cansas_reader_HDF5.py @ 92f586e9

Last change on this file since 92f586e9 was 61f329f0, checked in by krzywon, 7 years ago
Reset file reader class data state each time a new data file is loaded.
Property mode set to `100644`
File size: 26.9 KB

Line
1	"""
2	CanSAS 2D data reader for reading HDF5 formatted CanSAS files.
3	"""
4
5	import h5py
6	import numpy as np
7	import re
8	import os
9	import sys
10
11	from ..data_info import plottable_1D, plottable_2D,\
12	Data1D, Data2D, DataInfo, Process, Aperture, Collimation, \
13	TransmissionSpectrum, Detector
14	from ..data_info import combine_data_info_with_plottable
15	from ..loader_exceptions import FileContentsException, DefaultReaderException
16	from ..file_reader_base_class import FileReader, decode
17
18	def h5attr(node, key, default=None):
19	return decode(node.attrs.get(key, default))
20
21	class Reader(FileReader):
22	"""
23	A class for reading in CanSAS v2.0 data files. The existing iteration opens
24	Mantid generated HDF5 formatted files with file extension .h5/.H5. Any
25	number of data sets may be present within the file and any dimensionality
26	of data may be used. Currently 1D and 2D SAS data sets are supported, but
27	future implementations will include 1D and 2D SESANS data.
28
29	Any number of SASdata sets may be present in a SASentry and the data within
30	can be either 1D I(Q) or 2D I(Qx, Qy).
31
32	Also supports reading NXcanSAS formatted HDF5 files
33
34	:Dependencies:
35	The CanSAS HDF5 reader requires h5py => v2.5.0 or later.
36	"""
37
38	# CanSAS version
39	cansas_version = 2.0
40	# Logged warnings or messages
41	logging = None
42	# List of errors for the current data set
43	errors = None
44	# Raw file contents to be processed
45	raw_data = None
46	# List of plottable1D objects that should be linked to the current_datainfo
47	data1d = None
48	# List of plottable2D objects that should be linked to the current_datainfo
49	data2d = None
50	# Data type name
51	type_name = "CanSAS 2.0"
52	# Wildcards
53	type = ["CanSAS 2.0 HDF5 Files (.h5)\|.h5"]
54	# List of allowed extensions
55	ext = ['.h5', '.H5']
56	# Flag to bypass extension check
57	allow_all = True
58
59	def get_file_contents(self):
60	"""
61	This is the general read method that all SasView data_loaders must have.
62
63	:param filename: A path for an HDF5 formatted CanSAS 2D data file.
64	:return: List of Data1D/2D objects and/or a list of errors.
65	"""
66	# Reinitialize when loading a new data file to reset all class variables
67	self.reset_state()
68
69	filename = self.f_open.name
70	self.f_open.close() # IO handled by h5py
71
72	# Check that the file exists
73	if os.path.isfile(filename):
74	basename = os.path.basename(filename)
75	_, extension = os.path.splitext(basename)
76	# If the file type is not allowed, return empty list
77	if extension in self.ext or self.allow_all:
78	# Load the data file
79	try:
80	self.raw_data = h5py.File(filename, 'r')
81	except Exception as e:
82	if extension not in self.ext:
83	msg = "CanSAS2.0 HDF5 Reader could not load file {}".format(basename + extension)
84	raise DefaultReaderException(msg)
85	raise FileContentsException(e.message)
86	try:
87	# Read in all child elements of top level SASroot
88	self.read_children(self.raw_data, [])
89	# Add the last data set to the list of outputs
90	self.add_data_set()
91	except Exception as exc:
92	raise FileContentsException(exc.message)
93	finally:
94	# Close the data file
95	self.raw_data.close()
96
97	for dataset in self.output:
98	if isinstance(dataset, Data1D):
99	if dataset.x.size < 5:
100	self.output = []
101	raise FileContentsException("Fewer than 5 data points found.")
102
103	def reset_state(self):
104	"""
105	Create the reader object and define initial states for class variables
106	"""
107	super(Reader, self).reset_state()
108	self.data1d = []
109	self.data2d = []
110	self.raw_data = None
111	self.errors = set()
112	self.logging = []
113	self.parent_class = u''
114	self.detector = Detector()
115	self.collimation = Collimation()
116	self.aperture = Aperture()
117	self.process = Process()
118	self.trans_spectrum = TransmissionSpectrum()
119
120	def read_children(self, data, parent_list):
121	"""
122	A recursive method for stepping through the hierarchical data file.
123
124	:param data: h5py Group object of any kind
125	:param parent: h5py Group parent name
126	"""
127
128	# Loop through each element of the parent and process accordingly
129	for key in data.keys():
130	# Get all information for the current key
131	value = data.get(key)
132	class_name = h5attr(value, u'canSAS_class')
133	if class_name is None:
134	class_name = h5attr(value, u'NX_class')
135	if class_name is not None:
136	class_prog = re.compile(class_name)
137	else:
138	class_prog = re.compile(value.name)
139
140	if isinstance(value, h5py.Group):
141	# Set parent class before recursion
142	self.parent_class = class_name
143	parent_list.append(key)
144	# If a new sasentry, store the current data sets and create
145	# a fresh Data1D/2D object
146	if class_prog.match(u'SASentry'):
147	self.add_data_set(key)
148	elif class_prog.match(u'SASdata'):
149	self._initialize_new_data_set(parent_list)
150	# Recursion step to access data within the group
151	self.read_children(value, parent_list)
152	# Reset parent class when returning from recursive method
153	self.parent_class = class_name
154	self.add_intermediate()
155	parent_list.remove(key)
156
157	elif isinstance(value, h5py.Dataset):
158	# If this is a dataset, store the data appropriately
159	data_set = data[key][:]
160	unit = self._get_unit(value)
161
162	# I and Q Data
163	if key == u'I':
164	if isinstance(self.current_dataset, plottable_2D):
165	self.current_dataset.data = data_set
166	self.current_dataset.zaxis("Intensity", unit)
167	else:
168	self.current_dataset.y = data_set.flatten()
169	self.current_dataset.yaxis("Intensity", unit)
170	continue
171	elif key == u'Idev':
172	if isinstance(self.current_dataset, plottable_2D):
173	self.current_dataset.err_data = data_set.flatten()
174	else:
175	self.current_dataset.dy = data_set.flatten()
176	continue
177	elif key == u'Q':
178	self.current_dataset.xaxis("Q", unit)
179	if isinstance(self.current_dataset, plottable_2D):
180	self.current_dataset.q = data_set.flatten()
181	else:
182	self.current_dataset.x = data_set.flatten()
183	continue
184	elif key == u'Qdev':
185	self.current_dataset.dx = data_set.flatten()
186	continue
187	elif key == u'dQw':
188	self.current_dataset.dxw = data_set.flatten()
189	continue
190	elif key == u'dQl':
191	self.current_dataset.dxl = data_set.flatten()
192	continue
193	elif key == u'Qy':
194	self.current_dataset.yaxis("Q_y", unit)
195	self.current_dataset.qy_data = data_set.flatten()
196	continue
197	elif key == u'Qydev':
198	self.current_dataset.dqy_data = data_set.flatten()
199	continue
200	elif key == u'Qx':
201	self.current_dataset.xaxis("Q_x", unit)
202	self.current_dataset.qx_data = data_set.flatten()
203	continue
204	elif key == u'Qxdev':
205	self.current_dataset.dqx_data = data_set.flatten()
206	continue
207	elif key == u'Mask':
208	self.current_dataset.mask = data_set.flatten()
209	continue
210	# Transmission Spectrum
211	elif (key == u'T'
212	and self.parent_class == u'SAStransmission_spectrum'):
213	self.trans_spectrum.transmission = data_set.flatten()
214	continue
215	elif (key == u'Tdev'
216	and self.parent_class == u'SAStransmission_spectrum'):
217	self.trans_spectrum.transmission_deviation = \
218	data_set.flatten()
219	continue
220	elif (key == u'lambda'
221	and self.parent_class == u'SAStransmission_spectrum'):
222	self.trans_spectrum.wavelength = data_set.flatten()
223	continue
224
225	for data_point in data_set:
226	if isinstance(data_point, np.ndarray):
227	if data_point.dtype.char == 'S':
228	data_point = decode(bytes(data_point))
229	else:
230	data_point = decode(data_point)
231	# Top Level Meta Data
232	if key == u'definition':
233	self.current_datainfo.meta_data['reader'] = data_point
234	elif key == u'run':
235	self.current_datainfo.run.append(data_point)
236	try:
237	run_name = h5attr(value, 'name')
238	run_dict = {data_point: run_name}
239	self.current_datainfo.run_name = run_dict
240	except Exception:
241	pass
242	elif key == u'title':
243	self.current_datainfo.title = data_point
244	elif key == u'SASnote':
245	self.current_datainfo.notes.append(data_point)
246
247	# Sample Information
248	# CanSAS 2.0 format
249	elif key == u'Title' and self.parent_class == u'SASsample':
250	self.current_datainfo.sample.name = data_point
251	# NXcanSAS format
252	elif key == u'name' and self.parent_class == u'SASsample':
253	self.current_datainfo.sample.name = data_point
254	# NXcanSAS format
255	elif key == u'ID' and self.parent_class == u'SASsample':
256	self.current_datainfo.sample.name = data_point
257	elif (key == u'thickness'
258	and self.parent_class == u'SASsample'):
259	self.current_datainfo.sample.thickness = data_point
260	elif (key == u'temperature'
261	and self.parent_class == u'SASsample'):
262	self.current_datainfo.sample.temperature = data_point
263	elif (key == u'transmission'
264	and self.parent_class == u'SASsample'):
265	self.current_datainfo.sample.transmission = data_point
266	elif (key == u'x_position'
267	and self.parent_class == u'SASsample'):
268	self.current_datainfo.sample.position.x = data_point
269	elif (key == u'y_position'
270	and self.parent_class == u'SASsample'):
271	self.current_datainfo.sample.position.y = data_point
272	elif key == u'pitch' and self.parent_class == u'SASsample':
273	self.current_datainfo.sample.orientation.x = data_point
274	elif key == u'yaw' and self.parent_class == u'SASsample':
275	self.current_datainfo.sample.orientation.y = data_point
276	elif key == u'roll' and self.parent_class == u'SASsample':
277	self.current_datainfo.sample.orientation.z = data_point
278	elif (key == u'details'
279	and self.parent_class == u'SASsample'):
280	self.current_datainfo.sample.details.append(data_point)
281
282	# Instrumental Information
283	elif (key == u'name'
284	and self.parent_class == u'SASinstrument'):
285	self.current_datainfo.instrument = data_point
286	elif key == u'name' and self.parent_class == u'SASdetector':
287	self.detector.name = data_point
288	elif key == u'SDD' and self.parent_class == u'SASdetector':
289	self.detector.distance = float(data_point)
290	self.detector.distance_unit = unit
291	elif (key == u'slit_length'
292	and self.parent_class == u'SASdetector'):
293	self.detector.slit_length = float(data_point)
294	self.detector.slit_length_unit = unit
295	elif (key == u'x_position'
296	and self.parent_class == u'SASdetector'):
297	self.detector.offset.x = float(data_point)
298	self.detector.offset_unit = unit
299	elif (key == u'y_position'
300	and self.parent_class == u'SASdetector'):
301	self.detector.offset.y = float(data_point)
302	self.detector.offset_unit = unit
303	elif (key == u'pitch'
304	and self.parent_class == u'SASdetector'):
305	self.detector.orientation.x = float(data_point)
306	self.detector.orientation_unit = unit
307	elif key == u'roll' and self.parent_class == u'SASdetector':
308	self.detector.orientation.z = float(data_point)
309	self.detector.orientation_unit = unit
310	elif key == u'yaw' and self.parent_class == u'SASdetector':
311	self.detector.orientation.y = float(data_point)
312	self.detector.orientation_unit = unit
313	elif (key == u'beam_center_x'
314	and self.parent_class == u'SASdetector'):
315	self.detector.beam_center.x = float(data_point)
316	self.detector.beam_center_unit = unit
317	elif (key == u'beam_center_y'
318	and self.parent_class == u'SASdetector'):
319	self.detector.beam_center.y = float(data_point)
320	self.detector.beam_center_unit = unit
321	elif (key == u'x_pixel_size'
322	and self.parent_class == u'SASdetector'):
323	self.detector.pixel_size.x = float(data_point)
324	self.detector.pixel_size_unit = unit
325	elif (key == u'y_pixel_size'
326	and self.parent_class == u'SASdetector'):
327	self.detector.pixel_size.y = float(data_point)
328	self.detector.pixel_size_unit = unit
329	elif (key == u'distance'
330	and self.parent_class == u'SAScollimation'):
331	self.collimation.length = data_point
332	self.collimation.length_unit = unit
333	elif (key == u'name'
334	and self.parent_class == u'SAScollimation'):
335	self.collimation.name = data_point
336	elif (key == u'shape'
337	and self.parent_class == u'SASaperture'):
338	self.aperture.shape = data_point
339	elif (key == u'x_gap'
340	and self.parent_class == u'SASaperture'):
341	self.aperture.size.x = data_point
342	elif (key == u'y_gap'
343	and self.parent_class == u'SASaperture'):
344	self.aperture.size.y = data_point
345
346	# Process Information
347	elif (key == u'Title'
348	and self.parent_class == u'SASprocess'): # CanSAS 2.0
349	self.process.name = data_point
350	elif (key == u'name'
351	and self.parent_class == u'SASprocess'): # NXcanSAS
352	self.process.name = data_point
353	elif (key == u'description'
354	and self.parent_class == u'SASprocess'):
355	self.process.description = data_point
356	elif key == u'date' and self.parent_class == u'SASprocess':
357	self.process.date = data_point
358	elif key == u'term' and self.parent_class == u'SASprocess':
359	self.process.term = data_point
360	elif self.parent_class == u'SASprocess':
361	self.process.notes.append(data_point)
362
363	# Source
364	elif (key == u'wavelength'
365	and self.parent_class == u'SASdata'):
366	self.current_datainfo.source.wavelength = data_point
367	self.current_datainfo.source.wavelength_unit = unit
368	elif (key == u'incident_wavelength'
369	and self.parent_class == 'SASsource'):
370	self.current_datainfo.source.wavelength = data_point
371	self.current_datainfo.source.wavelength_unit = unit
372	elif (key == u'wavelength_max'
373	and self.parent_class == u'SASsource'):
374	self.current_datainfo.source.wavelength_max = data_point
375	self.current_datainfo.source.wavelength_max_unit = unit
376	elif (key == u'wavelength_min'
377	and self.parent_class == u'SASsource'):
378	self.current_datainfo.source.wavelength_min = data_point
379	self.current_datainfo.source.wavelength_min_unit = unit
380	elif (key == u'incident_wavelength_spread'
381	and self.parent_class == u'SASsource'):
382	self.current_datainfo.source.wavelength_spread = \
383	data_point
384	self.current_datainfo.source.wavelength_spread_unit = \
385	unit
386	elif (key == u'beam_size_x'
387	and self.parent_class == u'SASsource'):
388	self.current_datainfo.source.beam_size.x = data_point
389	self.current_datainfo.source.beam_size_unit = unit
390	elif (key == u'beam_size_y'
391	and self.parent_class == u'SASsource'):
392	self.current_datainfo.source.beam_size.y = data_point
393	self.current_datainfo.source.beam_size_unit = unit
394	elif (key == u'beam_shape'
395	and self.parent_class == u'SASsource'):
396	self.current_datainfo.source.beam_shape = data_point
397	elif (key == u'radiation'
398	and self.parent_class == u'SASsource'):
399	self.current_datainfo.source.radiation = data_point
400	elif (key == u'transmission'
401	and self.parent_class == u'SASdata'):
402	self.current_datainfo.sample.transmission = data_point
403
404	# Everything else goes in meta_data
405	else:
406	new_key = self._create_unique_key(
407	self.current_datainfo.meta_data, key)
408	self.current_datainfo.meta_data[new_key] = data_point
409
410	else:
411	# I don't know if this reachable code
412	self.errors.add("ShouldNeverHappenException")
413
414	def add_intermediate(self):
415	"""
416	This method stores any intermediate objects within the final data set
417	after fully reading the set.
418
419	:param parent: The NXclass name for the h5py Group object that just
420	finished being processed
421	"""
422
423	if self.parent_class == u'SASprocess':
424	self.current_datainfo.process.append(self.process)
425	self.process = Process()
426	elif self.parent_class == u'SASdetector':
427	self.current_datainfo.detector.append(self.detector)
428	self.detector = Detector()
429	elif self.parent_class == u'SAStransmission_spectrum':
430	self.current_datainfo.trans_spectrum.append(self.trans_spectrum)
431	self.trans_spectrum = TransmissionSpectrum()
432	elif self.parent_class == u'SAScollimation':
433	self.current_datainfo.collimation.append(self.collimation)
434	self.collimation = Collimation()
435	elif self.parent_class == u'SASaperture':
436	self.collimation.aperture.append(self.aperture)
437	self.aperture = Aperture()
438	elif self.parent_class == u'SASdata':
439	if isinstance(self.current_dataset, plottable_2D):
440	self.data2d.append(self.current_dataset)
441	elif isinstance(self.current_dataset, plottable_1D):
442	self.data1d.append(self.current_dataset)
443
444	def final_data_cleanup(self):
445	"""
446	Does some final cleanup and formatting on self.current_datainfo and
447	all data1D and data2D objects and then combines the data and info into
448	Data1D and Data2D objects
449	"""
450	# Type cast data arrays to float64
451	if len(self.current_datainfo.trans_spectrum) > 0:
452	spectrum_list = []
453	for spectrum in self.current_datainfo.trans_spectrum:
454	spectrum.transmission = np.delete(spectrum.transmission, [0])
455	spectrum.transmission = spectrum.transmission.astype(np.float64)
456	spectrum.transmission_deviation = np.delete(
457	spectrum.transmission_deviation, [0])
458	spectrum.transmission_deviation = \
459	spectrum.transmission_deviation.astype(np.float64)
460	spectrum.wavelength = np.delete(spectrum.wavelength, [0])
461	spectrum.wavelength = spectrum.wavelength.astype(np.float64)
462	if len(spectrum.transmission) > 0:
463	spectrum_list.append(spectrum)
464	self.current_datainfo.trans_spectrum = spectrum_list
465
466	# Append errors to dataset and reset class errors
467	self.current_datainfo.errors = self.errors
468	self.errors.clear()
469
470	# Combine all plottables with datainfo and append each to output
471	# Type cast data arrays to float64 and find min/max as appropriate
472	for dataset in self.data2d:
473	zeros = np.ones(dataset.data.size, dtype=bool)
474	try:
475	for i in range(0, dataset.mask.size - 1):
476	zeros[i] = dataset.mask[i]
477	except:
478	self.errors.add(sys.exc_value)
479	dataset.mask = zeros
480	# Calculate the actual Q matrix
481	try:
482	if dataset.q_data.size <= 1:
483	dataset.q_data = np.sqrt(dataset.qx_data
484	* dataset.qx_data
485	+ dataset.qy_data
486	* dataset.qy_data)
487	except:
488	dataset.q_data = None
489
490	if dataset.data.ndim == 2:
491	(n_rows, n_cols) = dataset.data.shape
492	dataset.y_bins = dataset.qy_data[0::n_cols]
493	dataset.x_bins = dataset.qx_data[:n_cols]
494	dataset.data = dataset.data.flatten()
495	self.current_dataset = dataset
496	self.send_to_output()
497
498	for dataset in self.data1d:
499	self.current_dataset = dataset
500	self.send_to_output()
501
502	def add_data_set(self, key=""):
503	"""
504	Adds the current_dataset to the list of outputs after preforming final
505	processing on the data and then calls a private method to generate a
506	new data set.
507
508	:param key: NeXus group name for current tree level
509	"""
510
511	if self.current_datainfo and self.current_dataset:
512	self.final_data_cleanup()
513	self.data1d = []
514	self.data2d = []
515	self.current_datainfo = DataInfo()
516
517
518	def _initialize_new_data_set(self, parent_list=None):
519	"""
520	A private class method to generate a new 1D or 2D data object based on
521	the type of data within the set. Outside methods should call
522	add_data_set() to be sure any existing data is stored properly.
523
524	:param parent_list: List of names of parent elements
525	"""
526
527	if parent_list is None:
528	parent_list = []
529	if self._find_intermediate(parent_list, "Qx"):
530	self.current_dataset = plottable_2D()
531	else:
532	x = np.array(0)
533	y = np.array(0)
534	self.current_dataset = plottable_1D(x, y)
535	self.current_datainfo.filename = self.raw_data.filename
536
537	def _find_intermediate(self, parent_list, basename=""):
538	"""
539	A private class used to find an entry by either using a direct key or
540	knowing the approximate basename.
541
542	:param parent_list: List of parents nodes in the HDF5 file
543	:param basename: Approximate name of an entry to search for
544	:return:
545	"""
546
547	entry = False
548	key_prog = re.compile(basename)
549	top = self.raw_data
550	for parent in parent_list:
551	top = top.get(parent)
552	for key in top.keys():
553	if key_prog.match(key):
554	entry = True
555	break
556	return entry
557
558	def _create_unique_key(self, dictionary, name, numb=0):
559	"""
560	Create a unique key value for any dictionary to prevent overwriting
561	Recurses until a unique key value is found.
562
563	:param dictionary: A dictionary with any number of entries
564	:param name: The index of the item to be added to dictionary
565	:param numb: The number to be appended to the name, starts at 0
566	:return: The new name for the dictionary entry
567	"""
568	if dictionary.get(name) is not None:
569	numb += 1
570	name = name.split("_")[0]
571	name += "_{0}".format(numb)
572	name = self._create_unique_key(dictionary, name, numb)
573	return name
574
575	def _get_unit(self, value):
576	"""
577	Find the unit for a particular value within the h5py dictionary
578
579	:param value: attribute dictionary for a particular value set
580	:return: unit for the value passed to the method
581	"""
582	unit = h5attr(value, u'units')
583	if unit is None:
584	unit = h5attr(value, u'unit')
585	# Convert the unit formats
586	if unit == "1/A":
587	unit = "A^{-1}"
588	elif unit == "1/cm":
589	unit = "cm^{-1}"
590	return unit

Note: See TracBrowser for help on using the repository browser.

SasView

source: sasview/src/sas/sascalc/dataloader/readers/cansas_reader_HDF5.py @ 92f586e9

Download in other formats: