From 7c12e4976337a063301be260cb3954bf4303f5e0 Mon Sep 17 00:00:00 2001
From: AndreaGuarracino
Date: Sat, 26 Sep 2020 12:12:17 +0200
Subject: script for processing the metadata of the ESR samples; moved
 delete_entries_on_arvados script in scripts directory

---
 scripts/delete_entries_on_arvados.py | 34 ++++++++++++++++++++++++++++++++++
 1 file changed, 34 insertions(+)
 create mode 100644 scripts/delete_entries_on_arvados.py

(limited to 'scripts/delete_entries_on_arvados.py')

diff --git a/scripts/delete_entries_on_arvados.py b/scripts/delete_entries_on_arvados.py
new file mode 100644
index 0000000..d281456
--- /dev/null
+++ b/scripts/delete_entries_on_arvados.py
@@ -0,0 +1,34 @@
+import sys
+import arvados
+import arvados.collection
+
+from datetime import datetime
+
+date_time_str = '2020-08-20'
+date_time_obj = datetime.strptime(date_time_str, '%Y-%m-%d')
+
+api = arvados.api()
+keepclient = arvados.keep.KeepClient(api_client=api)
+
+validated = arvados.util.list_all(api.collections().list, filters=[
+    ["owner_uuid", "=", sys.argv[1]],
+#    ["properties.status", "=", "validated"]
+])
+
+# validated.sort(key=lambda v: v["portable_data_hash"])
+
+num_sample_deleted = 0
+for item in validated:
+    sequence_label = item['properties']["sequence_label"]
+
+    # The SRA samples start with SRR or ERR
+    if not sequence_label.startswith('SRR') and not sequence_label.startswith('ERR'):
+        created_at_obj = datetime.strptime(item["created_at"], '%Y-%m-%dT%H:%M:%S.%fZ')
+        # print(item, created_at_obj)
+
+        if created_at_obj < date_time_obj:
+            api.collections().delete(uuid=item['current_version_uuid']).execute()
+            num_sample_deleted += 1
+            print(sequence_label)
+
+print('num_sample_deleted: {}'.format(num_sample_deleted))
-- 
cgit v1.2.3