def __init__(self, parent=None):
super(ThreadedTasks, self).__init__(parent)
self.selection_task = inthread(SelectionTask)(impatient=True)
self.correlograms_task = inprocess(CorrelogramsTask)(
impatient=True, use_master_thread=False)
# HACK: the similarity matrix view does not appear to update on
# some versions of Mac+Qt, but it seems to work with inthread
if sys.platform == 'darwin':
self.similarity_matrix_task = inthread(SimilarityMatrixTask)(
impatient=True)
else:
self.similarity_matrix_task = inprocess(SimilarityMatrixTask)(
impatient=True, use_master_thread=False)
def __init__(self, parent=None):
super(ThreadedTasks, self).__init__(parent)
self.selection_task = inthread(SelectionTask)(
impatient=True)
self.correlograms_task = inprocess(CorrelogramsTask)(
impatient=True, use_master_thread=False)
# HACK: the similarity matrix view does not appear to update on
# some versions of Mac+Qt, but it seems to work with inthread
if sys.platform == 'darwin':
self.similarity_matrix_task = inthread(SimilarityMatrixTask)(
impatient=True)
else:
self.similarity_matrix_task = inprocess(SimilarityMatrixTask)(
impatient=True, use_master_thread=False)
def __init__(self, folder=None):
# get list of images in the folder
if len(sys.argv) > 1:
folder = sys.argv[1]
else:
folder = '.'
self.folder = folder
self.files = sorted(filter(lambda f: f.lower().endswith('.jpg'), os.listdir(folder)))
self.cache = {}
self.n = len(self.files)
# Number of images to keep forward/backward in cache.
self.steps = 2
self.loader = inprocess(Loader)()
self.nav = Navigator(self.n, self.steps)
self.set_index(0)
def test_tasksinprocess_state():
tasks = qtools.inprocess(TestTasks)(7)
tasks.operation(3, 4, coeff=2)
x = tasks.kwarg0
assert x == 7
tasks.join()
def test_tasksinprocess_constructor():
tasks = qtools.inprocess(TestTasks)(7)
tasks.operation(3, 4, coeff=2)
tasks.join()
def load(self, filename, fileindex=1, probefile=None):#, progressbar=None):
# load XML
self.holder = DataHolder()
try:
path = get_actual_filename(filename, 'xml', None)
params = parse_xml(path, fileindex=fileindex)
except Exception as e:
raise Exception(("The XML file was not found and the data cannot "
"be loaded."))
# klusters tests
nchannels = params['nchannels']
nsamples = params['nsamples']
fetdim = params['fetdim']
freq = params['rate']
self.filename = filename
self.fileindex = fileindex
# if filename.endswith('_spiky'):
# filename = filename.replace('_spiky', '')
# spiky = True
# else:
# spiky = False
# FEATURES
# -------------------------------------------------
# features = load_text_fast(filename + ".fet.%d" % fileindex, np.int32, skiprows=1)
path = get_actual_filename(filename, 'fet', fileindex)
features, headers = load_text_pandas(path, np.int32, skiprows=1, returnheaders=True)
features = np.array(features, dtype=np.float32)
# Find out the number of extra features.
nfet = int(headers[0])
# HACK: sometimes, problem with penultimate column due to double white space
if features.shape[1] != nfet:
features = np.hstack((features[:,:-2], features[:,-1].reshape((-1, 1))))
nextrafet = nfet - fetdim * nchannels
# HACK: there are either 1 or 5 dimensions more than fetdim*nchannels
# we can't be sure so we first try 1, if it does not work we try 5
# try:
features = features.reshape((-1, nfet))
# except:
# raise Exception(("The FET file was not found and the data cannot "
# "be loaded."))
# log_debug("The number of columns is not fetdim (%d) x nchannels (%d) + 1." \
# % (fetdim, nchannels))
# try:
# features = features.reshape((-1, fetdim * nchannels + 5))
# except:
# log_debug("The number of columns is not fetdim (%d) x nchannels (%d) + 5, so I'm confused and I can't continue. Sorry :(" \
# % (fetdim, nchannels))
nspikes = features.shape[0]
# CLUSTERS
# -------------------------------------------------
try:
# if spiky:
# path = filename + "_spiky.clu.%d" % fileindex
# else:
# path = filename + ".clu.%d" % fileindex
path = get_actual_filename(filename, 'clu', fileindex)
# clusters = load_text(path, np.int32)
clusters = load_text_pandas(path, np.int32)
except Exception as e:
log_warn("CLU file '%s' not found" % filename)
clusters = np.zeros(nspikes + 1, dtype=np.int32)
clusters[0] = 1
# nclusters = clusters[0]
clusters = clusters[1:]
# if progressbar:
# progressbar.setValue(1)
ssignals.emit(self, 'FileLoading', .2)
# get the spiketimes
spiketimes = features[:,-1].copy()
# remove the last column in features, containing the spiketimes
# features = features[:,:nchannels * fetdim]
nextrafet = features.shape[1] - nchannels * fetdim
# normalize normal features
m = features[:,:-nextrafet].min()
M = features[:,:-nextrafet].max()
# force symmetry
vx = max(np.abs(m), np.abs(M))
m, M = -vx, vx
features[:,:-nextrafet] = -1+2*(features[:,:-nextrafet]-m)/(M-m)
# normalize extra features
m = features[:,-nextrafet:].min()
M = features[:,-nextrafet:].max()
# # force symmetry
# vx = max(np.abs(m), np.abs(M))
# m, M = -vx, vx
features[:,-nextrafet:] = -1+2*(features[:,-nextrafet:]-m)/(M-m)
# if progressbar:
# progressbar.setValue(2)
ssignals.emit(self, 'FileLoading', .4)
# MASKS
# -------------------------------------------------
# first: try fmask
try:
# masks = load_text(filename + ".fmask.%d" % fileindex, np.float32, skiprows=1)
path = get_actual_filename(filename, 'fmask', fileindex)
masks = load_text_pandas(path, np.float32, skiprows=1)
self.holder.masks_complete = masks
masks = masks[:,:-1:fetdim]
# masks = masks[::fetdim]
except Exception as e:
try:
# otherwise, try mask
# masks = load_text(filename + ".mask.%d" % fileindex, np.float32, skiprows=1)
path = get_actual_filename(filename, 'mask', fileindex)
masks = load_text_pandas(path, np.float32, skiprows=1)
# masks = masks[:,:-1:fetdim]
self.holder.masks_complete = masks
masks = masks[:,:-1:fetdim]
# masks = masks[::fetdim]
except:
# finally, warning and default masks (everything to 1)
log_warn("MASK file '%s' not found" % filename)
masks = np.ones((nspikes, nchannels))
self.holder.masks_complete = np.ones(features.shape)
# if progressbar:
# progressbar.setValue(3)
ssignals.emit(self, 'FileLoading', .6)
# WAVEFORMS
# -------------------------------------------------
try:
path = get_actual_filename(filename, 'spk', fileindex)
waveforms = load_binary(path)
# print waveforms.shape
# print (nspikes, nsamples, nchannels)
# DEBUG
# nchannels = 32
# print waveforms.shape
# print nspikes * nsamples * nchannels
waveforms = waveforms.reshape((nspikes, nsamples, nchannels))
except IOError as e:
log_warn("SPK file '%s' not found" % filename)
waveforms = np.zeros((nspikes, nsamples, nchannels))
# if progressbar:
# progressbar.setValue(4)
ssignals.emit(self, 'FileLoading', .8)
self.holder.freq = freq
self.holder.nspikes = nspikes
self.holder.nchannels = nchannels
self.holder.nextrafet = nextrafet
# construct spike times from random interspike interval
self.holder.spiketimes = spiketimes
self.holder.duration = spiketimes[-1] / float(self.holder.freq)
# normalize waveforms at once
waveforms = (waveforms - waveforms.mean())
waveforms = waveforms / np.abs(waveforms).max()
self.holder.waveforms = waveforms
self.holder.waveforms_info = dict(nsamples=nsamples)
self.holder.fetdim = fetdim
self.holder.features = features
self.holder.masks = masks
self.holder.clusters = clusters
# create the groups info object
# Default groups
# GROUPS
# --------------------------------------
try:
path = get_actual_filename(filename, 'groups', fileindex)
info = load_pickle(path)
clusters_info = info['clusters_info']
groups_info = info['groups_info']
except:
groups_info = {
0: dict(groupidx=0, name='Noise', color=0, spkcount=0),
1: dict(groupidx=1, name='Multi-unit', color=1, spkcount=0),
2: dict(groupidx=2, name='Good', color=2, spkcount=nspikes),
}
clusters_info = get_clusters_info(clusters, groupidx=2)
nclusters = len(clusters_info)
self.holder.nclusters = nclusters
self.holder.clusters_info = dict(
clusters_info=clusters_info,
groups_info=groups_info,
)
# c = Counter(clusters)
# self.holder.clusters_counter = c
probe = None
try:
if probefile:
probe = np.loadtxt(probefile)
except Exception as e:
print(str(e))
self.holder.probe = dict(positions=probe)
# cross correlograms
nsamples_correlograms = 20
self.holder.correlograms_info = dict(nsamples=nsamples_correlograms)
# self.holder.correlation_matrix = rdn.rand(nclusters, nclusters)
# self.holder.correlation_matrix = np.array([[]])
# features =
# self.holder.correlation_matrix = correlation_matrix(features, clusters)
# self.holder.correlation_matrix_queue = CorrelationMatrixQueue(self.holder)
# self.holder.correlation_matrix_queue.process()
# TASKS.add('correlation_matrix_queue', CorrelationMatrixQueue, self.holder)
tasks.TASKS.correlation_matrix_queue = inprocess(tasks.CorrelationMatrixQueue)(self.holder)
tasks.TASKS.correlation_matrix_queue.process()
return self.holder
def test_tasksinprocess_state():
tasks = qtools.inprocess(TestTasks)(7)
tasks.operation(3, 4, coeff=2)
x = tasks.kwarg0
assert x == 7
tasks.join()
def test_tasksinprocess_constructor():
tasks = qtools.inprocess(TestTasks)(7)
tasks.operation(3, 4, coeff=2)
tasks.join()