def main(args):
    pargs = parse_args(args)
    logging_setup(pargs.verbose, get_local_path(pargs.dataset_name,'result','result.log'))
    
    print '[STARTED]'
    local_extract(pargs.dataset_name, pargs.extractor, 
                  pargs.timeout, pargs.retry_failed, pargs.skip_existing)
    print '[DONE]'
def _verify_args(args):
    # verify arguments provoded by argparse and
    # return the path to the output directory
    
    # printing arguments
    print 'dataset type: %s' % args.dataset_type
    print 'dataset name: %s' % args.dataset_name
    
    #validate dataset name
    if not check_local_path(args.dataset_name):
        print 'error: this dataset does not exist'
        sys.exit(-1)
    
    # validate path argument
    if args.path and not os.path.exists(args.path):
        print 'error: path does not exist'
        sys.exit(-1)
        
    output_dir = args.path or get_local_path(args.dataset_name)
    print 'output directory: %s' % output_dir
    return output_dir
def _verify_args(args):
    # verify arguments provoded by argparse and
    # return the path to the output directory

    # printing arguments
    print 'dataset type: %s' % args.dataset_type
    print 'dataset name: %s' % args.dataset_name

    #validate dataset name
    if not check_local_path(args.dataset_name):
        print 'error: this dataset does not exist'
        sys.exit(-1)

    # validate path argument
    if args.path and not os.path.exists(args.path):
        print 'error: path does not exist'
        sys.exit(-1)

    output_dir = args.path or get_local_path(args.dataset_name)
    print 'output directory: %s' % output_dir
    return output_dir
 def __init__(self, output_dir, dataset_name):   
     self.dataset_name = dataset_name
     self._dataset_dir = get_local_path(dataset_name)
     self._output_dir = output_dir
     self.meta_data_list = [] # list to be serialized
 def __init__(self, output_dir, dataset_name):
     self.dataset_name = dataset_name
     self._dataset_dir = get_local_path(dataset_name)
     self._output_dir = output_dir
     self.meta_data_list = []  # list to be serialized