fix str in atoms.arrays

tests/conftest.py

@@ -1,3 +1,4 @@
+import ase.build
 import ase.collections
 import numpy as np
 import pytest
@@ -132,3 +133,16 @@ def s22_no_ascii() -> list[ase.Atoms]:
         atoms.info["config"] = "βγ"
         images.append(atoms)
     return images
+
+
+@pytest.fixture
+def frames_with_residuenames() -> list[ase.Atoms]:
+    water = ase.build.molecule("H2O")
+    # typical PDB array data
+    water.arrays["residuenames"] = np.array(["H2O"] * len(water))
+    water.arrays["atomtypes"] = np.array(["γO", "βH", "βH"])
+
+    ethane = ase.build.molecule("C2H6")
+    ethane.arrays["residuenames"] = np.array(["C2H6"] * len(ethane))
+    ethane.arrays["atomtypes"] = np.array(["γC", "βH", "βH", "βH", "βH", "βH"])
+    return [water, ethane]

tests/test_string_arrays.py

@@ -0,0 +1,14 @@
+import numpy.testing as npt
+
+import znh5md
+
+
+def test_info_non_ascii(tmp_path, frames_with_residuenames):
+    io = znh5md.IO(tmp_path / "test.h5")
+    # io = znh5md.IO("test.h5")
+    io.extend(frames_with_residuenames)
+
+    for a, b in zip(io, frames_with_residuenames):
+        for key in b.arrays:
+            npt.assert_array_equal(a.arrays[key], b.arrays[key])
+            # test json per atom?

tests/test_znh5md.py

@@ -3,3 +3,10 @@
 
 def test_version():
     assert znh5md.__version__ == "0.3.7"
+
+
+def test_creator(tmp_path):
+    io = znh5md.IO(tmp_path / "test.h5")
+    # These are the defaults
+    assert io.creator == "znh5md"
+    assert io.creator_version == znh5md.__version__

znh5md/config.py

@@ -0,0 +1,4 @@
+import numpy as np
+
+NUMERIC_FILL_VALUE = np.nan
+STRING_FILL_VALUE = b""

znh5md/format.py

@@ -218,6 +218,12 @@ def extract_atoms_data(atoms: Atoms, use_ase_calc: bool = True) -> ASEData:  # n
         if use_ase_calc and key in all_properties:
             raise ValueError(f"Key {key} is reserved for ASE calculator results.")
         if key not in ASE_TO_H5MD:
+            if isinstance(value, np.ndarray):
+                # check if dtype is <U1 or <U2
+                if value.dtype.kind == "U":
+                    value = np.array(
+                        [x.encode() for x in value.tolist()], dtype=NUMPY_STRING_DTYPE
+                    )
             particles[key] = value
 
     time: Optional[float] = atoms.info.get(CustomINFOData.h5md_time.name, None)

znh5md/io.py

@@ -1,5 +1,6 @@
 import contextlib
 import dataclasses
+import importlib.metadata
 import json
 import os
 import pathlib
@@ -17,6 +18,8 @@
 import znh5md.format as fmt
 from znh5md import utils
 
+__version__ = importlib.metadata.version("znh5md")
+
 # TODO: use pint to convert the units in the h5md file to ase units
 
 
@@ -41,8 +44,8 @@ class IO(MutableSequence):
     save_units: bool = True  # Export ASE units into the H5MD file
     author: str = "N/A"
     author_email: str = "N/A"
-    creator: str = "N/A"
-    creator_version: str = "N/A"
+    creator: str = "znh5md"
+    creator_version: str = __version__
     particle_group: Optional[str] = None
     compression: Optional[str] = "gzip"
     compression_opts: Optional[int] = None

znh5md/utils.py

@@ -3,6 +3,8 @@
 import numpy as np
 from ase.calculators.singlepoint import SinglePointCalculator
 
+from znh5md.config import NUMERIC_FILL_VALUE, STRING_FILL_VALUE
+
 NUMPY_STRING_DTYPE = np.dtype("S512")
 
 
@@ -32,9 +34,18 @@ def concatenate_varying_shape_arrays(arrays: list[np.ndarray]) -> np.ndarray:
     max_n_particles = max(x.shape[0] for x in arrays)
     dimensions = arrays[0].shape[1:]
 
-    result = np.full((len(arrays), max_n_particles, *dimensions), np.nan)
-    for i, x in enumerate(arrays):
-        result[i, : x.shape[0], ...] = x
+    if arrays[0].dtype == NUMPY_STRING_DTYPE:
+        result = np.full(
+            (len(arrays), max_n_particles), STRING_FILL_VALUE, dtype=NUMPY_STRING_DTYPE
+        )
+        for i, x in enumerate(arrays):
+            result[i, : x.shape[0]] = x
+    else:
+        result = np.full(
+            (len(arrays), max_n_particles, *dimensions), NUMERIC_FILL_VALUE
+        )
+        for i, x in enumerate(arrays):
+            result[i, : x.shape[0], ...] = x
     return result
 
 
@@ -58,6 +69,8 @@ def remove_nan_rows(array: np.ndarray) -> np.ndarray | None:
     1
 
     """
+    if isinstance(array, np.ndarray) and array.dtype == object:
+        return np.array([x.decode() for x in array if x != STRING_FILL_VALUE])
     if np.isnan(array).all():
         return None
     if len(np.shape(array)) == 0: