util.c

关于tor匿名通信的源代码

  if (string < eos) {
    size_t line_len = (eos-string) + pCurLen + 2;
    char *line = tor_malloc(line_len);
    memcpy(line, prefixCur, pCurLen);
    memcpy(line+pCurLen, string, eos-string);
    line[line_len-2] = '\n';
    line[line_len-1] = '\0';
    smartlist_add(out, line);
  }
}

/* =====
 * Time
 * ===== */

/** Return the number of microseconds elapsed between *start and *end.
 */
long
tv_udiff(const struct timeval *start, const struct timeval *end)
{
  long udiff;
  long secdiff = end->tv_sec - start->tv_sec;

  if (labs(secdiff+1) > LONG_MAX/1000000) {
    log_warn(LD_GENERAL, "comparing times too far apart.");
    return LONG_MAX;
  }

  udiff = secdiff*1000000L + (end->tv_usec - start->tv_usec);
  return udiff;
}

/** Yield true iff <b>y</b> is a leap-year. */
#define IS_LEAPYEAR(y) (!(y % 4) && ((y % 100) || !(y % 400)))
/** Helper: Return the number of leap-days between Jan 1, y1 and Jan 1, y2. */
static int
n_leapdays(int y1, int y2)
{
  --y1;
  --y2;
  return (y2/4 - y1/4) - (y2/100 - y1/100) + (y2/400 - y1/400);
}
/** Number of days per month in non-leap year; used by tor_timegm. */
static const int days_per_month[] =
  { 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31};

/** Return a time_t given a struct tm.  The result is given in GMT, and
 * does not account for leap seconds.
 */
time_t
tor_timegm(struct tm *tm)
{
  /* This is a pretty ironclad timegm implementation, snarfed from Python2.2.
   * It's way more brute-force than fiddling with tzset().
   */
  time_t ret;
  unsigned long year, days, hours, minutes;
  int i;
  year = tm->tm_year + 1900;
  if (year < 1970 || tm->tm_mon < 0 || tm->tm_mon > 11) {
    log_warn(LD_BUG, "Out-of-range argument to tor_timegm");
    return -1;
  }
  tor_assert(year < INT_MAX);
  days = 365 * (year-1970) + n_leapdays(1970,(int)year);
  for (i = 0; i < tm->tm_mon; ++i)
    days += days_per_month[i];
  if (tm->tm_mon > 1 && IS_LEAPYEAR(year))
    ++days;
  days += tm->tm_mday - 1;
  hours = days*24 + tm->tm_hour;

  minutes = hours*60 + tm->tm_min;
  ret = minutes*60 + tm->tm_sec;
  return ret;
}

/* strftime is locale-specific, so we need to replace those parts */
static const char *WEEKDAY_NAMES[] =
  { "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat" };
static const char *MONTH_NAMES[] =
  { "Jan", "Feb", "Mar", "Apr", "May", "Jun",
    "Jul", "Aug", "Sep", "Oct", "Nov", "Dec" };

/** Set <b>buf</b> to the RFC1123 encoding of the GMT value of <b>t</b>.
 * The buffer must be at least RFC1123_TIME_LEN+1 bytes long.
 *
 * (RFC1123 format is Fri, 29 Sep 2006 15:54:20 GMT)
 */
void
format_rfc1123_time(char *buf, time_t t)
{
  struct tm tm;

  tor_gmtime_r(&t, &tm);

  strftime(buf, RFC1123_TIME_LEN+1, "___, %d ___ %Y %H:%M:%S GMT", &tm);
  tor_assert(tm.tm_wday >= 0);
  tor_assert(tm.tm_wday <= 6);
  memcpy(buf, WEEKDAY_NAMES[tm.tm_wday], 3);
  tor_assert(tm.tm_wday >= 0);
  tor_assert(tm.tm_mon <= 11);
  memcpy(buf+8, MONTH_NAMES[tm.tm_mon], 3);
}

/** Parse the the RFC1123 encoding of some time (in GMT) from <b>buf</b>,
 * and store the result in *<b>t</b>.
 *
 * Return 0 on succcess, -1 on failure.
*/
int
parse_rfc1123_time(const char *buf, time_t *t)
{
  struct tm tm;
  char month[4];
  char weekday[4];
  int i, m;

  if (strlen(buf) != RFC1123_TIME_LEN)
    return -1;
  memset(&tm, 0, sizeof(tm));
  if (sscanf(buf, "%3s, %d %3s %d %d:%d:%d GMT", weekday,
             &tm.tm_mday, month, &tm.tm_year, &tm.tm_hour,
             &tm.tm_min, &tm.tm_sec) < 7) {
    char *esc = esc_for_log(buf);
    log_warn(LD_GENERAL, "Got invalid RFC1123 time %s", esc);
    tor_free(esc);
    return -1;
  }

  m = -1;
  for (i = 0; i < 12; ++i) {
    if (!strcmp(month, MONTH_NAMES[i])) {
      m = i;
      break;
    }
  }
  if (m<0) {
    char *esc = esc_for_log(buf);
    log_warn(LD_GENERAL, "Got invalid RFC1123 time %s: No such month", esc);
    tor_free(esc);
    return -1;
  }
  tm.tm_mon = m;

  if (tm.tm_year < 1970) {
    char *esc = esc_for_log(buf);
    log_warn(LD_GENERAL,
             "Got invalid RFC1123 time %s. (Before 1970)", esc);
    tor_free(esc);
    return -1;
  }
  tm.tm_year -= 1900;

  *t = tor_timegm(&tm);
  return 0;
}

/** Set <b>buf</b> to the ISO8601 encoding of the local value of <b>t</b>.
 * The buffer must be at least ISO_TIME_LEN+1 bytes long.
 *
 * (ISO8601 format is 2006-10-29 10:57:20)
 */
void
format_local_iso_time(char *buf, time_t t)
{
  struct tm tm;
  strftime(buf, ISO_TIME_LEN+1, "%Y-%m-%d %H:%M:%S", tor_localtime_r(&t, &tm));
}

/** Set <b>buf</b> to the ISO8601 encoding of the GMT value of <b>t</b>.
 * The buffer must be at least ISO_TIME_LEN+1 bytes long.
 */
void
format_iso_time(char *buf, time_t t)
{
  struct tm tm;
  strftime(buf, ISO_TIME_LEN+1, "%Y-%m-%d %H:%M:%S", tor_gmtime_r(&t, &tm));
}

/** Given an ISO-formatted UTC time value (after the epoch) in <b>cp</b>,
 * parse it and store its value in *<b>t</b>.  Return 0 on success, -1 on
 * failure.  Ignore extraneous stuff in <b>cp</b> separated by whitespace from
 * the end of the time string. */
int
parse_iso_time(const char *cp, time_t *t)
{
  struct tm st_tm;
#ifdef HAVE_STRPTIME
  if (!strptime(cp, "%Y-%m-%d %H:%M:%S", &st_tm)) {
    log_warn(LD_GENERAL, "ISO time was unparseable by strptime"); return -1;
  }
#else
  unsigned int year=0, month=0, day=0, hour=100, minute=100, second=100;
  if (sscanf(cp, "%u-%u-%u %u:%u:%u", &year, &month,
                &day, &hour, &minute, &second) < 6) {
    log_warn(LD_GENERAL, "ISO time was unparseable"); return -1;
  }
  if (year < 1970 || month < 1 || month > 12 || day < 1 || day > 31 ||
          hour > 23 || minute > 59 || second > 61) {
    log_warn(LD_GENERAL, "ISO time was nonsensical"); return -1;
  }
  st_tm.tm_year = year-1900;
  st_tm.tm_mon = month-1;
  st_tm.tm_mday = day;
  st_tm.tm_hour = hour;
  st_tm.tm_min = minute;
  st_tm.tm_sec = second;
#endif
  if (st_tm.tm_year < 70) {
    char *esc = esc_for_log(cp);
    log_warn(LD_GENERAL, "Got invalid ISO time %s. (Before 1970)", esc);
    tor_free(esc);
    return -1;
  }
  *t = tor_timegm(&st_tm);
  return 0;
}

/** Given a <b>date</b> in one of the three formats allowed by HTTP (ugh),
 * parse it into <b>tm</b>.  Return 0 on success, negative on failure. */
int
parse_http_time(const char *date, struct tm *tm)
{
  const char *cp;
  char month[4];
  char wkday[4];
  int i;

  tor_assert(tm);
  memset(tm, 0, sizeof(*tm));

  /* First, try RFC1123 or RFC850 format: skip the weekday.  */
  if ((cp = strchr(date, ','))) {
    ++cp;
    if (sscanf(date, "%2d %3s %4d %2d:%2d:%2d GMT",
               &tm->tm_mday, month, &tm->tm_year,
               &tm->tm_hour, &tm->tm_min, &tm->tm_sec) == 6) {
      /* rfc1123-date */
      tm->tm_year -= 1900;
    } else if (sscanf(date, "%2d-%3s-%2d %2d:%2d:%2d GMT",
                      &tm->tm_mday, month, &tm->tm_year,
                      &tm->tm_hour, &tm->tm_min, &tm->tm_sec) == 6) {
      /* rfc850-date */
    } else {
      return -1;
    }
  } else {
    /* No comma; possibly asctime() format. */
    if (sscanf(date, "%3s %3s %2d %2d:%2d:%2d %4d",
               wkday, month, &tm->tm_mday,
               &tm->tm_hour, &tm->tm_min, &tm->tm_sec, &tm->tm_year) == 7) {
      tm->tm_year -= 1900;
    } else {
      return -1;
    }
  }

  month[3] = '\0';
  /* Okay, now decode the month. */
  for (i = 0; i < 12; ++i) {
    if (!strcasecmp(MONTH_NAMES[i], month)) {
      tm->tm_mon = i+1;
    }
  }

  if (tm->tm_year < 0 ||
      tm->tm_mon < 1  || tm->tm_mon > 12 ||
      tm->tm_mday < 0 || tm->tm_mday > 31 ||
      tm->tm_hour < 0 || tm->tm_hour > 23 ||
      tm->tm_min < 0  || tm->tm_min > 59 ||
      tm->tm_sec < 0  || tm->tm_sec > 61)
    return -1; /* Out of range, or bad month. */

  return 0;
}

/** Given an <b>interval</b> in seconds, try to write it to the
 * <b>out_len</b>-byte buffer in <b>out</b> in a human-readable form.
 * Return 0 on success, -1 on failure.
 */
int
format_time_interval(char *out, size_t out_len, long interval)
{
  /* We only report seconds if there's no hours. */
  long sec = 0, min = 0, hour = 0, day = 0;
  if (interval < 0)
    interval = -interval;

  if (interval >= 86400) {
    day = interval / 86400;
    interval %= 86400;
  }
  if (interval >= 3600) {
    hour = interval / 3600;
    interval %= 3600;
  }
  if (interval >= 60) {
    min = interval / 60;
    interval %= 60;
  }
  sec = interval;

  if (day) {
    return tor_snprintf(out, out_len, "%ld days, %ld hours, %ld minutes",
                        day, hour, min);
  } else if (hour) {
    return tor_snprintf(out, out_len, "%ld hours, %ld minutes", hour, min);
  } else if (min) {
    return tor_snprintf(out, out_len, "%ld minutes, %ld seconds", min, sec);
  } else {
    return tor_snprintf(out, out_len, "%ld seconds", sec);
  }
}

/* =====
 * Fuzzy time
 * XXXX021 Use this consistently or rip it out.
 * ===== */

/* In a perfect world, everybody would run ntp, and ntp would be perfect, so
 * if we wanted to know "Is the current time before time X?" we could just say
 * "time(NULL) < X".
 *
 * But unfortunately, many users are running Tor in an imperfect world, on
 * even more imperfect computers.  Hence, we need to track time oddly.  We
 * model the user's computer as being "skewed" from accurate time by
 * -<b>ftime_skew</b> seconds, such that our best guess of the current time is
 * time(NULL)+ftime_skew.  We also assume that our measurements of time may
 * have up to <b>ftime_slop</b> seconds of inaccuracy; IOW, our window of
 * estimate for the current time is now + ftime_skew +/- ftime_slop.
 */
static int ftime_skew = 0;
static int ftime_slop = 60;
void
ftime_set_maximum_sloppiness(int seconds)
{
  tor_assert(seconds >= 0);
  ftime_slop = seconds;
}
void
ftime_set_estimated_skew(int seconds)
{
  ftime_skew = seconds;
}
#if 0
void
ftime_get_window(time_t now, ftime_t *ft_out)
{
  ft_out->earliest = now + ftime_skew - ftime_slop;
  ft_out->latest =  now + ftime_skew + ftime_slop;
}
#endif
int
ftime_maybe_after(time_t now, time_t when)
{
  /* It may be after when iff the latest possible current time is after when */
  return (now + ftime_skew + ftime_slop) >= when;
}
int
ftime_maybe_before(time_t now, time_t when)
{
  /* It may be before when iff the earliest possible current time is before */
  return (now + ftime_skew - ftime_slop) < when;
}
int
ftime_definitely_after(time_t now, time_t when)
{
  /* It is definitely after when if the earliest time it could be is still
   * after when. */
  return (now + ftime_skew - ftime_slop) >= when;
}
int
ftime_definitely_before(time_t now, time_t when)
{
  /* It is definitely before when if the latest time it could be is still
   * before when. */
  return (now + ftime_skew + ftime_slop) < when;
}

/* =====
 * File helpers
 * ===== */

/** Write <b>count</b> bytes from <b>buf</b> to <b>fd</b>.  <b>isSocket</b>
 * must be 1 if fd was returned by socket() or accept(), and 0 if fd
 * was returned by open().  Return the number of bytes written, or -1
 * on error.  Only use if fd is a blocking fd.  */
int
write_all(int fd, const char *buf, size_t count, int isSocket)
{
  size_t written = 0;
  ssize_t result;
  tor_assert(count < INT_MAX); /*XXXX021 make returnval an ssize_t */

  while (written != count) {
    if (isSocket)
      result = tor_socket_send(fd, buf+written, count-written, 0);
    else
      result = write(fd, buf+written, count-written);
    if (result<0)
      return -1;
    written += result;
  }
  return (int)count;
}

/** Read from <b>fd</b> to <b>buf</b>, until we get <b>count</b> bytes
 * or reach the end of the file. <b>isSocket</b> must be 1 if fd
 * was returned by socket() or accept(), and 0 if fd was returned by
 * open().  Return the number of bytes read, or -1 on error. Only use
 * if fd is a blocking fd. */
int
read_all(int fd, char *buf, size_t count, int isSocket)
{
  /*XXXX021 return ssize_t. */
  size_t numread = 0;
  ssize_t result;

  if (count > SIZE_T_CEILING || count > INT_MAX)
    return -1;

  while (numread != count) {
    if (isSocket)
      result = tor_socket_recv(fd, buf+numread, count-numread, 0);
    else
      result = read(fd, buf+numread, count-numread);
    if (result<0)
      return -1;
    else if (result == 0)
      break;
    numread += result;
  }
  return (int)numread;
}

/*
 *    Filesystem operations.
 */

/** Clean up <b>name</b> so that we can use it in a call to "stat".  On Unix,
 * we do nothing.  On Windows, we remove a trailing slash, unless the path is
 * the root of a disk. */
static void
clean_name_for_stat(char *name)
{
#ifdef MS_WINDOWS
  size_t len = strlen(name);
  if (!len)
    return;
  if (name[len-1]=='\\' || name[len-1]=='/') {
    if (len == 1 || (len==3 && name[1]==':'))
      return;
    name[len-1]='\0';
  }
#else
  (void)name;
#endif
}

/** Return FN_ERROR if filename can't be read, FN_NOENT if it doesn't
 * exist, FN_FILE if it is a regular file, or FN_DIR if it's a
 * directory.  On FN_ERROR, sets errno. */
file_status_t
file_status(const char *fname)
{
  struct stat st;
  char *f;
  int r;