def get_iframe(self): self.has_iframe, self.el_iframe = self.get_el(self.xpath_pdf) if self.has_iframe: self.pdf_url = norm_url(self.el_iframe.get_attribute("src")) else: self.driver.save_screenshot(self.pdf_file + ".png") return self.has_iframe
def get_pdf_url(self): html_a = self.html_tree.xpath(self.xpath_pdf_url) if len(html_a) == 0: print("\tPDF link for ", self.page_url, " not found") found = False self.pdf_url = None else: self.pdf_url = norm_url(html_a[0].attrib["href"]) found = True return found, self.pdf_url