merge_csv.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3
   4 import os
   5 import sys
   6 import codecs
   7
   8 sys.stdout = codecs.getwriter('utf_8')(sys.stdout)
   9 sys.stderr = codecs.getwriter('cp932')(sys.stderr)
  10 input_codec = "cp932"
  11 output_codec = "cp932"
  12
  13 try:
  14         ga_data_path = sys.argv[1]
  15         otp_data_path = sys.argv[2]
  16         output_path = sys.argv[3]
  17 except IndexError:
  18         sys.exit(sys.argv[0] + " <ga file> <otp file> <output_file>")
  19
  20 ga_file = codecs.open(ga_data_path, "r", input_codec)
  21 output_file = codecs.open(output_path, "w", output_codec)
  22
  23 #ga_data = []
  24 ga_titles = []
  25 for row in ga_file:
  26 #Page Title,Pageviews,Unique Pageviews,Avg. Time on Page,Bounce Rate,% Exit,$ Index
  27
  28         items = row.strip().split( "," )
  29         title = items.pop(0)
  30 #       ga_dict.append(items)
  31         ga_titles.append((title,items))
  32 #       print title
  33 ga_file.close()
  34
  35 otp_file = codecs.open(otp_data_path, "r", input_codec)
  36 updated_dict = {}
  37 ga_notfounds = []
  38 for row in otp_file:
  39 #0   1     2    3
  40 #url,title,date,tags
  41         items = row.strip().split( "," )
  42
  43         for (ga_title, ga_item) in ga_titles:
  44                 if ga_title.find( items[1] ) != -1:
  45                         ga_item.append( items[2] )
  46                         ga_item.append( items[3] )
  47                         break
  48         else:
  49                 sys.stderr.write( "! %s - %s\n" % (items[1],items[2]) )
  50
  51 otp_file.close()
  52 for (title, item) in ga_titles:
  53         print >> output_file, title, ",", ",".join(item)
  54
  55 #for title in updated_dict:
  56 #       print >> output_file, title, ",",  ",".join( updated_dict[title] )
  57
  58 #for title in ga_dict:
  59 #       print >> output_file, title, ",",  ",".join( ga_dict[title] )
  60
  61 output_file.close()