sort_out_dates_for_path-o-gen.py

from __future__ import division
import datetime
import random
## inhandle is formatted 'sample_id\tdate', no header
inhandle = '/Users/flashton/Desktop/sample_dates'

def read_file(inhandle):
	res_dict = {}
	with open(inhandle) as fi:
		lines = fi.readlines()
		lines = [x.strip() for x in lines]
		for line in lines:
			split_line = line.split('\t')
			res_dict[split_line[0]] = split_line[1]
	return res_dict

def calc_year_proportion(sample_date_dict):
	## this is going to assume your data is in dd/mm/yyyy format - if it isn't you need to change lines 22 and 24 of this script
	res_dict = {}
	for each in sample_date_dict:
		# print each, sample_date_dict[each]
		split_date = sample_date_dict[each].split('/')
		split_date = [int(x) for x in split_date]
		date = datetime.date(split_date[2], split_date[1], split_date[0])
		# print (date.timetuple().tm_yday / 365)
		# print date.year
		final_date = date.year + (date.timetuple().tm_yday / 365)
		
		if str(final_date) in res_dict:
			# print
			# print each, final_date
			final_date = final_date + (random.random() / 100000)
			# print each, final_date
			# if each == '2309_H14146069601-1':
				# print final_date
			res_dict[str(final_date)] = each
		else:
			res_dict[str(final_date)] = each

	return res_dict




def main(inhandle):
	sample_date_dict = read_file(inhandle)
	# print sample_date_dict['2309_H14146069601-1']
	# print len(sample_date_dict)
	year_prop_date = calc_year_proportion(sample_date_dict)
	# print len(year_prop_date)
	for each in year_prop_date:
		print year_prop_date[each] + '\t' + each


main(inhandle)