add_date_to_log.py

from convert_to_datetime import convert_to_datetime


def add_date_to_log(line):
    """ add YYMMDDhh to the beginning of the log.
    Argument:
    line (hadoop log line)
    """
    arr = line.rstrip().split()
    date_string = ' '.join(arr[0:2])
    dt = convert_to_datetime(date_string)
    hour_tag = "{0}{1}{2}".format(dt.year, dt.day, dt.hour)
    return "{0} {1}".format(hour_tag, line)

if __name__ == '__main__':
    nn_line = "2014-01-17 12:21:44,327 INFO org.apache.hadoop.hdfs.StateChange: BLOCK* addStoredBlock: blockMap updated: 192.168.0.1:50010 is added to blk_-2607434453651253239_8145228 size 125334"
    print("NN log: {0}".format(nn_line))
    print("add_yeardayhour_to_log: {0}".format(add_yeardayhour_to_log(nn_line)))