def parse_search_result(self, element, block_xpath, sample):
search_result = Component()
search_result.type = "SEARCH_RESULT"
search_result.alignment = "LEFT"
block_xpath = self.extract_xpath(block_xpath)
page_url_xpath = self.extract_xpath(
sample.page_url.xpath)[len(block_xpath):]
search_result.page_url = self.get_attr(
element.xpath(self.combine_xpath(page_url_xpath, True)),
sample.page_url.attr)
title_xpath = self.extract_xpath(sample.title.xpath)[len(block_xpath):]
search_result.title = self.get_attr(
element.xpath(self.combine_xpath(title_xpath, True)),
sample.title.attr)
snippet_xpath = self.extract_xpath(
sample.snippet.xpath)[len(block_xpath):]
search_result.snippet = self.get_attr(
element.xpath(self.combine_xpath(snippet_xpath, True)),
sample.snippet.attr)
view_url_xpath = self.extract_xpath(
sample.view_url.xpath)[len(block_xpath):]
search_result.view_url = self.get_attr(
element.xpath(self.combine_xpath(view_url_xpath, True)),
sample.view_url.attr)
return search_result
def parse_image(self, element):
image = Component()
image.type = "IMAGE"
image.alignment = "LEFT"
image.page_url = self.get_from_page(element, "./a", "href")
image.view_url = self.get_from_page(element, "./cite", "title")
image.title = self.get_from_page(element, ".", "string")
return image
def parse_adv(self, element):
adv = Component()
adv.type = "ADV"
adv.alignment = "LEFT"
adv.page_url = self.get_from_page(element, "./h3/a", "href")
adv.title = self.get_from_page(element, "./h3/a", "string")
adv.snippet = self.get_from_page(element, "./div[2]", "string")
adv.view_url = self.get_from_page(element, "./div/cite", "string")
return adv
def parse_search_result(self, element):
search_result = Component()
search_result.type = "SEARCH_RESULT"
search_result.alignment = "LEFT"
search_result.page_url = self.get_from_page(element, "./h3/a", "href")
search_result.title = self.get_from_page(element, "./h3/a", "string")
search_result.snippet = self.get_from_page(element, "./div/span", "strings")
search_result.view_url = self.get_from_page(element, "./div/div/cite", "string")
return search_result
def parse_adv(self, element):
adv = Component()
adv.type = "ADV"
adv.alignment = "LEFT"
adv.page_url = self.get_from_page(element, "./h2/a", "href")
adv.title = self.get_from_page(element, "./h2/a", "string")
adv.snippet = self.get_from_page(element, "./div[@class='organic__content-wrapper clearfix']/div[1]", "string")
adv.view_url = self.get_from_page(element, "./div[@class='organic__subtitle typo typo_type_greenurl']/div[1]/a", "string")
return adv