def __init__(self, your_object):
self.your_object = your_object
if self.your_object.isfits:
logger.debug(f'Calculating dada size and data step for the fits files')
self.list_of_subints = self.your_object.specinfo.num_subint.astype('int')
if len(self.list_of_subints) > 1:
# if there is more than one files see how many subints we need to read so that the data is equally split
self.subint_steps = int(find_gcd(self.list_of_subints))
else:
# if there is just one large file, read it in chunks
self.subint_steps = int(np.max(np.prod(np.unique((primes(self.list_of_subints))))))
self.dada_size = self.subint_steps * self.your_object.your_header.nchans * self.your_object.specinfo.spectra_per_subint # * self.your_object.nbits / 8 # bytes
self.data_step = int(self.subint_steps * self.your_object.specinfo.spectra_per_subint)
else:
nsamp_gulp = 2 ** 18
logger.debug(f'Calculating dada size and data step for the filterbank file')
if self.your_object.your_header.nspectra < nsamp_gulp:
self.dada_size = self.your_object.your_header.nspectra * self.your_object.your_header.nchans * self.your_object.your_header.nbits / 8 # bytes
self.data_step = int(self.your_object.your_header.nspectra)
else:
self.data_step = int(closest_divisor(self.your_object.your_header.nspectra, nsamp_gulp))
self.dada_size = self.data_step * self.your_object.your_header.nchans * self.your_object.your_header.nbits / 8 # bytes
self.dada_key = hex(np.random.randint(0, 16 ** 4))
def __init__(
self,
your_object,
nstart=0,
nsamp=None,
c_min=None,
c_max=None,
outdir=None,
outname=None,
flag_rfi=False,
progress=True,
spectral_kurtosis_sigma=4,
savgol_frequency_window=15,
savgol_sigma=4,
gulp=None,
zero_dm_subt=False,
time_decimation_factor=1,
frequency_decimation_factor=1,
replacement_policy="mean",
):
self.your_object = your_object
self.nstart = nstart
if nsamp is None:
self.nsamp = self.your_object.your_header.nspectra
else:
self.nsamp = nsamp
self.c_min = c_min
self.c_max = c_max
self.time_decimation_factor = time_decimation_factor
self.frequency_decimation_factor = frequency_decimation_factor
if self.time_decimation_factor > 1:
raise NotImplementedError("We have not implemented this feature yet.")
if self.frequency_decimation_factor > 1:
raise NotImplementedError("We have not implemented this feature yet.")
self.replacement_policy = replacement_policy
if self.replacement_policy not in ["mean", "median", "zero"]:
raise ValueError(
f"replacement_policy can only be 'mean', 'median' or 'zero'."
)
self.outdir = outdir
self.outname = outname
self.flag_rfi = flag_rfi
self.progress = progress
self.sk_sig = spectral_kurtosis_sigma
self.sg_fw = savgol_frequency_window
self.sg_sig = savgol_sigma
self.zero_dm_subt = zero_dm_subt
self.data = None
self.dada_is_set = False
if gulp is not None:
self.gulp = gulp
else:
# this logic fails if the number of samples is a prime number.
p = np.sort(primes(self.nsamp))[::-1]
if len(p) > 1:
cumprods = np.cumprod(p)
self.gulp = int(cumprods[len(cumprods) // 2])
else:
self.gulp = self.nsamp
if self.gulp > self.nsamp:
self.gulp = self.nsamp
original_dir, orig_basename = os.path.split(
self.your_object.your_header.filename
)
if not self.outname:
name, ext = os.path.splitext(orig_basename)
if ext == ".fits":
temp = name.split("_")
if len(temp) > 1:
self.outname = "_".join(temp[:-1]) + "_converted"
else:
self.outname = name + "_converted"
else:
self.outname = name + "_converted"
if self.outdir is None:
self.outdir = original_dir
logging.debug("Writer Attributes:-")
for arg, value in sorted(vars(self).items()):
logging.debug("Attribute %s: %r", arg, value)