merge_csv.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3
   4 import os, sys
   5 import codecs
   6
   7 sys.stdout = codecs.getwriter('utf_8')(sys.stdout)
   8 sys.stderr = codecs.getwriter('utf_8')(sys.stderr)
   9
  10 try:
  11         ga_data_path = sys.argv[1]
  12         otp_data_path = sys.argv[2]
  13 except IndexError:
  14         sys.exit(sys.argv[0] + " <ga file> <otp file>")
  15
  16 ga_file = codecs.open(ga_data_path, "r", "utf_8")
  17
  18 ga_dict = {}
  19 ga_titles = []
  20 for row in ga_file:
  21 #Page Title,Pageviews,Unique Pageviews,Avg. Time on Page,Bounce Rate,% Exit,$ Index
  22
  23         items = row.strip().split( "," )
  24         title = items.pop(0)
  25         ga_dict[title] = items
  26         ga_titles.append(title)
  27 #       print title
  28 ga_file.close()
  29
  30 otp_file = codecs.open(otp_data_path, "r", "utf_8")
  31 updated_dict = {}
  32 ga_notfounds = []
  33 for row in otp_file:
  34 #0   1     2      3   4        5    6
  35 #url,title,editor,PVs,comments,date,tags
  36         items = row.strip().split( "," )
  37
  38         for ga_title in ga_titles:
  39                 if ga_title.find( items[1] ) != -1:
  40                         ga_info = ga_dict.pop(ga_title)
  41                         ga_info.append( items[5] )
  42                         ga_info.append( items[6] )
  43                         updated_dict[ga_title] = ga_info
  44                         break
  45         else:
  46                 sys.stderr.write( "! %s - %s\n" % (items[1],items[5]) )
  47
  48 otp_file.close()
  49 for title in updated_dict:
  50         print title, ",",  ",".join( updated_dict[title] )
  51
  52 for title in ga_dict:
  53         print title, ",",  ",".join( ga_dict[title] )