1 /* Multibyte character I/O: macros for multi-byte encodings.
2 Copyright (C) 2001, 2005 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 2, or (at your option)
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, write to the Free Software Foundation,
16 Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
18 /* Written by Mitsuru Chinen <mchinen@yamato.ibm.com>
19 and Bruno Haible <bruno@clisp.org>. */
21 /* The macros in this file implement multi-byte character input from a
25 is the type for multibyte character input stream, usable for variable
29 is the type for multibyte character or EOF, usable for variable
32 mbf_init (mbf, stream)
33 initializes the MB_FILE for reading from stream.
36 reads the next multibyte character from mbf and stores it in mbc.
39 returns true if mbc represents the EOF value.
41 Here are the function prototypes of the macros.
43 extern void mbf_init (mb_file_t mbf, FILE *stream);
44 extern void mbf_getc (mbf_char_t mbc, mb_file_t mbf);
45 extern bool mb_iseof (const mbf_char_t mbc);
56 /* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before
58 BSD/OS 4.1 has a bug: <stdio.h> and <time.h> must be included before
71 unsigned int bufcount;
72 char buf[MBCHAR_BUF_SIZE];
73 struct mbchar pushback;
77 mbfile_multi_getc (struct mbchar *mbc, struct mbfile_multi *mbf)
81 /* If EOF has already been seen, don't use getc. This matters if
82 mbf->fp is connected to an interactive tty. */
86 /* Return character pushed back, if there is one. */
87 if (mbf->have_pushback)
89 mb_copy (mbc, &mbf->pushback);
90 mbf->have_pushback = false;
94 /* Before using mbrtowc, we need at least one byte. */
95 if (mbf->bufcount == 0)
97 int c = getc (mbf->fp);
100 mbf->eof_seen = true;
103 mbf->buf[0] = (unsigned char) c;
107 /* Handle most ASCII characters quickly, without calling mbrtowc(). */
108 if (mbf->bufcount == 1 && mbsinit (&mbf->state) && is_basic (mbf->buf[0]))
110 /* These characters are part of the basic character set. ISO C 99
111 guarantees that their wide character code is identical to their
113 mbc->wc = mbc->buf[0] = mbf->buf[0];
114 mbc->wc_valid = true;
115 mbc->ptr = &mbc->buf[0];
121 /* Use mbrtowc on an increasing number of bytes. Read only as many bytes
122 from mbf->fp as needed. This is needed to give reasonable interactive
123 behaviour when mbf->fp is connected to an interactive tty. */
126 /* We don't know whether the 'mbrtowc' function updates the state when
127 it returns -2, - this is the ISO C 99 and glibc-2.2 behaviour - or
128 not - amended ANSI C, glibc-2.1 and Solaris 2.7 behaviour. We
129 don't have an autoconf test for this, yet.
130 The new behaviour would allow us to feed the bytes one by one into
131 mbrtowc. But the old behaviour forces us to feed all bytes since
132 the end of the last character into mbrtowc. Since we want to retry
133 with more bytes when mbrtowc returns -2, we must backup the state
134 before calling mbrtowc, because implementations with the new
135 behaviour will clobber it. */
136 mbstate_t backup_state = mbf->state;
138 bytes = mbrtowc (&mbc->wc, &mbf->buf[0], mbf->bufcount, &mbf->state);
140 if (bytes == (size_t) -1)
142 /* An invalid multibyte sequence was encountered. */
143 /* Return a single byte. */
145 mbc->wc_valid = false;
148 else if (bytes == (size_t) -2)
150 /* An incomplete multibyte character. */
151 mbf->state = backup_state;
152 if (mbf->bufcount == MBCHAR_BUF_SIZE)
154 /* An overlong incomplete multibyte sequence was encountered. */
155 /* Return a single byte. */
157 mbc->wc_valid = false;
162 /* Read one more byte and retry mbrtowc. */
163 int c = getc (mbf->fp);
166 /* An incomplete multibyte character at the end. */
167 mbf->eof_seen = true;
168 bytes = mbf->bufcount;
169 mbc->wc_valid = false;
172 mbf->buf[mbf->bufcount] = (unsigned char) c;
180 /* A null wide character was encountered. */
182 assert (mbf->buf[0] == '\0');
183 assert (mbc->wc == 0);
185 mbc->wc_valid = true;
190 /* Return the multibyte sequence mbf->buf[0..bytes-1]. */
191 mbc->ptr = &mbc->buf[0];
192 memcpy (&mbc->buf[0], &mbf->buf[0], bytes);
195 mbf->bufcount -= bytes;
196 if (mbf->bufcount > 0)
198 /* It's not worth calling memmove() for so few bytes. */
199 unsigned int count = mbf->bufcount;
200 char *p = &mbf->buf[0];
212 /* An mbchar_t with bytes == 0 is used to indicate EOF. */
215 mbc->wc_valid = false;
220 mbfile_multi_ungetc (const struct mbchar *mbc, struct mbfile_multi *mbf)
222 mb_copy (&mbf->pushback, mbc);
223 mbf->have_pushback = true;
226 typedef struct mbfile_multi mb_file_t;
228 typedef mbchar_t mbf_char_t;
230 #define mbf_init(mbf, stream) \
231 ((mbf).fp = (stream), \
232 (mbf).eof_seen = false, \
233 (mbf).have_pushback = false, \
234 memset (&(mbf).state, '\0', sizeof (mbstate_t)), \
237 #define mbf_getc(mbc, mbf) mbfile_multi_getc (&(mbc), &(mbf))
239 #define mbf_ungetc(mbc, mbf) mbfile_multi_ungetc (&(mbc), &(mbf))
241 #define mb_iseof(mbc) ((mbc).bytes == 0)
243 #endif /* _MBFILE_H */