Submitted by:            Xi Ruoyao <xry111@xry111.site>
Date:                    2026-04-21
Initial Package Version: 9.11
Upstream Status:         Rejected
Origin:                  https://src.fedoraproject.org/rpms/coreutils/raw/0095d94d91e4/f/coreutils-i18n.patch
                         (originally for 9.8, rebased for 9.9,
                          edited for 9.10 and then 9.11)
Description:             Fixes i18n issues with pr and sort.

diff -Naur -i coreutils-9.10/src/pr.c coreutils-9.10.patched/src/pr.c
--- coreutils-9.10/src/pr.c	2026-01-31 16:20:52.000000000 -0600
+++ coreutils-9.10.patched/src/pr.c	2026-02-07 18:28:47.472251064 -0600
@@ -311,6 +311,24 @@
 
 #include <getopt.h>
 #include <sys/types.h>
+
+/* Get MB_LEN_MAX.  */
+#include <limits.h>
+/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC
+   installation; work around this configuration error.  */
+#if !defined MB_LEN_MAX || MB_LEN_MAX == 1
+# define MB_LEN_MAX 16
+#endif
+
+/* Get MB_CUR_MAX.  */
+#include <stdlib.h>
+
+/* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>.  */
+/* Get mbstate_t, mbrtowc(), wcwidth().  */
+#if HAVE_WCHAR_H
+# include <wchar.h>
+#endif
+
 #include "system.h"
 #include "c-ctype.h"
 #include "fadvise.h"
@@ -324,6 +342,18 @@
 #include "xstrtol-error.h"
 #include "xdectoint.h"
 
+/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t.  */
+#if HAVE_MBRTOWC && defined mbstate_t
+# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0)
+#endif
+
+#ifndef HAVE_DECL_WCWIDTH
+"this configure-time declaration test was not run"
+#endif
+#if !HAVE_DECL_WCWIDTH
+extern int wcwidth ();
+#endif
+
 /* The official name of this program (e.g., no 'g' prefix).  */
 #define PROGRAM_NAME "pr"
 
@@ -416,7 +446,20 @@
 
 typedef struct COLUMN COLUMN;
 
-static int char_to_clump (char c);
+/* Funtion pointers to switch functions for single byte locale or for
+   multibyte locale. If multibyte functions do not exist in your sysytem,
+   these pointers always point the function for single byte locale. */
+static void (*print_char) (char c);
+static int (*char_to_clump) (char c);
+
+/* Functions for single byte locale. */
+static void print_char_single (char c);
+static int char_to_clump_single (char c);
+
+/* Functions for multibyte locale. */
+static void print_char_multi (char c);
+static int char_to_clump_multi (char c);
+
 static bool read_line (COLUMN *p);
 static bool print_page (void);
 static bool print_stored (COLUMN *p);
@@ -427,6 +470,7 @@
 static void add_line_number (COLUMN *p);
 static int getoptnum (char const *n_str, int min, char const *errfmt);
 static void getoptarg (char *arg, char switch_char, char *character,
+                       int *character_length, int *character_width,
                        int *number);
 static void print_files (int number_of_files, char **av);
 static void init_parameters (int number_of_files);
@@ -440,7 +484,6 @@
 static void pad_down (unsigned int lines);
 static void read_rest_of_line (COLUMN *p);
 static void skip_read (COLUMN *p, int column_number);
-static void print_char (char c);
 static void cleanup (void);
 static void print_sep_string (void);
 static void separator_string (char const *optarg_S);
@@ -452,7 +495,7 @@
    we store the leftmost columns contiguously in buff.
    To print a line from buff, get the index of the first character
    from line_vector[i], and print up to line_vector[i + 1]. */
-static char *buff;
+static unsigned char *buff;
 
 /* Index of the position in buff where the next character
    will be stored. */
@@ -556,7 +599,7 @@
 static bool untabify_input = false;
 
 /* (-e) The input tab character. */
-static char input_tab_char = '\t';
+static char input_tab_char[MB_LEN_MAX] = "\t";
 
 /* (-e) Tabstops are at chars_per_tab, 2*chars_per_tab, 3*chars_per_tab, ...
    where the leftmost column is 1. */
@@ -566,7 +609,10 @@
 static bool tabify_output = false;
 
 /* (-i) The output tab character. */
-static char output_tab_char = '\t';
+static char output_tab_char[MB_LEN_MAX] = "\t";
+
+/* (-i) The byte length of output tab character. */
+static int output_tab_char_length = 1;
 
 /* (-i) The width of the output tab. */
 static int chars_per_output_tab = 8;
@@ -636,7 +682,13 @@
 static bool numbered_lines = false;
 
 /* (-n) Character which follows each line number. */
-static char number_separator = '\t';
+static char number_separator[MB_LEN_MAX] = "\t";
+
+/* (-n) The byte length of the character which follows each line number. */
+static int number_separator_length = 1;
+
+/* (-n) The character width of the character which follows each line number. */
+static int number_separator_width = 0;
 
 /* (-n) line counting starts with 1st line of input file (not with 1st
    line of 1st page printed). */
@@ -689,6 +741,7 @@
    -a|COLUMN|-m is a 'space' and with the -J option a 'tab'. */
 static char const *col_sep_string = "";
 static int col_sep_length = 0;
+static int col_sep_width = 0;
 static char *column_separator = (char *) " ";
 static char *line_separator = (char *) "\t";
 
@@ -851,6 +904,13 @@
     integer_overflow ();
   col_sep_length = len;
   col_sep_string = optarg_S;
+
+#if HAVE_MBRTOWC
+  if (MB_CUR_MAX > 1)
+    col_sep_width = mbswidth (col_sep_string, 0);
+  else
+#endif
+    col_sep_width = col_sep_length;
 }
 
 int
@@ -875,6 +935,21 @@
 
   atexit (close_stdout);
 
+/* Define which functions are used, the ones for single byte locale or the ones
+   for multibyte locale. */
+#if HAVE_MBRTOWC
+  if (MB_CUR_MAX > 1)
+    {
+      print_char = print_char_multi;
+      char_to_clump = char_to_clump_multi;
+    }
+  else
+#endif
+    {
+      print_char = print_char_single;
+      char_to_clump = char_to_clump_single;
+    }
+
   n_files = 0;
   file_names = (argc > 1
                 ? xnmalloc (argc - 1, sizeof (char *))
@@ -951,8 +1026,12 @@
           break;
         case 'e':
           if (optarg)
-            getoptarg (optarg, 'e', &input_tab_char,
-                       &chars_per_input_tab);
+            {
+              int dummy_length, dummy_width;
+
+              getoptarg (optarg, 'e', input_tab_char, &dummy_length,
+                         &dummy_width, &chars_per_input_tab);
+            }
           /* Could check tab width > 0. */
           untabify_input = true;
           break;
@@ -965,8 +1044,12 @@
           break;
         case 'i':
           if (optarg)
-            getoptarg (optarg, 'i', &output_tab_char,
-                       &chars_per_output_tab);
+            {
+              int dummy_width;
+
+              getoptarg (optarg, 'i', output_tab_char, &output_tab_char_length,
+                         &dummy_width, &chars_per_output_tab);
+            }
           /* Could check tab width > 0. */
           tabify_output = true;
           break;
@@ -985,8 +1068,8 @@
         case 'n':
           numbered_lines = true;
           if (optarg)
-            getoptarg (optarg, 'n', &number_separator,
-                       &chars_per_number);
+            getoptarg (optarg, 'n', number_separator, &number_separator_length,
+                       &number_separator_width, &chars_per_number);
           break;
         case 'N':
           skip_count = false;
@@ -1012,6 +1095,7 @@
           /* Reset an additional input of -s, -S dominates -s */
           col_sep_string = "";
           col_sep_length = 0;
+          col_sep_width = 0;
           use_col_separator = true;
           if (optarg)
             separator_string (optarg);
@@ -1167,7 +1251,8 @@
    a number. */
 
 static void
-getoptarg (char *arg, char switch_char, char *character, int *number)
+getoptarg (char *arg, char switch_char, char *character, int *character_length,
+           int *character_width, int *number)
 {
   if (!*arg)
     {
@@ -1176,7 +1261,41 @@
     }
 
   if (!c_isdigit (*arg))
-    *character = *arg++;
+    {
+#ifdef HAVE_MBRTOWC
+      if (MB_CUR_MAX > 1)        /* for multibyte locale. */
+        {
+          wchar_t wc;
+          size_t mblength;
+          int width;
+          mbstate_t state = {'\0'};
+
+          mblength = mbrtowc (&wc, arg, strnlen(arg, MB_LEN_MAX), &state);
+
+          if (mblength == (size_t)-1 || mblength == (size_t)-2)
+            {
+              *character_length = 1;
+              *character_width = 1;
+            }
+          else
+            {
+              *character_length = (mblength < 1) ? 1 : mblength;
+              width = wcwidth (wc);
+              *character_width = (width < 0) ? 0 : width;
+            }
+
+          strncpy (character, arg, *character_length);
+          arg += *character_length;
+        }
+      else                        /* for single byte locale. */
+#endif
+        {
+          *character = *arg++;
+          *character_length = 1;
+          *character_width = 1;
+        }
+    }
+
   if (*arg)
     {
       long int tmp_long;
@@ -1205,6 +1324,11 @@
 init_parameters (int number_of_files)
 {
   int chars_used_by_number = 0;
+  int mb_len = 1;
+#if HAVE_MBRTOWC
+  if (MB_CUR_MAX > 1)
+    mb_len = MB_LEN_MAX;
+#endif
 
   lines_per_body = lines_per_page - lines_per_header - lines_per_footer;
   if (lines_per_body <= 0)
@@ -1242,7 +1366,7 @@
           else
             col_sep_string = column_separator;
 
-          col_sep_length = 1;
+          col_sep_length = col_sep_width = 1;
           use_col_separator = true;
         }
       /* It's rather pointless to define a TAB separator with column
@@ -1274,11 +1398,11 @@
              + TAB_WIDTH (chars_per_input_tab, chars_per_number);   */
 
       /* Estimate chars_per_text without any margin and keep it constant. */
-      if (number_separator == '\t')
+      if (number_separator[0] == '\t')
         number_width = (chars_per_number
                         + TAB_WIDTH (chars_per_default_tab, chars_per_number));
       else
-        number_width = chars_per_number + 1;
+        number_width = chars_per_number + number_separator_width;
 
       /* The number is part of the column width unless we are
          printing files in parallel. */
@@ -1287,7 +1411,7 @@
     }
 
   int sep_chars, useful_chars;
-  if (ckd_mul (&sep_chars, columns - 1, col_sep_length))
+  if (ckd_mul (&sep_chars, columns - 1, col_sep_width))
     sep_chars = INT_MAX;
   if (ckd_sub (&useful_chars, chars_per_line - chars_used_by_number,
                sep_chars))
@@ -1310,7 +1434,7 @@
      We've to use 8 as the lower limit, if we use chars_per_default_tab = 8
      to expand a tab which is not an input_tab-char. */
   free (clump_buff);
-  clump_buff = xmalloc (MAX (8, chars_per_input_tab));
+  clump_buff = xmalloc (mb_len * MAX (8, chars_per_input_tab));
 }
 
 /* Open the necessary files,
@@ -1416,7 +1540,7 @@
 
   /* Enlarge p->start_position of first column to use the same form of
      padding_not_printed with all columns. */
-  h = h + col_sep_length;
+  h = h + col_sep_width;
 
   /* This loop takes care of all but the rightmost column. */
 
@@ -1450,7 +1574,7 @@
         }
       else
         {
-          h = h_next + col_sep_length;
+          h = h_next + col_sep_width;
           h_next = h + chars_per_column;
         }
     }
@@ -1750,9 +1874,9 @@
 align_column (COLUMN *p)
 {
   padding_not_printed = p->start_position;
-  if (col_sep_length < padding_not_printed)
+  if (col_sep_width < padding_not_printed)
     {
-      pad_across_to (padding_not_printed - col_sep_length);
+      pad_across_to (padding_not_printed - col_sep_width);
       padding_not_printed = ANYWHERE;
     }
 
@@ -2029,13 +2153,13 @@
       /* May be too generous. */
       buff = xpalloc (buff, &buff_allocated, 1, -1, sizeof *buff);
     }
-  buff[buff_current++] = c;
+  buff[buff_current++] = (unsigned char) c;
 }
 
 static void
 add_line_number (COLUMN *p)
 {
-  int i;
+  int i, j;
   char *s;
   int num_width;
 
@@ -2052,22 +2176,24 @@
       /* Tabification is assumed for multiple columns, also for n-separators,
          but 'default n-separator = TAB' hasn't been given priority over
          equal column_width also specified by POSIX. */
-      if (number_separator == '\t')
+      if (number_separator[0] == '\t')
         {
           i = number_width - chars_per_number;
           while (i-- > 0)
             (p->char_func) (' ');
         }
       else
-        (p->char_func) (number_separator);
+        for (j = 0; j < number_separator_length; j++)
+          (p->char_func) (number_separator[j]);
     }
   else
     /* To comply with POSIX, we avoid any expansion of default TAB
        separator with a single column output. No column_width requirement
        has to be considered. */
     {
-      (p->char_func) (number_separator);
-      if (number_separator == '\t')
+      for (j = 0; j < number_separator_length; j++)
+        (p->char_func) (number_separator[j]);
+      if (number_separator[0] == '\t')
         output_position = POS_AFTER_TAB (chars_per_output_tab,
                           output_position);
     }
@@ -2226,7 +2352,7 @@
   while (goal - h_old > 1
          && (h_new = POS_AFTER_TAB (chars_per_output_tab, h_old)) <= goal)
     {
-      putchar (output_tab_char);
+      fwrite (output_tab_char, sizeof(char), output_tab_char_length, stdout);
       h_old = h_new;
     }
   while (++h_old <= goal)
@@ -2246,6 +2372,7 @@
 {
   char const *s = col_sep_string;
   int l = col_sep_length;
+  int not_space_flag;
 
   if (separators_not_printed <= 0)
     {
@@ -2257,6 +2384,7 @@
     {
       for (; separators_not_printed > 0; --separators_not_printed)
         {
+          not_space_flag = 0;
           while (l-- > 0)
             {
               /* 3 types of sep_strings: spaces only, spaces and chars,
@@ -2270,12 +2398,15 @@
                 }
               else
                 {
+                  not_space_flag = 1;
                   if (spaces_not_printed > 0)
                     print_white_space ();
                   putchar (*s++);
-                  ++output_position;
                 }
             }
+          if (not_space_flag)
+            output_position += col_sep_width;
+
           /* sep_string ends with some spaces */
           if (spaces_not_printed > 0)
             print_white_space ();
@@ -2306,7 +2437,7 @@
    required number of tabs and spaces. */
 
 static void
-print_char (char c)
+print_char_single (char c)
 {
   if (tabify_output)
     {
@@ -2330,6 +2461,74 @@
   putchar (c);
 }
 
+#ifdef HAVE_MBRTOWC
+static void
+print_char_multi (char c)
+{
+  static size_t mbc_pos = 0;
+  static char mbc[MB_LEN_MAX] = {'\0'};
+  static mbstate_t state = {'\0'};
+  mbstate_t state_bak;
+  wchar_t wc;
+  size_t mblength;
+  int width;
+
+  if (tabify_output)
+    {
+      state_bak = state;
+      mbc[mbc_pos++] = c;
+      mblength = mbrtowc (&wc, mbc, mbc_pos, &state);
+
+      while (mbc_pos > 0)
+        {
+          switch (mblength)
+            {
+            case (size_t)-2:
+              state = state_bak;
+              return;
+
+            case (size_t)-1:
+              state = state_bak;
+              ++output_position;
+              putchar (mbc[0]);
+              memmove (mbc, mbc + 1, MB_CUR_MAX - 1);
+              --mbc_pos;
+              break;
+
+            case 0:
+              mblength = 1;
+
+            default:
+              if (wc == L' ')
+                {
+                  memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength);
+                  --mbc_pos;
+                  ++spaces_not_printed;
+                  return;
+                }
+              else if (spaces_not_printed > 0)
+                print_white_space ();
+
+              /* Nonprintables are assumed to have width 0, except L'\b'. */
+              if ((width = wcwidth (wc)) < 1)
+                {
+                  if (wc == L'\b')
+                    --output_position;
+                }
+              else
+                output_position += width;
+
+              fwrite (mbc, sizeof(char), mblength, stdout);
+              memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength);
+              mbc_pos -= mblength;
+            }
+        }
+      return;
+    }
+  putchar (c);
+}
+#endif
+
 /* Skip to page PAGE before printing.
    PAGE may be larger than total number of pages. */
 
@@ -2506,9 +2705,9 @@
           align_empty_cols = false;
         }
 
-      if (col_sep_length < padding_not_printed)
+      if (col_sep_width < padding_not_printed)
         {
-          pad_across_to (padding_not_printed - col_sep_length);
+          pad_across_to (padding_not_printed - col_sep_width);
           padding_not_printed = ANYWHERE;
         }
 
@@ -2577,7 +2776,7 @@
   COLUMN *q;
 
   int line = p->current_line++;
-  char *first = &buff[line_vector[line]];
+  unsigned char *first = &buff[line_vector[line]];
   /* FIXME
      UMR: Uninitialized memory read:
      * This is occurring while in:
@@ -2589,7 +2788,7 @@
      xmalloc        [xmalloc.c:94]
      init_store_cols [pr.c:1648]
      */
-  char *last = &buff[line_vector[line + 1]];
+  unsigned char *last = &buff[line_vector[line + 1]];
 
   pad_vertically = true;
 
@@ -2609,9 +2808,9 @@
         }
     }
 
-  if (col_sep_length < padding_not_printed)
+  if (col_sep_width < padding_not_printed)
     {
-      pad_across_to (padding_not_printed - col_sep_length);
+      pad_across_to (padding_not_printed - col_sep_width);
       padding_not_printed = ANYWHERE;
     }
 
@@ -2624,8 +2823,8 @@
   if (spaces_not_printed == 0)
     {
       output_position = p->start_position + end_vector[line];
-      if (p->start_position - col_sep_length == chars_per_margin)
-        output_position -= col_sep_length;
+      if (p->start_position - col_sep_width == chars_per_margin)
+        output_position -= col_sep_width;
     }
 
   return true;
@@ -2644,7 +2843,7 @@
    number of characters is 1.) */
 
 static int
-char_to_clump (char c)
+char_to_clump_single (char c)
 {
   unsigned char uc = c;
   char *s = clump_buff;
@@ -2653,10 +2852,10 @@
   int chars;
   int chars_per_c = 8;
 
-  if (c == input_tab_char)
+  if (c == input_tab_char[0])
     chars_per_c = chars_per_input_tab;
 
-  if (c == input_tab_char || c == '\t')
+  if (c == input_tab_char[0] || c == '\t')
     {
       width = TAB_WIDTH (chars_per_c, input_position);
 
@@ -2737,6 +2936,164 @@
   return chars;
 }
 
+#ifdef HAVE_MBRTOWC
+static int
+char_to_clump_multi (char c)
+{
+  static size_t mbc_pos = 0;
+  static char mbc[MB_LEN_MAX] = {'\0'};
+  static mbstate_t state = {'\0'};
+  mbstate_t state_bak;
+  wchar_t wc;
+  size_t mblength;
+  int wc_width;
+  register char *s = clump_buff;
+  register int i, j;
+  char esc_buff[4];
+  int width;
+  int chars;
+  int chars_per_c = 8;
+
+  state_bak = state;
+  mbc[mbc_pos++] = c;
+  mblength = mbrtowc (&wc, mbc, mbc_pos, &state);
+
+  width = 0;
+  chars = 0;
+  while (mbc_pos > 0)
+    {
+      switch (mblength)
+        {
+        case (size_t)-2:
+          state = state_bak;
+          return 0;
+
+        case (size_t)-1:
+          state = state_bak;
+          mblength = 1;
+
+          if (use_esc_sequence || use_cntrl_prefix)
+            {
+              width = +4;
+              chars = +4;
+              *s++ = '\\';
+              sprintf (esc_buff, "%03o", (unsigned char) mbc[0]);
+              for (i = 0; i <= 2; ++i)
+                *s++ = (int) esc_buff[i];
+            }
+          else
+            {
+              width += 1;
+              chars += 1;
+              *s++ = mbc[0];
+            }
+          break;
+
+        case 0:
+          mblength = 1;
+                /* Fall through */
+
+        default:
+          if (memcmp (mbc, input_tab_char, mblength) == 0)
+            chars_per_c = chars_per_input_tab;
+
+          if (memcmp (mbc, input_tab_char, mblength) == 0 || c == '\t')
+            {
+              int  width_inc;
+
+              width_inc = TAB_WIDTH (chars_per_c, input_position);
+              width += width_inc;
+
+              if (untabify_input)
+                {
+                  for (i = width_inc; i; --i)
+                    *s++ = ' ';
+                  chars += width_inc;
+                }
+              else
+                {
+                  for (i = 0; i <  mblength; i++)
+                    *s++ = mbc[i];
+                  chars += mblength;
+                }
+            }
+          else if ((wc_width = wcwidth (wc)) < 1)
+            {
+              if (use_esc_sequence)
+                {
+                  for (i = 0; i < mblength; i++)
+                    {
+                      width += 4;
+                      chars += 4;
+                      *s++ = '\\';
+                      sprintf (esc_buff, "%03o", (unsigned char) mbc[i]);
+                      for (j = 0; j <= 2; ++j)
+                        *s++ = (int) esc_buff[j];
+                    }
+                }
+              else if (use_cntrl_prefix)
+                {
+                  if (wc < 0200)
+                    {
+                      width += 2;
+                      chars += 2;
+                      *s++ = '^';
+                      *s++ = wc ^ 0100;
+                    }
+                  else
+                    {
+                      for (i = 0; i < mblength; i++)
+                        {
+                          width += 4;
+                          chars += 4;
+                          *s++ = '\\';
+                          sprintf (esc_buff, "%03o", (unsigned char) mbc[i]);
+                          for (j = 0; j <= 2; ++j)
+                            *s++ = (int) esc_buff[j];
+                        }
+                    }
+                }
+              else if (wc == L'\b')
+                {
+                  width += -1;
+                  chars += 1;
+                  *s++ = c;
+                }
+              else
+                {
+                  width += 0;
+                  chars += mblength;
+                  for (i = 0; i < mblength; i++)
+                    *s++ = mbc[i];
+                }
+            }
+          else
+            {
+              width += wc_width;
+              chars += mblength;
+              for (i = 0; i < mblength; i++)
+                *s++ = mbc[i];
+            }
+        }
+      memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength);
+      mbc_pos -= mblength;
+    }
+
+  /* Too many backspaces must put us in position 0 -- never negative. */
+  if (width < 0 && input_position == 0)
+    {
+      chars = 0;
+      input_position = 0;
+    }
+  else if (width < 0 && input_position <= -width)
+    input_position = 0;
+  else
+   input_position += width;
+
+  return chars;
+}
+#endif
+
 /* We've just printed some files and need to clean up things before
    looking for more options and printing the next batch of files.
 
diff -Naur -i coreutils-9.10/src/sort.c coreutils-9.10.patched/src/sort.c
--- coreutils-9.10/src/sort.c	2026-01-21 07:51:40.000000000 -0600
+++ coreutils-9.10.patched/src/sort.c	2026-02-07 18:28:47.472920908 -0600
@@ -29,6 +29,14 @@
 #include <sys/wait.h>
 #include <signal.h>
 #include <spawn.h>
+#if HAVE_WCHAR_H
+# include <wchar.h>
+#endif
+/* Get isw* functions. */
+#if HAVE_WCTYPE_H
+# include <wctype.h>
+#endif
+
 #include "system.h"
 #include "argmatch.h"
 #include "assure.h"
@@ -148,14 +156,39 @@
 /* We currently ignore multi-byte grouping chars.  */
 static bool thousands_sep_ignored;
 
+/* True if -f is specified.  */
+static bool folding;
+
 /* Nonzero if the corresponding locales are hard.  */
 static bool hard_LC_COLLATE;
-#if HAVE_NL_LANGINFO
+#if HAVE_LANGINFO_CODESET
 static bool hard_LC_TIME;
 #endif
 
 #define NONZERO(x) ((x) != 0)
 
+/* get a multibyte character's byte length. */
+#define GET_BYTELEN_OF_CHAR(LIM, PTR, MBLENGTH, STATE)                        \
+  do                                                                        \
+    {                                                                        \
+      wchar_t wc;                                                        \
+      mbstate_t state_bak;                                                \
+                                                                        \
+      state_bak = STATE;                                                \
+      mblength = mbrtowc (&wc, PTR, LIM - PTR, &STATE);                        \
+                                                                        \
+      switch (MBLENGTH)                                                        \
+        {                                                                \
+        case (size_t)-1:                                                \
+        case (size_t)-2:                                                \
+          STATE = state_bak;                                                \
+                /* Fall through. */                                        \
+        case 0:                                                                \
+          MBLENGTH = 1;                                                        \
+      }                                                                        \
+    }                                                                        \
+  while (0)
+
 /* The kind of blanks for '-b' to skip in various options. */
 enum blanktype { bl_start, bl_end, bl_both };
 
@@ -332,13 +365,11 @@
 /* An int value outside char range.  */
 enum { NON_CHAR = CHAR_MAX + 1 };
 
-/* If TAB has this value, blanks separate fields.  */
-enum { TAB_DEFAULT = CHAR_MAX + 1 };
-
-/* Tab character separating fields.  If TAB_DEFAULT, then fields are
+/* Tab character separating fields.  If tab_length is 0, then fields are
    separated by the empty string between a non-blank character and a blank
    character. */
-static int tab = TAB_DEFAULT;
+static char tab[MB_LEN_MAX + 1];
+static size_t tab_length = 0;
 
 /* Flag to remove consecutive duplicate lines from the output.
    Only the last of a sequence of equal lines will be output. */
@@ -374,6 +405,46 @@
 static struct tempnode *volatile temphead;
 static struct tempnode *volatile *temptail = &temphead;
 
+/* Function pointers. */
+static void
+(*inittables) (void);
+static char *
+(*begfield) (const struct line*, const struct keyfield *);
+static char *
+(*limfield) (const struct line*, const struct keyfield *);
+static void
+(*skipblanks) (char **ptr, char *lim);
+static int
+(*getmonth) (char const *, size_t, char **);
+static int
+(*keycompare) (const struct line *, const struct line *);
+static int
+(*numcompare) (const char *, const char *);
+
+/* Test for white space multibyte character.
+   Set LENGTH the byte length of investigated multibyte character. */
+#if HAVE_MBRTOWC
+static int
+ismbblank (const char *str, size_t len, size_t *length)
+{
+  size_t mblength;
+  wchar_t wc;
+  mbstate_t state;
+
+  memset (&state, '\0', sizeof(mbstate_t));
+  mblength = mbrtowc (&wc, str, len, &state);
+
+  if (mblength == (size_t)-1 || mblength == (size_t)-2)
+    {
+      *length = 1;
+      return 0;
+    }
+
+  *length = (mblength < 1) ? 1 : mblength;
+  return iswblank (wc) || wc == '\n';
+}
+#endif
+
 /* Clean up any remaining temporary files.  */
 
 static void
@@ -1381,7 +1452,7 @@
   free (node);
 }
 
-#if HAVE_NL_LANGINFO
+#if HAVE_LANGINFO_CODESET
 
 static int
 struct_month_cmp (void const *m1, void const *m2)
@@ -1396,7 +1467,7 @@
 /* Initialize the character class tables. */
 
 static void
-inittables (void)
+inittables_uni (void)
 {
   for (size_t i = 0; i < UCHAR_LIM; ++i)
     {
@@ -1406,7 +1477,7 @@
       fold_toupper[i] = toupper (i);
     }
 
-#if HAVE_NL_LANGINFO
+#if HAVE_LANGINFO_CODESET
   /* If we're not in the "C" locale, read different names for months.  */
   if (hard_LC_TIME)
     {
@@ -1486,6 +1557,84 @@
     xstrtol_fatal (e, oi, c, long_options, s);
 }
 
+#if HAVE_MBRTOWC
+static void
+inittables_mb (void)
+{
+  int i, j, k, l;
+  char *name, *s, *lc_time, *lc_ctype;
+  size_t s_len, mblength;
+  char mbc[MB_LEN_MAX];
+  wchar_t wc, pwc;
+  mbstate_t state_mb, state_wc;
+
+  lc_time = setlocale (LC_TIME, "");
+  if (lc_time)
+    lc_time = xstrdup (lc_time);
+
+  lc_ctype = setlocale (LC_CTYPE, "");
+  if (lc_ctype)
+    lc_ctype = xstrdup (lc_ctype);
+
+  if (lc_time && lc_ctype)
+    /* temporarily set LC_CTYPE to match LC_TIME, so that we can convert
+     * the names of months to upper case */
+    setlocale (LC_CTYPE, lc_time);
+
+  for (i = 0; i < MONTHS_PER_YEAR; i++)
+    {
+      s = (char *) nl_langinfo (ABMON_1 + i);
+      s_len = strlen (s);
+      monthtab[i].name = name = (char *) xmalloc (s_len + 1);
+      monthtab[i].val = i + 1;
+
+      memset (&state_mb, '\0', sizeof (mbstate_t));
+      memset (&state_wc, '\0', sizeof (mbstate_t));
+
+      for (j = 0; j < s_len;)
+        {
+          if (!ismbblank (s + j, s_len - j, &mblength))
+            break;
+          j += mblength;
+        }
+
+      for (k = 0; j < s_len;)
+        {
+          mblength = mbrtowc (&wc, (s + j), (s_len - j), &state_mb);
+          assert (mblength != (size_t)-1 && mblength != (size_t)-2);
+          if (mblength == 0)
+            break;
+
+          pwc = towupper (wc);
+          if (pwc == wc)
+            {
+              memcpy (mbc, s + j, mblength);
+              j += mblength;
+            }
+          else
+            {
+              j += mblength;
+              mblength = wcrtomb (mbc, pwc, &state_wc);
+              assert (mblength != (size_t)0 && mblength != (size_t)-1);
+            }
+
+          for (l = 0; l < mblength; l++)
+            name[k++] = mbc[l];
+        }
+      name[k] = '\0';
+    }
+  qsort ((void *) monthtab, MONTHS_PER_YEAR,
+      sizeof (struct month), struct_month_cmp);
+
+  if (lc_time && lc_ctype)
+    /* restore the original locales */
+    setlocale (LC_CTYPE, lc_ctype);
+
+  free (lc_ctype);
+  free (lc_time);
+}
+#endif
+
 /* Specify the amount of main memory to use when sorting.  */
 static void
 specify_sort_size (int oi, char c, char const *s)
@@ -1712,7 +1861,7 @@
    by KEY in LINE. */
 
 static char *
-begfield (struct line const *line, struct keyfield const *key)
+begfield_uni (const struct line *line, const struct keyfield *key)
 {
   char *ptr = line->text, *lim = ptr + line->length - 1;
   size_t sword = key->sword;
@@ -1721,10 +1870,10 @@
   /* The leading field separator itself is included in a field when -t
      is absent.  */
 
-  if (tab != TAB_DEFAULT)
+  if (tab_length)
     while (ptr < lim && sword--)
       {
-        char *sep = memchr (ptr, tab, lim - ptr);
+        char *sep = memchr (ptr, tab[0], lim - ptr);
         ptr = sep ? sep : lim;
         if (ptr < lim)
           ++ptr;
@@ -1754,12 +1903,71 @@
   return ptr;
 }
 
+#if HAVE_MBRTOWC
+static char *
+begfield_mb (const struct line *line, const struct keyfield *key)
+{
+  int i;
+  char *ptr = line->text, *lim = ptr + line->length - 1;
+  size_t sword = key->sword;
+  size_t schar = key->schar;
+  size_t mblength;
+  mbstate_t state;
+
+  memset (&state, '\0', sizeof(mbstate_t));
+
+  if (tab_length)
+    while (ptr < lim && sword--)
+      {
+        while (ptr < lim && memcmp (ptr, tab, tab_length) != 0)
+          {
+            GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state);
+            ptr += mblength;
+          }
+        if (ptr < lim)
+          {
+            GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state);
+            ptr += mblength;
+          }
+      }
+  else
+    while (ptr < lim && sword--)
+      {
+        while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength))
+          ptr += mblength;
+        if (ptr < lim)
+          {
+            GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state);
+            ptr += mblength;
+          }
+        while (ptr < lim && !ismbblank (ptr, lim - ptr, &mblength))
+          ptr += mblength;
+      }
+
+  if (key->skipsblanks)
+    while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength))
+      ptr += mblength;
+
+  for (i = 0; i < schar; i++)
+    {
+      GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state);
+
+      if (ptr + mblength > lim)
+        break;
+      else
+        ptr += mblength;
+    }
+
+  return ptr;
+}
+#endif
+
 /* Return the limit of (a pointer to the first character after) the field
    in LINE specified by KEY. */
 
 ATTRIBUTE_PURE
 static char *
-limfield (struct line const *line, struct keyfield const *key)
+limfield_uni (struct line const *line, struct keyfield const *key)
 {
   char *ptr = line->text, *lim = ptr + line->length - 1;
   size_t eword = key->eword, echar = key->echar;
@@ -1774,10 +1982,10 @@
      'beginning' is the first character following the delimiting TAB.
      Otherwise, leave PTR pointing at the first 'blank' character after
      the preceding field.  */
-  if (tab != TAB_DEFAULT)
+  if (tab_length)
     while (ptr < lim && eword--)
       {
-        char *sep = memchr (ptr, tab, lim - ptr);
+        char *sep = memchr (ptr, tab[0], lim - ptr);
         ptr = sep ? sep : lim;
         if (ptr < lim && (eword || echar))
           ++ptr;
@@ -1823,10 +2031,10 @@
      */
 
   /* Make LIM point to the end of (one byte past) the current field.  */
-  if (tab != TAB_DEFAULT)
+  if (tab_length)
     {
       char *newlim;
-      newlim = memchr (ptr, tab, lim - ptr);
+      newlim = memchr (ptr, tab[0], lim - ptr);
       if (newlim)
         lim = newlim;
     }
@@ -1861,6 +2069,130 @@
   return ptr;
 }
 
+#if HAVE_MBRTOWC
+static char * _GL_ATTRIBUTE_PURE
+limfield_mb (const struct line *line, const struct keyfield *key)
+{
+  char *ptr = line->text, *lim = ptr + line->length - 1;
+  size_t eword = key->eword, echar = key->echar;
+  int i;
+  size_t mblength;
+  mbstate_t state;
+
+  if (echar == 0)
+    eword++; /* skip all of end field. */
+
+  memset (&state, '\0', sizeof(mbstate_t));
+
+  if (tab_length)
+    while (ptr < lim && eword--)
+      {
+        while (ptr < lim && memcmp (ptr, tab, tab_length) != 0)
+          {
+            GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state);
+            ptr += mblength;
+          }
+        if (ptr < lim && (eword | echar))
+          {
+            GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state);
+            ptr += mblength;
+          }
+      }
+  else
+    while (ptr < lim && eword--)
+      {
+        while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength))
+          ptr += mblength;
+        if (ptr < lim)
+          {
+            GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state);
+            ptr += mblength;
+          }
+        while (ptr < lim && !ismbblank (ptr, lim - ptr, &mblength))
+          ptr += mblength;
+      }
+
+
+# ifdef POSIX_UNSPECIFIED
+  /* Make LIM point to the end of (one byte past) the current field.  */
+  if (tab_length)
+    {
+      char *newlim, *p;
+
+      newlim = NULL;
+      for (p = ptr; p < lim;)
+         {
+          if (memcmp (p, tab, tab_length) == 0)
+            {
+              newlim = p;
+              break;
+            }
+
+          GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state);
+          p += mblength;
+        }
+    }
+  else
+    {
+      char *newlim;
+      newlim = ptr;
+
+      while (newlim < lim && ismbblank (newlim, lim - newlim, &mblength))
+        newlim += mblength;
+      if (ptr < lim)
+        {
+          GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state);
+          ptr += mblength;
+        }
+      while (newlim < lim && !ismbblank (newlim, lim - newlim, &mblength))
+        newlim += mblength;
+      lim = newlim;
+    }
+# endif
+
+  if (echar != 0)
+  {
+    /* If we're skipping leading blanks, don't start counting characters
+     *      until after skipping past any leading blanks.  */
+    if (key->skipeblanks)
+      while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength))
+        ptr += mblength;
+
+    memset (&state, '\0', sizeof(mbstate_t));
+
+    /* Advance PTR by ECHAR (if possible), but no further than LIM.  */
+    for (i = 0; i < echar; i++)
+     {
+        GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state);
+
+        if (ptr + mblength > lim)
+          break;
+        else
+          ptr += mblength;
+      }
+  }
+
+  return ptr;
+}
+#endif
+
+static void
+skipblanks_uni (char **ptr, char *lim)
+{
+  while (*ptr < lim && blanks[to_uchar (**ptr)])
+    ++(*ptr);
+}
+
+#if HAVE_MBRTOWC
+static void
+skipblanks_mb (char **ptr, char *lim)
+{
+  size_t mblength;
+  while (*ptr < lim && ismbblank (*ptr, lim - *ptr, &mblength))
+    (*ptr) += mblength;
+}
+#endif
+
 /* Fill BUF reading from FP, moving buf->left bytes from the end
    of buf->buf to the beginning first.  If EOF is reached and the
    file wasn't terminated by a newline, supply one.  Set up BUF's line
@@ -1947,8 +2279,22 @@
                   else
                     {
                       if (key->skipsblanks)
-                        while (blanks[to_uchar (*line_start)])
-                          line_start++;
+                        {
+#if HAVE_MBRTOWC
+                          if (MB_CUR_MAX > 1)
+                            {
+                              size_t mblength;
+                              while (line_start < line->keylim &&
+                                     ismbblank (line_start,
+                                                line->keylim - line_start,
+                                                &mblength))
+                                line_start += mblength;
+                            }
+                          else
+#endif
+                          while (blanks[to_uchar (*line_start)])
+                            line_start++;
+                        }
                       line->keybeg = line_start;
                     }
                 }
@@ -2086,12 +2432,10 @@
 
 ATTRIBUTE_PURE
 static int
-human_numcompare (char const *a, char const *b)
+human_numcompare (char *a, char *b)
 {
-  while (blanks[to_uchar (*a)])
-    a++;
-  while (blanks[to_uchar (*b)])
-    b++;
+  skipblanks(&a, a + strlen(a));
+  skipblanks(&b, b + strlen(b));
 
   int diff = find_unit_order (a) - find_unit_order (b);
   return (diff ? diff : strnumcmp (a, b, decimal_point, thousands_sep));
@@ -2103,7 +2447,7 @@
 
 ATTRIBUTE_PURE
 static int
-numcompare (char const *a, char const *b)
+numcompare_uni (const char *a, const char *b)
 {
   while (blanks[to_uchar (*a)])
     a++;
@@ -2113,6 +2457,25 @@
   return strnumcmp (a, b, decimal_point, thousands_sep);
 }
 
+#if HAVE_MBRTOWC
+static int
+numcompare_mb (const char *a, const char *b)
+{
+  size_t mblength, len;
+  len = strlen (a); /* okay for UTF-8 */
+  while (*a && ismbblank (a, len > MB_CUR_MAX ? MB_CUR_MAX : len, &mblength))
+    {
+      a += mblength;
+      len -= mblength;
+    }
+  len = strlen (b); /* okay for UTF-8 */
+  while (*b && ismbblank (b, len > MB_CUR_MAX ? MB_CUR_MAX : len, &mblength))
+    b += mblength;
+
+  return strnumcmp (a, b, decimal_point, thousands_sep);
+}
+#endif /* HAV_EMBRTOWC */
+
 static int
 nan_compare (long double a, long double b)
 {
@@ -2154,7 +2517,7 @@
    Return 0 if the name in S is not recognized.  */
 
 static int
-getmonth (char const *month, char **ea)
+getmonth_uni (char const *month, size_t len, char **ea)
 {
   size_t lo = 0;
   size_t hi = MONTHS_PER_YEAR;
@@ -2493,15 +2856,14 @@
           char saved = *lim;
           *lim = '\0';
 
-          while (blanks[to_uchar (*beg)])
-            beg++;
+          skipblanks (&beg, lim);
 
           char *tighter_lim = beg;
 
           if (lim < beg)
             tighter_lim = lim;
           else if (key->month)
-            getmonth (beg, &tighter_lim);
+            getmonth (beg, lim-beg, &tighter_lim);
           else if (key->general_numeric)
             ignore_value (strtold (beg, &tighter_lim));
           else if (key->numeric || key->human_numeric)
@@ -2646,7 +3008,7 @@
       /* Warn about significant leading blanks.  */
       bool implicit_skip = key_numeric (key) || key->month;
       bool line_offset = key->eword == 0 && key->echar != 0; /* -k1.x,1.y  */
-      if (!zero_width && !gkey_only && tab == TAB_DEFAULT && !line_offset
+      if (!zero_width && !gkey_only && !tab_length && !line_offset
           && ((!key->skipsblanks && !implicit_skip)
               || (!key->skipsblanks && key->schar)
               || (!key->skipeblanks && key->echar)))
@@ -2694,9 +3056,9 @@
   bool number_locale_warned = false;
   if (basic_numeric_field_span)
     {
-      if (tab == TAB_DEFAULT
-          ? thousands_sep != NON_CHAR && (isblank (to_uchar (thousands_sep)))
-          : tab == thousands_sep)
+      if (tab_length
+          ? tab[0] == thousands_sep
+          : thousands_sep != NON_CHAR && (isblank (to_uchar (thousands_sep))))
         {
           error (0, 0,
                  _("field separator %s is treated as a "
@@ -2707,9 +3069,9 @@
     }
   if (basic_numeric_field_span || general_numeric_field_span)
     {
-      if (tab == TAB_DEFAULT
-          ? thousands_sep != NON_CHAR && (isblank (to_uchar (decimal_point)))
-          : tab == decimal_point)
+      if (tab_length
+          ? tab[0] == decimal_point
+          : thousands_sep != NON_CHAR && (isblank (to_uchar (decimal_point))))
         {
           error (0, 0,
                  _("field separator %s is treated as a "
@@ -2717,19 +3079,19 @@
                  quote (((char []) {decimal_point, 0})));
           number_locale_warned = true;
         }
-      else if (tab == '-')
+      else if (tab_length && tab[0] == '-')
         {
           error (0, 0,
                  _("field separator %s is treated as a "
                    "minus sign in numbers"),
-                 quote (((char []) {tab, 0})));
+                 quote (((char []) {tab[0], 0})));
         }
-      else if (general_numeric_field_span && tab == '+')
+      else if (general_numeric_field_span && tab_length && tab[0] == '+')
         {
           error (0, 0,
                  _("field separator %s is treated as a "
                    "plus sign in numbers"),
-                 quote (((char []) {tab, 0})));
+                 quote (((char []) {tab[0], 0})));
         }
     }
 
@@ -2781,11 +3143,87 @@
   return reversed ? _GL_CMP (0, diff) : diff;
 }
 
+#if HAVE_MBRTOWC
+static int
+getmonth_mb (const char *s, size_t len, char **ea)
+{
+  char *month;
+  register size_t i;
+  register int lo = 0, hi = MONTHS_PER_YEAR, result;
+  char *tmp;
+  size_t wclength, mblength;
+  const char *pp;
+  const wchar_t *wpp;
+  wchar_t *month_wcs;
+  mbstate_t state;
+
+  while (len > 0 && ismbblank (s, len, &mblength))
+    {
+      s += mblength;
+      len -= mblength;
+    }
+
+  if (len == 0)
+    return 0;
+
+  if (SIZE_MAX - len < 1)
+    xalloc_die ();
+
+  month = (char *) xnmalloc (len + 1, MB_CUR_MAX);
+
+  pp = tmp = (char *) xnmalloc (len + 1, MB_CUR_MAX);
+  memcpy (tmp, s, len);
+  tmp[len] = '\0';
+  wpp = month_wcs = (wchar_t *) xnmalloc (len + 1, sizeof (wchar_t));
+  memset (&state, '\0', sizeof (mbstate_t));
+
+  wclength = mbsrtowcs (month_wcs, &pp, len + 1, &state);
+  if (wclength == (size_t)-1 || pp != NULL)
+    error (SORT_FAILURE, 0, _("Invalid multibyte input %s."), quote(s));
+
+  for (i = 0; i < wclength; i++)
+    {
+      month_wcs[i] = towupper(month_wcs[i]);
+      if (iswblank (month_wcs[i]))
+        {
+          month_wcs[i] = L'\0';
+          break;
+        }
+    }
+
+  mblength = wcsrtombs (month, &wpp, (len + 1) * MB_CUR_MAX, &state);
+  assert (mblength != (-1) && wpp == NULL);
+
+  do
+    {
+      int ix = (lo + hi) / 2;
+
+      if (strncmp (month, monthtab[ix].name, strlen (monthtab[ix].name)) < 0)
+        hi = ix;
+      else
+        lo = ix;
+    }
+  while (hi - lo > 1);
+
+  result = (!strncmp (month, monthtab[lo].name, strlen (monthtab[lo].name))
+      ? monthtab[lo].val : 0);
+
+  if (ea && result)
+     *ea = (char*) s + strlen (monthtab[lo].name);
+
+  free (month);
+  free (tmp);
+  free (month_wcs);
+
+  return result;
+}
+#endif
+
 /* Compare two lines A and B trying every key in sequence until there
    are no more keys or a difference is found. */
 
 static int
-keycompare (struct line const *a, struct line const *b)
+keycompare_uni (const struct line *a, const struct line *b)
 {
   struct keyfield *key = keylist;
 
@@ -2866,7 +3304,7 @@
           else if (key->human_numeric)
             diff = human_numcompare (ta, tb);
           else if (key->month)
-            diff = getmonth (ta, NULL) - getmonth (tb, NULL);
+            diff = getmonth (ta, tlena, NULL) - getmonth (tb, tlenb, NULL);
           else if (key->random)
             diff = compare_random (ta, tlena, tb, tlenb);
           else if (key->version)
@@ -2976,6 +3414,211 @@
   return diff_reversed (diff, key->reverse);
 }
 
+#if HAVE_MBRTOWC
+static int
+keycompare_mb (const struct line *a, const struct line *b)
+{
+  struct keyfield *key = keylist;
+
+  /* For the first iteration only, the key positions have been
+     precomputed for us. */
+  char *texta = a->keybeg;
+  char *textb = b->keybeg;
+  char *lima = a->keylim;
+  char *limb = b->keylim;
+
+  size_t mblength_a, mblength_b;
+  wchar_t wc_a, wc_b;
+  mbstate_t state_a, state_b;
+
+  int diff = 0;
+
+  memset (&state_a, '\0', sizeof(mbstate_t));
+  memset (&state_b, '\0', sizeof(mbstate_t));
+  /* Ignore keys with start after end.  */
+  if (a->keybeg - a->keylim > 0)
+    return 0;
+
+
+              /* Ignore and/or translate chars before comparing.  */
+# define IGNORE_CHARS(NEW_LEN, LEN, TEXT, COPY, WC, MBLENGTH, STATE)        \
+  do                                                                        \
+    {                                                                        \
+      wchar_t uwc;                                                        \
+      char mbc[MB_LEN_MAX];                                                \
+      mbstate_t state_wc;                                                \
+                                                                        \
+      for (NEW_LEN = i = 0; i < LEN;)                                        \
+        {                                                                \
+          mbstate_t state_bak;                                                \
+                                                                        \
+          state_bak = STATE;                                                \
+          MBLENGTH = mbrtowc (&WC, TEXT + i, LEN - i, &STATE);                \
+                                                                        \
+          if (MBLENGTH == (size_t)-2 || MBLENGTH == (size_t)-1                \
+              || MBLENGTH == 0)                                                \
+            {                                                                \
+              if (MBLENGTH == (size_t)-2 || MBLENGTH == (size_t)-1)        \
+                STATE = state_bak;                                        \
+              if (!ignore)                                                \
+                COPY[NEW_LEN++] = TEXT[i];                                \
+              i++;                                                         \
+              continue;                                                        \
+            }                                                                \
+                                                                        \
+          if (ignore)                                                        \
+            {                                                                \
+              if ((ignore == nonprinting && !iswprint (WC))                \
+                   || (ignore == nondictionary                                \
+                       && !iswalnum (WC) && !iswblank (WC)))                \
+                {                                                        \
+                  i += MBLENGTH;                                        \
+                  continue;                                                \
+                }                                                        \
+            }                                                                \
+                                                                        \
+          if (translate)                                                \
+            {                                                                \
+                                                                        \
+              uwc = towupper(WC);                                        \
+              if (WC == uwc)                                                \
+                {                                                        \
+                  memcpy (mbc, TEXT + i, MBLENGTH);                        \
+                  i += MBLENGTH;                                        \
+                }                                                        \
+              else                                                        \
+                {                                                        \
+                  i += MBLENGTH;                                        \
+                  WC = uwc;                                                \
+                  memset (&state_wc, '\0', sizeof (mbstate_t));                \
+                                                                        \
+                  MBLENGTH = wcrtomb (mbc, WC, &state_wc);                \
+                  assert (MBLENGTH != (size_t)-1 && MBLENGTH != 0);        \
+                }                                                        \
+                                                                        \
+              for (j = 0; j < MBLENGTH; j++)                                \
+                COPY[NEW_LEN++] = mbc[j];                                \
+            }                                                                \
+          else                                                                \
+            for (j = 0; j < MBLENGTH; j++)                                \
+              COPY[NEW_LEN++] = TEXT[i++];                                \
+        }                                                                \
+      COPY[NEW_LEN] = '\0';                                                \
+    }                                                                        \
+  while (0)
+
+      /* Actually compare the fields. */
+
+  for (;;)
+    {
+      /* Find the lengths. */
+      size_t lena = lima <= texta ? 0 : lima - texta;
+      size_t lenb = limb <= textb ? 0 : limb - textb;
+
+      char enda IF_LINT (= 0);
+      char endb IF_LINT (= 0);
+
+      char const *translate = key->translate;
+      bool const *ignore = key->ignore;
+
+      if (ignore || translate)
+        {
+          if (SIZE_MAX - lenb - 2 < lena)
+            xalloc_die ();
+          char *copy_a = (char *) xnmalloc (lena + lenb + 2, MB_CUR_MAX);
+          char *copy_b = copy_a + lena * MB_CUR_MAX + 1;
+          size_t new_len_a, new_len_b;
+          size_t i, j;
+
+          IGNORE_CHARS (new_len_a, lena, texta, copy_a,
+                        wc_a, mblength_a, state_a);
+          IGNORE_CHARS (new_len_b, lenb, textb, copy_b,
+                        wc_b, mblength_b, state_b);
+          texta = copy_a; textb = copy_b;
+          lena = new_len_a; lenb = new_len_b;
+        }
+      else
+        {
+          /* Use the keys in-place, temporarily null-terminated.  */
+          enda = texta[lena]; texta[lena] = '\0';
+          endb = textb[lenb]; textb[lenb] = '\0';
+        }
+
+      if (key->random)
+        diff = compare_random (texta, lena, textb, lenb);
+      else if (key->numeric | key->general_numeric | key->human_numeric)
+        {
+          char savea = *lima, saveb = *limb;
+
+          *lima = *limb = '\0';
+          diff = (key->numeric ? numcompare (texta, textb)
+                  : key->general_numeric ? general_numcompare (texta, textb)
+                  : human_numcompare (texta, textb));
+          *lima = savea, *limb = saveb;
+        }
+      else if (key->version)
+        diff = filevercmp (texta, textb);
+      else if (key->month)
+        diff = getmonth (texta, lena, NULL) - getmonth (textb, lenb, NULL);
+      else if (lena == 0)
+        diff = - NONZERO (lenb);
+      else if (lenb == 0)
+        diff = 1;
+      else if (hard_LC_COLLATE && !folding)
+        {
+          diff = xmemcoll0 (texta, lena + 1, textb, lenb + 1);
+        }
+      else
+        {
+          diff = memcmp (texta, textb, MIN (lena, lenb));
+          if (diff == 0)
+            diff = lena < lenb ? -1 : lena != lenb;
+        }
+
+      if (ignore || translate)
+        free (texta);
+      else
+        {
+          texta[lena] = enda;
+          textb[lenb] = endb;
+        }
+
+      if (diff)
+        goto not_equal;
+
+      key = key->next;
+      if (! key)
+        break;
+
+      /* Find the beginning and limit of the next field.  */
+      if (key->eword != -1)
+        lima = limfield (a, key), limb = limfield (b, key);
+      else
+        lima = a->text + a->length - 1, limb = b->text + b->length - 1;
+
+      if (key->sword != -1)
+        texta = begfield (a, key), textb = begfield (b, key);
+      else
+        {
+          texta = a->text, textb = b->text;
+          if (key->skipsblanks)
+            {
+              while (texta < lima && ismbblank (texta, lima - texta, &mblength_a))
+                texta += mblength_a;
+              while (textb < limb && ismbblank (textb, limb - textb, &mblength_b))
+                textb += mblength_b;
+            }
+        }
+    }
+
+not_equal:
+  if (key && key->reverse)
+    return -diff;
+  else
+    return diff;
+}
+#endif
+
 /* Compare two lines A and B, returning negative, zero, or positive
    depending on whether A compares less than, equal to, or greater than B. */
 
@@ -3003,7 +3646,7 @@
     diff = - NONZERO (blen);
   else if (blen == 0)
     diff = 1;
-  else if (hard_LC_COLLATE)
+  else if (hard_LC_COLLATE && !folding)
     {
       /* xmemcoll0 is a performance enhancement as
          it will not unconditionally write '\0' after the
@@ -4381,6 +5024,7 @@
           break;
         case 'f':
           key->translate = fold_toupper;
+          folding = true;
           break;
         case 'g':
           key->general_numeric = true;
@@ -4460,7 +5104,7 @@
   initialize_exit_failure (SORT_FAILURE);
 
   hard_LC_COLLATE = hard_locale (LC_COLLATE);
-#if HAVE_NL_LANGINFO
+#if HAVE_LANGINFO_CODESET
   hard_LC_TIME = hard_locale (LC_TIME);
 #endif
 
@@ -4483,5 +5127,28 @@
       thousands_sep = NON_CHAR;
   }
 
+#if HAVE_MBRTOWC
+  if (MB_CUR_MAX > 1)
+    {
+      inittables = inittables_mb;
+      begfield = begfield_mb;
+      limfield = limfield_mb;
+      skipblanks = skipblanks_mb;
+      getmonth = getmonth_mb;
+      keycompare = keycompare_mb;
+      numcompare = numcompare_mb;
+    }
+  else
+#endif
+    {
+      inittables = inittables_uni;
+      begfield = begfield_uni;
+      limfield = limfield_uni;
+      skipblanks = skipblanks_uni;
+      getmonth = getmonth_uni;
+      keycompare = keycompare_uni;
+      numcompare = numcompare_uni;
+    }
+
   inittables ();
 
@@ -4729,13 +5396,34 @@
 
         case 't':
           {
-            char newtab = optarg[0];
-            if (! newtab)
+            char newtab[MB_LEN_MAX + 1];
+            size_t newtab_length = 1;
+            strncpy (newtab, optarg, MB_LEN_MAX);
+            if (! newtab[0])
               error (SORT_FAILURE, 0, _("empty tab"));
-            if (optarg[1])
+#if HAVE_MBRTOWC
+            if (MB_CUR_MAX > 1)
+              {
+                wchar_t wc;
+                mbstate_t state;
+
+                memset (&state, '\0', sizeof (mbstate_t));
+                newtab_length = mbrtowc (&wc, newtab, strnlen (newtab,
+                                                               MB_LEN_MAX),
+                                         &state);
+                switch (newtab_length)
+                  {
+                  case (size_t) -1:
+                  case (size_t) -2:
+                  case 0:
+                    newtab_length = 1;
+                  }
+              }
+#endif
+            if (newtab_length == 1 && optarg[1])
               {
                 if (streq (optarg, "\\0"))
-                  newtab = '\0';
+                  newtab[0] = '\0';
                 else
                   {
                     /* Provoke with 'sort -txx'.  Complain about
@@ -4746,9 +5434,11 @@
                            quote (optarg));
                   }
               }
-            if (tab != TAB_DEFAULT && tab != newtab)
+            if (tab_length && (tab_length != newtab_length
+                        || memcmp (tab, newtab, tab_length) != 0))
               error (SORT_FAILURE, 0, _("incompatible tabs"));
-            tab = newtab;
+            memcpy (tab, newtab, newtab_length);
+            tab_length = newtab_length;
           }
           break;
 
 
diff -Naur -i coreutils-9.10/tests/Coreutils.pm coreutils-9.10.patched/tests/Coreutils.pm
--- coreutils-9.10/tests/Coreutils.pm	2026-01-19 06:16:08.000000000 -0600
+++ coreutils-9.10.patched/tests/Coreutils.pm	2026-02-07 18:28:47.473865675 -0600
@@ -269,6 +269,9 @@
       # Yes, this is an arbitrary limit.  If it causes trouble,
       # consider removing it.
       my $max = 30;
+      # The downstream i18n multi-byte tests have a "-mb" suffix.
+      # Therefore add 3 to the maximum test name length.
+      $max += 3;
       if ($max < length $test_name)
         {
           warn "$program_name: $test_name: test name is too long (> $max)\n";
diff -Naur -i coreutils-9.10/tests/i18n/sort.sh coreutils-9.10.patched/tests/i18n/sort.sh
--- coreutils-9.10/tests/i18n/sort.sh	1969-12-31 18:00:00.000000000 -0600
+++ coreutils-9.10.patched/tests/i18n/sort.sh	2026-02-07 18:28:47.473995440 -0600
@@ -0,0 +1,29 @@
+#!/bin/sh
+# Verify sort's multi-byte support.
+
+. "${srcdir=.}/tests/init.sh"; path_prepend_ ./src
+print_ver_ sort
+
+export LC_ALL=en_US.UTF-8
+locale -k LC_CTYPE | grep -q "charmap.*UTF-8" \
+  || skip_ "No UTF-8 locale available"
+
+# Enable heap consistency checkng on older systems
+export MALLOC_CHECK_=2
+
+
+# check buffer overflow issue due to
+# expanding multi-byte representation due to case conversion
+# https://bugzilla.suse.com/show_bug.cgi?id=928749
+cat <<EOF > exp
+.
+ɑ
+EOF
+cat <<EOF | sort -f > out || fail=1
+.
+ɑ
+EOF
+compare exp out || { fail=1; cat out; }
+
+
+Exit $fail
diff -Naur -i coreutils-9.10/tests/local.mk coreutils-9.10.patched/tests/local.mk
--- coreutils-9.10/tests/local.mk	2026-01-30 07:30:40.000000000 -0600
+++ coreutils-9.10.patched/tests/local.mk	2026-02-07 18:28:47.474030251 -0600
@@ -423,6 +423,8 @@
   tests/sort/sort-field-limit.sh		\
   tests/sort/sort-files0-from.pl		\
   tests/sort/sort-float.sh			\
+  tests/misc/sort-mb-tests.sh			\
+  tests/i18n/sort.sh				\
   tests/sort/sort-h-thousands-sep.sh		\
   tests/sort/sort-locale.sh			\
   tests/sort/sort-merge.pl			\
diff -Naur -i coreutils-9.10/tests/misc/sort-mb-tests.sh coreutils-9.10.patched/tests/misc/sort-mb-tests.sh
--- coreutils-9.10/tests/misc/sort-mb-tests.sh	1969-12-31 18:00:00.000000000 -0600
+++ coreutils-9.10.patched/tests/misc/sort-mb-tests.sh	2026-02-07 18:28:47.474193452 -0600
@@ -0,0 +1,45 @@
+#!/bin/sh
+# Verify sort's multi-byte support.
+
+. "${srcdir=.}/tests/init.sh"; path_prepend_ ./src
+print_ver_ sort
+
+export LC_ALL=en_US.UTF-8
+locale -k LC_CTYPE | grep -q "charmap.*UTF-8" \
+  || skip_ "No UTF-8 locale available"
+
+
+cat <<EOF > exp
+Banana＠5
+Apple＠10
+Citrus＠20
+Cherry＠30
+EOF
+
+cat <<EOF | sort -t ＠ -k2 -n > out || fail=1
+Apple＠10
+Banana＠5
+Citrus＠20
+Cherry＠30
+EOF
+
+compare exp out || { fail=1; cat out; }
+
+
+cat <<EOF > exp
+Citrus＠ＡＡ20＠＠5
+Cherry＠ＡＡ30＠＠10
+Apple＠ＡＡ10＠＠20
+Banana＠ＡＡ5＠＠30
+EOF
+
+cat <<EOF | sort -t ＠ -k4 -n > out || fail=1
+Apple＠ＡＡ10＠＠20
+Banana＠ＡＡ5＠＠30
+Citrus＠ＡＡ20＠＠5
+Cherry＠ＡＡ30＠＠10
+EOF
+
+compare exp out || { fail=1; cat out; }
+
+Exit $fail
diff -Naur -i coreutils-9.10/tests/pr/pr-tests.pl coreutils-9.10.patched/tests/pr/pr-tests.pl
--- coreutils-9.10/tests/pr/pr-tests.pl	2026-01-19 06:16:08.000000000 -0600
+++ coreutils-9.10.patched/tests/pr/pr-tests.pl	2026-02-07 18:28:47.474231855 -0600
@@ -24,6 +24,15 @@
 my $prog = 'pr';
 my $normalize_strerror = "s/': .*/'/";
 
+my $mb_locale;
+#Uncomment the following line to enable multibyte tests
+$mb_locale = $ENV{LOCALE_FR_UTF8};
+! defined $mb_locale || $mb_locale eq 'none'
+  and $mb_locale = 'C';
+
+my $try = "Try \`$prog --help' for more information.\n";
+my $inval = "$prog: invalid byte, character or field list\n$try";
+
 my @tv = (
 
 # -b option is no longer an official option. But it's still working to
@@ -515,8 +524,48 @@
     {IN=>"x\tx\tx\tx\tx\nx\tx\tx\tx\tx\n"},
      {OUT=>"x\tx\tx\tx\tx\tx\tx\tx\tx\tx\n"} ];
 
+# Add _POSIX2_VERSION=199209 to the environment of each test
+# that uses an old-style option like +1.
+if ($mb_locale ne 'C')
+  {
+    # Duplicate each test vector, appending "-mb" to the test name and
+    # inserting {ENV => "LC_ALL=$mb_locale"} in the copy, so that we
+    # provide coverage for the distro-added multi-byte code paths.
+    my @new;
+    foreach my $t (@Tests)
+      {
+        my @new_t = @$t;
+        my $test_name = shift @new_t;
+
+        # Depending on whether pr is multi-byte-patched,
+        # it emits different diagnostics:
+        #   non-MB: invalid byte or field list
+        #   MB:     invalid byte, character or field list
+        # Adjust the expected error output accordingly.
+        if (grep {ref $_ eq 'HASH' && exists $_->{ERR} && $_->{ERR} eq $inval}
+            (@new_t))
+          {
+            my $sub = {ERR_SUBST => 's/, character//'};
+            push @new_t, $sub;
+            push @$t, $sub;
+          }
+        #temporarily skip some failing tests
+        next if ($test_name =~ "col-0" or $test_name =~ "col-inval" or $test_name =~ "asan1");
+        push @new, ["$test_name-mb", @new_t, {ENV => "LC_ALL=$mb_locale"}];
+      }
+    push @Tests, @new;
+  }
+
 @Tests = triple_test \@Tests;
 
+# Remember that triple_test creates from each test with exactly one "IN"
+# file two more tests (.p and .r suffix on name) corresponding to reading
+# input from a file and from a pipe.  The pipe-reading test would fail
+# due to a race condition about 1 in 20 times.
+# Remove the IN_PIPE version of the "output-is-input" test above.
+# The others aren't susceptible because they have three inputs each.
+@Tests = grep {$_->[0] ne 'output-is-input.p'} @Tests;
+
 my $save_temps = $ENV{DEBUG};
 my $verbose = $ENV{VERBOSE};
 
diff -Naur -i coreutils-9.10/tests/sort/sort-merge.pl coreutils-9.10.patched/tests/sort/sort-merge.pl
--- coreutils-9.10/tests/sort/sort-merge.pl	2026-01-19 06:16:08.000000000 -0600
+++ coreutils-9.10.patched/tests/sort/sort-merge.pl	2026-02-07 18:28:47.474282863 -0600
@@ -26,6 +26,15 @@
 # Turn off localization of executable's output.
 @ENV{qw(LANGUAGE LANG LC_ALL)} = ('C') x 3;
 
+my $mb_locale;
+# uncommented according to upstream commit enabling multibyte paths
+$mb_locale = $ENV{LOCALE_FR_UTF8};
+! defined $mb_locale || $mb_locale eq 'none'
+ and $mb_locale = 'C';
+
+my $try = "Try \`$prog --help' for more information.\n";
+my $inval = "$prog: invalid byte, character or field list\n$try";
+
 # three empty files and one that says 'foo'
 my @inputs = (+(map{{IN=> {"empty$_"=> ''}}}1..3), {IN=> {foo=> "foo\n"}});
 
@@ -77,6 +86,39 @@
         {OUT=>$big_input}],
     );
 
+# Add _POSIX2_VERSION=199209 to the environment of each test
+# that uses an old-style option like +1.
+if ($mb_locale ne 'C')
+  {
+    # Duplicate each test vector, appending "-mb" to the test name and
+    # inserting {ENV => "LC_ALL=$mb_locale"} in the copy, so that we
+    # provide coverage for the distro-added multi-byte code paths.
+    my @new;
+    foreach my $t (@Tests)
+      {
+        my @new_t = @$t;
+        my $test_name = shift @new_t;
+
+        # Depending on whether sort is multi-byte-patched,
+        # it emits different diagnostics:
+        #   non-MB: invalid byte or field list
+        #   MB:     invalid byte, character or field list
+        # Adjust the expected error output accordingly.
+        if (grep {ref $_ eq 'HASH' && exists $_->{ERR} && $_->{ERR} eq $inval}
+            (@new_t))
+          {
+            my $sub = {ERR_SUBST => 's/, character//'};
+            push @new_t, $sub;
+            push @$t, $sub;
+          }
+        next if ($test_name =~ "nmerge-.");
+        push @new, ["$test_name-mb", @new_t, {ENV => "LC_ALL=$mb_locale"}];
+      }
+    push @Tests, @new;
+  }
+
+@Tests = triple_test \@Tests;
+
 my $save_temps = $ENV{DEBUG};
 my $verbose = $ENV{VERBOSE};
 
diff -Naur -i coreutils-9.10/tests/sort/sort.pl coreutils-9.10.patched/tests/sort/sort.pl
--- coreutils-9.10/tests/sort/sort.pl	2026-01-19 06:16:08.000000000 -0600
+++ coreutils-9.10.patched/tests/sort/sort.pl	2026-02-07 18:28:47.474324591 -0600
@@ -24,10 +24,15 @@
 # Turn off localization of executable's output.
 @ENV{qw(LANGUAGE LANG LC_ALL)} = ('C') x 3;
 
-my $mb_locale = $ENV{LOCALE_FR_UTF8};
+my $mb_locale;
+#Comment out next line to disable multibyte tests
+$mb_locale = $ENV{LOCALE_FR_UTF8};
 ! defined $mb_locale || $mb_locale eq 'none'
   and $mb_locale = 'C';
 
+my $try = "Try \`$prog --help' for more information.\n";
+my $inval = "$prog: invalid byte, character or field list\n$try";
+
 # Since each test is run with a file name and with redirected stdin,
 # the name in the diagnostic is either the file name or "-".
 # Normalize each diagnostic to use '-'.
@@ -428,6 +433,38 @@
       }
   }
 
+if ($mb_locale ne 'C')
+   {
+    # Duplicate each test vector, appending "-mb" to the test name and
+    # inserting {ENV => "LC_ALL=$mb_locale"} in the copy, so that we
+    # provide coverage for the distro-added multi-byte code paths.
+    my @new;
+    foreach my $t (@Tests)
+       {
+        my @new_t = @$t;
+        my $test_name = shift @new_t;
+
+        # Depending on whether sort is multi-byte-patched,
+        # it emits different diagnostics:
+        #   non-MB: invalid byte or field list
+        #   MB:     invalid byte, character or field list
+        # Adjust the expected error output accordingly.
+        if (grep {ref $_ eq 'HASH' && exists $_->{ERR} && $_->{ERR} eq $inval}
+            (@new_t))
+          {
+            my $sub = {ERR_SUBST => 's/, character//'};
+            push @new_t, $sub;
+            push @$t, $sub;
+          }
+        #disable several failing tests until investigation, disable all tests with envvars set
+        next if (grep {ref $_ eq 'HASH' && exists $_->{ENV}} (@new_t));
+        next if ($test_name =~ "18g" or $test_name =~ "sort-numeric" or $test_name =~ "08[ab]" or $test_name =~ "03[def]" or $test_name =~ "h4" or $test_name =~ "n1" or $test_name =~ "2[01]a");
+        next if ($test_name =~ "11[ab]"); # avoid FP: expected result differs to MB result due to collation rules.
+        push @new, ["$test_name-mb", @new_t, {ENV => "LC_ALL=$mb_locale"}];
+       }
+    push @Tests, @new;
+   }
+
 @Tests = triple_test \@Tests;
 
 # Remember that triple_test creates from each test with exactly one "IN"
@@ -437,6 +474,7 @@
 # Remove the IN_PIPE version of the "output-is-input" test above.
 # The others aren't susceptible because they have three inputs each.
 @Tests = grep {$_->[0] ne 'output-is-input.p'} @Tests;
+@Tests = grep {$_->[0] ne 'output-is-input-mb.p'} @Tests;
 
 my $save_temps = $ENV{DEBUG};
 my $verbose = $ENV{VERBOSE};