adds function that lists files send to s3 in the last two weeks

aim/digifeeds/functions.py

@@ -1,6 +1,8 @@
 from aim.services import S
 import boto3
 from pathlib import Path
+from rclone_python import rclone
+from datetime import datetime
 
 
 def list_barcodes_in_input_bucket():
@@ -13,3 +15,17 @@ def list_barcodes_in_input_bucket():
     response = s3.list_objects_v2(Bucket=S.digifeeds_s3_bucket, Prefix=prefix)
     barcodes = [Path(object["Key"]).stem for object in response["Contents"]]
     return barcodes
+
+
+def barcodes_added_in_last_two_weeks():
+    files = rclone.ls(
+        path=f"{S.digifeeds_s3_rclone_remote}:{S.digifeeds_s3_input_path}",
+        args=["--max-age 14d"],
+    )
+    output = []
+    for file in files:
+        barcode = file["Name"].split(".")[0]
+        date = datetime.fromisoformat(file["ModTime"]).strftime("%Y-%m-%d")
+        output.append([date, barcode])
+
+    return output

tests/digifeeds/test_functions.py

@@ -0,0 +1,33 @@
+import json
+from aim.digifeeds.functions import rclone, barcodes_added_in_last_two_weeks
+
+
+def test_barcodes_added_in_last_two_weeks(mocker):
+    ls_data_raw = """
+    [
+      {
+        "Path": "35112203951670.zip",
+        "Name": "35112203951670.zip",
+        "Size": 554562627,
+        "MimeType": "application/zip",
+        "ModTime": "2024-12-14T02:01:05.093051502-05:00",
+        "IsDir": false,
+        "Tier": "STANDARD"
+      },
+      {
+        "Path": "39015004707009.zip",
+        "Name": "39015004707009.zip",
+        "Size": 232895588,
+        "MimeType": "application/zip",
+        "ModTime": "2024-12-14T02:02:29.111076546-05:00",
+        "IsDir": false,
+        "Tier": "STANDARD"
+      }
+    ]
+"""
+    mocker.patch.object(rclone, "ls", return_value=json.loads(ls_data_raw))
+    output = barcodes_added_in_last_two_weeks()
+    assert output == [
+        ["2024-12-14", "35112203951670"],
+        ["2024-12-14", "39015004707009"],
+    ]