +121,10 @@ def do(self):
             process_status=Archives.ACCESSIONING_STARTED
         ):
             try:
+                FH.update_bag_info(
+                    join(settings.STORAGE_ROOT_DIR, archive.machine_file_identifier),
+                    {"Origin": "aurora"}
+                )
                 tar_filename = "{}.tar.gz".format(archive.machine_file_identifier)
                 FH.make_tarfile(
                     join(settings.STORAGE_ROOT_DIR, tar_filename),
@@ -166,6 +170,7 @@ def do(self):
                 )
 
                 archive.process_status = Archives.DELIVERED
+                print(archive.machine_file_identifier)
                 archive.save()
             except Exception as e:
                 print("Error delivering transfer {}: {}".format(archive, str(e)))

diff --git a/aurora/bag_transfer/lib/files_helper.py b/aurora/bag_transfer/lib/files_helper.py
@@ -7,6 +7,8 @@
 import psutil
 import pwd
 
+import bagit
+
 
 def open_files_list():
     """Return a list of files open on the linux system"""
@@ -265,3 +267,11 @@ def chown_path_to_root(file_path):
 def make_tarfile(output_filename, source_dir):
     with tarfile.open(output_filename, "w:gz") as tar:
         tar.add(source_dir, arcname=os.path.basename(source_dir))
+
+
+def update_bag_info(bag_path, data):
+    """Adds metadata to `bag-info.txt`"""
+    bag = bagit.Bag(bag_path)
+    for k,v in data.items():
+        bag.info[k] = v
+    bag.save()
diff --git a/aurora/bag_transfer/test/test_cron.py b/aurora/bag_transfer/test/test_cron.py
@@ -2,6 +2,7 @@
 import pwd
 import random
 
+import bagit
 from django.test import TransactionTestCase, Client
 from django.conf import settings
 
@@ -29,11 +30,20 @@ def test_cron(self):
             helpers.create_target_bags(ref[0], settings.TEST_BAGS_DIR, self.orgs[0], username=self.user.username)
         discovered = DiscoverTransfers().do()
         self.assertIsNot(False, discovered)
+
         for archive in Archives.objects.filter(process_status=Archives.VALIDATED):
             archive.process_status = Archives.ACCESSIONING_STARTED
             archive.save()
         delivered = DeliverTransfers().do()
         self.assertIsNot(False, delivered)
+        self.assertEqual(len(Archives.objects.filter(process_status=Archives.ACCESSIONING_STARTED)), 0)
+        self.assertEqual(
+            len(Archives.objects.filter(process_status=Archives.DELIVERED)),
+            len(os.listdir(settings.DELIVERY_QUEUE_DIR))
+        )
+        for bag_path in os.listdir(settings.DELIVERY_QUEUE_DIR):
+            bag = bagit.Bag(os.path.join(settings.DELIVERY_QUEUE_DIR, bag_path))
+            self.assertTrue('Origin' in bag.bag_info)
 
     def tearDown(self):
         helpers.delete_test_orgs(self.orgs)