From 1a2c516f49516ce5092b8af791b18dc5ac63bde0 Mon Sep 17 00:00:00 2001 From: PAN Lei Date: Thu, 20 Apr 2023 20:33:16 +0200 Subject: [PATCH] add extract_name_with_pinyinlist.py --- src/person_name/extract_name_with_pinyinlist.py | 2 +- src/utils/excel_reader.py | 13 +++++++------ 2 files changed, 8 insertions(+), 7 deletions(-) diff --git a/src/person_name/extract_name_with_pinyinlist.py b/src/person_name/extract_name_with_pinyinlist.py index abf1b20..af5c1c3 100644 --- a/src/person_name/extract_name_with_pinyinlist.py +++ b/src/person_name/extract_name_with_pinyinlist.py @@ -93,7 +93,7 @@ def write_new_contacts_to_excel(valid_contacts: list): if __name__ == '__main__': db_manager = MongoDbManager() - mail_list = db_manager.get_destination_emails()[1001:1200] + mail_list = db_manager.get_destination_emails()[1201:1500] # mail_list = db_manager.get_destination_emails()[50:200] generate_contacts = [] pinyin_name_list = read_pinyin_list_from_file() diff --git a/src/utils/excel_reader.py b/src/utils/excel_reader.py index 4989c37..f7eb231 100644 --- a/src/utils/excel_reader.py +++ b/src/utils/excel_reader.py @@ -39,6 +39,7 @@ class ExcelHelper: print(user_agent_list) def read_contacts(self, file_name=CONTACT_LIST_FILE) -> 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 = [] @@ -60,8 +61,8 @@ class ExcelHelper: contact_list.append(contact) return contact_list - def check_contact_list(self): - contact_list = self.read_contacts() + def check_contact_list(self, file_name=CONTACT_LIST_FILE): + contact_list = self.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) @@ -227,12 +228,12 @@ def save_mails_to_db(): if __name__ == '__main__': # excel_reader = ExcelHelper() - # contacts = excel_reader.read_names("/Users/lpan/Downloads/10_rambler_ru_23_03_2023.xlsx") + # 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() - save_mails_to_db() + excel_reader = ExcelHelper() + excel_reader.check_contact_list("/Users/lpan/Desktop/contact_email_valid.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, 64):