def scrape_versions(self, chamber, session): chamber_letter = {"upper": "s", "lower": "h"}[chamber] versions_url = "ftp://ftp.cga.ct.gov/%s/tob/%s/" % (session, chamber_letter) with self.urlopen(versions_url) as page: files = parse_directory_listing(page) for f in files: match = re.match(r"^\d{4,4}([A-Z]+-\d{5,5})-(R\d\d)", f.filename) bill_id = match.group(1).replace("-", "") try: bill = self.bills[bill_id] except KeyError: continue url = versions_url + f.filename bill.add_version(match.group(2), url)
def scrape_versions(self, chamber, session): chamber_letter = {'upper': 's', 'lower': 'h'}[chamber] versions_url = "ftp://ftp.cga.ct.gov/%s/tob/%s/" % (session, chamber_letter) with self.urlopen(versions_url) as page: files = parse_directory_listing(page) for f in files: match = re.match(r'^\d{4,4}([A-Z]+-\d{5,5})-(R\d\d)', f.filename) bill_id = match.group(1).replace('-', '') try: bill = self.bills[bill_id] except KeyError: continue url = versions_url + f.filename bill.add_version(match.group(2), url)