adds a list barcodes in input bucket method to digifeeds cli

aim/cli/digifeeds.py

@@ -1,7 +1,9 @@
 import typer
 from aim.digifeeds.add_to_db import add_to_db as add_to_digifeeds_db
+from aim.digifeeds.list_barcodes_in_bucket import list_barcodes_in_bucket
 from aim.digifeeds.database import models, main
 
+
 app = typer.Typer()
 
 
@@ -15,3 +17,8 @@ def add_to_db(barcode: str):
 def load_statuses():
     with main.SessionLocal() as db_session:
         models.load_statuses(session=db_session)
+
+
+@app.command()
+def list_barcodes_in_input_bucket():
+    print(list_barcodes_in_bucket(), end="")

aim/digifeeds/list_barcodes_in_bucket.py

@@ -0,0 +1,19 @@
+import boto3
+from aim.services import S
+
+
+def list_barcodes_in_bucket():
+    s3 = boto3.client(
+        "s3",
+        aws_access_key_id=S.digifeeds_s3_access_key,
+        aws_secret_access_key=S.digifeeds_s3_secret_access_key,
+    )
+    prefix = S.digifeeds_s3_input_path + "/"
+    response = s3.list_objects_v2(
+        Bucket=S.digifeeds_s3_bucket,
+        Prefix=prefix,
+        Delimiter="/",
+    )
+    paths = [object["Prefix"] for object in response["CommonPrefixes"]]
+    barcodes = [path.split("/")[1] for path in paths]
+    return barcodes

aim/services.py

@@ -12,6 +12,10 @@
         ("alma_api_url", str),
         ("digifeeds_api_url", str),
         ("digifeeds_set_id", str),
+        ("digifeeds_s3_access_key", str),
+        ("digifeeds_s3_secret_access_key", str),
+        ("digifeeds_s3_bucket", str),
+        ("digifeeds_s3_input_path", str),
     ],
 )
 
@@ -29,4 +33,11 @@
     digifeeds_set_id=os.getenv("DIGIFEEDS_SET_ID") or "digifeeds_set_id",
     alma_api_key=os.getenv("ALMA_API_KEY") or "alma_api_key",
     alma_api_url="https://api-na.hosted.exlibrisgroup.com/almaws/v1",
+    digifeeds_s3_access_key=os.getenv("DIGIFEEDS_S3_ACCESS_KEY")
+    or "digifeeds_s3_access_key",
+    digifeeds_s3_secret_access_key=os.getenv("DIGIFEEDS_S3_SECRET_ACCESS_KEY")
+    or "digifeeds_s3_secret_access_key",
+    digifeeds_s3_bucket=os.getenv("DIGIFEEDS_S3_BUCKET") or "digifeeds_s3_bucket",
+    digifeeds_s3_input_path=os.getenv("DIGIFEEDS_S3_INPUT_PATH")
+    or "path_to_input_barcodes",
 )

pyproject.toml

@@ -18,6 +18,7 @@ fastapi = {extras = ["standard"], version = "^0.114.2"}
 httpx = "^0.27.2"
 alembic = "^1.13.2"
 typer = "^0.12.5"
+boto3 = "^1.35.28"
 
 
 [tool.poetry.group.dev.dependencies]
@@ -27,6 +28,7 @@ ruff = "^0.6.6"
 responses = "^0.25.3"
 pytest-socket = "^0.7.0"
 pytest-mock = "^3.14.0"
+moto = {extras = ["s3"], version = "^5.0.15"}
 
 
 [tool.pytest.ini_options]

tests/cli/test_digifeeds.py

@@ -60,3 +60,14 @@ def test_load_statuses(mocker):
     assert session_local_mock.call_count == 1
     assert load_statuse_mock.call_count == 1
     assert result.exit_code == 0
+
+
+def test_list_barcodes_in_input_bucket(mocker):
+    list_barcodes_mock = mocker.patch(
+        "aim.cli.digifeeds.list_barcodes_in_bucket",
+        return_value=["barcode1", "barcode2"],
+    )
+    result = runner.invoke(app, ["digifeeds", "list-barcodes-in-input-bucket"])
+    assert list_barcodes_mock.call_count == 1
+    assert result.exit_code == 0
+    assert "['barcode1', 'barcode2']" == result.stdout

tests/digifeeds/test_list_barcodes_in_bucket.py

@@ -0,0 +1,26 @@
+import boto3
+from moto import mock_aws
+from aim.services import S
+from aim.digifeeds.list_barcodes_in_bucket import list_barcodes_in_bucket
+
+
+@mock_aws
+def test_list_barcodes_in_bucket():
+    conn = boto3.resource(
+        "s3",
+        aws_access_key_id=S.digifeeds_s3_access_key,
+        aws_secret_access_key=S.digifeeds_s3_secret_access_key,
+    )
+    conn.create_bucket(Bucket=S.digifeeds_s3_bucket)
+
+    barcode1 = conn.Object(
+        S.digifeeds_s3_bucket, f"{S.digifeeds_s3_input_path}/barcode1/some_file.txt"
+    )
+    barcode1.put(Body="some text")
+    barcode2 = conn.Object(
+        S.digifeeds_s3_bucket, f"{S.digifeeds_s3_input_path}/barcode2/some_file.txt"
+    )
+    barcode2.put(Body="some text")
+
+    result = list_barcodes_in_bucket()
+    assert result == ["barcode1", "barcode2"]