Mercurial > emacs
annotate src/character.c @ 90765:bde48639c22d
*** empty log message ***
author | Kenichi Handa <handa@m17n.org> |
---|---|
date | Fri, 16 Feb 2007 06:37:36 +0000 |
parents | 2951f3d44eba |
children | b74794d7f11a |
rev | line source |
---|---|
88363 | 1 /* Basic character support. |
2 Copyright (C) 1995, 1997, 1998, 2001 Electrotechnical Laboratory, JAPAN. | |
89483 | 3 Licensed to the Free Software Foundation. |
90424 | 4 Copyright (C) 2001, 2005, 2006 Free Software Foundation, Inc. |
5 Copyright (C) 2003, 2006 | |
88363 | 6 National Institute of Advanced Industrial Science and Technology (AIST) |
7 Registration Number H13PRO009 | |
8 | |
9 This file is part of GNU Emacs. | |
10 | |
11 GNU Emacs is free software; you can redistribute it and/or modify | |
12 it under the terms of the GNU General Public License as published by | |
13 the Free Software Foundation; either version 2, or (at your option) | |
14 any later version. | |
15 | |
16 GNU Emacs is distributed in the hope that it will be useful, | |
17 but WITHOUT ANY WARRANTY; without even the implied warranty of | |
18 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
19 GNU General Public License for more details. | |
20 | |
21 You should have received a copy of the GNU General Public License | |
22 along with GNU Emacs; see the file COPYING. If not, write to | |
23 the Free Software Foundation, Inc., 59 Temple Place - Suite 330, | |
24 Boston, MA 02111-1307, USA. */ | |
25 | |
26 /* At first, see the document in `character.h' to understand the code | |
27 in this file. */ | |
28 | |
29 #ifdef emacs | |
30 #include <config.h> | |
31 #endif | |
32 | |
33 #include <stdio.h> | |
34 | |
35 #ifdef emacs | |
36 | |
37 #include <sys/types.h> | |
38 #include "lisp.h" | |
39 #include "character.h" | |
40 #include "buffer.h" | |
41 #include "charset.h" | |
42 #include "composite.h" | |
43 #include "disptab.h" | |
44 | |
45 #else /* not emacs */ | |
46 | |
47 #include "mulelib.h" | |
48 | |
49 #endif /* emacs */ | |
50 | |
51 Lisp_Object Qcharacterp; | |
52 | |
53 /* Vector of translation table ever defined. | |
54 ID of a translation table is used to index this vector. */ | |
55 Lisp_Object Vtranslation_table_vector; | |
56 | |
57 /* A char-table for characters which may invoke auto-filling. */ | |
58 Lisp_Object Vauto_fill_chars; | |
59 | |
60 Lisp_Object Qauto_fill_chars; | |
61 | |
89888 | 62 /* Char-table of information about which character to unify to which |
63 Unicode character. */ | |
88363 | 64 Lisp_Object Vchar_unify_table; |
65 | |
66 /* A char-table. An element is non-nil iff the corresponding | |
67 character has a printable glyph. */ | |
68 Lisp_Object Vprintable_chars; | |
69 | |
70 /* A char-table. An elemnent is a column-width of the corresponding | |
71 character. */ | |
72 Lisp_Object Vchar_width_table; | |
73 | |
74 /* A char-table. An element is a symbol indicating the direction | |
75 property of corresponding character. */ | |
76 Lisp_Object Vchar_direction_table; | |
77 | |
89020
b0277093a5f2
(_fetch_multibyte_char_len): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88990
diff
changeset
|
78 /* Variable used locally in the macro FETCH_MULTIBYTE_CHAR. */ |
88363 | 79 unsigned char *_fetch_multibyte_char_p; |
80 | |
88913
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
81 /* Char table of scripts. */ |
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
82 Lisp_Object Vchar_script_table; |
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
83 |
90403
81f1bbab281f
(Vscript_representative_chars): New variable.
Kenichi Handa <handa@m17n.org>
parents:
90256
diff
changeset
|
84 /* Alist of scripts vs representative characters. */ |
81f1bbab281f
(Vscript_representative_chars): New variable.
Kenichi Handa <handa@m17n.org>
parents:
90256
diff
changeset
|
85 Lisp_Object Vscript_representative_chars; |
81f1bbab281f
(Vscript_representative_chars): New variable.
Kenichi Handa <handa@m17n.org>
parents:
90256
diff
changeset
|
86 |
88913
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
87 static Lisp_Object Qchar_script_table; |
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
88 |
89054
8a6da305ac40
(unibyte_to_multibyte_table): New variable.
Kenichi Handa <handa@m17n.org>
parents:
89029
diff
changeset
|
89 /* Mapping table from unibyte chars to multibyte chars. */ |
8a6da305ac40
(unibyte_to_multibyte_table): New variable.
Kenichi Handa <handa@m17n.org>
parents:
89029
diff
changeset
|
90 int unibyte_to_multibyte_table[256]; |
88878
a0ae1a5876c7
(Vscript_alist): New variable.
Kenichi Handa <handa@m17n.org>
parents:
88853
diff
changeset
|
91 |
90020
680f4ba0cdc9
(unibyte_has_multibyte_table): New variable.
Kenichi Handa <handa@m17n.org>
parents:
89911
diff
changeset
|
92 /* Nth element is 1 iff unibyte char N can be mapped to a multibyte |
680f4ba0cdc9
(unibyte_has_multibyte_table): New variable.
Kenichi Handa <handa@m17n.org>
parents:
89911
diff
changeset
|
93 char. */ |
680f4ba0cdc9
(unibyte_has_multibyte_table): New variable.
Kenichi Handa <handa@m17n.org>
parents:
89911
diff
changeset
|
94 char unibyte_has_multibyte_table[256]; |
680f4ba0cdc9
(unibyte_has_multibyte_table): New variable.
Kenichi Handa <handa@m17n.org>
parents:
89911
diff
changeset
|
95 |
88363 | 96 |
97 | |
89888 | 98 /* Store multibyte form of character C at P. If C has modifier bits, |
99 handle them appropriately. */ | |
100 | |
88363 | 101 int |
89181
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
102 char_string (c, p) |
88363 | 103 int c; |
88427
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
104 unsigned char *p; |
88363 | 105 { |
106 int bytes; | |
107 | |
89181
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
108 if (c & CHAR_MODIFIER_MASK) |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
109 { |
89538
c719d9f420d1
(char_string): Use ASCII_CHAR_P instead of SINGLE_BYTE_CHAR_P.
Kenichi Handa <handa@m17n.org>
parents:
89511
diff
changeset
|
110 /* As an non-ASCII character can't have modifier bits, we just |
c719d9f420d1
(char_string): Use ASCII_CHAR_P instead of SINGLE_BYTE_CHAR_P.
Kenichi Handa <handa@m17n.org>
parents:
89511
diff
changeset
|
111 ignore the bits. */ |
c719d9f420d1
(char_string): Use ASCII_CHAR_P instead of SINGLE_BYTE_CHAR_P.
Kenichi Handa <handa@m17n.org>
parents:
89511
diff
changeset
|
112 if (ASCII_CHAR_P ((c & ~CHAR_MODIFIER_MASK))) |
89181
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
113 { |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
114 /* For Meta, Shift, and Control modifiers, we need special care. */ |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
115 if (c & CHAR_META) |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
116 { |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
117 /* Move the meta bit to the right place for a string. */ |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
118 c = (c & ~CHAR_META) | 0x80; |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
119 } |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
120 if (c & CHAR_SHIFT) |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
121 { |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
122 /* Shift modifier is valid only with [A-Za-z]. */ |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
123 if ((c & 0377) >= 'A' && (c & 0377) <= 'Z') |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
124 c &= ~CHAR_SHIFT; |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
125 else if ((c & 0377) >= 'a' && (c & 0377) <= 'z') |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
126 c = (c & ~CHAR_SHIFT) - ('a' - 'A'); |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
127 } |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
128 if (c & CHAR_CTL) |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
129 { |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
130 /* Simulate the code in lread.c. */ |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
131 /* Allow `\C- ' and `\C-?'. */ |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
132 if (c == (CHAR_CTL | ' ')) |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
133 c = 0; |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
134 else if (c == (CHAR_CTL | '?')) |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
135 c = 127; |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
136 /* ASCII control chars are made from letters (both cases), |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
137 as well as the non-letters within 0100...0137. */ |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
138 else if ((c & 0137) >= 0101 && (c & 0137) <= 0132) |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
139 c &= (037 | (~0177 & ~CHAR_CTL)); |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
140 else if ((c & 0177) >= 0100 && (c & 0177) <= 0137) |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
141 c &= (037 | (~0177 & ~CHAR_CTL)); |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
142 } |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
143 } |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
144 |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
145 /* If C still has any modifier bits, just ignore it. */ |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
146 c &= ~CHAR_MODIFIER_MASK; |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
147 } |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
148 |
88363 | 149 MAYBE_UNIFY_CHAR (c); |
150 | |
89181
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
151 if (c <= MAX_3_BYTE_CHAR) |
88363 | 152 { |
153 bytes = CHAR_STRING (c, p); | |
154 } | |
155 else if (c <= MAX_4_BYTE_CHAR) | |
156 { | |
157 p[0] = (0xF0 | (c >> 18)); | |
158 p[1] = (0x80 | ((c >> 12) & 0x3F)); | |
159 p[2] = (0x80 | ((c >> 6) & 0x3F)); | |
160 p[3] = (0x80 | (c & 0x3F)); | |
161 bytes = 4; | |
162 } | |
89181
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
163 else if (c <= MAX_5_BYTE_CHAR) |
88363 | 164 { |
165 p[0] = 0xF8; | |
166 p[1] = (0x80 | ((c >> 18) & 0x0F)); | |
167 p[2] = (0x80 | ((c >> 12) & 0x3F)); | |
168 p[3] = (0x80 | ((c >> 6) & 0x3F)); | |
169 p[4] = (0x80 | (c & 0x3F)); | |
170 bytes = 5; | |
171 } | |
89181
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
172 else |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
173 { |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
174 c = CHAR_TO_BYTE8 (c); |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
175 bytes = BYTE8_STRING (c, p); |
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
176 } |
88427
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
177 |
88363 | 178 return bytes; |
179 } | |
180 | |
181 | |
89888 | 182 /* Return a character whose multibyte form is at P. Set LEN is not |
183 NULL, it must be a pointer to integer. In that case, set *LEN to | |
184 the byte length of the multibyte form. If ADVANCED is not NULL, is | |
185 must be a pointer to unsigned char. In that case, set *ADVANCED to | |
186 the ending address (i.e. the starting address of the next | |
187 character) of the multibyte form. */ | |
188 | |
88363 | 189 int |
89181
be75d5934738
(char_string): Renamed from
Kenichi Handa <handa@m17n.org>
parents:
89054
diff
changeset
|
190 string_char (p, advanced, len) |
88878
a0ae1a5876c7
(Vscript_alist): New variable.
Kenichi Handa <handa@m17n.org>
parents:
88853
diff
changeset
|
191 const unsigned char *p; |
a0ae1a5876c7
(Vscript_alist): New variable.
Kenichi Handa <handa@m17n.org>
parents:
88853
diff
changeset
|
192 const unsigned char **advanced; |
88363 | 193 int *len; |
194 { | |
88427
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
195 int c; |
88878
a0ae1a5876c7
(Vscript_alist): New variable.
Kenichi Handa <handa@m17n.org>
parents:
88853
diff
changeset
|
196 const unsigned char *saved_p = p; |
88363 | 197 |
198 if (*p < 0x80 || ! (*p & 0x20) || ! (*p & 0x10)) | |
199 { | |
200 c = STRING_CHAR_ADVANCE (p); | |
201 } | |
202 else if (! (*p & 0x08)) | |
203 { | |
204 c = ((((p)[0] & 0xF) << 18) | |
205 | (((p)[1] & 0x3F) << 12) | |
206 | (((p)[2] & 0x3F) << 6) | |
207 | ((p)[3] & 0x3F)); | |
208 p += 4; | |
209 } | |
210 else | |
211 { | |
212 c = ((((p)[1] & 0x3F) << 18) | |
213 | (((p)[2] & 0x3F) << 12) | |
214 | (((p)[3] & 0x3F) << 6) | |
215 | ((p)[4] & 0x3F)); | |
216 p += 5; | |
217 } | |
218 | |
219 MAYBE_UNIFY_CHAR (c); | |
220 | |
221 if (len) | |
222 *len = p - saved_p; | |
223 if (advanced) | |
224 *advanced = p; | |
225 return c; | |
226 } | |
227 | |
228 | |
229 /* Translate character C by translation table TABLE. If C is | |
230 negative, translate a character specified by CHARSET and CODE. If | |
231 no translation is found in TABLE, return the untranslated | |
89757
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
232 character. If TABLE is a list, elements are char tables. In this |
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
233 case, translace C by all tables. */ |
88363 | 234 |
235 int | |
236 translate_char (table, c) | |
237 Lisp_Object table; | |
238 int c; | |
239 { | |
89757
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
240 if (CHAR_TABLE_P (table)) |
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
241 { |
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
242 Lisp_Object ch; |
88363 | 243 |
89757
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
244 ch = CHAR_TABLE_REF (table, c); |
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
245 if (CHARACTERP (ch)) |
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
246 c = XINT (ch); |
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
247 } |
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
248 else |
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
249 { |
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
250 for (; CONSP (table); table = XCDR (table)) |
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
251 c = translate_char (XCAR (table), c); |
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
252 } |
482c15940000
(translate_char): Accept list of translation tables.
Kenichi Handa <handa@m17n.org>
parents:
89538
diff
changeset
|
253 return c; |
88363 | 254 } |
255 | |
256 /* Convert the multibyte character C to unibyte 8-bit character based | |
88983
2d504d707ce6
(unibyte_char_to_multibyte): Refer to
Kenichi Handa <handa@m17n.org>
parents:
88947
diff
changeset
|
257 on the current value of charset_unibyte. If dimension of |
2d504d707ce6
(unibyte_char_to_multibyte): Refer to
Kenichi Handa <handa@m17n.org>
parents:
88947
diff
changeset
|
258 charset_unibyte is more than one, return (C & 0xFF). |
88363 | 259 |
260 The argument REV_TBL is now ignored. It will be removed in the | |
261 future. */ | |
262 | |
263 int | |
264 multibyte_char_to_unibyte (c, rev_tbl) | |
265 int c; | |
266 Lisp_Object rev_tbl; | |
267 { | |
89054
8a6da305ac40
(unibyte_to_multibyte_table): New variable.
Kenichi Handa <handa@m17n.org>
parents:
89029
diff
changeset
|
268 struct charset *charset; |
8a6da305ac40
(unibyte_to_multibyte_table): New variable.
Kenichi Handa <handa@m17n.org>
parents:
89029
diff
changeset
|
269 unsigned c1; |
88363 | 270 |
89054
8a6da305ac40
(unibyte_to_multibyte_table): New variable.
Kenichi Handa <handa@m17n.org>
parents:
89029
diff
changeset
|
271 if (CHAR_BYTE8_P (c)) |
8a6da305ac40
(unibyte_to_multibyte_table): New variable.
Kenichi Handa <handa@m17n.org>
parents:
89029
diff
changeset
|
272 return CHAR_TO_BYTE8 (c); |
8a6da305ac40
(unibyte_to_multibyte_table): New variable.
Kenichi Handa <handa@m17n.org>
parents:
89029
diff
changeset
|
273 charset = CHARSET_FROM_ID (charset_unibyte); |
8a6da305ac40
(unibyte_to_multibyte_table): New variable.
Kenichi Handa <handa@m17n.org>
parents:
89029
diff
changeset
|
274 c1 = ENCODE_CHAR (charset, c); |
88363 | 275 return ((c1 != CHARSET_INVALID_CODE (charset)) ? c1 : c & 0xFF); |
276 } | |
277 | |
90761
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
278 /* Like multibyte_char_to_unibyte, but return -1 if C is not supported |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
279 by charset_unibyte. */ |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
280 |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
281 int |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
282 multibyte_char_to_unibyte_safe (c) |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
283 int c; |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
284 { |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
285 struct charset *charset; |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
286 unsigned c1; |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
287 |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
288 if (CHAR_BYTE8_P (c)) |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
289 return CHAR_TO_BYTE8 (c); |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
290 charset = CHARSET_FROM_ID (charset_unibyte); |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
291 c1 = ENCODE_CHAR (charset, c); |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
292 return ((c1 != CHARSET_INVALID_CODE (charset)) ? c1 : -1); |
2951f3d44eba
(multibyte_char_to_unibyte_safe): New function.
Kenichi Handa <handa@m17n.org>
parents:
90424
diff
changeset
|
293 } |
88363 | 294 |
295 DEFUN ("characterp", Fcharacterp, Scharacterp, 1, 2, 0, | |
296 doc: /* Return non-nil if OBJECT is a character. */) | |
297 (object, ignore) | |
298 Lisp_Object object, ignore; | |
299 { | |
300 return (CHARACTERP (object) ? Qt : Qnil); | |
301 } | |
302 | |
303 DEFUN ("max-char", Fmax_char, Smax_char, 0, 0, 0, | |
304 doc: /* Return the character of the maximum code. */) | |
305 () | |
306 { | |
307 return make_number (MAX_CHAR); | |
308 } | |
309 | |
310 DEFUN ("unibyte-char-to-multibyte", Funibyte_char_to_multibyte, | |
311 Sunibyte_char_to_multibyte, 1, 1, 0, | |
312 doc: /* Convert the unibyte character CH to multibyte character. | |
313 The multibyte character is a result of decoding CH by | |
88990
16317690f974
(Funibyte_char_to_multibyte): Doc fix.
Dave Love <fx@gnu.org>
parents:
88983
diff
changeset
|
314 the current unibyte charset (see `unibyte-charset'). */) |
88363 | 315 (ch) |
316 Lisp_Object ch; | |
317 { | |
318 int c; | |
319 struct charset *charset; | |
320 | |
321 CHECK_CHARACTER (ch); | |
322 c = XFASTINT (ch); | |
323 if (c >= 0400) | |
324 error ("Invalid unibyte character: %d", c); | |
88983
2d504d707ce6
(unibyte_char_to_multibyte): Refer to
Kenichi Handa <handa@m17n.org>
parents:
88947
diff
changeset
|
325 charset = CHARSET_FROM_ID (charset_unibyte); |
88363 | 326 c = DECODE_CHAR (charset, c); |
327 if (c < 0) | |
88947
894b8bce3208
(Funibyte_char_to_multibyte): If C can't be decoded
Kenichi Handa <handa@m17n.org>
parents:
88926
diff
changeset
|
328 c = BYTE8_TO_CHAR (XFASTINT (ch)); |
88363 | 329 return make_number (c); |
330 } | |
331 | |
332 DEFUN ("multibyte-char-to-unibyte", Fmultibyte_char_to_unibyte, | |
333 Smultibyte_char_to_unibyte, 1, 1, 0, | |
334 doc: /* Convert the multibyte character CH to unibyte character.\n\ | |
335 The unibyte character is a result of encoding CH by | |
336 the current primary charset (value of `charset-primary'). */) | |
337 (ch) | |
338 Lisp_Object ch; | |
339 { | |
340 int c; | |
341 | |
342 CHECK_CHARACTER (ch); | |
343 c = XFASTINT (ch); | |
88947
894b8bce3208
(Funibyte_char_to_multibyte): If C can't be decoded
Kenichi Handa <handa@m17n.org>
parents:
88926
diff
changeset
|
344 c = CHAR_TO_BYTE8 (c); |
894b8bce3208
(Funibyte_char_to_multibyte): If C can't be decoded
Kenichi Handa <handa@m17n.org>
parents:
88926
diff
changeset
|
345 return make_number (c); |
88363 | 346 } |
347 | |
348 DEFUN ("char-bytes", Fchar_bytes, Schar_bytes, 1, 1, 0, | |
349 doc: /* Return 1 regardless of the argument CHAR. | |
350 This is now an obsolete function. We keep it just for backward compatibility. */) | |
351 (ch) | |
352 Lisp_Object ch; | |
353 { | |
354 CHECK_CHARACTER (ch); | |
355 return make_number (1); | |
356 } | |
357 | |
358 DEFUN ("char-width", Fchar_width, Schar_width, 1, 1, 0, | |
359 doc: /* Return width of CHAR when displayed in the current buffer. | |
360 The width is measured by how many columns it occupies on the screen. | |
361 Tab is taken to occupy `tab-width' columns. */) | |
362 (ch) | |
363 Lisp_Object ch; | |
364 { | |
365 Lisp_Object disp; | |
366 int c, width; | |
367 struct Lisp_Char_Table *dp = buffer_display_table (); | |
368 | |
369 CHECK_CHARACTER (ch); | |
370 c = XINT (ch); | |
371 | |
372 /* Get the way the display table would display it. */ | |
373 disp = dp ? DISP_CHAR_VECTOR (dp, c) : Qnil; | |
374 | |
375 if (VECTORP (disp)) | |
376 width = ASIZE (disp); | |
377 else | |
378 width = CHAR_WIDTH (c); | |
379 | |
380 return make_number (width); | |
381 } | |
382 | |
383 /* Return width of string STR of length LEN when displayed in the | |
384 current buffer. The width is measured by how many columns it | |
385 occupies on the screen. If PRECISION > 0, return the width of | |
386 longest substring that doesn't exceed PRECISION, and set number of | |
387 characters and bytes of the substring in *NCHARS and *NBYTES | |
388 respectively. */ | |
389 | |
88427
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
390 int |
88363 | 391 c_string_width (str, len, precision, nchars, nbytes) |
89483 | 392 const unsigned char *str; |
88363 | 393 int precision, *nchars, *nbytes; |
394 { | |
395 int i = 0, i_byte = 0; | |
396 int width = 0; | |
397 struct Lisp_Char_Table *dp = buffer_display_table (); | |
398 | |
399 while (i_byte < len) | |
400 { | |
401 int bytes, thiswidth; | |
402 Lisp_Object val; | |
403 int c = STRING_CHAR_AND_LENGTH (str + i_byte, len - i_byte, bytes); | |
404 | |
405 if (dp) | |
406 { | |
407 val = DISP_CHAR_VECTOR (dp, c); | |
408 if (VECTORP (val)) | |
409 thiswidth = XVECTOR (val)->size; | |
410 else | |
411 thiswidth = CHAR_WIDTH (c); | |
412 } | |
413 else | |
414 { | |
415 thiswidth = CHAR_WIDTH (c); | |
416 } | |
417 | |
418 if (precision > 0 | |
419 && (width + thiswidth > precision)) | |
420 { | |
421 *nchars = i; | |
422 *nbytes = i_byte; | |
423 return width; | |
424 } | |
425 i++; | |
426 i_byte += bytes; | |
427 width += thiswidth; | |
428 } | |
429 | |
430 if (precision > 0) | |
431 { | |
432 *nchars = i; | |
433 *nbytes = i_byte; | |
434 } | |
435 | |
436 return width; | |
437 } | |
438 | |
88427
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
439 /* Return width of string STR of length LEN when displayed in the |
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
440 current buffer. The width is measured by how many columns it |
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
441 occupies on the screen. */ |
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
442 |
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
443 int |
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
444 strwidth (str, len) |
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
445 unsigned char *str; |
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
446 int len; |
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
447 { |
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
448 return c_string_width (str, len, -1, NULL, NULL); |
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
449 } |
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
450 |
88363 | 451 /* Return width of Lisp string STRING when displayed in the current |
452 buffer. The width is measured by how many columns it occupies on | |
453 the screen while paying attention to compositions. If PRECISION > | |
454 0, return the width of longest substring that doesn't exceed | |
455 PRECISION, and set number of characters and bytes of the substring | |
456 in *NCHARS and *NBYTES respectively. */ | |
457 | |
458 int | |
459 lisp_string_width (string, precision, nchars, nbytes) | |
460 Lisp_Object string; | |
461 int precision, *nchars, *nbytes; | |
462 { | |
89483 | 463 int len = SCHARS (string); |
90256
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
464 /* This set multibyte to 0 even if STRING is multibyte when it |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
465 contains only ascii and eight-bit-graphic, but that's |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
466 intentional. */ |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
467 int multibyte = len < SBYTES (string); |
89483 | 468 unsigned char *str = SDATA (string); |
88363 | 469 int i = 0, i_byte = 0; |
470 int width = 0; | |
471 struct Lisp_Char_Table *dp = buffer_display_table (); | |
472 | |
473 while (i < len) | |
474 { | |
475 int chars, bytes, thiswidth; | |
476 Lisp_Object val; | |
477 int cmp_id; | |
89511
bae1254b8517
(lisp_string_width): Make ignore and end EMACS_INT.
Dave Love <fx@gnu.org>
parents:
89483
diff
changeset
|
478 EMACS_INT ignore, end; |
88363 | 479 |
480 if (find_composition (i, -1, &ignore, &end, &val, string) | |
481 && ((cmp_id = get_composition_id (i, i_byte, end - i, val, string)) | |
482 >= 0)) | |
483 { | |
484 thiswidth = composition_table[cmp_id]->width; | |
485 chars = end - i; | |
486 bytes = string_char_to_byte (string, end) - i_byte; | |
487 } | |
488 else | |
489 { | |
90256
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
490 int c; |
88363 | 491 |
90256
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
492 if (multibyte) |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
493 c = STRING_CHAR_AND_LENGTH (str + i_byte, len - i_byte, bytes); |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
494 else |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
495 c = str[i_byte], bytes = 1; |
88363 | 496 chars = 1; |
90256
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
497 if (dp) |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
498 { |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
499 val = DISP_CHAR_VECTOR (dp, c); |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
500 if (VECTORP (val)) |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
501 thiswidth = XVECTOR (val)->size; |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
502 else |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
503 thiswidth = CHAR_WIDTH (c); |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
504 } |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
505 else |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
506 { |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
507 thiswidth = CHAR_WIDTH (c); |
d4755c5aa57d
(lisp_string_width): Check multibyteness of STRING.
Kenichi Handa <handa@m17n.org>
parents:
90159
diff
changeset
|
508 } |
88363 | 509 } |
510 | |
511 if (precision > 0 | |
512 && (width + thiswidth > precision)) | |
513 { | |
514 *nchars = i; | |
515 *nbytes = i_byte; | |
516 return width; | |
517 } | |
518 i += chars; | |
519 i_byte += bytes; | |
520 width += thiswidth; | |
521 } | |
522 | |
523 if (precision > 0) | |
524 { | |
525 *nchars = i; | |
526 *nbytes = i_byte; | |
527 } | |
528 | |
529 return width; | |
530 } | |
531 | |
532 DEFUN ("string-width", Fstring_width, Sstring_width, 1, 1, 0, | |
533 doc: /* Return width of STRING when displayed in the current buffer. | |
534 Width is measured by how many columns it occupies on the screen. | |
535 When calculating width of a multibyte character in STRING, | |
536 only the base leading-code is considered; the validity of | |
537 the following bytes is not checked. Tabs in STRING are always | |
538 taken to occupy `tab-width' columns. */) | |
539 (str) | |
540 Lisp_Object str; | |
541 { | |
542 Lisp_Object val; | |
543 | |
544 CHECK_STRING (str); | |
545 XSETFASTINT (val, lisp_string_width (str, -1, NULL, NULL)); | |
546 return val; | |
547 } | |
548 | |
549 DEFUN ("char-direction", Fchar_direction, Schar_direction, 1, 1, 0, | |
550 doc: /* Return the direction of CHAR. | |
551 The returned value is 0 for left-to-right and 1 for right-to-left. */) | |
552 (ch) | |
553 Lisp_Object ch; | |
554 { | |
555 int c; | |
556 | |
557 CHECK_CHARACTER (ch); | |
558 c = XINT (ch); | |
559 return CHAR_TABLE_REF (Vchar_direction_table, c); | |
560 } | |
561 | |
562 /* Return the number of characters in the NBYTES bytes at PTR. | |
563 This works by looking at the contents and checking for multibyte | |
564 sequences while assuming that there's no invalid sequence. | |
565 However, if the current buffer has enable-multibyte-characters = | |
566 nil, we treat each byte as a character. */ | |
567 | |
568 int | |
569 chars_in_text (ptr, nbytes) | |
89483 | 570 const unsigned char *ptr; |
88363 | 571 int nbytes; |
572 { | |
573 /* current_buffer is null at early stages of Emacs initialization. */ | |
574 if (current_buffer == 0 | |
575 || NILP (current_buffer->enable_multibyte_characters)) | |
576 return nbytes; | |
577 | |
578 return multibyte_chars_in_text (ptr, nbytes); | |
579 } | |
580 | |
581 /* Return the number of characters in the NBYTES bytes at PTR. | |
582 This works by looking at the contents and checking for multibyte | |
583 sequences while assuming that there's no invalid sequence. It | |
584 ignores enable-multibyte-characters. */ | |
585 | |
586 int | |
587 multibyte_chars_in_text (ptr, nbytes) | |
89483 | 588 const unsigned char *ptr; |
88363 | 589 int nbytes; |
590 { | |
89483 | 591 const unsigned char *endp = ptr + nbytes; |
88363 | 592 int chars = 0; |
593 | |
594 while (ptr < endp) | |
595 { | |
596 int len = MULTIBYTE_LENGTH (ptr, endp); | |
597 | |
598 if (len == 0) | |
599 abort (); | |
600 ptr += len; | |
601 chars++; | |
602 } | |
603 | |
604 return chars; | |
605 } | |
606 | |
607 /* Parse unibyte text at STR of LEN bytes as a multibyte text, count | |
608 characters and bytes in it, and store them in *NCHARS and *NBYTES | |
609 respectively. On counting bytes, pay attention to that 8-bit | |
610 characters not constructing a valid multibyte sequence are | |
611 represented by 2-byte in a multibyte text. */ | |
612 | |
613 void | |
614 parse_str_as_multibyte (str, len, nchars, nbytes) | |
89483 | 615 const unsigned char *str; |
88363 | 616 int len, *nchars, *nbytes; |
617 { | |
89483 | 618 const unsigned char *endp = str + len; |
88363 | 619 int n, chars = 0, bytes = 0; |
620 | |
621 if (len >= MAX_MULTIBYTE_LENGTH) | |
622 { | |
89483 | 623 const unsigned char *adjusted_endp = endp - MAX_MULTIBYTE_LENGTH; |
88363 | 624 while (str < adjusted_endp) |
625 { | |
626 if ((n = MULTIBYTE_LENGTH_NO_CHECK (str)) > 0) | |
627 str += n, bytes += n; | |
628 else | |
629 str++, bytes += 2; | |
630 chars++; | |
631 } | |
632 } | |
633 while (str < endp) | |
634 { | |
635 if ((n = MULTIBYTE_LENGTH (str, endp)) > 0) | |
636 str += n, bytes += n; | |
637 else | |
638 str++, bytes += 2; | |
639 chars++; | |
640 } | |
641 | |
642 *nchars = chars; | |
643 *nbytes = bytes; | |
644 return; | |
645 } | |
646 | |
647 /* Arrange unibyte text at STR of NBYTES bytes as a multibyte text. | |
648 It actually converts only such 8-bit characters that don't contruct | |
649 a multibyte sequence to multibyte forms of Latin-1 characters. If | |
650 NCHARS is nonzero, set *NCHARS to the number of characters in the | |
651 text. It is assured that we can use LEN bytes at STR as a work | |
652 area and that is enough. Return the number of bytes of the | |
653 resulting text. */ | |
654 | |
655 int | |
656 str_as_multibyte (str, len, nbytes, nchars) | |
657 unsigned char *str; | |
658 int len, nbytes, *nchars; | |
659 { | |
660 unsigned char *p = str, *endp = str + nbytes; | |
661 unsigned char *to; | |
662 int chars = 0; | |
663 int n; | |
664 | |
665 if (nbytes >= MAX_MULTIBYTE_LENGTH) | |
666 { | |
667 unsigned char *adjusted_endp = endp - MAX_MULTIBYTE_LENGTH; | |
668 while (p < adjusted_endp | |
669 && (n = MULTIBYTE_LENGTH_NO_CHECK (p)) > 0) | |
670 p += n, chars++; | |
671 } | |
672 while ((n = MULTIBYTE_LENGTH (p, endp)) > 0) | |
673 p += n, chars++; | |
674 if (nchars) | |
675 *nchars = chars; | |
676 if (p == endp) | |
677 return nbytes; | |
678 | |
679 to = p; | |
680 nbytes = endp - p; | |
681 endp = str + len; | |
682 safe_bcopy ((char *) p, (char *) (endp - nbytes), nbytes); | |
683 p = endp - nbytes; | |
684 | |
685 if (nbytes >= MAX_MULTIBYTE_LENGTH) | |
686 { | |
687 unsigned char *adjusted_endp = endp - MAX_MULTIBYTE_LENGTH; | |
688 while (p < adjusted_endp) | |
689 { | |
690 if ((n = MULTIBYTE_LENGTH_NO_CHECK (p)) > 0) | |
691 { | |
692 while (n--) | |
693 *to++ = *p++; | |
694 } | |
695 else | |
696 { | |
697 int c = *p++; | |
698 c = BYTE8_TO_CHAR (c); | |
699 to += CHAR_STRING (c, to); | |
700 } | |
701 } | |
702 chars++; | |
703 } | |
704 while (p < endp) | |
705 { | |
706 if ((n = MULTIBYTE_LENGTH (p, endp)) > 0) | |
707 { | |
708 while (n--) | |
709 *to++ = *p++; | |
89483 | 710 } |
88363 | 711 else |
712 { | |
713 int c = *p++; | |
714 c = BYTE8_TO_CHAR (c); | |
715 to += CHAR_STRING (c, to); | |
716 } | |
717 chars++; | |
718 } | |
719 if (nchars) | |
720 *nchars = chars; | |
721 return (to - str); | |
722 } | |
723 | |
724 /* Parse unibyte string at STR of LEN bytes, and return the number of | |
725 bytes it may ocupy when converted to multibyte string by | |
726 `str_to_multibyte'. */ | |
727 | |
728 int | |
729 parse_str_to_multibyte (str, len) | |
730 unsigned char *str; | |
731 int len; | |
732 { | |
733 unsigned char *endp = str + len; | |
734 int bytes; | |
735 | |
736 for (bytes = 0; str < endp; str++) | |
737 bytes += (*str < 0x80) ? 1 : 2; | |
738 return bytes; | |
739 } | |
740 | |
741 | |
742 /* Convert unibyte text at STR of NBYTES bytes to a multibyte text | |
743 that contains the same single-byte characters. It actually | |
744 converts all 8-bit characters to multibyte forms. It is assured | |
745 that we can use LEN bytes at STR as a work area and that is | |
746 enough. */ | |
747 | |
748 int | |
749 str_to_multibyte (str, len, bytes) | |
750 unsigned char *str; | |
751 int len, bytes; | |
752 { | |
753 unsigned char *p = str, *endp = str + bytes; | |
754 unsigned char *to; | |
755 | |
756 while (p < endp && *p < 0x80) p++; | |
757 if (p == endp) | |
758 return bytes; | |
759 to = p; | |
760 bytes = endp - p; | |
761 endp = str + len; | |
762 safe_bcopy ((char *) p, (char *) (endp - bytes), bytes); | |
763 p = endp - bytes; | |
89483 | 764 while (p < endp) |
88363 | 765 { |
766 int c = *p++; | |
767 | |
768 if (c >= 0x80) | |
769 c = BYTE8_TO_CHAR (c); | |
770 to += CHAR_STRING (c, to); | |
771 } | |
772 return (to - str); | |
773 } | |
774 | |
775 /* Arrange multibyte text at STR of LEN bytes as a unibyte text. It | |
776 actually converts characters in the range 0x80..0xFF to | |
777 unibyte. */ | |
778 | |
779 int | |
780 str_as_unibyte (str, bytes) | |
781 unsigned char *str; | |
782 int bytes; | |
783 { | |
88878
a0ae1a5876c7
(Vscript_alist): New variable.
Kenichi Handa <handa@m17n.org>
parents:
88853
diff
changeset
|
784 const unsigned char *p = str, *endp = str + bytes; |
a0ae1a5876c7
(Vscript_alist): New variable.
Kenichi Handa <handa@m17n.org>
parents:
88853
diff
changeset
|
785 unsigned char *to; |
88363 | 786 int c, len; |
787 | |
788 while (p < endp) | |
789 { | |
790 c = *p; | |
791 len = BYTES_BY_CHAR_HEAD (c); | |
792 if (CHAR_BYTE8_HEAD_P (c)) | |
793 break; | |
794 p += len; | |
795 } | |
88878
a0ae1a5876c7
(Vscript_alist): New variable.
Kenichi Handa <handa@m17n.org>
parents:
88853
diff
changeset
|
796 to = str + (p - str); |
89483 | 797 while (p < endp) |
88363 | 798 { |
799 c = *p; | |
800 len = BYTES_BY_CHAR_HEAD (c); | |
801 if (CHAR_BYTE8_HEAD_P (c)) | |
802 { | |
803 c = STRING_CHAR_ADVANCE (p); | |
804 *to++ = CHAR_TO_BYTE8 (c); | |
805 } | |
806 else | |
807 { | |
808 while (len--) *to++ = *p++; | |
809 } | |
810 } | |
811 return (to - str); | |
812 } | |
813 | |
814 int | |
815 string_count_byte8 (string) | |
816 Lisp_Object string; | |
817 { | |
818 int multibyte = STRING_MULTIBYTE (string); | |
89483 | 819 int nbytes = SBYTES (string); |
820 unsigned char *p = SDATA (string); | |
88363 | 821 unsigned char *pend = p + nbytes; |
822 int count = 0; | |
823 int c, len; | |
824 | |
825 if (multibyte) | |
826 while (p < pend) | |
827 { | |
828 c = *p; | |
829 len = BYTES_BY_CHAR_HEAD (c); | |
830 | |
831 if (CHAR_BYTE8_HEAD_P (c)) | |
832 count++; | |
833 p += len; | |
834 } | |
835 else | |
836 while (p < pend) | |
837 { | |
838 if (*p++ >= 0x80) | |
839 count++; | |
840 } | |
841 return count; | |
842 } | |
843 | |
844 | |
845 Lisp_Object | |
846 string_escape_byte8 (string) | |
847 Lisp_Object string; | |
848 { | |
89483 | 849 int nchars = SCHARS (string); |
850 int nbytes = SBYTES (string); | |
88363 | 851 int multibyte = STRING_MULTIBYTE (string); |
852 int byte8_count; | |
88878
a0ae1a5876c7
(Vscript_alist): New variable.
Kenichi Handa <handa@m17n.org>
parents:
88853
diff
changeset
|
853 const unsigned char *src, *src_end; |
a0ae1a5876c7
(Vscript_alist): New variable.
Kenichi Handa <handa@m17n.org>
parents:
88853
diff
changeset
|
854 unsigned char *dst; |
88363 | 855 Lisp_Object val; |
856 int c, len; | |
857 | |
858 if (multibyte && nchars == nbytes) | |
859 return string; | |
860 | |
861 byte8_count = string_count_byte8 (string); | |
862 | |
863 if (byte8_count == 0) | |
864 return string; | |
865 | |
866 if (multibyte) | |
867 /* Convert 2-byte sequence of byte8 chars to 4-byte octal. */ | |
88526
6bac4ae1c326
(string_escape_byte8): Make multibyte string with correct size.
Kenichi Handa <handa@m17n.org>
parents:
88427
diff
changeset
|
868 val = make_uninit_multibyte_string (nchars + byte8_count * 3, |
88363 | 869 nbytes + byte8_count * 2); |
870 else | |
871 /* Convert 1-byte sequence of byte8 chars to 4-byte octal. */ | |
872 val = make_uninit_string (nbytes + byte8_count * 3); | |
873 | |
89483 | 874 src = SDATA (string); |
88363 | 875 src_end = src + nbytes; |
89483 | 876 dst = SDATA (val); |
88363 | 877 if (multibyte) |
878 while (src < src_end) | |
879 { | |
880 c = *src; | |
881 len = BYTES_BY_CHAR_HEAD (c); | |
882 | |
883 if (CHAR_BYTE8_HEAD_P (c)) | |
884 { | |
885 c = STRING_CHAR_ADVANCE (src); | |
886 c = CHAR_TO_BYTE8 (c); | |
88427
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
887 sprintf ((char *) dst, "\\%03o", c); |
88363 | 888 dst += 4; |
889 } | |
890 else | |
891 while (len--) *dst++ = *src++; | |
892 } | |
893 else | |
894 while (src < src_end) | |
895 { | |
896 c = *src++; | |
897 if (c >= 0x80) | |
898 { | |
88427
1a3aec316071
(c_string_width): Add return type `int'.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
899 sprintf ((char *) dst, "\\%03o", c); |
88363 | 900 dst += 4; |
901 } | |
902 else | |
903 *dst++ = c; | |
904 } | |
905 return val; | |
906 } | |
907 | |
908 | |
89483 | 909 DEFUN ("string", Fstring, Sstring, 0, MANY, 0, |
88363 | 910 doc: /* |
88640 | 911 Concatenate all the argument characters and make the result a string. |
912 usage: (string &rest CHARACTERS) */) | |
88363 | 913 (n, args) |
914 int n; | |
915 Lisp_Object *args; | |
916 { | |
917 int i; | |
918 unsigned char *buf = (unsigned char *) alloca (MAX_MULTIBYTE_LENGTH * n); | |
919 unsigned char *p = buf; | |
920 int c; | |
921 | |
922 for (i = 0; i < n; i++) | |
923 { | |
924 CHECK_CHARACTER (args[i]); | |
925 c = XINT (args[i]); | |
926 p += CHAR_STRING (c, p); | |
927 } | |
928 | |
929 return make_string_from_bytes ((char *) buf, n, p - buf); | |
930 } | |
931 | |
932 void | |
933 init_character_once () | |
934 { | |
935 } | |
936 | |
937 #ifdef emacs | |
938 | |
939 void | |
940 syms_of_character () | |
941 { | |
942 DEFSYM (Qcharacterp, "characterp"); | |
943 DEFSYM (Qauto_fill_chars, "auto-fill-chars"); | |
944 | |
945 staticpro (&Vchar_unify_table); | |
946 Vchar_unify_table = Qnil; | |
947 | |
948 defsubr (&Smax_char); | |
949 defsubr (&Scharacterp); | |
950 defsubr (&Sunibyte_char_to_multibyte); | |
951 defsubr (&Smultibyte_char_to_unibyte); | |
952 defsubr (&Schar_bytes); | |
953 defsubr (&Schar_width); | |
954 defsubr (&Sstring_width); | |
955 defsubr (&Schar_direction); | |
956 defsubr (&Sstring); | |
957 | |
958 DEFVAR_LISP ("translation-table-vector", &Vtranslation_table_vector, | |
959 doc: /* | |
88926
f026b7e4fc61
(syms_of_character) <translation-table-vector>: Doc
Dave Love <fx@gnu.org>
parents:
88913
diff
changeset
|
960 Vector recording all translation tables ever defined. |
f026b7e4fc61
(syms_of_character) <translation-table-vector>: Doc
Dave Love <fx@gnu.org>
parents:
88913
diff
changeset
|
961 Each element is a pair (SYMBOL . TABLE) relating the table to the |
f026b7e4fc61
(syms_of_character) <translation-table-vector>: Doc
Dave Love <fx@gnu.org>
parents:
88913
diff
changeset
|
962 symbol naming it. The ID of a translation table is an index into this vector. */); |
88363 | 963 Vtranslation_table_vector = Fmake_vector (make_number (16), Qnil); |
964 | |
965 DEFVAR_LISP ("auto-fill-chars", &Vauto_fill_chars, | |
966 doc: /* | |
967 A char-table for characters which invoke auto-filling. | |
968 Such characters have value t in this table. */); | |
969 Vauto_fill_chars = Fmake_char_table (Qauto_fill_chars, Qnil); | |
88853
503c60238368
(syms_of_character): Fix CHAR_TABLE_SET call.
Dave Love <fx@gnu.org>
parents:
88743
diff
changeset
|
970 CHAR_TABLE_SET (Vauto_fill_chars, ' ', Qt); |
503c60238368
(syms_of_character): Fix CHAR_TABLE_SET call.
Dave Love <fx@gnu.org>
parents:
88743
diff
changeset
|
971 CHAR_TABLE_SET (Vauto_fill_chars, '\n', Qt); |
88363 | 972 |
973 DEFVAR_LISP ("char-width-table", &Vchar_width_table, | |
974 doc: /* | |
975 A char-table for width (columns) of each character. */); | |
976 Vchar_width_table = Fmake_char_table (Qnil, make_number (1)); | |
88743
4570f1a4c391
(syms_of_character): Setup Vchar_width_table for
Kenichi Handa <handa@m17n.org>
parents:
88716
diff
changeset
|
977 char_table_set_range (Vchar_width_table, 0x80, 0x9F, make_number (4)); |
4570f1a4c391
(syms_of_character): Setup Vchar_width_table for
Kenichi Handa <handa@m17n.org>
parents:
88716
diff
changeset
|
978 char_table_set_range (Vchar_width_table, MAX_5_BYTE_CHAR + 1, MAX_CHAR, |
4570f1a4c391
(syms_of_character): Setup Vchar_width_table for
Kenichi Handa <handa@m17n.org>
parents:
88716
diff
changeset
|
979 make_number (4)); |
88363 | 980 |
981 DEFVAR_LISP ("char-direction-table", &Vchar_direction_table, | |
982 doc: /* A char-table for direction of each character. */); | |
983 Vchar_direction_table = Fmake_char_table (Qnil, make_number (1)); | |
984 | |
985 DEFVAR_LISP ("printable-chars", &Vprintable_chars, | |
986 doc: /* A char-table for each printable character. */); | |
88716
f617b1bf301b
(syms_of_character): Set the default value of
Kenichi Handa <handa@m17n.org>
parents:
88640
diff
changeset
|
987 Vprintable_chars = Fmake_char_table (Qnil, Qnil); |
89029
ad8f56efa7c6
(syms_of_character): Setup Vprintable_chars.
Kenichi Handa <handa@m17n.org>
parents:
89020
diff
changeset
|
988 Fset_char_table_range (Vprintable_chars, |
ad8f56efa7c6
(syms_of_character): Setup Vprintable_chars.
Kenichi Handa <handa@m17n.org>
parents:
89020
diff
changeset
|
989 Fcons (make_number (32), make_number (126)), Qt); |
ad8f56efa7c6
(syms_of_character): Setup Vprintable_chars.
Kenichi Handa <handa@m17n.org>
parents:
89020
diff
changeset
|
990 Fset_char_table_range (Vprintable_chars, |
ad8f56efa7c6
(syms_of_character): Setup Vprintable_chars.
Kenichi Handa <handa@m17n.org>
parents:
89020
diff
changeset
|
991 Fcons (make_number (160), |
ad8f56efa7c6
(syms_of_character): Setup Vprintable_chars.
Kenichi Handa <handa@m17n.org>
parents:
89020
diff
changeset
|
992 make_number (MAX_5_BYTE_CHAR)), Qt); |
88878
a0ae1a5876c7
(Vscript_alist): New variable.
Kenichi Handa <handa@m17n.org>
parents:
88853
diff
changeset
|
993 |
88913
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
994 DEFVAR_LISP ("char-script-table", &Vchar_script_table, |
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
995 doc: /* Char table of script symbols. |
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
996 It has one extra slot whose value is a list of script symbols. */); |
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
997 |
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
998 /* Intern this now in case it isn't already done. |
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
999 Setting this variable twice is harmless. |
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
1000 But don't staticpro it here--that is done in alloc.c. */ |
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
1001 Qchar_table_extra_slots = intern ("char-table-extra-slots"); |
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
1002 DEFSYM (Qchar_script_table, "char-script-table"); |
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
1003 Fput (Qchar_script_table, Qchar_table_extra_slots, make_number (1)); |
76074dea1258
(Vscript_alist): This variable deleted.
Kenichi Handa <handa@m17n.org>
parents:
88878
diff
changeset
|
1004 Vchar_script_table = Fmake_char_table (Qchar_script_table, Qnil); |
90403
81f1bbab281f
(Vscript_representative_chars): New variable.
Kenichi Handa <handa@m17n.org>
parents:
90256
diff
changeset
|
1005 |
81f1bbab281f
(Vscript_representative_chars): New variable.
Kenichi Handa <handa@m17n.org>
parents:
90256
diff
changeset
|
1006 DEFVAR_LISP ("script-representative-chars", &Vscript_representative_chars, |
81f1bbab281f
(Vscript_representative_chars): New variable.
Kenichi Handa <handa@m17n.org>
parents:
90256
diff
changeset
|
1007 doc: /* Alist of scripts vs the representative characters. */); |
81f1bbab281f
(Vscript_representative_chars): New variable.
Kenichi Handa <handa@m17n.org>
parents:
90256
diff
changeset
|
1008 Vscript_representative_chars = Qnil; |
88363 | 1009 } |
1010 | |
1011 #endif /* emacs */ | |
89911 | 1012 |
1013 /* arch-tag: b6665960-3c3d-4184-85cd-af4318197999 | |
1014 (do not change this comment) */ |