def test_nan(make_rampmodel, make_darkmodel):
'''Verify that when a dark has NaNs, these are correctly assumed as zero and the PIXELDQ is set properly'''
# size of integration
nints = 1
ngroups = 10
xsize = 200
ysize = 200
# create raw input data for step
dm_ramp = make_rampmodel(nints, ngroups, ysize, xsize)
dm_ramp.meta.exposure.nframes = 1
dm_ramp.meta.exposure.groupgap = 0
# populate data array of science cube
for i in range(0, ngroups - 1):
dm_ramp.data[0, i, :, :] = i
# create dark reference file model with more frames than science data
refgroups = 15
dark = make_darkmodel(refgroups, ysize, xsize)
# populate data array of reference file
for i in range(0, refgroups - 1):
dark.data[0, i] = i * 0.1
# set NaN in dark file
dark.data[0, 5, 100, 100] = np.nan
# apply correction
outfile, avg_dark = darkcorr(dm_ramp, dark)
# test that the NaN dark reference pixel was set to 0 (nothing subtracted)
assert outfile.data[0, 5, 100, 100] == 5.0
def test_frame_avg(make_rampmodel, make_darkmodel):
'''Check that if NFRAMES>1 or GROUPGAP>0, the frame-averaged dark data are
subtracted group-by-group from science data groups and the ERR arrays are not modified'''
# size of integration
nints = 1
ngroups = 5
xsize = 1032
ysize = 1024
# create raw input data for step
dm_ramp = make_rampmodel(nints, ngroups, ysize, xsize)
dm_ramp.meta.exposure.nframes = 4
dm_ramp.meta.exposure.groupgap = 0
# populate data array of science cube
for i in range(0, ngroups - 1):
dm_ramp.data[:, i] = i + 1
# create dark reference file model
refgroups = 20 # This needs to be 20 groups for the calculations to work
dark = make_darkmodel(refgroups, ysize, xsize)
# populate data array of reference file
for i in range(0, refgroups - 1):
dark.data[0, i] = i * 0.1
# apply correction
outfile, avg_dark = darkcorr(dm_ramp, dark)
# dark frames should be averaged in groups of 4 frames
# this will result in average values of 0.15, 0.55, 0.95, and 1.35
# these values are then subtracted from frame values of 1, 2, 3 and 4
assert outfile.data[0, 0, 500, 500] == pytest.approx(0.85)
assert outfile.data[0, 1, 500, 500] == pytest.approx(1.45)
assert outfile.data[0, 2, 500, 500] == pytest.approx(2.05)
assert outfile.data[0, 3, 500, 500] == pytest.approx(2.65)
# check that the error array is not modified.
np.testing.assert_array_equal(outfile.err[:, :], 0)
def test_dq_combine(make_rampmodel, make_darkmodel):
'''Verify that the DQ array of the dark is correctly combined with the PIXELDQ array of the science data.'''
# size of integration
nints = 1
ngroups = 5
xsize = 200
ysize = 200
# create raw input data for step
dm_ramp = make_rampmodel(nints, ngroups, ysize, xsize)
dm_ramp.meta.exposure.nframes = 1
dm_ramp.meta.exposure.groupgap = 0
# populate data array of science cube
for i in range(1, ngroups - 1):
dm_ramp.data[0, i, :, :] = i
# create dark reference file model with more frames than science data
refgroups = 7
dark = make_darkmodel(refgroups, ysize, xsize)
jump_det = dqflags.pixel['JUMP_DET']
saturated = dqflags.pixel['SATURATED']
do_not_use = dqflags.pixel['DO_NOT_USE']
# populate dq flags of sci pixeldq and reference dq
dm_ramp.pixeldq[50, 50] = jump_det
dm_ramp.pixeldq[50, 51] = saturated
dark.dq[0, 0, 50, 50] = do_not_use
dark.dq[0, 0, 50, 51] = do_not_use
# run correction step
outfile, avg_dark = darkcorr(dm_ramp, dark)
# check that dq flags were correctly added
assert outfile.pixeldq[50, 50] == np.bitwise_or(jump_det, do_not_use)
assert outfile.pixeldq[50, 51] == np.bitwise_or(saturated, do_not_use)
def test_sub_by_frame(make_rampmodel, make_darkmodel):
'''Check that if NFRAMES=1 and GROUPGAP=0 for the science data, the dark reference data are
directly subtracted frame by frame'''
# size of integration
nints = 1
ngroups = 10
xsize = 200
ysize = 200
# create raw input data for step
dm_ramp = make_rampmodel(nints, ngroups, ysize, xsize)
dm_ramp.meta.exposure.nframes = 1
dm_ramp.meta.exposure.groupgap = 0
# populate data array of science cube
for i in range(0, ngroups - 1):
dm_ramp.data[0, i] = i
# create dark reference file model with more frames than science data
refgroups = 15
dark = make_darkmodel(refgroups, ysize, xsize)
# populate data array of reference file
for i in range(0, refgroups - 1):
dark.data[0, i] = i * 0.1
# apply correction
outfile, avg_dark = darkcorr(dm_ramp, dark)
# remove the single dimension at start of file (1, 30, 1032, 1024) so comparison in assert works
outdata = np.squeeze(outfile.data)
# check that the dark file is subtracted frame by frame from the science data
diff = dm_ramp.data[0] - dark.data[0, :ngroups]
# test that the output data file is equal to the difference found when subtracting ref file from sci file
np.testing.assert_array_equal(
outdata, diff, err_msg='dark file should be subtracted from sci file ')
def test_more_sci_frames(make_rampmodel, make_darkmodel):
'''Check that data is unchanged if there are more frames in the science
data is than in the dark reference file and verify that when the dark is not applied,
the data is correctly flagged as such'''
# size of integration
nints = 1
ngroups = 7
xsize = 200
ysize = 200
# create raw input data for step
dm_ramp = make_rampmodel(nints, ngroups, ysize, xsize)
dm_ramp.meta.exposure.nframes = 1
dm_ramp.meta.exposure.groupgap = 0
# populate data array of science cube
for i in range(0, ngroups - 1):
dm_ramp.data[0, i] = i
refgroups = 5
# create dark reference file model with fewer frames than science data
dark = make_darkmodel(refgroups, ysize, xsize)
# populate data array of reference file
for i in range(0, refgroups - 1):
dark.data[0, i] = i * 0.1
# apply correction
out_data, avg_data = darkcorr(dm_ramp, dark)
# test that the science data are not changed; input file = output file
np.testing.assert_array_equal(dm_ramp.data, out_data.data)
# get dark correction status from header
# darkstatus = outfile.meta.cal_step.dark_sub
darkstatus = out_data.cal_step
assert darkstatus == 'SKIPPED'
def test_2_int(make_rampmodel, make_darkmodel):
'''Verify the dark correction is done by integration for MIRI observations'''
# size of integration
nints = 2
ngroups = 10
xsize = 200
ysize = 200
# create raw input data for step
dm_ramp = make_rampmodel(nints, ngroups, ysize, xsize)
dm_ramp.meta.exposure.nframes = 1
dm_ramp.meta.exposure.groupgap = 0
# populate data array of science cube
for i in range(0, ngroups - 1):
dm_ramp.data[:, i] = i
# create dark reference file model with more frames than science data
refgroups = 15
dark = make_darkmodel(refgroups, ysize, xsize)
# populate data array of reference file
for i in range(0, refgroups - 1):
dark.data[0, i] = i * 0.1
dark.data[1, i] = i * 0.2
# run correction
outfile, avg_dark = darkcorr(dm_ramp, dark)
# check that the dark file is subtracted frame by frame from the science data
diff = dm_ramp.data[0] - dark.data[0, :ngroups]
diff_int2 = dm_ramp.data[1] - dark.data[1, :ngroups]
# test that the output data file is equal to the difference found when subtracting ref file from sci file
np.testing.assert_array_equal(outfile.data[0], diff)
np.testing.assert_array_equal(outfile.data[1], diff_int2)