commit bash-20090528 snapshot

lib/readline/display.c

@@ -733,6 +733,7 @@ rl_redisplay ()
   if (MB_CUR_MAX > 1 && rl_byte_oriented == 0)
     {
       memset (&ps, 0, sizeof (mbstate_t));
+      /* XXX - what if wc_bytes ends up <= 0? check for MB_INVALIDCH */
       wc_bytes = mbrtowc (&wc, rl_line_buffer, rl_end, &ps);
     }
   else
@@ -893,6 +894,7 @@ rl_redisplay ()
       if (MB_CUR_MAX > 1 && rl_byte_oriented == 0)
 	{
 	  in += wc_bytes;
+	  /* XXX - what if wc_bytes ends up <= 0? check for MB_INVALIDCH */
 	  wc_bytes = mbrtowc (&wc, rl_line_buffer + in, rl_end - in, &ps);
 	}
       else
@@ -1189,9 +1191,11 @@ rl_redisplay ()
       if (t < out)
 	line[t - 1] = '>';
 
-      if (!rl_display_fixed || forced_display || lmargin != last_lmargin)
+      if (rl_display_fixed == 0 || forced_display || lmargin != last_lmargin)
 	{
 	  forced_display = 0;
+	  o_cpos = _rl_last_c_pos;
+	  cpos_adjusted = 0;
 	  update_line (&visible_line[last_lmargin],
 		       &invisible_line[lmargin],
 		       0,
@@ -1199,6 +1203,13 @@ rl_redisplay ()
 		       _rl_screenwidth + (lmargin ? 0 : wrap_offset),
 		       0);
 
+	  if ((MB_CUR_MAX > 1 && rl_byte_oriented == 0) &&
+	      cpos_adjusted == 0 &&
+	      _rl_last_c_pos != o_cpos &&
+	      _rl_last_c_pos > wrap_offset &&
+	      o_cpos < prompt_last_invisible)
+		_rl_last_c_pos -= prompt_invis_chars_first_line;	/* XXX - was wrap_offset */
+
 	  /* If the visible new line is shorter than the old, but the number
 	     of invisible characters is greater, and we are at the end of
 	     the new line, we need to clear to eol. */

lib/readline/display.c~

@@ -733,6 +733,7 @@ rl_redisplay ()
   if (MB_CUR_MAX > 1 && rl_byte_oriented == 0)
     {
       memset (&ps, 0, sizeof (mbstate_t));
+      /* XXX - what if wc_bytes ends up <= 0? check for MB_INVALIDCH */
       wc_bytes = mbrtowc (&wc, rl_line_buffer, rl_end, &ps);
     }
   else
@@ -893,6 +894,7 @@ rl_redisplay ()
       if (MB_CUR_MAX > 1 && rl_byte_oriented == 0)
 	{
 	  in += wc_bytes;
+	  /* XXX - what if wc_bytes ends up <= 0? check for MB_INVALIDCH */
 	  wc_bytes = mbrtowc (&wc, rl_line_buffer + in, rl_end - in, &ps);
 	}
       else
@@ -1189,9 +1191,11 @@ rl_redisplay ()
       if (t < out)
 	line[t - 1] = '>';
 
-      if (!rl_display_fixed || forced_display || lmargin != last_lmargin)
+      if (rl_display_fixed == 0 || forced_display || lmargin != last_lmargin)
 	{
 	  forced_display = 0;
+	  o_cpos = _rl_last_c_pos;
+	  cpos_adjusted = 0;
 	  update_line (&visible_line[last_lmargin],
 		       &invisible_line[lmargin],
 		       0,
@@ -1199,6 +1203,13 @@ rl_redisplay ()
 		       _rl_screenwidth + (lmargin ? 0 : wrap_offset),
 		       0);
 
+	  if (linenum == 0 && (MB_CUR_MAX > 1 && rl_byte_oriented == 0) &&
+	      cpos_adjusted == 0 &&
+	      _rl_last_c_pos != o_cpos &&
+	      _rl_last_c_pos > wrap_offset &&
+	      o_cpos < prompt_last_invisible)
+		_rl_last_c_pos -= prompt_invis_chars_first_line;	/* XXX - was wrap_offset */
+
 	  /* If the visible new line is shorter than the old, but the number
 	     of invisible characters is greater, and we are at the end of
 	     the new line, we need to clear to eol. */
@@ -1895,10 +1906,8 @@ _rl_move_cursor_relative (new, data)
   woff = WRAP_OFFSET (_rl_last_v_pos, wrap_offset);
   cpos = _rl_last_c_pos;
 
-#if 0
   if (cpos == 0 && cpos == new)
     return;
-#endif
 
 #if defined (HANDLE_MULTIBYTE)
   /* If we have multibyte characters, NEW is indexed by the buffer point in

lib/readline/mbutil.c

@@ -77,7 +77,7 @@ _rl_find_next_mbchar_internal (string, seed, count, find_non_zero)
      char *string;
      int seed, count, find_non_zero;
 {
-  size_t tmp;
+  size_t tmp, len;
   mbstate_t ps;
   int point;
   wchar_t wc;
@@ -99,7 +99,10 @@ _rl_find_next_mbchar_internal (string, seed, count, find_non_zero)
 
   while (count > 0)  
     {
-      tmp = mbrtowc (&wc, string+point, strlen(string + point), &ps);
+      len = strlen (string + point);
+      if (len == 0)
+	break;
+      tmp = mbrtowc (&wc, string+point, len, &ps);
       if (MB_INVALIDCH ((size_t)tmp))
 	{
 	  /* invalid bytes. assume a byte represents a character */

lib/readline/mbutil.c~

@@ -0,0 +1,372 @@
+/* mbutil.c -- readline multibyte character utility functions */
+
+/* Copyright (C) 2001-2009 Free Software Foundation, Inc.
+
+   This file is part of the GNU Readline Library (Readline), a library
+   for reading lines of text with interactive input and history editing.      
+
+   Readline 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 3 of the License, or
+   (at your option) any later version.
+
+   Readline 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 Readline.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+#define READLINE_LIBRARY
+
+#if defined (HAVE_CONFIG_H)
+#  include <config.h>
+#endif
+
+#include <sys/types.h>
+#include <fcntl.h>
+#include "posixjmp.h"
+
+#if defined (HAVE_UNISTD_H)
+#  include <unistd.h>	   /* for _POSIX_VERSION */
+#endif /* HAVE_UNISTD_H */
+
+#if defined (HAVE_STDLIB_H)
+#  include <stdlib.h>
+#else
+#  include "ansi_stdlib.h"
+#endif /* HAVE_STDLIB_H */
+
+#include <stdio.h>
+#include <ctype.h>
+
+/* System-specific feature definitions and include files. */
+#include "rldefs.h"
+#include "rlmbutil.h"
+
+#if defined (TIOCSTAT_IN_SYS_IOCTL)
+#  include <sys/ioctl.h>
+#endif /* TIOCSTAT_IN_SYS_IOCTL */
+
+/* Some standard library routines. */
+#include "readline.h"
+
+#include "rlprivate.h"
+#include "xmalloc.h"
+
+/* Declared here so it can be shared between the readline and history
+   libraries. */
+#if defined (HANDLE_MULTIBYTE)
+int rl_byte_oriented = 0;
+#else
+int rl_byte_oriented = 1;
+#endif
+
+/* **************************************************************** */
+/*								    */
+/*		Multibyte Character Utility Functions		    */
+/*								    */
+/* **************************************************************** */
+
+#if defined(HANDLE_MULTIBYTE)
+
+static int
+_rl_find_next_mbchar_internal (string, seed, count, find_non_zero)
+     char *string;
+     int seed, count, find_non_zero;
+{
+  size_t tmp, len;
+  mbstate_t ps;
+  int point;
+  wchar_t wc;
+
+  tmp = 0;
+
+  memset(&ps, 0, sizeof (mbstate_t));
+  if (seed < 0)
+    seed = 0;
+  if (count <= 0)
+    return seed;
+
+  point = seed + _rl_adjust_point (string, seed, &ps);
+  /* if this is true, means that seed was not pointing to a byte indicating
+     the beginning of a multibyte character.  Correct the point and consume
+     one char. */
+  if (seed < point)
+    count--;
+
+  while (count > 0)  
+    {
+      tmp = mbrtowc (&wc, string+point, strlen(string + point), &ps);
+      if (MB_INVALIDCH ((size_t)tmp))
+	{
+	  /* invalid bytes. assume a byte represents a character */
+	  point++;
+	  count--;
+	  /* reset states. */
+	  memset(&ps, 0, sizeof(mbstate_t));
+	}
+      else if (MB_NULLWCH (tmp))
+	break;			/* found wide '\0' */
+      else
+	{
+	  /* valid bytes */
+	  point += tmp;
+	  if (find_non_zero)
+	    {
+	      if (wcwidth (wc) == 0)
+		continue;
+	      else
+		count--;
+	    }
+	  else
+	    count--;
+	}
+    }
+
+  if (find_non_zero)
+    {
+      tmp = mbrtowc (&wc, string + point, strlen (string + point), &ps);
+      while (MB_NULLWCH (tmp) == 0 && MB_INVALIDCH (tmp) == 0 && wcwidth (wc) == 0)
+	{
+	  point += tmp;
+	  tmp = mbrtowc (&wc, string + point, strlen (string + point), &ps);
+	}
+    }
+
+  return point;
+}
+
+static int
+_rl_find_prev_mbchar_internal (string, seed, find_non_zero)
+     char *string;
+     int seed, find_non_zero;
+{
+  mbstate_t ps;
+  int prev, non_zero_prev, point, length;
+  size_t tmp;
+  wchar_t wc;
+
+  memset(&ps, 0, sizeof(mbstate_t));
+  length = strlen(string);
+  
+  if (seed < 0)
+    return 0;
+  else if (length < seed)
+    return length;
+
+  prev = non_zero_prev = point = 0;
+  while (point < seed)
+    {
+      tmp = mbrtowc (&wc, string + point, length - point, &ps);
+      if (MB_INVALIDCH ((size_t)tmp))
+	{
+	  /* in this case, bytes are invalid or shorted to compose
+	     multibyte char, so assume that the first byte represents
+	     a single character anyway. */
+	  tmp = 1;
+	  /* clear the state of the byte sequence, because
+	     in this case effect of mbstate is undefined  */
+	  memset(&ps, 0, sizeof (mbstate_t));
+
+	  /* Since we're assuming that this byte represents a single
+	     non-zero-width character, don't forget about it. */
+	  prev = point;
+	}
+      else if (MB_NULLWCH (tmp))
+	break;			/* Found '\0' char.  Can this happen? */
+      else
+	{
+	  if (find_non_zero)
+	    {
+	      if (wcwidth (wc) != 0)
+		prev = point;
+	    }
+	  else
+	    prev = point;  
+	}
+
+      point += tmp;
+    }
+
+  return prev;
+}
+
+/* return the number of bytes parsed from the multibyte sequence starting
+   at src, if a non-L'\0' wide character was recognized. It returns 0, 
+   if a L'\0' wide character was recognized. It  returns (size_t)(-1), 
+   if an invalid multibyte sequence was encountered. It returns (size_t)(-2) 
+   if it couldn't parse a complete  multibyte character.  */
+int
+_rl_get_char_len (src, ps)
+     char *src;
+     mbstate_t *ps;
+{
+  size_t tmp;
+
+  tmp = mbrlen((const char *)src, (size_t)strlen (src), ps);
+  if (tmp == (size_t)(-2))
+    {
+      /* shorted to compose multibyte char */
+      if (ps)
+	memset (ps, 0, sizeof(mbstate_t));
+      return -2;
+    }
+  else if (tmp == (size_t)(-1))
+    {
+      /* invalid to compose multibyte char */
+      /* initialize the conversion state */
+      if (ps)
+	memset (ps, 0, sizeof(mbstate_t));
+      return -1;
+    }
+  else if (tmp == (size_t)0)
+    return 0;
+  else
+    return (int)tmp;
+}
+
+/* compare the specified two characters. If the characters matched,
+   return 1. Otherwise return 0. */
+int
+_rl_compare_chars (buf1, pos1, ps1, buf2, pos2, ps2)
+     char *buf1;
+     int pos1;
+     mbstate_t *ps1;
+     char *buf2;
+     int pos2;
+     mbstate_t *ps2;
+{
+  int i, w1, w2;
+
+  if ((w1 = _rl_get_char_len (&buf1[pos1], ps1)) <= 0 || 
+	(w2 = _rl_get_char_len (&buf2[pos2], ps2)) <= 0 ||
+	(w1 != w2) ||
+	(buf1[pos1] != buf2[pos2]))
+    return 0;
+
+  for (i = 1; i < w1; i++)
+    if (buf1[pos1+i] != buf2[pos2+i])
+      return 0;
+
+  return 1;
+}
+
+/* adjust pointed byte and find mbstate of the point of string.
+   adjusted point will be point <= adjusted_point, and returns
+   differences of the byte(adjusted_point - point).
+   if point is invalied (point < 0 || more than string length),
+   it returns -1 */
+int
+_rl_adjust_point(string, point, ps)
+     char *string;
+     int point;
+     mbstate_t *ps;
+{
+  size_t tmp = 0;
+  int length;
+  int pos = 0;
+
+  length = strlen(string);
+  if (point < 0)
+    return -1;
+  if (length < point)
+    return -1;
+  
+  while (pos < point)
+    {
+      tmp = mbrlen (string + pos, length - pos, ps);
+      if (MB_INVALIDCH ((size_t)tmp))
+	{
+	  /* in this case, bytes are invalid or shorted to compose
+	     multibyte char, so assume that the first byte represents
+	     a single character anyway. */
+	  pos++;
+	  /* clear the state of the byte sequence, because
+	     in this case effect of mbstate is undefined  */
+	  if (ps)
+	    memset (ps, 0, sizeof (mbstate_t));
+	}
+      else if (MB_NULLWCH (tmp))
+	pos++;
+      else
+	pos += tmp;
+    }
+
+  return (pos - point);
+}
+
+int
+_rl_is_mbchar_matched (string, seed, end, mbchar, length)
+     char *string;
+     int seed, end;
+     char *mbchar;
+     int length;
+{
+  int i;
+
+  if ((end - seed) < length)
+    return 0;
+
+  for (i = 0; i < length; i++)
+    if (string[seed + i] != mbchar[i])
+      return 0;
+  return 1;
+}
+
+wchar_t
+_rl_char_value (buf, ind)
+     char *buf;
+     int ind;
+{
+  size_t tmp;
+  wchar_t wc;
+  mbstate_t ps;
+  int l;
+
+  if (MB_LEN_MAX == 1 || rl_byte_oriented)
+    return ((wchar_t) buf[ind]);
+  l = strlen (buf);
+  if (ind >= l - 1)
+    return ((wchar_t) buf[ind]);
+  memset (&ps, 0, sizeof (mbstate_t));
+  tmp = mbrtowc (&wc, buf + ind, l - ind, &ps);
+  if (MB_INVALIDCH (tmp) || MB_NULLWCH (tmp))  
+    return ((wchar_t) buf[ind]);
+  return wc;
+}
+#endif /* HANDLE_MULTIBYTE */
+
+/* Find next `count' characters started byte point of the specified seed.
+   If flags is MB_FIND_NONZERO, we look for non-zero-width multibyte
+   characters. */
+#undef _rl_find_next_mbchar
+int
+_rl_find_next_mbchar (string, seed, count, flags)
+     char *string;
+     int seed, count, flags;
+{
+#if defined (HANDLE_MULTIBYTE)
+  return _rl_find_next_mbchar_internal (string, seed, count, flags);
+#else
+  return (seed + count);
+#endif
+}
+
+/* Find previous character started byte point of the specified seed.
+   Returned point will be point <= seed.  If flags is MB_FIND_NONZERO,
+   we look for non-zero-width multibyte characters. */
+#undef _rl_find_prev_mbchar
+int
+_rl_find_prev_mbchar (string, seed, flags)
+     char *string;
+     int seed, flags;
+{
+#if defined (HANDLE_MULTIBYTE)
+  return _rl_find_prev_mbchar_internal (string, seed, flags);
+#else
+  return ((seed == 0) ? seed : seed - 1);
+#endif
+}

lib/readline/text.c

@@ -1251,6 +1251,7 @@ rl_change_case (count, op)
   wchar_t wc, nwc;
   char mb[MB_LEN_MAX+1];
   int mlen;
+  size_t m;
   mbstate_t mps;
 #endif
 
@@ -1303,7 +1304,11 @@ rl_change_case (count, op)
 #if defined (HANDLE_MULTIBYTE)
       else
 	{
-	  mbrtowc (&wc, rl_line_buffer + start, end - start, &mps);
+	  m = mbrtowc (&wc, rl_line_buffer + start, end - start, &mps);
+	  if (MB_INVALIDCH (m))
+	    wc = (wchar_t)rl_line_buffer[start];
+	  else if (MB_NULLWCH (m))
+	    wc = L'\0';
 	  nwc = (nop == UpCase) ? _rl_to_wupper (wc) : _rl_to_wlower (wc);
 	  if  (nwc != wc)	/*  just skip unchanged characters */
 	    {

lib/readline/vi_mode.c

@@ -769,6 +769,7 @@ _rl_vi_change_mbchar_case (count)
   wchar_t wc;
   char mb[MB_LEN_MAX+1];
   int mlen, p;
+  size_t m;
   mbstate_t ps;
 
   memset (&ps, 0, sizeof (mbstate_t));
@@ -776,7 +777,11 @@ _rl_vi_change_mbchar_case (count)
     count--;
   while (count-- && rl_point < rl_end)
     {
-      mbrtowc (&wc, rl_line_buffer + rl_point, rl_end - rl_point, &ps);
+      m = mbrtowc (&wc, rl_line_buffer + rl_point, rl_end - rl_point, &ps);
+      if (MB_INVALIDCH (m))
+	wc = (wchar_t)rl_line_buffer[rl_point];
+      else if (MB_NULLWCH (m))
+        wc = L'\0';
       if (iswupper (wc))
 	wc = towlower (wc);
       else if (iswlower (wc))

lib/readline/vi_mode.c~

@@ -131,9 +131,10 @@ static int _rl_vi_callback_char_search PARAMS((_rl_callback_generic_arg *));
 void
 _rl_vi_initialize_line ()
 {
-  register int i;
+  register int i, n;
 
-  for (i = 0; i < sizeof (vi_mark_chars) / sizeof (int); i++)
+  n = sizeof (vi_mark_chars) / sizeof (vi_mark_chars[0]);
+  for (i = 0; i < n; i++)
     vi_mark_chars[i] = -1;
 
   RL_UNSETSTATE(RL_STATE_VICMDONCE);

lib/sh/Makefile.in

@@ -526,6 +526,10 @@ casemod.o: ${topdir}/bashintl.h ${LIBINTL_H} ${BASHINCDIR}/gettext.h
 
 fdprintf.o: ${BASHINCDIR}/stdc.h
 
-input_avail.o: ${topdir}/bashansi.h ${BASHINCDIR}/ansi_stdlib.h
+input_avail.o: ${topdir}/bashansi.h ${BASHINCDIR}/ansi_stdlib.h${topdir}/bashansi.h ${BASHINCDIR}/ansi_stdlib.h${topdir}/bashansi.h ${BASHINCDIR}/ansi_stdlib.h${topdir}/bashansi.h ${BASHINCDIR}/ansi_stdlib.h
 input_avail.o: ${BASHINCDIR}/stdc.h
-input_avail.o: ${topdir}/xmalloc.h
+input_avail.o: ${topdir}/xmalloc.h ${BASHINCDIR}/posixselect.h
+
+mktime.o: ${topdir}/bashansi.h ${BASHINCDIR}/ansi_stdlib.h
+mktime.o: ${BASHINCDIR}/stdc.h
+

lib/sh/Makefile.in~

@@ -91,7 +91,7 @@ CSOURCES = clktck.c clock.c getcwd.c getenv.c oslib.c setlinebuf.c \
 	   strtoll.c strtoull.c strtoimax.c strtoumax.c memset.c strstr.c \
 	   mktime.c strftime.c mbschr.c zcatfd.c zmapfd.c winsize.c eaccess.c \
 	   wcsdup.c fpurge.c zgetline.c mbscmp.c uconvert.c ufuncs.c \
-	   casemod.c fdprintf.c input_avail.c
+	   casemod.c fdprintf.c input_avail.c mbscasecmp.c
 
 # The header files for this library.
 HSOURCES = 
@@ -105,7 +105,7 @@ OBJECTS = clktck.o clock.o getenv.o oslib.o setlinebuf.o strnlen.o \
 	  strtrans.o snprintf.o mailstat.o fmtulong.o \
 	  fmtullong.o fmtumax.o zcatfd.o zmapfd.o winsize.o wcsdup.o \
 	  fpurge.o zgetline.o mbscmp.o uconvert.o ufuncs.o casemod.o \
-	  fdprintf.o input_avail.o ${LIBOBJS}
+	  fdprintf.o input_avail.o mbscasecmp.o ${LIBOBJS}
 
 SUPPORT = Makefile
 
@@ -153,6 +153,7 @@ input_avail.o: input_avail.c
 itos.o: itos.c
 mailstat.o: mailstat.c
 makepath.o: makepath.c
+mbscasecmp.o: mbscasecmp.c
 mbscmp.o: mbscmp.c
 memset.o: memset.c
 mktime.o: mktime.c
@@ -221,6 +222,7 @@ input_avail.o: ${BUILD_DIR}/config.h
 itos.o: ${BUILD_DIR}/config.h
 mailstat.o: ${BUILD_DIR}/config.h
 makepath.o: ${BUILD_DIR}/config.h
+mbscasecmp.o: ${BUILD_DIR}/config.h
 mbscmp.o: ${BUILD_DIR}/config.h
 memset.o: ${BUILD_DIR}/config.h
 mktime.o: ${BUILD_DIR}/config.h
@@ -507,6 +509,10 @@ zgetline.o: ${BASHINCDIR}/stdc.h
 zgetline.o: ${topdir}/xmalloc.h
 zgetline.o: ${topdir}/bashtypes.h
 
+mbscasecmp.o: ${topdir}/bashansi.h ${BASHINCDIR}/ansi_stdlib.h
+mbscasecmp.o: ${BASHINCDIR}/stdc.h
+mbscasecmp.o: ${topdir}/xmalloc.h
+
 mbscmp.o: ${topdir}/bashansi.h ${BASHINCDIR}/ansi_stdlib.h
 mbscmp.o: ${BASHINCDIR}/stdc.h
 mbscmp.o: ${topdir}/xmalloc.h
@@ -522,4 +528,4 @@ fdprintf.o: ${BASHINCDIR}/stdc.h
 
 input_avail.o: ${topdir}/bashansi.h ${BASHINCDIR}/ansi_stdlib.h
 input_avail.o: ${BASHINCDIR}/stdc.h
-input_avail.o: ${topdir}/xmalloc.h
+input_avail.o: ${topdir}/xmalloc.h ${BASHINCDIR}/posixselect.h

lib/sh/casemod.c

@@ -107,6 +107,7 @@ sh_modcase (string, pat, flags)
   wchar_t nwc;
   char mb[MB_LEN_MAX+1];
   int mlen;
+  size_t m;
   mbstate_t state;
 #endif
 
@@ -210,7 +211,11 @@ sh_modcase (string, pat, flags)
 #if defined (HANDLE_MULTIBYTE)
       else
 	{
-	  mbrtowc (&wc, string + start, end - start, &state);
+	  m = mbrtowc (&wc, string + start, end - start, &state);
+	  if (MB_INVALIDCH (m))
+	    wc = (wchar_t)string[start];
+	  else if (MB_NULLWCH (m))
+	    wc = L'\0';
 	  switch (nop)
 	  {
 	  default:

lib/sh/casemod.c~

@@ -0,0 +1,243 @@
+/* casemod.c -- functions to change case of strings */
+
+/* Copyright (C) 2008,2009 Free Software Foundation, Inc.
+
+   This file is part of GNU Bash, the Bourne Again SHell.
+
+   Bash 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 3 of the License, or
+   (at your option) any later version.
+
+   Bash 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 Bash.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+#if defined (HAVE_CONFIG_H)
+#  include <config.h>
+#endif
+
+#if defined (HAVE_UNISTD_H)
+#  include <unistd.h>
+#endif /* HAVE_UNISTD_H */
+
+#include <stdc.h>
+
+#include <bashansi.h>
+#include <bashintl.h>
+#include <bashtypes.h>
+
+#include <stdio.h>
+#include <ctype.h>
+#include <xmalloc.h>
+
+#include <shmbutil.h>
+#include <chartypes.h>
+
+#include <glob/strmatch.h>
+
+#define _to_wupper(wc)	(iswlower (wc) ? towupper (wc) : (wc))
+#define _to_wlower(wc)	(iswupper (wc) ? towlower (wc) : (wc))
+
+#if !defined (HANDLE_MULTIBYTE)
+#  define cval(s, i)	((s)[(i)])
+#  define iswalnum(c)	(isalnum(c))
+#  define TOGGLE(x)	(ISUPPER (x) ? tolower (x) : (TOUPPER (x)))
+#else
+#  define TOGGLE(x)	(iswupper (x) ? towlower (x) : (_to_wupper(x)))
+#endif
+
+/* These must agree with the defines in externs.h */
+#define CASE_NOOP	0x0000
+#define CASE_LOWER	0x0001
+#define CASE_UPPER	0x0002
+#define CASE_CAPITALIZE	0x0004
+#define CASE_UNCAP	0x0008
+#define CASE_TOGGLE	0x0010
+#define CASE_TOGGLEALL	0x0020
+#define CASE_UPFIRST	0x0040
+#define CASE_LOWFIRST	0x0080
+
+#define CASE_USEWORDS	0x1000		/* modify behavior to act on words in passed string */
+
+extern char *substring __P((char *, int, int));
+
+#if defined (HANDLE_MULTIBYTE)
+static wchar_t
+cval (s, i)
+     char *s;
+     int i;
+{
+  size_t tmp;
+  wchar_t wc;
+  int l;
+  mbstate_t mps;  
+
+  if (MB_CUR_MAX == 1)
+    return ((wchar_t)s[i]);
+  l = strlen (s);
+  if (i >= (l - 1))
+    return ((wchar_t)s[i]);
+  memset (&mps, 0, sizeof (mbstate_t));
+  tmp = mbrtowc (&wc, s + i, l - i, &mps);
+  if (MB_INVALIDCH (tmp) || MB_NULLWCH (tmp))
+    return ((wchar_t)s[i]);
+  return wc;  
+}
+#endif
+
+/* Modify the case of characters in STRING matching PAT based on the value of
+   FLAGS.  If PAT is null, modify the case of each character */
+char *
+sh_modcase (string, pat, flags)
+     const char *string;
+     char *pat;
+     int flags;
+{
+  int start, next, end;
+  int inword, c, nc, nop, match, usewords;
+  char *ret, *s;
+  wchar_t wc;
+#if defined (HANDLE_MULTIBYTE)
+  wchar_t nwc;
+  char mb[MB_LEN_MAX+1];
+  int mlen, m;
+  mbstate_t state;
+#endif
+
+#if defined (HANDLE_MULTIBYTE)
+  memset (&state, 0, sizeof (mbstate_t));
+#endif
+
+  start = 0;
+  end = strlen (string);
+
+  ret = (char *)xmalloc (end + 1);
+  strcpy (ret, string);
+
+  /* See if we are supposed to split on alphanumerics and operate on each word */
+  usewords = (flags & CASE_USEWORDS);
+  flags &= ~CASE_USEWORDS;
+
+  inword = 0;
+  while (start < end)
+    {
+      wc = cval (ret, start);
+
+      if (iswalnum (wc) == 0)
+	{
+	  inword = 0;
+	  ADVANCE_CHAR (ret, end, start);
+	  continue;
+	}
+
+      if (pat)
+	{
+	  next = start;
+	  ADVANCE_CHAR (ret, end, next);
+	  s = substring (ret, start, next);
+	  match = strmatch (pat, s, FNM_EXTMATCH) != FNM_NOMATCH;
+	  free (s);
+	  if (match == 0)
+            {
+              start = next;
+              inword = 1;
+              continue;
+            }
+	}
+
+      /* XXX - for now, the toggling operators work on the individual
+	 words in the string, breaking on alphanumerics.  Should I
+	 leave the capitalization operators to do that also? */
+      if (flags == CASE_CAPITALIZE)
+	{
+	  if (usewords)
+	    nop = inword ? CASE_LOWER : CASE_UPPER;
+	  else
+	    nop = (start > 0) ? CASE_LOWER : CASE_UPPER;
+	  inword = 1;
+	}
+      else if (flags == CASE_UNCAP)
+	{
+	  if (usewords)
+	    nop = inword ? CASE_UPPER : CASE_LOWER;
+	  else
+	    nop = (start > 0) ? CASE_UPPER : CASE_LOWER;
+	  inword = 1;
+	}
+      else if (flags == CASE_UPFIRST)
+ 	{
+ 	  if (usewords)
+	    nop = inword ? CASE_NOOP : CASE_UPPER;
+	  else
+	    nop = (start > 0) ? CASE_NOOP : CASE_UPPER;
+ 	  inword = 1;
+ 	}
+      else if (flags == CASE_LOWFIRST)
+ 	{
+ 	  if (usewords)
+	    nop = inword ? CASE_NOOP : CASE_LOWER;
+	  else
+	    nop = (start > 0) ? CASE_NOOP : CASE_LOWER;
+ 	  inword = 1;
+ 	}
+      else if (flags == CASE_TOGGLE)
+	{
+	  nop = inword ? CASE_NOOP : CASE_TOGGLE;
+	  inword = 1;
+	}
+      else
+	nop = flags;
+
+      if (MB_CUR_MAX == 1 || isascii (wc))
+	{
+	  switch (nop)
+	  {
+	  default:
+	  case CASE_NOOP:  nc = wc; break;
+	  case CASE_UPPER:  nc = TOUPPER (wc); break;
+	  case CASE_LOWER:  nc = TOLOWER (wc); break;
+	  case CASE_TOGGLEALL:
+	  case CASE_TOGGLE: nc = TOGGLE (wc); break;
+	  }
+	  ret[start] = nc;
+	}
+#if defined (HANDLE_MULTIBYTE)
+      else
+	{
+	  m = mbrtowc (&wc, string + start, end - start, &state);
+	  if (MB_INVALIDCH (m))
+	    wc = (wchar_t)string[start];
+	  else if (MB_NULLWCH (m))
+	    wc = L'\0';
+	  switch (nop)
+	  {
+	  default:
+	  case CASE_NOOP:  nwc = wc; break;
+	  case CASE_UPPER:  nwc = TOUPPER (wc); break;
+	  case CASE_LOWER:  nwc = TOLOWER (wc); break;
+	  case CASE_TOGGLEALL:
+	  case CASE_TOGGLE: nwc = TOGGLE (wc); break;
+	  }
+	  if  (nwc != wc)	/*  just skip unchanged characters */
+	    {
+	      mlen = wcrtomb (mb, nwc, &state);
+	      if (mlen > 0)
+		mb[mlen] = '\0';
+	      /* Assume the same width */
+	      strncpy (ret + start, mb, mlen);
+	    }
+	}
+#endif
+
+      /*  This assumes that the upper and lower case versions are the same width. */
+      ADVANCE_CHAR (ret, end, start);
+    }
+
+  return ret;
+}