def __init__(self, schema, last_time=None, types=None, projects=None, config=None):
self.schema = schema
self.last_time = parse_interval(last_time) if isinstance(last_time, basestring) else last_time
self.types = types
self.projects = projects
self._log_counter = itertools.count(1)
self.shotgun = get_shotgun('sgapi', config=config)
self.loop_controller = LoopController()
class Scanner(object):
def __init__(self, schema, last_time=None, types=None, projects=None, config=None):
self.schema = schema
self.last_time = parse_interval(last_time) if isinstance(last_time, basestring) else last_time
self.types = types
self.projects = projects
self._log_counter = itertools.count(1)
self.shotgun = get_shotgun('sgapi', config=config)
self.loop_controller = LoopController()
def scan(self, interval=None):
interval = parse_interval(interval) if interval else None
sleep_target = time.time()
while True:
log_globals.meta = {'scan': next(self._log_counter)}
scan_start = datetime.datetime.utcnow() # would be great if this matched the sleep target
self._scan()
self.last_time = scan_start - datetime.timedelta(seconds=1) # need some leeway
if not interval:
break
# Figure out the next time
while sleep_target < time.time():
sleep_target += interval
delay = sleep_target - time.time()
log.info('sleeping %ds until next scan' % delay)
self.loop_controller.sleep(delay)
def _scan(self):
base_filters = []
if self.last_time:
if isinstance(self.last_time, (int, float)):
self.last_time = datetime.datetime.utcnow() - datetime.timedelta(seconds=self.last_time)
base_filters.append(('updated_at', 'greater_than', self.last_time))
log.info('scan starting')
counts = {}
for entity_type in sorted(self.schema._entity_types.itervalues(), key=lambda e: e.type_name):
if self.types and entity_type.type_name not in self.types:
continue
#log.info('scanning %ss' % entity_type.type_name)
filters = base_filters[:]
if self.projects and entity_type.type_name not in ('ApiUser', 'HumanUser'):
# Need to make sure the project field actually exists.
project_field = entity_type.fields['project']
if project_field and project_field.is_cached():
filters.append(('project', 'in', [{'type': 'Project', 'id': pid} for pid in self.projects]))
return_fields = sorted(name for name, field in entity_type.fields.iteritems() if field.is_cached())
for entity in self._find_active_and_retired(entity_type.type_name, filters, return_fields, threads=1, per_page=100):
for key in return_fields:
value = entity.get(key)
if isinstance(value, datetime.datetime):
entity[key] = value.isoformat() + 'Z'
log.info('updating %s %s %d%s' % (
'active' if entity['_active'] else 'retired',
entity['type'],
entity['id'],
' "%s"' % entity['name'][:40] if entity.get('name') else ''
))
self.schema.create_or_update(entity.pop('type'), entity, create_with_id=True)
counts[entity_type.type_name] = counts.get(entity_type.type_name, 0) + 1
summary = ', '.join('%d %ss' % (count, name) for name, count in sorted(counts.iteritems()) if count)
log.info('scan completed; updated %s' % (summary or 'nothing'))
def _find_active_and_retired(self, *args, **kwargs):
for active in True, False:
kwargs['retired_only'] = not active
for e in self.shotgun.find(*args, **kwargs):
e['_active'] = active
yield e
