/* Copyright (C) 1996-2018 Free Software Foundation, Inc.
   This file is part of the GNU C Library.
   Contributed by Ulrich Drepper <drepper@gnu.org>, 1996.

   This program 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; version 2 of the License, or
   (at your option) any later version.

   This program 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 this program; if not, see <http://www.gnu.org/licenses/>.  */

#ifdef HAVE_CONFIG_H
# include <config.h>
#endif

#include <ctype.h>
#include <errno.h>
#include <libintl.h>
#include <limits.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <stdint.h>

#include "localedef.h"
#include "linereader.h"
#include "charmap.h"
#include "charmap-dir.h"

#include <assert.h>


/* Define the lookup function.  */
#include "charmap-kw.h"


/* Prototypes for local functions.  */
static struct charmap_t *parse_charmap (struct linereader *cmfile,
					int verbose, int be_quiet);
static void new_width (struct linereader *cmfile, struct charmap_t *result,
		       const char *from, const char *to,
		       unsigned long int width);
static void charmap_new_char (struct linereader *lr, struct charmap_t *cm,
			      size_t nbytes, unsigned char *bytes,
			      const char *from, const char *to,
			      int decimal_ellipsis, int step);


bool enc_not_ascii_compatible;


#ifdef NEED_NULL_POINTER
static const char *null_pointer;
#endif

static struct linereader *
cmlr_open (const char *directory, const char *name, kw_hash_fct_t hf)
{
  FILE *fp;

  fp = charmap_open (directory, name);
  if (fp == NULL)
    return NULL;
  else
    {
      size_t dlen = strlen (directory);
      int add_slash = (dlen == 0 || directory[dlen - 1] != '/');
      size_t nlen = strlen (name);
      char *pathname;
      char *p;

      pathname = alloca (dlen + add_slash + nlen + 1);
      p = stpcpy (pathname, directory);
      if (add_slash)
	*p++ = '/';
      stpcpy (p, name);

      return lr_create (fp, pathname, hf);
    }
}

struct charmap_t *
charmap_read (const char *filename, int verbose, int error_not_found,
	      int be_quiet, int use_default)
{
  struct charmap_t *result = NULL;

  if (filename != NULL)
    {
      struct linereader *cmfile;

      /* First try the name as found in the parameter.  */
      cmfile = lr_open (filename, charmap_hash);
      if (cmfile == NULL)
	{
	  /* No successful.  So start looking through the directories
	     in the I18NPATH if this is a simple name.  */
	  if (strchr (filename, '/') == NULL)
	    {
	      char *i18npath = getenv ("I18NPATH");
	      if (i18npath != NULL && *i18npath != '\0')
		{
		  const size_t pathlen = strlen (i18npath);
		  char i18npathbuf[pathlen + 1];
		  char path[pathlen + sizeof ("/charmaps")];
		  char *next;
		  i18npath = memcpy (i18npathbuf, i18npath, pathlen + 1);

		  while (cmfile == NULL
			 && (next = strsep (&i18npath, ":")) != NULL)
		    {
		      stpcpy (stpcpy (path, next), "/charmaps");
		      cmfile = cmlr_open (path, filename, charmap_hash);

		      if (cmfile == NULL)
			/* Try without the "/charmaps" part.  */
			cmfile = cmlr_open (next, filename, charmap_hash);
		    }
		}

	      if (cmfile == NULL)
		/* Try the default directory.  */
		cmfile = cmlr_open (CHARMAP_PATH, filename, charmap_hash);
	    }
	}

      if (cmfile != NULL)
	result = parse_charmap (cmfile, verbose, be_quiet);

      if (result == NULL && error_not_found)
	record_error (0, errno,
		      _("character map file `%s' not found"),
		      filename);
    }

  if (result == NULL && filename != NULL && strchr (filename, '/') == NULL)
    {
      /* OK, one more try.  We also accept the names given to the
	 character sets in the files.  Sometimes they differ from the
	 file name.  */
      CHARMAP_DIR *dir;

      dir = charmap_opendir (CHARMAP_PATH);
      if (dir != NULL)
	{
	  const char *dirent;

	  while ((dirent = charmap_readdir (dir)) != NULL)
	    {
	      char **aliases;
	      char **p;
	      int found;

	      aliases = charmap_aliases (CHARMAP_PATH, dirent);
	      found = 0;
	      for (p = aliases; *p; p++)
		if (strcasecmp (*p, filename) == 0)
		  {
		    found = 1;
		    break;
		  }
	      charmap_free_aliases (aliases);

	      if (found)
		{
		  struct linereader *cmfile;

		  cmfile = cmlr_open (CHARMAP_PATH, dirent, charmap_hash);
		  if (cmfile != NULL)
		    result = parse_charmap (cmfile, verbose, be_quiet);

		  break;
		}
	    }

	  charmap_closedir (dir);
	}
    }

  if (result == NULL && DEFAULT_CHARMAP != NULL)
    {
      struct linereader *cmfile;

      cmfile = cmlr_open (CHARMAP_PATH, DEFAULT_CHARMAP, charmap_hash);
      if (cmfile != NULL)
	result = parse_charmap (cmfile, verbose, be_quiet);

      if (result == NULL)
	record_error (4, errno,
		      _("default character map file `%s' not found"),
		      DEFAULT_CHARMAP);
    }

  if (result != NULL && result->code_set_name == NULL)
    /* The input file does not specify a code set name.  This
       shouldn't happen but we should cope with it.  */
    result->code_set_name = basename (filename);

  /* Test of ASCII compatibility of locale encoding.

     Verify that the encoding to be used in a locale is ASCII compatible,
     at least for the graphic characters, excluding the control characters,
     '$' and '@'.  This constraint comes from an ISO C 99 restriction.

     ISO C 99 section 7.17.(2) (about wchar_t):
       the null character shall have the code value zero and each member of
       the basic character set shall have a code value equal to its value
       when used as the lone character in an integer character constant.
     ISO C 99 section 5.2.1.(3):
       Both the basic source and basic execution character sets shall have
       the following members: the 26 uppercase letters of the Latin alphabet
            A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
       the 26 lowercase letters of the Latin alphabet
            a b c d e f g h i j k l m n o p q r s t u v w x y z
       the 10 decimal digits
            0 1 2 3 4 5 6 7 8 9
       the following 29 graphic characters
            ! " # % & ' ( ) * + , - . / : ; < = > ? [ \ ] ^ _ { | } ~
       the space character, and control characters representing horizontal
       tab, vertical tab, and form feed.

     Therefore, for all members of the "basic character set", the 'char' code
     must have the same value as the 'wchar_t' code, which in glibc is the
     same as the Unicode code, which for all of the enumerated characters
     is identical to the ASCII code. */
  if (result != NULL && use_default)
    {
      static const char basic_charset[] =
	{
	  'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M',
	  'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z',
	  'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm',
	  'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z',
	  '0', '1', '2', '3', '4', '5', '6', '7', '8', '9',
	  '!', '"', '#', '%', '&', '\'', '(', ')', '*', '+', ',', '-',
	  '.', '/', ':', ';', '<', '=', '>', '?', '[', '\\', ']', '^',
	  '_', '{', '|', '}', '~', ' ', '\t', '\v', '\f', '\0'
	};
      int failed = 0;
      const char *p = basic_charset;

      do
	{
	  struct charseq *seq = charmap_find_symbol (result, p, 1);

	  if (seq == NULL || seq->ucs4 != (uint32_t) *p)
	    failed = 1;
	}
      while (*p++ != '\0');

      if (failed)
	{
	  /* A user may disable the ASCII compatibility warning check,
	     but we must remember that the encoding is not ASCII
	     compatible, since it may have other implications.  Later
	     we will set _NL_CTYPE_MAP_TO_NONASCII from this value.  */
	  if (warn_ascii)
	    record_warning (_(
"character map `%s' is not ASCII compatible, locale not ISO C compliant "
"[--no-warnings=ascii]"),
			    result->code_set_name);
	  enc_not_ascii_compatible = true;
	}
    }

  return result;
}


static struct charmap_t *
parse_charmap (struct linereader *cmfile, int verbose, int be_quiet)
{
  struct charmap_t *result;
  int state;
  enum token_t expected_tok = tok_error;
  const char *expected_str = NULL;
  char *from_name = NULL;
  char *to_name = NULL;
  enum token_t ellipsis = 0;
  int step = 1;

  /* We don't want symbolic names in string to be translated.  */
  cmfile->translate_strings = 0;

  /* Allocate room for result.  */
  result = (struct charmap_t *) xmalloc (sizeof (struct charmap_t));
  memset (result, '\0', sizeof (struct charmap_t));
  /* The default DEFAULT_WIDTH is 1.  */
  result->width_default = 1;

#define obstack_chunk_alloc malloc
#define obstack_chunk_free free
  obstack_init (&result->mem_pool);

  if (init_hash (&result->char_table, 256)
      || init_hash (&result->byte_table, 256))
    {
      free (result);
      return NULL;
    }

  /* We use a state machine to describe the charmap description file
     format.  */
  state = 1;
  while (1)
    {
      /* What's on?  */
      struct token *now = lr_token (cmfile, NULL, NULL, NULL, verbose);
      enum token_t nowtok = now->tok;
      struct token *arg;

      if (nowtok == tok_eof)
	break;

      switch (state)
	{
	case 1:
	  /* The beginning.  We expect the special declarations, EOL or
	     `CHARMAP'.  */
	  if (nowtok == tok_eol)
	    /* Ignore empty lines.  */
	    continue;

	  if (nowtok == tok_charmap)
	    {
	      from_name = NULL;
	      to_name = NULL;

	      /* We have to set up the real work.  Fill in some
		 default values.  */
	      if (result->mb_cur_max == 0)
		result->mb_cur_max = 1;
	      if (result->mb_cur_min == 0)
		result->mb_cur_min = result->mb_cur_max;
	      if (result->mb_cur_min > result->mb_cur_max)
		{
		  record_error (0, 0, _("\
%s: <mb_cur_max> must be greater than <mb_cur_min>\n"),
				cmfile->fname);

		  result->mb_cur_min = result->mb_cur_max;
		}

	      lr_ignore_rest (cmfile, 1);

	      state = 2;
	      continue;
	    }

	  if (nowtok != tok_code_set_name && nowtok != tok_mb_cur_max
	      && nowtok != tok_mb_cur_min && nowtok != tok_escape_char
	      && nowtok != tok_comment_char && nowtok != tok_g0esc
	      && nowtok != tok_g1esc && nowtok != tok_g2esc
	      && nowtok != tok_g3esc && nowtok != tok_repertoiremap
	      && nowtok != tok_include)
	    {
	      lr_error (cmfile, _("syntax error in prolog: %s"),
			_("invalid definition"));

	      lr_ignore_rest (cmfile, 0);
	      continue;
	    }

	  /* We know that we need an argument.  */
	  arg = lr_token (cmfile, NULL, NULL, NULL, verbose);

	  switch (nowtok)
	    {
	    case tok_code_set_name:
	    case tok_repertoiremap:
	      if (arg->tok != tok_ident && arg->tok != tok_string)
		{
		badarg:
		  lr_error (cmfile, _("syntax error in prolog: %s"),
			    _("bad argument"));

		  lr_ignore_rest (cmfile, 0);
		  continue;
		}

	      if (nowtok == tok_code_set_name)
		result->code_set_name = obstack_copy0 (&result->mem_pool,
						       arg->val.str.startmb,
						       arg->val.str.lenmb);
	      else
		result->repertoiremap = obstack_copy0 (&result->mem_pool,
						       arg->val.str.startmb,
						       arg->val.str.lenmb);

	      lr_ignore_rest (cmfile, 1);
	      continue;

	    case tok_mb_cur_max:
	    case tok_mb_cur_min:
	      if (arg->tok != tok_number)
		goto badarg;

	      if ((nowtok == tok_mb_cur_max
		       && result->mb_cur_max != 0)
		      || (nowtok == tok_mb_cur_max
			  && result->mb_cur_max != 0))
		lr_error (cmfile, _("duplicate definition of <%s>"),
			  nowtok == tok_mb_cur_min
			  ? "mb_cur_min" : "mb_cur_max");

	      if (arg->val.num < 1)
		{
		  lr_error (cmfile,
			    _("value for <%s> must be 1 or greater"),
			    nowtok == tok_mb_cur_min
			    ? "mb_cur_min" : "mb_cur_max");

		  lr_ignore_rest (cmfile, 0);
		  continue;
		}
	      if ((nowtok == tok_mb_cur_max && result->mb_cur_min != 0
		   && (int) arg->val.num < result->mb_cur_min)
		  || (nowtok == tok_mb_cur_min && result->mb_cur_max != 0
		      && (int) arg->val.num > result->mb_cur_max))
		{
		  lr_error (cmfile, _("\
value of <%s> must be greater or equal than the value of <%s>"),
			    "mb_cur_max", "mb_cur_min");

		  lr_ignore_rest (cmfile, 0);
		  continue;
		}

	      if (nowtok == tok_mb_cur_max)
		result->mb_cur_max = arg->val.num;
	      else
		result->mb_cur_min = arg->val.num;

	      lr_ignore_rest (cmfile, 1);
	      continue;

	    case tok_escape_char:
	    case tok_comment_char:
	      if (arg->tok != tok_ident)
		goto badarg;

	      if (arg->val.str.lenmb != 1)
		{
		  lr_error (cmfile, _("\
argument to <%s> must be a single character"),
			    nowtok == tok_escape_char ? "escape_char"
						      : "comment_char");

		  lr_ignore_rest (cmfile, 0);
		  continue;
		}

	      if (nowtok == tok_escape_char)
		cmfile->escape_char = *arg->val.str.startmb;
	      else
		cmfile->comment_char = *arg->val.str.startmb;

	      lr_ignore_rest (cmfile, 1);
	      continue;

	    case tok_g0esc:
	    case tok_g1esc:
	    case tok_g2esc:
	    case tok_g3esc:
	    case tok_escseq:
	      lr_ignore_rest (cmfile, 0); /* XXX */
	      continue;

	    case tok_include:
	      lr_error (cmfile, _("\
character sets with locking states are not supported"));
	      exit (4);

	    default:
	      /* Cannot happen.  */
	      assert (! "Should not happen");
	    }
	  break;

	case 2:
	  /* We have seen `CHARMAP' and now are in the body.  Each line
	     must have the format "%s %s %s\n" or "%s...%s %s %s\n".  */
	  if (nowtok == tok_eol)
	    /* Ignore empty lines.  */
	    continue;

	  if (nowtok == tok_end)
	    {
	      expected_tok = tok_charmap;
	      expected_str = "CHARMAP";
	      state = 90;
	      continue;
	    }

	  if (nowtok != tok_bsymbol && nowtok != tok_ucs4)
	    {
	      lr_error (cmfile, _("syntax error in %s definition: %s"),
			"CHARMAP", _("no symbolic name given"));

	      lr_ignore_rest (cmfile, 0);
	      continue;
	    }

	  /* If the previous line was not completely correct free the
	     used memory.  */
	  if (from_name != NULL)
	    obstack_free (&result->mem_pool, from_name);

	  if (nowtok == tok_bsymbol)
	    from_name = (char *) obstack_copy0 (&result->mem_pool,
						now->val.str.startmb,
						now->val.str.lenmb);
	  else
	    {
	      obstack_printf (&result->mem_pool, "U%08X",
			      cmfile->token.val.ucs4);
	      obstack_1grow (&result->mem_pool, '\0');
	      from_name = (char *) obstack_finish (&result->mem_pool);
	    }
	  to_name = NULL;

	  state = 3;
	  continue;

	case 3:
	  /* We have two possibilities: We can see an ellipsis or an
	     encoding value.  */
	  if (nowtok == tok_ellipsis3 || nowtok == tok_ellipsis4
	      || nowtok == tok_ellipsis2 || nowtok == tok_ellipsis4_2
	      || nowtok == tok_ellipsis2_2)
	    {
	      ellipsis = nowtok;
	      if (nowtok == tok_ellipsis4_2)
		{
		  step = 2;
		  nowtok = tok_ellipsis4;
		}
	      else if (nowtok == tok_ellipsis2_2)
		{
		  step = 2;
		  nowtok = tok_ellipsis2;
		}
	      state = 4;
	      continue;
	    }
	  /* FALLTHROUGH */

	case 5:
	  if (nowtok != tok_charcode)
	    {
	      lr_error (cmfile, _("syntax error in %s definition: %s"),
			"CHARMAP", _("invalid encoding given"));

	      lr_ignore_rest (cmfile, 0);

	      state = 2;
	      continue;
	    }

	  if (now->val.charcode.nbytes < result->mb_cur_min)
	    lr_error (cmfile, _("too few bytes in character encoding"));
	  else if (now->val.charcode.nbytes > result->mb_cur_max)
	    lr_error (cmfile, _("too many bytes in character encoding"));
	  else
	    charmap_new_char (cmfile, result, now->val.charcode.nbytes,
			      now->val.charcode.bytes, from_name, to_name,
			      ellipsis != tok_ellipsis2, step);

	  /* Ignore trailing comment silently.  */
	  lr_ignore_rest (cmfile, 0);

	  from_name = NULL;
	  to_name = NULL;
	  ellipsis = tok_none;
	  step = 1;

	  state = 2;
	  continue;

	case 4:
	  if (nowtok != tok_bsymbol && nowtok != tok_ucs4)
	    {
	      lr_error (cmfile, _("syntax error in %s definition: %s"),
			"CHARMAP",
			_("no symbolic name given for end of range"));

	      lr_ignore_rest (cmfile, 0);
	      continue;
	    }

	  /* Copy the to-name in a safe place.  */
	  if (nowtok == tok_bsymbol)
	    to_name = (char *) obstack_copy0 (&result->mem_pool,
					      cmfile->token.val.str.startmb,
					      cmfile->token.val.str.lenmb);
	  else
	    {
	      obstack_printf (&result->mem_pool, "U%08X",
			      cmfile->token.val.ucs4);
	      obstack_1grow (&result->mem_pool, '\0');
	      to_name = (char *) obstack_finish (&result->mem_pool);
	    }

	  state = 5;
	  continue;

	case 90:
	  if (nowtok != expected_tok)
	    lr_error (cmfile, _("\
%1$s: definition does not end with `END %1$s'"), expected_str);

	  lr_ignore_rest (cmfile, nowtok == expected_tok);
	  state = 91;
	  continue;

	case 91:
	  /* Waiting for WIDTH... */
	  if (nowtok == tok_eol)
	    /* Ignore empty lines.  */
	    continue;

	  if (nowtok == tok_width_default)
	    {
	      state = 92;
	      continue;
	    }

	  if (nowtok == tok_width)
	    {
	      lr_ignore_rest (cmfile, 1);
	      state = 93;
	      continue;
	    }

	  if (nowtok == tok_width_variable)
	    {
	      lr_ignore_rest (cmfile, 1);
	      state = 98;
	      continue;
	    }

	  lr_error (cmfile, _("\
only WIDTH definitions are allowed to follow the CHARMAP definition"));

	  lr_ignore_rest (cmfile, 0);
	  continue;

	case 92:
	  if (nowtok != tok_number)
	    lr_error (cmfile, _("value for %s must be an integer"),
		      "WIDTH_DEFAULT");
	  else
	    result->width_default = now->val.num;

	  lr_ignore_rest (cmfile, nowtok == tok_number);

	  state = 91;
	  continue;

	case 93:
	  /* We now expect `END WIDTH' or lines of the format "%s %d\n" or
	     "%s...%s %d\n".  */
	  if (nowtok == tok_eol)
	    /* ignore empty lines.  */
	    continue;

	  if (nowtok == tok_end)
	    {
	      expected_tok = tok_width;
	      expected_str = "WIDTH";
	      state = 90;
	      continue;
	    }

	  if (nowtok != tok_bsymbol && nowtok != tok_ucs4)
	    {
	      lr_error (cmfile, _("syntax error in %s definition: %s"),
			"WIDTH", _("no symbolic name given"));

	      lr_ignore_rest (cmfile, 0);
	      continue;
	    }

	  if (from_name != NULL)
	    obstack_free (&result->mem_pool, from_name);

	  if (nowtok == tok_bsymbol)
	    from_name = (char *) obstack_copy0 (&result->mem_pool,
						now->val.str.startmb,
						now->val.str.lenmb);
	  else
	    {
	      obstack_printf (&result->mem_pool, "U%08X",
			      cmfile->token.val.ucs4);
	      obstack_1grow (&result->mem_pool, '\0');
	      from_name = (char *) obstack_finish (&result->mem_pool);
	    }

	  to_name = NULL;

	  state = 94;
	  continue;

	case 94:
	  if (nowtok == tok_ellipsis3)
	    {
	      state = 95;
	      continue;
	    }

	case 96:
	  if (nowtok != tok_number)
	    lr_error (cmfile, _("value for %s must be an integer"),
		      "WIDTH");
	  else
	    {
	      /* Store width for chars.  */
	      new_width (cmfile, result, from_name, to_name, now->val.num);

	      from_name = NULL;
	      to_name = NULL;
	    }

	  lr_ignore_rest (cmfile, nowtok == tok_number);

	  state = 93;
	  continue;

	case 95:
	  if (nowtok != tok_bsymbol && nowtok != tok_ucs4)
	    {
	      lr_error (cmfile, _("syntax error in %s definition: %s"),
			"WIDTH", _("no symbolic name given for end of range"));

	      lr_ignore_rest (cmfile, 0);

	      state = 93;
	      continue;
	    }

	  if (nowtok == tok_bsymbol)
	    to_name = (char *) obstack_copy0 (&result->mem_pool,
					      now->val.str.startmb,
					      now->val.str.lenmb);
	  else
	    {
	      obstack_printf (&result->mem_pool, "U%08X",
			      cmfile->token.val.ucs4);
	      obstack_1grow (&result->mem_pool, '\0');
	      to_name = (char *) obstack_finish (&result->mem_pool);
	    }

	  state = 96;
	  continue;

	case 98:
	  /* We now expect `END WIDTH_VARIABLE' or lines of the format
	     "%s\n" or "%s...%s\n".  */
	  if (nowtok == tok_eol)
	    /* ignore empty lines.  */
	    continue;

	  if (nowtok == tok_end)
	    {
	      expected_tok = tok_width_variable;
	      expected_str = "WIDTH_VARIABLE";
	      state = 90;
	      continue;
	    }

	  if (nowtok != tok_bsymbol && nowtok != tok_ucs4)
	    {
	      lr_error (cmfile, _("syntax error in %s definition: %s"),
			"WIDTH_VARIABLE", _("no symbolic name given"));

	      lr_ignore_rest (cmfile, 0);

	      continue;
	    }

	  if (from_name != NULL)
	    obstack_free (&result->mem_pool, from_name);

	  if (nowtok == tok_bsymbol)
	    from_name = (char *) obstack_copy0 (&result->mem_pool,
						now->val.str.startmb,
						now->val.str.lenmb);
	  else
	    {
	      obstack_printf (&result->mem_pool, "U%08X",
			      cmfile->token.val.ucs4);
	      obstack_1grow (&result->mem_pool, '\0');
	      from_name = (char *) obstack_finish (&result->mem_pool);
	    }
	  to_name = NULL;

	  state = 99;
	  continue;

	case 99:
	  if (nowtok == tok_ellipsis3)
	    state = 100;

	  /* Store info.  */
	  from_name = NULL;

	  /* Warn */
	  state = 98;
	  continue;

	case 100:
	  if (nowtok != tok_bsymbol && nowtok != tok_ucs4)
	    {
	      lr_error (cmfile, _("syntax error in %s definition: %s"),
			"WIDTH_VARIABLE",
			_("no symbolic name given for end of range"));
	      lr_ignore_rest (cmfile, 0);
	      continue;
	    }

	  if (nowtok == tok_bsymbol)
	    to_name = (char *) obstack_copy0 (&result->mem_pool,
					      now->val.str.startmb,
					      now->val.str.lenmb);
	  else
	    {
	      obstack_printf (&result->mem_pool, "U%08X",
			      cmfile->token.val.ucs4);
	      obstack_1grow (&result->mem_pool, '\0');
	      to_name = (char *) obstack_finish (&result->mem_pool);
	    }

	  /* XXX Enter value into table.  */

	  lr_ignore_rest (cmfile, 1);

	  state = 98;
	  continue;

	default:
	  record_error (5, 0, _("%s: error in state machine"),
			__FILE__);
	  /* NOTREACHED */
	}
      break;
    }

  if (state != 91)
    record_error (0, 0, _("%s: premature end of file"),
		  cmfile->fname);

  lr_close (cmfile);

  return result;
}


static void
new_width (struct linereader *cmfile, struct charmap_t *result,
	   const char *from, const char *to, unsigned long int width)
{
  struct charseq *from_val;
  struct charseq *to_val;

  from_val = charmap_find_value (result, from, strlen (from));
  if (from_val == NULL)
    {
      lr_error (cmfile, _("unknown character `%s'"), from);
      return;
    }

  if (to == NULL)
    to_val = from_val;
  else
    {
      to_val = charmap_find_value (result, to, strlen (to));
      if (to_val == NULL)
	{
	  lr_error (cmfile, _("unknown character `%s'"), to);
	  return;
	}

      /* Make sure the number of bytes for the end points of the range
	 is correct.  */
      if (from_val->nbytes != to_val->nbytes)
	{
	  lr_error (cmfile, _("\
number of bytes for byte sequence of beginning and end of range not the same: %d vs %d"),
		    from_val->nbytes, to_val->nbytes);
	  return;
	}
    }

  if (result->nwidth_rules >= result->nwidth_rules_max)
    {
      size_t new_size = result->nwidth_rules + 32;
      struct width_rule *new_rules =
	(struct width_rule *) obstack_alloc (&result->mem_pool,
					     (new_size
					      * sizeof (struct width_rule)));

      memcpy (new_rules, result->width_rules,
	      result->nwidth_rules_max * sizeof (struct width_rule));

      result->width_rules = new_rules;
      result->nwidth_rules_max = new_size;
    }

  result->width_rules[result->nwidth_rules].from = from_val;
  result->width_rules[result->nwidth_rules].to = to_val;
  result->width_rules[result->nwidth_rules].width = (unsigned int) width;
  ++result->nwidth_rules;
}


struct charseq *
charmap_find_value (const struct charmap_t *cm, const char *name, size_t len)
{
  void *result;

  return (find_entry ((hash_table *) &cm->char_table, name, len, &result)
	  < 0 ? NULL : (struct charseq *) result);
}


static void
charmap_new_char (struct linereader *lr, struct charmap_t *cm,
		  size_t nbytes, unsigned char *bytes,
		  const char *from, const char *to,
		  int decimal_ellipsis, int step)
{
  hash_table *ht = &cm->char_table;
  hash_table *bt = &cm->byte_table;
  struct obstack *ob = &cm->mem_pool;
  char *from_end;
  char *to_end;
  const char *cp;
  int prefix_len, len1, len2;
  unsigned int from_nr, to_nr, cnt;
  struct charseq *newp;

  len1 = strlen (from);

  if (to == NULL)
    {
      newp = (struct charseq *) obstack_alloc (ob, sizeof (*newp) + nbytes);
      newp->nbytes = nbytes;
      memcpy (newp->bytes, bytes, nbytes);
      newp->name = from;

      newp->ucs4 = UNINITIALIZED_CHAR_VALUE;
      if ((from[0] == 'U' || from[0] == 'P') && (len1 == 5 || len1 == 9))
	{
	  /* Maybe the name is of the form `Uxxxx' or `Uxxxxxxxx' where
	     xxxx and xxxxxxxx are hexadecimal numbers.  In this case
	     we use the value of xxxx or xxxxxxxx as the UCS4 value of
	     this character and we don't have to consult the repertoire
	     map.

	     If the name is of the form `Pxxxx' or `Pxxxxxxxx' the xxxx
	     and xxxxxxxx also give the code point in UCS4 but this must
	     be in the private, i.e., unassigned, area.  This should be
	     used for characters which do not (yet) have an equivalent
	     in ISO 10646 and Unicode.  */
	  char *endp;

	  errno = 0;
	  newp->ucs4 = strtoul (from + 1, &endp, 16);
	  if (endp - from != len1
	      || (newp->ucs4 == ~((uint32_t) 0) && errno == ERANGE)
	      || newp->ucs4 >= 0x80000000)
	    /* This wasn't successful.  Signal this name cannot be a
	       correct UCS value.  */
	    newp->ucs4 = UNINITIALIZED_CHAR_VALUE;
	}

      insert_entry (ht, from, len1, newp);
      insert_entry (bt, newp->bytes, nbytes, newp);
      /* Please note that it isn't a bug if a symbol is defined more
	 than once.  All later definitions are simply discarded.  */
      return;
    }

  /* We have a range: the names must have names with equal prefixes
     and an equal number of digits, where the second number is greater
     or equal than the first.  */
  len2 = strlen (to);

  if (len1 != len2)
    {
    illegal_range:
      lr_error (lr, _("invalid names for character range"));
      return;
    }

  cp = &from[len1 - 1];
  if (decimal_ellipsis)
    while (isdigit (*cp) && cp >= from)
      --cp;
  else
    while (isxdigit (*cp) && cp >= from)
      {
	if (!isdigit (*cp) && !isupper (*cp))
	  lr_error (lr, _("\
hexadecimal range format should use only capital characters"));
	--cp;
      }

  prefix_len = (cp - from) + 1;

  if (cp == &from[len1 - 1] || strncmp (from, to, prefix_len) != 0)
    goto illegal_range;

  errno = 0;
  from_nr = strtoul (&from[prefix_len], &from_end, decimal_ellipsis ? 10 : 16);
  if (*from_end != '\0' || (from_nr == UINT_MAX && errno == ERANGE)
      || ((to_nr = strtoul (&to[prefix_len], &to_end,
			    decimal_ellipsis ? 10 : 16)) == UINT_MAX
	  && errno == ERANGE)
      || *to_end != '\0')
    {
      lr_error (lr, _("<%s> and <%s> are invalid names for range"), from, to);
      return;
    }

  if (from_nr > to_nr)
    {
      lr_error (lr, _("upper limit in range is smaller than lower limit"));
      return;
    }

  for (cnt = from_nr; cnt <= to_nr; cnt += step)
    {
      char *name_end;
      obstack_printf (ob, decimal_ellipsis ? "%.*s%0*d" : "%.*s%0*X",
		      prefix_len, from, len1 - prefix_len, cnt);
      obstack_1grow (ob, '\0');
      name_end = obstack_finish (ob);

      newp = (struct charseq *) obstack_alloc (ob, sizeof (*newp) + nbytes);
      newp->nbytes = nbytes;
      memcpy (newp->bytes, bytes, nbytes);
      newp->name = name_end;

      newp->ucs4 = UNINITIALIZED_CHAR_VALUE;
      if ((name_end[0] == 'U' || name_end[0] == 'P')
	  && (len1 == 5 || len1 == 9))
	{
	  /* Maybe the name is of the form `Uxxxx' or `Uxxxxxxxx' where
	     xxxx and xxxxxxxx are hexadecimal numbers.  In this case
	     we use the value of xxxx or xxxxxxxx as the UCS4 value of
	     this character and we don't have to consult the repertoire
	     map.

	     If the name is of the form `Pxxxx' or `Pxxxxxxxx' the xxxx
	     and xxxxxxxx also give the code point in UCS4 but this must
	     be in the private, i.e., unassigned, area.  This should be
	     used for characters which do not (yet) have an equivalent
	     in ISO 10646 and Unicode.  */
	  char *endp;

	  errno = 0;
	  newp->ucs4 = strtoul (name_end + 1, &endp, 16);
	  if (endp - name_end != len1
	      || (newp->ucs4 == ~((uint32_t) 0) && errno == ERANGE)
	      || newp->ucs4 >= 0x80000000)
	    /* This wasn't successful.  Signal this name cannot be a
	       correct UCS value.  */
	    newp->ucs4 = UNINITIALIZED_CHAR_VALUE;
	}

      insert_entry (ht, name_end, len1, newp);
      insert_entry (bt, newp->bytes, nbytes, newp);
      /* Please note we don't examine the return value since it is no error
	 if we have two definitions for a symbol.  */

      /* Increment the value in the byte sequence.  */
      if (++bytes[nbytes - 1] == '\0')
	{
	  int b = nbytes - 2;

	  do
	    if (b < 0)
	      {
		lr_error (lr,
			  _("resulting bytes for range not representable."));
		return;
	      }
	  while (++bytes[b--] == 0);
	}
    }
}


struct charseq *
charmap_find_symbol (const struct charmap_t *cm, const char *bytes,
		     size_t nbytes)
{
  void *result;

  return (find_entry ((hash_table *) &cm->byte_table, bytes, nbytes, &result)
	  < 0 ? NULL : (struct charseq *) result);
}