# coding=utf-8 import csv import time from selenium import webdriver MAX_RETRY_TIME = 150 URL = "https://emweb.securities.eastmoney.com/pc_hsf10/pages/index.html?type=web&code=%s&color=b#/cwfx" def get_company_finance(stock_code_list): browser = webdriver.Firefox() for stock_code in stock_code_list: code = stock_code[0] name = stock_code[1] company_name = stock_code[2] stock_code = stock_code[0] if stock_code[0] == '6': stock_code = 'SH' + stock_code elif stock_code[0] == '8' or stock_code[0] == '4': stock_code = 'BJ' + stock_code else: stock_code = 'SZ' + stock_code url = URL % stock_code browser.execute_script('window.open("%s")' % url) retry_time = 0 while True: try: if retry_time < MAX_RETRY_TIME: browser.switch_to.window(browser.window_handles[1]) zcfzb_button = browser.find_elements_by_css_selector('div.section.zcfzb ul:last-child li:nth-child(2)')[0] lrb_button = browser.find_elements_by_css_selector('div.section.lrb ul:last-child li:nth-child(2)')[0] browser.execute_script("arguments[0].click();", zcfzb_button) time.sleep(0.2) browser.execute_script("arguments[0].click();", lrb_button) time.sleep(0.2) break else: print("年报按钮获取失败") return except: retry_time += 1 time.sleep(0.2) flag = 0 retry_time = 0 while True: try: if retry_time < MAX_RETRY_TIME: if "银行" in company_name or "信托" in company_name: # 资产总额 TotalAssets_2022 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(43) td:nth-child(2) span')[ 0].text TotalAssets_2021 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(43) td:nth-child(3) span')[ 0].text TotalAssets_2020 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(43) td:nth-child(4) span')[ 0].text # 销售收入 SalesProceeds_2022 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(2) td:nth-child(2) span')[0].text SalesProceeds_2021 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(2) td:nth-child(3) span')[0].text SalesProceeds_2020 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(2) td:nth-child(4) span')[0].text # 利润总额 LRZE_2022 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(35) td:nth-child(2) span')[0].text LRZE_2021 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(35) td:nth-child(3) span')[0].text LRZE_2020 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(35) td:nth-child(4) span')[0].text # 净利润 JLR_2022 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(39) td:nth-child(2) span')[0].text JLR_2021 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(39) td:nth-child(3) span')[0].text JLR_2020 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(39) td:nth-child(4) span')[0].text # 负债总计 FZZJ_2022 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(74) td:nth-child(2) span')[ 0].text FZZJ_2021 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(74) td:nth-child(3) span')[ 0].text FZZJ_2020 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(74) td:nth-child(4) span')[ 0].text # 纳税额 Ratal_2022 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(59) td:nth-child(2) span')[ 0].text Ratal_2021 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(59) td:nth-child(3) span')[ 0].text Ratal_2020 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(59) td:nth-child(4) span')[ 0].text break elif "金融" in company_name or "证券" in company_name or "期货" in company_name: # 资产总额 TotalAssets_2022 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(44) td:nth-child(2) span')[ 0].text TotalAssets_2021 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(44) td:nth-child(3) span')[ 0].text TotalAssets_2020 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(44) td:nth-child(4) span')[ 0].text # 销售收入 SalesProceeds_2022 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(2) td:nth-child(2) span')[0].text SalesProceeds_2021 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(2) td:nth-child(3) span')[0].text SalesProceeds_2020 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(2) td:nth-child(4) span')[0].text # 利润总额 LRZE_2022 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(40) td:nth-child(2) span')[0].text LRZE_2021 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(40) td:nth-child(3) span')[0].text LRZE_2020 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(40) td:nth-child(4) span')[0].text # 净利润 JLR_2022 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(42) td:nth-child(2) span')[0].text JLR_2021 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(42) td:nth-child(3) span')[0].text JLR_2020 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(42) td:nth-child(4) span')[0].text # 负债总计 FZZJ_2022 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(76) td:nth-child(2) span')[ 0].text FZZJ_2021 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(76) td:nth-child(3) span')[ 0].text FZZJ_2020 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(76) td:nth-child(4) span')[ 0].text # 纳税额 Ratal_2022 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(61) td:nth-child(2) span')[ 0].text Ratal_2021 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(61) td:nth-child(3) span')[ 0].text Ratal_2020 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(61) td:nth-child(4) span')[ 0].text break elif "保险" in company_name: # 资产总额 TotalAssets_2022 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(51) td:nth-child(2) span')[ 0].text TotalAssets_2021 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(51) td:nth-child(3) span')[ 0].text TotalAssets_2020 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(51) td:nth-child(4) span')[ 0].text # 销售收入 SalesProceeds_2022 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(2) td:nth-child(2) span')[0].text SalesProceeds_2021 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(2) td:nth-child(3) span')[0].text SalesProceeds_2020 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(2) td:nth-child(4) span')[0].text # 利润总额 LRZE_2022 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(53) td:nth-child(2) span')[0].text LRZE_2021 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(53) td:nth-child(3) span')[0].text LRZE_2020 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(53) td:nth-child(4) span')[0].text # 净利润 JLR_2022 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(57) td:nth-child(2) span')[0].text JLR_2021 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(57) td:nth-child(3) span')[0].text JLR_2020 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(57) td:nth-child(4) span')[0].text # 负债总计 FZZJ_2022 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(98) td:nth-child(2) span')[ 0].text FZZJ_2021 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(98) td:nth-child(3) span')[ 0].text FZZJ_2020 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(98) td:nth-child(4) span')[ 0].text # 纳税额 Ratal_2022 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(72) td:nth-child(2) span')[ 0].text Ratal_2021 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(72) td:nth-child(3) span')[ 0].text Ratal_2020 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(72) td:nth-child(4) span')[ 0].text break else: # 资产总额 TotalAssets_2022 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(69) td:nth-child(2) span')[ 0].text TotalAssets_2021 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(69) td:nth-child(3) span')[ 0].text TotalAssets_2020 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(69) td:nth-child(4) span')[ 0].text # 销售收入 SalesProceeds_2022 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(2) td:nth-child(2) span')[ 0].text SalesProceeds_2021 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(2) td:nth-child(3) span')[ 0].text SalesProceeds_2020 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(2) td:nth-child(4) span')[ 0].text # 利润总额 LRZE_2022 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(48) td:nth-child(2) span')[ 0].text LRZE_2021 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(48) td:nth-child(3) span')[ 0].text LRZE_2020 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(48) td:nth-child(4) span')[ 0].text # 净利润 JLR_2022 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(52) td:nth-child(2) span')[ 0].text JLR_2021 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(52) td:nth-child(3) span')[ 0].text JLR_2020 = \ browser.find_elements_by_css_selector('div.lrb_table tr:nth-child(52) td:nth-child(4) span')[ 0].text # 负债总计 FZZJ_2022 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(126) td:nth-child(2) span')[ 0].text FZZJ_2021 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(126) td:nth-child(3) span')[ 0].text FZZJ_2020 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(126) td:nth-child(4) span')[ 0].text # 纳税额 Ratal_2022 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(88) td:nth-child(2) span')[ 0].text Ratal_2021 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(88) td:nth-child(3) span')[ 0].text Ratal_2020 = \ browser.find_elements_by_css_selector('div.zcfzb_table tr:nth-child(88) td:nth-child(4) span')[ 0].text break else: print("数据获取失败") return except: if retry_time > 25: with open('company_error.csv', 'a', newline='') as newfile: writer = csv.writer(newfile) company_a = [code,name,company_name] writer.writerow(company_a) newfile.close() flag = 1 break retry_time += 1 time.sleep(0.2) if flag != 1: company_finance_details = {"股票代码":code,"股票名称":name,"公司名称":company_name, "资产总额2022":TotalAssets_2022,"资产总额2021":TotalAssets_2021,"资产总额2020":TotalAssets_2020, "销售收入2022":SalesProceeds_2022,"销售收入2021":SalesProceeds_2021,"销售收入2020":SalesProceeds_2020, "利润总额2022":LRZE_2022,"利润总额2021":LRZE_2021,"利润总额2020":LRZE_2020, "净利润2022":JLR_2022,"净利润2021":JLR_2021,"净利润2020":JLR_2020, "负债总计2022":FZZJ_2022,"负债总计2021":FZZJ_2021,"负债总计2020":FZZJ_2020, "纳税额2022":Ratal_2022,"纳税额2021":Ratal_2021,"纳税额2020":Ratal_2020 } print(company_finance_details) company_finance_details_csv = [code,name,company_name, TotalAssets_2022,TotalAssets_2021,TotalAssets_2020, SalesProceeds_2022,SalesProceeds_2021,SalesProceeds_2020, LRZE_2022,LRZE_2021,LRZE_2020, JLR_2022,JLR_2021,JLR_2020, FZZJ_2022,FZZJ_2021,FZZJ_2020, Ratal_2022,Ratal_2021,Ratal_2020] with open('company_finance_details.csv', 'a', newline='') as newfile: writer = csv.writer(newfile) writer.writerow(company_finance_details_csv) newfile.close() browser.close() browser.switch_to.window(browser.window_handles[0]) else: browser.close() browser.switch_to.window(browser.window_handles[0]) browser.close() file = 'company.csv' with open(file,'r') as file: code_list = csv.reader(file,delimiter=',') stock_code_list = [] for code in code_list: stock_code_list.append(code) del stock_code_list[0] get_company_finance(stock_code_list)