def test_fail_invalid_namefunc(namefunc):
"""We do basic checking on namefunc"""
# The contents shouldn't be checked
model = '/dev/null'
with pytest.raises(ValueError) as ve:
xspec.parse_xspec_model_description(model, namefunc)
assert str(ve.value) == 'namefunc must be a callable which takes and returns a string'
def test_fail_missing_parameters():
"""We have less parameters than required"""
model = StringIO("""apec 3 0. 1.e20 C_apec add 0
Abundanc " " 1. 0. 0. 5. 5. -0.001
""")
with pytest.raises(ValueError) as ve:
xspec.parse_xspec_model_description(model)
assert str(ve.value) == 'model=apec missing 2 parameters'
def test_fail_invalid_model_line():
"""There is some check of a model line"""
model = StringIO("""apec 3 0. 1.e20 C_apec add
Abundanc " " 1. 0. 0. 5. 5. -0.001
""")
with pytest.raises(ValueError) as ve:
xspec.parse_xspec_model_description(model)
assert str(ve.value).startswith('Expected: modelname npars ')
def test_fail_periodic_parameter():
"""We may add support for these in the future.
"""
model = StringIO("""apec 1 0. 1.e20 C_apec add 0
Abundanc " " 1. 0. 0. 5. 5. -0.001 P
""")
with pytest.raises(ValueError) as ve:
xspec.parse_xspec_model_description(model)
assert str(ve.value).startswith('Periodic parameters are unsupported; model=apec:\n')
def test_fail_invalid_basic_parameter():
"""There is a check of model types.
It might be better to just skip these.
"""
model = StringIO("""apec 1 0. 1.e20 C_apec sub 0
Abundanc " " 1.
""")
with pytest.raises(ValueError) as ve:
xspec.parse_xspec_model_description(model)
assert str(ve.value).startswith('Expected 6 values after units; model=apec\n')
def test_fail_unknown_model():
"""There is some check of model types
It mmight be better to just skip these.
"""
model = StringIO("""apec 1 0. 1.e20 C_apec sub 0
Abundanc " " 1. 0. 0. 5. 5. -0.001
""")
with pytest.raises(ValueError) as ve:
xspec.parse_xspec_model_description(model)
assert str(ve.value).startswith('Unexpected model type sub in:\n')
def test_create_model_additive():
"""Fake up an additive model"""
model = StringIO("""apec 1 0. 1.e20 C_apec add 0
kT keV 1. 0.008 0.008 64.0 64.0 .01
""")
parsed = xspec.parse_xspec_model_description(model)
converted = xspec.create_xspec_code(parsed)
# Very basic checks of the output
#
python = converted.python.split('\n')
compiled = converted.compiled.split('\n')
assert 'class XSapec(XSAdditiveModel):' in python
assert ' _calc = _models.C_apec' in python
assert " def __init__(self, name='apec'):" in python
assert " self.kT = XSParameter(name, 'kT', 1.0, min=0.008, max=64.0, hard_min=0.008, hard_max=64.0, units='keV')" in python
assert " self.norm = Parameter(name, 'norm', 1.0, min=0.0, max=1e+24, hard_min=0.0, hard_max=1e+24)" in python
assert ' XSAdditiveModel.__init__(self, name, (self.kT,self.norm))' in python
assert converted.compiled.find('\nextern "C" {\n\n}\n') > -1
assert ' XSPECMODELFCT_C_NORM(C_apec, 2),' in compiled
def test_parse_model_dat(path):
"""Can we parse the model.dat file (via string or pathlike)
There's limited checking of the data since we can't
guarantee it won't change significantly over time.
"""
from sherpa.astro.xspec import get_xspath_manager
path = get_xspath_manager()
mfile = os.path.join(path, 'model.dat')
if path:
mfile = pathlib.PurePath(mfile)
parsed = xspec.parse_xspec_model_description(mfile)
assert len(parsed) > 0
# Check we only have Add, Mul, Con, and Acn types
#
mtypes = set()
for model in parsed:
mtypes.add(model.modeltype)
assert mtypes == set(["Add", "Mul", "Con", "Acn"])
def test_warn_parse_repeated_parname(caplog):
"""Check we warn if the parameter name is repeated"""
model = StringIO("""apec 3 0. 1.e20 C_apec add 0
Abundanc " " 1. 0. 0. 5. 5. -0.001
kT keV 1. 0.008 0.008 64.0 64.0 .01
abundanc " " 0. -0.999 -0.999 10. 10. -0.01
""")
assert len(caplog.records) == 0
xspec.parse_xspec_model_description(model)
assert len(caplog.records) == 1
lname, lvl, msg = caplog.record_tuples[0]
assert lname == 'sherpa.astro.utils.xspec'
assert lvl == logging.WARNING
assert msg == 'model=apec re-uses parameter name abundanc'
def test_fail_toomany_parameters():
"""We have more parameters than required
This actually errors out as it tries to parse the extra
parameter as a model (that is, this is only checked for
indirectly).
"""
model = StringIO("""apec 1 0. 1.e20 C_apec add 0
Abundanc " " 1. 0. 0. 5. 5. -0.001
Redshift " " 0. -0.999 -0.999 10. 10. -0.01
""")
# As we don't have an explicit check the actual error depends on the
# next line so it's not worth checking the message.
#
with pytest.raises(ValueError):
xspec.parse_xspec_model_description(model)
def test_convert_model_dat():
"""Can we convert the model.dat file?
This is more just an existence test (i.e. does it work) with
limited checking of the output. Checking the warnings that
are created by parsing/converting the XSPEC model.dat is
tricky to do reliably as it depends on the XSPEC version,
so we do not check them (there are specific tests above).
"""
from sherpa.astro.xspec import get_xspath_manager
path = get_xspath_manager()
mfile = os.path.join(path, 'model.dat')
# We do not check the return value, just that it runs
parsed = xspec.parse_xspec_model_description(mfile)
xspec.create_xspec_code(parsed)
def test_skip_repeated_model(caplog):
"""Technically we could use the same model but with different
parameters but there was a bug in a version of XSPEC where
the same model was used incorrectly, so check we handle this.
We only skip it when we come to creating the code.
"""
model = StringIO("""apec 1 0. 1.e20 C_foo mul 0
Abundanc " " 1. 0. 0. 5. 5. -0.001
fred 0 0. 1.e20 C_foo mul 0
""")
parsed = xspec.parse_xspec_model_description(model)
assert len(parsed) == 2
assert len(caplog.records) == 0
# As an extra benefit as we skip both models we can check the
# 'no valid models' error.
#
with pytest.raises(ValueError) as ve:
xspec.create_xspec_code(parsed)
assert str(ve.value) == 'No supported models were found!'
assert len(caplog.records) == 2
lname, lvl, msg = caplog.record_tuples[0]
assert lname == 'sherpa.astro.utils.xspec'
assert lvl == logging.WARNING
assert msg == 'Skipping model apec as it calls foo which is used by 2 different models'
lname, lvl, msg = caplog.record_tuples[1]
assert lname == 'sherpa.astro.utils.xspec'
assert lvl == logging.WARNING
assert msg == 'Skipping model fred as it calls foo which is used by 2 different models'
def test_create_model_convolution():
"""Fake up a convolution model"""
model = StringIO("""rgsxsrc 1 0. 1.e20 rgsxsrc con 0
order " " -1. -3. -3. -1. -1. -1
""")
parsed = xspec.parse_xspec_model_description(model)
converted = xspec.create_xspec_code(parsed)
# Very basic checks of the output
#
python = converted.python.split('\n')
compiled = converted.compiled.split('\n')
assert 'class XSrgsxsrc(XSConvolutionKernel):' in python
assert ' _calc = _models.rgsxsrc' in python
assert " def __init__(self, name='rgsxsrc'):" in python
assert " self.order = XSParameter(name, 'order', -1.0, min=-3.0, max=-1.0, hard_min=-3.0, hard_max=-1.0, frozen=True)" in python
assert ' XSConvolutionKernel.__init__(self, name, (self.order,))' in python
assert ' void rgsxsrc_(float* ear, int* ne, float* param, int* ifl, float* photar, float* photer);' in compiled
assert ' XSPECMODELFCT_CON_F77(rgsxsrc, 1),' in compiled
def test_create_model_multiplicative():
"""Fake up a multplicative model"""
model = StringIO("""abcd 1 0. 1.e20 foos mul 0
nH cm^-3 1.0 1.e-6 1.e-5 1.e19 1.e20 -0.01
""")
parsed = xspec.parse_xspec_model_description(model)
converted = xspec.create_xspec_code(parsed)
# Very basic checks of the output
#
python = converted.python.split('\n')
compiled = converted.compiled.split('\n')
assert 'class XSabcd(XSMultiplicativeModel):' in python
assert ' _calc = _models.foos' in python
assert " def __init__(self, name='abcd'):" in python
assert " self.nH = XSParameter(name, 'nH', 1.0, min=1e-05, max=1e+19, hard_min=1e-06, hard_max=1e+20, frozen=True, units='cm^-3')" in python
assert ' XSMultiplicativeModel.__init__(self, name, (self.nH,))' in python
assert ' void foos_(float* ear, int* ne, float* param, int* ifl, float* photar, float* photer);' in compiled
assert ' XSPECMODELFCT(foos, 1),' in compiled
except TypeError:
continue
if name not in seen:
print(f"Unable to find {name} in model.dat")
if __name__ == "__main__":
# Too lazy to use python argument module
hard = True
argv = []
for arg in sys.argv[1:]:
if arg == '--nohard':
hard = False
else:
argv.append(arg)
if len(argv) != 1:
sys.stderr.write(f"Usage: {sys.argv[0]} infile\n")
sys.stderr.write(" --nohard do not check hard ranges\n\n")
sys.exit(1)
# This errors out in case of significant model-support issues
# (e.g. a periodic model parameter) but can also just be an
# issue with a poorly-specified interchange format (the model.dat
# file) which may just need changes to this routine.
#
models = parse_xspec_model_description(argv[0])
compare_xspec_models(models, hard=hard)
"""
for amodel in models:
if amodel.name == modelname:
model = amodel
break
else:
raise ValueError(f"Unknown model name: {modelname}")
code = xspec.create_xspec_code([model])
print("# C++ code for sherpa/astro/xspec/src/_xspec.cc\n")
print(code.compiled)
print("\n# Python code for sherpa/astro/xspec/__init__.py\n")
print(code.python)
if __name__ == "__main__":
if len(sys.argv) != 3:
sys.stderr.write(f"Usage: {sys.argv[0]} infile model\n")
sys.exit(1)
# This errors out in case of significant model-support issues
# (e.g. a periodic model parameter) but can also just be an
# issue with a poorly-specified interchange format (the model.dat
# file) which may just need changes to this routine.
#
models = xspec.parse_xspec_model_description(sys.argv[1])
create_xspec_model(models, sys.argv[2])
def test_parse_multiplicative_model():
"""Can we parse a Multiplicative model"""
model = StringIO("""zdust 4 0. 1.e20 mszdst mul 0 1
$method " " 1 1 1 3 3 -0.01
E_BmV " " 0.1 0.0 0.0 100. 100. 0.01
*redshift " " 0.0
break Hz 2.42E17 1.E10 1.E15 1.E19 1.E25 1E10
""")
parsed = xspec.parse_xspec_model_description(model)
assert len(parsed) == 1
model = parsed[0]
assert model.name == 'zdust'
assert model.clname == 'XSzdust'
assert model.modeltype == 'Mul'
assert model.flags == pytest.approx([0, 1])
assert model.elo == pytest.approx(0)
assert model.ehi == pytest.approx(1e20)
assert model.initString is None
assert model.funcname == 'mszdst'
assert model.language == 'Fortran - single precision'
assert len(model.pars) == 4
assert model.pars[0].paramtype == 'Switch'
assert model.pars[0].name == 'method'
assert model.pars[0].units is None
assert model.pars[0].default == pytest.approx(1)
assert model.pars[0].softmin == pytest.approx(1)
assert model.pars[0].hardmin == pytest.approx(1)
assert model.pars[0].softmax == pytest.approx(3)
assert model.pars[0].hardmax == pytest.approx(3)
assert model.pars[1].paramtype == 'Basic'
assert model.pars[1].name == 'E_BmV'
assert not model.pars[1].frozen
assert model.pars[1].units is None
assert model.pars[1].default == pytest.approx(0.1)
assert model.pars[1].softmin == pytest.approx(0.0)
assert model.pars[1].hardmin == pytest.approx(0.0)
assert model.pars[1].softmax == pytest.approx(100.0)
assert model.pars[1].hardmax == pytest.approx(100.0)
assert model.pars[2].paramtype == 'Scale'
assert model.pars[2].name == 'redshift'
assert model.pars[2].units is None
assert model.pars[2].default == pytest.approx(0)
assert model.pars[2].softmin is None
assert model.pars[2].hardmin is None
assert model.pars[2].softmax is None
assert model.pars[2].hardmax is None
assert model.pars[3].paramtype == 'Basic'
assert model.pars[3].name == 'break_'
assert not model.pars[3].frozen
assert model.pars[3].units == 'Hz'
assert model.pars[3].default == pytest.approx(2.42e17)
assert model.pars[3].softmin == pytest.approx(1e15)
assert model.pars[3].hardmin == pytest.approx(1e10)
assert model.pars[3].softmax == pytest.approx(1e19)
assert model.pars[3].hardmax == pytest.approx(1e25)
def test_parse_additive_model():
"""Can we parse an Additive model"""
model = StringIO("""apec 3 0. 1.e20 C_apec add 0
kT keV 1. 0.008 0.008 64.0 64.0 .01
Abundanc " " 1. 0. 0. 5. 5. -0.001
Redshift " " 0. -0.999 -0.999 10. 10. -0.01
""")
parsed = xspec.parse_xspec_model_description(model)
assert len(parsed) == 1
model = parsed[0]
assert model.name == 'apec'
assert model.clname == 'XSapec'
assert model.modeltype == 'Add'
assert model.flags == pytest.approx([0])
assert model.elo == pytest.approx(0)
assert model.ehi == pytest.approx(1e20)
assert model.initString is None
assert model.funcname == 'apec'
assert model.language == 'C++ style'
assert len(model.pars) == 4
for par in model.pars:
assert par.paramtype == 'Basic'
assert model.pars[0].name == 'kT'
assert not model.pars[0].frozen
assert model.pars[0].units == 'keV'
assert model.pars[0].default == pytest.approx(1)
assert model.pars[0].softmin == pytest.approx(0.008)
assert model.pars[0].hardmin == pytest.approx(0.008)
assert model.pars[0].softmax == pytest.approx(64.0)
assert model.pars[0].hardmax == pytest.approx(64.0)
assert model.pars[1].name == 'Abundanc'
assert model.pars[1].frozen
assert model.pars[1].units is None
assert model.pars[1].default == pytest.approx(1)
assert model.pars[1].softmin == pytest.approx(0.0)
assert model.pars[1].hardmin == pytest.approx(0.0)
assert model.pars[1].softmax == pytest.approx(5.0)
assert model.pars[1].hardmax == pytest.approx(5.0)
assert model.pars[2].name == 'Redshift'
assert model.pars[2].frozen
assert model.pars[2].units is None
assert model.pars[2].default == pytest.approx(0)
assert model.pars[2].softmin == pytest.approx(-0.999)
assert model.pars[2].hardmin == pytest.approx(-0.999)
assert model.pars[2].softmax == pytest.approx(10.0)
assert model.pars[2].hardmax == pytest.approx(10.0)
assert model.pars[3].name == 'norm'
assert not model.pars[3].frozen
assert model.pars[3].units is None
assert model.pars[3].default == pytest.approx(1)
assert model.pars[3].softmin == pytest.approx(0)
assert model.pars[3].hardmin == pytest.approx(0)
assert model.pars[3].softmax == pytest.approx(1e24)
assert model.pars[3].hardmax == pytest.approx(1e24)