use encode for non-ascii

.gitignore

@@ -162,3 +162,4 @@ cython_debug/
 .idea/
 
 .vscode/
+*.h5

tests/conftest.py

@@ -123,3 +123,12 @@ def s22_illegal_calc_results() -> list[ase.Atoms]:
 def water() -> list[ase.Atoms]:
     """Get a dataset without positions."""
     return [ase.Atoms("H2O")]
+
+
+@pytest.fixture
+def s22_no_ascii() -> list[ase.Atoms]:
+    images = []
+    for atoms in ase.collections.s22:
+        atoms.info["config"] = "βγ"
+        images.append(atoms)
+    return images

tests/test_non_ascii.py

@@ -0,0 +1,10 @@
+import znh5md
+
+
+def test_info_non_ascii(tmp_path, s22_no_ascii):
+    # io = znh5md.IO(tmp_path / "test.h5")
+    io = znh5md.IO("test.h5")
+    io.extend(s22_no_ascii)
+
+    for a, b in zip(io, s22_no_ascii):
+        assert a.info["config"] == b.info["config"] == "βγ"

tests/test_znh5md.py

@@ -2,4 +2,4 @@
 
 
 def test_version():
-    assert znh5md.__version__ == "0.3.6"
+    assert znh5md.__version__ == "0.3.7"

znh5md/format.py

@@ -193,7 +193,6 @@ def extract_atoms_data(atoms: Atoms, use_ase_calc: bool = True) -> ASEData:  # n
             key = "force" if key == "forces" else key
             uses_calc.append(key)
             value = np.array(value) if isinstance(value, (int, float, list)) else value
-
             if value.ndim > 1 and value.shape[0] == len(atomic_numbers):
                 particles[key] = value
             else:
@@ -206,9 +205,11 @@ def extract_atoms_data(atoms: Atoms, use_ase_calc: bool = True) -> ASEData:  # n
             if isinstance(value, str):
                 if len(value) > NUMPY_STRING_DTYPE.itemsize:
                     raise ValueError(f"String {key} is too long to be stored.")
-                info_data[key] = np.array(value, dtype=NUMPY_STRING_DTYPE)
+                info_data[key] = np.array(value.encode(), dtype=NUMPY_STRING_DTYPE)
             elif isinstance(value, dict):
-                info_data[key] = np.array(json.dumps(value), dtype=NUMPY_STRING_DTYPE)
+                info_data[key] = np.array(
+                    json.dumps(value).encode(), dtype=NUMPY_STRING_DTYPE
+                )
                 metadata[key] = {"unit": None, "calc": False, "type": "json"}
             else:
                 info_data[key] = value