def fetch_filings(afterdate: str, beforedate: str, case_num_prefix: str) -> List[str]:
"Get filing case numbers between afterdate and beforedate and starting with case_num_prefix."
logger.info(
f"Scraping case numbers between {afterdate} and {beforedate} "
f"for prefix {case_num_prefix}..."
)
for tries in range(1, 11):
try:
filings_page_content = fetch_page.query_filings(
afterdate, beforedate, case_num_prefix
)
filings_soup = BeautifulSoup(filings_page_content, "html.parser")
filings_case_nums_list, query_needs_splitting = get_filing_case_nums(filings_soup)
break
except:
if tries == 10:
logger.error(f"Failed to find case numbers on all 10 attempts.")
# handle case of too many results (200 results means that the search cut off)
if query_needs_splitting:
try:
end_of_first_range, start_of_second_range = split_date_range(
afterdate, beforedate
)
filings_case_nums_list = fetch_filings(
afterdate, end_of_first_range, case_num_prefix
) + fetch_filings(start_of_second_range, beforedate, case_num_prefix)
except ValueError:
logger.error(
f"The search returned {len(filings_case_nums_list)} results but there's nothing "
"the code can do because beforedate and afterdate are the same.\n"
"Case details will be scraped for these results.\n"
)
# some logging to make sure results look good - could remove
logger.info(f"Found {len(filings_case_nums_list)} case numbers.")
if len(filings_case_nums_list) > 5:
logger.info(
f"Results preview: {filings_case_nums_list[0]}, {filings_case_nums_list[1]}, "
f"..., {filings_case_nums_list[-1]}\n"
)
else:
logger.info(f"Results: {', '.join(filings_case_nums_list)}\n")
return filings_case_nums_list
def test_fetch_filings_page(self):
fetched = fetch_page.query_filings(
afterdate="6-1-2020", beforedate="6-30-2020", case_num_prefix="J1-CV-20*"
)
assert "J1-CV-20-001773" in fetched