check dataset locks

Author: atrisovic

dataverse.py

@@ -1,13 +1,13 @@
-from pyDataverse.api import NativeApi, DataAccessApi
-from pyDataverse.models import Datafile
-from os.path import isdir, join
-from time import sleep
-from os import walk 
 import argparse
+from time import sleep
+from os.path import join
+from os import walk
 import requests
-import sys
+from pyDataverse.api import NativeApi
+from pyDataverse.models import Datafile
 
 def parse_arguments():
+    """ Parses cmd-line arguments """
     parser = argparse.ArgumentParser()
 
     # Mandatory arguments
@@ -18,65 +18,72 @@ def parse_arguments():
 
     # Optional arguments
     parser.add_argument("-d", "--dir", help="Uploads only a specific dir.")
-    
-    args = parser.parse_args()
-    return args
-    
-
-def check_dataset_lock(dataset_dbid):
-    query_str = '/datasets/' + str(dataset_dbid) + '/locks'
-    params = {}
-    resp = api.get_request(query_str, params=params, auth=True)
-    locks = resp.json()['data']
-    if (locks):
-        print('Lock found for dataset id ' + str(dataset_dbid) + '... sleeping...')
-        time.sleep(2)
-        check_dataset_lock(dataset_dbid)
-
-
-if __name__ == '__main__':        
+
+    args_ = parser.parse_args()
+    return args_
+
+
+def check_dataset_lock(num):
+    """ Gives Dataverse server more time for upload """
+    if num <= 1:
+        print('Lock found for dataset id ' + \
+          str(dataset_dbid) + '\nTry again later!')
+        return
+
+    query_str = dataverse_server + \
+         '/api/datasets/' + str(dataset_dbid) + '/locks/'
+    resp_ = requests.get(query_str, auth = (token, ""))
+    locks = resp_.json()['data']
+
+    if bool(locks):
+        print('Lock found for dataset id ' + \
+           str(dataset_dbid) + '\n... sleeping...')
+        sleep(2)
+        check_dataset_lock(num-1)
+    return
+
+
+if __name__ == '__main__':
     args = parse_arguments()
-    dataverse_server = args.server.strip("/") 
-    api = NativeApi(dataverse_server , args.token)
-    data_api = DataAccessApi(dataverse_server)
-    
-    # the following deletes all the files in the dataset 
-    
+    token = args.token
+    dataverse_server = args.server.strip("/")
+    api = NativeApi(dataverse_server , token)
+
+    # the following deletes all the files in the dataset
     dataset = api.get_dataset(args.doi)
     files_list = dataset.json()['data']['latestVersion']['files']
-    
+    dataset_dbid = dataset.json()['data']['id']
+
     delete_api = dataverse_server + \
-                 '/dvn/api/data-deposit/v1.1/swordv2/edit-media/file/' 
+        '/dvn/api/data-deposit/v1.1/swordv2/edit-media/file/'
     for f in files_list:
-       fileid = f["dataFile"]["id"]
-       resp = requests.delete(
-           delete_api + str(fileid), \
-           auth = (args.token  , ""))
-    
-    # the following adds all files from the repository to Dataverse 
-    
+        fileid = f["dataFile"]["id"]
+        resp = requests.delete(
+            delete_api + str(fileid), \
+            auth = (token  , ""))
+
+    # the following adds all files from the repository to Dataverse
     path = join('repo',args.dir) if args.dir else 'repo'
+
     for root, subdirs, files in walk(path):
-      if '.git' in subdirs:
-         subdirs.remove('.git')
-      if '.github' in subdirs:
-         subdirs.remove('.github')
-      for f in files:
-         df = Datafile() 
-         df.set({
-             "pid" : args.doi,
-             "filename" : f,
-             "directoryLabel": root[5:], 
-             "description" : \
-               "Uploaded with GitHub Action from {}.".format(
-                 args.repo), 
-              })
-         resp = api.upload_datafile(
-                 args.doi, join(root,f), df.json())
-         check_dataset_lock(
-                 resp['data']['files'][0]['dataFile']['id'])
-    
+        if '.git' in subdirs:
+            subdirs.remove('.git')
+        if '.github' in subdirs:
+            subdirs.remove('.github')
+        for f in files:
+            df = Datafile()
+            df.set({
+                "pid" : args.doi,
+                "filename" : f,
+                "directoryLabel": root[5:],
+                "description" : \
+                  "Uploaded with GitHub Action from {}.".format(
+                    args.repo),
+                })
+            resp = api.upload_datafile(
+                args.doi, join(root,f), df.json())
+            check_dataset_lock(5)
+
     # publish updated dataset
-    
     resp = api.publish_dataset(args.doi, release_type="major")