gistfile1.txt

#!/usr/bin/env python

from datetime import datetime
from os import listdir
import re

dailyTotals = {}
timeStamps = []
logDir = "/Users/thomascate/s3logs"

logFiles = listdir(logDir)
for logFile in logFiles:
  fullPath = logDir + "/" + logFile
  fileObj = open(fullPath, 'r')
  fileContents = fileObj.readlines()
  for logLine in fileContents:

    splitLine = logLine.split(' ',)
    if re.search("REST.GET.OBJECT", logLine) and splitLine[14] != "-":
      bytes = splitLine[14]

      if logFile[0:7] in dailyTotals:
        dailyTotals[logFile[0:7]] = dailyTotals[logFile[0:7]] + int(bytes)
      else:
        dailyTotals[logFile[0:7]] = int(bytes)

  fileObj.close

for timeStamp in dailyTotals.iterkeys():
  timeStamps.append(timeStamp)

for timeStamp in sorted(timeStamps):
  print timeStamp, dailyTotals[timeStamp]/1024/1024/1024, "GB"