def db_prepare(self):
# @TODO: race condition - use transactions (pool.runInteraction())
query = (select([Resources]).where(Resources.c.name == self.data['name']))
resource = yield tx_pool.runQuery(query)
if not resource:
query = (ResourceMeta.insert().values(
recursive_sizes=self.data['options']['recursive_foldersizes'],
web_user_agent=self.data['options']['user-agent'],
auth_user=self.data['options']['auth_user'],
auth_pass=self.data['options']['auth_pass']
).returning(ResourceMeta.c.id))
meta_id = yield tx_pool.runQuery(query)
meta_id = meta_id[0].id
query = (Resources.insert().values(
name=self.data['name'],
address=self.data['address'],
port=self.data['options']['port'],
protocol=FileProtocols().id_by_name(self.data['method']),
display_url=self.data['options']['display_url'],
date_crawl_start=datetime.now(),
basepath=self.data['basepath'],
meta_id=meta_id
).returning(Resources.c.id))
resource_id = yield tx_pool.runQuery(query)
self.data['resource_id'] = resource_id[0].id
query = (select([Resources]).where(Resources.c.id == self.data['resource_id']))
resource = yield tx_pool.runQuery(query)
resource = resource[0]
resource_meta = yield tx_pool.runQuery(select([ResourceMeta]).where(ResourceMeta.c.id == resource.meta_id))
resource_meta = resource_meta[0]
if resource_meta.busy:
raise Exception('This resource is already being crawled')
self.data['resource_id'] = resource.id
self.data['resource_meta_id'] = resource.meta_id
self.data['resource'] = resource
self.data['resource_meta'] = resource_meta
self.db_busy_crawling_toggle()
query = (Resources.update().where(Resources.c.id == resource.id).values(date_crawl_start=datetime.now()))
yield tx_pool.runOperation(query)
query = (Resources.update().where(Resources.c.id == resource.id).values(basepath=self.data['basepath']))
yield tx_pool.runOperation(query)
query = (Resources.update().where(Resources.c.id == resource.id).values(address=self.data['address']))
yield tx_pool.runOperation(query)
def db_finalize(self):
query = (Resources.update().where(Resources.c.id == self.data['resource_id']).values(date_crawl_end=datetime.now()))
yield tx_pool.runOperation(query)
query = (Files.delete().where(Files.c.resource_id == self.data['resource_id']))
yield tx_pool.runOperation(query)
query = (Files.update().where(Files.c.resource_id == '-%s' % str(self.data['resource_id'])).values(resource_id=self.data['resource_id']))
yield tx_pool.runOperation(query)
query = (ResourceMeta.update().where(ResourceMeta.c.id == self.data['resource_meta_id']).values(file_count=self.resource.db_files_inserts))
yield tx_pool.runOperation(query)
if 'recursive_foldersizes' in self.data['options']:
yield self.db_calculate_foldersizes()
self.db_calculate_filedistribution(resource_id=self.data['resource_id'])
self.db_busy_crawling_toggle()
