def on_data(self, ispan):
idata = ispan.data
if self.data_format == 'filterbank':
if len(idata.shape) == 3:
idata.tofile(self.ofile)
else:
for b in range(idata.shape[0]):
idata[b].tofile(self.ofiles[b])
elif self.data_format == 'timeseries':
ndim = len(idata.shape)
if self.pol_axis != ndim - 1:
perm = list(range(ndim))
del perm[self.pol_axis]
perm.append(self.pol_axis)
idata = transpose(idata, perm)
if ndim == 2:
idata.tofile(self.ofile)
else:
for d in range(idata.shape[0]):
idata[d].tofile(self.ofiles[d])
elif self.data_format == 'pulseprofile':
time_unix = self.t0 + ispan.frame_offset * self.dt
filename = self.filename + '.%017.6f.tim' % time_unix
with open(filename, 'wb') as ofile:
self.sigproc_hdr['tstart'] += self.sigproc_hdr['tsamp']
sigproc.write_header(self.sigproc_hdr, ofile)
idata.tofile(ofile)
else:
raise ValueError("Internal error: Unknown data format!")
def on_data(self, ispan):
idata = ispan.data
if self.data_format == 'filterbank':
if len(idata.shape) == 3:
idata.tofile(self.ofile)
else:
for b in xrange(idata.shape[0]):
idata[b].tofile(self.ofiles[b])
elif self.data_format == 'timeseries':
if len(idata.shape) == 2:
idata.tofile(self.ofile)
else:
for d in xrange(idata.shape[0]):
idata[d].tofile(self.ofiles[d])
elif self.data_format == 'pulseprofile':
time_unix = self.t0 + ispan.frame_offset * self.dt
filename = self.filename + '.%017.6f.tim' % time_unix
with open(filename, 'wb') as ofile:
self.sigproc_hdr['tstart'] += self.sigproc_hdr['tsamp']
sigproc.write_header(self.sigproc_hdr, ofile)
idata.tofile(ofile)
else:
raise ValueError("Internal error: Unknown data format!")
def on_sequence(self, iseq):
ihdr = iseq.header
itensor = ihdr['_tensor']
axnames = list(itensor['labels'])
shape = list(itensor['shape'])
scales = list(itensor['scales'])
units = list(itensor['units'])
ndim = len(shape)
dtype = DataType(itensor['dtype'])
sigproc_hdr = {}
_copy_item_if_exists(sigproc_hdr, ihdr, 'source_name')
_copy_item_if_exists(sigproc_hdr, ihdr, 'rawdatafile')
_copy_item_if_exists(sigproc_hdr, ihdr, 'az_start')
_copy_item_if_exists(sigproc_hdr, ihdr, 'za_start')
_copy_item_if_exists(sigproc_hdr, ihdr, 'raj', 'src_raj')
_copy_item_if_exists(sigproc_hdr, ihdr, 'dej', 'src_dej')
if 'telescope' in ihdr:
sigproc_hdr['telescope_id'] = sigproc.telescope2id(
ihdr['telescope'])
if 'machine' in ihdr:
sigproc_hdr['machine_id'] = sigproc.machine2id(ihdr['machine'])
_copy_item_if_exists(sigproc_hdr, ihdr, 'telescope_id')
_copy_item_if_exists(sigproc_hdr, ihdr, 'machine_id')
_copy_item_if_exists(sigproc_hdr, ihdr, 'ibeam')
_copy_item_if_exists(sigproc_hdr, ihdr, 'nbeams')
sigproc_hdr['nbits'] = dtype.itemsize_bits
_copy_item_if_exists(sigproc_hdr, ihdr, 'barycentric')
_copy_item_if_exists(sigproc_hdr, ihdr, 'pulsarcentric')
if dtype.is_integer and dtype.is_signed:
sigproc_hdr['signed'] = True
if 'coord_frame' in ihdr:
coord_frame = ihdr['coord_frame']
else:
coord_frame = None
sigproc_hdr['pulsarcentric'] = (coord_frame == 'pulsarcentric')
sigproc_hdr['barycentric'] = (coord_frame == 'barycentric')
filename = os.path.join(self.path, ihdr['name'])
if ndim >= 3 and axnames[-3:] == ['time', 'pol', 'freq']:
self.data_format = 'filterbank'
assert (dtype.is_real)
sigproc_hdr['data_type'] = 1
sigproc_hdr['nifs'] = shape[-2]
sigproc_hdr['nchans'] = shape[-1]
sigproc_hdr['tstart'] = _unix2mjd(scales[-3][0])
sigproc_hdr['tsamp'] = convert_units(scales[-3][1], units[-3], 's')
sigproc_hdr['fch1'] = convert_units(scales[-1][0], units[-1],
'MHz')
sigproc_hdr['foff'] = convert_units(scales[-1][1], units[-1],
'MHz')
if 'refdm' in ihdr:
sigproc_hdr['refdm'] = convert_units(ihdr['refdm'],
ihdr['refdm_units'],
'pc cm^-3')
if ndim == 3:
filename += '.fil'
self.ofile = open(filename, 'wb')
sigproc.write_header(sigproc_hdr, self.ofile)
elif ndim == 4:
if axnames[-4] != 'beam':
raise ValueError("Expected first axis to be 'beam'"
" got '%s'" % axnames[-4])
nbeam = shape[-4]
sigproc_hdr['nbeams'] = nbeam
filenames = [
filename + '.%06iof.%06i.fil' % (b + 1, nbeam)
for b in range(nbeam)
]
self.ofiles = [open(fname, 'wb') for fname in filenames]
for b in range(nbeam):
sigproc_hdr['ibeam'] = b
sigproc.write_header(sigproc_hdr, self.ofiles[b])
else:
raise ValueError("Too many dimensions")
elif ndim >= 2 and 'time' in axnames and 'pol' in axnames:
pol_axis = axnames.index('pol')
if pol_axis != ndim - 1:
# Need to move pol axis
# Note: We support this because it tends to be convenient
# for rest of the pipeline to operate with pol being
# the first dim, and doing the transpose on the fly
# inside this block is unlikely to cost much relative
# to disk perf (and it's free if npol==1).
axnames.append(axnames[pol_axis])
del axnames[pol_axis]
shape.append(shape[pol_axis])
del shape[pol_axis]
scales.append(scales[pol_axis])
del scales[pol_axis]
units.append(units[pol_axis])
del units[pol_axis]
self.pol_axis = pol_axis
self.data_format = 'timeseries'
assert (dtype.is_real)
sigproc_hdr['data_type'] = 2
sigproc_hdr['nchans'] = 1
sigproc_hdr['nifs'] = shape[-2]
sigproc_hdr['tstart'] = _unix2mjd(scales[-2][0])
sigproc_hdr['tsamp'] = convert_units(scales[-2][1], units[-2], 's')
if 'cfreq' in ihdr and 'bw' in ihdr:
sigproc_hdr['fch1'] = convert_units(ihdr['cfreq'],
ihdr['cfreq_units'], 'MHz')
sigproc_hdr['foff'] = convert_units(ihdr['bw'],
ihdr['bw_units'], 'MHz')
# TODO: Write ndim separate output files, each with its own refdm
if ndim == 2:
if 'refdm' in ihdr:
sigproc_hdr['refdm'] = convert_units(
ihdr['refdm'], ihdr['refdm_units'], 'pc cm^-3')
filename += '.tim'
self.ofile = open(filename, 'wb')
sigproc.write_header(sigproc_hdr, self.ofile)
elif ndim == 3:
if axnames[-3] != 'dispersion':
raise ValueError("Expected first axis to be 'dispersion'"
" got '%s'" % axnames[-3])
ndm = shape[-3]
dm0 = scales[-3][0]
ddm = scales[-3][1]
dms = [dm0 + ddm * d for d in range(ndm)]
dms = [convert_units(dm, units[-3], 'pc cm^-3') for dm in dms]
filenames = [filename + '.%09.2f.tim' % dm for dm in dms]
self.ofiles = [open(fname, 'wb') for fname in filenames]
for d, dm in enumerate(dms):
sigproc_hdr['refdm'] = dm
sigproc.write_header(sigproc_hdr, self.ofiles[d])
else:
raise ValueError("Too many dimensions")
elif ndim == 4 and axnames[-3:] == ['pol', 'freq', 'phase']:
self.data_format = 'pulseprofile'
assert (dtype.is_real)
sigproc_hdr['data_type'] = 2
sigproc_hdr['nifs'] = shape[-3]
sigproc_hdr['nchans'] = shape[-2]
sigproc_hdr['nbins'] = shape[-1]
sigproc_hdr['tstart'] = _unix2mjd(scales[-4][0])
sigproc_hdr['tsamp'] = convert_units(scales[-4][1], units[-4], 's')
sigproc_hdr['fch1'] = convert_units(scales[-2][0], units[-2],
'MHz')
sigproc_hdr['foff'] = convert_units(scales[-2][1], units[-2],
'MHz')
if 'refdm' in ihdr:
sigproc_hdr['refdm'] = convert_units(ihdr['refdm'],
ihdr['refdm_units'],
'pc cm^-3')
_copy_item_if_exists(sigproc_hdr, ihdr, 'npuls')
self.filename = filename
self.sigproc_hdr = sigproc_hdr
self.t0 = scales[-4][0]
self.dt = scales[-4][1]
else:
raise ValueError(
"Axis labels do not correspond to a known data format: " +
str(axnames) + "\nKnown formats are:" +
"\n [time, pol, freq]\n [beam, time, pol]\n" +
" [time, pol]\n [dispersion, time, pol]\n" +
" [pol, freq, phase]")
def on_sequence(self, iseq):
ihdr = iseq.header
itensor = ihdr['_tensor']
axnames = tuple(itensor['labels'])
shape = itensor['shape']
scales = itensor['scales']
units = itensor['units']
ndim = len(shape)
dtype = DataType(itensor['dtype'])
sigproc_hdr = {}
copy_item_if_exists(sigproc_hdr, ihdr, 'source_name')
copy_item_if_exists(sigproc_hdr, ihdr, 'rawdatafile')
copy_item_if_exists(sigproc_hdr, ihdr, 'az_start')
copy_item_if_exists(sigproc_hdr, ihdr, 'za_start')
copy_item_if_exists(sigproc_hdr, ihdr, 'raj', 'src_raj')
copy_item_if_exists(sigproc_hdr, ihdr, 'dej', 'src_dej')
if 'telescope' in ihdr:
sigproc_hdr['telescope_id'] = sigproc.telescope2id(
ihdr['telescope'])
if 'machine' in ihdr:
sigproc_hdr['machine_id'] = sigproc.machine2id(ihdr['machine'])
copy_item_if_exists(sigproc_hdr, ihdr, 'telescope_id')
copy_item_if_exists(sigproc_hdr, ihdr, 'machine_id')
copy_item_if_exists(sigproc_hdr, ihdr, 'ibeam')
copy_item_if_exists(sigproc_hdr, ihdr, 'nbeams')
sigproc_hdr['nbits'] = dtype.itemsize_bits
copy_item_if_exists(sigproc_hdr, ihdr, 'barycentric')
copy_item_if_exists(sigproc_hdr, ihdr, 'pulsarcentric')
if dtype.is_integer and dtype.is_signed:
sigproc_hdr['signed'] = True
if 'coord_frame' in ihdr:
coord_frame = ihdr['coord_frame']
else:
coord_frame = None
sigproc_hdr['pulsarcentric'] = (coord_frame == 'pulsarcentric')
sigproc_hdr['barycentric'] = (coord_frame == 'barycentric')
filename = os.path.join(self.path, ihdr['name'])
if ndim >= 3 and axnames[-3:] == ('time', 'pol', 'freq'):
self.data_format = 'filterbank'
assert (dtype.is_real)
sigproc_hdr['data_type'] = 1
sigproc_hdr['nifs'] = shape[-2]
sigproc_hdr['nchans'] = shape[-1]
sigproc_hdr['tstart'] = unix2mjd(scales[-3][0])
sigproc_hdr['tsamp'] = convert_units(scales[-3][1], units[-3], 's')
sigproc_hdr['fch1'] = convert_units(scales[-1][0], units[-1],
'MHz')
sigproc_hdr['foff'] = convert_units(scales[-1][1], units[-1],
'MHz')
if 'refdm' in ihdr:
sigproc_hdr['refdm'] = convert_units(ihdr['refdm'],
ihdr['refdm_units'],
'pc cm^-3')
if ndim == 3:
filename += '.fil'
self.ofile = open(filename, 'wb')
sigproc.write_header(sigproc_hdr, self.ofile)
elif ndim == 4:
if axnames[-4] != 'beam':
raise ValueError("Expected first axis to be 'beam'")
nbeam = shape[-4]
sigproc_hdr['nbeams'] = nbeam
filenames = [
filename + '.%06iof.%06i.fil' % (b + 1, nbeam)
for b in xrange(nbeam)
]
self.ofiles = [open(fname, 'wb') for fname in filenames]
for b in xrange(nbeam):
sigproc_hdr['ibeam'] = b
sigproc.write_header(sigproc_hdr, self.ofiles[b])
else:
raise ValueError("Too many dimensions")
elif ndim >= 2 and axnames[-2:] == ('time', 'pol'):
self.data_format = 'timeseries'
assert (dtype.is_real)
sigproc_hdr['data_type'] = 2
sigproc_hdr['nchans'] = 1
sigproc_hdr['nifs'] = shape[-2]
sigproc_hdr['tstart'] = unix2mjd(scales[-2][0])
sigproc_hdr['tsamp'] = convert_units(scales[-2][1], units[-2], 's')
if 'cfreq' in ihdr and 'bw' in ihdr:
sigproc_hdr['fch1'] = convert_units(ihdr['cfreq'],
ihdr['cfreq_units'], 'MHz')
sigproc_hdr['foff'] = convert_units(ihdr['bw'],
ihdr['bw_units'], 'MHz')
# TODO: Write ndim separate output files, each with its own refdm
if ndim == 2:
if 'refdm' in ihdr:
sigproc_hdr['refdm'] = convert_units(
ihdr['refdm'], ihdr['refdm_units'], 'pc cm^-3')
filename += '.tim'
self.ofile = open(filename, 'wb')
sigproc.write_header(sigproc_hdr, self.ofile)
elif ndim == 3:
if axnames[-3] != 'dispersion measure':
raise ValueError(
"Expected first axis to be 'dispersion measure'")
ndm = shape[-3]
dm0 = scales[-3][0]
ddm = scales[-3][1]
dms = [dm0 + ddm * d for d in xrange(ndm)]
dms = [convert_units(dm, units[-3], 'pc cm^-3') for dm in dms]
filenames = [filename + '.%09.2f.tim' % dm for dm in dms]
self.ofiles = [open(fname, 'wb') for fname in filenames]
for d, dm in enumerate(dms):
sigproc_hdr['refdm'] = dm
sigproc.write_header(sigproc_hdr, self.ofiles[d])
else:
raise ValueError("Too many dimensions")
elif ndim == 4 and axnames[-3:] == ('pol', 'freq', 'phase'):
self.data_format = 'pulseprofile'
assert (dtype.is_real)
sigproc_hdr['data_type'] = 2
sigproc_hdr['nifs'] = shape[-3]
sigproc_hdr['nchans'] = shape[-2]
sigproc_hdr['nbins'] = shape[-1]
sigproc_hdr['tstart'] = unix2mjd(scales[-4][0])
sigproc_hdr['tsamp'] = convert_units(scales[-4][1], units[-4], 's')
sigproc_hdr['fch1'] = convert_units(scales[-2][0], units[-2],
'MHz')
sigproc_hdr['foff'] = convert_units(scales[-2][1], units[-2],
'MHz')
if 'refdm' in ihdr:
sigproc_hdr['refdm'] = convert_units(ihdr['refdm'],
ihdr['refdm_units'],
'pc cm^-3')
copy_item_if_exists(sigproc_hdr, ihdr, 'npuls')
self.filename = filename
self.sigproc_hdr = sigproc_hdr
self.t0 = scales[-4][0]
self.dt = scales[-4][1]
else:
raise ValueError(
"Axis labels do not correspond to a known data format: " +
str(axnames))