#!/usr/bin/python

import re
import subprocess
import sys

remove_local1 = "rm -f /tmp/morphlines.conf"

process = subprocess.Popen(remove_local1.split(), stdout=subprocess.PIPE)
output, error = process.communicate()

hdfs_move_from_hdfs = "hdfs dfs -getmerge /tmp/oracle/metadata/columns /tmp/col.txt"
hdfs_move_to_hdfs = "hdfs dfs -moveFromLocal /tmp/morphlines_oracle.conf /tmp/oracle/"


#hdfs_move_from_hdfs = "hdfs dfs -getmerge " + \
#    str(sys.argv[1]) + " /tmp/col.txt"

#hdfs_move_to_hdfs = "hdfs dfs -moveFromLocal /tmp/new_morphlines.conf " + \
#    str(sys.argv[2])



#hdfs_move_from_hdfs_morphlines = "hdfs dfs -get " + \
#    str(sys.argv[3]) + " /tmp/morphlines.txt"


hdfs_move_from_hdfs_morphlines = "hdfs dfs -get /user/admin/scripts/morphlines_oracle.conf /tmp/morphlines.conf"



remove_local = "rm -f /tmp/col.txt"

process = subprocess.Popen(hdfs_move_from_hdfs.split(), stdout=subprocess.PIPE)
output, error = process.communicate()

process = subprocess.Popen(hdfs_move_from_hdfs_morphlines.split(), stdout=subprocess.PIPE)
output, error = process.communicate()


f = open('/tmp/col.txt', 'r')
n = f.read()
b = n.split()
f.close()

if 'database_ip' not in b:
    b.append('database_ip')

if 'database_name' not in b:
    b.append('database_name')

if 'id' not in b:
    b.append('id')

if 'table_name' not in b:
    b.append('table_name')

c = sorted(b)
p = []
for word in c:
    line = '            ' + word + ' : /' + word
    p.append(line)

big_line = '\n'.join(p)

morph = open('/tmp/morphlines.conf', 'r+')
morph_readed_file = morph.read()

find_peace = re.findall(r'paths : {\n([\S\s]*?)\n.*}', morph_readed_file)

new_morph = re.sub(find_peace[0], big_line, morph_readed_file)

new_file = open('/tmp/morphlines_oracle.conf', 'w+')
new_file.write(new_morph)
new_file.close()

process = subprocess.Popen(hdfs_move_to_hdfs.split(), stdout=subprocess.PIPE)
output, error = process.communicate()

process = subprocess.Popen(remove_local.split(), stdout=subprocess.PIPE)
output, error = process.communicate()