def get_product(self, brand, source_id, pid, url):
try:
print ('get_product', int(time.time()))
driver = None
driver = of_utils.create_chrome_driver()
driver.get(url)
if pid == -1: # entry
return self.parse_entry(driver)
else: # product
return self.parse_product(driver)
finally:
driver.quit()
def proc(self, brand, source_id, pid, url):
driver = None
try:
driver = of_utils.create_chrome_driver()
driver.get(url)
try:
if pid == -1: # entry
result = self.parse_entry(driver)
else: # product
result = self.parse_product(driver)
status = of_config.status_finished
except Exception as e:
result = traceback.format_exc()
status = of_config.status_failed
self.write_products(brand, url, source_id, pid, status, result)
except Exception as e:
self.logger.exception(traceback.format_exc())
finally:
if driver:
driver.quit()
import sys
import traceback
sys.path.append('.')
import of_utils
def parse_entry(driver):
elements = of_utils.find_elements_by_css_selector(
driver, 'div.product > div.inner > figure > a')
return [element.get_attribute('href').strip() for element in elements]
if __name__ == '__main__':
driver = None
try:
driver = of_utils.create_chrome_driver()
driver.get('https://www.giuseppezanotti.cn/woman/shoes')
products = parse_entry(driver)
print(products)
print(len(products))
except Exception as e:
print(e)
print(traceback.format_exc())
finally:
if driver:
driver.quit()