wget/src/cmpt.c

/* Replacements for routines missing on some systems.
   Copyright (C) 1995-2005 Free Software Foundation, Inc.

This file is part of GNU Wget.

GNU Wget is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.

GNU Wget is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with Wget; if not, write to the Free Software
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.

In addition, as a special exception, the Free Software Foundation
gives permission to link the code of its release of Wget with the
OpenSSL project's "OpenSSL" library (or with modified versions of it
that use the same license as the "OpenSSL" library), and distribute
the linked executables.  You must obey the GNU General Public License
in all respects for all of the code used other than "OpenSSL".  If you
modify this file, you may extend this exception to your version of the
file, but you are not obligated to do so.  If you do not wish to do
so, delete this exception statement from your version.  */

#include <config.h>

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#ifdef HAVE_UNISTD_H
# include <unistd.h>
#endif

#include <errno.h>

#include "wget.h"

/* Some systems lack certain functions normally taken for granted.
   For example, Windows doesn't have strptime, and some systems lack
   strcasecmp and strncasecmp.  This file should contain fallback
   implementations of the missing functions.  It should *not* define
   new Wget-specific functions -- those should placed in utils.c or
   elsewhere.  */

/* strcasecmp and strncasecmp apparently originated with BSD 4.4.
   SUSv3 seems to be the only standard out there (that I can find)
   that requires their existence, so there are systems that lack them
   still in use.  Note that these don't get defined under Windows
   because mswindows.h defines them to the equivalent Windows
   functions stricmp and strnicmp.  */

#ifndef HAVE_STRCASECMP
/* From GNU libc.  */
/* Compare S1 and S2, ignoring case, returning less than, equal to or
   greater than zero if S1 is lexiographically less than,
   equal to or greater than S2.  */
int
strcasecmp (const char *s1, const char *s2)
{
  register const unsigned char *p1 = (const unsigned char *) s1;
  register const unsigned char *p2 = (const unsigned char *) s2;
  unsigned char c1, c2;

  if (p1 == p2)
    return 0;

  do
    {
      c1 = TOLOWER (*p1++);
      c2 = TOLOWER (*p2++);
      if (c1 == '\0')
	break;
    }
  while (c1 == c2);

  return c1 - c2;
}
#endif /* not HAVE_STRCASECMP */

#ifndef HAVE_STRNCASECMP
/* From GNU libc.  */
/* Compare no more than N characters of S1 and S2,
   ignoring case, returning less than, equal to or
   greater than zero if S1 is lexicographically less
   than, equal to or greater than S2.  */
int
strncasecmp (const char *s1, const char *s2, size_t n)
{
  register const unsigned char *p1 = (const unsigned char *) s1;
  register const unsigned char *p2 = (const unsigned char *) s2;
  unsigned char c1, c2;

  if (p1 == p2 || n == 0)
    return 0;

  do
    {
      c1 = TOLOWER (*p1++);
      c2 = TOLOWER (*p2++);
      if (c1 == '\0' || c1 != c2)
	return c1 - c2;
    } while (--n > 0);

  return c1 - c2;
}
#endif /* not HAVE_STRNCASECMP */

/* strpbrk is a BSD 4.3 function mandated by POSIX and C99 and present
   on Windows.  It might be missing from older Unixes of non-BSD
   ancestry or possibly from non-Unix systems Wget gets ported to, so
   we add it here.  */

#ifndef HAVE_STRPBRK
/* Find the first ocurrence in S of any character in ACCEPT.  */
char *
strpbrk (const char *s, const char *accept)
{
  while (*s != '\0')
    {
      const char *a = accept;
      while (*a != '\0')
	if (*a++ == *s)
	  return (char *) s;
      ++s;
    }

  return 0;
}
#endif /* HAVE_STRPBRK */

/* mktime is a BSD 4.3 function also required by POSIX and C99.  I
   don't know if there is a widely used system that lacks it, so it
   might be a candidate for removal.  */

#ifndef HAVE_MKTIME
/* From GNU libc 2.0.  */

/* Copyright (C) 1993, 1994, 1995, 1996, 1997 Free Software Foundation, Inc.
   This file is part of the GNU C Library.
   Contributed by Paul Eggert (eggert@twinsun.com).  */

#ifdef _LIBC
# define HAVE_LIMITS_H 1
# define HAVE_LOCALTIME_R 1
# define STDC_HEADERS 1
#endif

/* Assume that leap seconds are possible, unless told otherwise.
   If the host has a `zic' command with a `-L leapsecondfilename' option,
   then it supports leap seconds; otherwise it probably doesn't.  */
#ifndef LEAP_SECONDS_POSSIBLE
# define LEAP_SECONDS_POSSIBLE 1
#endif

#ifndef __P
# define __P(args) args
#endif  /* Not __P.  */

#ifndef CHAR_BIT
# define CHAR_BIT 8
#endif

#ifndef INT_MIN
# define INT_MIN (~0 << (sizeof (int) * CHAR_BIT - 1))
#endif
#ifndef INT_MAX
# define INT_MAX (~0 - INT_MIN)
#endif

#ifndef TIME_T_MIN
/* The outer cast to time_t works around a bug in Cray C 5.0.3.0.  */
# define TIME_T_MIN ((time_t) \
		    (0 < (time_t) -1 ? (time_t) 0 \
		     : ~ (time_t) 0 << (sizeof (time_t) * CHAR_BIT - 1)))
#endif
#ifndef TIME_T_MAX
# define TIME_T_MAX (~ (time_t) 0 - TIME_T_MIN)
#endif

#define TM_YEAR_BASE 1900
#define EPOCH_YEAR 1970

#ifndef __isleap
/* Nonzero if YEAR is a leap year (every 4 years,
   except every 100th isn't, and every 400th is).  */
# define __isleap(year)	\
  ((year) % 4 == 0 && ((year) % 100 != 0 || (year) % 400 == 0))
#endif

/* How many days come before each month (0-12).  */
/* __mon_yday[][] is common to mktime and strptime implementations.
   --abbotti */
const unsigned short int __mon_yday[2][13];
#ifndef NEED_MON_YDAY
# define NEED_MON_YDAY
#endif

static time_t ydhms_tm_diff __P ((int, int, int, int, int, const struct tm *));
time_t __mktime_internal __P ((struct tm *,
			       struct tm *(*) (const time_t *, struct tm *),
			       time_t *));


#ifdef _LIBC
# define localtime_r __localtime_r
#else
# if ! HAVE_LOCALTIME_R && ! defined localtime_r
/* Approximate localtime_r as best we can in its absence.  */
#  define localtime_r my_mktime_localtime_r
static struct tm *localtime_r __P ((const time_t *, struct tm *));
static struct tm *
localtime_r (t, tp)
     const time_t *t;
     struct tm *tp;
{
  struct tm *l = localtime (t);
  if (! l)
    return 0;
  *tp = *l;
  return tp;
}
# endif /* ! HAVE_LOCALTIME_R && ! defined (localtime_r) */
#endif /* ! _LIBC */


/* Yield the difference between (YEAR-YDAY HOUR:MIN:SEC) and (*TP),
   measured in seconds, ignoring leap seconds.
   YEAR uses the same numbering as TM->tm_year.
   All values are in range, except possibly YEAR.
   If overflow occurs, yield the low order bits of the correct answer.  */
static time_t
ydhms_tm_diff (year, yday, hour, min, sec, tp)
     int year, yday, hour, min, sec;
     const struct tm *tp;
{
  /* Compute intervening leap days correctly even if year is negative.
     Take care to avoid int overflow.  time_t overflow is OK, since
     only the low order bits of the correct time_t answer are needed.
     Don't convert to time_t until after all divisions are done, since
     time_t might be unsigned.  */
  int a4 = (year >> 2) + (TM_YEAR_BASE >> 2) - ! (year & 3);
  int b4 = (tp->tm_year >> 2) + (TM_YEAR_BASE >> 2) - ! (tp->tm_year & 3);
  int a100 = a4 / 25 - (a4 % 25 < 0);
  int b100 = b4 / 25 - (b4 % 25 < 0);
  int a400 = a100 >> 2;
  int b400 = b100 >> 2;
  int intervening_leap_days = (a4 - b4) - (a100 - b100) + (a400 - b400);
  time_t years = year - (time_t) tp->tm_year;
  time_t days = (365 * years + intervening_leap_days
		 + (yday - tp->tm_yday));
  return (60 * (60 * (24 * days + (hour - tp->tm_hour))
		+ (min - tp->tm_min))
	  + (sec - tp->tm_sec));
}


static time_t localtime_offset;

/* Convert *TP to a time_t value.  */
time_t
mktime (tp)
     struct tm *tp;
{
#ifdef _LIBC
  /* POSIX.1 8.1.1 requires that whenever mktime() is called, the
     time zone names contained in the external variable `tzname' shall
     be set as if the tzset() function had been called.  */
  __tzset ();
#endif

  return __mktime_internal (tp, localtime_r, &localtime_offset);
}

/* Convert *TP to a time_t value, inverting
   the monotonic and mostly-unit-linear conversion function CONVERT.
   Use *OFFSET to keep track of a guess at the offset of the result,
   compared to what the result would be for UTC without leap seconds.
   If *OFFSET's guess is correct, only one CONVERT call is needed.  */
time_t
__mktime_internal (tp, convert, offset)
     struct tm *tp;
     struct tm *(*convert) __P ((const time_t *, struct tm *));
     time_t *offset;
{
  time_t t, dt, t0;
  struct tm tm;

  /* The maximum number of probes (calls to CONVERT) should be enough
     to handle any combinations of time zone rule changes, solar time,
     and leap seconds.  Posix.1 prohibits leap seconds, but some hosts
     have them anyway.  */
  int remaining_probes = 4;

  /* Time requested.  Copy it in case CONVERT modifies *TP; this can
     occur if TP is localtime's returned value and CONVERT is localtime.  */
  int sec = tp->tm_sec;
  int min = tp->tm_min;
  int hour = tp->tm_hour;
  int mday = tp->tm_mday;
  int mon = tp->tm_mon;
  int year_requested = tp->tm_year;
  int isdst = tp->tm_isdst;

  /* Ensure that mon is in range, and set year accordingly.  */
  int mon_remainder = mon % 12;
  int negative_mon_remainder = mon_remainder < 0;
  int mon_years = mon / 12 - negative_mon_remainder;
  int year = year_requested + mon_years;

  /* The other values need not be in range:
     the remaining code handles minor overflows correctly,
     assuming int and time_t arithmetic wraps around.
     Major overflows are caught at the end.  */

  /* Calculate day of year from year, month, and day of month.
     The result need not be in range.  */
  int yday = ((__mon_yday[__isleap (year + TM_YEAR_BASE)]
	       [mon_remainder + 12 * negative_mon_remainder])
	      + mday - 1);

  int sec_requested = sec;
#if LEAP_SECONDS_POSSIBLE
  /* Handle out-of-range seconds specially,
     since ydhms_tm_diff assumes every minute has 60 seconds.  */
  if (sec < 0)
    sec = 0;
  if (59 < sec)
    sec = 59;
#endif

  /* Invert CONVERT by probing.  First assume the same offset as last time.
     Then repeatedly use the error to improve the guess.  */

  tm.tm_year = EPOCH_YEAR - TM_YEAR_BASE;
  tm.tm_yday = tm.tm_hour = tm.tm_min = tm.tm_sec = 0;
  t0 = ydhms_tm_diff (year, yday, hour, min, sec, &tm);

  for (t = t0 + *offset;
       (dt = ydhms_tm_diff (year, yday, hour, min, sec, (*convert) (&t, &tm)));
       t += dt)
    if (--remaining_probes == 0)
      return -1;

  /* Check whether tm.tm_isdst has the requested value, if any.  */
  if (0 <= isdst && 0 <= tm.tm_isdst)
    {
      int dst_diff = (isdst != 0) - (tm.tm_isdst != 0);
      if (dst_diff)
	{
	  /* Move two hours in the direction indicated by the disagreement,
	     probe some more, and switch to a new time if found.
	     The largest known fallback due to daylight savings is two hours:
	     once, in Newfoundland, 1988-10-30 02:00 -> 00:00.  */
	  time_t ot = t - 2 * 60 * 60 * dst_diff;
	  while (--remaining_probes != 0)
	    {
	      struct tm otm;
	      if (! (dt = ydhms_tm_diff (year, yday, hour, min, sec,
					 (*convert) (&ot, &otm))))
		{
		  t = ot;
		  tm = otm;
		  break;
		}
	      if ((ot += dt) == t)
		break;  /* Avoid a redundant probe.  */
	    }
	}
    }

  *offset = t - t0;

#if LEAP_SECONDS_POSSIBLE
  if (sec_requested != tm.tm_sec)
    {
      /* Adjust time to reflect the tm_sec requested, not the normalized value.
	 Also, repair any damage from a false match due to a leap second.  */
      t += sec_requested - sec + (sec == 0 && tm.tm_sec == 60);
      (*convert) (&t, &tm);
    }
#endif

  if (TIME_T_MAX / INT_MAX / 366 / 24 / 60 / 60 < 3)
    {
      /* time_t isn't large enough to rule out overflows in ydhms_tm_diff,
	 so check for major overflows.  A gross check suffices,
	 since if t has overflowed, it is off by a multiple of
	 TIME_T_MAX - TIME_T_MIN + 1.  So ignore any component of
	 the difference that is bounded by a small value.  */

      double dyear = (double) year_requested + mon_years - tm.tm_year;
      double dday = 366 * dyear + mday;
      double dsec = 60 * (60 * (24 * dday + hour) + min) + sec_requested;

      if (TIME_T_MAX / 3 - TIME_T_MIN / 3 < (dsec < 0 ? - dsec : dsec))
	return -1;
    }

  *tp = tm;
  return t;
}

#ifdef weak_alias
weak_alias (mktime, timelocal)
#endif
#endif /* not HAVE_MKTIME */

/* strptime is required by POSIX, but it is missing from Windows,
   which means we must keep a fallback implementation.  It is
   reportedly missing or broken on many older systems as well.  */

#ifndef HAVE_STRPTIME
/* From GNU libc 2.1.3.  */
/* Ulrich, thanks for helping me out with this!  --hniksic  */

/* strptime - Convert a string representation of time to a time value.
   Copyright (C) 1996, 1997, 1998, 1999, 2000 Free Software Foundation, Inc.
   This file is part of the GNU C Library.
   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1996.  */

/* XXX This version of the implementation is not really complete.
   Some of the fields cannot add information alone.  But if seeing
   some of them in the same format (such as year, week and weekday)
   this is enough information for determining the date.  */

#ifndef __P
# define __P(args) args
#endif /* not __P */

#if ! HAVE_LOCALTIME_R && ! defined localtime_r
# ifdef _LIBC
#  define localtime_r __localtime_r
# else
/* Approximate localtime_r as best we can in its absence.  */
#  define localtime_r my_localtime_r
static struct tm *localtime_r __P ((const time_t *, struct tm *));
static struct tm *
localtime_r (t, tp)
     const time_t *t;
     struct tm *tp;
{
  struct tm *l = localtime (t);
  if (! l)
    return 0;
  *tp = *l;
  return tp;
}
# endif /* ! _LIBC */
#endif /* ! HAVE_LOCALTIME_R && ! defined (localtime_r) */


#define match_char(ch1, ch2) if (ch1 != ch2) return NULL
#if defined __GNUC__ && __GNUC__ >= 2
# define match_string(cs1, s2) \
  ({ size_t len = strlen (cs1);						      \
     int result = strncasecmp ((cs1), (s2), len) == 0;			      \
     if (result) (s2) += len;						      \
     result; })
#else
/* Oh come on.  Get a reasonable compiler.  */
# define match_string(cs1, s2) \
  (strncasecmp ((cs1), (s2), strlen (cs1)) ? 0 : ((s2) += strlen (cs1), 1))
#endif
/* We intentionally do not use isdigit() for testing because this will
   lead to problems with the wide character version.  */
#define get_number(from, to, n) \
  do {									      \
    int __n = n;							      \
    val = 0;								      \
    while (*rp == ' ')							      \
      ++rp;								      \
    if (*rp < '0' || *rp > '9')						      \
      return NULL;							      \
    do {								      \
      val *= 10;							      \
      val += *rp++ - '0';						      \
    } while (--__n > 0 && val * 10 <= to && *rp >= '0' && *rp <= '9');	      \
    if (val < from || val > to)						      \
      return NULL;							      \
  } while (0)
#ifdef _NL_CURRENT
/* Added check for __GNUC__ extensions here for Wget. --abbotti */
# if defined __GNUC__ && __GNUC__ >= 2
#  define get_alt_number(from, to, n) \
  ({									      \
    __label__ do_normal;						      \
    if (*decided != raw)						      \
      {									      \
	const char *alts = _NL_CURRENT (LC_TIME, ALT_DIGITS);		      \
	int __n = n;							      \
	int any = 0;							      \
	while (*rp == ' ')						      \
	  ++rp;								      \
	val = 0;							      \
	do {								      \
	  val *= 10;							      \
	  while (*alts != '\0')						      \
	    {								      \
	      size_t len = strlen (alts);				      \
	      if (strncasecmp (alts, rp, len) == 0)			      \
	        break;							      \
	      alts += len + 1;						      \
	      ++val;							      \
	    }								      \
	  if (*alts == '\0')						      \
	    {								      \
	      if (*decided == not && ! any)				      \
		goto do_normal;						      \
	      /* If we haven't read anything it's an error.  */		      \
	      if (! any)						      \
		return NULL;						      \
	      /* Correct the premature multiplication.  */		      \
	      val /= 10;						      \
	      break;							      \
	    }								      \
	  else								      \
	    *decided = loc;						      \
	} while (--__n > 0 && val * 10 <= to);				      \
	if (val < from || val > to)					      \
	  return NULL;							      \
      }									      \
    else								      \
      {									      \
       do_normal:							      \
        get_number (from, to, n);					      \
      }									      \
    0;									      \
  })
# else
#  define get_alt_number(from, to, n) \
  do {
    if (*decided != raw)						      \
      {									      \
	const char *alts = _NL_CURRENT (LC_TIME, ALT_DIGITS);		      \
	int __n = n;							      \
	int any = 0;							      \
	while (*rp == ' ')						      \
	  ++rp;								      \
	val = 0;							      \
	do {								      \
	  val *= 10;							      \
	  while (*alts != '\0')						      \
	    {								      \
	      size_t len = strlen (alts);				      \
	      if (strncasecmp (alts, rp, len) == 0)			      \
	        break;							      \
	      alts += len + 1;						      \
	      ++val;							      \
	    }								      \
	  if (*alts == '\0')						      \
	    {								      \
	      if (*decided == not && ! any)				      \
		goto do_normal;						      \
	      /* If we haven't read anything it's an error.  */		      \
	      if (! any)						      \
		return NULL;						      \
	      /* Correct the premature multiplication.  */		      \
	      val /= 10;						      \
	      break;							      \
	    }								      \
	  else								      \
	    *decided = loc;						      \
	} while (--__n > 0 && val * 10 <= to);				      \
	if (val < from || val > to)					      \
	  return NULL;							      \
      }									      \
    else								      \
      {									      \
       do_normal:							      \
        get_number (from, to, n);					      \
      }									      \
  } while (0)
# endif /* defined __GNUC__ && __GNUC__ >= 2 */
#else
# define get_alt_number(from, to, n) \
  /* We don't have the alternate representation.  */			      \
  get_number(from, to, n)
#endif
#define recursive(new_fmt) \
  (*(new_fmt) != '\0'							      \
   && (rp = strptime_internal (rp, (new_fmt), tm, decided)) != NULL)


#ifdef _LIBC
/* This is defined in locale/C-time.c in the GNU libc.  */
extern const struct locale_data _nl_C_LC_TIME;
extern const unsigned short int __mon_yday[2][13];

# define weekday_name (&_nl_C_LC_TIME.values[_NL_ITEM_INDEX (DAY_1)].string)
# define ab_weekday_name \
  (&_nl_C_LC_TIME.values[_NL_ITEM_INDEX (ABDAY_1)].string)
# define month_name (&_nl_C_LC_TIME.values[_NL_ITEM_INDEX (MON_1)].string)
# define ab_month_name (&_nl_C_LC_TIME.values[_NL_ITEM_INDEX (ABMON_1)].string)
# define HERE_D_T_FMT (_nl_C_LC_TIME.values[_NL_ITEM_INDEX (D_T_FMT)].string)
# define HERE_D_FMT (_nl_C_LC_TIME.values[_NL_ITEM_INDEX (D_FMT)].string)
# define HERE_AM_STR (_nl_C_LC_TIME.values[_NL_ITEM_INDEX (AM_STR)].string)
# define HERE_PM_STR (_nl_C_LC_TIME.values[_NL_ITEM_INDEX (PM_STR)].string)
# define HERE_T_FMT_AMPM \
  (_nl_C_LC_TIME.values[_NL_ITEM_INDEX (T_FMT_AMPM)].string)
# define HERE_T_FMT (_nl_C_LC_TIME.values[_NL_ITEM_INDEX (T_FMT)].string)

# define strncasecmp(s1, s2, n) __strncasecmp (s1, s2, n)
#else
static char const weekday_name[][10] =
  {
    "Sunday", "Monday", "Tuesday", "Wednesday",
    "Thursday", "Friday", "Saturday"
  };
static char const ab_weekday_name[][4] =
  {
    "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"
  };
static char const month_name[][10] =
  {
    "January", "February", "March", "April", "May", "June",
    "July", "August", "September", "October", "November", "December"
  };
static char const ab_month_name[][4] =
  {
    "Jan", "Feb", "Mar", "Apr", "May", "Jun",
    "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
  };
# define HERE_D_T_FMT "%a %b %e %H:%M:%S %Y"
# define HERE_D_FMT "%m/%d/%y"
# define HERE_AM_STR "AM"
# define HERE_PM_STR "PM"
# define HERE_T_FMT_AMPM "%I:%M:%S %p"
# define HERE_T_FMT "%H:%M:%S"

/* __mon_yday[][] is common to mktime and strptime implementations.
   --abbotti */
const unsigned short int __mon_yday[2][13];
# ifndef NEED_MON_YDAY
#  define NEED_MON_YDAY
# endif
#endif

/* Status of lookup: do we use the locale data or the raw data?  */
enum locale_status { not, loc, raw };


#ifndef __isleap
/* Nonzero if YEAR is a leap year (every 4 years,
   except every 100th isn't, and every 400th is).  */
# define __isleap(year)	\
  ((year) % 4 == 0 && ((year) % 100 != 0 || (year) % 400 == 0))
#endif

/* Compute the day of the week.  */
static void
day_of_the_week (struct tm *tm)
{
  /* We know that January 1st 1970 was a Thursday (= 4).  Compute the
     the difference between this data in the one on TM and so determine
     the weekday.  */
  int corr_year = 1900 + tm->tm_year - (tm->tm_mon < 2);
  int wday = (-473
	      + (365 * (tm->tm_year - 70))
	      + (corr_year / 4)
	      - ((corr_year / 4) / 25) + ((corr_year / 4) % 25 < 0)
	      + (((corr_year / 4) / 25) / 4)
	      + __mon_yday[0][tm->tm_mon]
	      + tm->tm_mday - 1);
  tm->tm_wday = ((wday % 7) + 7) % 7;
}

/* Compute the day of the year.  */
static void
day_of_the_year (struct tm *tm)
{
  tm->tm_yday = (__mon_yday[__isleap (1900 + tm->tm_year)][tm->tm_mon]
		 + (tm->tm_mday - 1));
}

static char *
#ifdef _LIBC
internal_function
#endif
strptime_internal __P ((const char *buf, const char *format, struct tm *tm,
			enum locale_status *decided));

static char *
#ifdef _LIBC
internal_function
#endif
strptime_internal (rp, fmt, tm, decided)
     const char *rp;
     const char *fmt;
     struct tm *tm;
     enum locale_status *decided;
{
#ifdef _NL_CURRENT
  const char *rp_backup;
#endif
  int cnt;
  size_t val;
  int have_I, is_pm;
  int century, want_century;
  int have_wday, want_xday;
  int have_yday;
  int have_mon, have_mday;

  have_I = is_pm = 0;
  century = -1;
  want_century = 0;
  have_wday = want_xday = have_yday = have_mon = have_mday = 0;

  while (*fmt != '\0')
    {
      /* A white space in the format string matches 0 more or white
	 space in the input string.  */
      if (ISSPACE (*fmt))
	{
	  while (ISSPACE (*rp))
	    ++rp;
	  ++fmt;
	  continue;
	}

      /* Any character but `%' must be matched by the same character
	 in the iput string.  */
      if (*fmt != '%')
	{
	  match_char (*fmt++, *rp++);
	  continue;
	}

      ++fmt;
#ifndef _NL_CURRENT
      /* We need this for handling the `E' modifier.  */
    start_over:
#endif

#ifdef _NL_CURRENT
      /* Make back up of current processing pointer.  */
      rp_backup = rp;
#endif

      switch (*fmt++)
	{
	case '%':
	  /* Match the `%' character itself.  */
	  match_char ('%', *rp++);
	  break;
	case 'a':
	case 'A':
	  /* Match day of week.  */
	  for (cnt = 0; cnt < 7; ++cnt)
	    {
#ifdef _NL_CURRENT
	      if (*decided !=raw)
		{
		  if (match_string (_NL_CURRENT (LC_TIME, DAY_1 + cnt), rp))
		    {
		      if (*decided == not
			  && strcmp (_NL_CURRENT (LC_TIME, DAY_1 + cnt),
				     weekday_name[cnt]))
			*decided = loc;
		      break;
		    }
		  if (match_string (_NL_CURRENT (LC_TIME, ABDAY_1 + cnt), rp))
		    {
		      if (*decided == not
			  && strcmp (_NL_CURRENT (LC_TIME, ABDAY_1 + cnt),
				     ab_weekday_name[cnt]))
			*decided = loc;
		      break;
		    }
		}
#endif
	      if (*decided != loc
		  && (match_string (weekday_name[cnt], rp)
		      || match_string (ab_weekday_name[cnt], rp)))
		{
		  *decided = raw;
		  break;
		}
	    }
	  if (cnt == 7)
	    /* Does not match a weekday name.  */
	    return NULL;
	  tm->tm_wday = cnt;
	  have_wday = 1;
	  break;
	case 'b':
	case 'B':
	case 'h':
	  /* Match month name.  */
	  for (cnt = 0; cnt < 12; ++cnt)
	    {
#ifdef _NL_CURRENT
	      if (*decided !=raw)
		{
		  if (match_string (_NL_CURRENT (LC_TIME, MON_1 + cnt), rp))
		    {
		      if (*decided == not
			  && strcmp (_NL_CURRENT (LC_TIME, MON_1 + cnt),
				     month_name[cnt]))
			*decided = loc;
		      break;
		    }
		  if (match_string (_NL_CURRENT (LC_TIME, ABMON_1 + cnt), rp))
		    {
		      if (*decided == not
			  && strcmp (_NL_CURRENT (LC_TIME, ABMON_1 + cnt),
				     ab_month_name[cnt]))
			*decided = loc;
		      break;
		    }
		}
#endif
	      if (match_string (month_name[cnt], rp)
		  || match_string (ab_month_name[cnt], rp))
		{
		  *decided = raw;
		  break;
		}
	    }
	  if (cnt == 12)
	    /* Does not match a month name.  */
	    return NULL;
	  tm->tm_mon = cnt;
	  want_xday = 1;
	  break;
	case 'c':
	  /* Match locale's date and time format.  */
#ifdef _NL_CURRENT
	  if (*decided != raw)
	    {
	      if (!recursive (_NL_CURRENT (LC_TIME, D_T_FMT)))
		{
		  if (*decided == loc)
		    return NULL;
		  else
		    rp = rp_backup;
		}
	      else
		{
		  if (*decided == not &&
		      strcmp (_NL_CURRENT (LC_TIME, D_T_FMT), HERE_D_T_FMT))
		    *decided = loc;
		  want_xday = 1;
		  break;
		}
	      *decided = raw;
	    }
#endif
	  if (!recursive (HERE_D_T_FMT))
	    return NULL;
	  want_xday = 1;
	  break;
	case 'C':
	  /* Match century number.  */
	  get_number (0, 99, 2);
	  century = val;
	  want_xday = 1;
	  break;
	case 'd':
	case 'e':
	  /* Match day of month.  */
	  get_number (1, 31, 2);
	  tm->tm_mday = val;
	  have_mday = 1;
	  want_xday = 1;
	  break;
	case 'F':
	  if (!recursive ("%Y-%m-%d"))
	    return NULL;
	  want_xday = 1;
	  break;
	case 'x':
#ifdef _NL_CURRENT
	  if (*decided != raw)
	    {
	      if (!recursive (_NL_CURRENT (LC_TIME, D_FMT)))
		{
		  if (*decided == loc)
		    return NULL;
		  else
		    rp = rp_backup;
		}
	      else
		{
		  if (*decided == not
		      && strcmp (_NL_CURRENT (LC_TIME, D_FMT), HERE_D_FMT))
		    *decided = loc;
		  want_xday = 1;
		  break;
		}
	      *decided = raw;
	    }
#endif
	  /* Fall through.  */
	case 'D':
	  /* Match standard day format.  */
	  if (!recursive (HERE_D_FMT))
	    return NULL;
	  want_xday = 1;
	  break;
	case 'k':
	case 'H':
	  /* Match hour in 24-hour clock.  */
	  get_number (0, 23, 2);
	  tm->tm_hour = val;
	  have_I = 0;
	  break;
	case 'I':
	  /* Match hour in 12-hour clock.  */
	  get_number (1, 12, 2);
	  tm->tm_hour = val % 12;
	  have_I = 1;
	  break;
	case 'j':
	  /* Match day number of year.  */
	  get_number (1, 366, 3);
	  tm->tm_yday = val - 1;
	  have_yday = 1;
	  break;
	case 'm':
	  /* Match number of month.  */
	  get_number (1, 12, 2);
	  tm->tm_mon = val - 1;
	  have_mon = 1;
	  want_xday = 1;
	  break;
	case 'M':
	  /* Match minute.  */
	  get_number (0, 59, 2);
	  tm->tm_min = val;
	  break;
	case 'n':
	case 't':
	  /* Match any white space.  */
	  while (ISSPACE (*rp))
	    ++rp;
	  break;
	case 'p':
	  /* Match locale's equivalent of AM/PM.  */
#ifdef _NL_CURRENT
	  if (*decided != raw)
	    {
	      if (match_string (_NL_CURRENT (LC_TIME, AM_STR), rp))
		{
		  if (strcmp (_NL_CURRENT (LC_TIME, AM_STR), HERE_AM_STR))
		    *decided = loc;
		  break;
		}
	      if (match_string (_NL_CURRENT (LC_TIME, PM_STR), rp))
		{
		  if (strcmp (_NL_CURRENT (LC_TIME, PM_STR), HERE_PM_STR))
		    *decided = loc;
		  is_pm = 1;
		  break;
		}
	      *decided = raw;
	    }
#endif
	  if (!match_string (HERE_AM_STR, rp))
	    if (match_string (HERE_PM_STR, rp))
	      is_pm = 1;
	    else
	      return NULL;
	  break;
	case 'r':
#ifdef _NL_CURRENT
	  if (*decided != raw)
	    {
	      if (!recursive (_NL_CURRENT (LC_TIME, T_FMT_AMPM)))
		{
		  if (*decided == loc)
		    return NULL;
		  else
		    rp = rp_backup;
		}
	      else
		{
		  if (*decided == not &&
		      strcmp (_NL_CURRENT (LC_TIME, T_FMT_AMPM),
			      HERE_T_FMT_AMPM))
		    *decided = loc;
		  break;
		}
	      *decided = raw;
	    }
#endif
	  if (!recursive (HERE_T_FMT_AMPM))
	    return NULL;
	  break;
	case 'R':
	  if (!recursive ("%H:%M"))
	    return NULL;
	  break;
	case 's':
	  {
	    /* The number of seconds may be very high so we cannot use
	       the `get_number' macro.  Instead read the number
	       character for character and construct the result while
	       doing this.  */
	    time_t secs = 0;
	    if (*rp < '0' || *rp > '9')
	      /* We need at least one digit.  */
	      return NULL;

	    do
	      {
		secs *= 10;
		secs += *rp++ - '0';
	      }
	    while (*rp >= '0' && *rp <= '9');

	    if (localtime_r (&secs, tm) == NULL)
	      /* Error in function.  */
	      return NULL;
	  }
	  break;
	case 'S':
	  get_number (0, 61, 2);
	  tm->tm_sec = val;
	  break;
	case 'X':
#ifdef _NL_CURRENT
	  if (*decided != raw)
	    {
	      if (!recursive (_NL_CURRENT (LC_TIME, T_FMT)))
		{
		  if (*decided == loc)
		    return NULL;
		  else
		    rp = rp_backup;
		}
	      else
		{
		  if (strcmp (_NL_CURRENT (LC_TIME, T_FMT), HERE_T_FMT))
		    *decided = loc;
		  break;
		}
	      *decided = raw;
	    }
#endif
	  /* Fall through.  */
	case 'T':
	  if (!recursive (HERE_T_FMT))
	    return NULL;
	  break;
	case 'u':
	  get_number (1, 7, 1);
	  tm->tm_wday = val % 7;
	  have_wday = 1;
	  break;
	case 'g':
	  get_number (0, 99, 2);
	  /* XXX This cannot determine any field in TM.  */
	  break;
	case 'G':
	  if (*rp < '0' || *rp > '9')
	    return NULL;
	  /* XXX Ignore the number since we would need some more
	     information to compute a real date.  */
	  do
	    ++rp;
	  while (*rp >= '0' && *rp <= '9');
	  break;
	case 'U':
	case 'V':
	case 'W':
	  get_number (0, 53, 2);
	  /* XXX This cannot determine any field in TM without some
	     information.  */
	  break;
	case 'w':
	  /* Match number of weekday.  */
	  get_number (0, 6, 1);
	  tm->tm_wday = val;
	  have_wday = 1;
	  break;
	case 'y':
	  /* Match year within century.  */
	  get_number (0, 99, 2);
	  /* The "Year 2000: The Millennium Rollover" paper suggests that
	     values in the range 69-99 refer to the twentieth century.  */
	  tm->tm_year = val >= 69 ? val : val + 100;
	  /* Indicate that we want to use the century, if specified.  */
	  want_century = 1;
	  want_xday = 1;
	  break;
	case 'Y':
	  /* Match year including century number.  */
	  get_number (0, 9999, 4);
	  tm->tm_year = val - 1900;
	  want_century = 0;
	  want_xday = 1;
	  break;
	case 'Z':
	  /* XXX How to handle this?  */
	  break;
	case 'E':
#ifdef _NL_CURRENT
	  switch (*fmt++)
	    {
	    case 'c':
	      /* Match locale's alternate date and time format.  */
	      if (*decided != raw)
		{
		  const char *fmt = _NL_CURRENT (LC_TIME, ERA_D_T_FMT);

		  if (*fmt == '\0')
		    fmt = _NL_CURRENT (LC_TIME, D_T_FMT);

		  if (!recursive (fmt))
		    {
		      if (*decided == loc)
			return NULL;
		      else
			rp = rp_backup;
		    }
		  else
		    {
		      if (strcmp (fmt, HERE_D_T_FMT))
			*decided = loc;
		      want_xday = 1;
		      break;
		    }
		  *decided = raw;
		}
	      /* The C locale has no era information, so use the
		 normal representation.  */
	      if (!recursive (HERE_D_T_FMT))
		return NULL;
	      want_xday = 1;
	      break;
	    case 'C':
	    case 'y':
	    case 'Y':
	      /* Match name of base year in locale's alternate
		 representation.  */
	      /* XXX This is currently not implemented.  It should
		 use the value _NL_CURRENT (LC_TIME, ERA).  */
	      break;
	    case 'x':
	      if (*decided != raw)
		{
		  const char *fmt = _NL_CURRENT (LC_TIME, ERA_D_FMT);

		  if (*fmt == '\0')
		    fmt = _NL_CURRENT (LC_TIME, D_FMT);

		  if (!recursive (fmt))
		    {
		      if (*decided == loc)
			return NULL;
		      else
			rp = rp_backup;
		    }
		  else
		    {
		      if (strcmp (fmt, HERE_D_FMT))
			*decided = loc;
		      break;
		    }
		  *decided = raw;
		}
	      if (!recursive (HERE_D_FMT))
		return NULL;
	      break;
	    case 'X':
	      if (*decided != raw)
		{
		  const char *fmt = _NL_CURRENT (LC_TIME, ERA_T_FMT);

		  if (*fmt == '\0')
		    fmt = _NL_CURRENT (LC_TIME, T_FMT);

		  if (!recursive (fmt))
		    {
		      if (*decided == loc)
			return NULL;
		      else
			rp = rp_backup;
		    }
		  else
		    {
		      if (strcmp (fmt, HERE_T_FMT))
			*decided = loc;
		      break;
		    }
		  *decided = raw;
		}
	      if (!recursive (HERE_T_FMT))
		return NULL;
	      break;
	    default:
	      return NULL;
	    }
	  break;
#else
	  /* We have no information about the era format.  Just use
	     the normal format.  */
	  if (*fmt != 'c' && *fmt != 'C' && *fmt != 'y' && *fmt != 'Y'
	      && *fmt != 'x' && *fmt != 'X')
	    /* This is an illegal format.  */
	    return NULL;

	  goto start_over;
#endif
	case 'O':
	  switch (*fmt++)
	    {
	    case 'd':
	    case 'e':
	      /* Match day of month using alternate numeric symbols.  */
	      get_alt_number (1, 31, 2);
	      tm->tm_mday = val;
	      have_mday = 1;
	      want_xday = 1;
	      break;
	    case 'H':
	      /* Match hour in 24-hour clock using alternate numeric
		 symbols.  */
	      get_alt_number (0, 23, 2);
	      tm->tm_hour = val;
	      have_I = 0;
	      break;
	    case 'I':
	      /* Match hour in 12-hour clock using alternate numeric
		 symbols.  */
	      get_alt_number (1, 12, 2);
	      tm->tm_hour = val - 1;
	      have_I = 1;
	      break;
	    case 'm':
	      /* Match month using alternate numeric symbols.  */
	      get_alt_number (1, 12, 2);
	      tm->tm_mon = val - 1;
	      have_mon = 1;
	      want_xday = 1;
	      break;
	    case 'M':
	      /* Match minutes using alternate numeric symbols.  */
	      get_alt_number (0, 59, 2);
	      tm->tm_min = val;
	      break;
	    case 'S':
	      /* Match seconds using alternate numeric symbols.  */
	      get_alt_number (0, 61, 2);
	      tm->tm_sec = val;
	      break;
	    case 'U':
	    case 'V':
	    case 'W':
	      get_alt_number (0, 53, 2);
	      /* XXX This cannot determine any field in TM without
		 further information.  */
	      break;
	    case 'w':
	      /* Match number of weekday using alternate numeric symbols.  */
	      get_alt_number (0, 6, 1);
	      tm->tm_wday = val;
	      have_wday = 1;
	      break;
	    case 'y':
	      /* Match year within century using alternate numeric symbols.  */
	      get_alt_number (0, 99, 2);
	      tm->tm_year = val >= 69 ? val : val + 100;
	      want_xday = 1;
	      break;
	    default:
	      return NULL;
	    }
	  break;
	default:
	  return NULL;
	}
    }

  if (have_I && is_pm)
    tm->tm_hour += 12;

  if (century != -1)
    {
      if (want_century)
	tm->tm_year = tm->tm_year % 100 + (century - 19) * 100;
      else
	/* Only the century, but not the year.  Strange, but so be it.  */
	tm->tm_year = (century - 19) * 100;
    }

  if (want_xday && !have_wday) {
      if ( !(have_mon && have_mday) && have_yday)  {
	  /* we don't have tm_mon and/or tm_mday, compute them */
	  int t_mon = 0;
	  while (__mon_yday[__isleap(1900 + tm->tm_year)][t_mon] <= tm->tm_yday)
	      t_mon++;
	  if (!have_mon)
	      tm->tm_mon = t_mon - 1;
	  if (!have_mday)
	      tm->tm_mday = tm->tm_yday - __mon_yday[__isleap(1900 + tm->tm_year)][t_mon - 1] + 1;
      }
      day_of_the_week (tm);
  }
  if (want_xday && !have_yday)
    day_of_the_year (tm);

  return (char *) rp;
}


char *
strptime (buf, format, tm)
     const char *buf;
     const char *format;
     struct tm *tm;
{
  enum locale_status decided;
#ifdef _NL_CURRENT
  decided = not;
#else
  decided = raw;
#endif
  return strptime_internal (buf, format, tm, &decided);
}
#endif /* not HAVE_STRPTIME */

#ifdef NEED_MON_YDAY
/* __mon_yday[][] is common to mktime and strptime implementations.
   --abbotti */
const unsigned short int __mon_yday[2][13] =
  {
    /* Normal years.  */
    { 0, 31, 59, 90, 120, 151, 181, 212, 243, 273, 304, 334, 365 },
    /* Leap years.  */
    { 0, 31, 60, 91, 121, 152, 182, 213, 244, 274, 305, 335, 366 }
  };
#endif

/* fnmatch is required by POSIX, but we include an implementation for
   the sake of systems that don't have it, most notably Windows.  Some
   systems do have fnmatch, but Apache's installation process installs
   its own fnmatch.h (incompatible with the system one!) in a system
   include directory, effectively rendering fnmatch unusable.  This
   has been fixed with Apache 2, where fnmatch has been moved to apr
   and given a prefix, but many systems out there are still (as of
   this writing in 2005) broken and we must cater to them.

   Additionally, according to anecdotal evidence and conventional
   wisdom I lack courage to challenge, many implementations of fnmatch
   are notoriously buggy and unreliable.  So we use our version by
   default, except when compiling under systems where fnmatch is known
   to work (currently on GNU libc-based systems and Solaris.)  */

#ifndef SYSTEM_FNMATCH

#define	__FNM_FLAGS	(FNM_PATHNAME | FNM_NOESCAPE | FNM_PERIOD)

/* Match STRING against the filename pattern PATTERN, returning zero
   if it matches, FNM_NOMATCH if not.  This implementation comes from
   an earlier version of GNU Bash.  (It doesn't make sense to update
   it with a newer version because those versions add a lot of
   features Wget doesn't use or care about.)  */

int
fnmatch (const char *pattern, const char *string, int flags)
{
  register const char *p = pattern, *n = string;
  register char c;

  if ((flags & ~__FNM_FLAGS) != 0)
    {
      errno = EINVAL;
      return (-1);
    }

  while ((c = *p++) != '\0')
    {
      switch (c)
	{
	case '?':
	  if (*n == '\0')
	    return (FNM_NOMATCH);
	  else if ((flags & FNM_PATHNAME) && *n == '/')
	    return (FNM_NOMATCH);
	  else if ((flags & FNM_PERIOD) && *n == '.' &&
		   (n == string || ((flags & FNM_PATHNAME) && n[-1] == '/')))
	    return (FNM_NOMATCH);
	  break;

	case '\\':
	  if (!(flags & FNM_NOESCAPE))
	    c = *p++;
	  if (*n != c)
	    return (FNM_NOMATCH);
	  break;

	case '*':
	  if ((flags & FNM_PERIOD) && *n == '.' &&
	      (n == string || ((flags & FNM_PATHNAME) && n[-1] == '/')))
	    return (FNM_NOMATCH);

	  for (c = *p++; c == '?' || c == '*'; c = *p++, ++n)
	    if (((flags & FNM_PATHNAME) && *n == '/') ||
		(c == '?' && *n == '\0'))
	      return (FNM_NOMATCH);

	  if (c == '\0')
	    return (0);

	  {
	    char c1 = (!(flags & FNM_NOESCAPE) && c == '\\') ? *p : c;
	    for (--p; *n != '\0'; ++n)
	      if ((c == '[' || *n == c1) &&
		  fnmatch (p, n, flags & ~FNM_PERIOD) == 0)
		return (0);
	    return (FNM_NOMATCH);
	  }

	case '[':
	  {
	    /* Nonzero if the sense of the character class is
	       inverted.  */
	    register int not;

	    if (*n == '\0')
	      return (FNM_NOMATCH);

	    if ((flags & FNM_PERIOD) && *n == '.' &&
		(n == string || ((flags & FNM_PATHNAME) && n[-1] == '/')))
	      return (FNM_NOMATCH);

	    /* Make sure there is a closing `]'.  If there isn't,
	       the `[' is just a character to be matched.  */
	    {
	      register const char *np;

	      for (np = p; np && *np && *np != ']'; np++);

	      if (np && !*np)
		{
		  if (*n != '[')
		    return (FNM_NOMATCH);
		  goto next_char;
		}
	    }

	    not = (*p == '!' || *p == '^');
	    if (not)
	      ++p;

	    c = *p++;
	    while (1)
	      {
		register char cstart = c, cend = c;

		if (!(flags & FNM_NOESCAPE) && c == '\\')
		  cstart = cend = *p++;

		if (c == '\0')
		  /* [ (unterminated) loses.  */
		  return (FNM_NOMATCH);

		c = *p++;

		if ((flags & FNM_PATHNAME) && c == '/')
		  /* [/] can never match.  */
		  return (FNM_NOMATCH);

		if (c == '-' && *p != ']')
		  {
		    cend = *p++;
		    if (!(flags & FNM_NOESCAPE) && cend == '\\')
		      cend = *p++;
		    if (cend == '\0')
		      return (FNM_NOMATCH);
		    c = *p++;
		  }

		if (*n >= cstart && *n <= cend)
		  goto matched;

		if (c == ']')
		  break;
	      }
	    if (!not)
	      return (FNM_NOMATCH);

	  next_char:
	    break;

	  matched:
	    /* Skip the rest of the [...] that already matched.  */
	    while (c != ']')
	      {
		if (c == '\0')
		  /* [... (unterminated) loses.  */
		  return (FNM_NOMATCH);

		c = *p++;
		if (!(flags & FNM_NOESCAPE) && c == '\\')
		  /* 1003.2d11 is unclear if this is right.  %%% */
		  ++p;
	      }
	    if (not)
	      return (FNM_NOMATCH);
	  }
	  break;

	default:
	  if (c != *n)
	    return (FNM_NOMATCH);
	}

      ++n;
    }

  if (*n == '\0')
    return (0);

  return (FNM_NOMATCH);
}

#endif /* not SYSTEM_FNMATCH */