reconcile

dsc/workflows/opencourseware.py

@@ -1,5 +1,6 @@
 import json
 import zipfile
+from collections import defaultdict
 from collections.abc import Iterator
 from typing import Any
 
@@ -19,6 +20,29 @@ class OpenCourseWare(Workflow):
     workflow_name: str = "opencourseware"
     metadata_mapping_path: str = "dsc/workflows/metadata_mapping/opencourseware.json"
 
+    def _build_bitstream_dict(self) -> dict:
+        """Build dictionary of bitstreams.
+
+        This method will look for zip files within the designated 'batch' folder
+        of the S3 bucket (i.e., self.batch_path). In the case of OpenCourseWare
+        deposits, this method expects:
+            * a single (1) zipped file per item identifier
+            * filename of zipped file must correspond to item identifier
+              (i.e., '<item_identifier>'.zip)
+        """
+        s3_client = S3Client()
+        bitstreams = list(
+            s3_client.files_iter(
+                bucket=self.s3_bucket, prefix=self.batch_path, file_type=".zip"
+            )
+        )
+        bitstream_dict: dict[str, list[str]] = defaultdict(list)
+        for bitstream in bitstreams:
+            file_name = bitstream.split("/")[-1]
+            item_identifier = file_name.replace(".zip", "")
+            bitstream_dict[item_identifier].append(bitstream)
+        return bitstream_dict
+
     def item_metadata_iter(self) -> Iterator[dict[str, Any]]:
         """Yield source metadata from metadata JSON file in the zip file.
 

tests/test_workflow_opencourseware.py

@@ -22,6 +22,30 @@
 ]
 
 
+def test_workflow_ocw_build_bitstream_dict(
+    mocked_s3, opencourseware_workflow_instance, s3_client
+):
+    s3_client.put_file(
+        file_content="",
+        bucket="dsc",
+        key="opencourseware/batch-aaa/123.zip",
+    )
+    s3_client.put_file(
+        file_content="",
+        bucket="dsc",
+        key="opencourseware/batch-aaa/124.zip",
+    )
+    s3_client.put_file(
+        file_content="",
+        bucket="dsc",
+        key="opencourseware/batch-aaa/ignore_me.txt",
+    )
+    assert opencourseware_workflow_instance._build_bitstream_dict() == {
+        "123": ["opencourseware/batch-aaa/123.zip"],
+        "124": ["opencourseware/batch-aaa/124.zip"],
+    }
+
+
 @patch("zipfile.ZipFile")
 @patch("smart_open.open")
 def test_workflow_ocw_extract_metadata_from_zip_file_success(