MailtoDataEncode.py

import json
import os, sys
import hashlib
import re

from mailto import mailto
from MailtoManage import MailtoManage
MAXIMUM_BLOCKSIZE_TO_READ = 65535
EMAIL_ADDRESS_REGEX = "\A[a-z0-9!#$%&'*+/=?^_`{|}~-]+(?:\.[a-z0-9!#$%&'*+/=?^_`{|}~-]+)*@(?:[a-z0-9](?:[a-z0-9-]*[a-z0-9])?\.)+[a-z0-9](?:[a-z0-9-]*[a-z0-9])?\z"

# description: generate the "emaildata_lookup.json" file: 
# {
#     "00018744d84b8ae35942b3b670d5fb734cd2eac2bf4c2e09ffba916472f3b018":"tim.barton@tabcorp.com.au",
#     "014b64ed9d69d16cf8d9c4c170c9502d6b137e1d8c7b4e51ff43638b756d486f":"henrietta.thomas@tabcorp.com.au",
#     "01bbb7fccf0ff2e7cfc09a1cd1ce04de8ed4c40ff24b11e063b8cd577e3202c1":"paul.moore@scientificgames.com",
#     "02866c1cfc9c619ed90f31a474d3726ee7653b2fa3b2d0586211dd8425086d59":"trose@bytecraft.com.au",
#     "02a56382213dd53ebd4db85b5cbc70861d9d4c46388a37f222f1003e628298b5":"clowe@agtslots.com",
#     "041663e3349b3dde0a259efd171839a814ea170c97caa89f9e239b09b660b7ed":"operational.compliance@tattsgroup.com",
#     "04a1ede4597722a1318119d768ebf75fc0088b57e7a1189d24da7af2ffc1984b":"dugald.peacock@tabcorp.com.au",
#     "06a7d77988c4bf00867f5559b5dde67e0107645e69e159deca5149d8146ca780":"dlstaffrecords@tattsgroup.com",
# ...
# }

# As well as the 'emaildata_v2.json' file: 
# {
#     "Car Booking":[
#         "8c165a9c8d5a7362b45c51c47eb09158086fa579eaaae2620c7572881707ca38"
#     ],
#     "Car Booking_cc":[
#         "e6a027809eaf55716e052e4439953574d519b881bbfdd81487f998f25804eb1b"
#     ],
#     "Casino Reef":[
#         "c2c28fe5615f9a842cfa3dc7d0274d231177558a33b78625f31de954486c510c",
#         "12d09858d17f2daaa7b4b3443a73a2bc039e94e5e2b69dc21e1dce888b522416"
#     ],
#     "Casino Reef Datafiles":[
#         "c2c28fe5615f9a842cfa3dc7d0274d231177558a33b78625f31de954486c510c",
#         "12d09858d17f2daaa7b4b3443a73a2bc039e94e5e2b69dc21e1dce888b522416"
#     ],
# ...
# }

# output: 2 x files named: "sample_emaildata_lookup.json" and "sample_emaildata_v2.json"
# you will need to do a diff/compare with the exisiting lookup and emaildata files

# input: file is the file generated by the 'iNetToMailtoFormat.py' script. by default this is called "sample_converted.json"

class MailtoDataEncode: 

    def __init__(self, fname):
        self.filename = fname
        # Generate the Loookup File for Hashes : Email Addresses
        self.json_data = mailto.ReadJSONfile(self, self.filename) 
        self.email_address_hashed = self.ProcessEmailGroups(self.json_data)
        print(json.dumps(self.email_address_hashed, sort_keys=True, indent=4, separators=(',',':'))) # write to disk. 
        self.WriteDatatoFile(self.email_address_hashed, "sample_emaildata_lookup.json")
        
        # Converts emaildata.json to replace email addresses with Hashes
        self.emailgroups_data_w_hash = self.ConvertEmailData(self.json_data, self.email_address_hashed) 
        print(json.dumps(self.emailgroups_data_w_hash, sort_keys=True, indent=4, separators=(',',':'))) # write to disk.         
        self.WriteDatatoFile(self.emailgroups_data_w_hash, "sample_emaildata_v2.json")

        # Note: Let MailtoManage generate emaildata_v2.json_sigs.json file. 

        # Test Retrieves email address for the following email group. 
        test_mail_group = "OLGR Casino Inspectorate - Star Casino - Brisbane" 
        assert(self.GetEmailAddressListString_from_EmailGroup(test_mail_group) == 'brisbanesouth@justice.qld.gov.au')
    
    def GetEmailAddressListString_from_EmailGroup(self, email_group_str): 
        email_hashes = self.emailgroups_data_w_hash[email_group_str]
        email_addresses = list() 
        for h in email_hashes: 
            email_addresses.append(self.email_address_hashed[h]) 
        
        return ";".join(email_addresses)
      
    def ProcessEmailGroups(self, jd): 
        email_address_dict = dict()
        for _,email_list in jd.items(): 
            for eaddr in email_list:              
                
                # string ';' in email address
                eaddr = re.sub(';', '', eaddr.lower())                
                # filter out just email address
                match = re.findall(r'[\w\.-]+@[\w\.-]+', eaddr) 

                hash_str = self.HashStr(match[0]) # hash the matched string            
                email_address_dict[hash_str] = match[0] # update
            
        # print(json.dumps(email_address_dict, sort_keys=True, indent=4, separators=(',',':'))) # write to disk. 
        return(email_address_dict) 

    def WriteDatatoFile(self, data, fname):
        with open(fname,'w+') as json_file:
            json.dump(data, json_file, sort_keys=True, indent=4, separators=(',',':'))
        # messagebox.showinfo("Backup Complete", "Backup of " + self.filename + ", has been saved as: " + output_backup_fname)

    def ConvertEmailData(self, email_data, email_hash_data):
        new_email_data = dict() 
        
        for email_group, email_list in email_data.items(): 
            email_hash_list = list()
            for email in email_list: 
                # string ';' in email address
                email = re.sub(';', '', email.lower())
                
                # filter out just email address
                match = re.findall(r'[\w\.-]+@[\w\.-]+', email) 
                
                if len(match) > 0: 
                    for hash, hashed_email in email_hash_data.items(): 
                        if hashed_email == match[0]: 
                            email_hash_list.append(hash) 
            
                    new_email_data[email_group] = sorted(email_hash_list)

        # print(json.dumps(new_email_data, sort_keys=True, indent=4, separators=(',',':'))) # write to disk. 
        return new_email_data
        
    def HashStr(self, s): 
        hash_obj = hashlib.sha256(str(s).encode('utf-8'))        
        return hash_obj.hexdigest()

def main():
    input_file = "sample_converted.json"
    # input_file is a JSON file format that needs to be formatted as a dict 
    # to support mailto script. 
    app = MailtoDataEncode(input_file)

if __name__ == "__main__": main()