Files
appointment_tool/src/utils/excel_reader.py
T
2024-10-03 23:42:33 +02:00

212 lines
8.4 KiB
Python
Executable File

import json
import random
import string
import pandas as pandas
import validators as validators
import xlsxwriter
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.contacts.generate_random_passport_id import get_random_passport_id_number
fr_phone_number_prefix = ['73', '74', '75', '76', '77', '78', '79']
chinnese_number_prefix = ['13', '15', '18']
def read_links_to_click(file_path):
links_info_in_json = pandas.read_excel(file_path).to_json(orient='records')
# print(links_info_in_json)
for item in json.loads(links_info_in_json):
link = item['link']
if validators.url(link):
print(link)
MONGO_STORE_MANAGER.save_links_to_validate(link, "")
else:
print("error on link " + link)
def read_contacts(file_name) -> list:
print("read file " + file_name)
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)])
ip_country = "FR"
if contact_dict.get('ip_country') is not None:
ip_country = contact_dict['ip_country']
store = "random"
if contact_dict.get('store') is not None:
store = contact_dict['store']
ua = ""
if contact_dict.get('ua') is not None:
ua = contact_dict['ua']
contact = ContactPojo(phone_number=contact_dict['phone'],
last_name=last_name,
first_name=first_name,
passport_number=contact_dict['passport'],
mail=contact_dict['email'], store=store)
if contact_dict.get('serial') is not None:
serial = contact_dict['serial']
contact.serial = serial
contact.ip_country = ip_country
contact.ua = ua
contact_list.append(contact)
return contact_list
class ExcelHelper:
def __init__(self):
self._df = pandas.Series()
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 check_contact_list(self, file_name):
contact_list = read_contacts(file_name)
for contact in contact_list:
if contact.first_name is None or len(contact.first_name) == 0:
print("error in firstName for " + contact.mail)
if contact.last_name is None or len(contact.last_name) == 0:
print("error in last_name for " + contact.mail)
if contact.phone is None or len(contact.phone) == 0:
print("error in phone for " + contact.mail)
if contact.passport is None or len(contact.passport) == 0:
print("error in passport_number for " + contact.mail)
if contact.mail is None or len(contact.mail) == 0:
print("error in mail for " + contact.phone_number)
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['email']:
mail = contact_dict['email'].strip()
pwd = contact_dict['password']
contact = MailAddress(mail, pwd)
contact_list.append(contact)
return contact_list
def read_names(self, file_name) -> 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) -> 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_fr_phone_numbers():
length = 7 # number of characters in the string.
ran = ''.join(random.choices(string.digits, k=length))
_phone_number = random.choice(fr_phone_number_prefix) + str(ran)
return _phone_number
def get_random_cn_phone_numbers():
length = 8 # number of characters in the string.
ran = ''.join(random.choices(string.digits, k=length))
_phone_number = random.choice(fr_phone_number_prefix) + str(ran)
prefix = random.choice(chinnese_number_prefix)
return prefix + _phone_number
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 save_mails_to_db():
excel_reader = ExcelHelper()
emails = excel_reader.read_email_pojo("~/Downloads/邮箱及密码.xlsx")
print(emails)
for mail in emails:
MONGO_STORE_MANAGER.save_destinary_emails(mail)
if __name__ == '__main__':
# excel_reader = ExcelHelper()
# contacts = excel_reader.read_names("/Users/lpan/Downloads/gmail_10.xlsx")
# print(contacts)
# write_new_contacts_to_excel(valid_contacts=contacts)
# excel_reader = ExcelHelper()
# excel_reader.check_contact_list("/Users/lpan/Desktop/contact_email_valid.xlsx")
# read_links_to_click("/Users/lpan/Downloads/links.xlsx")
save_mails_to_db()
# for mail in excel_reader.read_mails_and_pwd():
# MONGO_STORE_MANAGER.insert_email(mail)
# for i in range(1, 300):
# print(get_random_fr_phone_numbers())