"""Implements loading exported HDF files from Igor."""
import copy
import os
from pathlib import Path
import h5py
import numpy as np
import xarray as xr
from arpes.endstations import SESEndstation
from arpes.load_pxt import read_single_pxt
from arpes.provenance import provenance_from_file
from arpes.repair import negate_energy
__all__ = ("IgorExportEndstation",)
[docs]class IgorExportEndstation(SESEndstation):
"""Implements loading exported HDF files for ARPES data from Igor."""
PRINCIPAL_NAME = "Igor"
ALIASES = [
"igor",
"igor-export",
]
RENAME_KEYS = {}
def load_single_frame(self, frame_path: str = None, scan_desc: dict = None, **kwargs):
"""HDF files are all inclusive, so we just need to load one file per scan."""
_, ext = os.path.splitext(frame_path)
if "nc" in ext or "h5" in ext:
# was converted to hdf5/NetCDF format with Conrad's Igor scripts
scan_desc = copy.deepcopy(scan_desc)
scan_desc["path"] = frame_path
return self.load_SES_h5(scan_desc=scan_desc, robust_dimension_labels=True, **kwargs)
# it's given by SES PXT files
pxt_data = negate_energy(read_single_pxt(frame_path))
return xr.Dataset({"spectrum": pxt_data}, attrs=pxt_data.attrs)
def load_SES_h5(
self, scan_desc: dict = None, robust_dimension_labels=False, **kwargs
) -> xr.Dataset:
"""Imports an hdf5 dataset exported from Igor.
In particular, this handles data that was originally generated by a Scienta spectrometer
in the SESb format. In order to understand the structure of these files have a look at Conrad's
saveSESDataset in Igor Pro.
Args:
scan_desc: Dictionary with extra information to attach to the xr.Dataset, must contain the location
of the file
Returns:
The loaded data.
"""
scan_desc = copy.deepcopy(scan_desc)
data_loc = scan_desc.get("path", scan_desc.get("file"))
p = Path(data_loc)
if not p.exists():
import arpes.config
data_loc = os.path.join(arpes.config.DATA_PATH, data_loc)
# wave_note = shim_wave_note(data_loc)
wave_note = ""
f = h5py.File(data_loc, "r")
primary_dataset_name = list(f)[0]
# This is bugged for the moment in h5py due to an inability to read fixed length unicode strings
# wave_note = f['/' + primary_dataset_name].attrs['IGORWaveNote']
dimension_labels = list(f["/" + primary_dataset_name].attrs["IGORWaveDimensionLabels"][0])
# print(list(f['/' + primary_dataset_name].attrs.keys()))
if any(x == "" for x in dimension_labels):
# print(dimension_labels)
if not robust_dimension_labels:
raise ValueError(
"Missing dimension labels. Use robust_dimension_labels=True to override"
)
else:
used_blanks = 0
for i in range(len(dimension_labels)):
if dimension_labels[i] == "":
dimension_labels[i] = "missing{}".format(used_blanks)
used_blanks += 1
# print(dimension_labels)
scaling = f["/" + primary_dataset_name].attrs["IGORWaveScaling"][-len(dimension_labels) :]
raw_data = f["/" + primary_dataset_name][:]
scaling = [
np.linspace(scale[1], scale[1] + scale[0] * raw_data.shape[i], raw_data.shape[i])
for i, scale in enumerate(scaling)
]
dataset_contents = {}
attrs = scan_desc.pop("note", {})
attrs.update(wave_note)
built_coords = dict(zip(dimension_labels, scaling))
deg_to_rad_coords = {"theta", "beta", "phi"}
# the hemisphere axis is handled below
built_coords = {
k: c * (np.pi / 180) if k in deg_to_rad_coords else c for k, c in built_coords.items()
}
deg_to_rad_attrs = {"theta", "beta", "alpha", "chi"}
for angle_attr in deg_to_rad_attrs:
if angle_attr in attrs:
attrs[angle_attr] = float(attrs[angle_attr]) * np.pi / 180
dataset_contents["spectrum"] = xr.DataArray(
raw_data,
coords=built_coords,
dims=dimension_labels,
attrs=attrs,
)
provenance_from_file(
dataset_contents["spectrum"],
data_loc,
{"what": "Loaded SES dataset from HDF5.", "by": "load_SES"},
)
return xr.Dataset(
dataset_contents,
attrs={**scan_desc, "name": primary_dataset_name},
)