fix path standardization when bucket name not in path

chanzuckerberg · Oct 16, 2024 · 655f103 · 655f103
1 parent 31ec71a
commit 655f103
Showing 1 changed file with 5 additions and 1 deletion.
diff --git a/ingestion_tools/scripts/common/config.py b/ingestion_tools/scripts/common/config.py
@@ -251,4 +251,8 @@ def to_formatted_path(self, path: str) -> str:
         """
         Returns the s3 key, without the bucket name, for the given path. Helpful for formatting paths in metadata.
         """
-        return os.path.relpath(path, self.output_prefix) if path else None
+        if path:
+            if path.startswith(self.output_prefix):
+                return os.path.relpath(path, self.output_prefix)
+            return path
+        return None