issue_monthly.py

import datetime as datetime
import os 

def organize_monthly(path):
	opened_per_month_path = path + 'opened-per-month/'
	closed_per_month_path = path + 'closed-per-month/'
	
	if not os.path.exists(closed_per_month_path):
		os.makedirs(closed_per_month_path)
		
	if not os.path.exists(opened_per_month_path):
		os.makedirs(opened_per_month_path)
				
	for file in os.listdir(path):
		if file.endswith('.txt'):
			this_file = open(os.path.join(path,file))
			this_data = this_file.readlines()[1:]
			opened,closed = collect_issues(this_data)
			organize_and_write_file(opened,opened_per_month_path,str(file))	
			organize_and_write_file(closed,closed_per_month_path,str(file))
	
def collect_issues(this_data):
	open_list = []
	close_list = []
		
	for line in this_data:
		line = line.strip()
		line = line.split(',') 
		open_raw  = line[4]
		close_raw = line[5]
		
		open_value = open_raw.split('/')
		open_date  = datetime_it(open_value[2],open_value[1],open_value[0])
		open_list.append(open_date)
		
		if '-' in close_raw:
			close_raw_dates = close_raw.split('-')
			for date in close_raw_dates:
				this_close_value = date.split('/')
				this_close_date = datetime_it(this_close_value[2],this_close_value[1],this_close_value[0])
				close_list.append(this_close_date)
		else:
			if not 'None' in close_raw:
				close_value = close_raw.split('/')
				close_date = datetime_it(close_value[2],close_value[1],close_value[0])
				close_list.append(close_date)
	return open_list,close_list	

def datetime_it(day,month,year):
	date = datetime.datetime(int(day),int(month),int(year))
	return date

def organize_and_write_file(data,folder_path,file_name):
	this_file = open(os.path.join(folder_path + file_name),'w')
	
	years = []
	for date in data: 
		if date.year not in years:
			years.append(date.year)
			
	months = [[list() for month in range(12)] for year in range(len(years))]
	
	for current_year, current_month_list in zip(years, months):
		for current_month_number, month in zip(range(1,13), current_month_list): 
			for date in data:
				if date.year == current_year:
					if date.month == current_month_number:
						month.append(date.day)
						
	for year,month in zip(years,months):
		this_file.write(str(year))
		this_file.write('\n')
		for month_number, month_data in zip(range(1,13), month):
			this_file.write(str(month_number) + '->' + str(len(month_data)))  
			this_file.write('\n')

issue_folder_path = raw_input('Please specify the folder where are the data generated by IssueSpider.py:')
organize_monthly(issue_folder_path)