1/*
2 * Copyright (C) 1984-2000 Mark Nudelman
3 *
4 * You may distribute under the terms of either the GNU General Public
5 * License or the Less License, as specified in the README file.
6 *
7 * For more information about less, or for information on how to
8 * contact the author, see the README file.
9 */
10
11
12/*
13 * Functions to define the character set
14 * and do things specific to the character set.
15 */
16
17#include "less.h"
18#if HAVE_LOCALE
19#include <locale.h>
20#include <ctype.h>
21#endif
22
23public int utf_mode = 0;
24
25/*
26 * Predefined character sets,
27 * selected by the LESSCHARSET environment variable.
28 */
29struct charset {
30 char *name;
31 int *p_flag;
32 char *desc;
33} charsets[] = {
34 { "ascii", NULL, "8bcccbcc18b95.b" },
35 { "dos", NULL, "8bcccbcc12bc5b95.b." },
36 { "ebcdic", NULL, "5bc6bcc7bcc41b.9b7.9b5.b..8b6.10b6.b9.7b9.8b8.17b3.3b9.7b9.8b8.6b10.b.b.b." },
37 { "iso8859", NULL, "8bcccbcc18b95.33b." },
38 { "koi8-r", NULL, "8bcccbcc18b95.b128." },
39 { "latin1", NULL, "8bcccbcc18b95.33b." },
40 { "next", NULL, "8bcccbcc18b95.bb125.bb" },
41 { "utf-8", &utf_mode, "8bcccbcc18b." },
42 { NULL, NULL, NULL }
43};
44
45#define IS_BINARY_CHAR 01
46#define IS_CONTROL_CHAR 02
47
48static char chardef[256];
49static char *binfmt = NULL;
50public int binattr = AT_STANDOUT;
51
52
53/*
54 * Define a charset, given a description string.
55 * The string consists of 256 letters,
56 * one for each character in the charset.
57 * If the string is shorter than 256 letters, missing letters
58 * are taken to be identical to the last one.
59 * A decimal number followed by a letter is taken to be a
60 * repetition of the letter.
61 *
62 * Each letter is one of:
63 * . normal character
64 * b binary character
65 * c control character
66 */
67 static void
68ichardef(s)
69 char *s;
70{
71 register char *cp;
72 register int n;
73 register char v;
74
75 n = 0;
76 v = 0;
77 cp = chardef;
78 while (*s != '\0')
79 {
80 switch (*s++)
81 {
82 case '.':
83 v = 0;
84 break;
85 case 'c':
86 v = IS_CONTROL_CHAR;
87 break;
88 case 'b':
89 v = IS_BINARY_CHAR|IS_CONTROL_CHAR;
90 break;
91
92 case '0': case '1': case '2': case '3': case '4':
93 case '5': case '6': case '7': case '8': case '9':
94 n = (10 * n) + (s[-1] - '0');
95 continue;
96
97 default:
98 error("invalid chardef", NULL_PARG);
99 quit(QUIT_ERROR);
100 /*NOTREACHED*/
101 }
102
103 do
104 {
105 if (cp >= chardef + sizeof(chardef))
106 {
107 error("chardef longer than 256", NULL_PARG);
108 quit(QUIT_ERROR);
109 /*NOTREACHED*/
110 }
111 *cp++ = v;
112 } while (--n > 0);
113 n = 0;
114 }
115
116 while (cp < chardef + sizeof(chardef))
117 *cp++ = v;
118}
119
120/*
121 * Define a charset, given a charset name.
122 * The valid charset names are listed in the "charsets" array.
123 */
124 static int
125icharset(name)
126 register char *name;
127{
128 register struct charset *p;
129
130 if (name == NULL || *name == '\0')
131 return (0);
132
133 for (p = charsets; p->name != NULL; p++)
134 {
135 if (strcmp(name, p->name) == 0)
136 {
137 ichardef(p->desc);
138 if (p->p_flag != NULL)
139 *(p->p_flag) = 1;
140 return (1);
141 }
142 }
143
144 error("invalid charset name", NULL_PARG);
145 quit(QUIT_ERROR);
146 /*NOTREACHED*/
147}
148
149#if HAVE_LOCALE
150/*
151 * Define a charset, given a locale name.
152 */
153 static void
154ilocale()
155{
156 register int c;
157
158 setlocale(LC_ALL, "");
159 for (c = 0; c < (int) sizeof(chardef); c++)
160 {
161 if (isprint(c))
162 chardef[c] = 0;
163 else if (iscntrl(c))
164 chardef[c] = IS_CONTROL_CHAR;
165 else
166 chardef[c] = IS_BINARY_CHAR|IS_CONTROL_CHAR;
167 }
168}
169#endif
170
171/*
172 * Define the printing format for control chars.
173 */
174 public void
175setbinfmt(s)
176 char *s;
177{
178 if (s == NULL || *s == '\0')
179 s = "*s<%X>";
180 /*
181 * Select the attributes if it starts with "*".
182 */
183 if (*s == '*')
184 {
185 switch (s[1])
186 {
187 case 'd': binattr = AT_BOLD; break;
188 case 'k': binattr = AT_BLINK; break;
189 case 's': binattr = AT_STANDOUT; break;
190 case 'u': binattr = AT_UNDERLINE; break;
191 default: binattr = AT_NORMAL; break;
192 }
193 s += 2;
194 }
195 binfmt = s;
196}
197
198/*
199 * Initialize charset data structures.
200 */
201 public void
202init_charset()
203{
204 register char *s;
205
206 s = lgetenv("LESSBINFMT");
207 setbinfmt(s);
208
209 /*
210 * See if environment variable LESSCHARSET is defined.
211 */
212 s = lgetenv("LESSCHARSET");
213 if (icharset(s))
214 return;
215 /*
216 * LESSCHARSET is not defined: try LESSCHARDEF.
217 */
218 s = lgetenv("LESSCHARDEF");
219 if (s != NULL && *s != '\0')
220 {
221 ichardef(s);
222 return;
223 }
224
225#if HAVE_STRSTR
226 /*
227 * Check whether LC_ALL, LC_CTYPE or LANG look like UTF-8 is used.
228 */
229 if ((s = lgetenv("LC_ALL")) != NULL ||
230 (s = lgetenv("LC_CTYPE")) != NULL ||
231 (s = lgetenv("LANG")) != NULL)
232 {
233 if (strstr(s, "UTF-8") != NULL || strstr(s, "utf-8") != NULL)
234 if (icharset("utf-8"))
235 return;
236 }
237#endif
238
239#if HAVE_LOCALE
240 /*
241 * Use setlocale.
242 */
243 ilocale();
244#else
245 /*
246 * Default to "latin1".
247 */
248 (void) icharset("latin1");
249#endif
250}
251
252/*
253 * Is a given character a "binary" character?
254 */
255 public int
256binary_char(c)
257 unsigned char c;
258{
259 c &= 0377;
260 return (chardef[c] & IS_BINARY_CHAR);
261}
262
263/*
264 * Is a given character a "control" character?
265 */
266 public int
267control_char(c)
268 int c;
269{
270 c &= 0377;
271 return (chardef[c] & IS_CONTROL_CHAR);
272}
273
274/*
275 * Return the printable form of a character.
276 * For example, in the "ascii" charset '\3' is printed as "^C".
277 */
278 public char *
279prchar(c)
280 int c;
281{
282 static char buf[8];
283
284 c &= 0377;
285 if (!control_char(c))
286 sprintf(buf, "%c", c);
287 else if (c == ESC)
288 sprintf(buf, "ESC");
289 else if (c < 128 && !control_char(c ^ 0100))
290 sprintf(buf, "^%c", c ^ 0100);
291 else
292 sprintf(buf, binfmt, c);
293 return (buf);
294}