def download_caselist(search_criteria, index, csv_file): cases = dict() wenshu = Spider.WenShu() wenshu.set_search_criteria(search_criteria) cases = wenshu.get_case_list(index) csv_file.write(cases) print('dump {} items into {}'.format(len(cases['doc_id']), cases)) print('Sleep 10s.....') time.sleep(2)
def download_all_caselist(search_criteria, max_page): cases = dict() wenshu = Spider.WenShu() wenshu.set_search_criteria(search_criteria) for index in range(1, max_page + 1): tmp_case_list = wenshu.get_case_list(index) if not cases: cases = tmp_case_list else: for key, value in tmp_case_list.items(): cases[key] += value print(f"{cases}") return cases
def download_all_caselist(search_criteria, max_page, csv_file): cases = dict() wenshu = Spider.WenShu() wenshu.set_search_criteria(search_criteria) for index in range(1, max_page + 1): tmp_case_list = wenshu.get_case_list(index) if not cases: cases = tmp_case_list else: for key, value in tmp_case_list.items(): cases[key] += value csv_file.write(cases) print('dump {} items into {}'.format(len(cases['doc_id']), cases)) print('Sleep 10s.....') time.sleep(2)
def get_court_list(mid_court_list): wenshu = Spider.WenShu() for mid_court in mid_court_list: wenshu.get_court_list(mid_court)
def download_case(case_id): wenshu = Spider.WenShu() return wenshu.get_case(case_id)
def get_total_number(search_criteria): wenshu = Spider.WenShu() wenshu.set_search_criteria(search_criteria) return wenshu.get_total_item_number()
def download_case(case_id): w = Spider.WenShu() return w.get_case(case_id)