def sigma_mask(data,error= None,sigma=3,Verbose= False):
if type(error) == type(None):
error = np.zeros(len(data))
calcaverage = sigmacut.calcaverageclass()
calcaverage.calcaverage_sigmacutloop(data,Nsigma=sigma
,median_firstiteration=True,saveused=True)
if Verbose:
print("mean:%f (uncertainty:%f)" % (calcaverage.mean,calcaverage.mean_err))
return calcaverage.clipped
def sigmacut_meandev(pixels, mask=None):
sigmacut = calcaverageclass()
pixels = np.ravel(pixels)
initmask = np.isnan(pixels)
if mask != None:
initmask = np.logical_or(initmask, np.ravel(mask))
sigmacut.calcaverage_sigmacutloop(pixels,
mask=initmask,
Nsigma=3.0,
verbose=0)
return sigmacut.mean, sigmacut.stdev
def sigma_clipped_mean(pix):
pix = np.ravel(pix)
goodpts = ~np.isnan(pix)
nanpts = np.isnan(pix)
pixclass = sigmacut.calcaverageclass()
pixclass.calcaverage_sigmacutloop(pix,
Nsigma=3,
saveused=True,
mask=nanpts)
finalmask = np.logical_or(nanpts, pixclass.clipped)
pixclass.calcaverage_sigmacutloop(pix, Nsigma=0, mask=finalmask)
return pixclass.mean, pixclass.stdev
def make_rdnoise(self):
if self.verbose:
print('Loading {}'.format(infile))
#Read in input fits file using astropy
with fits.open(self.infile) as h:
self.data = h[1].data
self.hdr = h[1].header
self.hdr0 = h[0].header
#Keep only the first integration
if len(self.data.shape) == 4:
print('WARNING: Input data cube has 4 dimensions.')
print('Extracting the first integration only and continuing.')
self.data = self.data[0, :, :, :]
elif len(self.data.shape) < 3:
print 'ERROR: data cube has less than 3 dimensions!!! Quitting!'
sys.exit(0)
#Make sure the data are full frame
Ngroups, ny, nx = self.data.shape
if (ny, nx) != (2048, 2048):
print("WARNING: input data from {} appear not to be full frame!".
format(self.infile))
print("x,y dimensions are {}x{}".format(nx, ny))
sys.exit()
if self.verbose:
print('Number of groups: {}'.format(Ngroups))
#Set step sizes
stepsizex = self.stepsizex
stepsizey = self.stepsizey
if stepsizex is None:
stepsizex = self.boxsizex
if stepsizey is None:
stepsizey = self.boxsizey
halfstepsizex = int(0.5 * stepsizex)
halfstepsizey = int(0.5 * stepsizey)
# get the xy limits.
if (self.xmin is None):
xmin = 0
else:
xmin = self.xmin
if (self.xmax is None):
xmax = self.data.shape[2]
else:
xmax = self.xmax
if (self.ymin is None):
ymin = 0
else:
ymin = self.ymin
if (self.ymax is None):
ymax = self.data.shape[1]
else:
ymax = self.ymax
if self.verbose:
print('xmin, xmax, ymin, ymax: {}, {}, {}, {}'.format(
xmin, xmax, ymin, ymax))
sigmacut = calcaverageclass()
#Create a series of CDS frames
diffim = self.data[1:Ngroups, :, :] - self.data[0:Ngroups - 1, :, :]
# define output matrices
self.im_rdnoise = scipy.zeros((self.data.shape[1], self.data.shape[2]),
dtype=float)
self.im_rdnoise_err = scipy.zeros(self.im_rdnoise.shape, dtype=float)
im_Nused = scipy.zeros(self.im_rdnoise.shape, dtype=int)
im_Pskipped = scipy.zeros(self.im_rdnoise.shape, dtype=float)
# load mask
if self.bpm is not None:
print('Loading mask file {}'.format(self.bpm))
with fits.open(self.bpm) as bpmhdu:
self.bpmdata = bpmhdu[1].data
if self.bpmdata.shape != (2048, 2048):
print(
"WARNING! Bad pixel mask shape is incorrect or was improperly read!"
)
sys.exit()
mask = scipy.where(
scipy.logical_and(self.bpmdata, self.bpmval) > 0, True, False)
# mask the overscan (THIS ASSUMES WE ARE WORKING ON FULL FRAME DATA)
mask[0:4, :] = 1
mask[2044:2048, :] = 1
mask[:, 0:4] = 1
mask[:, 2044:2048] = 1
else:
mask = scipy.zeros((self.data.shape[1], self.data.shape[2]),
dtype=np.uint16)
#load epoxy mask if needed
detector = self.hdr0['DETECTOR']
if detector in [
'NRCA1', 'NRCA3', 'NRCA4', 'NRCALONG', 'NRCB1', 'NRCB4'
]:
epoxymask = self.load_epoxy_mask(detector)
#invert the epoxy mask if requested
if self.invert_epoxy:
print(
"Inverting the epoxy mask to work within the epoxy void region."
)
epoxymask = epoxymask - 1
epoxymask[epoxymask == -1] = 1
#add the epoxy mask to the bpm
mask = scipy.logical_or(mask, epoxymask)
#create an inverted epoxy mask for later when filling in readnoise values
#inv_epoxymask = epoxymask.astype(bool)
#inv_epoxymask = np.invert(inv_epoxymask)
x = xmin + halfstepsizex
OneoverSQR2 = 1.0 / math.sqrt(2)
xtransitions = [512, 2 * 512, 3 * 512]
if self.gmin is None:
gmin = 0
else:
gmin = self.gmin
if self.gmax is None:
gmax = diffim.shape[0]
else:
gmax = self.gmax
if gmax > diffim.shape[0]:
gmax = diffim.shape[0]
#lists of starting and stopping indexes for the filling later
xfills = []
xfille = []
yfills = []
yfille = []
grange = range(gmin, gmax)
while x < xmax:
# define the x1,x2 of the box for stats
x1 = x - int(0.5 * self.boxsizex)
if x1 < 0: x1 = 0
if self.forcexylimits and (x1 < xmin): x1 = xmin
x2 = int(x + max(1, 0.5 * self.boxsizex))
if x2 > self.data.shape[2]: x2 = self.data.shape[2]
if self.forcexylimits and (x2 > xmax): x2 = xmax
# make sure that the box contains only data from the same amp!!!
for xtransition in xtransitions:
if x >= xtransition and x1 < xtransition: x1 = xtransition
if x < xtransition and x2 >= xtransition: x2 = xtransition
#print('!!!x',x,0.5*self.boxsizex,x1,x2)
y = ymin + halfstepsizey
while y < ymax:
# define the y1,y2 of the box for stats
y1 = y - int(0.5 * self.boxsizey)
if y1 < 0: y1 = 0
if self.forcexylimits and (y1 < ymin): y1 = ymin
y2 = int(y + max(1, 0.5 * self.boxsizey))
if y2 > self.data.shape[2]: y2 = self.data.shape[1]
if self.forcexylimits and (y2 > ymax): y2 = ymax
if self.verbose:
if (x % 64) == 0 and (y == 0):
print('(x,y)=(%d,%d) box:%d:%d,%d:%d' %
(x, y, x1, x2, y1, y2))
stdevs = []
Nused = []
Pskipped = []
for g in grange:
sigmacut.calcaverage_sigmacutloop(diffim[g, y1:y2, x1:x2],
mask=mask[y1:y2, x1:x2],
Nsigma=3.0,
verbose=0)
if self.verbose:
print('x:%d y:%d g:%d' % (x, y, g), sigmacut.__str__())
if sigmacut.converged and sigmacut.Nused > self.Npixmin and 100.0 * sigmacut.Nskipped / (
sigmacut.Nused +
sigmacut.Nskipped) < self.Pclipmax:
stdevs.append(sigmacut.stdev)
Nused.append(sigmacut.Nused)
Pskipped.append(100.0 * sigmacut.Nskipped /
(sigmacut.Nused + sigmacut.Nskipped))
if len(stdevs) > 1:
sigmacut.calcaverage_sigmacutloop(np.array(stdevs),
Nsigma=3.0,
verbose=0)
if sigmacut.converged:
self.im_rdnoise[y, x] = sigmacut.mean * OneoverSQR2
self.im_rdnoise_err[
y, x] = sigmacut.mean_err * OneoverSQR2
if self.verbose:
print('x:%d y:%d average' % (x, y), sigmacut.__str__())
im_Nused[y, x] = scipy.median(Nused)
im_Pskipped[y, x] = scipy.median(Pskipped)
elif len(stdevs) == 1:
self.im_rdnoise[y, x] = sigmacut.stdev * OneoverSQR2
self.im_rdnoise_err[y,
x] = sigmacut.stdev_err * OneoverSQR2
im_Nused[y, x] = 1
im_Pskipped[y, x] = 0.0
if self.fill:
xx1 = x - halfstepsizex
for xtransition in xtransitions:
if x - stepsizex < xtransition and x >= xtransition:
xx1 = xtransition
if x - stepsizex < 0: xx1 = 0
if xx1 < x1: xx1 = x1
if xx1 < 0: xx1 = 0
xx2 = x + max(1, halfstepsizex)
for xtransition in xtransitions:
if x + stepsizex >= xtransition and x < xtransition:
xx2 = xtransition
if x + stepsizex > self.data.shape[2]:
xx2 = self.data.shape[2]
if xx2 > x2: xx2 = x2
if xx2 > self.data.shape[2]: xx2 = self.data.shape[2]
yy1 = y - halfstepsizey
if y - stepsizey < 0: yy1 = 0
if yy1 < y1: yy1 = y1
if yy1 < 0: yy1 = 0
yy2 = y + max(1, halfstepsizey)
if y + stepsizey > self.data.shape[1]:
yy2 = self.data.shape[1]
if yy2 > y2: yy2 = y2
if yy2 > self.data.shape[1]: yy2 = self.data.shape[1]
#save the x and y coordinates for filling in missing data later
xfills.append(xx1)
xfille.append(xx2)
yfills.append(yy1)
yfille.append(yy2)
if len(stdevs) > 0:
self.im_rdnoise[yy1:yy2, xx1:xx2] = self.im_rdnoise[y,
x]
self.im_rdnoise_err[yy1:yy2,
xx1:xx2] = self.im_rdnoise_err[y,
x]
im_Nused[yy1:yy2, xx1:xx2] = im_Nused[y, x]
im_Pskipped[yy1:yy2, xx1:xx2] = im_Pskipped[y, x]
y += stepsizey
x += stepsizex
#fill in the gaps in the map (i.e. the partial boxes that contain less than the minimum number
#of pixels needed for calculating the readnoise
for iter in range(self.fill_iterations):
#print('BEGINNING FILL ITERATION {}'.format(iter))
self.im_rdnoise = self.fill_empty_regions(self.im_rdnoise, xfills,
xfille, yfills, yfille)
self.im_rdnoise_err = self.fill_empty_regions(
self.im_rdnoise_err, xfills, xfille, yfills, yfille)
#mask out the pixels in the low epoxy area
if detector in [
'NRCA1', 'NRCA3', 'NRCA4', 'NRCALONG', 'NRCB1', 'NRCB4'
]:
self.im_rdnoise[epoxymask.astype(bool)] = 0.
self.im_rdnoise_err[epoxymask.astype(bool)] = 0.
#save output file
outfilename = self.save_readnoise_file(self.im_rdnoise)
#redcat team checks
#subprocess.call(['fitsverify',outfilename])
return outfilename, self.im_rdnoise
#---------------------------------------------------------------------
#Next, quantify how well the quadrant-to-quadrant offsets were removed
#by looking at the quad-to-quad offsets in the subtracted data
with fits.open(origfile) as h:
orig = h[1].data
with fits.open(refeofile) as h:
refeo = h[1].data
ints, groups, yd, xd = orig.shape
origmeans = np.zeros((ints, groups, 4))
origmeanerrs = np.zeros((ints, groups, 4))
refeomeans = np.zeros((ints, groups, 4))
refeomeanerrs = np.zeros((ints, groups, 4))
qstart = [0, 512, 1024, 1536, 2040]
sig = sigmacut.calcaverageclass()
for integration in xrange(ints):
for group in xrange(groups):
for quad in xrange(4):
#use sigma-clipped mean
sig.calcaverage_sigmacutloop(orig[integration, group, 4:2040,
qstart[quad]:qstart[quad + 1]],
Nsigma=3.0,
verbose=0)
origmeans[integration, group, quad] = sig.mean
origmeanerrs[integration, group, quad] = sig.mean_err
sig.calcaverage_sigmacutloop(refeo[integration, group, 4:2040,
qstart[quad]:qstart[quad + 1]],
Nsigma=3.0,
verbose=0)
refeomeans[integration, group, quad] = sig.mean
def run(self):
#read in listfile
files = []
with open(self.listfile) as f:
for line in f:
if len(line) > 2:
files.append(line.strip())
#read in all of the individual readnoise maps
numfiles = len(files)
for i, file in enumerate(files):
if self.verbose:
print("Reading in " + file)
map = ReadnoiseModel(file)
mapdata = map.data
if i == 0:
all_maps = np.zeros(
(len(files), mapdata.shape[0], mapdata.shape[1]))
all_maps[i, :, :] = mapdata
print("Individual readnoise maps read in")
print("Beginning averaging")
#now calculate the sigma-clipped mean readnoise for each pixel
#could make this smarter by having it look for boxes....might be
#tough in the case of epoxy voids that are irregularly shaped
readnoise_map = np.zeros(mapdata.shape)
readnoise_err = np.zeros(mapdata.shape)
for y in xrange(mapdata.shape[0]):
for x in xrange(mapdata.shape[1]):
pixdata = all_maps[:, y, x]
sigmacut = calcaverageclass()
sigmacut.calcaverage_sigmacutloop(pixdata)
readnoise_map[y, x] = sigmacut.mean
readnoise_err[y, x] = sigmacut.stdev
#convert to readnoise for a CDS pair
readnoise_map = readnoise_map * np.sqrt(2)
#find the gain reference file if needed
if self.gainfile == None:
self.gainfile = self.find_gainfile(files[0])
print("Using gain file: " + self.gainfile)
#save the mean readnoise file in ADU, so we don't need to
#repeat the steps above when the gain reference file changes
mapadu = map
mapadu.data = readnoise_map
mapadu = self.header_prep(mapadu, files)
#save the readnoise reference file
if self.outfile == None:
self.outfile = self.listfile + '_MEAN_READNOISE_MAP.fits'
dot = self.outfile.rfind('.')
adufile = self.outfile[0:dot] + '_ADU_FOR_CDS.fits'
#if the specified output file exists, delete it.
if os.path.isfile(adufile):
os.remove(adufile)
print("Saving mean CDS readnoise map in units of ADU as " + adufile)
mapadu.save(adufile)
print("Converting to electrons")
#read in gain file, using SSB's GainModel
gainmodel = GainModel(self.gainfile)
gain = gainmodel.data
#Now convert from ADU to electrons using the gain, and from single-frame to CDS
#by multiplying by sqrt(2)
readnoise_map = readnoise_map * gain
#put the readnoise_map back into the previously-opened ReadnoiseModel instance
map.data = readnoise_map
#take care of all the details for the reference file header keywords
#Since these files have already gone through mkimreadnoise_ssboutput.py, all that should
#need to be updated is the HISTORY section, with the names of all the input files.
print("Preparing header")
map = self.header_prep(map, files)
#save the readnoise reference file
if self.outfile == None:
self.outfile = self.listfile + '_MEAN_READNOISE_MAP_electrons_FOR_CDS.fits'
#if the specified output file exists, delete it.
if os.path.isfile(self.outfile):
os.remove(self.outfile)
print("Saving mean CDS readnoise image in electrons as " +
self.outfile)
map.save(self.outfile)
def mkimrdnoise(self,infile,outfilebasename,pattern='.fits',format='g%03d'):
if self.verbose:
print 'Loading ',infile
#self.data, self.hdr = pyfits.getdata(infile, 0, header=True)
#astropy.fits
with fits.open(infile) as h:
self.data = h[1].data
self.hdr = h[1].header
self.hdr0 = h[0].header
#read in via RampModel
#inramp = NRCRampModel()
#indata = inramp.loadimage(infile)
#self.data = indata.data
#self.hdr = ??indata.header
#self.hdr0 = ??indata.header
if len(self.data.shape)==4:
print('WARNING: Input data cube has 4 dimensions.')
print('Extracting the first integration only and continuing.')
self.data = self.data[0,:,:,:]
elif len(self.data.shape) < 3:
print 'ERROR: data cube has less than 3 dimensions!!! Quitting!'
sys.exit(0)
Ngroups = self.data.shape[0]
if self.verbose:
print 'Ngroups ',Ngroups
#outfilebasename=re.sub('\.fits$','',outfilebasename)
stepsizex = self.stepsizex
stepsizey = self.stepsizey
if stepsizex == None: stepsizex = self.boxsizex
if stepsizey == None: stepsizey = self.boxsizey
halfstepsizex = int(0.5*stepsizex)
halfstepsizey = int(0.5*stepsizey)
# get the xy limits.
if (self.xmin==None): xmin=0
else: xmin=self.xmin
if (self.xmax==None): xmax=self.data.shape[2]
else: xmax=self.xmax
if (self.ymin==None): ymin=0
else: ymin=self.ymin
if (self.ymax==None): ymax=self.data.shape[1]
else: ymax=self.ymax
if self.verbose:
print 'xmin, xmax, ymin, ymax:',xmin,xmax,ymin,ymax
sigmacut = calcaverageclass()
diffim = self.data[1:Ngroups,:,:]- self.data[0:Ngroups-1,:,:]
# define output matrices
im_rdnoise = scipy.zeros((self.data.shape[1],self.data.shape[2]), dtype=float)
im_rdnoise_err = scipy.zeros(im_rdnoise.shape, dtype=float)
im_Nused = scipy.zeros(im_rdnoise.shape, dtype=int)
im_Pskipped = scipy.zeros(im_rdnoise.shape, dtype=float)
# load mask
if self.bpm != None:
print 'loading ',self.bpm
print('BPM read in via astropy for the moment, to avoid MaskModel bug!!!!!')
#self.bpmdata = pyfits.getdata(self.bpm, 0).astype('int_')
with fits.open(self.bpm) as bpmhdu:
self.bpmdata = bpmhdu[1].data
#bpminstance = MaskModel(self.bpm)
#self.bpmdata = bpminstance.dq
if self.bpmdata.shape != (2048,2048):
print("WARNING! BPM shape is incorrect or was improperly read!")
sys.exit()
#print self.bpmdata
mask = scipy.where(scipy.logical_and(self.bpmdata,self.bpmval)>0 ,True,False)
# mask the overscan
mask[0:4,:]=1
mask[2044:2048,:]=1
mask[:,0:4]=1
mask[:,2044:2048]=1
#print mask
else:
mask = scipy.zeros((self.data.shape[1],self.data.shape[2]),dtype=np.uint16)
#load epoxy mask if needed
detector = self.hdr0['DETECTOR']
if detector in ['NRCA1','NRCA3','NRCA4','NRCALONG','NRCB1','NRCB4']:
epoxymask = self.load_epoxy_mask(detector)
#invert the epoxy mask if requested
if self.invert_epoxy:
print("Inverting the epoxy mask to work within the epoxy void region.")
epoxymask = epoxymask - 1
epoxymask[epoxymask == -1] = 1
#add the epoxy mask to the bpm
mask = scipy.logical_or(mask,epoxymask)
#create an inverted epoxy mask for later when filling in readnoise values
inv_epoxymask = epoxymask.astype(bool)
inv_epoxymask = np.invert(inv_epoxymask)
x=xmin + halfstepsizex
OneoverSQR2 = 1.0/math.sqrt(2)
xtransitions = [512,2*512,3*512]
if self.gmin==None:
gmin=0
else:
gmin=self.gmin
if self.gmax==None:
gmax=diffim.shape[0]
else:
gmax=self.gmax
if gmax>diffim.shape[0]:
gmax=diffim.shape[0]
#lists of starting and stopping indexes for the filling later
xfills = []
xfille = []
yfills = []
yfille = []
grange = xrange(gmin,gmax)
while x<xmax:
# define the x1,x2 of the box for stats
x1 = x-int(0.5*self.boxsizex)
if x1<0: x1=0
if self.forcexylimits and (x1<xmin): x1=xmin
x2 = int(x+max(1,0.5*self.boxsizex))
if x2>self.data.shape[2]: x2=self.data.shape[2]
if self.forcexylimits and (x2>xmax): x2=xmax
# make sure that the box contains only data from the same amp!!!
for xtransition in xtransitions:
if x>=xtransition and x1<xtransition:x1=xtransition
if x< xtransition and x2>=xtransition:x2=xtransition
print '!!!x',x,0.5*self.boxsizex,x1,x2
y=ymin + halfstepsizey
while y<ymax:
# define the y1,y2 of the box for stats
y1 = y-int(0.5*self.boxsizey)
if y1<0: y1=0
if self.forcexylimits and (y1<ymin): y1=ymin
y2 = int(y+max(1,0.5*self.boxsizey))
if y2>self.data.shape[2]: y2=self.data.shape[1]
if self.forcexylimits and (y2>ymax): y2=ymax
if (x % 64) == 0 and (y == 0):
print '(x,y)=(%d,%d) box:%d:%d,%d:%d' % (x,y,x1,x2,y1,y2)
stdevs = []
Nused = []
Pskipped = []
for g in grange:
sigmacut.calcaverage_sigmacutloop(diffim[g,y1:y2,x1:x2],mask=mask[y1:y2,x1:x2],Nsigma=3.0,verbose=0)
if self.verbose:
print 'x:%d y:%d g:%d' % (x,y,g),sigmacut.__str__()
if sigmacut.converged and sigmacut.Nused>self.Npixmin and 100.0*sigmacut.Nskipped/(sigmacut.Nused+sigmacut.Nskipped)<self.Pclipmax:
stdevs.append(sigmacut.stdev)
Nused.append(sigmacut.Nused)
Pskipped.append(100.0*sigmacut.Nskipped/(sigmacut.Nused+sigmacut.Nskipped))
#if len(stdevs)>0:
if len(stdevs)>1:
sigmacut.calcaverage_sigmacutloop(np.array(stdevs),Nsigma=3.0,verbose=0)
if sigmacut.converged:
im_rdnoise[y,x]=sigmacut.mean*OneoverSQR2
im_rdnoise_err[y,x]=sigmacut.mean_err*OneoverSQR2
if self.verbose:
print 'x:%d y:%d average' % (x,y),sigmacut.__str__()
im_Nused[y,x]=scipy.median(Nused)
im_Pskipped[y,x]=scipy.median(Pskipped)
elif len(stdevs) == 1:
im_rdnoise[y,x]=sigmacut.stdev*OneoverSQR2
im_rdnoise_err[y,x]=sigmacut.stdev_err*OneoverSQR2
im_Nused[y,x]=1
im_Pskipped[y,x]=0.0
if self.fill:
xx1=x-halfstepsizex
for xtransition in xtransitions:
if x-stepsizex<xtransition and x>=xtransition:xx1=xtransition
if x-stepsizex<0:xx1=0
if xx1<x1:xx1=x1
if xx1<0:xx1=0
xx2=x+max(1,halfstepsizex)
for xtransition in xtransitions:
if x+stepsizex>=xtransition and x<xtransition:
xx2=xtransition
if x+stepsizex>self.data.shape[2]: xx2=self.data.shape[2]
if xx2>x2:xx2=x2
if xx2>self.data.shape[2]: xx2=self.data.shape[2]
yy1=y-halfstepsizey
if y-stepsizey<0:yy1=0
if yy1<y1:yy1=y1
if yy1<0:yy1=0
yy2=y+max(1,halfstepsizey)
if y+stepsizey>self.data.shape[1]: yy2=self.data.shape[1]
if yy2>y2:yy2=y2
if yy2>self.data.shape[1]: yy2=self.data.shape[1]
#save the x and y coordinates for filling in missing data later
xfills.append(xx1)
xfille.append(xx2)
yfills.append(yy1)
yfille.append(yy2)
if len(stdevs)>0:
#submask = inv_epoxymask[yy1:yy2,xx1:xx2]
im_rdnoise[yy1:yy2,xx1:xx2]=im_rdnoise[y,x]#*submask
im_rdnoise_err[yy1:yy2,xx1:xx2]=im_rdnoise_err[y,x]#*submask
im_Nused[yy1:yy2,xx1:xx2]=im_Nused[y,x]
im_Pskipped[yy1:yy2,xx1:xx2]=im_Pskipped[y,x]
#####
y+=stepsizey
x+=stepsizex
#fill in the gaps in the map (i.e. the partial boxes that contain less than the minimum number
#of pixels needed for calculating the readnoise
#save pre-filled image
#h0=fits.PrimaryHDU(im_rdnoise)
#hhlist = fits.HDUList([h0])
#tmpfile = 'before_filling_'+infile
#if os.path.isfile(tmpfile):
# os.rename(tmpfile,'outside_'+tmpfile)
#hhlist.writeto(tmpfile)
#goo = im_rdnoise > 0
#print(np.min(im_rdnoise[goo]),np.max(im_rdnoise[goo]))
#print("BEFORE FILLING EDGES, FILE SAVED TO before_filling*fits")
for iter in xrange(self.fill_iterations):
print('BEGINNING FILL ITERATION {}'.format(iter))
im_rdnoise = self.fill_empty_regions(im_rdnoise,xfills,xfille,yfills,yfille)
im_rdnoise_err = self.fill_empty_regions(im_rdnoise_err,xfills,xfille,yfills,yfille)
#print("AFTER FILLING EDGES, FILE SAVED TO after_filling*fits")
#h0=fits.PrimaryHDU(im_rdnoise)
#hhlist = fits.HDUList([h0])
#tmpfile = 'after_filling_'+infile
#if os.path.isfile(tmpfile):
# os.rename(tmpfile,'outside_'+tmpfile)
#hhlist.writeto(tmpfile)
#goo = im_rdnoise > 0
#print(np.min(im_rdnoise[goo]),np.max(im_rdnoise[goo]))
#now mask out the pixels in the epoxymask
#print("why is this not being done????")
#l0=fits.PrimaryHDU(epoxymask)
#ll=fits.HDUList([l0])
#ll.writeto('test.fits',clobber=True)
if detector in ['NRCA1','NRCA3','NRCA4','NRCALONG','NRCB1','NRCB4']:
im_rdnoise[epoxymask.astype(bool)] = 0.
im_rdnoise_err[epoxymask.astype(bool)] = 0.
#l0=fits.PrimaryHDU(im_rdnoise)
#ll=fits.HDUList([l0])
#ll.writeto('test2.fits',clobber=True)
#stop
#save output file
outfilename = self.save_readnoise_file(im_rdnoise,infile,outfilebasename)
#fits format checks
check_ssb = fits.open(outfilename)
print(check_ssb.info())
print(check_ssb['SCI'].data)
print(check_ssb['SCI'].data[500,500])
print(im_rdnoise[500,500])
#redcat team checks
subprocess.call(['fitsverify',outfilename])
