feat: add mcstas h5 file loader

Author: jokasimr

src/ess/estia/mcstas.py

@@ -1,6 +1,9 @@
+import h5py
 import numpy as np
 import scipp as sc
 
+from ess.reflectometry.load import load_h5
+
 
 def parse_metadata_ascii(lines):
     data = {}
@@ -50,12 +53,26 @@ def parse_events_ascii(lines):
 
 
 def parse_events_h5(f):
-    pass
-
+    if isinstance(f, str):
+        with h5py.File(f) as ff:
+            return parse_events_h5(ff)
 
-'''
-def parse_events_h5(f):
-    f['entry1/data']
-    events = load_nx(f, 'NXentry/NXdetector/NXdata')
-    parameters = load_nx(f, 'NXentry/simulation/Param')
-'''
+    data, events, params = load_h5(
+        f,
+        'NXentry/NXdetector/NXdata',
+        'NXentry/NXdetector/NXdata/events',
+        'NXentry/simulation/Param',
+    )
+    da = sc.DataArray(
+        sc.array(dims=['events'], values=events[:, 0], variances=events[:, 0] ** 2),
+    )
+    for i, label in enumerate(data.attrs["ylabel"].decode().strip().split(' ')):
+        if label == 'p':
+            continue
+        da.coords[label] = sc.array(dims=['events'], values=events[:, i])
+    for k, v in params.items():
+        v = v[0]
+        if isinstance(v, bytes):
+            v = v.decode()
+        da.coords[k] = sc.scalar(v)
+    return da

src/ess/reflectometry/load.py

@@ -1,6 +1,6 @@
 # SPDX-License-Identifier: BSD-3-Clause
 # Copyright (c) 2024 Scipp contributors (https://github.com/scipp)
-
+import h5py
 import sciline
 import scipp as sc
 import scippnexus as snx
@@ -37,6 +37,34 @@ def _unique_child_group(
     return next(iter(children.values()))  # type: ignore[return-value]
 
 
+def load_h5(group: h5py.Group | str, *paths: str):
+    if isinstance(group, str):
+        with h5py.File(group) as group:
+            yield from load_h5(group, *paths)
+        return
+    for path in paths:
+        g = group
+        for p in path.strip('/').split('/'):
+            g = _unique_child_group_h5(g, p) if p.startswith('NX') else g.get(p)
+        yield g
+
+
+def _unique_child_group_h5(
+    group: h5py.Group,
+    nx_class: str,
+) -> h5py.Group | None:
+    out = None
+    for v in group.values():
+        if v.attrs.get("NX_class") == nx_class.encode():
+            if out is None:
+                out = v
+            else:
+                raise ValueError(
+                    f'Expected exactly one {nx_class} group, but found more'
+                )
+    return out
+
+
 def save_reference(pl: sciline.Pipeline, fname: str):
     pl.compute(ReducedReference).save_hdf5(fname)
     return fname