Added abort_multipart_uploads.py utility script

Author: Scott Collins

src/pds/ingress/client/abort_multipart_uploads.py

@@ -0,0 +1,74 @@
+"""
+abort_multipart_uploads.py
+
+Utility script to abort unclosed multipart uploads in an S3 bucket.
+Partially generated by GitHub Copilot.
+"""
+import json
+import subprocess
+import sys
+
+
+def list_multipart_uploads(bucket):
+    """
+    List unclosed multipart uploads in an S3 bucket.
+
+    Parameters
+    ----------
+    bucket : str
+        Name of the S3 bucket.
+
+    Returns
+    -------
+    upload_list : list
+        List of dictionaries containing information about unclosed multipart uploads.
+
+    """
+    cmd = ["aws", "s3api", "list-multipart-uploads", "--bucket", bucket, "--output", "json"]
+    result = subprocess.run(cmd, capture_output=True, text=True, check=True)
+    upload_ids = json.loads(result.stdout)
+    print("Found %d unclosed multipart uploads in bucket '%s'." % (len(upload_ids.get("Uploads", [])), bucket))
+    return upload_ids["Uploads"]
+
+
+def abort_multipart_upload(bucket, key, upload_id):
+    """
+    Abort a specific multipart upload in an S3 bucket.
+
+    Parameters
+    ----------
+    bucket : str
+        Name of the S3 bucket.
+    key : str
+        S3 key for the object being multipart uploaded.
+    upload_id : dict
+        Dictionary containing the upload ID and other metadata.
+
+    """
+    cmd = ["aws", "s3api", "abort-multipart-upload", "--bucket", bucket, "--key", key, "--upload-id", upload_id]
+    subprocess.run(cmd, check=True)
+    print(f"Aborted multipart upload: {key=} {upload_id=}")
+
+
+def main():
+    """Main function to abort multipart uploads in an S3 bucket."""
+    if len(sys.argv) != 2:
+        print("Usage: python abort_multipart_uploads.py <bucket>")
+        sys.exit(1)
+
+    bucket = sys.argv[1]
+
+    try:
+        upload_ids = list_multipart_uploads(bucket)
+        if not upload_ids:
+            print("No multipart uploads found for the specified bucket.")
+            return
+        for upload_id in upload_ids:
+            abort_multipart_upload(bucket, upload_id["Key"], upload_id["UploadId"])
+    except subprocess.CalledProcessError as e:
+        print("Error:", e.stderr)
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    main()