def test_run_outlier_only(mk_tmp_dirs):
"""Test a basic run"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
file_copy(
path.join(DATAPATH, 'two_member_spec3_asn.json'),
tmp_data_path
)
asn_path = path.join(tmp_data_path, 'two_member_spec3_asn.json')
with open(asn_path) as fp:
asn = load_asn(fp)
for product in asn['products']:
for member in product['members']:
file_copy(
path.join(DATAPATH, 'level2b_twoslit', member['expname']),
tmp_data_path
)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec3_default.cfg'),
asn_path,
'--steps.mrs_imatch.skip=true',
'--steps.outlier_detection.save_results=true',
'--steps.resample_spec.skip=true',
'--steps.cube_build.skip=true',
'--steps.extract_1d.skip=true',
]
Step.from_cmdline(args)
assert False
def test_run_full(mk_tmp_dirs):
"""Test a full run"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
file_copy(
path.join(DATAPATH, 'two_member_spec3_asn.json'),
tmp_data_path
)
asn_path = path.join(tmp_data_path, 'two_member_spec3_asn.json')
with open(asn_path) as fp:
asn = load_asn(fp)
for product in asn['products']:
for member in product['members']:
file_copy(
path.join(DATAPATH, 'level2b_twoslit', member['expname']),
tmp_data_path
)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec3_default.cfg'),
asn_path,
]
Step.from_cmdline(args)
assert False
def test_run_full(mk_tmp_dirs):
"""Test a full run"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
file_copy(
path.join(DATAPATH, 'two_member_spec3_asn.json'),
tmp_data_path
)
asn_path = path.join(tmp_data_path, 'two_member_spec3_asn.json')
with open(asn_path) as fp:
asn = load_asn(fp)
for product in asn['products']:
for member in product['members']:
file_copy(
path.join(DATAPATH, 'level2b_twoslit', member['expname']),
tmp_data_path
)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec3_default.cfg'),
asn_path,
]
Step.from_cmdline(args)
assert False
def test_run_outlier_only(mk_tmp_dirs):
"""Test a basic run"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
file_copy(path.join(DATAPATH, 'two_member_spec3_asn.json'), tmp_data_path)
asn_path = path.join(tmp_data_path, 'two_member_spec3_asn.json')
with open(asn_path) as fp:
asn = load_asn(fp)
for product in asn['products']:
for member in product['members']:
file_copy(
path.join(DATAPATH, 'level2b_twoslit', member['expname']),
tmp_data_path)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec3_default.cfg'),
asn_path,
'--steps.mrs_imatch.skip=true',
'--steps.outlier_detection.save_results=true',
'--steps.resample_spec.skip=true',
'--steps.cube_build.skip=true',
'--steps.extract_1d.skip=true',
]
Step.from_cmdline(args)
assert False
def test_run_msaflagging(mk_tmp_dirs, caplog):
"""Test msa flagging operation"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy msa config files from DATAPATH to
# current working directory
file_copy(path.join(DATAPATH, 'jw95065006001_0_msa_twoslit.fits'), '.')
asn_path = update_asn_basedir(path.join(
DATAPATH, 'mos_udf_g235m_twoslit_spec2_asn.json'),
root=path.join(DATAPATH, 'level2a_twoslit'))
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec2_basic.cfg'), asn_path,
'--steps.msa_flagging.skip=false'
]
Step.from_cmdline(args)
assert 'Step msa_flagging running with args' in caplog.text
assert 'Step msa_flagging done' in caplog.text
with open(asn_path) as fp:
asn = load_asn(fp)
for product in asn['products']:
prod_name = product['name'] + '_cal.fits'
assert path.isfile(prod_name)
def install_my_kernel_spec(user=False):
with TemporaryDirectory() as td:
os.chmod(td, 0o755) # Starts off as 700, not user readable
path_of_file = dirname(abspath(__file__))
file_copy(path_of_file + "/mikrokosmoskernel/kernel.js", td)
file_copy(path_of_file + "/mikrokosmoskernel/kernel.json", td)
print('Installing Jupyter kernel spec')
install_kernel_spec(td, 'IMikrokosmos', user=user, replace=True)
def upgrade_thread():
global in_progress
global file_count, manifest_checksum, manifest_size, manifest_url, version
try:
common.log("Firmware upgrade start")
gc.collect()
manifest_pathname = "manifest"
shutil.path_remove(manifest_pathname)
# TODO: Remove all "*_new" and "*_old"
aiko.web_client.http_get_file(manifest_url, manifest_pathname)
# TODO: Verify "manifest_pathname" actual file size versus "manifest_size"
# TODO: Verify "manifest_pathname" checksum
top_level_files = []
url_prefix = manifest_url.rpartition("/")[0]
with open(manifest_pathname, "r") as manifest_file:
file_index = 0
for line in manifest_file.readlines():
file_index += 1
file_checksum, file_size, filepath = line.split()
url_suffix = filepath.partition("/")[-1]
file_url = "/".join([url_prefix, url_suffix])
pathname = url_suffix.partition("/")
if not pathname[0] in top_level_files:
top_level_files.append(pathname[0])
pathname = "".join([pathname[0] + "_new"] + list(pathname[1:]))
print(file_url + " --> " + pathname)
common.log("Firmware get ... %d of %d" % (file_index, file_count))
aiko.web_client.http_get_file(file_url, pathname)
# TODO: Verify actual file size versus size stated in the "manifest"
# TODO: Verify actual file checksum
shutil.path_remove(manifest_pathname)
shutil.file_copy("configuration/net.py", "configuration_new/net.py")
shutil.file_copy("configuration/keys.db", "configuration_new/keys.db")
common.log("Firmware install")
for file in top_level_files:
try:
print("Rename %s to %s" % (file + "_new", file))
shutil.path_remove(file)
os.rename(file + "_new", file)
except OSError:
print("OSError")
common.log("Firmware upgrade success !")
common.log("Please reboot :)")
except Exception as exception:
common.log("Firmware upgrade failed :(")
import sys
sys.print_exception(exception)
finally:
in_progress = False
version = None
def install_my_kernel_spec(user=True):
with TemporaryDirectory() as td:
os.chmod(td, 0o755) # Starts off as 700, not user readable
with open(os.path.join(td, 'kernel.json'), 'w') as f:
json.dump(kernel_json, f, sort_keys=True)
path_of_file = dirname( abspath(__file__) ) + "/resources/"
file_copy(path_of_file + "logo-32x32.png", td )
file_copy(path_of_file + "logo-64x64.png", td )
print('Installing IPython kernel spec')
install_kernel_spec(td, 'Singular', user=user, replace=True)
def install_my_kernel_spec(user=True):
with TemporaryDirectory() as td:
os.chmod(td, 0o755) # Starts off as 700, not user readable
with open(os.path.join(td, 'kernel.json'), 'w') as f:
json.dump(kernel_json, f, sort_keys=True)
# TODO: Copy resources once they're specified
path_of_file = dirname(abspath(__file__)) + "/resources/"
file_copy(path_of_file + "logo-32x32.png", td)
file_copy(path_of_file + "logo-64x64.png", td)
print('Installing Jupyter kernel spec from')
install_kernel_spec(td, 'gap', user=user, replace=True)
def test_run_extract_1d_resample_mock(mk_tmp_dirs):
"""Test only the extraction step. Should produce nothing
because extraction requires resampling
"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
file_copy(
path.join(DATAPATH, 'two_member_spec3_asn.json'),
tmp_data_path
)
asn_path = path.join(tmp_data_path, 'two_member_spec3_asn.json')
with open(asn_path) as fp:
asn = load_asn(fp)
for product in asn['products']:
for member in product['members']:
file_copy(
path.join(DATAPATH, 'level2b_twoslit', member['expname']),
tmp_data_path
)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec3_mock.cfg'),
asn_path,
'--steps.mrs_imatch.skip=true',
'--steps.outlier_detection.skip=true',
'--steps.cube_build.skip=true',
]
Step.from_cmdline(args)
# Though the calibration is not run, the conversion to
# source base has occured. Check
with open(asn_path) as fd:
asn = load_asn(fd)
product_name_template = asn['products'][0]['name']
product_name_glob = product_name_template.format(
source_id='s0000[14]',
) + '_cal.fits'
assert len(glob(product_name_glob)) == 2
product_name_glob = product_name_template.format(
source_id='s0000[14]',
) + '_s2d.fits'
assert len(glob(product_name_glob)) == 2
product_name_glob = product_name_template.format(
source_id='s0000[14]',
) + '_x1d.fits'
assert len(glob(product_name_glob)) == 2
def test_run_extract_1d_resample_mock(mk_tmp_dirs):
"""Test only the extraction step. Should produce nothing
because extraction requires resampling
"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
file_copy(
path.join(DATAPATH, 'two_member_spec3_asn.json'),
tmp_data_path
)
asn_path = path.join(tmp_data_path, 'two_member_spec3_asn.json')
with open(asn_path) as fp:
asn = load_asn(fp)
for product in asn['products']:
for member in product['members']:
file_copy(
path.join(DATAPATH, 'level2b_twoslit', member['expname']),
tmp_data_path
)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec3_mock.cfg'),
asn_path,
'--steps.mrs_imatch.skip=true',
'--steps.outlier_detection.skip=true',
'--steps.cube_build.skip=true',
]
Step.from_cmdline(args)
# Though the calibration is not run, the conversion to
# source base has occured. Check
with open(asn_path) as fd:
asn = load_asn(fd)
product_name_template = asn['products'][0]['name']
product_name_glob = product_name_template.format(
source_id='s0000[14]',
) + '_cal.fits'
assert len(glob(product_name_glob)) == 2
product_name_glob = product_name_template.format(
source_id='s0000[14]',
) + '_s2d.fits'
assert len(glob(product_name_glob)) == 2
product_name_glob = product_name_template.format(
source_id='s0000[14]',
) + '_x1d.fits'
assert len(glob(product_name_glob)) == 2
def test_msa_missing_nofail(mk_tmp_dirs, caplog):
"""Test MSA missing failure"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
input_file = 'F170LP-G235M_MOS_observation-6-c0e0_001_DN_NRS1_mod.fits'
file_copy(path.join(DATAPATH, 'level2a_twoslit', input_file),
tmp_data_path)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec2_basic.cfg'),
path.join(tmp_data_path, input_file), '--fail_on_exception=false'
]
Step.from_cmdline(args)
assert 'Unable to open MSA FITS file (MSAMETFL)' in caplog.text
def test_msa_missing_skip(mk_tmp_dirs, caplog):
"""Test MSA missing failure"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
input_file = 'F170LP-G235M_MOS_observation-6-c0e0_001_DN_NRS1_mod.fits'
file_copy(path.join(DATAPATH, 'level2a_twoslit', input_file),
tmp_data_path)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec2_basic.cfg'),
path.join(tmp_data_path, input_file), '--steps.assign_wcs.skip=true'
]
Step.from_cmdline(args)
assert 'Aborting remaining processing for this exposure.' in caplog.text
def test_run_nosteps(mk_tmp_dirs):
"""Test where no steps execute"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
file_copy(
path.join(DATAPATH, 'two_member_spec3_asn.json'),
tmp_data_path
)
asn_path = path.join(tmp_data_path, 'two_member_spec3_asn.json')
with open(asn_path) as fp:
asn = load_asn(fp)
for product in asn['products']:
for member in product['members']:
file_copy(
path.join(DATAPATH, 'level2b_twoslit', member['expname']),
tmp_data_path
)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec3_default.cfg'),
asn_path,
'--steps.mrs_imatch.skip=true',
'--steps.outlier_detection.skip=true',
'--steps.resample_spec.skip=true',
'--steps.cube_build.skip=true',
'--steps.extract_1d.skip=true',
]
Step.from_cmdline(args)
# Check for the Source-based cal name.
with open(asn_path) as fp:
asn = load_asn(fp)
product_name_template = asn['products'][0]['name']
product_name_glob = product_name_template.format(
source_id='s0000[14]',
) + '_cal.fits'
assert len(glob(product_name_glob)) == 2
# Check that no other products built
files = glob('*s3d*')
files.extend(glob('*s2d*'))
files.extend(glob('*x1d*'))
assert not files
def test_run_nosteps(mk_tmp_dirs):
"""Test where no steps execute"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
file_copy(
path.join(DATAPATH, 'two_member_spec3_asn.json'),
tmp_data_path
)
asn_path = path.join(tmp_data_path, 'two_member_spec3_asn.json')
with open(asn_path) as fp:
asn = load_asn(fp)
for product in asn['products']:
for member in product['members']:
file_copy(
path.join(DATAPATH, 'level2b_twoslit', member['expname']),
tmp_data_path
)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec3_default.cfg'),
asn_path,
'--steps.mrs_imatch.skip=true',
'--steps.outlier_detection.skip=true',
'--steps.resample_spec.skip=true',
'--steps.cube_build.skip=true',
'--steps.extract_1d.skip=true',
]
Step.from_cmdline(args)
# Check for the Source-based cal name.
with open(asn_path) as fp:
asn = load_asn(fp)
product_name_template = asn['products'][0]['name']
product_name_glob = product_name_template.format(
source_id='s0000[14]',
) + '_cal.fits'
assert len(glob(product_name_glob)) == 2
# Check that no other products built
files = glob('*s3d*')
files.extend(glob('*s2d*'))
files.extend(glob('*x1d*'))
assert not files
def test_run_outlier_only_mock(mk_tmp_dirs):
"""Test a basic run"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
file_copy(
path.join(DATAPATH, 'two_member_spec3_asn.json'),
tmp_data_path
)
asn_path = path.join(tmp_data_path, 'two_member_spec3_asn.json')
with open(asn_path) as fp:
asn = load_asn(fp)
for product in asn['products']:
for member in product['members']:
file_copy(
path.join(DATAPATH, 'level2b_twoslit', member['expname']),
tmp_data_path
)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec3_mock.cfg'),
asn_path,
'--steps.mrs_imatch.skip=true',
'--steps.resample_spec.skip=true',
'--steps.cube_build.skip=true',
'--steps.extract_1d.skip=true',
]
Step.from_cmdline(args)
# Check for the Source-based cal name.
with open(asn_path) as fp:
asn = load_asn(fp)
product_name_template = asn['products'][0]['name']
product_name_glob = product_name_template.format(
source_id='s0000[14]',
) + '_cal.fits'
assert len(glob(product_name_glob)) == 2
# Check for the outlier resutls
product_name_glob = product_name_template.format(
source_id='s0000[14]',
) + '_crj.fits'
assert len(glob(product_name_glob)) == 2
def test_run_outlier_only_mock(mk_tmp_dirs):
"""Test a basic run"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
file_copy(
path.join(DATAPATH, 'two_member_spec3_asn.json'),
tmp_data_path
)
asn_path = path.join(tmp_data_path, 'two_member_spec3_asn.json')
with open(asn_path) as fp:
asn = load_asn(fp)
for product in asn['products']:
for member in product['members']:
file_copy(
path.join(DATAPATH, 'level2b_twoslit', member['expname']),
tmp_data_path
)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec3_mock.cfg'),
asn_path,
'--steps.mrs_imatch.skip=true',
'--steps.resample_spec.skip=true',
'--steps.cube_build.skip=true',
'--steps.extract_1d.skip=true',
]
Step.from_cmdline(args)
# Check for the Source-based cal name.
with open(asn_path) as fp:
asn = load_asn(fp)
product_name_template = asn['products'][0]['name']
product_name_glob = product_name_template.format(
source_id='s0000[14]',
) + '_cal.fits'
assert len(glob(product_name_glob)) == 2
# Check for the outlier resutls
product_name_glob = product_name_template.format(
source_id='s0000[14]',
) + '_crj.fits'
assert len(glob(product_name_glob)) == 2
def _store_output_data(
data_object_path,
data_directory,
output_to_store,
results_by_id,
):
"""Collects all of the simulation to store, and saves it into a directory
whose path will be passed to the storage backend to process.
Parameters
----------
data_object_path: str
The file path to serialize the data object to.
data_directory: str
The path of the directory to store ancillary data in.
output_to_store: BaseStoredData
An object which contains `ProtocolPath`s pointing to the
data to store.
results_by_id: dict of ProtocolPath and any
The results of the protocols which formed the property
estimation workflow.
"""
makedirs(data_directory, exist_ok=True)
for attribute_name in output_to_store.get_attributes(StorageAttribute):
attribute = getattr(output_to_store.__class__, attribute_name)
attribute_value = getattr(output_to_store, attribute_name)
if not isinstance(attribute_value, ProtocolPath):
continue
attribute_value = results_by_id[attribute_value]
if issubclass(attribute.type_hint, FilePath):
file_copy(attribute_value, data_directory)
attribute_value = path.basename(attribute_value)
setattr(output_to_store, attribute_name, attribute_value)
with open(data_object_path, "w") as file:
json.dump(output_to_store, file, cls=TypedJSONEncoder)
def install_my_kernel_spec(user=True):
with TemporaryDirectory() as td:
os.chmod(td, 0o755) # Starts off as 700, not user readable
with open(os.path.join(td, 'kernel.json'), 'w') as f:
json.dump(kernel_json, f, sort_keys=True)
path_of_file = dirname( abspath(__file__) ) + "/resources/"
filenames=[ "Detector.js", "three.js", "kernel.js" ]
filenames_renderer=[ "CanvasRenderer.js", "Projector.js" ]
filenames_control=[ "TrackballControls.js" ]
for i in filenames:
file_copy(path_of_file + i, td )
os.mkdir( td + "renderers", mode=755 )
for i in filenames_renderer:
file_copy(path_of_file + "renderers/" + i, td + "renderers" )
os.mkdir( td + "controls", mode=755 )
for i in filenames_control:
file_copy(path_of_file + "controls/" + i, td + "controls" )
file_copy(path_of_file + "logo-32x32.png", td )
file_copy(path_of_file + "logo-64x64.png", td )
print('Installing IPython kernel spec')
install_kernel_spec(td, 'polymake', user=user, replace=True)
def test_run_msaflagging(mk_tmp_dirs, caplog):
"""Test msa flagging operation"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
file_copy(path.join(DATAPATH, 'jw95065006001_0_msa_twoslit.fits'),
tmp_data_path)
file_copy(path.join(DATAPATH, 'mos_udf_g235m_twoslit_spec2_asn.json'),
tmp_data_path)
asn_path = path.join(tmp_data_path, 'mos_udf_g235m_twoslit_spec2_asn.json')
with open(asn_path) as fp:
asn = load_asn(fp)
for product in asn['products']:
for member in product['members']:
file_copy(
path.join(DATAPATH, 'level2a_twoslit', member['expname']),
tmp_data_path)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec2_basic.cfg'), asn_path,
'--steps.msa_flagging.skip=false'
]
Step.from_cmdline(args)
assert 'Step msa_flagging running with args' in caplog.text
assert 'Step msa_flagging done' in caplog.text
for product in asn['products']:
prod_name = product['name'] + '_cal.fits'
assert path.isfile(prod_name)
def test_msa_missing_nofail(mk_tmp_dirs, caplog):
"""Test MSA missing failure"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
input_file = 'F170LP-G235M_MOS_observation-6-c0e0_001_DN_NRS1_mod.fits'
file_copy(
path.join(
DATAPATH,
'level2a_twoslit',
input_file
),
tmp_data_path
)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec2_basic.cfg'),
path.join(tmp_data_path, input_file),
'--fail_on_exception=false'
]
Step.from_cmdline(args)
assert 'Unable to open MSA FITS file (MSAMETFL)' in caplog.text
def test_msa_missing_skip(mk_tmp_dirs, caplog):
"""Test MSA missing failure"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
input_file = 'F170LP-G235M_MOS_observation-6-c0e0_001_DN_NRS1_mod.fits'
file_copy(
path.join(
DATAPATH,
'level2a_twoslit',
input_file
),
tmp_data_path
)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec2_basic.cfg'),
path.join(tmp_data_path, input_file),
'--steps.assign_wcs.skip=true'
]
Step.from_cmdline(args)
assert 'Aborting remaining processing for this exposure.' in caplog.text
def add_song(file, artist, song_name, date, *tags):
try:
fileDetails = file.split(".")
file_name = fileDetails[0]
type = fileDetails[1].lower()
id = str(uuid4())
if type not in accepted_types:
raise "Unregonized file format!"
# Format params
artist = artist.replace("_", " ")
song_name = artist.replace("_", " ")
# Check date
if match('(\d{2})[./-](\d{2})[./-](\d{4})$', date) is None:
raise Exception("Invalid Date")
# Copy file and rename it to ID.type
file_copy(file, "Storage")
os.rename(f"Storage/{file}", f"Storage/{id}.{type}")
data = {
"ID": id,
"file name": file_name,
"type": type,
"artist": artist,
"song name": song_name,
"date": date,
"tags": tags
}
db.insert(data)
return f"Success! Song id: {id}"
except Exception as error:
return f"Failure! {error}"
def run(self):
step = 0
size = len(self.data)
while step < size:
print '.',
#quito los salto de linea
path = self.data[step].replace('\n', '')
filename = path.split('\\')[-1]
try:
npath = os.path.join(self.folder, filename)
file_copy(path, npath)
self.gauge.step()
self.blist.insert(tk.END, "[%s] %s" %(str(step+1).zfill(3), filename))
except:
time = datetime.datetime.now().strftime("[%d-%m-%Y %H:%M:%S]")
text = "%s - ERROR AL COPIAR: %s \n" %(time, path)
self.log.write(text)
self.blist.insert(tk.END, "[%s] ERROR %s" %(str(step+1).zfill(3), filename))
#time.sleep(10) #duermo el hilo
step += 1
self.is_run = False
def install_my_kernel_spec(user=True):
with TemporaryDirectory() as td:
os.chmod(td, 0o755) # Starts off as 700, not user readable
with open(os.path.join(td, 'kernel.json'), 'w') as f:
json.dump(kernel_json, f, sort_keys=True)
path_of_file = dirname(
abspath(__file__)) + "/jupyter_kernel_polymake/resources/"
filenames = [
"three.js", "Detector.js", "controls/TrackballControls.js",
"renderers/SVGRenderer.js", "renderers/CanvasRenderer.js",
"renderers/Projector.js", "menu.svg", "close.svg"
]
for i in filenames:
file_copy(path_of_file + i, td)
file_copy(path_of_file + "kernel.js", td)
file_copy(path_of_file + "logo-32x32.png", td)
file_copy(path_of_file + "logo-64x64.png", td)
print('Installing jupyter kernel spec for polymake')
install_kernel_spec(td, 'polymake', user=user, replace=True)
def test_run_msaflagging(mk_tmp_dirs, caplog):
"""Test msa flagging operation"""
tmp_current_path, tmp_data_path, tmp_config_path = mk_tmp_dirs
# Copy necessary data to the tmp_data_path
file_copy(
path.join(DATAPATH, 'jw95065006001_0_msa_twoslit.fits'),
tmp_data_path
)
file_copy(
path.join(DATAPATH, 'mos_udf_g235m_twoslit_spec2_asn.json'),
tmp_data_path
)
asn_path = path.join(tmp_data_path, 'mos_udf_g235m_twoslit_spec2_asn.json')
with open(asn_path) as fp:
asn = load_asn(fp)
for product in asn['products']:
for member in product['members']:
file_copy(
path.join(DATAPATH, 'level2a_twoslit', member['expname']),
tmp_data_path
)
args = [
path.join(SCRIPT_DATA_PATH, 'calwebb_spec2_basic.cfg'),
asn_path,
'--steps.msa_flagging.skip=false'
]
Step.from_cmdline(args)
assert 'Step msa_flagging running with args' in caplog.text
assert 'Step msa_flagging done' in caplog.text
for product in asn['products']:
prod_name = product['name'] + '_cal.fits'
assert path.isfile(prod_name)
def tag_learn_test(**kwargs): # 90201
# tag_files args:
input_path: str = kwargs['input_parses']
output_path: str = kwargs['output_grammar'] # ['out_path'] ?
corpus: str = kwargs['output_grammar']
cp: str
log = {}
if 'input_grammar' not in kwargs:
rulez, re01 = learn(**kwargs)
log.update(re01)
new_dict_path = re01['grammar_file']
else: # tag and learn
# print('else: tag and learn')
#?kwargs['out_path'] = kwargs['output_grammar'] # used in tag_files only
# if 'out_path' in kwargs:
# out_path : str = kwargs['out_path']
# del kwargs['out_path'] # tag_files uses kwargs['output_grammar'] instead
key_dict_path: str = kwargs['input_grammar'] # dict for tagging
re02 = tag_files(**kwargs)
log.update(re02)
#-kwargs['input_parses'] = re1['tagger_output_path'] + '/tagged_ull'
kwargs['input_parses'] = output_path + '/tagged_ull'
check_dir(kwargs['input_parses'], False, 'max')
#-kwargs['output_grammar'] = kwargs['out_path']
rulez, re03 = learn(**kwargs) # rulez: dict FIXME: return
log.update(re03)
# Decode .dict:
new_dict_path = re03['grammar_file']
with open(new_dict_path, 'r') as f:
d: list = f.read().splitlines() # TODO? split at dict2list?
tagged_dict_path = file_copy(new_dict_path, new_dict_path + '.tagged')
with open(key_dict_path, 'r') as f:
kd: list = f.read().splitlines() # TODO? split at dict2list?
clusters: dict = dict2lists(kd, **kwargs)
with open(new_dict_path, 'w') as f:
f.write(decode_dict(d, clusters))
# TODO: single def to decode dict, input -- 2*strings:
# with open(key_dict_path, 'r') as f: kd = f.read() # string
# with open(new_dict_path, 'r') as f: d = f.read() # string
# decoded_dict: str = decode_dict_new(d, kd)
# decoded
#-check:
#-with open(new_dict_path, 'r') as f: tmp = f.read().splitlines()
#-print(tmp[-7:])
# TODO: decode cat_tree.txt
cat_tree_file = re03['cat_tree_file']
with open(cat_tree_file, 'r') as f:
tree = f.read()
tagged_cat_tree_path = file_copy(cat_tree_file,
cat_tree_file + '.tagged')
with open(cat_tree_file, 'w') as f:
f.write(decode_cat_tree(tree, kd, **kwargs))
# TODO: Test Grammar with decoded .dict
# pa, f1, p, pq: parse-ability, F-measure, precision, parse quality
pa, f1, p, pq = pqa_meter(new_dict_path, '', '', '', **kwargs)
# op,cp,rp = '' ยป use kwargs['out_path'], corpus_path, reference_path
# TODO: log.update(a, f1, p, q)
# print('pa, f1, p, pq:', pa, f1, p, pq)
# TODO: replace pqa_meter with a local function: re = pqa(**kwargs)
# TODO: decode & return rulez? return .dict converted to a string?
# TODO: return line []?
return log['grammar_rules'], pa, f1, log # rulez, log
def _store_output_data(
data_object_path,
data_directory,
output_to_store,
results_by_id,
):
"""Collects all of the simulation to store, and saves it into a directory
whose path will be passed to the storage backend to process.
Parameters
----------
data_object_path: str
The file path to serialize the data object to.
data_directory: str
The path of the directory to store ancillary data in.
output_to_store: BaseStoredData
An object which contains `ProtocolPath`s pointing to the
data to store.
results_by_id: dict of ProtocolPath and any
The results of the protocols which formed the property
estimation workflow.
"""
makedirs(data_directory, exist_ok=True)
for attribute_name in output_to_store.get_attributes(StorageAttribute):
attribute = getattr(output_to_store.__class__, attribute_name)
attribute_value = getattr(output_to_store, attribute_name)
if isinstance(attribute_value, ProtocolPath):
# Strip any nested attribute accessors before retrieving the result
property_name = attribute_value.property_name.split(
".")[0].split("[")[0]
result_path = ProtocolPath(property_name,
*attribute_value.protocol_ids)
result = results_by_id[result_path]
if result_path != attribute_value:
result = get_nested_attribute(
{property_name: result}, attribute_value.property_name)
attribute_value = result
# Do not store gradient information for observables as this information
# is very workflow / context specific.
if isinstance(attribute_value,
(Observable, ObservableArray, ObservableFrame)):
attribute_value.clear_gradients()
if issubclass(attribute.type_hint, FilePath):
file_copy(attribute_value, data_directory)
attribute_value = path.basename(attribute_value)
setattr(output_to_store, attribute_name, attribute_value)
with open(data_object_path, "w") as file:
json.dump(output_to_store, file, cls=TypedJSONEncoder)
def install_by_version(cls, gadgets, kata_runtime_type,
http_proxy=None, https_proxy=None, no_proxy=None, verbose=False):
"""Install Kata-containers with specified version.
Args:
gadgets: Kata-containers gadgets (e.g. kata-containers).
kata_runtime_type: Runtime of Kata (e.g. qemu/clh/...).
http_proxy: HTTP proxy.
https_proxy: HTTPS proxy.
no_proxy: Domains which should be visited without proxy.
verbose: Verbose or not.
Returns:
Boolean indicating whether Kata-containers is successfully installed or not.
"""
stdout, stderr = verbose_func.verbose_output(verbose)
kata_static_tar_file = config.kata_static_tar_file % gadgets[0]['version']
kata_static_save_path = config.runtime_data_dir + kata_static_tar_file
kata_static_tar = Path(kata_static_save_path)
# 1. download kata tar if necessary
if not kata_static_tar.exists():
color_print.debug(
'{kata_tar} is going to be downloaded'.format(
kata_tar=kata_static_tar_file))
kata_static_url = (
config.kata_static_url_prefix %
gadgets[0]['version']) + kata_static_tar_file
proxies = {
'http': http_proxy,
'https': https_proxy,
'no_proxy': no_proxy,
}
cls.download_file(
url=kata_static_url,
save_path=kata_static_save_path,
proxies=proxies)
else:
color_print.debug(
'{kata_tar} has been downloaded'.format(
kata_tar=kata_static_tar_file))
# 2. decompress
color_print.debug(
'decompressing files into {dest}'.format(
dest=config.kata_tar_decompress_dest))
rmtree(path=config.kata_tar_decompress_dest, ignore_errors=True)
system_func.mkdir_if_not_exist(config.kata_tar_decompress_dest)
# use --strip-components=3 because `opt/kata/` path from tar are not needed
# also, we should not just decompress files into `/` root path
# which might cause risks
temp_cmd = 'tar xf {file} -C {dest} --strip-components=3'.format(
file=kata_static_save_path, dest=config.kata_tar_decompress_dest)
try:
subprocess.run(
temp_cmd.split(),
stdout=stdout,
stderr=stderr,
check=True)
except subprocess.CalledProcessError:
color_print.error(
'failed to decompress {kata_tar}'.format(
kata_tar=kata_static_tar_file))
return False
# 3. copy files
color_print.debug(
'copying files to {kata_config_dir}'.format(
kata_config_dir=config.kata_config_dir))
rmtree(path=config.kata_config_dir, ignore_errors=True)
system_func.mkdir_if_not_exist(config.kata_config_dir)
for file in glob.glob(
config.kata_tar_decompress_dest + 'share/defaults/kata-containers/*'):
file_copy(
src=file,
dst=config.kata_config_dir,
follow_symlinks=False)
# 4. configure runtime type
color_print.debug(
'configuring kata runtime (type: {runtime_type})'.format(
runtime_type=kata_runtime_type))
kata_configuration_file = Path(
'{kata_config_dir}/configuration.toml'.format(kata_config_dir=config.kata_config_dir))
if kata_configuration_file.exists():
kata_configuration_file.unlink()
kata_configuration_file.symlink_to(
'{kata_config_dir}/configuration-{runtime_type}.toml'.format(
kata_config_dir=config.kata_config_dir,
runtime_type=kata_runtime_type))
# [5]. if docker is installed,
# modify docker's configuration and restart docker
# currently, metarget only supports docker
# in the future more CRIs will be supported
# see
# https://github.com/kata-containers/documentation/blob/master/how-to/run-kata-with-k8s.md
color_print.debug('configuring docker with kata-containers')
if not cls._configure_docker_with_kata(
base_dir=config.kata_tar_decompress_dest):
color_print.error(
'failed to configure docker with kata-containers')
return False
return cls.reload_and_restart_docker(verbose=verbose)
def generate_train_test_set_helper(sample_dir,
target_dir,
cross_val_num=5,
seed=42,
has_just_feature_matrix=False,
dataset_name=""):
sample_dir_path = Path(sample_dir)
target_dir_path = Path(target_dir)
if not dataset_name:
dataset_name = sample_dir_path.stem
# guess class label file
class_label_file = MccImsAnalysis.guess_class_label_extension(
sample_dir_path)
class_labels = MccImsAnalysis.parse_class_labels(class_label_file)
available_raw_files = sorted(sample_dir_path.glob("*_ims.csv"))
available_preprocessed_files = sorted(
sample_dir_path.glob("*_ims_preprocessed.csv"))
available_pdrs = sorted(
sample_dir_path.glob("*_peak_detection_result.csv"))
# make sure we know which files we're missing so we can get them - too many files is not a problem - subset is allowed - or if we need to edit class labels
not_available_raw_files = []
if available_raw_files:
raw_files = []
for arw in available_raw_files:
raw_files.append(arw.name)
raw_file_set = set(raw_files)
for fn in class_labels.keys():
if fn not in raw_file_set:
not_available_raw_files.append(fn)
print(f"Missing raw files: {not_available_raw_files}")
not_available_preproc_files = []
if available_preprocessed_files:
preproc_files = []
for apf in available_preprocessed_files:
raw_name = apf.name.split("_ims.csv_")[0]
raw_name += "_ims.csv"
preproc_files.append(raw_name)
preproc_file_set = set(preproc_files)
for fn in class_labels:
if fn not in preproc_file_set:
not_available_preproc_files.append(
f"{fn[:-4]}_preprocessed.csv")
print(f"Missing preprocessed files: {not_available_preproc_files}")
# approximate search, dont want to spell out all pdr_names
not_available_pdr_files = []
if available_pdrs:
av_pdrs = []
for apdr in available_pdrs:
raw_name = apdr.name.split("_ims.csv_")[0]
raw_name += "_ims.csv"
av_pdrs.append(raw_name)
av_pdr_set = set(av_pdrs)
for fn in class_labels:
if fn not in av_pdr_set:
not_available_pdr_files.append(
f"{fn[:-4]}_peak_detection_result.csv")
print(f"Missing peak detection result: {not_available_pdr_files}")
if not_available_raw_files or not_available_preproc_files or not_available_pdr_files:
raise ValueError(
"Class labels needs to be adjusted or missing files added.")
# check if we have a layer_file
potential_layers = [
str(filename) for filename in sample_dir_path.glob("*")
if (str.endswith(str(filename), "layer.csv")
or str.endswith(str(filename), "layer.xls"))
]
print(
f"Preparing dataset for {Counter(class_labels.values())} using {cross_val_num}-fold cross validation splits."
)
X = [k for k in class_labels.keys()]
y = [v for v in class_labels.values()]
# class_labels[m.filename]
test_fraction = 1. / cross_val_num
train_df, test_df = split_labels_ratio(class_labels,
train_val_fraction=1 -
test_fraction,
seed=seed)
train_dir = str(target_dir_path) + "/" + f"train_{dataset_name}/"
test_dir = str(target_dir_path) + "/" + f"test_{dataset_name}/"
print(f"Deleting {train_dir} and {test_dir}")
# delete train and test dir if already exisitent
if Path(train_dir).exists():
rmtree(train_dir, ignore_errors=True)
if Path(test_dir).exists():
rmtree(test_dir, ignore_errors=True)
# TODO also remove exsiting results, such as peak_detection_results, feature matrices
print(f"Creating {train_dir} and {test_dir}")
Path(train_dir).mkdir(parents=True)
Path(test_dir).mkdir(parents=True)
tr_class_label_fn = Path(train_dir) / Path(class_label_file).name
te_class_label_fn = Path(test_dir) / Path(class_label_file).name
train_df[['name', 'label']].to_csv(tr_class_label_fn, sep=",", index=False)
test_df[['name', 'label']].to_csv(te_class_label_fn, sep=",", index=False)
# check if it has peak detection results
pdrs = sorted(sample_dir_path.glob("*_peak_detection_result.csv"))
# distribute into train and test list
train_name_set = set(train_df['name'].values)
test_name_set = set(test_df['name'].values)
cannot_copy = []
for pdr in pdrs:
raw_fn_pre = pdr.name.split("_ims.csv")[0]
raw_fn = raw_fn_pre + "_ims.csv"
new_fn = ""
if raw_fn in train_name_set:
new_fn = Path(train_dir) / pdr.name
elif raw_fn in test_name_set:
new_fn = Path(test_dir) / pdr.name
else:
cannot_copy.append(pdr)
# copy to destination
if new_fn:
file_copy(pdr, new_fn)
if cannot_copy:
print(f"{len(cannot_copy)} PDRs not in either index.",
f"{cannot_copy}")
if has_just_feature_matrix:
# write feature matrix
potential_feature_matrices = sample_dir_path.glob(
"*_feature_matrix.csv")
for fn in potential_feature_matrices:
try:
fm = pd.read_csv(fn, index_col=0)
tr_fm = fm.loc[fm.index.intersection(train_df['name'])]
te_fm = fm.loc[fm.index.intersection(test_df['name'])]
tr_fm_fn = Path(train_dir) / "train_feature_matrix.csv"
te_fm_fn = Path(test_dir) / "test_feature_matrix.csv"
tr_fm.to_csv(tr_fm_fn)
te_fm.to_csv(te_fm_fn)
print(f"Created feature matrices {tr_fm_fn} and {te_fm_fn}")
# also implement for other branches - pdr and preprocessed
for t_dir, t_fm, t_cl, in zip(
[train_dir, test_dir], [tr_fm_fn, te_fm_fn],
[tr_class_label_fn, te_class_label_fn]):
t_dir_path = Path(t_dir)
t_dir_name = t_dir_path.stem
zip_path_tr = t_dir_path / f"{t_dir_name}.zip"
with ZipFile(zip_path_tr, 'w', ZIP_DEFLATED) as trzip:
trzip.write(
t_fm, t_fm.name
) # needs to exist as file object on disk to to write to zip
trzip.write(t_cl, t_cl.name)
except ValueError:
# go until no more potential candidates, which should be just one anyways
pass
else:
# copy files to target dirs - only works if raw files are actually there, not always the case - eg if there's just results
raw_files_not_copied = []
for fn in train_df.name:
file_path = Path(sample_dir_path) / fn
new_path = Path(train_dir) / fn
if file_path.exists():
file_copy(file_path, dst=new_path)
else:
raw_files_not_copied.append(file_path)
# same for test set
for fn in test_df.name:
file_path = Path(sample_dir_path) / fn
new_path = Path(test_dir) / fn
if file_path.exists():
file_copy(file_path, dst=new_path)
else:
raw_files_not_copied.append(file_path)
if raw_files_not_copied:
print(
f"Didn't copy {len(raw_files_not_copied)} raw files - as not found in source directory."
)
# also consider featureXML files
feature_xmls_fns = filter_feature_xmls(sample_dir_path)
cannot_copy = []
for feature_xml_fn_ in feature_xmls_fns:
feature_xml_fn = Path(feature_xml_fn_)
# split_fn - so can campare with class labels
# need to get file ending of original raw file to match with class labels
# could be MZML or MZXML
raw_fn_pre = feature_xml_fn.name.split("ML_output.featureXML")[0]
raw_fn = raw_fn_pre + "ML"
new_fn = ""
if raw_fn in train_name_set:
new_fn = Path(train_dir) / feature_xml_fn.name
elif raw_fn in test_name_set:
new_fn = Path(test_dir) / feature_xml_fn.name
else:
cannot_copy.append(feature_xml_fn)
# copy to destination
if new_fn:
file_copy(feature_xml_fn, new_fn)
if feature_xmls_fns:
print(
f"Copied {len(feature_xmls_fns) - len(cannot_copy)}/{len(feature_xmls_fns)} featureXML files."
)
if cannot_copy:
print(f"{len(cannot_copy)} featureXML not in either index.",
f"{cannot_copy}")
# guess layer file and copy to target dir too
if potential_layers:
potential_layer_file = potential_layers[0]
layer_name = Path(potential_layer_file).stem + Path(
potential_layer_file).suffix
file_copy(potential_layers[0], dst=str(train_dir) + "/" + layer_name)
file_copy(potential_layers[0], dst=str(test_dir) + "/" + layer_name)
print(f"{'|' * 40}\nFinished preparation of {dataset_name}\n")
return train_df, test_df
def generate_full_candy_classes(plot_params, file_params, preprocessing_steps,
evaluation_params_dict):
all_files = glob.glob(file_params['data_dir'] + "*_ims.csv")
class_labels = {
r[0]: r[1]
for r in np.loadtxt(file_params['data_dir'] + "class_labels.csv",
delimiter=",",
dtype=str,
skiprows=1)
}
from collections import OrderedDict
class_labels = OrderedDict(class_labels)
analysis = MccImsAnalysis(
[MccImsMeasurement(f) for f in all_files],
preprocessing_steps, [],
performance_measure_parameters=evaluation_params_dict,
class_label_file=file_params['label_filename'],
dataset_name='full_candy',
dir_level="")
for m in analysis.measurements:
class_label = class_labels.get(m.filename)
m.set_class_label(class_label)
from sklearn.model_selection import train_test_split
from shutil import copy as file_copy
from shutil import rmtree
from pathlib import Path
X = [k for k in class_labels.keys()]
y = [v for v in class_labels.values()]
# class_labels[m.filename]
X_train, X_test, y_train, y_test = train_test_split(X,
y,
test_size=0.20,
random_state=42)
train_df = pd.DataFrame({"name": X_train, "candy": y_train})
test_df = pd.DataFrame({"name": X_test, "candy": y_test})
train_dir = file_params['data_dir'] + "train_full_candy/"
test_dir = file_params['data_dir'] + "test_full_candy/"
# delete train and test dir
if Path(train_dir).exists():
rmtree(train_dir, ignore_errors=True)
if Path(test_dir).exists():
rmtree(test_dir, ignore_errors=True)
# create directory
Path(train_dir).mkdir()
Path(test_dir).mkdir()
# create class_label file
train_df[['name', 'candy']].to_csv(train_dir + "class_labels.csv", sep=",")
test_df[['name', 'candy']].to_csv(test_dir + "class_labels.csv", sep=",")
# copy files to target dirs
for fn in train_df.name:
file_path = file_params['data_dir'] + fn
new_path = train_dir + fn
file_copy(file_path, dst=new_path)
# same for test set
for fn in test_df.name:
file_path = file_params['data_dir'] + fn
new_path = test_dir + fn
file_copy(file_path, dst=new_path)
def main():
'''Main function to get a fresh scope.'''
args = parse_arguments()
# Setup functions for dry run if necessary, else start logging
if args.dry:
print('Performing dry run!')
setup_dry_run()
else:
# Create root directory for all the results
os.mkdir(args.outdir)
# Setup logging
logging.basicConfig(filename=os.path.join(args.outdir, 'stats.log'),
level=logging.INFO)
# Models we are applying
models, models2 = load_models(args.modeldir, args.powerset)
if len(models) == 0 and not args.none:
print('Warning, no model is being used! Use --none to force')
return
# Save some interesting data in global stats
write_stats(args, models, args.powerset)
logger_stats = logging.getLogger('stats')
logger_other = logging.getLogger('other')
# If provided, copy configuration file
# TODO also, read values from the config file instead of using cli arguments
if args.config is not None:
cfg = os.path.join(args.outdir, 'configuration.ini')
file_copy(args.config, cfg)
args.config = cfg # Replace old choice
if not args.noask:
print(
f'This is the time to review your configuration file in {cfg}')
print(
f'Generation counts will be {args.shortg} (short) and {args.longg} (long)'
)
input(f'Press Enter when ready to go.')
# Save config file to stats, for reference
with zopen(cfg, 'rt') as cfgfp:
global_stats['configuration.ini'] = cfgfp.read()
# TODO
# per poter fare le analisi sul tempo, bisogna avere una media delle semantiche
# dopo tutti i run. per non recuperare le semantiche al momento delle analisi dati
# conviene fare qui la media e la produzione di un file di output medio che sia facilmente usabile nelle analisi
for r in range(args.runs):
print(f'Performing run {r}')
# Prepare output directory for this run
outdir = get_run_path(args.outdir, r) # somepath/sim += /sim{r}
os.mkdir(outdir)
# Prepare dataset in somepath/sim/sim{r}/dataset
# This single run will have the datafile partitioned in k folds
dataset = Dataset(args.datafile, args.k_folds, outdir)
cons, cons_msg = dataset.is_consistent()
if not cons:
# Emit a signal if K does not evenly partition the dataset
print('Warning! Selected K cannot produce consistent semantics!')
print(cons_msg)
logi('run.dataset', cons_msg)
dataset.generate_folds(True)
# Model selection
if args.all:
bm = len(models2) - 1 # Last combination
t_tot = 0 # No time spent
elif args.none:
bm = 0
t_tot = 0 # No time spent
else:
if args.powerset:
bm, t_tot = run_powerset(args, outdir, models2, dataset)
else:
bm, t_tot = run_set(args, outdir, models2, dataset)
# Get actual
best_models = models2[bm]
# Save selection time
global_stats['sel_time'] = global_stats.get('sel_time', 0) + t_tot
global_stats.setdefault('sel_times', []).append(t_tot)
logi('stats.selection.walltimes',
f'Time for running selection: {t_tot}')
# Increment best model usage
global_stats['best_models'] = global_stats.get(
'best_models', Counter()) + Counter({str(bm): 1})
# Save combination
global_stats.setdefault('bm_hist', []).append(bm)
logi('stats.selection.models.best', f'{bm} {best_models}')
print('Performing long run with best models', models2[bm])
# Prepare simulation, storing data in somepath/sim/sim{r}/longrun
forrest = Forrest(f'longrun', args.algorithm, models2[bm], dataset,
args.k_folds, outdir, args.bindir, args.config)
# Run simulation
k_fits, k_timing, avg_sem_train, avg_sem_test = forrest.run(args.longg)
# Write average semantic data
forrest.save_files(avg_sem_train, avg_sem_test)
# Save logs and stats
logi('stats.longrun.cv.fitness.average',
f'Average CV: {row_average(k_fits)}')
t_tot = sum(k_timing) # Total time for executing K-fold CV
logi('stats.longrun.walltimes', f'Total time for longruns: {t_tot}')
global_stats['lon_time'] = global_stats.get('lon_time', 0) + t_tot
global_stats.setdefault('lon_times', []).append(t_tot)
logi('stats.selection.models.frequency', f'{global_stats["best_models"]}')
with zopen(os.path.join(args.outdir, 'stats.json'), 'wt') as statfile:
#for k, v in global_stats.items():
# print(f'Writing stat {k} = {v}')
json.dump(global_stats, statfile, indent=4, sort_keys=True)
