ENH: add _load_requirements class attribute to Dataset classes

yt/data_objects/static_output.py

@@ -142,7 +142,7 @@ def ireq(self, value):
 
 
 class Dataset(abc.ABC):
-
+    _load_requirements: List[str] = []
     default_fluid_type = "gas"
     default_field = ("gas", "density")
     fluid_types: Tuple[FieldType, ...] = ("gas", "deposit", "index")
@@ -346,6 +346,13 @@ def force_periodicity(self, val=True):
             raise TypeError("force_periodicity expected a boolean.")
         self._force_periodicity = val
 
+    @classmethod
+    def _missing_extra_requirements(cls) -> List[str]:
+        # return a list of optional dependencies that are
+        # needed for the present class to function, but currently missing.
+        # returning an empty list means that all requirements are met
+        return [name for name in cls._load_requirements if not find_spec(name)]
+
     # abstract methods require implementation in subclasses
     @classmethod
     @abc.abstractmethod

yt/frontends/_skeleton/data_structures.py

@@ -1,5 +1,6 @@
 import os
 import weakref
+from typing import List
 
 import numpy as np
 
@@ -76,6 +77,9 @@ class SkeletonDataset(Dataset):
     _index_class = SkeletonHierarchy
     _field_info_class = SkeletonFieldInfo
 
+    # names of additional modules that may be required to load data from disk
+    _load_requirements: List[str] = []
+
     def __init__(
         self,
         filename,

yt/frontends/arepo/data_structures.py

@@ -8,6 +8,7 @@
 
 
 class ArepoHDF5Dataset(GadgetHDF5Dataset):
+    _load_requirements = ["h5py"]
     _field_info_class = ArepoFieldInfo
 
     def __init__(

yt/frontends/cf_radial/data_structures.py

@@ -80,6 +80,7 @@ def _populate_grid_objects(self):
 
 
 class CFRadialDataset(Dataset):
+    _load_requirements = ["xarray", "pyart"]
     _index_class = CFRadialHierarchy
     _field_info_class = CFRadialFieldInfo
 

yt/frontends/chimera/data_structures.py

@@ -195,6 +195,7 @@ def _setup_data_io(self):
 
 
 class ChimeraDataset(Dataset):
+    _load_requirements = ["h5py"]
     _index_class = ChimeraUNSIndex  # ChimeraHierarchy
     _field_info_class = ChimeraFieldInfo
 

yt/frontends/cholla/data_structures.py

@@ -61,6 +61,7 @@ def _populate_grid_objects(self):
 
 
 class ChollaDataset(Dataset):
+    _load_requirements = ["h5py"]
     _index_class = ChollaHierarchy
     _field_info_class = ChollaFieldInfo
 

yt/frontends/chombo/data_structures.py

@@ -237,6 +237,7 @@ def _reconstruct_parent_child(self):
 
 
 class ChomboDataset(Dataset):
+    _load_requirements = ["h5py"]
     _index_class = ChomboHierarchy
     _field_info_class: Type[FieldInfoContainer] = ChomboFieldInfo
 
@@ -641,7 +642,7 @@ def _read_particles(self):
 
 
 class Orion2Dataset(ChomboDataset):
-
+    _load_requirements = ["h5py"]
     _index_class = Orion2Hierarchy
     _field_info_class = Orion2FieldInfo
 
@@ -759,7 +760,7 @@ def __init__(self, ds, dataset_type="chombo_hdf5"):
 
 
 class ChomboPICDataset(ChomboDataset):
-
+    _load_requirements = ["h5py"]
     _index_class = ChomboPICHierarchy
     _field_info_class = ChomboPICFieldInfo3D
 

yt/frontends/eagle/data_structures.py

@@ -12,6 +12,7 @@
 
 
 class EagleDataset(GadgetHDF5Dataset):
+    _load_requirements = ["h5py"]
     _particle_mass_name = "Mass"
     _field_info_class: Type[FieldInfoContainer] = OWLSFieldInfo
     _time_readin_ = "Time"
@@ -71,6 +72,7 @@ def _is_valid(cls, filename, *args, **kwargs):
 
 
 class EagleNetworkDataset(EagleDataset):
+    _load_requirements = ["h5py"]
     _particle_mass_name = "Mass"
     _field_info_class = EagleNetworkFieldInfo
     _time_readin = "Time"

yt/frontends/enzo/data_structures.py

@@ -668,6 +668,7 @@ class EnzoDataset(Dataset):
     Enzo-specific output, set at a fixed time.
     """
 
+    _load_requirements = ["h5py", "libconf"]
     _index_class = EnzoHierarchy
     _field_info_class = EnzoFieldInfo
 

yt/frontends/enzo_e/data_structures.py

@@ -286,6 +286,7 @@ class EnzoEDataset(Dataset):
     Enzo-E-specific output, set at a fixed time.
     """
 
+    _load_requirements = ["h5py", "libconf"]
     refine_by = 2
     _index_class = EnzoEHierarchy
     _field_info_class = EnzoEFieldInfo

yt/frontends/exodus_ii/data_structures.py

@@ -46,6 +46,7 @@ def _detect_output_fields(self):
 
 
 class ExodusIIDataset(Dataset):
+    _load_requirements = ["netCDF4"]
     _index_class = ExodusIIUnstructuredIndex
     _field_info_class = ExodusIIFieldInfo
 

yt/frontends/fits/data_structures.py

@@ -315,6 +315,7 @@ def check_sky_coords(filename, ndim):
 
 
 class FITSDataset(Dataset):
+    _load_requirements = ["astropy"]
     _index_class = FITSHierarchy
     _field_info_class: Type[FieldInfoContainer] = FITSFieldInfo
     _dataset_type = "fits"
@@ -573,6 +574,7 @@ def find_axes(axis_names, prefixes):
 
 
 class YTFITSDataset(FITSDataset):
+    _load_requirements = ["astropy"]
     _field_info_class = YTFITSFieldInfo
 
     def _parse_parameter_file(self):
@@ -655,6 +657,7 @@ def _is_valid(cls, filename, *args, **kwargs):
 
 
 class SkyDataFITSDataset(FITSDataset):
+    _load_requirements = ["astropy"]
     _field_info_class = WCSFITSFieldInfo
 
     def _determine_wcs(self):
@@ -741,6 +744,7 @@ def _domain_decomp(self):
 
 
 class SpectralCubeFITSDataset(SkyDataFITSDataset):
+    _load_requirements = ["astropy"]
     _index_class = SpectralCubeFITSHierarchy
 
     def __init__(
@@ -860,6 +864,7 @@ def _parse_index(self):
 
 
 class EventsFITSDataset(SkyDataFITSDataset):
+    _load_requirements = ["astropy"]
     _index_class = EventsFITSHierarchy
 
     def __init__(

yt/frontends/flash/data_structures.py

@@ -164,6 +164,7 @@ def _populate_grid_objects(self):
 
 
 class FLASHDataset(Dataset):
+    _load_requirements = ["h5py"]
     _index_class: Type[Index] = FLASHHierarchy
     _field_info_class = FLASHFieldInfo
     _handle = None
@@ -484,6 +485,7 @@ class FLASHParticleFile(ParticleFile):
 
 
 class FLASHParticleDataset(FLASHDataset):
+    _load_requirements = ["h5py"]
     _index_class = ParticleIndex
     filter_bbox = False
     _file_class = FLASHParticleFile

yt/frontends/gadget/data_structures.py

@@ -574,6 +574,7 @@ class GadgetHDF5File(ParticleFile):
 
 
 class GadgetHDF5Dataset(GadgetDataset):
+    _load_requirements = ["h5py"]
     _file_class = GadgetHDF5File
     _index_class = SPHParticleIndex
     _field_info_class: Type[FieldInfoContainer] = GadgetFieldInfo

yt/frontends/gadget_fof/data_structures.py

@@ -137,6 +137,7 @@ def _read_particle_positions(self, ptype, f=None):
 
 
 class GadgetFOFDataset(ParticleDataset):
+    _load_requirements = ["h5py"]
     _index_class = GadgetFOFParticleIndex
     _file_class = GadgetFOFHDF5File
     _field_info_class = GadgetFOFFieldInfo

yt/frontends/gamer/data_structures.py

@@ -201,6 +201,7 @@ def _validate_parent_children_relationship(self):
 
 
 class GAMERDataset(Dataset):
+    _load_requirements = ["h5py"]
     _index_class = GAMERHierarchy
     _field_info_class = GAMERFieldInfo
     _handle = None

yt/frontends/gdf/data_structures.py

@@ -148,6 +148,7 @@ def _get_grid_children(self, grid):
 
 
 class GDFDataset(Dataset):
+    _load_requirements = ["h5py"]
     _index_class = GDFHierarchy
     _field_info_class = GDFFieldInfo
 

yt/frontends/gizmo/data_structures.py

@@ -7,6 +7,7 @@
 
 
 class GizmoDataset(GadgetHDF5Dataset):
+    _load_requirements = ["h5py"]
     _field_info_class = GizmoFieldInfo
 
     @classmethod

yt/frontends/halo_catalog/data_structures.py

@@ -98,6 +98,7 @@ class YTHaloCatalogDataset(SavedDataset):
     in yt_astro_analysis.
     """
 
+    _load_requirements = ["h5py"]
     _index_class = ParticleIndex
     _file_class = YTHaloCatalogFile
     _field_info_class = YTHaloCatalogFieldInfo

yt/frontends/http_stream/data_structures.py

@@ -15,6 +15,7 @@ class HTTPParticleFile(ParticleFile):
 
 
 class HTTPStreamDataset(ParticleDataset):
+    _load_requirements = ["requests"]
     _index_class = ParticleIndex
     _file_class = HTTPParticleFile
     _field_info_class = SPHFieldInfo

yt/frontends/moab/data_structures.py

@@ -47,6 +47,7 @@ def _count_grids(self):
 
 
 class MoabHex8Dataset(Dataset):
+    _load_requirements = ["h5py"]
     _index_class = MoabHex8Hierarchy
     _field_info_class = MoabFieldInfo
     periodicity = (False, False, False)

yt/frontends/nc4_cm1/data_structures.py

@@ -64,6 +64,7 @@ def _populate_grid_objects(self):
 
 
 class CM1Dataset(Dataset):
+    _load_requirements = ["netCDF4"]
     _index_class = CM1Hierarchy
     _field_info_class = CM1FieldInfo
 

yt/frontends/open_pmd/data_structures.py

@@ -431,6 +431,7 @@ class OpenPMDDataset(Dataset):
     - particle and mesh positions are *absolute* with respect to the simulation origin.
     """
 
+    _load_requirements = ["h5py"]
     _index_class = OpenPMDHierarchy
     _field_info_class = OpenPMDFieldInfo
 
@@ -656,6 +657,7 @@ def _load(self, it, **kwargs):
 
 
 class OpenPMDGroupBasedDataset(Dataset):
+    _load_requirements = ["h5py"]
     _index_class = OpenPMDHierarchy
     _field_info_class = OpenPMDFieldInfo
 

yt/frontends/owls/data_structures.py

@@ -9,6 +9,7 @@
 
 
 class OWLSDataset(GadgetHDF5Dataset):
+    _load_requirements = ["h5py"]
     _particle_mass_name = "Mass"
     _field_info_class = OWLSFieldInfo
     _time_readin = "Time_GYR"

yt/frontends/owls_subfind/data_structures.py

@@ -79,6 +79,7 @@ def __init__(self, ds, io, filename, file_id, bounds):
 
 
 class OWLSSubfindDataset(ParticleDataset):
+    _load_requirements = ["h5py"]
     _index_class = OWLSSubfindParticleIndex
     _file_class = OWLSSubfindHDF5File
     _field_info_class = OWLSSubfindFieldInfo

yt/frontends/ramses/data_structures.py

@@ -697,6 +697,7 @@ def print_stats(self):
 
 
 class RAMSESDataset(Dataset):
+    _load_requirements = ["f90nml"]
     _index_class = RAMSESIndex
     _field_info_class = RAMSESFieldInfo
     gamma = 1.4  # This will get replaced on hydro_fn open

yt/frontends/sdf/data_structures.py

@@ -24,6 +24,7 @@ class SDFFile(ParticleFile):
 
 
 class SDFDataset(ParticleDataset):
+    _load_requirements = ["requests"]
     _index_class = ParticleIndex
     _file_class = SDFFile
     _field_info_class = SDFFieldInfo

yt/frontends/swift/data_structures.py

@@ -13,6 +13,7 @@ class SwiftParticleFile(ParticleFile):
 
 
 class SwiftDataset(SPHDataset):
+    _load_requirements = ["h5py"]
     _index_class = SPHParticleIndex
     _field_info_class = SPHFieldInfo
     _file_class = SwiftParticleFile

yt/frontends/ytdata/data_structures.py

@@ -234,6 +234,7 @@ def __init__(self, ds, io, filename, file_id, range):
 class YTDataContainerDataset(YTDataset):
     """Dataset for saved geometric data containers."""
 
+    _load_requirements = ["h5py"]
     _index_class = ParticleIndex
     _file_class = YTDataHDF5File
     _field_info_class: Type[FieldInfoContainer] = YTDataContainerFieldInfo
@@ -316,6 +317,8 @@ def _is_valid(cls, filename, *args, **kwargs):
 class YTDataLightRayDataset(YTDataContainerDataset):
     """Dataset for saved LightRay objects."""
 
+    _load_requirements = ["h5py"]
+
     def _parse_parameter_file(self):
         super()._parse_parameter_file()
         self._restore_light_ray_solution()
@@ -357,6 +360,7 @@ def _is_valid(cls, filename, *args, **kwargs):
 class YTSpatialPlotDataset(YTDataContainerDataset):
     """Dataset for saved slices and projections."""
 
+    _load_requirements = ["h5py"]
     _field_info_class = YTGridFieldInfo
 
     def __init__(self, *args, **kwargs):
@@ -477,6 +481,7 @@ def _populate_grid_objects(self):
 class YTGridDataset(YTDataset):
     """Dataset for saved covering grids, arbitrary grids, and FRBs."""
 
+    _load_requirements = ["h5py"]
     _index_class: Type[Index] = YTGridHierarchy
     _field_info_class = YTGridFieldInfo
     _dataset_type = "ytgridhdf5"
@@ -709,6 +714,7 @@ def _read_fluid_fields(self, fields, dobj, chunk=None):
 class YTNonspatialDataset(YTGridDataset):
     """Dataset for general array data."""
 
+    _load_requirements = ["h5py"]
     _index_class = YTNonspatialHierarchy
     _field_info_class = YTGridFieldInfo
     _dataset_type = "ytnonspatialhdf5"
@@ -777,6 +783,7 @@ def _is_valid(cls, filename, *args, **kwargs):
 class YTProfileDataset(YTNonspatialDataset):
     """Dataset for saved profile objects."""
 
+    _load_requirements = ["h5py"]
     fluid_types = ("data", "gas", "standard_deviation")
 
     def __init__(self, filename, unit_system="cgs"):
@@ -926,6 +933,8 @@ def __getitem__(self, field):
 class YTClumpTreeDataset(YTNonspatialDataset):
     """Dataset for saved clump-finder data."""
 
+    _load_requirements = ["h5py"]
+
     def __init__(self, filename, unit_system="cgs"):
         super().__init__(filename, unit_system=unit_system)
         self._load_tree()