Local file to S3 upload.

example/download.yaml

@@ -12,6 +12,24 @@
 - url: https://drive.google.com/uc?id=10ojJffrPSl12OMcu4gyx0fak2CNu6qOs
   local_name: gdrive_test_2.txt
 
+- base_url: https://rest.uniprot.org/uniprotkb/
+  api: uniprot
+  url: 'gdrive:1Ai52w4fu4XPu5w4wdE8y3rymEgN0BDqe'
+  local_name: 'uniprot_genome_features'
+  fields: [
+    "organism_id",
+    "id", 
+    "accession",
+    "protein_name",
+    "ec",
+    "ft_binding"
+  ]
+  keywords: [
+    "Reference+proteome"
+  ]
+  size: 500
+  batch_size: 1
+  test: True
 # - url: https://www.ebi.ac.uk/chembl/elk/es/
 #   api: elasticsearch
 #   query_file: example/query.json

kghub_downloader/download_utils.py

@@ -20,6 +20,8 @@
 from google.cloud.storage.blob import Blob
 from typing import List, Optional
 import gdown
+import boto3
+from botocore.exceptions import NoCredentialsError
 
 GDOWN_MAP = {"gdrive": "https://drive.google.com/uc?id="}
 
@@ -188,9 +190,20 @@ def mirror_to_bucket(local_file, bucket_url, remote_file) -> None:
             blob.upload_from_filename(local_file)
 
         elif bucket_url.startswith("s3://"):
-            raise ValueError("Currently, only Google Cloud storage is supported.")
-            # bashCommand = f"aws s3 cp {outfile} {mirror}"
-            # subprocess.run(bashCommand.split())
+            # Create an S3 client
+            s3 = boto3.client('s3')
+
+            try:
+                # Upload the file
+                s3.upload_file(local_file, bucket_name, remote_file)
+                print(f"File {local_file} uploaded to {bucket_name}/{remote_file}")
+                return True
+            except FileNotFoundError:
+                print(f"The file {local_file} was not found")
+                return False
+            except NoCredentialsError:
+                print("Credentials not available")
+                return False
 
         else:
             raise ValueError("Currently, only Google Cloud storage is supported.")