def test_check_input_ext():
test_filename = 'help'
ext = 'py'
out_filename = utils.check_input_ext(test_filename, ext)
assert str(out_filename) == f'{test_filename}.{ext}'
test_filename = 'help.txt.gz'
assert str(utils.check_input_ext(test_filename, ext)) == 'help.py'
def phys2bids(filename,
info=False,
indir='.',
outdir='.',
heur_file=None,
sub=None,
ses=None,
chtrig=0,
chsel=None,
num_timepoints_expected=None,
tr=None,
thr=None,
pad=9,
ch_name=[],
yml='',
debug=False,
quiet=False):
"""
Run main workflow of phys2bids.
Runs the parser, does some checks on input, then imports
the right interface file to read the input. If only info is required,
it returns a summary onscreen.
Otherwise, it operates on the input to return a .tsv.gz file, possibly
in BIDS format.
Raises
------
NotImplementedError
If the file extension is not supported yet.
"""
# Check options to make them internally coherent pt. I
# #!# This can probably be done while parsing?
outdir = os.path.abspath(outdir)
os.makedirs(outdir, exist_ok=True)
os.makedirs(os.path.join(outdir, 'code'), exist_ok=True)
conversion_path = os.path.join(outdir, 'code', 'conversion')
os.makedirs(conversion_path, exist_ok=True)
# Create logfile name
basename = 'phys2bids_'
extension = 'tsv'
isotime = datetime.datetime.now().strftime('%Y-%m-%dT%H%M%S')
logname = os.path.join(conversion_path,
(basename + isotime + '.' + extension))
# Set logging format
log_formatter = logging.Formatter(
'%(asctime)s\t%(name)-12s\t%(levelname)-8s\t%(message)s',
datefmt='%Y-%m-%dT%H:%M:%S')
# Set up logging file and open it for writing
log_handler = logging.FileHandler(logname)
log_handler.setFormatter(log_formatter)
sh = logging.StreamHandler()
if quiet:
logging.basicConfig(level=logging.WARNING,
handlers=[log_handler, sh],
format='%(levelname)-10s %(message)s')
elif debug:
logging.basicConfig(level=logging.DEBUG,
handlers=[log_handler, sh],
format='%(levelname)-10s %(message)s')
else:
logging.basicConfig(level=logging.INFO,
handlers=[log_handler, sh],
format='%(levelname)-10s %(message)s')
version_number = _version.get_versions()['version']
LGR.info(f'Currently running phys2bids version {version_number}')
LGR.info(f'Input file is {filename}')
# Save call.sh
arg_str = ' '.join(sys.argv[1:])
call_str = f'phys2bids {arg_str}'
f = open(os.path.join(conversion_path, 'call.sh'), "a")
f.write(f'#!bin/bash \n{call_str}')
f.close()
# Check options to make them internally coherent pt. II
# #!# This can probably be done while parsing?
indir = os.path.abspath(indir)
if chtrig < 0:
raise Exception(
'Wrong trigger channel. Channel indexing starts with 0!')
filename, ftype = utils.check_input_type(filename, indir)
if heur_file:
heur_file = utils.check_input_ext(heur_file, '.py')
utils.check_file_exists(heur_file)
infile = os.path.join(indir, filename)
utils.check_file_exists(infile)
if isinstance(num_timepoints_expected, int):
num_timepoints_expected = [num_timepoints_expected]
if isinstance(tr, (int, float)):
tr = [tr]
if tr is not None and num_timepoints_expected is not None:
# If tr and ntp were specified, check that tr is either length one or ntp.
if len(num_timepoints_expected) != len(tr) and len(tr) != 1:
raise Exception('Number of sequence types listed with TR '
'doesn\'t match expected number of runs in '
'the session')
# Read file!
LGR.info(f'Reading the file {infile}')
if ftype == 'acq':
from phys2bids.io import load_acq
phys_in = load_acq(infile, chtrig)
elif ftype == 'txt':
from phys2bids.io import load_txt
phys_in = load_txt(infile, chtrig)
elif ftype == 'mat':
from phys2bids.io import load_mat
phys_in = load_mat(infile, chtrig)
LGR.info('Checking that units of measure are BIDS compatible')
for index, unit in enumerate(phys_in.units):
phys_in.units[index] = bids.bidsify_units(unit)
LGR.info('Reading infos')
phys_in.print_info(filename)
# #!# Here the function viz.plot_channel should be called
viz.plot_all(phys_in.ch_name, phys_in.timeseries, phys_in.units,
phys_in.freq, infile, conversion_path)
# If only info were asked, end here.
if info:
return
# The next few lines remove the undesired channels from phys_in.
if chsel:
LGR.info('Dropping unselected channels')
for i in reversed(range(0, phys_in.ch_amount)):
if i not in chsel:
phys_in.delete_at_index(i)
# If requested, change channel names.
if ch_name:
LGR.info('Renaming channels with given names')
phys_in.rename_channels(ch_name)
# Checking acquisition type via user's input
if tr is not None and num_timepoints_expected is not None:
# Multi-run acquisition type section
# Check list length, more than 1 means multi-run
if len(num_timepoints_expected) > 1:
# if multi-run of same sequence type, pad list with ones
# and multiply array with user's input
if len(tr) == 1:
tr = np.ones(len(num_timepoints_expected)) * tr[0]
# Sum of values in ntp_list should be equivalent to num_timepoints_found
phys_in.check_trigger_amount(
thr=thr,
num_timepoints_expected=sum(num_timepoints_expected),
tr=1)
# Check that sum of tp expected is equivalent to num_timepoints_found,
# if it passes call slice4phys
if phys_in.num_timepoints_found != sum(num_timepoints_expected):
raise Exception('The number of triggers found is different '
'than expected. Better stop now than break '
'something.')
# slice the recording based on user's entries
# !!! ATTENTION: PHYS_IN GETS OVERWRITTEN AS DICTIONARY
phys_in = slice4phys(phys_in, num_timepoints_expected, tr,
phys_in.thr, pad)
# returns a dictionary in the form {run_idx: phys_in[startpoint, endpoint]}
# save a figure for each run | give the right acquisition parameters for runs
fileprefix = os.path.join(
conversion_path,
os.path.splitext(os.path.basename(filename))[0])
for i, run in enumerate(phys_in.keys()):
plot_fileprefix = f'{fileprefix}_0{run}'
viz.export_trigger_plot(phys_in[run], chtrig, plot_fileprefix,
tr[i], num_timepoints_expected[i],
filename, sub, ses)
# Single run acquisition type, or : nothing to split workflow
else:
# Run analysis on trigger channel to get first timepoint
# and the time offset.
phys_in.check_trigger_amount(thr, num_timepoints_expected[0],
tr[0])
# save a figure of the trigger
fileprefix = os.path.join(
conversion_path,
os.path.splitext(os.path.basename(filename))[0])
viz.export_trigger_plot(phys_in, chtrig, fileprefix, tr[0],
num_timepoints_expected[0], filename, sub,
ses)
# Reassign phys_in as dictionary
# !!! ATTENTION: PHYS_IN GETS OVERWRITTEN AS DICTIONARY
phys_in = {1: phys_in}
else:
LGR.warning('Skipping trigger pulse count. If you want to run it, '
'call phys2bids using both "-ntp" and "-tr" arguments')
# !!! ATTENTION: PHYS_IN GETS OVERWRITTEN AS DICTIONARY
phys_in = {1: phys_in}
# The next few lines create a dictionary of different BlueprintInput
# objects, one for each unique frequency for each run in phys_in
# they also save the amount of runs and unique frequencies
run_amount = len(phys_in)
uniq_freq_list = set(phys_in[1].freq)
freq_amount = len(uniq_freq_list)
if freq_amount > 1:
LGR.info(f'Found {freq_amount} different frequencies in input!')
if run_amount > 1:
LGR.info(f'Found {run_amount} different scans in input!')
LGR.info(f'Preparing {freq_amount*run_amount} output files.')
# Create phys_out dict that will have a blueprint object for each different frequency
phys_out = {}
if heur_file is not None and sub is not None:
LGR.info(f'Preparing BIDS output using {heur_file}')
# If heuristics are used, init a dict of arguments to pass to use_heuristic
heur_args = {
'heur_file': heur_file,
'sub': sub,
'ses': ses,
'filename': filename,
'outdir': outdir,
'run': '',
'record_label': ''
}
# Generate participants.tsv file if it doesn't exist already.
# Update the file if the subject is not in the file.
# Do not update if the subject is already in the file.
bids.participants_file(outdir, yml, sub)
# Generate dataset_description.json file if it doesn't exist already.
bids.dataset_description_file(outdir)
# Generate README file if it doesn't exist already.
bids.readme_file(outdir)
cp(
heur_file,
os.path.join(
conversion_path,
os.path.splitext(os.path.basename(heur_file))[0] + '.py'))
elif heur_file is not None and sub is None:
LGR.warning('While "-heur" was specified, option "-sub" was not.\n'
'Skipping BIDS formatting.')
# Export a (set of) phys_out for each element in phys_in
# run keys start from 1 (human friendly)
for run in phys_in.keys():
for uniq_freq in uniq_freq_list:
# Initialise the key for the (possibly huge amount of) dictionary entries
key = f'{run}_{uniq_freq}'
# copy the phys_in object to the new dict entry
phys_out[key] = deepcopy(phys_in[run])
# this counter will take into account how many channels are eliminated
count = 0
# for each channel in the original phys_in object
# take the frequency
for idx, i in enumerate(phys_in[run].freq):
# if that frequency is different than the frequency of the phys_obj entry
if i != uniq_freq:
# eliminate that channel from the dict since we only want channels
# with the same frequency
phys_out[key].delete_at_index(idx - count)
# take into acount the elimination so in the next eliminated channel we
# eliminate correctly
count += 1
# Also create a BlueprintOutput object for each unique frequency found.
# Populate it with the corresponding blueprint input and replace it
# in the dictionary.
# Add time channel in the proper frequency.
if uniq_freq != phys_in[run].freq[0]:
phys_out[key].ch_name.insert(0, phys_in[run].ch_name[0])
phys_out[key].units.insert(0, phys_in[run].units[0])
phys_out[key].timeseries.insert(
0,
np.linspace(phys_in[run].timeseries[0][0],
phys_in[run].timeseries[0][-1],
num=phys_out[key].timeseries[0].shape[0]))
# Add trigger channel in the proper frequency.
if uniq_freq != phys_in[run].freq[chtrig]:
phys_out[key].ch_name.insert(1, phys_in[run].ch_name[chtrig])
phys_out[key].units.insert(1, phys_in[run].units[chtrig])
phys_out[key].timeseries.insert(
1,
np.interp(phys_out[key].timeseries[0],
phys_in[run].timeseries[0],
phys_in[run].timeseries[chtrig]))
phys_out[key] = BlueprintOutput.init_from_blueprint(phys_out[key])
# Preparing output parameters: name and folder.
for uniq_freq in uniq_freq_list:
key = f'{run}_{uniq_freq}'
# If possible, prepare bids renaming.
if heur_file is not None and sub is not None:
# Add run info to heur_args if more than one run is present
if run_amount > 1:
heur_args['run'] = f'{run:02d}'
# Append "recording-freq" to filename if more than one freq
if freq_amount > 1:
heur_args['record_label'] = f'{uniq_freq:.0f}Hz'
phys_out[key].filename = bids.use_heuristic(**heur_args)
# If any filename exists already because of multirun, append labels
# But warn about the non-validity of this BIDS-like name.
if run_amount > 1:
if any([
phys.filename == phys_out[key].filename
for phys in phys_out.values()
]):
phys_out[key].filename = (f'{phys_out[key].filename}'
f'_take-{run}')
LGR.warning(
'Identified multiple outputs with the same name.\n'
'Appending fake label to avoid overwriting.\n'
'!!! ATTENTION !!! the output is not BIDS compliant.\n'
'Please check heuristics to solve the problem.')
else:
phys_out[key].filename = os.path.join(
outdir,
os.path.splitext(os.path.basename(filename))[0])
# Append "run" to filename if more than one run
if run_amount > 1:
phys_out[
key].filename = f'{phys_out[key].filename}_{run:02d}'
# Append "freq" to filename if more than one freq
if freq_amount > 1:
phys_out[
key].filename = f'{phys_out[key].filename}_{uniq_freq:.0f}Hz'
LGR.info(f'Exporting files for run {run} freq {uniq_freq}')
np.savetxt(phys_out[key].filename + '.tsv.gz',
phys_out[key].timeseries,
fmt='%.8e',
delimiter='\t')
print_json(phys_out[key].filename, phys_out[key].freq,
phys_out[key].start_time, phys_out[key].ch_name)
print_summary(
filename, num_timepoints_expected,
phys_in[run].num_timepoints_found, uniq_freq,
phys_out[key].start_time,
os.path.join(
conversion_path,
os.path.splitext(os.path.basename(
phys_out[key].filename))[0]))
def phys2bids(filename,
info=False,
indir='.',
outdir='.',
heur_file=None,
sub=None,
ses=None,
chtrig=0,
chsel=None,
num_timepoints_expected=0,
tr=1,
thr=None,
ch_name=[],
chplot='',
debug=False,
quiet=False):
"""
Main workflow of phys2bids.
Runs the parser, does some checks on input, then imports
the right interface file to read the input. If only info is required,
it returns a summary onscreen.
Otherwise, it operates on the input to return a .tsv.gz file, possibily
in BIDS format.
Raises
------
NotImplementedError
If the file extension is not supported yet.
"""
# Check options to make them internally coherent pt. I
# #!# This can probably be done while parsing?
outdir = utils.check_input_dir(outdir)
utils.path_exists_or_make_it(outdir)
# Create logfile name
basename = 'phys2bids_'
extension = 'tsv'
isotime = datetime.datetime.now().strftime('%Y-%m-%dT%H%M%S')
logname = os.path.join(outdir, (basename + isotime + '.' + extension))
# Set logging format
log_formatter = logging.Formatter(
'%(asctime)s\t%(name)-12s\t%(levelname)-8s\t%(message)s',
datefmt='%Y-%m-%dT%H:%M:%S')
# Set up logging file and open it for writing
log_handler = logging.FileHandler(logname)
log_handler.setFormatter(log_formatter)
sh = logging.StreamHandler()
if quiet:
logging.basicConfig(level=logging.WARNING, handlers=[log_handler, sh])
elif debug:
logging.basicConfig(level=logging.DEBUG, handlers=[log_handler, sh])
else:
logging.basicConfig(level=logging.INFO, handlers=[log_handler, sh])
version_number = _version.get_versions()['version']
LGR.info(f'Currently running phys2bids version {version_number}')
LGR.info(f'Input file is {filename}')
# Check options to make them internally coherent pt. II
# #!# This can probably be done while parsing?
indir = utils.check_input_dir(indir)
filename, ftype = utils.check_input_type(filename, indir)
if heur_file:
heur_file = utils.check_input_ext(heur_file, '.py')
utils.check_file_exists(heur_file)
infile = os.path.join(indir, filename)
utils.check_file_exists(infile)
# Read file!
if ftype == 'acq':
from phys2bids.interfaces.acq import populate_phys_input
elif ftype == 'txt':
from phys2bids.interfaces.txt import populate_phys_input
else:
# #!# We should add a logger here.
raise NotImplementedError('Currently unsupported file type.')
LGR.info(f'Reading the file {infile}')
phys_in = populate_phys_input(infile, chtrig)
LGR.info('Reading infos')
phys_in.print_info(filename)
# #!# Here the function viz.plot_channel should be called
if chplot != '' or info:
viz.plot_all(phys_in.ch_name, phys_in.timeseries, phys_in.units,
phys_in.freq, infile, chplot)
# If only info were asked, end here.
if info:
return
# Create trigger plot. If possible, to have multiple outputs in the same
# place, adds sub and ses label.
if tr != 0 and num_timepoints_expected != 0:
# Run analysis on trigger channel to get first timepoint and the time offset.
# #!# Get option of no trigger! (which is wrong practice or Respiract)
phys_in.check_trigger_amount(chtrig, thr, num_timepoints_expected, tr)
LGR.info('Plot trigger')
plot_path = os.path.join(
outdir,
os.path.splitext(os.path.basename(filename))[0])
if sub:
plot_path += f'_sub-{sub}'
if ses:
plot_path += f'_ses-{ses}'
viz.plot_trigger(phys_in.timeseries[0], phys_in.timeseries[chtrig],
plot_path, tr, phys_in.thr, num_timepoints_expected,
filename)
else:
LGR.warning('Skipping trigger pulse count. If you want to run it, '
'call phys2bids using "-ntp" and "-tr" arguments')
# The next few lines remove the undesired channels from phys_in.
if chsel:
LGR.info('Dropping unselected channels')
for i in reversed(range(0, phys_in.ch_amount)):
if i not in chsel:
phys_in.delete_at_index(i)
# If requested, change channel names.
if ch_name:
LGR.info('Renaming channels with given names')
phys_in.rename_channels(ch_name)
# The next few lines create a dictionary of different BlueprintInput
# objects, one for each unique frequency in phys_in
uniq_freq_list = set(phys_in.freq)
output_amount = len(uniq_freq_list)
if output_amount > 1:
LGR.warning(f'Found {output_amount} different frequencies in input!')
LGR.info(f'Preparing {output_amount} output files.')
phys_out = {} # create phys_out dict that will have a
# blueprint object per frequency
# for each different frequency
for uniq_freq in uniq_freq_list:
# copy the phys_in object to the new dict entry
phys_out[uniq_freq] = deepcopy(phys_in)
# this counter will take into account how many channels are eliminated
count = 0
# for each channel in the original phys_in object
# take the frequency
for idx, i in enumerate(phys_in.freq):
# if that frequency is different than the frequency of the phys_obj entry
if i != uniq_freq:
# eliminate that channel from the dict since we only want channels
# with the same frequency
phys_out[uniq_freq].delete_at_index(idx - count)
# take into acount the elimination so in the next eliminated channel we
# eliminate correctly
count += 1
# Also create a BlueprintOutput object for each unique frequency found.
# Populate it with the corresponding blueprint input and replace it
# in the dictionary.
phys_out[uniq_freq] = BlueprintOutput.init_from_blueprint(
phys_out[uniq_freq])
if heur_file and sub:
LGR.info(f'Preparing BIDS output using {heur_file}')
elif heur_file and not sub:
LGR.warning(f'While "-heur" was specified, option "-sub" was not.\n'
f'Skipping BIDS formatting.')
# Preparing output parameters: name and folder.
for uniq_freq in uniq_freq_list:
# If possible, prepare bids renaming.
if heur_file and sub:
if output_amount > 1:
# Add "recording-freq" to filename if more than one freq
outfile = use_heuristic(heur_file, sub, ses, filename, outdir,
uniq_freq)
else:
outfile = use_heuristic(heur_file, sub, ses, filename, outdir)
else:
outfile = os.path.join(
outdir,
os.path.splitext(os.path.basename(filename))[0])
if output_amount > 1:
# Append "freq" to filename if more than one freq
outfile = f'{outfile}_{uniq_freq}'
LGR.info(f'Exporting files for freq {uniq_freq}')
savetxt(outfile + '.tsv.gz',
phys_out[uniq_freq].timeseries,
fmt='%.8e',
delimiter='\t')
print_json(outfile, phys_out[uniq_freq].freq,
phys_out[uniq_freq].start_time, phys_out[uniq_freq].ch_name)
print_summary(filename, num_timepoints_expected,
phys_in.num_timepoints_found, uniq_freq,
phys_out[uniq_freq].start_time, outfile)
def _main(argv=None):
"""
Main workflow of phys2bids.
Runs the parser, does some checks on input, then imports
the right interface file to read the input. If only info is required,
it returns a summary onscreen.
Otherwise, it operates on the input to return a .tsv.gz file, possibily
in BIDS format.
"""
options = _get_parser().parse_args(argv)
# Check options to make them internally coherent
# #!# This can probably be done while parsing?
options.indir = utils.check_input_dir(options.indir)
options.outdir = utils.check_input_dir(options.outdir)
options.filename, ftype = utils.check_input_type(options.filename,
options.indir)
if options.heur_file:
options.heur_file = utils.check_input_ext(options.heur_file, '.py')
utils.check_file_exists(options.heur_file)
infile = os.path.join(options.indir, options.filename)
utils.check_file_exists(infile)
outfile = os.path.join(
options.outdir,
os.path.splitext(os.path.basename(options.filename))[0])
# Read file!
if ftype == 'acq':
from phys2bids.interfaces.acq import populate_phys_input
elif ftype == 'txt':
from phys2bids.interfaces.txt import populate_phys_input
else:
# #!# We should add a logger here.
raise NotImplementedError('Currently unsupported file type.')
print('Reading the file')
phys_in = populate_phys_input(infile, options.chtrig)
print('Reading infos')
phys_in.print_info(options.filename)
# #!# Here the function viz.plot_channel should be called
# for the desired channels.
# If only info were asked, end here.
if options.info:
return
# Run analysis on trigger channel to get first timepoint and the time offset.
# #!# Get option of no trigger! (which is wrong practice or Respiract)
phys_in.check_trigger_amount(options.thr, options.num_timepoints_expected,
options.tr)
# Create output folder if necessary
print('Checking that the output folder exists')
utils.path_exists_or_make_it(options.outdir)
# Create trigger plot. If possible, to have multiple outputs in the same
# place, adds sub and ses label.
print('Plot trigger')
plot_path = deepcopy(outfile)
if options.sub:
plot_path += f'_sub-{options.sub}'
if options.ses:
plot_path += f'_sub-{options.ses}'
viz.plot_trigger(phys_in.timeseries[0], phys_in.timeseries[1], plot_path,
options)
# The next few lines remove the undesired channels from phys_in.
if options.chsel:
print('Dropping unselected channels')
for i in reversed(range(0, phys_in.ch_amount)):
if i not in options.chsel:
phys_in.delete_at_index(i)
# If requested, change channel names.
if options.ch_name:
print('Renaming channels with given names')
phys_in.rename_channels(options.ch_name)
# The next few lines create a dictionary of different BlueprintInput
# objects, one for each unique frequency in phys_in
uniq_freq_list = set(phys_in.freq)
output_amount = len(uniq_freq_list)
if output_amount > 1:
print(f'Found {output_amount} different frequencies in input!')
print(f'Preparing {output_amount} output files.')
phys_out = {}
for uniq_freq in uniq_freq_list:
phys_out[uniq_freq] = deepcopy(phys_in)
for i in reversed(phys_in.freq):
if i != uniq_freq:
phys_out[uniq_freq].delete_at_index(phys_in.ch_amount - i - 1)
# Also create a BlueprintOutput object for each unique frequency found.
# Populate it with the corresponding blueprint input and replace it
# in the dictionary.
phys_out[uniq_freq] = BlueprintOutput.init_from_blueprint(
phys_out[uniq_freq])
if options.heur_file and options.sub:
print(f'Preparing BIDS output using {options.heur_file}')
elif options.heur_file and not options.sub:
print(f'While "-heur" was specified, option "-sub" was not.\n'
f'Skipping BIDS formatting.')
for uniq_freq in uniq_freq_list:
# If possible, prepare bids renaming.
if options.heur_file and options.sub:
if output_amount > 1:
# Add "recording-freq" to filename if more than one freq
outfile = use_heuristic(options.heur_file, options.sub,
options.ses, options.filename,
options.outdir, uniq_freq)
else:
outfile = use_heuristic(options.heur_file, options.sub,
options.ses, options.filename,
options.outdir)
elif output_amount > 1:
# Append "freq" to filename if more than one freq
outfile = f'outfile_{uniq_freq}'
print(f'Exporting files for freq {uniq_freq}')
savetxt(outfile + '.tsv.gz',
phys_out[uniq_freq].timeseries,
fmt='%.8e',
delimiter='\t')
print_json(outfile, phys_out[uniq_freq].freq,
phys_out[uniq_freq].start_time, phys_out[uniq_freq].ch_name)
print_summary(options.filename, options.num_timepoints_expected,
phys_in.num_timepoints_found, uniq_freq,
phys_out[uniq_freq].start_time, outfile)