mcabber/mcabber/utf8.c
author Mikael Berthe <mikael@lilotux.net>
Mon, 22 Mar 2010 21:32:11 +0100
changeset 1811 e6d355e50d7a
parent 1668 41c26b7d2890
child 2267 dda89a3801c4
permissions -rw-r--r--
Update Vim modelines
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
930
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
     1
/*
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
     2
 * utf8.c       -- UTF-8 routines
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
     3
 *
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
     4
 * Copyright (C) 2006 Reimar Döffinger <Reimar.Doeffinger@stud.uni-karlsruhe.de>
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
     5
 *
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
     6
 * This program is free software; you can redistribute it and/or modify
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
     7
 * it under the terms of the GNU General Public License as published by
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
     8
 * the Free Software Foundation; either version 2 of the License, or (at
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
     9
 * your option) any later version.
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    10
 *
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    11
 * This program is distributed in the hope that it will be useful, but
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    12
 * WITHOUT ANY WARRANTY; without even the implied warranty of
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    13
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    14
 * General Public License for more details.
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    15
 *
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    16
 * You should have received a copy of the GNU General Public License
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    17
 * along with this program; if not, write to the Free Software
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    18
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    19
 * USA
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    20
 */
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    21
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    22
#include "utf8.h"
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    23
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    24
char *prev_char(char *str, const char *limit)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    25
{
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    26
  if (str <= limit)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    27
    return str;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    28
  str--;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    29
  if (utf8_mode)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    30
    while ((str > limit) && ((*str & 0xc0) == 0x80))
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    31
      str--;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    32
  return str;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    33
}
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    34
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    35
char *next_char(char *str)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    36
{
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    37
  if (!*str)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    38
    return str;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    39
  str++;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    40
  if (utf8_mode)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    41
    while ((*str & 0xc0) == 0x80)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    42
      str++;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    43
  return str;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    44
}
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    45
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    46
unsigned get_char(const char *str)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    47
{
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    48
  unsigned char *strp = (unsigned char *)str;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    49
  unsigned c = *strp++;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    50
  unsigned mask = 0x80;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    51
  int len = -1;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    52
  if (!utf8_mode)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    53
    return c;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    54
  while (c & mask) {
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    55
    mask >>= 1;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    56
    len++;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    57
  }
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    58
  if (len <= 0 || len > 4)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    59
    goto no_utf8;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    60
  c &= mask - 1;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    61
  while ((*strp & 0xc0) == 0x80) {
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    62
    if (len-- <= 0)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    63
      goto no_utf8;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    64
    c = (c << 6) | (*strp++ & 0x3f);
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    65
  }
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    66
  if (len)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    67
    goto no_utf8;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    68
  return c;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    69
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    70
no_utf8:
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    71
  return *str;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    72
}
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    73
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    74
char *put_char(char *str, unsigned c)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    75
{
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    76
  int mask = 0xffffffc0;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    77
  int i = 4;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    78
  char code[5];
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    79
  if (!utf8_mode || c < 128) {
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    80
    *str++ = c;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    81
    return str;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    82
  }
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    83
  while (c & mask) {
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    84
    code[i--] = 0x80 | (c & 0x3f);
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    85
    c >>= 6;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    86
    mask >>= 1;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    87
    if (i < 0) {
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    88
      *str++ = '?';
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    89
      return str;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    90
    }
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    91
  }
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    92
  code[i] = (mask << 1) | c;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    93
  for (; i < 5; i++)
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    94
    *str++ = code[i];
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    95
  return str;
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    96
}
a75f7a13df7b UTF-8 terminal support (Reimar Döffinger)
Mikael Berthe <mikael@lilotux.net>
parents:
diff changeset
    97
1811
e6d355e50d7a Update Vim modelines
Mikael Berthe <mikael@lilotux.net>
parents: 1668
diff changeset
    98
/* vim: set et cindent cinoptions=>2\:2(0 ts=2 sw=2:  For Vim users... */