def get_download_url(self, detail_page_driver):
download_url = None
try:
ele = detail_page_driver.find_element_by_xpath(self.DOWNLOAD_URL)
except selenium_exception.NoSuchElementException as e:
try:
detail_page_driver.refresh()
except selenium_exception.TimeoutException as e:
pass
try:
ele = detail_page_driver.find_element_by_xpath(self.DOWNLOAD_URL)
except selenium_exception.NoSuchElementException as e:
return download_url
return util.href2url(self.SOURCE, ele.get_attribute('href'), util.get_current_url(detail_page_driver))
def get_detail_urls(self, list_page_driver):
detail_urls = []
eles = list_page_driver.find_elements_by_xpath(self.DETAIL_URLS)
if len(eles) == 0:
try:
list_page_driver.refresh()
except selenium_exception.TimeoutException as e:
pass
eles = list_page_driver.find_elements_by_xpath(self.DETAIL_URLS)
for ele in eles:
detail_url = util.href2url(self.SOURCE, ele.get_attribute('href'), util.get_current_url(list_page_driver))
if detail_url:
detail_urls.append(detail_url)
return detail_urls
def get_download_url(self,detail_page_driver):
time.sleep(1.5)
download_url = None
try:
ele = detail_page_driver.find_element_by_xpath(self.DOWNLOAD_URL)
except selenium_exception.NoSuchElementException as e:
try:
detail_page_driver.refresh()
except selenium_exception.TimeoutException as e:
pass
try:
ele = detail_page_driver.find_element_by_xpath(self.DOWNLOAD_URL)
except selenium_exception.NoSuchElementException as e:
return download_url
return util.href2url(self.SOURCE,ele.get_attribute('href'),util.get_current_url(detail_page_driver))
def get_detail_urls(self, list_page_driver):
detail_urls = []
eles = list_page_driver.find_elements_by_xpath(self.DETAIL_URLS)
if len(eles) == 0:
try:
list_page_driver.refresh()
except selenium_exception.TimeoutException as e:
pass
eles = list_page_driver.find_elements_by_xpath(self.DETAIL_URLS)
for ele in eles:
detail_url = util.href2url(self.SOURCE, ele.get_attribute('href'),
util.get_current_url(list_page_driver))
if detail_url:
detail_urls.append(detail_url)
return detail_urls