def get_reviews_url_asin(self, goods_html):
result = ''
xpath_list = [
'//*[text()="Customer Reviews"]/../..//a[contains(text(), "customer reviews")]/@href',
]
result_list = ReviewsParser.get_new_data(xpath_list=xpath_list,
html_code=goods_html)
if len(result_list) > 0:
pattern_list = [
re.compile('product-reviews/([A-Za-z0-9]{10,10})/ref'),
]
result_list1 = ReviewsParser.get_new_data(
pattern_list=pattern_list, html_code=result_list[0])
if len(result_list1) > 0:
result = result_list1[0]
print('get_reviews_url.result: ', result)
return result
def looking_something(self, html_code):
not_found_patterns = [
re.compile('The Web address you entered is not a functioning',
re.S),
re.compile('Looking for something', re.S),
]
not_found = ReviewsParser.get_new_data(pattern_list=not_found_patterns,
html_code=html_code)
if len(not_found) > 0:
return True
return False