Mercurial > emacs
annotate src/charset.c @ 23696:ee5224740210
(BSD_PGRPS): Add #undef.
author | Karl Heuer <kwzh@gnu.org> |
---|---|
date | Wed, 11 Nov 1998 18:58:45 +0000 |
parents | 5f039c506e17 |
children | 3371f3bd2057 |
rev | line source |
---|---|
18341
33e78cc7f058
Change copyright notices.
Richard M. Stallman <rms@gnu.org>
parents:
18003
diff
changeset
|
1 /* Basic multilingual character support. |
20705
a949dc65fbea
(nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents:
20636
diff
changeset
|
2 Copyright (C) 1995, 1997, 1998 Electrotechnical Laboratory, JAPAN. |
18341
33e78cc7f058
Change copyright notices.
Richard M. Stallman <rms@gnu.org>
parents:
18003
diff
changeset
|
3 Licensed to the Free Software Foundation. |
17052 | 4 |
17071 | 5 This file is part of GNU Emacs. |
6 | |
7 GNU Emacs is free software; you can redistribute it and/or modify | |
8 it under the terms of the GNU General Public License as published by | |
9 the Free Software Foundation; either version 2, or (at your option) | |
10 any later version. | |
17052 | 11 |
17071 | 12 GNU Emacs is distributed in the hope that it will be useful, |
13 but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 GNU General Public License for more details. | |
17052 | 16 |
17071 | 17 You should have received a copy of the GNU General Public License |
18 along with GNU Emacs; see the file COPYING. If not, write to | |
19 the Free Software Foundation, Inc., 59 Temple Place - Suite 330, | |
20 Boston, MA 02111-1307, USA. */ | |
17052 | 21 |
22 /* At first, see the document in `charset.h' to understand the code in | |
23 this file. */ | |
24 | |
25 #include <stdio.h> | |
26 | |
27 #ifdef emacs | |
28 | |
29 #include <sys/types.h> | |
30 #include <config.h> | |
31 #include "lisp.h" | |
32 #include "buffer.h" | |
33 #include "charset.h" | |
34 #include "coding.h" | |
17843 | 35 #include "disptab.h" |
17052 | 36 |
37 #else /* not emacs */ | |
38 | |
39 #include "mulelib.h" | |
40 | |
41 #endif /* emacs */ | |
42 | |
43 Lisp_Object Qcharset, Qascii, Qcomposition; | |
44 | |
45 /* Declaration of special leading-codes. */ | |
46 int leading_code_composition; /* for composite characters */ | |
47 int leading_code_private_11; /* for private DIMENSION1 of 1-column */ | |
48 int leading_code_private_12; /* for private DIMENSION1 of 2-column */ | |
49 int leading_code_private_21; /* for private DIMENSION2 of 1-column */ | |
50 int leading_code_private_22; /* for private DIMENSION2 of 2-column */ | |
51 | |
52 /* Declaration of special charsets. */ | |
53 int charset_ascii; /* ASCII */ | |
54 int charset_composition; /* for a composite character */ | |
55 int charset_latin_iso8859_1; /* ISO8859-1 (Latin-1) */ | |
56 int charset_jisx0208_1978; /* JISX0208.1978 (Japanese Kanji old set) */ | |
57 int charset_jisx0208; /* JISX0208.1983 (Japanese Kanji) */ | |
58 int charset_katakana_jisx0201; /* JISX0201.Kana (Japanese Katakana) */ | |
59 int charset_latin_jisx0201; /* JISX0201.Roman (Japanese Roman) */ | |
60 int charset_big5_1; /* Big5 Level 1 (Chinese Traditional) */ | |
61 int charset_big5_2; /* Big5 Level 2 (Chinese Traditional) */ | |
62 | |
20720
f084bb8d535f
(min_composite_char): New variable.
Kenichi Handa <handa@m17n.org>
parents:
20705
diff
changeset
|
63 int min_composite_char; |
f084bb8d535f
(min_composite_char): New variable.
Kenichi Handa <handa@m17n.org>
parents:
20705
diff
changeset
|
64 |
17052 | 65 Lisp_Object Qcharset_table; |
66 | |
67 /* A char-table containing information of each character set. */ | |
68 Lisp_Object Vcharset_table; | |
69 | |
70 /* A vector of charset symbol indexed by charset-id. This is used | |
71 only for returning charset symbol from C functions. */ | |
72 Lisp_Object Vcharset_symbol_table; | |
73 | |
74 /* A list of charset symbols ever defined. */ | |
75 Lisp_Object Vcharset_list; | |
76 | |
22185
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
77 /* Vector of translation table ever defined. |
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
78 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
|
79 Lisp_Object Vtranslation_table_vector; |
20720
f084bb8d535f
(min_composite_char): New variable.
Kenichi Handa <handa@m17n.org>
parents:
20705
diff
changeset
|
80 |
23489
09201cebfa21
(Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents:
23427
diff
changeset
|
81 /* 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
|
82 Lisp_Object Vauto_fill_chars; |
09201cebfa21
(Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents:
23427
diff
changeset
|
83 |
09201cebfa21
(Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents:
23427
diff
changeset
|
84 Lisp_Object Qauto_fill_chars; |
09201cebfa21
(Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents:
23427
diff
changeset
|
85 |
17052 | 86 /* Tables used by macros BYTES_BY_CHAR_HEAD and WIDTH_BY_CHAR_HEAD. */ |
87 int bytes_by_char_head[256]; | |
88 int width_by_char_head[256]; | |
89 | |
90 /* Mapping table from ISO2022's charset (specified by DIMENSION, | |
91 CHARS, and FINAL-CHAR) to Emacs' charset. */ | |
92 int iso_charset_table[2][2][128]; | |
93 | |
17186
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
94 /* Table of pointers to the structure `cmpchar_info' indexed by |
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
95 CMPCHAR-ID. */ |
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
96 struct cmpchar_info **cmpchar_table; |
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
97 /* The current size of `cmpchar_table'. */ |
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
98 static int cmpchar_table_size; |
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
99 /* Number of the current composite characters. */ |
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
100 int n_cmpchars; |
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
101 |
17052 | 102 /* Variables used locally in the macro FETCH_MULTIBYTE_CHAR. */ |
103 unsigned char *_fetch_multibyte_char_p; | |
104 int _fetch_multibyte_char_len; | |
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. */ |
a949dc65fbea
(nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents:
20636
diff
changeset
|
107 int nonascii_insert_offset; |
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 | 116 #define min(X, Y) ((X) < (Y) ? (X) : (Y)) |
117 #define max(X, Y) ((X) > (Y) ? (X) : (Y)) | |
118 | |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
119 void |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
120 invalid_character (c) |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
121 int c; |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
122 { |
23397
2fe04027242e
(invalid_character): Prepend `0' to octal
Kenichi Handa <handa@m17n.org>
parents:
23210
diff
changeset
|
123 error ("Invalid character: 0%o, %d, 0x%x", c, c, c); |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
124 } |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
125 |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
126 |
17052 | 127 /* Set STR a pointer to the multi-byte form of the character C. If C |
128 is not a composite character, the multi-byte form is set in WORKBUF | |
129 and STR points WORKBUF. The caller should allocate at least 4-byte | |
130 area at WORKBUF in advance. Returns the length of the multi-byte | |
17834
e154b943bbba
(non_ascii_char_to_string): Signal error if the
Kenichi Handa <handa@m17n.org>
parents:
17818
diff
changeset
|
131 form. If C is an invalid character to have a multi-byte form, |
e154b943bbba
(non_ascii_char_to_string): Signal error if the
Kenichi Handa <handa@m17n.org>
parents:
17818
diff
changeset
|
132 signal an error. |
17052 | 133 |
134 Use macro `CHAR_STRING (C, WORKBUF, STR)' instead of calling this | |
135 function directly if C can be an ASCII character. */ | |
136 | |
137 int | |
138 non_ascii_char_to_string (c, workbuf, str) | |
139 int c; | |
140 unsigned char *workbuf, **str; | |
141 { | |
17322
a7d9b8e167ca
(non_ascii_char_to_string, update_charset_table,
Kenichi Handa <handa@m17n.org>
parents:
17186
diff
changeset
|
142 int charset, c1, c2; |
17052 | 143 |
144 if (COMPOSITE_CHAR_P (c)) | |
145 { | |
146 int cmpchar_id = COMPOSITE_CHAR_ID (c); | |
147 | |
148 if (cmpchar_id < n_cmpchars) | |
149 { | |
150 *str = cmpchar_table[cmpchar_id]->data; | |
151 return cmpchar_table[cmpchar_id]->len; | |
152 } | |
153 else | |
154 { | |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
155 invalid_character (c); |
17052 | 156 } |
157 } | |
158 | |
159 SPLIT_NON_ASCII_CHAR (c, charset, c1, c2); | |
17834
e154b943bbba
(non_ascii_char_to_string): Signal error if the
Kenichi Handa <handa@m17n.org>
parents:
17818
diff
changeset
|
160 if (!charset |
e154b943bbba
(non_ascii_char_to_string): Signal error if the
Kenichi Handa <handa@m17n.org>
parents:
17818
diff
changeset
|
161 || ! CHARSET_DEFINED_P (charset) |
e154b943bbba
(non_ascii_char_to_string): Signal error if the
Kenichi Handa <handa@m17n.org>
parents:
17818
diff
changeset
|
162 || c1 >= 0 && c1 < 32 |
e154b943bbba
(non_ascii_char_to_string): Signal error if the
Kenichi Handa <handa@m17n.org>
parents:
17818
diff
changeset
|
163 || c2 >= 0 && c2 < 32) |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
164 invalid_character (c); |
17052 | 165 |
166 *str = workbuf; | |
167 *workbuf++ = CHARSET_LEADING_CODE_BASE (charset); | |
168 if (*workbuf = CHARSET_LEADING_CODE_EXT (charset)) | |
169 workbuf++; | |
170 *workbuf++ = c1 | 0x80; | |
17322
a7d9b8e167ca
(non_ascii_char_to_string, update_charset_table,
Kenichi Handa <handa@m17n.org>
parents:
17186
diff
changeset
|
171 if (c2 >= 0) |
17052 | 172 *workbuf++ = c2 | 0x80; |
173 | |
174 return (workbuf - *str); | |
175 } | |
176 | |
177 /* Return a non-ASCII character of which multi-byte form is at STR of | |
178 length LEN. If ACTUAL_LEN is not NULL, the actual length of the | |
22185
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
179 multibyte form is set to the address ACTUAL_LEN. |
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
180 |
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
181 If exclude_tail_garbage is nonzero, ACTUAL_LEN excludes gabage |
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
182 bytes following the non-ASCII character. |
17052 | 183 |
184 Use macro `STRING_CHAR (STR, LEN)' instead of calling this function | |
185 directly if STR can hold an ASCII character. */ | |
186 | |
21514 | 187 int |
22185
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
188 string_to_non_ascii_char (str, len, actual_len, exclude_tail_garbage) |
21419
95aae2ff5fcd
(string_to_non_ascii_char, multibyte_form_length)
Karl Heuer <kwzh@gnu.org>
parents:
21255
diff
changeset
|
189 const unsigned char *str; |
22185
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
190 int len, *actual_len, exclude_tail_garbage; |
17052 | 191 { |
192 int charset; | |
193 unsigned char c1, c2; | |
23654
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
194 int c, bytes; |
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
195 const unsigned char *begp = str; |
17052 | 196 |
23654
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
197 c = *str++; |
21445
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
198 bytes = 1; |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
199 |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
200 if (BASE_LEADING_CODE_P (c)) |
23654
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
201 do { |
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
202 while (bytes < len && ! CHAR_HEAD_P (begp[bytes])) bytes++; |
21445
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
203 |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
204 if (c == LEADING_CODE_COMPOSITION) |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
205 { |
23654
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
206 int cmpchar_id = str_cmpchar_id (begp, bytes); |
21445
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
207 |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
208 if (cmpchar_id >= 0) |
22333
8d612a46dac6
(string_to_non_ascii_char): Don't check the arg
Kenichi Handa <handa@m17n.org>
parents:
22289
diff
changeset
|
209 { |
8d612a46dac6
(string_to_non_ascii_char): Don't check the arg
Kenichi Handa <handa@m17n.org>
parents:
22289
diff
changeset
|
210 c = MAKE_COMPOSITE_CHAR (cmpchar_id); |
23665
53553d8b253e
(string_to_non_ascii_char): Return correct length for
Kenichi Handa <handa@m17n.org>
parents:
23656
diff
changeset
|
211 str += cmpchar_table[cmpchar_id]->len - 1; |
22333
8d612a46dac6
(string_to_non_ascii_char): Don't check the arg
Kenichi Handa <handa@m17n.org>
parents:
22289
diff
changeset
|
212 } |
23665
53553d8b253e
(string_to_non_ascii_char): Return correct length for
Kenichi Handa <handa@m17n.org>
parents:
23656
diff
changeset
|
213 else |
53553d8b253e
(string_to_non_ascii_char): Return correct length for
Kenichi Handa <handa@m17n.org>
parents:
23656
diff
changeset
|
214 str += bytes - 1; |
21445
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
215 } |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
216 else |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
217 { |
23654
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
218 const unsigned char *endp = begp + bytes; |
21445
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
219 int charset = c, c1, c2 = 0; |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
220 |
23654
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
221 if (str >= endp) break; |
23649
9e74943ac418
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23628
diff
changeset
|
222 if (c >= LEADING_CODE_PRIVATE_11 && c <= LEADING_CODE_PRIVATE_22) |
21445
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
223 { |
23654
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
224 charset = *str++; |
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
225 if (str < endp) |
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
226 c1 = *str++ & 0x7F; |
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
227 else |
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
228 c1 = charset, charset = c; |
21445
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
229 } |
23654
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
230 else |
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
231 c1 = *str++ & 0x7f; |
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
232 if (CHARSET_DEFINED_P (charset) |
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
233 && CHARSET_DIMENSION (charset) == 2 |
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
234 && str < endp) |
23665
53553d8b253e
(string_to_non_ascii_char): Return correct length for
Kenichi Handa <handa@m17n.org>
parents:
23656
diff
changeset
|
235 c2 = *str++ & 0x7F; |
23654
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
236 c = MAKE_NON_ASCII_CHAR (charset, c1, c2); |
21445
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
237 } |
23654
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
238 } while (0); |
17052 | 239 |
240 if (actual_len) | |
23654
e70afb86651a
(string_to_non_ascii_char): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23649
diff
changeset
|
241 *actual_len = exclude_tail_garbage ? str - begp : bytes; |
17052 | 242 return c; |
243 } | |
244 | |
245 /* Return the length of the multi-byte form at string STR of length LEN. */ | |
246 int | |
247 multibyte_form_length (str, len) | |
21419
95aae2ff5fcd
(string_to_non_ascii_char, multibyte_form_length)
Karl Heuer <kwzh@gnu.org>
parents:
21255
diff
changeset
|
248 const unsigned char *str; |
17052 | 249 int len; |
250 { | |
21445
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
251 int bytes = 1; |
17052 | 252 |
21445
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
253 if (BASE_LEADING_CODE_P (*str)) |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
254 while (bytes < len && ! CHAR_HEAD_P (str[bytes])) bytes++; |
17052 | 255 |
21445
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
256 return bytes; |
17052 | 257 } |
258 | |
259 /* Check if string STR of length LEN contains valid multi-byte form of | |
260 a character. If valid, charset and position codes of the character | |
261 is set at *CHARSET, *C1, and *C2, and return 0. If not valid, | |
262 return -1. This should be used only in the macro SPLIT_STRING | |
263 which checks range of STR in advance. */ | |
264 | |
21514 | 265 int |
17052 | 266 split_non_ascii_string (str, len, charset, c1, c2) |
21419
95aae2ff5fcd
(string_to_non_ascii_char, multibyte_form_length)
Karl Heuer <kwzh@gnu.org>
parents:
21255
diff
changeset
|
267 register const unsigned char *str; |
95aae2ff5fcd
(string_to_non_ascii_char, multibyte_form_length)
Karl Heuer <kwzh@gnu.org>
parents:
21255
diff
changeset
|
268 register unsigned char *c1, *c2; |
17052 | 269 register int len, *charset; |
270 { | |
271 register unsigned int cs = *str++; | |
272 | |
273 if (cs == LEADING_CODE_COMPOSITION) | |
274 { | |
275 int cmpchar_id = str_cmpchar_id (str - 1, len); | |
276 | |
277 if (cmpchar_id < 0) | |
278 return -1; | |
279 *charset = cs, *c1 = cmpchar_id >> 7, *c2 = cmpchar_id & 0x7F; | |
280 } | |
281 else if ((cs < LEADING_CODE_PRIVATE_11 || (cs = *str++) >= 0xA0) | |
282 && CHARSET_DEFINED_P (cs)) | |
283 { | |
284 *charset = cs; | |
285 if (*str < 0xA0) | |
286 return -1; | |
287 *c1 = (*str++) & 0x7F; | |
288 if (CHARSET_DIMENSION (cs) == 2) | |
289 { | |
290 if (*str < 0xA0) | |
291 return -1; | |
292 *c2 = (*str++) & 0x7F; | |
293 } | |
294 } | |
295 else | |
296 return -1; | |
297 return 0; | |
298 } | |
299 | |
22185
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
300 /* 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
|
301 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
|
302 (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
|
303 found in TABLE, return C. */ |
21514 | 304 int |
22121
6e79a15594ab
Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents:
21718
diff
changeset
|
305 translate_char (table, c, charset, c1, c2) |
17727
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
306 Lisp_Object table; |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
307 int c, charset, c1, c2; |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
308 { |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
309 Lisp_Object ch; |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
310 int alt_charset, alt_c1, alt_c2, dimension; |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
311 |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
312 if (c < 0) c = MAKE_CHAR (charset, c1, c2); |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
313 if (!CHAR_TABLE_P (table) |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
314 || (ch = Faref (table, make_number (c)), !INTEGERP (ch)) |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
315 || XINT (ch) < 0) |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
316 return c; |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
317 |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
318 SPLIT_CHAR (XFASTINT (ch), alt_charset, alt_c1, alt_c2); |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
319 dimension = CHARSET_DIMENSION (alt_charset); |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
320 if (dimension == 1 && alt_c1 > 0 || dimension == 2 && alt_c2 > 0) |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
321 /* CH is not a generic character, just return it. */ |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
322 return XFASTINT (ch); |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
323 |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
324 /* 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
|
325 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
|
326 if (charset < 0) |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
327 SPLIT_CHAR (c, charset, c1, c2); |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
328 if (dimension != CHARSET_DIMENSION (charset)) |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
329 /* 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
|
330 return c; |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
331 return MAKE_CHAR (alt_charset, c1, c2); |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
332 } |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
333 |
21034
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
334 /* 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
|
335 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
|
336 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
|
337 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
|
338 |
21514 | 339 int |
20705
a949dc65fbea
(nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents:
20636
diff
changeset
|
340 unibyte_char_to_multibyte (c) |
a949dc65fbea
(nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents:
20636
diff
changeset
|
341 int c; |
a949dc65fbea
(nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents:
20636
diff
changeset
|
342 { |
23154
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
343 if (c < 0400) |
20705
a949dc65fbea
(nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents:
20636
diff
changeset
|
344 { |
21034
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
345 int c_save = c; |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
346 |
22121
6e79a15594ab
Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents:
21718
diff
changeset
|
347 if (! NILP (Vnonascii_translation_table)) |
23154
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
348 { |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
349 c = XINT (Faref (Vnonascii_translation_table, make_number (c))); |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
350 if (c >= 0400 && ! VALID_MULTIBYTE_CHAR_P (c)) |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
351 c = c_save + DEFAULT_NONASCII_INSERT_OFFSET; |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
352 } |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
353 else if (c >= 0240 && nonascii_insert_offset > 0) |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
354 { |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
355 c += nonascii_insert_offset; |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
356 if (c < 0400 || ! VALID_MULTIBYTE_CHAR_P (c)) |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
357 c = c_save + DEFAULT_NONASCII_INSERT_OFFSET; |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
358 } |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
359 else if (c >= 0240) |
21034
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
360 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
|
361 } |
a949dc65fbea
(nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents:
20636
diff
changeset
|
362 return c; |
a949dc65fbea
(nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents:
20636
diff
changeset
|
363 } |
22927
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
364 |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
365 |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
366 /* 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
|
367 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
|
368 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
|
369 Vnonascii_translation_table, i.e. what given by: |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
370 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
|
371 |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
372 int |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
373 multibyte_char_to_unibyte (c, rev_tbl) |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
374 int c; |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
375 Lisp_Object rev_tbl; |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
376 { |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
377 if (!SINGLE_BYTE_CHAR_P (c)) |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
378 { |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
379 int c_save = c; |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
380 |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
381 if (! CHAR_TABLE_P (rev_tbl) |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
382 && CHAR_TABLE_P (Vnonascii_translation_table)) |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
383 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
|
384 make_number (0)); |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
385 if (CHAR_TABLE_P (rev_tbl)) |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
386 { |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
387 Lisp_Object temp; |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
388 temp = Faref (rev_tbl, make_number (c)); |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
389 if (INTEGERP (temp)) |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
390 c = XINT (temp); |
23154
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
391 if (c >= 256) |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
392 c = (c_save & 0177) + 0200; |
22927
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
393 } |
23154
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
394 else |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
395 { |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
396 if (nonascii_insert_offset > 0) |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
397 c -= nonascii_insert_offset; |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
398 if (c < 128 || c >= 256) |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
399 c = (c_save & 0177) + 0200; |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
400 } |
22927
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
401 } |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
402 |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
403 return c; |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
404 } |
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
405 |
20705
a949dc65fbea
(nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents:
20636
diff
changeset
|
406 |
17052 | 407 /* Update the table Vcharset_table with the given arguments (see the |
408 document of `define-charset' for the meaning of each argument). | |
409 Several other table contents are also updated. The caller should | |
410 check the validity of CHARSET-ID and the remaining arguments in | |
411 advance. */ | |
412 | |
413 void | |
414 update_charset_table (charset_id, dimension, chars, width, direction, | |
415 iso_final_char, iso_graphic_plane, | |
416 short_name, long_name, description) | |
417 Lisp_Object charset_id, dimension, chars, width, direction; | |
418 Lisp_Object iso_final_char, iso_graphic_plane; | |
419 Lisp_Object short_name, long_name, description; | |
420 { | |
421 int charset = XINT (charset_id); | |
422 int bytes; | |
423 unsigned char leading_code_base, leading_code_ext; | |
424 | |
17322
a7d9b8e167ca
(non_ascii_char_to_string, update_charset_table,
Kenichi Handa <handa@m17n.org>
parents:
17186
diff
changeset
|
425 if (NILP (CHARSET_TABLE_ENTRY (charset))) |
a7d9b8e167ca
(non_ascii_char_to_string, update_charset_table,
Kenichi Handa <handa@m17n.org>
parents:
17186
diff
changeset
|
426 CHARSET_TABLE_ENTRY (charset) |
a7d9b8e167ca
(non_ascii_char_to_string, update_charset_table,
Kenichi Handa <handa@m17n.org>
parents:
17186
diff
changeset
|
427 = Fmake_vector (make_number (CHARSET_MAX_IDX), Qnil); |
17052 | 428 |
429 /* Get byte length of multibyte form, base leading-code, and | |
430 extended leading-code of the charset. See the comment under the | |
431 title "GENERAL NOTE on CHARACTER SET (CHARSET)" in charset.h. */ | |
432 bytes = XINT (dimension); | |
433 if (charset < MIN_CHARSET_PRIVATE_DIMENSION1) | |
434 { | |
435 /* Official charset, it doesn't have an extended leading-code. */ | |
436 if (charset != CHARSET_ASCII) | |
437 bytes += 1; /* For a base leading-code. */ | |
438 leading_code_base = charset; | |
439 leading_code_ext = 0; | |
440 } | |
441 else | |
442 { | |
443 /* Private charset. */ | |
444 bytes += 2; /* For base and extended leading-codes. */ | |
445 leading_code_base | |
446 = (charset < LEADING_CODE_EXT_12 | |
447 ? LEADING_CODE_PRIVATE_11 | |
448 : (charset < LEADING_CODE_EXT_21 | |
449 ? LEADING_CODE_PRIVATE_12 | |
450 : (charset < LEADING_CODE_EXT_22 | |
451 ? LEADING_CODE_PRIVATE_21 | |
452 : LEADING_CODE_PRIVATE_22))); | |
453 leading_code_ext = charset; | |
454 } | |
455 | |
23628
4a0a107fac9e
(string_to_non_ascii_char): Change the check for the
Kenichi Handa <handa@m17n.org>
parents:
23576
diff
changeset
|
456 if (BYTES_BY_CHAR_HEAD (leading_code_base) != bytes) |
4a0a107fac9e
(string_to_non_ascii_char): Change the check for the
Kenichi Handa <handa@m17n.org>
parents:
23576
diff
changeset
|
457 error ("Invalid dimension for the charset-ID %d", charset); |
4a0a107fac9e
(string_to_non_ascii_char): Change the check for the
Kenichi Handa <handa@m17n.org>
parents:
23576
diff
changeset
|
458 |
17052 | 459 CHARSET_TABLE_INFO (charset, CHARSET_ID_IDX) = charset_id; |
460 CHARSET_TABLE_INFO (charset, CHARSET_BYTES_IDX) = make_number (bytes); | |
461 CHARSET_TABLE_INFO (charset, CHARSET_DIMENSION_IDX) = dimension; | |
462 CHARSET_TABLE_INFO (charset, CHARSET_CHARS_IDX) = chars; | |
463 CHARSET_TABLE_INFO (charset, CHARSET_WIDTH_IDX) = width; | |
464 CHARSET_TABLE_INFO (charset, CHARSET_DIRECTION_IDX) = direction; | |
465 CHARSET_TABLE_INFO (charset, CHARSET_LEADING_CODE_BASE_IDX) | |
466 = make_number (leading_code_base); | |
467 CHARSET_TABLE_INFO (charset, CHARSET_LEADING_CODE_EXT_IDX) | |
468 = make_number (leading_code_ext); | |
469 CHARSET_TABLE_INFO (charset, CHARSET_ISO_FINAL_CHAR_IDX) = iso_final_char; | |
470 CHARSET_TABLE_INFO (charset, CHARSET_ISO_GRAPHIC_PLANE_IDX) | |
471 = iso_graphic_plane; | |
472 CHARSET_TABLE_INFO (charset, CHARSET_SHORT_NAME_IDX) = short_name; | |
473 CHARSET_TABLE_INFO (charset, CHARSET_LONG_NAME_IDX) = long_name; | |
474 CHARSET_TABLE_INFO (charset, CHARSET_DESCRIPTION_IDX) = description; | |
475 CHARSET_TABLE_INFO (charset, CHARSET_PLIST_IDX) = Qnil; | |
476 | |
477 { | |
478 /* If we have already defined a charset which has the same | |
479 DIMENSION, CHARS and ISO-FINAL-CHAR but the different | |
480 DIRECTION, we must update the entry REVERSE-CHARSET of both | |
481 charsets. If there's no such charset, the value of the entry | |
482 is set to nil. */ | |
483 int i; | |
484 | |
17186
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
485 for (i = 0; i <= MAX_CHARSET; i++) |
17052 | 486 if (!NILP (CHARSET_TABLE_ENTRY (i))) |
487 { | |
488 if (CHARSET_DIMENSION (i) == XINT (dimension) | |
489 && CHARSET_CHARS (i) == XINT (chars) | |
490 && CHARSET_ISO_FINAL_CHAR (i) == XINT (iso_final_char) | |
491 && CHARSET_DIRECTION (i) != XINT (direction)) | |
492 { | |
493 CHARSET_TABLE_INFO (charset, CHARSET_REVERSE_CHARSET_IDX) | |
494 = make_number (i); | |
495 CHARSET_TABLE_INFO (i, CHARSET_REVERSE_CHARSET_IDX) = charset_id; | |
496 break; | |
497 } | |
498 } | |
17186
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
499 if (i > MAX_CHARSET) |
17052 | 500 /* No such a charset. */ |
501 CHARSET_TABLE_INFO (charset, CHARSET_REVERSE_CHARSET_IDX) | |
502 = make_number (-1); | |
503 } | |
504 | |
505 if (charset != CHARSET_ASCII | |
506 && charset < MIN_CHARSET_PRIVATE_DIMENSION1) | |
507 { | |
508 width_by_char_head[leading_code_base] = XINT (width); | |
509 | |
510 /* Update table emacs_code_class. */ | |
511 emacs_code_class[charset] = (bytes == 2 | |
512 ? EMACS_leading_code_2 | |
513 : (bytes == 3 | |
514 ? EMACS_leading_code_3 | |
515 : EMACS_leading_code_4)); | |
516 } | |
517 | |
518 /* Update table iso_charset_table. */ | |
519 if (ISO_CHARSET_TABLE (dimension, chars, iso_final_char) < 0) | |
520 ISO_CHARSET_TABLE (dimension, chars, iso_final_char) = charset; | |
521 } | |
522 | |
523 #ifdef emacs | |
524 | |
525 /* Return charset id of CHARSET_SYMBOL, or return -1 if CHARSET_SYMBOL | |
526 is invalid. */ | |
527 int | |
528 get_charset_id (charset_symbol) | |
529 Lisp_Object charset_symbol; | |
530 { | |
531 Lisp_Object val; | |
532 int charset; | |
533 | |
534 return ((SYMBOLP (charset_symbol) | |
535 && (val = Fget (charset_symbol, Qcharset), VECTORP (val)) | |
536 && (charset = XINT (XVECTOR (val)->contents[CHARSET_ID_IDX]), | |
537 CHARSET_VALID_P (charset))) | |
538 ? charset : -1); | |
539 } | |
540 | |
541 /* Return an identification number for a new private charset of | |
542 DIMENSION and WIDTH. If there's no more room for the new charset, | |
543 return 0. */ | |
544 Lisp_Object | |
545 get_new_private_charset_id (dimension, width) | |
546 int dimension, width; | |
547 { | |
548 int charset, from, to; | |
549 | |
550 if (dimension == 1) | |
551 { | |
552 if (width == 1) | |
553 from = LEADING_CODE_EXT_11, to = LEADING_CODE_EXT_12; | |
554 else | |
555 from = LEADING_CODE_EXT_12, to = LEADING_CODE_EXT_21; | |
556 } | |
557 else | |
558 { | |
559 if (width == 1) | |
560 from = LEADING_CODE_EXT_21, to = LEADING_CODE_EXT_22; | |
561 else | |
20720
f084bb8d535f
(min_composite_char): New variable.
Kenichi Handa <handa@m17n.org>
parents:
20705
diff
changeset
|
562 from = LEADING_CODE_EXT_22, to = LEADING_CODE_EXT_MAX + 1; |
17052 | 563 } |
564 | |
565 for (charset = from; charset < to; charset++) | |
566 if (!CHARSET_DEFINED_P (charset)) break; | |
567 | |
568 return make_number (charset < to ? charset : 0); | |
569 } | |
570 | |
571 DEFUN ("define-charset", Fdefine_charset, Sdefine_charset, 3, 3, 0, | |
572 "Define CHARSET-ID as the identification number of CHARSET with INFO-VECTOR.\n\ | |
17727
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
573 If CHARSET-ID is nil, it is decided automatically, which means CHARSET is\n\ |
17052 | 574 treated as a private charset.\n\ |
575 INFO-VECTOR is a vector of the format:\n\ | |
576 [DIMENSION CHARS WIDTH DIRECTION ISO-FINAL-CHAR ISO-GRAPHIC-PLANE\n\ | |
577 SHORT-NAME LONG-NAME DESCRIPTION]\n\ | |
578 The meanings of each elements is as follows:\n\ | |
579 DIMENSION (integer) is the number of bytes to represent a character: 1 or 2.\n\ | |
580 CHARS (integer) is the number of characters in a dimension: 94 or 96.\n\ | |
581 WIDTH (integer) is the number of columns a character in the charset\n\ | |
582 occupies on the screen: one of 0, 1, and 2.\n\ | |
583 \n\ | |
584 DIRECTION (integer) is the rendering direction of characters in the\n\ | |
22774
a0909e59257d
(Fdefine_charset): Fix doc-string.
Kenichi Handa <handa@m17n.org>
parents:
22739
diff
changeset
|
585 charset when rendering. If 0, render from left to right, else\n\ |
a0909e59257d
(Fdefine_charset): Fix doc-string.
Kenichi Handa <handa@m17n.org>
parents:
22739
diff
changeset
|
586 render from right to left.\n\ |
17052 | 587 \n\ |
588 ISO-FINAL-CHAR (character) is the final character of the\n\ | |
589 corresponding ISO 2022 charset.\n\ | |
590 \n\ | |
591 ISO-GRAPHIC-PLANE (integer) is the graphic plane to be invoked\n\ | |
592 while encoding to variants of ISO 2022 coding system, one of the\n\ | |
593 following: 0/graphic-plane-left(GL), 1/graphic-plane-right(GR).\n\ | |
594 \n\ | |
595 SHORT-NAME (string) is the short name to refer to the charset.\n\ | |
596 \n\ | |
597 LONG-NAME (string) is the long name to refer to the charset.\n\ | |
598 \n\ | |
599 DESCRIPTION (string) is the description string of the charset.") | |
600 (charset_id, charset_symbol, info_vector) | |
601 Lisp_Object charset_id, charset_symbol, info_vector; | |
602 { | |
603 Lisp_Object *vec; | |
604 | |
605 if (!NILP (charset_id)) | |
606 CHECK_NUMBER (charset_id, 0); | |
607 CHECK_SYMBOL (charset_symbol, 1); | |
608 CHECK_VECTOR (info_vector, 2); | |
609 | |
610 if (! NILP (charset_id)) | |
611 { | |
612 if (! CHARSET_VALID_P (XINT (charset_id))) | |
613 error ("Invalid CHARSET: %d", XINT (charset_id)); | |
614 else if (CHARSET_DEFINED_P (XINT (charset_id))) | |
615 error ("Already defined charset: %d", XINT (charset_id)); | |
616 } | |
617 | |
618 vec = XVECTOR (info_vector)->contents; | |
619 if (XVECTOR (info_vector)->size != 9 | |
620 || !INTEGERP (vec[0]) || !(XINT (vec[0]) == 1 || XINT (vec[0]) == 2) | |
621 || !INTEGERP (vec[1]) || !(XINT (vec[1]) == 94 || XINT (vec[1]) == 96) | |
622 || !INTEGERP (vec[2]) || !(XINT (vec[2]) == 1 || XINT (vec[2]) == 2) | |
623 || !INTEGERP (vec[3]) || !(XINT (vec[3]) == 0 || XINT (vec[3]) == 1) | |
624 || !INTEGERP (vec[4]) || !(XINT (vec[4]) >= '0' && XINT (vec[4]) <= '~') | |
625 || !INTEGERP (vec[5]) || !(XINT (vec[5]) == 0 || XINT (vec[5]) == 1) | |
626 || !STRINGP (vec[6]) | |
627 || !STRINGP (vec[7]) | |
628 || !STRINGP (vec[8])) | |
629 error ("Invalid info-vector argument for defining charset %s", | |
630 XSYMBOL (charset_symbol)->name->data); | |
631 | |
632 if (NILP (charset_id)) | |
633 { | |
634 charset_id = get_new_private_charset_id (XINT (vec[0]), XINT (vec[2])); | |
635 if (XINT (charset_id) == 0) | |
636 error ("There's no room for a new private charset %s", | |
637 XSYMBOL (charset_symbol)->name->data); | |
638 } | |
639 | |
640 update_charset_table (charset_id, vec[0], vec[1], vec[2], vec[3], | |
641 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
|
642 Fput (charset_symbol, Qcharset, CHARSET_TABLE_ENTRY (XINT (charset_id))); |
17052 | 643 CHARSET_SYMBOL (XINT (charset_id)) = charset_symbol; |
644 Vcharset_list = Fcons (charset_symbol, Vcharset_list); | |
645 return Qnil; | |
646 } | |
647 | |
22703
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
648 DEFUN ("generic-character-list", Fgeneric_character_list, |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
649 Sgeneric_character_list, 0, 0, 0, |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
650 "Return a list of all possible generic characters.\n\ |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
651 It includes a generic character for a charset not yet defined.") |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
652 () |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
653 { |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
654 return Vgeneric_character_list; |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
655 } |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
656 |
20435
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
657 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
|
658 Sget_unused_iso_final_char, 2, 2, 0, |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
659 "Return an unsed ISO's final char for a charset of DIMENISION and CHARS.\n\ |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
660 DIMENSION is the number of bytes to represent a character: 1 or 2.\n\ |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
661 CHARS is the number of characters in a dimension: 94 or 96.\n\ |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
662 \n\ |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
663 This final char is for private use, thus the range is `0' (48) .. `?' (63).\n\ |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
664 If there's no unused final char for the specified kind of charset,\n\ |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
665 return nil.") |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
666 (dimension, chars) |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
667 Lisp_Object dimension, chars; |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
668 { |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
669 int final_char; |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
670 |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
671 CHECK_NUMBER (dimension, 0); |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
672 CHECK_NUMBER (chars, 1); |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
673 if (XINT (dimension) != 1 && XINT (dimension) != 2) |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
674 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
|
675 XINT (dimension)); |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
676 if (XINT (chars) != 94 && XINT (chars) != 96) |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
677 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
|
678 XINT (chars)); |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
679 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
|
680 { |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
681 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
|
682 break; |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
683 } |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
684 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
|
685 } |
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
686 |
17052 | 687 DEFUN ("declare-equiv-charset", Fdeclare_equiv_charset, Sdeclare_equiv_charset, |
688 4, 4, 0, | |
689 "Declare a charset of DIMENSION, CHARS, FINAL-CHAR is the same as CHARSET.\n\ | |
690 CHARSET should be defined by `defined-charset' in advance.") | |
691 (dimension, chars, final_char, charset_symbol) | |
692 Lisp_Object dimension, chars, final_char, charset_symbol; | |
693 { | |
694 int charset; | |
695 | |
696 CHECK_NUMBER (dimension, 0); | |
697 CHECK_NUMBER (chars, 1); | |
698 CHECK_NUMBER (final_char, 2); | |
699 CHECK_SYMBOL (charset_symbol, 3); | |
700 | |
701 if (XINT (dimension) != 1 && XINT (dimension) != 2) | |
702 error ("Invalid DIMENSION %d, it should be 1 or 2", XINT (dimension)); | |
703 if (XINT (chars) != 94 && XINT (chars) != 96) | |
704 error ("Invalid CHARS %d, it should be 94 or 96", XINT (chars)); | |
705 if (XINT (final_char) < '0' || XFASTINT (final_char) > '~') | |
706 error ("Invalid FINAL-CHAR %c, it should be `0'..`~'", XINT (chars)); | |
707 if ((charset = get_charset_id (charset_symbol)) < 0) | |
708 error ("Invalid charset %s", XSYMBOL (charset_symbol)->name->data); | |
709 | |
710 ISO_CHARSET_TABLE (dimension, chars, final_char) = charset; | |
711 return Qnil; | |
712 } | |
713 | |
714 /* Return number of different charsets in STR of length LEN. In | |
715 addition, for each found charset N, CHARSETS[N] is set 1. The | |
18003
ac148c939dec
(Ffind_charset_string): The variable CHARSETS is
Kenichi Handa <handa@m17n.org>
parents:
17880
diff
changeset
|
716 caller should allocate CHARSETS (MAX_CHARSET + 1 elements) in advance. |
22168
2156ac6d054f
(find_charset_in_str): New arg CMPCHARP.
Kenichi Handa <handa@m17n.org>
parents:
22121
diff
changeset
|
717 It may lookup a translation table TABLE if supplied. |
2156ac6d054f
(find_charset_in_str): New arg CMPCHARP.
Kenichi Handa <handa@m17n.org>
parents:
22121
diff
changeset
|
718 |
2156ac6d054f
(find_charset_in_str): New arg CMPCHARP.
Kenichi Handa <handa@m17n.org>
parents:
22121
diff
changeset
|
719 If CMPCHARP is nonzero and some composite character is found, |
2156ac6d054f
(find_charset_in_str): New arg CMPCHARP.
Kenichi Handa <handa@m17n.org>
parents:
22121
diff
changeset
|
720 CHARSETS[128] is also set 1 and the returned number is incremented |
2156ac6d054f
(find_charset_in_str): New arg CMPCHARP.
Kenichi Handa <handa@m17n.org>
parents:
22121
diff
changeset
|
721 by 1. */ |
17052 | 722 |
723 int | |
22168
2156ac6d054f
(find_charset_in_str): New arg CMPCHARP.
Kenichi Handa <handa@m17n.org>
parents:
22121
diff
changeset
|
724 find_charset_in_str (str, len, charsets, table, cmpcharp) |
17880
06f58d8a59ba
(Ffind_charset_region): Change the declaration of
Kenichi Handa <handa@m17n.org>
parents:
17863
diff
changeset
|
725 unsigned char *str; |
06f58d8a59ba
(Ffind_charset_region): Change the declaration of
Kenichi Handa <handa@m17n.org>
parents:
17863
diff
changeset
|
726 int len, *charsets; |
17727
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
727 Lisp_Object table; |
22168
2156ac6d054f
(find_charset_in_str): New arg CMPCHARP.
Kenichi Handa <handa@m17n.org>
parents:
22121
diff
changeset
|
728 int cmpcharp; |
17052 | 729 { |
20228
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
730 register int num = 0, c; |
17052 | 731 |
17727
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
732 if (! CHAR_TABLE_P (table)) |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
733 table = Qnil; |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
734 |
17052 | 735 while (len > 0) |
736 { | |
19983
a62008636710
(find_charset_in_str): Return also charsets in
Kenichi Handa <handa@m17n.org>
parents:
19444
diff
changeset
|
737 int bytes, charset; |
20228
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
738 c = *str; |
17727
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
739 |
20228
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
740 if (c == LEADING_CODE_COMPOSITION) |
19983
a62008636710
(find_charset_in_str): Return also charsets in
Kenichi Handa <handa@m17n.org>
parents:
19444
diff
changeset
|
741 { |
20228
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
742 int cmpchar_id = str_cmpchar_id (str, len); |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
743 GLYPH *glyph; |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
744 |
22168
2156ac6d054f
(find_charset_in_str): New arg CMPCHARP.
Kenichi Handa <handa@m17n.org>
parents:
22121
diff
changeset
|
745 if (cmpchar_id >= 0) |
19983
a62008636710
(find_charset_in_str): Return also charsets in
Kenichi Handa <handa@m17n.org>
parents:
19444
diff
changeset
|
746 { |
23427
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
747 struct cmpchar_info *cmp_p = cmpchar_table[cmpchar_id]; |
20228
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
748 int i; |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
749 |
23427
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
750 for (i = 0; i < cmp_p->glyph_len; i++) |
20228
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
751 { |
23427
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
752 c = cmp_p->glyph[i]; |
20228
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
753 if (!NILP (table)) |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
754 { |
22121
6e79a15594ab
Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents:
21718
diff
changeset
|
755 if ((c = translate_char (table, c, 0, 0, 0)) < 0) |
23427
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
756 c = cmp_p->glyph[i]; |
20228
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
757 } |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
758 if ((charset = CHAR_CHARSET (c)) < 0) |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
759 charset = CHARSET_ASCII; |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
760 if (!charsets[charset]) |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
761 { |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
762 charsets[charset] = 1; |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
763 num += 1; |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
764 } |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
765 } |
23427
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
766 str += cmp_p->len; |
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
767 len -= cmp_p->len; |
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
768 if (cmpcharp && !charsets[CHARSET_COMPOSITION]) |
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
769 { |
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
770 charsets[CHARSET_COMPOSITION] = 1; |
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
771 num += 1; |
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
772 } |
20228
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
773 continue; |
19983
a62008636710
(find_charset_in_str): Return also charsets in
Kenichi Handa <handa@m17n.org>
parents:
19444
diff
changeset
|
774 } |
20228
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
775 |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
776 charset = CHARSET_ASCII; |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
777 bytes = 1; |
19983
a62008636710
(find_charset_in_str): Return also charsets in
Kenichi Handa <handa@m17n.org>
parents:
19444
diff
changeset
|
778 } |
a62008636710
(find_charset_in_str): Return also charsets in
Kenichi Handa <handa@m17n.org>
parents:
19444
diff
changeset
|
779 else |
17727
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
780 { |
20228
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
781 c = STRING_CHAR_AND_LENGTH (str, len, bytes); |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
782 if (! NILP (table)) |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
783 { |
22121
6e79a15594ab
Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents:
21718
diff
changeset
|
784 int c1 = translate_char (table, c, 0, 0, 0); |
20228
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
785 if (c1 >= 0) |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
786 c = c1; |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
787 } |
acb7aa0aa71f
(find_charset_in_str): Handle the case that STR
Kenichi Handa <handa@m17n.org>
parents:
20188
diff
changeset
|
788 charset = CHAR_CHARSET (c); |
17727
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
789 } |
17052 | 790 |
791 if (!charsets[charset]) | |
792 { | |
793 charsets[charset] = 1; | |
794 num += 1; | |
795 } | |
796 str += bytes; | |
797 len -= bytes; | |
798 } | |
799 return num; | |
800 } | |
801 | |
802 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
|
803 2, 3, 0, |
17052 | 804 "Return a list of charsets in the region between BEG and END.\n\ |
17727
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
805 BEG and END are buffer positions.\n\ |
23427
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
806 If the region contains any composite character,\n\ |
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
807 `composition' is included in the returned list.\n\ |
22121
6e79a15594ab
Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents:
21718
diff
changeset
|
808 Optional arg TABLE if non-nil is a translation table to look up.") |
17727
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
809 (beg, end, table) |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
810 Lisp_Object beg, end, table; |
17052 | 811 { |
17880
06f58d8a59ba
(Ffind_charset_region): Change the declaration of
Kenichi Handa <handa@m17n.org>
parents:
17863
diff
changeset
|
812 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
|
813 int from, from_byte, to, stop, stop_byte, i; |
17052 | 814 Lisp_Object val; |
815 | |
816 validate_region (&beg, &end); | |
817 from = XFASTINT (beg); | |
818 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
|
819 |
23210
53c330e846d0
(Ffind_charset_region): Optimization for unibyte buffer.
Kenichi Handa <handa@m17n.org>
parents:
23154
diff
changeset
|
820 if (NILP (current_buffer->enable_multibyte_characters)) |
53c330e846d0
(Ffind_charset_region): Optimization for unibyte buffer.
Kenichi Handa <handa@m17n.org>
parents:
23154
diff
changeset
|
821 return (from == to |
53c330e846d0
(Ffind_charset_region): Optimization for unibyte buffer.
Kenichi Handa <handa@m17n.org>
parents:
23154
diff
changeset
|
822 ? Qnil |
53c330e846d0
(Ffind_charset_region): Optimization for unibyte buffer.
Kenichi Handa <handa@m17n.org>
parents:
23154
diff
changeset
|
823 : Fcons (Qascii, Qnil)); |
53c330e846d0
(Ffind_charset_region): Optimization for unibyte buffer.
Kenichi Handa <handa@m17n.org>
parents:
23154
diff
changeset
|
824 |
17052 | 825 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
|
826 { |
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
827 stop = GPT; |
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
828 stop_byte = GPT_BYTE; |
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
829 } |
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
830 else |
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
831 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
|
832 |
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
833 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
|
834 |
17880
06f58d8a59ba
(Ffind_charset_region): Change the declaration of
Kenichi Handa <handa@m17n.org>
parents:
17863
diff
changeset
|
835 bzero (charsets, (MAX_CHARSET + 1) * sizeof (int)); |
17052 | 836 while (1) |
837 { | |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
838 find_charset_in_str (BYTE_POS_ADDR (from_byte), stop_byte - from_byte, |
23427
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
839 charsets, table, 1); |
17052 | 840 if (stop < to) |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
841 { |
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
842 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
|
843 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
|
844 } |
17052 | 845 else |
846 break; | |
847 } | |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
848 |
17052 | 849 val = Qnil; |
17186
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
850 for (i = MAX_CHARSET; i >= 0; i--) |
17052 | 851 if (charsets[i]) |
852 val = Fcons (CHARSET_SYMBOL (i), val); | |
853 return val; | |
854 } | |
855 | |
856 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
|
857 1, 2, 0, |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
858 "Return a list of charsets in STR.\n\ |
23427
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
859 If the string contains any composite characters,\n\ |
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
860 `composition' is included in the returned list.\n\ |
22121
6e79a15594ab
Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents:
21718
diff
changeset
|
861 Optional arg TABLE if non-nil is a translation table to look up.") |
17727
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
862 (str, table) |
9d39361ce928
(unify_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
17366
diff
changeset
|
863 Lisp_Object str, table; |
17052 | 864 { |
18003
ac148c939dec
(Ffind_charset_string): The variable CHARSETS is
Kenichi Handa <handa@m17n.org>
parents:
17880
diff
changeset
|
865 int charsets[MAX_CHARSET + 1]; |
17052 | 866 int i; |
867 Lisp_Object val; | |
868 | |
869 CHECK_STRING (str, 0); | |
20614
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
870 |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
871 if (! STRING_MULTIBYTE (str)) |
23210
53c330e846d0
(Ffind_charset_region): Optimization for unibyte buffer.
Kenichi Handa <handa@m17n.org>
parents:
23154
diff
changeset
|
872 return (XSTRING (str)->size == 0 |
53c330e846d0
(Ffind_charset_region): Optimization for unibyte buffer.
Kenichi Handa <handa@m17n.org>
parents:
23154
diff
changeset
|
873 ? Qnil |
53c330e846d0
(Ffind_charset_region): Optimization for unibyte buffer.
Kenichi Handa <handa@m17n.org>
parents:
23154
diff
changeset
|
874 : Fcons (Qascii, Qnil)); |
20614
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
875 |
18003
ac148c939dec
(Ffind_charset_string): The variable CHARSETS is
Kenichi Handa <handa@m17n.org>
parents:
17880
diff
changeset
|
876 bzero (charsets, (MAX_CHARSET + 1) * sizeof (int)); |
21244
50929073a0ba
Use STRING_BYTES and SET_STRING_BYTES.
Richard M. Stallman <rms@gnu.org>
parents:
21063
diff
changeset
|
877 find_charset_in_str (XSTRING (str)->data, STRING_BYTES (XSTRING (str)), |
23427
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
878 charsets, table, 1); |
17052 | 879 val = Qnil; |
17186
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
880 for (i = MAX_CHARSET; i >= 0; i--) |
17052 | 881 if (charsets[i]) |
882 val = Fcons (CHARSET_SYMBOL (i), val); | |
883 return val; | |
884 } | |
885 | |
886 DEFUN ("make-char-internal", Fmake_char_internal, Smake_char_internal, 1, 3, 0, | |
17186
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
887 "") |
17052 | 888 (charset, code1, code2) |
889 Lisp_Object charset, code1, code2; | |
890 { | |
891 CHECK_NUMBER (charset, 0); | |
892 | |
893 if (NILP (code1)) | |
894 XSETFASTINT (code1, 0); | |
895 else | |
896 CHECK_NUMBER (code1, 1); | |
897 if (NILP (code2)) | |
898 XSETFASTINT (code2, 0); | |
899 else | |
900 CHECK_NUMBER (code2, 2); | |
901 | |
902 if (!CHARSET_DEFINED_P (XINT (charset))) | |
903 error ("Invalid charset: %d", XINT (charset)); | |
904 | |
905 return make_number (MAKE_CHAR (XINT (charset), XINT (code1), XINT (code2))); | |
906 } | |
907 | |
908 DEFUN ("split-char", Fsplit_char, Ssplit_char, 1, 1, 0, | |
909 "Return list of charset and one or two position-codes of CHAR.") | |
910 (ch) | |
911 Lisp_Object ch; | |
912 { | |
913 Lisp_Object val; | |
17322
a7d9b8e167ca
(non_ascii_char_to_string, update_charset_table,
Kenichi Handa <handa@m17n.org>
parents:
17186
diff
changeset
|
914 int charset, c1, c2; |
17052 | 915 |
916 CHECK_NUMBER (ch, 0); | |
917 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
|
918 return (c2 >= 0 |
17052 | 919 ? Fcons (CHARSET_SYMBOL (charset), |
920 Fcons (make_number (c1), Fcons (make_number (c2), Qnil))) | |
921 : Fcons (CHARSET_SYMBOL (charset), Fcons (make_number (c1), Qnil))); | |
922 } | |
923 | |
924 DEFUN ("char-charset", Fchar_charset, Schar_charset, 1, 1, 0, | |
925 "Return charset of CHAR.") | |
926 (ch) | |
927 Lisp_Object ch; | |
928 { | |
929 CHECK_NUMBER (ch, 0); | |
930 | |
931 return CHARSET_SYMBOL (CHAR_CHARSET (XINT (ch))); | |
932 } | |
933 | |
21445
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
934 DEFUN ("charset-after", Fcharset_after, Scharset_after, 0, 1, 0, |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
935 "Return charset of a character in current buffer at position POS.\n\ |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
936 If POS is nil, it defauls to the current point.") |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
937 (pos) |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
938 Lisp_Object pos; |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
939 { |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
940 register int pos_byte, c, charset; |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
941 register unsigned char *p; |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
942 |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
943 if (NILP (pos)) |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
944 pos_byte = PT_BYTE; |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
945 else if (MARKERP (pos)) |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
946 pos_byte = marker_byte_position (pos); |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
947 else |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
948 { |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
949 CHECK_NUMBER (pos, 0); |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
950 pos_byte = CHAR_TO_BYTE (XINT (pos)); |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
951 } |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
952 p = BYTE_POS_ADDR (pos_byte); |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
953 c = STRING_CHAR (p, Z_BYTE - pos_byte); |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
954 charset = CHAR_CHARSET (c); |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
955 return CHARSET_SYMBOL (charset); |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
956 } |
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
957 |
17052 | 958 DEFUN ("iso-charset", Fiso_charset, Siso_charset, 3, 3, 0, |
20833
cf404f800d2c
(Fiso_charset): Doc-string modified.
Kenichi Handa <handa@m17n.org>
parents:
20720
diff
changeset
|
959 "Return charset of ISO's specification DIMENSION, CHARS, and FINAL-CHAR.\n\ |
cf404f800d2c
(Fiso_charset): Doc-string modified.
Kenichi Handa <handa@m17n.org>
parents:
20720
diff
changeset
|
960 \n\ |
cf404f800d2c
(Fiso_charset): Doc-string modified.
Kenichi Handa <handa@m17n.org>
parents:
20720
diff
changeset
|
961 ISO 2022's designation sequence (escape sequence) distinguishes charsets\n\ |
cf404f800d2c
(Fiso_charset): Doc-string modified.
Kenichi Handa <handa@m17n.org>
parents:
20720
diff
changeset
|
962 by their DIMENSION, CHARS, and FINAL-CHAR,\n\ |
cf404f800d2c
(Fiso_charset): Doc-string modified.
Kenichi Handa <handa@m17n.org>
parents:
20720
diff
changeset
|
963 where as Emacs distinguishes them by charset symbol.\n\ |
cf404f800d2c
(Fiso_charset): Doc-string modified.
Kenichi Handa <handa@m17n.org>
parents:
20720
diff
changeset
|
964 See the documentation of the function `charset-info' for the meanings of\n\ |
cf404f800d2c
(Fiso_charset): Doc-string modified.
Kenichi Handa <handa@m17n.org>
parents:
20720
diff
changeset
|
965 DIMENSION, CHARS, and FINAL-CHAR.") |
17052 | 966 (dimension, chars, final_char) |
967 Lisp_Object dimension, chars, final_char; | |
968 { | |
969 int charset; | |
970 | |
971 CHECK_NUMBER (dimension, 0); | |
972 CHECK_NUMBER (chars, 1); | |
973 CHECK_NUMBER (final_char, 2); | |
974 | |
975 if ((charset = ISO_CHARSET_TABLE (dimension, chars, final_char)) < 0) | |
976 return Qnil; | |
977 return CHARSET_SYMBOL (charset); | |
978 } | |
979 | |
20152
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
980 /* If GENERICP is nonzero, return nonzero iff C is a valid normal or |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
981 generic character. If GENERICP is zero, return nonzero iff C is a |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
982 valid normal character. Do not call this function directly, |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
983 instead use macro CHAR_VALID_P. */ |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
984 int |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
985 char_valid_p (c, genericp) |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
986 int c, genericp; |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
987 { |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
988 int charset, c1, c2; |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
989 |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
990 if (c < 0) |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
991 return 0; |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
992 if (SINGLE_BYTE_CHAR_P (c)) |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
993 return 1; |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
994 SPLIT_NON_ASCII_CHAR (c, charset, c1, c2); |
23688
5f039c506e17
(char_valid_p): Handle a composite character correctly.
Kenichi Handa <handa@m17n.org>
parents:
23685
diff
changeset
|
995 if (charset != CHARSET_COMPOSITION && !CHARSET_DEFINED_P (charset)) |
20152
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
996 return 0; |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
997 return (c < MIN_CHAR_COMPOSITION |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
998 ? ((c & CHAR_FIELD1_MASK) /* i.e. dimension of C is two. */ |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
999 ? (genericp && c1 == 0 && c2 == 0 |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1000 || c1 >= 32 && c2 >= 32) |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1001 : (genericp && c1 == 0 |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1002 || c1 >= 32)) |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1003 : c < MIN_CHAR_COMPOSITION + n_cmpchars); |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1004 } |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1005 |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1006 DEFUN ("char-valid-p", Fchar_valid_p, Schar_valid_p, 1, 2, 0, |
20636
5ab6701a2b4b
(Fchar_valid_p): Fix bare newlines in doc string.
Dave Love <fx@gnu.org>
parents:
20614
diff
changeset
|
1007 "Return t if OBJECT is a valid normal character.\n\ |
5ab6701a2b4b
(Fchar_valid_p): Fix bare newlines in doc string.
Dave Love <fx@gnu.org>
parents:
20614
diff
changeset
|
1008 If optional arg GENERICP is non-nil, also return t if OBJECT is\n\ |
20152
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1009 a valid generic character.") |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1010 (object, genericp) |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1011 Lisp_Object object, genericp; |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1012 { |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1013 if (! NATNUMP (object)) |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1014 return Qnil; |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1015 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
|
1016 } |
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1017 |
21034
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1018 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
|
1019 Sunibyte_char_to_multibyte, 1, 1, 0, |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1020 "Convert the unibyte character CH to multibyte character.\n\ |
22185
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
1021 The conversion is done based on `nonascii-translation-table' (which see)\n\ |
21715
98c5740d198b
(Funibyte_char_to_multibyte): Doc fix.
Richard M. Stallman <rms@gnu.org>
parents:
21519
diff
changeset
|
1022 or `nonascii-insert-offset' (which see).") |
21034
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1023 (ch) |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1024 Lisp_Object ch; |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1025 { |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1026 int c; |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1027 |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1028 CHECK_NUMBER (ch, 0); |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1029 c = XINT (ch); |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1030 if (c < 0 || c >= 0400) |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1031 error ("Invalid unibyte character: %d", c); |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1032 c = unibyte_char_to_multibyte (c); |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1033 if (c < 0) |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1034 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
|
1035 return make_number (c); |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1036 } |
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1037 |
23060
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1038 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
|
1039 Smultibyte_char_to_unibyte, 1, 1, 0, |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1040 "Convert the multibyte character CH to unibyte character.\n\ |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1041 The conversion is done based on `nonascii-translation-table' (which see)\n\ |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1042 or `nonascii-insert-offset' (which see).") |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1043 (ch) |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1044 Lisp_Object ch; |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1045 { |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1046 int c; |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1047 |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1048 CHECK_NUMBER (ch, 0); |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1049 c = XINT (ch); |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1050 if (c < 0) |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1051 error ("Invalid multibyte character: %d", c); |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1052 c = multibyte_char_to_unibyte (c, Qnil); |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1053 if (c < 0) |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1054 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
|
1055 return make_number (c); |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1056 } |
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1057 |
17052 | 1058 DEFUN ("char-bytes", Fchar_bytes, Schar_bytes, 1, 1, 0, |
23140
17ad9953ff42
(Fchar_bytes): Add \n\ in doc-string.
Kenichi Handa <handa@m17n.org>
parents:
23131
diff
changeset
|
1059 "Return 1 regardless of the argument CHAR.\n\ |
23576 | 1060 This is now an obsolete function. We keep it just for backward compatibility.") |
17052 | 1061 (ch) |
1062 Lisp_Object ch; | |
1063 { | |
1064 Lisp_Object val; | |
1065 | |
1066 CHECK_NUMBER (ch, 0); | |
23131
f7e486faa1ad
(Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents:
23060
diff
changeset
|
1067 return make_number (1); |
f7e486faa1ad
(Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents:
23060
diff
changeset
|
1068 } |
f7e486faa1ad
(Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents:
23060
diff
changeset
|
1069 |
f7e486faa1ad
(Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents:
23060
diff
changeset
|
1070 /* 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
|
1071 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
|
1072 int |
f7e486faa1ad
(Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents:
23060
diff
changeset
|
1073 char_bytes (c) |
f7e486faa1ad
(Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents:
23060
diff
changeset
|
1074 int c; |
f7e486faa1ad
(Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents:
23060
diff
changeset
|
1075 { |
f7e486faa1ad
(Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents:
23060
diff
changeset
|
1076 int bytes; |
f7e486faa1ad
(Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents:
23060
diff
changeset
|
1077 |
f7e486faa1ad
(Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents:
23060
diff
changeset
|
1078 if (COMPOSITE_CHAR_P (c)) |
17052 | 1079 { |
23131
f7e486faa1ad
(Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents:
23060
diff
changeset
|
1080 unsigned int id = COMPOSITE_CHAR_ID (c); |
17052 | 1081 |
1082 bytes = (id < n_cmpchars ? cmpchar_table[id]->len : 1); | |
1083 } | |
1084 else | |
1085 { | |
23131
f7e486faa1ad
(Fchar_bytes): Now always return 1.
Kenichi Handa <handa@m17n.org>
parents:
23060
diff
changeset
|
1086 int charset = CHAR_CHARSET (c); |
17052 | 1087 |
1088 bytes = CHARSET_DEFINED_P (charset) ? CHARSET_BYTES (charset) : 1; | |
1089 } | |
1090 | |
23576 | 1091 return bytes; |
17052 | 1092 } |
1093 | |
1094 /* Return the width of character of which multi-byte form starts with | |
1095 C. The width is measured by how many columns occupied on the | |
1096 screen when displayed in the current buffer. */ | |
1097 | |
1098 #define ONE_BYTE_CHAR_WIDTH(c) \ | |
1099 (c < 0x20 \ | |
1100 ? (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
|
1101 ? XFASTINT (current_buffer->tab_width) \ |
17052 | 1102 : (c == '\n' ? 0 : (NILP (current_buffer->ctl_arrow) ? 4 : 2))) \ |
1103 : (c < 0x7f \ | |
1104 ? 1 \ | |
1105 : (c == 0x7F \ | |
1106 ? (NILP (current_buffer->ctl_arrow) ? 4 : 2) \ | |
1107 : ((! NILP (current_buffer->enable_multibyte_characters) \ | |
1108 && BASE_LEADING_CODE_P (c)) \ | |
1109 ? WIDTH_BY_CHAR_HEAD (c) \ | |
22121
6e79a15594ab
Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents:
21718
diff
changeset
|
1110 : 4)))) |
17052 | 1111 |
1112 DEFUN ("char-width", Fchar_width, Schar_width, 1, 1, 0, | |
1113 "Return width of CHAR when displayed in the current buffer.\n\ | |
1114 The width is measured by how many columns it occupies on the screen.") | |
1115 (ch) | |
1116 Lisp_Object ch; | |
1117 { | |
17818
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1118 Lisp_Object val, disp; |
17052 | 1119 int c; |
18537
6a726658dfa9
(Fchar_width): Check if the current buffer has display
Kenichi Handa <handa@m17n.org>
parents:
18341
diff
changeset
|
1120 struct Lisp_Char_Table *dp = buffer_display_table (); |
17052 | 1121 |
1122 CHECK_NUMBER (ch, 0); | |
1123 | |
17818
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1124 c = XINT (ch); |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1125 |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1126 /* 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
|
1127 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
|
1128 |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1129 if (VECTORP (disp)) |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1130 XSETINT (val, XVECTOR (disp)->size); |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1131 else if (SINGLE_BYTE_CHAR_P (c)) |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1132 XSETINT (val, ONE_BYTE_CHAR_WIDTH (c)); |
17052 | 1133 else if (COMPOSITE_CHAR_P (c)) |
1134 { | |
1135 int id = COMPOSITE_CHAR_ID (XFASTINT (ch)); | |
1136 XSETFASTINT (val, (id < n_cmpchars ? cmpchar_table[id]->width : 0)); | |
1137 } | |
1138 else | |
1139 { | |
1140 int charset = CHAR_CHARSET (c); | |
1141 | |
1142 XSETFASTINT (val, CHARSET_WIDTH (charset)); | |
1143 } | |
1144 return val; | |
1145 } | |
1146 | |
1147 /* Return width of string STR of length LEN when displayed in the | |
1148 current buffer. The width is measured by how many columns it | |
1149 occupies on the screen. */ | |
17818
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1150 |
17052 | 1151 int |
1152 strwidth (str, len) | |
1153 unsigned char *str; | |
1154 int len; | |
1155 { | |
1156 unsigned char *endp = str + len; | |
1157 int width = 0; | |
20362
67bcb068070c
(strwidth): Remove extra argument to buffer_display_table.
Kenichi Handa <handa@m17n.org>
parents:
20228
diff
changeset
|
1158 struct Lisp_Char_Table *dp = buffer_display_table (); |
17052 | 1159 |
17818
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1160 while (str < endp) |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1161 { |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1162 if (*str == LEADING_CODE_COMPOSITION) |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1163 { |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1164 int id = str_cmpchar_id (str, endp - str); |
17052 | 1165 |
17818
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1166 if (id < 0) |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1167 { |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1168 width += 4; |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1169 str++; |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1170 } |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1171 else |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1172 { |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1173 width += cmpchar_table[id]->width; |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1174 str += cmpchar_table[id]->len; |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1175 } |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1176 } |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1177 else |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1178 { |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1179 Lisp_Object disp; |
21063
44845df088bc
(strwidth): Make it work for invalid bytes sequence.
Kenichi Handa <handa@m17n.org>
parents:
21034
diff
changeset
|
1180 int thislen; |
44845df088bc
(strwidth): Make it work for invalid bytes sequence.
Kenichi Handa <handa@m17n.org>
parents:
21034
diff
changeset
|
1181 int c = STRING_CHAR_AND_LENGTH (str, endp - str, thislen); |
17818
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1182 |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1183 /* Get the way the display table would display it. */ |
17863
bc8774c7999b
(strwidth): Fix previous change.
Richard M. Stallman <rms@gnu.org>
parents:
17843
diff
changeset
|
1184 if (dp) |
bc8774c7999b
(strwidth): Fix previous change.
Richard M. Stallman <rms@gnu.org>
parents:
17843
diff
changeset
|
1185 disp = DISP_CHAR_VECTOR (dp, c); |
bc8774c7999b
(strwidth): Fix previous change.
Richard M. Stallman <rms@gnu.org>
parents:
17843
diff
changeset
|
1186 else |
bc8774c7999b
(strwidth): Fix previous change.
Richard M. Stallman <rms@gnu.org>
parents:
17843
diff
changeset
|
1187 disp = Qnil; |
17818
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1188 |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1189 if (VECTORP (disp)) |
21063
44845df088bc
(strwidth): Make it work for invalid bytes sequence.
Kenichi Handa <handa@m17n.org>
parents:
21034
diff
changeset
|
1190 width += XVECTOR (disp)->size; |
17818
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1191 else |
21063
44845df088bc
(strwidth): Make it work for invalid bytes sequence.
Kenichi Handa <handa@m17n.org>
parents:
21034
diff
changeset
|
1192 width += ONE_BYTE_CHAR_WIDTH (*str); |
17818
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1193 |
21063
44845df088bc
(strwidth): Make it work for invalid bytes sequence.
Kenichi Handa <handa@m17n.org>
parents:
21034
diff
changeset
|
1194 str += thislen; |
17818
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1195 } |
76c6740bd736
(strwidth, Fchar_width): Handle display table.
Richard M. Stallman <rms@gnu.org>
parents:
17727
diff
changeset
|
1196 } |
17052 | 1197 return width; |
1198 } | |
1199 | |
1200 DEFUN ("string-width", Fstring_width, Sstring_width, 1, 1, 0, | |
1201 "Return width of STRING when displayed in the current buffer.\n\ | |
1202 Width is measured by how many columns it occupies on the screen.\n\ | |
19384 | 1203 When calculating width of a multibyte character in STRING,\n\ |
1204 only the base leading-code is considered; the validity of\n\ | |
1205 the following bytes is not checked.") | |
17052 | 1206 (str) |
1207 Lisp_Object str; | |
1208 { | |
1209 Lisp_Object val; | |
1210 | |
1211 CHECK_STRING (str, 0); | |
21244
50929073a0ba
Use STRING_BYTES and SET_STRING_BYTES.
Richard M. Stallman <rms@gnu.org>
parents:
21063
diff
changeset
|
1212 XSETFASTINT (val, strwidth (XSTRING (str)->data, |
50929073a0ba
Use STRING_BYTES and SET_STRING_BYTES.
Richard M. Stallman <rms@gnu.org>
parents:
21063
diff
changeset
|
1213 STRING_BYTES (XSTRING (str)))); |
17052 | 1214 return val; |
1215 } | |
1216 | |
1217 DEFUN ("char-direction", Fchar_direction, Schar_direction, 1, 1, 0, | |
1218 "Return the direction of CHAR.\n\ | |
1219 The returned value is 0 for left-to-right and 1 for right-to-left.") | |
1220 (ch) | |
1221 Lisp_Object ch; | |
1222 { | |
1223 int charset; | |
1224 | |
1225 CHECK_NUMBER (ch, 0); | |
1226 charset = CHAR_CHARSET (XFASTINT (ch)); | |
1227 if (!CHARSET_DEFINED_P (charset)) | |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1228 invalid_character (XINT (ch)); |
17052 | 1229 return CHARSET_TABLE_INFO (charset, CHARSET_DIRECTION_IDX); |
1230 } | |
1231 | |
19415
309b9a0d8a45
(Fchars_in_region): Renamed form Fcount_chars_region.
Kenichi Handa <handa@m17n.org>
parents:
19384
diff
changeset
|
1232 DEFUN ("chars-in-region", Fchars_in_region, Schars_in_region, 2, 2, 0, |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1233 "Return number of characters between BEG and END.") |
19384 | 1234 (beg, end) |
1235 Lisp_Object beg, end; | |
1236 { | |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1237 int from, to; |
19384 | 1238 |
22739
f0c8af6c2dfd
(Fchars_in_region): Coerce markers.
Richard M. Stallman <rms@gnu.org>
parents:
22703
diff
changeset
|
1239 CHECK_NUMBER_COERCE_MARKER (beg, 0); |
f0c8af6c2dfd
(Fchars_in_region): Coerce markers.
Richard M. Stallman <rms@gnu.org>
parents:
22703
diff
changeset
|
1240 CHECK_NUMBER_COERCE_MARKER (end, 1); |
f0c8af6c2dfd
(Fchars_in_region): Coerce markers.
Richard M. Stallman <rms@gnu.org>
parents:
22703
diff
changeset
|
1241 |
19384 | 1242 from = min (XFASTINT (beg), XFASTINT (end)); |
19444
752afe97eaa4
(Fchars_in_region): Fix gap handling.
Richard M. Stallman <rms@gnu.org>
parents:
19415
diff
changeset
|
1243 to = max (XFASTINT (beg), XFASTINT (end)); |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1244 |
21519
9b26f9300d41
(Fchars_in_region): Fix mixing of Lisp_Object and int.
Andreas Schwab <schwab@suse.de>
parents:
21514
diff
changeset
|
1245 return make_number (to - from); |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1246 } |
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1247 |
20614
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1248 /* 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
|
1249 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
|
1250 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
|
1251 we treat each byte as a character. */ |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1252 |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1253 int |
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1254 chars_in_text (ptr, nbytes) |
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1255 unsigned char *ptr; |
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1256 int nbytes; |
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1257 { |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1258 unsigned char *endp, c; |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1259 int chars; |
19384 | 1260 |
20614
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1261 /* 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
|
1262 if (current_buffer == 0 |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1263 || 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
|
1264 return nbytes; |
20188
d4ad4463384d
(Fchars_in_string): Pay attention to
Kenichi Handa <handa@m17n.org>
parents:
20152
diff
changeset
|
1265 |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1266 endp = ptr + nbytes; |
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1267 chars = 0; |
19384 | 1268 |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1269 while (ptr < endp) |
19384 | 1270 { |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1271 c = *ptr++; |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1272 |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1273 if (BASE_LEADING_CODE_P (c)) |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1274 while (ptr < endp && ! CHAR_HEAD_P (*ptr)) ptr++; |
19384 | 1275 chars++; |
1276 } | |
1277 | |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1278 return chars; |
19384 | 1279 } |
1280 | |
20614
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1281 /* 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
|
1282 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
|
1283 It ignores enable-multibyte-characters. */ |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1284 |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1285 int |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1286 multibyte_chars_in_text (ptr, nbytes) |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1287 unsigned char *ptr; |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1288 int nbytes; |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1289 { |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1290 unsigned char *endp, c; |
20614
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1291 int chars; |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1292 |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1293 endp = ptr + nbytes; |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1294 chars = 0; |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1295 |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1296 while (ptr < endp) |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1297 { |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1298 c = *ptr++; |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1299 |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1300 if (BASE_LEADING_CODE_P (c)) |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1301 while (ptr < endp && ! CHAR_HEAD_P (*ptr)) ptr++; |
20614
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1302 chars++; |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1303 } |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1304 |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1305 return chars; |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1306 } |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1307 |
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1308 DEFUN ("string", Fstring, Sstring, 1, MANY, 0, |
17052 | 1309 "Concatenate all the argument characters and make the result a string.") |
17366
f7809ff2c65f
(ONE_BYTE_CHAR_WIDTH, Fconcat_chars): Don't use Lisp_Object as integer.
Karl Heuer <kwzh@gnu.org>
parents:
17322
diff
changeset
|
1310 (n, args) |
f7809ff2c65f
(ONE_BYTE_CHAR_WIDTH, Fconcat_chars): Don't use Lisp_Object as integer.
Karl Heuer <kwzh@gnu.org>
parents:
17322
diff
changeset
|
1311 int n; |
17052 | 1312 Lisp_Object *args; |
1313 { | |
17366
f7809ff2c65f
(ONE_BYTE_CHAR_WIDTH, Fconcat_chars): Don't use Lisp_Object as integer.
Karl Heuer <kwzh@gnu.org>
parents:
17322
diff
changeset
|
1314 int i; |
17052 | 1315 unsigned char *buf |
17834
e154b943bbba
(non_ascii_char_to_string): Signal error if the
Kenichi Handa <handa@m17n.org>
parents:
17818
diff
changeset
|
1316 = (unsigned char *) alloca (MAX_LENGTH_OF_MULTI_BYTE_FORM * n); |
17052 | 1317 unsigned char *p = buf; |
1318 Lisp_Object val; | |
1319 | |
1320 for (i = 0; i < n; i++) | |
1321 { | |
1322 int c, len; | |
1323 unsigned char *str; | |
1324 | |
1325 if (!INTEGERP (args[i])) | |
20720
f084bb8d535f
(min_composite_char): New variable.
Kenichi Handa <handa@m17n.org>
parents:
20705
diff
changeset
|
1326 CHECK_NUMBER (args[i], 0); |
17052 | 1327 c = XINT (args[i]); |
1328 len = CHAR_STRING (c, p, str); | |
1329 if (p != str) | |
1330 /* C is a composite character. */ | |
1331 bcopy (str, p, len); | |
1332 p += len; | |
1333 } | |
1334 | |
23427
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
1335 /* Here, we can't use make_string_from_bytes because of byte |
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
1336 combining problem. */ |
4acd4faa545a
(Fstring): Call make_string instead of
Kenichi Handa <handa@m17n.org>
parents:
23397
diff
changeset
|
1337 val = make_string (buf, p - buf); |
17052 | 1338 return val; |
1339 } | |
1340 | |
1341 #endif /* emacs */ | |
1342 | |
1343 /*** Composite characters staffs ***/ | |
1344 | |
1345 /* Each composite character is identified by CMPCHAR-ID which is | |
1346 assigned when Emacs needs the character code of the composite | |
1347 character (e.g. when displaying it on the screen). See the | |
1348 document "GENERAL NOTE on COMPOSITE CHARACTER" in `charset.h' how a | |
1349 composite character is represented in Emacs. */ | |
1350 | |
1351 /* If `static' is defined, it means that it is defined to null string. */ | |
1352 #ifndef static | |
1353 /* The following function is copied from lread.c. */ | |
1354 static int | |
1355 hash_string (ptr, len) | |
1356 unsigned char *ptr; | |
1357 int len; | |
1358 { | |
1359 register unsigned char *p = ptr; | |
1360 register unsigned char *end = p + len; | |
1361 register unsigned char c; | |
1362 register int hash = 0; | |
1363 | |
1364 while (p != end) | |
1365 { | |
1366 c = *p++; | |
1367 if (c >= 0140) c -= 40; | |
1368 hash = ((hash<<3) + (hash>>28) + c); | |
1369 } | |
1370 return hash & 07777777777; | |
1371 } | |
1372 #endif | |
1373 | |
1374 #define CMPCHAR_HASH_TABLE_SIZE 0xFFF | |
1375 | |
1376 static int *cmpchar_hash_table[CMPCHAR_HASH_TABLE_SIZE]; | |
1377 | |
1378 /* Each element of `cmpchar_hash_table' is a pointer to an array of | |
1379 integer, where the 1st element is the size of the array, the 2nd | |
1380 element is how many elements are actually used in the array, and | |
1381 the remaining elements are CMPCHAR-IDs of composite characters of | |
1382 the same hash value. */ | |
1383 #define CMPCHAR_HASH_SIZE(table) table[0] | |
1384 #define CMPCHAR_HASH_USED(table) table[1] | |
1385 #define CMPCHAR_HASH_CMPCHAR_ID(table, i) table[i] | |
1386 | |
1387 /* Return CMPCHAR-ID of the composite character in STR of the length | |
1388 LEN. If the composite character has not yet been registered, | |
1389 register it in `cmpchar_table' and assign new CMPCHAR-ID. This | |
1390 is the sole function for assigning CMPCHAR-ID. */ | |
1391 int | |
1392 str_cmpchar_id (str, len) | |
21419
95aae2ff5fcd
(string_to_non_ascii_char, multibyte_form_length)
Karl Heuer <kwzh@gnu.org>
parents:
21255
diff
changeset
|
1393 const unsigned char *str; |
17052 | 1394 int len; |
1395 { | |
1396 int hash_idx, *hashp; | |
1397 unsigned char *buf; | |
1398 int embedded_rule; /* 1 if composition rule is embedded. */ | |
1399 int chars; /* number of components. */ | |
1400 int i; | |
1401 struct cmpchar_info *cmpcharp; | |
1402 | |
1403 /* The second byte 0xFF means compostion rule is embedded. */ | |
1404 embedded_rule = (str[1] == 0xFF); | |
1405 | |
1406 /* At first, get the actual length of the composite character. */ | |
1407 { | |
21419
95aae2ff5fcd
(string_to_non_ascii_char, multibyte_form_length)
Karl Heuer <kwzh@gnu.org>
parents:
21255
diff
changeset
|
1408 const unsigned char *p, *endp = str + 1, *lastp = str + len; |
17052 | 1409 int bytes; |
1410 | |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1411 while (endp < lastp && ! CHAR_HEAD_P (*endp)) endp++; |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1412 if (endp - str < 5) |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1413 /* Any composite char have at least 5-byte length. */ |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1414 return -1; |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1415 |
17052 | 1416 chars = 0; |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1417 p = str + 1; |
17052 | 1418 while (p < endp) |
1419 { | |
23685
14c75dfaaa44
(str_cmpchar_id): Check the byte sequence for
Kenichi Handa <handa@m17n.org>
parents:
23668
diff
changeset
|
1420 if (embedded_rule) |
14c75dfaaa44
(str_cmpchar_id): Check the byte sequence for
Kenichi Handa <handa@m17n.org>
parents:
23668
diff
changeset
|
1421 { |
14c75dfaaa44
(str_cmpchar_id): Check the byte sequence for
Kenichi Handa <handa@m17n.org>
parents:
23668
diff
changeset
|
1422 p++; |
14c75dfaaa44
(str_cmpchar_id): Check the byte sequence for
Kenichi Handa <handa@m17n.org>
parents:
23668
diff
changeset
|
1423 if (p >= endp) |
14c75dfaaa44
(str_cmpchar_id): Check the byte sequence for
Kenichi Handa <handa@m17n.org>
parents:
23668
diff
changeset
|
1424 return -1; |
14c75dfaaa44
(str_cmpchar_id): Check the byte sequence for
Kenichi Handa <handa@m17n.org>
parents:
23668
diff
changeset
|
1425 } |
17052 | 1426 /* No need of checking if *P is 0xA0 because |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1427 BYTES_BY_CHAR_HEAD (0x80) surely returns 2. */ |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1428 p += BYTES_BY_CHAR_HEAD (*p - 0x20); |
17052 | 1429 chars++; |
1430 } | |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1431 if (p > endp || chars < 2 || chars > MAX_COMPONENT_COUNT) |
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1432 /* Invalid components. */ |
17052 | 1433 return -1; |
20933
5ceea9d50194
(invalid_character): New function.
Kenichi Handa <handa@m17n.org>
parents:
20890
diff
changeset
|
1434 len = p - str; |
17052 | 1435 } |
1436 hash_idx = hash_string (str, len) % CMPCHAR_HASH_TABLE_SIZE; | |
1437 hashp = cmpchar_hash_table[hash_idx]; | |
1438 | |
1439 /* Then, look into the hash table. */ | |
1440 if (hashp != NULL) | |
1441 /* Find the correct one among composite characters of the same | |
1442 hash value. */ | |
1443 for (i = 2; i < CMPCHAR_HASH_USED (hashp); i++) | |
1444 { | |
1445 cmpcharp = cmpchar_table[CMPCHAR_HASH_CMPCHAR_ID (hashp, i)]; | |
1446 if (len == cmpcharp->len | |
1447 && ! bcmp (str, cmpcharp->data, len)) | |
1448 return CMPCHAR_HASH_CMPCHAR_ID (hashp, i); | |
1449 } | |
1450 | |
1451 /* We have to register the composite character in cmpchar_table. */ | |
17186
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
1452 if (n_cmpchars > (CHAR_FIELD2_MASK | CHAR_FIELD3_MASK)) |
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
1453 /* No, we have no more room for a new composite character. */ |
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
1454 return -1; |
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
1455 |
17052 | 1456 /* Make the entry in hash table. */ |
1457 if (hashp == NULL) | |
1458 { | |
1459 /* Make a table for 8 composite characters initially. */ | |
1460 hashp = (cmpchar_hash_table[hash_idx] | |
1461 = (int *) xmalloc (sizeof (int) * (2 + 8))); | |
1462 CMPCHAR_HASH_SIZE (hashp) = 10; | |
1463 CMPCHAR_HASH_USED (hashp) = 2; | |
1464 } | |
1465 else if (CMPCHAR_HASH_USED (hashp) >= CMPCHAR_HASH_SIZE (hashp)) | |
1466 { | |
1467 CMPCHAR_HASH_SIZE (hashp) += 8; | |
1468 hashp = (cmpchar_hash_table[hash_idx] | |
1469 = (int *) xrealloc (hashp, | |
1470 sizeof (int) * CMPCHAR_HASH_SIZE (hashp))); | |
1471 } | |
1472 CMPCHAR_HASH_CMPCHAR_ID (hashp, CMPCHAR_HASH_USED (hashp)) = n_cmpchars; | |
1473 CMPCHAR_HASH_USED (hashp)++; | |
1474 | |
1475 /* Set information of the composite character in cmpchar_table. */ | |
1476 if (cmpchar_table_size == 0) | |
1477 { | |
1478 /* This is the first composite character to be registered. */ | |
1479 cmpchar_table_size = 256; | |
1480 cmpchar_table | |
1481 = (struct cmpchar_info **) xmalloc (sizeof (cmpchar_table[0]) | |
1482 * cmpchar_table_size); | |
1483 } | |
1484 else if (cmpchar_table_size <= n_cmpchars) | |
1485 { | |
1486 cmpchar_table_size += 256; | |
1487 cmpchar_table | |
1488 = (struct cmpchar_info **) xrealloc (cmpchar_table, | |
1489 sizeof (cmpchar_table[0]) | |
1490 * cmpchar_table_size); | |
1491 } | |
1492 | |
1493 cmpcharp = (struct cmpchar_info *) xmalloc (sizeof (struct cmpchar_info)); | |
1494 | |
1495 cmpcharp->len = len; | |
1496 cmpcharp->data = (unsigned char *) xmalloc (len + 1); | |
1497 bcopy (str, cmpcharp->data, len); | |
1498 cmpcharp->data[len] = 0; | |
1499 cmpcharp->glyph_len = chars; | |
1500 cmpcharp->glyph = (GLYPH *) xmalloc (sizeof (GLYPH) * chars); | |
1501 if (embedded_rule) | |
1502 { | |
1503 cmpcharp->cmp_rule = (unsigned char *) xmalloc (chars); | |
1504 cmpcharp->col_offset = (float *) xmalloc (sizeof (float) * chars); | |
1505 } | |
1506 else | |
1507 { | |
1508 cmpcharp->cmp_rule = NULL; | |
1509 cmpcharp->col_offset = NULL; | |
1510 } | |
1511 | |
1512 /* Setup GLYPH data and composition rules (if any) so as not to make | |
1513 them every time on displaying. */ | |
1514 { | |
1515 unsigned char *bufp; | |
1516 int width; | |
1517 float leftmost = 0.0, rightmost = 1.0; | |
1518 | |
1519 if (embedded_rule) | |
1520 /* At first, col_offset[N] is set to relative to col_offset[0]. */ | |
1521 cmpcharp->col_offset[0] = 0; | |
1522 | |
1523 for (i = 0, bufp = cmpcharp->data + 1; i < chars; i++) | |
1524 { | |
1525 if (embedded_rule) | |
1526 cmpcharp->cmp_rule[i] = *bufp++; | |
1527 | |
1528 if (*bufp == 0xA0) /* This is an ASCII character. */ | |
1529 { | |
1530 cmpcharp->glyph[i] = FAST_MAKE_GLYPH ((*++bufp & 0x7F), 0); | |
1531 width = 1; | |
1532 bufp++; | |
1533 } | |
1534 else /* Multibyte character. */ | |
1535 { | |
1536 /* Make `bufp' point normal multi-byte form temporally. */ | |
1537 *bufp -= 0x20; | |
1538 cmpcharp->glyph[i] | |
22185
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
1539 = FAST_MAKE_GLYPH (string_to_non_ascii_char (bufp, 4, 0, 0), 0); |
17052 | 1540 width = WIDTH_BY_CHAR_HEAD (*bufp); |
1541 *bufp += 0x20; | |
1542 bufp += BYTES_BY_CHAR_HEAD (*bufp - 0x20); | |
1543 } | |
1544 | |
1545 if (embedded_rule && i > 0) | |
1546 { | |
1547 /* Reference points (global_ref and new_ref) are | |
1548 encoded as below: | |
1549 | |
1550 0--1--2 -- ascent | |
1551 | | | |
1552 | | | |
1553 | 4 -+--- center | |
1554 -- 3 5 -- baseline | |
1555 | | | |
1556 6--7--8 -- descent | |
1557 | |
1558 Now, we calculate the column offset of the new glyph | |
1559 from the left edge of the first glyph. This can avoid | |
1560 the same calculation everytime displaying this | |
1561 composite character. */ | |
1562 | |
1563 /* Reference points of global glyph and new glyph. */ | |
1564 int global_ref = (cmpcharp->cmp_rule[i] - 0xA0) / 9; | |
1565 int new_ref = (cmpcharp->cmp_rule[i] - 0xA0) % 9; | |
1566 /* Column offset relative to the first glyph. */ | |
1567 float left = (leftmost | |
1568 + (global_ref % 3) * (rightmost - leftmost) / 2.0 | |
1569 - (new_ref % 3) * width / 2.0); | |
1570 | |
1571 cmpcharp->col_offset[i] = left; | |
1572 if (left < leftmost) | |
1573 leftmost = left; | |
1574 if (left + width > rightmost) | |
1575 rightmost = left + width; | |
1576 } | |
1577 else | |
1578 { | |
1579 if (width > rightmost) | |
1580 rightmost = width; | |
1581 } | |
1582 } | |
1583 if (embedded_rule) | |
1584 { | |
1585 /* Now col_offset[N] are relative to the left edge of the | |
1586 first component. Make them relative to the left edge of | |
1587 overall glyph. */ | |
1588 for (i = 0; i < chars; i++) | |
1589 cmpcharp->col_offset[i] -= leftmost; | |
1590 /* Make rightmost holds width of overall glyph. */ | |
1591 rightmost -= leftmost; | |
1592 } | |
1593 | |
1594 cmpcharp->width = rightmost; | |
1595 if (cmpcharp->width < rightmost) | |
1596 /* To get a ceiling integer value. */ | |
1597 cmpcharp->width++; | |
1598 } | |
1599 | |
1600 cmpchar_table[n_cmpchars] = cmpcharp; | |
1601 | |
1602 return n_cmpchars++; | |
1603 } | |
1604 | |
23668
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1605 /* Return the Nth element of the composite character C. If NOERROR is |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1606 nonzero, return 0 on error condition (C is an invalid composite |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1607 charcter, or N is out of range). */ |
17052 | 1608 int |
23668
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1609 cmpchar_component (c, n, noerror) |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1610 int c, n, noerror; |
17052 | 1611 { |
1612 int id = COMPOSITE_CHAR_ID (c); | |
1613 | |
23668
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1614 if (id < 0 || id >= n_cmpchars) |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1615 { |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1616 /* C is not a valid composite character. */ |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1617 if (noerror) return 0; |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1618 error ("Invalid composite character: %d", c) ; |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1619 } |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1620 if (n >= cmpchar_table[id]->glyph_len) |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1621 { |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1622 /* No such component. */ |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1623 if (noerror) return 0; |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1624 args_out_of_range (make_number (c), make_number (n)); |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1625 } |
17052 | 1626 /* No face data is stored in glyph code. */ |
1627 return ((int) (cmpchar_table[id]->glyph[n])); | |
1628 } | |
1629 | |
1630 DEFUN ("cmpcharp", Fcmpcharp, Scmpcharp, 1, 1, 0, | |
1631 "T if CHAR is a composite character.") | |
1632 (ch) | |
1633 Lisp_Object ch; | |
1634 { | |
1635 CHECK_NUMBER (ch, 0); | |
1636 return (COMPOSITE_CHAR_P (XINT (ch)) ? Qt : Qnil); | |
1637 } | |
1638 | |
1639 DEFUN ("composite-char-component", Fcmpchar_component, Scmpchar_component, | |
1640 2, 2, 0, | |
23668
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1641 "Return the Nth component character of composite character CHARACTER.") |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1642 (character, n) |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1643 Lisp_Object character, n; |
17052 | 1644 { |
23668
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1645 int id; |
17052 | 1646 |
1647 CHECK_NUMBER (character, 0); | |
23668
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1648 CHECK_NUMBER (n, 1); |
17052 | 1649 |
23668
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1650 return (make_number (cmpchar_component (XINT (character), XINT (n), 0))); |
17052 | 1651 } |
1652 | |
1653 DEFUN ("composite-char-composition-rule", Fcmpchar_cmp_rule, Scmpchar_cmp_rule, | |
1654 2, 2, 0, | |
23668
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1655 "Return the Nth composition rule of composite character CHARACTER.\n\ |
17121
f8e2cd94c274
(Fcmpchar_cmp_rule): Escape newlines in docstring.
Kenichi Handa <handa@m17n.org>
parents:
17071
diff
changeset
|
1656 The returned rule is for composing the Nth component\n\ |
23668
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1657 on the (N-1)th component.\n\ |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1658 If CHARACTER should be composed relatively or N is 0, return 255.") |
17121
f8e2cd94c274
(Fcmpchar_cmp_rule): Escape newlines in docstring.
Kenichi Handa <handa@m17n.org>
parents:
17071
diff
changeset
|
1659 (character, n) |
f8e2cd94c274
(Fcmpchar_cmp_rule): Escape newlines in docstring.
Kenichi Handa <handa@m17n.org>
parents:
17071
diff
changeset
|
1660 Lisp_Object character, n; |
17052 | 1661 { |
23668
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1662 int id; |
17052 | 1663 |
1664 CHECK_NUMBER (character, 0); | |
17121
f8e2cd94c274
(Fcmpchar_cmp_rule): Escape newlines in docstring.
Kenichi Handa <handa@m17n.org>
parents:
17071
diff
changeset
|
1665 CHECK_NUMBER (n, 1); |
17052 | 1666 |
1667 id = COMPOSITE_CHAR_ID (XINT (character)); | |
1668 if (id < 0 || id >= n_cmpchars) | |
1669 error ("Invalid composite character: %d", XINT (character)); | |
23668
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1670 if (XINT (n) < 0 || XINT (n) >= cmpchar_table[id]->glyph_len) |
17121
f8e2cd94c274
(Fcmpchar_cmp_rule): Escape newlines in docstring.
Kenichi Handa <handa@m17n.org>
parents:
17071
diff
changeset
|
1671 args_out_of_range (character, n); |
17052 | 1672 |
23668
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1673 return make_number (cmpchar_table[id]->cmp_rule |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1674 ? cmpchar_table[id]->cmp_rule[XINT (n)] |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1675 : 255); |
17052 | 1676 } |
1677 | |
1678 DEFUN ("composite-char-composition-rule-p", Fcmpchar_cmp_rule_p, | |
1679 Scmpchar_cmp_rule_p, 1, 1, 0, | |
1680 "Return non-nil if composite character CHARACTER contains a embedded rule.") | |
1681 (character) | |
1682 Lisp_Object character; | |
1683 { | |
1684 int id; | |
1685 | |
1686 CHECK_NUMBER (character, 0); | |
1687 id = COMPOSITE_CHAR_ID (XINT (character)); | |
1688 if (id < 0 || id >= n_cmpchars) | |
1689 error ("Invalid composite character: %d", XINT (character)); | |
1690 | |
1691 return (cmpchar_table[id]->cmp_rule ? Qt : Qnil); | |
1692 } | |
1693 | |
1694 DEFUN ("composite-char-component-count", Fcmpchar_cmp_count, | |
1695 Scmpchar_cmp_count, 1, 1, 0, | |
1696 "Return number of compoents of composite character CHARACTER.") | |
1697 (character) | |
1698 Lisp_Object character; | |
1699 { | |
1700 int id; | |
1701 | |
1702 CHECK_NUMBER (character, 0); | |
1703 id = COMPOSITE_CHAR_ID (XINT (character)); | |
1704 if (id < 0 || id >= n_cmpchars) | |
1705 error ("Invalid composite character: %d", XINT (character)); | |
1706 | |
1707 return (make_number (cmpchar_table[id]->glyph_len)); | |
1708 } | |
1709 | |
1710 DEFUN ("compose-string", Fcompose_string, Scompose_string, | |
1711 1, 1, 0, | |
1712 "Return one char string composed from all characters in STRING.") | |
1713 (str) | |
1714 Lisp_Object str; | |
1715 { | |
1716 unsigned char buf[MAX_LENGTH_OF_MULTI_BYTE_FORM], *p, *pend, *ptemp; | |
1717 int len, i; | |
1718 | |
1719 CHECK_STRING (str, 0); | |
1720 | |
1721 buf[0] = LEADING_CODE_COMPOSITION; | |
1722 p = XSTRING (str)->data; | |
21244
50929073a0ba
Use STRING_BYTES and SET_STRING_BYTES.
Richard M. Stallman <rms@gnu.org>
parents:
21063
diff
changeset
|
1723 pend = p + STRING_BYTES (XSTRING (str)); |
17052 | 1724 i = 1; |
1725 while (p < pend) | |
1726 { | |
23685
14c75dfaaa44
(str_cmpchar_id): Check the byte sequence for
Kenichi Handa <handa@m17n.org>
parents:
23668
diff
changeset
|
1727 if (*p < 0x20) /* control code */ |
17052 | 1728 error ("Invalid component character: %d", *p); |
1729 else if (*p < 0x80) /* ASCII */ | |
1730 { | |
1731 if (i + 2 >= MAX_LENGTH_OF_MULTI_BYTE_FORM) | |
1732 error ("Too long string to be composed: %s", XSTRING (str)->data); | |
1733 /* Prepend an ASCII charset indicator 0xA0, set MSB of the | |
1734 code itself. */ | |
1735 buf[i++] = 0xA0; | |
1736 buf[i++] = *p++ + 0x80; | |
1737 } | |
1738 else if (*p == LEADING_CODE_COMPOSITION) /* composite char */ | |
1739 { | |
1740 /* Already composed. Eliminate the heading | |
1741 LEADING_CODE_COMPOSITION, keep the remaining bytes | |
1742 unchanged. */ | |
1743 p++; | |
23668
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1744 if (*p == 255) |
23faf76ec325
(cmpchar_component): New arg NOERROR. Check
Kenichi Handa <handa@m17n.org>
parents:
23665
diff
changeset
|
1745 error ("Can't compose a rule-based composition character"); |
17052 | 1746 ptemp = p; |
20534
066251bcf9f7
(Fchars_in_string): Update call to CHAR_HEAD_P.
Richard M. Stallman <rms@gnu.org>
parents:
20435
diff
changeset
|
1747 while (! CHAR_HEAD_P (*p)) p++; |
23685
14c75dfaaa44
(str_cmpchar_id): Check the byte sequence for
Kenichi Handa <handa@m17n.org>
parents:
23668
diff
changeset
|
1748 if (str_cmpchar_id (ptemp - 1, p - ptemp + 1) < 0) |
14c75dfaaa44
(str_cmpchar_id): Check the byte sequence for
Kenichi Handa <handa@m17n.org>
parents:
23668
diff
changeset
|
1749 error ("Can't compose an invalid composition character"); |
17052 | 1750 if (i + (p - ptemp) >= MAX_LENGTH_OF_MULTI_BYTE_FORM) |
1751 error ("Too long string to be composed: %s", XSTRING (str)->data); | |
1752 bcopy (ptemp, buf + i, p - ptemp); | |
1753 i += p - ptemp; | |
1754 } | |
1755 else /* multibyte char */ | |
1756 { | |
1757 /* Add 0x20 to the base leading-code, keep the remaining | |
1758 bytes unchanged. */ | |
23685
14c75dfaaa44
(str_cmpchar_id): Check the byte sequence for
Kenichi Handa <handa@m17n.org>
parents:
23668
diff
changeset
|
1759 int c = STRING_CHAR_AND_CHAR_LENGTH (p, pend - p, len); |
14c75dfaaa44
(str_cmpchar_id): Check the byte sequence for
Kenichi Handa <handa@m17n.org>
parents:
23668
diff
changeset
|
1760 |
14c75dfaaa44
(str_cmpchar_id): Check the byte sequence for
Kenichi Handa <handa@m17n.org>
parents:
23668
diff
changeset
|
1761 if (len <= 1 || ! CHAR_VALID_P (c, 0)) |
14c75dfaaa44
(str_cmpchar_id): Check the byte sequence for
Kenichi Handa <handa@m17n.org>
parents:
23668
diff
changeset
|
1762 error ("Can't compose an invalid character"); |
17052 | 1763 if (i + len >= MAX_LENGTH_OF_MULTI_BYTE_FORM) |
1764 error ("Too long string to be composed: %s", XSTRING (str)->data); | |
1765 bcopy (p, buf + i, len); | |
1766 buf[i] += 0x20; | |
1767 p += len, i += len; | |
1768 } | |
1769 } | |
1770 | |
1771 if (i < 5) | |
1772 /* STR contains only one character, which can't be composed. */ | |
1773 error ("Too short string to be composed: %s", XSTRING (str)->data); | |
1774 | |
21255
3d75330ce275
(Fstring): Use make_string_from_bytes.
Richard M. Stallman <rms@gnu.org>
parents:
21244
diff
changeset
|
1775 return make_string_from_bytes (buf, 1, i); |
17052 | 1776 } |
1777 | |
1778 | |
21514 | 1779 int |
17052 | 1780 charset_id_internal (charset_name) |
1781 char *charset_name; | |
1782 { | |
22927
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
1783 Lisp_Object val; |
17052 | 1784 |
22927
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
1785 val= Fget (intern (charset_name), Qcharset); |
17052 | 1786 if (!VECTORP (val)) |
1787 error ("Charset %s is not defined", charset_name); | |
1788 | |
1789 return (XINT (XVECTOR (val)->contents[0])); | |
1790 } | |
1791 | |
1792 DEFUN ("setup-special-charsets", Fsetup_special_charsets, | |
1793 Ssetup_special_charsets, 0, 0, 0, "Internal use only.") | |
1794 () | |
1795 { | |
1796 charset_latin_iso8859_1 = charset_id_internal ("latin-iso8859-1"); | |
1797 charset_jisx0208_1978 = charset_id_internal ("japanese-jisx0208-1978"); | |
1798 charset_jisx0208 = charset_id_internal ("japanese-jisx0208"); | |
1799 charset_katakana_jisx0201 = charset_id_internal ("katakana-jisx0201"); | |
1800 charset_latin_jisx0201 = charset_id_internal ("latin-jisx0201"); | |
1801 charset_big5_1 = charset_id_internal ("chinese-big5-1"); | |
1802 charset_big5_2 = charset_id_internal ("chinese-big5-2"); | |
1803 return Qnil; | |
1804 } | |
1805 | |
21514 | 1806 void |
17052 | 1807 init_charset_once () |
1808 { | |
1809 int i, j, k; | |
1810 | |
1811 staticpro (&Vcharset_table); | |
1812 staticpro (&Vcharset_symbol_table); | |
22703
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1813 staticpro (&Vgeneric_character_list); |
17052 | 1814 |
1815 /* This has to be done here, before we call Fmake_char_table. */ | |
1816 Qcharset_table = intern ("charset-table"); | |
1817 staticpro (&Qcharset_table); | |
1818 | |
1819 /* Intern this now in case it isn't already done. | |
1820 Setting this variable twice is harmless. | |
1821 But don't staticpro it here--that is done in alloc.c. */ | |
1822 Qchar_table_extra_slots = intern ("char-table-extra-slots"); | |
1823 | |
1824 /* Now we are ready to set up this property, so we can | |
1825 create the charset table. */ | |
1826 Fput (Qcharset_table, Qchar_table_extra_slots, make_number (0)); | |
1827 Vcharset_table = Fmake_char_table (Qcharset_table, Qnil); | |
1828 | |
17186
65bc06706714
Adjusted for the change of MAX_CHARSET.
Kenichi Handa <handa@m17n.org>
parents:
17121
diff
changeset
|
1829 Vcharset_symbol_table = Fmake_vector (make_number (MAX_CHARSET + 1), Qnil); |
17052 | 1830 |
1831 /* Setup tables. */ | |
1832 for (i = 0; i < 2; i++) | |
1833 for (j = 0; j < 2; j++) | |
1834 for (k = 0; k < 128; k++) | |
1835 iso_charset_table [i][j][k] = -1; | |
1836 | |
1837 bzero (cmpchar_hash_table, sizeof cmpchar_hash_table); | |
1838 cmpchar_table_size = n_cmpchars = 0; | |
1839 | |
23656
031653c4634a
(init_charset_once): Fix previous change.
Kenichi Handa <handa@m17n.org>
parents:
23654
diff
changeset
|
1840 for (i = 0; i < 256; i++) |
17052 | 1841 BYTES_BY_CHAR_HEAD (i) = 1; |
23628
4a0a107fac9e
(string_to_non_ascii_char): Change the check for the
Kenichi Handa <handa@m17n.org>
parents:
23576
diff
changeset
|
1842 for (i = MIN_CHARSET_OFFICIAL_DIMENSION1; |
4a0a107fac9e
(string_to_non_ascii_char): Change the check for the
Kenichi Handa <handa@m17n.org>
parents:
23576
diff
changeset
|
1843 i <= MAX_CHARSET_OFFICIAL_DIMENSION1; i++) |
4a0a107fac9e
(string_to_non_ascii_char): Change the check for the
Kenichi Handa <handa@m17n.org>
parents:
23576
diff
changeset
|
1844 BYTES_BY_CHAR_HEAD (i) = 2; |
4a0a107fac9e
(string_to_non_ascii_char): Change the check for the
Kenichi Handa <handa@m17n.org>
parents:
23576
diff
changeset
|
1845 for (i = MIN_CHARSET_OFFICIAL_DIMENSION2; |
4a0a107fac9e
(string_to_non_ascii_char): Change the check for the
Kenichi Handa <handa@m17n.org>
parents:
23576
diff
changeset
|
1846 i <= MAX_CHARSET_OFFICIAL_DIMENSION2; i++) |
4a0a107fac9e
(string_to_non_ascii_char): Change the check for the
Kenichi Handa <handa@m17n.org>
parents:
23576
diff
changeset
|
1847 BYTES_BY_CHAR_HEAD (i) = 3; |
17052 | 1848 BYTES_BY_CHAR_HEAD (LEADING_CODE_PRIVATE_11) = 3; |
1849 BYTES_BY_CHAR_HEAD (LEADING_CODE_PRIVATE_12) = 3; | |
1850 BYTES_BY_CHAR_HEAD (LEADING_CODE_PRIVATE_21) = 4; | |
1851 BYTES_BY_CHAR_HEAD (LEADING_CODE_PRIVATE_22) = 4; | |
23628
4a0a107fac9e
(string_to_non_ascii_char): Change the check for the
Kenichi Handa <handa@m17n.org>
parents:
23576
diff
changeset
|
1852 /* The followings don't reflect the actual bytes, but just to tell |
17052 | 1853 that it is a start of a multibyte character. */ |
1854 BYTES_BY_CHAR_HEAD (LEADING_CODE_COMPOSITION) = 2; | |
23628
4a0a107fac9e
(string_to_non_ascii_char): Change the check for the
Kenichi Handa <handa@m17n.org>
parents:
23576
diff
changeset
|
1855 BYTES_BY_CHAR_HEAD (0x9E) = 2; |
4a0a107fac9e
(string_to_non_ascii_char): Change the check for the
Kenichi Handa <handa@m17n.org>
parents:
23576
diff
changeset
|
1856 BYTES_BY_CHAR_HEAD (0x9F) = 2; |
17052 | 1857 |
1858 for (i = 0; i < 128; i++) | |
1859 WIDTH_BY_CHAR_HEAD (i) = 1; | |
1860 for (; i < 256; i++) | |
1861 WIDTH_BY_CHAR_HEAD (i) = 4; | |
1862 WIDTH_BY_CHAR_HEAD (LEADING_CODE_PRIVATE_11) = 1; | |
1863 WIDTH_BY_CHAR_HEAD (LEADING_CODE_PRIVATE_12) = 2; | |
1864 WIDTH_BY_CHAR_HEAD (LEADING_CODE_PRIVATE_21) = 1; | |
1865 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
|
1866 |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1867 { |
22927
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
1868 Lisp_Object val; |
22703
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1869 |
22927
209acf11198c
(multibyte_char_to_unibyte): New function.
Kenichi Handa <handa@m17n.org>
parents:
22774
diff
changeset
|
1870 val = Qnil; |
22703
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1871 for (i = 0x81; i < 0x90; i++) |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1872 val = Fcons (make_number ((i - 0x70) << 7), val); |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1873 for (; i < 0x9A; i++) |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1874 val = Fcons (make_number ((i - 0x8F) << 14), val); |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1875 for (i = 0xA0; i < 0xF0; i++) |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1876 val = Fcons (make_number ((i - 0x70) << 7), val); |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1877 for (; i < 0xFF; i++) |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1878 val = Fcons (make_number ((i - 0xE0) << 14), val); |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1879 val = Fcons (make_number (GENERIC_COMPOSITION_CHAR), val); |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1880 Vgeneric_character_list = Fnreverse (val); |
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1881 } |
23154
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
1882 |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
1883 nonascii_insert_offset = 0; |
a9cbf64603ca
(unibyte_char_to_multibyte):
Kenichi Handa <handa@m17n.org>
parents:
23140
diff
changeset
|
1884 Vnonascii_translation_table = Qnil; |
17052 | 1885 } |
1886 | |
1887 #ifdef emacs | |
1888 | |
21514 | 1889 void |
17052 | 1890 syms_of_charset () |
1891 { | |
1892 Qascii = intern ("ascii"); | |
1893 staticpro (&Qascii); | |
1894 | |
1895 Qcharset = intern ("charset"); | |
1896 staticpro (&Qcharset); | |
1897 | |
1898 /* Define ASCII charset now. */ | |
1899 update_charset_table (make_number (CHARSET_ASCII), | |
1900 make_number (1), make_number (94), | |
1901 make_number (1), | |
1902 make_number (0), | |
1903 make_number ('B'), | |
1904 make_number (0), | |
1905 build_string ("ASCII"), | |
1906 build_string ("ASCII"), | |
1907 build_string ("ASCII (ISO646 IRV)")); | |
1908 CHARSET_SYMBOL (CHARSET_ASCII) = Qascii; | |
1909 Fput (Qascii, Qcharset, CHARSET_TABLE_ENTRY (CHARSET_ASCII)); | |
1910 | |
1911 Qcomposition = intern ("composition"); | |
1912 staticpro (&Qcomposition); | |
1913 CHARSET_SYMBOL (CHARSET_COMPOSITION) = Qcomposition; | |
1914 | |
23489
09201cebfa21
(Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents:
23427
diff
changeset
|
1915 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
|
1916 staticpro (&Qauto_fill_chars); |
09201cebfa21
(Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents:
23427
diff
changeset
|
1917 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
|
1918 |
17052 | 1919 defsubr (&Sdefine_charset); |
22703
15e0119749cf
(Vgeneric_character_list): New variable.
Kenichi Handa <handa@m17n.org>
parents:
22524
diff
changeset
|
1920 defsubr (&Sgeneric_character_list); |
20435
099812ea310f
(Fget_unused_iso_final_char): New function.
Kenichi Handa <handa@m17n.org>
parents:
20362
diff
changeset
|
1921 defsubr (&Sget_unused_iso_final_char); |
17052 | 1922 defsubr (&Sdeclare_equiv_charset); |
1923 defsubr (&Sfind_charset_region); | |
1924 defsubr (&Sfind_charset_string); | |
1925 defsubr (&Smake_char_internal); | |
1926 defsubr (&Ssplit_char); | |
1927 defsubr (&Schar_charset); | |
21445
4c0b4a1025cd
(string_to_non_ascii_char): Include garbage bytes (if
Kenichi Handa <handa@m17n.org>
parents:
21419
diff
changeset
|
1928 defsubr (&Scharset_after); |
17052 | 1929 defsubr (&Siso_charset); |
20152
9a3343c36e83
(char_valid_p): New function
Kenichi Handa <handa@m17n.org>
parents:
19983
diff
changeset
|
1930 defsubr (&Schar_valid_p); |
21034
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1931 defsubr (&Sunibyte_char_to_multibyte); |
23060
5036ee3de0ac
(Fmultibyte_char_to_unibyte): New function.
Richard M. Stallman <rms@gnu.org>
parents:
22927
diff
changeset
|
1932 defsubr (&Smultibyte_char_to_unibyte); |
17052 | 1933 defsubr (&Schar_bytes); |
1934 defsubr (&Schar_width); | |
1935 defsubr (&Sstring_width); | |
1936 defsubr (&Schar_direction); | |
19415
309b9a0d8a45
(Fchars_in_region): Renamed form Fcount_chars_region.
Kenichi Handa <handa@m17n.org>
parents:
19384
diff
changeset
|
1937 defsubr (&Schars_in_region); |
20614
a4800f9842e4
(multibyte_chars_in_text): New function.
Richard M. Stallman <rms@gnu.org>
parents:
20575
diff
changeset
|
1938 defsubr (&Sstring); |
17052 | 1939 defsubr (&Scmpcharp); |
1940 defsubr (&Scmpchar_component); | |
1941 defsubr (&Scmpchar_cmp_rule); | |
1942 defsubr (&Scmpchar_cmp_rule_p); | |
1943 defsubr (&Scmpchar_cmp_count); | |
1944 defsubr (&Scompose_string); | |
1945 defsubr (&Ssetup_special_charsets); | |
1946 | |
1947 DEFVAR_LISP ("charset-list", &Vcharset_list, | |
1948 "List of charsets ever defined."); | |
1949 Vcharset_list = Fcons (Qascii, Qnil); | |
1950 | |
22185
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
1951 DEFVAR_LISP ("translation-table-vector", &Vtranslation_table_vector, |
22121
6e79a15594ab
Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents:
21718
diff
changeset
|
1952 "Vector of cons cell of a symbol and translation table ever defined.\n\ |
6e79a15594ab
Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents:
21718
diff
changeset
|
1953 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
|
1954 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
|
1955 |
17052 | 1956 DEFVAR_INT ("leading-code-composition", &leading_code_composition, |
1957 "Leading-code of composite characters."); | |
1958 leading_code_composition = LEADING_CODE_COMPOSITION; | |
1959 | |
1960 DEFVAR_INT ("leading-code-private-11", &leading_code_private_11, | |
1961 "Leading-code of private TYPE9N charset of column-width 1."); | |
1962 leading_code_private_11 = LEADING_CODE_PRIVATE_11; | |
1963 | |
1964 DEFVAR_INT ("leading-code-private-12", &leading_code_private_12, | |
1965 "Leading-code of private TYPE9N charset of column-width 2."); | |
1966 leading_code_private_12 = LEADING_CODE_PRIVATE_12; | |
1967 | |
1968 DEFVAR_INT ("leading-code-private-21", &leading_code_private_21, | |
1969 "Leading-code of private TYPE9Nx9N charset of column-width 1."); | |
1970 leading_code_private_21 = LEADING_CODE_PRIVATE_21; | |
1971 | |
1972 DEFVAR_INT ("leading-code-private-22", &leading_code_private_22, | |
1973 "Leading-code of private TYPE9Nx9N charset of column-width 2."); | |
1974 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
|
1975 |
a949dc65fbea
(nonascii_insert_offset): Variable moved from cmds.c.
Richard M. Stallman <rms@gnu.org>
parents:
20636
diff
changeset
|
1976 DEFVAR_INT ("nonascii-insert-offset", &nonascii_insert_offset, |
21034
f3128abaf8aa
(DEFAULT_NONASCII_INSERT_OFFSET): Macro definition is
Kenichi Handa <handa@m17n.org>
parents:
20933
diff
changeset
|
1977 "Offset for converting non-ASCII unibyte codes 0240...0377 to multibyte.\n\ |
20890
c7cfd531cf2b
(Vnonascii_translate_table): New variable.
Richard M. Stallman <rms@gnu.org>
parents:
20833
diff
changeset
|
1978 This is used for converting unibyte text to multibyte,\n\ |
c7cfd531cf2b
(Vnonascii_translate_table): New variable.
Richard M. Stallman <rms@gnu.org>
parents:
20833
diff
changeset
|
1979 and for inserting character codes specified by number.\n\n\ |
21718
361e50e075d8
(syms_of_charset): Doc fix for nonascii-insert-offset.
Richard M. Stallman <rms@gnu.org>
parents:
21715
diff
changeset
|
1980 This serves to convert a Latin-1 or similar 8-bit character code\n\ |
361e50e075d8
(syms_of_charset): Doc fix for nonascii-insert-offset.
Richard M. Stallman <rms@gnu.org>
parents:
21715
diff
changeset
|
1981 to the corresponding Emacs multibyte character code.\n\ |
361e50e075d8
(syms_of_charset): Doc fix for nonascii-insert-offset.
Richard M. Stallman <rms@gnu.org>
parents:
21715
diff
changeset
|
1982 Typically the value should be (- (make-char CHARSET 0) 128),\n\ |
361e50e075d8
(syms_of_charset): Doc fix for nonascii-insert-offset.
Richard M. Stallman <rms@gnu.org>
parents:
21715
diff
changeset
|
1983 for your choice of character set.\n\ |
22185
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
1984 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
|
1985 nonascii_insert_offset = 0; |
20720
f084bb8d535f
(min_composite_char): New variable.
Kenichi Handa <handa@m17n.org>
parents:
20705
diff
changeset
|
1986 |
22121
6e79a15594ab
Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents:
21718
diff
changeset
|
1987 DEFVAR_LISP ("nonascii-translation-table", &Vnonascii_translation_table, |
22185
80a2aa51a6e1
Change term "character translation table" to
Kenichi Handa <handa@m17n.org>
parents:
22168
diff
changeset
|
1988 "Translation table to convert non-ASCII unibyte codes to multibyte.\n\ |
20890
c7cfd531cf2b
(Vnonascii_translate_table): New variable.
Richard M. Stallman <rms@gnu.org>
parents:
20833
diff
changeset
|
1989 This is used for converting unibyte text to multibyte,\n\ |
c7cfd531cf2b
(Vnonascii_translate_table): New variable.
Richard M. Stallman <rms@gnu.org>
parents:
20833
diff
changeset
|
1990 and for inserting character codes specified by number.\n\n\ |
c7cfd531cf2b
(Vnonascii_translate_table): New variable.
Richard M. Stallman <rms@gnu.org>
parents:
20833
diff
changeset
|
1991 Conversion is performed only when multibyte characters are enabled,\n\ |
c7cfd531cf2b
(Vnonascii_translate_table): New variable.
Richard M. Stallman <rms@gnu.org>
parents:
20833
diff
changeset
|
1992 and it serves to convert a Latin-1 or similar 8-bit character code\n\ |
c7cfd531cf2b
(Vnonascii_translate_table): New variable.
Richard M. Stallman <rms@gnu.org>
parents:
20833
diff
changeset
|
1993 to the corresponding Emacs character code.\n\n\ |
22289
141f0392fb81
(syms_of_charset): Put \n\ at eol of docstring.
Karl Heuer <kwzh@gnu.org>
parents:
22185
diff
changeset
|
1994 If this is nil, `nonascii-insert-offset' is used instead.\n\ |
22121
6e79a15594ab
Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents:
21718
diff
changeset
|
1995 See also the docstring of `make-translation-table'."); |
6e79a15594ab
Change terms unify/unification to
Kenichi Handa <handa@m17n.org>
parents:
21718
diff
changeset
|
1996 Vnonascii_translation_table = Qnil; |
20890
c7cfd531cf2b
(Vnonascii_translate_table): New variable.
Richard M. Stallman <rms@gnu.org>
parents:
20833
diff
changeset
|
1997 |
20720
f084bb8d535f
(min_composite_char): New variable.
Kenichi Handa <handa@m17n.org>
parents:
20705
diff
changeset
|
1998 DEFVAR_INT ("min-composite-char", &min_composite_char, |
f084bb8d535f
(min_composite_char): New variable.
Kenichi Handa <handa@m17n.org>
parents:
20705
diff
changeset
|
1999 "Minimum character code of a composite character."); |
f084bb8d535f
(min_composite_char): New variable.
Kenichi Handa <handa@m17n.org>
parents:
20705
diff
changeset
|
2000 min_composite_char = MIN_CHAR_COMPOSITION; |
23489
09201cebfa21
(Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents:
23427
diff
changeset
|
2001 |
09201cebfa21
(Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents:
23427
diff
changeset
|
2002 DEFVAR_LISP ("auto-fill-chars", &Vauto_fill_chars, |
09201cebfa21
(Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents:
23427
diff
changeset
|
2003 "A char-table for characters which invoke auto-filling.\n\ |
09201cebfa21
(Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents:
23427
diff
changeset
|
2004 Such characters has value t in this table."); |
09201cebfa21
(Vauto_fill_chars, Qauto_fill_chars): New variables.
Kenichi Handa <handa@m17n.org>
parents:
23427
diff
changeset
|
2005 Vauto_fill_chars = Fmake_char_table (Qauto_fill_chars, Qnil); |
23576 | 2006 CHAR_TABLE_SET (Vauto_fill_chars, make_number (' '), Qt); |
2007 CHAR_TABLE_SET (Vauto_fill_chars, make_number ('\n'), Qt); | |
17052 | 2008 } |
2009 | |
2010 #endif /* emacs */ |