hive_to_csv

#!/usr/bin/python

# pluo, Apr 1, 2013
# PATH=$PATH:/home/dataproc/bi/lib/python # add python lib to path
# hive_to_mysql --hive_table='dim_listing_vw' --hive_db='bi' --hive_query='select count(*) from dim_listing_vw where agent_email is not null or agent_phone is not null;'
# hive_to_mysql --hive_table='lkp_state' --mysql_create_table='Y' 

import legoo
import sys
from optparse import OptionParser

def main():
  usage = """%prog [options] 
# sample: export query result to /data/tmp/dim_time.csv
[hive_to_csv --hive_node='namenode1' --hive_db='bi' --hive_table='dim_time' --hive_query='select * from dim_time limit 1000' --csv_dir='/data/tmp/']

# sample: export query result to tsv /data/tmp/dim_time.csv
[hive_to_csv --hive_node='namenode1' --hive_db='bi' --hive_table='dim_time' --csv_dir='/tmp/' --csv_file='dim_time2.csv']
"""

  # create new parser object
  parser = OptionParser(usage=usage)

  # register hive options
  parser.add_option("--hive_node", dest="hive_node",
                    help="source hive node. default: [namenode1]", default='namenode1')
  parser.add_option("--hive_db", dest="hive_db",
                    help="source hive database. default: [staging]", default='staging')
  parser.add_option("--hive_table", dest="hive_table",
                    help="source hive table name")
  parser.add_option("--hive_query", dest="hive_query",
                    help="Free form query results to be exported")
  parser.add_option("--mapred_job_priority", dest="mapred_job_priority",
                  help="OPTIONAL: map reduce job priority [VERY_HIGH, HIGH, NORMAL, LOW, VERY_LOW]",  default='NORMAL')
  parser.add_option("--csv_dir", dest="csv_dir",
                    help="dir for tsv", default='/data/tmp/')
  parser.add_option("--csv_file", dest="csv_file",
                    help="export hive [table | query results] to tsv")
  parser.add_option("-q", "--quiet", "--silent",  dest="quiet",
                    help="OPTIONAL: suppress messages to stdout. default: [N]", default='N')
  parser.add_option("-d", "--debug", dest="debug", help="OPTIONAL: debug flag [Y|N], default: [N]", default='N')

  (options, args) = parser.parse_args()

  legoo.hive_to_csv(hive_node           = options.hive_node, \
                    hive_db             = options.hive_db, \
                    hive_table          = options.hive_table, \
                    hive_query          = options.hive_query, \
                    mapred_job_priority = options.mapred_job_priority, \
                    csv_dir             = options.csv_dir, \
                    csv_file            = options.csv_file, \
                    quiet               = options.quiet, \
                    debug               = options.debug
                  )

if __name__ == '__main__':
  main()