def parse_wizard_image(self, element, block_xpath, sample):
wizard = Component()
wizard.type = "WIZARD"
wizard.wizard_type = "WIZARD_IMAGE"
wizard.alignment = "LEFT"
block_xpath = self.extract_xpath(block_xpath)
inner_xpath = self.extract_xpath(sample.media_links[0].xpath)
for img in sample.media_links:
inner_xpath = self.great_common_prefix(
inner_xpath, self.extract_xpath(img.xpath))
inner_xpath = self.combine_xpath(inner_xpath[len(block_xpath):], True)
wizard.media_links = list()
img_list = element.xpath(inner_xpath)
for img in img_list:
wizard.media_links.append(
self.get_attr(img, sample.media_links[0].attr))
page_url_xpath = self.extract_xpath(
sample.page_url.xpath)[len(block_xpath):]
wizard.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):]
wizard.title = self.get_attr(
element.xpath(self.combine_xpath(title_xpath, True)),
sample.title.attr)
return wizard
def parse_wizard_news(self, element):
wizard = Component()
wizard.type = "WIZARD"
wizard.wizard_type = "WIZARD_NEWS"
wizard.alignment = "LEFT"
wizard.page_url = self.get_from_page(element, ".", "href")
wizard.title = self.get_from_page(element, ".", "string")
return wizard
def parse_wizard_image(self, element):
wizard = Component()
wizard.type = "WIZARD"
wizard.wizard_type = "WIZARD_IMAGE"
wizard.alignment = "LEFT"
img_list = element.xpath("./div/a/img")
wizard.media_links = list()
for img in img_list:
wizard.media_links.append(self.get_from_page(img, ".", "src"))
wizard.page_url = self.get_from_page(element, "./h3/a", "href")
wizard.title = self.get_from_page(element, "./h3/a", "string")
return wizard
def parse_wizard_news(self, element, block_xpath, sample):
wizard = Component()
wizard.type = "WIZARD"
wizard.wizard_type = "WIZARD_NEWS"
wizard.alignment = "LEFT"
block_xpath = self.extract_xpath(block_xpath)
page_url_xpath = self.extract_xpath(
sample.page_url.xpath)[len(block_xpath):]
wizard.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):]
wizard.title = self.get_attr(
element.xpath(self.combine_xpath(title_xpath, True)),
sample.title.attr)
return wizard