def login_ffp(driver, username=username, password=password):
while True:
driver.get(ffp_url)
driver.implicitly_wait(10)
try:
driver.find_element_by_id('acceptButton').click()
except:
pass
username_input = driver.find_element_by_id('j_username')
password_input = driver.find_element_by_id('j_password')
captcha_input = driver.find_element_by_id('j_verify')
username_input.clear()
driver.implicitly_wait(1)
username_input.send_keys(username)
driver.implicitly_wait(1)
password_input.send_keys(password)
driver.implicitly_wait(1)
captcha_image_url = captcha_input.find_element_by_xpath(
'..').find_element_by_tag_name('img').get_attribute('src')
captcha_code = get_captcha_image(captcha_image_url)
captcha_input.send_keys(captcha_code)
driver.implicitly_wait(1)
driver.find_element_by_id('loginButton').click()
driver.implicitly_wait(10)
if 'error' in driver.current_url:
choice = yes_no_box('登录失败，是否重新登录？', 'ICS登录')
if choice == '是':
continue
else:
alert_box('您正在退出程序，感谢使用', '退出程序')
os._exit(0)
driver.find_element_by_id('row_customer').click()
driver.implicitly_wait(10)
return driver
def main():
start_date = date_box('请输入开始日期', '开始日期')
end_date = date_box('请输入结束日期', '结束 日期')
date_list = create_date_list(start_date, end_date)
if not date_list:
alert_box('日期填写有误', '错误')
os._exit(0)
st_data = get_st_data(date_list, start_date, end_date)
sales_data = get_sales_data(start_date, end_date)
data = compare_data(st_data, sales_data)
describe_data(data)
def open_app(app_path):
try:
os.startfile(app_path)
time.sleep(3)
keyboard_press('enter')
time.sleep(3)
keyboard_press('enter')
return True
except:
alert_box('无法打开应用程序%s，请检查程序路径是否正确' % app_path, '打开应用程序失败')
os._exit(0)
def get_sales_data(start_date, end_date):
if '%s_%s_sales_total.csv' % (start_date,
end_date) in os.listdir(sales_dir):
try:
sales_data = pd.read_csv('%s%s%s_%s_sales_total.csv' %
(sales_dir, os.sep, start_date, end_date),
encoding='gbk',
converters={'关联票号': str})
except:
sales_data = pd.read_csv('%s%s%s_%s_sales_total.csv' %
(sales_dir, os.sep, start_date, end_date),
converters={'关联票号': str})
return sales_data
sales_data = pd.DataFrame(None)
for root, _, files in os.walk(sales_dir):
for file in files:
sales_data_temp = pd.read_excel(os.path.join(root, file),
skiprows=3,
converters={'关联票号': str})
if sales_data_temp.shape[0] <= 4:
continue
sales_data_temp.drop(sales_data_temp.head(2).index, inplace=True)
# CAN
# sales_data_temp = sales_data_temp[(sales_data_temp['起飞机场'] == 'CAN') & (sales_data_temp['销售单位'] == 'SMARTSERVICE系统') & (sales_data_temp['下单渠道'] == 'SS') & (sales_data_temp['EMD类型'].isin(['柜台升舱', '一人多座', '休息室升舱', '机场柜台升舱']) == True)]
# URC
sales_data_temp = sales_data_temp[
(sales_data_temp['起飞机场'] == 'URC')
& (sales_data_temp['销售单位'] == 'SMARTSERVICE系统') &
(sales_data_temp['下单渠道'] == 'SS') &
(sales_data_temp['EMD类型'].isin(
['柜台升舱', '一人多座', '休息室升舱', '机场柜台升舱']) == True)]
sales_data = pd.concat([sales_data, sales_data_temp],
ignore_index=True,
join='outer')
if sales_data.shape[0] == 0:
alert_box('未提取到销售数据', '错误')
os._exit(0)
sales_data.fillna('', inplace=True)
sales_data = sales_data[['EMD类型', '关联票号', '支付价', '扣减里程/配额', '订单下单人姓名']]
sales_data.to_csv('%s%s%s_%s_sales_total.csv' %
(sales_dir, os.sep, start_date, end_date))
return sales_data
def get_ff_list(date):
try:
ff_list = pd.read_excel('%s%s%s.xlsx' %
(ics_processed_dir, os.sep, date),
converters={
'flt_num': str,
'ff_number': str
})
except:
alert_box(
'未找到当天航班的xlsx文档，请检查。注：文档需从SOC预报中直接复制入Excel表格，并以YYYY-MM-DD_XXX.xlsx命名，其中XXX为场站三字码。',
'退出程序')
os._exit(0)
ff_list = ff_list[(ff_list['ff_airline'] == 'CZ')
& (ff_list['ff_number'].isna() == False) &
(ff_list['ff_number'] != '')]
if len(ff_list) == 0:
alert_box('会员列表无数据，请确认。', '退出程序')
os._exit(0)
ff_list.fillna('', inplace=True)
return ff_list
def check_file(file):
if not os.path.exists(file):
alert_box('缺少 %s 文件' % file, '错误')
os._exit(0)
def get_st_data(date_list, start_date, end_date):
if '%s_%s_st_total.csv' % (start_date, end_date) in os.listdir(task_dir):
try:
st_data = pd.read_csv('%s%s%s_%s_st_total.csv' %
(task_dir, os.sep, start_date, end_date),
encoding='gbk',
converters={
'会员卡号': str,
'飞行日期': str,
'OC航班号': str,
'电子客票号': str
})
except:
st_data = pd.read_csv('%s%s%s_%s_st_total.csv' %
(task_dir, os.sep, start_date, end_date),
converters={
'会员卡号': str,
'飞行日期': str,
'OC航班号': str,
'电子客票号': str
})
return st_data
st_data = pd.DataFrame(None)
files = os.listdir(task_dir)
for date in date_list:
for file in files:
if date in file:
try:
st_data_temp = pd.read_csv('%s%s%s' %
(task_dir, os.sep, file),
encoding='gbk',
converters={
'会员卡号': str,
'飞行日期': str,
'OC航班号': str,
'电子客票号': str
})
except:
st_data_temp = pd.read_csv('%s%s%s' %
(task_dir, os.sep, file),
converters={
'会员卡号': str,
'飞行日期': str,
'OC航班号': str,
'电子客票号': str
})
if st_data_temp.shape[0] != 0 and st_data_temp is not None:
st_data_temp.replace('\t', '', regex=True, inplace=True)
st_data_temp['飞行日期'] = pd.to_datetime(
st_data_temp['飞行日期']).dt.strftime('%Y-%m-%d')
st_data_temp.fillna('', inplace=True)
# CAN
# st_data_temp = st_data_temp[st_data_temp['航段始发机场'] == 'CAN']
# URC
st_data_temp = st_data_temp[st_data_temp['航段始发机场'] ==
'URC']
st_data = pd.concat([st_data, st_data_temp],
ignore_index=True,
join='outer')
else:
continue
if st_data.shape[0] == 0:
alert_box('未提取到st数据', '错误')
os._exit(0)
st_data.fillna('', inplace=True)
st_data = st_data[[
'UID', '航段', '飞行日期', '航段性质', '会员级别', 'OC母舱位', '电子客票号', '查询', '备注'
]]
st_data.to_csv('%s%s%s_%s_st_total.csv' %
(task_dir, os.sep, start_date, end_date))
return st_data
'flt_date', 'station', 'flt_num',
'name', 'seat_class', 'group_',
'ff_number', 'ff_level',
'available_miles', 'inbound'
])
structured_data.to_excel('%s%s%s_ffp.xlsx' %
(ics_results_dir, os.sep, date))
def logout_ffp(driver):
driver.find_element_by_id('arrow').click()
driver.implicitly_wait(2)
driver.find_element_by_id('logout').click()
driver.implicitly_wait(2)
if __name__ == "__main__":
date = get_date()
ff_list = get_ff_list(date)
try:
driver = open_driver()
login_ffp(driver)
unstructured_data = main_crawler(driver, ff_list)
data_structuring(unstructured_data, date)
alert_box('数据处理完毕，请从%s文件夹查看，感谢使用！' % ics_results_dir, '退出程序')
except:
alert_box('程序出现问题，正在退出程序，感谢使用！', '退出程序')
finally:
logout_ffp(driver)
driver.quit()