Function url_parse_http_regex

src/proxy/hdrs/URL.cc:1526–1589

}

ParseResult
url_parse_http_regex(HdrHeap *heap, URLImpl *url, const char **start, const char *end, bool copy_strings)
{
  const char *cur = *start;
  const char *host_end;

  // Do a quick check for "://" - our only format check.
  if (end - cur > 3 && (((':' ^ *cur) | ('/' ^ cur[1]) | ('/' ^ cur[2])) == 0)) {
    cur += 3;
  } else if (':' == *cur && (++cur >= end || ('/' == *cur && (++cur >= end || ('/' == *cur && ++cur >= end))))) {
    return PARSE_RESULT_ERROR;
  }

  // Grab everything until EOS or slash.
  const char *base = cur;
  cur              = static_cast<const char *>(memchr(cur, '/', end - cur));
  if (cur) {
    host_end = cur;
    // Remove all preceding slashes
    while (cur < end && *cur == '/') {
      cur++;
    }
  } else {
    host_end = cur = end;
  }

  // Did we find something for the host?
  if (base != host_end) {
    const char *port     = nullptr;
    int         port_len = 0;

    // Check for port. Search from the end stopping on the first non-digit
    // or more than 5 digits and a delimiter.
    port                   = host_end - 1;
    const char *port_limit = host_end - 6;
    if (port_limit < base) {
      port_limit = base; // don't go past start.
    }

    while (port >= port_limit && isdigit(*port)) {
      --port;
    }

    // A port if we're still in the host area and we found a ':' as
    // the immediately preceeding character.
    if (port >= base && ':' == *port) {
      port_len = host_end - port - 1; // must compute this first.
      host_end = port;                // then point at colon.
      ++port;                         // drop colon from port.
      url->set_port(heap, port, port_len, copy_strings);
    }

    // Now we can set the host.
    url->set_host(heap, base, host_end - base, copy_strings);
  }

  // path is anything that's left.
  if (cur < end) {