fix

phitrann · Mar 8, 2024 · 83a80c2 · 83a80c2
1 parent 4f2b06a
commit 83a80c2
Show file tree

Hide file tree

Showing 6 changed files with 1,413 additions and 1,427 deletions.
diff --git a/.github/workflows/main.yaml b/.github/workflows/main.yaml
diff --git a/07032024/07032024.csv b/07032024/07032024.csv
@@ -0,0 +1,68 @@
+,Số hiệu chuyến bay,Khởi hành,Đến,Airbus,SkyBOSS_Business,SkyBOSS,Deluxe,Eco,Thuế_phí
+0,VJ640,2024-03-07 22:35:00,2024-03-07 23:50:00,Airbus A321,,3992000.0000000005,2738000.0,1968000.0,593.0
+1,VJ622,2024-03-08 07:25:00,2024-03-08 08:45:00,Airbus A320,,3992000.0000000005,2892000.0,2122000.0,593.0
+2,VJ626,2024-03-08 11:05:00,2024-03-08 12:25:00,Airbus A320,,3992000.0000000005,3024000.0,2254000.0,593.0
+3,VJ630,2024-03-08 14:40:00,2024-03-08 16:00:00,Airbus A321,,3992000.0000000005,2892000.0,2122000.0,593.0
+4,VJ632,2024-03-08 17:20:00,2024-03-08 18:35:00,Airbus A320,,3992000.0000000005,3145000.0,2375000.0,593.0
+5,VJ638,2023-03-08 20:55:00,2023-03-08 22:15:00,Airbus A320,,,,,593.0
+6,VJ640,2024-03-08 22:35:00,2024-03-08 23:50:00,Airbus A321,,3992000.0000000005,2595000.0,1825000.0,593.0
+7,VJ622,2024-03-09 07:25:00,2024-03-09 08:45:00,Airbus A320,,3992000.0000000005,3112000.0,2342000.0,593.0
+8,VJ626,2023-03-09 11:05:00,2023-03-09 12:25:00,Airbus A320,,,,,593.0
+9,VJ630,2024-03-09 14:40:00,2024-03-09 16:00:00,Airbus A321,,3992000.0000000005,2892000.0,2122000.0,593.0
+10,VJ632,2024-03-09 17:20:00,2024-03-09 18:35:00,Airbus A320,,3992000.0000000005,2342000.0,1572000.0,593.0
+11,VJ634,2024-03-09 19:25:00,2024-03-09 20:45:00,Airbus A321,,3992000.0000000005,2177000.0,1407000.0,593.0
+12,VJ638,2024-03-09 20:15:00,2024-03-09 21:30:00,Airbus A321,,3992000.0000000005,2034000.0,1264000.0,593.0
+13,VJ640,2024-03-09 22:35:00,2024-03-09 23:55:00,Airbus A321,,3992000.0000000005,2034000.0,1264000.0,593.0
+14,VJ622,2024-03-10 07:25:00,2024-03-10 08:45:00,Airbus A320,,3992000.0000000005,2342000.0,1572000.0,593.0
+15,VJ626,2024-03-10 11:05:00,2024-03-10 12:25:00,Airbus A321,,3992000.0000000005,2463000.0,1693000.0,593.0
+16,VJ630,2024-03-10 14:40:00,2024-03-10 16:00:00,Airbus A321,,3992000.0000000005,2463000.0,1693000.0,593.0
+17,VJ632,2024-03-10 17:10:00,2024-03-10 18:30:00,Airbus A320,,3992000.0000000005,2463000.0,1693000.0,593.0
+18,VJ634,2024-03-10 19:25:00,2024-03-10 20:45:00,Airbus A321,,3992000.0000000005,2034000.0,1264000.0,593.0
+19,VJ638,2024-03-10 20:10:00,2024-03-10 21:30:00,Airbus A320,,3992000.0000000005,2177000.0,1407000.0,593.0
+20,VJ640,2024-03-10 22:35:00,2024-03-10 23:55:00,Airbus A321,,3992000.0000000005,2034000.0,1264000.0,593.0
+21,VJ622,2024-03-11 07:25:00,2024-03-11 08:45:00,Airbus A320,,3992000.0000000005,2595000.0,1825000.0,593.0
+22,VJ626,2024-03-11 11:05:00,2024-03-11 12:25:00,Airbus A320,,3992000.0000000005,2595000.0,1825000.0,593.0
+23,VJ630,2024-03-11 14:40:00,2024-03-11 16:00:00,Airbus A321,,3992000.0000000005,2463000.0,1693000.0,593.0
+24,VJ632,2024-03-11 17:20:00,2024-03-11 18:35:00,Airbus A320,,3992000.0000000005,2342000.0,1572000.0,593.0
+25,VJ634,2024-03-11 19:25:00,2024-03-11 20:45:00,Airbus A320,,3992000.0000000005,2034000.0,1264000.0,593.0
+26,VJ636,2024-03-11 20:15:00,2024-03-11 21:35:00,Airbus A320,,3992000.0000000005,2034000.0,1264000.0,593.0
+27,VJ622,2024-03-12 07:25:00,2024-03-12 08:45:00,Airbus A320,,3992000.0000000005,2342000.0,1572000.0,593.0
+28,VJ626,2024-03-12 11:05:00,2024-03-12 12:25:00,Airbus A320,,3992000.0000000005,2595000.0,1825000.0,593.0
+29,VJ630,2024-03-12 14:40:00,2024-03-12 16:00:00,Airbus A321,,3992000.0000000005,2463000.0,1693000.0,593.0
+30,VJ632,2024-03-12 17:10:00,2024-03-12 18:30:00,Airbus A320,,3992000.0000000005,2177000.0,1407000.0,593.0
+31,VJ634,2024-03-12 19:25:00,2024-03-12 20:45:00,Airbus A321,,3992000.0000000005,2034000.0,1264000.0,593.0
+32,VJ636,2024-03-12 20:15:00,2024-03-12 21:35:00,Airbus A321,,3992000.0000000005,1902000.0,1132000.0,593.0
+33,VJ622,2024-03-13 07:25:00,2024-03-13 08:45:00,Airbus A320,,3992000.0000000005,2342000.0,1572000.0,593.0
+34,VJ630,2024-03-13 14:40:00,2024-03-13 16:00:00,Airbus A321,,3992000.0000000005,2892000.0,2122000.0,593.0
+35,VJ632,2024-03-13 16:55:00,2024-03-13 18:15:00,Airbus A320,,3992000.0000000005,2342000.0,1572000.0,593.0
+36,VJ634,2024-03-13 19:25:00,2024-03-13 20:45:00,Airbus A321,,3992000.0000000005,2034000.0,1264000.0,593.0
+37,VJ636,2024-03-13 20:15:00,2024-03-13 21:35:00,Airbus A320,,3992000.0000000005,2034000.0,1264000.0,593.0
+38,VJ622,2024-03-14 07:25:00,2024-03-14 08:45:00,Airbus A320,,3992000.0000000005,2463000.0,1693000.0,593.0
+39,VJ626,2024-03-14 11:05:00,2024-03-14 12:25:00,Airbus A320,,3992000.0000000005,2738000.0,1968000.0,593.0
+40,VJ630,2024-03-14 14:40:00,2024-03-14 16:00:00,Airbus A321,,3992000.0000000005,2463000.0,1693000.0,593.0
+41,VJ632,2024-03-14 17:20:00,2024-03-14 18:35:00,Airbus A320,,3992000.0000000005,2463000.0,1693000.0,593.0
+42,VJ634,2024-03-14 19:25:00,2024-03-14 20:45:00,Airbus A320,,3992000.0000000005,2463000.0,1693000.0,593.0
+43,VJ636,2024-03-14 20:15:00,2024-03-14 21:35:00,Airbus A321,,3992000.0000000005,2463000.0,1693000.0,593.0
+44,VJ622,2024-03-15 07:25:00,2024-03-15 08:45:00,Airbus A320,,3992000.0000000005,2892000.0,2122000.0,593.0
+45,VJ626,2024-03-15 11:05:00,2024-03-15 12:25:00,Airbus A320,,3992000.0000000005,2947000.0,2177000.0,593.0
+46,VJ630,2024-03-15 14:40:00,2024-03-15 16:00:00,Airbus A321,,3992000.0000000005,2892000.0,2122000.0,593.0
+47,VJ632,2024-03-15 16:55:00,2024-03-15 18:15:00,Airbus A320,,3992000.0000000005,3024000.0,2254000.0,593.0
+48,VJ634,2024-03-15 19:25:00,2024-03-15 20:45:00,Airbus A321,,3992000.0000000005,2463000.0,1693000.0,593.0
+49,VJ636,2024-03-15 21:10:00,2024-03-15 22:30:00,Airbus A320,,3992000.0000000005,2463000.0,1693000.0,593.0
+50,VJ622,2024-03-16 07:25:00,2024-03-16 08:45:00,Airbus A320,,3992000.0000000005,2463000.0,1693000.0,593.0
+51,VJ626,2024-03-16 11:05:00,2024-03-16 12:25:00,Airbus A320,,3992000.0000000005,2595000.0,1825000.0,593.0
+52,VJ630,2024-03-16 14:40:00,2024-03-16 16:00:00,Airbus A321,,3992000.0000000005,2463000.0,1693000.0,593.0
+53,VJ632,2024-03-16 17:20:00,2024-03-16 18:35:00,Airbus A320,,3992000.0000000005,2342000.0,1572000.0,593.0
+54,VJ634,2024-03-16 19:25:00,2024-03-16 20:45:00,Airbus A321,,3992000.0000000005,2034000.0,1264000.0,593.0
+55,VJ622,2024-03-17 07:25:00,2024-03-17 08:45:00,Airbus A320,,3992000.0000000005,2177000.0,1407000.0,593.0
+56,VJ626,2024-03-17 11:05:00,2024-03-17 12:25:00,Airbus A321,,3992000.0000000005,2463000.0,1693000.0,593.0
+57,VJ630,2024-03-17 14:40:00,2024-03-17 16:00:00,Airbus A321,,3992000.0000000005,2463000.0,1693000.0,593.0
+58,VJ632,2024-03-17 17:10:00,2024-03-17 18:30:00,Airbus A320,,3992000.0000000005,2342000.0,1572000.0,593.0
+59,VJ634,2024-03-17 19:25:00,2024-03-17 20:45:00,Airbus A321,,3992000.0000000005,2034000.0,1264000.0,593.0
+60,VJ636,2024-03-17 20:15:00,2024-03-17 21:35:00,Airbus A321,,3992000.0000000005,1902000.0,1132000.0,593.0
+61,VJ622,2024-03-18 07:25:00,2024-03-18 08:45:00,Airbus A320,,3992000.0000000005,3255000.0,2485000.0,593.0
+62,VJ626,2024-03-18 11:05:00,2024-03-18 12:25:00,Airbus A320,,3992000.0000000005,2892000.0,2122000.0,593.0
+63,VJ630,2024-03-18 14:40:00,2024-03-18 16:00:00,Airbus A321,,3992000.0000000005,2595000.0,1825000.0,593.0
+64,VJ632,2024-03-18 17:20:00,2024-03-18 18:35:00,Airbus A320,,3992000.0000000005,2463000.0,1693000.0,593.0
+65,VJ634,2024-03-18 19:25:00,2024-03-18 20:45:00,Airbus A321,,3992000.0000000005,2177000.0,1407000.0,593.0
+66,VJ636,2024-03-18 20:15:00,2024-03-18 21:35:00,Airbus A320,,3992000.0000000005,2177000.0,1407000.0,593.0
diff --git a/07032024/giave.png b/07032024/giave.png
diff --git a/changelogs.txt b/changelogs.txt
@@ -1,3 +1,4 @@
 - Cập nhật ngày 20, 21/4/2023 update trường hợp các chuyến bay hết chỗ và lấy được toàn bộ muibox.
 - Cập nhật ngày 24/6/2023 update tắt quảng cáo mới (2 quảng cáo), accept cookie, thuế phí bị tính riêng.
 - Cập nhật ngày 25/6/2023 fix lỗi không lấy được ngày hết chỗ.
+- Cập nhật ngày 07/03/2024 fix lỗi tắt popup quảng cáo.
diff --git a/crawler.py b/crawler.py
@@ -0,0 +1,260 @@
+from selenium import webdriver
+from selenium.webdriver.common.keys import Keys
+from selenium.webdriver.common.by import By
+from selenium.webdriver.support import expected_conditions as EC
+from selenium.webdriver.common.action_chains import ActionChains
+from bs4 import BeautifulSoup
+import re # Regular expressionc
+from datetime import date, timedelta
+from time import sleep
+from selenium.webdriver.chrome.service import Service
+from webdriver_manager.chrome import ChromeDriverManager
+from selenium.webdriver.support.ui import WebDriverWait
+from selenium.common.exceptions import StaleElementReferenceException
+from selenium.common.exceptions import ElementNotInteractableException
+
+import pandas as pd
+import numpy as np
+import matplotlib.pyplot as plt
+
+# Mở Chrome và cho full màn hình
+# Tự tải phiên bản ChromeDriver phù hợp với phiên bản trên máy
+browser = webdriver.Chrome(service=Service(ChromeDriverManager().install())) #webdriver.Chrome("chromedriver.exe")
+browser.maximize_window()
+# Set thời gian chờ mặc định là 10 giây
+browser.implicitly_wait(10)
+# Mở trang chủ của Vietjet
+url = 'https://www.vietjetair.com'
+browser.get(url)
+
+wait = WebDriverWait(browser, 10)
+
+# Tắt quảng cáo 
+wait.until(EC.element_to_be_clickable((By.XPATH, '//*[@id="onesignal-slidedown-cancel-button"]'))).click()
+
+close_buttons = wait.until(EC.presence_of_all_elements_located((By.CSS_SELECTOR, 'button.MuiIconButton-root[aria-label="close"]')))
+for close_button in close_buttons[1:]:
+    close_button.click()
+
+# Chọn sử dụng cookie
+wait.until(EC.element_to_be_clickable((By.XPATH, '//div[@role="dialog"]//button'))).click()
+
+# Chọn vé một chiều
+wait.until(EC.element_to_be_clickable((
+    By.XPATH, '//span[contains(text(),"Một chiều")]'))).click()
+
+# Chọn điểm khởi hành là TP HCM 
+wait.until(EC.element_to_be_clickable((By.XPATH, '//*[@class="MuiInputBase-input MuiOutlinedInput-input"]'))).click()
+# selectOrigin = browser.find_element(By.XPATH, '//*[@class="MuiInputBase-input MuiOutlinedInput-input"]')
+# selectOrigin.click()
+# sleep(3)
+
+# Chọn sân bay Tân Sơn Nhất
+wait.until(EC.element_to_be_clickable((By.XPATH, '//*[contains(text(),"Sân bay Tân Sơn Nhất")]'))).click()
+
+# Chọn sân bay Đà Nẵng
+try:
+    wait.until(EC.element_to_be_clickable((By.XPATH, '//*[contains(text(),"Sân bay Đà Nẵng")]'))).click()
+except:
+    wait.until(EC.element_to_be_clickable((By.XPATH, '//*[contains(text(),"Sân bay Đà Nẵng")]'))).click()
+
+# Chọn ngày tháng 
+today = date.today()
+tomorrow = today + timedelta(1)
+tomorrow_year = str(tomorrow.year)
+tomorrow_month = tomorrow.strftime('%m')
+tomorrow_day = str(tomorrow.day)
+
+
+# Chuỗi ngày tháng để click
+str_time = 'tháng '+ tomorrow_month + ' ' + tomorrow_year 
+
+wait.until(EC.presence_of_element_located((
+    By.XPATH, '//*[contains(text(),"{a}")]//following-sibling::div[2]//descendant::span[contains(text(),"{b}")]'.format(a = str_time, b = tomorrow_day)))).click()
+
+# Click Tìm chuyến bay
+submit = wait.until(EC.visibility_of_element_located((By.XPATH, '(//*[contains(text(), "Tìm chuyến bay")])[2]'))) 
+try: 
+    submit.click()
+except:
+    submit.click()
+
+from lxml import html
+# Chọn số ngày cần lấy dữ liệu
+nums_date = 12
+tree = BeautifulSoup(browser.page_source, 'html.parser')
+
+def find_by_xpath(element_source,xpath_expression):
+    try: 
+        root = html.fromstring(element_source)
+        return root.xpath(xpath_expression) 
+    except:
+        return []
+
+def check_seat(browser, rows, type, i): # kiểm tra xem còn chỗ hay không: type = hạng vé (tương ứng là 1, 2, 3, 4 )
+    check = browser.find_element(By.XPATH, rows + f'[{i+1}]/div/div/div[2]/div[{type}]//p[contains(text(), "Hết chỗ")]')
+    try: 
+        if check: # Nếu hết chỗ thì return về xpath tương ứng
+            return rows + f'[{i+1}]/div/div/div[2]/div[{type}]/div//p/text()'
+    except: return rows + f'[{i+1}]/div/div/div[2]/div[{type}]/div/p[1]/text()'
+
+def click_with_js(element):
+    """Clicks an element using JavaScript"""
+    browser.execute_script("arguments[0].click();", element)
+
+
+class display_to_be_flex(object):
+    """An expectation for checking that an element's display property has been changed to 'flex'."""
+    def __init__(self, locator):
+        self.locator = locator
+
+    def __call__(self, driver):
+        try:
+            element = EC.presence_of_element_located(self.locator)(driver)
+            if driver.execute_script("return getComputedStyle(arguments[0]).display;", element) == "flex":
+                return element
+            else:
+                return False
+        except StaleElementReferenceException:
+            return False
+
+
+data = []
+root = '//*[@id="root"]/div[1]/div[2]/div/div/div/div[1]/div/div/div[4]/div[2]/div[1]/div'
+def get_rows(browser, addr):
+    sleep(1)
+
+    # chờ để thông tin các chuyến bay được load đầy đủ và hiển thị trong css
+    wait.until(display_to_be_flex((By.XPATH, '//*[@id="root"]/div[1]/div[2]/div/div/div/div[1]/div/div/div[4]/div[2]/div[2]')))
+    wait.until(EC.visibility_of_all_elements_located((By.XPATH, root)))
+    # browser.execute_script("document.querySelector('#my-button').click();")
+
+    # scroll xuống cuối trang để thấy hết các muibox element
+    browser.execute_script("window.scrollTo(0, document.body.scrollHeight);")
+    sleep(1)
+    # scroll lại lên đầu trang để bắt đầu crawl
+    browser.execute_script("window.scrollTo(0, 0)")
+    n_rows = len(find_by_xpath(browser.page_source, root))
+    print(n_rows)
+    for i in range(n_rows):
+        print(i, end = ' ')
+        click = True
+
+        ID = find_by_xpath(browser.page_source, 
+                        root + f'[{i+1}]/div/div/div[1]/div[1]/span/span/text()')[0]
+
+        # Thay đổi xpath để lấy text đúng trong trường hợp hết chỗ: div/p[1]/text() -> div//p//text()
+        skyBoss_business = find_by_xpath(browser.page_source, root + f'[{i+1}]/div/div/div[2]/div[1]/div//p//text()')[0]
+        skyBoss = find_by_xpath(browser.page_source, root + f'[{i+1}]/div/div/div[2]/div[2]/div//p//text()')[0]
+        Deluxe = find_by_xpath(browser.page_source, root + f'[{i+1}]/div/div/div[2]/div[3]/div//p//text()')[0]
+        Eco = find_by_xpath(browser.page_source, root + f'[{i+1}]/div/div/div[2]/div[4]/div//p//text()')[0]
+        if skyBoss_business == skyBoss == Deluxe == Eco == 'Hết chỗ':
+            click = False
+
+        row_elem = wait.until(EC.element_to_be_clickable((By.XPATH, root + f'[{i+1}]/div/div[1]/div[1]')))
+        if click:
+            click_with_js(row_elem)
+        sleep(0.01)
+
+        # scroll tới ô cần click
+        # browser.execute_script('return arguments[0].scrollIntoView(true);',
+        #                        wait.until(EC.element_to_be_clickable((By.XPATH, rows + f'[{i+1}]/div/div[1]/div[1]'))))
+
+
+        # scroll tới ô hiện ra thông tin
+        print('click = ', click)
+        if click:
+            browser.execute_script('return arguments[0].scrollIntoView(true);',
+                           wait.until(EC.element_to_be_clickable((By.XPATH, root + f'[{i+1}]/div/div[2]'))))
+        else: 
+            pass
+
+        hours = find_by_xpath(browser.page_source, root + f'[{i+1}]/div/div/div[1]/div[2]/text()')
+        date = find_by_xpath(browser.page_source, addr + '/div/div/div/p[2]/text()')[0]
+
+
+        # find the element with the string "Tổng tiền" using XPath
+        # total_fee = find_by_xpath(browser.page_source, '//h4[text()="Tổng tiền"]/following-sibling::h4/text()')
+        # print(total_fee)
+
+        start_ = find_by_xpath(browser.page_source, root + 
+    f'[{i+1}]/div/div[2]/div/div/div/div/div/div/div[2]/div/div[1]/div/div[2]/p[1][1]/text()')[0] if click else hours[0][:-1] + ' ' + date
+        end_ = find_by_xpath(browser.page_source, root + 
+    f'[{i+1}]/div/div[2]/div/div/div/div/div/div/div[2]/div/div[2]/div/div[2]/p[1][1]/text()')[0] if click else hours[1][1:] + ' ' + date
+        airbus = ''.join(element for element in find_by_xpath(browser.page_source, root + f'[{i+1}]/div/div/div[1]/div[3]/span/text()'))
+        # wait.until(EC.presence_of_element_located((By.XPATH, root + f'[{i+1}]/div/div[1]/div[1]'))).click()
+
+        data.append({'Số hiệu chuyến bay': ID, 'Khởi hành': start_, 'Đến': end_,'Airbus':airbus,
+                     'SkyBOSS_Business':skyBoss_business, 'SkyBOSS': skyBoss, 'Deluxe': Deluxe, 'Eco': Eco})
+        if click:
+            click_with_js(row_elem)
+    print('Đã load xong...\n')
+    # scroll lên đầu trang để chuyển qua ngày khác
+    browser.find_element(By.TAG_NAME, "body").send_keys(Keys.CONTROL + Keys.HOME)
+    return data
+
+slick_track = tree.find("div", class_="slick-track") 
+count = 0
+rows = '//*[@id="root"]/div[1]/div[2]/div/div/div/div[1]/div/div/div[4]/div[2]/div[1]/div'
+
+childrens = list(slick_track.children)[:nums_date]
+
+if find_by_xpath(browser.page_source, '//h5[contains(text(), "Không tìm thấy chuyến bay nào cho lựa chọn của bạn. Quay lại để chọn ngày khác.")]'):
+    childrens = list(slick_track.children)[1:nums_date+1]
+
+for child in childrens:
+    # print(f'{(int(child.attrs["data-index"])+1)*"."}')
+    sleep(1)
+    # if wait.until(EC.visibility_of_element_located((By.XPATH, '(//p[contains(text(), "Hết chỗ")])'))):
+    #     browser.execute_script("window.scrollTo(0, document.body.scrollHeight);")
+    #     wait.until(EC.visibility_of_element_located((By.XPATH, f'(//p[contains(text(), "Hết chỗ")]/ancestor::div/ancestor::div/following-sibling::div/div/p[contains(text(), "{tomorrow_day}")])'))).click()
+    #     wait.until(EC.visibility_of_element_located((By.XPATH, '//span[contains(text(), "Đi tiếp")]'))).click()
+
+    # browser.execute_script("window.scrollTo(0,0);")
+    addr = f'//*[@data-index="{int(child.attrs["data-index"])}"]'
+    next_day = wait.until(EC.element_to_be_clickable((By.XPATH, f'//*[@data-index="{int(child.attrs["data-index"])}"]')))
+    try:
+        next_day.click()
+    except:
+        next_day.click()
+
+    # print(len(find_by_xpath(browser.page_source, '//*[@id="root"]/div[1]/div[2]/div/div/div/div[1]/div/div/div[4]/div[2]/div[1]/div')))
+    count += len(find_by_xpath(browser.page_source, '//*[@id="root"]/div[1]/div[2]/div/div/div/div[1]/div/div/div[4]/div[2]/div[1]/div'))
+    # get_rows_selenium(browser)
+    get_rows(browser, addr)
+print('Tổng số chuyến bay: ', count)
+
+#########################################################################################################################################
+import re 
+import os
+
+flight = pd.DataFrame(data)
+ticket_class = ['SkyBOSS_Business', 'SkyBOSS', 'Deluxe', 'Eco']
+for i in ticket_class:
+    flight[i] = flight[i].apply(lambda x: np.nan if x =='Hết chỗ' else x.replace(',', '') + 'e3').astype(float)
+
+pattern = r'(\d{1,2})\s+tháng\s+(\d{1,2})'
+flight['Khởi hành'] = flight['Khởi hành'].str.replace('( \(Giờ địa phương\))',
+    '', regex=True).str.replace(pattern, lambda x: f"{int(x.group(1)):02}/{int(x.group(2)):02}/2023", regex=True).str.replace(',', '', regex=True)
+flight['Đến'] = flight['Đến'].str.replace('( \(Giờ địa phương\))',
+    '', regex=True).str.replace(pattern, lambda x: f"{int(x.group(1)):02}/{int(x.group(2)):02}/2023", regex=True).str.replace(',', '', regex=True)
+
+str_time = '%H:%M %d/%m/%Y'
+flight['Khởi hành'] = pd.to_datetime(flight['Khởi hành'], format=str_time)
+flight['Đến'] = pd.to_datetime(flight['Đến'], format=str_time)
+
+fee = 593000.0
+flight[['SkyBOSS_Business', 'SkyBOSS', 'Deluxe', 'Eco']] = flight[['SkyBOSS_Business', 'SkyBOSS', 'Deluxe', 'Eco']] * 1.1+ fee
+flight['Thuế_phí'] = 593.0
+
+path = today.strftime('%d%m%Y')
+
+if not os.path.exists(path):
+    # Create the directory
+    os.makedirs(path)
+    print(f"Directory '{path}' created successfully.")
+else:
+    print(f"Directory '{path}' already exists.")
+
+
+flight.to_csv(path + f'/{path}.csv')