Cleanup script also clears errors for revalidate

Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>

1 files changed, 20 insertions, 2 deletions

diff --git a/scripts/cleanup.py b/scripts/cleanup.py
index f4bd0b4..6a82659 100644
--- a/scripts/cleanup.py
+++ b/scripts/cleanup.py
@@ -3,18 +3,36 @@ import arvados.util
 
 api = arvados.api()
 
-patterns = [
+delete_patterns = [
     "%missing%`collection_location`%",
     "%missing%`technology`%",
     "%missing%`host_species`%",
     "%QC fail: alignment%",
     "%does not look like a valid URI%",
+    "%Duplicate of%"
     ]
 
-for p in patterns:
+revalidate_patterns = [
+    "%missing%`license`%"
+]
+
+for p in delete_patterns:
     c = arvados.util.list_all(api.collections().list, filters=[
         ["owner_uuid", "=", "lugli-j7d0g-n5clictpuvwk8aa"],
         ["properties.errors", "like", p]])
     for i in c:
         print("trashing %s %s" % (i["uuid"], i["properties"].get("sequence_label")))
         api.collections().delete(uuid=i["uuid"]).execute()
+
+for p in revalidate_patterns:
+    c = arvados.util.list_all(api.collections().list, filters=[
+        ["owner_uuid", "=", "lugli-j7d0g-n5clictpuvwk8aa"],
+        ["properties.errors", "like", p]])
+    for i in c:
+        print("clearing status %s %s" % (i["uuid"], i["properties"].get("sequence_label")))
+        pr = i["properties"]
+        if "status" in pr:
+            del pr["status"]
+        if "errors" in pr:
+            del pr["errors"]
+        api.collections().update(uuid=i["uuid"], body={"properties": pr}).execute()