def data(self): result = [] for row in sel.elements("./div/table/tbody/tr/td", root=self.root): # Regular expressions? Boring. desc, date = sel.text(row).encode("utf-8").strip().rsplit("\n", 1) date = date.split(":", 1)[-1].strip() date = parsetime.from_iso_with_utc(date) result.append((desc, date)) return result
def data(self): result = [] for row in sel.elements("./div/table/tbody/tr/td", root=self.root): # Regular expressions? Boring. desc, date = sel.text(row).encode("utf-8").strip().rsplit("\n", 1) date = date.split(":", 1)[-1].strip() date = parsetime.from_iso_with_utc(date) url_source = sel.element("./..", root=row).get_attribute("onclick") getter_script = re.sub(r"^window.location\s*=\s*([^;]+;)", "return \\1", url_source) try: url = sel.execute_script(getter_script) except sel.WebDriverException: url = None result.append((desc, date, url)) return result