1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
|
# Authors: Alexandre Gramfort <alexandre.gramfort@telecom-paristech.fr>
# Matti Hamalainen <msh@nmr.mgh.harvard.edu>
#
# License: BSD (3-clause)
import struct
import os
import gzip
import numpy as np
from scipy import linalg
from .constants import FIFF
from ..externals.six import text_type
from ..externals.jdcal import jd2jcal
class Tag(object):
"""Tag in FIF tree structure
Parameters
----------
kind : int
Kind of Tag.
type_ : int
Type of Tag.
size : int
Size in bytes.
int : next
Position of next Tag.
pos : int
Position of Tag is the original file.
"""
def __init__(self, kind, type_, size, next, pos=None):
self.kind = int(kind)
self.type = int(type_)
self.size = int(size)
self.next = int(next)
self.pos = pos if pos is not None else next
self.pos = int(self.pos)
self.data = None
def __repr__(self):
out = ("kind: %s - type: %s - size: %s - next: %s - pos: %s"
% (self.kind, self.type, self.size, self.next, self.pos))
if hasattr(self, 'data'):
out += " - data: %s" % self.data
out += "\n"
return out
def __cmp__(self, tag):
is_equal = (self.kind == tag.kind and
self.type == tag.type and
self.size == tag.size and
self.next == tag.next and
self.pos == tag.pos and
self.data == tag.data)
if is_equal:
return 0
else:
return 1
def read_big(fid, size=None):
"""Function to read large chunks of data (>16MB) Windows-friendly
Parameters
----------
fid : file
Open file to read from.
size : int or None
Number of bytes to read. If None, the whole file is read.
Returns
-------
buf : bytes
The data.
Notes
-----
Windows (argh) can't handle reading large chunks of data, so we
have to do it piece-wise, possibly related to:
http://stackoverflow.com/questions/4226941
Examples
--------
This code should work for normal files and .gz files:
>>> import numpy as np
>>> import gzip, os, tempfile, shutil
>>> fname = tempfile.mkdtemp()
>>> fname_gz = os.path.join(fname, 'temp.gz')
>>> fname = os.path.join(fname, 'temp.bin')
>>> randgen = np.random.RandomState(9)
>>> x = randgen.randn(3000000) # > 16MB data
>>> with open(fname, 'wb') as fid: x.tofile(fid)
>>> with open(fname, 'rb') as fid: y = np.fromstring(read_big(fid))
>>> assert np.all(x == y)
>>> fid_gz = gzip.open(fname_gz, 'wb')
>>> _ = fid_gz.write(x.tostring())
>>> fid_gz.close()
>>> fid_gz = gzip.open(fname_gz, 'rb')
>>> y = np.fromstring(read_big(fid_gz))
>>> assert np.all(x == y)
>>> shutil.rmtree(os.path.dirname(fname))
>>> fid_gz.close()
"""
# buf_size is chosen as a largest working power of 2 (16 MB):
buf_size = 16777216
if size is None:
# it's not possible to get .gz uncompressed file size
if not isinstance(fid, gzip.GzipFile):
size = os.fstat(fid.fileno()).st_size - fid.tell()
if size is not None:
# Use pre-buffering method
segments = np.r_[np.arange(0, size, buf_size), size]
buf = bytearray(b' ' * size)
for start, end in zip(segments[:-1], segments[1:]):
data = fid.read(int(end - start))
if len(data) != end - start:
raise ValueError('Read error')
buf[start:end] = data
buf = bytes(buf)
else:
# Use presumably less efficient concatenating method
buf = [b'']
new = fid.read(buf_size)
while len(new) > 0:
buf.append(new)
new = fid.read(buf_size)
buf = b''.join(buf)
return buf
def read_tag_info(fid):
"""Read Tag info (or header)
"""
s = fid.read(4 * 4)
if len(s) == 0:
return None
tag = Tag(*struct.unpack(">iiii", s))
if tag.next == 0:
fid.seek(tag.size, 1)
elif tag.next > 0:
fid.seek(tag.next, 0)
return tag
def _fromstring_rows(fid, tag_size, dtype=None, shape=None, rlims=None):
"""Helper for getting a range of rows from a large tag"""
if shape is not None:
item_size = np.dtype(dtype).itemsize
if not len(shape) == 2:
raise ValueError('Only implemented for 2D matrices')
if not np.prod(shape) == tag_size / item_size:
raise ValueError('Wrong shape specified')
if not len(rlims) == 2:
raise ValueError('rlims must have two elements')
n_row_out = rlims[1] - rlims[0]
if n_row_out <= 0:
raise ValueError('rlims must yield at least one output')
row_size = item_size * shape[1]
# # of bytes to skip at the beginning, # to read, where to end
start_skip = int(rlims[0] * row_size)
read_size = int(n_row_out * row_size)
end_pos = int(fid.tell() + tag_size)
# Move the pointer ahead to the read point
fid.seek(start_skip, 1)
# Do the reading
out = np.fromstring(fid.read(read_size), dtype=dtype)
# Move the pointer ahead to the end of the tag
fid.seek(end_pos)
else:
out = np.fromstring(fid.read(tag_size), dtype=dtype)
return out
def _loc_to_trans(loc):
"""Helper to convert loc vector to coil_trans"""
# deal with nasty OSX Anaconda bug by casting to float64
loc = loc.astype(np.float64)
coil_trans = np.concatenate([loc.reshape(4, 3).T[:, [1, 2, 3, 0]],
np.array([0, 0, 0, 1]).reshape(1, 4)])
return coil_trans
def read_tag(fid, pos=None, shape=None, rlims=None):
"""Read a Tag from a file at a given position
Parameters
----------
fid : file
The open FIF file descriptor.
pos : int
The position of the Tag in the file.
shape : tuple | None
If tuple, the shape of the stored matrix. Only to be used with
data stored as a vector (not implemented for matrices yet).
rlims : tuple | None
If tuple, the first and last rows to retrieve. Note that data are
assumed to be stored row-major in the file. Only to be used with
data stored as a vector (not implemented for matrices yet).
Returns
-------
tag : Tag
The Tag read.
"""
if pos is not None:
fid.seek(pos, 0)
s = fid.read(4 * 4)
tag = Tag(*struct.unpack(">iIii", s))
#
# The magic hexadecimal values
#
is_matrix = 4294901760 # ffff0000
matrix_coding_dense = 16384 # 4000
matrix_coding_CCS = 16400 # 4010
matrix_coding_RCS = 16416 # 4020
data_type = 65535 # ffff
#
if tag.size > 0:
matrix_coding = is_matrix & tag.type
if matrix_coding != 0:
matrix_coding = matrix_coding >> 16
# This should be easy to implement (see _fromstring_rows)
# if we need it, but for now, it's not...
if shape is not None:
raise ValueError('Row reading not implemented for matrices '
'yet')
# Matrices
if matrix_coding == matrix_coding_dense:
# Find dimensions and return to the beginning of tag data
pos = fid.tell()
fid.seek(tag.size - 4, 1)
ndim = int(np.fromstring(fid.read(4), dtype='>i4'))
fid.seek(-(ndim + 1) * 4, 1)
dims = np.fromstring(fid.read(4 * ndim), dtype='>i4')[::-1]
#
# Back to where the data start
#
fid.seek(pos, 0)
if ndim > 3:
raise Exception('Only 2 or 3-dimensional matrices are '
'supported at this time')
matrix_type = data_type & tag.type
if matrix_type == FIFF.FIFFT_INT:
tag.data = np.fromstring(read_big(fid, 4 * dims.prod()),
dtype='>i4').reshape(dims)
elif matrix_type == FIFF.FIFFT_JULIAN:
tag.data = np.fromstring(read_big(fid, 4 * dims.prod()),
dtype='>i4').reshape(dims)
elif matrix_type == FIFF.FIFFT_FLOAT:
tag.data = np.fromstring(read_big(fid, 4 * dims.prod()),
dtype='>f4').reshape(dims)
elif matrix_type == FIFF.FIFFT_DOUBLE:
tag.data = np.fromstring(read_big(fid, 8 * dims.prod()),
dtype='>f8').reshape(dims)
elif matrix_type == FIFF.FIFFT_COMPLEX_FLOAT:
data = np.fromstring(read_big(fid, 4 * 2 * dims.prod()),
dtype='>f4')
# Note: we need the non-conjugate transpose here
tag.data = (data[::2] + 1j * data[1::2]).reshape(dims)
elif matrix_type == FIFF.FIFFT_COMPLEX_DOUBLE:
data = np.fromstring(read_big(fid, 8 * 2 * dims.prod()),
dtype='>f8')
# Note: we need the non-conjugate transpose here
tag.data = (data[::2] + 1j * data[1::2]).reshape(dims)
else:
raise Exception('Cannot handle matrix of type %d yet'
% matrix_type)
elif matrix_coding in (matrix_coding_CCS, matrix_coding_RCS):
from scipy import sparse
# Find dimensions and return to the beginning of tag data
pos = fid.tell()
fid.seek(tag.size - 4, 1)
ndim = int(np.fromstring(fid.read(4), dtype='>i4'))
fid.seek(-(ndim + 2) * 4, 1)
dims = np.fromstring(fid.read(4 * (ndim + 1)), dtype='>i4')
if ndim != 2:
raise Exception('Only two-dimensional matrices are '
'supported at this time')
# Back to where the data start
fid.seek(pos, 0)
nnz = int(dims[0])
nrow = int(dims[1])
ncol = int(dims[2])
sparse_data = np.fromstring(fid.read(4 * nnz), dtype='>f4')
shape = (dims[1], dims[2])
if matrix_coding == matrix_coding_CCS:
# CCS
sparse.csc_matrix()
sparse_indices = np.fromstring(fid.read(4 * nnz),
dtype='>i4')
sparse_ptrs = np.fromstring(fid.read(4 * (ncol + 1)),
dtype='>i4')
tag.data = sparse.csc_matrix((sparse_data, sparse_indices,
sparse_ptrs), shape=shape)
else:
# RCS
sparse_indices = np.fromstring(fid.read(4 * nnz),
dtype='>i4')
sparse_ptrs = np.fromstring(fid.read(4 * (nrow + 1)),
dtype='>i4')
tag.data = sparse.csr_matrix((sparse_data, sparse_indices,
sparse_ptrs), shape=shape)
else:
raise Exception('Cannot handle other than dense or sparse '
'matrices yet')
else:
# All other data types
# Simple types
if tag.type == FIFF.FIFFT_BYTE:
tag.data = _fromstring_rows(fid, tag.size, dtype=">B1",
shape=shape, rlims=rlims)
elif tag.type == FIFF.FIFFT_SHORT:
tag.data = _fromstring_rows(fid, tag.size, dtype=">i2",
shape=shape, rlims=rlims)
elif tag.type == FIFF.FIFFT_INT:
tag.data = _fromstring_rows(fid, tag.size, dtype=">i4",
shape=shape, rlims=rlims)
elif tag.type == FIFF.FIFFT_USHORT:
tag.data = _fromstring_rows(fid, tag.size, dtype=">u2",
shape=shape, rlims=rlims)
elif tag.type == FIFF.FIFFT_UINT:
tag.data = _fromstring_rows(fid, tag.size, dtype=">u4",
shape=shape, rlims=rlims)
elif tag.type == FIFF.FIFFT_FLOAT:
tag.data = _fromstring_rows(fid, tag.size, dtype=">f4",
shape=shape, rlims=rlims)
elif tag.type == FIFF.FIFFT_DOUBLE:
tag.data = _fromstring_rows(fid, tag.size, dtype=">f8",
shape=shape, rlims=rlims)
elif tag.type == FIFF.FIFFT_STRING:
tag.data = _fromstring_rows(fid, tag.size, dtype=">c",
shape=shape, rlims=rlims)
# Always decode to unicode.
td = tag.data.tostring().decode('utf-8', 'ignore')
tag.data = text_type(td)
elif tag.type == FIFF.FIFFT_DAU_PACK16:
tag.data = _fromstring_rows(fid, tag.size, dtype=">i2",
shape=shape, rlims=rlims)
elif tag.type == FIFF.FIFFT_COMPLEX_FLOAT:
# data gets stored twice as large
if shape is not None:
shape = (shape[0], shape[1] * 2)
tag.data = _fromstring_rows(fid, tag.size, dtype=">f4",
shape=shape, rlims=rlims)
tag.data = tag.data[::2] + 1j * tag.data[1::2]
elif tag.type == FIFF.FIFFT_COMPLEX_DOUBLE:
# data gets stored twice as large
if shape is not None:
shape = (shape[0], shape[1] * 2)
tag.data = _fromstring_rows(fid, tag.size, dtype=">f8",
shape=shape, rlims=rlims)
tag.data = tag.data[::2] + 1j * tag.data[1::2]
#
# Structures
#
elif tag.type == FIFF.FIFFT_ID_STRUCT:
tag.data = dict()
tag.data['version'] = int(np.fromstring(fid.read(4),
dtype=">i4"))
tag.data['version'] = int(np.fromstring(fid.read(4),
dtype=">i4"))
tag.data['machid'] = np.fromstring(fid.read(8), dtype=">i4")
tag.data['secs'] = int(np.fromstring(fid.read(4), dtype=">i4"))
tag.data['usecs'] = int(np.fromstring(fid.read(4),
dtype=">i4"))
elif tag.type == FIFF.FIFFT_DIG_POINT_STRUCT:
tag.data = dict()
tag.data['kind'] = int(np.fromstring(fid.read(4), dtype=">i4"))
tag.data['ident'] = int(np.fromstring(fid.read(4),
dtype=">i4"))
tag.data['r'] = np.fromstring(fid.read(12), dtype=">f4")
tag.data['coord_frame'] = FIFF.FIFFV_COORD_UNKNOWN
elif tag.type == FIFF.FIFFT_COORD_TRANS_STRUCT:
tag.data = dict()
tag.data['from'] = int(np.fromstring(fid.read(4), dtype=">i4"))
tag.data['to'] = int(np.fromstring(fid.read(4), dtype=">i4"))
rot = np.fromstring(fid.read(36), dtype=">f4").reshape(3, 3)
move = np.fromstring(fid.read(12), dtype=">f4")
tag.data['trans'] = np.r_[np.c_[rot, move],
np.array([[0], [0], [0], [1]]).T]
#
# Skip over the inverse transformation
# It is easier to just use inverse of trans in Matlab
#
fid.seek(12 * 4, 1)
elif tag.type == FIFF.FIFFT_CH_INFO_STRUCT:
d = dict()
d['scanno'] = int(np.fromstring(fid.read(4), dtype=">i4"))
d['logno'] = int(np.fromstring(fid.read(4), dtype=">i4"))
d['kind'] = int(np.fromstring(fid.read(4), dtype=">i4"))
d['range'] = float(np.fromstring(fid.read(4), dtype=">f4"))
d['cal'] = float(np.fromstring(fid.read(4), dtype=">f4"))
d['coil_type'] = int(np.fromstring(fid.read(4), dtype=">i4"))
#
# Read the coil coordinate system definition
#
d['loc'] = np.fromstring(fid.read(48), dtype=">f4")
d['coil_trans'] = None
d['eeg_loc'] = None
d['coord_frame'] = FIFF.FIFFV_COORD_UNKNOWN
tag.data = d
#
# Convert loc into a more useful format
#
loc = tag.data['loc']
kind = tag.data['kind']
if kind in [FIFF.FIFFV_MEG_CH, FIFF.FIFFV_REF_MEG_CH]:
tag.data['coil_trans'] = _loc_to_trans(loc)
tag.data['coord_frame'] = FIFF.FIFFV_COORD_DEVICE
elif tag.data['kind'] == FIFF.FIFFV_EEG_CH:
# deal with nasty OSX Anaconda bug by casting to float64
loc = loc.astype(np.float64)
if linalg.norm(loc[3:6]) > 0.:
tag.data['eeg_loc'] = np.c_[loc[0:3], loc[3:6]]
else:
tag.data['eeg_loc'] = loc[0:3][:, np.newaxis].copy()
tag.data['coord_frame'] = FIFF.FIFFV_COORD_HEAD
#
# Unit and exponent
#
tag.data['unit'] = int(np.fromstring(fid.read(4), dtype=">i4"))
tag.data['unit_mul'] = int(np.fromstring(fid.read(4),
dtype=">i4"))
#
# Handle the channel name
#
ch_name = np.fromstring(fid.read(16), dtype=">c")
ch_name = ch_name[:np.argmax(ch_name == b'')].tostring()
# Use unicode or bytes depending on Py2/3
tag.data['ch_name'] = str(ch_name.decode())
elif tag.type == FIFF.FIFFT_OLD_PACK:
offset = float(np.fromstring(fid.read(4), dtype=">f4"))
scale = float(np.fromstring(fid.read(4), dtype=">f4"))
tag.data = np.fromstring(fid.read(tag.size - 8), dtype=">h2")
tag.data = scale * tag.data + offset
elif tag.type == FIFF.FIFFT_DIR_ENTRY_STRUCT:
tag.data = list()
for _ in range(tag.size // 16 - 1):
s = fid.read(4 * 4)
tag.data.append(Tag(*struct.unpack(">iIii", s)))
elif tag.type == FIFF.FIFFT_JULIAN:
tag.data = int(np.fromstring(fid.read(4), dtype=">i4"))
tag.data = jd2jcal(tag.data)
else:
raise Exception('Unimplemented tag data type %s' % tag.type)
if tag.next != FIFF.FIFFV_NEXT_SEQ:
# f.seek(tag.next,0)
fid.seek(tag.next, 1) # XXX : fix? pb when tag.next < 0
return tag
def find_tag(fid, node, findkind):
"""Find Tag in an open FIF file descriptor
"""
for p in range(node['nent']):
if node['directory'][p].kind == findkind:
return read_tag(fid, node['directory'][p].pos)
tag = None
return tag
def has_tag(node, kind):
"""Does the node contains a Tag of a given kind?
"""
for d in node['directory']:
if d.kind == kind:
return True
return False
|
