adds option to keep data from bootstrap sort

- uses shutil.copy instead of shutil.move

bootstrap.py

@@ -120,7 +120,7 @@ def sort(args):
             with open(filepath, 'rb') as fd:
                 for chunk in iter(lambda: fd.read(1048577 * hash_.block_size), ''):
                     hash_.update(chunk)
-        status, detail = util.insert_file(db.acquisitions, None, None, filepath, hash_.hexdigest(), data_path, quarantine_path)
+        status, detail = util.insert_file(db.acquisitions, None, None, filepath, hash_.hexdigest(), data_path, quarantine_path, args.keep)
         if status != 200:
             print detail
 
@@ -221,6 +221,7 @@ sort_parser = subparsers.add_parser(
         formatter_class=argparse.RawDescriptionHelpFormatter,
         )
 sort_parser.add_argument('-q', '--quick', action='store_true', help='omit computing of file checksums')
+sort_parser.add_argument('-k', '--keep', action='store_true', help='do not remove the files after sorting')
 sort_parser.add_argument('db_uri', help='database URI')
 sort_parser.add_argument('path', help='filesystem path to data')
 sort_parser.add_argument('sort_path', help='filesystem path to sorted data')
@@ -236,6 +237,7 @@ dbinitsort_parser.add_argument('db_uri', help='database URI')
 dbinitsort_parser.add_argument('path', help='filesystem path to data')
 dbinitsort_parser.add_argument('sort_path', help='filesystem path to sorted data')
 dbinitsort_parser.add_argument('-q', '--quick', action='store_true', help='omit computing of file checksums')
+dbinitsort_parser.add_argument('-k', '--keep', action='store_true', help='do not remove the files after sorting')
 dbinitsort_parser.add_argument('-j', '--json', help='JSON file container users and groups')
 dbinitsort_parser.add_argument('-f', '--force', action='store_true', help='wipe out any existing db data')
 dbinitsort_parser.set_defaults(func=dbinitsort)

util.py

@@ -15,7 +15,7 @@ import scitran.data
 PROJECTION_FIELDS = ['timestamp', 'permissions', 'public']
 
 
-def insert_file(dbc, _id, file_info, filepath, digest, data_path, quarantine_path):
+def insert_file(dbc, _id, file_info, filepath, digest, data_path, quarantine_path, keep=False):
     filename = os.path.basename(filepath)
     if _id is None:
         try:
@@ -62,7 +62,10 @@ def insert_file(dbc, _id, file_info, filepath, digest, data_path, quarantine_pat
     success = dbc.update(file_spec, {'$set': {'files.$': file_info}})
     if not success['updatedExisting']:
         dbc.update({'_id': _id}, {'$push': {'files': file_info}})
-    shutil.move(filepath, container_path + '/' + filename)
+    if not keep:
+        shutil.move(filepath, container_path + '/' + filename)
+    else:
+        shutil.copy(filepath, container_path + '/' + filename)
     log.debug('Done        %s' % os.path.basename(filepath)) # must use filepath, since filename is updated for sorted files
     return 200, 'Success'