Merge pull request #124 from ManuelHu/patch-1

legend-exp · Dec 8, 2024 · 96b68d5 · 96b68d5
2 parents 94cb680 + b44409d
commit 96b68d5
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 0 deletions.
diff --git a/src/lgdo/utils.py b/src/lgdo/utils.py
@@ -49,6 +49,14 @@ def get_element_type(obj: object) -> str:
         return "complex"
     if kind in ["S", "U"]:
         return "string"
+    if (
+        kind == "O"
+        and dt.metadata is not None
+        and dt.metadata.get("vlen", None) in (str, bytes)
+    ):
+        # variable length strings in HDF5 are read as numpy object arrays in h5py.
+        # see also h5py.check_vlen_dtype.
+        return "string"
 
     # couldn't figure it out
     msg = "cannot determine lgdo element_type for object of type"

diff --git a/tests/test_lgdo_utils.py b/tests/test_lgdo_utils.py
@@ -1,11 +1,15 @@
 from __future__ import annotations
 
+import h5py
 import numpy as np
 
 from lgdo import utils
 
 
 def test_get_element_type():
+    # variable length HD5 string datatype.
+    h5py_str_dtype = h5py.string_dtype(encoding="ascii", length=None)
+
     objs = [
         ("hi", "string"),
         (True, "bool"),
@@ -16,6 +20,7 @@ def test_get_element_type():
         (1 + 1j, "complex"),
         (b"hi", "string"),
         (np.array(["hi"]), "string"),
+        (np.array([b"hi"], h5py_str_dtype), "string"),
     ]
 
     for obj, name in objs: