1 /* Multibyte character I/O: macros for multi-byte encodings.
2 Copyright (C) 2001, 2005, 2009-2012 Free Software Foundation, Inc.
4 This program is free software: you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation; either version 3 of the License, or
7 (at your option) any later version.
9 This program 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 General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program. If not, see <http://www.gnu.org/licenses/>. */
17 /* Written by Mitsuru Chinen <mchinen@yamato.ibm.com>
18 and Bruno Haible <bruno@clisp.org>. */
20 /* The macros in this file implement multi-byte character input from a
24 is the type for multibyte character input stream, usable for variable
28 is the type for multibyte character or EOF, usable for variable
31 mbf_init (mbf, stream)
32 initializes the MB_FILE for reading from stream.
35 reads the next multibyte character from mbf and stores it in mbc.
38 returns true if mbc represents the EOF value.
40 Here are the function prototypes of the macros.
42 extern void mbf_init (mb_file_t mbf, FILE *stream);
43 extern void mbf_getc (mbf_char_t mbc, mb_file_t mbf);
44 extern bool mb_iseof (const mbf_char_t mbc);
55 /* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before
57 BSD/OS 4.1 has a bug: <stdio.h> and <time.h> must be included before
65 _GL_INLINE_HEADER_BEGIN
67 # define MBFILE_INLINE _GL_INLINE
75 unsigned int bufcount;
76 char buf[MBCHAR_BUF_SIZE];
77 struct mbchar pushback;
81 mbfile_multi_getc (struct mbchar *mbc, struct mbfile_multi *mbf)
85 /* If EOF has already been seen, don't use getc. This matters if
86 mbf->fp is connected to an interactive tty. */
90 /* Return character pushed back, if there is one. */
91 if (mbf->have_pushback)
93 mb_copy (mbc, &mbf->pushback);
94 mbf->have_pushback = false;
98 /* Before using mbrtowc, we need at least one byte. */
99 if (mbf->bufcount == 0)
101 int c = getc (mbf->fp);
104 mbf->eof_seen = true;
107 mbf->buf[0] = (unsigned char) c;
111 /* Handle most ASCII characters quickly, without calling mbrtowc(). */
112 if (mbf->bufcount == 1 && mbsinit (&mbf->state) && is_basic (mbf->buf[0]))
114 /* These characters are part of the basic character set. ISO C 99
115 guarantees that their wide character code is identical to their
117 mbc->wc = mbc->buf[0] = mbf->buf[0];
118 mbc->wc_valid = true;
119 mbc->ptr = &mbc->buf[0];
125 /* Use mbrtowc on an increasing number of bytes. Read only as many bytes
126 from mbf->fp as needed. This is needed to give reasonable interactive
127 behaviour when mbf->fp is connected to an interactive tty. */
130 /* We don't know whether the 'mbrtowc' function updates the state when
131 it returns -2, - this is the ISO C 99 and glibc-2.2 behaviour - or
132 not - amended ANSI C, glibc-2.1 and Solaris 2.7 behaviour. We
133 don't have an autoconf test for this, yet.
134 The new behaviour would allow us to feed the bytes one by one into
135 mbrtowc. But the old behaviour forces us to feed all bytes since
136 the end of the last character into mbrtowc. Since we want to retry
137 with more bytes when mbrtowc returns -2, we must backup the state
138 before calling mbrtowc, because implementations with the new
139 behaviour will clobber it. */
140 mbstate_t backup_state = mbf->state;
142 bytes = mbrtowc (&mbc->wc, &mbf->buf[0], mbf->bufcount, &mbf->state);
144 if (bytes == (size_t) -1)
146 /* An invalid multibyte sequence was encountered. */
147 /* Return a single byte. */
149 mbc->wc_valid = false;
152 else if (bytes == (size_t) -2)
154 /* An incomplete multibyte character. */
155 mbf->state = backup_state;
156 if (mbf->bufcount == MBCHAR_BUF_SIZE)
158 /* An overlong incomplete multibyte sequence was encountered. */
159 /* Return a single byte. */
161 mbc->wc_valid = false;
166 /* Read one more byte and retry mbrtowc. */
167 int c = getc (mbf->fp);
170 /* An incomplete multibyte character at the end. */
171 mbf->eof_seen = true;
172 bytes = mbf->bufcount;
173 mbc->wc_valid = false;
176 mbf->buf[mbf->bufcount] = (unsigned char) c;
184 /* A null wide character was encountered. */
186 assert (mbf->buf[0] == '\0');
187 assert (mbc->wc == 0);
189 mbc->wc_valid = true;
194 /* Return the multibyte sequence mbf->buf[0..bytes-1]. */
195 mbc->ptr = &mbc->buf[0];
196 memcpy (&mbc->buf[0], &mbf->buf[0], bytes);
199 mbf->bufcount -= bytes;
200 if (mbf->bufcount > 0)
202 /* It's not worth calling memmove() for so few bytes. */
203 unsigned int count = mbf->bufcount;
204 char *p = &mbf->buf[0];
216 /* An mbchar_t with bytes == 0 is used to indicate EOF. */
219 mbc->wc_valid = false;
224 mbfile_multi_ungetc (const struct mbchar *mbc, struct mbfile_multi *mbf)
226 mb_copy (&mbf->pushback, mbc);
227 mbf->have_pushback = true;
230 typedef struct mbfile_multi mb_file_t;
232 typedef mbchar_t mbf_char_t;
234 #define mbf_init(mbf, stream) \
235 ((mbf).fp = (stream), \
236 (mbf).eof_seen = false, \
237 (mbf).have_pushback = false, \
238 memset (&(mbf).state, '\0', sizeof (mbstate_t)), \
241 #define mbf_getc(mbc, mbf) mbfile_multi_getc (&(mbc), &(mbf))
243 #define mbf_ungetc(mbc, mbf) mbfile_multi_ungetc (&(mbc), &(mbf))
245 #define mb_iseof(mbc) ((mbc).bytes == 0)
247 _GL_INLINE_HEADER_BEGIN
249 #endif /* _MBFILE_H */