annotate src/charset.c @ 85265:4fe6aac1d449

(ebrowse-tree-mode): Disable undo in the BROWSE buffer.
author Eli Zaretskii <eliz@gnu.org>
date Sat, 13 Oct 2007 12:55:00 +0000
parents 1677cf1c2509
children fc2bcd2a8aad
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
18341
33e78cc7f058 Change copyright notices.
Richard M. Stallman <rms@gnu.org>
parents: 18003
diff changeset
1 /* Basic multilingual character support.
68651
3bd95f4f2941 Update years in copyright notice; nfc.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents: 67841
diff changeset
2 Copyright (C) 2001, 2002, 2003, 2004, 2005,
75348
3d45362f1d38 Add 2007 to copyright years.
Glenn Morris <rgm@gnu.org>
parents: 74605
diff changeset
3 2006, 2007 Free Software Foundation, Inc.
74605
6ee41fdd69ff Update AIST copyright years.
Kenichi Handa <handa@m17n.org>
parents: 68651
diff changeset
4 Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
75348
3d45362f1d38 Add 2007 to copyright years.
Glenn Morris <rgm@gnu.org>
parents: 74605
diff changeset
5 2005, 2006, 2007
67658
547f2b420666 Copyright fixed
Kenichi Handa <handa@m17n.org>
parents: 67039
diff changeset
6 National Institute of Advanced Industrial Science and Technology (AIST)
547f2b420666 Copyright fixed
Kenichi Handa <handa@m17n.org>
parents: 67039
diff changeset
7 Registration Number H14PRO021
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
8
17071
70194012fb3a Fix FSF address in comment.
Karl Heuer <kwzh@gnu.org>
parents: 17052
diff changeset
9 This file is part of GNU Emacs.
70194012fb3a Fix FSF address in comment.
Karl Heuer <kwzh@gnu.org>
parents: 17052
diff changeset
10
70194012fb3a Fix FSF address in comment.
Karl Heuer <kwzh@gnu.org>
parents: 17052
diff changeset
11 GNU Emacs is free software; you can redistribute it and/or modify
70194012fb3a Fix FSF address in comment.
Karl Heuer <kwzh@gnu.org>
parents: 17052
diff changeset
12 it under the terms of the GNU General Public License as published by
78313
2b9404c2f01f Remove license from trivial file.
Glenn Morris <rgm@gnu.org>
parents: 75348
diff changeset
13 the Free Software Foundation; either version 3, or (at your option)
17071
70194012fb3a Fix FSF address in comment.
Karl Heuer <kwzh@gnu.org>
parents: 17052
diff changeset
14 any later version.
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
15
17071
70194012fb3a Fix FSF address in comment.
Karl Heuer <kwzh@gnu.org>
parents: 17052
diff changeset
16 GNU Emacs is distributed in the hope that it will be useful,
70194012fb3a Fix FSF address in comment.
Karl Heuer <kwzh@gnu.org>
parents: 17052
diff changeset
17 but WITHOUT ANY WARRANTY; without even the implied warranty of
70194012fb3a Fix FSF address in comment.
Karl Heuer <kwzh@gnu.org>
parents: 17052
diff changeset
18 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
70194012fb3a Fix FSF address in comment.
Karl Heuer <kwzh@gnu.org>
parents: 17052
diff changeset
19 GNU General Public License for more details.
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
20
17071
70194012fb3a Fix FSF address in comment.
Karl Heuer <kwzh@gnu.org>
parents: 17052
diff changeset
21 You should have received a copy of the GNU General Public License
70194012fb3a Fix FSF address in comment.
Karl Heuer <kwzh@gnu.org>
parents: 17052
diff changeset
22 along with GNU Emacs; see the file COPYING. If not, write to
64084
a8fa7c632ee4 Update FSF's address.
Lute Kamstra <lute@gnu.org>
parents: 62010
diff changeset
23 the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
a8fa7c632ee4 Update FSF's address.
Lute Kamstra <lute@gnu.org>
parents: 62010
diff changeset
24 Boston, MA 02110-1301, USA. */
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
25
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
26 /* At first, see the document in `charset.h' to understand the code in
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
27 this file. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
28
26088
b7aa6ac26872 Add support for large files, 64-bit Solaris, system locale codings.
Paul Eggert <eggert@twinsun.com>
parents: 25611
diff changeset
29 #ifdef emacs
b7aa6ac26872 Add support for large files, 64-bit Solaris, system locale codings.
Paul Eggert <eggert@twinsun.com>
parents: 25611
diff changeset
30 #include <config.h>
b7aa6ac26872 Add support for large files, 64-bit Solaris, system locale codings.
Paul Eggert <eggert@twinsun.com>
parents: 25611
diff changeset
31 #endif
b7aa6ac26872 Add support for large files, 64-bit Solaris, system locale codings.
Paul Eggert <eggert@twinsun.com>
parents: 25611
diff changeset
32
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
33 #include <stdio.h>
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
34
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
35 #ifdef emacs
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
36
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
37 #include <sys/types.h>
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
38 #include "lisp.h"
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
39 #include "buffer.h"
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
40 #include "charset.h"
31420
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
41 #include "composite.h"
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
42 #include "coding.h"
17843
f03ff9d069c6 Include disptab.h.
Kenichi Handa <handa@m17n.org>
parents: 17834
diff changeset
43 #include "disptab.h"
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
44
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
45 #else /* not emacs */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
46
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
47 #include "mulelib.h"
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
48
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
49 #endif /* emacs */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
50
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
51 Lisp_Object Qcharset, Qascii, Qeight_bit_control, Qeight_bit_graphic;
23883
0ca2fa58ca7b (Qunknown): New variable.
Kenichi Handa <handa@m17n.org>
parents: 23767
diff changeset
52 Lisp_Object Qunknown;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
53
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
54 /* Declaration of special leading-codes. */
44322
fca902769bf4 (Fstring): Allow 0 arguments.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 43713
diff changeset
55 EMACS_INT leading_code_private_11; /* for private DIMENSION1 of 1-column */
fca902769bf4 (Fstring): Allow 0 arguments.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 43713
diff changeset
56 EMACS_INT leading_code_private_12; /* for private DIMENSION1 of 2-column */
fca902769bf4 (Fstring): Allow 0 arguments.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 43713
diff changeset
57 EMACS_INT leading_code_private_21; /* for private DIMENSION2 of 1-column */
fca902769bf4 (Fstring): Allow 0 arguments.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 43713
diff changeset
58 EMACS_INT leading_code_private_22; /* for private DIMENSION2 of 2-column */
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
59
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
60 /* Declaration of special charsets. The values are set by
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
61 Fsetup_special_charsets. */
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
62 int charset_latin_iso8859_1; /* ISO8859-1 (Latin-1) */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
63 int charset_jisx0208_1978; /* JISX0208.1978 (Japanese Kanji old set) */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
64 int charset_jisx0208; /* JISX0208.1983 (Japanese Kanji) */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
65 int charset_katakana_jisx0201; /* JISX0201.Kana (Japanese Katakana) */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
66 int charset_latin_jisx0201; /* JISX0201.Roman (Japanese Roman) */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
67 int charset_big5_1; /* Big5 Level 1 (Chinese Traditional) */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
68 int charset_big5_2; /* Big5 Level 2 (Chinese Traditional) */
66375
6680788d2e7f (charset_mule_unicode_0100_24ff)
Kenichi Handa <handa@m17n.org>
parents: 64770
diff changeset
69 int charset_mule_unicode_0100_24ff;
6680788d2e7f (charset_mule_unicode_0100_24ff)
Kenichi Handa <handa@m17n.org>
parents: 64770
diff changeset
70 int charset_mule_unicode_2500_33ff;
6680788d2e7f (charset_mule_unicode_0100_24ff)
Kenichi Handa <handa@m17n.org>
parents: 64770
diff changeset
71 int charset_mule_unicode_e000_ffff;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
72
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
73 Lisp_Object Qcharset_table;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
74
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
75 /* A char-table containing information of each character set. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
76 Lisp_Object Vcharset_table;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
77
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
78 /* A vector of charset symbol indexed by charset-id. This is used
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
79 only for returning charset symbol from C functions. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
80 Lisp_Object Vcharset_symbol_table;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
81
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
82 /* A list of charset symbols ever defined. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
83 Lisp_Object Vcharset_list;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
84
22185
80a2aa51a6e1 Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents: 22168
diff changeset
85 /* Vector of translation table ever defined.
80a2aa51a6e1 Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents: 22168
diff changeset
86 ID of a translation table is used to index this vector. */
80a2aa51a6e1 Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents: 22168
diff changeset
87 Lisp_Object Vtranslation_table_vector;
20720
f084bb8d535f (min_composite_char): New variable.
Kenichi Handa <handa@m17n.org>
parents: 20705
diff changeset
88
23489
09201cebfa21 (Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents: 23427
diff changeset
89 /* A char-table for characters which may invoke auto-filling. */
09201cebfa21 (Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents: 23427
diff changeset
90 Lisp_Object Vauto_fill_chars;
09201cebfa21 (Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents: 23427
diff changeset
91
09201cebfa21 (Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents: 23427
diff changeset
92 Lisp_Object Qauto_fill_chars;
09201cebfa21 (Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents: 23427
diff changeset
93
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
94 /* Tables used by macros BYTES_BY_CHAR_HEAD and WIDTH_BY_CHAR_HEAD. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
95 int bytes_by_char_head[256];
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
96 int width_by_char_head[256];
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
97
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
98 /* Mapping table from ISO2022's charset (specified by DIMENSION,
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
99 CHARS, and FINAL-CHAR) to Emacs' charset. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
100 int iso_charset_table[2][2][128];
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
101
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
102 /* Variables used locally in the macro FETCH_MULTIBYTE_CHAR. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
103 unsigned char *_fetch_multibyte_char_p;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
104 int _fetch_multibyte_char_len;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
105
20705
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
106 /* Offset to add to a non-ASCII value when inserting it. */
43713
f92c4d87863a Change defvar_int def and vars to use EMACS_INT instead of just int.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 43144
diff changeset
107 EMACS_INT nonascii_insert_offset;
20705
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
108
20890
c7cfd531cf2b (Vnonascii_translate_table): New variable.
Richard M. Stallman <rms@gnu.org>
parents: 20833
diff changeset
109 /* Translation table for converting non-ASCII unibyte characters
c7cfd531cf2b (Vnonascii_translate_table): New variable.
Richard M. Stallman <rms@gnu.org>
parents: 20833
diff changeset
110 to multibyte codes, or nil. */
22121
6e79a15594ab Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents: 21718
diff changeset
111 Lisp_Object Vnonascii_translation_table;
20890
c7cfd531cf2b (Vnonascii_translate_table): New variable.
Richard M. Stallman <rms@gnu.org>
parents: 20833
diff changeset
112
22703
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
113 /* List of all possible generic characters. */
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
114 Lisp_Object Vgeneric_character_list;
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
115
19384
6ce37633c7f9 Many doc fixes.
Richard M. Stallman <rms@gnu.org>
parents: 19025
diff changeset
116
20933
5ceea9d50194 (invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents: 20890
diff changeset
117 void
5ceea9d50194 (invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents: 20890
diff changeset
118 invalid_character (c)
5ceea9d50194 (invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents: 20890
diff changeset
119 int c;
5ceea9d50194 (invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents: 20890
diff changeset
120 {
67039
2022f6e307e1 (invalid_character): Use Lisp-readable syntax
Juri Linkov <juri@jurta.org>
parents: 66375
diff changeset
121 error ("Invalid character: %d, #o%o, #x%x", c, c, c);
20933
5ceea9d50194 (invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents: 20890
diff changeset
122 }
5ceea9d50194 (invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents: 20890
diff changeset
123
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
124 /* Parse string STR of length LENGTH and fetch information of a
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
125 character at STR. Set BYTES to the byte length the character
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
126 occupies, CHARSET, C1, C2 to proper values of the character. */
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
127
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
128 #define SPLIT_MULTIBYTE_SEQ(str, length, bytes, charset, c1, c2) \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
129 do { \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
130 (c1) = *(str); \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
131 (bytes) = BYTES_BY_CHAR_HEAD (c1); \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
132 if ((bytes) == 1) \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
133 (charset) = ASCII_BYTE_P (c1) ? CHARSET_ASCII : CHARSET_8_BIT_GRAPHIC; \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
134 else if ((bytes) == 2) \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
135 { \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
136 if ((c1) == LEADING_CODE_8_BIT_CONTROL) \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
137 (charset) = CHARSET_8_BIT_CONTROL, (c1) = (str)[1] - 0x20; \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
138 else \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
139 (charset) = (c1), (c1) = (str)[1] & 0x7F; \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
140 } \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
141 else if ((bytes) == 3) \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
142 { \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
143 if ((c1) < LEADING_CODE_PRIVATE_11) \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
144 (charset) = (c1), (c1) = (str)[1] & 0x7F, (c2) = (str)[2] & 0x7F; \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
145 else \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
146 (charset) = (str)[1], (c1) = (str)[2] & 0x7F; \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
147 } \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
148 else \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
149 (charset) = (str)[1], (c1) = (str)[2] & 0x7F, (c2) = (str)[3] & 0x7F; \
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
150 } while (0)
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
151
44878
c62f10c00a76 (CHAR_COMPONENTS_VALID_P): Tell in a comment that some invalid pairs are
Eli Zaretskii <eliz@gnu.org>
parents: 44322
diff changeset
152 /* 1 if CHARSET, C1, and C2 compose a valid character, else 0.
c62f10c00a76 (CHAR_COMPONENTS_VALID_P): Tell in a comment that some invalid pairs are
Eli Zaretskii <eliz@gnu.org>
parents: 44322
diff changeset
153 Note that this intentionally allows invalid components, such
c62f10c00a76 (CHAR_COMPONENTS_VALID_P): Tell in a comment that some invalid pairs are
Eli Zaretskii <eliz@gnu.org>
parents: 44322
diff changeset
154 as 0xA0 0xA0, because there exist many files that contain
c62f10c00a76 (CHAR_COMPONENTS_VALID_P): Tell in a comment that some invalid pairs are
Eli Zaretskii <eliz@gnu.org>
parents: 44322
diff changeset
155 such invalid byte sequences, especially in EUC-GB. */
25584
01026ea61dcc Lots of comments fixed.
Kenichi Handa <handa@m17n.org>
parents: 25506
diff changeset
156 #define CHAR_COMPONENTS_VALID_P(charset, c1, c2) \
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
157 ((charset) == CHARSET_ASCII \
25611
9c9d13747ffb (CHAR_COMPONENTS_VALID_P): Fix for ASCII.
Kenichi Handa <handa@m17n.org>
parents: 25610
diff changeset
158 ? ((c1) >= 0 && (c1) <= 0x7F) \
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
159 : ((charset) == CHARSET_8_BIT_CONTROL \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
160 ? ((c1) >= 0x80 && (c1) <= 0x9F) \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
161 : ((charset) == CHARSET_8_BIT_GRAPHIC \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
162 ? ((c1) >= 0x80 && (c1) <= 0xFF) \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
163 : (CHARSET_DIMENSION (charset) == 1 \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
164 ? ((c1) >= 0x20 && (c1) <= 0x7F) \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
165 : ((c1) >= 0x20 && (c1) <= 0x7F \
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
166 && (c2) >= 0x20 && (c2) <= 0x7F)))))
20933
5ceea9d50194 (invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents: 20890
diff changeset
167
26844
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
168 /* Store multi-byte form of the character C in STR. The caller should
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
169 allocate at least 4-byte area at STR in advance. Returns the
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
170 length of the multi-byte form. If C is an invalid character code,
38394
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
171 return -1. */
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
172
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
173 int
38394
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
174 char_to_string_1 (c, str)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
175 int c;
26844
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
176 unsigned char *str;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
177 {
26844
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
178 unsigned char *p = str;
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
179
25250
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
180 if (c & CHAR_MODIFIER_MASK) /* This includes the case C is negative. */
23767
1dfb6d38354f (non_ascii_char_to_string): If C has modifier bits,
Kenichi Handa <handa@m17n.org>
parents: 23720
diff changeset
181 {
25250
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
182 /* Multibyte character can't have a modifier bit. */
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
183 if (! SINGLE_BYTE_CHAR_P ((c & ~CHAR_MODIFIER_MASK)))
38394
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
184 return -1;
25250
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
185
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
186 /* For Meta, Shift, and Control modifiers, we need special care. */
23767
1dfb6d38354f (non_ascii_char_to_string): If C has modifier bits,
Kenichi Handa <handa@m17n.org>
parents: 23720
diff changeset
187 if (c & CHAR_META)
25250
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
188 {
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
189 /* Move the meta bit to the right place for a string. */
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
190 c = (c & ~CHAR_META) | 0x80;
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
191 }
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
192 if (c & CHAR_SHIFT)
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
193 {
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
194 /* Shift modifier is valid only with [A-Za-z]. */
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
195 if ((c & 0377) >= 'A' && (c & 0377) <= 'Z')
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
196 c &= ~CHAR_SHIFT;
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
197 else if ((c & 0377) >= 'a' && (c & 0377) <= 'z')
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
198 c = (c & ~CHAR_SHIFT) - ('a' - 'A');
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
199 }
23767
1dfb6d38354f (non_ascii_char_to_string): If C has modifier bits,
Kenichi Handa <handa@m17n.org>
parents: 23720
diff changeset
200 if (c & CHAR_CTL)
25250
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
201 {
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
202 /* Simulate the code in lread.c. */
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
203 /* Allow `\C- ' and `\C-?'. */
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
204 if (c == (CHAR_CTL | ' '))
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
205 c = 0;
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
206 else if (c == (CHAR_CTL | '?'))
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
207 c = 127;
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
208 /* ASCII control chars are made from letters (both cases),
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
209 as well as the non-letters within 0100...0137. */
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
210 else if ((c & 0137) >= 0101 && (c & 0137) <= 0132)
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
211 c &= (037 | (~0177 & ~CHAR_CTL));
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
212 else if ((c & 0177) >= 0100 && (c & 0177) <= 0137)
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
213 c &= (037 | (~0177 & ~CHAR_CTL));
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
214 }
2228d0a678ef (non_ascii_char_to_string): Handle modifier bits as
Kenichi Handa <handa@m17n.org>
parents: 23978
diff changeset
215
36521
4c211a545642 (char_to_string): If a single byte char has modifier
Kenichi Handa <handa@m17n.org>
parents: 35999
diff changeset
216 /* If C still has any modifier bits, just ignore it. */
4c211a545642 (char_to_string): If a single byte char has modifier
Kenichi Handa <handa@m17n.org>
parents: 35999
diff changeset
217 c &= ~CHAR_MODIFIER_MASK;
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
218 }
49600
23a1cea22d13 Trailing whitespace deleted.
Juanma Barranquero <lekktu@gmail.com>
parents: 48317
diff changeset
219
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
220 if (SINGLE_BYTE_CHAR_P (c))
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
221 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
222 if (ASCII_BYTE_P (c) || c >= 0xA0)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
223 *p++ = c;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
224 else
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
225 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
226 *p++ = LEADING_CODE_8_BIT_CONTROL;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
227 *p++ = c + 0x20;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
228 }
23767
1dfb6d38354f (non_ascii_char_to_string): If C has modifier bits,
Kenichi Handa <handa@m17n.org>
parents: 23720
diff changeset
229 }
31176
3aeaabf6fcad *** empty log message ***
Kenichi Handa <handa@m17n.org>
parents: 31175
diff changeset
230 else if (CHAR_VALID_P (c, 0))
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
231 {
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
232 int charset, c1, c2;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
233
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
234 SPLIT_CHAR (c, charset, c1, c2);
26844
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
235
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
236 if (charset >= LEADING_CODE_EXT_11)
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
237 *p++ = (charset < LEADING_CODE_EXT_12
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
238 ? LEADING_CODE_PRIVATE_11
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
239 : (charset < LEADING_CODE_EXT_21
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
240 ? LEADING_CODE_PRIVATE_12
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
241 : (charset < LEADING_CODE_EXT_22
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
242 ? LEADING_CODE_PRIVATE_21
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
243 : LEADING_CODE_PRIVATE_22)));
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
244 *p++ = charset;
41898
9965a3b28660 (char_to_string_1, translate_char, Fdefine_charset): Add parentheses around
Pavel Janík <Pavel@Janik.cz>
parents: 40656
diff changeset
245 if ((c1 > 0 && c1 < 32) || (c2 > 0 && c2 < 32))
38394
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
246 return -1;
26844
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
247 if (c1)
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
248 {
26844
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
249 *p++ = c1 | 0x80;
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
250 if (c2 > 0)
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
251 *p++ = c2 | 0x80;
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
252 }
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
253 }
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
254 else
38394
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
255 return -1;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
256
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
257 return (p - str);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
258 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
259
38394
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
260
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
261 /* Store multi-byte form of the character C in STR. The caller should
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
262 allocate at least 4-byte area at STR in advance. Returns the
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
263 length of the multi-byte form. If C is an invalid character code,
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
264 signal an error.
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
265
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
266 Use macro `CHAR_STRING (C, STR)' instead of calling this function
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
267 directly if C can be an ASCII character. */
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
268
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
269 int
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
270 char_to_string (c, str)
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
271 int c;
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
272 unsigned char *str;
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
273 {
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
274 int len;
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
275 len = char_to_string_1 (c, str);
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
276 if (len == -1)
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
277 invalid_character (c);
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
278 return len;
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
279 }
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
280
245ae371bd4d (char_to_string_1): Extracted from char_to_string.
Gerd Moellmann <gerd@gnu.org>
parents: 36521
diff changeset
281
25584
01026ea61dcc Lots of comments fixed.
Kenichi Handa <handa@m17n.org>
parents: 25506
diff changeset
282 /* Return the non-ASCII character corresponding to multi-byte form at
01026ea61dcc Lots of comments fixed.
Kenichi Handa <handa@m17n.org>
parents: 25506
diff changeset
283 STR of length LEN. If ACTUAL_LEN is not NULL, store the byte
01026ea61dcc Lots of comments fixed.
Kenichi Handa <handa@m17n.org>
parents: 25506
diff changeset
284 length of the multibyte form in *ACTUAL_LEN.
22185
80a2aa51a6e1 Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents: 22168
diff changeset
285
26844
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
286 Use macros STRING_CHAR or STRING_CHAR_AND_LENGTH instead of calling
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
287 this function directly if you want ot handle ASCII characters as
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
288 well. */
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
289
21514
fa9ff387d260 Fix -Wimplicit warnings.
Andreas Schwab <schwab@suse.de>
parents: 21445
diff changeset
290 int
26844
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
291 string_to_char (str, len, actual_len)
21419
95aae2ff5fcd (string_to_non_ascii_char, multibyte_form_length)
Karl Heuer <kwzh@gnu.org>
parents: 21255
diff changeset
292 const unsigned char *str;
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
293 int len, *actual_len;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
294 {
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
295 int c, bytes, charset, c1, c2;
21445
4c0b4a1025cd (string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents: 21419
diff changeset
296
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
297 SPLIT_MULTIBYTE_SEQ (str, len, bytes, charset, c1, c2);
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
298 c = MAKE_CHAR (charset, c1, c2);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
299 if (actual_len)
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
300 *actual_len = bytes;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
301 return c;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
302 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
303
25584
01026ea61dcc Lots of comments fixed.
Kenichi Handa <handa@m17n.org>
parents: 25506
diff changeset
304 /* Return the length of the multi-byte form at string STR of length LEN.
01026ea61dcc Lots of comments fixed.
Kenichi Handa <handa@m17n.org>
parents: 25506
diff changeset
305 Use the macro MULTIBYTE_FORM_LENGTH instead. */
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
306 int
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
307 multibyte_form_length (str, len)
21419
95aae2ff5fcd (string_to_non_ascii_char, multibyte_form_length)
Karl Heuer <kwzh@gnu.org>
parents: 21255
diff changeset
308 const unsigned char *str;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
309 int len;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
310 {
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
311 int bytes;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
312
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
313 PARSE_MULTIBYTE_SEQ (str, len, bytes);
21445
4c0b4a1025cd (string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents: 21419
diff changeset
314 return bytes;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
315 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
316
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
317 /* Check multibyte form at string STR of length LEN and set variables
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
318 pointed by CHARSET, C1, and C2 to charset and position codes of the
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
319 character at STR, and return 0. If there's no multibyte character,
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
320 return -1. This should be used only in the macro SPLIT_STRING
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
321 which checks range of STR in advance. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
322
21514
fa9ff387d260 Fix -Wimplicit warnings.
Andreas Schwab <schwab@suse.de>
parents: 21445
diff changeset
323 int
26844
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
324 split_string (str, len, charset, c1, c2)
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
325 const unsigned char *str;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
326 unsigned char *c1, *c2;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
327 int len, *charset;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
328 {
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
329 register int bytes, cs, code1, code2 = -1;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
330
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
331 SPLIT_MULTIBYTE_SEQ (str, len, bytes, cs, code1, code2);
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
332 if (cs == CHARSET_ASCII)
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
333 return -1;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
334 *charset = cs;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
335 *c1 = code1;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
336 *c2 = code2;
26327
74566592a2d6 (split_non_ascii_string): Define return value.
Dave Love <fx@gnu.org>
parents: 26088
diff changeset
337 return 0;
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
338 }
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
339
78501
1677cf1c2509 Replace `iff' in comments.
Glenn Morris <rgm@gnu.org>
parents: 78313
diff changeset
340 /* Return 1 if character C has valid printable glyph.
25584
01026ea61dcc Lots of comments fixed.
Kenichi Handa <handa@m17n.org>
parents: 25506
diff changeset
341 Use the macro CHAR_PRINTABLE_P instead. */
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
342 int
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
343 char_printable_p (c)
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
344 int c;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
345 {
34985
cbfe4fc97ebe (char_printable_p): Remove unused variable `chars'.
Eli Zaretskii <eliz@gnu.org>
parents: 34761
diff changeset
346 int charset, c1, c2;
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
347
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
348 if (ASCII_BYTE_P (c))
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
349 return 1;
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
350 else if (SINGLE_BYTE_CHAR_P (c))
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
351 return 0;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
352 else if (c >= MAX_CHAR)
26844
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
353 return 0;
49600
23a1cea22d13 Trailing whitespace deleted.
Juanma Barranquero <lekktu@gmail.com>
parents: 48317
diff changeset
354
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
355 SPLIT_CHAR (c, charset, c1, c2);
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
356 if (! CHARSET_DEFINED_P (charset))
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
357 return 0;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
358 if (CHARSET_CHARS (charset) == 94
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
359 ? c1 <= 32 || c1 >= 127
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
360 : c1 < 32)
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
361 return 0;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
362 if (CHARSET_DIMENSION (charset) == 2
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
363 && (CHARSET_CHARS (charset) == 94
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
364 ? c2 <= 32 || c2 >= 127
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
365 : c2 < 32))
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
366 return 0;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
367 return 1;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
368 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
369
22185
80a2aa51a6e1 Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents: 22168
diff changeset
370 /* Translate character C by translation table TABLE. If C
22121
6e79a15594ab Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents: 21718
diff changeset
371 is negative, translate a character specified by CHARSET, C1, and C2
6e79a15594ab Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents: 21718
diff changeset
372 (C1 and C2 are code points of the character). If no translation is
6e79a15594ab Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents: 21718
diff changeset
373 found in TABLE, return C. */
21514
fa9ff387d260 Fix -Wimplicit warnings.
Andreas Schwab <schwab@suse.de>
parents: 21445
diff changeset
374 int
22121
6e79a15594ab Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents: 21718
diff changeset
375 translate_char (table, c, charset, c1, c2)
17727
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
376 Lisp_Object table;
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
377 int c, charset, c1, c2;
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
378 {
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
379 Lisp_Object ch;
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
380 int alt_charset, alt_c1, alt_c2, dimension;
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
381
25610
5f9f35400f70 Reset MSBs of arguments of MAKE_CHAR.
Kenichi Handa <handa@m17n.org>
parents: 25584
diff changeset
382 if (c < 0) c = MAKE_CHAR (charset, (c1 & 0x7F) , (c2 & 0x7F));
17727
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
383 if (!CHAR_TABLE_P (table)
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
384 || (ch = Faref (table, make_number (c)), !NATNUMP (ch)))
17727
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
385 return c;
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
386
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
387 SPLIT_CHAR (XFASTINT (ch), alt_charset, alt_c1, alt_c2);
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
388 dimension = CHARSET_DIMENSION (alt_charset);
41898
9965a3b28660 (char_to_string_1, translate_char, Fdefine_charset): Add parentheses around
Pavel Janík <Pavel@Janik.cz>
parents: 40656
diff changeset
389 if ((dimension == 1 && alt_c1 > 0) || (dimension == 2 && alt_c2 > 0))
17727
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
390 /* CH is not a generic character, just return it. */
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
391 return XFASTINT (ch);
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
392
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
393 /* Since CH is a generic character, we must return a specific
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
394 charater which has the same position codes as C from CH. */
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
395 if (charset < 0)
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
396 SPLIT_CHAR (c, charset, c1, c2);
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
397 if (dimension != CHARSET_DIMENSION (charset))
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
398 /* We can't make such a character because of dimension mismatch. */
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
399 return c;
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
400 return MAKE_CHAR (alt_charset, c1, c2);
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
401 }
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
402
21034
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
403 /* Convert the unibyte character C to multibyte based on
22121
6e79a15594ab Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents: 21718
diff changeset
404 Vnonascii_translation_table or nonascii_insert_offset. If they can't
21034
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
405 convert C to a valid multibyte character, convert it based on
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
406 DEFAULT_NONASCII_INSERT_OFFSET which makes C a Latin-1 character. */
20705
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
407
21514
fa9ff387d260 Fix -Wimplicit warnings.
Andreas Schwab <schwab@suse.de>
parents: 21445
diff changeset
408 int
20705
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
409 unibyte_char_to_multibyte (c)
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
410 int c;
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
411 {
23928
b718a88bb2b1 (unibyte_char_to_multibyte): Don't convert 7-bit ASCII characters via
Eli Zaretskii <eliz@gnu.org>
parents: 23883
diff changeset
412 if (c < 0400 && c >= 0200)
20705
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
413 {
21034
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
414 int c_save = c;
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
415
22121
6e79a15594ab Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents: 21718
diff changeset
416 if (! NILP (Vnonascii_translation_table))
23154
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
417 {
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
418 c = XINT (Faref (Vnonascii_translation_table, make_number (c)));
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
419 if (c >= 0400 && ! char_valid_p (c, 0))
23154
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
420 c = c_save + DEFAULT_NONASCII_INSERT_OFFSET;
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
421 }
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
422 else if (c >= 0240 && nonascii_insert_offset > 0)
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
423 {
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
424 c += nonascii_insert_offset;
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
425 if (c < 0400 || ! char_valid_p (c, 0))
23154
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
426 c = c_save + DEFAULT_NONASCII_INSERT_OFFSET;
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
427 }
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
428 else if (c >= 0240)
21034
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
429 c = c_save + DEFAULT_NONASCII_INSERT_OFFSET;
20705
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
430 }
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
431 return c;
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
432 }
22927
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
433
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
434
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
435 /* Convert the multibyte character C to unibyte 8-bit character based
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
436 on Vnonascii_translation_table or nonascii_insert_offset. If
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
437 REV_TBL is non-nil, it should be a reverse table of
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
438 Vnonascii_translation_table, i.e. what given by:
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
439 Fchar_table_extra_slot (Vnonascii_translation_table, make_number (0)) */
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
440
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
441 int
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
442 multibyte_char_to_unibyte (c, rev_tbl)
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
443 int c;
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
444 Lisp_Object rev_tbl;
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
445 {
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
446 if (!SINGLE_BYTE_CHAR_P (c))
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
447 {
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
448 int c_save = c;
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
449
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
450 if (! CHAR_TABLE_P (rev_tbl)
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
451 && CHAR_TABLE_P (Vnonascii_translation_table))
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
452 rev_tbl = Fchar_table_extra_slot (Vnonascii_translation_table,
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
453 make_number (0));
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
454 if (CHAR_TABLE_P (rev_tbl))
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
455 {
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
456 Lisp_Object temp;
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
457 temp = Faref (rev_tbl, make_number (c));
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
458 if (INTEGERP (temp))
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
459 c = XINT (temp);
23154
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
460 if (c >= 256)
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
461 c = (c_save & 0177) + 0200;
22927
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
462 }
23154
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
463 else
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
464 {
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
465 if (nonascii_insert_offset > 0)
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
466 c -= nonascii_insert_offset;
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
467 if (c < 128 || c >= 256)
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
468 c = (c_save & 0177) + 0200;
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
469 }
22927
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
470 }
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
471
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
472 return c;
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
473 }
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
474
20705
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
475
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
476 /* Update the table Vcharset_table with the given arguments (see the
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
477 document of `define-charset' for the meaning of each argument).
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
478 Several other table contents are also updated. The caller should
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
479 check the validity of CHARSET-ID and the remaining arguments in
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
480 advance. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
481
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
482 void
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
483 update_charset_table (charset_id, dimension, chars, width, direction,
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
484 iso_final_char, iso_graphic_plane,
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
485 short_name, long_name, description)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
486 Lisp_Object charset_id, dimension, chars, width, direction;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
487 Lisp_Object iso_final_char, iso_graphic_plane;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
488 Lisp_Object short_name, long_name, description;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
489 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
490 int charset = XINT (charset_id);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
491 int bytes;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
492 unsigned char leading_code_base, leading_code_ext;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
493
17322
a7d9b8e167ca (non_ascii_char_to_string, update_charset_table,
Kenichi Handa <handa@m17n.org>
parents: 17186
diff changeset
494 if (NILP (CHARSET_TABLE_ENTRY (charset)))
a7d9b8e167ca (non_ascii_char_to_string, update_charset_table,
Kenichi Handa <handa@m17n.org>
parents: 17186
diff changeset
495 CHARSET_TABLE_ENTRY (charset)
a7d9b8e167ca (non_ascii_char_to_string, update_charset_table,
Kenichi Handa <handa@m17n.org>
parents: 17186
diff changeset
496 = Fmake_vector (make_number (CHARSET_MAX_IDX), Qnil);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
497
29139
c1c227f62493 (update_charset_table): Accept nil in LONG_NAME and
Kenichi Handa <handa@m17n.org>
parents: 29095
diff changeset
498 if (NILP (long_name))
c1c227f62493 (update_charset_table): Accept nil in LONG_NAME and
Kenichi Handa <handa@m17n.org>
parents: 29095
diff changeset
499 long_name = short_name;
c1c227f62493 (update_charset_table): Accept nil in LONG_NAME and
Kenichi Handa <handa@m17n.org>
parents: 29095
diff changeset
500 if (NILP (description))
c1c227f62493 (update_charset_table): Accept nil in LONG_NAME and
Kenichi Handa <handa@m17n.org>
parents: 29095
diff changeset
501 description = long_name;
c1c227f62493 (update_charset_table): Accept nil in LONG_NAME and
Kenichi Handa <handa@m17n.org>
parents: 29095
diff changeset
502
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
503 /* Get byte length of multibyte form, base leading-code, and
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
504 extended leading-code of the charset. See the comment under the
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
505 title "GENERAL NOTE on CHARACTER SET (CHARSET)" in charset.h. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
506 bytes = XINT (dimension);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
507 if (charset < MIN_CHARSET_PRIVATE_DIMENSION1)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
508 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
509 /* Official charset, it doesn't have an extended leading-code. */
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
510 if (charset != CHARSET_ASCII && charset != CHARSET_8_BIT_GRAPHIC)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
511 bytes += 1; /* For a base leading-code. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
512 leading_code_base = charset;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
513 leading_code_ext = 0;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
514 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
515 else
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
516 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
517 /* Private charset. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
518 bytes += 2; /* For base and extended leading-codes. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
519 leading_code_base
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
520 = (charset < LEADING_CODE_EXT_12
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
521 ? LEADING_CODE_PRIVATE_11
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
522 : (charset < LEADING_CODE_EXT_21
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
523 ? LEADING_CODE_PRIVATE_12
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
524 : (charset < LEADING_CODE_EXT_22
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
525 ? LEADING_CODE_PRIVATE_21
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
526 : LEADING_CODE_PRIVATE_22)));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
527 leading_code_ext = charset;
29538
c0c34d18e8a0 (update_charset_table): Update the table
Kenichi Handa <handa@m17n.org>
parents: 29269
diff changeset
528 if (BYTES_BY_CHAR_HEAD (leading_code_base) != bytes)
c0c34d18e8a0 (update_charset_table): Update the table
Kenichi Handa <handa@m17n.org>
parents: 29269
diff changeset
529 error ("Invalid dimension for the charset-ID %d", charset);
c0c34d18e8a0 (update_charset_table): Update the table
Kenichi Handa <handa@m17n.org>
parents: 29269
diff changeset
530 }
23628
4a0a107fac9e (string_to_non_ascii_char): Change the check for the
Kenichi Handa <handa@m17n.org>
parents: 23576
diff changeset
531
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
532 CHARSET_TABLE_INFO (charset, CHARSET_ID_IDX) = charset_id;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
533 CHARSET_TABLE_INFO (charset, CHARSET_BYTES_IDX) = make_number (bytes);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
534 CHARSET_TABLE_INFO (charset, CHARSET_DIMENSION_IDX) = dimension;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
535 CHARSET_TABLE_INFO (charset, CHARSET_CHARS_IDX) = chars;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
536 CHARSET_TABLE_INFO (charset, CHARSET_WIDTH_IDX) = width;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
537 CHARSET_TABLE_INFO (charset, CHARSET_DIRECTION_IDX) = direction;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
538 CHARSET_TABLE_INFO (charset, CHARSET_LEADING_CODE_BASE_IDX)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
539 = make_number (leading_code_base);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
540 CHARSET_TABLE_INFO (charset, CHARSET_LEADING_CODE_EXT_IDX)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
541 = make_number (leading_code_ext);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
542 CHARSET_TABLE_INFO (charset, CHARSET_ISO_FINAL_CHAR_IDX) = iso_final_char;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
543 CHARSET_TABLE_INFO (charset, CHARSET_ISO_GRAPHIC_PLANE_IDX)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
544 = iso_graphic_plane;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
545 CHARSET_TABLE_INFO (charset, CHARSET_SHORT_NAME_IDX) = short_name;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
546 CHARSET_TABLE_INFO (charset, CHARSET_LONG_NAME_IDX) = long_name;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
547 CHARSET_TABLE_INFO (charset, CHARSET_DESCRIPTION_IDX) = description;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
548 CHARSET_TABLE_INFO (charset, CHARSET_PLIST_IDX) = Qnil;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
549
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
550 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
551 /* If we have already defined a charset which has the same
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
552 DIMENSION, CHARS and ISO-FINAL-CHAR but the different
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
553 DIRECTION, we must update the entry REVERSE-CHARSET of both
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
554 charsets. If there's no such charset, the value of the entry
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
555 is set to nil. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
556 int i;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
557
17186
65bc06706714 Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents: 17121
diff changeset
558 for (i = 0; i <= MAX_CHARSET; i++)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
559 if (!NILP (CHARSET_TABLE_ENTRY (i)))
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
560 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
561 if (CHARSET_DIMENSION (i) == XINT (dimension)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
562 && CHARSET_CHARS (i) == XINT (chars)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
563 && CHARSET_ISO_FINAL_CHAR (i) == XINT (iso_final_char)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
564 && CHARSET_DIRECTION (i) != XINT (direction))
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
565 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
566 CHARSET_TABLE_INFO (charset, CHARSET_REVERSE_CHARSET_IDX)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
567 = make_number (i);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
568 CHARSET_TABLE_INFO (i, CHARSET_REVERSE_CHARSET_IDX) = charset_id;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
569 break;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
570 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
571 }
17186
65bc06706714 Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents: 17121
diff changeset
572 if (i > MAX_CHARSET)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
573 /* No such a charset. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
574 CHARSET_TABLE_INFO (charset, CHARSET_REVERSE_CHARSET_IDX)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
575 = make_number (-1);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
576 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
577
29538
c0c34d18e8a0 (update_charset_table): Update the table
Kenichi Handa <handa@m17n.org>
parents: 29269
diff changeset
578 if (charset != CHARSET_ASCII && charset != CHARSET_8_BIT_GRAPHIC
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
579 && charset < MIN_CHARSET_PRIVATE_DIMENSION1)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
580 {
29538
c0c34d18e8a0 (update_charset_table): Update the table
Kenichi Handa <handa@m17n.org>
parents: 29269
diff changeset
581 bytes_by_char_head[leading_code_base] = bytes;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
582 width_by_char_head[leading_code_base] = XINT (width);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
583
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
584 /* Update table emacs_code_class. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
585 emacs_code_class[charset] = (bytes == 2
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
586 ? EMACS_leading_code_2
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
587 : (bytes == 3
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
588 ? EMACS_leading_code_3
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
589 : EMACS_leading_code_4));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
590 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
591
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
592 /* Update table iso_charset_table. */
29717
acaa36b47f50 fix up more Lisp_Object/int conversion issues
Ken Raeburn <raeburn@raeburn.org>
parents: 29538
diff changeset
593 if (XINT (iso_final_char) >= 0
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
594 && ISO_CHARSET_TABLE (dimension, chars, iso_final_char) < 0)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
595 ISO_CHARSET_TABLE (dimension, chars, iso_final_char) = charset;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
596 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
597
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
598 #ifdef emacs
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
599
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
600 /* Return charset id of CHARSET_SYMBOL, or return -1 if CHARSET_SYMBOL
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
601 is invalid. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
602 int
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
603 get_charset_id (charset_symbol)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
604 Lisp_Object charset_symbol;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
605 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
606 Lisp_Object val;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
607 int charset;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
608
43144
69a3fef3cb46 (get_charset_id): Use if-else instead of ?:.
Eli Zaretskii <eliz@gnu.org>
parents: 42162
diff changeset
609 /* This originally used a ?: operator, but reportedly the HP-UX
69a3fef3cb46 (get_charset_id): Use if-else instead of ?:.
Eli Zaretskii <eliz@gnu.org>
parents: 42162
diff changeset
610 compiler version HP92453-01 A.10.32.22 miscompiles that. */
69a3fef3cb46 (get_charset_id): Use if-else instead of ?:.
Eli Zaretskii <eliz@gnu.org>
parents: 42162
diff changeset
611 if (SYMBOLP (charset_symbol)
69a3fef3cb46 (get_charset_id): Use if-else instead of ?:.
Eli Zaretskii <eliz@gnu.org>
parents: 42162
diff changeset
612 && VECTORP (val = Fget (charset_symbol, Qcharset))
69a3fef3cb46 (get_charset_id): Use if-else instead of ?:.
Eli Zaretskii <eliz@gnu.org>
parents: 42162
diff changeset
613 && CHARSET_VALID_P (charset =
69a3fef3cb46 (get_charset_id): Use if-else instead of ?:.
Eli Zaretskii <eliz@gnu.org>
parents: 42162
diff changeset
614 XINT (XVECTOR (val)->contents[CHARSET_ID_IDX])))
69a3fef3cb46 (get_charset_id): Use if-else instead of ?:.
Eli Zaretskii <eliz@gnu.org>
parents: 42162
diff changeset
615 return charset;
69a3fef3cb46 (get_charset_id): Use if-else instead of ?:.
Eli Zaretskii <eliz@gnu.org>
parents: 42162
diff changeset
616 else
69a3fef3cb46 (get_charset_id): Use if-else instead of ?:.
Eli Zaretskii <eliz@gnu.org>
parents: 42162
diff changeset
617 return -1;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
618 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
619
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
620 /* Return an identification number for a new private charset of
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
621 DIMENSION and WIDTH. If there's no more room for the new charset,
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
622 return 0. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
623 Lisp_Object
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
624 get_new_private_charset_id (dimension, width)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
625 int dimension, width;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
626 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
627 int charset, from, to;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
628
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
629 if (dimension == 1)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
630 {
33629
d07521d90a8e (get_new_private_charset_id): Don't limit CHARSET_ID by WIDTH.
Kenichi Handa <handa@m17n.org>
parents: 32380
diff changeset
631 from = LEADING_CODE_EXT_11;
d07521d90a8e (get_new_private_charset_id): Don't limit CHARSET_ID by WIDTH.
Kenichi Handa <handa@m17n.org>
parents: 32380
diff changeset
632 to = LEADING_CODE_EXT_21;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
633 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
634 else
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
635 {
33629
d07521d90a8e (get_new_private_charset_id): Don't limit CHARSET_ID by WIDTH.
Kenichi Handa <handa@m17n.org>
parents: 32380
diff changeset
636 from = LEADING_CODE_EXT_21;
d07521d90a8e (get_new_private_charset_id): Don't limit CHARSET_ID by WIDTH.
Kenichi Handa <handa@m17n.org>
parents: 32380
diff changeset
637 to = LEADING_CODE_EXT_MAX + 1;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
638 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
639
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
640 for (charset = from; charset < to; charset++)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
641 if (!CHARSET_DEFINED_P (charset)) break;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
642
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
643 return make_number (charset < to ? charset : 0);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
644 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
645
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
646 DEFUN ("define-charset", Fdefine_charset, Sdefine_charset, 3, 3, 0,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
647 doc: /* Define CHARSET-ID as the identification number of CHARSET with INFO-VECTOR.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
648 If CHARSET-ID is nil, it is decided automatically, which means CHARSET is
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
649 treated as a private charset.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
650 INFO-VECTOR is a vector of the format:
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
651 [DIMENSION CHARS WIDTH DIRECTION ISO-FINAL-CHAR ISO-GRAPHIC-PLANE
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
652 SHORT-NAME LONG-NAME DESCRIPTION]
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
653 The meanings of each elements is as follows:
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
654 DIMENSION (integer) is the number of bytes to represent a character: 1 or 2.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
655 CHARS (integer) is the number of characters in a dimension: 94 or 96.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
656 WIDTH (integer) is the number of columns a character in the charset
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
657 occupies on the screen: one of 0, 1, and 2.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
658
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
659 DIRECTION (integer) is the rendering direction of characters in the
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
660 charset when rendering. If 0, render from left to right, else
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
661 render from right to left.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
662
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
663 ISO-FINAL-CHAR (character) is the final character of the
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
664 corresponding ISO 2022 charset.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
665 It may be -1 if the charset is internal use only.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
666
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
667 ISO-GRAPHIC-PLANE (integer) is the graphic plane to be invoked
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
668 while encoding to variants of ISO 2022 coding system, one of the
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
669 following: 0/graphic-plane-left(GL), 1/graphic-plane-right(GR).
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
670 It may be -1 if the charset is internal use only.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
671
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
672 SHORT-NAME (string) is the short name to refer to the charset.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
673
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
674 LONG-NAME (string) is the long name to refer to the charset.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
675
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
676 DESCRIPTION (string) is the description string of the charset. */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
677 (charset_id, charset_symbol, info_vector)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
678 Lisp_Object charset_id, charset_symbol, info_vector;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
679 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
680 Lisp_Object *vec;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
681
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
682 if (!NILP (charset_id))
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
683 CHECK_NUMBER (charset_id);
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
684 CHECK_SYMBOL (charset_symbol);
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
685 CHECK_VECTOR (info_vector);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
686
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
687 if (! NILP (charset_id))
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
688 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
689 if (! CHARSET_VALID_P (XINT (charset_id)))
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
690 error ("Invalid CHARSET: %d", XINT (charset_id));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
691 else if (CHARSET_DEFINED_P (XINT (charset_id)))
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
692 error ("Already defined charset: %d", XINT (charset_id));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
693 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
694
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
695 vec = XVECTOR (info_vector)->contents;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
696 if (XVECTOR (info_vector)->size != 9
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
697 || !INTEGERP (vec[0]) || !(XINT (vec[0]) == 1 || XINT (vec[0]) == 2)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
698 || !INTEGERP (vec[1]) || !(XINT (vec[1]) == 94 || XINT (vec[1]) == 96)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
699 || !INTEGERP (vec[2]) || !(XINT (vec[2]) == 1 || XINT (vec[2]) == 2)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
700 || !INTEGERP (vec[3]) || !(XINT (vec[3]) == 0 || XINT (vec[3]) == 1)
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
701 || !INTEGERP (vec[4])
41898
9965a3b28660 (char_to_string_1, translate_char, Fdefine_charset): Add parentheses around
Pavel Janík <Pavel@Janik.cz>
parents: 40656
diff changeset
702 || !(XINT (vec[4]) == -1 || (XINT (vec[4]) >= '0' && XINT (vec[4]) <= '~'))
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
703 || !INTEGERP (vec[5])
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
704 || !(XINT (vec[5]) == -1 || XINT (vec[5]) == 0 || XINT (vec[5]) == 1)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
705 || !STRINGP (vec[6])
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
706 || !STRINGP (vec[7])
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
707 || !STRINGP (vec[8]))
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
708 error ("Invalid info-vector argument for defining charset %s",
46370
40db0673e6f0 Most uses of XSTRING combined with STRING_BYTES or indirection changed to
Ken Raeburn <raeburn@raeburn.org>
parents: 46056
diff changeset
709 SDATA (SYMBOL_NAME (charset_symbol)));
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
710
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
711 if (NILP (charset_id))
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
712 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
713 charset_id = get_new_private_charset_id (XINT (vec[0]), XINT (vec[2]));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
714 if (XINT (charset_id) == 0)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
715 error ("There's no room for a new private charset %s",
46370
40db0673e6f0 Most uses of XSTRING combined with STRING_BYTES or indirection changed to
Ken Raeburn <raeburn@raeburn.org>
parents: 46056
diff changeset
716 SDATA (SYMBOL_NAME (charset_symbol)));
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
717 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
718
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
719 update_charset_table (charset_id, vec[0], vec[1], vec[2], vec[3],
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
720 vec[4], vec[5], vec[6], vec[7], vec[8]);
17322
a7d9b8e167ca (non_ascii_char_to_string, update_charset_table,
Kenichi Handa <handa@m17n.org>
parents: 17186
diff changeset
721 Fput (charset_symbol, Qcharset, CHARSET_TABLE_ENTRY (XINT (charset_id)));
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
722 CHARSET_SYMBOL (XINT (charset_id)) = charset_symbol;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
723 Vcharset_list = Fcons (charset_symbol, Vcharset_list);
42162
478d5ff60075 (Fdefine_charset): Call Fupdate_coding_systems_internal.
Richard M. Stallman <rms@gnu.org>
parents: 41898
diff changeset
724 Fupdate_coding_systems_internal ();
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
725 return Qnil;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
726 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
727
22703
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
728 DEFUN ("generic-character-list", Fgeneric_character_list,
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
729 Sgeneric_character_list, 0, 0, 0,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
730 doc: /* Return a list of all possible generic characters.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
731 It includes a generic character for a charset not yet defined. */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
732 ()
22703
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
733 {
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
734 return Vgeneric_character_list;
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
735 }
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
736
20435
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
737 DEFUN ("get-unused-iso-final-char", Fget_unused_iso_final_char,
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
738 Sget_unused_iso_final_char, 2, 2, 0,
55505
7abf0e14a59e (Fget_unused_iso_final_char): Fix typos in docstring.
Juanma Barranquero <lekktu@gmail.com>
parents: 53528
diff changeset
739 doc: /* Return an unused ISO's final char for a charset of DIMENSION and CHARS.
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
740 DIMENSION is the number of bytes to represent a character: 1 or 2.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
741 CHARS is the number of characters in a dimension: 94 or 96.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
742
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
743 This final char is for private use, thus the range is `0' (48) .. `?' (63).
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
744 If there's no unused final char for the specified kind of charset,
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
745 return nil. */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
746 (dimension, chars)
20435
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
747 Lisp_Object dimension, chars;
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
748 {
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
749 int final_char;
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
750
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
751 CHECK_NUMBER (dimension);
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
752 CHECK_NUMBER (chars);
20435
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
753 if (XINT (dimension) != 1 && XINT (dimension) != 2)
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
754 error ("Invalid charset dimension %d, it should be 1 or 2",
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
755 XINT (dimension));
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
756 if (XINT (chars) != 94 && XINT (chars) != 96)
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
757 error ("Invalid charset chars %d, it should be 94 or 96",
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
758 XINT (chars));
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
759 for (final_char = '0'; final_char <= '?'; final_char++)
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
760 {
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
761 if (ISO_CHARSET_TABLE (dimension, chars, make_number (final_char)) < 0)
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
762 break;
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
763 }
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
764 return (final_char <= '?' ? make_number (final_char) : Qnil);
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
765 }
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
766
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
767 DEFUN ("declare-equiv-charset", Fdeclare_equiv_charset, Sdeclare_equiv_charset,
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
768 4, 4, 0,
53528
dbd73534e01e (Fdeclare_equiv_charset): Fix docstring.
Kenichi Handa <handa@m17n.org>
parents: 52401
diff changeset
769 doc: /* Declare an equivalent charset for ISO-2022 decoding.
dbd73534e01e (Fdeclare_equiv_charset): Fix docstring.
Kenichi Handa <handa@m17n.org>
parents: 52401
diff changeset
770
dbd73534e01e (Fdeclare_equiv_charset): Fix docstring.
Kenichi Handa <handa@m17n.org>
parents: 52401
diff changeset
771 On decoding by an ISO-2022 base coding system, when a charset
dbd73534e01e (Fdeclare_equiv_charset): Fix docstring.
Kenichi Handa <handa@m17n.org>
parents: 52401
diff changeset
772 specified by DIMENSION, CHARS, and FINAL-CHAR is designated, behave as
dbd73534e01e (Fdeclare_equiv_charset): Fix docstring.
Kenichi Handa <handa@m17n.org>
parents: 52401
diff changeset
773 if CHARSET is designated instead. */)
dbd73534e01e (Fdeclare_equiv_charset): Fix docstring.
Kenichi Handa <handa@m17n.org>
parents: 52401
diff changeset
774 (dimension, chars, final_char, charset)
dbd73534e01e (Fdeclare_equiv_charset): Fix docstring.
Kenichi Handa <handa@m17n.org>
parents: 52401
diff changeset
775 Lisp_Object dimension, chars, final_char, charset;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
776 {
53528
dbd73534e01e (Fdeclare_equiv_charset): Fix docstring.
Kenichi Handa <handa@m17n.org>
parents: 52401
diff changeset
777 int charset_id;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
778
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
779 CHECK_NUMBER (dimension);
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
780 CHECK_NUMBER (chars);
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
781 CHECK_NUMBER (final_char);
53528
dbd73534e01e (Fdeclare_equiv_charset): Fix docstring.
Kenichi Handa <handa@m17n.org>
parents: 52401
diff changeset
782 CHECK_SYMBOL (charset);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
783
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
784 if (XINT (dimension) != 1 && XINT (dimension) != 2)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
785 error ("Invalid DIMENSION %d, it should be 1 or 2", XINT (dimension));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
786 if (XINT (chars) != 94 && XINT (chars) != 96)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
787 error ("Invalid CHARS %d, it should be 94 or 96", XINT (chars));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
788 if (XINT (final_char) < '0' || XFASTINT (final_char) > '~')
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
789 error ("Invalid FINAL-CHAR %c, it should be `0'..`~'", XINT (chars));
53528
dbd73534e01e (Fdeclare_equiv_charset): Fix docstring.
Kenichi Handa <handa@m17n.org>
parents: 52401
diff changeset
790 if ((charset_id = get_charset_id (charset)) < 0)
dbd73534e01e (Fdeclare_equiv_charset): Fix docstring.
Kenichi Handa <handa@m17n.org>
parents: 52401
diff changeset
791 error ("Invalid charset %s", SDATA (SYMBOL_NAME (charset)));
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
792
53528
dbd73534e01e (Fdeclare_equiv_charset): Fix docstring.
Kenichi Handa <handa@m17n.org>
parents: 52401
diff changeset
793 ISO_CHARSET_TABLE (dimension, chars, final_char) = charset_id;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
794 return Qnil;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
795 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
796
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
797 /* Return information about charsets in the text at PTR of NBYTES
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
798 bytes, which are NCHARS characters. The value is:
29094
8b1a19b318e9 Fix comment for find_charset_in_text.
Kenichi Handa <handa@m17n.org>
parents: 29003
diff changeset
799
29095
39e94c12f469 Fix previous change (typo).
Kenichi Handa <handa@m17n.org>
parents: 29094
diff changeset
800 0: Each character is represented by one byte. This is always
29094
8b1a19b318e9 Fix comment for find_charset_in_text.
Kenichi Handa <handa@m17n.org>
parents: 29003
diff changeset
801 true for unibyte text.
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
802 1: No charsets other than ascii eight-bit-control,
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
803 eight-bit-graphic, and latin-1 are found.
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
804 2: Otherwise.
22168
2156ac6d054f (find_charset_in_str): New arg CMPCHARP.
Kenichi Handa <handa@m17n.org>
parents: 22121
diff changeset
805
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
806 In addition, if CHARSETS is nonzero, for each found charset N, set
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
807 CHARSETS[N] to 1. For that, callers should allocate CHARSETS
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
808 (MAX_CHARSET + 1 elements) in advance. It may lookup a translation
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
809 table TABLE if supplied. For invalid charsets, set CHARSETS[1] to
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
810 1 (note that there's no charset whose ID is 1). */
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
811
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
812 int
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
813 find_charset_in_text (ptr, nchars, nbytes, charsets, table)
46460
876356a4a4ca (find_charset_in_text, c_string_width):
Ken Raeburn <raeburn@raeburn.org>
parents: 46370
diff changeset
814 const unsigned char *ptr;
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
815 int nchars, nbytes, *charsets;
17727
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
816 Lisp_Object table;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
817 {
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
818 if (nchars == nbytes)
23883
0ca2fa58ca7b (Qunknown): New variable.
Kenichi Handa <handa@m17n.org>
parents: 23767
diff changeset
819 {
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
820 if (charsets && nbytes > 0)
23883
0ca2fa58ca7b (Qunknown): New variable.
Kenichi Handa <handa@m17n.org>
parents: 23767
diff changeset
821 {
46460
876356a4a4ca (find_charset_in_text, c_string_width):
Ken Raeburn <raeburn@raeburn.org>
parents: 46370
diff changeset
822 const unsigned char *endp = ptr + nbytes;
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
823 int maskbits = 0;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
824
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
825 while (ptr < endp && maskbits != 7)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
826 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
827 maskbits |= (*ptr < 0x80 ? 1 : *ptr < 0xA0 ? 2 : 4);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
828 ptr++;
49600
23a1cea22d13 Trailing whitespace deleted.
Juanma Barranquero <lekktu@gmail.com>
parents: 48317
diff changeset
829 }
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
830
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
831 if (maskbits & 1)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
832 charsets[CHARSET_ASCII] = 1;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
833 if (maskbits & 2)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
834 charsets[CHARSET_8_BIT_CONTROL] = 1;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
835 if (maskbits & 4)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
836 charsets[CHARSET_8_BIT_GRAPHIC] = 1;
23883
0ca2fa58ca7b (Qunknown): New variable.
Kenichi Handa <handa@m17n.org>
parents: 23767
diff changeset
837 }
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
838 return 0;
23883
0ca2fa58ca7b (Qunknown): New variable.
Kenichi Handa <handa@m17n.org>
parents: 23767
diff changeset
839 }
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
840 else
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
841 {
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
842 int return_val = 1;
26844
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
843 int bytes, charset, c1, c2;
20228
acb7aa0aa71f (find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents: 20188
diff changeset
844
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
845 if (! CHAR_TABLE_P (table))
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
846 table = Qnil;
20228
acb7aa0aa71f (find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents: 20188
diff changeset
847
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
848 while (nchars-- > 0)
17727
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
849 {
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
850 SPLIT_MULTIBYTE_SEQ (ptr, len, bytes, charset, c1, c2);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
851 ptr += bytes;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
852
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
853 if (!CHARSET_DEFINED_P (charset))
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
854 charset = 1;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
855 else if (! NILP (table))
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
856 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
857 int c = translate_char (table, -1, charset, c1, c2);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
858 if (c >= 0)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
859 charset = CHAR_CHARSET (c);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
860 }
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
861
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
862 if (return_val == 1
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
863 && charset != CHARSET_ASCII
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
864 && charset != CHARSET_8_BIT_CONTROL
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
865 && charset != CHARSET_8_BIT_GRAPHIC
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
866 && charset != charset_latin_iso8859_1)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
867 return_val = 2;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
868
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
869 if (charsets)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
870 charsets[charset] = 1;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
871 else if (return_val == 2)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
872 break;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
873 }
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
874 return return_val;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
875 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
876 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
877
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
878 DEFUN ("find-charset-region", Ffind_charset_region, Sfind_charset_region,
17727
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
879 2, 3, 0,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
880 doc: /* Return a list of charsets in the region between BEG and END.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
881 BEG and END are buffer positions.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
882 Optional arg TABLE if non-nil is a translation table to look up.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
883
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
884 If the region contains invalid multibyte characters,
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
885 `unknown' is included in the returned list.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
886
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
887 If the current buffer is unibyte, the returned list may contain
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
888 only `ascii', `eight-bit-control', and `eight-bit-graphic'. */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
889 (beg, end, table)
17727
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
890 Lisp_Object beg, end, table;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
891 {
17880
06f58d8a59ba (Ffind_charset_region): Change the declaration of
Kenichi Handa <handa@m17n.org>
parents: 17863
diff changeset
892 int charsets[MAX_CHARSET + 1];
20534
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
893 int from, from_byte, to, stop, stop_byte, i;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
894 Lisp_Object val;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
895
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
896 validate_region (&beg, &end);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
897 from = XFASTINT (beg);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
898 stop = to = XFASTINT (end);
20534
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
899
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
900 if (from < GPT && GPT < to)
20534
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
901 {
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
902 stop = GPT;
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
903 stop_byte = GPT_BYTE;
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
904 }
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
905 else
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
906 stop_byte = CHAR_TO_BYTE (stop);
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
907
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
908 from_byte = CHAR_TO_BYTE (from);
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
909
17880
06f58d8a59ba (Ffind_charset_region): Change the declaration of
Kenichi Handa <handa@m17n.org>
parents: 17863
diff changeset
910 bzero (charsets, (MAX_CHARSET + 1) * sizeof (int));
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
911 while (1)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
912 {
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
913 find_charset_in_text (BYTE_POS_ADDR (from_byte), stop - from,
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
914 stop_byte - from_byte, charsets, table);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
915 if (stop < to)
20534
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
916 {
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
917 from = stop, from_byte = stop_byte;
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
918 stop = to, stop_byte = CHAR_TO_BYTE (stop);
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
919 }
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
920 else
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
921 break;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
922 }
20534
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
923
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
924 val = Qnil;
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
925 if (charsets[1])
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
926 val = Fcons (Qunknown, val);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
927 for (i = MAX_CHARSET; i >= MIN_CHARSET_OFFICIAL_DIMENSION1; i--)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
928 if (charsets[i])
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
929 val = Fcons (CHARSET_SYMBOL (i), val);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
930 if (charsets[0])
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
931 val = Fcons (Qascii, val);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
932 return val;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
933 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
934
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
935 DEFUN ("find-charset-string", Ffind_charset_string, Sfind_charset_string,
17727
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
936 1, 2, 0,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
937 doc: /* Return a list of charsets in STR.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
938 Optional arg TABLE if non-nil is a translation table to look up.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
939
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
940 If the string contains invalid multibyte characters,
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
941 `unknown' is included in the returned list.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
942
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
943 If STR is unibyte, the returned list may contain
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
944 only `ascii', `eight-bit-control', and `eight-bit-graphic'. */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
945 (str, table)
17727
9d39361ce928 (unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 17366
diff changeset
946 Lisp_Object str, table;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
947 {
18003
ac148c939dec (Ffind_charset_string): The variable CHARSETS is
Kenichi Handa <handa@m17n.org>
parents: 17880
diff changeset
948 int charsets[MAX_CHARSET + 1];
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
949 int i;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
950 Lisp_Object val;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
951
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
952 CHECK_STRING (str);
20614
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
953
18003
ac148c939dec (Ffind_charset_string): The variable CHARSETS is
Kenichi Handa <handa@m17n.org>
parents: 17880
diff changeset
954 bzero (charsets, (MAX_CHARSET + 1) * sizeof (int));
46370
40db0673e6f0 Most uses of XSTRING combined with STRING_BYTES or indirection changed to
Ken Raeburn <raeburn@raeburn.org>
parents: 46056
diff changeset
955 find_charset_in_text (SDATA (str), SCHARS (str),
40db0673e6f0 Most uses of XSTRING combined with STRING_BYTES or indirection changed to
Ken Raeburn <raeburn@raeburn.org>
parents: 46056
diff changeset
956 SBYTES (str), charsets, table);
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
957
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
958 val = Qnil;
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
959 if (charsets[1])
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
960 val = Fcons (Qunknown, val);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
961 for (i = MAX_CHARSET; i >= MIN_CHARSET_OFFICIAL_DIMENSION1; i--)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
962 if (charsets[i])
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
963 val = Fcons (CHARSET_SYMBOL (i), val);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
964 if (charsets[0])
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
965 val = Fcons (Qascii, val);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
966 return val;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
967 }
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
968
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
969
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
970 DEFUN ("make-char-internal", Fmake_char_internal, Smake_char_internal, 1, 3, 0,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
971 doc: /* Return a character made from arguments.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
972 Internal use only. */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
973 (charset, code1, code2)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
974 Lisp_Object charset, code1, code2;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
975 {
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
976 int charset_id, c1, c2;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
977
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
978 CHECK_NUMBER (charset);
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
979 charset_id = XINT (charset);
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
980 if (!CHARSET_DEFINED_P (charset_id))
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
981 error ("Invalid charset ID: %d", XINT (charset));
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
982
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
983 if (NILP (code1))
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
984 c1 = 0;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
985 else
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
986 {
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
987 CHECK_NUMBER (code1);
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
988 c1 = XINT (code1);
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
989 }
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
990 if (NILP (code2))
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
991 c2 = 0;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
992 else
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
993 {
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
994 CHECK_NUMBER (code2);
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
995 c2 = XINT (code2);
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
996 }
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
997
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
998 if (charset_id == CHARSET_ASCII)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
999 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1000 if (c1 < 0 || c1 > 0x7F)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1001 goto invalid_code_posints;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1002 return make_number (c1);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1003 }
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1004 else if (charset_id == CHARSET_8_BIT_CONTROL)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1005 {
34761
b4244c3295bb (Fmake_char_internal): If CHARSET doesn't have a
Kenichi Handa <handa@m17n.org>
parents: 33629
diff changeset
1006 if (NILP (code1))
b4244c3295bb (Fmake_char_internal): If CHARSET doesn't have a
Kenichi Handa <handa@m17n.org>
parents: 33629
diff changeset
1007 c1 = 0x80;
b4244c3295bb (Fmake_char_internal): If CHARSET doesn't have a
Kenichi Handa <handa@m17n.org>
parents: 33629
diff changeset
1008 else if (c1 < 0x80 || c1 > 0x9F)
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1009 goto invalid_code_posints;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1010 return make_number (c1);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1011 }
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1012 else if (charset_id == CHARSET_8_BIT_GRAPHIC)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1013 {
34761
b4244c3295bb (Fmake_char_internal): If CHARSET doesn't have a
Kenichi Handa <handa@m17n.org>
parents: 33629
diff changeset
1014 if (NILP (code1))
b4244c3295bb (Fmake_char_internal): If CHARSET doesn't have a
Kenichi Handa <handa@m17n.org>
parents: 33629
diff changeset
1015 c1 = 0xA0;
b4244c3295bb (Fmake_char_internal): If CHARSET doesn't have a
Kenichi Handa <handa@m17n.org>
parents: 33629
diff changeset
1016 else if (c1 < 0xA0 || c1 > 0xFF)
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1017 goto invalid_code_posints;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1018 return make_number (c1);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1019 }
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1020 else if (c1 < 0 || c1 > 0xFF || c2 < 0 || c2 > 0xFF)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1021 goto invalid_code_posints;
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1022 c1 &= 0x7F;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1023 c2 &= 0x7F;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1024 if (c1 == 0
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1025 ? c2 != 0
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1026 : (c2 == 0
28515
75bd73426842 * charset.c (Fmake_char_internal): CHAR_COMPONENTS_VALID_P takes a
Ken Raeburn <raeburn@raeburn.org>
parents: 27791
diff changeset
1027 ? !CHAR_COMPONENTS_VALID_P (charset_id, c1, 0x20)
75bd73426842 * charset.c (Fmake_char_internal): CHAR_COMPONENTS_VALID_P takes a
Ken Raeburn <raeburn@raeburn.org>
parents: 27791
diff changeset
1028 : !CHAR_COMPONENTS_VALID_P (charset_id, c1, c2)))
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1029 goto invalid_code_posints;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1030 return make_number (MAKE_CHAR (charset_id, c1, c2));
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1031
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1032 invalid_code_posints:
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1033 error ("Invalid code points for charset ID %d: %d %d", charset_id, c1, c2);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1034 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1035
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1036 DEFUN ("split-char", Fsplit_char, Ssplit_char, 1, 1, 0,
55505
7abf0e14a59e (Fget_unused_iso_final_char): Fix typos in docstring.
Juanma Barranquero <lekktu@gmail.com>
parents: 53528
diff changeset
1037 doc: /* Return list of charset and one or two position-codes of CH.
7abf0e14a59e (Fget_unused_iso_final_char): Fix typos in docstring.
Juanma Barranquero <lekktu@gmail.com>
parents: 53528
diff changeset
1038 If CH is invalid as a character code,
7abf0e14a59e (Fget_unused_iso_final_char): Fix typos in docstring.
Juanma Barranquero <lekktu@gmail.com>
parents: 53528
diff changeset
1039 return a list of symbol `unknown' and CH. */)
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1040 (ch)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1041 Lisp_Object ch;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1042 {
23883
0ca2fa58ca7b (Qunknown): New variable.
Kenichi Handa <handa@m17n.org>
parents: 23767
diff changeset
1043 int c, charset, c1, c2;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1044
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
1045 CHECK_NUMBER (ch);
23883
0ca2fa58ca7b (Qunknown): New variable.
Kenichi Handa <handa@m17n.org>
parents: 23767
diff changeset
1046 c = XFASTINT (ch);
0ca2fa58ca7b (Qunknown): New variable.
Kenichi Handa <handa@m17n.org>
parents: 23767
diff changeset
1047 if (!CHAR_VALID_P (c, 1))
0ca2fa58ca7b (Qunknown): New variable.
Kenichi Handa <handa@m17n.org>
parents: 23767
diff changeset
1048 return Fcons (Qunknown, Fcons (ch, Qnil));
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1049 SPLIT_CHAR (XFASTINT (ch), charset, c1, c2);
17322
a7d9b8e167ca (non_ascii_char_to_string, update_charset_table,
Kenichi Handa <handa@m17n.org>
parents: 17186
diff changeset
1050 return (c2 >= 0
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1051 ? Fcons (CHARSET_SYMBOL (charset),
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1052 Fcons (make_number (c1), Fcons (make_number (c2), Qnil)))
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1053 : Fcons (CHARSET_SYMBOL (charset), Fcons (make_number (c1), Qnil)));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1054 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1055
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1056 DEFUN ("char-charset", Fchar_charset, Schar_charset, 1, 1, 0,
55505
7abf0e14a59e (Fget_unused_iso_final_char): Fix typos in docstring.
Juanma Barranquero <lekktu@gmail.com>
parents: 53528
diff changeset
1057 doc: /* Return charset of CH. */)
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1058 (ch)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1059 Lisp_Object ch;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1060 {
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
1061 CHECK_NUMBER (ch);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1062
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1063 return CHARSET_SYMBOL (CHAR_CHARSET (XINT (ch)));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1064 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1065
21445
4c0b4a1025cd (string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents: 21419
diff changeset
1066 DEFUN ("charset-after", Fcharset_after, Scharset_after, 0, 1, 0,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1067 doc: /* Return charset of a character in the current buffer at position POS.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1068 If POS is nil, it defauls to the current point.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1069 If POS is out of range, the value is nil. */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1070 (pos)
21445
4c0b4a1025cd (string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents: 21419
diff changeset
1071 Lisp_Object pos;
4c0b4a1025cd (string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents: 21419
diff changeset
1072 {
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1073 Lisp_Object ch;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1074 int charset;
21445
4c0b4a1025cd (string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents: 21419
diff changeset
1075
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1076 ch = Fchar_after (pos);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1077 if (! INTEGERP (ch))
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1078 return ch;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1079 charset = CHAR_CHARSET (XINT (ch));
21445
4c0b4a1025cd (string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents: 21419
diff changeset
1080 return CHARSET_SYMBOL (charset);
4c0b4a1025cd (string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents: 21419
diff changeset
1081 }
4c0b4a1025cd (string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents: 21419
diff changeset
1082
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1083 DEFUN ("iso-charset", Fiso_charset, Siso_charset, 3, 3, 0,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1084 doc: /* Return charset of ISO's specification DIMENSION, CHARS, and FINAL-CHAR.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1085
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1086 ISO 2022's designation sequence (escape sequence) distinguishes charsets
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1087 by their DIMENSION, CHARS, and FINAL-CHAR,
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1088 where as Emacs distinguishes them by charset symbol.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1089 See the documentation of the function `charset-info' for the meanings of
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1090 DIMENSION, CHARS, and FINAL-CHAR. */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1091 (dimension, chars, final_char)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1092 Lisp_Object dimension, chars, final_char;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1093 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1094 int charset;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1095
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
1096 CHECK_NUMBER (dimension);
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
1097 CHECK_NUMBER (chars);
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
1098 CHECK_NUMBER (final_char);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1099
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1100 if ((charset = ISO_CHARSET_TABLE (dimension, chars, final_char)) < 0)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1101 return Qnil;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1102 return CHARSET_SYMBOL (charset);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1103 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1104
78501
1677cf1c2509 Replace `iff' in comments.
Glenn Morris <rgm@gnu.org>
parents: 78313
diff changeset
1105 /* If GENERICP is nonzero, return nonzero if C is a valid normal or
1677cf1c2509 Replace `iff' in comments.
Glenn Morris <rgm@gnu.org>
parents: 78313
diff changeset
1106 generic character. If GENERICP is zero, return nonzero if C is a
20152
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1107 valid normal character. Do not call this function directly,
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1108 instead use macro CHAR_VALID_P. */
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1109 int
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1110 char_valid_p (c, genericp)
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1111 int c, genericp;
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1112 {
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1113 int charset, c1, c2;
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1114
31175
738330ba020f (char_to_string): Check the character validity.
Kenichi Handa <handa@m17n.org>
parents: 30673
diff changeset
1115 if (c < 0 || c >= MAX_CHAR)
20152
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1116 return 0;
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1117 if (SINGLE_BYTE_CHAR_P (c))
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1118 return 1;
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1119 SPLIT_CHAR (c, charset, c1, c2);
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1120 if (genericp)
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1121 {
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1122 if (c1)
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1123 {
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1124 if (c2 <= 0) c2 = 0x20;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1125 }
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1126 else
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1127 {
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1128 if (c2 <= 0) c1 = c2 = 0x20;
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1129 }
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1130 }
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1131 return (CHARSET_DEFINED_P (charset)
25584
01026ea61dcc Lots of comments fixed.
Kenichi Handa <handa@m17n.org>
parents: 25506
diff changeset
1132 && CHAR_COMPONENTS_VALID_P (charset, c1, c2));
20152
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1133 }
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1134
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1135 DEFUN ("char-valid-p", Fchar_valid_p, Schar_valid_p, 1, 2, 0,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1136 doc: /* Return t if OBJECT is a valid normal character.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1137 If optional arg GENERICP is non-nil, also return t if OBJECT is
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1138 a valid generic character. */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1139 (object, genericp)
20152
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1140 Lisp_Object object, genericp;
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1141 {
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1142 if (! NATNUMP (object))
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1143 return Qnil;
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1144 return (CHAR_VALID_P (XFASTINT (object), !NILP (genericp)) ? Qt : Qnil);
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1145 }
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1146
21034
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1147 DEFUN ("unibyte-char-to-multibyte", Funibyte_char_to_multibyte,
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1148 Sunibyte_char_to_multibyte, 1, 1, 0,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1149 doc: /* Convert the unibyte character CH to multibyte character.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1150 The conversion is done based on `nonascii-translation-table' (which see)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1151 or `nonascii-insert-offset' (which see). */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1152 (ch)
21034
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1153 Lisp_Object ch;
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1154 {
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1155 int c;
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1156
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
1157 CHECK_NUMBER (ch);
21034
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1158 c = XINT (ch);
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1159 if (c < 0 || c >= 0400)
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1160 error ("Invalid unibyte character: %d", c);
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1161 c = unibyte_char_to_multibyte (c);
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1162 if (c < 0)
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1163 error ("Can't convert to multibyte character: %d", XINT (ch));
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1164 return make_number (c);
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1165 }
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1166
23060
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1167 DEFUN ("multibyte-char-to-unibyte", Fmultibyte_char_to_unibyte,
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1168 Smultibyte_char_to_unibyte, 1, 1, 0,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1169 doc: /* Convert the multibyte character CH to unibyte character.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1170 The conversion is done based on `nonascii-translation-table' (which see)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1171 or `nonascii-insert-offset' (which see). */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1172 (ch)
23060
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1173 Lisp_Object ch;
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1174 {
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1175 int c;
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1176
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
1177 CHECK_NUMBER (ch);
23060
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1178 c = XINT (ch);
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1179 if (! CHAR_VALID_P (c, 0))
23060
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1180 error ("Invalid multibyte character: %d", c);
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1181 c = multibyte_char_to_unibyte (c, Qnil);
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1182 if (c < 0)
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1183 error ("Can't convert to unibyte character: %d", XINT (ch));
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1184 return make_number (c);
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1185 }
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1186
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1187 DEFUN ("char-bytes", Fchar_bytes, Schar_bytes, 1, 1, 0,
55505
7abf0e14a59e (Fget_unused_iso_final_char): Fix typos in docstring.
Juanma Barranquero <lekktu@gmail.com>
parents: 53528
diff changeset
1188 doc: /* Return 1 regardless of the argument CH. */)
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1189 (ch)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1190 Lisp_Object ch;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1191 {
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
1192 CHECK_NUMBER (ch);
23131
f7e486faa1ad (Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents: 23060
diff changeset
1193 return make_number (1);
f7e486faa1ad (Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents: 23060
diff changeset
1194 }
f7e486faa1ad (Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents: 23060
diff changeset
1195
f7e486faa1ad (Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents: 23060
diff changeset
1196 /* Return how many bytes C will occupy in a multibyte buffer.
f7e486faa1ad (Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents: 23060
diff changeset
1197 Don't call this function directly, instead use macro CHAR_BYTES. */
f7e486faa1ad (Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents: 23060
diff changeset
1198 int
f7e486faa1ad (Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents: 23060
diff changeset
1199 char_bytes (c)
f7e486faa1ad (Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents: 23060
diff changeset
1200 int c;
f7e486faa1ad (Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents: 23060
diff changeset
1201 {
26844
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
1202 int charset;
23131
f7e486faa1ad (Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents: 23060
diff changeset
1203
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1204 if (ASCII_BYTE_P (c) || (c & ~((1 << CHARACTERBITS) -1)))
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1205 return 1;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1206 if (SINGLE_BYTE_CHAR_P (c) && c >= 0xA0)
23767
1dfb6d38354f (non_ascii_char_to_string): If C has modifier bits,
Kenichi Handa <handa@m17n.org>
parents: 23720
diff changeset
1207 return 1;
1dfb6d38354f (non_ascii_char_to_string): If C has modifier bits,
Kenichi Handa <handa@m17n.org>
parents: 23720
diff changeset
1208
26844
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
1209 charset = CHAR_CHARSET (c);
9e9a0735c824 In this entry, just `Modified' means that codes for a
Kenichi Handa <handa@m17n.org>
parents: 26632
diff changeset
1210 return (CHARSET_DEFINED_P (charset) ? CHARSET_BYTES (charset) : 1);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1211 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1212
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1213 /* Return the width of character of which multi-byte form starts with
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1214 C. The width is measured by how many columns occupied on the
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1215 screen when displayed in the current buffer. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1216
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1217 #define ONE_BYTE_CHAR_WIDTH(c) \
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1218 (c < 0x20 \
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1219 ? (c == '\t' \
17366
f7809ff2c65f (ONE_BYTE_CHAR_WIDTH, Fconcat_chars): Don't use Lisp_Object as integer.
Karl Heuer <kwzh@gnu.org>
parents: 17322
diff changeset
1220 ? XFASTINT (current_buffer->tab_width) \
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1221 : (c == '\n' ? 0 : (NILP (current_buffer->ctl_arrow) ? 4 : 2))) \
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1222 : (c < 0x7f \
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1223 ? 1 \
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1224 : (c == 0x7F \
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1225 ? (NILP (current_buffer->ctl_arrow) ? 4 : 2) \
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1226 : ((! NILP (current_buffer->enable_multibyte_characters) \
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1227 && BASE_LEADING_CODE_P (c)) \
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1228 ? WIDTH_BY_CHAR_HEAD (c) \
22121
6e79a15594ab Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents: 21718
diff changeset
1229 : 4))))
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1230
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1231 DEFUN ("char-width", Fchar_width, Schar_width, 1, 1, 0,
55505
7abf0e14a59e (Fget_unused_iso_final_char): Fix typos in docstring.
Juanma Barranquero <lekktu@gmail.com>
parents: 53528
diff changeset
1232 doc: /* Return width of CH when displayed in the current buffer.
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1233 The width is measured by how many columns it occupies on the screen.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1234 Tab is taken to occupy `tab-width' columns. */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1235 (ch)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1236 Lisp_Object ch;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1237 {
17818
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1238 Lisp_Object val, disp;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1239 int c;
18537
6a726658dfa9 (Fchar_width): Check if the current buffer has display
Kenichi Handa <handa@m17n.org>
parents: 18341
diff changeset
1240 struct Lisp_Char_Table *dp = buffer_display_table ();
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1241
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
1242 CHECK_NUMBER (ch);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1243
17818
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1244 c = XINT (ch);
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1245
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1246 /* Get the way the display table would display it. */
18537
6a726658dfa9 (Fchar_width): Check if the current buffer has display
Kenichi Handa <handa@m17n.org>
parents: 18341
diff changeset
1247 disp = dp ? DISP_CHAR_VECTOR (dp, c) : Qnil;
17818
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1248
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1249 if (VECTORP (disp))
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1250 XSETINT (val, XVECTOR (disp)->size);
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1251 else if (SINGLE_BYTE_CHAR_P (c))
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1252 XSETINT (val, ONE_BYTE_CHAR_WIDTH (c));
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1253 else
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1254 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1255 int charset = CHAR_CHARSET (c);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1256
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1257 XSETFASTINT (val, CHARSET_WIDTH (charset));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1258 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1259 return val;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1260 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1261
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1262 /* Return width of string STR of length LEN when displayed in the
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1263 current buffer. The width is measured by how many columns it
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1264 occupies on the screen. */
17818
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1265
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1266 int
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1267 strwidth (str, len)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1268 unsigned char *str;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1269 int len;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1270 {
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1271 return c_string_width (str, len, -1, NULL, NULL);
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1272 }
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1273
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1274 /* Return width of string STR of length LEN when displayed in the
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1275 current buffer. The width is measured by how many columns it
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1276 occupies on the screen. If PRECISION > 0, return the width of
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1277 longest substring that doesn't exceed PRECISION, and set number of
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1278 characters and bytes of the substring in *NCHARS and *NBYTES
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1279 respectively. */
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1280
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1281 int
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1282 c_string_width (str, len, precision, nchars, nbytes)
46460
876356a4a4ca (find_charset_in_text, c_string_width):
Ken Raeburn <raeburn@raeburn.org>
parents: 46370
diff changeset
1283 const unsigned char *str;
48317
8ce6ddfd61d0 Declare all args (per C99).
Dave Love <fx@gnu.org>
parents: 46460
diff changeset
1284 int len, precision, *nchars, *nbytes;
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1285 {
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1286 int i = 0, i_byte = 0;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1287 int width = 0;
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1288 int chars;
20362
67bcb068070c (strwidth): Remove extra argument to buffer_display_table.
Kenichi Handa <handa@m17n.org>
parents: 20228
diff changeset
1289 struct Lisp_Char_Table *dp = buffer_display_table ();
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1290
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1291 while (i_byte < len)
17818
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1292 {
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1293 int bytes, thiswidth;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1294 Lisp_Object val;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1295
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1296 if (dp)
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1297 {
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1298 int c = STRING_CHAR_AND_LENGTH (str + i_byte, len - i_byte, bytes);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1299
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1300 chars = 1;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1301 val = DISP_CHAR_VECTOR (dp, c);
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1302 if (VECTORP (val))
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1303 thiswidth = XVECTOR (val)->size;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1304 else
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1305 thiswidth = ONE_BYTE_CHAR_WIDTH (str[i_byte]);
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1306 }
17818
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1307 else
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1308 {
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1309 chars = 1;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1310 PARSE_MULTIBYTE_SEQ (str + i_byte, len - i_byte, bytes);
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1311 thiswidth = ONE_BYTE_CHAR_WIDTH (str[i_byte]);
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1312 }
17818
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1313
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1314 if (precision > 0
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1315 && (width + thiswidth > precision))
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1316 {
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1317 *nchars = i;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1318 *nbytes = i_byte;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1319 return width;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1320 }
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1321 i++;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1322 i_byte += bytes;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1323 width += thiswidth;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1324 }
17818
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1325
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1326 if (precision > 0)
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1327 {
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1328 *nchars = i;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1329 *nbytes = i_byte;
17818
76c6740bd736 (strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents: 17727
diff changeset
1330 }
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1331
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1332 return width;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1333 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1334
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1335 /* Return width of Lisp string STRING when displayed in the current
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1336 buffer. The width is measured by how many columns it occupies on
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1337 the screen while paying attention to compositions. If PRECISION >
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1338 0, return the width of longest substring that doesn't exceed
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1339 PRECISION, and set number of characters and bytes of the substring
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1340 in *NCHARS and *NBYTES respectively. */
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1341
31420
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1342 int
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1343 lisp_string_width (string, precision, nchars, nbytes)
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1344 Lisp_Object string;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1345 int precision, *nchars, *nbytes;
31420
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1346 {
46370
40db0673e6f0 Most uses of XSTRING combined with STRING_BYTES or indirection changed to
Ken Raeburn <raeburn@raeburn.org>
parents: 46056
diff changeset
1347 int len = SCHARS (string);
40db0673e6f0 Most uses of XSTRING combined with STRING_BYTES or indirection changed to
Ken Raeburn <raeburn@raeburn.org>
parents: 46056
diff changeset
1348 int len_byte = SBYTES (string);
67841
53d14610ec35 (lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents: 67658
diff changeset
1349 /* This set multibyte to 0 even if STRING is multibyte when it
53d14610ec35 (lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents: 67658
diff changeset
1350 contains only ascii and eight-bit-graphic, but that's
53d14610ec35 (lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents: 67658
diff changeset
1351 intentional. */
53d14610ec35 (lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents: 67658
diff changeset
1352 int multibyte = len < len_byte;
46460
876356a4a4ca (find_charset_in_text, c_string_width):
Ken Raeburn <raeburn@raeburn.org>
parents: 46370
diff changeset
1353 const unsigned char *str = SDATA (string);
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1354 int i = 0, i_byte = 0;
31420
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1355 int width = 0;
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1356 struct Lisp_Char_Table *dp = buffer_display_table ();
31420
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1357
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1358 while (i < len)
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1359 {
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1360 int chars, bytes, thiswidth;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1361 Lisp_Object val;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1362 int cmp_id;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1363 int ignore, end;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1364
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1365 if (find_composition (i, -1, &ignore, &end, &val, string)
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1366 && ((cmp_id = get_composition_id (i, i_byte, end - i, val, string))
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1367 >= 0))
31420
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1368 {
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1369 thiswidth = composition_table[cmp_id]->width;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1370 chars = end - i;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1371 bytes = string_char_to_byte (string, end) - i_byte;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1372 }
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1373 else if (dp)
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1374 {
67841
53d14610ec35 (lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents: 67658
diff changeset
1375 int c;
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1376
67841
53d14610ec35 (lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents: 67658
diff changeset
1377 if (multibyte)
53d14610ec35 (lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents: 67658
diff changeset
1378 c = STRING_CHAR_AND_LENGTH (str + i_byte, len - i_byte, bytes);
53d14610ec35 (lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents: 67658
diff changeset
1379 else
53d14610ec35 (lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents: 67658
diff changeset
1380 c = str[i_byte], bytes = 1;
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1381 chars = 1;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1382 val = DISP_CHAR_VECTOR (dp, c);
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1383 if (VECTORP (val))
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1384 thiswidth = XVECTOR (val)->size;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1385 else
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1386 thiswidth = ONE_BYTE_CHAR_WIDTH (str[i_byte]);
31420
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1387 }
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1388 else
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1389 {
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1390 chars = 1;
67841
53d14610ec35 (lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents: 67658
diff changeset
1391 if (multibyte)
53d14610ec35 (lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents: 67658
diff changeset
1392 PARSE_MULTIBYTE_SEQ (str + i_byte, len_byte - i_byte, bytes);
53d14610ec35 (lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents: 67658
diff changeset
1393 else
53d14610ec35 (lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents: 67658
diff changeset
1394 bytes = 1;
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1395 thiswidth = ONE_BYTE_CHAR_WIDTH (str[i_byte]);
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1396 }
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1397
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1398 if (precision > 0
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1399 && (width + thiswidth > precision))
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1400 {
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1401 *nchars = i;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1402 *nbytes = i_byte;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1403 return width;
31420
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1404 }
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1405 i += chars;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1406 i_byte += bytes;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1407 width += thiswidth;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1408 }
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1409
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1410 if (precision > 0)
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1411 {
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1412 *nchars = i;
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1413 *nbytes = i_byte;
31420
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1414 }
35407
916718dc2950 (strwidth): Use c_string_width.
Kenichi Handa <handa@m17n.org>
parents: 34985
diff changeset
1415
31420
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1416 return width;
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1417 }
126c87fe1aad Include composite.h
Kenichi Handa <handa@m17n.org>
parents: 31176
diff changeset
1418
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1419 DEFUN ("string-width", Fstring_width, Sstring_width, 1, 1, 0,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1420 doc: /* Return width of STRING when displayed in the current buffer.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1421 Width is measured by how many columns it occupies on the screen.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1422 When calculating width of a multibyte character in STRING,
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1423 only the base leading-code is considered; the validity of
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1424 the following bytes is not checked. Tabs in STRING are always
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1425 taken to occupy `tab-width' columns. */)
55505
7abf0e14a59e (Fget_unused_iso_final_char): Fix typos in docstring.
Juanma Barranquero <lekktu@gmail.com>
parents: 53528
diff changeset
1426 (string)
7abf0e14a59e (Fget_unused_iso_final_char): Fix typos in docstring.
Juanma Barranquero <lekktu@gmail.com>
parents: 53528
diff changeset
1427 Lisp_Object string;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1428 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1429 Lisp_Object val;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1430
55505
7abf0e14a59e (Fget_unused_iso_final_char): Fix typos in docstring.
Juanma Barranquero <lekktu@gmail.com>
parents: 53528
diff changeset
1431 CHECK_STRING (string);
7abf0e14a59e (Fget_unused_iso_final_char): Fix typos in docstring.
Juanma Barranquero <lekktu@gmail.com>
parents: 53528
diff changeset
1432 XSETFASTINT (val, lisp_string_width (string, -1, NULL, NULL));
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1433 return val;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1434 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1435
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1436 DEFUN ("char-direction", Fchar_direction, Schar_direction, 1, 1, 0,
55505
7abf0e14a59e (Fget_unused_iso_final_char): Fix typos in docstring.
Juanma Barranquero <lekktu@gmail.com>
parents: 53528
diff changeset
1437 doc: /* Return the direction of CH.
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1438 The returned value is 0 for left-to-right and 1 for right-to-left. */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1439 (ch)
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1440 Lisp_Object ch;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1441 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1442 int charset;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1443
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
1444 CHECK_NUMBER (ch);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1445 charset = CHAR_CHARSET (XFASTINT (ch));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1446 if (!CHARSET_DEFINED_P (charset))
20933
5ceea9d50194 (invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents: 20890
diff changeset
1447 invalid_character (XINT (ch));
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1448 return CHARSET_TABLE_INFO (charset, CHARSET_DIRECTION_IDX);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1449 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1450
20614
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1451 /* Return the number of characters in the NBYTES bytes at PTR.
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1452 This works by looking at the contents and checking for multibyte sequences.
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1453 However, if the current buffer has enable-multibyte-characters = nil,
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1454 we treat each byte as a character. */
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1455
20534
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
1456 int
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
1457 chars_in_text (ptr, nbytes)
46460
876356a4a4ca (find_charset_in_text, c_string_width):
Ken Raeburn <raeburn@raeburn.org>
parents: 46370
diff changeset
1458 const unsigned char *ptr;
20534
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
1459 int nbytes;
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
1460 {
20614
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1461 /* current_buffer is null at early stages of Emacs initialization. */
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1462 if (current_buffer == 0
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1463 || NILP (current_buffer->enable_multibyte_characters))
20534
066251bcf9f7 (Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents: 20435
diff changeset
1464 return nbytes;
20188
d4ad4463384d (Fchars_in_string): Pay attention to
Kenichi Handa <handa@m17n.org>
parents: 20152
diff changeset
1465
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1466 return multibyte_chars_in_text (ptr, nbytes);
19384
6ce37633c7f9 Many doc fixes.
Richard M. Stallman <rms@gnu.org>
parents: 19025
diff changeset
1467 }
6ce37633c7f9 Many doc fixes.
Richard M. Stallman <rms@gnu.org>
parents: 19025
diff changeset
1468
20614
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1469 /* Return the number of characters in the NBYTES bytes at PTR.
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1470 This works by looking at the contents and checking for multibyte sequences.
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1471 It ignores enable-multibyte-characters. */
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1472
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1473 int
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1474 multibyte_chars_in_text (ptr, nbytes)
46460
876356a4a4ca (find_charset_in_text, c_string_width):
Ken Raeburn <raeburn@raeburn.org>
parents: 46370
diff changeset
1475 const unsigned char *ptr;
20614
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1476 int nbytes;
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1477 {
46460
876356a4a4ca (find_charset_in_text, c_string_width):
Ken Raeburn <raeburn@raeburn.org>
parents: 46370
diff changeset
1478 const unsigned char *endp;
25506
ba0ad349e792 (SPLIT_COMPOSITE_SEQ): New macro.
Kenichi Handa <handa@m17n.org>
parents: 25250
diff changeset
1479 int chars, bytes;
20614
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1480
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1481 endp = ptr + nbytes;
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1482 chars = 0;
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1483
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1484 while (ptr < endp)
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1485 {
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1486 PARSE_MULTIBYTE_SEQ (ptr, endp - ptr, bytes);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1487 ptr += bytes;
20614
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1488 chars++;
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1489 }
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1490
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1491 return chars;
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1492 }
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1493
35478
340a9e4aeb29 comment fixes
Dave Love <fx@gnu.org>
parents: 35407
diff changeset
1494 /* Parse unibyte text at STR of LEN bytes as multibyte text, and
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1495 count the numbers of characters and bytes in it. On counting
35478
340a9e4aeb29 comment fixes
Dave Love <fx@gnu.org>
parents: 35407
diff changeset
1496 bytes, pay attention to the fact that 8-bit characters in the range
340a9e4aeb29 comment fixes
Dave Love <fx@gnu.org>
parents: 35407
diff changeset
1497 0x80..0x9F are represented by 2 bytes in multibyte text. */
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1498 void
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1499 parse_str_as_multibyte (str, len, nchars, nbytes)
46460
876356a4a4ca (find_charset_in_text, c_string_width):
Ken Raeburn <raeburn@raeburn.org>
parents: 46370
diff changeset
1500 const unsigned char *str;
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1501 int len, *nchars, *nbytes;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1502 {
46460
876356a4a4ca (find_charset_in_text, c_string_width):
Ken Raeburn <raeburn@raeburn.org>
parents: 46370
diff changeset
1503 const unsigned char *endp = str + len;
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1504 int n, chars = 0, bytes = 0;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1505
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1506 while (str < endp)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1507 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1508 if (UNIBYTE_STR_AS_MULTIBYTE_P (str, endp - str, n))
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1509 str += n, bytes += n;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1510 else
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1511 str++, bytes += 2;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1512 chars++;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1513 }
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1514 *nchars = chars;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1515 *nbytes = bytes;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1516 return;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1517 }
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1518
35478
340a9e4aeb29 comment fixes
Dave Love <fx@gnu.org>
parents: 35407
diff changeset
1519 /* Arrange unibyte text at STR of NBYTES bytes as multibyte text.
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1520 It actually converts only 8-bit characters in the range 0x80..0x9F
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1521 that don't contruct multibyte characters to multibyte forms. If
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1522 NCHARS is nonzero, set *NCHARS to the number of characters in the
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1523 text. It is assured that we can use LEN bytes at STR as a work
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1524 area and that is enough. Return the number of bytes of the
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1525 resulting text. */
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1526
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1527 int
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1528 str_as_multibyte (str, len, nbytes, nchars)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1529 unsigned char *str;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1530 int len, nbytes, *nchars;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1531 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1532 unsigned char *p = str, *endp = str + nbytes;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1533 unsigned char *to;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1534 int chars = 0;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1535 int n;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1536
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1537 while (p < endp && UNIBYTE_STR_AS_MULTIBYTE_P (p, endp - p, n))
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1538 p += n, chars++;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1539 if (nchars)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1540 *nchars = chars;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1541 if (p == endp)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1542 return nbytes;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1543
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1544 to = p;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1545 nbytes = endp - p;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1546 endp = str + len;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1547 safe_bcopy (p, endp - nbytes, nbytes);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1548 p = endp - nbytes;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1549 while (p < endp)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1550 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1551 if (UNIBYTE_STR_AS_MULTIBYTE_P (p, endp - p, n))
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1552 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1553 while (n--)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1554 *to++ = *p++;
49600
23a1cea22d13 Trailing whitespace deleted.
Juanma Barranquero <lekktu@gmail.com>
parents: 48317
diff changeset
1555 }
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1556 else
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1557 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1558 *to++ = LEADING_CODE_8_BIT_CONTROL;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1559 *to++ = *p++ + 0x20;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1560 }
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1561 chars++;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1562 }
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1563 if (nchars)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1564 *nchars = chars;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1565 return (to - str);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1566 }
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1567
35950
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1568 /* Parse unibyte string at STR of LEN bytes, and return the number of
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1569 bytes it may ocupy when converted to multibyte string by
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1570 `str_to_multibyte'. */
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1571
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1572 int
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1573 parse_str_to_multibyte (str, len)
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1574 unsigned char *str;
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1575 int len;
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1576 {
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1577 unsigned char *endp = str + len;
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1578 int bytes;
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1579
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1580 for (bytes = 0; str < endp; str++)
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1581 bytes += (*str < 0x80 || *str >= 0xA0) ? 1 : 2;
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1582 return bytes;
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1583 }
d2c753cf5c34 (parse_str_to_multibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 35478
diff changeset
1584
35478
340a9e4aeb29 comment fixes
Dave Love <fx@gnu.org>
parents: 35407
diff changeset
1585 /* Convert unibyte text at STR of NBYTES bytes to multibyte text
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1586 that contains the same single-byte characters. It actually
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1587 converts all 8-bit characters to multibyte forms. It is assured
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1588 that we can use LEN bytes at STR as a work area and that is
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1589 enough. */
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1590
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1591 int
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1592 str_to_multibyte (str, len, bytes)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1593 unsigned char *str;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1594 int len, bytes;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1595 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1596 unsigned char *p = str, *endp = str + bytes;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1597 unsigned char *to;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1598
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1599 while (p < endp && (*p < 0x80 || *p >= 0xA0)) p++;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1600 if (p == endp)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1601 return bytes;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1602 to = p;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1603 bytes = endp - p;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1604 endp = str + len;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1605 safe_bcopy (p, endp - bytes, bytes);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1606 p = endp - bytes;
49600
23a1cea22d13 Trailing whitespace deleted.
Juanma Barranquero <lekktu@gmail.com>
parents: 48317
diff changeset
1607 while (p < endp)
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1608 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1609 if (*p < 0x80 || *p >= 0xA0)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1610 *to++ = *p++;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1611 else
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1612 *to++ = LEADING_CODE_8_BIT_CONTROL, *to++ = *p++ + 0x20;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1613 }
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1614 return (to - str);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1615 }
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1616
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1617 /* Arrange multibyte text at STR of LEN bytes as a unibyte text. It
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1618 actually converts only 8-bit characters in the range 0x80..0x9F to
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1619 unibyte forms. */
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1620
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1621 int
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1622 str_as_unibyte (str, bytes)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1623 unsigned char *str;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1624 int bytes;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1625 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1626 unsigned char *p = str, *endp = str + bytes;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1627 unsigned char *to = str;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1628
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1629 while (p < endp && *p != LEADING_CODE_8_BIT_CONTROL) p++;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1630 to = p;
49600
23a1cea22d13 Trailing whitespace deleted.
Juanma Barranquero <lekktu@gmail.com>
parents: 48317
diff changeset
1631 while (p < endp)
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1632 {
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1633 if (*p == LEADING_CODE_8_BIT_CONTROL)
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1634 *to++ = *(p + 1) - 0x20, p += 2;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1635 else
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1636 *to++ = *p++;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1637 }
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1638 return (to - str);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1639 }
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1640
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1641
44322
fca902769bf4 (Fstring): Allow 0 arguments.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 43713
diff changeset
1642 DEFUN ("string", Fstring, Sstring, 0, MANY, 0,
40115
78a7615cd863 (Fstring): Add usage: string to doc string.
Miles Bader <miles@gnu.org>
parents: 40103
diff changeset
1643 doc: /* Concatenate all the argument characters and make the result a string.
78a7615cd863 (Fstring): Add usage: string to doc string.
Miles Bader <miles@gnu.org>
parents: 40103
diff changeset
1644 usage: (string &rest CHARACTERS) */)
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1645 (n, args)
17366
f7809ff2c65f (ONE_BYTE_CHAR_WIDTH, Fconcat_chars): Don't use Lisp_Object as integer.
Karl Heuer <kwzh@gnu.org>
parents: 17322
diff changeset
1646 int n;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1647 Lisp_Object *args;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1648 {
56190
f7711455b84f (Fstring): Use SAFE_ALLOCA.
Kim F. Storm <storm@cua.dk>
parents: 55505
diff changeset
1649 int i, bufsize;
f7711455b84f (Fstring): Use SAFE_ALLOCA.
Kim F. Storm <storm@cua.dk>
parents: 55505
diff changeset
1650 unsigned char *buf, *p;
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1651 int c;
35997
d04b544d2a18 (Fstring): If all arguments are less than 256, return a unibyte
Kenichi Handa <handa@m17n.org>
parents: 35950
diff changeset
1652 int multibyte = 0;
56190
f7711455b84f (Fstring): Use SAFE_ALLOCA.
Kim F. Storm <storm@cua.dk>
parents: 55505
diff changeset
1653 Lisp_Object ret;
f7711455b84f (Fstring): Use SAFE_ALLOCA.
Kim F. Storm <storm@cua.dk>
parents: 55505
diff changeset
1654 USE_SAFE_ALLOCA;
f7711455b84f (Fstring): Use SAFE_ALLOCA.
Kim F. Storm <storm@cua.dk>
parents: 55505
diff changeset
1655
f7711455b84f (Fstring): Use SAFE_ALLOCA.
Kim F. Storm <storm@cua.dk>
parents: 55505
diff changeset
1656 bufsize = MAX_MULTIBYTE_LENGTH * n;
f7711455b84f (Fstring): Use SAFE_ALLOCA.
Kim F. Storm <storm@cua.dk>
parents: 55505
diff changeset
1657 SAFE_ALLOCA (buf, unsigned char *, bufsize);
f7711455b84f (Fstring): Use SAFE_ALLOCA.
Kim F. Storm <storm@cua.dk>
parents: 55505
diff changeset
1658 p = buf;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1659
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1660 for (i = 0; i < n; i++)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1661 {
40656
cdfd4d09b79a Update usage of CHECK_ macros (remove unused second argument).
Pavel Janík <Pavel@Janik.cz>
parents: 40115
diff changeset
1662 CHECK_NUMBER (args[i]);
35997
d04b544d2a18 (Fstring): If all arguments are less than 256, return a unibyte
Kenichi Handa <handa@m17n.org>
parents: 35950
diff changeset
1663 if (!multibyte && !SINGLE_BYTE_CHAR_P (XFASTINT (args[i])))
d04b544d2a18 (Fstring): If all arguments are less than 256, return a unibyte
Kenichi Handa <handa@m17n.org>
parents: 35950
diff changeset
1664 multibyte = 1;
d04b544d2a18 (Fstring): If all arguments are less than 256, return a unibyte
Kenichi Handa <handa@m17n.org>
parents: 35950
diff changeset
1665 }
d04b544d2a18 (Fstring): If all arguments are less than 256, return a unibyte
Kenichi Handa <handa@m17n.org>
parents: 35950
diff changeset
1666
d04b544d2a18 (Fstring): If all arguments are less than 256, return a unibyte
Kenichi Handa <handa@m17n.org>
parents: 35950
diff changeset
1667 for (i = 0; i < n; i++)
d04b544d2a18 (Fstring): If all arguments are less than 256, return a unibyte
Kenichi Handa <handa@m17n.org>
parents: 35950
diff changeset
1668 {
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1669 c = XINT (args[i]);
35997
d04b544d2a18 (Fstring): If all arguments are less than 256, return a unibyte
Kenichi Handa <handa@m17n.org>
parents: 35950
diff changeset
1670 if (multibyte)
d04b544d2a18 (Fstring): If all arguments are less than 256, return a unibyte
Kenichi Handa <handa@m17n.org>
parents: 35950
diff changeset
1671 p += CHAR_STRING (c, p);
d04b544d2a18 (Fstring): If all arguments are less than 256, return a unibyte
Kenichi Handa <handa@m17n.org>
parents: 35950
diff changeset
1672 else
35999
6b1e74f1fcce 2001-02-09 ShengHuo ZHU <zsh@cs.rochester.edu>
ShengHuo ZHU <zsh@cs.rochester.edu>
parents: 35997
diff changeset
1673 *p++ = c;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1674 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1675
56190
f7711455b84f (Fstring): Use SAFE_ALLOCA.
Kim F. Storm <storm@cua.dk>
parents: 55505
diff changeset
1676 ret = make_string_from_bytes (buf, n, p - buf);
57726
66e97a54985f Fix SAFE_FREE calls. Replace SAFE_FREE_LISP calls.
Kim F. Storm <storm@cua.dk>
parents: 56190
diff changeset
1677 SAFE_FREE ();
56190
f7711455b84f (Fstring): Use SAFE_ALLOCA.
Kim F. Storm <storm@cua.dk>
parents: 55505
diff changeset
1678
f7711455b84f (Fstring): Use SAFE_ALLOCA.
Kim F. Storm <storm@cua.dk>
parents: 55505
diff changeset
1679 return ret;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1680 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1681
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1682 #endif /* emacs */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1683
21514
fa9ff387d260 Fix -Wimplicit warnings.
Andreas Schwab <schwab@suse.de>
parents: 21445
diff changeset
1684 int
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1685 charset_id_internal (charset_name)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1686 char *charset_name;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1687 {
22927
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
1688 Lisp_Object val;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1689
22927
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
1690 val= Fget (intern (charset_name), Qcharset);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1691 if (!VECTORP (val))
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1692 error ("Charset %s is not defined", charset_name);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1693
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1694 return (XINT (XVECTOR (val)->contents[0]));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1695 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1696
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1697 DEFUN ("setup-special-charsets", Fsetup_special_charsets,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1698 Ssetup_special_charsets, 0, 0, 0, doc: /* Internal use only. */)
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1699 ()
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1700 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1701 charset_latin_iso8859_1 = charset_id_internal ("latin-iso8859-1");
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1702 charset_jisx0208_1978 = charset_id_internal ("japanese-jisx0208-1978");
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1703 charset_jisx0208 = charset_id_internal ("japanese-jisx0208");
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1704 charset_katakana_jisx0201 = charset_id_internal ("katakana-jisx0201");
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1705 charset_latin_jisx0201 = charset_id_internal ("latin-jisx0201");
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1706 charset_big5_1 = charset_id_internal ("chinese-big5-1");
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1707 charset_big5_2 = charset_id_internal ("chinese-big5-2");
66375
6680788d2e7f (charset_mule_unicode_0100_24ff)
Kenichi Handa <handa@m17n.org>
parents: 64770
diff changeset
1708 charset_mule_unicode_0100_24ff
6680788d2e7f (charset_mule_unicode_0100_24ff)
Kenichi Handa <handa@m17n.org>
parents: 64770
diff changeset
1709 = charset_id_internal ("mule-unicode-0100-24ff");
6680788d2e7f (charset_mule_unicode_0100_24ff)
Kenichi Handa <handa@m17n.org>
parents: 64770
diff changeset
1710 charset_mule_unicode_2500_33ff
6680788d2e7f (charset_mule_unicode_0100_24ff)
Kenichi Handa <handa@m17n.org>
parents: 64770
diff changeset
1711 = charset_id_internal ("mule-unicode-2500-33ff");
6680788d2e7f (charset_mule_unicode_0100_24ff)
Kenichi Handa <handa@m17n.org>
parents: 64770
diff changeset
1712 charset_mule_unicode_e000_ffff
6680788d2e7f (charset_mule_unicode_0100_24ff)
Kenichi Handa <handa@m17n.org>
parents: 64770
diff changeset
1713 = charset_id_internal ("mule-unicode-e000-ffff");
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1714 return Qnil;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1715 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1716
21514
fa9ff387d260 Fix -Wimplicit warnings.
Andreas Schwab <schwab@suse.de>
parents: 21445
diff changeset
1717 void
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1718 init_charset_once ()
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1719 {
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1720 int i, j, k;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1721
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1722 staticpro (&Vcharset_table);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1723 staticpro (&Vcharset_symbol_table);
22703
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1724 staticpro (&Vgeneric_character_list);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1725
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1726 /* This has to be done here, before we call Fmake_char_table. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1727 Qcharset_table = intern ("charset-table");
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1728 staticpro (&Qcharset_table);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1729
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1730 /* Intern this now in case it isn't already done.
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1731 Setting this variable twice is harmless.
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1732 But don't staticpro it here--that is done in alloc.c. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1733 Qchar_table_extra_slots = intern ("char-table-extra-slots");
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1734
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1735 /* Now we are ready to set up this property, so we can
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1736 create the charset table. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1737 Fput (Qcharset_table, Qchar_table_extra_slots, make_number (0));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1738 Vcharset_table = Fmake_char_table (Qcharset_table, Qnil);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1739
23883
0ca2fa58ca7b (Qunknown): New variable.
Kenichi Handa <handa@m17n.org>
parents: 23767
diff changeset
1740 Qunknown = intern ("unknown");
0ca2fa58ca7b (Qunknown): New variable.
Kenichi Handa <handa@m17n.org>
parents: 23767
diff changeset
1741 staticpro (&Qunknown);
0ca2fa58ca7b (Qunknown): New variable.
Kenichi Handa <handa@m17n.org>
parents: 23767
diff changeset
1742 Vcharset_symbol_table = Fmake_vector (make_number (MAX_CHARSET + 1),
0ca2fa58ca7b (Qunknown): New variable.
Kenichi Handa <handa@m17n.org>
parents: 23767
diff changeset
1743 Qunknown);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1744
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1745 /* Setup tables. */
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1746 for (i = 0; i < 2; i++)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1747 for (j = 0; j < 2; j++)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1748 for (k = 0; k < 128; k++)
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1749 iso_charset_table [i][j][k] = -1;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1750
23656
031653c4634a (init_charset_once): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents: 23654
diff changeset
1751 for (i = 0; i < 256; i++)
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1752 bytes_by_char_head[i] = 1;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1753 bytes_by_char_head[LEADING_CODE_PRIVATE_11] = 3;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1754 bytes_by_char_head[LEADING_CODE_PRIVATE_12] = 3;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1755 bytes_by_char_head[LEADING_CODE_PRIVATE_21] = 4;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1756 bytes_by_char_head[LEADING_CODE_PRIVATE_22] = 4;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1757
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1758 for (i = 0; i < 128; i++)
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1759 width_by_char_head[i] = 1;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1760 for (; i < 256; i++)
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1761 width_by_char_head[i] = 4;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1762 width_by_char_head[LEADING_CODE_PRIVATE_11] = 1;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1763 width_by_char_head[LEADING_CODE_PRIVATE_12] = 2;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1764 width_by_char_head[LEADING_CODE_PRIVATE_21] = 1;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1765 width_by_char_head[LEADING_CODE_PRIVATE_22] = 2;
22703
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1766
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1767 {
22927
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
1768 Lisp_Object val;
22703
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1769
22927
209acf11198c (multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents: 22774
diff changeset
1770 val = Qnil;
22703
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1771 for (i = 0x81; i < 0x90; i++)
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1772 val = Fcons (make_number ((i - 0x70) << 7), val);
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1773 for (; i < 0x9A; i++)
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1774 val = Fcons (make_number ((i - 0x8F) << 14), val);
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1775 for (i = 0xA0; i < 0xF0; i++)
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1776 val = Fcons (make_number ((i - 0x70) << 7), val);
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1777 for (; i < 0xFF; i++)
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1778 val = Fcons (make_number ((i - 0xE0) << 14), val);
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1779 Vgeneric_character_list = Fnreverse (val);
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1780 }
23154
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
1781
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
1782 nonascii_insert_offset = 0;
a9cbf64603ca (unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents: 23140
diff changeset
1783 Vnonascii_translation_table = Qnil;
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1784 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1785
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1786 #ifdef emacs
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1787
21514
fa9ff387d260 Fix -Wimplicit warnings.
Andreas Schwab <schwab@suse.de>
parents: 21445
diff changeset
1788 void
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1789 syms_of_charset ()
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1790 {
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1791 Qcharset = intern ("charset");
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1792 staticpro (&Qcharset);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1793
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1794 Qascii = intern ("ascii");
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1795 staticpro (&Qascii);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1796
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1797 Qeight_bit_control = intern ("eight-bit-control");
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1798 staticpro (&Qeight_bit_control);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1799
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1800 Qeight_bit_graphic = intern ("eight-bit-graphic");
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1801 staticpro (&Qeight_bit_graphic);
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1802
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1803 /* Define special charsets ascii, eight-bit-control, and
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1804 eight-bit-graphic. */
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1805 update_charset_table (make_number (CHARSET_ASCII),
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1806 make_number (1), make_number (94),
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1807 make_number (1),
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1808 make_number (0),
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1809 make_number ('B'),
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1810 make_number (0),
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1811 build_string ("ASCII"),
29139
c1c227f62493 (update_charset_table): Accept nil in LONG_NAME and
Kenichi Handa <handa@m17n.org>
parents: 29095
diff changeset
1812 Qnil, /* same as above */
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1813 build_string ("ASCII (ISO646 IRV)"));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1814 CHARSET_SYMBOL (CHARSET_ASCII) = Qascii;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1815 Fput (Qascii, Qcharset, CHARSET_TABLE_ENTRY (CHARSET_ASCII));
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1816
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1817 update_charset_table (make_number (CHARSET_8_BIT_CONTROL),
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1818 make_number (1), make_number (96),
29269
452835795e0f (syms_of_charset): Revert previous change.
Eli Zaretskii <eliz@gnu.org>
parents: 29257
diff changeset
1819 make_number (4),
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1820 make_number (0),
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1821 make_number (-1),
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1822 make_number (-1),
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1823 build_string ("8-bit control code (0x80..0x9F)"),
29139
c1c227f62493 (update_charset_table): Accept nil in LONG_NAME and
Kenichi Handa <handa@m17n.org>
parents: 29095
diff changeset
1824 Qnil, /* same as above */
c1c227f62493 (update_charset_table): Accept nil in LONG_NAME and
Kenichi Handa <handa@m17n.org>
parents: 29095
diff changeset
1825 Qnil); /* same as above */
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1826 CHARSET_SYMBOL (CHARSET_8_BIT_CONTROL) = Qeight_bit_control;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1827 Fput (Qeight_bit_control, Qcharset,
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1828 CHARSET_TABLE_ENTRY (CHARSET_8_BIT_CONTROL));
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1829
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1830 update_charset_table (make_number (CHARSET_8_BIT_GRAPHIC),
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1831 make_number (1), make_number (96),
29269
452835795e0f (syms_of_charset): Revert previous change.
Eli Zaretskii <eliz@gnu.org>
parents: 29257
diff changeset
1832 make_number (4),
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1833 make_number (0),
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1834 make_number (-1),
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1835 make_number (-1),
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1836 build_string ("8-bit graphic char (0xA0..0xFF)"),
29139
c1c227f62493 (update_charset_table): Accept nil in LONG_NAME and
Kenichi Handa <handa@m17n.org>
parents: 29095
diff changeset
1837 Qnil, /* same as above */
c1c227f62493 (update_charset_table): Accept nil in LONG_NAME and
Kenichi Handa <handa@m17n.org>
parents: 29095
diff changeset
1838 Qnil); /* same as above */
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1839 CHARSET_SYMBOL (CHARSET_8_BIT_GRAPHIC) = Qeight_bit_graphic;
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1840 Fput (Qeight_bit_graphic, Qcharset,
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1841 CHARSET_TABLE_ENTRY (CHARSET_8_BIT_GRAPHIC));
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1842
23489
09201cebfa21 (Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents: 23427
diff changeset
1843 Qauto_fill_chars = intern ("auto-fill-chars");
09201cebfa21 (Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents: 23427
diff changeset
1844 staticpro (&Qauto_fill_chars);
09201cebfa21 (Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents: 23427
diff changeset
1845 Fput (Qauto_fill_chars, Qchar_table_extra_slots, make_number (0));
09201cebfa21 (Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents: 23427
diff changeset
1846
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1847 defsubr (&Sdefine_charset);
22703
15e0119749cf (Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents: 22524
diff changeset
1848 defsubr (&Sgeneric_character_list);
20435
099812ea310f (Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents: 20362
diff changeset
1849 defsubr (&Sget_unused_iso_final_char);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1850 defsubr (&Sdeclare_equiv_charset);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1851 defsubr (&Sfind_charset_region);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1852 defsubr (&Sfind_charset_string);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1853 defsubr (&Smake_char_internal);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1854 defsubr (&Ssplit_char);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1855 defsubr (&Schar_charset);
21445
4c0b4a1025cd (string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents: 21419
diff changeset
1856 defsubr (&Scharset_after);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1857 defsubr (&Siso_charset);
20152
9a3343c36e83 (char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents: 19983
diff changeset
1858 defsubr (&Schar_valid_p);
21034
f3128abaf8aa (DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents: 20933
diff changeset
1859 defsubr (&Sunibyte_char_to_multibyte);
23060
5036ee3de0ac (Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents: 22927
diff changeset
1860 defsubr (&Smultibyte_char_to_unibyte);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1861 defsubr (&Schar_bytes);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1862 defsubr (&Schar_width);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1863 defsubr (&Sstring_width);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1864 defsubr (&Schar_direction);
20614
a4800f9842e4 (multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents: 20575
diff changeset
1865 defsubr (&Sstring);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1866 defsubr (&Ssetup_special_charsets);
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1867
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1868 DEFVAR_LISP ("charset-list", &Vcharset_list,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1869 doc: /* List of charsets ever defined. */);
29003
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1870 Vcharset_list = Fcons (Qascii, Fcons (Qeight_bit_control,
72eafb39ec65 (Qeight_bit_control, Qeight_bit_graphic): New
Kenichi Handa <handa@m17n.org>
parents: 28818
diff changeset
1871 Fcons (Qeight_bit_graphic, Qnil)));
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1872
22185
80a2aa51a6e1 Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents: 22168
diff changeset
1873 DEFVAR_LISP ("translation-table-vector", &Vtranslation_table_vector,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1874 doc: /* Vector of cons cell of a symbol and translation table ever defined.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1875 An ID of a translation table is an index of this vector. */);
22185
80a2aa51a6e1 Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents: 22168
diff changeset
1876 Vtranslation_table_vector = Fmake_vector (make_number (16), Qnil);
20720
f084bb8d535f (min_composite_char): New variable.
Kenichi Handa <handa@m17n.org>
parents: 20705
diff changeset
1877
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1878 DEFVAR_INT ("leading-code-private-11", &leading_code_private_11,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1879 doc: /* Leading-code of private TYPE9N charset of column-width 1. */);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1880 leading_code_private_11 = LEADING_CODE_PRIVATE_11;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1881
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1882 DEFVAR_INT ("leading-code-private-12", &leading_code_private_12,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1883 doc: /* Leading-code of private TYPE9N charset of column-width 2. */);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1884 leading_code_private_12 = LEADING_CODE_PRIVATE_12;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1885
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1886 DEFVAR_INT ("leading-code-private-21", &leading_code_private_21,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1887 doc: /* Leading-code of private TYPE9Nx9N charset of column-width 1. */);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1888 leading_code_private_21 = LEADING_CODE_PRIVATE_21;
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1889
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1890 DEFVAR_INT ("leading-code-private-22", &leading_code_private_22,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1891 doc: /* Leading-code of private TYPE9Nx9N charset of column-width 2. */);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1892 leading_code_private_22 = LEADING_CODE_PRIVATE_22;
20705
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
1893
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
1894 DEFVAR_INT ("nonascii-insert-offset", &nonascii_insert_offset,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1895 doc: /* Offset for converting non-ASCII unibyte codes 0240...0377 to multibyte.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1896 This is used for converting unibyte text to multibyte,
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1897 and for inserting character codes specified by number.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1898
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1899 This serves to convert a Latin-1 or similar 8-bit character code
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1900 to the corresponding Emacs multibyte character code.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1901 Typically the value should be (- (make-char CHARSET 0) 128),
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1902 for your choice of character set.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1903 If `nonascii-translation-table' is non-nil, it overrides this variable. */);
20705
a949dc65fbea (nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents: 20636
diff changeset
1904 nonascii_insert_offset = 0;
20720
f084bb8d535f (min_composite_char): New variable.
Kenichi Handa <handa@m17n.org>
parents: 20705
diff changeset
1905
22121
6e79a15594ab Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents: 21718
diff changeset
1906 DEFVAR_LISP ("nonascii-translation-table", &Vnonascii_translation_table,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1907 doc: /* Translation table to convert non-ASCII unibyte codes to multibyte.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1908 This is used for converting unibyte text to multibyte,
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1909 and for inserting character codes specified by number.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1910
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1911 Conversion is performed only when multibyte characters are enabled,
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1912 and it serves to convert a Latin-1 or similar 8-bit character code
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1913 to the corresponding Emacs character code.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1914
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1915 If this is nil, `nonascii-insert-offset' is used instead.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1916 See also the docstring of `make-translation-table'. */);
22121
6e79a15594ab Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents: 21718
diff changeset
1917 Vnonascii_translation_table = Qnil;
20890
c7cfd531cf2b (Vnonascii_translate_table): New variable.
Richard M. Stallman <rms@gnu.org>
parents: 20833
diff changeset
1918
23489
09201cebfa21 (Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents: 23427
diff changeset
1919 DEFVAR_LISP ("auto-fill-chars", &Vauto_fill_chars,
40103
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1920 doc: /* A char-table for characters which invoke auto-filling.
6b389fb978bc Change doc-string comments to `new style' [w/`doc:' keyword].
Pavel Janík <Pavel@Janik.cz>
parents: 39682
diff changeset
1921 Such characters have value t in this table. */);
23489
09201cebfa21 (Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents: 23427
diff changeset
1922 Vauto_fill_chars = Fmake_char_table (Qauto_fill_chars, Qnil);
23576
00ee2e0c8976 (Fchar_bytes): Doc fix.
Andreas Schwab <schwab@suse.de>
parents: 23489
diff changeset
1923 CHAR_TABLE_SET (Vauto_fill_chars, make_number (' '), Qt);
00ee2e0c8976 (Fchar_bytes): Doc fix.
Andreas Schwab <schwab@suse.de>
parents: 23489
diff changeset
1924 CHAR_TABLE_SET (Vauto_fill_chars, make_number ('\n'), Qt);
17052
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1925 }
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1926
d0d7b244b1d0 Initial revision
Karl Heuer <kwzh@gnu.org>
parents:
diff changeset
1927 #endif /* emacs */
52401
695cf19ef79e Add arch taglines
Miles Bader <miles@gnu.org>
parents: 49600
diff changeset
1928
695cf19ef79e Add arch taglines
Miles Bader <miles@gnu.org>
parents: 49600
diff changeset
1929 /* arch-tag: 66a89b8d-4c28-47d3-9ca1-56f78440d69f
695cf19ef79e Add arch taglines
Miles Bader <miles@gnu.org>
parents: 49600
diff changeset
1930 (do not change this comment) */