7
|
1 /* vi:set ts=8 sts=4 sw=4:
|
|
2 *
|
|
3 * VIM - Vi IMproved by Bram Moolenaar
|
|
4 * Multibyte extensions partly by Sung-Hoon Baek
|
|
5 *
|
|
6 * Do ":help uganda" in Vim to read copying and usage conditions.
|
|
7 * Do ":help credits" in Vim to see a list of people who contributed.
|
|
8 * See README.txt for an overview of the Vim source code.
|
|
9 */
|
|
10 /*
|
|
11 * mbyte.c: Code specifically for handling multi-byte characters.
|
|
12 *
|
|
13 * The encoding used in the core is set with 'encoding'. When 'encoding' is
|
|
14 * changed, the following four variables are set (for speed).
|
|
15 * Currently these types of character encodings are supported:
|
|
16 *
|
|
17 * "enc_dbcs" When non-zero it tells the type of double byte character
|
|
18 * encoding (Chinese, Korean, Japanese, etc.).
|
|
19 * The cell width on the display is equal to the number of
|
|
20 * bytes. (exception: DBCS_JPNU with first byte 0x8e)
|
|
21 * Recognizing the first or second byte is difficult, it
|
|
22 * requires checking a byte sequence from the start.
|
|
23 * "enc_utf8" When TRUE use Unicode characters in UTF-8 encoding.
|
|
24 * The cell width on the display needs to be determined from
|
|
25 * the character value.
|
|
26 * Recognizing bytes is easy: 0xxx.xxxx is a single-byte
|
|
27 * char, 10xx.xxxx is a trailing byte, 11xx.xxxx is a leading
|
|
28 * byte of a multi-byte character.
|
|
29 * To make things complicated, up to two composing characters
|
|
30 * are allowed. These are drawn on top of the first char.
|
|
31 * For most editing the sequence of bytes with composing
|
|
32 * characters included is considered to be one character.
|
|
33 * "enc_unicode" When 2 use 16-bit Unicode characters (or UTF-16).
|
|
34 * When 4 use 32-but Unicode characters.
|
|
35 * Internally characters are stored in UTF-8 encoding to
|
|
36 * avoid NUL bytes. Conversion happens when doing I/O.
|
|
37 * "enc_utf8" will also be TRUE.
|
|
38 *
|
|
39 * "has_mbyte" is set when "enc_dbcs" or "enc_utf8" is non-zero.
|
|
40 *
|
|
41 * If none of these is TRUE, 8-bit bytes are used for a character. The
|
|
42 * encoding isn't currently specified (TODO).
|
|
43 *
|
|
44 * 'encoding' specifies the encoding used in the core. This is in registers,
|
|
45 * text manipulation, buffers, etc. Conversion has to be done when characters
|
|
46 * in another encoding are received or send:
|
|
47 *
|
|
48 * clipboard
|
|
49 * ^
|
|
50 * | (2)
|
|
51 * V
|
|
52 * +---------------+
|
|
53 * (1) | | (3)
|
|
54 * keyboard ----->| core |-----> display
|
|
55 * | |
|
|
56 * +---------------+
|
|
57 * ^
|
|
58 * | (4)
|
|
59 * V
|
|
60 * file
|
|
61 *
|
|
62 * (1) Typed characters arrive in the current locale. Conversion is to be
|
|
63 * done when 'encoding' is different from 'termencoding'.
|
|
64 * (2) Text will be made available with the encoding specified with
|
|
65 * 'encoding'. If this is not sufficient, system-specific conversion
|
|
66 * might be required.
|
|
67 * (3) For the GUI the correct font must be selected, no conversion done.
|
|
68 * Otherwise, conversion is to be done when 'encoding' differs from
|
|
69 * 'termencoding'. (Different in the GTK+ 2 port -- 'termencoding'
|
|
70 * is always used for both input and output and must always be set to
|
|
71 * "utf-8". gui_mch_init() does this automatically.)
|
|
72 * (4) The encoding of the file is specified with 'fileencoding'. Conversion
|
|
73 * is to be done when it's different from 'encoding'.
|
|
74 *
|
|
75 * The viminfo file is a special case: Only text is converted, not file names.
|
|
76 * Vim scripts may contain an ":encoding" command. This has an effect for
|
|
77 * some commands, like ":menutrans"
|
|
78 */
|
|
79
|
|
80 #include "vim.h"
|
|
81
|
|
82 #ifdef WIN32UNIX
|
|
83 # ifndef WIN32_LEAN_AND_MEAN
|
|
84 # define WIN32_LEAN_AND_MEAN
|
|
85 # endif
|
|
86 # include <windows.h>
|
|
87 # ifdef WIN32
|
|
88 # undef WIN32 /* Some windows.h define WIN32, we don't want that here. */
|
|
89 # endif
|
|
90 #endif
|
|
91
|
|
92 #if (defined(WIN3264) || defined(WIN32UNIX)) && !defined(__MINGW32__)
|
|
93 # include <winnls.h>
|
|
94 #endif
|
|
95
|
|
96 #ifdef FEAT_GUI_X11
|
|
97 # include <X11/Intrinsic.h>
|
|
98 #endif
|
|
99 #ifdef X_LOCALE
|
|
100 #include <X11/Xlocale.h>
|
|
101 #endif
|
|
102
|
661
|
103 #if defined(FEAT_GUI_GTK) && defined(FEAT_XIM) && defined(HAVE_GTK2)
|
7
|
104 # include <gdk/gdkkeysyms.h>
|
|
105 # ifdef WIN3264
|
|
106 # include <gdk/gdkwin32.h>
|
|
107 # else
|
|
108 # include <gdk/gdkx.h>
|
|
109 # endif
|
|
110 #endif
|
|
111
|
|
112 #ifdef HAVE_WCHAR_H
|
|
113 # include <wchar.h>
|
|
114 #endif
|
|
115
|
|
116 #if 0
|
|
117 /* This has been disabled, because several people reported problems with the
|
|
118 * wcwidth() and iswprint() library functions, esp. for Hebrew. */
|
|
119 # ifdef __STDC_ISO_10646__
|
|
120 # define USE_WCHAR_FUNCTIONS
|
|
121 # endif
|
|
122 #endif
|
|
123
|
|
124 #if defined(FEAT_MBYTE) || defined(PROTO)
|
|
125
|
|
126 static int enc_canon_search __ARGS((char_u *name));
|
|
127 static int dbcs_char2len __ARGS((int c));
|
|
128 static int dbcs_char2bytes __ARGS((int c, char_u *buf));
|
474
|
129 static int dbcs_ptr2len __ARGS((char_u *p));
|
7
|
130 static int dbcs_char2cells __ARGS((int c));
|
|
131 static int dbcs_ptr2char __ARGS((char_u *p));
|
|
132
|
|
133 /* Lookup table to quickly get the length in bytes of a UTF-8 character from
|
|
134 * the first byte of a UTF-8 string. Bytes which are illegal when used as the
|
|
135 * first byte have a one, because these will be used separately. */
|
|
136 static char utf8len_tab[256] =
|
|
137 {
|
|
138 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
|
|
139 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
|
|
140 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
|
|
141 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
|
|
142 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /*bogus*/
|
|
143 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /*bogus*/
|
|
144 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
|
|
145 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6,1,1,
|
|
146 };
|
|
147
|
|
148 /*
|
|
149 * XIM often causes trouble. Define XIM_DEBUG to get a log of XIM callbacks
|
|
150 * in the "xim.log" file.
|
|
151 */
|
|
152 /* #define XIM_DEBUG */
|
|
153 #ifdef XIM_DEBUG
|
|
154 static void
|
|
155 xim_log(char *s, ...)
|
|
156 {
|
|
157 va_list arglist;
|
|
158 static FILE *fd = NULL;
|
|
159
|
|
160 if (fd == (FILE *)-1)
|
|
161 return;
|
|
162 if (fd == NULL)
|
|
163 {
|
531
|
164 fd = mch_fopen("xim.log", "w");
|
7
|
165 if (fd == NULL)
|
|
166 {
|
|
167 EMSG("Cannot open xim.log");
|
|
168 fd = (FILE *)-1;
|
|
169 return;
|
|
170 }
|
|
171 }
|
|
172
|
|
173 va_start(arglist, s);
|
|
174 vfprintf(fd, s, arglist);
|
|
175 va_end(arglist);
|
|
176 }
|
|
177 #endif
|
|
178
|
|
179 #endif
|
|
180
|
|
181 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO)
|
|
182 /*
|
|
183 * Canonical encoding names and their properties.
|
|
184 * "iso-8859-n" is handled by enc_canonize() directly.
|
|
185 */
|
|
186 static struct
|
|
187 { char *name; int prop; int codepage;}
|
|
188 enc_canon_table[] =
|
|
189 {
|
|
190 #define IDX_LATIN_1 0
|
|
191 {"latin1", ENC_8BIT + ENC_LATIN1, 1252},
|
|
192 #define IDX_ISO_2 1
|
|
193 {"iso-8859-2", ENC_8BIT, 0},
|
|
194 #define IDX_ISO_3 2
|
|
195 {"iso-8859-3", ENC_8BIT, 0},
|
|
196 #define IDX_ISO_4 3
|
|
197 {"iso-8859-4", ENC_8BIT, 0},
|
|
198 #define IDX_ISO_5 4
|
|
199 {"iso-8859-5", ENC_8BIT, 0},
|
|
200 #define IDX_ISO_6 5
|
|
201 {"iso-8859-6", ENC_8BIT, 0},
|
|
202 #define IDX_ISO_7 6
|
|
203 {"iso-8859-7", ENC_8BIT, 0},
|
407
|
204 #define IDX_ISO_8 7
|
7
|
205 {"iso-8859-8", ENC_8BIT, 0},
|
407
|
206 #define IDX_ISO_9 8
|
7
|
207 {"iso-8859-9", ENC_8BIT, 0},
|
407
|
208 #define IDX_ISO_10 9
|
7
|
209 {"iso-8859-10", ENC_8BIT, 0},
|
407
|
210 #define IDX_ISO_11 10
|
7
|
211 {"iso-8859-11", ENC_8BIT, 0},
|
407
|
212 #define IDX_ISO_13 11
|
7
|
213 {"iso-8859-13", ENC_8BIT, 0},
|
407
|
214 #define IDX_ISO_14 12
|
7
|
215 {"iso-8859-14", ENC_8BIT, 0},
|
407
|
216 #define IDX_ISO_15 13
|
26
|
217 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0},
|
407
|
218 #define IDX_KOI8_R 14
|
7
|
219 {"koi8-r", ENC_8BIT, 0},
|
407
|
220 #define IDX_KOI8_U 15
|
7
|
221 {"koi8-u", ENC_8BIT, 0},
|
407
|
222 #define IDX_UTF8 16
|
7
|
223 {"utf-8", ENC_UNICODE, 0},
|
407
|
224 #define IDX_UCS2 17
|
7
|
225 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0},
|
407
|
226 #define IDX_UCS2LE 18
|
7
|
227 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0},
|
407
|
228 #define IDX_UTF16 19
|
7
|
229 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0},
|
407
|
230 #define IDX_UTF16LE 20
|
7
|
231 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0},
|
407
|
232 #define IDX_UCS4 21
|
7
|
233 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0},
|
407
|
234 #define IDX_UCS4LE 22
|
7
|
235 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0},
|
407
|
236
|
|
237 /* For debugging DBCS encoding on Unix. */
|
|
238 #define IDX_DEBUG 23
|
7
|
239 {"debug", ENC_DBCS, DBCS_DEBUG},
|
407
|
240 #define IDX_EUC_JP 24
|
7
|
241 {"euc-jp", ENC_DBCS, DBCS_JPNU},
|
407
|
242 #define IDX_SJIS 25
|
7
|
243 {"sjis", ENC_DBCS, DBCS_JPN},
|
407
|
244 #define IDX_EUC_KR 26
|
7
|
245 {"euc-kr", ENC_DBCS, DBCS_KORU},
|
407
|
246 #define IDX_EUC_CN 27
|
7
|
247 {"euc-cn", ENC_DBCS, DBCS_CHSU},
|
407
|
248 #define IDX_EUC_TW 28
|
7
|
249 {"euc-tw", ENC_DBCS, DBCS_CHTU},
|
407
|
250 #define IDX_BIG5 29
|
7
|
251 {"big5", ENC_DBCS, DBCS_CHT},
|
407
|
252
|
|
253 /* MS-DOS and MS-Windows codepages are included here, so that they can be
|
|
254 * used on Unix too. Most of them are similar to ISO-8859 encodings, but
|
|
255 * not exactly the same. */
|
|
256 #define IDX_CP437 30
|
|
257 {"cp437", ENC_8BIT, 437}, /* like iso-8859-1 */
|
|
258 #define IDX_CP737 31
|
|
259 {"cp737", ENC_8BIT, 737}, /* like iso-8859-7 */
|
|
260 #define IDX_CP775 32
|
|
261 {"cp775", ENC_8BIT, 775}, /* Baltic */
|
|
262 #define IDX_CP850 33
|
|
263 {"cp850", ENC_8BIT, 850}, /* like iso-8859-4 */
|
|
264 #define IDX_CP852 34
|
|
265 {"cp852", ENC_8BIT, 852}, /* like iso-8859-1 */
|
|
266 #define IDX_CP855 35
|
|
267 {"cp855", ENC_8BIT, 855}, /* like iso-8859-2 */
|
|
268 #define IDX_CP857 36
|
|
269 {"cp857", ENC_8BIT, 857}, /* like iso-8859-5 */
|
|
270 #define IDX_CP860 37
|
|
271 {"cp860", ENC_8BIT, 860}, /* like iso-8859-9 */
|
|
272 #define IDX_CP861 38
|
|
273 {"cp861", ENC_8BIT, 861}, /* like iso-8859-1 */
|
|
274 #define IDX_CP862 39
|
|
275 {"cp862", ENC_8BIT, 862}, /* like iso-8859-1 */
|
|
276 #define IDX_CP863 40
|
|
277 {"cp863", ENC_8BIT, 863}, /* like iso-8859-8 */
|
|
278 #define IDX_CP865 41
|
|
279 {"cp865", ENC_8BIT, 865}, /* like iso-8859-1 */
|
|
280 #define IDX_CP866 42
|
|
281 {"cp866", ENC_8BIT, 866}, /* like iso-8859-5 */
|
|
282 #define IDX_CP869 43
|
|
283 {"cp869", ENC_8BIT, 869}, /* like iso-8859-7 */
|
|
284 #define IDX_CP874 44
|
|
285 {"cp874", ENC_8BIT, 874}, /* Thai */
|
|
286 #define IDX_CP932 45
|
|
287 {"cp932", ENC_DBCS, DBCS_JPN},
|
|
288 #define IDX_CP936 46
|
|
289 {"cp936", ENC_DBCS, DBCS_CHS},
|
|
290 #define IDX_CP949 47
|
|
291 {"cp949", ENC_DBCS, DBCS_KOR},
|
|
292 #define IDX_CP950 48
|
|
293 {"cp950", ENC_DBCS, DBCS_CHT},
|
|
294 #define IDX_CP1250 49
|
|
295 {"cp1250", ENC_8BIT, 1250}, /* Czech, Polish, etc. */
|
|
296 #define IDX_CP1251 50
|
|
297 {"cp1251", ENC_8BIT, 1251}, /* Cyrillic */
|
|
298 /* cp1252 is considered to be equal to latin1 */
|
|
299 #define IDX_CP1253 51
|
|
300 {"cp1253", ENC_8BIT, 1253}, /* Greek */
|
|
301 #define IDX_CP1254 52
|
|
302 {"cp1254", ENC_8BIT, 1254}, /* Turkish */
|
|
303 #define IDX_CP1255 53
|
|
304 {"cp1255", ENC_8BIT, 1255}, /* Hebrew */
|
|
305 #define IDX_CP1256 54
|
|
306 {"cp1256", ENC_8BIT, 1256}, /* Arabic */
|
|
307 #define IDX_CP1257 55
|
|
308 {"cp1257", ENC_8BIT, 1257}, /* Baltic */
|
|
309 #define IDX_CP1258 56
|
|
310 {"cp1258", ENC_8BIT, 1258}, /* Vietnamese */
|
|
311
|
|
312 #define IDX_MACROMAN 57
|
|
313 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, /* Mac OS */
|
890
|
314 #define IDX_DECMCS 58
|
|
315 {"dec-mcs", ENC_8BIT, 0}, /* DEC MCS */
|
|
316 #define IDX_HPROMAN8 59
|
|
317 {"hp-roman8", ENC_8BIT, 0}, /* HP Roman8 */
|
|
318 #define IDX_COUNT 60
|
7
|
319 };
|
|
320
|
|
321 /*
|
|
322 * Aliases for encoding names.
|
|
323 */
|
|
324 static struct
|
|
325 { char *name; int canon;}
|
|
326 enc_alias_table[] =
|
|
327 {
|
|
328 {"ansi", IDX_LATIN_1},
|
|
329 {"iso-8859-1", IDX_LATIN_1},
|
|
330 {"latin2", IDX_ISO_2},
|
|
331 {"latin3", IDX_ISO_3},
|
|
332 {"latin4", IDX_ISO_4},
|
|
333 {"cyrillic", IDX_ISO_5},
|
|
334 {"arabic", IDX_ISO_6},
|
|
335 {"greek", IDX_ISO_7},
|
|
336 #ifdef WIN3264
|
|
337 {"hebrew", IDX_CP1255},
|
|
338 #else
|
|
339 {"hebrew", IDX_ISO_8},
|
|
340 #endif
|
|
341 {"latin5", IDX_ISO_9},
|
|
342 {"turkish", IDX_ISO_9}, /* ? */
|
|
343 {"latin6", IDX_ISO_10},
|
|
344 {"nordic", IDX_ISO_10}, /* ? */
|
|
345 {"thai", IDX_ISO_11}, /* ? */
|
|
346 {"latin7", IDX_ISO_13},
|
|
347 {"latin8", IDX_ISO_14},
|
|
348 {"latin9", IDX_ISO_15},
|
|
349 {"utf8", IDX_UTF8},
|
|
350 {"unicode", IDX_UCS2},
|
|
351 {"ucs2", IDX_UCS2},
|
|
352 {"ucs2be", IDX_UCS2},
|
|
353 {"ucs-2be", IDX_UCS2},
|
|
354 {"ucs2le", IDX_UCS2LE},
|
|
355 {"utf16", IDX_UTF16},
|
|
356 {"utf16be", IDX_UTF16},
|
|
357 {"utf-16be", IDX_UTF16},
|
|
358 {"utf16le", IDX_UTF16LE},
|
|
359 {"ucs4", IDX_UCS4},
|
|
360 {"ucs4be", IDX_UCS4},
|
|
361 {"ucs-4be", IDX_UCS4},
|
|
362 {"ucs4le", IDX_UCS4LE},
|
1540
|
363 {"utf32", IDX_UCS4},
|
|
364 {"utf-32", IDX_UCS4},
|
|
365 {"utf32be", IDX_UCS4},
|
|
366 {"utf-32be", IDX_UCS4},
|
|
367 {"utf32le", IDX_UCS4LE},
|
|
368 {"utf-32le", IDX_UCS4LE},
|
7
|
369 {"932", IDX_CP932},
|
|
370 {"949", IDX_CP949},
|
|
371 {"936", IDX_CP936},
|
932
|
372 {"gbk", IDX_CP936},
|
7
|
373 {"950", IDX_CP950},
|
|
374 {"eucjp", IDX_EUC_JP},
|
|
375 {"unix-jis", IDX_EUC_JP},
|
|
376 {"ujis", IDX_EUC_JP},
|
|
377 {"shift-jis", IDX_SJIS},
|
|
378 {"euckr", IDX_EUC_KR},
|
|
379 {"5601", IDX_EUC_KR}, /* Sun: KS C 5601 */
|
|
380 {"euccn", IDX_EUC_CN},
|
|
381 {"gb2312", IDX_EUC_CN},
|
|
382 {"euctw", IDX_EUC_TW},
|
|
383 #if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS)
|
|
384 {"japan", IDX_CP932},
|
|
385 {"korea", IDX_CP949},
|
|
386 {"prc", IDX_CP936},
|
|
387 {"chinese", IDX_CP936},
|
|
388 {"taiwan", IDX_CP950},
|
|
389 {"big5", IDX_CP950},
|
|
390 #else
|
|
391 {"japan", IDX_EUC_JP},
|
|
392 {"korea", IDX_EUC_KR},
|
|
393 {"prc", IDX_EUC_CN},
|
|
394 {"chinese", IDX_EUC_CN},
|
|
395 {"taiwan", IDX_EUC_TW},
|
|
396 {"cp950", IDX_BIG5},
|
|
397 {"950", IDX_BIG5},
|
|
398 #endif
|
|
399 {"mac", IDX_MACROMAN},
|
890
|
400 {"mac-roman", IDX_MACROMAN},
|
7
|
401 {NULL, 0}
|
|
402 };
|
|
403
|
|
404 #ifndef CP_UTF8
|
|
405 # define CP_UTF8 65001 /* magic number from winnls.h */
|
|
406 #endif
|
|
407
|
|
408 /*
|
|
409 * Find encoding "name" in the list of canonical encoding names.
|
|
410 * Returns -1 if not found.
|
|
411 */
|
|
412 static int
|
|
413 enc_canon_search(name)
|
|
414 char_u *name;
|
|
415 {
|
|
416 int i;
|
|
417
|
|
418 for (i = 0; i < IDX_COUNT; ++i)
|
|
419 if (STRCMP(name, enc_canon_table[i].name) == 0)
|
|
420 return i;
|
|
421 return -1;
|
|
422 }
|
|
423
|
|
424 #endif
|
|
425
|
|
426 #if defined(FEAT_MBYTE) || defined(PROTO)
|
|
427
|
|
428 /*
|
|
429 * Find canonical encoding "name" in the list and return its properties.
|
|
430 * Returns 0 if not found.
|
|
431 */
|
|
432 int
|
|
433 enc_canon_props(name)
|
|
434 char_u *name;
|
|
435 {
|
|
436 int i;
|
|
437
|
|
438 i = enc_canon_search(name);
|
|
439 if (i >= 0)
|
|
440 return enc_canon_table[i].prop;
|
|
441 #ifdef WIN3264
|
|
442 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2]))
|
|
443 {
|
|
444 CPINFO cpinfo;
|
|
445
|
|
446 /* Get info on this codepage to find out what it is. */
|
|
447 if (GetCPInfo(atoi(name + 2), &cpinfo) != 0)
|
|
448 {
|
|
449 if (cpinfo.MaxCharSize == 1) /* some single-byte encoding */
|
|
450 return ENC_8BIT;
|
|
451 if (cpinfo.MaxCharSize == 2
|
|
452 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0))
|
|
453 /* must be a DBCS encoding */
|
|
454 return ENC_DBCS;
|
|
455 }
|
|
456 return 0;
|
|
457 }
|
|
458 #endif
|
|
459 if (STRNCMP(name, "2byte-", 6) == 0)
|
|
460 return ENC_DBCS;
|
|
461 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0)
|
|
462 return ENC_8BIT;
|
|
463 return 0;
|
|
464 }
|
|
465
|
|
466 /*
|
|
467 * Set up for using multi-byte characters.
|
|
468 * Called in three cases:
|
|
469 * - by main() to initialize (p_enc == NULL)
|
|
470 * - by set_init_1() after 'encoding' was set to its default.
|
|
471 * - by do_set() when 'encoding' has been set.
|
|
472 * p_enc must have been passed through enc_canonize() already.
|
|
473 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags.
|
|
474 * Fills mb_bytelen_tab[] and returns NULL when there are no problems.
|
|
475 * When there is something wrong: Returns an error message and doesn't change
|
|
476 * anything.
|
|
477 */
|
|
478 char_u *
|
|
479 mb_init()
|
|
480 {
|
|
481 int i;
|
|
482 int idx;
|
|
483 int n;
|
|
484 int enc_dbcs_new = 0;
|
|
485 #if defined(USE_ICONV) && !defined(WIN3264) && !defined(WIN32UNIX) \
|
|
486 && !defined(MACOS)
|
|
487 # define LEN_FROM_CONV
|
|
488 vimconv_T vimconv;
|
|
489 char_u *p;
|
|
490 #endif
|
|
491
|
|
492 if (p_enc == NULL)
|
|
493 {
|
|
494 /* Just starting up: set the whole table to one's. */
|
|
495 for (i = 0; i < 256; ++i)
|
|
496 mb_bytelen_tab[i] = 1;
|
|
497 input_conv.vc_type = CONV_NONE;
|
|
498 input_conv.vc_factor = 1;
|
|
499 output_conv.vc_type = CONV_NONE;
|
|
500 return NULL;
|
|
501 }
|
|
502
|
|
503 #ifdef WIN3264
|
|
504 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2]))
|
|
505 {
|
|
506 CPINFO cpinfo;
|
|
507
|
|
508 /* Get info on this codepage to find out what it is. */
|
|
509 if (GetCPInfo(atoi(p_enc + 2), &cpinfo) != 0)
|
|
510 {
|
|
511 if (cpinfo.MaxCharSize == 1)
|
|
512 {
|
|
513 /* some single-byte encoding */
|
|
514 enc_unicode = 0;
|
|
515 enc_utf8 = FALSE;
|
|
516 }
|
|
517 else if (cpinfo.MaxCharSize == 2
|
|
518 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0))
|
|
519 {
|
|
520 /* must be a DBCS encoding, check below */
|
|
521 enc_dbcs_new = atoi(p_enc + 2);
|
|
522 }
|
|
523 else
|
|
524 goto codepage_invalid;
|
|
525 }
|
|
526 else if (GetLastError() == ERROR_INVALID_PARAMETER)
|
|
527 {
|
|
528 codepage_invalid:
|
|
529 return (char_u *)N_("E543: Not a valid codepage");
|
|
530 }
|
|
531 }
|
|
532 #endif
|
|
533 else if (STRNCMP(p_enc, "8bit-", 5) == 0
|
|
534 || STRNCMP(p_enc, "iso-8859-", 9) == 0)
|
|
535 {
|
|
536 /* Accept any "8bit-" or "iso-8859-" name. */
|
|
537 enc_unicode = 0;
|
|
538 enc_utf8 = FALSE;
|
|
539 }
|
|
540 else if (STRNCMP(p_enc, "2byte-", 6) == 0)
|
|
541 {
|
|
542 #ifdef WIN3264
|
|
543 /* Windows: accept only valid codepage numbers, check below. */
|
|
544 if (p_enc[6] != 'c' || p_enc[7] != 'p'
|
|
545 || (enc_dbcs_new = atoi(p_enc + 8)) == 0)
|
|
546 return e_invarg;
|
|
547 #else
|
|
548 /* Unix: accept any "2byte-" name, assume current locale. */
|
|
549 enc_dbcs_new = DBCS_2BYTE;
|
|
550 #endif
|
|
551 }
|
|
552 else if ((idx = enc_canon_search(p_enc)) >= 0)
|
|
553 {
|
|
554 i = enc_canon_table[idx].prop;
|
|
555 if (i & ENC_UNICODE)
|
|
556 {
|
|
557 /* Unicode */
|
|
558 enc_utf8 = TRUE;
|
|
559 if (i & (ENC_2BYTE | ENC_2WORD))
|
|
560 enc_unicode = 2;
|
|
561 else if (i & ENC_4BYTE)
|
|
562 enc_unicode = 4;
|
|
563 else
|
|
564 enc_unicode = 0;
|
|
565 }
|
|
566 else if (i & ENC_DBCS)
|
|
567 {
|
|
568 /* 2byte, handle below */
|
|
569 enc_dbcs_new = enc_canon_table[idx].codepage;
|
|
570 }
|
|
571 else
|
|
572 {
|
|
573 /* Must be 8-bit. */
|
|
574 enc_unicode = 0;
|
|
575 enc_utf8 = FALSE;
|
|
576 }
|
|
577 }
|
|
578 else /* Don't know what encoding this is, reject it. */
|
|
579 return e_invarg;
|
|
580
|
|
581 if (enc_dbcs_new != 0)
|
|
582 {
|
|
583 #ifdef WIN3264
|
|
584 /* Check if the DBCS code page is OK. */
|
|
585 if (!IsValidCodePage(enc_dbcs_new))
|
|
586 goto codepage_invalid;
|
|
587 #endif
|
|
588 enc_unicode = 0;
|
|
589 enc_utf8 = FALSE;
|
|
590 }
|
|
591 enc_dbcs = enc_dbcs_new;
|
|
592 has_mbyte = (enc_dbcs != 0 || enc_utf8);
|
|
593
|
|
594 #ifdef WIN3264
|
|
595 enc_codepage = encname2codepage(p_enc);
|
26
|
596 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0);
|
7
|
597 #endif
|
|
598
|
492
|
599 /* Detect an encoding that uses latin1 characters. */
|
|
600 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0
|
|
601 || STRCMP(p_enc, "iso-8859-15") == 0);
|
|
602
|
7
|
603 /*
|
|
604 * Set the function pointers.
|
|
605 */
|
|
606 if (enc_utf8)
|
|
607 {
|
474
|
608 mb_ptr2len = utfc_ptr2len;
|
7
|
609 mb_char2len = utf_char2len;
|
|
610 mb_char2bytes = utf_char2bytes;
|
|
611 mb_ptr2cells = utf_ptr2cells;
|
|
612 mb_char2cells = utf_char2cells;
|
|
613 mb_off2cells = utf_off2cells;
|
|
614 mb_ptr2char = utf_ptr2char;
|
|
615 mb_head_off = utf_head_off;
|
|
616 }
|
|
617 else if (enc_dbcs != 0)
|
|
618 {
|
474
|
619 mb_ptr2len = dbcs_ptr2len;
|
7
|
620 mb_char2len = dbcs_char2len;
|
|
621 mb_char2bytes = dbcs_char2bytes;
|
|
622 mb_ptr2cells = dbcs_ptr2cells;
|
|
623 mb_char2cells = dbcs_char2cells;
|
|
624 mb_off2cells = dbcs_off2cells;
|
|
625 mb_ptr2char = dbcs_ptr2char;
|
|
626 mb_head_off = dbcs_head_off;
|
|
627 }
|
|
628 else
|
|
629 {
|
474
|
630 mb_ptr2len = latin_ptr2len;
|
7
|
631 mb_char2len = latin_char2len;
|
|
632 mb_char2bytes = latin_char2bytes;
|
|
633 mb_ptr2cells = latin_ptr2cells;
|
|
634 mb_char2cells = latin_char2cells;
|
|
635 mb_off2cells = latin_off2cells;
|
|
636 mb_ptr2char = latin_ptr2char;
|
|
637 mb_head_off = latin_head_off;
|
|
638 }
|
|
639
|
|
640 /*
|
|
641 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN().
|
|
642 */
|
|
643 #ifdef LEN_FROM_CONV
|
|
644 /* When 'encoding' is different from the current locale mblen() won't
|
|
645 * work. Use conversion to "utf-8" instead. */
|
|
646 vimconv.vc_type = CONV_NONE;
|
|
647 if (enc_dbcs)
|
|
648 {
|
|
649 p = enc_locale();
|
|
650 if (p == NULL || STRCMP(p, p_enc) != 0)
|
|
651 {
|
|
652 convert_setup(&vimconv, p_enc, (char_u *)"utf-8");
|
|
653 vimconv.vc_fail = TRUE;
|
|
654 }
|
|
655 vim_free(p);
|
|
656 }
|
|
657 #endif
|
|
658
|
|
659 for (i = 0; i < 256; ++i)
|
|
660 {
|
|
661 /* Our own function to reliably check the length of UTF-8 characters,
|
|
662 * independent of mblen(). */
|
|
663 if (enc_utf8)
|
|
664 n = utf8len_tab[i];
|
|
665 else if (enc_dbcs == 0)
|
|
666 n = 1;
|
|
667 else
|
|
668 {
|
|
669 #if defined(WIN3264) || defined(WIN32UNIX)
|
|
670 /* enc_dbcs is set by setting 'fileencoding'. It becomes a Windows
|
|
671 * CodePage identifier, which we can pass directly in to Windows
|
|
672 * API */
|
|
673 n = IsDBCSLeadByteEx(enc_dbcs, (BYTE)i) ? 2 : 1;
|
|
674 #else
|
1030
|
675 # if defined(MACOS) || defined(__amigaos4__)
|
7
|
676 /*
|
|
677 * if mblen() is not available, character which MSB is turned on
|
|
678 * are treated as leading byte character. (note : This assumption
|
|
679 * is not always true.)
|
|
680 */
|
|
681 n = (i & 0x80) ? 2 : 1;
|
|
682 # else
|
|
683 char buf[MB_MAXBYTES];
|
|
684 # ifdef X_LOCALE
|
|
685 # ifndef mblen
|
|
686 # define mblen _Xmblen
|
|
687 # endif
|
|
688 # endif
|
|
689 if (i == NUL) /* just in case mblen() can't handle "" */
|
|
690 n = 1;
|
|
691 else
|
|
692 {
|
|
693 buf[0] = i;
|
|
694 buf[1] = 0;
|
|
695 #ifdef LEN_FROM_CONV
|
|
696 if (vimconv.vc_type != CONV_NONE)
|
|
697 {
|
|
698 /*
|
|
699 * string_convert() should fail when converting the first
|
|
700 * byte of a double-byte character.
|
|
701 */
|
|
702 p = string_convert(&vimconv, (char_u *)buf, NULL);
|
|
703 if (p != NULL)
|
|
704 {
|
|
705 vim_free(p);
|
|
706 n = 1;
|
|
707 }
|
|
708 else
|
|
709 n = 2;
|
|
710 }
|
|
711 else
|
|
712 #endif
|
|
713 {
|
|
714 /*
|
|
715 * mblen() should return -1 for invalid (means the leading
|
|
716 * multibyte) character. However there are some platforms
|
|
717 * where mblen() returns 0 for invalid character.
|
|
718 * Therefore, following condition includes 0.
|
|
719 */
|
418
|
720 (void)mblen(NULL, 0); /* First reset the state. */
|
7
|
721 if (mblen(buf, (size_t)1) <= 0)
|
|
722 n = 2;
|
|
723 else
|
|
724 n = 1;
|
|
725 }
|
|
726 }
|
|
727 # endif
|
|
728 #endif
|
|
729 }
|
|
730
|
|
731 mb_bytelen_tab[i] = n;
|
|
732 }
|
|
733
|
|
734 #ifdef LEN_FROM_CONV
|
|
735 convert_setup(&vimconv, NULL, NULL);
|
|
736 #endif
|
|
737
|
|
738 /* The cell width depends on the type of multi-byte characters. */
|
|
739 (void)init_chartab();
|
|
740
|
|
741 /* When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] */
|
|
742 screenalloc(FALSE);
|
|
743
|
|
744 /* When using Unicode, set default for 'fileencodings'. */
|
|
745 if (enc_utf8 && !option_was_set((char_u *)"fencs"))
|
|
746 set_string_option_direct((char_u *)"fencs", -1,
|
694
|
747 (char_u *)"ucs-bom,utf-8,default,latin1", OPT_FREE, 0);
|
|
748
|
7
|
749 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT)
|
|
750 /* GNU gettext 0.10.37 supports this feature: set the codeset used for
|
|
751 * translated messages independently from the current locale. */
|
|
752 (void)bind_textdomain_codeset(VIMPACKAGE,
|
|
753 enc_utf8 ? "utf-8" : (char *)p_enc);
|
|
754 #endif
|
|
755
|
23
|
756 #ifdef WIN32
|
|
757 /* When changing 'encoding' while starting up, then convert the command
|
|
758 * line arguments from the active codepage to 'encoding'. */
|
|
759 if (starting != 0)
|
|
760 fix_arg_enc();
|
|
761 #endif
|
|
762
|
7
|
763 #ifdef FEAT_AUTOCMD
|
|
764 /* Fire an autocommand to let people do custom font setup. This must be
|
|
765 * after Vim has been setup for the new encoding. */
|
|
766 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf);
|
|
767 #endif
|
|
768
|
740
|
769 #ifdef FEAT_SPELL
|
236
|
770 /* Need to reload spell dictionaries */
|
|
771 spell_reload();
|
|
772 #endif
|
|
773
|
7
|
774 return NULL;
|
|
775 }
|
|
776
|
|
777 /*
|
|
778 * Return the size of the BOM for the current buffer:
|
|
779 * 0 - no BOM
|
|
780 * 2 - UCS-2 or UTF-16 BOM
|
|
781 * 4 - UCS-4 BOM
|
|
782 * 3 - UTF-8 BOM
|
|
783 */
|
|
784 int
|
|
785 bomb_size()
|
|
786 {
|
|
787 int n = 0;
|
|
788
|
|
789 if (curbuf->b_p_bomb && !curbuf->b_p_bin)
|
|
790 {
|
|
791 if (*curbuf->b_p_fenc == NUL)
|
|
792 {
|
|
793 if (enc_utf8)
|
|
794 {
|
|
795 if (enc_unicode != 0)
|
|
796 n = enc_unicode;
|
|
797 else
|
|
798 n = 3;
|
|
799 }
|
|
800 }
|
|
801 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0)
|
|
802 n = 3;
|
|
803 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0
|
|
804 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0)
|
|
805 n = 2;
|
|
806 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0)
|
|
807 n = 4;
|
|
808 }
|
|
809 return n;
|
|
810 }
|
|
811
|
|
812 /*
|
|
813 * Get class of pointer:
|
|
814 * 0 for blank or NUL
|
|
815 * 1 for punctuation
|
|
816 * 2 for an (ASCII) word character
|
|
817 * >2 for other word characters
|
|
818 */
|
|
819 int
|
|
820 mb_get_class(p)
|
|
821 char_u *p;
|
|
822 {
|
|
823 if (MB_BYTE2LEN(p[0]) == 1)
|
|
824 {
|
|
825 if (p[0] == NUL || vim_iswhite(p[0]))
|
|
826 return 0;
|
|
827 if (vim_iswordc(p[0]))
|
|
828 return 2;
|
|
829 return 1;
|
|
830 }
|
|
831 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL)
|
|
832 return dbcs_class(p[0], p[1]);
|
|
833 if (enc_utf8)
|
|
834 return utf_class(utf_ptr2char(p));
|
|
835 return 0;
|
|
836 }
|
|
837
|
|
838 /*
|
|
839 * Get class of a double-byte character. This always returns 3 or bigger.
|
|
840 * TODO: Should return 1 for punctuation.
|
|
841 */
|
|
842 int
|
|
843 dbcs_class(lead, trail)
|
|
844 unsigned lead;
|
|
845 unsigned trail;
|
|
846 {
|
|
847 switch (enc_dbcs)
|
|
848 {
|
|
849 /* please add classfy routine for your language in here */
|
|
850
|
|
851 case DBCS_JPNU: /* ? */
|
|
852 case DBCS_JPN:
|
|
853 {
|
|
854 /* JIS code classification */
|
|
855 unsigned char lb = lead;
|
|
856 unsigned char tb = trail;
|
|
857
|
|
858 /* convert process code to JIS */
|
|
859 # if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS)
|
|
860 /* process code is SJIS */
|
|
861 if (lb <= 0x9f)
|
|
862 lb = (lb - 0x81) * 2 + 0x21;
|
|
863 else
|
|
864 lb = (lb - 0xc1) * 2 + 0x21;
|
|
865 if (tb <= 0x7e)
|
|
866 tb -= 0x1f;
|
|
867 else if (tb <= 0x9e)
|
|
868 tb -= 0x20;
|
|
869 else
|
|
870 {
|
|
871 tb -= 0x7e;
|
|
872 lb += 1;
|
|
873 }
|
|
874 # else
|
|
875 /*
|
|
876 * XXX: Code page identification can not use with all
|
|
877 * system! So, some other encoding information
|
|
878 * will be needed.
|
|
879 * In japanese: SJIS,EUC,UNICODE,(JIS)
|
|
880 * Note that JIS-code system don't use as
|
|
881 * process code in most system because it uses
|
|
882 * escape sequences(JIS is context depend encoding).
|
|
883 */
|
|
884 /* assume process code is JAPANESE-EUC */
|
|
885 lb &= 0x7f;
|
|
886 tb &= 0x7f;
|
|
887 # endif
|
|
888 /* exceptions */
|
|
889 switch (lb << 8 | tb)
|
|
890 {
|
|
891 case 0x2121: /* ZENKAKU space */
|
|
892 return 0;
|
|
893 case 0x2122: /* KU-TEN (Japanese comma) */
|
|
894 case 0x2123: /* TOU-TEN (Japanese period) */
|
|
895 case 0x2124: /* ZENKAKU comma */
|
|
896 case 0x2125: /* ZENKAKU period */
|
|
897 return 1;
|
|
898 case 0x213c: /* prolongedsound handled as KATAKANA */
|
|
899 return 13;
|
|
900 }
|
|
901 /* sieved by KU code */
|
|
902 switch (lb)
|
|
903 {
|
|
904 case 0x21:
|
|
905 case 0x22:
|
|
906 /* special symbols */
|
|
907 return 10;
|
|
908 case 0x23:
|
|
909 /* alpha-numeric */
|
|
910 return 11;
|
|
911 case 0x24:
|
|
912 /* hiragana */
|
|
913 return 12;
|
|
914 case 0x25:
|
|
915 /* katakana */
|
|
916 return 13;
|
|
917 case 0x26:
|
|
918 /* greek */
|
|
919 return 14;
|
|
920 case 0x27:
|
|
921 /* russian */
|
|
922 return 15;
|
|
923 case 0x28:
|
|
924 /* lines */
|
|
925 return 16;
|
|
926 default:
|
|
927 /* kanji */
|
|
928 return 17;
|
|
929 }
|
|
930 }
|
|
931
|
|
932 case DBCS_KORU: /* ? */
|
|
933 case DBCS_KOR:
|
|
934 {
|
|
935 /* KS code classification */
|
|
936 unsigned char c1 = lead;
|
|
937 unsigned char c2 = trail;
|
|
938
|
|
939 /*
|
|
940 * 20 : Hangul
|
|
941 * 21 : Hanja
|
|
942 * 22 : Symbols
|
|
943 * 23 : Alpha-numeric/Roman Letter (Full width)
|
|
944 * 24 : Hangul Letter(Alphabet)
|
|
945 * 25 : Roman Numeral/Greek Letter
|
|
946 * 26 : Box Drawings
|
|
947 * 27 : Unit Symbols
|
|
948 * 28 : Circled/Parenthesized Letter
|
|
949 * 29 : Hirigana/Katakana
|
|
950 * 30 : Cyrillic Letter
|
|
951 */
|
|
952
|
|
953 if (c1 >= 0xB0 && c1 <= 0xC8)
|
|
954 /* Hangul */
|
|
955 return 20;
|
|
956 #if defined(WIN3264) || defined(WIN32UNIX)
|
|
957 else if (c1 <= 0xA0 || c2 <= 0xA0)
|
|
958 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */
|
|
959 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE
|
|
960 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0
|
|
961 */
|
|
962 return 20;
|
|
963 #endif
|
|
964
|
|
965 else if (c1 >= 0xCA && c1 <= 0xFD)
|
|
966 /* Hanja */
|
|
967 return 21;
|
|
968 else switch (c1)
|
|
969 {
|
|
970 case 0xA1:
|
|
971 case 0xA2:
|
|
972 /* Symbols */
|
|
973 return 22;
|
|
974 case 0xA3:
|
|
975 /* Alpha-numeric */
|
|
976 return 23;
|
|
977 case 0xA4:
|
|
978 /* Hangul Letter(Alphabet) */
|
|
979 return 24;
|
|
980 case 0xA5:
|
|
981 /* Roman Numeral/Greek Letter */
|
|
982 return 25;
|
|
983 case 0xA6:
|
|
984 /* Box Drawings */
|
|
985 return 26;
|
|
986 case 0xA7:
|
|
987 /* Unit Symbols */
|
|
988 return 27;
|
|
989 case 0xA8:
|
|
990 case 0xA9:
|
|
991 if (c2 <= 0xAF)
|
|
992 return 25; /* Roman Letter */
|
|
993 else if (c2 >= 0xF6)
|
|
994 return 22; /* Symbols */
|
|
995 else
|
|
996 /* Circled/Parenthesized Letter */
|
|
997 return 28;
|
|
998 case 0xAA:
|
|
999 case 0xAB:
|
|
1000 /* Hirigana/Katakana */
|
|
1001 return 29;
|
|
1002 case 0xAC:
|
|
1003 /* Cyrillic Letter */
|
|
1004 return 30;
|
|
1005 }
|
|
1006 }
|
|
1007 default:
|
|
1008 break;
|
|
1009 }
|
|
1010 return 3;
|
|
1011 }
|
|
1012
|
|
1013 /*
|
|
1014 * mb_char2len() function pointer.
|
|
1015 * Return length in bytes of character "c".
|
|
1016 * Returns 1 for a single-byte character.
|
|
1017 */
|
|
1018 /* ARGSUSED */
|
|
1019 int
|
|
1020 latin_char2len(c)
|
|
1021 int c;
|
|
1022 {
|
|
1023 return 1;
|
|
1024 }
|
|
1025
|
|
1026 static int
|
|
1027 dbcs_char2len(c)
|
|
1028 int c;
|
|
1029 {
|
|
1030 if (c >= 0x100)
|
|
1031 return 2;
|
|
1032 return 1;
|
|
1033 }
|
|
1034
|
|
1035 /*
|
|
1036 * mb_char2bytes() function pointer.
|
|
1037 * Convert a character to its bytes.
|
|
1038 * Returns the length in bytes.
|
|
1039 */
|
|
1040 int
|
|
1041 latin_char2bytes(c, buf)
|
|
1042 int c;
|
|
1043 char_u *buf;
|
|
1044 {
|
|
1045 buf[0] = c;
|
|
1046 return 1;
|
|
1047 }
|
|
1048
|
|
1049 static int
|
|
1050 dbcs_char2bytes(c, buf)
|
|
1051 int c;
|
|
1052 char_u *buf;
|
|
1053 {
|
|
1054 if (c >= 0x100)
|
|
1055 {
|
|
1056 buf[0] = (unsigned)c >> 8;
|
|
1057 buf[1] = c;
|
221
|
1058 /* Never use a NUL byte, it causes lots of trouble. It's an invalid
|
|
1059 * character anyway. */
|
|
1060 if (buf[1] == NUL)
|
|
1061 buf[1] = '\n';
|
7
|
1062 return 2;
|
|
1063 }
|
|
1064 buf[0] = c;
|
|
1065 return 1;
|
|
1066 }
|
|
1067
|
|
1068 /*
|
474
|
1069 * mb_ptr2len() function pointer.
|
7
|
1070 * Get byte length of character at "*p" but stop at a NUL.
|
|
1071 * For UTF-8 this includes following composing characters.
|
|
1072 * Returns 0 when *p is NUL.
|
|
1073 *
|
|
1074 */
|
|
1075 int
|
474
|
1076 latin_ptr2len(p)
|
7
|
1077 char_u *p;
|
|
1078 {
|
|
1079 return MB_BYTE2LEN(*p);
|
|
1080 }
|
|
1081
|
|
1082 static int
|
474
|
1083 dbcs_ptr2len(p)
|
7
|
1084 char_u *p;
|
|
1085 {
|
|
1086 int len;
|
|
1087
|
|
1088 /* Check if second byte is not missing. */
|
|
1089 len = MB_BYTE2LEN(*p);
|
|
1090 if (len == 2 && p[1] == NUL)
|
|
1091 len = 1;
|
|
1092 return len;
|
|
1093 }
|
|
1094
|
|
1095 struct interval
|
|
1096 {
|
|
1097 unsigned short first;
|
|
1098 unsigned short last;
|
|
1099 };
|
|
1100 static int intable __ARGS((struct interval *table, size_t size, int c));
|
|
1101
|
|
1102 /*
|
|
1103 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]".
|
|
1104 */
|
|
1105 static int
|
|
1106 intable(table, size, c)
|
|
1107 struct interval *table;
|
|
1108 size_t size;
|
|
1109 int c;
|
|
1110 {
|
|
1111 int mid, bot, top;
|
|
1112
|
|
1113 /* first quick check for Latin1 etc. characters */
|
|
1114 if (c < table[0].first)
|
|
1115 return FALSE;
|
|
1116
|
|
1117 /* binary search in table */
|
|
1118 bot = 0;
|
835
|
1119 top = (int)(size / sizeof(struct interval) - 1);
|
7
|
1120 while (top >= bot)
|
|
1121 {
|
|
1122 mid = (bot + top) / 2;
|
|
1123 if (table[mid].last < c)
|
|
1124 bot = mid + 1;
|
|
1125 else if (table[mid].first > c)
|
|
1126 top = mid - 1;
|
|
1127 else
|
|
1128 return TRUE;
|
|
1129 }
|
|
1130 return FALSE;
|
|
1131 }
|
|
1132
|
|
1133 /*
|
|
1134 * For UTF-8 character "c" return 2 for a double-width character, 1 for others.
|
|
1135 * Returns 4 or 6 for an unprintable character.
|
|
1136 * Is only correct for characters >= 0x80.
|
|
1137 * When p_ambw is "double", return 2 for a character with East Asian Width
|
|
1138 * class 'A'(mbiguous).
|
|
1139 */
|
|
1140 int
|
|
1141 utf_char2cells(c)
|
|
1142 int c;
|
|
1143 {
|
|
1144 /* sorted list of non-overlapping intervals of East Asian Ambiguous
|
|
1145 * characters, generated with:
|
|
1146 * "uniset +WIDTH-A -cat=Me -cat=Mn -cat=Cf c" */
|
|
1147 static struct interval ambiguous[] = {
|
|
1148 {0x00A1, 0x00A1}, {0x00A4, 0x00A4}, {0x00A7, 0x00A8},
|
|
1149 {0x00AA, 0x00AA}, {0x00AE, 0x00AE}, {0x00B0, 0x00B4},
|
|
1150 {0x00B6, 0x00BA}, {0x00BC, 0x00BF}, {0x00C6, 0x00C6},
|
|
1151 {0x00D0, 0x00D0}, {0x00D7, 0x00D8}, {0x00DE, 0x00E1},
|
|
1152 {0x00E6, 0x00E6}, {0x00E8, 0x00EA}, {0x00EC, 0x00ED},
|
|
1153 {0x00F0, 0x00F0}, {0x00F2, 0x00F3}, {0x00F7, 0x00FA},
|
|
1154 {0x00FC, 0x00FC}, {0x00FE, 0x00FE}, {0x0101, 0x0101},
|
|
1155 {0x0111, 0x0111}, {0x0113, 0x0113}, {0x011B, 0x011B},
|
|
1156 {0x0126, 0x0127}, {0x012B, 0x012B}, {0x0131, 0x0133},
|
|
1157 {0x0138, 0x0138}, {0x013F, 0x0142}, {0x0144, 0x0144},
|
|
1158 {0x0148, 0x014B}, {0x014D, 0x014D}, {0x0152, 0x0153},
|
|
1159 {0x0166, 0x0167}, {0x016B, 0x016B}, {0x01CE, 0x01CE},
|
|
1160 {0x01D0, 0x01D0}, {0x01D2, 0x01D2}, {0x01D4, 0x01D4},
|
|
1161 {0x01D6, 0x01D6}, {0x01D8, 0x01D8}, {0x01DA, 0x01DA},
|
|
1162 {0x01DC, 0x01DC}, {0x0251, 0x0251}, {0x0261, 0x0261},
|
|
1163 {0x02C4, 0x02C4}, {0x02C7, 0x02C7}, {0x02C9, 0x02CB},
|
|
1164 {0x02CD, 0x02CD}, {0x02D0, 0x02D0}, {0x02D8, 0x02DB},
|
|
1165 {0x02DD, 0x02DD}, {0x02DF, 0x02DF}, {0x0391, 0x03A1},
|
|
1166 {0x03A3, 0x03A9}, {0x03B1, 0x03C1}, {0x03C3, 0x03C9},
|
|
1167 {0x0401, 0x0401}, {0x0410, 0x044F}, {0x0451, 0x0451},
|
|
1168 {0x2010, 0x2010}, {0x2013, 0x2016}, {0x2018, 0x2019},
|
|
1169 {0x201C, 0x201D}, {0x2020, 0x2022}, {0x2024, 0x2027},
|
|
1170 {0x2030, 0x2030}, {0x2032, 0x2033}, {0x2035, 0x2035},
|
|
1171 {0x203B, 0x203B}, {0x203E, 0x203E}, {0x2074, 0x2074},
|
|
1172 {0x207F, 0x207F}, {0x2081, 0x2084}, {0x20AC, 0x20AC},
|
|
1173 {0x2103, 0x2103}, {0x2105, 0x2105}, {0x2109, 0x2109},
|
|
1174 {0x2113, 0x2113}, {0x2116, 0x2116}, {0x2121, 0x2122},
|
|
1175 {0x2126, 0x2126}, {0x212B, 0x212B}, {0x2153, 0x2154},
|
|
1176 {0x215B, 0x215E}, {0x2160, 0x216B}, {0x2170, 0x2179},
|
|
1177 {0x2190, 0x2199}, {0x21B8, 0x21B9}, {0x21D2, 0x21D2},
|
|
1178 {0x21D4, 0x21D4}, {0x21E7, 0x21E7}, {0x2200, 0x2200},
|
|
1179 {0x2202, 0x2203}, {0x2207, 0x2208}, {0x220B, 0x220B},
|
|
1180 {0x220F, 0x220F}, {0x2211, 0x2211}, {0x2215, 0x2215},
|
|
1181 {0x221A, 0x221A}, {0x221D, 0x2220}, {0x2223, 0x2223},
|
|
1182 {0x2225, 0x2225}, {0x2227, 0x222C}, {0x222E, 0x222E},
|
|
1183 {0x2234, 0x2237}, {0x223C, 0x223D}, {0x2248, 0x2248},
|
|
1184 {0x224C, 0x224C}, {0x2252, 0x2252}, {0x2260, 0x2261},
|
|
1185 {0x2264, 0x2267}, {0x226A, 0x226B}, {0x226E, 0x226F},
|
|
1186 {0x2282, 0x2283}, {0x2286, 0x2287}, {0x2295, 0x2295},
|
|
1187 {0x2299, 0x2299}, {0x22A5, 0x22A5}, {0x22BF, 0x22BF},
|
|
1188 {0x2312, 0x2312}, {0x2460, 0x24E9}, {0x24EB, 0x254B},
|
|
1189 {0x2550, 0x2573}, {0x2580, 0x258F}, {0x2592, 0x2595},
|
|
1190 {0x25A0, 0x25A1}, {0x25A3, 0x25A9}, {0x25B2, 0x25B3},
|
|
1191 {0x25B6, 0x25B7}, {0x25BC, 0x25BD}, {0x25C0, 0x25C1},
|
|
1192 {0x25C6, 0x25C8}, {0x25CB, 0x25CB}, {0x25CE, 0x25D1},
|
|
1193 {0x25E2, 0x25E5}, {0x25EF, 0x25EF}, {0x2605, 0x2606},
|
|
1194 {0x2609, 0x2609}, {0x260E, 0x260F}, {0x2614, 0x2615},
|
|
1195 {0x261C, 0x261C}, {0x261E, 0x261E}, {0x2640, 0x2640},
|
|
1196 {0x2642, 0x2642}, {0x2660, 0x2661}, {0x2663, 0x2665},
|
|
1197 {0x2667, 0x266A}, {0x266C, 0x266D}, {0x266F, 0x266F},
|
|
1198 {0x273D, 0x273D}, {0x2776, 0x277F}, {0xE000, 0xF8FF},
|
|
1199 {0xFFFD, 0xFFFD}, /* {0xF0000, 0xFFFFD}, {0x100000, 0x10FFFD} */
|
|
1200 };
|
|
1201
|
|
1202 if (c >= 0x100)
|
|
1203 {
|
|
1204 #ifdef USE_WCHAR_FUNCTIONS
|
|
1205 /*
|
|
1206 * Assume the library function wcwidth() works better than our own
|
|
1207 * stuff. It should return 1 for ambiguous width chars!
|
|
1208 */
|
|
1209 int n = wcwidth(c);
|
|
1210
|
|
1211 if (n < 0)
|
|
1212 return 6; /* unprintable, displays <xxxx> */
|
|
1213 if (n > 1)
|
|
1214 return n;
|
|
1215 #else
|
|
1216 if (!utf_printable(c))
|
|
1217 return 6; /* unprintable, displays <xxxx> */
|
|
1218 if (c >= 0x1100
|
|
1219 && (c <= 0x115f /* Hangul Jamo */
|
|
1220 || c == 0x2329
|
|
1221 || c == 0x232a
|
|
1222 || (c >= 0x2e80 && c <= 0xa4cf
|
|
1223 && c != 0x303f) /* CJK ... Yi */
|
|
1224 || (c >= 0xac00 && c <= 0xd7a3) /* Hangul Syllables */
|
|
1225 || (c >= 0xf900 && c <= 0xfaff) /* CJK Compatibility
|
|
1226 Ideographs */
|
|
1227 || (c >= 0xfe30 && c <= 0xfe6f) /* CJK Compatibility Forms */
|
|
1228 || (c >= 0xff00 && c <= 0xff60) /* Fullwidth Forms */
|
|
1229 || (c >= 0xffe0 && c <= 0xffe6)
|
|
1230 || (c >= 0x20000 && c <= 0x2fffd)
|
|
1231 || (c >= 0x30000 && c <= 0x3fffd)))
|
|
1232 return 2;
|
|
1233 #endif
|
|
1234 }
|
|
1235
|
|
1236 /* Characters below 0x100 are influenced by 'isprint' option */
|
|
1237 else if (c >= 0x80 && !vim_isprintc(c))
|
|
1238 return 4; /* unprintable, displays <xx> */
|
|
1239
|
|
1240 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c))
|
|
1241 return 2;
|
|
1242
|
|
1243 return 1;
|
|
1244 }
|
|
1245
|
|
1246 /*
|
|
1247 * mb_ptr2cells() function pointer.
|
|
1248 * Return the number of display cells character at "*p" occupies.
|
|
1249 * This doesn't take care of unprintable characters, use ptr2cells() for that.
|
|
1250 */
|
|
1251 /*ARGSUSED*/
|
|
1252 int
|
|
1253 latin_ptr2cells(p)
|
|
1254 char_u *p;
|
|
1255 {
|
|
1256 return 1;
|
|
1257 }
|
|
1258
|
|
1259 int
|
|
1260 utf_ptr2cells(p)
|
|
1261 char_u *p;
|
|
1262 {
|
|
1263 int c;
|
|
1264
|
|
1265 /* Need to convert to a wide character. */
|
|
1266 if (*p >= 0x80)
|
|
1267 {
|
|
1268 c = utf_ptr2char(p);
|
|
1269 /* An illegal byte is displayed as <xx>. */
|
474
|
1270 if (utf_ptr2len(p) == 1 || c == NUL)
|
7
|
1271 return 4;
|
|
1272 /* If the char is ASCII it must be an overlong sequence. */
|
|
1273 if (c < 0x80)
|
|
1274 return char2cells(c);
|
|
1275 return utf_char2cells(c);
|
|
1276 }
|
|
1277 return 1;
|
|
1278 }
|
|
1279
|
|
1280 int
|
|
1281 dbcs_ptr2cells(p)
|
|
1282 char_u *p;
|
|
1283 {
|
|
1284 /* Number of cells is equal to number of bytes, except for euc-jp when
|
|
1285 * the first byte is 0x8e. */
|
|
1286 if (enc_dbcs == DBCS_JPNU && *p == 0x8e)
|
|
1287 return 1;
|
|
1288 return MB_BYTE2LEN(*p);
|
|
1289 }
|
|
1290
|
|
1291 /*
|
|
1292 * mb_char2cells() function pointer.
|
|
1293 * Return the number of display cells character "c" occupies.
|
|
1294 * Only takes care of multi-byte chars, not "^C" and such.
|
|
1295 */
|
|
1296 /*ARGSUSED*/
|
|
1297 int
|
|
1298 latin_char2cells(c)
|
|
1299 int c;
|
|
1300 {
|
|
1301 return 1;
|
|
1302 }
|
|
1303
|
|
1304 static int
|
|
1305 dbcs_char2cells(c)
|
|
1306 int c;
|
|
1307 {
|
|
1308 /* Number of cells is equal to number of bytes, except for euc-jp when
|
|
1309 * the first byte is 0x8e. */
|
|
1310 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e)
|
|
1311 return 1;
|
|
1312 /* use the first byte */
|
|
1313 return MB_BYTE2LEN((unsigned)c >> 8);
|
|
1314 }
|
|
1315
|
|
1316 /*
|
|
1317 * mb_off2cells() function pointer.
|
|
1318 * Return number of display cells for char at ScreenLines[off].
|
1378
|
1319 * We make sure that the offset used is less than "max_off".
|
7
|
1320 */
|
|
1321 /*ARGSUSED*/
|
|
1322 int
|
1378
|
1323 latin_off2cells(off, max_off)
|
7
|
1324 unsigned off;
|
1378
|
1325 unsigned max_off;
|
7
|
1326 {
|
|
1327 return 1;
|
|
1328 }
|
|
1329
|
|
1330 int
|
1378
|
1331 dbcs_off2cells(off, max_off)
|
7
|
1332 unsigned off;
|
1378
|
1333 unsigned max_off;
|
7
|
1334 {
|
1378
|
1335 /* never check beyond end of the line */
|
|
1336 if (off >= max_off)
|
|
1337 return 1;
|
|
1338
|
7
|
1339 /* Number of cells is equal to number of bytes, except for euc-jp when
|
|
1340 * the first byte is 0x8e. */
|
|
1341 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e)
|
|
1342 return 1;
|
|
1343 return MB_BYTE2LEN(ScreenLines[off]);
|
|
1344 }
|
|
1345
|
|
1346 int
|
1378
|
1347 utf_off2cells(off, max_off)
|
7
|
1348 unsigned off;
|
1378
|
1349 unsigned max_off;
|
7
|
1350 {
|
1378
|
1351 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1;
|
7
|
1352 }
|
|
1353
|
|
1354 /*
|
|
1355 * mb_ptr2char() function pointer.
|
|
1356 * Convert a byte sequence into a character.
|
|
1357 */
|
|
1358 int
|
|
1359 latin_ptr2char(p)
|
|
1360 char_u *p;
|
|
1361 {
|
|
1362 return *p;
|
|
1363 }
|
|
1364
|
|
1365 static int
|
|
1366 dbcs_ptr2char(p)
|
|
1367 char_u *p;
|
|
1368 {
|
|
1369 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL)
|
|
1370 return (p[0] << 8) + p[1];
|
|
1371 return *p;
|
|
1372 }
|
|
1373
|
|
1374 /*
|
|
1375 * Convert a UTF-8 byte sequence to a wide character.
|
|
1376 * If the sequence is illegal or truncated by a NUL the first byte is
|
|
1377 * returned.
|
|
1378 * Does not include composing characters, of course.
|
|
1379 */
|
|
1380 int
|
|
1381 utf_ptr2char(p)
|
|
1382 char_u *p;
|
|
1383 {
|
|
1384 int len;
|
|
1385
|
|
1386 if (p[0] < 0x80) /* be quick for ASCII */
|
|
1387 return p[0];
|
|
1388
|
|
1389 len = utf8len_tab[p[0]];
|
1658
|
1390 if (len > 1 && (p[1] & 0xc0) == 0x80)
|
7
|
1391 {
|
|
1392 if (len == 2)
|
|
1393 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f);
|
|
1394 if ((p[2] & 0xc0) == 0x80)
|
|
1395 {
|
|
1396 if (len == 3)
|
|
1397 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6)
|
|
1398 + (p[2] & 0x3f);
|
|
1399 if ((p[3] & 0xc0) == 0x80)
|
|
1400 {
|
|
1401 if (len == 4)
|
|
1402 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12)
|
|
1403 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f);
|
|
1404 if ((p[4] & 0xc0) == 0x80)
|
|
1405 {
|
|
1406 if (len == 5)
|
|
1407 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18)
|
|
1408 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6)
|
|
1409 + (p[4] & 0x3f);
|
|
1410 if ((p[5] & 0xc0) == 0x80 && len == 6)
|
|
1411 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24)
|
|
1412 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12)
|
|
1413 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f);
|
|
1414 }
|
|
1415 }
|
|
1416 }
|
|
1417 }
|
|
1418 /* Illegal value, just return the first byte */
|
|
1419 return p[0];
|
|
1420 }
|
|
1421
|
|
1422 /*
|
|
1423 * Get character at **pp and advance *pp to the next character.
|
|
1424 * Note: composing characters are skipped!
|
|
1425 */
|
|
1426 int
|
|
1427 mb_ptr2char_adv(pp)
|
|
1428 char_u **pp;
|
|
1429 {
|
|
1430 int c;
|
|
1431
|
|
1432 c = (*mb_ptr2char)(*pp);
|
474
|
1433 *pp += (*mb_ptr2len)(*pp);
|
|
1434 return c;
|
|
1435 }
|
|
1436
|
|
1437 /*
|
|
1438 * Get character at **pp and advance *pp to the next character.
|
|
1439 * Note: composing characters are returned as separate characters.
|
|
1440 */
|
|
1441 int
|
|
1442 mb_cptr2char_adv(pp)
|
|
1443 char_u **pp;
|
|
1444 {
|
|
1445 int c;
|
|
1446
|
|
1447 c = (*mb_ptr2char)(*pp);
|
|
1448 if (enc_utf8)
|
|
1449 *pp += utf_ptr2len(*pp);
|
|
1450 else
|
|
1451 *pp += (*mb_ptr2len)(*pp);
|
7
|
1452 return c;
|
|
1453 }
|
|
1454
|
|
1455 #if defined(FEAT_ARABIC) || defined(PROTO)
|
|
1456 /*
|
|
1457 * Check whether we are dealing with Arabic combining characters.
|
|
1458 * Note: these are NOT really composing characters!
|
|
1459 */
|
|
1460 int
|
|
1461 arabic_combine(one, two)
|
|
1462 int one; /* first character */
|
|
1463 int two; /* character just after "one" */
|
|
1464 {
|
|
1465 if (one == a_LAM)
|
|
1466 return arabic_maycombine(two);
|
|
1467 return FALSE;
|
|
1468 }
|
|
1469
|
|
1470 /*
|
|
1471 * Check whether we are dealing with a character that could be regarded as an
|
|
1472 * Arabic combining character, need to check the character before this.
|
|
1473 */
|
|
1474 int
|
|
1475 arabic_maycombine(two)
|
|
1476 int two;
|
|
1477 {
|
|
1478 if (p_arshape && !p_tbidi)
|
|
1479 return (two == a_ALEF_MADDA
|
|
1480 || two == a_ALEF_HAMZA_ABOVE
|
|
1481 || two == a_ALEF_HAMZA_BELOW
|
|
1482 || two == a_ALEF);
|
|
1483 return FALSE;
|
|
1484 }
|
|
1485
|
|
1486 /*
|
|
1487 * Check if the character pointed to by "p2" is a composing character when it
|
|
1488 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which
|
|
1489 * behaves like a composing character.
|
|
1490 */
|
|
1491 int
|
|
1492 utf_composinglike(p1, p2)
|
|
1493 char_u *p1;
|
|
1494 char_u *p2;
|
|
1495 {
|
|
1496 int c2;
|
|
1497
|
|
1498 c2 = utf_ptr2char(p2);
|
|
1499 if (utf_iscomposing(c2))
|
|
1500 return TRUE;
|
|
1501 if (!arabic_maycombine(c2))
|
|
1502 return FALSE;
|
|
1503 return arabic_combine(utf_ptr2char(p1), c2);
|
|
1504 }
|
|
1505 #endif
|
|
1506
|
|
1507 /*
|
1205
|
1508 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO
|
7
|
1509 * composing characters.
|
|
1510 */
|
|
1511 int
|
714
|
1512 utfc_ptr2char(p, pcc)
|
7
|
1513 char_u *p;
|
714
|
1514 int *pcc; /* return: composing chars, last one is 0 */
|
7
|
1515 {
|
|
1516 int len;
|
|
1517 int c;
|
|
1518 int cc;
|
714
|
1519 int i = 0;
|
7
|
1520
|
|
1521 c = utf_ptr2char(p);
|
474
|
1522 len = utf_ptr2len(p);
|
714
|
1523
|
7
|
1524 /* Only accept a composing char when the first char isn't illegal. */
|
|
1525 if ((len > 1 || *p < 0x80)
|
|
1526 && p[len] >= 0x80
|
|
1527 && UTF_COMPOSINGLIKE(p, p + len))
|
|
1528 {
|
714
|
1529 cc = utf_ptr2char(p + len);
|
|
1530 for (;;)
|
|
1531 {
|
|
1532 pcc[i++] = cc;
|
|
1533 if (i == MAX_MCO)
|
|
1534 break;
|
|
1535 len += utf_ptr2len(p + len);
|
|
1536 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len)))
|
|
1537 break;
|
|
1538 }
|
7
|
1539 }
|
714
|
1540
|
|
1541 if (i < MAX_MCO) /* last composing char must be 0 */
|
|
1542 pcc[i] = 0;
|
|
1543
|
7
|
1544 return c;
|
|
1545 }
|
|
1546
|
|
1547 /*
|
1205
|
1548 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO
|
7
|
1549 * composing characters. Use no more than p[maxlen].
|
|
1550 */
|
|
1551 int
|
714
|
1552 utfc_ptr2char_len(p, pcc, maxlen)
|
7
|
1553 char_u *p;
|
714
|
1554 int *pcc; /* return: composing chars, last one is 0 */
|
7
|
1555 int maxlen;
|
|
1556 {
|
|
1557 int len;
|
|
1558 int c;
|
|
1559 int cc;
|
714
|
1560 int i = 0;
|
7
|
1561
|
|
1562 c = utf_ptr2char(p);
|
474
|
1563 len = utf_ptr2len_len(p, maxlen);
|
7
|
1564 /* Only accept a composing char when the first char isn't illegal. */
|
|
1565 if ((len > 1 || *p < 0x80)
|
|
1566 && len < maxlen
|
|
1567 && p[len] >= 0x80
|
|
1568 && UTF_COMPOSINGLIKE(p, p + len))
|
|
1569 {
|
714
|
1570 cc = utf_ptr2char(p + len);
|
|
1571 for (;;)
|
|
1572 {
|
|
1573 pcc[i++] = cc;
|
|
1574 if (i == MAX_MCO)
|
|
1575 break;
|
|
1576 len += utf_ptr2len_len(p + len, maxlen - len);
|
|
1577 if (len >= maxlen
|
|
1578 || p[len] < 0x80
|
|
1579 || !utf_iscomposing(cc = utf_ptr2char(p + len)))
|
|
1580 break;
|
|
1581 }
|
7
|
1582 }
|
714
|
1583
|
|
1584 if (i < MAX_MCO) /* last composing char must be 0 */
|
|
1585 pcc[i] = 0;
|
|
1586
|
7
|
1587 return c;
|
|
1588 }
|
|
1589
|
|
1590 /*
|
|
1591 * Convert the character at screen position "off" to a sequence of bytes.
|
|
1592 * Includes the composing characters.
|
|
1593 * "buf" must at least have the length MB_MAXBYTES.
|
|
1594 * Returns the produced number of bytes.
|
|
1595 */
|
|
1596 int
|
|
1597 utfc_char2bytes(off, buf)
|
|
1598 int off;
|
|
1599 char_u *buf;
|
|
1600 {
|
|
1601 int len;
|
714
|
1602 int i;
|
7
|
1603
|
|
1604 len = utf_char2bytes(ScreenLinesUC[off], buf);
|
714
|
1605 for (i = 0; i < Screen_mco; ++i)
|
7
|
1606 {
|
714
|
1607 if (ScreenLinesC[i][off] == 0)
|
|
1608 break;
|
|
1609 len += utf_char2bytes(ScreenLinesC[i][off], buf + len);
|
7
|
1610 }
|
|
1611 return len;
|
|
1612 }
|
|
1613
|
|
1614 /*
|
|
1615 * Get the length of a UTF-8 byte sequence, not including any following
|
|
1616 * composing characters.
|
|
1617 * Returns 0 for "".
|
|
1618 * Returns 1 for an illegal byte sequence.
|
|
1619 */
|
|
1620 int
|
474
|
1621 utf_ptr2len(p)
|
7
|
1622 char_u *p;
|
|
1623 {
|
|
1624 int len;
|
|
1625 int i;
|
|
1626
|
|
1627 if (*p == NUL)
|
|
1628 return 0;
|
|
1629 len = utf8len_tab[*p];
|
|
1630 for (i = 1; i < len; ++i)
|
|
1631 if ((p[i] & 0xc0) != 0x80)
|
|
1632 return 1;
|
|
1633 return len;
|
|
1634 }
|
|
1635
|
|
1636 /*
|
|
1637 * Return length of UTF-8 character, obtained from the first byte.
|
|
1638 * "b" must be between 0 and 255!
|
|
1639 */
|
|
1640 int
|
|
1641 utf_byte2len(b)
|
|
1642 int b;
|
|
1643 {
|
|
1644 return utf8len_tab[b];
|
|
1645 }
|
|
1646
|
|
1647 /*
|
|
1648 * Get the length of UTF-8 byte sequence "p[size]". Does not include any
|
|
1649 * following composing characters.
|
|
1650 * Returns 1 for "".
|
1487
|
1651 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.).
|
7
|
1652 * Returns number > "size" for an incomplete byte sequence.
|
|
1653 */
|
|
1654 int
|
474
|
1655 utf_ptr2len_len(p, size)
|
7
|
1656 char_u *p;
|
|
1657 int size;
|
|
1658 {
|
|
1659 int len;
|
|
1660 int i;
|
1487
|
1661 int m;
|
7
|
1662
|
|
1663 if (*p == NUL)
|
|
1664 return 1;
|
1487
|
1665 m = len = utf8len_tab[*p];
|
7
|
1666 if (len > size)
|
1487
|
1667 m = size; /* incomplete byte sequence. */
|
|
1668 for (i = 1; i < m; ++i)
|
7
|
1669 if ((p[i] & 0xc0) != 0x80)
|
|
1670 return 1;
|
|
1671 return len;
|
|
1672 }
|
|
1673
|
|
1674 /*
|
|
1675 * Return the number of bytes the UTF-8 encoding of the character at "p" takes.
|
|
1676 * This includes following composing characters.
|
|
1677 */
|
|
1678 int
|
474
|
1679 utfc_ptr2len(p)
|
7
|
1680 char_u *p;
|
|
1681 {
|
|
1682 int len;
|
318
|
1683 int b0 = *p;
|
7
|
1684 #ifdef FEAT_ARABIC
|
|
1685 int prevlen;
|
|
1686 #endif
|
|
1687
|
318
|
1688 if (b0 == NUL)
|
7
|
1689 return 0;
|
318
|
1690 if (b0 < 0x80 && p[1] < 0x80) /* be quick for ASCII */
|
7
|
1691 return 1;
|
|
1692
|
|
1693 /* Skip over first UTF-8 char, stopping at a NUL byte. */
|
474
|
1694 len = utf_ptr2len(p);
|
7
|
1695
|
|
1696 /* Check for illegal byte. */
|
318
|
1697 if (len == 1 && b0 >= 0x80)
|
7
|
1698 return 1;
|
|
1699
|
|
1700 /*
|
|
1701 * Check for composing characters. We can handle only the first two, but
|
|
1702 * skip all of them (otherwise the cursor would get stuck).
|
|
1703 */
|
|
1704 #ifdef FEAT_ARABIC
|
|
1705 prevlen = 0;
|
|
1706 #endif
|
|
1707 for (;;)
|
|
1708 {
|
|
1709 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len))
|
|
1710 return len;
|
|
1711
|
|
1712 /* Skip over composing char */
|
|
1713 #ifdef FEAT_ARABIC
|
|
1714 prevlen = len;
|
|
1715 #endif
|
474
|
1716 len += utf_ptr2len(p + len);
|
7
|
1717 }
|
|
1718 }
|
|
1719
|
|
1720 /*
|
|
1721 * Return the number of bytes the UTF-8 encoding of the character at "p[size]"
|
|
1722 * takes. This includes following composing characters.
|
|
1723 * Returns 1 for an illegal char or an incomplete byte sequence.
|
|
1724 */
|
|
1725 int
|
474
|
1726 utfc_ptr2len_len(p, size)
|
7
|
1727 char_u *p;
|
|
1728 int size;
|
|
1729 {
|
|
1730 int len;
|
|
1731 #ifdef FEAT_ARABIC
|
|
1732 int prevlen;
|
|
1733 #endif
|
|
1734
|
|
1735 if (*p == NUL)
|
|
1736 return 0;
|
|
1737 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) /* be quick for ASCII */
|
|
1738 return 1;
|
|
1739
|
|
1740 /* Skip over first UTF-8 char, stopping at a NUL byte. */
|
474
|
1741 len = utf_ptr2len_len(p, size);
|
7
|
1742
|
|
1743 /* Check for illegal byte and incomplete byte sequence. */
|
|
1744 if ((len == 1 && p[0] >= 0x80) || len > size)
|
|
1745 return 1;
|
|
1746
|
|
1747 /*
|
|
1748 * Check for composing characters. We can handle only the first two, but
|
|
1749 * skip all of them (otherwise the cursor would get stuck).
|
|
1750 */
|
|
1751 #ifdef FEAT_ARABIC
|
|
1752 prevlen = 0;
|
|
1753 #endif
|
|
1754 while (len < size)
|
|
1755 {
|
1658
|
1756 int len_next_char;
|
|
1757
|
|
1758 if (p[len] < 0x80)
|
|
1759 break;
|
|
1760
|
|
1761 /*
|
|
1762 * Next character length should not go beyond size to ensure that
|
|
1763 * UTF_COMPOSINGLIKE(...) does not read beyond size.
|
|
1764 */
|
|
1765 len_next_char = utf_ptr2len_len(p + len, size - len);
|
|
1766 if (len_next_char > size - len)
|
|
1767 break;
|
|
1768
|
|
1769 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len))
|
7
|
1770 break;
|
|
1771
|
|
1772 /* Skip over composing char */
|
|
1773 #ifdef FEAT_ARABIC
|
|
1774 prevlen = len;
|
|
1775 #endif
|
1658
|
1776 len += len_next_char;
|
7
|
1777 }
|
|
1778 return len;
|
|
1779 }
|
|
1780
|
|
1781 /*
|
|
1782 * Return the number of bytes the UTF-8 encoding of character "c" takes.
|
|
1783 * This does not include composing characters.
|
|
1784 */
|
|
1785 int
|
|
1786 utf_char2len(c)
|
|
1787 int c;
|
|
1788 {
|
|
1789 if (c < 0x80)
|
|
1790 return 1;
|
|
1791 if (c < 0x800)
|
|
1792 return 2;
|
|
1793 if (c < 0x10000)
|
|
1794 return 3;
|
|
1795 if (c < 0x200000)
|
|
1796 return 4;
|
|
1797 if (c < 0x4000000)
|
|
1798 return 5;
|
|
1799 return 6;
|
|
1800 }
|
|
1801
|
|
1802 /*
|
|
1803 * Convert Unicode character "c" to UTF-8 string in "buf[]".
|
|
1804 * Returns the number of bytes.
|
|
1805 * This does not include composing characters.
|
|
1806 */
|
|
1807 int
|
|
1808 utf_char2bytes(c, buf)
|
|
1809 int c;
|
|
1810 char_u *buf;
|
|
1811 {
|
|
1812 if (c < 0x80) /* 7 bits */
|
|
1813 {
|
|
1814 buf[0] = c;
|
|
1815 return 1;
|
|
1816 }
|
|
1817 if (c < 0x800) /* 11 bits */
|
|
1818 {
|
|
1819 buf[0] = 0xc0 + ((unsigned)c >> 6);
|
|
1820 buf[1] = 0x80 + (c & 0x3f);
|
|
1821 return 2;
|
|
1822 }
|
|
1823 if (c < 0x10000) /* 16 bits */
|
|
1824 {
|
|
1825 buf[0] = 0xe0 + ((unsigned)c >> 12);
|
|
1826 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f);
|
|
1827 buf[2] = 0x80 + (c & 0x3f);
|
|
1828 return 3;
|
|
1829 }
|
|
1830 if (c < 0x200000) /* 21 bits */
|
|
1831 {
|
|
1832 buf[0] = 0xf0 + ((unsigned)c >> 18);
|
|
1833 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f);
|
|
1834 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f);
|
|
1835 buf[3] = 0x80 + (c & 0x3f);
|
|
1836 return 4;
|
|
1837 }
|
|
1838 if (c < 0x4000000) /* 26 bits */
|
|
1839 {
|
|
1840 buf[0] = 0xf8 + ((unsigned)c >> 24);
|
|
1841 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f);
|
|
1842 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f);
|
|
1843 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f);
|
|
1844 buf[4] = 0x80 + (c & 0x3f);
|
|
1845 return 5;
|
|
1846 }
|
|
1847 /* 31 bits */
|
|
1848 buf[0] = 0xfc + ((unsigned)c >> 30);
|
|
1849 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f);
|
|
1850 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f);
|
|
1851 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f);
|
|
1852 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f);
|
|
1853 buf[5] = 0x80 + (c & 0x3f);
|
|
1854 return 6;
|
|
1855 }
|
|
1856
|
|
1857 /*
|
|
1858 * Return TRUE if "c" is a composing UTF-8 character. This means it will be
|
|
1859 * drawn on top of the preceding character.
|
|
1860 * Based on code from Markus Kuhn.
|
|
1861 */
|
|
1862 int
|
|
1863 utf_iscomposing(c)
|
|
1864 int c;
|
|
1865 {
|
|
1866 /* sorted list of non-overlapping intervals */
|
|
1867 static struct interval combining[] =
|
|
1868 {
|
|
1869 {0x0300, 0x034f}, {0x0360, 0x036f}, {0x0483, 0x0486}, {0x0488, 0x0489},
|
|
1870 {0x0591, 0x05a1}, {0x05a3, 0x05b9}, {0x05bb, 0x05bd}, {0x05bf, 0x05bf},
|
|
1871 {0x05c1, 0x05c2}, {0x05c4, 0x05c4}, {0x0610, 0x0615}, {0x064b, 0x0658},
|
|
1872 {0x0670, 0x0670}, {0x06d6, 0x06dc}, {0x06de, 0x06e4}, {0x06e7, 0x06e8},
|
|
1873 {0x06ea, 0x06ed}, {0x0711, 0x0711}, {0x0730, 0x074a}, {0x07a6, 0x07b0},
|
|
1874 {0x0901, 0x0903}, {0x093c, 0x093c}, {0x093e, 0x094d}, {0x0951, 0x0954},
|
|
1875 {0x0962, 0x0963}, {0x0981, 0x0983}, {0x09bc, 0x09bc}, {0x09be, 0x09c4},
|
|
1876 {0x09c7, 0x09c8}, {0x09cb, 0x09cd}, {0x09d7, 0x09d7}, {0x09e2, 0x09e3},
|
|
1877 {0x0a01, 0x0a03}, {0x0a3c, 0x0a3c}, {0x0a3e, 0x0a42}, {0x0a47, 0x0a48},
|
|
1878 {0x0a4b, 0x0a4d}, {0x0a70, 0x0a71}, {0x0a81, 0x0a83}, {0x0abc, 0x0abc},
|
|
1879 {0x0abe, 0x0ac5}, {0x0ac7, 0x0ac9}, {0x0acb, 0x0acd}, {0x0ae2, 0x0ae3},
|
|
1880 {0x0b01, 0x0b03}, {0x0b3c, 0x0b3c}, {0x0b3e, 0x0b43}, {0x0b47, 0x0b48},
|
|
1881 {0x0b4b, 0x0b4d}, {0x0b56, 0x0b57}, {0x0b82, 0x0b82}, {0x0bbe, 0x0bc2},
|
|
1882 {0x0bc6, 0x0bc8}, {0x0bca, 0x0bcd}, {0x0bd7, 0x0bd7}, {0x0c01, 0x0c03},
|
|
1883 {0x0c3e, 0x0c44}, {0x0c46, 0x0c48}, {0x0c4a, 0x0c4d}, {0x0c55, 0x0c56},
|
|
1884 {0x0c82, 0x0c83}, {0x0cbc, 0x0cbc}, {0x0cbe, 0x0cc4}, {0x0cc6, 0x0cc8},
|
|
1885 {0x0cca, 0x0ccd}, {0x0cd5, 0x0cd6}, {0x0d02, 0x0d03}, {0x0d3e, 0x0d43},
|
|
1886 {0x0d46, 0x0d48}, {0x0d4a, 0x0d4d}, {0x0d57, 0x0d57}, {0x0d82, 0x0d83},
|
|
1887 {0x0dca, 0x0dca}, {0x0dcf, 0x0dd4}, {0x0dd6, 0x0dd6}, {0x0dd8, 0x0ddf},
|
|
1888 {0x0df2, 0x0df3}, {0x0e31, 0x0e31}, {0x0e34, 0x0e3a}, {0x0e47, 0x0e4e},
|
|
1889 {0x0eb1, 0x0eb1}, {0x0eb4, 0x0eb9}, {0x0ebb, 0x0ebc}, {0x0ec8, 0x0ecd},
|
|
1890 {0x0f18, 0x0f19}, {0x0f35, 0x0f35}, {0x0f37, 0x0f37}, {0x0f39, 0x0f39},
|
|
1891 {0x0f3e, 0x0f3f}, {0x0f71, 0x0f84}, {0x0f86, 0x0f87}, {0x0f90, 0x0f97},
|
|
1892 {0x0f99, 0x0fbc}, {0x0fc6, 0x0fc6}, {0x102c, 0x1032}, {0x1036, 0x1039},
|
|
1893 {0x1056, 0x1059}, {0x1712, 0x1714}, {0x1732, 0x1734}, {0x1752, 0x1753},
|
|
1894 {0x1772, 0x1773}, {0x17b6, 0x17d3}, {0x17dd, 0x17dd}, {0x180b, 0x180d},
|
|
1895 {0x18a9, 0x18a9}, {0x1920, 0x192b}, {0x1930, 0x193b}, {0x20d0, 0x20ea},
|
|
1896 {0x302a, 0x302f}, {0x3099, 0x309a}, {0xfb1e, 0xfb1e}, {0xfe00, 0xfe0f},
|
|
1897 {0xfe20, 0xfe23},
|
|
1898 };
|
|
1899
|
|
1900 return intable(combining, sizeof(combining), c);
|
|
1901 }
|
|
1902
|
|
1903 /*
|
|
1904 * Return TRUE for characters that can be displayed in a normal way.
|
|
1905 * Only for characters of 0x100 and above!
|
|
1906 */
|
|
1907 int
|
|
1908 utf_printable(c)
|
|
1909 int c;
|
|
1910 {
|
|
1911 #ifdef USE_WCHAR_FUNCTIONS
|
|
1912 /*
|
|
1913 * Assume the iswprint() library function works better than our own stuff.
|
|
1914 */
|
|
1915 return iswprint(c);
|
|
1916 #else
|
|
1917 /* Sorted list of non-overlapping intervals.
|
|
1918 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */
|
|
1919 static struct interval nonprint[] =
|
|
1920 {
|
|
1921 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e},
|
|
1922 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb},
|
|
1923 {0xfffe, 0xffff}
|
|
1924 };
|
|
1925
|
|
1926 return !intable(nonprint, sizeof(nonprint), c);
|
|
1927 #endif
|
|
1928 }
|
|
1929
|
|
1930 /*
|
|
1931 * Get class of a Unicode character.
|
|
1932 * 0: white space
|
|
1933 * 1: punctuation
|
|
1934 * 2 or bigger: some class of word character.
|
|
1935 */
|
|
1936 int
|
|
1937 utf_class(c)
|
|
1938 int c;
|
|
1939 {
|
|
1940 /* sorted list of non-overlapping intervals */
|
|
1941 static struct clinterval
|
|
1942 {
|
|
1943 unsigned short first;
|
|
1944 unsigned short last;
|
|
1945 unsigned short class;
|
|
1946 } classes[] =
|
|
1947 {
|
|
1948 {0x037e, 0x037e, 1}, /* Greek question mark */
|
|
1949 {0x0387, 0x0387, 1}, /* Greek ano teleia */
|
|
1950 {0x055a, 0x055f, 1}, /* Armenian punctuation */
|
|
1951 {0x0589, 0x0589, 1}, /* Armenian full stop */
|
|
1952 {0x05be, 0x05be, 1},
|
|
1953 {0x05c0, 0x05c0, 1},
|
|
1954 {0x05c3, 0x05c3, 1},
|
|
1955 {0x05f3, 0x05f4, 1},
|
|
1956 {0x060c, 0x060c, 1},
|
|
1957 {0x061b, 0x061b, 1},
|
|
1958 {0x061f, 0x061f, 1},
|
|
1959 {0x066a, 0x066d, 1},
|
|
1960 {0x06d4, 0x06d4, 1},
|
|
1961 {0x0700, 0x070d, 1}, /* Syriac punctuation */
|
|
1962 {0x0964, 0x0965, 1},
|
|
1963 {0x0970, 0x0970, 1},
|
|
1964 {0x0df4, 0x0df4, 1},
|
|
1965 {0x0e4f, 0x0e4f, 1},
|
|
1966 {0x0e5a, 0x0e5b, 1},
|
|
1967 {0x0f04, 0x0f12, 1},
|
|
1968 {0x0f3a, 0x0f3d, 1},
|
|
1969 {0x0f85, 0x0f85, 1},
|
|
1970 {0x104a, 0x104f, 1}, /* Myanmar punctuation */
|
|
1971 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */
|
|
1972 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */
|
|
1973 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */
|
|
1974 {0x1680, 0x1680, 0},
|
|
1975 {0x169b, 0x169c, 1},
|
|
1976 {0x16eb, 0x16ed, 1},
|
|
1977 {0x1735, 0x1736, 1},
|
|
1978 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */
|
|
1979 {0x1800, 0x180a, 1}, /* Mongolian punctuation */
|
|
1980 {0x2000, 0x200b, 0}, /* spaces */
|
|
1981 {0x200c, 0x2027, 1}, /* punctuation and symbols */
|
|
1982 {0x2028, 0x2029, 0},
|
|
1983 {0x202a, 0x202e, 1}, /* punctuation and symbols */
|
|
1984 {0x202f, 0x202f, 0},
|
|
1985 {0x2030, 0x205e, 1}, /* punctuation and symbols */
|
|
1986 {0x205f, 0x205f, 0},
|
|
1987 {0x2060, 0x27ff, 1}, /* punctuation and symbols */
|
|
1988 {0x2070, 0x207f, 0x2070}, /* superscript */
|
1593
|
1989 {0x2080, 0x2094, 0x2080}, /* subscript */
|
|
1990 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */
|
|
1991 {0x2800, 0x28ff, 0x2800}, /* braille */
|
|
1992 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */
|
7
|
1993 {0x29d8, 0x29db, 1},
|
|
1994 {0x29fc, 0x29fd, 1},
|
|
1995 {0x3000, 0x3000, 0}, /* ideographic space */
|
|
1996 {0x3001, 0x3020, 1}, /* ideographic punctuation */
|
|
1997 {0x3030, 0x3030, 1},
|
|
1998 {0x303d, 0x303d, 1},
|
|
1999 {0x3040, 0x309f, 0x3040}, /* Hiragana */
|
|
2000 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */
|
|
2001 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */
|
|
2002 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */
|
|
2003 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */
|
|
2004 {0xfd3e, 0xfd3f, 1},
|
|
2005 {0xfe30, 0xfe6b, 1}, /* punctuation forms */
|
|
2006 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */
|
|
2007 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */
|
|
2008 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */
|
|
2009 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */
|
|
2010 };
|
|
2011 int bot = 0;
|
|
2012 int top = sizeof(classes) / sizeof(struct clinterval) - 1;
|
|
2013 int mid;
|
|
2014
|
|
2015 /* First quick check for Latin1 characters, use 'iskeyword'. */
|
|
2016 if (c < 0x100)
|
|
2017 {
|
335
|
2018 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0)
|
7
|
2019 return 0; /* blank */
|
|
2020 if (vim_iswordc(c))
|
|
2021 return 2; /* word character */
|
|
2022 return 1; /* punctuation */
|
|
2023 }
|
|
2024
|
|
2025 /* binary search in table */
|
|
2026 while (top >= bot)
|
|
2027 {
|
|
2028 mid = (bot + top) / 2;
|
|
2029 if (classes[mid].last < c)
|
|
2030 bot = mid + 1;
|
|
2031 else if (classes[mid].first > c)
|
|
2032 top = mid - 1;
|
|
2033 else
|
|
2034 return (int)classes[mid].class;
|
|
2035 }
|
|
2036
|
|
2037 /* most other characters are "word" characters */
|
|
2038 return 2;
|
|
2039 }
|
|
2040
|
|
2041 /*
|
|
2042 * Code for Unicode case-dependent operations. Based on notes in
|
|
2043 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt
|
|
2044 * This code uses simple case folding, not full case folding.
|
|
2045 */
|
|
2046
|
|
2047 /*
|
|
2048 * The following table is built by foldExtract.pl < CaseFolding.txt .
|
|
2049 * It must be in numeric order, because we use binary search on it.
|
|
2050 * An entry such as {0x41,0x5a,1,32} means that UCS-4 characters in the range
|
|
2051 * from 0x41 to 0x5a inclusive, stepping by 1, are folded by adding 32.
|
|
2052 */
|
|
2053
|
|
2054 typedef struct
|
|
2055 {
|
|
2056 int rangeStart;
|
|
2057 int rangeEnd;
|
|
2058 int step;
|
|
2059 int offset;
|
|
2060 } convertStruct;
|
|
2061
|
297
|
2062 static convertStruct foldCase[] =
|
7
|
2063 {
|
|
2064 {0x41,0x5a,1,32}, {0xc0,0xd6,1,32}, {0xd8,0xde,1,32},
|
|
2065 {0x100,0x12e,2,1}, {0x130,0x130,-1,-199}, {0x132,0x136,2,1},
|
|
2066 {0x139,0x147,2,1}, {0x14a,0x176,2,1}, {0x178,0x178,-1,-121},
|
|
2067 {0x179,0x17d,2,1}, {0x181,0x181,-1,210}, {0x182,0x184,2,1},
|
|
2068 {0x186,0x186,-1,206}, {0x187,0x187,-1,1}, {0x189,0x18a,1,205},
|
|
2069 {0x18b,0x18b,-1,1}, {0x18e,0x18e,-1,79}, {0x18f,0x18f,-1,202},
|
|
2070 {0x190,0x190,-1,203}, {0x191,0x191,-1,1}, {0x193,0x193,-1,205},
|
|
2071 {0x194,0x194,-1,207}, {0x196,0x196,-1,211}, {0x197,0x197,-1,209},
|
|
2072 {0x198,0x198,-1,1}, {0x19c,0x19c,-1,211}, {0x19d,0x19d,-1,213},
|
|
2073 {0x19f,0x19f,-1,214}, {0x1a0,0x1a4,2,1}, {0x1a6,0x1a6,-1,218},
|
|
2074 {0x1a7,0x1a7,-1,1}, {0x1a9,0x1a9,-1,218}, {0x1ac,0x1ac,-1,1},
|
|
2075 {0x1ae,0x1ae,-1,218}, {0x1af,0x1af,-1,1}, {0x1b1,0x1b2,1,217},
|
|
2076 {0x1b3,0x1b5,2,1}, {0x1b7,0x1b7,-1,219}, {0x1b8,0x1bc,4,1},
|
|
2077 {0x1c4,0x1c4,-1,2}, {0x1c5,0x1c5,-1,1}, {0x1c7,0x1c7,-1,2},
|
|
2078 {0x1c8,0x1c8,-1,1}, {0x1ca,0x1ca,-1,2}, {0x1cb,0x1db,2,1},
|
|
2079 {0x1de,0x1ee,2,1}, {0x1f1,0x1f1,-1,2}, {0x1f2,0x1f4,2,1},
|
|
2080 {0x1f6,0x1f6,-1,-97}, {0x1f7,0x1f7,-1,-56}, {0x1f8,0x21e,2,1},
|
|
2081 {0x220,0x220,-1,-130}, {0x222,0x232,2,1}, {0x386,0x386,-1,38},
|
|
2082 {0x388,0x38a,1,37}, {0x38c,0x38c,-1,64}, {0x38e,0x38f,1,63},
|
|
2083 {0x391,0x3a1,1,32}, {0x3a3,0x3ab,1,32}, {0x3d8,0x3ee,2,1},
|
|
2084 {0x3f4,0x3f4,-1,-60}, {0x3f7,0x3f7,-1,1}, {0x3f9,0x3f9,-1,-7},
|
|
2085 {0x3fa,0x3fa,-1,1}, {0x400,0x40f,1,80}, {0x410,0x42f,1,32},
|
|
2086 {0x460,0x480,2,1}, {0x48a,0x4be,2,1}, {0x4c1,0x4cd,2,1},
|
|
2087 {0x4d0,0x4f4,2,1}, {0x4f8,0x500,8,1}, {0x502,0x50e,2,1},
|
|
2088 {0x531,0x556,1,48}, {0x1e00,0x1e94,2,1}, {0x1ea0,0x1ef8,2,1},
|
|
2089 {0x1f08,0x1f0f,1,-8}, {0x1f18,0x1f1d,1,-8}, {0x1f28,0x1f2f,1,-8},
|
|
2090 {0x1f38,0x1f3f,1,-8}, {0x1f48,0x1f4d,1,-8}, {0x1f59,0x1f5f,2,-8},
|
|
2091 {0x1f68,0x1f6f,1,-8}, {0x1f88,0x1f8f,1,-8}, {0x1f98,0x1f9f,1,-8},
|
|
2092 {0x1fa8,0x1faf,1,-8}, {0x1fb8,0x1fb9,1,-8}, {0x1fba,0x1fbb,1,-74},
|
|
2093 {0x1fbc,0x1fbc,-1,-9}, {0x1fc8,0x1fcb,1,-86}, {0x1fcc,0x1fcc,-1,-9},
|
|
2094 {0x1fd8,0x1fd9,1,-8}, {0x1fda,0x1fdb,1,-100}, {0x1fe8,0x1fe9,1,-8},
|
|
2095 {0x1fea,0x1feb,1,-112}, {0x1fec,0x1fec,-1,-7}, {0x1ff8,0x1ff9,1,-128},
|
|
2096 {0x1ffa,0x1ffb,1,-126}, {0x1ffc,0x1ffc,-1,-9}, {0x2126,0x2126,-1,-7517},
|
|
2097 {0x212a,0x212a,-1,-8383}, {0x212b,0x212b,-1,-8262},
|
|
2098 {0x2160,0x216f,1,16}, {0x24b6,0x24cf,1,26}, {0xff21,0xff3a,1,32},
|
|
2099 {0x10400,0x10427,1,40}
|
|
2100 };
|
|
2101
|
|
2102 static int utf_convert(int a, convertStruct table[], int tableSize);
|
|
2103
|
|
2104 /*
|
|
2105 * Generic conversion function for case operations.
|
|
2106 * Return the converted equivalent of "a", which is a UCS-4 character. Use
|
|
2107 * the given conversion "table". Uses binary search on "table".
|
|
2108 */
|
|
2109 static int
|
|
2110 utf_convert(a, table, tableSize)
|
|
2111 int a;
|
|
2112 convertStruct table[];
|
|
2113 int tableSize;
|
|
2114 {
|
|
2115 int start, mid, end; /* indices into table */
|
|
2116
|
|
2117 start = 0;
|
|
2118 end = tableSize / sizeof(convertStruct);
|
|
2119 while (start < end)
|
|
2120 {
|
|
2121 /* need to search further */
|
|
2122 mid = (end + start) /2;
|
|
2123 if (table[mid].rangeEnd < a)
|
|
2124 start = mid + 1;
|
|
2125 else
|
|
2126 end = mid;
|
|
2127 }
|
|
2128 if (table[start].rangeStart <= a && a <= table[start].rangeEnd
|
|
2129 && (a - table[start].rangeStart) % table[start].step == 0)
|
|
2130 return (a + table[start].offset);
|
|
2131 else
|
|
2132 return a;
|
|
2133 }
|
|
2134
|
|
2135 /*
|
|
2136 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses
|
|
2137 * simple case folding.
|
|
2138 */
|
|
2139 int
|
|
2140 utf_fold(a)
|
|
2141 int a;
|
|
2142 {
|
|
2143 return utf_convert(a, foldCase, sizeof(foldCase));
|
|
2144 }
|
|
2145
|
|
2146 /*
|
|
2147 * The following tables are built by upperLowerExtract.pl < UnicodeData.txt .
|
|
2148 * They must be in numeric order, because we use binary search on them.
|
|
2149 * An entry such as {0x41,0x5a,1,32} means that UCS-4 characters in the range
|
|
2150 * from 0x41 to 0x5a inclusive, stepping by 1, are switched to lower (for
|
|
2151 * example) by adding 32.
|
|
2152 */
|
297
|
2153 static convertStruct toLower[] =
|
7
|
2154 {
|
|
2155 {0x41,0x5a,1,32}, {0xc0,0xd6,1,32}, {0xd8,0xde,1,32},
|
|
2156 {0x100,0x12e,2,1}, {0x130,0x130,-1,-199}, {0x132,0x136,2,1},
|
|
2157 {0x139,0x147,2,1}, {0x14a,0x176,2,1}, {0x178,0x178,-1,-121},
|
|
2158 {0x179,0x17d,2,1}, {0x181,0x181,-1,210}, {0x182,0x184,2,1},
|
|
2159 {0x186,0x186,-1,206}, {0x187,0x187,-1,1}, {0x189,0x18a,1,205},
|
|
2160 {0x18b,0x18b,-1,1}, {0x18e,0x18e,-1,79}, {0x18f,0x18f,-1,202},
|
|
2161 {0x190,0x190,-1,203}, {0x191,0x191,-1,1}, {0x193,0x193,-1,205},
|
|
2162 {0x194,0x194,-1,207}, {0x196,0x196,-1,211}, {0x197,0x197,-1,209},
|
|
2163 {0x198,0x198,-1,1}, {0x19c,0x19c,-1,211}, {0x19d,0x19d,-1,213},
|
|
2164 {0x19f,0x19f,-1,214}, {0x1a0,0x1a4,2,1}, {0x1a6,0x1a6,-1,218},
|
|
2165 {0x1a7,0x1a7,-1,1}, {0x1a9,0x1a9,-1,218}, {0x1ac,0x1ac,-1,1},
|
|
2166 {0x1ae,0x1ae,-1,218}, {0x1af,0x1af,-1,1}, {0x1b1,0x1b2,1,217},
|
|
2167 {0x1b3,0x1b5,2,1}, {0x1b7,0x1b7,-1,219}, {0x1b8,0x1bc,4,1},
|
|
2168 {0x1c4,0x1ca,3,2}, {0x1cd,0x1db,2,1}, {0x1de,0x1ee,2,1},
|
|
2169 {0x1f1,0x1f1,-1,2}, {0x1f4,0x1f4,-1,1}, {0x1f6,0x1f6,-1,-97},
|
|
2170 {0x1f7,0x1f7,-1,-56}, {0x1f8,0x21e,2,1}, {0x220,0x220,-1,-130},
|
|
2171 {0x222,0x232,2,1}, {0x386,0x386,-1,38}, {0x388,0x38a,1,37},
|
|
2172 {0x38c,0x38c,-1,64}, {0x38e,0x38f,1,63}, {0x391,0x3a1,1,32},
|
|
2173 {0x3a3,0x3ab,1,32}, {0x3d8,0x3ee,2,1}, {0x3f4,0x3f4,-1,-60},
|
|
2174 {0x3f7,0x3f7,-1,1}, {0x3f9,0x3f9,-1,-7}, {0x3fa,0x3fa,-1,1},
|
|
2175 {0x400,0x40f,1,80}, {0x410,0x42f,1,32}, {0x460,0x480,2,1},
|
|
2176 {0x48a,0x4be,2,1}, {0x4c1,0x4cd,2,1}, {0x4d0,0x4f4,2,1},
|
|
2177 {0x4f8,0x500,8,1}, {0x502,0x50e,2,1}, {0x531,0x556,1,48},
|
|
2178 {0x1e00,0x1e94,2,1}, {0x1ea0,0x1ef8,2,1}, {0x1f08,0x1f0f,1,-8},
|
|
2179 {0x1f18,0x1f1d,1,-8}, {0x1f28,0x1f2f,1,-8}, {0x1f38,0x1f3f,1,-8},
|
|
2180 {0x1f48,0x1f4d,1,-8}, {0x1f59,0x1f5f,2,-8}, {0x1f68,0x1f6f,1,-8},
|
|
2181 {0x1fb8,0x1fb9,1,-8}, {0x1fba,0x1fbb,1,-74}, {0x1fc8,0x1fcb,1,-86},
|
|
2182 {0x1fd8,0x1fd9,1,-8}, {0x1fda,0x1fdb,1,-100}, {0x1fe8,0x1fe9,1,-8},
|
|
2183 {0x1fea,0x1feb,1,-112}, {0x1fec,0x1fec,-1,-7}, {0x1ff8,0x1ff9,1,-128},
|
|
2184 {0x1ffa,0x1ffb,1,-126}, {0x2126,0x2126,-1,-7517}, {0x212a,0x212a,-1,-8383},
|
|
2185 {0x212b,0x212b,-1,-8262}, {0xff21,0xff3a,1,32}, {0x10400,0x10427,1,40}
|
|
2186 };
|
|
2187
|
297
|
2188 static convertStruct toUpper[] =
|
7
|
2189 {
|
|
2190 {0x61,0x7a,1,-32}, {0xb5,0xb5,-1,743}, {0xe0,0xf6,1,-32},
|
|
2191 {0xf8,0xfe,1,-32}, {0xff,0xff,-1,121}, {0x101,0x12f,2,-1},
|
|
2192 {0x131,0x131,-1,-232}, {0x133,0x137,2,-1}, {0x13a,0x148,2,-1},
|
|
2193 {0x14b,0x177,2,-1}, {0x17a,0x17e,2,-1}, {0x17f,0x17f,-1,-300},
|
|
2194 {0x183,0x185,2,-1}, {0x188,0x18c,4,-1}, {0x192,0x192,-1,-1},
|
|
2195 {0x195,0x195,-1,97}, {0x199,0x199,-1,-1}, {0x19e,0x19e,-1,130},
|
|
2196 {0x1a1,0x1a5,2,-1}, {0x1a8,0x1ad,5,-1}, {0x1b0,0x1b4,4,-1},
|
|
2197 {0x1b6,0x1b9,3,-1}, {0x1bd,0x1bd,-1,-1}, {0x1bf,0x1bf,-1,56},
|
|
2198 {0x1c5,0x1c6,1,-1}, {0x1c8,0x1c9,1,-1}, {0x1cb,0x1cc,1,-1},
|
|
2199 {0x1ce,0x1dc,2,-1}, {0x1dd,0x1dd,-1,-79}, {0x1df,0x1ef,2,-1},
|
|
2200 {0x1f2,0x1f3,1,-1}, {0x1f5,0x1f9,4,-1}, {0x1fb,0x21f,2,-1},
|
|
2201 {0x223,0x233,2,-1}, {0x253,0x253,-1,-210}, {0x254,0x254,-1,-206},
|
|
2202 {0x256,0x257,1,-205}, {0x259,0x259,-1,-202}, {0x25b,0x25b,-1,-203},
|
|
2203 {0x260,0x260,-1,-205}, {0x263,0x263,-1,-207}, {0x268,0x268,-1,-209},
|
|
2204 {0x269,0x26f,6,-211}, {0x272,0x272,-1,-213}, {0x275,0x275,-1,-214},
|
|
2205 {0x280,0x283,3,-218}, {0x288,0x288,-1,-218}, {0x28a,0x28b,1,-217},
|
|
2206 {0x292,0x292,-1,-219}, {0x3ac,0x3ac,-1,-38}, {0x3ad,0x3af,1,-37},
|
|
2207 {0x3b1,0x3c1,1,-32}, {0x3c2,0x3c2,-1,-31}, {0x3c3,0x3cb,1,-32},
|
|
2208 {0x3cc,0x3cc,-1,-64}, {0x3cd,0x3ce,1,-63}, {0x3d0,0x3d0,-1,-62},
|
|
2209 {0x3d1,0x3d1,-1,-57}, {0x3d5,0x3d5,-1,-47}, {0x3d6,0x3d6,-1,-54},
|
|
2210 {0x3d9,0x3ef,2,-1}, {0x3f0,0x3f0,-1,-86}, {0x3f1,0x3f1,-1,-80},
|
|
2211 {0x3f2,0x3f2,-1,7}, {0x3f5,0x3f5,-1,-96}, {0x3f8,0x3fb,3,-1},
|
|
2212 {0x430,0x44f,1,-32}, {0x450,0x45f,1,-80}, {0x461,0x481,2,-1},
|
|
2213 {0x48b,0x4bf,2,-1}, {0x4c2,0x4ce,2,-1}, {0x4d1,0x4f5,2,-1},
|
|
2214 {0x4f9,0x501,8,-1}, {0x503,0x50f,2,-1}, {0x561,0x586,1,-48},
|
|
2215 {0x1e01,0x1e95,2,-1}, {0x1e9b,0x1e9b,-1,-59}, {0x1ea1,0x1ef9,2,-1},
|
|
2216 {0x1f00,0x1f07,1,8}, {0x1f10,0x1f15,1,8}, {0x1f20,0x1f27,1,8},
|
|
2217 {0x1f30,0x1f37,1,8}, {0x1f40,0x1f45,1,8}, {0x1f51,0x1f57,2,8},
|
|
2218 {0x1f60,0x1f67,1,8}, {0x1f70,0x1f71,1,74}, {0x1f72,0x1f75,1,86},
|
|
2219 {0x1f76,0x1f77,1,100}, {0x1f78,0x1f79,1,128}, {0x1f7a,0x1f7b,1,112},
|
|
2220 {0x1f7c,0x1f7d,1,126}, {0x1f80,0x1f87,1,8}, {0x1f90,0x1f97,1,8},
|
|
2221 {0x1fa0,0x1fa7,1,8}, {0x1fb0,0x1fb1,1,8}, {0x1fb3,0x1fb3,-1,9},
|
|
2222 {0x1fbe,0x1fbe,-1,-7205}, {0x1fc3,0x1fc3,-1,9}, {0x1fd0,0x1fd1,1,8},
|
|
2223 {0x1fe0,0x1fe1,1,8}, {0x1fe5,0x1fe5,-1,7}, {0x1ff3,0x1ff3,-1,9},
|
|
2224 {0xff41,0xff5a,1,-32}, {0x10428,0x1044f,1,-40}
|
|
2225 };
|
|
2226
|
|
2227 /*
|
|
2228 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use
|
|
2229 * simple case folding.
|
|
2230 */
|
|
2231 int
|
|
2232 utf_toupper(a)
|
|
2233 int a;
|
|
2234 {
|
|
2235 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */
|
|
2236 if (a < 128 && (cmp_flags & CMP_KEEPASCII))
|
|
2237 return TOUPPER_ASC(a);
|
|
2238
|
856
|
2239 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__)
|
|
2240 /* If towupper() is available and handles Unicode, use it. */
|
7
|
2241 if (!(cmp_flags & CMP_INTERNAL))
|
|
2242 return towupper(a);
|
|
2243 #endif
|
|
2244
|
|
2245 /* For characters below 128 use locale sensitive toupper(). */
|
|
2246 if (a < 128)
|
|
2247 return TOUPPER_LOC(a);
|
|
2248
|
|
2249 /* For any other characters use the above mapping table. */
|
|
2250 return utf_convert(a, toUpper, sizeof(toUpper));
|
|
2251 }
|
|
2252
|
|
2253 int
|
|
2254 utf_islower(a)
|
|
2255 int a;
|
|
2256 {
|
|
2257 return (utf_toupper(a) != a);
|
|
2258 }
|
|
2259
|
|
2260 /*
|
|
2261 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use
|
|
2262 * simple case folding.
|
|
2263 */
|
|
2264 int
|
|
2265 utf_tolower(a)
|
|
2266 int a;
|
|
2267 {
|
|
2268 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */
|
|
2269 if (a < 128 && (cmp_flags & CMP_KEEPASCII))
|
|
2270 return TOLOWER_ASC(a);
|
|
2271
|
856
|
2272 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__)
|
257
|
2273 /* If towlower() is available and handles Unicode, use it. */
|
7
|
2274 if (!(cmp_flags & CMP_INTERNAL))
|
|
2275 return towlower(a);
|
|
2276 #endif
|
|
2277
|
|
2278 /* For characters below 128 use locale sensitive tolower(). */
|
|
2279 if (a < 128)
|
|
2280 return TOLOWER_LOC(a);
|
|
2281
|
|
2282 /* For any other characters use the above mapping table. */
|
|
2283 return utf_convert(a, toLower, sizeof(toLower));
|
|
2284 }
|
|
2285
|
|
2286 int
|
|
2287 utf_isupper(a)
|
|
2288 int a;
|
|
2289 {
|
|
2290 return (utf_tolower(a) != a);
|
|
2291 }
|
|
2292
|
|
2293 /*
|
|
2294 * Version of strnicmp() that handles multi-byte characters.
|
|
2295 * Needed for Big5, Sjift-JIS and UTF-8 encoding. Other DBCS encodings can
|
|
2296 * probably use strnicmp(), because there are no ASCII characters in the
|
|
2297 * second byte.
|
|
2298 * Returns zero if s1 and s2 are equal (ignoring case), the difference between
|
|
2299 * two characters otherwise.
|
|
2300 */
|
|
2301 int
|
303
|
2302 mb_strnicmp(s1, s2, nn)
|
7
|
2303 char_u *s1, *s2;
|
303
|
2304 size_t nn;
|
7
|
2305 {
|
|
2306 int i, j, l;
|
|
2307 int cdiff;
|
308
|
2308 int incomplete = FALSE;
|
835
|
2309 int n = (int)nn;
|
7
|
2310
|
|
2311 for (i = 0; i < n; i += l)
|
|
2312 {
|
|
2313 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */
|
|
2314 return 0;
|
|
2315 if (enc_utf8)
|
|
2316 {
|
|
2317 l = utf_byte2len(s1[i]);
|
|
2318 if (l > n - i)
|
308
|
2319 {
|
7
|
2320 l = n - i; /* incomplete character */
|
308
|
2321 incomplete = TRUE;
|
|
2322 }
|
7
|
2323 /* Check directly first, it's faster. */
|
|
2324 for (j = 0; j < l; ++j)
|
1052
|
2325 {
|
7
|
2326 if (s1[i + j] != s2[i + j])
|
|
2327 break;
|
1052
|
2328 if (s1[i + j] == 0)
|
|
2329 /* Both stings have the same bytes but are incomplete or
|
|
2330 * have illegal bytes, accept them as equal. */
|
|
2331 l = j;
|
|
2332 }
|
7
|
2333 if (j < l)
|
|
2334 {
|
|
2335 /* If one of the two characters is incomplete return -1. */
|
308
|
2336 if (incomplete || i + utf_byte2len(s2[i]) > n)
|
7
|
2337 return -1;
|
|
2338 cdiff = utf_fold(utf_ptr2char(s1 + i))
|
|
2339 - utf_fold(utf_ptr2char(s2 + i));
|
|
2340 if (cdiff != 0)
|
|
2341 return cdiff;
|
|
2342 }
|
|
2343 }
|
|
2344 else
|
|
2345 {
|
474
|
2346 l = (*mb_ptr2len)(s1 + i);
|
7
|
2347 if (l <= 1)
|
|
2348 {
|
|
2349 /* Single byte: first check normally, then with ignore case. */
|
|
2350 if (s1[i] != s2[i])
|
|
2351 {
|
1347
|
2352 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]);
|
7
|
2353 if (cdiff != 0)
|
|
2354 return cdiff;
|
|
2355 }
|
|
2356 }
|
|
2357 else
|
|
2358 {
|
|
2359 /* For non-Unicode multi-byte don't ignore case. */
|
|
2360 if (l > n - i)
|
|
2361 l = n - i;
|
|
2362 cdiff = STRNCMP(s1 + i, s2 + i, l);
|
|
2363 if (cdiff != 0)
|
|
2364 return cdiff;
|
|
2365 }
|
|
2366 }
|
|
2367 }
|
|
2368 return 0;
|
|
2369 }
|
|
2370
|
|
2371 /*
|
|
2372 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what
|
|
2373 * 'encoding' has been set to.
|
|
2374 */
|
|
2375 void
|
|
2376 show_utf8()
|
|
2377 {
|
|
2378 int len;
|
273
|
2379 int rlen = 0;
|
7
|
2380 char_u *line;
|
|
2381 int clen;
|
|
2382 int i;
|
|
2383
|
|
2384 /* Get the byte length of the char under the cursor, including composing
|
|
2385 * characters. */
|
|
2386 line = ml_get_cursor();
|
474
|
2387 len = utfc_ptr2len(line);
|
7
|
2388 if (len == 0)
|
|
2389 {
|
|
2390 MSG("NUL");
|
|
2391 return;
|
|
2392 }
|
|
2393
|
|
2394 clen = 0;
|
|
2395 for (i = 0; i < len; ++i)
|
|
2396 {
|
|
2397 if (clen == 0)
|
|
2398 {
|
|
2399 /* start of (composing) character, get its length */
|
|
2400 if (i > 0)
|
273
|
2401 {
|
|
2402 STRCPY(IObuff + rlen, "+ ");
|
|
2403 rlen += 2;
|
|
2404 }
|
474
|
2405 clen = utf_ptr2len(line + i);
|
7
|
2406 }
|
273
|
2407 sprintf((char *)IObuff + rlen, "%02x ", line[i]);
|
7
|
2408 --clen;
|
835
|
2409 rlen += (int)STRLEN(IObuff + rlen);
|
273
|
2410 if (rlen > IOSIZE - 20)
|
|
2411 break;
|
7
|
2412 }
|
|
2413
|
|
2414 msg(IObuff);
|
|
2415 }
|
|
2416
|
|
2417 /*
|
|
2418 * mb_head_off() function pointer.
|
|
2419 * Return offset from "p" to the first byte of the character it points into.
|
|
2420 * Returns 0 when already at the first byte of a character.
|
|
2421 */
|
|
2422 /*ARGSUSED*/
|
|
2423 int
|
|
2424 latin_head_off(base, p)
|
|
2425 char_u *base;
|
|
2426 char_u *p;
|
|
2427 {
|
|
2428 return 0;
|
|
2429 }
|
|
2430
|
|
2431 int
|
|
2432 dbcs_head_off(base, p)
|
|
2433 char_u *base;
|
|
2434 char_u *p;
|
|
2435 {
|
|
2436 char_u *q;
|
|
2437
|
|
2438 /* It can't be a trailing byte when not using DBCS, at the start of the
|
|
2439 * string or the previous byte can't start a double-byte. */
|
|
2440 if (p <= base || MB_BYTE2LEN(p[-1]) == 1)
|
|
2441 return 0;
|
|
2442
|
|
2443 /* This is slow: need to start at the base and go forward until the
|
|
2444 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */
|
|
2445 q = base;
|
|
2446 while (q < p)
|
474
|
2447 q += dbcs_ptr2len(q);
|
7
|
2448 return (q == p) ? 0 : 1;
|
|
2449 }
|
|
2450
|
|
2451 /*
|
|
2452 * Special version of dbcs_head_off() that works for ScreenLines[], where
|
|
2453 * single-width DBCS_JPNU characters are stored separately.
|
|
2454 */
|
|
2455 int
|
|
2456 dbcs_screen_head_off(base, p)
|
|
2457 char_u *base;
|
|
2458 char_u *p;
|
|
2459 {
|
|
2460 char_u *q;
|
|
2461
|
|
2462 /* It can't be a trailing byte when not using DBCS, at the start of the
|
|
2463 * string or the previous byte can't start a double-byte.
|
|
2464 * For euc-jp an 0x8e byte in the previous cell always means we have a
|
|
2465 * lead byte in the current cell. */
|
|
2466 if (p <= base
|
|
2467 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e)
|
|
2468 || MB_BYTE2LEN(p[-1]) == 1)
|
|
2469 return 0;
|
|
2470
|
|
2471 /* This is slow: need to start at the base and go forward until the
|
|
2472 * byte we are looking for. Return 1 when we went past it, 0 otherwise.
|
|
2473 * For DBCS_JPNU look out for 0x8e, which means the second byte is not
|
|
2474 * stored as the next byte. */
|
|
2475 q = base;
|
|
2476 while (q < p)
|
|
2477 {
|
|
2478 if (enc_dbcs == DBCS_JPNU && *q == 0x8e)
|
|
2479 ++q;
|
|
2480 else
|
474
|
2481 q += dbcs_ptr2len(q);
|
7
|
2482 }
|
|
2483 return (q == p) ? 0 : 1;
|
|
2484 }
|
|
2485
|
|
2486 int
|
|
2487 utf_head_off(base, p)
|
|
2488 char_u *base;
|
|
2489 char_u *p;
|
|
2490 {
|
|
2491 char_u *q;
|
|
2492 char_u *s;
|
|
2493 int c;
|
|
2494 #ifdef FEAT_ARABIC
|
|
2495 char_u *j;
|
|
2496 #endif
|
|
2497
|
|
2498 if (*p < 0x80) /* be quick for ASCII */
|
|
2499 return 0;
|
|
2500
|
|
2501 /* Skip backwards over trailing bytes: 10xx.xxxx
|
|
2502 * Skip backwards again if on a composing char. */
|
|
2503 for (q = p; ; --q)
|
|
2504 {
|
|
2505 /* Move s to the last byte of this char. */
|
|
2506 for (s = q; (s[1] & 0xc0) == 0x80; ++s)
|
|
2507 ;
|
|
2508 /* Move q to the first byte of this char. */
|
|
2509 while (q > base && (*q & 0xc0) == 0x80)
|
|
2510 --q;
|
|
2511 /* Check for illegal sequence. Do allow an illegal byte after where we
|
|
2512 * started. */
|
|
2513 if (utf8len_tab[*q] != (int)(s - q + 1)
|
|
2514 && utf8len_tab[*q] != (int)(p - q + 1))
|
|
2515 return 0;
|
|
2516
|
|
2517 if (q <= base)
|
|
2518 break;
|
|
2519
|
|
2520 c = utf_ptr2char(q);
|
|
2521 if (utf_iscomposing(c))
|
|
2522 continue;
|
|
2523
|
|
2524 #ifdef FEAT_ARABIC
|
|
2525 if (arabic_maycombine(c))
|
|
2526 {
|
|
2527 /* Advance to get a sneak-peak at the next char */
|
|
2528 j = q;
|
|
2529 --j;
|
|
2530 /* Move j to the first byte of this char. */
|
|
2531 while (j > base && (*j & 0xc0) == 0x80)
|
|
2532 --j;
|
|
2533 if (arabic_combine(utf_ptr2char(j), c))
|
|
2534 continue;
|
|
2535 }
|
|
2536 #endif
|
|
2537 break;
|
|
2538 }
|
|
2539
|
|
2540 return (int)(p - q);
|
|
2541 }
|
|
2542
|
140
|
2543 #if defined(FEAT_EVAL) || defined(PROTO)
|
7
|
2544 /*
|
99
|
2545 * Copy a character from "*fp" to "*tp" and advance the pointers.
|
|
2546 */
|
|
2547 void
|
|
2548 mb_copy_char(fp, tp)
|
|
2549 char_u **fp;
|
|
2550 char_u **tp;
|
|
2551 {
|
474
|
2552 int l = (*mb_ptr2len)(*fp);
|
99
|
2553
|
|
2554 mch_memmove(*tp, *fp, (size_t)l);
|
|
2555 *tp += l;
|
|
2556 *fp += l;
|
|
2557 }
|
140
|
2558 #endif
|
99
|
2559
|
|
2560 /*
|
7
|
2561 * Return the offset from "p" to the first byte of a character. When "p" is
|
|
2562 * at the start of a character 0 is returned, otherwise the offset to the next
|
|
2563 * character. Can start anywhere in a stream of bytes.
|
|
2564 */
|
|
2565 int
|
|
2566 mb_off_next(base, p)
|
|
2567 char_u *base;
|
|
2568 char_u *p;
|
|
2569 {
|
|
2570 int i;
|
|
2571 int j;
|
|
2572
|
|
2573 if (enc_utf8)
|
|
2574 {
|
|
2575 if (*p < 0x80) /* be quick for ASCII */
|
|
2576 return 0;
|
|
2577
|
|
2578 /* Find the next character that isn't 10xx.xxxx */
|
|
2579 for (i = 0; (p[i] & 0xc0) == 0x80; ++i)
|
|
2580 ;
|
|
2581 if (i > 0)
|
|
2582 {
|
|
2583 /* Check for illegal sequence. */
|
|
2584 for (j = 0; p - j > base; ++j)
|
|
2585 if ((p[-j] & 0xc0) != 0x80)
|
|
2586 break;
|
|
2587 if (utf8len_tab[p[-j]] != i + j)
|
|
2588 return 0;
|
|
2589 }
|
|
2590 return i;
|
|
2591 }
|
|
2592
|
|
2593 /* Only need to check if we're on a trail byte, it doesn't matter if we
|
|
2594 * want the offset to the next or current character. */
|
|
2595 return (*mb_head_off)(base, p);
|
|
2596 }
|
|
2597
|
|
2598 /*
|
|
2599 * Return the offset from "p" to the last byte of the character it points
|
|
2600 * into. Can start anywhere in a stream of bytes.
|
|
2601 */
|
|
2602 int
|
|
2603 mb_tail_off(base, p)
|
|
2604 char_u *base;
|
|
2605 char_u *p;
|
|
2606 {
|
|
2607 int i;
|
|
2608 int j;
|
|
2609
|
|
2610 if (*p == NUL)
|
|
2611 return 0;
|
|
2612
|
|
2613 if (enc_utf8)
|
|
2614 {
|
|
2615 /* Find the last character that is 10xx.xxxx */
|
|
2616 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i)
|
|
2617 ;
|
|
2618 /* Check for illegal sequence. */
|
|
2619 for (j = 0; p - j > base; ++j)
|
|
2620 if ((p[-j] & 0xc0) != 0x80)
|
|
2621 break;
|
|
2622 if (utf8len_tab[p[-j]] != i + j + 1)
|
|
2623 return 0;
|
|
2624 return i;
|
|
2625 }
|
|
2626
|
|
2627 /* It can't be the first byte if a double-byte when not using DBCS, at the
|
|
2628 * end of the string or the byte can't start a double-byte. */
|
|
2629 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1)
|
|
2630 return 0;
|
|
2631
|
|
2632 /* Return 1 when on the lead byte, 0 when on the tail byte. */
|
|
2633 return 1 - dbcs_head_off(base, p);
|
|
2634 }
|
|
2635
|
776
|
2636 /*
|
|
2637 * Find the next illegal byte sequence.
|
|
2638 */
|
|
2639 void
|
|
2640 utf_find_illegal()
|
|
2641 {
|
|
2642 pos_T pos = curwin->w_cursor;
|
|
2643 char_u *p;
|
|
2644 int len;
|
|
2645 vimconv_T vimconv;
|
|
2646 char_u *tofree = NULL;
|
|
2647
|
|
2648 vimconv.vc_type = CONV_NONE;
|
|
2649 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT))
|
|
2650 {
|
|
2651 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file,
|
|
2652 * possibly a utf-8 file with illegal bytes. Setup for conversion
|
|
2653 * from utf-8 to 'fileencoding'. */
|
|
2654 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc);
|
|
2655 }
|
|
2656
|
|
2657 #ifdef FEAT_VIRTUALEDIT
|
|
2658 curwin->w_cursor.coladd = 0;
|
|
2659 #endif
|
|
2660 for (;;)
|
|
2661 {
|
|
2662 p = ml_get_cursor();
|
|
2663 if (vimconv.vc_type != CONV_NONE)
|
|
2664 {
|
|
2665 vim_free(tofree);
|
|
2666 tofree = string_convert(&vimconv, p, NULL);
|
|
2667 if (tofree == NULL)
|
|
2668 break;
|
|
2669 p = tofree;
|
|
2670 }
|
|
2671
|
|
2672 while (*p != NUL)
|
|
2673 {
|
|
2674 /* Illegal means that there are not enough trail bytes (checked by
|
|
2675 * utf_ptr2len()) or too many of them (overlong sequence). */
|
|
2676 len = utf_ptr2len(p);
|
|
2677 if (*p >= 0x80 && (len == 1
|
|
2678 || utf_char2len(utf_ptr2char(p)) != len))
|
|
2679 {
|
|
2680 if (vimconv.vc_type == CONV_NONE)
|
835
|
2681 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor());
|
776
|
2682 else
|
|
2683 {
|
|
2684 int l;
|
|
2685
|
835
|
2686 len = (int)(p - tofree);
|
776
|
2687 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l)
|
|
2688 {
|
|
2689 l = utf_ptr2len(p);
|
|
2690 curwin->w_cursor.col += l;
|
|
2691 }
|
|
2692 }
|
|
2693 goto theend;
|
|
2694 }
|
|
2695 p += len;
|
|
2696 }
|
|
2697 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count)
|
|
2698 break;
|
|
2699 ++curwin->w_cursor.lnum;
|
|
2700 curwin->w_cursor.col = 0;
|
|
2701 }
|
|
2702
|
|
2703 /* didn't find it: don't move and beep */
|
|
2704 curwin->w_cursor = pos;
|
|
2705 beep_flush();
|
|
2706
|
|
2707 theend:
|
|
2708 vim_free(tofree);
|
|
2709 convert_setup(&vimconv, NULL, NULL);
|
|
2710 }
|
|
2711
|
26
|
2712 #if defined(HAVE_GTK2) || defined(PROTO)
|
|
2713 /*
|
|
2714 * Return TRUE if string "s" is a valid utf-8 string.
|
|
2715 * When "end" is NULL stop at the first NUL.
|
|
2716 * When "end" is positive stop there.
|
|
2717 */
|
|
2718 int
|
|
2719 utf_valid_string(s, end)
|
|
2720 char_u *s;
|
|
2721 char_u *end;
|
|
2722 {
|
|
2723 int l;
|
|
2724 char_u *p = s;
|
|
2725
|
|
2726 while (end == NULL ? *p != NUL : p < end)
|
|
2727 {
|
|
2728 if ((*p & 0xc0) == 0x80)
|
|
2729 return FALSE; /* invalid lead byte */
|
|
2730 l = utf8len_tab[*p];
|
|
2731 if (end != NULL && p + l > end)
|
|
2732 return FALSE; /* incomplete byte sequence */
|
|
2733 ++p;
|
|
2734 while (--l > 0)
|
|
2735 if ((*p++ & 0xc0) != 0x80)
|
|
2736 return FALSE; /* invalid trail byte */
|
|
2737 }
|
|
2738 return TRUE;
|
|
2739 }
|
|
2740 #endif
|
|
2741
|
7
|
2742 #if defined(FEAT_GUI) || defined(PROTO)
|
|
2743 /*
|
|
2744 * Special version of mb_tail_off() for use in ScreenLines[].
|
|
2745 */
|
|
2746 int
|
|
2747 dbcs_screen_tail_off(base, p)
|
|
2748 char_u *base;
|
|
2749 char_u *p;
|
|
2750 {
|
|
2751 /* It can't be the first byte if a double-byte when not using DBCS, at the
|
|
2752 * end of the string or the byte can't start a double-byte.
|
|
2753 * For euc-jp an 0x8e byte always means we have a lead byte in the current
|
|
2754 * cell. */
|
|
2755 if (*p == NUL || p[1] == NUL
|
|
2756 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)
|
|
2757 || MB_BYTE2LEN(*p) == 1)
|
|
2758 return 0;
|
|
2759
|
|
2760 /* Return 1 when on the lead byte, 0 when on the tail byte. */
|
|
2761 return 1 - dbcs_screen_head_off(base, p);
|
|
2762 }
|
|
2763 #endif
|
|
2764
|
|
2765 /*
|
|
2766 * If the cursor moves on an trail byte, set the cursor on the lead byte.
|
|
2767 * Thus it moves left if necessary.
|
|
2768 * Return TRUE when the cursor was adjusted.
|
|
2769 */
|
|
2770 void
|
|
2771 mb_adjust_cursor()
|
|
2772 {
|
|
2773 mb_adjustpos(&curwin->w_cursor);
|
|
2774 }
|
|
2775
|
|
2776 /*
|
|
2777 * Adjust position "*lp" to point to the first byte of a multi-byte character.
|
|
2778 * If it points to a tail byte it's moved backwards to the head byte.
|
|
2779 */
|
|
2780 void
|
|
2781 mb_adjustpos(lp)
|
|
2782 pos_T *lp;
|
|
2783 {
|
|
2784 char_u *p;
|
|
2785
|
|
2786 if (lp->col > 0
|
|
2787 #ifdef FEAT_VIRTUALEDIT
|
|
2788 || lp->coladd > 1
|
|
2789 #endif
|
|
2790 )
|
|
2791 {
|
|
2792 p = ml_get(lp->lnum);
|
|
2793 lp->col -= (*mb_head_off)(p, p + lp->col);
|
|
2794 #ifdef FEAT_VIRTUALEDIT
|
|
2795 /* Reset "coladd" when the cursor would be on the right half of a
|
|
2796 * double-wide character. */
|
|
2797 if (lp->coladd == 1
|
|
2798 && p[lp->col] != TAB
|
|
2799 && vim_isprintc((*mb_ptr2char)(p + lp->col))
|
|
2800 && ptr2cells(p + lp->col) > 1)
|
|
2801 lp->coladd = 0;
|
|
2802 #endif
|
|
2803 }
|
|
2804 }
|
|
2805
|
|
2806 /*
|
|
2807 * Return a pointer to the character before "*p", if there is one.
|
|
2808 */
|
|
2809 char_u *
|
|
2810 mb_prevptr(line, p)
|
|
2811 char_u *line; /* start of the string */
|
|
2812 char_u *p;
|
|
2813 {
|
|
2814 if (p > line)
|
39
|
2815 mb_ptr_back(line, p);
|
7
|
2816 return p;
|
|
2817 }
|
|
2818
|
|
2819 /*
|
474
|
2820 * Return the character length of "str". Each multi-byte character (with
|
|
2821 * following composing characters) counts as one.
|
7
|
2822 */
|
|
2823 int
|
|
2824 mb_charlen(str)
|
|
2825 char_u *str;
|
|
2826 {
|
500
|
2827 char_u *p = str;
|
|
2828 int count;
|
|
2829
|
|
2830 if (p == NULL)
|
7
|
2831 return 0;
|
|
2832
|
500
|
2833 for (count = 0; *p != NUL; count++)
|
|
2834 p += (*mb_ptr2len)(p);
|
7
|
2835
|
|
2836 return count;
|
|
2837 }
|
|
2838
|
740
|
2839 #if defined(FEAT_SPELL) || defined(PROTO)
|
500
|
2840 /*
|
|
2841 * Like mb_charlen() but for a string with specified length.
|
|
2842 */
|
|
2843 int
|
|
2844 mb_charlen_len(str, len)
|
|
2845 char_u *str;
|
|
2846 int len;
|
|
2847 {
|
|
2848 char_u *p = str;
|
|
2849 int count;
|
|
2850
|
|
2851 for (count = 0; *p != NUL && p < str + len; count++)
|
|
2852 p += (*mb_ptr2len)(p);
|
|
2853
|
|
2854 return count;
|
|
2855 }
|
|
2856 #endif
|
|
2857
|
7
|
2858 /*
|
|
2859 * Try to un-escape a multi-byte character.
|
|
2860 * Used for the "to" and "from" part of a mapping.
|
|
2861 * Return the un-escaped string if it is a multi-byte character, and advance
|
|
2862 * "pp" to just after the bytes that formed it.
|
|
2863 * Return NULL if no multi-byte char was found.
|
|
2864 */
|
|
2865 char_u *
|
|
2866 mb_unescape(pp)
|
|
2867 char_u **pp;
|
|
2868 {
|
|
2869 static char_u buf[MB_MAXBYTES + 1];
|
|
2870 int n, m = 0;
|
|
2871 char_u *str = *pp;
|
|
2872
|
|
2873 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI
|
|
2874 * KS_EXTRA KE_CSI to CSI. */
|
|
2875 for (n = 0; str[n] != NUL && m <= MB_MAXBYTES; ++n)
|
|
2876 {
|
|
2877 if (str[n] == K_SPECIAL
|
|
2878 && str[n + 1] == KS_SPECIAL
|
|
2879 && str[n + 2] == KE_FILLER)
|
|
2880 {
|
|
2881 buf[m++] = K_SPECIAL;
|
|
2882 n += 2;
|
|
2883 }
|
1495
|
2884 else if ((str[n] == K_SPECIAL
|
7
|
2885 # ifdef FEAT_GUI
|
1495
|
2886 || str[n] == CSI
|
|
2887 # endif
|
|
2888 )
|
7
|
2889 && str[n + 1] == KS_EXTRA
|
|
2890 && str[n + 2] == (int)KE_CSI)
|
|
2891 {
|
|
2892 buf[m++] = CSI;
|
|
2893 n += 2;
|
|
2894 }
|
|
2895 else if (str[n] == K_SPECIAL
|
|
2896 # ifdef FEAT_GUI
|
|
2897 || str[n] == CSI
|
|
2898 # endif
|
|
2899 )
|
|
2900 break; /* a special key can't be a multibyte char */
|
|
2901 else
|
|
2902 buf[m++] = str[n];
|
|
2903 buf[m] = NUL;
|
|
2904
|
|
2905 /* Return a multi-byte character if it's found. An illegal sequence
|
|
2906 * will result in a 1 here. */
|
474
|
2907 if ((*mb_ptr2len)(buf) > 1)
|
7
|
2908 {
|
|
2909 *pp = str + n + 1;
|
|
2910 return buf;
|
|
2911 }
|
|
2912 }
|
|
2913 return NULL;
|
|
2914 }
|
|
2915
|
|
2916 /*
|
|
2917 * Return TRUE if the character at "row"/"col" on the screen is the left side
|
|
2918 * of a double-width character.
|
|
2919 * Caller must make sure "row" and "col" are not invalid!
|
|
2920 */
|
|
2921 int
|
|
2922 mb_lefthalve(row, col)
|
|
2923 int row;
|
|
2924 int col;
|
|
2925 {
|
|
2926 #ifdef FEAT_HANGULIN
|
|
2927 if (composing_hangul)
|
|
2928 return TRUE;
|
|
2929 #endif
|
1378
|
2930 return (*mb_off2cells)(LineOffset[row] + col,
|
|
2931 LineOffset[row] + screen_Columns) > 1;
|
7
|
2932 }
|
|
2933
|
|
2934 /*
|
1698
|
2935 * Correct a position on the screen, if it's the right half of a double-wide
|
|
2936 * char move it to the left half. Returns the corrected column.
|
7
|
2937 */
|
|
2938 int
|
|
2939 mb_fix_col(col, row)
|
|
2940 int col;
|
|
2941 int row;
|
|
2942 {
|
|
2943 col = check_col(col);
|
|
2944 row = check_row(row);
|
|
2945 if (has_mbyte && ScreenLines != NULL && col > 0
|
|
2946 && ((enc_dbcs
|
|
2947 && ScreenLines[LineOffset[row] + col] != NUL
|
|
2948 && dbcs_screen_head_off(ScreenLines + LineOffset[row],
|
|
2949 ScreenLines + LineOffset[row] + col))
|
|
2950 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0)))
|
1620
|
2951 return col - 1;
|
7
|
2952 return col;
|
|
2953 }
|
|
2954 #endif
|
|
2955
|
|
2956 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO)
|
|
2957 static int enc_alias_search __ARGS((char_u *name));
|
|
2958
|
|
2959 /*
|
|
2960 * Skip the Vim specific head of a 'encoding' name.
|
|
2961 */
|
|
2962 char_u *
|
|
2963 enc_skip(p)
|
|
2964 char_u *p;
|
|
2965 {
|
|
2966 if (STRNCMP(p, "2byte-", 6) == 0)
|
|
2967 return p + 6;
|
|
2968 if (STRNCMP(p, "8bit-", 5) == 0)
|
|
2969 return p + 5;
|
|
2970 return p;
|
|
2971 }
|
|
2972
|
|
2973 /*
|
|
2974 * Find the canonical name for encoding "enc".
|
|
2975 * When the name isn't recognized, returns "enc" itself, but with all lower
|
|
2976 * case characters and '_' replaced with '-'.
|
|
2977 * Returns an allocated string. NULL for out-of-memory.
|
|
2978 */
|
|
2979 char_u *
|
|
2980 enc_canonize(enc)
|
|
2981 char_u *enc;
|
|
2982 {
|
|
2983 char_u *r;
|
|
2984 char_u *p, *s;
|
|
2985 int i;
|
|
2986
|
39
|
2987 # ifdef FEAT_MBYTE
|
|
2988 if (STRCMP(enc, "default") == 0)
|
|
2989 {
|
|
2990 /* Use the default encoding as it's found by set_init_1(). */
|
|
2991 r = get_encoding_default();
|
|
2992 if (r == NULL)
|
|
2993 r = (char_u *)"latin1";
|
|
2994 return vim_strsave(r);
|
|
2995 }
|
|
2996 # endif
|
|
2997
|
1205
|
2998 /* copy "enc" to allocated memory, with room for two '-' */
|
7
|
2999 r = alloc((unsigned)(STRLEN(enc) + 3));
|
|
3000 if (r != NULL)
|
|
3001 {
|
|
3002 /* Make it all lower case and replace '_' with '-'. */
|
|
3003 p = r;
|
|
3004 for (s = enc; *s != NUL; ++s)
|
|
3005 {
|
|
3006 if (*s == '_')
|
|
3007 *p++ = '-';
|
|
3008 else
|
|
3009 *p++ = TOLOWER_ASC(*s);
|
|
3010 }
|
|
3011 *p = NUL;
|
|
3012
|
|
3013 /* Skip "2byte-" and "8bit-". */
|
|
3014 p = enc_skip(r);
|
|
3015
|
481
|
3016 /* Change "microsoft-cp" to "cp". Used in some spell files. */
|
|
3017 if (STRNCMP(p, "microsoft-cp", 12) == 0)
|
1620
|
3018 STRMOVE(p, p + 10);
|
481
|
3019
|
7
|
3020 /* "iso8859" -> "iso-8859" */
|
|
3021 if (STRNCMP(p, "iso8859", 7) == 0)
|
|
3022 {
|
1620
|
3023 STRMOVE(p + 4, p + 3);
|
7
|
3024 p[3] = '-';
|
|
3025 }
|
|
3026
|
|
3027 /* "iso-8859n" -> "iso-8859-n" */
|
|
3028 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-')
|
|
3029 {
|
1620
|
3030 STRMOVE(p + 9, p + 8);
|
7
|
3031 p[8] = '-';
|
|
3032 }
|
|
3033
|
|
3034 /* "latin-N" -> "latinN" */
|
|
3035 if (STRNCMP(p, "latin-", 6) == 0)
|
1620
|
3036 STRMOVE(p + 5, p + 6);
|
7
|
3037
|
|
3038 if (enc_canon_search(p) >= 0)
|
|
3039 {
|
|
3040 /* canonical name can be used unmodified */
|
|
3041 if (p != r)
|
1620
|
3042 STRMOVE(r, p);
|
7
|
3043 }
|
|
3044 else if ((i = enc_alias_search(p)) >= 0)
|
|
3045 {
|
|
3046 /* alias recognized, get canonical name */
|
|
3047 vim_free(r);
|
|
3048 r = vim_strsave((char_u *)enc_canon_table[i].name);
|
|
3049 }
|
|
3050 }
|
|
3051 return r;
|
|
3052 }
|
|
3053
|
|
3054 /*
|
|
3055 * Search for an encoding alias of "name".
|
|
3056 * Returns -1 when not found.
|
|
3057 */
|
|
3058 static int
|
|
3059 enc_alias_search(name)
|
|
3060 char_u *name;
|
|
3061 {
|
|
3062 int i;
|
|
3063
|
|
3064 for (i = 0; enc_alias_table[i].name != NULL; ++i)
|
|
3065 if (STRCMP(name, enc_alias_table[i].name) == 0)
|
|
3066 return enc_alias_table[i].canon;
|
|
3067 return -1;
|
|
3068 }
|
|
3069 #endif
|
|
3070
|
|
3071 #if defined(FEAT_MBYTE) || defined(PROTO)
|
|
3072
|
|
3073 #ifdef HAVE_LANGINFO_H
|
|
3074 # include <langinfo.h>
|
|
3075 #endif
|
|
3076
|
|
3077 /*
|
|
3078 * Get the canonicalized encoding of the current locale.
|
|
3079 * Returns an allocated string when successful, NULL when not.
|
|
3080 */
|
|
3081 char_u *
|
|
3082 enc_locale()
|
|
3083 {
|
|
3084 #ifndef WIN3264
|
|
3085 char *s;
|
|
3086 char *p;
|
|
3087 int i;
|
|
3088 #endif
|
|
3089 char buf[50];
|
|
3090 #ifdef WIN3264
|
|
3091 long acp = GetACP();
|
|
3092
|
|
3093 if (acp == 1200)
|
|
3094 STRCPY(buf, "ucs-2le");
|
407
|
3095 else if (acp == 1252) /* cp1252 is used as latin1 */
|
7
|
3096 STRCPY(buf, "latin1");
|
|
3097 else
|
|
3098 sprintf(buf, "cp%ld", acp);
|
|
3099 #else
|
|
3100 # ifdef HAVE_NL_LANGINFO_CODESET
|
|
3101 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL)
|
|
3102 # endif
|
167
|
3103 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE)
|
7
|
3104 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL)
|
167
|
3105 # endif
|
7
|
3106 if ((s = getenv("LC_ALL")) == NULL || *s == NUL)
|
|
3107 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL)
|
|
3108 s = getenv("LANG");
|
|
3109
|
|
3110 if (s == NULL || *s == NUL)
|
|
3111 return FAIL;
|
|
3112
|
|
3113 /* The most generic locale format is:
|
|
3114 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]]
|
|
3115 * If there is a '.' remove the part before it.
|
|
3116 * if there is something after the codeset, remove it.
|
|
3117 * Make the name lowercase and replace '_' with '-'.
|
|
3118 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn",
|
|
3119 * "ko_KR.EUC" == "euc-kr"
|
|
3120 */
|
|
3121 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL)
|
|
3122 {
|
|
3123 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0
|
|
3124 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_')
|
|
3125 {
|
|
3126 /* copy "XY.EUC" to "euc-XY" to buf[10] */
|
|
3127 STRCPY(buf + 10, "euc-");
|
|
3128 buf[14] = p[-2];
|
|
3129 buf[15] = p[-1];
|
|
3130 buf[16] = 0;
|
|
3131 s = buf + 10;
|
|
3132 }
|
|
3133 else
|
|
3134 s = p + 1;
|
|
3135 }
|
840
|
3136 for (i = 0; s[i] != NUL && s + i < buf + sizeof(buf) - 1; ++i)
|
7
|
3137 {
|
|
3138 if (s[i] == '_' || s[i] == '-')
|
|
3139 buf[i] = '-';
|
|
3140 else if (isalnum((int)s[i]))
|
|
3141 buf[i] = TOLOWER_ASC(s[i]);
|
|
3142 else
|
|
3143 break;
|
|
3144 }
|
|
3145 buf[i] = NUL;
|
|
3146 #endif
|
|
3147
|
|
3148 return enc_canonize((char_u *)buf);
|
|
3149 }
|
|
3150
|
|
3151 #if defined(WIN3264) || defined(PROTO)
|
|
3152 /*
|
|
3153 * Convert an encoding name to an MS-Windows codepage.
|
|
3154 * Returns zero if no codepage can be figured out.
|
|
3155 */
|
|
3156 int
|
|
3157 encname2codepage(name)
|
|
3158 char_u *name;
|
|
3159 {
|
|
3160 int cp;
|
|
3161 char_u *p = name;
|
|
3162 int idx;
|
|
3163
|
|
3164 if (STRNCMP(p, "8bit-", 5) == 0)
|
|
3165 p += 5;
|
|
3166 else if (STRNCMP(p_enc, "2byte-", 6) == 0)
|
|
3167 p += 6;
|
|
3168
|
|
3169 if (p[0] == 'c' && p[1] == 'p')
|
|
3170 cp = atoi(p + 2);
|
|
3171 else if ((idx = enc_canon_search(p)) >= 0)
|
|
3172 cp = enc_canon_table[idx].codepage;
|
|
3173 else
|
|
3174 return 0;
|
|
3175 if (IsValidCodePage(cp))
|
|
3176 return cp;
|
|
3177 return 0;
|
|
3178 }
|
|
3179 #endif
|
|
3180
|
|
3181 # if defined(USE_ICONV) || defined(PROTO)
|
|
3182
|
|
3183 static char_u *iconv_string __ARGS((vimconv_T *vcp, char_u *str, int slen, int *unconvlenp));
|
|
3184
|
|
3185 /*
|
|
3186 * Call iconv_open() with a check if iconv() works properly (there are broken
|
|
3187 * versions).
|
|
3188 * Returns (void *)-1 if failed.
|
|
3189 * (should return iconv_t, but that causes problems with prototypes).
|
|
3190 */
|
|
3191 void *
|
|
3192 my_iconv_open(to, from)
|
|
3193 char_u *to;
|
|
3194 char_u *from;
|
|
3195 {
|
|
3196 iconv_t fd;
|
|
3197 #define ICONV_TESTLEN 400
|
|
3198 char_u tobuf[ICONV_TESTLEN];
|
|
3199 char *p;
|
|
3200 size_t tolen;
|
|
3201 static int iconv_ok = -1;
|
|
3202
|
|
3203 if (iconv_ok == FALSE)
|
|
3204 return (void *)-1; /* detected a broken iconv() previously */
|
|
3205
|
|
3206 #ifdef DYNAMIC_ICONV
|
|
3207 /* Check if the iconv.dll can be found. */
|
|
3208 if (!iconv_enabled(TRUE))
|
|
3209 return (void *)-1;
|
|
3210 #endif
|
|
3211
|
|
3212 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from));
|
|
3213
|
|
3214 if (fd != (iconv_t)-1 && iconv_ok == -1)
|
|
3215 {
|
|
3216 /*
|
|
3217 * Do a dummy iconv() call to check if it actually works. There is a
|
|
3218 * version of iconv() on Linux that is broken. We can't ignore it,
|
|
3219 * because it's wide-spread. The symptoms are that after outputting
|
|
3220 * the initial shift state the "to" pointer is NULL and conversion
|
|
3221 * stops for no apparent reason after about 8160 characters.
|
|
3222 */
|
|
3223 p = (char *)tobuf;
|
|
3224 tolen = ICONV_TESTLEN;
|
|
3225 (void)iconv(fd, NULL, NULL, &p, &tolen);
|
|
3226 if (p == NULL)
|
|
3227 {
|
|
3228 iconv_ok = FALSE;
|
|
3229 iconv_close(fd);
|
|
3230 fd = (iconv_t)-1;
|
|
3231 }
|
|
3232 else
|
|
3233 iconv_ok = TRUE;
|
|
3234 }
|
|
3235
|
|
3236 return (void *)fd;
|
|
3237 }
|
|
3238
|
|
3239 /*
|
|
3240 * Convert the string "str[slen]" with iconv().
|
|
3241 * If "unconvlenp" is not NULL handle the string ending in an incomplete
|
|
3242 * sequence and set "*unconvlenp" to the length of it.
|
|
3243 * Returns the converted string in allocated memory. NULL for an error.
|
|
3244 */
|
|
3245 static char_u *
|
|
3246 iconv_string(vcp, str, slen, unconvlenp)
|
|
3247 vimconv_T *vcp;
|
|
3248 char_u *str;
|
|
3249 int slen;
|
|
3250 int *unconvlenp;
|
|
3251 {
|
|
3252 const char *from;
|
|
3253 size_t fromlen;
|
|
3254 char *to;
|
|
3255 size_t tolen;
|
|
3256 size_t len = 0;
|
|
3257 size_t done = 0;
|
|
3258 char_u *result = NULL;
|
|
3259 char_u *p;
|
|
3260 int l;
|
|
3261
|
|
3262 from = (char *)str;
|
|
3263 fromlen = slen;
|
|
3264 for (;;)
|
|
3265 {
|
|
3266 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG)
|
|
3267 {
|
|
3268 /* Allocate enough room for most conversions. When re-allocating
|
|
3269 * increase the buffer size. */
|
|
3270 len = len + fromlen * 2 + 40;
|
|
3271 p = alloc((unsigned)len);
|
|
3272 if (p != NULL && done > 0)
|
|
3273 mch_memmove(p, result, done);
|
|
3274 vim_free(result);
|
|
3275 result = p;
|
|
3276 if (result == NULL) /* out of memory */
|
|
3277 break;
|
|
3278 }
|
|
3279
|
|
3280 to = (char *)result + done;
|
|
3281 tolen = len - done - 2;
|
|
3282 /* Avoid a warning for systems with a wrong iconv() prototype by
|
|
3283 * casting the second argument to void *. */
|
|
3284 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen)
|
|
3285 != (size_t)-1)
|
|
3286 {
|
|
3287 /* Finished, append a NUL. */
|
|
3288 *to = NUL;
|
|
3289 break;
|
|
3290 }
|
|
3291
|
|
3292 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded
|
|
3293 * iconv library may use one of them. */
|
|
3294 if (!vcp->vc_fail && unconvlenp != NULL
|
|
3295 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL))
|
|
3296 {
|
|
3297 /* Handle an incomplete sequence at the end. */
|
|
3298 *to = NUL;
|
835
|
3299 *unconvlenp = (int)fromlen;
|
7
|
3300 break;
|
|
3301 }
|
|
3302
|
|
3303 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded
|
|
3304 * iconv library may use one of them. */
|
|
3305 else if (!vcp->vc_fail
|
|
3306 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ
|
|
3307 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL))
|
|
3308 {
|
|
3309 /* Can't convert: insert a '?' and skip a character. This assumes
|
|
3310 * conversion from 'encoding' to something else. In other
|
|
3311 * situations we don't know what to skip anyway. */
|
|
3312 *to++ = '?';
|
|
3313 if ((*mb_ptr2cells)((char_u *)from) > 1)
|
|
3314 *to++ = '?';
|
221
|
3315 if (enc_utf8)
|
835
|
3316 l = utfc_ptr2len_len((char_u *)from, (int)fromlen);
|
221
|
3317 else
|
|
3318 {
|
474
|
3319 l = (*mb_ptr2len)((char_u *)from);
|
227
|
3320 if (l > (int)fromlen)
|
835
|
3321 l = (int)fromlen;
|
221
|
3322 }
|
7
|
3323 from += l;
|
|
3324 fromlen -= l;
|
|
3325 }
|
|
3326 else if (ICONV_ERRNO != ICONV_E2BIG)
|
|
3327 {
|
|
3328 /* conversion failed */
|
|
3329 vim_free(result);
|
|
3330 result = NULL;
|
|
3331 break;
|
|
3332 }
|
|
3333 /* Not enough room or skipping illegal sequence. */
|
|
3334 done = to - (char *)result;
|
|
3335 }
|
|
3336 return result;
|
|
3337 }
|
|
3338
|
|
3339 # if defined(DYNAMIC_ICONV) || defined(PROTO)
|
|
3340 /*
|
|
3341 * Dynamically load the "iconv.dll" on Win32.
|
|
3342 */
|
|
3343
|
|
3344 #ifndef DYNAMIC_ICONV /* just generating prototypes */
|
|
3345 # define HINSTANCE int
|
|
3346 #endif
|
297
|
3347 static HINSTANCE hIconvDLL = 0;
|
|
3348 static HINSTANCE hMsvcrtDLL = 0;
|
7
|
3349
|
|
3350 # ifndef DYNAMIC_ICONV_DLL
|
|
3351 # define DYNAMIC_ICONV_DLL "iconv.dll"
|
|
3352 # define DYNAMIC_ICONV_DLL_ALT "libiconv.dll"
|
|
3353 # endif
|
|
3354 # ifndef DYNAMIC_MSVCRT_DLL
|
|
3355 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll"
|
|
3356 # endif
|
|
3357
|
|
3358 /*
|
|
3359 * Try opening the iconv.dll and return TRUE if iconv() can be used.
|
|
3360 */
|
|
3361 int
|
|
3362 iconv_enabled(verbose)
|
|
3363 int verbose;
|
|
3364 {
|
|
3365 if (hIconvDLL != 0 && hMsvcrtDLL != 0)
|
|
3366 return TRUE;
|
|
3367 hIconvDLL = LoadLibrary(DYNAMIC_ICONV_DLL);
|
|
3368 if (hIconvDLL == 0) /* sometimes it's called libiconv.dll */
|
|
3369 hIconvDLL = LoadLibrary(DYNAMIC_ICONV_DLL_ALT);
|
|
3370 if (hIconvDLL != 0)
|
|
3371 hMsvcrtDLL = LoadLibrary(DYNAMIC_MSVCRT_DLL);
|
|
3372 if (hIconvDLL == 0 || hMsvcrtDLL == 0)
|
|
3373 {
|
|
3374 /* Only give the message when 'verbose' is set, otherwise it might be
|
|
3375 * done whenever a conversion is attempted. */
|
|
3376 if (verbose && p_verbose > 0)
|
292
|
3377 {
|
|
3378 verbose_enter();
|
7
|
3379 EMSG2(_(e_loadlib),
|
|
3380 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL);
|
292
|
3381 verbose_leave();
|
|
3382 }
|
7
|
3383 iconv_end();
|
|
3384 return FALSE;
|
|
3385 }
|
|
3386
|
344
|
3387 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv");
|
|
3388 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open");
|
|
3389 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close");
|
|
3390 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl");
|
|
3391 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno");
|
7
|
3392 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL
|
|
3393 || iconvctl == NULL || iconv_errno == NULL)
|
|
3394 {
|
|
3395 iconv_end();
|
|
3396 if (verbose && p_verbose > 0)
|
292
|
3397 {
|
|
3398 verbose_enter();
|
7
|
3399 EMSG2(_(e_loadfunc), "for libiconv");
|
292
|
3400 verbose_leave();
|
|
3401 }
|
7
|
3402 return FALSE;
|
|
3403 }
|
|
3404 return TRUE;
|
|
3405 }
|
|
3406
|
|
3407 void
|
|
3408 iconv_end()
|
|
3409 {
|
|
3410 /* Don't use iconv() when inputting or outputting characters. */
|
|
3411 if (input_conv.vc_type == CONV_ICONV)
|
|
3412 convert_setup(&input_conv, NULL, NULL);
|
|
3413 if (output_conv.vc_type == CONV_ICONV)
|
|
3414 convert_setup(&output_conv, NULL, NULL);
|
|
3415
|
|
3416 if (hIconvDLL != 0)
|
|
3417 FreeLibrary(hIconvDLL);
|
|
3418 if (hMsvcrtDLL != 0)
|
|
3419 FreeLibrary(hMsvcrtDLL);
|
|
3420 hIconvDLL = 0;
|
|
3421 hMsvcrtDLL = 0;
|
|
3422 }
|
|
3423 # endif /* DYNAMIC_ICONV */
|
|
3424 # endif /* USE_ICONV */
|
|
3425
|
|
3426 #endif /* FEAT_MBYTE */
|
|
3427
|
|
3428 #if defined(FEAT_XIM) || defined(PROTO)
|
|
3429
|
|
3430 # ifdef FEAT_GUI_GTK
|
|
3431 static int xim_has_preediting INIT(= FALSE); /* IM current status */
|
|
3432
|
|
3433 /*
|
|
3434 * Set preedit_start_col to the current cursor position.
|
|
3435 */
|
|
3436 static void
|
|
3437 init_preedit_start_col(void)
|
|
3438 {
|
|
3439 if (State & CMDLINE)
|
|
3440 preedit_start_col = cmdline_getvcol_cursor();
|
|
3441 else if (curwin != NULL)
|
|
3442 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL);
|
|
3443 /* Prevent that preediting marks the buffer as changed. */
|
|
3444 xim_changed_while_preediting = curbuf->b_changed;
|
|
3445 }
|
|
3446 # endif
|
|
3447
|
|
3448 # if defined(HAVE_GTK2) && !defined(PROTO)
|
|
3449
|
|
3450 static int im_is_active = FALSE; /* IM is enabled for current mode */
|
1668
|
3451 static int preedit_is_active = FALSE;
|
7
|
3452 static int im_preedit_cursor = 0; /* cursor offset in characters */
|
|
3453 static int im_preedit_trailing = 0; /* number of characters after cursor */
|
|
3454
|
|
3455 static unsigned long im_commit_handler_id = 0;
|
|
3456 static unsigned int im_activatekey_keyval = GDK_VoidSymbol;
|
|
3457 static unsigned int im_activatekey_state = 0;
|
|
3458
|
|
3459 void
|
|
3460 im_set_active(int active)
|
|
3461 {
|
|
3462 int was_active;
|
|
3463
|
|
3464 was_active = !!im_is_active;
|
|
3465 im_is_active = (active && !p_imdisable);
|
|
3466
|
|
3467 if (im_is_active != was_active)
|
|
3468 xim_reset();
|
|
3469 }
|
|
3470
|
|
3471 void
|
|
3472 xim_set_focus(int focus)
|
|
3473 {
|
|
3474 if (xic != NULL)
|
|
3475 {
|
|
3476 if (focus)
|
|
3477 gtk_im_context_focus_in(xic);
|
|
3478 else
|
|
3479 gtk_im_context_focus_out(xic);
|
|
3480 }
|
|
3481 }
|
|
3482
|
|
3483 void
|
|
3484 im_set_position(int row, int col)
|
|
3485 {
|
|
3486 if (xic != NULL)
|
|
3487 {
|
|
3488 GdkRectangle area;
|
|
3489
|
|
3490 area.x = FILL_X(col);
|
|
3491 area.y = FILL_Y(row);
|
|
3492 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1);
|
|
3493 area.height = gui.char_height;
|
|
3494
|
|
3495 gtk_im_context_set_cursor_location(xic, &area);
|
|
3496 }
|
|
3497 }
|
|
3498
|
|
3499 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */
|
|
3500 void
|
|
3501 xim_set_preedit(void)
|
|
3502 {
|
|
3503 im_set_position(gui.row, gui.col);
|
|
3504 }
|
|
3505 # endif
|
|
3506
|
|
3507 static void
|
|
3508 im_add_to_input(char_u *str, int len)
|
|
3509 {
|
|
3510 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */
|
|
3511 if (input_conv.vc_type != CONV_NONE)
|
|
3512 {
|
|
3513 str = string_convert(&input_conv, str, &len);
|
|
3514 g_return_if_fail(str != NULL);
|
|
3515 }
|
|
3516
|
|
3517 add_to_input_buf_csi(str, len);
|
|
3518
|
|
3519 if (input_conv.vc_type != CONV_NONE)
|
|
3520 vim_free(str);
|
|
3521
|
|
3522 if (p_mh) /* blank out the pointer if necessary */
|
|
3523 gui_mch_mousehide(TRUE);
|
|
3524 }
|
|
3525
|
|
3526 static void
|
|
3527 im_delete_preedit(void)
|
|
3528 {
|
|
3529 char_u bskey[] = {CSI, 'k', 'b'};
|
|
3530 char_u delkey[] = {CSI, 'k', 'D'};
|
|
3531
|
|
3532 if (State & NORMAL)
|
|
3533 {
|
|
3534 im_preedit_cursor = 0;
|
|
3535 return;
|
|
3536 }
|
|
3537 for (; im_preedit_cursor > 0; --im_preedit_cursor)
|
|
3538 add_to_input_buf(bskey, (int)sizeof(bskey));
|
|
3539
|
|
3540 for (; im_preedit_trailing > 0; --im_preedit_trailing)
|
|
3541 add_to_input_buf(delkey, (int)sizeof(delkey));
|
|
3542 }
|
|
3543
|
941
|
3544 /*
|
|
3545 * Move the cursor left by "num_move_back" characters.
|
|
3546 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for
|
|
3547 * these K_LEFT keys.
|
|
3548 */
|
7
|
3549 static void
|
|
3550 im_correct_cursor(int num_move_back)
|
|
3551 {
|
|
3552 char_u backkey[] = {CSI, 'k', 'l'};
|
|
3553
|
|
3554 if (State & NORMAL)
|
|
3555 return;
|
|
3556 # ifdef FEAT_RIGHTLEFT
|
|
3557 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl)
|
|
3558 backkey[2] = 'r';
|
|
3559 # endif
|
|
3560 for (; num_move_back > 0; --num_move_back)
|
|
3561 add_to_input_buf(backkey, (int)sizeof(backkey));
|
|
3562 }
|
|
3563
|
|
3564 static int xim_expected_char = NUL;
|
|
3565 static int xim_ignored_char = FALSE;
|
|
3566
|
|
3567 /*
|
|
3568 * Update the mode and cursor while in an IM callback.
|
|
3569 */
|
|
3570 static void
|
|
3571 im_show_info(void)
|
|
3572 {
|
|
3573 int old_vgetc_busy;
|
822
|
3574
|
7
|
3575 old_vgetc_busy = vgetc_busy;
|
|
3576 vgetc_busy = TRUE;
|
|
3577 showmode();
|
|
3578 vgetc_busy = old_vgetc_busy;
|
|
3579 setcursor();
|
|
3580 out_flush();
|
|
3581 }
|
|
3582
|
|
3583 /*
|
|
3584 * Callback invoked when the user finished preediting.
|
|
3585 * Put the final string into the input buffer.
|
|
3586 */
|
|
3587 /*ARGSUSED0*/
|
|
3588 static void
|
|
3589 im_commit_cb(GtkIMContext *context, const gchar *str, gpointer data)
|
|
3590 {
|
|
3591 int slen = (int)STRLEN(str);
|
|
3592 int add_to_input = TRUE;
|
|
3593 int clen;
|
|
3594 int len = slen;
|
|
3595 int commit_with_preedit = TRUE;
|
|
3596 char_u *im_str, *p;
|
|
3597
|
|
3598 #ifdef XIM_DEBUG
|
|
3599 xim_log("im_commit_cb(): %s\n", str);
|
|
3600 #endif
|
|
3601
|
|
3602 /* The imhangul module doesn't reset the preedit string before
|
|
3603 * committing. Call im_delete_preedit() to work around that. */
|
|
3604 im_delete_preedit();
|
|
3605
|
|
3606 /* Indicate that preediting has finished. */
|
|
3607 if (preedit_start_col == MAXCOL)
|
|
3608 {
|
|
3609 init_preedit_start_col();
|
|
3610 commit_with_preedit = FALSE;
|
|
3611 }
|
|
3612
|
|
3613 /* The thing which setting "preedit_start_col" to MAXCOL means that
|
|
3614 * "preedit_start_col" will be set forcely when calling
|
|
3615 * preedit_changed_cb() next time.
|
|
3616 * "preedit_start_col" should not reset with MAXCOL on this part. Vim
|
|
3617 * is simulating the preediting by using add_to_input_str(). when
|
|
3618 * preedit begin immediately before committed, the typebuf is not
|
|
3619 * flushed to screen, then it can't get correct "preedit_start_col".
|
|
3620 * Thus, it should calculate the cells by adding cells of the committed
|
|
3621 * string. */
|
|
3622 if (input_conv.vc_type != CONV_NONE)
|
|
3623 {
|
|
3624 im_str = string_convert(&input_conv, (char_u *)str, &len);
|
|
3625 g_return_if_fail(im_str != NULL);
|
|
3626 }
|
|
3627 else
|
|
3628 im_str = (char_u *)str;
|
|
3629 clen = 0;
|
474
|
3630 for (p = im_str; p < im_str + len; p += (*mb_ptr2len)(p))
|
7
|
3631 clen += (*mb_ptr2cells)(p);
|
|
3632 if (input_conv.vc_type != CONV_NONE)
|
|
3633 vim_free(im_str);
|
|
3634 preedit_start_col += clen;
|
|
3635
|
|
3636 /* Is this a single character that matches a keypad key that's just
|
|
3637 * been pressed? If so, we don't want it to be entered as such - let
|
|
3638 * us carry on processing the raw keycode so that it may be used in
|
|
3639 * mappings as <kSomething>. */
|
|
3640 if (xim_expected_char != NUL)
|
|
3641 {
|
|
3642 /* We're currently processing a keypad or other special key */
|
|
3643 if (slen == 1 && str[0] == xim_expected_char)
|
|
3644 {
|
|
3645 /* It's a match - don't do it here */
|
|
3646 xim_ignored_char = TRUE;
|
|
3647 add_to_input = FALSE;
|
|
3648 }
|
|
3649 else
|
|
3650 {
|
|
3651 /* Not a match */
|
|
3652 xim_ignored_char = FALSE;
|
|
3653 }
|
|
3654 }
|
|
3655
|
|
3656 if (add_to_input)
|
|
3657 im_add_to_input((char_u *)str, slen);
|
|
3658
|
|
3659 /* Inserting chars while "im_is_active" is set does not cause a change of
|
|
3660 * buffer. When the chars are committed the buffer must be marked as
|
|
3661 * changed. */
|
|
3662 if (!commit_with_preedit)
|
|
3663 preedit_start_col = MAXCOL;
|
|
3664
|
|
3665 /* This flag is used in changed() at next call. */
|
|
3666 xim_changed_while_preediting = TRUE;
|
|
3667
|
|
3668 if (gtk_main_level() > 0)
|
|
3669 gtk_main_quit();
|
|
3670 }
|
|
3671
|
|
3672 /*
|
|
3673 * Callback invoked after start to the preedit.
|
|
3674 */
|
|
3675 /*ARGSUSED*/
|
|
3676 static void
|
|
3677 im_preedit_start_cb(GtkIMContext *context, gpointer data)
|
|
3678 {
|
|
3679 #ifdef XIM_DEBUG
|
|
3680 xim_log("im_preedit_start_cb()\n");
|
|
3681 #endif
|
|
3682
|
|
3683 im_is_active = TRUE;
|
1668
|
3684 preedit_is_active = TRUE;
|
7
|
3685 gui_update_cursor(TRUE, FALSE);
|
1668
|
3686 im_show_info();
|
7
|
3687 }
|
|
3688
|
|
3689 /*
|
|
3690 * Callback invoked after end to the preedit.
|
|
3691 */
|
|
3692 /*ARGSUSED*/
|
|
3693 static void
|
|
3694 im_preedit_end_cb(GtkIMContext *context, gpointer data)
|
|
3695 {
|
|
3696 #ifdef XIM_DEBUG
|
|
3697 xim_log("im_preedit_end_cb()\n");
|
|
3698 #endif
|
|
3699 im_delete_preedit();
|
|
3700
|
|
3701 /* Indicate that preediting has finished */
|
|
3702 preedit_start_col = MAXCOL;
|
|
3703 xim_has_preediting = FALSE;
|
|
3704
|
1620
|
3705 #if 0
|
|
3706 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was
|
1668
|
3707 * switched off unintentionally. We now use preedit_is_active (added by
|
|
3708 * SungHyun Nam). */
|
7
|
3709 im_is_active = FALSE;
|
1620
|
3710 #endif
|
1668
|
3711 preedit_is_active = FALSE;
|
7
|
3712 gui_update_cursor(TRUE, FALSE);
|
|
3713 im_show_info();
|
|
3714 }
|
|
3715
|
|
3716 /*
|
|
3717 * Callback invoked after changes to the preedit string. If the preedit
|
|
3718 * string was empty before, remember the preedit start column so we know
|
|
3719 * where to apply feedback attributes. Delete the previous preedit string
|
|
3720 * if there was one, save the new preedit cursor offset, and put the new
|
|
3721 * string into the input buffer.
|
|
3722 *
|
|
3723 * TODO: The pragmatic "put into input buffer" approach used here has
|
|
3724 * several fundamental problems:
|
|
3725 *
|
|
3726 * - The characters in the preedit string are subject to remapping.
|
|
3727 * That's broken, only the finally committed string should be remapped.
|
|
3728 *
|
|
3729 * - There is a race condition involved: The retrieved value for the
|
|
3730 * current cursor position will be wrong if any unprocessed characters
|
|
3731 * are still queued in the input buffer.
|
|
3732 *
|
|
3733 * - Due to the lack of synchronization between the file buffer in memory
|
|
3734 * and any typed characters, it's practically impossible to implement the
|
|
3735 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM
|
|
3736 * modules for languages such as Thai are likely to rely on this feature
|
|
3737 * for proper operation.
|
|
3738 *
|
|
3739 * Conclusions: I think support for preediting needs to be moved to the
|
|
3740 * core parts of Vim. Ideally, until it has been committed, the preediting
|
|
3741 * string should only be displayed and not affect the buffer content at all.
|
|
3742 * The question how to deal with the synchronization issue still remains.
|
|
3743 * Circumventing the input buffer is probably not desirable. Anyway, I think
|
|
3744 * implementing "retrieve_surrounding" is the only hard problem.
|
|
3745 *
|
|
3746 * One way to solve all of this in a clean manner would be to queue all key
|
|
3747 * press/release events "as is" in the input buffer, and apply the IM filtering
|
|
3748 * at the receiving end of the queue. This, however, would have a rather large
|
|
3749 * impact on the code base. If there is an easy way to force processing of all
|
|
3750 * remaining input from within the "retrieve_surrounding" signal handler, this
|
|
3751 * might not be necessary. Gotta ask on vim-dev for opinions.
|
|
3752 */
|
|
3753 /*ARGSUSED1*/
|
|
3754 static void
|
|
3755 im_preedit_changed_cb(GtkIMContext *context, gpointer data)
|
|
3756 {
|
|
3757 char *preedit_string = NULL;
|
|
3758 int cursor_index = 0;
|
|
3759 int num_move_back = 0;
|
|
3760 char_u *str;
|
|
3761 char_u *p;
|
|
3762 int i;
|
|
3763
|
|
3764 gtk_im_context_get_preedit_string(context,
|
|
3765 &preedit_string, NULL,
|
|
3766 &cursor_index);
|
|
3767
|
|
3768 #ifdef XIM_DEBUG
|
|
3769 xim_log("im_preedit_changed_cb(): %s\n", preedit_string);
|
|
3770 #endif
|
|
3771
|
|
3772 g_return_if_fail(preedit_string != NULL); /* just in case */
|
|
3773
|
|
3774 /* If preedit_start_col is MAXCOL set it to the current cursor position. */
|
|
3775 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0')
|
|
3776 {
|
|
3777 xim_has_preediting = TRUE;
|
|
3778
|
|
3779 /* Urgh, this breaks if the input buffer isn't empty now */
|
|
3780 init_preedit_start_col();
|
|
3781 }
|
|
3782 else if (cursor_index == 0 && preedit_string[0] == '\0')
|
|
3783 {
|
941
|
3784 xim_has_preediting = FALSE;
|
7
|
3785
|
|
3786 /* If at the start position (after typing backspace)
|
|
3787 * preedit_start_col must be reset. */
|
|
3788 preedit_start_col = MAXCOL;
|
|
3789 }
|
|
3790
|
|
3791 im_delete_preedit();
|
|
3792
|
|
3793 /*
|
|
3794 * Compute the end of the preediting area: "preedit_end_col".
|
|
3795 * According to the documentation of gtk_im_context_get_preedit_string(),
|
|
3796 * the cursor_pos output argument returns the offset in bytes. This is
|
|
3797 * unfortunately not true -- real life shows the offset is in characters,
|
|
3798 * and the GTK+ source code agrees with me. Will file a bug later.
|
|
3799 */
|
|
3800 if (preedit_start_col != MAXCOL)
|
|
3801 preedit_end_col = preedit_start_col;
|
|
3802 str = (char_u *)preedit_string;
|
|
3803 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i)
|
|
3804 {
|
|
3805 int is_composing;
|
|
3806
|
|
3807 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p)));
|
|
3808 /*
|
|
3809 * These offsets are used as counters when generating <BS> and <Del>
|
|
3810 * to delete the preedit string. So don't count composing characters
|
|
3811 * unless 'delcombine' is enabled.
|
|
3812 */
|
|
3813 if (!is_composing || p_deco)
|
|
3814 {
|
|
3815 if (i < cursor_index)
|
|
3816 ++im_preedit_cursor;
|
|
3817 else
|
|
3818 ++im_preedit_trailing;
|
|
3819 }
|
|
3820 if (!is_composing && i >= cursor_index)
|
|
3821 {
|
|
3822 /* This is essentially the same as im_preedit_trailing, except
|
|
3823 * composing characters are not counted even if p_deco is set. */
|
|
3824 ++num_move_back;
|
|
3825 }
|
|
3826 if (preedit_start_col != MAXCOL)
|
|
3827 preedit_end_col += utf_ptr2cells(p);
|
|
3828 }
|
|
3829
|
|
3830 if (p > str)
|
|
3831 {
|
|
3832 im_add_to_input(str, (int)(p - str));
|
|
3833 im_correct_cursor(num_move_back);
|
|
3834 }
|
|
3835
|
|
3836 g_free(preedit_string);
|
|
3837
|
|
3838 if (gtk_main_level() > 0)
|
|
3839 gtk_main_quit();
|
|
3840 }
|
|
3841
|
|
3842 /*
|
|
3843 * Translate the Pango attributes at iter to Vim highlighting attributes.
|
|
3844 * Ignore attributes not supported by Vim highlighting. This shouldn't have
|
|
3845 * too much impact -- right now we handle even more attributes than necessary
|
|
3846 * for the IM modules I tested with.
|
|
3847 */
|
|
3848 static int
|
|
3849 translate_pango_attributes(PangoAttrIterator *iter)
|
|
3850 {
|
|
3851 PangoAttribute *attr;
|
|
3852 int char_attr = HL_NORMAL;
|
|
3853
|
|
3854 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE);
|
|
3855 if (attr != NULL && ((PangoAttrInt *)attr)->value
|
|
3856 != (int)PANGO_UNDERLINE_NONE)
|
|
3857 char_attr |= HL_UNDERLINE;
|
|
3858
|
|
3859 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT);
|
|
3860 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD)
|
|
3861 char_attr |= HL_BOLD;
|
|
3862
|
|
3863 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE);
|
|
3864 if (attr != NULL && ((PangoAttrInt *)attr)->value
|
|
3865 != (int)PANGO_STYLE_NORMAL)
|
|
3866 char_attr |= HL_ITALIC;
|
|
3867
|
|
3868 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND);
|
|
3869 if (attr != NULL)
|
|
3870 {
|
|
3871 const PangoColor *color = &((PangoAttrColor *)attr)->color;
|
|
3872
|
|
3873 /* Assume inverse if black background is requested */
|
|
3874 if ((color->red | color->green | color->blue) == 0)
|
|
3875 char_attr |= HL_INVERSE;
|
|
3876 }
|
|
3877
|
|
3878 return char_attr;
|
|
3879 }
|
|
3880
|
|
3881 /*
|
|
3882 * Retrieve the highlighting attributes at column col in the preedit string.
|
|
3883 * Return -1 if not in preediting mode or if col is out of range.
|
|
3884 */
|
|
3885 int
|
|
3886 im_get_feedback_attr(int col)
|
|
3887 {
|
|
3888 char *preedit_string = NULL;
|
|
3889 PangoAttrList *attr_list = NULL;
|
|
3890 int char_attr = -1;
|
|
3891
|
|
3892 if (xic == NULL)
|
|
3893 return char_attr;
|
|
3894
|
|
3895 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL);
|
|
3896
|
|
3897 if (preedit_string != NULL && attr_list != NULL)
|
|
3898 {
|
944
|
3899 int idx;
|
7
|
3900
|
|
3901 /* Get the byte index as used by PangoAttrIterator */
|
944
|
3902 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col)
|
|
3903 idx += utfc_ptr2len((char_u *)preedit_string + idx);
|
|
3904
|
|
3905 if (preedit_string[idx] != '\0')
|
7
|
3906 {
|
|
3907 PangoAttrIterator *iter;
|
|
3908 int start, end;
|
|
3909
|
|
3910 char_attr = HL_NORMAL;
|
|
3911 iter = pango_attr_list_get_iterator(attr_list);
|
|
3912
|
|
3913 /* Extract all relevant attributes from the list. */
|
|
3914 do
|
|
3915 {
|
|
3916 pango_attr_iterator_range(iter, &start, &end);
|
|
3917
|
944
|
3918 if (idx >= start && idx < end)
|
7
|
3919 char_attr |= translate_pango_attributes(iter);
|
|
3920 }
|
|
3921 while (pango_attr_iterator_next(iter));
|
|
3922
|
|
3923 pango_attr_iterator_destroy(iter);
|
|
3924 }
|
|
3925 }
|
|
3926
|
|
3927 if (attr_list != NULL)
|
|
3928 pango_attr_list_unref(attr_list);
|
|
3929 g_free(preedit_string);
|
|
3930
|
|
3931 return char_attr;
|
|
3932 }
|
|
3933
|
|
3934 void
|
|
3935 xim_init(void)
|
|
3936 {
|
|
3937 #ifdef XIM_DEBUG
|
|
3938 xim_log("xim_init()\n");
|
|
3939 #endif
|
|
3940
|
|
3941 g_return_if_fail(gui.drawarea != NULL);
|
|
3942 g_return_if_fail(gui.drawarea->window != NULL);
|
|
3943
|
|
3944 xic = gtk_im_multicontext_new();
|
|
3945 g_object_ref(xic);
|
|
3946
|
|
3947 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit",
|
|
3948 G_CALLBACK(&im_commit_cb), NULL);
|
|
3949 g_signal_connect(G_OBJECT(xic), "preedit_changed",
|
|
3950 G_CALLBACK(&im_preedit_changed_cb), NULL);
|
|
3951 g_signal_connect(G_OBJECT(xic), "preedit_start",
|
|
3952 G_CALLBACK(&im_preedit_start_cb), NULL);
|
|
3953 g_signal_connect(G_OBJECT(xic), "preedit_end",
|
|
3954 G_CALLBACK(&im_preedit_end_cb), NULL);
|
|
3955
|
|
3956 gtk_im_context_set_client_window(xic, gui.drawarea->window);
|
|
3957 }
|
|
3958
|
|
3959 void
|
|
3960 im_shutdown(void)
|
|
3961 {
|
|
3962 #ifdef XIM_DEBUG
|
|
3963 xim_log("im_shutdown()\n");
|
|
3964 #endif
|
|
3965
|
|
3966 if (xic != NULL)
|
|
3967 {
|
|
3968 gtk_im_context_focus_out(xic);
|
|
3969 g_object_unref(xic);
|
|
3970 xic = NULL;
|
|
3971 }
|
|
3972 im_is_active = FALSE;
|
|
3973 im_commit_handler_id = 0;
|
|
3974 preedit_start_col = MAXCOL;
|
|
3975 xim_has_preediting = FALSE;
|
|
3976 }
|
|
3977
|
|
3978 /*
|
|
3979 * Convert the string argument to keyval and state for GdkEventKey.
|
|
3980 * If str is valid return TRUE, otherwise FALSE.
|
|
3981 *
|
|
3982 * See 'imactivatekey' for documentation of the format.
|
|
3983 */
|
|
3984 static int
|
|
3985 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state)
|
|
3986 {
|
|
3987 const char *mods_end;
|
|
3988 unsigned tmp_keyval;
|
|
3989 unsigned tmp_state = 0;
|
|
3990
|
|
3991 mods_end = strrchr(str, '-');
|
|
3992 mods_end = (mods_end != NULL) ? mods_end + 1 : str;
|
|
3993
|
|
3994 /* Parse modifier keys */
|
|
3995 while (str < mods_end)
|
|
3996 switch (*str++)
|
|
3997 {
|
|
3998 case '-': break;
|
|
3999 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break;
|
|
4000 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break;
|
|
4001 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break;
|
|
4002 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break;
|
|
4003 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break;
|
|
4004 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break;
|
|
4005 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break;
|
|
4006 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break;
|
|
4007 default:
|
|
4008 return FALSE;
|
|
4009 }
|
|
4010
|
|
4011 tmp_keyval = gdk_keyval_from_name(str);
|
|
4012
|
|
4013 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol)
|
|
4014 return FALSE;
|
|
4015
|
|
4016 if (keyval != NULL)
|
|
4017 *keyval = tmp_keyval;
|
|
4018 if (state != NULL)
|
|
4019 *state = tmp_state;
|
|
4020
|
|
4021 return TRUE;
|
|
4022 }
|
|
4023
|
|
4024 /*
|
|
4025 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an
|
|
4026 * empty string is also regarded as valid.
|
|
4027 *
|
|
4028 * Note: The numerical key value of p_imak is cached if it was valid; thus
|
|
4029 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever
|
|
4030 * 'imak' changes. This is currently the case but not obvious -- should
|
|
4031 * probably rename the function for clarity.
|
|
4032 */
|
|
4033 int
|
|
4034 im_xim_isvalid_imactivate(void)
|
|
4035 {
|
|
4036 if (p_imak[0] == NUL)
|
|
4037 {
|
|
4038 im_activatekey_keyval = GDK_VoidSymbol;
|
|
4039 im_activatekey_state = 0;
|
|
4040 return TRUE;
|
|
4041 }
|
|
4042
|
|
4043 return im_string_to_keyval((const char *)p_imak,
|
|
4044 &im_activatekey_keyval,
|
|
4045 &im_activatekey_state);
|
|
4046 }
|
|
4047
|
|
4048 static void
|
|
4049 im_synthesize_keypress(unsigned int keyval, unsigned int state)
|
|
4050 {
|
|
4051 GdkEventKey *event;
|
|
4052
|
|
4053 # ifdef HAVE_GTK_MULTIHEAD
|
|
4054 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS);
|
|
4055 g_object_ref(gui.drawarea->window); /* unreffed by gdk_event_free() */
|
|
4056 # else
|
|
4057 event = (GdkEventKey *)g_malloc0((gulong)sizeof(GdkEvent));
|
|
4058 event->type = GDK_KEY_PRESS;
|
|
4059 # endif
|
|
4060 event->window = gui.drawarea->window;
|
|
4061 event->send_event = TRUE;
|
|
4062 event->time = GDK_CURRENT_TIME;
|
|
4063 event->state = state;
|
|
4064 event->keyval = keyval;
|
|
4065 event->hardware_keycode = /* needed for XIM */
|
|
4066 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval);
|
|
4067 event->length = 0;
|
|
4068 event->string = NULL;
|
|
4069
|
|
4070 gtk_im_context_filter_keypress(xic, event);
|
|
4071
|
|
4072 /* For consistency, also send the corresponding release event. */
|
|
4073 event->type = GDK_KEY_RELEASE;
|
|
4074 event->send_event = FALSE;
|
|
4075 gtk_im_context_filter_keypress(xic, event);
|
|
4076
|
|
4077 # ifdef HAVE_GTK_MULTIHEAD
|
|
4078 gdk_event_free((GdkEvent *)event);
|
|
4079 # else
|
|
4080 g_free(event);
|
|
4081 # endif
|
|
4082 }
|
|
4083
|
|
4084 void
|
|
4085 xim_reset(void)
|
|
4086 {
|
|
4087 if (xic != NULL)
|
|
4088 {
|
|
4089 /*
|
|
4090 * The third-party imhangul module (and maybe others too) ignores
|
|
4091 * gtk_im_context_reset() or at least doesn't reset the active state.
|
|
4092 * Thus sending imactivatekey would turn it off if it was on before,
|
|
4093 * which is clearly not what we want. Fortunately we can work around
|
|
4094 * that for imhangul by sending GDK_Escape, but I don't know if it
|
|
4095 * works with all IM modules that support an activation key :/
|
|
4096 *
|
|
4097 * An alternative approach would be to destroy the IM context and
|
|
4098 * recreate it. But that means loading/unloading the IM module on
|
|
4099 * every mode switch, which causes a quite noticable delay even on
|
|
4100 * my rather fast box...
|
|
4101 * *
|
|
4102 * Moreover, there are some XIM which cannot respond to
|
|
4103 * im_synthesize_keypress(). we hope that they reset by
|
|
4104 * xim_shutdown().
|
|
4105 */
|
|
4106 if (im_activatekey_keyval != GDK_VoidSymbol && im_is_active)
|
|
4107 im_synthesize_keypress(GDK_Escape, 0U);
|
|
4108
|
|
4109 gtk_im_context_reset(xic);
|
|
4110
|
|
4111 /*
|
|
4112 * HACK for Ami: This sequence of function calls makes Ami handle
|
1205
|
4113 * the IM reset graciously, without breaking loads of other stuff.
|
7
|
4114 * It seems to force English mode as well, which is exactly what we
|
|
4115 * want because it makes the Ami status display work reliably.
|
|
4116 */
|
|
4117 gtk_im_context_set_use_preedit(xic, FALSE);
|
|
4118
|
|
4119 if (p_imdisable)
|
|
4120 im_shutdown();
|
|
4121 else
|
|
4122 {
|
|
4123 gtk_im_context_set_use_preedit(xic, TRUE);
|
|
4124 xim_set_focus(gui.in_focus);
|
|
4125
|
|
4126 if (im_activatekey_keyval != GDK_VoidSymbol)
|
|
4127 {
|
|
4128 if (im_is_active)
|
|
4129 {
|
|
4130 g_signal_handler_block(xic, im_commit_handler_id);
|
|
4131 im_synthesize_keypress(im_activatekey_keyval,
|
|
4132 im_activatekey_state);
|
|
4133 g_signal_handler_unblock(xic, im_commit_handler_id);
|
|
4134 }
|
|
4135 }
|
|
4136 else
|
|
4137 {
|
|
4138 im_shutdown();
|
|
4139 xim_init();
|
|
4140 xim_set_focus(gui.in_focus);
|
|
4141 }
|
|
4142 }
|
|
4143 }
|
|
4144
|
|
4145 preedit_start_col = MAXCOL;
|
|
4146 xim_has_preediting = FALSE;
|
|
4147 }
|
|
4148
|
|
4149 int
|
|
4150 xim_queue_key_press_event(GdkEventKey *event, int down)
|
|
4151 {
|
|
4152 if (down)
|
|
4153 {
|
|
4154 /*
|
|
4155 * Workaround GTK2 XIM 'feature' that always converts keypad keys to
|
|
4156 * chars., even when not part of an IM sequence (ref. feature of
|
|
4157 * gdk/gdkkeyuni.c).
|
|
4158 * Flag any keypad keys that might represent a single char.
|
|
4159 * If this (on its own - i.e., not part of an IM sequence) is
|
|
4160 * committed while we're processing one of these keys, we can ignore
|
|
4161 * that commit and go ahead & process it ourselves. That way we can
|
|
4162 * still distinguish keypad keys for use in mappings.
|
1620
|
4163 * Also add GDK_space to make <S-Space> work.
|
7
|
4164 */
|
|
4165 switch (event->keyval)
|
|
4166 {
|
|
4167 case GDK_KP_Add: xim_expected_char = '+'; break;
|
|
4168 case GDK_KP_Subtract: xim_expected_char = '-'; break;
|
|
4169 case GDK_KP_Divide: xim_expected_char = '/'; break;
|
|
4170 case GDK_KP_Multiply: xim_expected_char = '*'; break;
|
|
4171 case GDK_KP_Decimal: xim_expected_char = '.'; break;
|
|
4172 case GDK_KP_Equal: xim_expected_char = '='; break;
|
|
4173 case GDK_KP_0: xim_expected_char = '0'; break;
|
|
4174 case GDK_KP_1: xim_expected_char = '1'; break;
|
|
4175 case GDK_KP_2: xim_expected_char = '2'; break;
|
|
4176 case GDK_KP_3: xim_expected_char = '3'; break;
|
|
4177 case GDK_KP_4: xim_expected_char = '4'; break;
|
|
4178 case GDK_KP_5: xim_expected_char = '5'; break;
|
|
4179 case GDK_KP_6: xim_expected_char = '6'; break;
|
|
4180 case GDK_KP_7: xim_expected_char = '7'; break;
|
|
4181 case GDK_KP_8: xim_expected_char = '8'; break;
|
|
4182 case GDK_KP_9: xim_expected_char = '9'; break;
|
1620
|
4183 case GDK_space: xim_expected_char = ' '; break;
|
7
|
4184 default: xim_expected_char = NUL;
|
|
4185 }
|
|
4186 xim_ignored_char = FALSE;
|
|
4187 }
|
|
4188
|
|
4189 /*
|
|
4190 * When typing fFtT, XIM may be activated. Thus it must pass
|
|
4191 * gtk_im_context_filter_keypress() in Normal mode.
|
|
4192 * And while doing :sh too.
|
|
4193 */
|
|
4194 if (xic != NULL && !p_imdisable
|
|
4195 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0)
|
|
4196 {
|
|
4197 /*
|
|
4198 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is
|
|
4199 * always aware of the current status of IM, and can even emulate
|
|
4200 * the activation key for modules that don't support one.
|
|
4201 */
|
|
4202 if (event->keyval == im_activatekey_keyval
|
|
4203 && (event->state & im_activatekey_state) == im_activatekey_state)
|
|
4204 {
|
|
4205 unsigned int state_mask;
|
|
4206
|
|
4207 /* Require the state of the 3 most used modifiers to match exactly.
|
|
4208 * Otherwise e.g. <S-C-space> would be unusable for other purposes
|
|
4209 * if the IM activate key is <S-space>. */
|
|
4210 state_mask = im_activatekey_state;
|
|
4211 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK
|
|
4212 | (int)GDK_MOD1_MASK);
|
|
4213
|
|
4214 if ((event->state & state_mask) != im_activatekey_state)
|
|
4215 return FALSE;
|
|
4216
|
|
4217 /* Don't send it a second time on GDK_KEY_RELEASE. */
|
|
4218 if (event->type != GDK_KEY_PRESS)
|
|
4219 return TRUE;
|
|
4220
|
781
|
4221 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE))
|
7
|
4222 {
|
|
4223 im_set_active(FALSE);
|
|
4224
|
|
4225 /* ":lmap" mappings exists, toggle use of mappings. */
|
|
4226 State ^= LANGMAP;
|
|
4227 if (State & LANGMAP)
|
|
4228 {
|
|
4229 curbuf->b_p_iminsert = B_IMODE_NONE;
|
|
4230 State &= ~LANGMAP;
|
|
4231 }
|
|
4232 else
|
|
4233 {
|
|
4234 curbuf->b_p_iminsert = B_IMODE_LMAP;
|
|
4235 State |= LANGMAP;
|
|
4236 }
|
|
4237 return TRUE;
|
|
4238 }
|
|
4239
|
|
4240 return gtk_im_context_filter_keypress(xic, event);
|
|
4241 }
|
|
4242
|
|
4243 /* Don't filter events through the IM context if IM isn't active
|
|
4244 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module
|
|
4245 * not doing anything before the activation key was sent. */
|
|
4246 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active)
|
|
4247 {
|
|
4248 int imresult = gtk_im_context_filter_keypress(xic, event);
|
|
4249
|
|
4250 /* Some XIM send following sequence:
|
|
4251 * 1. preedited string.
|
|
4252 * 2. committed string.
|
|
4253 * 3. line changed key.
|
|
4254 * 4. preedited string.
|
|
4255 * 5. remove preedited string.
|
|
4256 * if 3, Vim can't move back the above line for 5.
|
|
4257 * thus, this part should not parse the key. */
|
|
4258 if (!imresult && preedit_start_col != MAXCOL
|
|
4259 && event->keyval == GDK_Return)
|
|
4260 {
|
|
4261 im_synthesize_keypress(GDK_Return, 0U);
|
|
4262 return FALSE;
|
|
4263 }
|
|
4264
|
|
4265 /* If XIM tried to commit a keypad key as a single char.,
|
|
4266 * ignore it so we can use the keypad key 'raw', for mappings. */
|
|
4267 if (xim_expected_char != NUL && xim_ignored_char)
|
|
4268 /* We had a keypad key, and XIM tried to thieve it */
|
|
4269 return FALSE;
|
|
4270
|
|
4271 /* Normal processing */
|
|
4272 return imresult;
|
|
4273 }
|
|
4274 }
|
|
4275
|
|
4276 return FALSE;
|
|
4277 }
|
|
4278
|
|
4279 int
|
|
4280 im_get_status(void)
|
|
4281 {
|
|
4282 return im_is_active;
|
|
4283 }
|
|
4284
|
|
4285 # else /* !HAVE_GTK2 */
|
|
4286
|
|
4287 static int xim_is_active = FALSE; /* XIM should be active in the current
|
|
4288 mode */
|
|
4289 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */
|
|
4290 #ifdef FEAT_GUI_X11
|
|
4291 static XIMStyle input_style;
|
|
4292 static int status_area_enabled = TRUE;
|
|
4293 #endif
|
|
4294
|
|
4295 #ifdef FEAT_GUI_GTK
|
|
4296 # ifdef WIN3264
|
|
4297 # include <gdk/gdkwin32.h>
|
|
4298 # else
|
|
4299 # include <gdk/gdkx.h>
|
|
4300 # endif
|
|
4301 #else
|
|
4302 # ifdef PROTO
|
|
4303 /* Define a few things to be able to generate prototypes while not configured
|
|
4304 * for GTK. */
|
|
4305 # define GSList int
|
|
4306 # define gboolean int
|
|
4307 typedef int GdkEvent;
|
|
4308 typedef int GdkEventKey;
|
|
4309 # define GdkIC int
|
|
4310 # endif
|
|
4311 #endif
|
|
4312
|
574
|
4313 #if defined(FEAT_GUI_GTK) || defined(PROTO)
|
7
|
4314 static int preedit_buf_len = 0;
|
|
4315 static int xim_can_preediting INIT(= FALSE); /* XIM in showmode() */
|
|
4316 static int xim_input_style;
|
|
4317 #ifndef FEAT_GUI_GTK
|
|
4318 # define gboolean int
|
|
4319 #endif
|
|
4320 static gboolean use_status_area = 0;
|
|
4321
|
|
4322 static int im_xim_str2keycode __ARGS((unsigned int *code, unsigned int *state));
|
|
4323 static void im_xim_send_event_imactivate __ARGS((void));
|
|
4324
|
|
4325 /*
|
|
4326 * Convert string to keycode and state for XKeyEvent.
|
|
4327 * When string is valid return OK, when invalid return FAIL.
|
|
4328 *
|
|
4329 * See 'imactivatekey' documentation for the format.
|
|
4330 */
|
|
4331 static int
|
|
4332 im_xim_str2keycode(code, state)
|
|
4333 unsigned int *code;
|
|
4334 unsigned int *state;
|
|
4335 {
|
|
4336 int retval = OK;
|
|
4337 int len;
|
|
4338 unsigned keycode = 0, keystate = 0;
|
|
4339 Window window;
|
|
4340 Display *display;
|
|
4341 char_u *flag_end;
|
|
4342 char_u *str;
|
|
4343
|
|
4344 if (*p_imak != NUL)
|
|
4345 {
|
|
4346 len = STRLEN(p_imak);
|
|
4347 for (flag_end = p_imak + len - 1;
|
|
4348 flag_end > p_imak && *flag_end != '-'; --flag_end)
|
|
4349 ;
|
|
4350
|
|
4351 /* Parse modifier keys */
|
|
4352 for (str = p_imak; str < flag_end; ++str)
|
|
4353 {
|
|
4354 switch (*str)
|
|
4355 {
|
|
4356 case 's': case 'S':
|
|
4357 keystate |= ShiftMask;
|
|
4358 break;
|
|
4359 case 'l': case 'L':
|
|
4360 keystate |= LockMask;
|
|
4361 break;
|
|
4362 case 'c': case 'C':
|
|
4363 keystate |= ControlMask;
|
|
4364 break;
|
|
4365 case '1':
|
|
4366 keystate |= Mod1Mask;
|
|
4367 break;
|
|
4368 case '2':
|
|
4369 keystate |= Mod2Mask;
|
|
4370 break;
|
|
4371 case '3':
|
|
4372 keystate |= Mod3Mask;
|
|
4373 break;
|
|
4374 case '4':
|
|
4375 keystate |= Mod4Mask;
|
|
4376 break;
|
|
4377 case '5':
|
|
4378 keystate |= Mod5Mask;
|
|
4379 break;
|
|
4380 case '-':
|
|
4381 break;
|
|
4382 default:
|
|
4383 retval = FAIL;
|
|
4384 }
|
|
4385 }
|
|
4386 if (*str == '-')
|
|
4387 ++str;
|
|
4388
|
|
4389 /* Get keycode from string. */
|
|
4390 gui_get_x11_windis(&window, &display);
|
|
4391 if (display)
|
|
4392 keycode = XKeysymToKeycode(display, XStringToKeysym((char *)str));
|
|
4393 if (keycode == 0)
|
|
4394 retval = FAIL;
|
|
4395
|
|
4396 if (code != NULL)
|
|
4397 *code = keycode;
|
|
4398 if (state != NULL)
|
|
4399 *state = keystate;
|
|
4400 }
|
|
4401 return retval;
|
|
4402 }
|
|
4403
|
|
4404 static void
|
|
4405 im_xim_send_event_imactivate()
|
|
4406 {
|
|
4407 /* Force turn on preedit state by symulate keypress event.
|
|
4408 * Keycode and state is specified by 'imactivatekey'.
|
|
4409 */
|
|
4410 XKeyEvent ev;
|
|
4411
|
|
4412 gui_get_x11_windis(&ev.window, &ev.display);
|
|
4413 ev.root = RootWindow(ev.display, DefaultScreen(ev.display));
|
|
4414 ev.subwindow = None;
|
|
4415 ev.time = CurrentTime;
|
|
4416 ev.x = 1;
|
|
4417 ev.y = 1;
|
|
4418 ev.x_root = 1;
|
|
4419 ev.y_root = 1;
|
|
4420 ev.same_screen = 1;
|
|
4421 ev.type = KeyPress;
|
|
4422 if (im_xim_str2keycode(&ev.keycode, &ev.state) == OK)
|
|
4423 XSendEvent(ev.display, ev.window, 1, KeyPressMask, (XEvent*)&ev);
|
|
4424 }
|
|
4425
|
|
4426 /*
|
|
4427 * Return TRUE if 'imactivatekey' has a valid value.
|
|
4428 */
|
|
4429 int
|
|
4430 im_xim_isvalid_imactivate()
|
|
4431 {
|
|
4432 return im_xim_str2keycode(NULL, NULL) == OK;
|
|
4433 }
|
|
4434 #endif /* FEAT_GUI_GTK */
|
|
4435
|
|
4436 /*
|
|
4437 * Switch using XIM on/off. This is used by the code that changes "State".
|
|
4438 */
|
|
4439 void
|
|
4440 im_set_active(active)
|
|
4441 int active;
|
|
4442 {
|
|
4443 if (xic == NULL)
|
|
4444 return;
|
|
4445
|
|
4446 /* If 'imdisable' is set, XIM is never active. */
|
|
4447 if (p_imdisable)
|
|
4448 active = FALSE;
|
574
|
4449 #if !defined (FEAT_GUI_GTK)
|
7
|
4450 else if (input_style & XIMPreeditPosition)
|
|
4451 /* There is a problem in switching XIM off when preediting is used,
|
|
4452 * and it is not clear how this can be solved. For now, keep XIM on
|
|
4453 * all the time, like it was done in Vim 5.8. */
|
|
4454 active = TRUE;
|
|
4455 #endif
|
|
4456
|
|
4457 /* Remember the active state, it is needed when Vim gets keyboard focus. */
|
|
4458 xim_is_active = active;
|
|
4459
|
|
4460 #ifdef FEAT_GUI_GTK
|
|
4461 /* When 'imactivatekey' has valid key-string, try to control XIM preedit
|
|
4462 * state. When 'imactivatekey' has no or invalid string, try old XIM
|
|
4463 * focus control.
|
|
4464 */
|
|
4465 if (*p_imak != NUL)
|
|
4466 {
|
|
4467 /* BASIC STRATEGY:
|
|
4468 * Destroy old Input Context (XIC), and create new one. New XIC
|
|
4469 * would have a state of preedit that is off. When argument:active
|
|
4470 * is false, that's all. Else argument:active is true, send a key
|
|
4471 * event specified by 'imactivatekey' to activate XIM preedit state.
|
|
4472 */
|
|
4473
|
|
4474 xim_is_active = TRUE; /* Disable old XIM focus control */
|
|
4475 /* If we can monitor preedit state with preedit callback functions,
|
|
4476 * try least creation of new XIC.
|
|
4477 */
|
|
4478 if (xim_input_style & (int)GDK_IM_PREEDIT_CALLBACKS)
|
|
4479 {
|
|
4480 if (xim_can_preediting && !active)
|
|
4481 {
|
|
4482 /* Force turn off preedit state. With some IM
|
|
4483 * implementations, we cannot turn off preedit state by
|
|
4484 * symulate keypress event. It is why using such a method
|
|
4485 * that destroy old IC (input context), and create new one.
|
|
4486 * When create new IC, its preedit state is usually off.
|
|
4487 */
|
|
4488 xim_reset();
|
|
4489 xim_set_focus(FALSE);
|
|
4490 gdk_ic_destroy(xic);
|
|
4491 xim_init();
|
|
4492 xim_can_preediting = FALSE;
|
|
4493 }
|
|
4494 else if (!xim_can_preediting && active)
|
|
4495 im_xim_send_event_imactivate();
|
|
4496 }
|
|
4497 else
|
|
4498 {
|
|
4499 /* First, force destroy old IC, and create new one. It
|
|
4500 * symulates "turning off preedit state".
|
|
4501 */
|
|
4502 xim_set_focus(FALSE);
|
|
4503 gdk_ic_destroy(xic);
|
|
4504 xim_init();
|
|
4505 xim_can_preediting = FALSE;
|
|
4506
|
|
4507 /* 2nd, when requested to activate IM, symulate this by sending
|
|
4508 * the event.
|
|
4509 */
|
|
4510 if (active)
|
|
4511 {
|
|
4512 im_xim_send_event_imactivate();
|
|
4513 xim_can_preediting = TRUE;
|
|
4514 }
|
|
4515 }
|
|
4516 }
|
|
4517 else
|
|
4518 {
|
|
4519 # ifndef XIMPreeditUnKnown
|
|
4520 /* X11R5 doesn't have these, it looks safe enough to define here. */
|
|
4521 typedef unsigned long XIMPreeditState;
|
|
4522 # define XIMPreeditUnKnown 0L
|
|
4523 # define XIMPreeditEnable 1L
|
|
4524 # define XIMPreeditDisable (1L<<1)
|
|
4525 # define XNPreeditState "preeditState"
|
|
4526 # endif
|
|
4527 XIMPreeditState preedit_state = XIMPreeditUnKnown;
|
|
4528 XVaNestedList preedit_attr;
|
|
4529 XIC pxic;
|
|
4530
|
|
4531 preedit_attr = XVaCreateNestedList(0,
|
|
4532 XNPreeditState, &preedit_state,
|
|
4533 NULL);
|
|
4534 pxic = ((GdkICPrivate *)xic)->xic;
|
|
4535
|
|
4536 if (!XGetICValues(pxic, XNPreeditAttributes, preedit_attr, NULL))
|
|
4537 {
|
|
4538 XFree(preedit_attr);
|
|
4539 preedit_attr = XVaCreateNestedList(0,
|
|
4540 XNPreeditState,
|
|
4541 active ? XIMPreeditEnable : XIMPreeditDisable,
|
|
4542 NULL);
|
|
4543 XSetICValues(pxic, XNPreeditAttributes, preedit_attr, NULL);
|
|
4544 xim_can_preediting = active;
|
|
4545 xim_is_active = active;
|
|
4546 }
|
|
4547 XFree(preedit_attr);
|
|
4548 }
|
|
4549 if (xim_input_style & XIMPreeditCallbacks)
|
|
4550 {
|
|
4551 preedit_buf_len = 0;
|
|
4552 init_preedit_start_col();
|
|
4553 }
|
|
4554 #else
|
|
4555 # if 0
|
|
4556 /* When had tested kinput2 + canna + Athena GUI version with
|
|
4557 * 'imactivatekey' is "s-space", im_xim_send_event_imactivate() did not
|
|
4558 * work correctly. It just inserted one space. I don't know why we
|
|
4559 * couldn't switch state of XIM preediting. This is reason why these
|
|
4560 * codes are commented out.
|
|
4561 */
|
|
4562 /* First, force destroy old IC, and create new one. It symulates
|
|
4563 * "turning off preedit state".
|
|
4564 */
|
|
4565 xim_set_focus(FALSE);
|
|
4566 XDestroyIC(xic);
|
|
4567 xic = NULL;
|
|
4568 xim_init();
|
|
4569
|
|
4570 /* 2nd, when requested to activate IM, symulate this by sending the
|
|
4571 * event.
|
|
4572 */
|
|
4573 if (active)
|
|
4574 im_xim_send_event_imactivate();
|
|
4575 # endif
|
|
4576 #endif
|
|
4577 xim_set_preedit();
|
|
4578 }
|
|
4579
|
|
4580 /*
|
|
4581 * Adjust using XIM for gaining or losing keyboard focus. Also called when
|
|
4582 * "xim_is_active" changes.
|
|
4583 */
|
|
4584 void
|
|
4585 xim_set_focus(focus)
|
|
4586 int focus;
|
|
4587 {
|
|
4588 if (xic == NULL)
|
|
4589 return;
|
|
4590
|
|
4591 /*
|
|
4592 * XIM only gets focus when the Vim window has keyboard focus and XIM has
|
|
4593 * been set active for the current mode.
|
|
4594 */
|
|
4595 if (focus && xim_is_active)
|
|
4596 {
|
|
4597 if (!xim_has_focus)
|
|
4598 {
|
|
4599 xim_has_focus = TRUE;
|
|
4600 #ifdef FEAT_GUI_GTK
|
|
4601 gdk_im_begin(xic, gui.drawarea->window);
|
|
4602 #else
|
|
4603 XSetICFocus(xic);
|
|
4604 #endif
|
|
4605 }
|
|
4606 }
|
|
4607 else
|
|
4608 {
|
|
4609 if (xim_has_focus)
|
|
4610 {
|
|
4611 xim_has_focus = FALSE;
|
|
4612 #ifdef FEAT_GUI_GTK
|
|
4613 gdk_im_end();
|
|
4614 #else
|
|
4615 XUnsetICFocus(xic);
|
|
4616 #endif
|
|
4617 }
|
|
4618 }
|
|
4619 }
|
|
4620
|
|
4621 /*ARGSUSED*/
|
|
4622 void
|
|
4623 im_set_position(row, col)
|
|
4624 int row;
|
|
4625 int col;
|
|
4626 {
|
|
4627 xim_set_preedit();
|
|
4628 }
|
|
4629
|
|
4630 /*
|
|
4631 * Set the XIM to the current cursor position.
|
|
4632 */
|
|
4633 void
|
|
4634 xim_set_preedit()
|
|
4635 {
|
|
4636 if (xic == NULL)
|
|
4637 return;
|
|
4638
|
|
4639 xim_set_focus(TRUE);
|
|
4640
|
|
4641 #ifdef FEAT_GUI_GTK
|
|
4642 if (gdk_im_ready())
|
|
4643 {
|
|
4644 int attrmask;
|
|
4645 GdkICAttr *attr;
|
|
4646
|
|
4647 if (!xic_attr)
|
|
4648 return;
|
|
4649
|
|
4650 attr = xic_attr;
|
|
4651 attrmask = 0;
|
|
4652
|
|
4653 # ifdef FEAT_XFONTSET
|
|
4654 if ((xim_input_style & (int)GDK_IM_PREEDIT_POSITION)
|
|
4655 && gui.fontset != NOFONTSET
|
|
4656 && gui.fontset->type == GDK_FONT_FONTSET)
|
|
4657 {
|
|
4658 if (!xim_has_focus)
|
|
4659 {
|
|
4660 if (attr->spot_location.y >= 0)
|
|
4661 {
|
|
4662 attr->spot_location.x = 0;
|
|
4663 attr->spot_location.y = -100;
|
|
4664 attrmask |= (int)GDK_IC_SPOT_LOCATION;
|
|
4665 }
|
|
4666 }
|
|
4667 else
|
|
4668 {
|
|
4669 gint width, height;
|
|
4670
|
|
4671 if (attr->spot_location.x != TEXT_X(gui.col)
|
|
4672 || attr->spot_location.y != TEXT_Y(gui.row))
|
|
4673 {
|
|
4674 attr->spot_location.x = TEXT_X(gui.col);
|
|
4675 attr->spot_location.y = TEXT_Y(gui.row);
|
|
4676 attrmask |= (int)GDK_IC_SPOT_LOCATION;
|
|
4677 }
|
|
4678
|
|
4679 gdk_window_get_size(gui.drawarea->window, &width, &height);
|
|
4680 width -= 2 * gui.border_offset;
|
|
4681 height -= 2 * gui.border_offset;
|
|
4682 if (xim_input_style & (int)GDK_IM_STATUS_AREA)
|
|
4683 height -= gui.char_height;
|
|
4684 if (attr->preedit_area.width != width
|
|
4685 || attr->preedit_area.height != height)
|
|
4686 {
|
|
4687 attr->preedit_area.x = gui.border_offset;
|
|
4688 attr->preedit_area.y = gui.border_offset;
|
|
4689 attr->preedit_area.width = width;
|
|
4690 attr->preedit_area.height = height;
|
|
4691 attrmask |= (int)GDK_IC_PREEDIT_AREA;
|
|
4692 }
|
|
4693
|
|
4694 if (attr->preedit_fontset != gui.current_font)
|
|
4695 {
|
|
4696 attr->preedit_fontset = gui.current_font;
|
|
4697 attrmask |= (int)GDK_IC_PREEDIT_FONTSET;
|
|
4698 }
|
|
4699 }
|
|
4700 }
|
|
4701 # endif /* FEAT_XFONTSET */
|
|
4702
|
|
4703 if (xim_fg_color == INVALCOLOR)
|
|
4704 {
|
|
4705 xim_fg_color = gui.def_norm_pixel;
|
|
4706 xim_bg_color = gui.def_back_pixel;
|
|
4707 }
|
|
4708 if (attr->preedit_foreground.pixel != xim_fg_color)
|
|
4709 {
|
|
4710 attr->preedit_foreground.pixel = xim_fg_color;
|
|
4711 attrmask |= (int)GDK_IC_PREEDIT_FOREGROUND;
|
|
4712 }
|
|
4713 if (attr->preedit_background.pixel != xim_bg_color)
|
|
4714 {
|
|
4715 attr->preedit_background.pixel = xim_bg_color;
|
|
4716 attrmask |= (int)GDK_IC_PREEDIT_BACKGROUND;
|
|
4717 }
|
|
4718
|
|
4719 if (attrmask != 0)
|
|
4720 gdk_ic_set_attr(xic, attr, (GdkICAttributesType)attrmask);
|
|
4721 }
|
|
4722 #else /* FEAT_GUI_GTK */
|
|
4723 {
|
|
4724 XVaNestedList attr_list;
|
|
4725 XRectangle spot_area;
|
|
4726 XPoint over_spot;
|
|
4727 int line_space;
|
|
4728
|
|
4729 if (!xim_has_focus)
|
|
4730 {
|
|
4731 /* hide XIM cursor */
|
|
4732 over_spot.x = 0;
|
|
4733 over_spot.y = -100; /* arbitrary invisible position */
|
|
4734 attr_list = (XVaNestedList) XVaCreateNestedList(0,
|
|
4735 XNSpotLocation,
|
|
4736 &over_spot,
|
|
4737 NULL);
|
|
4738 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL);
|
|
4739 XFree(attr_list);
|
|
4740 return;
|
|
4741 }
|
|
4742
|
|
4743 if (input_style & XIMPreeditPosition)
|
|
4744 {
|
|
4745 if (xim_fg_color == INVALCOLOR)
|
|
4746 {
|
|
4747 xim_fg_color = gui.def_norm_pixel;
|
|
4748 xim_bg_color = gui.def_back_pixel;
|
|
4749 }
|
|
4750 over_spot.x = TEXT_X(gui.col);
|
|
4751 over_spot.y = TEXT_Y(gui.row);
|
|
4752 spot_area.x = 0;
|
|
4753 spot_area.y = 0;
|
|
4754 spot_area.height = gui.char_height * Rows;
|
|
4755 spot_area.width = gui.char_width * Columns;
|
|
4756 line_space = gui.char_height;
|
|
4757 attr_list = (XVaNestedList) XVaCreateNestedList(0,
|
|
4758 XNSpotLocation, &over_spot,
|
|
4759 XNForeground, (Pixel) xim_fg_color,
|
|
4760 XNBackground, (Pixel) xim_bg_color,
|
|
4761 XNArea, &spot_area,
|
|
4762 XNLineSpace, line_space,
|
|
4763 NULL);
|
|
4764 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL))
|
|
4765 EMSG(_("E284: Cannot set IC values"));
|
|
4766 XFree(attr_list);
|
|
4767 }
|
|
4768 }
|
|
4769 #endif /* FEAT_GUI_GTK */
|
|
4770 }
|
|
4771
|
|
4772 /*
|
|
4773 * Set up the status area.
|
|
4774 *
|
|
4775 * This should use a separate Widget, but that seems not possible, because
|
|
4776 * preedit_area and status_area should be set to the same window as for the
|
|
4777 * text input. Unfortunately this means the status area pollutes the text
|
|
4778 * window...
|
|
4779 */
|
|
4780 void
|
|
4781 xim_set_status_area()
|
|
4782 {
|
|
4783 if (xic == NULL)
|
|
4784 return;
|
|
4785
|
|
4786 #ifdef FEAT_GUI_GTK
|
|
4787 # if defined(FEAT_XFONTSET)
|
|
4788 if (use_status_area)
|
|
4789 {
|
|
4790 GdkICAttr *attr;
|
|
4791 int style;
|
|
4792 gint width, height;
|
|
4793 GtkWidget *widget;
|
|
4794 int attrmask;
|
|
4795
|
|
4796 if (!xic_attr)
|
|
4797 return;
|
|
4798
|
|
4799 attr = xic_attr;
|
|
4800 attrmask = 0;
|
|
4801 style = (int)gdk_ic_get_style(xic);
|
|
4802 if ((style & (int)GDK_IM_STATUS_MASK) == (int)GDK_IM_STATUS_AREA)
|
|
4803 {
|
|
4804 if (gui.fontset != NOFONTSET
|
|
4805 && gui.fontset->type == GDK_FONT_FONTSET)
|
|
4806 {
|
|
4807 widget = gui.mainwin;
|
|
4808 gdk_window_get_size(widget->window, &width, &height);
|
|
4809
|
|
4810 attrmask |= (int)GDK_IC_STATUS_AREA;
|
|
4811 attr->status_area.x = 0;
|
|
4812 attr->status_area.y = height - gui.char_height - 1;
|
|
4813 attr->status_area.width = width;
|
|
4814 attr->status_area.height = gui.char_height;
|
|
4815 }
|
|
4816 }
|
|
4817 if (attrmask != 0)
|
|
4818 gdk_ic_set_attr(xic, attr, (GdkICAttributesType)attrmask);
|
|
4819 }
|
|
4820 # endif
|
|
4821 #else
|
|
4822 {
|
|
4823 XVaNestedList preedit_list = 0, status_list = 0, list = 0;
|
|
4824 XRectangle pre_area, status_area;
|
|
4825
|
|
4826 if (input_style & XIMStatusArea)
|
|
4827 {
|
|
4828 if (input_style & XIMPreeditArea)
|
|
4829 {
|
|
4830 XRectangle *needed_rect;
|
|
4831
|
|
4832 /* to get status_area width */
|
|
4833 status_list = XVaCreateNestedList(0, XNAreaNeeded,
|
|
4834 &needed_rect, NULL);
|
|
4835 XGetICValues(xic, XNStatusAttributes, status_list, NULL);
|
|
4836 XFree(status_list);
|
|
4837
|
|
4838 status_area.width = needed_rect->width;
|
|
4839 }
|
|
4840 else
|
|
4841 status_area.width = gui.char_width * Columns;
|
|
4842
|
|
4843 status_area.x = 0;
|
|
4844 status_area.y = gui.char_height * Rows + gui.border_offset;
|
|
4845 if (gui.which_scrollbars[SBAR_BOTTOM])
|
|
4846 status_area.y += gui.scrollbar_height;
|
|
4847 #ifdef FEAT_MENU
|
|
4848 if (gui.menu_is_active)
|
|
4849 status_area.y += gui.menu_height;
|
|
4850 #endif
|
|
4851 status_area.height = gui.char_height;
|
|
4852 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL);
|
|
4853 }
|
|
4854 else
|
|
4855 {
|
|
4856 status_area.x = 0;
|
|
4857 status_area.y = gui.char_height * Rows + gui.border_offset;
|
|
4858 if (gui.which_scrollbars[SBAR_BOTTOM])
|
|
4859 status_area.y += gui.scrollbar_height;
|
|
4860 #ifdef FEAT_MENU
|
|
4861 if (gui.menu_is_active)
|
|
4862 status_area.y += gui.menu_height;
|
|
4863 #endif
|
|
4864 status_area.width = 0;
|
|
4865 status_area.height = gui.char_height;
|
|
4866 }
|
|
4867
|
|
4868 if (input_style & XIMPreeditArea) /* off-the-spot */
|
|
4869 {
|
|
4870 pre_area.x = status_area.x + status_area.width;
|
|
4871 pre_area.y = gui.char_height * Rows + gui.border_offset;
|
|
4872 pre_area.width = gui.char_width * Columns - pre_area.x;
|
|
4873 if (gui.which_scrollbars[SBAR_BOTTOM])
|
|
4874 pre_area.y += gui.scrollbar_height;
|
|
4875 #ifdef FEAT_MENU
|
|
4876 if (gui.menu_is_active)
|
|
4877 pre_area.y += gui.menu_height;
|
|
4878 #endif
|
|
4879 pre_area.height = gui.char_height;
|
|
4880 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL);
|
|
4881 }
|
|
4882 else if (input_style & XIMPreeditPosition) /* over-the-spot */
|
|
4883 {
|
|
4884 pre_area.x = 0;
|
|
4885 pre_area.y = 0;
|
|
4886 pre_area.height = gui.char_height * Rows;
|
|
4887 pre_area.width = gui.char_width * Columns;
|
|
4888 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL);
|
|
4889 }
|
|
4890
|
|
4891 if (preedit_list && status_list)
|
|
4892 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list,
|
|
4893 XNStatusAttributes, status_list, NULL);
|
|
4894 else if (preedit_list)
|
|
4895 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list,
|
|
4896 NULL);
|
|
4897 else if (status_list)
|
|
4898 list = XVaCreateNestedList(0, XNStatusAttributes, status_list,
|
|
4899 NULL);
|
|
4900 else
|
|
4901 list = NULL;
|
|
4902
|
|
4903 if (list)
|
|
4904 {
|
|
4905 XSetICValues(xic, XNVaNestedList, list, NULL);
|
|
4906 XFree(list);
|
|
4907 }
|
|
4908 if (status_list)
|
|
4909 XFree(status_list);
|
|
4910 if (preedit_list)
|
|
4911 XFree(preedit_list);
|
|
4912 }
|
|
4913 #endif
|
|
4914 }
|
|
4915
|
574
|
4916 #if defined(FEAT_GUI_X11) || defined(FEAT_GUI_GTK)
|
7
|
4917 static char e_xim[] = N_("E285: Failed to create input context");
|
|
4918 #endif
|
|
4919
|
|
4920 #if defined(FEAT_GUI_X11) || defined(PROTO)
|
|
4921 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(sun)
|
|
4922 # define USE_X11R6_XIM
|
|
4923 # endif
|
|
4924
|
|
4925 static int xim_real_init __ARGS((Window x11_window, Display *x11_display));
|
|
4926
|
|
4927
|
|
4928 #ifdef USE_X11R6_XIM
|
|
4929 static void xim_instantiate_cb __ARGS((Display *display, XPointer client_data, XPointer call_data));
|
|
4930 static void xim_destroy_cb __ARGS((XIM im, XPointer client_data, XPointer call_data));
|
|
4931
|
|
4932 /*ARGSUSED*/
|
|
4933 static void
|
|
4934 xim_instantiate_cb(display, client_data, call_data)
|
|
4935 Display *display;
|
|
4936 XPointer client_data;
|
|
4937 XPointer call_data;
|
|
4938 {
|
|
4939 Window x11_window;
|
|
4940 Display *x11_display;
|
|
4941
|
|
4942 #ifdef XIM_DEBUG
|
|
4943 xim_log("xim_instantiate_cb()\n");
|
|
4944 #endif
|
|
4945
|
|
4946 gui_get_x11_windis(&x11_window, &x11_display);
|
|
4947 if (display != x11_display)
|
|
4948 return;
|
|
4949
|
|
4950 xim_real_init(x11_window, x11_display);
|
811
|
4951 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH);
|
7
|
4952 if (xic != NULL)
|
|
4953 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL,
|
|
4954 xim_instantiate_cb, NULL);
|
|
4955 }
|
|
4956
|
|
4957 /*ARGSUSED*/
|
|
4958 static void
|
|
4959 xim_destroy_cb(im, client_data, call_data)
|
|
4960 XIM im;
|
|
4961 XPointer client_data;
|
|
4962 XPointer call_data;
|
|
4963 {
|
|
4964 Window x11_window;
|
|
4965 Display *x11_display;
|
|
4966
|
|
4967 #ifdef XIM_DEBUG
|
|
4968 xim_log("xim_destroy_cb()\n");
|
|
4969 #endif
|
|
4970 gui_get_x11_windis(&x11_window, &x11_display);
|
|
4971
|
|
4972 xic = NULL;
|
|
4973 status_area_enabled = FALSE;
|
|
4974
|
811
|
4975 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH);
|
7
|
4976
|
|
4977 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL,
|
|
4978 xim_instantiate_cb, NULL);
|
|
4979 }
|
|
4980 #endif
|
|
4981
|
|
4982 void
|
|
4983 xim_init()
|
|
4984 {
|
|
4985 Window x11_window;
|
|
4986 Display *x11_display;
|
|
4987
|
|
4988 #ifdef XIM_DEBUG
|
|
4989 xim_log("xim_init()\n");
|
|
4990 #endif
|
|
4991
|
|
4992 gui_get_x11_windis(&x11_window, &x11_display);
|
|
4993
|
|
4994 xic = NULL;
|
|
4995
|
|
4996 if (xim_real_init(x11_window, x11_display))
|
|
4997 return;
|
|
4998
|
811
|
4999 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH);
|
7
|
5000
|
|
5001 #ifdef USE_X11R6_XIM
|
|
5002 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL,
|
|
5003 xim_instantiate_cb, NULL);
|
|
5004 #endif
|
|
5005 }
|
|
5006
|
|
5007 static int
|
|
5008 xim_real_init(x11_window, x11_display)
|
|
5009 Window x11_window;
|
|
5010 Display *x11_display;
|
|
5011 {
|
|
5012 int i;
|
|
5013 char *p,
|
|
5014 *s,
|
|
5015 *ns,
|
|
5016 *end,
|
|
5017 tmp[1024];
|
|
5018 #define IMLEN_MAX 40
|
|
5019 char buf[IMLEN_MAX + 7];
|
|
5020 XIM xim = NULL;
|
|
5021 XIMStyles *xim_styles;
|
|
5022 XIMStyle this_input_style = 0;
|
|
5023 Boolean found;
|
|
5024 XPoint over_spot;
|
|
5025 XVaNestedList preedit_list, status_list;
|
|
5026
|
|
5027 input_style = 0;
|
|
5028 status_area_enabled = FALSE;
|
|
5029
|
|
5030 if (xic != NULL)
|
|
5031 return FALSE;
|
|
5032
|
|
5033 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL)
|
|
5034 {
|
|
5035 strcpy(tmp, gui.rsrc_input_method);
|
|
5036 for (ns = s = tmp; ns != NULL && *s != NUL;)
|
|
5037 {
|
|
5038 s = (char *)skipwhite((char_u *)s);
|
|
5039 if (*s == NUL)
|
|
5040 break;
|
|
5041 if ((ns = end = strchr(s, ',')) == NULL)
|
|
5042 end = s + strlen(s);
|
|
5043 while (isspace(((char_u *)end)[-1]))
|
|
5044 end--;
|
|
5045 *end = NUL;
|
|
5046
|
|
5047 if (strlen(s) <= IMLEN_MAX)
|
|
5048 {
|
|
5049 strcpy(buf, "@im=");
|
|
5050 strcat(buf, s);
|
|
5051 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL
|
|
5052 && (xim = XOpenIM(x11_display, NULL, NULL, NULL))
|
|
5053 != NULL)
|
|
5054 break;
|
|
5055 }
|
|
5056
|
|
5057 s = ns + 1;
|
|
5058 }
|
|
5059 }
|
|
5060
|
|
5061 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL)
|
|
5062 xim = XOpenIM(x11_display, NULL, NULL, NULL);
|
|
5063
|
|
5064 /* This is supposed to be useful to obtain characters through
|
|
5065 * XmbLookupString() without really using a XIM. */
|
|
5066 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL
|
|
5067 && *p != NUL)
|
|
5068 xim = XOpenIM(x11_display, NULL, NULL, NULL);
|
|
5069
|
|
5070 if (xim == NULL)
|
|
5071 {
|
|
5072 /* Only give this message when verbose is set, because too many people
|
|
5073 * got this message when they didn't want to use a XIM. */
|
|
5074 if (p_verbose > 0)
|
292
|
5075 {
|
|
5076 verbose_enter();
|
7
|
5077 EMSG(_("E286: Failed to open input method"));
|
292
|
5078 verbose_leave();
|
|
5079 }
|
7
|
5080 return FALSE;
|
|
5081 }
|
|
5082
|
|
5083 #ifdef USE_X11R6_XIM
|
|
5084 {
|
|
5085 XIMCallback destroy_cb;
|
|
5086
|
|
5087 destroy_cb.callback = xim_destroy_cb;
|
|
5088 destroy_cb.client_data = NULL;
|
|
5089 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL))
|
|
5090 EMSG(_("E287: Warning: Could not set destroy callback to IM"));
|
|
5091 }
|
|
5092 #endif
|
|
5093
|
|
5094 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles)
|
|
5095 {
|
|
5096 EMSG(_("E288: input method doesn't support any style"));
|
|
5097 XCloseIM(xim);
|
|
5098 return FALSE;
|
|
5099 }
|
|
5100
|
|
5101 found = False;
|
|
5102 strcpy(tmp, gui.rsrc_preedit_type_name);
|
|
5103 for (s = tmp; s && !found; )
|
|
5104 {
|
|
5105 while (*s && isspace((unsigned char)*s))
|
|
5106 s++;
|
|
5107 if (!*s)
|
|
5108 break;
|
|
5109 if ((ns = end = strchr(s, ',')) != 0)
|
|
5110 ns++;
|
|
5111 else
|
|
5112 end = s + strlen(s);
|
|
5113 while (isspace((unsigned char)*end))
|
|
5114 end--;
|
|
5115 *end = '\0';
|
|
5116
|
|
5117 if (!strcmp(s, "OverTheSpot"))
|
|
5118 this_input_style = (XIMPreeditPosition | XIMStatusArea);
|
|
5119 else if (!strcmp(s, "OffTheSpot"))
|
|
5120 this_input_style = (XIMPreeditArea | XIMStatusArea);
|
|
5121 else if (!strcmp(s, "Root"))
|
|
5122 this_input_style = (XIMPreeditNothing | XIMStatusNothing);
|
|
5123
|
|
5124 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++)
|
|
5125 {
|
|
5126 if (this_input_style == xim_styles->supported_styles[i])
|
|
5127 {
|
|
5128 found = True;
|
|
5129 break;
|
|
5130 }
|
|
5131 }
|
|
5132 if (!found)
|
|
5133 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++)
|
|
5134 {
|
|
5135 if ((xim_styles->supported_styles[i] & this_input_style)
|
|
5136 == (this_input_style & ~XIMStatusArea))
|
|
5137 {
|
|
5138 this_input_style &= ~XIMStatusArea;
|
|
5139 found = True;
|
|
5140 break;
|
|
5141 }
|
|
5142 }
|
|
5143
|
|
5144 s = ns;
|
|
5145 }
|
|
5146 XFree(xim_styles);
|
|
5147
|
|
5148 if (!found)
|
|
5149 {
|
|
5150 /* Only give this message when verbose is set, because too many people
|
|
5151 * got this message when they didn't want to use a XIM. */
|
|
5152 if (p_verbose > 0)
|
292
|
5153 {
|
|
5154 verbose_enter();
|
7
|
5155 EMSG(_("E289: input method doesn't support my preedit type"));
|
292
|
5156 verbose_leave();
|
|
5157 }
|
7
|
5158 XCloseIM(xim);
|
|
5159 return FALSE;
|
|
5160 }
|
|
5161
|
|
5162 over_spot.x = TEXT_X(gui.col);
|
|
5163 over_spot.y = TEXT_Y(gui.row);
|
|
5164 input_style = this_input_style;
|
|
5165
|
|
5166 /* A crash was reported when trying to pass gui.norm_font as XNFontSet,
|
|
5167 * thus that has been removed. Hopefully the default works... */
|
|
5168 #ifdef FEAT_XFONTSET
|
|
5169 if (gui.fontset != NOFONTSET)
|
|
5170 {
|
|
5171 preedit_list = XVaCreateNestedList(0,
|
|
5172 XNSpotLocation, &over_spot,
|
|
5173 XNForeground, (Pixel)gui.def_norm_pixel,
|
|
5174 XNBackground, (Pixel)gui.def_back_pixel,
|
|
5175 XNFontSet, (XFontSet)gui.fontset,
|
|
5176 NULL);
|
|
5177 status_list = XVaCreateNestedList(0,
|
|
5178 XNForeground, (Pixel)gui.def_norm_pixel,
|
|
5179 XNBackground, (Pixel)gui.def_back_pixel,
|
|
5180 XNFontSet, (XFontSet)gui.fontset,
|
|
5181 NULL);
|
|
5182 }
|
|
5183 else
|
|
5184 #endif
|
|
5185 {
|
|
5186 preedit_list = XVaCreateNestedList(0,
|
|
5187 XNSpotLocation, &over_spot,
|
|
5188 XNForeground, (Pixel)gui.def_norm_pixel,
|
|
5189 XNBackground, (Pixel)gui.def_back_pixel,
|
|
5190 NULL);
|
|
5191 status_list = XVaCreateNestedList(0,
|
|
5192 XNForeground, (Pixel)gui.def_norm_pixel,
|
|
5193 XNBackground, (Pixel)gui.def_back_pixel,
|
|
5194 NULL);
|
|
5195 }
|
|
5196
|
|
5197 xic = XCreateIC(xim,
|
|
5198 XNInputStyle, input_style,
|
|
5199 XNClientWindow, x11_window,
|
|
5200 XNFocusWindow, gui.wid,
|
|
5201 XNPreeditAttributes, preedit_list,
|
|
5202 XNStatusAttributes, status_list,
|
|
5203 NULL);
|
|
5204 XFree(status_list);
|
|
5205 XFree(preedit_list);
|
|
5206 if (xic != NULL)
|
|
5207 {
|
|
5208 if (input_style & XIMStatusArea)
|
|
5209 {
|
|
5210 xim_set_status_area();
|
|
5211 status_area_enabled = TRUE;
|
|
5212 }
|
|
5213 else
|
811
|
5214 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH);
|
7
|
5215 }
|
|
5216 else
|
|
5217 {
|
|
5218 EMSG(_(e_xim));
|
|
5219 XCloseIM(xim);
|
|
5220 return FALSE;
|
|
5221 }
|
|
5222
|
|
5223 return TRUE;
|
|
5224 }
|
|
5225
|
|
5226 #endif /* FEAT_GUI_X11 */
|
|
5227
|
|
5228 #if defined(FEAT_GUI_GTK) || defined(PROTO)
|
|
5229
|
|
5230 # ifdef FEAT_XFONTSET
|
|
5231 static char e_overthespot[] = N_("E290: over-the-spot style requires fontset");
|
|
5232 # endif
|
|
5233
|
|
5234 # ifdef PROTO
|
|
5235 typedef int GdkIC;
|
|
5236 # endif
|
|
5237
|
|
5238 void
|
|
5239 xim_decide_input_style()
|
|
5240 {
|
|
5241 /* GDK_IM_STATUS_CALLBACKS was disabled, enabled it to allow Japanese
|
|
5242 * OverTheSpot. */
|
|
5243 int supported_style = (int)GDK_IM_PREEDIT_NONE |
|
|
5244 (int)GDK_IM_PREEDIT_NOTHING |
|
|
5245 (int)GDK_IM_PREEDIT_POSITION |
|
|
5246 (int)GDK_IM_PREEDIT_CALLBACKS |
|
|
5247 (int)GDK_IM_STATUS_CALLBACKS |
|
|
5248 (int)GDK_IM_STATUS_AREA |
|
|
5249 (int)GDK_IM_STATUS_NONE |
|
|
5250 (int)GDK_IM_STATUS_NOTHING;
|
|
5251
|
|
5252 #ifdef XIM_DEBUG
|
|
5253 xim_log("xim_decide_input_style()\n");
|
|
5254 #endif
|
|
5255
|
|
5256 if (!gdk_im_ready())
|
|
5257 xim_input_style = 0;
|
|
5258 else
|
|
5259 {
|
|
5260 if (gtk_major_version > 1
|
|
5261 || (gtk_major_version == 1
|
|
5262 && (gtk_minor_version > 2
|
|
5263 || (gtk_minor_version == 2 && gtk_micro_version >= 3))))
|
|
5264 use_status_area = TRUE;
|
|
5265 else
|
|
5266 {
|
|
5267 EMSG(_("E291: Your GTK+ is older than 1.2.3. Status area disabled"));
|
|
5268 use_status_area = FALSE;
|
|
5269 }
|
|
5270 #ifdef FEAT_XFONTSET
|
|
5271 if (gui.fontset == NOFONTSET || gui.fontset->type != GDK_FONT_FONTSET)
|
|
5272 #endif
|
|
5273 supported_style &= ~((int)GDK_IM_PREEDIT_POSITION
|
|
5274 | (int)GDK_IM_STATUS_AREA);
|
|
5275 if (!use_status_area)
|
|
5276 supported_style &= ~(int)GDK_IM_STATUS_AREA;
|
|
5277 xim_input_style = (int)gdk_im_decide_style((GdkIMStyle)supported_style);
|
|
5278 }
|
|
5279 }
|
|
5280
|
|
5281 /*ARGSUSED*/
|
|
5282 static void
|
|
5283 preedit_start_cbproc(XIC xic, XPointer client_data, XPointer call_data)
|
|
5284 {
|
|
5285 #ifdef XIM_DEBUG
|
|
5286 xim_log("xim_decide_input_style()\n");
|
|
5287 #endif
|
|
5288
|
|
5289 draw_feedback = NULL;
|
|
5290 xim_can_preediting = TRUE;
|
|
5291 xim_has_preediting = TRUE;
|
|
5292 gui_update_cursor(TRUE, FALSE);
|
|
5293 if (showmode() > 0)
|
|
5294 {
|
|
5295 setcursor();
|
|
5296 out_flush();
|
|
5297 }
|
|
5298 }
|
|
5299
|
|
5300 static void
|
|
5301 xim_back_delete(int n)
|
|
5302 {
|
|
5303 char_u str[3];
|
|
5304
|
|
5305 str[0] = CSI;
|
|
5306 str[1] = 'k';
|
|
5307 str[2] = 'b';
|
|
5308 while (n-- > 0)
|
|
5309 add_to_input_buf(str, 3);
|
|
5310 }
|
|
5311
|
|
5312 static GSList *key_press_event_queue = NULL;
|
|
5313 static gboolean processing_queued_event = FALSE;
|
|
5314
|
|
5315 /*ARGSUSED*/
|
|
5316 static void
|
|
5317 preedit_draw_cbproc(XIC xic, XPointer client_data, XPointer call_data)
|
|
5318 {
|
|
5319 XIMPreeditDrawCallbackStruct *draw_data;
|
|
5320 XIMText *text;
|
|
5321 char *src;
|
|
5322 GSList *event_queue;
|
|
5323
|
|
5324 #ifdef XIM_DEBUG
|
|
5325 xim_log("preedit_draw_cbproc()\n");
|
|
5326 #endif
|
|
5327
|
|
5328 draw_data = (XIMPreeditDrawCallbackStruct *) call_data;
|
|
5329 text = (XIMText *) draw_data->text;
|
|
5330
|
|
5331 if ((text == NULL && draw_data->chg_length == preedit_buf_len)
|
|
5332 || preedit_buf_len == 0)
|
|
5333 {
|
|
5334 init_preedit_start_col();
|
|
5335 vim_free(draw_feedback);
|
|
5336 draw_feedback = NULL;
|
|
5337 }
|
|
5338 if (draw_data->chg_length > 0)
|
|
5339 {
|
|
5340 int bs_cnt;
|
|
5341
|
|
5342 if (draw_data->chg_length > preedit_buf_len)
|
|
5343 bs_cnt = preedit_buf_len;
|
|
5344 else
|
|
5345 bs_cnt = draw_data->chg_length;
|
|
5346 xim_back_delete(bs_cnt);
|
|
5347 preedit_buf_len -= bs_cnt;
|
|
5348 }
|
|
5349 if (text != NULL)
|
|
5350 {
|
|
5351 int len;
|
|
5352 #ifdef FEAT_MBYTE
|
|
5353 char_u *buf = NULL;
|
|
5354 unsigned int nfeedback = 0;
|
|
5355 #endif
|
|
5356 char_u *ptr;
|
|
5357
|
|
5358 src = text->string.multi_byte;
|
|
5359 if (src != NULL && !text->encoding_is_wchar)
|
|
5360 {
|
|
5361 len = strlen(src);
|
|
5362 ptr = (char_u *)src;
|
|
5363 /* Avoid the enter for decision */
|
|
5364 if (*ptr == '\n')
|
|
5365 return;
|
|
5366
|
|
5367 #ifdef FEAT_MBYTE
|
|
5368 if (input_conv.vc_type != CONV_NONE
|
|
5369 && (buf = string_convert(&input_conv,
|
|
5370 (char_u *)src, &len)) != NULL)
|
|
5371 {
|
|
5372 /* Converted from 'termencoding' to 'encoding'. */
|
|
5373 add_to_input_buf_csi(buf, len);
|
|
5374 ptr = buf;
|
|
5375 }
|
|
5376 else
|
|
5377 #endif
|
|
5378 add_to_input_buf_csi((char_u *)src, len);
|
|
5379 /* Add count of character to preedit_buf_len */
|
|
5380 while (*ptr != NUL)
|
|
5381 {
|
|
5382 #ifdef FEAT_MBYTE
|
|
5383 if (draw_data->text->feedback != NULL)
|
|
5384 {
|
|
5385 if (draw_feedback == NULL)
|
|
5386 draw_feedback = (char *)alloc(draw_data->chg_first
|
|
5387 + text->length);
|
|
5388 else
|
|
5389 draw_feedback = realloc(draw_feedback,
|
|
5390 draw_data->chg_first + text->length);
|
|
5391 if (draw_feedback != NULL)
|
|
5392 {
|
|
5393 draw_feedback[nfeedback + draw_data->chg_first]
|
|
5394 = draw_data->text->feedback[nfeedback];
|
|
5395 nfeedback++;
|
|
5396 }
|
|
5397 }
|
|
5398 if (has_mbyte)
|
474
|
5399 ptr += (*mb_ptr2len)(ptr);
|
7
|
5400 else
|
|
5401 #endif
|
|
5402 ptr++;
|
|
5403 preedit_buf_len++;
|
|
5404 }
|
|
5405 #ifdef FEAT_MBYTE
|
|
5406 vim_free(buf);
|
|
5407 #endif
|
|
5408 preedit_end_col = MAXCOL;
|
|
5409 }
|
|
5410 }
|
|
5411 if (text != NULL || draw_data->chg_length > 0)
|
|
5412 {
|
|
5413 event_queue = key_press_event_queue;
|
|
5414 processing_queued_event = TRUE;
|
|
5415 while (event_queue != NULL && processing_queued_event)
|
|
5416 {
|
|
5417 GdkEvent *ev = event_queue->data;
|
|
5418
|
|
5419 gboolean *ret;
|
|
5420 gtk_signal_emit_by_name((GtkObject*)gui.mainwin, "key_press_event",
|
|
5421 ev, &ret);
|
|
5422 gdk_event_free(ev);
|
|
5423 event_queue = event_queue->next;
|
|
5424 }
|
|
5425 processing_queued_event = FALSE;
|
|
5426 if (key_press_event_queue)
|
|
5427 {
|
|
5428 g_slist_free(key_press_event_queue);
|
|
5429 key_press_event_queue = NULL;
|
|
5430 }
|
|
5431 }
|
|
5432 if (gtk_main_level() > 0)
|
|
5433 gtk_main_quit();
|
|
5434 }
|
|
5435
|
|
5436 /*
|
|
5437 * Retrieve the highlighting attributes at column col in the preedit string.
|
|
5438 * Return -1 if not in preediting mode or if col is out of range.
|
|
5439 */
|
|
5440 int
|
|
5441 im_get_feedback_attr(int col)
|
|
5442 {
|
|
5443 if (draw_feedback != NULL && col < preedit_buf_len)
|
|
5444 {
|
|
5445 if (draw_feedback[col] & XIMReverse)
|
|
5446 return HL_INVERSE;
|
|
5447 else if (draw_feedback[col] & XIMUnderline)
|
|
5448 return HL_UNDERLINE;
|
|
5449 else
|
|
5450 return hl_attr(HLF_V);
|
|
5451 }
|
|
5452
|
|
5453 return -1;
|
|
5454 }
|
|
5455
|
|
5456 /*ARGSUSED*/
|
|
5457 static void
|
|
5458 preedit_caret_cbproc(XIC xic, XPointer client_data, XPointer call_data)
|
|
5459 {
|
|
5460 #ifdef XIM_DEBUG
|
|
5461 xim_log("preedit_caret_cbproc()\n");
|
|
5462 #endif
|
|
5463 }
|
|
5464
|
|
5465 /*ARGSUSED*/
|
|
5466 static void
|
|
5467 preedit_done_cbproc(XIC xic, XPointer client_data, XPointer call_data)
|
|
5468 {
|
|
5469 #ifdef XIM_DEBUG
|
|
5470 xim_log("preedit_done_cbproc()\n");
|
|
5471 #endif
|
|
5472
|
|
5473 vim_free(draw_feedback);
|
|
5474 draw_feedback = NULL;
|
|
5475 xim_can_preediting = FALSE;
|
|
5476 xim_has_preediting = FALSE;
|
|
5477 gui_update_cursor(TRUE, FALSE);
|
|
5478 if (showmode() > 0)
|
|
5479 {
|
|
5480 setcursor();
|
|
5481 out_flush();
|
|
5482 }
|
|
5483 }
|
|
5484
|
|
5485 void
|
|
5486 xim_reset(void)
|
|
5487 {
|
|
5488 char *text;
|
|
5489
|
|
5490 #ifdef XIM_DEBUG
|
|
5491 xim_log("xim_reset()\n");
|
|
5492 #endif
|
|
5493
|
|
5494 if (xic != NULL)
|
|
5495 {
|
|
5496 text = XmbResetIC(((GdkICPrivate *)xic)->xic);
|
|
5497 if (text != NULL && !(xim_input_style & (int)GDK_IM_PREEDIT_CALLBACKS))
|
|
5498 add_to_input_buf_csi((char_u *)text, strlen(text));
|
|
5499 else
|
|
5500 preedit_buf_len = 0;
|
|
5501 if (text != NULL)
|
|
5502 XFree(text);
|
|
5503 }
|
|
5504 }
|
|
5505
|
|
5506 /*ARGSUSED*/
|
|
5507 int
|
|
5508 xim_queue_key_press_event(GdkEventKey *event, int down)
|
|
5509 {
|
|
5510 #ifdef XIM_DEBUG
|
|
5511 xim_log("xim_queue_key_press_event()\n");
|
|
5512 #endif
|
|
5513
|
|
5514 if (preedit_buf_len <= 0)
|
|
5515 return FALSE;
|
|
5516 if (processing_queued_event)
|
|
5517 processing_queued_event = FALSE;
|
|
5518
|
|
5519 key_press_event_queue = g_slist_append(key_press_event_queue,
|
|
5520 gdk_event_copy((GdkEvent *)event));
|
|
5521 return TRUE;
|
|
5522 }
|
|
5523
|
|
5524 /*ARGSUSED*/
|
|
5525 static void
|
|
5526 preedit_callback_setup(GdkIC *ic)
|
|
5527 {
|
|
5528 XIC xxic;
|
|
5529 XVaNestedList preedit_attr;
|
|
5530 XIMCallback preedit_start_cb;
|
|
5531 XIMCallback preedit_draw_cb;
|
|
5532 XIMCallback preedit_caret_cb;
|
|
5533 XIMCallback preedit_done_cb;
|
|
5534
|
|
5535 xxic = ((GdkICPrivate*)xic)->xic;
|
|
5536 preedit_start_cb.callback = (XIMProc)preedit_start_cbproc;
|
|
5537 preedit_draw_cb.callback = (XIMProc)preedit_draw_cbproc;
|
|
5538 preedit_caret_cb.callback = (XIMProc)preedit_caret_cbproc;
|
|
5539 preedit_done_cb.callback = (XIMProc)preedit_done_cbproc;
|
|
5540 preedit_attr
|
1598
|
5541 = XVaCreateNestedList(0,
|
7
|
5542 XNPreeditStartCallback, &preedit_start_cb,
|
|
5543 XNPreeditDrawCallback, &preedit_draw_cb,
|
|
5544 XNPreeditCaretCallback, &preedit_caret_cb,
|
|
5545 XNPreeditDoneCallback, &preedit_done_cb,
|
1598
|
5546 NULL);
|
|
5547 XSetICValues(xxic, XNPreeditAttributes, preedit_attr, NULL);
|
7
|
5548 XFree(preedit_attr);
|
|
5549 }
|
|
5550
|
|
5551 /*ARGSUSED*/
|
|
5552 static void
|
|
5553 reset_state_setup(GdkIC *ic)
|
|
5554 {
|
|
5555 #ifdef USE_X11R6_XIM
|
|
5556 /* don't change the input context when we call reset */
|
1598
|
5557 XSetICValues(((GdkICPrivate *)ic)->xic, XNResetState, XIMPreserveState,
|
|
5558 NULL);
|
7
|
5559 #endif
|
|
5560 }
|
|
5561
|
|
5562 void
|
|
5563 xim_init(void)
|
|
5564 {
|
|
5565 #ifdef XIM_DEBUG
|
|
5566 xim_log("xim_init()\n");
|
|
5567 #endif
|
|
5568
|
|
5569 xic = NULL;
|
|
5570 xic_attr = NULL;
|
|
5571
|
|
5572 if (!gdk_im_ready())
|
|
5573 {
|
|
5574 if (p_verbose > 0)
|
292
|
5575 {
|
|
5576 verbose_enter();
|
7
|
5577 EMSG(_("E292: Input Method Server is not running"));
|
292
|
5578 verbose_leave();
|
|
5579 }
|
7
|
5580 return;
|
|
5581 }
|
|
5582 if ((xic_attr = gdk_ic_attr_new()) != NULL)
|
|
5583 {
|
|
5584 #ifdef FEAT_XFONTSET
|
|
5585 gint width, height;
|
|
5586 #endif
|
|
5587 int mask;
|
|
5588 GdkColormap *colormap;
|
|
5589 GdkICAttr *attr = xic_attr;
|
|
5590 int attrmask = (int)GDK_IC_ALL_REQ;
|
|
5591 GtkWidget *widget = gui.drawarea;
|
|
5592
|
|
5593 attr->style = (GdkIMStyle)xim_input_style;
|
|
5594 attr->client_window = gui.mainwin->window;
|
|
5595
|
|
5596 if ((colormap = gtk_widget_get_colormap(widget)) !=
|
|
5597 gtk_widget_get_default_colormap())
|
|
5598 {
|
|
5599 attrmask |= (int)GDK_IC_PREEDIT_COLORMAP;
|
|
5600 attr->preedit_colormap = colormap;
|
|
5601 }
|
|
5602 attrmask |= (int)GDK_IC_PREEDIT_FOREGROUND;
|
|
5603 attrmask |= (int)GDK_IC_PREEDIT_BACKGROUND;
|
|
5604 attr->preedit_foreground = widget->style->fg[GTK_STATE_NORMAL];
|
|
5605 attr->preedit_background = widget->style->base[GTK_STATE_NORMAL];
|
|
5606
|
|
5607 #ifdef FEAT_XFONTSET
|
|
5608 if ((xim_input_style & (int)GDK_IM_PREEDIT_MASK)
|
|
5609 == (int)GDK_IM_PREEDIT_POSITION)
|
|
5610 {
|
|
5611 if (gui.fontset == NOFONTSET
|
|
5612 || gui.fontset->type != GDK_FONT_FONTSET)
|
|
5613 {
|
|
5614 EMSG(_(e_overthespot));
|
|
5615 }
|
|
5616 else
|
|
5617 {
|
|
5618 gdk_window_get_size(widget->window, &width, &height);
|
|
5619
|
|
5620 attrmask |= (int)GDK_IC_PREEDIT_POSITION_REQ;
|
|
5621 attr->spot_location.x = TEXT_X(0);
|
|
5622 attr->spot_location.y = TEXT_Y(0);
|
|
5623 attr->preedit_area.x = gui.border_offset;
|
|
5624 attr->preedit_area.y = gui.border_offset;
|
|
5625 attr->preedit_area.width = width - 2*gui.border_offset;
|
|
5626 attr->preedit_area.height = height - 2*gui.border_offset;
|
|
5627 attr->preedit_fontset = gui.fontset;
|
|
5628 }
|
|
5629 }
|
|
5630
|
|
5631 if ((xim_input_style & (int)GDK_IM_STATUS_MASK)
|
|
5632 == (int)GDK_IM_STATUS_AREA)
|
|
5633 {
|
|
5634 if (gui.fontset == NOFONTSET
|
|
5635 || gui.fontset->type != GDK_FONT_FONTSET)
|
|
5636 {
|
|
5637 EMSG(_(e_overthespot));
|
|
5638 }
|
|
5639 else
|
|
5640 {
|
|
5641 gdk_window_get_size(gui.mainwin->window, &width, &height);
|
|
5642 attrmask |= (int)GDK_IC_STATUS_AREA_REQ;
|
|
5643 attr->status_area.x = 0;
|
|
5644 attr->status_area.y = height - gui.char_height - 1;
|
|
5645 attr->status_area.width = width;
|
|
5646 attr->status_area.height = gui.char_height;
|
|
5647 attr->status_fontset = gui.fontset;
|
|
5648 }
|
|
5649 }
|
|
5650 else if ((xim_input_style & (int)GDK_IM_STATUS_MASK)
|
|
5651 == (int)GDK_IM_STATUS_CALLBACKS)
|
|
5652 {
|
|
5653 /* FIXME */
|
|
5654 }
|
|
5655 #endif
|
|
5656
|
|
5657 xic = gdk_ic_new(attr, (GdkICAttributesType)attrmask);
|
|
5658
|
|
5659 if (xic == NULL)
|
|
5660 EMSG(_(e_xim));
|
|
5661 else
|
|
5662 {
|
|
5663 mask = (int)gdk_window_get_events(widget->window);
|
|
5664 mask |= (int)gdk_ic_get_events(xic);
|
|
5665 gdk_window_set_events(widget->window, (GdkEventMask)mask);
|
|
5666 if (xim_input_style & (int)GDK_IM_PREEDIT_CALLBACKS)
|
|
5667 preedit_callback_setup(xic);
|
|
5668 reset_state_setup(xic);
|
|
5669 }
|
|
5670 }
|
|
5671 }
|
|
5672
|
|
5673 void
|
|
5674 im_shutdown(void)
|
|
5675 {
|
|
5676 #ifdef XIM_DEBUG
|
|
5677 xim_log("im_shutdown()\n");
|
|
5678 #endif
|
|
5679
|
|
5680 if (xic != NULL)
|
|
5681 {
|
|
5682 gdk_im_end();
|
|
5683 gdk_ic_destroy(xic);
|
|
5684 xic = NULL;
|
|
5685 }
|
|
5686 xim_is_active = FALSE;
|
|
5687 xim_can_preediting = FALSE;
|
|
5688 preedit_start_col = MAXCOL;
|
|
5689 xim_has_preediting = FALSE;
|
|
5690 }
|
|
5691
|
|
5692 #endif /* FEAT_GUI_GTK */
|
|
5693
|
|
5694 int
|
|
5695 xim_get_status_area_height()
|
|
5696 {
|
|
5697 #ifdef FEAT_GUI_GTK
|
|
5698 if (xim_input_style & (int)GDK_IM_STATUS_AREA)
|
|
5699 return gui.char_height;
|
129
|
5700 #else
|
7
|
5701 if (status_area_enabled)
|
|
5702 return gui.char_height;
|
|
5703 #endif
|
|
5704 return 0;
|
|
5705 }
|
|
5706
|
|
5707 /*
|
|
5708 * Get IM status. When IM is on, return TRUE. Else return FALSE.
|
|
5709 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is
|
|
5710 * active, when not having focus XIM may still be active (e.g., when using a
|
|
5711 * tear-off menu item).
|
|
5712 */
|
|
5713 int
|
|
5714 im_get_status()
|
|
5715 {
|
|
5716 # ifdef FEAT_GUI_GTK
|
|
5717 if (xim_input_style & (int)GDK_IM_PREEDIT_CALLBACKS)
|
|
5718 return xim_can_preediting;
|
|
5719 # endif
|
|
5720 return xim_has_focus;
|
|
5721 }
|
|
5722
|
|
5723 # endif /* !HAVE_GTK2 */
|
|
5724
|
1668
|
5725 # if defined(HAVE_GTK2) || defined(PROTO)
|
|
5726 int
|
|
5727 preedit_get_status(void)
|
|
5728 {
|
|
5729 return preedit_is_active;
|
|
5730 }
|
|
5731 # endif
|
|
5732
|
7
|
5733 # if defined(FEAT_GUI_GTK) || defined(PROTO)
|
|
5734 int
|
|
5735 im_is_preediting()
|
|
5736 {
|
|
5737 return xim_has_preediting;
|
|
5738 }
|
|
5739 # endif
|
|
5740 #endif /* FEAT_XIM */
|
|
5741
|
|
5742 #if defined(FEAT_MBYTE) || defined(PROTO)
|
|
5743
|
|
5744 /*
|
|
5745 * Setup "vcp" for conversion from "from" to "to".
|
|
5746 * The names must have been made canonical with enc_canonize().
|
|
5747 * vcp->vc_type must have been initialized to CONV_NONE.
|
|
5748 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8
|
|
5749 * instead).
|
|
5750 * Afterwards invoke with "from" and "to" equal to NULL to cleanup.
|
|
5751 * Return FAIL when conversion is not supported, OK otherwise.
|
|
5752 */
|
|
5753 int
|
|
5754 convert_setup(vcp, from, to)
|
|
5755 vimconv_T *vcp;
|
|
5756 char_u *from;
|
|
5757 char_u *to;
|
|
5758 {
|
|
5759 int from_prop;
|
|
5760 int to_prop;
|
|
5761
|
|
5762 /* Reset to no conversion. */
|
|
5763 # ifdef USE_ICONV
|
|
5764 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1)
|
|
5765 iconv_close(vcp->vc_fd);
|
|
5766 # endif
|
|
5767 vcp->vc_type = CONV_NONE;
|
|
5768 vcp->vc_factor = 1;
|
|
5769 vcp->vc_fail = FALSE;
|
|
5770
|
|
5771 /* No conversion when one of the names is empty or they are equal. */
|
|
5772 if (from == NULL || *from == NUL || to == NULL || *to == NUL
|
|
5773 || STRCMP(from, to) == 0)
|
|
5774 return OK;
|
|
5775
|
|
5776 from_prop = enc_canon_props(from);
|
|
5777 to_prop = enc_canon_props(to);
|
|
5778 if ((from_prop & ENC_LATIN1) && (to_prop & ENC_UNICODE))
|
|
5779 {
|
|
5780 /* Internal latin1 -> utf-8 conversion. */
|
|
5781 vcp->vc_type = CONV_TO_UTF8;
|
|
5782 vcp->vc_factor = 2; /* up to twice as long */
|
|
5783 }
|
26
|
5784 else if ((from_prop & ENC_LATIN9) && (to_prop & ENC_UNICODE))
|
|
5785 {
|
|
5786 /* Internal latin9 -> utf-8 conversion. */
|
|
5787 vcp->vc_type = CONV_9_TO_UTF8;
|
|
5788 vcp->vc_factor = 3; /* up to three as long (euro sign) */
|
|
5789 }
|
7
|
5790 else if ((from_prop & ENC_UNICODE) && (to_prop & ENC_LATIN1))
|
|
5791 {
|
|
5792 /* Internal utf-8 -> latin1 conversion. */
|
|
5793 vcp->vc_type = CONV_TO_LATIN1;
|
|
5794 }
|
26
|
5795 else if ((from_prop & ENC_UNICODE) && (to_prop & ENC_LATIN9))
|
|
5796 {
|
|
5797 /* Internal utf-8 -> latin9 conversion. */
|
|
5798 vcp->vc_type = CONV_TO_LATIN9;
|
|
5799 }
|
7
|
5800 #ifdef WIN3264
|
|
5801 /* Win32-specific codepage <-> codepage conversion without iconv. */
|
|
5802 else if (((from_prop & ENC_UNICODE) || encname2codepage(from) > 0)
|
|
5803 && ((to_prop & ENC_UNICODE) || encname2codepage(to) > 0))
|
|
5804 {
|
|
5805 vcp->vc_type = CONV_CODEPAGE;
|
|
5806 vcp->vc_factor = 2; /* up to twice as long */
|
|
5807 vcp->vc_cpfrom = (from_prop & ENC_UNICODE) ? 0 : encname2codepage(from);
|
|
5808 vcp->vc_cpto = (to_prop & ENC_UNICODE) ? 0 : encname2codepage(to);
|
|
5809 }
|
|
5810 #endif
|
|
5811 #ifdef MACOS_X
|
|
5812 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1))
|
|
5813 {
|
|
5814 vcp->vc_type = CONV_MAC_LATIN1;
|
|
5815 }
|
|
5816 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_UNICODE))
|
|
5817 {
|
|
5818 vcp->vc_type = CONV_MAC_UTF8;
|
|
5819 vcp->vc_factor = 2; /* up to twice as long */
|
|
5820 }
|
|
5821 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN))
|
|
5822 {
|
|
5823 vcp->vc_type = CONV_LATIN1_MAC;
|
|
5824 }
|
|
5825 else if ((from_prop & ENC_UNICODE) && (to_prop & ENC_MACROMAN))
|
|
5826 {
|
|
5827 vcp->vc_type = CONV_UTF8_MAC;
|
|
5828 }
|
|
5829 #endif
|
|
5830 # ifdef USE_ICONV
|
|
5831 else
|
|
5832 {
|
|
5833 /* Use iconv() for conversion. */
|
|
5834 vcp->vc_fd = (iconv_t)my_iconv_open(
|
|
5835 (to_prop & ENC_UNICODE) ? (char_u *)"utf-8" : to,
|
|
5836 (from_prop & ENC_UNICODE) ? (char_u *)"utf-8" : from);
|
|
5837 if (vcp->vc_fd != (iconv_t)-1)
|
|
5838 {
|
|
5839 vcp->vc_type = CONV_ICONV;
|
|
5840 vcp->vc_factor = 4; /* could be longer too... */
|
|
5841 }
|
|
5842 }
|
|
5843 # endif
|
|
5844 if (vcp->vc_type == CONV_NONE)
|
|
5845 return FAIL;
|
167
|
5846
|
7
|
5847 return OK;
|
|
5848 }
|
|
5849
|
|
5850 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \
|
|
5851 || defined(MSDOS) || defined(PROTO)
|
|
5852 /*
|
|
5853 * Do conversion on typed input characters in-place.
|
|
5854 * The input and output are not NUL terminated!
|
|
5855 * Returns the length after conversion.
|
|
5856 */
|
|
5857 int
|
|
5858 convert_input(ptr, len, maxlen)
|
|
5859 char_u *ptr;
|
|
5860 int len;
|
|
5861 int maxlen;
|
|
5862 {
|
|
5863 return convert_input_safe(ptr, len, maxlen, NULL, NULL);
|
|
5864 }
|
|
5865 #endif
|
|
5866
|
|
5867 /*
|
|
5868 * Like convert_input(), but when there is an incomplete byte sequence at the
|
|
5869 * end return that as an allocated string in "restp" and set "*restlenp" to
|
|
5870 * the length. If "restp" is NULL it is not used.
|
|
5871 */
|
|
5872 int
|
|
5873 convert_input_safe(ptr, len, maxlen, restp, restlenp)
|
|
5874 char_u *ptr;
|
|
5875 int len;
|
|
5876 int maxlen;
|
|
5877 char_u **restp;
|
|
5878 int *restlenp;
|
|
5879 {
|
|
5880 char_u *d;
|
|
5881 int dlen = len;
|
|
5882 int unconvertlen = 0;
|
|
5883
|
|
5884 d = string_convert_ext(&input_conv, ptr, &dlen,
|
|
5885 restp == NULL ? NULL : &unconvertlen);
|
|
5886 if (d != NULL)
|
|
5887 {
|
|
5888 if (dlen <= maxlen)
|
|
5889 {
|
|
5890 if (unconvertlen > 0)
|
|
5891 {
|
|
5892 /* Move the unconverted characters to allocated memory. */
|
|
5893 *restp = alloc(unconvertlen);
|
|
5894 if (*restp != NULL)
|
|
5895 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen);
|
|
5896 *restlenp = unconvertlen;
|
|
5897 }
|
|
5898 mch_memmove(ptr, d, dlen);
|
|
5899 }
|
|
5900 else
|
|
5901 /* result is too long, keep the unconverted text (the caller must
|
|
5902 * have done something wrong!) */
|
|
5903 dlen = len;
|
|
5904 vim_free(d);
|
|
5905 }
|
|
5906 return dlen;
|
|
5907 }
|
|
5908
|
|
5909 /*
|
|
5910 * Convert text "ptr[*lenp]" according to "vcp".
|
|
5911 * Returns the result in allocated memory and sets "*lenp".
|
|
5912 * When "lenp" is NULL, use NUL terminated strings.
|
|
5913 * Illegal chars are often changed to "?", unless vcp->vc_fail is set.
|
|
5914 * When something goes wrong, NULL is returned and "*lenp" is unchanged.
|
|
5915 */
|
|
5916 char_u *
|
|
5917 string_convert(vcp, ptr, lenp)
|
|
5918 vimconv_T *vcp;
|
|
5919 char_u *ptr;
|
|
5920 int *lenp;
|
|
5921 {
|
|
5922 return string_convert_ext(vcp, ptr, lenp, NULL);
|
|
5923 }
|
|
5924
|
|
5925 /*
|
|
5926 * Like string_convert(), but when "unconvlenp" is not NULL and there are is
|
|
5927 * an incomplete sequence at the end it is not converted and "*unconvlenp" is
|
|
5928 * set to the number of remaining bytes.
|
|
5929 */
|
|
5930 char_u *
|
|
5931 string_convert_ext(vcp, ptr, lenp, unconvlenp)
|
|
5932 vimconv_T *vcp;
|
|
5933 char_u *ptr;
|
|
5934 int *lenp;
|
|
5935 int *unconvlenp;
|
|
5936 {
|
|
5937 char_u *retval = NULL;
|
|
5938 char_u *d;
|
|
5939 int len;
|
|
5940 int i;
|
|
5941 int l;
|
|
5942 int c;
|
|
5943
|
|
5944 if (lenp == NULL)
|
|
5945 len = (int)STRLEN(ptr);
|
|
5946 else
|
|
5947 len = *lenp;
|
|
5948 if (len == 0)
|
|
5949 return vim_strsave((char_u *)"");
|
|
5950
|
|
5951 switch (vcp->vc_type)
|
|
5952 {
|
|
5953 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */
|
|
5954 retval = alloc(len * 2 + 1);
|
|
5955 if (retval == NULL)
|
|
5956 break;
|
|
5957 d = retval;
|
|
5958 for (i = 0; i < len; ++i)
|
|
5959 {
|
26
|
5960 c = ptr[i];
|
|
5961 if (c < 0x80)
|
|
5962 *d++ = c;
|
7
|
5963 else
|
|
5964 {
|
26
|
5965 *d++ = 0xc0 + ((unsigned)c >> 6);
|
|
5966 *d++ = 0x80 + (c & 0x3f);
|
7
|
5967 }
|
|
5968 }
|
|
5969 *d = NUL;
|
|
5970 if (lenp != NULL)
|
|
5971 *lenp = (int)(d - retval);
|
|
5972 break;
|
|
5973
|
26
|
5974 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */
|
|
5975 retval = alloc(len * 3 + 1);
|
|
5976 if (retval == NULL)
|
|
5977 break;
|
|
5978 d = retval;
|
|
5979 for (i = 0; i < len; ++i)
|
|
5980 {
|
|
5981 c = ptr[i];
|
|
5982 switch (c)
|
|
5983 {
|
|
5984 case 0xa4: c = 0x20ac; break; /* euro */
|
|
5985 case 0xa6: c = 0x0160; break; /* S hat */
|
|
5986 case 0xa8: c = 0x0161; break; /* S -hat */
|
|
5987 case 0xb4: c = 0x017d; break; /* Z hat */
|
|
5988 case 0xb8: c = 0x017e; break; /* Z -hat */
|
|
5989 case 0xbc: c = 0x0152; break; /* OE */
|
|
5990 case 0xbd: c = 0x0153; break; /* oe */
|
|
5991 case 0xbe: c = 0x0178; break; /* Y */
|
|
5992 }
|
|
5993 d += utf_char2bytes(c, d);
|
|
5994 }
|
|
5995 *d = NUL;
|
|
5996 if (lenp != NULL)
|
|
5997 *lenp = (int)(d - retval);
|
|
5998 break;
|
|
5999
|
7
|
6000 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */
|
26
|
6001 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */
|
7
|
6002 retval = alloc(len + 1);
|
|
6003 if (retval == NULL)
|
|
6004 break;
|
|
6005 d = retval;
|
|
6006 for (i = 0; i < len; ++i)
|
|
6007 {
|
474
|
6008 l = utf_ptr2len(ptr + i);
|
7
|
6009 if (l == 0)
|
|
6010 *d++ = NUL;
|
|
6011 else if (l == 1)
|
|
6012 {
|
|
6013 if (unconvlenp != NULL && utf8len_tab[ptr[i]] > len - i)
|
|
6014 {
|
|
6015 /* Incomplete sequence at the end. */
|
|
6016 *unconvlenp = len - i;
|
|
6017 break;
|
|
6018 }
|
|
6019 *d++ = ptr[i];
|
|
6020 }
|
|
6021 else
|
|
6022 {
|
|
6023 c = utf_ptr2char(ptr + i);
|
26
|
6024 if (vcp->vc_type == CONV_TO_LATIN9)
|
|
6025 switch (c)
|
|
6026 {
|
|
6027 case 0x20ac: c = 0xa4; break; /* euro */
|
|
6028 case 0x0160: c = 0xa6; break; /* S hat */
|
|
6029 case 0x0161: c = 0xa8; break; /* S -hat */
|
|
6030 case 0x017d: c = 0xb4; break; /* Z hat */
|
|
6031 case 0x017e: c = 0xb8; break; /* Z -hat */
|
|
6032 case 0x0152: c = 0xbc; break; /* OE */
|
|
6033 case 0x0153: c = 0xbd; break; /* oe */
|
|
6034 case 0x0178: c = 0xbe; break; /* Y */
|
|
6035 case 0xa4:
|
|
6036 case 0xa6:
|
|
6037 case 0xa8:
|
|
6038 case 0xb4:
|
|
6039 case 0xb8:
|
|
6040 case 0xbc:
|
|
6041 case 0xbd:
|
|
6042 case 0xbe: c = 0x100; break; /* not in latin9 */
|
|
6043 }
|
7
|
6044 if (!utf_iscomposing(c)) /* skip composing chars */
|
|
6045 {
|
|
6046 if (c < 0x100)
|
|
6047 *d++ = c;
|
|
6048 else if (vcp->vc_fail)
|
|
6049 {
|
|
6050 vim_free(retval);
|
|
6051 return NULL;
|
|
6052 }
|
|
6053 else
|
|
6054 {
|
|
6055 *d++ = 0xbf;
|
|
6056 if (utf_char2cells(c) > 1)
|
|
6057 *d++ = '?';
|
|
6058 }
|
|
6059 }
|
|
6060 i += l - 1;
|
|
6061 }
|
|
6062 }
|
|
6063 *d = NUL;
|
|
6064 if (lenp != NULL)
|
|
6065 *lenp = (int)(d - retval);
|
|
6066 break;
|
|
6067
|
765
|
6068 # ifdef MACOS_CONVERT
|
7
|
6069 case CONV_MAC_LATIN1:
|
|
6070 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail,
|
18
|
6071 'm', 'l', unconvlenp);
|
7
|
6072 break;
|
|
6073
|
|
6074 case CONV_LATIN1_MAC:
|
|
6075 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail,
|
18
|
6076 'l', 'm', unconvlenp);
|
7
|
6077 break;
|
|
6078
|
|
6079 case CONV_MAC_UTF8:
|
|
6080 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail,
|
18
|
6081 'm', 'u', unconvlenp);
|
7
|
6082 break;
|
|
6083
|
|
6084 case CONV_UTF8_MAC:
|
|
6085 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail,
|
18
|
6086 'u', 'm', unconvlenp);
|
7
|
6087 break;
|
|
6088 # endif
|
|
6089
|
|
6090 # ifdef USE_ICONV
|
|
6091 case CONV_ICONV: /* conversion with output_conv.vc_fd */
|
|
6092 retval = iconv_string(vcp, ptr, len, unconvlenp);
|
|
6093 if (retval != NULL && lenp != NULL)
|
|
6094 *lenp = (int)STRLEN(retval);
|
|
6095 break;
|
|
6096 # endif
|
|
6097 # ifdef WIN3264
|
|
6098 case CONV_CODEPAGE: /* codepage -> codepage */
|
|
6099 {
|
|
6100 int retlen;
|
|
6101 int tmp_len;
|
|
6102 short_u *tmp;
|
|
6103
|
|
6104 /* 1. codepage/UTF-8 -> ucs-2. */
|
|
6105 if (vcp->vc_cpfrom == 0)
|
|
6106 tmp_len = utf8_to_ucs2(ptr, len, NULL, NULL);
|
|
6107 else
|
|
6108 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 0,
|
|
6109 ptr, len, 0, 0);
|
|
6110 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len);
|
|
6111 if (tmp == NULL)
|
|
6112 break;
|
|
6113 if (vcp->vc_cpfrom == 0)
|
|
6114 utf8_to_ucs2(ptr, len, tmp, unconvlenp);
|
|
6115 else
|
|
6116 MultiByteToWideChar(vcp->vc_cpfrom, 0, ptr, len, tmp, tmp_len);
|
|
6117
|
|
6118 /* 2. ucs-2 -> codepage/UTF-8. */
|
|
6119 if (vcp->vc_cpto == 0)
|
|
6120 retlen = ucs2_to_utf8(tmp, tmp_len, NULL);
|
|
6121 else
|
|
6122 retlen = WideCharToMultiByte(vcp->vc_cpto, 0,
|
|
6123 tmp, tmp_len, 0, 0, 0, 0);
|
|
6124 retval = alloc(retlen + 1);
|
|
6125 if (retval != NULL)
|
|
6126 {
|
|
6127 if (vcp->vc_cpto == 0)
|
|
6128 ucs2_to_utf8(tmp, tmp_len, retval);
|
|
6129 else
|
|
6130 WideCharToMultiByte(vcp->vc_cpto, 0,
|
|
6131 tmp, tmp_len, retval, retlen, 0, 0);
|
|
6132 retval[retlen] = NUL;
|
|
6133 if (lenp != NULL)
|
|
6134 *lenp = retlen;
|
|
6135 }
|
|
6136 vim_free(tmp);
|
|
6137 break;
|
|
6138 }
|
|
6139 # endif
|
|
6140 }
|
|
6141
|
|
6142 return retval;
|
|
6143 }
|
|
6144 #endif
|