add upload from single dir

action.yml

@@ -11,6 +11,9 @@ inputs:
     DATAVERSE_DATASET_DOI:
       required: true
       description: "existing dataverse dataset doi"
+    GITHUB_DIR:
+      required: false
+      description: "upload only from specific dirs"
 
 runs:
     using: "composite"
@@ -37,4 +40,4 @@ runs:
             "${{inputs.DATAVERSE_SERVER}}" 
             "${{inputs.DATAVERSE_DATASET_DOI}}"
             $GITHUB_REPOSITORY
-            
+            -d "${{inputs.GITHUB_DIR}}" 

dataverse.py

@@ -1,52 +1,69 @@
 from pyDataverse.api import NativeApi, DataAccessApi
 from pyDataverse.models import Datafile
-from os.path import isdir, isfile, join
+from os.path import isdir, join
+from time import sleep
 from os import walk 
+import argparse
 import requests
-import hashlib
 import sys
 
-dataverse_token = sys.argv[1]
-dataverse_server = sys.argv[2].strip("/")
-dataverse_dataset_doi = sys.argv[3]
-github_repository = sys.argv[4]
+def parse_arguments():
+    parser = argparse.ArgumentParser()
 
-api = NativeApi(dataverse_server, dataverse_token)
-data_api = DataAccessApi(dataverse_server)
+    # Mandatory arguments
+    parser.add_argument("token", help="Dataverse token.")
+    parser.add_argument("server", help="Dataverse server.")
+    parser.add_argument("doi", help="Dataset DOI.")
+    parser.add_argument("repo", help="GitHub repository.")
 
-# the following deletes all the files in the dataset 
+    # Optional arguments
+    parser.add_argument("-d", "--dir", help="Uploads only a specific dir.")
+
+    args = parser.parse_args()
+    return args
+
+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 
+
+    dataset = api.get_dataset(args.doi)
+    files_list = dataset.json()['data']['latestVersion']['files']
+
+    delete_api = dataverse_server + \
+                 '/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 
+
+    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())
+         sleep(0.05) # give some time to upload 
+
+    # publish updated dataset
+
+    resp = api.publish_dataset(args.doi, release_type="major")
 
-dataset = api.get_dataset(dataverse_dataset_doi)
-files_list = dataset.json()['data']['latestVersion']['files']
-
-delete_api = dataverse_server + \
-             '/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 = (dataverse_token  , ""))
-
-# the following adds all files from the repository to Dataverse 
-
-for root, subdirs, files in walk('repo'):
-  if '.git' in subdirs:
-     subdirs.remove('.git')
-  if '.github' in subdirs:
-     subdirs.remove('.github')
-  for f in files:
-     df = Datafile() 
-     df.set({
-         "pid" : dataverse_dataset_doi,
-         "filename" : f,
-         "directoryLabel": root[5:], 
-         "description" : \
-           "Uploaded with GitHub Action from {}.".format(
-             github_repository), 
-          })
-     resp = api.upload_datafile(
-             dataverse_dataset_doi, join(root,f), df.json())
-
-# publish updated dataset
-
-resp = api.publish_dataset(dataverse_dataset_doi, release_type="major")