Only use files when computing model archive hash

Closes #174 (closed)

arkindex_worker/worker/training.py

@@ -30,9 +30,7 @@ FileSize = NewType("FileSize", int)
 
 
 @contextmanager
-def create_archive(
-    path: DirPath, use_parent_folder: bool = False
-) -> Tuple[Path, Hash, FileSize, Hash]:
+def create_archive(path: DirPath) -> Tuple[Path, Hash, FileSize, Hash]:
     """
     Create a tar archive from the files at the given location then compress it to a zst archive.
 
@@ -54,17 +52,21 @@ def create_archive(
     # Files hierarchy ifs kept in the archive.
     file_list = []
     with tarfile.open(path_to_tar_archive, "w") as tar:
-        if use_parent_folder:
-            # Only publish contents of the given folder without parent folder
-            tar.add(path, arcname=".")
-        else:
-            tar.add(path)
-        file_list = [member for member in tar.getnames() if os.path.isfile(member)]
+        for p in path.glob("**/*"):
+            x = p.relative_to(path)
+
+            tar.add(p, arcname=x, recursive=False)
+
+            if p.is_dir():
+                continue
+
+            file_list.append(p)
 
     # Sort by path
     file_list.sort()
     # Compute hash of the files
     for file_path in file_list:
+        print("BOOM PATH", file_path)
         with open(file_path, "rb") as file_data:
             for chunk in iter(lambda: file_data.read(CHUNK_SIZE), b""):
                 content_hasher.update(chunk)
@@ -100,6 +102,7 @@ class TrainingMixin(object):
         model_id: str,
         tag: Optional[str] = None,
         description: Optional[str] = None,
+        use_parent_folder=False,
     ):
         """
         This method creates a model archive and its associated hash,