def __init__(self,
dag_id=None,
cwl_workflow=None,
default_args={},
schedule_interval=None,
*args,
**kwargs):
self.top_task = None
self.bottom_task = None
self.cwlwf = self.quick_load_cwl(cwl_workflow)
kwargs.update({
"on_failure_callback":
kwargs.get("on_failure_callback", dag_on_failure),
"on_success_callback":
kwargs.get("on_success_callback", dag_on_success)
})
# parameters that cannot be overwritten in default_args
default_args.update({
'singularity':
conf_get_default('cwl', 'singularity',
default_args.get('singularity', False)),
'use_container':
conf_get_default('cwl', 'use_container',
default_args.get('use_container', True))
})
init_default_args = {
'start_date': days_ago(14),
'email_on_failure': False,
'email_on_retry': False,
'end_date': None,
'tmp_folder': conf_get_default('cwl', 'tmp_folder', '/tmp'),
'basedir': conf_get_default('cwl', 'tmp_folder', '/tmp'),
'no_match_user': conf_get_default('cwl', 'no_match_user', False),
'task_retries': conf_get_default('cwl', 'retry', 1),
'quiet': False,
'strict': False,
'on_error': 'continue',
'skip_schemas': True,
'cwl_workflow': cwl_workflow
}
init_default_args.update(default_args)
merged_default_args = get_default_args()
merged_default_args.update(init_default_args)
super(self.__class__, self).__init__(
dag_id=dag_id if dag_id else
urllib.parse.urldefrag(cwl_workflow)[0].split("/")[-1].replace(
".cwl", "").replace(".", "_dot_"),
default_args=merged_default_args,
schedule_interval=schedule_interval,
*args,
**kwargs)
def __init__(
self,
dag_id=None,
cwl_workflow=None,
default_args=None,
schedule_interval=None,
*args, **kwargs):
self.top_task = None
self.bottom_task = None
self.cwlwf = None
self.requirements = None
tmp_folder = conf_get_default('cwl', 'tmp_folder', '/tmp')
_default_args = {
'start_date': utcnow(),
'email_on_failure': False,
'email_on_retry': False,
'end_date': None,
'tmp_folder': tmp_folder,
'basedir': tmp_folder,
'print_deps': False,
'print_pre': False,
'print_rdf': False,
'print_dot': False,
'relative_deps': False,
'use_container': True,
'rm_container': True,
'enable_pull': True,
'preserve_environment': ["PATH"],
'preserve_entire_environment': False,
'print_input_deps': False,
'cachedir': None,
'rm_tmpdir': True,
'move_outputs': 'move',
'eval_timeout': 20,
'quiet': False,
'version': False,
'enable_dev': False,
'enable_ext': False,
'strict': False,
'rdf_serializer': None,
'tool_help': False,
'pack': False,
'on_error': 'continue',
'relax_path_checks': False,
'validate': False,
'compute_checksum': True,
'skip_schemas': True,
'no_match_user': False,
}
_default_args.update(default_args if default_args else {})
_d = get_default_args()
_d.update(_default_args)
self.cwl_workflow = cwl_workflow if cwl_workflow else _default_args["cwl_workflow"]
_dag_id = dag_id if dag_id else urllib.parse.urldefrag(self.cwl_workflow)[0].split("/")[-1] \
.replace(".cwl", "").replace(".", "_dot_")
super(self.__class__, self).__init__(dag_id=_dag_id,
default_args=_d,
schedule_interval=schedule_interval, *args, **kwargs)
def get_default_cwl_args(preset_cwl_args=None):
"""
Returns default arguments required by cwltool's functions with a few
parameters added and overwritten (required by CWL-Airflow). Defaults
can be preset through "preset_cwl_args" if provided. All new fields
from "preset_cwl_args" will be added to the returned results.
"""
preset_cwl_args = {} if preset_cwl_args is None else deepcopy(
preset_cwl_args)
# default arguments required by cwltool
required_cwl_args = get_default_args()
# update default arguments required by cwltool with those that were preset by user
required_cwl_args.update(preset_cwl_args)
# update default arguments required by cwltool with those that might
# be updated based on the higher priority of airflow configuration
# file. If airflow configuration file doesn't include correspondent
# parameters, use those that were preset by user, or defaults
required_cwl_args.update({
"tmp_folder":
get_dir(
conf_get("cwl", "tmp_folder",
preset_cwl_args.get("tmp_folder", CWL_TMP_FOLDER))),
"outputs_folder":
get_dir( # for CWL-Airflow to store outputs if "outputs_folder" is not overwritten in job
conf_get("cwl", "outputs_folder",
preset_cwl_args.get("outputs_folder",
CWL_OUTPUTS_FOLDER))),
"inputs_folder":
get_dir( # for CWL-Airflow to resolve relative locations for input files if job was loaded from parsed object
conf_get("cwl", "inputs_folder",
preset_cwl_args.get("inputs_folder", CWL_INPUTS_FOLDER))),
"pickle_folder":
get_dir( # for CWL-Airflow to store pickled workflows
conf_get("cwl", "pickle_folder",
preset_cwl_args.get("pickle_folder", CWL_PICKLE_FOLDER))),
"use_container":
conf_get(
"cwl",
"use_container",
preset_cwl_args.get(
"use_container",
CWL_USE_CONTAINER) # execute jobs in docker containers
),
"no_match_user":
conf_get(
"cwl",
"no_match_user",
preset_cwl_args.get(
"no_match_user", CWL_NO_MATCH_USER
) # disables passing the current uid to "docker run --user"
),
"skip_schemas":
conf_get(
"cwl",
"skip_schemas",
preset_cwl_args.get(
"skip_schemas", CWL_SKIP_SCHEMAS
) # it looks like this doesn't influence anything in the latest cwltool
),
"strict":
conf_get("cwl", "strict", preset_cwl_args.get("strict", CWL_STRICT)),
"quiet":
conf_get("cwl", "quiet", preset_cwl_args.get("quiet", CWL_QUIET)),
"rm_tmpdir":
preset_cwl_args.get(
"rm_tmpdir", CWL_RM_TMPDIR
), # even if we can set it in "preset_cwl_args" it's better not to change
"move_outputs":
preset_cwl_args.get(
"move_outputs", CWL_MOVE_OUTPUTS
), # even if we can set it in "preset_cwl_args" it's better not to change
"enable_dev":
preset_cwl_args.get(
"enable_dev", CWL_ENABLE_DEV
) # fails to run without it when creating workflow from tool. TODO: Ask Peter?
})
return required_cwl_args
def load_cwl(cwl_file):
load.loaders = {}
loading_context = cwltool.context.LoadingContext(get_default_args())
loading_context.construct_tool_object = default_make_tool
loading_context.resolver = tool_resolver
return load.load_tool(cwl_file, loading_context)
