file_sum.py

#!/usr/bin/env python

# Calculate sum of files older than X date
# File is large csv w/ format: filename,date time:time,x.xxx(mb)
# -- Cole Hocking

import csv
from datetime import datetime, timedelta
from sys import argv, exit
import os.path as opath

def validateFile(filei):
    """
    Check for file_exists
    """
    if opath.isfile(filei):
        return True
    else:
        print("%r not found" % (filei))
        exit(1)

def getFileData(file_name):
    """
    Calculates Files >= 7 years and their total size
    """
    #timedelta does not take years arg; arg=7*365
    seven_year_mkr = datetime.today() - timedelta(days=2555)
    #print(seven_year_mkr)
    #size of the files
    size_total = 0.000
    #number of files
    file_sum = 0
    line_no = 0
    with open(file_name, "r") as csvFile:
        csv_reader = csv.reader(csvFile, delimiter='|')
        for row in csv_reader:
            line_no += 1
            #print(line_no)
            last_accessed = datetime.strptime(row[1], '%m/%d/%Y %H:%M:%S')
            #greater than means younger than the marker
            if last_accessed > seven_year_mkr:
                continue
            #conditions here for files OLDER||= seven_year_mkr
            else:
                # strips commas for float normalization
                size_total += float(row[2].replace(',',''))
                file_sum += 1

    print("# of files >= 7 years: %d" % file_sum)
    print("Size of files >= 7 years: %r mb" % size_total)

def main(argv):

    if len(argv) != 2:
        print("Usage: ./file_sum.py <file-to-parse>")
        exit(1)

    assert(validateFile(argv[1]))
    getFileData(argv[1])

if __name__ == "__main__":
    main(argv)