projects.mako.cc - harrypotter-wikipedia-cdsw/blob - hpwp-trend.py

   1 import encoding_fix
   2
   3 from csv import DictReader
   4
   5 # read in the input file and count by day
   6 input_file = open("hp_wiki.tsv", 'r', encoding="utf-8")
   7
   8 edits_by_day = {}
   9 for row in DictReader(input_file, delimiter="\t"):
  10     day_string = row['timestamp'][0:10]
  11
  12     if day_string in edits_by_day:
  13         edits_by_day[day_string] = edits_by_day[day_string] + 1
  14     else:
  15         edits_by_day[day_string] = 1
  16
  17 input_file.close()
  18
  19 # output the counts by day
  20 output_file = open("hp_edits_by_day.tsv", "w", encoding='utf-8')
  21
  22 # write a header
  23 output_file.write("date\tedits\n")
  24
  25 # iterate through every day and print out data into the file
  26 for day_string in edits_by_day.keys():
  27     output_file.write("\t".join([day_string, str(edits_by_day[day_string])]) + "\n")
  28
  29 output_file.close()