WIP DXTBX compatibility with numpy 2.0

.azure-pipelines/ci-conda-env.txt

@@ -16,7 +16,7 @@ conda-forge::hdf5
 conda-forge::matplotlib-base
 conda-forge::mrcfile
 conda-forge::natsort
-conda-forge::numpy<2
+conda-forge::numpy
 conda-forge::nxmx
 conda-forge::orderedset
 conda-forge::pillow>=5.4.1

src/dxtbx/format/FormatHDF5EigerNearlyNexus.py

@@ -32,7 +32,7 @@ def find_entries(nx_file):
     if "entry" in nx_file:
         entry = nx_file["entry"]
         if "NX_class" in entry.attrs:
-            if entry.attrs["NX_class"] == np.string_("NXentry"):
+            if entry.attrs["NX_class"] == np.bytes_("NXentry"):
                 if "definition" not in entry:
                     return entry
     return None
@@ -49,7 +49,7 @@ def is_eiger_nearly_nexus_file(filename):
         if entry is not None:
             try:
                 return (
-                    np.string_("dectris eiger")
+                    np.bytes_("dectris eiger")
                     in entry["instrument"]["detector"]["description"][()].lower()
                 )
             except KeyError:
@@ -76,7 +76,7 @@ def create_scalar(handle, path, dtype, value):
             dataset[()] = value
 
         # Add NXmx definition
-        create_scalar(handle["entry"], "definition", "S4", np.string_("NXmx"))
+        create_scalar(handle["entry"], "definition", "S4", np.bytes_("NXmx"))
 
         # Add saturation value
         try:
@@ -100,7 +100,7 @@ def create_scalar(handle, path, dtype, value):
 
         # Add detector type
         create_scalar(
-            handle["entry/instrument/detector"], "type", "S5", np.string_("PIXEL")
+            handle["entry/instrument/detector"], "type", "S5", np.bytes_("PIXEL")
         )
 
         # Move the beam
@@ -111,7 +111,7 @@ def create_scalar(handle, path, dtype, value):
         module_path = "/entry/instrument/detector/module"
         # print "Creating detector module %s" % (module_path)
         group = handle.create_group(module_path)
-        group.attrs["NX_class"] = np.string_("NXdetector_module")
+        group.attrs["NX_class"] = np.bytes_("NXdetector_module")
 
         # Add a module index
         create_scalar(group, "module_index", "int64", 0)
@@ -174,13 +174,13 @@ def create_scalar(handle, path, dtype, value):
             "float32",
             handle["/entry/instrument/detector/x_pixel_size"][()],
         )
-        group["fast_pixel_direction"].attrs["transformation_type"] = np.string_(
+        group["fast_pixel_direction"].attrs["transformation_type"] = np.bytes_(
             "translation"
         )
         group["fast_pixel_direction"].attrs["vector"] = fast_axis
         group["fast_pixel_direction"].attrs["offset"] = (0, 0, 0)
-        group["fast_pixel_direction"].attrs["units"] = np.string_("m")
-        group["fast_pixel_direction"].attrs["depends_on"] = np.string_(depends_on)
+        group["fast_pixel_direction"].attrs["units"] = np.bytes_("m")
+        group["fast_pixel_direction"].attrs["depends_on"] = np.bytes_(depends_on)
 
         # Add slow_pixel_size dataset
         create_scalar(
@@ -189,29 +189,29 @@ def create_scalar(handle, path, dtype, value):
             "float32",
             handle["/entry/instrument/detector/y_pixel_size"][()],
         )
-        group["slow_pixel_direction"].attrs["transformation_type"] = np.string_(
+        group["slow_pixel_direction"].attrs["transformation_type"] = np.bytes_(
             "translation"
         )
         group["slow_pixel_direction"].attrs["vector"] = slow_axis
         group["slow_pixel_direction"].attrs["offset"] = (0, 0, 0)
-        group["slow_pixel_direction"].attrs["units"] = np.string_("m")
-        group["slow_pixel_direction"].attrs["depends_on"] = np.string_(depends_on)
+        group["slow_pixel_direction"].attrs["units"] = np.bytes_("m")
+        group["slow_pixel_direction"].attrs["depends_on"] = np.bytes_(depends_on)
 
         # Add module offset dataset
         # print "Set module offset to be zero relative to detector"
         create_scalar(group, "module_offset", "float32", 0)
-        group["module_offset"].attrs["transformation_type"] = np.string_("translation")
+        group["module_offset"].attrs["transformation_type"] = np.bytes_("translation")
         group["module_offset"].attrs["vector"] = (0, 0, 0)
         group["module_offset"].attrs["offset"] = (0, 0, 0)
-        group["module_offset"].attrs["units"] = np.string_("m")
-        group["module_offset"].attrs["depends_on"] = np.string_(depends_on)
+        group["module_offset"].attrs["units"] = np.bytes_("m")
+        group["module_offset"].attrs["depends_on"] = np.bytes_(depends_on)
 
         # Create detector depends_on
         create_scalar(
             handle["/entry/instrument/detector"],
             "depends_on",
             "S%d" % len(depends_on),
-            np.string_(depends_on),
+            np.bytes_(depends_on),
         )
 
         # Add detector position
@@ -228,22 +228,22 @@ def create_scalar(handle, path, dtype, value):
             )
         )
         group = handle.create_group("/entry/instrument/detector/transformations")
-        group.attrs["NX_class"] = np.string_("NXtransformations")
+        group.attrs["NX_class"] = np.bytes_("NXtransformations")
         create_scalar(group, "translation", "float32", detector_offset_vector.length())
-        group["translation"].attrs["transformation_type"] = np.string_("translation")
+        group["translation"].attrs["transformation_type"] = np.bytes_("translation")
         if detector_offset_vector.length() > 0:
             group["translation"].attrs["vector"] = detector_offset_vector.normalize()
         else:
             group["translation"].attrs["vector"] = detector_offset_vector
         group["translation"].attrs["offset"] = 0
-        group["translation"].attrs["units"] = np.string_("m")
-        group["translation"].attrs["depends_on"] = np.string_(".")
+        group["translation"].attrs["units"] = np.bytes_("m")
+        group["translation"].attrs["depends_on"] = np.bytes_(".")
 
         # Create goniometer transformations if not found
         if "/entry/sample/transformations" not in handle:
             # print "Creating group /entry/sample/transformation"
             group = handle.create_group("/entry/sample/transformations")
-            group.attrs["NX_class"] = np.string_("NXtransformations")
+            group.attrs["NX_class"] = np.bytes_("NXtransformations")
         else:
             group = handle["/entry/sample/transformations"]
 
@@ -274,11 +274,11 @@ def create_scalar(handle, path, dtype, value):
             for name in sorted(handle["/entry/data"]):
                 num_images += handle_orig_entry_properties[name]["length"]
             dataset = group.create_dataset("omega", (num_images,), dtype="float32")
-            dataset.attrs["units"] = np.string_("degree")
-            dataset.attrs["transformation_type"] = np.string_("rotation")
+            dataset.attrs["units"] = np.bytes_("degree")
+            dataset.attrs["transformation_type"] = np.bytes_("rotation")
             dataset.attrs["vector"] = default_axis
             dataset.attrs["offset"] = 0
-            dataset.attrs["depends_on"] = np.string_(".")
+            dataset.attrs["depends_on"] = np.bytes_(".")
             omega_range_average = handle[
                 "/entry/sample/goniometer/omega_range_average"
             ][()]
@@ -295,7 +295,7 @@ def create_scalar(handle, path, dtype, value):
                 handle["/entry/sample"],
                 "depends_on",
                 "S%d" % len(dataset.name),
-                np.string_(dataset.name),
+                np.bytes_(dataset.name),
             )
 
         # Change relative paths to absolute paths

src/dxtbx/format/FormatNexusJungfrauHack.py

@@ -106,11 +106,11 @@ def _setup_detector(self, detector, beam):
         detector_material = clean_string(str(material))
         material = {
             "Si": "Si",
-            np.string_("Si"): "Si",
-            np.string_("Silicon"): "Si",
-            np.string_("Sillicon"): "Si",
-            np.string_("CdTe"): "CdTe",
-            np.string_("GaAs"): "GaAs",
+            np.bytes_("Si"): "Si",
+            np.bytes_("Silicon"): "Si",
+            np.bytes_("Sillicon"): "Si",
+            np.bytes_("CdTe"): "CdTe",
+            np.bytes_("GaAs"): "GaAs",
         }.get(detector_material)
         if not material:
             raise RuntimeError("Unknown material: %s" % detector_material)

src/dxtbx/format/FormatSER.py

@@ -142,7 +142,7 @@ def read_emi(filename):
 
 
 def _parseEntry_emi(value):
-    """Auxiliary function to parse string entry to int, float or np.string_().
+    """Auxiliary function to parse string entry to int, float or np.bytes_().
     Parameters
     ----------
         value : str
@@ -162,7 +162,7 @@ def _parseEntry_emi(value):
             p = float(value)
         except ValueError:
             # if neither int nor float, stay with string
-            p = np.string_(str(value))
+            p = np.bytes_(str(value))
 
     return p
 

src/dxtbx/format/nexus.py

@@ -48,15 +48,15 @@
 NXNode = Union[h5py.File, h5py.Group]
 
 
-def h5str(h5_value: str | numpy.string_ | bytes) -> str:
+def h5str(h5_value: str | numpy.bytes_ | bytes) -> str:
     """
     Convert a value returned an h5py attribute to str.
 
-    h5py can return either a bytes-like (numpy.string_) or str object
+    h5py can return either a bytes-like (numpy.bytes_) or str object
     for attribute values depending on whether the value was written as
     fixed or variable length. This function collapses the two to str.
     """
-    if isinstance(h5_value, (numpy.string_, bytes)):
+    if isinstance(h5_value, (numpy.bytes_, bytes)):
         return h5_value.decode("utf-8")
     return h5_value
 
@@ -78,7 +78,7 @@ def dataset_as_flex(dataset, selection):
         if dataset.dtype in [
             numpy.half,
             numpy.single,
-            numpy.float_,
+            numpy.float64,
             numpy.float16,
             numpy.float32,
         ]:

src/dxtbx/format/nxmx_writer.py

@@ -171,7 +171,7 @@ def construct_entry(self):
             entry["end_time_estimated"] = self.params.nexus_details.end_time_estimated
 
         # --> definition
-        self._create_scalar(entry, "definition", "S4", np.string_("NXmx"))
+        self._create_scalar(entry, "definition", "S4", np.bytes_("NXmx"))
 
         # --> sample
         sample = self.handle["entry"].create_group("sample")
@@ -354,7 +354,7 @@ def recursive_setup_basis_dict(key, parent_name="", panel_id=0):
         det_group.attrs["NX_class"] = "NXdetector_group"
 
         det_group.create_dataset("group_index", data=list(range(1, 3)), dtype="i")
-        data = [np.string_("detector"), np.string_("detector")]
+        data = [np.bytes_("detector"), np.bytes_("detector")]
         det_group.create_dataset("group_names", (2,), data=data, dtype="S12")
         det_group.create_dataset("group_parent", (2,), data=[-1, 1], dtype="i")
         det_group.create_dataset("group_type", (2,), data=[1, 2], dtype="i")

tests/conftest.py

@@ -27,9 +27,7 @@ def nxmx_example():
 
         instrument = entry.create_group("instrument")
         instrument.attrs["NX_class"] = "NXinstrument"
-        name = instrument.create_dataset(
-            "name", data=np.string_("DIAMOND BEAMLINE I03")
-        )
+        name = instrument.create_dataset("name", data=np.bytes_("DIAMOND BEAMLINE I03"))
         name.attrs["short_name"] = "I03"
 
         beam = instrument.create_group("beam")

tests/format/test_FormatNXmxDLS16M.py

@@ -200,11 +200,11 @@ def test_understand(beamline, tmp_path):
     with h5py.File(nxs, mode="w") as fh:
         entry = fh.create_group("entry")
         instrument = entry.create_group("instrument")
-        instrument.attrs["short_name"] = np.string_(f"DLS {beamline}")
+        instrument.attrs["short_name"] = np.bytes_(f"DLS {beamline}")
         name = instrument.create_dataset(
-            "name", data=np.string_(f"DIAMOND BEAMLINE {beamline}")
+            "name", data=np.bytes_(f"DIAMOND BEAMLINE {beamline}")
         )
-        name.attrs["short_name"] = np.string_(f"DLS {beamline}")
+        name.attrs["short_name"] = np.bytes_(f"DLS {beamline}")
     assert FormatNXmxDLS16M.understand(nxs)
     assert FormatNXmxDLS.understand(nxs)
 
@@ -216,9 +216,9 @@ def test_understand_legacy(beamline, tmp_path):
     with h5py.File(nxs, mode="w") as fh:
         entry = fh.create_group("entry")
         instrument = entry.create_group("instrument")
-        instrument.attrs["short_name"] = np.string_(f"{beamline}")
-        name = instrument.create_dataset("name", data=np.string_(f"{beamline}"))
-        name.attrs["short_name"] = np.string_(f"{beamline}")
+        instrument.attrs["short_name"] = np.bytes_(f"{beamline}")
+        name = instrument.create_dataset("name", data=np.bytes_(f"{beamline}"))
+        name.attrs["short_name"] = np.bytes_(f"{beamline}")
     assert FormatNXmxDLS16M.understand(nxs)
     assert FormatNXmxDLS.understand(nxs)
 
@@ -236,9 +236,7 @@ def test_do_not_understand_i24(tmp_path):
     with h5py.File(nxs, mode="w") as fh:
         entry = fh.create_group("entry")
         instrument = entry.create_group("instrument")
-        instrument.attrs["short_name"] = np.string_("DLS I24")
-        name = instrument.create_dataset(
-            "name", data=np.string_("DIAMOND BEAMLINE I24")
-        )
-        name.attrs["short_name"] = np.string_("DLS I24")
+        instrument.attrs["short_name"] = np.bytes_("DLS I24")
+        name = instrument.create_dataset("name", data=np.bytes_("DIAMOND BEAMLINE I24"))
+        name.attrs["short_name"] = np.bytes_("DLS I24")
     assert not FormatNXmxDLS16M.understand(nxs)

tests/nexus/test_build_dxtbx_models.py

@@ -805,7 +805,7 @@ def test_dataset_as_flex_float():
 def test_dataset_as_flex_double():
     slices = ()
     np_double_types = (
-        np.float_,
+        np.float64,
         np.double,
         np.float64,
     )