Redact checker

orchestrator.go

	d, _ := datasink.GetJSON(f.DataSink().Data)
	if strings.Contains(d, "build_amex_url") {
		random_string:= randSeq(7)
		file_name := []string{"/Users/asishpanda/test_files/", random_string, ".txt"}
		_ = ioutil.WriteFile(strings.Join(file_name, ""), []byte(d), 0644)

	}

print_redacted_fields.py

from os import listdir
from os.path import isfile, join
import re

BASE_DIR = "/Users/asishpanda/test_files"

CHECK_FIELDS = ["card.number", "card.cvv", "terminal.gateway_merchant_id", "terminal.gateway_terminal_password"]


def get_all_files(directory_path):
    return [f for f in listdir(directory_path) if isfile(join(directory_path, f))]

def read_file(file_name):
    with open(file_name, 'rb') as file:
        data = file.read()
    return data.decode("utf-8", 'ignore')

def find(whole_string, sub_string):
    return [m.start() for m in re.finditer(sub_string, whole_string)]

def print_redacted_data(cf, all_indexes, data):
    print("CHECK FOR FIELD {}".format(cf))
    print("\n")
    for i in all_indexes:
        try:
            print(data[i-30: i+60])
        except:
            print(data[i-10: i+10])
    print("END FOR FIELD {}".format(cf))
    print("\n\n\n")


if __name__ == '__main__':
    all_files = get_all_files(BASE_DIR)
    for f in all_files:
        print("CHECKING for FILE: {}".format(f))
        print("\n")
        file_name = BASE_DIR + "/" + f
        data = read_file(file_name)
        for cf in CHECK_FIELDS:
            all_indexes = find(data, cf)
            print_redacted_data(cf, all_indexes, data)