def clean_place(self, places): new_places = [] for place in places: for k, v in place.iteritems(): if k == 'amenity': place[k] = helper.clean_lf(v, u', ') elif k == 'page_body': pass else: place[k] = helper.clean_lf(v) new_places.append(place) return new_places
def process_item(self, item, spider): if not isinstance(item, HotelItem): return item for k, v in item.iteritems(): if k == 'phone': if v and len(v): v = helper.SEL_RE_PHONE_NUMBER.findall(v[0]) item[k] = helper.rev_telephone(v[0] if len(v) else u'') elif k == 'page_breadcrumbs': if v and len(v): item[k] = v[:len(v) - 1] if v else [] else: item[k] = [] elif k == 'place': item[k] = self.clean_place(v) else: item[k] = helper.clean_lf(v) LocationEs.save(self.save_item_to_es(item)) return item