1 /* Multibyte character I/O: macros for multi-byte encodings.
2 Copyright (C) 2001, 2005, 2009-2011 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
70 unsigned int bufcount;
71 char buf[MBCHAR_BUF_SIZE];
72 struct mbchar pushback;
76 mbfile_multi_getc (struct mbchar *mbc, struct mbfile_multi *mbf)
80 /* If EOF has already been seen, don't use getc. This matters if
81 mbf->fp is connected to an interactive tty. */
85 /* Return character pushed back, if there is one. */
86 if (mbf->have_pushback)
88 mb_copy (mbc, &mbf->pushback);
89 mbf->have_pushback = false;
93 /* Before using mbrtowc, we need at least one byte. */
94 if (mbf->bufcount == 0)
96 int c = getc (mbf->fp);
102 mbf->buf[0] = (unsigned char) c;
106 /* Handle most ASCII characters quickly, without calling mbrtowc(). */
107 if (mbf->bufcount == 1 && mbsinit (&mbf->state) && is_basic (mbf->buf[0]))
109 /* These characters are part of the basic character set. ISO C 99
110 guarantees that their wide character code is identical to their
112 mbc->wc = mbc->buf[0] = mbf->buf[0];
113 mbc->wc_valid = true;
114 mbc->ptr = &mbc->buf[0];
120 /* Use mbrtowc on an increasing number of bytes. Read only as many bytes
121 from mbf->fp as needed. This is needed to give reasonable interactive
122 behaviour when mbf->fp is connected to an interactive tty. */
125 /* We don't know whether the 'mbrtowc' function updates the state when
126 it returns -2, - this is the ISO C 99 and glibc-2.2 behaviour - or
127 not - amended ANSI C, glibc-2.1 and Solaris 2.7 behaviour. We
128 don't have an autoconf test for this, yet.
129 The new behaviour would allow us to feed the bytes one by one into
130 mbrtowc. But the old behaviour forces us to feed all bytes since
131 the end of the last character into mbrtowc. Since we want to retry
132 with more bytes when mbrtowc returns -2, we must backup the state
133 before calling mbrtowc, because implementations with the new
134 behaviour will clobber it. */
135 mbstate_t backup_state = mbf->state;
137 bytes = mbrtowc (&mbc->wc, &mbf->buf[0], mbf->bufcount, &mbf->state);
139 if (bytes == (size_t) -1)
141 /* An invalid multibyte sequence was encountered. */
142 /* Return a single byte. */
144 mbc->wc_valid = false;
147 else if (bytes == (size_t) -2)
149 /* An incomplete multibyte character. */
150 mbf->state = backup_state;
151 if (mbf->bufcount == MBCHAR_BUF_SIZE)
153 /* An overlong incomplete multibyte sequence was encountered. */
154 /* Return a single byte. */
156 mbc->wc_valid = false;
161 /* Read one more byte and retry mbrtowc. */
162 int c = getc (mbf->fp);
165 /* An incomplete multibyte character at the end. */
166 mbf->eof_seen = true;
167 bytes = mbf->bufcount;
168 mbc->wc_valid = false;
171 mbf->buf[mbf->bufcount] = (unsigned char) c;
179 /* A null wide character was encountered. */
181 assert (mbf->buf[0] == '\0');
182 assert (mbc->wc == 0);
184 mbc->wc_valid = true;
189 /* Return the multibyte sequence mbf->buf[0..bytes-1]. */
190 mbc->ptr = &mbc->buf[0];
191 memcpy (&mbc->buf[0], &mbf->buf[0], bytes);
194 mbf->bufcount -= bytes;
195 if (mbf->bufcount > 0)
197 /* It's not worth calling memmove() for so few bytes. */
198 unsigned int count = mbf->bufcount;
199 char *p = &mbf->buf[0];
211 /* An mbchar_t with bytes == 0 is used to indicate EOF. */
214 mbc->wc_valid = false;
219 mbfile_multi_ungetc (const struct mbchar *mbc, struct mbfile_multi *mbf)
221 mb_copy (&mbf->pushback, mbc);
222 mbf->have_pushback = true;
225 typedef struct mbfile_multi mb_file_t;
227 typedef mbchar_t mbf_char_t;
229 #define mbf_init(mbf, stream) \
230 ((mbf).fp = (stream), \
231 (mbf).eof_seen = false, \
232 (mbf).have_pushback = false, \
233 memset (&(mbf).state, '\0', sizeof (mbstate_t)), \
236 #define mbf_getc(mbc, mbf) mbfile_multi_getc (&(mbc), &(mbf))
238 #define mbf_ungetc(mbc, mbf) mbfile_multi_ungetc (&(mbc), &(mbf))
240 #define mb_iseof(mbc) ((mbc).bytes == 0)
242 #endif /* _MBFILE_H */