[draft] EDPOPS-321 include block count in md5

solvebio/cli/data.py

@@ -295,7 +295,7 @@ def _object_exists(remote_parent, local_path, _client):
             return False
         else:
             # Check if the md5sum matches
-            local_md5 = md5sum(local_path)[0]
+            local_md5 = md5sum(local_path)
             remote_md5 = obj.get("md5")
             if remote_md5 and remote_md5 == local_md5:
                 return True
@@ -831,7 +831,7 @@ def _download_recursive(
         # Skip over files that match remote md5 checksum
         if os.path.exists(local_path):
             remote_md5 = remote_file.get("md5")
-            if remote_md5 and remote_md5 == md5sum(local_path)[0]:
+            if remote_md5 and remote_md5 == md5sum(local_path):
                 print("Skipping {} already in sync".format(local_path))
                 continue
 

solvebio/resource/object.py

@@ -600,7 +600,7 @@ def _upload_single_file(cls, obj, local_path, **kwargs):
         size = os.path.getsize(local_path)
 
         # Get MD5 for single part upload
-        local_md5, _ = md5sum(local_path, multipart_threshold=None)
+        local_md5 = md5sum(local_path, multipart_threshold=None)
 
         upload_url = obj.upload_url
 
@@ -934,10 +934,16 @@ def _upload_single_part(
                 if not chunk_data:
                     break
 
+                def md5_base64(data):
+                    import hashlib
+                    md5 = hashlib.md5(data).digest()
+                    return base64.b64encode(md5).decode("utf-8")
+
                 # Upload without requests-level retry (let our custom retry handle it)
                 session = requests.Session()
 
-                headers = {"Content-Length": str(len(chunk_data))}
+                headers = {"Content-Length": str(len(chunk_data)),
+                           "ContentMD5": md5_base64(chunk_data)}
 
                 # Calculate timeout based on part size
                 part_size_mb = len(chunk_data) / (1024 * 1024)

solvebio/utils/md5sum.py

@@ -36,4 +36,4 @@ def _read_chunks(f, chunk_size):
             for block in _read_chunks(f, multipart_chunksize):
                 md5.update(block)
 
-    return md5.hexdigest(), block_count
+    return f"{md5.hexdigest()}-{block_count}" if block_count else md5.hexdigest()