From d499901faaa07a48384dc8f3ae5a79de839334bc Mon Sep 17 00:00:00 2001
From: Arkadii Yakovets <ark@ccdatalab.org>
Date: Mon, 11 Dec 2023 09:50:43 -0800
Subject: [PATCH] Add a prototype of Sample::developmental_stage backfill
 script

---
 .../migrations/0075_sample_last_refreshed.py  | 18 ++++++++
 common/data_refinery_common/models/sample.py  |  4 ++
 .../commands/refresh_sample_metadata.py       | 43 +++++++++++++++++++
 3 files changed, 65 insertions(+)
 create mode 100644 common/data_refinery_common/migrations/0075_sample_last_refreshed.py
 create mode 100644 foreman/data_refinery_foreman/foreman/management/commands/refresh_sample_metadata.py

diff --git a/common/data_refinery_common/migrations/0075_sample_last_refreshed.py b/common/data_refinery_common/migrations/0075_sample_last_refreshed.py
new file mode 100644
index 000000000..f4f2d2289
--- /dev/null
+++ b/common/data_refinery_common/migrations/0075_sample_last_refreshed.py
@@ -0,0 +1,18 @@
+# Generated by Django 3.2.18 on 2023-12-08 00:45
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("data_refinery_common", "0074_sample_developmental_stage"),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name="sample",
+            name="last_refreshed",
+            field=models.DateTimeField(auto_now=True, null=True),
+        ),
+    ]
diff --git a/common/data_refinery_common/models/sample.py b/common/data_refinery_common/models/sample.py
index 9d2152496..1ae132e82 100644
--- a/common/data_refinery_common/models/sample.py
+++ b/common/data_refinery_common/models/sample.py
@@ -94,6 +94,10 @@ def __str__(self):
     created_at = models.DateTimeField(editable=False, default=timezone.now)
     last_modified = models.DateTimeField(default=timezone.now)
 
+    # Auxiliary field for tracking latest metadata update time.
+    # Originally added to support Sample::developmental_stage values backfilling.
+    last_refreshed = models.DateTimeField(auto_now=True, null=True)
+
     def save(self, *args, **kwargs):
         """On save, update timestamps"""
         current_time = timezone.now()
diff --git a/foreman/data_refinery_foreman/foreman/management/commands/refresh_sample_metadata.py b/foreman/data_refinery_foreman/foreman/management/commands/refresh_sample_metadata.py
new file mode 100644
index 000000000..c69f6ee30
--- /dev/null
+++ b/foreman/data_refinery_foreman/foreman/management/commands/refresh_sample_metadata.py
@@ -0,0 +1,43 @@
+import time
+
+from django.core.management.base import BaseCommand
+
+from data_refinery_common.logging import get_and_configure_logger
+from data_refinery_common.models import Sample
+from data_refinery_foreman.surveyor.sra import SraSurveyor
+
+logger = get_and_configure_logger(__name__)
+
+
+class Command(BaseCommand):
+    def add_arguments(self, parser):
+        parser.add_argument(
+            "--limit",
+            default=1000,
+            type=int,
+            help="Number of samples to refresh",
+        )
+        parser.add_argument(
+            "--source",
+            choices=("SRA",),
+            required=True,
+            type=str,
+            help="Source name (ARRAY_EXPRESS, GEO, SRA)",
+        )
+
+    def handle(self, *args, **options):
+        for sample in Sample.objects.filter(
+            developmental_stage__isnull=True,
+            last_refreshed__isnull=True,
+            source_database=options["source"],
+        ).order_by("id")[: options["limit"]]:
+            logger.info(f"Refreshing metadata for a sample {sample.accession_code}")
+            try:
+                _, sample_metadata = SraSurveyor.gather_all_metadata(sample.accession_code)
+                SraSurveyor._apply_harmonized_metadata_to_sample(sample_metadata)
+            except Exception as e:
+                logger.exception(e)
+            finally:
+                sample.save()
+
+            time.sleep(1)