Update download shared history cutoff time

ohsu-comp-bio · Mar 23, 2021 · a8139eb · a8139eb
1 parent 4ab287b
commit a8139eb
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 8 deletions.
diff --git a/cycif_db/galaxy_download/_core.py b/cycif_db/galaxy_download/_core.py
@@ -38,7 +38,6 @@ def download_datasets(destination, *datasets, server=None, api_key=None,
     if not gi:
         gi = galaxy_client(server=server, api_key=api_key)
     dataset_cli = galaxy.datasets.DatasetClient(gi)
-    his_cli = galaxy.histories.HistoryClient(gi)
 
     if isinstance(destination, str):
         destination = pathlib.Path(destination)
@@ -54,11 +53,9 @@ def download_datasets(destination, *datasets, server=None, api_key=None,
         dataset_cli.download_dataset(dataset_id, str(destination),
                                      use_default_filename=True)
 
-    annotation = {"server": server or gi.url[:-4]}
+    annotation = {"server": server or gi.url[:-3]}
     history_id = dataset_cli.show_dataset(dataset_id)['history_id']
-    history_username_and_slug = \
-        his_cli.show_history(history_id)['username_and_slug']
-    annotation['history_username_and_slug'] = history_username_and_slug
+    annotation['history_id'] = history_id
     annotation['datasets'] = datasets
 
     with open(destination.joinpath('annotation.txt'), 'w') as fp:
@@ -82,6 +79,11 @@ def find_markers_csv_and_quantification(his_client, history_id,
     --------
     None or tuple of dataset_ids ({quantification}, {markers_csv}).
     """
+    is_ok = his_client.get_histories(history_id, deleted=False)
+    if not is_ok:
+        log.error(f"The history `{history_id}` was deleted?")
+        return
+
     contents = his_client.show_history(history_id, contents=True,
                                        deleted=False, types=['dataset'])
     contents = [dataset for dataset in contents if dataset['state'] == 'ok']
@@ -117,7 +119,8 @@ def is_marker_csv(dataset_meta) -> bool:
                   if is_naivestate(dataset)]
 
     if not ns_dataset:
-        log.warn("Error: make sure the history is completed successfully! %s."
+        log.warn("Error: make sure the history runs cycif workflow and is "
+                 "completed successfully! %s."
                  % history_id)
         return
 

diff --git a/cycif_db/galaxy_download/_list_shared.py b/cycif_db/galaxy_download/_list_shared.py
@@ -25,7 +25,7 @@ class SharedGalaxy(GalaxyDriver):
     """
     def __init__(self, browser='Chrome', headless=True, server=None,
                  username=None, password=None, wait_time=10,
-                 cutoff_time='2021-01-01',
+                 cutoff_time='2021-02-06',
                  **kwargs) -> None:
         super().__init__(browser=browser, headless=headless, server=server,
                          username=username, password=password, **kwargs)

diff --git a/scripts/download_shared_histories.py b/scripts/download_shared_histories.py
@@ -18,6 +18,9 @@
 parser.add_argument(
     'destination', type=str,
     help="The folder to save the downloaded files.")
+parser.add_argument(
+    '-c', '--cutoff_time', type=str,
+    help="Download histories whose update time are later than cutoff time.")
 parser.add_argument(
     '-v', '--verbose', default=False, action='store_true',
     help="Show detailed log.")
@@ -32,6 +35,10 @@
 if args.debug:
     logging.basicConfig(level=logging.DEBUG)
 
-shared = SharedGalaxy(browser='Chrome', headless=True, cutoff_time='2021-01-13')
+cutoff_time = args.cutoff_time
+if not cutoff_time:
+    cutoff_time = '2021-02-06'
+
+shared = SharedGalaxy(browser='Chrome', headless=True, cutoff_time=cutoff_time)
 shared.download(args.destination, server=args.server, api_key=args.api_key)
 shared.quit()