#!/usr/bin/env python
#
# Method 1:
# Use this software to parse dataverse access log to locate annon file study
# download	
# Method 2:
# cat server_access_log.TIMESTAMP.txt | grep 200 | grep -i download | wc -l
#
# 20131121
# Minor fix, now it will take access log files with extra empty lines
##############################################################################

author = "YXNjaGVuYmFjaEBnbWFpbC5jb20="

import sys
import re
import base64


def ingestData():
	temp = []
	f=open('dvnAccess.txt')
	line = f.readline()
	while line != "\n":
		temp.append(line)
		line = f.readline()

	return temp

def processData(temp):
	p = re.compile('FileDownload')

	stor = []
	ans, mess = "", ""
	for a in temp:
		fields = a.split(' ')
		mess = fields[5]
		ans = re.search(p, mess)
		if ans != None:
			stor.append(a)
	
	print stor

def main():
	t = []
	t = ingestData()
	processData(t)

main()