def collect_output(datafile, min_points=1, mask_length=4):
""" collect all output from the flash program output directory created by
the flash program in flash_dir and calculate some additional stats on each flash
"""
import numpy as np
logger = logging.getLogger('FlashAutorunLogger')
# outdir = os.path.join(flash_dir,flash_output_dir)
# os.chdir(outdir)
lma = LMAdataFile(datafile, mask_length=mask_length)
# get the mapping from flash_ids to the points
order = np.argsort(lma.flash_id)
# In the case of no data in the file, lma.data.shape will have length zero, i.e., a 0-d array
if len(lma.data.shape) == 0:
# No data
flashes = []
else:
flid = lma.flash_id[order]
boundaries, = np.where(flid[1:] -
flid[:-1]) # where indices are nonzero
boundaries = np.hstack(([0], boundaries + 1))
all_data = lma.data[order]
max_idx = len(flid) #- 1
slice_lower_edges = tuple(boundaries)
slice_upper_edges = slice_lower_edges[1:] + (max_idx, )
slices = zip(slice_lower_edges, slice_upper_edges)
flashes = [Flash(all_data[slice(*sl)]) for sl in slices]
# calculate extra flash metadata, e.g., initation, centroid
logtext = "Calculating flash initation, centroid, area, etc. for %d flashes" % (
len(flashes), )
logger.info(logtext)
# print flashes[0].points.dtype
for fl in flashes:
header = ''.join(lma.header)
fl.metadata = FlashMetadata(header)
calculate_flash_stats(fl)
return lma, flashes
def collect_output(datafile, min_points=1, mask_length=4):
""" collect all output from the flash program output directory created by
the flash program in flash_dir and calculate some additional stats on each flash
"""
import numpy as np
logger = logging.getLogger('FlashAutorunLogger')
# outdir = os.path.join(flash_dir,flash_output_dir)
# os.chdir(outdir)
lma = LMAdataFile(datafile, mask_length=mask_length)
# get the mapping from flash_ids to the points
order = np.argsort(lma.flash_id)
# In the case of no data in the file, lma.data.shape will have length zero, i.e., a 0-d array
if len(lma.data.shape) == 0:
# No data
flashes = []
else:
flid = lma.flash_id[order]
boundaries, = np.where(flid[1:]-flid[:-1]) # where indices are nonzero
boundaries = np.hstack(([0], boundaries+1))
all_data = lma.data[order]
max_idx = len(flid) #- 1
slice_lower_edges = tuple(boundaries)
slice_upper_edges = slice_lower_edges[1:] + (max_idx,)
slices = zip(slice_lower_edges, slice_upper_edges)
flashes = [ Flash(all_data[slice(*sl)]) for sl in slices ]
# calculate extra flash metadata, e.g., initation, centroid
logtext = "Calculating flash initation, centroid, area, etc. for %d flashes" % (len(flashes), )
logger.info(logtext)
# print flashes[0].points.dtype
for fl in flashes:
header = ''.join(lma.header)
fl.metadata = FlashMetadata(header)
calculate_flash_stats(fl)
return lma, flashes
def create_flash_objs(lma, good_data):
""" lma is an LMAdataFile object. Its data instance gets overwritten with the sorted, qc'd, flash_id'd data.
very similar to collect_output in autorun_mflash
"""
logger = logging.getLogger('FlashAutorunLogger')
try:
while True:
(unique_labels, point_labels, all_IDs) = (yield)
# add flash_id column
empty_labels = np.empty_like(point_labels)
data = append_fields(good_data, ('flash_id',), (empty_labels,))
# all_IDs gives the index in the original data array to
# which each point_label corresponds
data['flash_id'][all_IDs] = point_labels
# In the case of no data in the file, lma.data.shape will have
# length zero, i.e., a 0-d array
if len(data.shape) == 0:
# No data
flashes = []
else:
# work first with non-singleton flashes
# to have strictly positive flash ids
print data.shape
singles = (data['flash_id'] == -1)
non_singleton = data[ np.logical_not(singles) ]
print non_singleton['flash_id'].shape
order = np.argsort(non_singleton['flash_id'])
ordered_data = non_singleton[order]
flid = ordered_data['flash_id']
if (flid.shape[0]>0):
max_flash_id = flid[-1]
else:
max_flash_id = 0
try:
assert max_flash_id == max(unique_labels)
except AssertionError:
print "Max flash ID {0} is not as expected from unique labels {1}".format(max_flash_id, max(unique_labels))
boundaries, = np.where(flid[1:]-flid[:-1]) # where indices are nonzero
boundaries = np.hstack(([0], boundaries+1))
max_idx = len(flid) #- 1
slice_lower_edges = tuple(boundaries)
slice_upper_edges = slice_lower_edges[1:] + (max_idx,)
slices = zip(slice_lower_edges, slice_upper_edges)
flashes = [ Flash(ordered_data[slice(*sl)]) for sl in slices ]
print "finished non-singletons"
# Now deal with the nonsingleton points.
# Each singleton point will have a high flash_id,
# starting with the previous maximum flash id.
singleton = data[singles]
n_singles = singleton.shape[0]
# this operation works on a view of the original data array,
# so it modifies the original data array
singleton['flash_id'] += max_flash_id + 1 + np.arange(n_singles, dtype=int)
singleton_flashes = [ Flash(singleton[i:i+1]) for i in range(n_singles)]
data[singles] = singleton
print "finished singletons"
flashes += singleton_flashes
logtext = "Calculating flash initation, centroid, area, etc. for %d flashes" % (len(flashes), )
logger.info(logtext)
# print flashes[0].points.dtype
for fl in flashes:
header = ''.join(lma.header)
fl.metadata = FlashMetadata(header)
calculate_flash_stats(fl)
# logger.info(fl.points.shape[0])
logger.info('finished setting flash metadata')
lma.raw_data = lma.data
lma.data = data
assert (lma.data['flash_id'].min() >=0) # this should be true since the singletons were modified in the original data array above
lma.sort_status = 'got some flashes'
except GeneratorExit:
lma.flash_objects = flashes
def create_flash_objs(lma, good_data):
""" lma is an LMAdataFile object. Its data instance gets overwritten with the sorted, qc'd, flash_id'd data.
very similar to collect_output in autorun_mflash
"""
logger = logging.getLogger('FlashAutorunLogger')
try:
while True:
(unique_labels, point_labels, all_IDs) = (yield)
# add flash_id column
empty_labels = np.empty_like(point_labels)
data = append_fields(good_data, ('flash_id',), (empty_labels,))
# all_IDs gives the index in the original data array to
# which each point_label corresponds
data['flash_id'][all_IDs] = point_labels
# In the case of no data in the file, lma.data.shape will have
# length zero, i.e., a 0-d array
if len(data.shape) == 0:
# No data
flashes = []
else:
# work first with non-singleton flashes
# to have strictly positive flash ids
print data.shape
singles = (data['flash_id'] == -1)
non_singleton = data[ np.logical_not(singles) ]
print non_singleton['flash_id'].shape
order = np.argsort(non_singleton['flash_id'])
ordered_data = non_singleton[order]
flid = ordered_data['flash_id']
if (flid.shape[0]>0):
max_flash_id = flid[-1]
else:
max_flash_id = 0
try:
assert max_flash_id == max(unique_labels)
except AssertionError:
print "Max flash ID {0} is not as expected from unique labels {1}".format(max_flash_id, max(unique_labels))
boundaries, = np.where(flid[1:]-flid[:-1]) # where indices are nonzero
boundaries = np.hstack(([0], boundaries+1))
max_idx = len(flid) #- 1
slice_lower_edges = tuple(boundaries)
slice_upper_edges = slice_lower_edges[1:] + (max_idx,)
slices = zip(slice_lower_edges, slice_upper_edges)
flashes = [ Flash(ordered_data[slice(*sl)]) for sl in slices ]
print "finished non-singletons"
# Now deal with the nonsingleton points.
# Each singleton point will have a high flash_id,
# starting with the previous maximum flash id.
singleton = data[singles]
n_singles = singleton.shape[0]
# this operation works on a view of the original data array,
# so it modifies the original data array
singleton['flash_id'] += max_flash_id + 1 + np.arange(n_singles, dtype=int)
singleton_flashes = [ Flash(singleton[i:i+1]) for i in range(n_singles)]
data[singles] = singleton
print "finished singletons"
flashes += singleton_flashes
logtext = "Calculating flash initation, centroid, area, etc. for %d flashes" % (len(flashes), )
logger.info(logtext)
# print flashes[0].points.dtype
for fl in flashes:
header = ''.join(lma.header)
fl.metadata = FlashMetadata(header)
calculate_flash_stats(fl)
# logger.info(fl.points.shape[0])
logger.info('finished setting flash metadata')
lma.raw_data = lma.data
lma.data = data
assert (lma.data['flash_id'].min() >=0) # this should be true since the singletons were modified in the original data array above
lma.sort_status = 'got some flashes'
except GeneratorExit:
lma.flash_objects = flashes