# NOTE: Before submission, change default/nargs values!!!!!!!

# This script detects backdating in a CSV log
# To view script input and options use -h

__author__ = 'XX'
__date__ = '20190419'
__version__ = '1.0'

# Import the necessary packages
import argparse
import os
from dateutil.parser import parse

# Declare variable to count detections of backdating
detection_counter = 0

# Read commandline arguments
ap = argparse.ArgumentParser()
ap.add_argument("lf", nargs='?', default='logfile.csv',
                help='Specify file path to CSV log to check')
ap.add_argument('-n', type=int, default=0,
                help='To sort log file first, provide column number of column that contains the record numbers')
ap.add_argument('-t', type=int, default=3,
                help='Column number of column that contains record timestamps. Default is column 1')
ap.add_argument('-s', action='store_true',
                help='Set flag if log has a header')
args = vars(ap.parse_args())

# Create non-array variables with values of parsed arguments
fp_log_file = args['lf']
column_seq_nr = args['n']
column_timestamp = args['t']
header = args['s']

# Determine first record in the list
if header:
    first_record_in_list = 1
else:
    first_record_in_list = 0

# Log file file path sanity checks
if fp_log_file is not None and not os.path.isfile(fp_log_file):
    print("ERROR: Log file specified does not exist.")
    exit(1)
if not fp_log_file.endswith('.csv'):
    print("ERROR: Log file specified is not a .csv file.")
    exit(1)

# Copy entries from the log file to an array
log_file = open(fp_log_file)
record_list = log_file.read().splitlines()
log_file.close()

# When required, sort log file on sequence number (does not work yet)
if column_seq_nr != 0:
    log_list = sorted(record_list, key=lambda row: float(row[column_seq_nr - 1].strip(',')))
    print('INFO: Log sorted on column', column_seq_nr)

# Get first record in list for the first check in the loop
previous_record = record_list[first_record_in_list]

print('INFO: Checking if timestamps increase in column', column_timestamp)

# Loop through entries in the log file, start at second record
for record in record_list[first_record_in_list + 1:]:

    # Timestamp column validity check
    try:
        timestamp = parse(record.split(',')[column_timestamp - 1])
        previous_timestamp = parse(previous_record.split(',')[column_timestamp - 1])
    except ValueError:
        print('ERROR: Column', column_timestamp, 'does not contain valid timestamp values')
        exit(1)

    # Check for the actual backdating
    if timestamp < previous_timestamp:
        detection_counter += 1
        print('[', detection_counter, '] BACKDATING DETECTED')
        print('Clock was set back between following records:')
        print(previous_record)
        print(record)

    # Safe current record to check next record against
    previous_record = record

# Print summary
if detection_counter == 0:
    print('DONE: NO BACKDATING DETECTED')
    exit(0)
else:
    print('DONE: BACKDATING WAS DETECTED', detection_counter, 'TIMES')
    exit(0)