1 /* Multibyte character data type.
2 Copyright (C) 2001, 2005-2007, 2009-2023 Free Software Foundation, Inc.
3
4 This file is free software: you can redistribute it and/or modify
5 it under the terms of the GNU Lesser General Public License as
6 published by the Free Software Foundation; either version 2.1 of the
7 License, or (at your option) any later version.
8
9 This file is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU Lesser General Public License for more details.
13
14 You should have received a copy of the GNU Lesser General Public License
15 along with this program. If not, see <https://www.gnu.org/licenses/>. */
16
17 /* Written by Bruno Haible <bruno@clisp.org>. */
18
19 /* A multibyte character is a short subsequence of a char* string,
20 representing a single 32-bit wide character.
21
22 We use multibyte characters instead of 32-bit wide characters because
23 of the following goals:
24 1) correct multibyte handling, i.e. operate according to the LC_CTYPE
25 locale,
26 2) ease of maintenance, i.e. the maintainer needs not know all details
27 of the ISO C 99 standard,
28 3) don't fail grossly if the input is not in the encoding set by the
29 locale, because often different encodings are in use in the same
30 countries (ISO-8859-1/UTF-8, EUC-JP/Shift_JIS, ...),
31 4) fast in the case of ASCII characters.
32
33 Multibyte characters are only accessed through the mb* macros.
34
35 mb_ptr (mbc)
36 return a pointer to the beginning of the multibyte sequence.
37
38 mb_len (mbc)
39 returns the number of bytes occupied by the multibyte sequence.
40 Always > 0.
41
42 mb_iseq (mbc, sc)
43 returns true if mbc is the standard ASCII character sc.
44
45 mb_isnul (mbc)
46 returns true if mbc is the nul character.
47
48 mb_cmp (mbc1, mbc2)
49 returns a positive, zero, or negative value depending on whether mbc1
50 sorts after, same or before mbc2.
51
52 mb_casecmp (mbc1, mbc2)
53 returns a positive, zero, or negative value depending on whether mbc1
54 sorts after, same or before mbc2, modulo upper/lowercase conversion.
55
56 mb_equal (mbc1, mbc2)
57 returns true if mbc1 and mbc2 are equal.
58
59 mb_caseequal (mbc1, mbc2)
60 returns true if mbc1 and mbc2 are equal modulo upper/lowercase conversion.
61
62 mb_isalnum (mbc)
63 returns true if mbc is alphanumeric.
64
65 mb_isalpha (mbc)
66 returns true if mbc is alphabetic.
67
68 mb_isascii(mbc)
69 returns true if mbc is plain ASCII.
70
71 mb_isblank (mbc)
72 returns true if mbc is a blank.
73
74 mb_iscntrl (mbc)
75 returns true if mbc is a control character.
76
77 mb_isdigit (mbc)
78 returns true if mbc is a decimal digit.
79
80 mb_isgraph (mbc)
81 returns true if mbc is a graphic character.
82
83 mb_islower (mbc)
84 returns true if mbc is lowercase.
85
86 mb_isprint (mbc)
87 returns true if mbc is a printable character.
88
89 mb_ispunct (mbc)
90 returns true if mbc is a punctuation character.
91
92 mb_isspace (mbc)
93 returns true if mbc is a space character.
94
95 mb_isupper (mbc)
96 returns true if mbc is uppercase.
97
98 mb_isxdigit (mbc)
99 returns true if mbc is a hexadecimal digit.
100
101 mb_width (mbc)
102 returns the number of columns on the output device occupied by mbc.
103 Always >= 0.
104
105 mb_putc (mbc, stream)
106 outputs mbc on stream, a byte oriented FILE stream opened for output.
107
108 mb_setascii (&mbc, sc)
109 assigns the standard ASCII character sc to mbc.
110 (Only available if the 'mbfile' module is in use.)
111
112 mb_copy (&destmbc, &srcmbc)
113 copies srcmbc to destmbc.
114
115 Here are the function prototypes of the macros.
116
117 extern const char * mb_ptr (const mbchar_t mbc);
118 extern size_t mb_len (const mbchar_t mbc);
119 extern bool mb_iseq (const mbchar_t mbc, char sc);
120 extern bool mb_isnul (const mbchar_t mbc);
121 extern int mb_cmp (const mbchar_t mbc1, const mbchar_t mbc2);
122 extern int mb_casecmp (const mbchar_t mbc1, const mbchar_t mbc2);
123 extern bool mb_equal (const mbchar_t mbc1, const mbchar_t mbc2);
124 extern bool mb_caseequal (const mbchar_t mbc1, const mbchar_t mbc2);
125 extern bool mb_isalnum (const mbchar_t mbc);
126 extern bool mb_isalpha (const mbchar_t mbc);
127 extern bool mb_isascii (const mbchar_t mbc);
128 extern bool mb_isblank (const mbchar_t mbc);
129 extern bool mb_iscntrl (const mbchar_t mbc);
130 extern bool mb_isdigit (const mbchar_t mbc);
131 extern bool mb_isgraph (const mbchar_t mbc);
132 extern bool mb_islower (const mbchar_t mbc);
133 extern bool mb_isprint (const mbchar_t mbc);
134 extern bool mb_ispunct (const mbchar_t mbc);
135 extern bool mb_isspace (const mbchar_t mbc);
136 extern bool mb_isupper (const mbchar_t mbc);
137 extern bool mb_isxdigit (const mbchar_t mbc);
138 extern int mb_width (const mbchar_t mbc);
139 extern void mb_putc (const mbchar_t mbc, FILE *stream);
140 extern void mb_setascii (mbchar_t *new, char sc);
141 extern void mb_copy (mbchar_t *new, const mbchar_t *old);
142 */
143
144 #ifndef _MBCHAR_H
145 #define _MBCHAR_H 1
146
147 /* This file uses _GL_INLINE_HEADER_BEGIN, _GL_INLINE. */
148 #if !_GL_CONFIG_H_INCLUDED
149 #error "Please include config.h first."
150 #endif
151
152 #include <string.h>
153 #include <uchar.h>
154
155 _GL_INLINE_HEADER_BEGIN
156 #ifndef MBCHAR_INLINE
157 # define MBCHAR_INLINE _GL_INLINE
158 #endif
159
160 /* The longest multibyte characters, nowadays, are 4 bytes long.
161 Regardless of the values of MB_CUR_MAX and MB_LEN_MAX. */
162 #define MBCHAR_BUF_SIZE 4
163
164 struct mbchar
165 {
166 const char *ptr; /* pointer to current character */
167 size_t bytes; /* number of bytes of current character, > 0 */
168 bool wc_valid; /* true if wc is a valid 32-bit wide character */
169 char32_t wc; /* if wc_valid: the current character */
170 #if defined GNULIB_MBFILE
171 char buf[MBCHAR_BUF_SIZE]; /* room for the bytes, used for file input only */
172 #endif
173 };
174
175 /* EOF (not a real character) is represented with bytes = 0 and
176 wc_valid = false. */
177
178 typedef struct mbchar mbchar_t;
179
180 /* Access the current character. */
181 #define mb_ptr(mbc) ((mbc).ptr)
182 #define mb_len(mbc) ((mbc).bytes)
183
184 /* Comparison of characters. */
185 #define mb_iseq(mbc, sc) ((mbc).wc_valid && (mbc).wc == (sc))
186 #define mb_isnul(mbc) ((mbc).wc_valid && (mbc).wc == 0)
187 #define mb_cmp(mbc1, mbc2) \
188 ((mbc1).wc_valid \
189 ? ((mbc2).wc_valid \
190 ? _GL_CMP ((mbc1).wc, (mbc2).wc) \
191 : -1) \
192 : ((mbc2).wc_valid \
193 ? 1 \
194 : (mbc1).bytes == (mbc2).bytes \
195 ? memcmp ((mbc1).ptr, (mbc2).ptr, (mbc1).bytes) \
196 : (mbc1).bytes < (mbc2).bytes \
197 ? (memcmp ((mbc1).ptr, (mbc2).ptr, (mbc1).bytes) > 0 ? 1 : -1) \
198 : (memcmp ((mbc1).ptr, (mbc2).ptr, (mbc2).bytes) >= 0 ? 1 : -1)))
199 #define mb_casecmp(mbc1, mbc2) \
200 ((mbc1).wc_valid \
201 ? ((mbc2).wc_valid \
202 ? _GL_CMP (c32tolower ((mbc1).wc), c32tolower ((mbc2).wc)) \
203 : -1) \
204 : ((mbc2).wc_valid \
205 ? 1 \
206 : (mbc1).bytes == (mbc2).bytes \
207 ? memcmp ((mbc1).ptr, (mbc2).ptr, (mbc1).bytes) \
208 : (mbc1).bytes < (mbc2).bytes \
209 ? (memcmp ((mbc1).ptr, (mbc2).ptr, (mbc1).bytes) > 0 ? 1 : -1) \
210 : (memcmp ((mbc1).ptr, (mbc2).ptr, (mbc2).bytes) >= 0 ? 1 : -1)))
211 #define mb_equal(mbc1, mbc2) \
212 ((mbc1).wc_valid && (mbc2).wc_valid \
213 ? (mbc1).wc == (mbc2).wc \
214 : (mbc1).bytes == (mbc2).bytes \
215 && memcmp ((mbc1).ptr, (mbc2).ptr, (mbc1).bytes) == 0)
216 #define mb_caseequal(mbc1, mbc2) \
217 ((mbc1).wc_valid && (mbc2).wc_valid \
218 ? c32tolower ((mbc1).wc) == c32tolower ((mbc2).wc) \
219 : (mbc1).bytes == (mbc2).bytes \
220 && memcmp ((mbc1).ptr, (mbc2).ptr, (mbc1).bytes) == 0)
221
222 /* <ctype.h>, <wctype.h> classification. */
223 #define mb_isascii(mbc) \
224 ((mbc).wc_valid && (mbc).wc >= 0 && (mbc).wc <= 127)
225 #define mb_isalnum(mbc) ((mbc).wc_valid && c32isalnum ((mbc).wc))
226 #define mb_isalpha(mbc) ((mbc).wc_valid && c32isalpha ((mbc).wc))
227 #define mb_isblank(mbc) ((mbc).wc_valid && c32isblank ((mbc).wc))
228 #define mb_iscntrl(mbc) ((mbc).wc_valid && c32iscntrl ((mbc).wc))
229 #define mb_isdigit(mbc) ((mbc).wc_valid && c32isdigit ((mbc).wc))
230 #define mb_isgraph(mbc) ((mbc).wc_valid && c32isgraph ((mbc).wc))
231 #define mb_islower(mbc) ((mbc).wc_valid && c32islower ((mbc).wc))
232 #define mb_isprint(mbc) ((mbc).wc_valid && c32isprint ((mbc).wc))
233 #define mb_ispunct(mbc) ((mbc).wc_valid && c32ispunct ((mbc).wc))
234 #define mb_isspace(mbc) ((mbc).wc_valid && c32isspace ((mbc).wc))
235 #define mb_isupper(mbc) ((mbc).wc_valid && c32isupper ((mbc).wc))
236 #define mb_isxdigit(mbc) ((mbc).wc_valid && c32isxdigit ((mbc).wc))
237
238 /* Extra <wchar.h> function. */
239
240 /* Unprintable characters appear as a small box of width 1. */
241 #define MB_UNPRINTABLE_WIDTH 1
242
243 MBCHAR_INLINE int
244 mb_width_aux (char32_t wc)
245 {
246 int w = c32width (wc);
247 /* For unprintable characters, arbitrarily return 0 for control characters
248 and MB_UNPRINTABLE_WIDTH otherwise. */
249 return (w >= 0 ? w : c32iscntrl (wc) ? 0 : MB_UNPRINTABLE_WIDTH);
250 }
251
252 #define mb_width(mbc) \
253 ((mbc).wc_valid ? mb_width_aux ((mbc).wc) : MB_UNPRINTABLE_WIDTH)
254
255 /* Output. */
256 #define mb_putc(mbc, stream) fwrite ((mbc).ptr, 1, (mbc).bytes, (stream))
257
258 #if defined GNULIB_MBFILE
259 /* Assignment. */
260 # define mb_setascii(mbc, sc) \
261 ((mbc)->ptr = (mbc)->buf, (mbc)->bytes = 1, (mbc)->wc_valid = 1, \
262 (mbc)->wc = (mbc)->buf[0] = (sc))
263 #endif
264
265 /* Copying a character. */
266 MBCHAR_INLINE void
267 mb_copy (mbchar_t *new_mbc, const mbchar_t *old_mbc)
268 {
269 #if defined GNULIB_MBFILE
270 if (old_mbc->ptr == &old_mbc->buf[0])
271 {
272 memcpy (&new_mbc->buf[0], &old_mbc->buf[0], old_mbc->bytes);
273 new_mbc->ptr = &new_mbc->buf[0];
274 }
275 else
276 #endif
277 new_mbc->ptr = old_mbc->ptr;
278 new_mbc->bytes = old_mbc->bytes;
279 if ((new_mbc->wc_valid = old_mbc->wc_valid))
280 new_mbc->wc = old_mbc->wc;
281 }
282
283
284 /* is_basic(c) tests whether the single-byte character c is
285 - in the ISO C "basic character set" or is one of '@', '$', and '`'
286 which ISO C 23 ยง 5.2.1.1.(1) guarantees to be single-byte and in
287 practice are safe to treat as basic in the execution character set,
288 or
289 - in the POSIX "portable character set", which
290 <https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap06.html>
291 equally guarantees to be single-byte.
292 This is a convenience function, and is in this file only to share code
293 between mbiter.h, mbuiter.h, and mbfile.h. */
294 #if (' ' == 32) && ('!' == 33) && ('"' == 34) && ('#' == 35) \
295 && ('$' == 36) && ('%' == 37) && ('&' == 38) && ('\'' == 39) \
296 && ('(' == 40) && (')' == 41) && ('*' == 42) && ('+' == 43) \
297 && (',' == 44) && ('-' == 45) && ('.' == 46) && ('/' == 47) \
298 && ('0' == 48) && ('1' == 49) && ('2' == 50) && ('3' == 51) \
299 && ('4' == 52) && ('5' == 53) && ('6' == 54) && ('7' == 55) \
300 && ('8' == 56) && ('9' == 57) && (':' == 58) && (';' == 59) \
301 && ('<' == 60) && ('=' == 61) && ('>' == 62) && ('?' == 63) \
302 && ('@' == 64) && ('A' == 65) && ('B' == 66) && ('C' == 67) \
303 && ('D' == 68) && ('E' == 69) && ('F' == 70) && ('G' == 71) \
304 && ('H' == 72) && ('I' == 73) && ('J' == 74) && ('K' == 75) \
305 && ('L' == 76) && ('M' == 77) && ('N' == 78) && ('O' == 79) \
306 && ('P' == 80) && ('Q' == 81) && ('R' == 82) && ('S' == 83) \
307 && ('T' == 84) && ('U' == 85) && ('V' == 86) && ('W' == 87) \
308 && ('X' == 88) && ('Y' == 89) && ('Z' == 90) && ('[' == 91) \
309 && ('\\' == 92) && (']' == 93) && ('^' == 94) && ('_' == 95) \
310 && ('`' == 96) && ('a' == 97) && ('b' == 98) && ('c' == 99) \
311 && ('d' == 100) && ('e' == 101) && ('f' == 102) && ('g' == 103) \
312 && ('h' == 104) && ('i' == 105) && ('j' == 106) && ('k' == 107) \
313 && ('l' == 108) && ('m' == 109) && ('n' == 110) && ('o' == 111) \
314 && ('p' == 112) && ('q' == 113) && ('r' == 114) && ('s' == 115) \
315 && ('t' == 116) && ('u' == 117) && ('v' == 118) && ('w' == 119) \
316 && ('x' == 120) && ('y' == 121) && ('z' == 122) && ('{' == 123) \
317 && ('|' == 124) && ('}' == 125) && ('~' == 126)
318 /* The character set is ISO-646, not EBCDIC. */
319 # define IS_BASIC_ASCII 1
320
321 /* All locale encodings (see localcharset.h) map the characters 0x00..0x7F
322 to U+0000..U+007F, like ASCII, except for
323 CP864 different mapping of '%'
324 SHIFT_JIS different mappings of 0x5C, 0x7E
325 JOHAB different mapping of 0x5C
326 However, these characters in the range 0x20..0x7E are in the ISO C
327 "basic character set" and in the POSIX "portable character set", which
328 ISO C and POSIX guarantee to be single-byte. Thus, locales with these
329 encodings are not POSIX compliant. And they are most likely not in use
330 any more (as of 2023). */
331 # define is_basic(c) ((unsigned char) (c) < 0x80)
332
333 #else
334
335 MBCHAR_INLINE bool
336 is_basic (char c)
337 {
338 switch (c)
339 {
340 case '\0':
341 case '\007': case '\010':
342 case '\t': case '\n': case '\v': case '\f': case '\r':
343 case ' ': case '!': case '"': case '#': case '$': case '%':
344 case '&': case '\'': case '(': case ')': case '*':
345 case '+': case ',': case '-': case '.': case '/':
346 case '0': case '1': case '2': case '3': case '4':
347 case '5': case '6': case '7': case '8': case '9':
348 case ':': case ';': case '<': case '=': case '>':
349 case '?': case '@':
350 case 'A': case 'B': case 'C': case 'D': case 'E':
351 case 'F': case 'G': case 'H': case 'I': case 'J':
352 case 'K': case 'L': case 'M': case 'N': case 'O':
353 case 'P': case 'Q': case 'R': case 'S': case 'T':
354 case 'U': case 'V': case 'W': case 'X': case 'Y':
355 case 'Z':
356 case '[': case '\\': case ']': case '^': case '_': case '`':
357 case 'a': case 'b': case 'c': case 'd': case 'e':
358 case 'f': case 'g': case 'h': case 'i': case 'j':
359 case 'k': case 'l': case 'm': case 'n': case 'o':
360 case 'p': case 'q': case 'r': case 's': case 't':
361 case 'u': case 'v': case 'w': case 'x': case 'y':
362 case 'z': case '{': case '|': case '}': case '~':
363 return 1;
364 default:
365 return 0;
366 }
367 }
368
369 #endif
370
371 _GL_INLINE_HEADER_END
372
373 #endif /* _MBCHAR_H */