Resolve issue where segy_to_mdio ingestion failed in trace ingestion phase if output contained a UPath with storage_options

Author: BrianMichell

src/mdio/api/io.py

@@ -30,7 +30,9 @@ def _normalize_path(path: UPath | Path | str) -> UPath:
 
 
 def _normalize_storage_options(path: UPath) -> dict[str, Any] | None:
-    return None if len(path.storage_options) == 0 else path.storage_options
+    # UPath.storage_options returns a read-only mappingproxy which cannot be pickled. Copy it into a
+    # plain dict so callers can safely pass it across process boundaries (e.g. spawned workers).
+    return None if len(path.storage_options) == 0 else dict(path.storage_options)
 
 
 def open_mdio(input_path: UPath | Path | str, chunks: T_Chunks = None) -> xr_Dataset:

tests/unit/test_io.py

@@ -0,0 +1,21 @@
+"""Tests for low-level MDIO API I/O helpers."""
+
+from __future__ import annotations
+
+from types import MappingProxyType
+
+from upath import UPath
+
+from mdio.api.io import _normalize_storage_options
+
+
+def test_normalize_storage_options_is_not_mappingproxy() -> None:
+    """Storage options must not be a mappingproxy.
+
+    `UPath.storage_options` returns a read-only ``mappingproxy`` that cannot be pickled. Blocked-I/O
+    ingestion passes these options into ``ProcessPoolExecutor`` initargs, so a mappingproxy breaks
+    spawned workers with ``TypeError: cannot pickle 'mappingproxy' object``.
+    """
+    storage_options = _normalize_storage_options(UPath("s3://bucket/key", key="access", secret="secret"))
+
+    assert not isinstance(storage_options, MappingProxyType)