Manipulate GitHub Archive data downloaded from Google BigQuery in order to make a graph in Excel. Transforms CSV of form: [watchers, repository_name, parsed_date] to form: [date, backbone count, angular.js count, ember.js count, knockout count]

gistfile1.py

import csv

names = sorted(['backbone', 'angular.js', 'ember.js', 'knockout'])

with open('in.csv') as csvfile:
  reader = csv.reader(csvfile, delimiter=',')
  dates = {}

  next(reader)
  for row in reader:
    count = row[0]
    framework = row[1]
    date = row[2]
    if date not in dates:
      dates[date] = {}

    dates[date][framework] = count


  header = 'date'
  for n in names:
    header += ',' + n + ''
  header += ''
  print(header)
  for date in sorted(dates):
    for n in names:
      if n not in dates[date]:
        dates[date][n] = ''
    line = '' + date + ''
    for name in names:
      line += ',' + dates[date][name] + ''
    print(line)