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()
def db_calculate_foldersizes(self):
log.msg("[%s] Recursively calculating folder sizes" % self.data['resource_id'])
dirs = ['/']
while dirs:
file_path = dirs[0]
query = (
select([Files.c.file_name, Files.c.file_path, Files.c.file_size, Files.c.file_isdir])
.where(Files.c.resource_id == self.data['resource_id'])
.where(Files.c.file_path == file_path))
all = yield tx_pool.runQuery(query)
for file_name in [z.file_name for z in all if z.file_isdir]:
query = (
select([func.sum(Files.c.file_size)])
.where(Files.c.resource_id == self.data['resource_id'])
.where(Files.c.file_path.like(file_path+file_name+'%')))
size = yield tx_pool.runQuery(query)
size = size[0]
if not size.sum_1:
size = 0
else:
size = long(size.sum_1)
query = (
Files.update()
.where(Files.c.resource_id == self.data['resource_id'])
.where(Files.c.file_path == file_path)
.where(Files.c.file_name == file_name)
.where(Files.c.file_isdir == True)
.values(file_size=size))
yield tx_pool.runOperation(query)
dirs.append('%s%s/' % (file_path, file_name))
dirs.pop(0)
log.msg("[%s] Recursively calculating folder sizes DONE" % self.data['resource_id'])