def _preprocess(self, return_params, kwargs):
prefix = 's3://'
aws_storage_bucket_name = auxiliary.read_option(self.config, 'aws', 'aws_storage_bucket_name')
aws_storage_head = os.path.join(prefix, aws_storage_bucket_name)
s3_app_input = auxiliary.read_option(self.config, 'preprocessor', 's3_app_input')
s3_app_output = auxiliary.read_option(self.config, 'preprocessor', 's3_app_output')
s3_app_scripts = auxiliary.read_option(self.config, 'preprocessor', 's3_app_scripts')
s3cfg_path = os.path.join(self.CLAUDE_ROOT, auxiliary.read_option(self.config, 'preprocessor', 's3cfg'))
group_id = kwargs['group_id']
input_param = kwargs['input_param']
producer = messaging.MessageProducer(
messaging.buildRabbitMQConnectionLink(address=return_params['return_address']),
return_params['return_queue'],
)
scripts_dir = scripts.__path__[0]
with auxiliary.TemporaryDirectory() as tmp_dir:
input_s3_path_full = os.path.join(aws_storage_head, kwargs['input_file'])
input_arch_name = os.path.basename(input_s3_path_full)
input_arch_path = os.path.join(tmp_dir, input_arch_name)
s3tools.s3cmdGET(s3cfg_path, input_s3_path_full, input_arch_path)
ready_archs = self._handle_local_arch(input_arch_path)
for arch in ready_archs:
app_type = AppParamsProvider(arch['app_type'], input_param)
script_name = app_type.getAppScriptName()
script_path = os.path.join(scripts_dir, script_name)
script_s3_path_full = os.path.join(*[aws_storage_head, s3_app_scripts, script_name])
s3tools.s3cmdPUT(s3cfg_path, script_path, script_s3_path_full)
app_input_path = arch['arch_path']
app_input_name = os.path.basename(app_input_path)
app_input_s3_path = os.path.join(s3_app_input, app_input_name)
app_input_s3_path_full = os.path.join(aws_storage_head, app_input_s3_path)
s3tools.s3cmdPUT(s3cfg_path, app_input_path, app_input_s3_path_full)
app_output_s3_path = os.path.join(s3_app_output, 'output_' + app_input_name)
app_output_s3_path_full = os.path.join(aws_storage_head, app_output_s3_path)
app_params = {
'app_input_file' : app_input_s3_path_full,
'app_output_file' : app_output_s3_path_full,
'app_script' : script_s3_path_full,
}
app_params.update(app_type.getAppSpecificParams())
message = messaging.createMessage(
'preprocessed',
group_id=group_id,
app_type=arch['app_type'],
app_name=arch['app_name'],
app_input_file=app_input_s3_path,
app_output_file=app_output_s3_path,
app_params=app_params,
)
producer.publish(message)
message = messaging.createMessage(
'preprocessing_completed',
input_id=kwargs['input_id'],
group_id=group_id,
)
producer.publish(message)
def main(config):
ts_start = datetime.now()
logger = auxiliary.getLogger()
s3cfg = auxiliary.read_option(config, 'general', 's3cfg')
app_input_s3_path_full = auxiliary.read_option(config, 'general', 'app_input_file')
app_output_s3_path_full = auxiliary.read_option(config, 'general', 'app_output_file')
licence = auxiliary.read_option(config, 'general', 'licence')
grok = auxiliary.read_option(config, 'general', 'grok')
hgs = auxiliary.read_option(config, 'general', 'hgs')
current_dir = os.getcwd()
app_input_name = os.path.basename(app_input_s3_path_full)
app_input_path = os.path.join(current_dir, app_input_name)
app_output_name = os.path.basename(app_output_s3_path_full)
app_output_path = os.path.join(current_dir, app_output_name)
logger.info('Starting %s script...\n' % (__name__))
logger.info('Configuration options:')
logger.info('Current directory: %s'% (current_dir))
logger.info('Input archive location: %s'% (app_input_s3_path_full))
logger.info('Output archive location: %s'% (app_output_s3_path_full))
logger.info('GROK location: %s'% (grok))
logger.info('HGS location: %s\n'% (hgs))
logger.info('Downloading input archive')
s3tools.s3cmdGET(s3cfg, app_input_s3_path_full, app_input_path, shellOutput=False)
hgs_file = auxiliary.seekZipForElement(app_input_path, AppTypes.HGS)[0]
working_dir = os.path.join(current_dir, os.path.split(hgs_file)[0])
root_dir = os.path.join(current_dir, auxiliary.getRootDir(hgs_file))
logger.info('Working directory: %s'% (working_dir))
logger.info('Extracting input archive...')
auxiliary.extractZip(app_input_path, current_dir)
os.chdir(working_dir)
cmd = '%s 1>&1 | tee -a %s' %(grok, 'grok.log')
logger.info('Executing GROK: %s' % (cmd))
auxiliary.execute(cmd)
cmd = '%s 1>&1 | tee -a %s' %(hgs, 'hgs.log')
logger.info('Executing HGS: %s' % (cmd))
auxiliary.execute(cmd)
os.chdir(current_dir)
logger.info('Creating output archive...')
auxiliary.createZip(app_output_path[:-4], root_dir)
logger.info('Uploading output archive')
#while not s3tools.s3cmdLS(s3cfg, app_output_s3_path_full, shellOutput=False):
s3tools.s3cmdPUT(s3cfg, app_output_path, app_output_s3_path_full, shellOutput=False)
logger.info('Done.')
ts_end = datetime.now()
ts_dif = ts_end - ts_start
print 'GROK & HGS execution + claud I/O operations took: %s seconds' % ts_dif.total_seconds()
print 'Executed on machine: %s' % messaging.getIPAdreess()
