def harvest_send_job_to_gather_queue(context, data_dict):
'''
Sends a harvest job to the gather queue.
:param id: the id of the harvest job
:type id: string
'''
log.info('Send job to gather queue: %r', data_dict)
job_id = logic.get_or_bust(data_dict, 'id')
job = toolkit.get_action('harvest_job_show')(
context, {'id': job_id})
check_access('harvest_send_job_to_gather_queue', context, job)
# gather queue
publisher = get_gather_publisher()
# Check the source is active
source = harvest_source_show(context, {'id': job['source_id']})
if not source['active']:
raise toolkit.ValidationError('Source is not active')
job_obj = HarvestJob.get(job['id'])
job_obj.status = job['status'] = u'Running'
job_obj.save()
publisher.send({'harvest_job_id': job['id']})
log.info('Sent job %s to the gather queue', job['id'])
return harvest_job_dictize(job_obj, context)
def harvest_jobs_run(context,data_dict):
log.info('Harvest job run: %r', data_dict)
check_access('harvest_jobs_run',context,data_dict)
source_id = data_dict.get('source_id',None)
# Check if there are pending harvest jobs
jobs = harvest_job_list(context,{'source_id':source_id,'status':u'New'})
log.info('Number of jobs: %i', len(jobs))
sent_jobs = []
if len(jobs) == 0:
log.info('No new harvest jobs.')
return sent_jobs # i.e. []
# Do not raise an exception as that will cause cron (which runs
# this) to produce an error email.
# Send each job to the gather queue
publisher = get_gather_publisher()
for job in jobs:
context['detailed'] = False
source = harvest_source_show(context,{'id':job['source']})
if source['active']:
publisher.send({'harvest_job_id': job['id']})
log.info('Sent job %s to the gather queue' % job['id'])
sent_jobs.append(job)
publisher.close()
# Record the running in harvest_status
log.info('%i jobs sent to the gather queue to be harvested', len(sent_jobs))
return sent_jobs
def harvest_jobs_run(context,data_dict):
log.info('Harvest job run: %r', data_dict)
check_access('harvest_jobs_run',context,data_dict)
source_id = data_dict.get('source_id',None)
# Check if there are pending harvest jobs
jobs = harvest_job_list(context,{'source_id':source_id,'status':u'New'})
if len(jobs) == 0:
log.info('No new harvest jobs.')
raise Exception('There are no new harvesting jobs')
# Send each job to the gather queue
publisher = get_gather_publisher()
sent_jobs = []
for job in jobs:
context['detailed'] = False
source = harvest_source_show(context,{'id':job['source']})
if source['active']:
publisher.send({'harvest_job_id': job['id']})
log.info('Sent job %s to the gather queue' % job['id'])
sent_jobs.append(job)
publisher.close()
return sent_jobs
def harvest_job_abort(context, data_dict):
'''
Aborts a harvest job. Given a harvest source_id, it looks for the latest
one and (assuming it not already Finished) marks it as Finished. It also
marks any of that source's harvest objects and (if not complete or error)
marks them "ERROR", so any left in limbo are cleaned up. Does not actually
stop running any queued harvest fetchs/objects.
:param source_id: the name or id of the harvest source with a job to abort
:type source_id: string
'''
check_access('harvest_job_abort', context, data_dict)
model = context['model']
source_id = data_dict.get('source_id')
source = harvest_source_show(context, {'id': source_id})
# HarvestJob set status to 'Finished'
# Don not use harvest_job_list since it can use a lot of memory
last_job = model.Session.query(HarvestJob) \
.filter_by(source_id=source['id']) \
.order_by(HarvestJob.created.desc()).first()
if not last_job:
raise NotFound('Error: source has no jobs')
job = get_action('harvest_job_show')(context,
{'id': last_job.id})
if job['status'] != 'Finished':
# i.e. New or Running
job_obj = HarvestJob.get(job['id'])
job_obj.status = new_status = 'Finished'
model.repo.commit_and_remove()
log.info('Harvest job changed status from "%s" to "%s"',
job['status'], new_status)
else:
log.info('Harvest job unchanged. Source %s status is: "%s"',
job['id'], job['status'])
# HarvestObjects set to ERROR
job_obj = HarvestJob.get(job['id'])
objs = job_obj.objects
for obj in objs:
if obj.state not in ('COMPLETE', 'ERROR'):
old_state = obj.state
obj.state = 'ERROR'
log.info('Harvest object changed state from "%s" to "%s": %s',
old_state, obj.state, obj.id)
else:
log.info('Harvest object not changed from "%s": %s',
obj.state, obj.id)
model.repo.commit_and_remove()
job_obj = HarvestJob.get(job['id'])
return harvest_job_dictize(job_obj, context)
def harvest_jobs_run(context, data_dict):
log.info("Harvest job run: %r", data_dict)
check_access("harvest_jobs_run", context, data_dict)
session = context["session"]
source_id = data_dict.get("source_id", None)
if not source_id:
_make_scheduled_jobs(context, data_dict)
context["return_objects"] = False
# Flag finished jobs as such
jobs = harvest_job_list(context, {"source_id": source_id, "status": u"Running"})
if len(jobs):
package_index = PackageSearchIndex()
for job in jobs:
if job["gather_finished"]:
objects = (
session.query(HarvestObject.id)
.filter(HarvestObject.harvest_job_id == job["id"])
.filter(and_((HarvestObject.state != u"COMPLETE"), (HarvestObject.state != u"ERROR")))
.order_by(HarvestObject.import_finished.desc())
)
if objects.count() == 0:
job_obj = HarvestJob.get(job["id"])
job_obj.status = u"Finished"
last_object = (
session.query(HarvestObject)
.filter(HarvestObject.harvest_job_id == job["id"])
.filter(HarvestObject.import_finished != None)
.order_by(HarvestObject.import_finished.desc())
.first()
)
if last_object:
job_obj.finished = last_object.import_finished
job_obj.save()
# Reindex the harvest source dataset so it has the latest
# status
if "extras_as_string" in context:
del context["extras_as_string"]
context.update({"validate": False, "ignore_auth": True})
package_dict = logic.get_action("package_show")(context, {"id": job_obj.source.id})
if package_dict:
package_index.index_package(package_dict)
# resubmit old redis tasks
resubmit_jobs()
# Check if there are pending harvest jobs
jobs = harvest_job_list(context, {"source_id": source_id, "status": u"New"})
if len(jobs) == 0:
log.info("No new harvest jobs.")
raise Exception("There are no new harvesting jobs")
# Send each job to the gather queue
publisher = get_gather_publisher()
sent_jobs = []
for job in jobs:
context["detailed"] = False
source = harvest_source_show(context, {"id": job["source_id"]})
if source["active"]:
job_obj = HarvestJob.get(job["id"])
job_obj.status = job["status"] = u"Running"
job_obj.save()
publisher.send({"harvest_job_id": job["id"]})
log.info("Sent job %s to the gather queue" % job["id"])
sent_jobs.append(job)
publisher.close()
return sent_jobs
def harvest_jobs_run(context,data_dict):
log.info('Harvest job run: %r', data_dict)
check_access('harvest_jobs_run',context,data_dict)
session = context['session']
source_id = data_dict.get('source_id',None)
if not source_id:
_make_scheduled_jobs(context, data_dict)
context['return_objects'] = False
# Flag finished jobs as such
jobs = harvest_job_list(context,{'source_id':source_id,'status':u'Running'})
if len(jobs):
package_index = PackageSearchIndex()
for job in jobs:
if job['gather_finished']:
objects = session.query(HarvestObject.id) \
.filter(HarvestObject.harvest_job_id==job['id']) \
.filter(and_((HarvestObject.state!=u'COMPLETE'),
(HarvestObject.state!=u'ERROR'))) \
.order_by(HarvestObject.import_finished.desc())
if objects.count() == 0:
job_obj = HarvestJob.get(job['id'])
job_obj.status = u'Finished'
last_object = session.query(HarvestObject) \
.filter(HarvestObject.harvest_job_id==job['id']) \
.filter(HarvestObject.import_finished!=None) \
.order_by(HarvestObject.import_finished.desc()) \
.first()
if last_object:
job_obj.finished = last_object.import_finished
job_obj.save()
# Reindex the harvest source dataset so it has the latest
# status
if 'extras_as_string'in context:
del context['extras_as_string']
context.update({'validate': False, 'ignore_auth': True})
package_dict = logic.get_action('package_show')(context,
{'id': job_obj.source.id})
if package_dict:
package_index.index_package(package_dict)
# resubmit old redis tasks
resubmit_jobs()
# Check if there are pending harvest jobs
jobs = harvest_job_list(context,{'source_id':source_id,'status':u'New'})
if len(jobs) == 0:
log.info('No new harvest jobs.')
raise Exception('There are no new harvesting jobs')
# Send each job to the gather queue
publisher = get_gather_publisher()
sent_jobs = []
for job in jobs:
context['detailed'] = False
source = harvest_source_show(context,{'id':job['source_id']})
if source['active']:
job_obj = HarvestJob.get(job['id'])
job_obj.status = job['status'] = u'Running'
job_obj.save()
publisher.send({'harvest_job_id': job['id']})
log.info('Sent job %s to the gather queue' % job['id'])
sent_jobs.append(job)
publisher.close()
return sent_jobs
def harvest_jobs_run(context,data_dict):
log.info('Harvest job run: %r', data_dict)
check_access('harvest_jobs_run',context,data_dict)
session = context['session']
source_id = data_dict.get('source_id',None)
if not source_id:
_make_scheduled_jobs(context, data_dict)
context['return_objects'] = False
# Flag finished jobs as such
jobs = harvest_job_list(context,{'source_id':source_id,'status':u'Running'})
if len(jobs):
for job in jobs:
if job['gather_finished']:
objects = session.query(HarvestObject.id) \
.filter(HarvestObject.harvest_job_id==job['id']) \
.filter(and_((HarvestObject.state!=u'COMPLETE'),
(HarvestObject.state!=u'ERROR'))) \
.order_by(HarvestObject.import_finished.desc())
if objects.count() == 0:
job_obj = HarvestJob.get(job['id'])
job_obj.status = u'Finished'
last_object = session.query(HarvestObject) \
.filter(HarvestObject.harvest_job_id==job['id']) \
.filter(HarvestObject.import_finished!=None) \
.order_by(HarvestObject.import_finished.desc()) \
.first()
if last_object:
job_obj.finished = last_object.import_finished
job_obj.save()
# Reindex the harvest source dataset so it has the latest
# status
get_action('harvest_source_reindex')(context,
{'id': job_obj.source.id})
# resubmit old redis tasks
resubmit_jobs()
# Check if there are pending harvest jobs
jobs = harvest_job_list(context,{'source_id':source_id,'status':u'New'})
if len(jobs) == 0:
log.info('No new harvest jobs.')
# raise Exception('There are no new harvesting jobs')
# Send each job to the gather queue
publisher = get_gather_publisher()
sent_jobs = []
for job in jobs:
context['detailed'] = False
source = harvest_source_show(context,{'id':job['source_id']})
if source['active']:
job_obj = HarvestJob.get(job['id'])
job_obj.status = job['status'] = u'Running'
job_obj.save()
publisher.send({'harvest_job_id': job['id']})
log.info('Sent job %s to the gather queue' % job['id'])
sent_jobs.append(job)
publisher.close()
return sent_jobs
def harvest_job_abort(context, data_dict):
'''
Aborts a harvest job. Given a harvest source_id, it looks for the latest
one and (assuming it not already Finished) marks it as Finished. It also
marks any of that source's harvest objects and (if not complete or error)
marks them "ERROR", so any left in limbo are cleaned up. Does not actually
stop running any queued harvest fetchs/objects.
Specify either id or source_id.
:param id: the job id to abort, or the id or name of the harvest source
with a job to abort
:type id: string
:param source_id: the name or id of the harvest source with a job to abort
:type source_id: string
'''
check_access('harvest_job_abort', context, data_dict)
model = context['model']
source_or_job_id = data_dict.get('source_id') or data_dict.get('id')
if source_or_job_id:
try:
source = harvest_source_show(context, {'id': source_or_job_id})
except NotFound:
job = get_action('harvest_job_show')(context, {
'id': source_or_job_id
})
else:
# HarvestJob set status to 'Aborted'
# Do not use harvest_job_list since it can use a lot of memory
# Get the most recent job for the source
job = model.Session.query(HarvestJob) \
.filter_by(source_id=source['id']) \
.order_by(HarvestJob.created.desc()).first()
if not job:
raise NotFound('Error: source has no jobs')
job_id = job.id
job = get_action('harvest_job_show')(context, {'id': job_id})
if job['status'] != 'Finished':
# i.e. New or Running
job_obj = HarvestJob.get(job['id'])
job_obj.status = new_status = 'Finished'
model.repo.commit_and_remove()
log.info('Harvest job changed status from "%s" to "%s"', job['status'],
new_status)
else:
log.info('Harvest job unchanged. Source %s status is: "%s"', job['id'],
job['status'])
# HarvestObjects set to ERROR
job_obj = HarvestJob.get(job['id'])
objs = job_obj.objects
for obj in objs:
if obj.state not in ('COMPLETE', 'ERROR'):
old_state = obj.state
obj.state = 'ERROR'
log.info('Harvest object changed state from "%s" to "%s": %s',
old_state, obj.state, obj.id)
else:
log.info('Harvest object not changed from "%s": %s', obj.state,
obj.id)
model.repo.commit_and_remove()
job_obj = HarvestJob.get(job['id'])
return harvest_job_dictize(job_obj, context)
def test_harvester(self):
"""
Test the harvester by running it for real with mocked requests.
We need to convert some blocks to helper functions or fixtures,
but this is an easy way to verify that a harvester does what it's
supposed to over the course of one or more runs, and we should
build on it for future tests.
"""
helpers.reset_db()
context = {}
context.setdefault('user', 'test_user')
context.setdefault('ignore_auth', True)
context['model'] = model
context['session'] = model.Session
user = {}
user['name'] = 'test_user'
user['email'] = '*****@*****.**'
user['password'] = '******'
helpers.call_action('user_create', context, **user)
org = {'name': 'test_org', 'url': 'https://www.example.com'}
owner_org = helpers.call_action('organization_create', context, **org)
config_dict = {
'source': 'esa_scihub',
'update_all': False,
'datasets_per_job': 10,
'timeout': 10,
'skip_raw': False
}
config = json.dumps(config_dict)
source = {
'url': 'http://www.scihub.org',
'name': 'scihub_test_harvester',
'owner_org': owner_org['id'],
'source_type': 'esasentinel',
'config': config
}
harvest_source_create(context, source)
source = harvest_source_show(context, {'id': 'scihub_test_harvester'})
job_dict = get_action('harvest_job_create')(context, {
'source_id': source['id']
})
job_obj = HarvestJob.get(job_dict['id'])
harvester = queue.get_harvester(source['source_type'])
with requests_mock.Mocker(real_http=True) as m:
m.register_uri('GET', '/dhus/search?q', text=self.raw_results)
lib.run_harvest_job(job_obj, harvester)
source = harvest_source_show(context, {'id': 'scihub_test_harvester'})
assert source['status']['last_job']['status'] == 'Finished'
assert source['status']['last_job']['stats']['added'] == 10
# Re-run the harvester
job_dict = get_action('harvest_job_create')(context, {
'source_id': source['id']
})
job_obj = HarvestJob.get(job_dict['id'])
harvester = queue.get_harvester(source['source_type'])
with requests_mock.Mocker(real_http=True) as m:
m.register_uri('GET', '/dhus/search?q', text=self.raw_results)
lib.run_harvest_job(job_obj, harvester)
source = harvest_source_show(context, {'id': 'scihub_test_harvester'})
assert source['status']['last_job']['status'] == 'Finished'
assert source['status']['last_job']['stats']['added'] == 0
assert source['status']['last_job']['stats']['updated'] == 0
# Re-run the harvester but force updates
config_dict = {
'source': 'esa_scihub',
'update_all': True,
'datasets_per_job': 10,
'timeout': 10,
'skip_raw': False
}
config = json.dumps(config_dict)
source['config'] = config
harvest_source_update(context, source)
job_dict = get_action('harvest_job_create')(context, {
'source_id': source['id']
})
job_obj = HarvestJob.get(job_dict['id'])
harvester = queue.get_harvester(source['source_type'])
with requests_mock.Mocker(real_http=True) as m:
m.register_uri('GET', '/dhus/search?q', text=self.raw_results)
lib.run_harvest_job(job_obj, harvester)
source = harvest_source_show(context, {'id': 'scihub_test_harvester'})
assert source['status']['last_job']['status'] == 'Finished'
assert source['status']['last_job']['stats']['added'] == 0
assert source['status']['last_job']['stats']['updated'] == 10
# Verify that the org now has 10 datasets now
org = helpers.call_action('organization_show', context,
**{'id': 'test_org'})
assert org['package_count'] == 10
def harvest_jobs_run(context, data_dict):
log.info('Harvest job run: %r', data_dict)
check_access('harvest_jobs_run', context, data_dict)
session = context['session']
source_id = data_dict.get('source_id', None)
if not source_id:
_make_scheduled_jobs(context, data_dict)
context['return_objects'] = False
# Flag finished jobs as such
jobs = harvest_job_list(context, {
'source_id': source_id,
'status': u'Running'
})
if len(jobs):
package_index = PackageSearchIndex()
for job in jobs:
if job['gather_finished']:
objects = session.query(HarvestObject.id) \
.filter(HarvestObject.harvest_job_id==job['id']) \
.filter(and_((HarvestObject.state!=u'COMPLETE'),
(HarvestObject.state!=u'ERROR'))) \
.order_by(HarvestObject.import_finished.desc())
if objects.count() == 0:
job_obj = HarvestJob.get(job['id'])
job_obj.status = u'Finished'
last_object = session.query(HarvestObject) \
.filter(HarvestObject.harvest_job_id==job['id']) \
.filter(HarvestObject.import_finished!=None) \
.order_by(HarvestObject.import_finished.desc()) \
.first()
if last_object:
job_obj.finished = last_object.import_finished
job_obj.save()
# recreate job for datajson collection or the like.
source = job_obj.source
source_config = json.loads(source.config or '{}')
datajson_collection = source_config.get(
'datajson_collection')
if datajson_collection == 'parents_run':
new_job = HarvestJob()
new_job.source = source
new_job.save()
source_config['datajson_collection'] = 'children_run'
source.config = json.dumps(source_config)
source.save()
elif datajson_collection:
# reset the key if 'children_run', or anything.
source_config.pop("datajson_collection", None)
source.config = json.dumps(source_config)
source.save()
# Reindex the harvest source dataset so it has the latest
# status
if 'extras_as_string' in context:
del context['extras_as_string']
context.update({'validate': False, 'ignore_auth': True})
package_dict = logic.get_action('package_show')(
context, {
'id': job_obj.source.id
})
if package_dict:
package_index.index_package(package_dict)
# resubmit old redis tasks
resubmit_jobs()
# Check if there are pending harvest jobs
jobs = harvest_job_list(context, {
'source_id': source_id,
'status': u'New'
})
if len(jobs) == 0:
log.info('No new harvest jobs.')
raise Exception('There are no new harvesting jobs')
# Send each job to the gather queue
publisher = get_gather_publisher()
sent_jobs = []
for job in jobs:
context['detailed'] = False
source = harvest_source_show(context, {'id': job['source_id']})
if source['active']:
job_obj = HarvestJob.get(job['id'])
job_obj.status = job['status'] = u'Running'
job_obj.save()
publisher.send({'harvest_job_id': job['id']})
log.info('Sent job %s to the gather queue' % job['id'])
sent_jobs.append(job)
publisher.close()
return sent_jobs
def distributed_harvest_jobs_run(context, data_dict):
log.info('Harvest job run: %r', data_dict)
check_access('distributed_harvest_jobs_run', context, data_dict)
session = context['session']
source_id = data_dict.get('source_id', None)
routing_key = data_dict.get('gather_routing_key', None)
exchange_name = data_dict.get('exchange_name', None)
fetch_routing_key = data_dict.get('fetch_routing_key', None)
if not source_id:
_make_scheduled_jobs(context, data_dict)
context['return_objects'] = False
# Flag finished jobs as such
jobs = harvest_job_list(context, {
'source_id': source_id,
'status': u'Running'
})
if len(jobs):
package_index = PackageSearchIndex()
for job in jobs:
if job['gather_finished']:
objects = session.query(HarvestObject.id) \
.filter(HarvestObject.harvest_job_id==job['id']) \
.filter(and_((HarvestObject.state!=u'COMPLETE'),
(HarvestObject.state!=u'ERROR'))) \
.order_by(HarvestObject.import_finished.desc())
if objects.count() == 0:
job_obj = HarvestJob.get(job['id'])
job_obj.status = u'Finished'
last_object = session.query(HarvestObject) \
.filter(HarvestObject.harvest_job_id==job['id']) \
.filter(HarvestObject.import_finished!=None) \
.order_by(HarvestObject.import_finished.desc()) \
.first()
if last_object:
job_obj.finished = last_object.import_finished
job_obj.save()
# Reindex the harvest source dataset so it has the latest
# status
if 'extras_as_string' in context:
del context['extras_as_string']
context.update({'validate': False, 'ignore_auth': True})
package_dict = logic.get_action('package_show')(
context, {
'id': job_obj.source.id
})
if package_dict:
package_index.index_package(package_dict)
# resubmit old redis tasks
resubmit_jobs()
# Check if there are pending harvest jobs
jobs = harvest_job_list(context, {
'source_id': source_id,
'status': u'New'
})
if len(jobs) == 0:
log.info('No new harvest jobs.')
raise Exception('There are no new harvesting jobs')
# Send each job to the gather queue
publisher = get_distributed_gather_publisher(exchange_name, routing_key)
sent_jobs = []
for job in jobs:
context['detailed'] = False
source = harvest_source_show(context, {'id': job['source_id']})
if source['active']:
job_obj = HarvestJob.get(job['id'])
job_obj.status = job['status'] = u'Running'
job_obj.save()
publisher.send({
'harvest_job_id': job['id'],
'exchange_name': exchange_name,
'fetch_routing_key': fetch_routing_key
})
log.info('Sent job %s to the gather queue' % job['id'])
sent_jobs.append(job)
publisher.close()
return sent_jobs
