def _get_protocol_and_headers(self, headerline, parts): headers = [] if headerline.startswith('filedesc://'): rec_type = 'warcinfo' else: rec_type = 'response' parts[3] = 'application/http;msgtype=response' headers.append(('WARC-Type', rec_type)) headers.append(('WARC-Record-ID', StatusAndHeadersParser.make_warc_id())) for name, value in zip(self.headernames, parts): if name == 'WARC-Date': value = timestamp_to_iso_date(value) if rec_type == 'warcinfo' and name == 'WARC-Target-URI': name = 'WARC-Filename' value = value[len('filedesc://'):] headers.append((name, value)) return ('WARC/1.0', headers)
def _make_warc_id(cls): return StatusAndHeadersParser.make_warc_id()