def collect_price(self, driver, current_url):
# Load a file with specified conditions.
rows = collectors_utils.read_file(self.data_dir)
for row in rows:
try:
WebDriverWait(driver,30).until(EC.visibility_of_element_located((By.XPATH,'//select[@id="st_airplace"]')))
except:
print('element invisible')
driver.quit()
# Enterring the data of the imported file.
replace_btn = driver.find_element_by_xpath('//div[@id="btnReverse"]')
dep = row['dep']
des = row['des']
# Departure locations
dep_box = driver.find_element_by_xpath('//select[@id="st_airplace"]')
dep_box, select_options_dic = self.fetch_select_options(dep_box)
dep_box.select_by_value(select_options_dic[dep])
# Destination locations
des_box = driver.find_element_by_xpath('//select[@id="ar_airplace"]')
des_box, select_options_dic = self.fetch_select_options(des_box)
des_box.select_by_value(select_options_dic[des])
## dep-date
dep_date = driver.find_element_by_xpath(f'//input[@id="datepicker"]')
# set date
start = row['start']
end = row['end']
start_date_check = collectors_utils.check_date(start)
end_date_check = collectors_utils.check_date(end)
if start_date_check:
start = start.replace('/','/')
else:
start = collectors_utils.get_date()[0]
if end_date_check:
end = end.replace('/','/')
else:
end = collectors_utils.get_date()[1]
## enter departure date
self.set_dep_date(start, dep_date, driver)
# click search button
driver.find_element_by_xpath('//div[@class="search_btn"]').click()
#wait
# fetch table header
table_header = self.fetch_table_header(driver)
date_obj = datetime.now()
dir_date = datetime.strftime(date_obj,'%Y%m%d')
##
dir_name = f'片道-{dep.replace("/","")}-{des.replace("/","")}-{dir_date}'
if os.path.exists(f'{self.data_dir}/{dir_name}'):
pass
else:
os.mkdir(f'{self.data_dir}/{dir_name}')
while True:
current_date_text = driver.find_element_by_xpath('//div[@class="date"]').text
y, m, d, _ = re.sub(r'[年月日\(]', ' ', current_date_text).split()
endsplit = end.split('/')
file_name = f"{self.data_dir}/{dir_name}/{m}-{d}_{endsplit[1]}-{endsplit[2]}"
table_contents = self.fetch_table_contents(driver, table_header)
if table_contents is not None:
##
if self.page_feed(driver, end, current_date_text):
##
break
else:
collectors_utils.write_file(file_name, table_header, table_contents)
else:
##
if self.page_feed(driver, end, current_date_text):
break
else:
pass
driver.get(current_url)
driver.quit()
def collect_price(self, driver, current_url):
# Load a file with specified conditions.
rows = collectors_utils.read_file(self.data_dir)
for row in rows:
try:
WebDriverWait(driver, 30).until(
EC.visibility_of_element_located(
(By.XPATH, '//select[@class="pl10 departure"]')))
except:
print('element invisible')
driver.quit()
# Enterring the data of the imported file.
dep = row['dep']
des = row['des']
# Departure locations
dep_box = driver.find_element_by_xpath(
'//select[@class="pl10 departure"]')
dep_box, select_options_dic = self.fetch_select_options(dep_box)
dep_box.select_by_value(select_options_dic[dep])
# Destination locations
des_box = driver.find_element_by_xpath(
'//select[@class="pl10 arrive"]')
des_box, select_options_dic = self.fetch_select_options(des_box)
des_box.select_by_value(select_options_dic[des])
## dep-date
dep_date = driver.find_element_by_xpath(
f'//input[@aria-labelledby="aria-label-departure-date"]')
# set date
start = row['start']
end = row['end']
start_date_check = collectors_utils.check_date(start)
end_date_check = collectors_utils.check_date(end)
if start_date_check:
start = start.replace('/', '/')
else:
start = collectors_utils.get_date()[0]
if end_date_check:
end = end.replace('/', '/')
else:
end = collectors_utils.get_date()[1]
## enter departure date
#self.set_dep_date(start, dep_date, driver)
search_date = start.replace("/", '/')
driver.execute_script(
f'arguments[0].setAttribute("value","{search_date}")',
dep_date)
# click search button
driver.find_element_by_xpath(
'//div[@class="decide-btn btn-grad-orange"]').click()
# fetch table header
table_header = self.fetch_table_header(driver)
date_obj = datetime.now()
dir_date = datetime.strftime(date_obj, '%Y%m%d')
##
dir_name = f'片道-{dep.replace("/","")}-{des.replace("/","")}-{dir_date}'
if os.path.exists(f'{self.data_dir}/{dir_name}'):
pass
else:
os.mkdir(f'{self.data_dir}/{dir_name}')
while True:
current_date = driver.find_element_by_xpath(
'//div[@class="lowest_price_list_cell tx-c active selectable"]/p[1]'
)
current_date_text = current_date.text
m, d, _ = current_date_text.replace('(', '/').split('/')
endsplit = end.split('/')
file_name = f"{self.data_dir}/{dir_name}/{m}-{d}_{endsplit[1]}-{endsplit[2]}"
table_contents = self.fetch_table_contents(
driver, table_header)
if table_contents is not None:
##
if self.page_feed(driver, end, current_date, m, d):
##
break
else:
collectors_utils.write_file(file_name, table_header,
table_contents)
else:
##
if self.page_feed(driver, end, current_date, m, d):
break
else:
pass
driver.get(current_url)
driver.quit()
def collect_price(self, driver, current_url, way_num):
# Load a file with specified conditions.
rows = collectors_utils.read_file(self.data_dir)
for row in rows:
try:
WebDriverWait(driver, 30).until(
EC.visibility_of_element_located(
(By.XPATH, '//select[@id="departure_airport"]')))
except:
print('element invisible')
driver.quit()
# Enterring the data of the imported file.
dep_box = driver.find_element_by_xpath(
'//select[@id="departure_airport"]')
replace_btn = driver.find_element_by_xpath(
'//button[@id="set_reverse"]')
des_box = driver.find_element_by_xpath(
'//select[@id="arrival_airport"]')
dep = row['dep']
des = row['des']
# Departure locations
dep_box, select_options_dic = self.fetch_select_options(dep_box)
dep_box.select_by_value(select_options_dic[dep])
# Destination locations
des_box, select_options_dic = self.fetch_select_options(des_box)
des_box.select_by_value(select_options_dic[des])
if way_num == 0:
# dep-date
dep_date = driver.find_element_by_xpath(
f'//div[@class="search__ctrl__date"]/div[@id="date"]/input[@id="js-DP"]'
)
else:
# dep-date
dep_date = driver.find_element_by_xpath(
f'//div[@class="search__ctrl__date"]/div[@id="date"]/input[@id="js-DP"]'
)
# des-date
des_date = driver.find_element_by_xpath(
f'//div[@class="search__ctrl__date__arrival"]/div[@id="date2"]/input[@id="js-DP2"]'
)
# set date
start = row['start']
end = row['end']
start_date_check = collectors_utils.check_date(start)
end_date_check = collectors_utils.check_date(end)
if start_date_check:
start = start.replace('/', '/')
else:
start = collectors_utils.get_date()[0]
if end_date_check:
end = end.replace('/', '/')
else:
end = collectors_utils.get_date()[1]
if way_num == 0:
# enter departure date
driver.execute_script(
f"arguments[0].value = '{start.replace('/','/')}'",
dep_date)
else:
# enter departure date
driver.execute_script(
f"arguments[0].value = '{start.replace('/','/')}'",
dep_date)
# enter destination date
driver.execute_script(
f"arguments[0].value = '{end.replace('/','/')}'",
des_date)
# click search button
driver.find_element_by_xpath(
'//button[@id="js-btnSearchTicket"]').click()
# fetch table header
table_header = self.fetch_table_header(driver)
date_obj = datetime.now()
dir_date = datetime.strftime(date_obj, '%Y%m%d')
if way_num == 0:
dir_name = f'片道-{dep}-{des}-{dir_date}'
else:
dir_name = f'往復-{dep}-{des}-{dir_date}'
if os.path.exists(f'{self.data_dir}/{dir_name}'):
pass
else:
os.mkdir(f'{self.data_dir}/{dir_name}')
while True:
try:
WebDriverWait(driver, 3).until(
EC.visibility_of_element_located(
(By.XPATH, '//p[@class="date"]')))
except:
print('element invisible')
current_date_text = driver.find_element_by_xpath(
'//p[@class="date"]').text
current_date = current_date_text.replace(
'\u3000|\u3000',
'-').replace(' - ', '-').replace(' (',
'(').replace('/', '-')
current_date = f"{current_date.split('-')[1]}-{current_date.split('-')[2]}"
if way_num == 0:
file_name = f'{self.data_dir}/{dir_name}/{current_date}'
else:
endsplit = end.split('/')
file_name = f"{self.data_dir}/{dir_name}/{current_date}_{endsplit[1]}-{endsplit[2]}"
table_contents = self.fetch_table_contents(
driver, table_header)
if table_contents is not None:
if self.page_feed(driver, current_date_text, end):
if way_num == 1:
endsplit = end.split('/')
last_day = int(endsplit[-1]) + 1
end = f'{endsplit[1]}-{last_day}'
file_name = f"{self.data_dir}/{dir_name}/{current_date}_{end}"
collectors_utils.write_file(
file_name, table_header, table_contents)
break
else:
collectors_utils.write_file(file_name, table_header,
table_contents)
else:
if self.page_feed(driver, current_date_text, end):
break
else:
pass
driver.get(current_url)
driver.quit()
def collect_price(self, driver, current_url, way_num):
# Load a file with specified conditions.
rows = collectors_utils.read_file(self.data_dir)
for row in rows:
try:
WebDriverWait(driver, 30).until(
EC.visibility_of_element_located(
(By.XPATH, '//select[@id="ID_F1Departure"]')))
except:
print('element invisible')
driver.quit()
# Enterring the data of the imported file.
dep_box = driver.find_element_by_xpath(
'//select[@id="ID_F1Departure"]')
replace_btn = driver.find_element_by_xpath(
'//li[@class="flight_replace replace_btn"]')
des_box = driver.find_element_by_xpath(
'//select[@id="ID_F1Destination"]')
dep = row['dep']
des = row['des']
# Departure locations
driver.execute_script("arguments[0].click()", dep_box)
driver.find_element_by_xpath(
f'//li[@data-airport_name="{dep}"]').click()
# Destination locations
driver.execute_script("arguments[0].click()", des_box)
time.sleep(1)
driver.find_element_by_xpath(
f'//span[@id="des_list"]/ul/li[@data-airport_name="{des}"]'
).click()
time.sleep(1)
if way_num == 0:
# dep-date
dep_date = driver.find_element_by_xpath(
f'//li[@class="flight_date to-date"]/span[@class="from-trigger"]/span[@class="text-dep-date"]'
)
# des-date
des_date = driver.find_element_by_xpath(
f'//li[@class="flight_date returnDay"]/span[@class="to-trigger"]/span[@class="text-des-date"]'
)
else:
# dep-date
dep_date = driver.find_element_by_xpath(
f'//li[@class="flight_date to-date"]/span[@class="from-trigger"]/span[@class="text-dep-date"]'
)
# set date
start = row['start']
end = row['end']
start_date_check = collectors_utils.check_date(start)
end_date_check = collectors_utils.check_date(end)
if start_date_check:
start = start.replace('/', '/')
else:
start = collectors_utils.get_date()[0]
if end_date_check:
end = end.replace('/', '/')
else:
end = collectors_utils.get_date()[1]
if way_num == 0:
# enter departure date
driver.execute_script(f"arguments[0].innerHTML = '{start}'",
dep_date)
# enter destination date
driver.execute_script(f"arguments[0].innerHTML = '{end}'",
des_date)
else:
# enter departure date
driver.execute_script(f"arguments[0].innerHTML = '{start}'",
dep_date)
# click search button
driver.find_element_by_xpath(
'//input[@class="search_button"]').click()
# fetch table header
table_header = self.fetch_table_header(driver)
date_obj = datetime.now()
dir_date = datetime.strftime(date_obj, '%Y%m%d')
if way_num == 0:
dir_name = f'往復-{dep}-{des}-{dir_date}'
else:
dir_name = f'片道-{dep}-{des}-{dir_date}'
if os.path.exists(f'{self.data_dir}/{dir_name}'):
pass
else:
os.mkdir(f'{self.data_dir}/{dir_name}')
while True:
current_date_text = driver.find_element_by_xpath(
'//div[@id="showSearchBoxButton"]').text
current_date = current_date_text.replace(
'\u3000|\u3000',
'-').replace(' - ', '-').replace(' (',
'(').replace('/', '-')
current_date = f"{current_date.split('-')[2]}-{current_date.split('-')[3]}"
if way_num == 0:
endsplit = end.split('/')
file_name = f"{self.data_dir}/{dir_name}/{current_date}_{endsplit[1]}-{endsplit[2]}"
else:
file_name = f'{self.data_dir}/{dir_name}/{current_date}'
table_contents = self.fetch_table_contents(
driver, table_header)
if table_contents is not None:
if self.page_feed(driver, end):
if way_num == 0:
endsplit = end.split('/')
last_day = int(endsplit[-1]) + 1
end = f'{endsplit[1]}-{last_day}'
file_name = f"{self.data_dir}/{dir_name}/{current_date}_{end}"
collectors_utils.write_file(
file_name, table_header, table_contents)
break
else:
collectors_utils.write_file(file_name, table_header,
table_contents)
else:
if self.page_feed(driver, end):
break
else:
pass
driver.get(current_url)
driver.quit()
def collect_price(self, driver, current_url):
# Load a file with specified conditions.
rows = collectors_utils.read_file(self.data_dir)
for row in rows:
try:
WebDriverWait(driver, 3).until(
EC.visibility_of_element_located(
(By.XPATH, '//*[@id="flight-origin-hp-flight"]')))
except:
print('element invisible')
# Enterring the data of the imported file.
dep = row['dep']
des = row['des']
# Departure locations
dep_box = driver.find_element_by_xpath(
'//*[@id="flight-origin-flp"]')
dep_box.clear()
dep_box.send_keys(dep)
time.sleep(1)
dep_box.clear()
dep_box.send_keys(dep)
# Destination locations
des_box = driver.find_element_by_xpath(
'//*[@id="flight-destination-flp"]')
des_box.clear()
des_box.send_keys(des)
time.sleep(1)
des_box.clear()
des_box.send_keys(des)
# set date
start = row['start']
end = row['end']
start_date_check = collectors_utils.check_date(start)
end_date_check = collectors_utils.check_date(end)
if start_date_check:
start = start.replace('/', '/')
else:
start = collectors_utils.get_date()[0]
if end_date_check:
end = end.replace('/', '/')
else:
end = collectors_utils.get_date()[1]
## enter departure date
#self.set_dep_date(start, dep_date, driver)
search_date = start.replace("/", '/')
self.set_date(driver, search_date)
# click search button
search_btn = driver.find_element_by_xpath(
'//label[@class="col search-btn-col"]/button')
driver.execute_script('arguments[0].click()', search_btn)
# fetch table header
table_header = [i + 1 for i in range(19)]
date_obj = datetime.now()
dir_date = datetime.strftime(date_obj, '%Y%m%d')
##
dir_name = f'片道-{dep.replace("/","")}-{des.replace("/","")}-{dir_date}'
if os.path.exists(f'{self.data_dir}/{dir_name}'):
pass
else:
os.mkdir(f'{self.data_dir}/{dir_name}')
while True:
current_date = driver.find_element_by_xpath(
'//div[@class="title-date-rtv"]')
current_date_text = current_date.text
m, d, _ = current_date_text.replace(' 月 ',
' ').replace(' 日 ',
' ').split()
endsplit = end.split('/')
file_name = f"{self.data_dir}/{dir_name}/{m}-{d}_{endsplit[1]}-{endsplit[2]}"
table_contents = self.fetch_table_contents(
driver, table_header)
if table_contents is not None:
##
if self.page_feed(driver, end, current_date, m, d):
##
break
else:
collectors_utils.write_file(file_name, table_header,
table_contents)
else:
##
if self.page_feed(driver, end, current_date, m, d):
break
else:
pass
driver.get(current_url)
driver.quit()