import json import random import string import pandas as pandas import xlsxwriter from src.config import CONTACT_LIST_FILE from src.db.mongo_manager import MONGO_STORE_MANAGER from src.pojo.contact_pojo import ContactPojo from src.pojo.mail.mail_pojo import MailAddress from src.utils.generate_random_passport_id import get_random_passport_id_number phone_number_prefix = ['6'] class ExcelHelper: def __init__(self): self._df = pandas.Series() def write_to_exel(self, file_name, data_list: list): new_df = pandas.Series(data_list) self._df = pandas.concat([self._df, new_df]) self._df.to_excel(file_name) def read_user_agens(self) -> list: user_agent_in_json = pandas.read_excel( "/Users/lpan/Documents/workspace/appointment_tool/docs/mobile_user_agent_list.xlsx").to_json( orient='records') user_agent_dict_list = json.loads(user_agent_in_json) user_agent_list = [] for user_agent_dict in user_agent_dict_list: user_agent_str = user_agent_dict['user_agent'] if 'Mozilla' in user_agent_str: if 'Android 5.1' in user_agent_str: if 'Mac OS' not in user_agent_str: user_agent_list.append(user_agent_dict['user_agent']) print(user_agent_list) def read_contacts(self, file_name=CONTACT_LIST_FILE) -> list: contact_list_in_json = pandas.read_excel(file_name).to_json(orient='records') contact_dict_list = json.loads(contact_list_in_json) contact_list = [] for contact_dict in contact_dict_list: if contact_dict['name']: raw_name = contact_dict['name'].strip() name = raw_name.split(' ') last_name = name[0] if len(name) == 2: first_name = name[-1] else: first_name = ''.join(name[1:len(name)]) contact = ContactPojo(phone_number=contact_dict['phone'], last_name=last_name, first_name=first_name, passport_number=contact_dict['passport'], mail=contact_dict['email']) contact_list.append(contact) return contact_list def read_mails_and_pwd(self, file_name='/Users/lpan/Desktop/163.xlsx'): contact_list = [] mail_list_in_json = pandas.read_excel(file_name).to_json(orient='records') contact_dict_list = json.loads(mail_list_in_json) for contact_dict in contact_dict_list: if contact_dict['mail']: mail = contact_dict['mail'].strip() pwd = contact_dict['password'] contact = MailAddress(mail, pwd) contact_list.append(contact) return contact_list def read_names(self, file_name=CONTACT_LIST_FILE) -> list: contact_list_in_json = pandas.read_excel(file_name).to_json(orient='records') contact_dict_list = json.loads(contact_list_in_json) contact_list = [] count = 2 for contact_dict in contact_dict_list: if contact_dict['name']: raw_name = contact_dict['name'].strip() name = raw_name.split(' ') if len(name) == 1: name = raw_name.split('\xa0') if len(name) == 1: print("error in " + str(name)) last_name = name[0] if len(name) == 2: first_name = name[-1] else: first_name = ''.join(name[1:len(name)]) contact = ContactPojo(phone_number="", last_name=last_name, first_name=first_name, passport_number="", mail="") if len(first_name) == 0: print("first_name is empty: position:" + str(count)) print(name) if len(last_name) == 0: print("last_name is empty: position:" + str(count)) count = count + 1 contact_list.append(contact) return contact_list def read_email_pojo(self, file_name=CONTACT_LIST_FILE) -> list: email_info_in_json = pandas.read_excel(file_name).to_json(orient='records') contact_dict_list = json.loads(email_info_in_json) contact_list = [] count = 0 for contact_dict in contact_dict_list: if contact_dict['email']: email = contact_dict['email'].strip() password = contact_dict['code'] email_destinaire = MailAddress(email, password) count = count + 1 contact_list.append(email_destinaire) return contact_list def get_random_phone_numbers(): length = 8 # number of characters in the string. ran = ''.join(random.choices(string.digits, k=length)) id_number = random.choice(phone_number_prefix) + str(ran) return id_number def generate_email_from_name(first_name: str, last_name: str) -> str: length = 2 # number of characters in the string. ran = ''.join(random.choices(string.digits, k=length)) separator = ['.', '_', ''] domains = ['gmail.com', 'hotmail.com', 'yahoo.com', 'aol.com', 'outlook.com', 'hotmail.fr', 'gmx.com', 'hotmail.com', 'yahoo.com', 'aol.com', 'hotmail.com'] email = "{}{}{}{}@{}".format(last_name.lower(), random.choice(separator), first_name.replace("-", "").replace("'", "").lower(), ran, random.choice(domains)) print(email) return email def get_random_id_number() -> str: # write_the_valid_profiles_to_excel() S = 8 # number of characters in the string. # call random.choices() string module to find the string in Uppercase + numeric data. ran = ''.join(random.choices(string.digits, k=S)) print("The randomly generated string is : 94" + str(ran)) # print the random data return ran def write_new_contacts_to_excel(valid_contacts: list, generate_passport=True): row = 0 col = 0 # Create a workbook and add a worksheet. workbook = xlsxwriter.Workbook('real_contacts_{}.xlsx'.format(len(valid_contacts))) header_data = ['name', 'phone', 'passport', 'email'] worksheet = workbook.add_worksheet() header_format = workbook.add_format({'bold': True}) for col_num, data in enumerate(header_data): worksheet.write(row, col_num, data, header_format) row = row + 1 for info in valid_contacts: info.phone = get_random_phone_numbers() info.passport = get_random_passport_id_number() info.mail = generate_email_from_name(info.first_name, info.last_name) # Iterate over the data and write it out row by row. worksheet.write(row, col, "{} {}".format(info.last_name, info.first_name)) worksheet.write(row, col + 1, info.phone) worksheet.write(row, col + 2, info.passport) worksheet.write(row, col + 3, info.mail) row += 1 workbook.close() def write_destinaire_email(valid_contacts: list, generate_passport=True): row = 0 col = 0 # Create a workbook and add a worksheet. workbook = xlsxwriter.Workbook('real_contacts_{}.xlsx'.format(len(valid_contacts))) header_data = ['name', 'phone', 'passport', 'email'] worksheet = workbook.add_worksheet() header_format = workbook.add_format({'bold': True}) for col_num, data in enumerate(header_data): worksheet.write(row, col_num, data, header_format) row = row + 1 for info in valid_contacts: info.phone = get_random_phone_numbers() info.passport = get_random_passport_id_number() info.mail = generate_email_from_name(info.first_name, info.last_name) # Iterate over the data and write it out row by row. worksheet.write(row, col, "{} {}".format(info.last_name, info.first_name)) worksheet.write(row, col + 1, info.phone) worksheet.write(row, col + 2, info.passport) worksheet.write(row, col + 3, info.mail) row += 1 workbook.close() def save_mails_to_db(): excel_reader = ExcelHelper() emails = excel_reader.read_email_pojo("/Users/lpan/Downloads/aol_mails_21.xlsx") print(emails) for mail in emails: MONGO_STORE_MANAGER.save_destinary_emails(mail) if __name__ == '__main__': excel_reader = ExcelHelper() excel_reader.read_user_agens() # contacts = excel_reader.read_names("/Users/lpan/Documents/rdv/backup_500.xlsx") # print(contacts) # write_new_contacts_to_excel(valid_contacts=contacts) # save_mails_to_db() # for mail in excel_reader.read_mails_and_pwd(): # MONGO_STORE_MANAGER.insert_email(mail) # for i in range(1, 64): # print(get_random_phone_numbers())