class JobManager:
def __init__(self, app):
self.app = app
self.dataset_manager = DatasetManager(app)
def get_accessible_job(self, trans, decoded_job_id):
job = trans.sa_session.query(trans.app.model.Job).filter(
trans.app.model.Job.id == decoded_job_id).first()
if job is None:
raise ObjectNotFound()
belongs_to_user = (job.user == trans.user) if job.user else (
job.session_id == trans.get_galaxy_session().id)
if not trans.user_is_admin and not belongs_to_user:
# Check access granted via output datasets.
if not job.output_datasets:
raise ItemAccessibilityException("Job has no output datasets.")
for data_assoc in job.output_datasets:
if not self.dataset_manager.is_accessible(
data_assoc.dataset.dataset, trans.user):
raise ItemAccessibilityException(
"You are not allowed to rerun this job.")
trans.sa_session.refresh(job)
return job
def stop(self, job, message=None):
if not job.finished:
job.mark_deleted(self.app.config.track_jobs_in_database)
self.app.model.context.current.flush()
self.app.job_manager.stop(job, message=message)
return True
else:
return False
class JobManager:
def __init__(self, app: StructuredApp):
self.app = app
self.dataset_manager = DatasetManager(app)
def job_lock(self) -> JobLock:
return JobLock(active=self.app.job_manager.job_lock)
def update_job_lock(self, job_lock: JobLock):
self.app.queue_worker.send_control_task(
'admin_job_lock',
kwargs={'job_lock': job_lock.active},
get_response=True)
return self.job_lock()
def get_accessible_job(self, trans, decoded_job_id):
job = trans.sa_session.query(trans.app.model.Job).filter(
trans.app.model.Job.id == decoded_job_id).first()
if job is None:
raise ObjectNotFound()
belongs_to_user = (
job.user_id == trans.user.id) if job.user_id and trans.user else (
job.session_id == trans.get_galaxy_session().id)
if not trans.user_is_admin and not belongs_to_user:
# Check access granted via output datasets.
if not job.output_datasets:
raise ItemAccessibilityException("Job has no output datasets.")
for data_assoc in job.output_datasets:
if not self.dataset_manager.is_accessible(
data_assoc.dataset.dataset, trans.user):
raise ItemAccessibilityException(
"You are not allowed to rerun this job.")
trans.sa_session.refresh(job)
return job
def stop(self, job, message=None):
if not job.finished:
job.mark_deleted(self.app.config.track_jobs_in_database)
self.app.model.session.flush()
self.app.job_manager.stop(job, message=message)
return True
else:
return False
class JobManager(object):
def __init__(self, app):
self.app = app
self.dataset_manager = DatasetManager(app)
def get_accessible_job(self, trans, decoded_job_id):
job = trans.sa_session.query(trans.app.model.Job).filter(trans.app.model.Job.id == decoded_job_id).first()
if job is None:
raise ObjectNotFound()
belongs_to_user = (job.user == trans.user) if job.user else (job.session_id == trans.get_galaxy_session().id)
if not trans.user_is_admin and not belongs_to_user:
# Check access granted via output datasets.
if not job.output_datasets:
raise ItemAccessibilityException("Job has no output datasets.")
for data_assoc in job.output_datasets:
if not self.dataset_manager.is_accessible(data_assoc.dataset.dataset, trans.user):
raise ItemAccessibilityException("You are not allowed to rerun this job.")
trans.sa_session.refresh(job)
return job
class DatasetDeserializerTestCase(BaseTestCase):
def set_up_managers(self):
super(DatasetDeserializerTestCase, self).set_up_managers()
self.dataset_manager = DatasetManager(self.app)
self.dataset_serializer = DatasetSerializer(self.app)
self.dataset_deserializer = DatasetDeserializer(self.app)
self.role_manager = RoleManager(self.app)
def test_deserialize_delete(self):
dataset = self.dataset_manager.create()
self.log('should raise when deserializing deleted from non-bool')
self.assertFalse(dataset.deleted)
self.assertRaises(exceptions.RequestParameterInvalidException,
self.dataset_deserializer.deserialize,
dataset,
data={'deleted': None})
self.assertFalse(dataset.deleted)
self.log('should be able to deserialize deleted from True')
self.dataset_deserializer.deserialize(dataset, data={'deleted': True})
self.assertTrue(dataset.deleted)
self.log(
'should be able to reverse by deserializing deleted from False')
self.dataset_deserializer.deserialize(dataset, data={'deleted': False})
self.assertFalse(dataset.deleted)
def test_deserialize_purge(self):
dataset = self.dataset_manager.create()
self.log('should raise when deserializing purged from non-bool')
self.assertRaises(exceptions.RequestParameterInvalidException,
self.dataset_deserializer.deserialize,
dataset,
data={'purged': None})
self.assertFalse(dataset.purged)
self.log('should be able to deserialize purged from True')
self.dataset_deserializer.deserialize(dataset, data={'purged': True})
self.assertTrue(dataset.purged)
# TODO: should this raise an error?
self.log(
'should NOT be able to deserialize purged from False (will remain True)'
)
self.dataset_deserializer.deserialize(dataset, data={'purged': False})
self.assertTrue(dataset.purged)
def test_deserialize_permissions(self):
dataset = self.dataset_manager.create()
who_manages = self.user_manager.create(**user2_data)
self.dataset_manager.permissions.manage.grant(dataset, who_manages)
existing_permissions = self.dataset_serializer.serialize_permissions(
dataset, 'permissions', user=who_manages)
existing_manage_permissions = existing_permissions['manage']
user3 = self.user_manager.create(**user3_data)
self.log(
'deserializing permissions from a non-dictionary should error')
not_a_dict = []
self.assertRaises(exceptions.RequestParameterInvalidException,
self.dataset_deserializer.deserialize,
dataset,
user=who_manages,
data={'permissions': not_a_dict})
self.log(
'deserializing permissions from a malformed dictionary should error'
)
self.assertRaises(exceptions.RequestParameterInvalidException,
self.dataset_deserializer.deserialize,
dataset,
user=who_manages,
data={'permissions': dict(nope=[], access=[])})
self.log('deserializing permissions with no manage roles should error')
self.assertRaises(exceptions.RequestParameterInvalidException,
self.dataset_deserializer.deserialize,
dataset,
user=who_manages,
data={'permissions': dict(manage=[], access=[])})
self.log(
'deserializing permissions using a non-managing user should error')
self.assertRaises(rbac_secured.DatasetManagePermissionFailedException,
self.dataset_deserializer.deserialize,
dataset,
user=user3,
data={'permissions': existing_permissions})
self.log(
'deserializing permissions with a single access should make the dataset private'
)
private_role = self.user_manager.private_role(who_manages)
private_role = private_role.to_dict(
value_mapper={'id': self.app.security.encode_id})
permissions = dict(manage=existing_manage_permissions,
access=[private_role['id']])
self.dataset_deserializer.deserialize(
dataset, user=who_manages, data={'permissions': permissions})
self.assertFalse(
self.dataset_manager.is_accessible(dataset, user=user3))
self.log(
'deserializing permissions manage should make the permissions available'
)
self.assertRaises(SkipAttribute,
self.dataset_serializer.serialize_permissions,
dataset,
'perms',
user=user3)
# now, have who_manages give a manage permission to user3
private_role = self.user_manager.private_role(user3)
new_manage_permissions = existing_manage_permissions + [
self.app.security.encode_id(private_role.id)
]
permissions = dict(manage=new_manage_permissions, access=[])
self.dataset_deserializer.deserialize(
dataset, user=who_manages, data={'permissions': permissions})
# deserializing for user3 shouldn't throw a skip bc they can manage
permissions = self.dataset_serializer.serialize_permissions(
dataset, 'perms', user=who_manages)
self.assertEqual(new_manage_permissions, permissions['manage'])
class JobController(BaseAPIController, UsesLibraryMixinItems):
def __init__(self, app):
super(JobController, self).__init__(app)
self.dataset_manager = DatasetManager(app)
self.job_search = JobSearch(app)
@expose_api
def index(self, trans, **kwd):
"""
index( trans, state=None, tool_id=None, history_id=None, date_range_min=None, date_range_max=None, user_details=False )
* GET /api/jobs:
return jobs for current user
!! if user is admin and user_details is True, then
return jobs for all galaxy users based on filtering - this is an extended service
:type state: string or list
:param state: limit listing of jobs to those that match one of the included states. If none, all are returned.
Valid Galaxy job states include:
'new', 'upload', 'waiting', 'queued', 'running', 'ok', 'error', 'paused', 'deleted', 'deleted_new'
:type tool_id: string or list
:param tool_id: limit listing of jobs to those that match one of the included tool_ids. If none, all are returned.
:type user_details: boolean
:param user_details: if true, and requestor is an admin, will return external job id and user email.
:type date_range_min: string '2014-01-01'
:param date_range_min: limit the listing of jobs to those updated on or after requested date
:type date_range_max: string '2014-12-31'
:param date_range_max: limit the listing of jobs to those updated on or before requested date
:type history_id: string
:param history_id: limit listing of jobs to those that match the history_id. If none, all are returned.
:rtype: list
:returns: list of dictionaries containing summary job information
"""
state = kwd.get('state', None)
is_admin = trans.user_is_admin()
user_details = kwd.get('user_details', False)
if is_admin:
query = trans.sa_session.query(trans.app.model.Job)
else:
query = trans.sa_session.query(trans.app.model.Job).filter(trans.app.model.Job.user == trans.user)
def build_and_apply_filters(query, objects, filter_func):
if objects is not None:
if isinstance(objects, string_types):
query = query.filter(filter_func(objects))
elif isinstance(objects, list):
t = []
for obj in objects:
t.append(filter_func(obj))
query = query.filter(or_(*t))
return query
query = build_and_apply_filters(query, state, lambda s: trans.app.model.Job.state == s)
query = build_and_apply_filters(query, kwd.get('tool_id', None), lambda t: trans.app.model.Job.tool_id == t)
query = build_and_apply_filters(query, kwd.get('tool_id_like', None), lambda t: trans.app.model.Job.tool_id.like(t))
query = build_and_apply_filters(query, kwd.get('date_range_min', None), lambda dmin: trans.app.model.Job.table.c.update_time >= dmin)
query = build_and_apply_filters(query, kwd.get('date_range_max', None), lambda dmax: trans.app.model.Job.table.c.update_time <= dmax)
history_id = kwd.get('history_id', None)
if history_id is not None:
try:
decoded_history_id = self.decode_id(history_id)
query = query.filter(trans.app.model.Job.history_id == decoded_history_id)
except Exception:
raise exceptions.ObjectAttributeInvalidException()
out = []
if kwd.get('order_by') == 'create_time':
order_by = trans.app.model.Job.create_time.desc()
else:
order_by = trans.app.model.Job.update_time.desc()
for job in query.order_by(order_by).all():
job_dict = job.to_dict('collection', system_details=is_admin)
j = self.encode_all_ids(trans, job_dict, True)
if user_details:
j['user_email'] = job.user.email
out.append(j)
return out
@expose_api_anonymous
def show(self, trans, id, **kwd):
"""
show( trans, id )
* GET /api/jobs/{id}:
return jobs for current user
:type id: string
:param id: Specific job id
:type full: boolean
:param full: whether to return extra information
:rtype: dictionary
:returns: dictionary containing full description of job data
"""
job = self.__get_job(trans, id)
is_admin = trans.user_is_admin()
job_dict = self.encode_all_ids(trans, job.to_dict('element', system_details=is_admin), True)
full_output = util.asbool(kwd.get('full', 'false'))
if full_output:
job_dict.update(dict(stderr=job.stderr, stdout=job.stdout))
if is_admin:
if job.user:
job_dict['user_email'] = job.user.email
else:
job_dict['user_email'] = None
def metric_to_dict(metric):
metric_name = metric.metric_name
metric_value = metric.metric_value
metric_plugin = metric.plugin
title, value = trans.app.job_metrics.format(metric_plugin, metric_name, metric_value)
return dict(
title=title,
value=value,
plugin=metric_plugin,
name=metric_name,
raw_value=str(metric_value),
)
job_dict['job_metrics'] = [metric_to_dict(metric) for metric in job.metrics]
return job_dict
@expose_api
def inputs(self, trans, id, **kwd):
"""
show( trans, id )
* GET /api/jobs/{id}/inputs
returns input datasets created by job
:type id: string
:param id: Encoded job id
:rtype: dictionary
:returns: dictionary containing input dataset associations
"""
job = self.__get_job(trans, id)
return self.__dictify_associations(trans, job.input_datasets, job.input_library_datasets)
@expose_api
def outputs(self, trans, id, **kwd):
"""
outputs( trans, id )
* GET /api/jobs/{id}/outputs
returns output datasets created by job
:type id: string
:param id: Encoded job id
:rtype: dictionary
:returns: dictionary containing output dataset associations
"""
job = self.__get_job(trans, id)
return self.__dictify_associations(trans, job.output_datasets, job.output_library_datasets)
@expose_api
def delete(self, trans, id, **kwd):
"""
delete( trans, id )
* Delete /api/jobs/{id}
cancels specified job
:type id: string
:param id: Encoded job id
"""
job = self.__get_job(trans, id)
if not job.finished:
job.mark_deleted(self.app.config.track_jobs_in_database)
trans.sa_session.flush()
self.app.job_manager.job_stop_queue.put(job.id)
return True
else:
return False
@expose_api_anonymous
def build_for_rerun(self, trans, id, **kwd):
"""
* GET /api/jobs/{id}/build_for_rerun
returns a tool input/param template prepopulated with this job's
information, suitable for rerunning or rendering parameters of the
job.
:type id: string
:param id: Encoded job id
:rtype: dictionary
:returns: dictionary containing output dataset associations
"""
job = self.__get_job(trans, id)
if not job:
raise exceptions.ObjectNotFound("Could not access job with id '%s'" % id)
tool = self.app.toolbox.get_tool(job.tool_id, kwd.get('tool_version') or job.tool_version)
if tool is None:
raise exceptions.ObjectNotFound("Requested tool not found")
if not tool.is_workflow_compatible:
raise exceptions.ConfigDoesNotAllowException("Tool '%s' cannot be rerun." % (job.tool_id))
return tool.to_json(trans, {}, job=job)
def __dictify_associations(self, trans, *association_lists):
rval = []
for association_list in association_lists:
rval.extend(self.__dictify_association(trans, a) for a in association_list)
return rval
def __dictify_association(self, trans, job_dataset_association):
dataset_dict = None
dataset = job_dataset_association.dataset
if dataset:
if isinstance(dataset, model.HistoryDatasetAssociation):
dataset_dict = dict(src="hda", id=trans.security.encode_id(dataset.id))
else:
dataset_dict = dict(src="ldda", id=trans.security.encode_id(dataset.id))
return dict(name=job_dataset_association.name, dataset=dataset_dict)
def __get_job(self, trans, id):
try:
decoded_job_id = self.decode_id(id)
except Exception:
raise exceptions.MalformedId()
job = trans.sa_session.query(trans.app.model.Job).filter(trans.app.model.Job.id == decoded_job_id).first()
if job is None:
raise exceptions.ObjectNotFound()
belongs_to_user = (job.user == trans.user) if job.user else (job.session_id == trans.get_galaxy_session().id)
if not trans.user_is_admin() and not belongs_to_user:
# Check access granted via output datasets.
if not job.output_datasets:
raise exceptions.ItemAccessibilityException("Job has no output datasets.")
for data_assoc in job.output_datasets:
if not self.dataset_manager.is_accessible(data_assoc.dataset.dataset, trans.user):
raise exceptions.ItemAccessibilityException("You are not allowed to rerun this job.")
return job
@expose_api
def create(self, trans, payload, **kwd):
""" See the create method in tools.py in order to submit a job. """
raise exceptions.NotImplemented('Please POST to /api/tools instead.')
@expose_api
def search(self, trans, payload, **kwd):
"""
search( trans, payload )
* POST /api/jobs/search:
return jobs for current user
:type payload: dict
:param payload: Dictionary containing description of requested job. This is in the same format as
a request to POST /apt/tools would take to initiate a job
:rtype: list
:returns: list of dictionaries containing summary job information of the jobs that match the requested job run
This method is designed to scan the list of previously run jobs and find records of jobs that had
the exact some input parameters and datasets. This can be used to minimize the amount of repeated work, and simply
recycle the old results.
"""
tool_id = payload.get('tool_id')
if tool_id is None:
raise exceptions.ObjectAttributeMissingException("No tool id")
tool = trans.app.toolbox.get_tool(tool_id)
if tool is None:
raise exceptions.ObjectNotFound("Requested tool not found")
if 'inputs' not in payload:
raise exceptions.ObjectAttributeMissingException("No inputs defined")
inputs = payload.get('inputs', {})
# Find files coming in as multipart file data and add to inputs.
for k, v in payload.items():
if k.startswith('files_') or k.startswith('__files_'):
inputs[k] = v
request_context = WorkRequestContext(app=trans.app, user=trans.user, history=trans.history)
all_params, all_errors, _, _ = tool.expand_incoming(trans=trans, incoming=inputs, request_context=request_context)
if any(all_errors):
return []
params_dump = [tool.params_to_strings(param, self.app, nested=True) for param in all_params]
jobs = []
for param_dump, param in zip(params_dump, all_params):
job = self.job_search.by_tool_input(trans=trans,
tool_id=tool_id,
tool_version=tool.version,
param=param,
param_dump=param_dump,
job_state=payload.get('state'))
if job:
jobs.append(job)
return [self.encode_all_ids(trans, single_job.to_dict('element'), True) for single_job in jobs]
@expose_api
def error(self, trans, id, **kwd):
"""
error( trans, id )
* POST /api/jobs/{id}/error
submits a bug report via the API.
:type id: string
:param id: Encoded job id
:rtype: dictionary
:returns: dictionary containing information regarding where the error report was sent.
"""
# Get dataset on which this error was triggered
try:
decoded_dataset_id = self.decode_id(kwd['dataset_id'])
except Exception:
raise exceptions.MalformedId()
dataset = trans.sa_session.query(trans.app.model.HistoryDatasetAssociation).get(decoded_dataset_id)
# Get job
job = self.__get_job(trans, id)
tool = trans.app.toolbox.get_tool(job.tool_id, tool_version=job.tool_version) or None
messages = trans.app.error_reports.default_error_plugin.submit_report(
dataset, job, tool, user_submission=True, user=trans.user,
email=kwd.get('email', trans.user.email),
message=kwd.get('message', None)
)
return {'messages': messages}
class JobController(BaseAPIController, UsesLibraryMixinItems):
def __init__(self, app):
super(JobController, self).__init__(app)
self.dataset_manager = DatasetManager(app)
self.job_search = JobSearch(app)
@expose_api
def index(self, trans, **kwd):
"""
index( trans, state=None, tool_id=None, history_id=None, date_range_min=None, date_range_max=None, user_details=False )
* GET /api/jobs:
return jobs for current user
!! if user is admin and user_details is True, then
return jobs for all galaxy users based on filtering - this is an extended service
:type state: string or list
:param state: limit listing of jobs to those that match one of the included states. If none, all are returned.
Valid Galaxy job states include:
'new', 'upload', 'waiting', 'queued', 'running', 'ok', 'error', 'paused', 'deleted', 'deleted_new'
:type tool_id: string or list
:param tool_id: limit listing of jobs to those that match one of the included tool_ids. If none, all are returned.
:type user_details: boolean
:param user_details: if true, and requestor is an admin, will return external job id and user email.
:type date_range_min: string '2014-01-01'
:param date_range_min: limit the listing of jobs to those updated on or after requested date
:type date_range_max: string '2014-12-31'
:param date_range_max: limit the listing of jobs to those updated on or before requested date
:type history_id: string
:param history_id: limit listing of jobs to those that match the history_id. If none, all are returned.
:rtype: list
:returns: list of dictionaries containing summary job information
"""
state = kwd.get('state', None)
is_admin = trans.user_is_admin()
user_details = kwd.get('user_details', False)
if is_admin:
query = trans.sa_session.query(trans.app.model.Job)
else:
query = trans.sa_session.query(trans.app.model.Job).filter(
trans.app.model.Job.user == trans.user)
def build_and_apply_filters(query, objects, filter_func):
if objects is not None:
if isinstance(objects, string_types):
query = query.filter(filter_func(objects))
elif isinstance(objects, list):
t = []
for obj in objects:
t.append(filter_func(obj))
query = query.filter(or_(*t))
return query
query = build_and_apply_filters(
query, state, lambda s: trans.app.model.Job.state == s)
query = build_and_apply_filters(
query, kwd.get('tool_id', None),
lambda t: trans.app.model.Job.tool_id == t)
query = build_and_apply_filters(
query, kwd.get('tool_id_like', None),
lambda t: trans.app.model.Job.tool_id.like(t))
query = build_and_apply_filters(
query, kwd.get('date_range_min', None),
lambda dmin: trans.app.model.Job.table.c.update_time >= dmin)
query = build_and_apply_filters(
query, kwd.get('date_range_max', None),
lambda dmax: trans.app.model.Job.table.c.update_time <= dmax)
history_id = kwd.get('history_id', None)
if history_id is not None:
try:
decoded_history_id = self.decode_id(history_id)
query = query.filter(
trans.app.model.Job.history_id == decoded_history_id)
except Exception:
raise exceptions.ObjectAttributeInvalidException()
out = []
if kwd.get('order_by') == 'create_time':
order_by = trans.app.model.Job.create_time.desc()
else:
order_by = trans.app.model.Job.update_time.desc()
for job in query.order_by(order_by).all():
job_dict = job.to_dict('collection', system_details=is_admin)
j = self.encode_all_ids(trans, job_dict, True)
if user_details:
j['user_email'] = job.user.email
out.append(j)
return out
@expose_api_anonymous
def show(self, trans, id, **kwd):
"""
show( trans, id )
* GET /api/jobs/{id}:
return jobs for current user
:type id: string
:param id: Specific job id
:type full: boolean
:param full: whether to return extra information
:rtype: dictionary
:returns: dictionary containing full description of job data
"""
job = self.__get_job(trans, id)
is_admin = trans.user_is_admin()
job_dict = self.encode_all_ids(
trans, job.to_dict('element', system_details=is_admin), True)
full_output = util.asbool(kwd.get('full', 'false'))
if full_output:
job_dict.update(dict(stderr=job.stderr, stdout=job.stdout))
if is_admin:
if job.user:
job_dict['user_email'] = job.user.email
else:
job_dict['user_email'] = None
def metric_to_dict(metric):
metric_name = metric.metric_name
metric_value = metric.metric_value
metric_plugin = metric.plugin
title, value = trans.app.job_metrics.format(
metric_plugin, metric_name, metric_value)
return dict(
title=title,
value=value,
plugin=metric_plugin,
name=metric_name,
raw_value=str(metric_value),
)
job_dict['job_metrics'] = [
metric_to_dict(metric) for metric in job.metrics
]
return job_dict
@expose_api
def inputs(self, trans, id, **kwd):
"""
show( trans, id )
* GET /api/jobs/{id}/inputs
returns input datasets created by job
:type id: string
:param id: Encoded job id
:rtype: dictionary
:returns: dictionary containing input dataset associations
"""
job = self.__get_job(trans, id)
return self.__dictify_associations(trans, job.input_datasets,
job.input_library_datasets)
@expose_api
def outputs(self, trans, id, **kwd):
"""
outputs( trans, id )
* GET /api/jobs/{id}/outputs
returns output datasets created by job
:type id: string
:param id: Encoded job id
:rtype: dictionary
:returns: dictionary containing output dataset associations
"""
job = self.__get_job(trans, id)
return self.__dictify_associations(trans, job.output_datasets,
job.output_library_datasets)
@expose_api
def delete(self, trans, id, **kwd):
"""
delete( trans, id )
* Delete /api/jobs/{id}
cancels specified job
:type id: string
:param id: Encoded job id
"""
job = self.__get_job(trans, id)
if not job.finished:
job.mark_deleted(self.app.config.track_jobs_in_database)
trans.sa_session.flush()
self.app.job_manager.job_stop_queue.put(job.id)
return True
else:
return False
@expose_api
def resume(self, trans, id, **kwd):
"""
* PUT /api/jobs/{id}/resume
Resumes a paused job
:type id: string
:param id: Encoded job id
:rtype: dictionary
:returns: dictionary containing output dataset associations
"""
job = self.__get_job(trans, id)
if not job:
raise exceptions.ObjectNotFound(
"Could not access job with id '%s'" % id)
if job.state == job.states.PAUSED:
job.resume()
else:
exceptions.RequestParameterInvalidException(
"Job with id '%s' is not paused" % (job.tool_id))
return self.__dictify_associations(trans, job.output_datasets,
job.output_library_datasets)
@expose_api_anonymous
def build_for_rerun(self, trans, id, **kwd):
"""
* GET /api/jobs/{id}/build_for_rerun
returns a tool input/param template prepopulated with this job's
information, suitable for rerunning or rendering parameters of the
job.
:type id: string
:param id: Encoded job id
:rtype: dictionary
:returns: dictionary containing output dataset associations
"""
job = self.__get_job(trans, id)
if not job:
raise exceptions.ObjectNotFound(
"Could not access job with id '%s'" % id)
tool = self.app.toolbox.get_tool(
job.tool_id,
kwd.get('tool_version') or job.tool_version)
if tool is None:
raise exceptions.ObjectNotFound("Requested tool not found")
if not tool.is_workflow_compatible:
raise exceptions.ConfigDoesNotAllowException(
"Tool '%s' cannot be rerun." % (job.tool_id))
return tool.to_json(trans, {}, job=job)
def __dictify_associations(self, trans, *association_lists):
rval = []
for association_list in association_lists:
rval.extend(
self.__dictify_association(trans, a) for a in association_list)
return rval
def __dictify_association(self, trans, job_dataset_association):
dataset_dict = None
dataset = job_dataset_association.dataset
if dataset:
if isinstance(dataset, model.HistoryDatasetAssociation):
dataset_dict = dict(src="hda",
id=trans.security.encode_id(dataset.id))
else:
dataset_dict = dict(src="ldda",
id=trans.security.encode_id(dataset.id))
return dict(name=job_dataset_association.name, dataset=dataset_dict)
def __get_job(self, trans, id):
try:
decoded_job_id = self.decode_id(id)
except Exception:
raise exceptions.MalformedId()
job = trans.sa_session.query(trans.app.model.Job).filter(
trans.app.model.Job.id == decoded_job_id).first()
if job is None:
raise exceptions.ObjectNotFound()
belongs_to_user = (job.user == trans.user) if job.user else (
job.session_id == trans.get_galaxy_session().id)
if not trans.user_is_admin() and not belongs_to_user:
# Check access granted via output datasets.
if not job.output_datasets:
raise exceptions.ItemAccessibilityException(
"Job has no output datasets.")
for data_assoc in job.output_datasets:
if not self.dataset_manager.is_accessible(
data_assoc.dataset.dataset, trans.user):
raise exceptions.ItemAccessibilityException(
"You are not allowed to rerun this job.")
return job
@expose_api
def create(self, trans, payload, **kwd):
""" See the create method in tools.py in order to submit a job. """
raise exceptions.NotImplemented('Please POST to /api/tools instead.')
@expose_api
def search(self, trans, payload, **kwd):
"""
search( trans, payload )
* POST /api/jobs/search:
return jobs for current user
:type payload: dict
:param payload: Dictionary containing description of requested job. This is in the same format as
a request to POST /apt/tools would take to initiate a job
:rtype: list
:returns: list of dictionaries containing summary job information of the jobs that match the requested job run
This method is designed to scan the list of previously run jobs and find records of jobs that had
the exact some input parameters and datasets. This can be used to minimize the amount of repeated work, and simply
recycle the old results.
"""
tool_id = payload.get('tool_id')
if tool_id is None:
raise exceptions.ObjectAttributeMissingException("No tool id")
tool = trans.app.toolbox.get_tool(tool_id)
if tool is None:
raise exceptions.ObjectNotFound("Requested tool not found")
if 'inputs' not in payload:
raise exceptions.ObjectAttributeMissingException(
"No inputs defined")
inputs = payload.get('inputs', {})
# Find files coming in as multipart file data and add to inputs.
for k, v in payload.items():
if k.startswith('files_') or k.startswith('__files_'):
inputs[k] = v
request_context = WorkRequestContext(app=trans.app,
user=trans.user,
history=trans.history)
all_params, all_errors, _, _ = tool.expand_incoming(
trans=trans, incoming=inputs, request_context=request_context)
if any(all_errors):
return []
params_dump = [
tool.params_to_strings(param, self.app, nested=True)
for param in all_params
]
jobs = []
for param_dump, param in zip(params_dump, all_params):
job = self.job_search.by_tool_input(trans=trans,
tool_id=tool_id,
tool_version=tool.version,
param=param,
param_dump=param_dump,
job_state=payload.get('state'))
if job:
jobs.append(job)
return [
self.encode_all_ids(trans, single_job.to_dict('element'), True)
for single_job in jobs
]
@expose_api_anonymous
def error(self, trans, id, **kwd):
"""
error( trans, id )
* POST /api/jobs/{id}/error
submits a bug report via the API.
:type id: string
:param id: Encoded job id
:rtype: dictionary
:returns: dictionary containing information regarding where the error report was sent.
"""
# Get dataset on which this error was triggered
try:
decoded_dataset_id = self.decode_id(kwd['dataset_id'])
except Exception:
raise exceptions.MalformedId()
dataset = trans.sa_session.query(
trans.app.model.HistoryDatasetAssociation).get(decoded_dataset_id)
# Get job
job = self.__get_job(trans, id)
tool = trans.app.toolbox.get_tool(
job.tool_id, tool_version=job.tool_version) or None
email = kwd.get('email')
if not email and not trans.anonymous:
email = trans.user.email
messages = trans.app.error_reports.default_error_plugin.submit_report(
dataset=dataset,
job=job,
tool=tool,
user_submission=True,
user=trans.user,
email=email,
message=kwd.get('message'))
return {'messages': messages}
class DatasetManagerTestCase( BaseTestCase ):
def set_up_managers( self ):
super( DatasetManagerTestCase, self ).set_up_managers()
self.dataset_mgr = DatasetManager( self.app )
def test_create( self ):
self.log( "should be able to create a new Dataset" )
dataset1 = self.dataset_mgr.create( self.trans )
self.assertIsInstance( dataset1, model.Dataset )
self.assertEqual( dataset1, self.trans.sa_session.query( model.Dataset ).get( dataset1.id ) )
def test_base( self ):
dataset1 = self.dataset_mgr.create( self.trans )
dataset2 = self.dataset_mgr.create( self.trans )
self.log( "should be able to query" )
datasets = self.trans.sa_session.query( model.Dataset ).all()
self.assertEqual( self.dataset_mgr.list( self.trans ), datasets )
self.assertEqual( self.dataset_mgr.one( self.trans, filters=( model.Dataset.id == dataset1.id ) ), dataset1 )
self.assertEqual( self.dataset_mgr.by_id( self.trans, dataset1.id ), dataset1 )
self.assertEqual( self.dataset_mgr.by_ids( self.trans, [ dataset2.id, dataset1.id ] ), [ dataset2, dataset1 ] )
self.log( "should be able to limit and offset" )
self.assertEqual( self.dataset_mgr.list( self.trans, limit=1 ), datasets[0:1] )
self.assertEqual( self.dataset_mgr.list( self.trans, offset=1 ), datasets[1:] )
self.assertEqual( self.dataset_mgr.list( self.trans, limit=1, offset=1 ), datasets[1:2] )
self.assertEqual( self.dataset_mgr.list( self.trans, limit=0 ), [] )
self.assertEqual( self.dataset_mgr.list( self.trans, offset=3 ), [] )
self.log( "should be able to order" )
self.assertEqual( self.dataset_mgr.list( self.trans, order_by=sqlalchemy.desc( model.Dataset.create_time ) ),
[ dataset2, dataset1 ] )
def test_delete( self ):
item1 = self.dataset_mgr.create( self.trans )
self.log( "should be able to delete and undelete an hda" )
self.assertFalse( item1.deleted )
self.assertEqual( self.dataset_mgr.delete( self.trans, item1 ), item1 )
self.assertTrue( item1.deleted )
self.assertEqual( self.dataset_mgr.undelete( self.trans, item1 ), item1 )
self.assertFalse( item1.deleted )
def test_purge_allowed( self ):
self.trans.app.config.allow_user_dataset_purge = True
item1 = self.dataset_mgr.create( self.trans )
self.log( "should purge an hda if config does allow" )
self.assertFalse( item1.purged )
self.assertEqual( self.dataset_mgr.purge( self.trans, item1 ), item1 )
self.assertTrue( item1.purged )
def test_purge_not_allowed( self ):
self.trans.app.config.allow_user_dataset_purge = False
item1 = self.dataset_mgr.create( self.trans )
self.log( "should raise an error when purging an hda if config does not allow" )
self.assertFalse( item1.purged )
self.assertRaises( exceptions.ConfigDoesNotAllowException, self.dataset_mgr.purge, self.trans, item1 )
self.assertFalse( item1.purged )
##TODO: I'm unclear as to how these work, so I'm kicking this down the road a bit....
#def test_access_permission( self ):
# owner = self.user_mgr.create( self.trans, **user2_data )
# dataset = self.dataset_mgr.create( self.trans )
# # giving one user access permission makes it non-public, removing access for anyone else
# self.dataset_mgr.give_access_permission( self.trans, dataset, owner )
#
# user3 = self.user_mgr.create( self.trans, **user3_data )
# user4 = self.user_mgr.create( self.trans,
# email='[email protected]', username='******', password=default_password )
#
# self.assertTrue( self.dataset_mgr.has_access_permission( self.trans, dataset, owner ) )
# self.assertFalse( self.dataset_mgr.has_access_permission( self.trans, dataset, user3 ) )
# self.assertFalse( self.dataset_mgr.has_access_permission( self.trans, dataset, user4 ) )
#
# # should be able to progressively add more roles without removing the previous
# self.dataset_mgr.give_access_permission( self.trans, dataset, user3 )
# self.assertTrue( self.dataset_mgr.has_access_permission( self.trans, dataset, user3 ) )
# self.assertTrue( self.dataset_mgr.has_access_permission( self.trans, dataset, owner ) )
# self.assertFalse( self.dataset_mgr.has_access_permission( self.trans, dataset, user4 ) )
#
# #self.assertTrue( self.dataset_mgr.is_accessible( self.trans, dataset, owner ) )
# #self.assertFalse( self.dataset_mgr.is_accessible( self.trans, dataset, non_owner ) )
def test_accessible( self ):
owner = self.user_mgr.create( self.trans, **user2_data )
non_owner = self.user_mgr.create( self.trans, **user3_data )
dataset = self.dataset_mgr.create( self.trans )
self.log( "(by default, dataset permissions are lax) should be accessible to all" )
for user in self.user_mgr.list( self.trans ):
self.assertTrue( self.dataset_mgr.is_accessible( self.trans, dataset, user ) )
class DatasetDeserializerTestCase( BaseTestCase ):
def set_up_managers( self ):
super( DatasetDeserializerTestCase, self ).set_up_managers()
self.dataset_manager = DatasetManager( self.app )
self.dataset_serializer = DatasetSerializer( self.app )
self.dataset_deserializer = DatasetDeserializer( self.app )
self.role_manager = RoleManager( self.app )
def test_deserialize_delete( self ):
dataset = self.dataset_manager.create()
self.log( 'should raise when deserializing deleted from non-bool' )
self.assertFalse( dataset.deleted )
self.assertRaises( exceptions.RequestParameterInvalidException,
self.dataset_deserializer.deserialize, dataset, data={ 'deleted': None } )
self.assertFalse( dataset.deleted )
self.log( 'should be able to deserialize deleted from True' )
self.dataset_deserializer.deserialize( dataset, data={ 'deleted': True } )
self.assertTrue( dataset.deleted )
self.log( 'should be able to reverse by deserializing deleted from False' )
self.dataset_deserializer.deserialize( dataset, data={ 'deleted': False } )
self.assertFalse( dataset.deleted )
def test_deserialize_purge( self ):
dataset = self.dataset_manager.create()
self.log( 'should raise when deserializing purged from non-bool' )
self.assertRaises( exceptions.RequestParameterInvalidException,
self.dataset_deserializer.deserialize, dataset, data={ 'purged': None } )
self.assertFalse( dataset.purged )
self.log( 'should be able to deserialize purged from True' )
self.dataset_deserializer.deserialize( dataset, data={ 'purged': True } )
self.assertTrue( dataset.purged )
# TODO: should this raise an error?
self.log( 'should NOT be able to deserialize purged from False (will remain True)' )
self.dataset_deserializer.deserialize( dataset, data={ 'purged': False } )
self.assertTrue( dataset.purged )
def test_deserialize_permissions( self ):
dataset = self.dataset_manager.create()
who_manages = self.user_manager.create( **user2_data )
self.dataset_manager.permissions.manage.grant( dataset, who_manages )
existing_permissions = self.dataset_serializer.serialize_permissions( dataset, 'permissions', user=who_manages )
existing_manage_permissions = existing_permissions[ 'manage' ]
user3 = self.user_manager.create( **user3_data )
self.log( 'deserializing permissions from a non-dictionary should error' )
not_a_dict = []
self.assertRaises( exceptions.RequestParameterInvalidException, self.dataset_deserializer.deserialize,
dataset, user=who_manages, data={ 'permissions': not_a_dict })
self.log( 'deserializing permissions from a malformed dictionary should error' )
self.assertRaises( exceptions.RequestParameterInvalidException, self.dataset_deserializer.deserialize,
dataset, user=who_manages, data={ 'permissions': dict( nope=[], access=[] ) })
self.log( 'deserializing permissions with no manage roles should error' )
self.assertRaises( exceptions.RequestParameterInvalidException, self.dataset_deserializer.deserialize,
dataset, user=who_manages, data={ 'permissions': dict( manage=[], access=[] ) })
self.log( 'deserializing permissions using a non-managing user should error' )
self.assertRaises( rbac_secured.DatasetManagePermissionFailedException, self.dataset_deserializer.deserialize,
dataset, user=user3, data={ 'permissions': existing_permissions })
self.log( 'deserializing permissions with a single access should make the dataset private' )
private_role = self.user_manager.private_role( who_manages )
private_role = private_role.to_dict( value_mapper={ 'id' : self.app.security.encode_id } )
permissions = dict( manage=existing_manage_permissions, access=[ private_role[ 'id' ] ] )
self.dataset_deserializer.deserialize( dataset, user=who_manages, data={
'permissions': permissions
})
self.assertFalse( self.dataset_manager.is_accessible( dataset, user=user3 ) )
self.log( 'deserializing permissions manage should make the permissions available' )
self.assertRaises( SkipAttribute, self.dataset_serializer.serialize_permissions,
dataset, 'perms', user=user3 )
# now, have who_manages give a manage permission to user3
private_role = self.user_manager.private_role( user3 )
new_manage_permissions = existing_manage_permissions + [ self.app.security.encode_id( private_role.id ) ]
permissions = dict( manage=new_manage_permissions, access=[] )
self.dataset_deserializer.deserialize( dataset, user=who_manages, data={
'permissions': permissions
})
# deserializing for user3 shouldn't throw a skip bc they can manage
permissions = self.dataset_serializer.serialize_permissions( dataset, 'perms', user=who_manages )
self.assertEqual( new_manage_permissions, permissions[ 'manage' ] )
