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
|
|
103 #if defined(FEAT_XIM) && defined(HAVE_GTK2)
|
|
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));
|
|
129 static int dbcs_ptr2len_check __ARGS((char_u *p));
|
|
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 {
|
|
164 fd = fopen("xim.log", "w");
|
|
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 */
|
|
314 #define IDX_COUNT 58
|
7
|
315 };
|
|
316
|
|
317 /*
|
|
318 * Aliases for encoding names.
|
|
319 */
|
|
320 static struct
|
|
321 { char *name; int canon;}
|
|
322 enc_alias_table[] =
|
|
323 {
|
|
324 {"ansi", IDX_LATIN_1},
|
|
325 {"iso-8859-1", IDX_LATIN_1},
|
|
326 {"latin2", IDX_ISO_2},
|
|
327 {"latin3", IDX_ISO_3},
|
|
328 {"latin4", IDX_ISO_4},
|
|
329 {"cyrillic", IDX_ISO_5},
|
|
330 {"arabic", IDX_ISO_6},
|
|
331 {"greek", IDX_ISO_7},
|
|
332 #ifdef WIN3264
|
|
333 {"hebrew", IDX_CP1255},
|
|
334 #else
|
|
335 {"hebrew", IDX_ISO_8},
|
|
336 #endif
|
|
337 {"latin5", IDX_ISO_9},
|
|
338 {"turkish", IDX_ISO_9}, /* ? */
|
|
339 {"latin6", IDX_ISO_10},
|
|
340 {"nordic", IDX_ISO_10}, /* ? */
|
|
341 {"thai", IDX_ISO_11}, /* ? */
|
|
342 {"latin7", IDX_ISO_13},
|
|
343 {"latin8", IDX_ISO_14},
|
|
344 {"latin9", IDX_ISO_15},
|
|
345 {"utf8", IDX_UTF8},
|
|
346 {"unicode", IDX_UCS2},
|
|
347 {"ucs2", IDX_UCS2},
|
|
348 {"ucs2be", IDX_UCS2},
|
|
349 {"ucs-2be", IDX_UCS2},
|
|
350 {"ucs2le", IDX_UCS2LE},
|
|
351 {"utf16", IDX_UTF16},
|
|
352 {"utf16be", IDX_UTF16},
|
|
353 {"utf-16be", IDX_UTF16},
|
|
354 {"utf16le", IDX_UTF16LE},
|
|
355 {"ucs4", IDX_UCS4},
|
|
356 {"ucs4be", IDX_UCS4},
|
|
357 {"ucs-4be", IDX_UCS4},
|
|
358 {"ucs4le", IDX_UCS4LE},
|
|
359 {"932", IDX_CP932},
|
|
360 {"949", IDX_CP949},
|
|
361 {"936", IDX_CP936},
|
|
362 {"950", IDX_CP950},
|
|
363 {"eucjp", IDX_EUC_JP},
|
|
364 {"unix-jis", IDX_EUC_JP},
|
|
365 {"ujis", IDX_EUC_JP},
|
|
366 {"shift-jis", IDX_SJIS},
|
|
367 {"euckr", IDX_EUC_KR},
|
|
368 {"5601", IDX_EUC_KR}, /* Sun: KS C 5601 */
|
|
369 {"euccn", IDX_EUC_CN},
|
|
370 {"gb2312", IDX_EUC_CN},
|
|
371 {"euctw", IDX_EUC_TW},
|
|
372 #if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS)
|
|
373 {"japan", IDX_CP932},
|
|
374 {"korea", IDX_CP949},
|
|
375 {"prc", IDX_CP936},
|
|
376 {"chinese", IDX_CP936},
|
|
377 {"taiwan", IDX_CP950},
|
|
378 {"big5", IDX_CP950},
|
|
379 #else
|
|
380 {"japan", IDX_EUC_JP},
|
|
381 {"korea", IDX_EUC_KR},
|
|
382 {"prc", IDX_EUC_CN},
|
|
383 {"chinese", IDX_EUC_CN},
|
|
384 {"taiwan", IDX_EUC_TW},
|
|
385 {"cp950", IDX_BIG5},
|
|
386 {"950", IDX_BIG5},
|
|
387 #endif
|
|
388 {"mac", IDX_MACROMAN},
|
|
389 {NULL, 0}
|
|
390 };
|
|
391
|
|
392 #ifndef CP_UTF8
|
|
393 # define CP_UTF8 65001 /* magic number from winnls.h */
|
|
394 #endif
|
|
395
|
|
396 /*
|
|
397 * Find encoding "name" in the list of canonical encoding names.
|
|
398 * Returns -1 if not found.
|
|
399 */
|
|
400 static int
|
|
401 enc_canon_search(name)
|
|
402 char_u *name;
|
|
403 {
|
|
404 int i;
|
|
405
|
|
406 for (i = 0; i < IDX_COUNT; ++i)
|
|
407 if (STRCMP(name, enc_canon_table[i].name) == 0)
|
|
408 return i;
|
|
409 return -1;
|
|
410 }
|
|
411
|
|
412 #endif
|
|
413
|
|
414 #if defined(FEAT_MBYTE) || defined(PROTO)
|
|
415
|
|
416 /*
|
|
417 * Find canonical encoding "name" in the list and return its properties.
|
|
418 * Returns 0 if not found.
|
|
419 */
|
|
420 int
|
|
421 enc_canon_props(name)
|
|
422 char_u *name;
|
|
423 {
|
|
424 int i;
|
|
425
|
|
426 i = enc_canon_search(name);
|
|
427 if (i >= 0)
|
|
428 return enc_canon_table[i].prop;
|
|
429 #ifdef WIN3264
|
|
430 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2]))
|
|
431 {
|
|
432 CPINFO cpinfo;
|
|
433
|
|
434 /* Get info on this codepage to find out what it is. */
|
|
435 if (GetCPInfo(atoi(name + 2), &cpinfo) != 0)
|
|
436 {
|
|
437 if (cpinfo.MaxCharSize == 1) /* some single-byte encoding */
|
|
438 return ENC_8BIT;
|
|
439 if (cpinfo.MaxCharSize == 2
|
|
440 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0))
|
|
441 /* must be a DBCS encoding */
|
|
442 return ENC_DBCS;
|
|
443 }
|
|
444 return 0;
|
|
445 }
|
|
446 #endif
|
|
447 if (STRNCMP(name, "2byte-", 6) == 0)
|
|
448 return ENC_DBCS;
|
|
449 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0)
|
|
450 return ENC_8BIT;
|
|
451 return 0;
|
|
452 }
|
|
453
|
|
454 /*
|
|
455 * Set up for using multi-byte characters.
|
|
456 * Called in three cases:
|
|
457 * - by main() to initialize (p_enc == NULL)
|
|
458 * - by set_init_1() after 'encoding' was set to its default.
|
|
459 * - by do_set() when 'encoding' has been set.
|
|
460 * p_enc must have been passed through enc_canonize() already.
|
|
461 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags.
|
|
462 * Fills mb_bytelen_tab[] and returns NULL when there are no problems.
|
|
463 * When there is something wrong: Returns an error message and doesn't change
|
|
464 * anything.
|
|
465 */
|
|
466 char_u *
|
|
467 mb_init()
|
|
468 {
|
|
469 int i;
|
|
470 int idx;
|
|
471 int n;
|
|
472 int enc_dbcs_new = 0;
|
|
473 #if defined(USE_ICONV) && !defined(WIN3264) && !defined(WIN32UNIX) \
|
|
474 && !defined(MACOS)
|
|
475 # define LEN_FROM_CONV
|
|
476 vimconv_T vimconv;
|
|
477 char_u *p;
|
|
478 #endif
|
|
479
|
|
480 if (p_enc == NULL)
|
|
481 {
|
|
482 /* Just starting up: set the whole table to one's. */
|
|
483 for (i = 0; i < 256; ++i)
|
|
484 mb_bytelen_tab[i] = 1;
|
|
485 input_conv.vc_type = CONV_NONE;
|
|
486 input_conv.vc_factor = 1;
|
|
487 output_conv.vc_type = CONV_NONE;
|
|
488 return NULL;
|
|
489 }
|
|
490
|
|
491 #ifdef WIN3264
|
|
492 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2]))
|
|
493 {
|
|
494 CPINFO cpinfo;
|
|
495
|
|
496 /* Get info on this codepage to find out what it is. */
|
|
497 if (GetCPInfo(atoi(p_enc + 2), &cpinfo) != 0)
|
|
498 {
|
|
499 if (cpinfo.MaxCharSize == 1)
|
|
500 {
|
|
501 /* some single-byte encoding */
|
|
502 enc_unicode = 0;
|
|
503 enc_utf8 = FALSE;
|
|
504 }
|
|
505 else if (cpinfo.MaxCharSize == 2
|
|
506 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0))
|
|
507 {
|
|
508 /* must be a DBCS encoding, check below */
|
|
509 enc_dbcs_new = atoi(p_enc + 2);
|
|
510 }
|
|
511 else
|
|
512 goto codepage_invalid;
|
|
513 }
|
|
514 else if (GetLastError() == ERROR_INVALID_PARAMETER)
|
|
515 {
|
|
516 codepage_invalid:
|
|
517 return (char_u *)N_("E543: Not a valid codepage");
|
|
518 }
|
|
519 }
|
|
520 #endif
|
|
521 else if (STRNCMP(p_enc, "8bit-", 5) == 0
|
|
522 || STRNCMP(p_enc, "iso-8859-", 9) == 0)
|
|
523 {
|
|
524 /* Accept any "8bit-" or "iso-8859-" name. */
|
|
525 enc_unicode = 0;
|
|
526 enc_utf8 = FALSE;
|
|
527 }
|
|
528 else if (STRNCMP(p_enc, "2byte-", 6) == 0)
|
|
529 {
|
|
530 #ifdef WIN3264
|
|
531 /* Windows: accept only valid codepage numbers, check below. */
|
|
532 if (p_enc[6] != 'c' || p_enc[7] != 'p'
|
|
533 || (enc_dbcs_new = atoi(p_enc + 8)) == 0)
|
|
534 return e_invarg;
|
|
535 #else
|
|
536 /* Unix: accept any "2byte-" name, assume current locale. */
|
|
537 enc_dbcs_new = DBCS_2BYTE;
|
|
538 #endif
|
|
539 }
|
|
540 else if ((idx = enc_canon_search(p_enc)) >= 0)
|
|
541 {
|
|
542 i = enc_canon_table[idx].prop;
|
|
543 if (i & ENC_UNICODE)
|
|
544 {
|
|
545 /* Unicode */
|
|
546 enc_utf8 = TRUE;
|
|
547 if (i & (ENC_2BYTE | ENC_2WORD))
|
|
548 enc_unicode = 2;
|
|
549 else if (i & ENC_4BYTE)
|
|
550 enc_unicode = 4;
|
|
551 else
|
|
552 enc_unicode = 0;
|
|
553 }
|
|
554 else if (i & ENC_DBCS)
|
|
555 {
|
|
556 /* 2byte, handle below */
|
|
557 enc_dbcs_new = enc_canon_table[idx].codepage;
|
|
558 }
|
|
559 else
|
|
560 {
|
|
561 /* Must be 8-bit. */
|
|
562 enc_unicode = 0;
|
|
563 enc_utf8 = FALSE;
|
|
564 }
|
|
565 }
|
|
566 else /* Don't know what encoding this is, reject it. */
|
|
567 return e_invarg;
|
|
568
|
|
569 if (enc_dbcs_new != 0)
|
|
570 {
|
|
571 #ifdef WIN3264
|
|
572 /* Check if the DBCS code page is OK. */
|
|
573 if (!IsValidCodePage(enc_dbcs_new))
|
|
574 goto codepage_invalid;
|
|
575 #endif
|
|
576 enc_unicode = 0;
|
|
577 enc_utf8 = FALSE;
|
|
578 }
|
|
579 enc_dbcs = enc_dbcs_new;
|
|
580 has_mbyte = (enc_dbcs != 0 || enc_utf8);
|
|
581
|
|
582 #ifdef WIN3264
|
|
583 enc_codepage = encname2codepage(p_enc);
|
26
|
584 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0);
|
7
|
585 #endif
|
|
586
|
|
587 /*
|
|
588 * Set the function pointers.
|
|
589 */
|
|
590 if (enc_utf8)
|
|
591 {
|
|
592 mb_ptr2len_check = utfc_ptr2len_check;
|
|
593 mb_char2len = utf_char2len;
|
|
594 mb_char2bytes = utf_char2bytes;
|
|
595 mb_ptr2cells = utf_ptr2cells;
|
|
596 mb_char2cells = utf_char2cells;
|
|
597 mb_off2cells = utf_off2cells;
|
|
598 mb_ptr2char = utf_ptr2char;
|
|
599 mb_head_off = utf_head_off;
|
|
600 }
|
|
601 else if (enc_dbcs != 0)
|
|
602 {
|
|
603 mb_ptr2len_check = dbcs_ptr2len_check;
|
|
604 mb_char2len = dbcs_char2len;
|
|
605 mb_char2bytes = dbcs_char2bytes;
|
|
606 mb_ptr2cells = dbcs_ptr2cells;
|
|
607 mb_char2cells = dbcs_char2cells;
|
|
608 mb_off2cells = dbcs_off2cells;
|
|
609 mb_ptr2char = dbcs_ptr2char;
|
|
610 mb_head_off = dbcs_head_off;
|
|
611 }
|
|
612 else
|
|
613 {
|
|
614 mb_ptr2len_check = latin_ptr2len_check;
|
|
615 mb_char2len = latin_char2len;
|
|
616 mb_char2bytes = latin_char2bytes;
|
|
617 mb_ptr2cells = latin_ptr2cells;
|
|
618 mb_char2cells = latin_char2cells;
|
|
619 mb_off2cells = latin_off2cells;
|
|
620 mb_ptr2char = latin_ptr2char;
|
|
621 mb_head_off = latin_head_off;
|
|
622 }
|
|
623
|
|
624 /*
|
|
625 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN().
|
|
626 */
|
|
627 #ifdef LEN_FROM_CONV
|
|
628 /* When 'encoding' is different from the current locale mblen() won't
|
|
629 * work. Use conversion to "utf-8" instead. */
|
|
630 vimconv.vc_type = CONV_NONE;
|
|
631 if (enc_dbcs)
|
|
632 {
|
|
633 p = enc_locale();
|
|
634 if (p == NULL || STRCMP(p, p_enc) != 0)
|
|
635 {
|
|
636 convert_setup(&vimconv, p_enc, (char_u *)"utf-8");
|
|
637 vimconv.vc_fail = TRUE;
|
|
638 }
|
|
639 vim_free(p);
|
|
640 }
|
|
641 #endif
|
|
642
|
|
643 for (i = 0; i < 256; ++i)
|
|
644 {
|
|
645 /* Our own function to reliably check the length of UTF-8 characters,
|
|
646 * independent of mblen(). */
|
|
647 if (enc_utf8)
|
|
648 n = utf8len_tab[i];
|
|
649 else if (enc_dbcs == 0)
|
|
650 n = 1;
|
|
651 else
|
|
652 {
|
|
653 #if defined(WIN3264) || defined(WIN32UNIX)
|
|
654 /* enc_dbcs is set by setting 'fileencoding'. It becomes a Windows
|
|
655 * CodePage identifier, which we can pass directly in to Windows
|
|
656 * API */
|
|
657 n = IsDBCSLeadByteEx(enc_dbcs, (BYTE)i) ? 2 : 1;
|
|
658 #else
|
|
659 # ifdef MACOS
|
|
660 /*
|
|
661 * if mblen() is not available, character which MSB is turned on
|
|
662 * are treated as leading byte character. (note : This assumption
|
|
663 * is not always true.)
|
|
664 */
|
|
665 n = (i & 0x80) ? 2 : 1;
|
|
666 # else
|
|
667 char buf[MB_MAXBYTES];
|
|
668 # ifdef X_LOCALE
|
|
669 # ifndef mblen
|
|
670 # define mblen _Xmblen
|
|
671 # endif
|
|
672 # endif
|
|
673 if (i == NUL) /* just in case mblen() can't handle "" */
|
|
674 n = 1;
|
|
675 else
|
|
676 {
|
|
677 buf[0] = i;
|
|
678 buf[1] = 0;
|
|
679 #ifdef LEN_FROM_CONV
|
|
680 if (vimconv.vc_type != CONV_NONE)
|
|
681 {
|
|
682 /*
|
|
683 * string_convert() should fail when converting the first
|
|
684 * byte of a double-byte character.
|
|
685 */
|
|
686 p = string_convert(&vimconv, (char_u *)buf, NULL);
|
|
687 if (p != NULL)
|
|
688 {
|
|
689 vim_free(p);
|
|
690 n = 1;
|
|
691 }
|
|
692 else
|
|
693 n = 2;
|
|
694 }
|
|
695 else
|
|
696 #endif
|
|
697 {
|
|
698 /*
|
|
699 * mblen() should return -1 for invalid (means the leading
|
|
700 * multibyte) character. However there are some platforms
|
|
701 * where mblen() returns 0 for invalid character.
|
|
702 * Therefore, following condition includes 0.
|
|
703 */
|
|
704 if (mblen(buf, (size_t)1) <= 0)
|
|
705 n = 2;
|
|
706 else
|
|
707 n = 1;
|
|
708 }
|
|
709 }
|
|
710 # endif
|
|
711 #endif
|
|
712 }
|
|
713
|
|
714 mb_bytelen_tab[i] = n;
|
|
715 }
|
|
716
|
|
717 #ifdef LEN_FROM_CONV
|
|
718 convert_setup(&vimconv, NULL, NULL);
|
|
719 #endif
|
|
720
|
|
721 /* The cell width depends on the type of multi-byte characters. */
|
|
722 (void)init_chartab();
|
|
723
|
|
724 /* When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] */
|
|
725 screenalloc(FALSE);
|
|
726
|
|
727 /* When using Unicode, set default for 'fileencodings'. */
|
|
728 if (enc_utf8 && !option_was_set((char_u *)"fencs"))
|
|
729 set_string_option_direct((char_u *)"fencs", -1,
|
39
|
730 (char_u *)"ucs-bom,utf-8,default,latin1", OPT_FREE);
|
7
|
731 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT)
|
|
732 /* GNU gettext 0.10.37 supports this feature: set the codeset used for
|
|
733 * translated messages independently from the current locale. */
|
|
734 (void)bind_textdomain_codeset(VIMPACKAGE,
|
|
735 enc_utf8 ? "utf-8" : (char *)p_enc);
|
|
736 #endif
|
|
737
|
23
|
738 #ifdef WIN32
|
|
739 /* When changing 'encoding' while starting up, then convert the command
|
|
740 * line arguments from the active codepage to 'encoding'. */
|
|
741 if (starting != 0)
|
|
742 fix_arg_enc();
|
|
743 #endif
|
|
744
|
7
|
745 #ifdef FEAT_AUTOCMD
|
|
746 /* Fire an autocommand to let people do custom font setup. This must be
|
|
747 * after Vim has been setup for the new encoding. */
|
|
748 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf);
|
|
749 #endif
|
|
750
|
11
|
751 #ifdef FEAT_GUI_KDE
|
|
752 if (gui.in_use)
|
|
753 gui_mch_update_codec();
|
|
754 #endif
|
|
755
|
236
|
756 #ifdef FEAT_SYN_HL
|
|
757 /* Need to reload spell dictionaries */
|
|
758 spell_reload();
|
|
759 #endif
|
|
760
|
7
|
761 return NULL;
|
|
762 }
|
|
763
|
|
764 /*
|
|
765 * Return the size of the BOM for the current buffer:
|
|
766 * 0 - no BOM
|
|
767 * 2 - UCS-2 or UTF-16 BOM
|
|
768 * 4 - UCS-4 BOM
|
|
769 * 3 - UTF-8 BOM
|
|
770 */
|
|
771 int
|
|
772 bomb_size()
|
|
773 {
|
|
774 int n = 0;
|
|
775
|
|
776 if (curbuf->b_p_bomb && !curbuf->b_p_bin)
|
|
777 {
|
|
778 if (*curbuf->b_p_fenc == NUL)
|
|
779 {
|
|
780 if (enc_utf8)
|
|
781 {
|
|
782 if (enc_unicode != 0)
|
|
783 n = enc_unicode;
|
|
784 else
|
|
785 n = 3;
|
|
786 }
|
|
787 }
|
|
788 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0)
|
|
789 n = 3;
|
|
790 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0
|
|
791 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0)
|
|
792 n = 2;
|
|
793 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0)
|
|
794 n = 4;
|
|
795 }
|
|
796 return n;
|
|
797 }
|
|
798
|
|
799 /*
|
|
800 * Get class of pointer:
|
|
801 * 0 for blank or NUL
|
|
802 * 1 for punctuation
|
|
803 * 2 for an (ASCII) word character
|
|
804 * >2 for other word characters
|
|
805 */
|
|
806 int
|
|
807 mb_get_class(p)
|
|
808 char_u *p;
|
|
809 {
|
|
810 if (MB_BYTE2LEN(p[0]) == 1)
|
|
811 {
|
|
812 if (p[0] == NUL || vim_iswhite(p[0]))
|
|
813 return 0;
|
|
814 if (vim_iswordc(p[0]))
|
|
815 return 2;
|
|
816 return 1;
|
|
817 }
|
|
818 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL)
|
|
819 return dbcs_class(p[0], p[1]);
|
|
820 if (enc_utf8)
|
|
821 return utf_class(utf_ptr2char(p));
|
|
822 return 0;
|
|
823 }
|
|
824
|
|
825 /*
|
|
826 * Get class of a double-byte character. This always returns 3 or bigger.
|
|
827 * TODO: Should return 1 for punctuation.
|
|
828 */
|
|
829 int
|
|
830 dbcs_class(lead, trail)
|
|
831 unsigned lead;
|
|
832 unsigned trail;
|
|
833 {
|
|
834 switch (enc_dbcs)
|
|
835 {
|
|
836 /* please add classfy routine for your language in here */
|
|
837
|
|
838 case DBCS_JPNU: /* ? */
|
|
839 case DBCS_JPN:
|
|
840 {
|
|
841 /* JIS code classification */
|
|
842 unsigned char lb = lead;
|
|
843 unsigned char tb = trail;
|
|
844
|
|
845 /* convert process code to JIS */
|
|
846 # if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS)
|
|
847 /* process code is SJIS */
|
|
848 if (lb <= 0x9f)
|
|
849 lb = (lb - 0x81) * 2 + 0x21;
|
|
850 else
|
|
851 lb = (lb - 0xc1) * 2 + 0x21;
|
|
852 if (tb <= 0x7e)
|
|
853 tb -= 0x1f;
|
|
854 else if (tb <= 0x9e)
|
|
855 tb -= 0x20;
|
|
856 else
|
|
857 {
|
|
858 tb -= 0x7e;
|
|
859 lb += 1;
|
|
860 }
|
|
861 # else
|
|
862 /*
|
|
863 * XXX: Code page identification can not use with all
|
|
864 * system! So, some other encoding information
|
|
865 * will be needed.
|
|
866 * In japanese: SJIS,EUC,UNICODE,(JIS)
|
|
867 * Note that JIS-code system don't use as
|
|
868 * process code in most system because it uses
|
|
869 * escape sequences(JIS is context depend encoding).
|
|
870 */
|
|
871 /* assume process code is JAPANESE-EUC */
|
|
872 lb &= 0x7f;
|
|
873 tb &= 0x7f;
|
|
874 # endif
|
|
875 /* exceptions */
|
|
876 switch (lb << 8 | tb)
|
|
877 {
|
|
878 case 0x2121: /* ZENKAKU space */
|
|
879 return 0;
|
|
880 case 0x2122: /* KU-TEN (Japanese comma) */
|
|
881 case 0x2123: /* TOU-TEN (Japanese period) */
|
|
882 case 0x2124: /* ZENKAKU comma */
|
|
883 case 0x2125: /* ZENKAKU period */
|
|
884 return 1;
|
|
885 case 0x213c: /* prolongedsound handled as KATAKANA */
|
|
886 return 13;
|
|
887 }
|
|
888 /* sieved by KU code */
|
|
889 switch (lb)
|
|
890 {
|
|
891 case 0x21:
|
|
892 case 0x22:
|
|
893 /* special symbols */
|
|
894 return 10;
|
|
895 case 0x23:
|
|
896 /* alpha-numeric */
|
|
897 return 11;
|
|
898 case 0x24:
|
|
899 /* hiragana */
|
|
900 return 12;
|
|
901 case 0x25:
|
|
902 /* katakana */
|
|
903 return 13;
|
|
904 case 0x26:
|
|
905 /* greek */
|
|
906 return 14;
|
|
907 case 0x27:
|
|
908 /* russian */
|
|
909 return 15;
|
|
910 case 0x28:
|
|
911 /* lines */
|
|
912 return 16;
|
|
913 default:
|
|
914 /* kanji */
|
|
915 return 17;
|
|
916 }
|
|
917 }
|
|
918
|
|
919 case DBCS_KORU: /* ? */
|
|
920 case DBCS_KOR:
|
|
921 {
|
|
922 /* KS code classification */
|
|
923 unsigned char c1 = lead;
|
|
924 unsigned char c2 = trail;
|
|
925
|
|
926 /*
|
|
927 * 20 : Hangul
|
|
928 * 21 : Hanja
|
|
929 * 22 : Symbols
|
|
930 * 23 : Alpha-numeric/Roman Letter (Full width)
|
|
931 * 24 : Hangul Letter(Alphabet)
|
|
932 * 25 : Roman Numeral/Greek Letter
|
|
933 * 26 : Box Drawings
|
|
934 * 27 : Unit Symbols
|
|
935 * 28 : Circled/Parenthesized Letter
|
|
936 * 29 : Hirigana/Katakana
|
|
937 * 30 : Cyrillic Letter
|
|
938 */
|
|
939
|
|
940 if (c1 >= 0xB0 && c1 <= 0xC8)
|
|
941 /* Hangul */
|
|
942 return 20;
|
|
943 #if defined(WIN3264) || defined(WIN32UNIX)
|
|
944 else if (c1 <= 0xA0 || c2 <= 0xA0)
|
|
945 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */
|
|
946 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE
|
|
947 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0
|
|
948 */
|
|
949 return 20;
|
|
950 #endif
|
|
951
|
|
952 else if (c1 >= 0xCA && c1 <= 0xFD)
|
|
953 /* Hanja */
|
|
954 return 21;
|
|
955 else switch (c1)
|
|
956 {
|
|
957 case 0xA1:
|
|
958 case 0xA2:
|
|
959 /* Symbols */
|
|
960 return 22;
|
|
961 case 0xA3:
|
|
962 /* Alpha-numeric */
|
|
963 return 23;
|
|
964 case 0xA4:
|
|
965 /* Hangul Letter(Alphabet) */
|
|
966 return 24;
|
|
967 case 0xA5:
|
|
968 /* Roman Numeral/Greek Letter */
|
|
969 return 25;
|
|
970 case 0xA6:
|
|
971 /* Box Drawings */
|
|
972 return 26;
|
|
973 case 0xA7:
|
|
974 /* Unit Symbols */
|
|
975 return 27;
|
|
976 case 0xA8:
|
|
977 case 0xA9:
|
|
978 if (c2 <= 0xAF)
|
|
979 return 25; /* Roman Letter */
|
|
980 else if (c2 >= 0xF6)
|
|
981 return 22; /* Symbols */
|
|
982 else
|
|
983 /* Circled/Parenthesized Letter */
|
|
984 return 28;
|
|
985 case 0xAA:
|
|
986 case 0xAB:
|
|
987 /* Hirigana/Katakana */
|
|
988 return 29;
|
|
989 case 0xAC:
|
|
990 /* Cyrillic Letter */
|
|
991 return 30;
|
|
992 }
|
|
993 }
|
|
994 default:
|
|
995 break;
|
|
996 }
|
|
997 return 3;
|
|
998 }
|
|
999
|
|
1000 /*
|
|
1001 * mb_char2len() function pointer.
|
|
1002 * Return length in bytes of character "c".
|
|
1003 * Returns 1 for a single-byte character.
|
|
1004 */
|
|
1005 /* ARGSUSED */
|
|
1006 int
|
|
1007 latin_char2len(c)
|
|
1008 int c;
|
|
1009 {
|
|
1010 return 1;
|
|
1011 }
|
|
1012
|
|
1013 static int
|
|
1014 dbcs_char2len(c)
|
|
1015 int c;
|
|
1016 {
|
|
1017 if (c >= 0x100)
|
|
1018 return 2;
|
|
1019 return 1;
|
|
1020 }
|
|
1021
|
|
1022 /*
|
|
1023 * mb_char2bytes() function pointer.
|
|
1024 * Convert a character to its bytes.
|
|
1025 * Returns the length in bytes.
|
|
1026 */
|
|
1027 int
|
|
1028 latin_char2bytes(c, buf)
|
|
1029 int c;
|
|
1030 char_u *buf;
|
|
1031 {
|
|
1032 buf[0] = c;
|
|
1033 return 1;
|
|
1034 }
|
|
1035
|
|
1036 static int
|
|
1037 dbcs_char2bytes(c, buf)
|
|
1038 int c;
|
|
1039 char_u *buf;
|
|
1040 {
|
|
1041 if (c >= 0x100)
|
|
1042 {
|
|
1043 buf[0] = (unsigned)c >> 8;
|
|
1044 buf[1] = c;
|
221
|
1045 /* Never use a NUL byte, it causes lots of trouble. It's an invalid
|
|
1046 * character anyway. */
|
|
1047 if (buf[1] == NUL)
|
|
1048 buf[1] = '\n';
|
7
|
1049 return 2;
|
|
1050 }
|
|
1051 buf[0] = c;
|
|
1052 return 1;
|
|
1053 }
|
|
1054
|
|
1055 /*
|
|
1056 * mb_ptr2len_check() function pointer.
|
|
1057 * Get byte length of character at "*p" but stop at a NUL.
|
|
1058 * For UTF-8 this includes following composing characters.
|
|
1059 * Returns 0 when *p is NUL.
|
|
1060 *
|
|
1061 */
|
|
1062 int
|
|
1063 latin_ptr2len_check(p)
|
|
1064 char_u *p;
|
|
1065 {
|
|
1066 return MB_BYTE2LEN(*p);
|
|
1067 }
|
|
1068
|
|
1069 static int
|
|
1070 dbcs_ptr2len_check(p)
|
|
1071 char_u *p;
|
|
1072 {
|
|
1073 int len;
|
|
1074
|
|
1075 /* Check if second byte is not missing. */
|
|
1076 len = MB_BYTE2LEN(*p);
|
|
1077 if (len == 2 && p[1] == NUL)
|
|
1078 len = 1;
|
|
1079 return len;
|
|
1080 }
|
|
1081
|
|
1082 struct interval
|
|
1083 {
|
|
1084 unsigned short first;
|
|
1085 unsigned short last;
|
|
1086 };
|
|
1087 static int intable __ARGS((struct interval *table, size_t size, int c));
|
|
1088
|
|
1089 /*
|
|
1090 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]".
|
|
1091 */
|
|
1092 static int
|
|
1093 intable(table, size, c)
|
|
1094 struct interval *table;
|
|
1095 size_t size;
|
|
1096 int c;
|
|
1097 {
|
|
1098 int mid, bot, top;
|
|
1099
|
|
1100 /* first quick check for Latin1 etc. characters */
|
|
1101 if (c < table[0].first)
|
|
1102 return FALSE;
|
|
1103
|
|
1104 /* binary search in table */
|
|
1105 bot = 0;
|
|
1106 top = size / sizeof(struct interval) - 1;
|
|
1107 while (top >= bot)
|
|
1108 {
|
|
1109 mid = (bot + top) / 2;
|
|
1110 if (table[mid].last < c)
|
|
1111 bot = mid + 1;
|
|
1112 else if (table[mid].first > c)
|
|
1113 top = mid - 1;
|
|
1114 else
|
|
1115 return TRUE;
|
|
1116 }
|
|
1117 return FALSE;
|
|
1118 }
|
|
1119
|
|
1120 /*
|
|
1121 * For UTF-8 character "c" return 2 for a double-width character, 1 for others.
|
|
1122 * Returns 4 or 6 for an unprintable character.
|
|
1123 * Is only correct for characters >= 0x80.
|
|
1124 * When p_ambw is "double", return 2 for a character with East Asian Width
|
|
1125 * class 'A'(mbiguous).
|
|
1126 */
|
|
1127 int
|
|
1128 utf_char2cells(c)
|
|
1129 int c;
|
|
1130 {
|
|
1131 /* sorted list of non-overlapping intervals of East Asian Ambiguous
|
|
1132 * characters, generated with:
|
|
1133 * "uniset +WIDTH-A -cat=Me -cat=Mn -cat=Cf c" */
|
|
1134 static struct interval ambiguous[] = {
|
|
1135 {0x00A1, 0x00A1}, {0x00A4, 0x00A4}, {0x00A7, 0x00A8},
|
|
1136 {0x00AA, 0x00AA}, {0x00AE, 0x00AE}, {0x00B0, 0x00B4},
|
|
1137 {0x00B6, 0x00BA}, {0x00BC, 0x00BF}, {0x00C6, 0x00C6},
|
|
1138 {0x00D0, 0x00D0}, {0x00D7, 0x00D8}, {0x00DE, 0x00E1},
|
|
1139 {0x00E6, 0x00E6}, {0x00E8, 0x00EA}, {0x00EC, 0x00ED},
|
|
1140 {0x00F0, 0x00F0}, {0x00F2, 0x00F3}, {0x00F7, 0x00FA},
|
|
1141 {0x00FC, 0x00FC}, {0x00FE, 0x00FE}, {0x0101, 0x0101},
|
|
1142 {0x0111, 0x0111}, {0x0113, 0x0113}, {0x011B, 0x011B},
|
|
1143 {0x0126, 0x0127}, {0x012B, 0x012B}, {0x0131, 0x0133},
|
|
1144 {0x0138, 0x0138}, {0x013F, 0x0142}, {0x0144, 0x0144},
|
|
1145 {0x0148, 0x014B}, {0x014D, 0x014D}, {0x0152, 0x0153},
|
|
1146 {0x0166, 0x0167}, {0x016B, 0x016B}, {0x01CE, 0x01CE},
|
|
1147 {0x01D0, 0x01D0}, {0x01D2, 0x01D2}, {0x01D4, 0x01D4},
|
|
1148 {0x01D6, 0x01D6}, {0x01D8, 0x01D8}, {0x01DA, 0x01DA},
|
|
1149 {0x01DC, 0x01DC}, {0x0251, 0x0251}, {0x0261, 0x0261},
|
|
1150 {0x02C4, 0x02C4}, {0x02C7, 0x02C7}, {0x02C9, 0x02CB},
|
|
1151 {0x02CD, 0x02CD}, {0x02D0, 0x02D0}, {0x02D8, 0x02DB},
|
|
1152 {0x02DD, 0x02DD}, {0x02DF, 0x02DF}, {0x0391, 0x03A1},
|
|
1153 {0x03A3, 0x03A9}, {0x03B1, 0x03C1}, {0x03C3, 0x03C9},
|
|
1154 {0x0401, 0x0401}, {0x0410, 0x044F}, {0x0451, 0x0451},
|
|
1155 {0x2010, 0x2010}, {0x2013, 0x2016}, {0x2018, 0x2019},
|
|
1156 {0x201C, 0x201D}, {0x2020, 0x2022}, {0x2024, 0x2027},
|
|
1157 {0x2030, 0x2030}, {0x2032, 0x2033}, {0x2035, 0x2035},
|
|
1158 {0x203B, 0x203B}, {0x203E, 0x203E}, {0x2074, 0x2074},
|
|
1159 {0x207F, 0x207F}, {0x2081, 0x2084}, {0x20AC, 0x20AC},
|
|
1160 {0x2103, 0x2103}, {0x2105, 0x2105}, {0x2109, 0x2109},
|
|
1161 {0x2113, 0x2113}, {0x2116, 0x2116}, {0x2121, 0x2122},
|
|
1162 {0x2126, 0x2126}, {0x212B, 0x212B}, {0x2153, 0x2154},
|
|
1163 {0x215B, 0x215E}, {0x2160, 0x216B}, {0x2170, 0x2179},
|
|
1164 {0x2190, 0x2199}, {0x21B8, 0x21B9}, {0x21D2, 0x21D2},
|
|
1165 {0x21D4, 0x21D4}, {0x21E7, 0x21E7}, {0x2200, 0x2200},
|
|
1166 {0x2202, 0x2203}, {0x2207, 0x2208}, {0x220B, 0x220B},
|
|
1167 {0x220F, 0x220F}, {0x2211, 0x2211}, {0x2215, 0x2215},
|
|
1168 {0x221A, 0x221A}, {0x221D, 0x2220}, {0x2223, 0x2223},
|
|
1169 {0x2225, 0x2225}, {0x2227, 0x222C}, {0x222E, 0x222E},
|
|
1170 {0x2234, 0x2237}, {0x223C, 0x223D}, {0x2248, 0x2248},
|
|
1171 {0x224C, 0x224C}, {0x2252, 0x2252}, {0x2260, 0x2261},
|
|
1172 {0x2264, 0x2267}, {0x226A, 0x226B}, {0x226E, 0x226F},
|
|
1173 {0x2282, 0x2283}, {0x2286, 0x2287}, {0x2295, 0x2295},
|
|
1174 {0x2299, 0x2299}, {0x22A5, 0x22A5}, {0x22BF, 0x22BF},
|
|
1175 {0x2312, 0x2312}, {0x2460, 0x24E9}, {0x24EB, 0x254B},
|
|
1176 {0x2550, 0x2573}, {0x2580, 0x258F}, {0x2592, 0x2595},
|
|
1177 {0x25A0, 0x25A1}, {0x25A3, 0x25A9}, {0x25B2, 0x25B3},
|
|
1178 {0x25B6, 0x25B7}, {0x25BC, 0x25BD}, {0x25C0, 0x25C1},
|
|
1179 {0x25C6, 0x25C8}, {0x25CB, 0x25CB}, {0x25CE, 0x25D1},
|
|
1180 {0x25E2, 0x25E5}, {0x25EF, 0x25EF}, {0x2605, 0x2606},
|
|
1181 {0x2609, 0x2609}, {0x260E, 0x260F}, {0x2614, 0x2615},
|
|
1182 {0x261C, 0x261C}, {0x261E, 0x261E}, {0x2640, 0x2640},
|
|
1183 {0x2642, 0x2642}, {0x2660, 0x2661}, {0x2663, 0x2665},
|
|
1184 {0x2667, 0x266A}, {0x266C, 0x266D}, {0x266F, 0x266F},
|
|
1185 {0x273D, 0x273D}, {0x2776, 0x277F}, {0xE000, 0xF8FF},
|
|
1186 {0xFFFD, 0xFFFD}, /* {0xF0000, 0xFFFFD}, {0x100000, 0x10FFFD} */
|
|
1187 };
|
|
1188
|
|
1189 if (c >= 0x100)
|
|
1190 {
|
|
1191 #ifdef USE_WCHAR_FUNCTIONS
|
|
1192 /*
|
|
1193 * Assume the library function wcwidth() works better than our own
|
|
1194 * stuff. It should return 1 for ambiguous width chars!
|
|
1195 */
|
|
1196 int n = wcwidth(c);
|
|
1197
|
|
1198 if (n < 0)
|
|
1199 return 6; /* unprintable, displays <xxxx> */
|
|
1200 if (n > 1)
|
|
1201 return n;
|
|
1202 #else
|
|
1203 if (!utf_printable(c))
|
|
1204 return 6; /* unprintable, displays <xxxx> */
|
|
1205 if (c >= 0x1100
|
|
1206 && (c <= 0x115f /* Hangul Jamo */
|
|
1207 || c == 0x2329
|
|
1208 || c == 0x232a
|
|
1209 || (c >= 0x2e80 && c <= 0xa4cf
|
|
1210 && c != 0x303f) /* CJK ... Yi */
|
|
1211 || (c >= 0xac00 && c <= 0xd7a3) /* Hangul Syllables */
|
|
1212 || (c >= 0xf900 && c <= 0xfaff) /* CJK Compatibility
|
|
1213 Ideographs */
|
|
1214 || (c >= 0xfe30 && c <= 0xfe6f) /* CJK Compatibility Forms */
|
|
1215 || (c >= 0xff00 && c <= 0xff60) /* Fullwidth Forms */
|
|
1216 || (c >= 0xffe0 && c <= 0xffe6)
|
|
1217 || (c >= 0x20000 && c <= 0x2fffd)
|
|
1218 || (c >= 0x30000 && c <= 0x3fffd)))
|
|
1219 return 2;
|
|
1220 #endif
|
|
1221 }
|
|
1222
|
|
1223 /* Characters below 0x100 are influenced by 'isprint' option */
|
|
1224 else if (c >= 0x80 && !vim_isprintc(c))
|
|
1225 return 4; /* unprintable, displays <xx> */
|
|
1226
|
|
1227 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c))
|
|
1228 return 2;
|
|
1229
|
|
1230 return 1;
|
|
1231 }
|
|
1232
|
|
1233 /*
|
|
1234 * mb_ptr2cells() function pointer.
|
|
1235 * Return the number of display cells character at "*p" occupies.
|
|
1236 * This doesn't take care of unprintable characters, use ptr2cells() for that.
|
|
1237 */
|
|
1238 /*ARGSUSED*/
|
|
1239 int
|
|
1240 latin_ptr2cells(p)
|
|
1241 char_u *p;
|
|
1242 {
|
|
1243 return 1;
|
|
1244 }
|
|
1245
|
|
1246 int
|
|
1247 utf_ptr2cells(p)
|
|
1248 char_u *p;
|
|
1249 {
|
|
1250 int c;
|
|
1251
|
|
1252 /* Need to convert to a wide character. */
|
|
1253 if (*p >= 0x80)
|
|
1254 {
|
|
1255 c = utf_ptr2char(p);
|
|
1256 /* An illegal byte is displayed as <xx>. */
|
|
1257 if (utf_ptr2len_check(p) == 1 || c == NUL)
|
|
1258 return 4;
|
|
1259 /* If the char is ASCII it must be an overlong sequence. */
|
|
1260 if (c < 0x80)
|
|
1261 return char2cells(c);
|
|
1262 return utf_char2cells(c);
|
|
1263 }
|
|
1264 return 1;
|
|
1265 }
|
|
1266
|
|
1267 int
|
|
1268 dbcs_ptr2cells(p)
|
|
1269 char_u *p;
|
|
1270 {
|
|
1271 /* Number of cells is equal to number of bytes, except for euc-jp when
|
|
1272 * the first byte is 0x8e. */
|
|
1273 if (enc_dbcs == DBCS_JPNU && *p == 0x8e)
|
|
1274 return 1;
|
|
1275 return MB_BYTE2LEN(*p);
|
|
1276 }
|
|
1277
|
|
1278 /*
|
|
1279 * mb_char2cells() function pointer.
|
|
1280 * Return the number of display cells character "c" occupies.
|
|
1281 * Only takes care of multi-byte chars, not "^C" and such.
|
|
1282 */
|
|
1283 /*ARGSUSED*/
|
|
1284 int
|
|
1285 latin_char2cells(c)
|
|
1286 int c;
|
|
1287 {
|
|
1288 return 1;
|
|
1289 }
|
|
1290
|
|
1291 static int
|
|
1292 dbcs_char2cells(c)
|
|
1293 int c;
|
|
1294 {
|
|
1295 /* Number of cells is equal to number of bytes, except for euc-jp when
|
|
1296 * the first byte is 0x8e. */
|
|
1297 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e)
|
|
1298 return 1;
|
|
1299 /* use the first byte */
|
|
1300 return MB_BYTE2LEN((unsigned)c >> 8);
|
|
1301 }
|
|
1302
|
|
1303 /*
|
|
1304 * mb_off2cells() function pointer.
|
|
1305 * Return number of display cells for char at ScreenLines[off].
|
|
1306 * Caller must make sure "off" and "off + 1" are valid!
|
|
1307 */
|
|
1308 /*ARGSUSED*/
|
|
1309 int
|
|
1310 latin_off2cells(off)
|
|
1311 unsigned off;
|
|
1312 {
|
|
1313 return 1;
|
|
1314 }
|
|
1315
|
|
1316 int
|
|
1317 dbcs_off2cells(off)
|
|
1318 unsigned off;
|
|
1319 {
|
|
1320 /* Number of cells is equal to number of bytes, except for euc-jp when
|
|
1321 * the first byte is 0x8e. */
|
|
1322 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e)
|
|
1323 return 1;
|
|
1324 return MB_BYTE2LEN(ScreenLines[off]);
|
|
1325 }
|
|
1326
|
|
1327 int
|
|
1328 utf_off2cells(off)
|
|
1329 unsigned off;
|
|
1330 {
|
|
1331 return ScreenLines[off + 1] == 0 ? 2 : 1;
|
|
1332 }
|
|
1333
|
|
1334 /*
|
|
1335 * mb_ptr2char() function pointer.
|
|
1336 * Convert a byte sequence into a character.
|
|
1337 */
|
|
1338 int
|
|
1339 latin_ptr2char(p)
|
|
1340 char_u *p;
|
|
1341 {
|
|
1342 return *p;
|
|
1343 }
|
|
1344
|
|
1345 static int
|
|
1346 dbcs_ptr2char(p)
|
|
1347 char_u *p;
|
|
1348 {
|
|
1349 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL)
|
|
1350 return (p[0] << 8) + p[1];
|
|
1351 return *p;
|
|
1352 }
|
|
1353
|
|
1354 /*
|
|
1355 * Convert a UTF-8 byte sequence to a wide character.
|
|
1356 * If the sequence is illegal or truncated by a NUL the first byte is
|
|
1357 * returned.
|
|
1358 * Does not include composing characters, of course.
|
|
1359 */
|
|
1360 int
|
|
1361 utf_ptr2char(p)
|
|
1362 char_u *p;
|
|
1363 {
|
|
1364 int len;
|
|
1365
|
|
1366 if (p[0] < 0x80) /* be quick for ASCII */
|
|
1367 return p[0];
|
|
1368
|
|
1369 len = utf8len_tab[p[0]];
|
|
1370 if ((p[1] & 0xc0) == 0x80)
|
|
1371 {
|
|
1372 if (len == 2)
|
|
1373 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f);
|
|
1374 if ((p[2] & 0xc0) == 0x80)
|
|
1375 {
|
|
1376 if (len == 3)
|
|
1377 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6)
|
|
1378 + (p[2] & 0x3f);
|
|
1379 if ((p[3] & 0xc0) == 0x80)
|
|
1380 {
|
|
1381 if (len == 4)
|
|
1382 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12)
|
|
1383 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f);
|
|
1384 if ((p[4] & 0xc0) == 0x80)
|
|
1385 {
|
|
1386 if (len == 5)
|
|
1387 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18)
|
|
1388 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6)
|
|
1389 + (p[4] & 0x3f);
|
|
1390 if ((p[5] & 0xc0) == 0x80 && len == 6)
|
|
1391 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24)
|
|
1392 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12)
|
|
1393 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f);
|
|
1394 }
|
|
1395 }
|
|
1396 }
|
|
1397 }
|
|
1398 /* Illegal value, just return the first byte */
|
|
1399 return p[0];
|
|
1400 }
|
|
1401
|
|
1402 /*
|
|
1403 * Get character at **pp and advance *pp to the next character.
|
|
1404 * Note: composing characters are skipped!
|
|
1405 */
|
|
1406 int
|
|
1407 mb_ptr2char_adv(pp)
|
|
1408 char_u **pp;
|
|
1409 {
|
|
1410 int c;
|
|
1411
|
|
1412 c = (*mb_ptr2char)(*pp);
|
|
1413 *pp += (*mb_ptr2len_check)(*pp);
|
|
1414 return c;
|
|
1415 }
|
|
1416
|
|
1417 #if defined(FEAT_ARABIC) || defined(PROTO)
|
|
1418 /*
|
|
1419 * Check whether we are dealing with Arabic combining characters.
|
|
1420 * Note: these are NOT really composing characters!
|
|
1421 */
|
|
1422 int
|
|
1423 arabic_combine(one, two)
|
|
1424 int one; /* first character */
|
|
1425 int two; /* character just after "one" */
|
|
1426 {
|
|
1427 if (one == a_LAM)
|
|
1428 return arabic_maycombine(two);
|
|
1429 return FALSE;
|
|
1430 }
|
|
1431
|
|
1432 /*
|
|
1433 * Check whether we are dealing with a character that could be regarded as an
|
|
1434 * Arabic combining character, need to check the character before this.
|
|
1435 */
|
|
1436 int
|
|
1437 arabic_maycombine(two)
|
|
1438 int two;
|
|
1439 {
|
|
1440 if (p_arshape && !p_tbidi)
|
|
1441 return (two == a_ALEF_MADDA
|
|
1442 || two == a_ALEF_HAMZA_ABOVE
|
|
1443 || two == a_ALEF_HAMZA_BELOW
|
|
1444 || two == a_ALEF);
|
|
1445 return FALSE;
|
|
1446 }
|
|
1447
|
|
1448 /*
|
|
1449 * Check if the character pointed to by "p2" is a composing character when it
|
|
1450 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which
|
|
1451 * behaves like a composing character.
|
|
1452 */
|
|
1453 int
|
|
1454 utf_composinglike(p1, p2)
|
|
1455 char_u *p1;
|
|
1456 char_u *p2;
|
|
1457 {
|
|
1458 int c2;
|
|
1459
|
|
1460 c2 = utf_ptr2char(p2);
|
|
1461 if (utf_iscomposing(c2))
|
|
1462 return TRUE;
|
|
1463 if (!arabic_maycombine(c2))
|
|
1464 return FALSE;
|
|
1465 return arabic_combine(utf_ptr2char(p1), c2);
|
|
1466 }
|
|
1467 #endif
|
|
1468
|
|
1469 /*
|
|
1470 * Convert a UTF-8 byte string to a wide chararacter. Also get up to two
|
|
1471 * composing characters.
|
|
1472 */
|
|
1473 int
|
|
1474 utfc_ptr2char(p, p1, p2)
|
|
1475 char_u *p;
|
|
1476 int *p1; /* return: first composing char or 0 */
|
|
1477 int *p2; /* return: second composing char or 0 */
|
|
1478 {
|
|
1479 int len;
|
|
1480 int c;
|
|
1481 int cc;
|
|
1482
|
|
1483 c = utf_ptr2char(p);
|
|
1484 len = utf_ptr2len_check(p);
|
|
1485 /* Only accept a composing char when the first char isn't illegal. */
|
|
1486 if ((len > 1 || *p < 0x80)
|
|
1487 && p[len] >= 0x80
|
|
1488 && UTF_COMPOSINGLIKE(p, p + len))
|
|
1489 {
|
|
1490 *p1 = utf_ptr2char(p + len);
|
|
1491 len += utf_ptr2len_check(p + len);
|
|
1492 if (p[len] >= 0x80 && utf_iscomposing(cc = utf_ptr2char(p + len)))
|
|
1493 *p2 = cc;
|
|
1494 else
|
|
1495 *p2 = 0;
|
|
1496 }
|
|
1497 else
|
|
1498 {
|
|
1499 *p1 = 0;
|
|
1500 *p2 = 0;
|
|
1501 }
|
|
1502 return c;
|
|
1503 }
|
|
1504
|
|
1505 /*
|
|
1506 * Convert a UTF-8 byte string to a wide chararacter. Also get up to two
|
|
1507 * composing characters. Use no more than p[maxlen].
|
|
1508 */
|
|
1509 int
|
|
1510 utfc_ptr2char_len(p, p1, p2, maxlen)
|
|
1511 char_u *p;
|
|
1512 int *p1; /* return: first composing char or 0 */
|
|
1513 int *p2; /* return: second composing char or 0 */
|
|
1514 int maxlen;
|
|
1515 {
|
|
1516 int len;
|
|
1517 int c;
|
|
1518 int cc;
|
|
1519
|
|
1520 c = utf_ptr2char(p);
|
|
1521 len = utf_ptr2len_check_len(p, maxlen);
|
|
1522 /* Only accept a composing char when the first char isn't illegal. */
|
|
1523 if ((len > 1 || *p < 0x80)
|
|
1524 && len < maxlen
|
|
1525 && p[len] >= 0x80
|
|
1526 && UTF_COMPOSINGLIKE(p, p + len))
|
|
1527 {
|
|
1528 *p1 = utf_ptr2char(p + len);
|
|
1529 len += utf_ptr2len_check_len(p + len, maxlen - len);
|
|
1530 if (len < maxlen
|
|
1531 && p[len] >= 0x80
|
|
1532 && utf_iscomposing(cc = utf_ptr2char(p + len)))
|
|
1533 *p2 = cc;
|
|
1534 else
|
|
1535 *p2 = 0;
|
|
1536 }
|
|
1537 else
|
|
1538 {
|
|
1539 *p1 = 0;
|
|
1540 *p2 = 0;
|
|
1541 }
|
|
1542 return c;
|
|
1543 }
|
|
1544
|
|
1545 /*
|
|
1546 * Convert the character at screen position "off" to a sequence of bytes.
|
|
1547 * Includes the composing characters.
|
|
1548 * "buf" must at least have the length MB_MAXBYTES.
|
|
1549 * Returns the produced number of bytes.
|
|
1550 */
|
|
1551 int
|
|
1552 utfc_char2bytes(off, buf)
|
|
1553 int off;
|
|
1554 char_u *buf;
|
|
1555 {
|
|
1556 int len;
|
|
1557
|
|
1558 len = utf_char2bytes(ScreenLinesUC[off], buf);
|
|
1559 if (ScreenLinesC1[off] != 0)
|
|
1560 {
|
|
1561 len += utf_char2bytes(ScreenLinesC1[off], buf + len);
|
|
1562 if (ScreenLinesC2[off] != 0)
|
|
1563 len += utf_char2bytes(ScreenLinesC2[off], buf + len);
|
|
1564 }
|
|
1565 return len;
|
|
1566 }
|
|
1567
|
|
1568 /*
|
|
1569 * Get the length of a UTF-8 byte sequence, not including any following
|
|
1570 * composing characters.
|
|
1571 * Returns 0 for "".
|
|
1572 * Returns 1 for an illegal byte sequence.
|
|
1573 */
|
|
1574 int
|
|
1575 utf_ptr2len_check(p)
|
|
1576 char_u *p;
|
|
1577 {
|
|
1578 int len;
|
|
1579 int i;
|
|
1580
|
|
1581 if (*p == NUL)
|
|
1582 return 0;
|
|
1583 len = utf8len_tab[*p];
|
|
1584 for (i = 1; i < len; ++i)
|
|
1585 if ((p[i] & 0xc0) != 0x80)
|
|
1586 return 1;
|
|
1587 return len;
|
|
1588 }
|
|
1589
|
|
1590 /*
|
|
1591 * Return length of UTF-8 character, obtained from the first byte.
|
|
1592 * "b" must be between 0 and 255!
|
|
1593 */
|
|
1594 int
|
|
1595 utf_byte2len(b)
|
|
1596 int b;
|
|
1597 {
|
|
1598 return utf8len_tab[b];
|
|
1599 }
|
|
1600
|
|
1601 /*
|
|
1602 * Get the length of UTF-8 byte sequence "p[size]". Does not include any
|
|
1603 * following composing characters.
|
|
1604 * Returns 1 for "".
|
|
1605 * Returns 1 for an illegal byte sequence.
|
|
1606 * Returns number > "size" for an incomplete byte sequence.
|
|
1607 */
|
|
1608 int
|
|
1609 utf_ptr2len_check_len(p, size)
|
|
1610 char_u *p;
|
|
1611 int size;
|
|
1612 {
|
|
1613 int len;
|
|
1614 int i;
|
|
1615
|
|
1616 if (*p == NUL)
|
|
1617 return 1;
|
|
1618 len = utf8len_tab[*p];
|
|
1619 if (len > size)
|
|
1620 return len; /* incomplete byte sequence. */
|
|
1621 for (i = 1; i < len; ++i)
|
|
1622 if ((p[i] & 0xc0) != 0x80)
|
|
1623 return 1;
|
|
1624 return len;
|
|
1625 }
|
|
1626
|
|
1627 /*
|
|
1628 * Return the number of bytes the UTF-8 encoding of the character at "p" takes.
|
|
1629 * This includes following composing characters.
|
|
1630 */
|
|
1631 int
|
|
1632 utfc_ptr2len_check(p)
|
|
1633 char_u *p;
|
|
1634 {
|
|
1635 int len;
|
318
|
1636 int b0 = *p;
|
7
|
1637 #ifdef FEAT_ARABIC
|
|
1638 int prevlen;
|
|
1639 #endif
|
|
1640
|
318
|
1641 if (b0 == NUL)
|
7
|
1642 return 0;
|
318
|
1643 if (b0 < 0x80 && p[1] < 0x80) /* be quick for ASCII */
|
7
|
1644 return 1;
|
|
1645
|
|
1646 /* Skip over first UTF-8 char, stopping at a NUL byte. */
|
|
1647 len = utf_ptr2len_check(p);
|
|
1648
|
|
1649 /* Check for illegal byte. */
|
318
|
1650 if (len == 1 && b0 >= 0x80)
|
7
|
1651 return 1;
|
|
1652
|
|
1653 /*
|
|
1654 * Check for composing characters. We can handle only the first two, but
|
|
1655 * skip all of them (otherwise the cursor would get stuck).
|
|
1656 */
|
|
1657 #ifdef FEAT_ARABIC
|
|
1658 prevlen = 0;
|
|
1659 #endif
|
|
1660 for (;;)
|
|
1661 {
|
|
1662 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len))
|
|
1663 return len;
|
|
1664
|
|
1665 /* Skip over composing char */
|
|
1666 #ifdef FEAT_ARABIC
|
|
1667 prevlen = len;
|
|
1668 #endif
|
|
1669 len += utf_ptr2len_check(p + len);
|
|
1670 }
|
|
1671 }
|
|
1672
|
|
1673 /*
|
|
1674 * Return the number of bytes the UTF-8 encoding of the character at "p[size]"
|
|
1675 * takes. This includes following composing characters.
|
|
1676 * Returns 1 for an illegal char or an incomplete byte sequence.
|
|
1677 */
|
|
1678 int
|
|
1679 utfc_ptr2len_check_len(p, size)
|
|
1680 char_u *p;
|
|
1681 int size;
|
|
1682 {
|
|
1683 int len;
|
|
1684 #ifdef FEAT_ARABIC
|
|
1685 int prevlen;
|
|
1686 #endif
|
|
1687
|
|
1688 if (*p == NUL)
|
|
1689 return 0;
|
|
1690 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) /* be quick for ASCII */
|
|
1691 return 1;
|
|
1692
|
|
1693 /* Skip over first UTF-8 char, stopping at a NUL byte. */
|
|
1694 len = utf_ptr2len_check_len(p, size);
|
|
1695
|
|
1696 /* Check for illegal byte and incomplete byte sequence. */
|
|
1697 if ((len == 1 && p[0] >= 0x80) || len > size)
|
|
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 while (len < size)
|
|
1708 {
|
|
1709 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len))
|
|
1710 break;
|
|
1711
|
|
1712 /* Skip over composing char */
|
|
1713 #ifdef FEAT_ARABIC
|
|
1714 prevlen = len;
|
|
1715 #endif
|
|
1716 len += utf_ptr2len_check_len(p + len, size - len);
|
|
1717 }
|
|
1718 return len;
|
|
1719 }
|
|
1720
|
|
1721 /*
|
|
1722 * Return the number of bytes the UTF-8 encoding of character "c" takes.
|
|
1723 * This does not include composing characters.
|
|
1724 */
|
|
1725 int
|
|
1726 utf_char2len(c)
|
|
1727 int c;
|
|
1728 {
|
|
1729 if (c < 0x80)
|
|
1730 return 1;
|
|
1731 if (c < 0x800)
|
|
1732 return 2;
|
|
1733 if (c < 0x10000)
|
|
1734 return 3;
|
|
1735 if (c < 0x200000)
|
|
1736 return 4;
|
|
1737 if (c < 0x4000000)
|
|
1738 return 5;
|
|
1739 return 6;
|
|
1740 }
|
|
1741
|
|
1742 /*
|
|
1743 * Convert Unicode character "c" to UTF-8 string in "buf[]".
|
|
1744 * Returns the number of bytes.
|
|
1745 * This does not include composing characters.
|
|
1746 */
|
|
1747 int
|
|
1748 utf_char2bytes(c, buf)
|
|
1749 int c;
|
|
1750 char_u *buf;
|
|
1751 {
|
|
1752 if (c < 0x80) /* 7 bits */
|
|
1753 {
|
|
1754 buf[0] = c;
|
|
1755 return 1;
|
|
1756 }
|
|
1757 if (c < 0x800) /* 11 bits */
|
|
1758 {
|
|
1759 buf[0] = 0xc0 + ((unsigned)c >> 6);
|
|
1760 buf[1] = 0x80 + (c & 0x3f);
|
|
1761 return 2;
|
|
1762 }
|
|
1763 if (c < 0x10000) /* 16 bits */
|
|
1764 {
|
|
1765 buf[0] = 0xe0 + ((unsigned)c >> 12);
|
|
1766 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f);
|
|
1767 buf[2] = 0x80 + (c & 0x3f);
|
|
1768 return 3;
|
|
1769 }
|
|
1770 if (c < 0x200000) /* 21 bits */
|
|
1771 {
|
|
1772 buf[0] = 0xf0 + ((unsigned)c >> 18);
|
|
1773 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f);
|
|
1774 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f);
|
|
1775 buf[3] = 0x80 + (c & 0x3f);
|
|
1776 return 4;
|
|
1777 }
|
|
1778 if (c < 0x4000000) /* 26 bits */
|
|
1779 {
|
|
1780 buf[0] = 0xf8 + ((unsigned)c >> 24);
|
|
1781 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f);
|
|
1782 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f);
|
|
1783 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f);
|
|
1784 buf[4] = 0x80 + (c & 0x3f);
|
|
1785 return 5;
|
|
1786 }
|
|
1787 /* 31 bits */
|
|
1788 buf[0] = 0xfc + ((unsigned)c >> 30);
|
|
1789 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f);
|
|
1790 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f);
|
|
1791 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f);
|
|
1792 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f);
|
|
1793 buf[5] = 0x80 + (c & 0x3f);
|
|
1794 return 6;
|
|
1795 }
|
|
1796
|
|
1797 /*
|
|
1798 * Return TRUE if "c" is a composing UTF-8 character. This means it will be
|
|
1799 * drawn on top of the preceding character.
|
|
1800 * Based on code from Markus Kuhn.
|
|
1801 */
|
|
1802 int
|
|
1803 utf_iscomposing(c)
|
|
1804 int c;
|
|
1805 {
|
|
1806 /* sorted list of non-overlapping intervals */
|
|
1807 static struct interval combining[] =
|
|
1808 {
|
|
1809 {0x0300, 0x034f}, {0x0360, 0x036f}, {0x0483, 0x0486}, {0x0488, 0x0489},
|
|
1810 {0x0591, 0x05a1}, {0x05a3, 0x05b9}, {0x05bb, 0x05bd}, {0x05bf, 0x05bf},
|
|
1811 {0x05c1, 0x05c2}, {0x05c4, 0x05c4}, {0x0610, 0x0615}, {0x064b, 0x0658},
|
|
1812 {0x0670, 0x0670}, {0x06d6, 0x06dc}, {0x06de, 0x06e4}, {0x06e7, 0x06e8},
|
|
1813 {0x06ea, 0x06ed}, {0x0711, 0x0711}, {0x0730, 0x074a}, {0x07a6, 0x07b0},
|
|
1814 {0x0901, 0x0903}, {0x093c, 0x093c}, {0x093e, 0x094d}, {0x0951, 0x0954},
|
|
1815 {0x0962, 0x0963}, {0x0981, 0x0983}, {0x09bc, 0x09bc}, {0x09be, 0x09c4},
|
|
1816 {0x09c7, 0x09c8}, {0x09cb, 0x09cd}, {0x09d7, 0x09d7}, {0x09e2, 0x09e3},
|
|
1817 {0x0a01, 0x0a03}, {0x0a3c, 0x0a3c}, {0x0a3e, 0x0a42}, {0x0a47, 0x0a48},
|
|
1818 {0x0a4b, 0x0a4d}, {0x0a70, 0x0a71}, {0x0a81, 0x0a83}, {0x0abc, 0x0abc},
|
|
1819 {0x0abe, 0x0ac5}, {0x0ac7, 0x0ac9}, {0x0acb, 0x0acd}, {0x0ae2, 0x0ae3},
|
|
1820 {0x0b01, 0x0b03}, {0x0b3c, 0x0b3c}, {0x0b3e, 0x0b43}, {0x0b47, 0x0b48},
|
|
1821 {0x0b4b, 0x0b4d}, {0x0b56, 0x0b57}, {0x0b82, 0x0b82}, {0x0bbe, 0x0bc2},
|
|
1822 {0x0bc6, 0x0bc8}, {0x0bca, 0x0bcd}, {0x0bd7, 0x0bd7}, {0x0c01, 0x0c03},
|
|
1823 {0x0c3e, 0x0c44}, {0x0c46, 0x0c48}, {0x0c4a, 0x0c4d}, {0x0c55, 0x0c56},
|
|
1824 {0x0c82, 0x0c83}, {0x0cbc, 0x0cbc}, {0x0cbe, 0x0cc4}, {0x0cc6, 0x0cc8},
|
|
1825 {0x0cca, 0x0ccd}, {0x0cd5, 0x0cd6}, {0x0d02, 0x0d03}, {0x0d3e, 0x0d43},
|
|
1826 {0x0d46, 0x0d48}, {0x0d4a, 0x0d4d}, {0x0d57, 0x0d57}, {0x0d82, 0x0d83},
|
|
1827 {0x0dca, 0x0dca}, {0x0dcf, 0x0dd4}, {0x0dd6, 0x0dd6}, {0x0dd8, 0x0ddf},
|
|
1828 {0x0df2, 0x0df3}, {0x0e31, 0x0e31}, {0x0e34, 0x0e3a}, {0x0e47, 0x0e4e},
|
|
1829 {0x0eb1, 0x0eb1}, {0x0eb4, 0x0eb9}, {0x0ebb, 0x0ebc}, {0x0ec8, 0x0ecd},
|
|
1830 {0x0f18, 0x0f19}, {0x0f35, 0x0f35}, {0x0f37, 0x0f37}, {0x0f39, 0x0f39},
|
|
1831 {0x0f3e, 0x0f3f}, {0x0f71, 0x0f84}, {0x0f86, 0x0f87}, {0x0f90, 0x0f97},
|
|
1832 {0x0f99, 0x0fbc}, {0x0fc6, 0x0fc6}, {0x102c, 0x1032}, {0x1036, 0x1039},
|
|
1833 {0x1056, 0x1059}, {0x1712, 0x1714}, {0x1732, 0x1734}, {0x1752, 0x1753},
|
|
1834 {0x1772, 0x1773}, {0x17b6, 0x17d3}, {0x17dd, 0x17dd}, {0x180b, 0x180d},
|
|
1835 {0x18a9, 0x18a9}, {0x1920, 0x192b}, {0x1930, 0x193b}, {0x20d0, 0x20ea},
|
|
1836 {0x302a, 0x302f}, {0x3099, 0x309a}, {0xfb1e, 0xfb1e}, {0xfe00, 0xfe0f},
|
|
1837 {0xfe20, 0xfe23},
|
|
1838 };
|
|
1839
|
|
1840 return intable(combining, sizeof(combining), c);
|
|
1841 }
|
|
1842
|
|
1843 /*
|
|
1844 * Return TRUE for characters that can be displayed in a normal way.
|
|
1845 * Only for characters of 0x100 and above!
|
|
1846 */
|
|
1847 int
|
|
1848 utf_printable(c)
|
|
1849 int c;
|
|
1850 {
|
|
1851 #ifdef USE_WCHAR_FUNCTIONS
|
|
1852 /*
|
|
1853 * Assume the iswprint() library function works better than our own stuff.
|
|
1854 */
|
|
1855 return iswprint(c);
|
|
1856 #else
|
|
1857 /* Sorted list of non-overlapping intervals.
|
|
1858 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */
|
|
1859 static struct interval nonprint[] =
|
|
1860 {
|
|
1861 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e},
|
|
1862 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb},
|
|
1863 {0xfffe, 0xffff}
|
|
1864 };
|
|
1865
|
|
1866 return !intable(nonprint, sizeof(nonprint), c);
|
|
1867 #endif
|
|
1868 }
|
|
1869
|
|
1870 /*
|
|
1871 * Get class of a Unicode character.
|
|
1872 * 0: white space
|
|
1873 * 1: punctuation
|
|
1874 * 2 or bigger: some class of word character.
|
|
1875 */
|
|
1876 int
|
|
1877 utf_class(c)
|
|
1878 int c;
|
|
1879 {
|
|
1880 /* sorted list of non-overlapping intervals */
|
|
1881 static struct clinterval
|
|
1882 {
|
|
1883 unsigned short first;
|
|
1884 unsigned short last;
|
|
1885 unsigned short class;
|
|
1886 } classes[] =
|
|
1887 {
|
|
1888 {0x037e, 0x037e, 1}, /* Greek question mark */
|
|
1889 {0x0387, 0x0387, 1}, /* Greek ano teleia */
|
|
1890 {0x055a, 0x055f, 1}, /* Armenian punctuation */
|
|
1891 {0x0589, 0x0589, 1}, /* Armenian full stop */
|
|
1892 {0x05be, 0x05be, 1},
|
|
1893 {0x05c0, 0x05c0, 1},
|
|
1894 {0x05c3, 0x05c3, 1},
|
|
1895 {0x05f3, 0x05f4, 1},
|
|
1896 {0x060c, 0x060c, 1},
|
|
1897 {0x061b, 0x061b, 1},
|
|
1898 {0x061f, 0x061f, 1},
|
|
1899 {0x066a, 0x066d, 1},
|
|
1900 {0x06d4, 0x06d4, 1},
|
|
1901 {0x0700, 0x070d, 1}, /* Syriac punctuation */
|
|
1902 {0x0964, 0x0965, 1},
|
|
1903 {0x0970, 0x0970, 1},
|
|
1904 {0x0df4, 0x0df4, 1},
|
|
1905 {0x0e4f, 0x0e4f, 1},
|
|
1906 {0x0e5a, 0x0e5b, 1},
|
|
1907 {0x0f04, 0x0f12, 1},
|
|
1908 {0x0f3a, 0x0f3d, 1},
|
|
1909 {0x0f85, 0x0f85, 1},
|
|
1910 {0x104a, 0x104f, 1}, /* Myanmar punctuation */
|
|
1911 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */
|
|
1912 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */
|
|
1913 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */
|
|
1914 {0x1680, 0x1680, 0},
|
|
1915 {0x169b, 0x169c, 1},
|
|
1916 {0x16eb, 0x16ed, 1},
|
|
1917 {0x1735, 0x1736, 1},
|
|
1918 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */
|
|
1919 {0x1800, 0x180a, 1}, /* Mongolian punctuation */
|
|
1920 {0x2000, 0x200b, 0}, /* spaces */
|
|
1921 {0x200c, 0x2027, 1}, /* punctuation and symbols */
|
|
1922 {0x2028, 0x2029, 0},
|
|
1923 {0x202a, 0x202e, 1}, /* punctuation and symbols */
|
|
1924 {0x202f, 0x202f, 0},
|
|
1925 {0x2030, 0x205e, 1}, /* punctuation and symbols */
|
|
1926 {0x205f, 0x205f, 0},
|
|
1927 {0x2060, 0x27ff, 1}, /* punctuation and symbols */
|
|
1928 {0x2070, 0x207f, 0x2070}, /* superscript */
|
|
1929 {0x2080, 0x208f, 0x2080}, /* subscript */
|
|
1930 {0x2983, 0x2998, 1},
|
|
1931 {0x29d8, 0x29db, 1},
|
|
1932 {0x29fc, 0x29fd, 1},
|
|
1933 {0x3000, 0x3000, 0}, /* ideographic space */
|
|
1934 {0x3001, 0x3020, 1}, /* ideographic punctuation */
|
|
1935 {0x3030, 0x3030, 1},
|
|
1936 {0x303d, 0x303d, 1},
|
|
1937 {0x3040, 0x309f, 0x3040}, /* Hiragana */
|
|
1938 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */
|
|
1939 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */
|
|
1940 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */
|
|
1941 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */
|
|
1942 {0xfd3e, 0xfd3f, 1},
|
|
1943 {0xfe30, 0xfe6b, 1}, /* punctuation forms */
|
|
1944 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */
|
|
1945 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */
|
|
1946 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */
|
|
1947 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */
|
|
1948 };
|
|
1949 int bot = 0;
|
|
1950 int top = sizeof(classes) / sizeof(struct clinterval) - 1;
|
|
1951 int mid;
|
|
1952
|
|
1953 /* First quick check for Latin1 characters, use 'iskeyword'. */
|
|
1954 if (c < 0x100)
|
|
1955 {
|
335
|
1956 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0)
|
7
|
1957 return 0; /* blank */
|
|
1958 if (vim_iswordc(c))
|
|
1959 return 2; /* word character */
|
|
1960 return 1; /* punctuation */
|
|
1961 }
|
|
1962
|
|
1963 /* binary search in table */
|
|
1964 while (top >= bot)
|
|
1965 {
|
|
1966 mid = (bot + top) / 2;
|
|
1967 if (classes[mid].last < c)
|
|
1968 bot = mid + 1;
|
|
1969 else if (classes[mid].first > c)
|
|
1970 top = mid - 1;
|
|
1971 else
|
|
1972 return (int)classes[mid].class;
|
|
1973 }
|
|
1974
|
|
1975 /* most other characters are "word" characters */
|
|
1976 return 2;
|
|
1977 }
|
|
1978
|
|
1979 /*
|
|
1980 * Code for Unicode case-dependent operations. Based on notes in
|
|
1981 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt
|
|
1982 * This code uses simple case folding, not full case folding.
|
|
1983 */
|
|
1984
|
|
1985 /*
|
|
1986 * The following table is built by foldExtract.pl < CaseFolding.txt .
|
|
1987 * It must be in numeric order, because we use binary search on it.
|
|
1988 * An entry such as {0x41,0x5a,1,32} means that UCS-4 characters in the range
|
|
1989 * from 0x41 to 0x5a inclusive, stepping by 1, are folded by adding 32.
|
|
1990 */
|
|
1991
|
|
1992 typedef struct
|
|
1993 {
|
|
1994 int rangeStart;
|
|
1995 int rangeEnd;
|
|
1996 int step;
|
|
1997 int offset;
|
|
1998 } convertStruct;
|
|
1999
|
297
|
2000 static convertStruct foldCase[] =
|
7
|
2001 {
|
|
2002 {0x41,0x5a,1,32}, {0xc0,0xd6,1,32}, {0xd8,0xde,1,32},
|
|
2003 {0x100,0x12e,2,1}, {0x130,0x130,-1,-199}, {0x132,0x136,2,1},
|
|
2004 {0x139,0x147,2,1}, {0x14a,0x176,2,1}, {0x178,0x178,-1,-121},
|
|
2005 {0x179,0x17d,2,1}, {0x181,0x181,-1,210}, {0x182,0x184,2,1},
|
|
2006 {0x186,0x186,-1,206}, {0x187,0x187,-1,1}, {0x189,0x18a,1,205},
|
|
2007 {0x18b,0x18b,-1,1}, {0x18e,0x18e,-1,79}, {0x18f,0x18f,-1,202},
|
|
2008 {0x190,0x190,-1,203}, {0x191,0x191,-1,1}, {0x193,0x193,-1,205},
|
|
2009 {0x194,0x194,-1,207}, {0x196,0x196,-1,211}, {0x197,0x197,-1,209},
|
|
2010 {0x198,0x198,-1,1}, {0x19c,0x19c,-1,211}, {0x19d,0x19d,-1,213},
|
|
2011 {0x19f,0x19f,-1,214}, {0x1a0,0x1a4,2,1}, {0x1a6,0x1a6,-1,218},
|
|
2012 {0x1a7,0x1a7,-1,1}, {0x1a9,0x1a9,-1,218}, {0x1ac,0x1ac,-1,1},
|
|
2013 {0x1ae,0x1ae,-1,218}, {0x1af,0x1af,-1,1}, {0x1b1,0x1b2,1,217},
|
|
2014 {0x1b3,0x1b5,2,1}, {0x1b7,0x1b7,-1,219}, {0x1b8,0x1bc,4,1},
|
|
2015 {0x1c4,0x1c4,-1,2}, {0x1c5,0x1c5,-1,1}, {0x1c7,0x1c7,-1,2},
|
|
2016 {0x1c8,0x1c8,-1,1}, {0x1ca,0x1ca,-1,2}, {0x1cb,0x1db,2,1},
|
|
2017 {0x1de,0x1ee,2,1}, {0x1f1,0x1f1,-1,2}, {0x1f2,0x1f4,2,1},
|
|
2018 {0x1f6,0x1f6,-1,-97}, {0x1f7,0x1f7,-1,-56}, {0x1f8,0x21e,2,1},
|
|
2019 {0x220,0x220,-1,-130}, {0x222,0x232,2,1}, {0x386,0x386,-1,38},
|
|
2020 {0x388,0x38a,1,37}, {0x38c,0x38c,-1,64}, {0x38e,0x38f,1,63},
|
|
2021 {0x391,0x3a1,1,32}, {0x3a3,0x3ab,1,32}, {0x3d8,0x3ee,2,1},
|
|
2022 {0x3f4,0x3f4,-1,-60}, {0x3f7,0x3f7,-1,1}, {0x3f9,0x3f9,-1,-7},
|
|
2023 {0x3fa,0x3fa,-1,1}, {0x400,0x40f,1,80}, {0x410,0x42f,1,32},
|
|
2024 {0x460,0x480,2,1}, {0x48a,0x4be,2,1}, {0x4c1,0x4cd,2,1},
|
|
2025 {0x4d0,0x4f4,2,1}, {0x4f8,0x500,8,1}, {0x502,0x50e,2,1},
|
|
2026 {0x531,0x556,1,48}, {0x1e00,0x1e94,2,1}, {0x1ea0,0x1ef8,2,1},
|
|
2027 {0x1f08,0x1f0f,1,-8}, {0x1f18,0x1f1d,1,-8}, {0x1f28,0x1f2f,1,-8},
|
|
2028 {0x1f38,0x1f3f,1,-8}, {0x1f48,0x1f4d,1,-8}, {0x1f59,0x1f5f,2,-8},
|
|
2029 {0x1f68,0x1f6f,1,-8}, {0x1f88,0x1f8f,1,-8}, {0x1f98,0x1f9f,1,-8},
|
|
2030 {0x1fa8,0x1faf,1,-8}, {0x1fb8,0x1fb9,1,-8}, {0x1fba,0x1fbb,1,-74},
|
|
2031 {0x1fbc,0x1fbc,-1,-9}, {0x1fc8,0x1fcb,1,-86}, {0x1fcc,0x1fcc,-1,-9},
|
|
2032 {0x1fd8,0x1fd9,1,-8}, {0x1fda,0x1fdb,1,-100}, {0x1fe8,0x1fe9,1,-8},
|
|
2033 {0x1fea,0x1feb,1,-112}, {0x1fec,0x1fec,-1,-7}, {0x1ff8,0x1ff9,1,-128},
|
|
2034 {0x1ffa,0x1ffb,1,-126}, {0x1ffc,0x1ffc,-1,-9}, {0x2126,0x2126,-1,-7517},
|
|
2035 {0x212a,0x212a,-1,-8383}, {0x212b,0x212b,-1,-8262},
|
|
2036 {0x2160,0x216f,1,16}, {0x24b6,0x24cf,1,26}, {0xff21,0xff3a,1,32},
|
|
2037 {0x10400,0x10427,1,40}
|
|
2038 };
|
|
2039
|
|
2040 static int utf_convert(int a, convertStruct table[], int tableSize);
|
|
2041
|
|
2042 /*
|
|
2043 * Generic conversion function for case operations.
|
|
2044 * Return the converted equivalent of "a", which is a UCS-4 character. Use
|
|
2045 * the given conversion "table". Uses binary search on "table".
|
|
2046 */
|
|
2047 static int
|
|
2048 utf_convert(a, table, tableSize)
|
|
2049 int a;
|
|
2050 convertStruct table[];
|
|
2051 int tableSize;
|
|
2052 {
|
|
2053 int start, mid, end; /* indices into table */
|
|
2054
|
|
2055 start = 0;
|
|
2056 end = tableSize / sizeof(convertStruct);
|
|
2057 while (start < end)
|
|
2058 {
|
|
2059 /* need to search further */
|
|
2060 mid = (end + start) /2;
|
|
2061 if (table[mid].rangeEnd < a)
|
|
2062 start = mid + 1;
|
|
2063 else
|
|
2064 end = mid;
|
|
2065 }
|
|
2066 if (table[start].rangeStart <= a && a <= table[start].rangeEnd
|
|
2067 && (a - table[start].rangeStart) % table[start].step == 0)
|
|
2068 return (a + table[start].offset);
|
|
2069 else
|
|
2070 return a;
|
|
2071 }
|
|
2072
|
|
2073 /*
|
|
2074 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses
|
|
2075 * simple case folding.
|
|
2076 */
|
|
2077 int
|
|
2078 utf_fold(a)
|
|
2079 int a;
|
|
2080 {
|
|
2081 return utf_convert(a, foldCase, sizeof(foldCase));
|
|
2082 }
|
|
2083
|
|
2084 /*
|
|
2085 * The following tables are built by upperLowerExtract.pl < UnicodeData.txt .
|
|
2086 * They must be in numeric order, because we use binary search on them.
|
|
2087 * An entry such as {0x41,0x5a,1,32} means that UCS-4 characters in the range
|
|
2088 * from 0x41 to 0x5a inclusive, stepping by 1, are switched to lower (for
|
|
2089 * example) by adding 32.
|
|
2090 */
|
297
|
2091 static convertStruct toLower[] =
|
7
|
2092 {
|
|
2093 {0x41,0x5a,1,32}, {0xc0,0xd6,1,32}, {0xd8,0xde,1,32},
|
|
2094 {0x100,0x12e,2,1}, {0x130,0x130,-1,-199}, {0x132,0x136,2,1},
|
|
2095 {0x139,0x147,2,1}, {0x14a,0x176,2,1}, {0x178,0x178,-1,-121},
|
|
2096 {0x179,0x17d,2,1}, {0x181,0x181,-1,210}, {0x182,0x184,2,1},
|
|
2097 {0x186,0x186,-1,206}, {0x187,0x187,-1,1}, {0x189,0x18a,1,205},
|
|
2098 {0x18b,0x18b,-1,1}, {0x18e,0x18e,-1,79}, {0x18f,0x18f,-1,202},
|
|
2099 {0x190,0x190,-1,203}, {0x191,0x191,-1,1}, {0x193,0x193,-1,205},
|
|
2100 {0x194,0x194,-1,207}, {0x196,0x196,-1,211}, {0x197,0x197,-1,209},
|
|
2101 {0x198,0x198,-1,1}, {0x19c,0x19c,-1,211}, {0x19d,0x19d,-1,213},
|
|
2102 {0x19f,0x19f,-1,214}, {0x1a0,0x1a4,2,1}, {0x1a6,0x1a6,-1,218},
|
|
2103 {0x1a7,0x1a7,-1,1}, {0x1a9,0x1a9,-1,218}, {0x1ac,0x1ac,-1,1},
|
|
2104 {0x1ae,0x1ae,-1,218}, {0x1af,0x1af,-1,1}, {0x1b1,0x1b2,1,217},
|
|
2105 {0x1b3,0x1b5,2,1}, {0x1b7,0x1b7,-1,219}, {0x1b8,0x1bc,4,1},
|
|
2106 {0x1c4,0x1ca,3,2}, {0x1cd,0x1db,2,1}, {0x1de,0x1ee,2,1},
|
|
2107 {0x1f1,0x1f1,-1,2}, {0x1f4,0x1f4,-1,1}, {0x1f6,0x1f6,-1,-97},
|
|
2108 {0x1f7,0x1f7,-1,-56}, {0x1f8,0x21e,2,1}, {0x220,0x220,-1,-130},
|
|
2109 {0x222,0x232,2,1}, {0x386,0x386,-1,38}, {0x388,0x38a,1,37},
|
|
2110 {0x38c,0x38c,-1,64}, {0x38e,0x38f,1,63}, {0x391,0x3a1,1,32},
|
|
2111 {0x3a3,0x3ab,1,32}, {0x3d8,0x3ee,2,1}, {0x3f4,0x3f4,-1,-60},
|
|
2112 {0x3f7,0x3f7,-1,1}, {0x3f9,0x3f9,-1,-7}, {0x3fa,0x3fa,-1,1},
|
|
2113 {0x400,0x40f,1,80}, {0x410,0x42f,1,32}, {0x460,0x480,2,1},
|
|
2114 {0x48a,0x4be,2,1}, {0x4c1,0x4cd,2,1}, {0x4d0,0x4f4,2,1},
|
|
2115 {0x4f8,0x500,8,1}, {0x502,0x50e,2,1}, {0x531,0x556,1,48},
|
|
2116 {0x1e00,0x1e94,2,1}, {0x1ea0,0x1ef8,2,1}, {0x1f08,0x1f0f,1,-8},
|
|
2117 {0x1f18,0x1f1d,1,-8}, {0x1f28,0x1f2f,1,-8}, {0x1f38,0x1f3f,1,-8},
|
|
2118 {0x1f48,0x1f4d,1,-8}, {0x1f59,0x1f5f,2,-8}, {0x1f68,0x1f6f,1,-8},
|
|
2119 {0x1fb8,0x1fb9,1,-8}, {0x1fba,0x1fbb,1,-74}, {0x1fc8,0x1fcb,1,-86},
|
|
2120 {0x1fd8,0x1fd9,1,-8}, {0x1fda,0x1fdb,1,-100}, {0x1fe8,0x1fe9,1,-8},
|
|
2121 {0x1fea,0x1feb,1,-112}, {0x1fec,0x1fec,-1,-7}, {0x1ff8,0x1ff9,1,-128},
|
|
2122 {0x1ffa,0x1ffb,1,-126}, {0x2126,0x2126,-1,-7517}, {0x212a,0x212a,-1,-8383},
|
|
2123 {0x212b,0x212b,-1,-8262}, {0xff21,0xff3a,1,32}, {0x10400,0x10427,1,40}
|
|
2124 };
|
|
2125
|
297
|
2126 static convertStruct toUpper[] =
|
7
|
2127 {
|
|
2128 {0x61,0x7a,1,-32}, {0xb5,0xb5,-1,743}, {0xe0,0xf6,1,-32},
|
|
2129 {0xf8,0xfe,1,-32}, {0xff,0xff,-1,121}, {0x101,0x12f,2,-1},
|
|
2130 {0x131,0x131,-1,-232}, {0x133,0x137,2,-1}, {0x13a,0x148,2,-1},
|
|
2131 {0x14b,0x177,2,-1}, {0x17a,0x17e,2,-1}, {0x17f,0x17f,-1,-300},
|
|
2132 {0x183,0x185,2,-1}, {0x188,0x18c,4,-1}, {0x192,0x192,-1,-1},
|
|
2133 {0x195,0x195,-1,97}, {0x199,0x199,-1,-1}, {0x19e,0x19e,-1,130},
|
|
2134 {0x1a1,0x1a5,2,-1}, {0x1a8,0x1ad,5,-1}, {0x1b0,0x1b4,4,-1},
|
|
2135 {0x1b6,0x1b9,3,-1}, {0x1bd,0x1bd,-1,-1}, {0x1bf,0x1bf,-1,56},
|
|
2136 {0x1c5,0x1c6,1,-1}, {0x1c8,0x1c9,1,-1}, {0x1cb,0x1cc,1,-1},
|
|
2137 {0x1ce,0x1dc,2,-1}, {0x1dd,0x1dd,-1,-79}, {0x1df,0x1ef,2,-1},
|
|
2138 {0x1f2,0x1f3,1,-1}, {0x1f5,0x1f9,4,-1}, {0x1fb,0x21f,2,-1},
|
|
2139 {0x223,0x233,2,-1}, {0x253,0x253,-1,-210}, {0x254,0x254,-1,-206},
|
|
2140 {0x256,0x257,1,-205}, {0x259,0x259,-1,-202}, {0x25b,0x25b,-1,-203},
|
|
2141 {0x260,0x260,-1,-205}, {0x263,0x263,-1,-207}, {0x268,0x268,-1,-209},
|
|
2142 {0x269,0x26f,6,-211}, {0x272,0x272,-1,-213}, {0x275,0x275,-1,-214},
|
|
2143 {0x280,0x283,3,-218}, {0x288,0x288,-1,-218}, {0x28a,0x28b,1,-217},
|
|
2144 {0x292,0x292,-1,-219}, {0x3ac,0x3ac,-1,-38}, {0x3ad,0x3af,1,-37},
|
|
2145 {0x3b1,0x3c1,1,-32}, {0x3c2,0x3c2,-1,-31}, {0x3c3,0x3cb,1,-32},
|
|
2146 {0x3cc,0x3cc,-1,-64}, {0x3cd,0x3ce,1,-63}, {0x3d0,0x3d0,-1,-62},
|
|
2147 {0x3d1,0x3d1,-1,-57}, {0x3d5,0x3d5,-1,-47}, {0x3d6,0x3d6,-1,-54},
|
|
2148 {0x3d9,0x3ef,2,-1}, {0x3f0,0x3f0,-1,-86}, {0x3f1,0x3f1,-1,-80},
|
|
2149 {0x3f2,0x3f2,-1,7}, {0x3f5,0x3f5,-1,-96}, {0x3f8,0x3fb,3,-1},
|
|
2150 {0x430,0x44f,1,-32}, {0x450,0x45f,1,-80}, {0x461,0x481,2,-1},
|
|
2151 {0x48b,0x4bf,2,-1}, {0x4c2,0x4ce,2,-1}, {0x4d1,0x4f5,2,-1},
|
|
2152 {0x4f9,0x501,8,-1}, {0x503,0x50f,2,-1}, {0x561,0x586,1,-48},
|
|
2153 {0x1e01,0x1e95,2,-1}, {0x1e9b,0x1e9b,-1,-59}, {0x1ea1,0x1ef9,2,-1},
|
|
2154 {0x1f00,0x1f07,1,8}, {0x1f10,0x1f15,1,8}, {0x1f20,0x1f27,1,8},
|
|
2155 {0x1f30,0x1f37,1,8}, {0x1f40,0x1f45,1,8}, {0x1f51,0x1f57,2,8},
|
|
2156 {0x1f60,0x1f67,1,8}, {0x1f70,0x1f71,1,74}, {0x1f72,0x1f75,1,86},
|
|
2157 {0x1f76,0x1f77,1,100}, {0x1f78,0x1f79,1,128}, {0x1f7a,0x1f7b,1,112},
|
|
2158 {0x1f7c,0x1f7d,1,126}, {0x1f80,0x1f87,1,8}, {0x1f90,0x1f97,1,8},
|
|
2159 {0x1fa0,0x1fa7,1,8}, {0x1fb0,0x1fb1,1,8}, {0x1fb3,0x1fb3,-1,9},
|
|
2160 {0x1fbe,0x1fbe,-1,-7205}, {0x1fc3,0x1fc3,-1,9}, {0x1fd0,0x1fd1,1,8},
|
|
2161 {0x1fe0,0x1fe1,1,8}, {0x1fe5,0x1fe5,-1,7}, {0x1ff3,0x1ff3,-1,9},
|
|
2162 {0xff41,0xff5a,1,-32}, {0x10428,0x1044f,1,-40}
|
|
2163 };
|
|
2164
|
|
2165 /*
|
|
2166 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use
|
|
2167 * simple case folding.
|
|
2168 */
|
|
2169 int
|
|
2170 utf_toupper(a)
|
|
2171 int a;
|
|
2172 {
|
|
2173 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */
|
|
2174 if (a < 128 && (cmp_flags & CMP_KEEPASCII))
|
|
2175 return TOUPPER_ASC(a);
|
|
2176
|
|
2177 #if defined(HAVE_TOWUPPER) && defined(__STDC__ISO_10646__)
|
|
2178 /* If towupper() is availble and handles Unicode, use it. */
|
|
2179 if (!(cmp_flags & CMP_INTERNAL))
|
|
2180 return towupper(a);
|
|
2181 #endif
|
|
2182
|
|
2183 /* For characters below 128 use locale sensitive toupper(). */
|
|
2184 if (a < 128)
|
|
2185 return TOUPPER_LOC(a);
|
|
2186
|
|
2187 /* For any other characters use the above mapping table. */
|
|
2188 return utf_convert(a, toUpper, sizeof(toUpper));
|
|
2189 }
|
|
2190
|
|
2191 int
|
|
2192 utf_islower(a)
|
|
2193 int a;
|
|
2194 {
|
|
2195 return (utf_toupper(a) != a);
|
|
2196 }
|
|
2197
|
|
2198 /*
|
|
2199 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use
|
|
2200 * simple case folding.
|
|
2201 */
|
|
2202 int
|
|
2203 utf_tolower(a)
|
|
2204 int a;
|
|
2205 {
|
|
2206 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */
|
|
2207 if (a < 128 && (cmp_flags & CMP_KEEPASCII))
|
|
2208 return TOLOWER_ASC(a);
|
|
2209
|
|
2210 #if defined(HAVE_TOWLOWER) && defined(__STDC__ISO_10646__)
|
257
|
2211 /* If towlower() is available and handles Unicode, use it. */
|
7
|
2212 if (!(cmp_flags & CMP_INTERNAL))
|
|
2213 return towlower(a);
|
|
2214 #endif
|
|
2215
|
|
2216 /* For characters below 128 use locale sensitive tolower(). */
|
|
2217 if (a < 128)
|
|
2218 return TOLOWER_LOC(a);
|
|
2219
|
|
2220 /* For any other characters use the above mapping table. */
|
|
2221 return utf_convert(a, toLower, sizeof(toLower));
|
|
2222 }
|
|
2223
|
|
2224 int
|
|
2225 utf_isupper(a)
|
|
2226 int a;
|
|
2227 {
|
|
2228 return (utf_tolower(a) != a);
|
|
2229 }
|
|
2230
|
|
2231 /*
|
|
2232 * Version of strnicmp() that handles multi-byte characters.
|
|
2233 * Needed for Big5, Sjift-JIS and UTF-8 encoding. Other DBCS encodings can
|
|
2234 * probably use strnicmp(), because there are no ASCII characters in the
|
|
2235 * second byte.
|
|
2236 * Returns zero if s1 and s2 are equal (ignoring case), the difference between
|
|
2237 * two characters otherwise.
|
|
2238 */
|
|
2239 int
|
303
|
2240 mb_strnicmp(s1, s2, nn)
|
7
|
2241 char_u *s1, *s2;
|
303
|
2242 size_t nn;
|
7
|
2243 {
|
|
2244 int i, j, l;
|
|
2245 int cdiff;
|
308
|
2246 int incomplete = FALSE;
|
303
|
2247 int n = nn;
|
7
|
2248
|
|
2249 for (i = 0; i < n; i += l)
|
|
2250 {
|
|
2251 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */
|
|
2252 return 0;
|
|
2253 if (enc_utf8)
|
|
2254 {
|
|
2255 l = utf_byte2len(s1[i]);
|
|
2256 if (l > n - i)
|
308
|
2257 {
|
7
|
2258 l = n - i; /* incomplete character */
|
308
|
2259 incomplete = TRUE;
|
|
2260 }
|
7
|
2261 /* Check directly first, it's faster. */
|
|
2262 for (j = 0; j < l; ++j)
|
|
2263 if (s1[i + j] != s2[i + j])
|
|
2264 break;
|
|
2265 if (j < l)
|
|
2266 {
|
|
2267 /* If one of the two characters is incomplete return -1. */
|
308
|
2268 if (incomplete || i + utf_byte2len(s2[i]) > n)
|
7
|
2269 return -1;
|
|
2270 cdiff = utf_fold(utf_ptr2char(s1 + i))
|
|
2271 - utf_fold(utf_ptr2char(s2 + i));
|
|
2272 if (cdiff != 0)
|
|
2273 return cdiff;
|
|
2274 }
|
|
2275 }
|
|
2276 else
|
|
2277 {
|
|
2278 l = (*mb_ptr2len_check)(s1 + i);
|
|
2279 if (l <= 1)
|
|
2280 {
|
|
2281 /* Single byte: first check normally, then with ignore case. */
|
|
2282 if (s1[i] != s2[i])
|
|
2283 {
|
|
2284 cdiff = TOLOWER_LOC(s1[i]) - TOLOWER_LOC(s2[i]);
|
|
2285 if (cdiff != 0)
|
|
2286 return cdiff;
|
|
2287 }
|
|
2288 }
|
|
2289 else
|
|
2290 {
|
|
2291 /* For non-Unicode multi-byte don't ignore case. */
|
|
2292 if (l > n - i)
|
|
2293 l = n - i;
|
|
2294 cdiff = STRNCMP(s1 + i, s2 + i, l);
|
|
2295 if (cdiff != 0)
|
|
2296 return cdiff;
|
|
2297 }
|
|
2298 }
|
|
2299 }
|
|
2300 return 0;
|
|
2301 }
|
|
2302
|
|
2303 /*
|
|
2304 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what
|
|
2305 * 'encoding' has been set to.
|
|
2306 */
|
|
2307 void
|
|
2308 show_utf8()
|
|
2309 {
|
|
2310 int len;
|
273
|
2311 int rlen = 0;
|
7
|
2312 char_u *line;
|
|
2313 int clen;
|
|
2314 int i;
|
|
2315
|
|
2316 /* Get the byte length of the char under the cursor, including composing
|
|
2317 * characters. */
|
|
2318 line = ml_get_cursor();
|
|
2319 len = utfc_ptr2len_check(line);
|
|
2320 if (len == 0)
|
|
2321 {
|
|
2322 MSG("NUL");
|
|
2323 return;
|
|
2324 }
|
|
2325
|
|
2326 clen = 0;
|
|
2327 for (i = 0; i < len; ++i)
|
|
2328 {
|
|
2329 if (clen == 0)
|
|
2330 {
|
|
2331 /* start of (composing) character, get its length */
|
|
2332 if (i > 0)
|
273
|
2333 {
|
|
2334 STRCPY(IObuff + rlen, "+ ");
|
|
2335 rlen += 2;
|
|
2336 }
|
7
|
2337 clen = utf_ptr2len_check(line + i);
|
|
2338 }
|
273
|
2339 sprintf((char *)IObuff + rlen, "%02x ", line[i]);
|
7
|
2340 --clen;
|
273
|
2341 rlen += STRLEN(IObuff + rlen);
|
|
2342 if (rlen > IOSIZE - 20)
|
|
2343 break;
|
7
|
2344 }
|
|
2345
|
|
2346 msg(IObuff);
|
|
2347 }
|
|
2348
|
|
2349 /*
|
|
2350 * mb_head_off() function pointer.
|
|
2351 * Return offset from "p" to the first byte of the character it points into.
|
|
2352 * Returns 0 when already at the first byte of a character.
|
|
2353 */
|
|
2354 /*ARGSUSED*/
|
|
2355 int
|
|
2356 latin_head_off(base, p)
|
|
2357 char_u *base;
|
|
2358 char_u *p;
|
|
2359 {
|
|
2360 return 0;
|
|
2361 }
|
|
2362
|
|
2363 int
|
|
2364 dbcs_head_off(base, p)
|
|
2365 char_u *base;
|
|
2366 char_u *p;
|
|
2367 {
|
|
2368 char_u *q;
|
|
2369
|
|
2370 /* It can't be a trailing byte when not using DBCS, at the start of the
|
|
2371 * string or the previous byte can't start a double-byte. */
|
|
2372 if (p <= base || MB_BYTE2LEN(p[-1]) == 1)
|
|
2373 return 0;
|
|
2374
|
|
2375 /* This is slow: need to start at the base and go forward until the
|
|
2376 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */
|
|
2377 q = base;
|
|
2378 while (q < p)
|
|
2379 q += dbcs_ptr2len_check(q);
|
|
2380 return (q == p) ? 0 : 1;
|
|
2381 }
|
|
2382
|
|
2383 #if defined(FEAT_CLIPBOARD) || defined(FEAT_GUI) || defined(FEAT_RIGHTLEFT) \
|
|
2384 || defined(PROTO)
|
|
2385 /*
|
|
2386 * Special version of dbcs_head_off() that works for ScreenLines[], where
|
|
2387 * single-width DBCS_JPNU characters are stored separately.
|
|
2388 */
|
|
2389 int
|
|
2390 dbcs_screen_head_off(base, p)
|
|
2391 char_u *base;
|
|
2392 char_u *p;
|
|
2393 {
|
|
2394 char_u *q;
|
|
2395
|
|
2396 /* It can't be a trailing byte when not using DBCS, at the start of the
|
|
2397 * string or the previous byte can't start a double-byte.
|
|
2398 * For euc-jp an 0x8e byte in the previous cell always means we have a
|
|
2399 * lead byte in the current cell. */
|
|
2400 if (p <= base
|
|
2401 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e)
|
|
2402 || MB_BYTE2LEN(p[-1]) == 1)
|
|
2403 return 0;
|
|
2404
|
|
2405 /* This is slow: need to start at the base and go forward until the
|
|
2406 * byte we are looking for. Return 1 when we went past it, 0 otherwise.
|
|
2407 * For DBCS_JPNU look out for 0x8e, which means the second byte is not
|
|
2408 * stored as the next byte. */
|
|
2409 q = base;
|
|
2410 while (q < p)
|
|
2411 {
|
|
2412 if (enc_dbcs == DBCS_JPNU && *q == 0x8e)
|
|
2413 ++q;
|
|
2414 else
|
|
2415 q += dbcs_ptr2len_check(q);
|
|
2416 }
|
|
2417 return (q == p) ? 0 : 1;
|
|
2418 }
|
|
2419 #endif
|
|
2420
|
|
2421 int
|
|
2422 utf_head_off(base, p)
|
|
2423 char_u *base;
|
|
2424 char_u *p;
|
|
2425 {
|
|
2426 char_u *q;
|
|
2427 char_u *s;
|
|
2428 int c;
|
|
2429 #ifdef FEAT_ARABIC
|
|
2430 char_u *j;
|
|
2431 #endif
|
|
2432
|
|
2433 if (*p < 0x80) /* be quick for ASCII */
|
|
2434 return 0;
|
|
2435
|
|
2436 /* Skip backwards over trailing bytes: 10xx.xxxx
|
|
2437 * Skip backwards again if on a composing char. */
|
|
2438 for (q = p; ; --q)
|
|
2439 {
|
|
2440 /* Move s to the last byte of this char. */
|
|
2441 for (s = q; (s[1] & 0xc0) == 0x80; ++s)
|
|
2442 ;
|
|
2443 /* Move q to the first byte of this char. */
|
|
2444 while (q > base && (*q & 0xc0) == 0x80)
|
|
2445 --q;
|
|
2446 /* Check for illegal sequence. Do allow an illegal byte after where we
|
|
2447 * started. */
|
|
2448 if (utf8len_tab[*q] != (int)(s - q + 1)
|
|
2449 && utf8len_tab[*q] != (int)(p - q + 1))
|
|
2450 return 0;
|
|
2451
|
|
2452 if (q <= base)
|
|
2453 break;
|
|
2454
|
|
2455 c = utf_ptr2char(q);
|
|
2456 if (utf_iscomposing(c))
|
|
2457 continue;
|
|
2458
|
|
2459 #ifdef FEAT_ARABIC
|
|
2460 if (arabic_maycombine(c))
|
|
2461 {
|
|
2462 /* Advance to get a sneak-peak at the next char */
|
|
2463 j = q;
|
|
2464 --j;
|
|
2465 /* Move j to the first byte of this char. */
|
|
2466 while (j > base && (*j & 0xc0) == 0x80)
|
|
2467 --j;
|
|
2468 if (arabic_combine(utf_ptr2char(j), c))
|
|
2469 continue;
|
|
2470 }
|
|
2471 #endif
|
|
2472 break;
|
|
2473 }
|
|
2474
|
|
2475 return (int)(p - q);
|
|
2476 }
|
|
2477
|
140
|
2478 #if defined(FEAT_EVAL) || defined(PROTO)
|
7
|
2479 /*
|
99
|
2480 * Copy a character from "*fp" to "*tp" and advance the pointers.
|
|
2481 */
|
|
2482 void
|
|
2483 mb_copy_char(fp, tp)
|
|
2484 char_u **fp;
|
|
2485 char_u **tp;
|
|
2486 {
|
|
2487 int l = (*mb_ptr2len_check)(*fp);
|
|
2488
|
|
2489 mch_memmove(*tp, *fp, (size_t)l);
|
|
2490 *tp += l;
|
|
2491 *fp += l;
|
|
2492 }
|
140
|
2493 #endif
|
99
|
2494
|
|
2495 /*
|
7
|
2496 * Return the offset from "p" to the first byte of a character. When "p" is
|
|
2497 * at the start of a character 0 is returned, otherwise the offset to the next
|
|
2498 * character. Can start anywhere in a stream of bytes.
|
|
2499 */
|
|
2500 int
|
|
2501 mb_off_next(base, p)
|
|
2502 char_u *base;
|
|
2503 char_u *p;
|
|
2504 {
|
|
2505 int i;
|
|
2506 int j;
|
|
2507
|
|
2508 if (enc_utf8)
|
|
2509 {
|
|
2510 if (*p < 0x80) /* be quick for ASCII */
|
|
2511 return 0;
|
|
2512
|
|
2513 /* Find the next character that isn't 10xx.xxxx */
|
|
2514 for (i = 0; (p[i] & 0xc0) == 0x80; ++i)
|
|
2515 ;
|
|
2516 if (i > 0)
|
|
2517 {
|
|
2518 /* Check for illegal sequence. */
|
|
2519 for (j = 0; p - j > base; ++j)
|
|
2520 if ((p[-j] & 0xc0) != 0x80)
|
|
2521 break;
|
|
2522 if (utf8len_tab[p[-j]] != i + j)
|
|
2523 return 0;
|
|
2524 }
|
|
2525 return i;
|
|
2526 }
|
|
2527
|
|
2528 /* Only need to check if we're on a trail byte, it doesn't matter if we
|
|
2529 * want the offset to the next or current character. */
|
|
2530 return (*mb_head_off)(base, p);
|
|
2531 }
|
|
2532
|
|
2533 /*
|
|
2534 * Return the offset from "p" to the last byte of the character it points
|
|
2535 * into. Can start anywhere in a stream of bytes.
|
|
2536 */
|
|
2537 int
|
|
2538 mb_tail_off(base, p)
|
|
2539 char_u *base;
|
|
2540 char_u *p;
|
|
2541 {
|
|
2542 int i;
|
|
2543 int j;
|
|
2544
|
|
2545 if (*p == NUL)
|
|
2546 return 0;
|
|
2547
|
|
2548 if (enc_utf8)
|
|
2549 {
|
|
2550 /* Find the last character that is 10xx.xxxx */
|
|
2551 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i)
|
|
2552 ;
|
|
2553 /* Check for illegal sequence. */
|
|
2554 for (j = 0; p - j > base; ++j)
|
|
2555 if ((p[-j] & 0xc0) != 0x80)
|
|
2556 break;
|
|
2557 if (utf8len_tab[p[-j]] != i + j + 1)
|
|
2558 return 0;
|
|
2559 return i;
|
|
2560 }
|
|
2561
|
|
2562 /* It can't be the first byte if a double-byte when not using DBCS, at the
|
|
2563 * end of the string or the byte can't start a double-byte. */
|
|
2564 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1)
|
|
2565 return 0;
|
|
2566
|
|
2567 /* Return 1 when on the lead byte, 0 when on the tail byte. */
|
|
2568 return 1 - dbcs_head_off(base, p);
|
|
2569 }
|
|
2570
|
26
|
2571 #if defined(HAVE_GTK2) || defined(PROTO)
|
|
2572 /*
|
|
2573 * Return TRUE if string "s" is a valid utf-8 string.
|
|
2574 * When "end" is NULL stop at the first NUL.
|
|
2575 * When "end" is positive stop there.
|
|
2576 */
|
|
2577 int
|
|
2578 utf_valid_string(s, end)
|
|
2579 char_u *s;
|
|
2580 char_u *end;
|
|
2581 {
|
|
2582 int l;
|
|
2583 char_u *p = s;
|
|
2584
|
|
2585 while (end == NULL ? *p != NUL : p < end)
|
|
2586 {
|
|
2587 if ((*p & 0xc0) == 0x80)
|
|
2588 return FALSE; /* invalid lead byte */
|
|
2589 l = utf8len_tab[*p];
|
|
2590 if (end != NULL && p + l > end)
|
|
2591 return FALSE; /* incomplete byte sequence */
|
|
2592 ++p;
|
|
2593 while (--l > 0)
|
|
2594 if ((*p++ & 0xc0) != 0x80)
|
|
2595 return FALSE; /* invalid trail byte */
|
|
2596 }
|
|
2597 return TRUE;
|
|
2598 }
|
|
2599 #endif
|
|
2600
|
7
|
2601 #if defined(FEAT_GUI) || defined(PROTO)
|
|
2602 /*
|
|
2603 * Special version of mb_tail_off() for use in ScreenLines[].
|
|
2604 */
|
|
2605 int
|
|
2606 dbcs_screen_tail_off(base, p)
|
|
2607 char_u *base;
|
|
2608 char_u *p;
|
|
2609 {
|
|
2610 /* It can't be the first byte if a double-byte when not using DBCS, at the
|
|
2611 * end of the string or the byte can't start a double-byte.
|
|
2612 * For euc-jp an 0x8e byte always means we have a lead byte in the current
|
|
2613 * cell. */
|
|
2614 if (*p == NUL || p[1] == NUL
|
|
2615 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)
|
|
2616 || MB_BYTE2LEN(*p) == 1)
|
|
2617 return 0;
|
|
2618
|
|
2619 /* Return 1 when on the lead byte, 0 when on the tail byte. */
|
|
2620 return 1 - dbcs_screen_head_off(base, p);
|
|
2621 }
|
|
2622 #endif
|
|
2623
|
|
2624 /*
|
|
2625 * If the cursor moves on an trail byte, set the cursor on the lead byte.
|
|
2626 * Thus it moves left if necessary.
|
|
2627 * Return TRUE when the cursor was adjusted.
|
|
2628 */
|
|
2629 void
|
|
2630 mb_adjust_cursor()
|
|
2631 {
|
|
2632 mb_adjustpos(&curwin->w_cursor);
|
|
2633 }
|
|
2634
|
|
2635 /*
|
|
2636 * Adjust position "*lp" to point to the first byte of a multi-byte character.
|
|
2637 * If it points to a tail byte it's moved backwards to the head byte.
|
|
2638 */
|
|
2639 void
|
|
2640 mb_adjustpos(lp)
|
|
2641 pos_T *lp;
|
|
2642 {
|
|
2643 char_u *p;
|
|
2644
|
|
2645 if (lp->col > 0
|
|
2646 #ifdef FEAT_VIRTUALEDIT
|
|
2647 || lp->coladd > 1
|
|
2648 #endif
|
|
2649 )
|
|
2650 {
|
|
2651 p = ml_get(lp->lnum);
|
|
2652 lp->col -= (*mb_head_off)(p, p + lp->col);
|
|
2653 #ifdef FEAT_VIRTUALEDIT
|
|
2654 /* Reset "coladd" when the cursor would be on the right half of a
|
|
2655 * double-wide character. */
|
|
2656 if (lp->coladd == 1
|
|
2657 && p[lp->col] != TAB
|
|
2658 && vim_isprintc((*mb_ptr2char)(p + lp->col))
|
|
2659 && ptr2cells(p + lp->col) > 1)
|
|
2660 lp->coladd = 0;
|
|
2661 #endif
|
|
2662 }
|
|
2663 }
|
|
2664
|
|
2665 /*
|
|
2666 * Return a pointer to the character before "*p", if there is one.
|
|
2667 */
|
|
2668 char_u *
|
|
2669 mb_prevptr(line, p)
|
|
2670 char_u *line; /* start of the string */
|
|
2671 char_u *p;
|
|
2672 {
|
|
2673 if (p > line)
|
39
|
2674 mb_ptr_back(line, p);
|
7
|
2675 return p;
|
|
2676 }
|
|
2677
|
|
2678 /*
|
312
|
2679 * Return the character length of "str". Each multi-byte character counts as
|
|
2680 * one.
|
7
|
2681 */
|
|
2682 int
|
|
2683 mb_charlen(str)
|
|
2684 char_u *str;
|
|
2685 {
|
|
2686 int count;
|
|
2687
|
|
2688 if (str == NULL)
|
|
2689 return 0;
|
|
2690
|
|
2691 for (count = 0; *str != NUL; count++)
|
|
2692 str += (*mb_ptr2len_check)(str);
|
|
2693
|
|
2694 return count;
|
|
2695 }
|
|
2696
|
|
2697 /*
|
|
2698 * Try to un-escape a multi-byte character.
|
|
2699 * Used for the "to" and "from" part of a mapping.
|
|
2700 * Return the un-escaped string if it is a multi-byte character, and advance
|
|
2701 * "pp" to just after the bytes that formed it.
|
|
2702 * Return NULL if no multi-byte char was found.
|
|
2703 */
|
|
2704 char_u *
|
|
2705 mb_unescape(pp)
|
|
2706 char_u **pp;
|
|
2707 {
|
|
2708 static char_u buf[MB_MAXBYTES + 1];
|
|
2709 int n, m = 0;
|
|
2710 char_u *str = *pp;
|
|
2711
|
|
2712 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI
|
|
2713 * KS_EXTRA KE_CSI to CSI. */
|
|
2714 for (n = 0; str[n] != NUL && m <= MB_MAXBYTES; ++n)
|
|
2715 {
|
|
2716 if (str[n] == K_SPECIAL
|
|
2717 && str[n + 1] == KS_SPECIAL
|
|
2718 && str[n + 2] == KE_FILLER)
|
|
2719 {
|
|
2720 buf[m++] = K_SPECIAL;
|
|
2721 n += 2;
|
|
2722 }
|
|
2723 # ifdef FEAT_GUI
|
|
2724 else if (str[n] == CSI
|
|
2725 && str[n + 1] == KS_EXTRA
|
|
2726 && str[n + 2] == (int)KE_CSI)
|
|
2727 {
|
|
2728 buf[m++] = CSI;
|
|
2729 n += 2;
|
|
2730 }
|
|
2731 # endif
|
|
2732 else if (str[n] == K_SPECIAL
|
|
2733 # ifdef FEAT_GUI
|
|
2734 || str[n] == CSI
|
|
2735 # endif
|
|
2736 )
|
|
2737 break; /* a special key can't be a multibyte char */
|
|
2738 else
|
|
2739 buf[m++] = str[n];
|
|
2740 buf[m] = NUL;
|
|
2741
|
|
2742 /* Return a multi-byte character if it's found. An illegal sequence
|
|
2743 * will result in a 1 here. */
|
|
2744 if ((*mb_ptr2len_check)(buf) > 1)
|
|
2745 {
|
|
2746 *pp = str + n + 1;
|
|
2747 return buf;
|
|
2748 }
|
|
2749 }
|
|
2750 return NULL;
|
|
2751 }
|
|
2752
|
|
2753 /*
|
|
2754 * Return TRUE if the character at "row"/"col" on the screen is the left side
|
|
2755 * of a double-width character.
|
|
2756 * Caller must make sure "row" and "col" are not invalid!
|
|
2757 */
|
|
2758 int
|
|
2759 mb_lefthalve(row, col)
|
|
2760 int row;
|
|
2761 int col;
|
|
2762 {
|
|
2763 #ifdef FEAT_HANGULIN
|
|
2764 if (composing_hangul)
|
|
2765 return TRUE;
|
|
2766 #endif
|
|
2767 if (enc_dbcs != 0)
|
|
2768 return dbcs_off2cells(LineOffset[row] + col) > 1;
|
|
2769 if (enc_utf8)
|
|
2770 return (col + 1 < Columns
|
|
2771 && ScreenLines[LineOffset[row] + col + 1] == 0);
|
|
2772 return FALSE;
|
|
2773 }
|
|
2774
|
|
2775 # if defined(FEAT_CLIPBOARD) || defined(FEAT_GUI) || defined(FEAT_RIGHTLEFT) \
|
|
2776 || defined(PROTO)
|
|
2777 /*
|
|
2778 * Correct a position on the screen, if it's the right halve of a double-wide
|
|
2779 * char move it to the left halve. Returns the corrected column.
|
|
2780 */
|
|
2781 int
|
|
2782 mb_fix_col(col, row)
|
|
2783 int col;
|
|
2784 int row;
|
|
2785 {
|
|
2786 col = check_col(col);
|
|
2787 row = check_row(row);
|
|
2788 if (has_mbyte && ScreenLines != NULL && col > 0
|
|
2789 && ((enc_dbcs
|
|
2790 && ScreenLines[LineOffset[row] + col] != NUL
|
|
2791 && dbcs_screen_head_off(ScreenLines + LineOffset[row],
|
|
2792 ScreenLines + LineOffset[row] + col))
|
|
2793 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0)))
|
|
2794 --col;
|
|
2795 return col;
|
|
2796 }
|
|
2797 # endif
|
|
2798 #endif
|
|
2799
|
|
2800 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO)
|
|
2801 static int enc_alias_search __ARGS((char_u *name));
|
|
2802
|
|
2803 /*
|
|
2804 * Skip the Vim specific head of a 'encoding' name.
|
|
2805 */
|
|
2806 char_u *
|
|
2807 enc_skip(p)
|
|
2808 char_u *p;
|
|
2809 {
|
|
2810 if (STRNCMP(p, "2byte-", 6) == 0)
|
|
2811 return p + 6;
|
|
2812 if (STRNCMP(p, "8bit-", 5) == 0)
|
|
2813 return p + 5;
|
|
2814 return p;
|
|
2815 }
|
|
2816
|
|
2817 /*
|
|
2818 * Find the canonical name for encoding "enc".
|
|
2819 * When the name isn't recognized, returns "enc" itself, but with all lower
|
|
2820 * case characters and '_' replaced with '-'.
|
|
2821 * Returns an allocated string. NULL for out-of-memory.
|
|
2822 */
|
|
2823 char_u *
|
|
2824 enc_canonize(enc)
|
|
2825 char_u *enc;
|
|
2826 {
|
|
2827 char_u *r;
|
|
2828 char_u *p, *s;
|
|
2829 int i;
|
|
2830
|
39
|
2831 # ifdef FEAT_MBYTE
|
|
2832 if (STRCMP(enc, "default") == 0)
|
|
2833 {
|
|
2834 /* Use the default encoding as it's found by set_init_1(). */
|
|
2835 r = get_encoding_default();
|
|
2836 if (r == NULL)
|
|
2837 r = (char_u *)"latin1";
|
|
2838 return vim_strsave(r);
|
|
2839 }
|
|
2840 # endif
|
|
2841
|
7
|
2842 /* copy "enc" to allocted memory, with room for two '-' */
|
|
2843 r = alloc((unsigned)(STRLEN(enc) + 3));
|
|
2844 if (r != NULL)
|
|
2845 {
|
|
2846 /* Make it all lower case and replace '_' with '-'. */
|
|
2847 p = r;
|
|
2848 for (s = enc; *s != NUL; ++s)
|
|
2849 {
|
|
2850 if (*s == '_')
|
|
2851 *p++ = '-';
|
|
2852 else
|
|
2853 *p++ = TOLOWER_ASC(*s);
|
|
2854 }
|
|
2855 *p = NUL;
|
|
2856
|
|
2857 /* Skip "2byte-" and "8bit-". */
|
|
2858 p = enc_skip(r);
|
|
2859
|
|
2860 /* "iso8859" -> "iso-8859" */
|
|
2861 if (STRNCMP(p, "iso8859", 7) == 0)
|
|
2862 {
|
|
2863 mch_memmove(p + 4, p + 3, STRLEN(p + 2));
|
|
2864 p[3] = '-';
|
|
2865 }
|
|
2866
|
|
2867 /* "iso-8859n" -> "iso-8859-n" */
|
|
2868 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-')
|
|
2869 {
|
|
2870 mch_memmove(p + 9, p + 8, STRLEN(p + 7));
|
|
2871 p[8] = '-';
|
|
2872 }
|
|
2873
|
|
2874 /* "latin-N" -> "latinN" */
|
|
2875 if (STRNCMP(p, "latin-", 6) == 0)
|
|
2876 mch_memmove(p + 5, p + 6, STRLEN(p + 5));
|
|
2877
|
|
2878 if (enc_canon_search(p) >= 0)
|
|
2879 {
|
|
2880 /* canonical name can be used unmodified */
|
|
2881 if (p != r)
|
|
2882 mch_memmove(r, p, STRLEN(p) + 1);
|
|
2883 }
|
|
2884 else if ((i = enc_alias_search(p)) >= 0)
|
|
2885 {
|
|
2886 /* alias recognized, get canonical name */
|
|
2887 vim_free(r);
|
|
2888 r = vim_strsave((char_u *)enc_canon_table[i].name);
|
|
2889 }
|
|
2890 }
|
|
2891 return r;
|
|
2892 }
|
|
2893
|
|
2894 /*
|
|
2895 * Search for an encoding alias of "name".
|
|
2896 * Returns -1 when not found.
|
|
2897 */
|
|
2898 static int
|
|
2899 enc_alias_search(name)
|
|
2900 char_u *name;
|
|
2901 {
|
|
2902 int i;
|
|
2903
|
|
2904 for (i = 0; enc_alias_table[i].name != NULL; ++i)
|
|
2905 if (STRCMP(name, enc_alias_table[i].name) == 0)
|
|
2906 return enc_alias_table[i].canon;
|
|
2907 return -1;
|
|
2908 }
|
|
2909 #endif
|
|
2910
|
|
2911 #if defined(FEAT_MBYTE) || defined(PROTO)
|
|
2912
|
|
2913 #ifdef HAVE_LANGINFO_H
|
|
2914 # include <langinfo.h>
|
|
2915 #endif
|
|
2916
|
|
2917 /*
|
|
2918 * Get the canonicalized encoding of the current locale.
|
|
2919 * Returns an allocated string when successful, NULL when not.
|
|
2920 */
|
|
2921 char_u *
|
|
2922 enc_locale()
|
|
2923 {
|
|
2924 #ifndef WIN3264
|
|
2925 char *s;
|
|
2926 char *p;
|
|
2927 int i;
|
|
2928 #endif
|
|
2929 char buf[50];
|
|
2930 #ifdef WIN3264
|
|
2931 long acp = GetACP();
|
|
2932
|
|
2933 if (acp == 1200)
|
|
2934 STRCPY(buf, "ucs-2le");
|
407
|
2935 else if (acp == 1252) /* cp1252 is used as latin1 */
|
7
|
2936 STRCPY(buf, "latin1");
|
|
2937 else
|
|
2938 sprintf(buf, "cp%ld", acp);
|
|
2939 #else
|
|
2940 # ifdef HAVE_NL_LANGINFO_CODESET
|
|
2941 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL)
|
|
2942 # endif
|
167
|
2943 # ifdef MACOS
|
|
2944 s = "utf-8";
|
|
2945 # else
|
|
2946 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE)
|
7
|
2947 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL)
|
167
|
2948 # endif
|
7
|
2949 if ((s = getenv("LC_ALL")) == NULL || *s == NUL)
|
|
2950 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL)
|
|
2951 s = getenv("LANG");
|
167
|
2952 # endif
|
7
|
2953
|
|
2954 if (s == NULL || *s == NUL)
|
|
2955 return FAIL;
|
|
2956
|
|
2957 /* The most generic locale format is:
|
|
2958 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]]
|
|
2959 * If there is a '.' remove the part before it.
|
|
2960 * if there is something after the codeset, remove it.
|
|
2961 * Make the name lowercase and replace '_' with '-'.
|
|
2962 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn",
|
|
2963 * "ko_KR.EUC" == "euc-kr"
|
|
2964 */
|
|
2965 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL)
|
|
2966 {
|
|
2967 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0
|
|
2968 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_')
|
|
2969 {
|
|
2970 /* copy "XY.EUC" to "euc-XY" to buf[10] */
|
|
2971 STRCPY(buf + 10, "euc-");
|
|
2972 buf[14] = p[-2];
|
|
2973 buf[15] = p[-1];
|
|
2974 buf[16] = 0;
|
|
2975 s = buf + 10;
|
|
2976 }
|
|
2977 else
|
|
2978 s = p + 1;
|
|
2979 }
|
|
2980 for (i = 0; s[i] != NUL && i < sizeof(buf) - 1; ++i)
|
|
2981 {
|
|
2982 if (s[i] == '_' || s[i] == '-')
|
|
2983 buf[i] = '-';
|
|
2984 else if (isalnum((int)s[i]))
|
|
2985 buf[i] = TOLOWER_ASC(s[i]);
|
|
2986 else
|
|
2987 break;
|
|
2988 }
|
|
2989 buf[i] = NUL;
|
|
2990 #endif
|
|
2991
|
|
2992 return enc_canonize((char_u *)buf);
|
|
2993 }
|
|
2994
|
|
2995 #if defined(WIN3264) || defined(PROTO)
|
|
2996 /*
|
|
2997 * Convert an encoding name to an MS-Windows codepage.
|
|
2998 * Returns zero if no codepage can be figured out.
|
|
2999 */
|
|
3000 int
|
|
3001 encname2codepage(name)
|
|
3002 char_u *name;
|
|
3003 {
|
|
3004 int cp;
|
|
3005 char_u *p = name;
|
|
3006 int idx;
|
|
3007
|
|
3008 if (STRNCMP(p, "8bit-", 5) == 0)
|
|
3009 p += 5;
|
|
3010 else if (STRNCMP(p_enc, "2byte-", 6) == 0)
|
|
3011 p += 6;
|
|
3012
|
|
3013 if (p[0] == 'c' && p[1] == 'p')
|
|
3014 cp = atoi(p + 2);
|
|
3015 else if ((idx = enc_canon_search(p)) >= 0)
|
|
3016 cp = enc_canon_table[idx].codepage;
|
|
3017 else
|
|
3018 return 0;
|
|
3019 if (IsValidCodePage(cp))
|
|
3020 return cp;
|
|
3021 return 0;
|
|
3022 }
|
|
3023 #endif
|
|
3024
|
|
3025 # if defined(USE_ICONV) || defined(PROTO)
|
|
3026
|
|
3027 static char_u *iconv_string __ARGS((vimconv_T *vcp, char_u *str, int slen, int *unconvlenp));
|
|
3028
|
|
3029 /*
|
|
3030 * Call iconv_open() with a check if iconv() works properly (there are broken
|
|
3031 * versions).
|
|
3032 * Returns (void *)-1 if failed.
|
|
3033 * (should return iconv_t, but that causes problems with prototypes).
|
|
3034 */
|
|
3035 void *
|
|
3036 my_iconv_open(to, from)
|
|
3037 char_u *to;
|
|
3038 char_u *from;
|
|
3039 {
|
|
3040 iconv_t fd;
|
|
3041 #define ICONV_TESTLEN 400
|
|
3042 char_u tobuf[ICONV_TESTLEN];
|
|
3043 char *p;
|
|
3044 size_t tolen;
|
|
3045 static int iconv_ok = -1;
|
|
3046
|
|
3047 if (iconv_ok == FALSE)
|
|
3048 return (void *)-1; /* detected a broken iconv() previously */
|
|
3049
|
|
3050 #ifdef DYNAMIC_ICONV
|
|
3051 /* Check if the iconv.dll can be found. */
|
|
3052 if (!iconv_enabled(TRUE))
|
|
3053 return (void *)-1;
|
|
3054 #endif
|
|
3055
|
|
3056 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from));
|
|
3057
|
|
3058 if (fd != (iconv_t)-1 && iconv_ok == -1)
|
|
3059 {
|
|
3060 /*
|
|
3061 * Do a dummy iconv() call to check if it actually works. There is a
|
|
3062 * version of iconv() on Linux that is broken. We can't ignore it,
|
|
3063 * because it's wide-spread. The symptoms are that after outputting
|
|
3064 * the initial shift state the "to" pointer is NULL and conversion
|
|
3065 * stops for no apparent reason after about 8160 characters.
|
|
3066 */
|
|
3067 p = (char *)tobuf;
|
|
3068 tolen = ICONV_TESTLEN;
|
|
3069 (void)iconv(fd, NULL, NULL, &p, &tolen);
|
|
3070 if (p == NULL)
|
|
3071 {
|
|
3072 iconv_ok = FALSE;
|
|
3073 iconv_close(fd);
|
|
3074 fd = (iconv_t)-1;
|
|
3075 }
|
|
3076 else
|
|
3077 iconv_ok = TRUE;
|
|
3078 }
|
|
3079
|
|
3080 return (void *)fd;
|
|
3081 }
|
|
3082
|
|
3083 /*
|
|
3084 * Convert the string "str[slen]" with iconv().
|
|
3085 * If "unconvlenp" is not NULL handle the string ending in an incomplete
|
|
3086 * sequence and set "*unconvlenp" to the length of it.
|
|
3087 * Returns the converted string in allocated memory. NULL for an error.
|
|
3088 */
|
|
3089 static char_u *
|
|
3090 iconv_string(vcp, str, slen, unconvlenp)
|
|
3091 vimconv_T *vcp;
|
|
3092 char_u *str;
|
|
3093 int slen;
|
|
3094 int *unconvlenp;
|
|
3095 {
|
|
3096 const char *from;
|
|
3097 size_t fromlen;
|
|
3098 char *to;
|
|
3099 size_t tolen;
|
|
3100 size_t len = 0;
|
|
3101 size_t done = 0;
|
|
3102 char_u *result = NULL;
|
|
3103 char_u *p;
|
|
3104 int l;
|
|
3105
|
|
3106 from = (char *)str;
|
|
3107 fromlen = slen;
|
|
3108 for (;;)
|
|
3109 {
|
|
3110 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG)
|
|
3111 {
|
|
3112 /* Allocate enough room for most conversions. When re-allocating
|
|
3113 * increase the buffer size. */
|
|
3114 len = len + fromlen * 2 + 40;
|
|
3115 p = alloc((unsigned)len);
|
|
3116 if (p != NULL && done > 0)
|
|
3117 mch_memmove(p, result, done);
|
|
3118 vim_free(result);
|
|
3119 result = p;
|
|
3120 if (result == NULL) /* out of memory */
|
|
3121 break;
|
|
3122 }
|
|
3123
|
|
3124 to = (char *)result + done;
|
|
3125 tolen = len - done - 2;
|
|
3126 /* Avoid a warning for systems with a wrong iconv() prototype by
|
|
3127 * casting the second argument to void *. */
|
|
3128 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen)
|
|
3129 != (size_t)-1)
|
|
3130 {
|
|
3131 /* Finished, append a NUL. */
|
|
3132 *to = NUL;
|
|
3133 break;
|
|
3134 }
|
|
3135
|
|
3136 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded
|
|
3137 * iconv library may use one of them. */
|
|
3138 if (!vcp->vc_fail && unconvlenp != NULL
|
|
3139 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL))
|
|
3140 {
|
|
3141 /* Handle an incomplete sequence at the end. */
|
|
3142 *to = NUL;
|
|
3143 *unconvlenp = fromlen;
|
|
3144 break;
|
|
3145 }
|
|
3146
|
|
3147 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded
|
|
3148 * iconv library may use one of them. */
|
|
3149 else if (!vcp->vc_fail
|
|
3150 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ
|
|
3151 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL))
|
|
3152 {
|
|
3153 /* Can't convert: insert a '?' and skip a character. This assumes
|
|
3154 * conversion from 'encoding' to something else. In other
|
|
3155 * situations we don't know what to skip anyway. */
|
|
3156 *to++ = '?';
|
|
3157 if ((*mb_ptr2cells)((char_u *)from) > 1)
|
|
3158 *to++ = '?';
|
221
|
3159 if (enc_utf8)
|
|
3160 l = utfc_ptr2len_check_len((char_u *)from, fromlen);
|
|
3161 else
|
|
3162 {
|
|
3163 l = (*mb_ptr2len_check)((char_u *)from);
|
227
|
3164 if (l > (int)fromlen)
|
221
|
3165 l = fromlen;
|
|
3166 }
|
7
|
3167 from += l;
|
|
3168 fromlen -= l;
|
|
3169 }
|
|
3170 else if (ICONV_ERRNO != ICONV_E2BIG)
|
|
3171 {
|
|
3172 /* conversion failed */
|
|
3173 vim_free(result);
|
|
3174 result = NULL;
|
|
3175 break;
|
|
3176 }
|
|
3177 /* Not enough room or skipping illegal sequence. */
|
|
3178 done = to - (char *)result;
|
|
3179 }
|
|
3180 return result;
|
|
3181 }
|
|
3182
|
|
3183 # if defined(DYNAMIC_ICONV) || defined(PROTO)
|
|
3184 /*
|
|
3185 * Dynamically load the "iconv.dll" on Win32.
|
|
3186 */
|
|
3187
|
|
3188 #ifndef DYNAMIC_ICONV /* just generating prototypes */
|
|
3189 # define HINSTANCE int
|
|
3190 #endif
|
297
|
3191 static HINSTANCE hIconvDLL = 0;
|
|
3192 static HINSTANCE hMsvcrtDLL = 0;
|
7
|
3193
|
|
3194 # ifndef DYNAMIC_ICONV_DLL
|
|
3195 # define DYNAMIC_ICONV_DLL "iconv.dll"
|
|
3196 # define DYNAMIC_ICONV_DLL_ALT "libiconv.dll"
|
|
3197 # endif
|
|
3198 # ifndef DYNAMIC_MSVCRT_DLL
|
|
3199 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll"
|
|
3200 # endif
|
|
3201
|
|
3202 /*
|
|
3203 * Try opening the iconv.dll and return TRUE if iconv() can be used.
|
|
3204 */
|
|
3205 int
|
|
3206 iconv_enabled(verbose)
|
|
3207 int verbose;
|
|
3208 {
|
|
3209 if (hIconvDLL != 0 && hMsvcrtDLL != 0)
|
|
3210 return TRUE;
|
|
3211 hIconvDLL = LoadLibrary(DYNAMIC_ICONV_DLL);
|
|
3212 if (hIconvDLL == 0) /* sometimes it's called libiconv.dll */
|
|
3213 hIconvDLL = LoadLibrary(DYNAMIC_ICONV_DLL_ALT);
|
|
3214 if (hIconvDLL != 0)
|
|
3215 hMsvcrtDLL = LoadLibrary(DYNAMIC_MSVCRT_DLL);
|
|
3216 if (hIconvDLL == 0 || hMsvcrtDLL == 0)
|
|
3217 {
|
|
3218 /* Only give the message when 'verbose' is set, otherwise it might be
|
|
3219 * done whenever a conversion is attempted. */
|
|
3220 if (verbose && p_verbose > 0)
|
292
|
3221 {
|
|
3222 verbose_enter();
|
7
|
3223 EMSG2(_(e_loadlib),
|
|
3224 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL);
|
292
|
3225 verbose_leave();
|
|
3226 }
|
7
|
3227 iconv_end();
|
|
3228 return FALSE;
|
|
3229 }
|
|
3230
|
344
|
3231 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv");
|
|
3232 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open");
|
|
3233 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close");
|
|
3234 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl");
|
|
3235 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno");
|
7
|
3236 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL
|
|
3237 || iconvctl == NULL || iconv_errno == NULL)
|
|
3238 {
|
|
3239 iconv_end();
|
|
3240 if (verbose && p_verbose > 0)
|
292
|
3241 {
|
|
3242 verbose_enter();
|
7
|
3243 EMSG2(_(e_loadfunc), "for libiconv");
|
292
|
3244 verbose_leave();
|
|
3245 }
|
7
|
3246 return FALSE;
|
|
3247 }
|
|
3248 return TRUE;
|
|
3249 }
|
|
3250
|
|
3251 void
|
|
3252 iconv_end()
|
|
3253 {
|
|
3254 /* Don't use iconv() when inputting or outputting characters. */
|
|
3255 if (input_conv.vc_type == CONV_ICONV)
|
|
3256 convert_setup(&input_conv, NULL, NULL);
|
|
3257 if (output_conv.vc_type == CONV_ICONV)
|
|
3258 convert_setup(&output_conv, NULL, NULL);
|
|
3259
|
|
3260 if (hIconvDLL != 0)
|
|
3261 FreeLibrary(hIconvDLL);
|
|
3262 if (hMsvcrtDLL != 0)
|
|
3263 FreeLibrary(hMsvcrtDLL);
|
|
3264 hIconvDLL = 0;
|
|
3265 hMsvcrtDLL = 0;
|
|
3266 }
|
|
3267 # endif /* DYNAMIC_ICONV */
|
|
3268 # endif /* USE_ICONV */
|
|
3269
|
|
3270 #endif /* FEAT_MBYTE */
|
|
3271
|
|
3272 #if defined(FEAT_XIM) || defined(PROTO)
|
|
3273
|
|
3274 # ifdef FEAT_GUI_GTK
|
|
3275 static int xim_has_preediting INIT(= FALSE); /* IM current status */
|
|
3276
|
|
3277 /*
|
|
3278 * Set preedit_start_col to the current cursor position.
|
|
3279 */
|
|
3280 static void
|
|
3281 init_preedit_start_col(void)
|
|
3282 {
|
|
3283 if (State & CMDLINE)
|
|
3284 preedit_start_col = cmdline_getvcol_cursor();
|
|
3285 else if (curwin != NULL)
|
|
3286 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL);
|
|
3287 /* Prevent that preediting marks the buffer as changed. */
|
|
3288 xim_changed_while_preediting = curbuf->b_changed;
|
|
3289 }
|
|
3290 # endif
|
|
3291
|
|
3292 # if defined(HAVE_GTK2) && !defined(PROTO)
|
|
3293
|
|
3294 static int im_is_active = FALSE; /* IM is enabled for current mode */
|
|
3295 static int im_preedit_cursor = 0; /* cursor offset in characters */
|
|
3296 static int im_preedit_trailing = 0; /* number of characters after cursor */
|
|
3297
|
|
3298 static unsigned long im_commit_handler_id = 0;
|
|
3299 static unsigned int im_activatekey_keyval = GDK_VoidSymbol;
|
|
3300 static unsigned int im_activatekey_state = 0;
|
|
3301
|
|
3302 void
|
|
3303 im_set_active(int active)
|
|
3304 {
|
|
3305 int was_active;
|
|
3306
|
|
3307 was_active = !!im_is_active;
|
|
3308 im_is_active = (active && !p_imdisable);
|
|
3309
|
|
3310 if (im_is_active != was_active)
|
|
3311 xim_reset();
|
|
3312 }
|
|
3313
|
|
3314 void
|
|
3315 xim_set_focus(int focus)
|
|
3316 {
|
|
3317 if (xic != NULL)
|
|
3318 {
|
|
3319 if (focus)
|
|
3320 gtk_im_context_focus_in(xic);
|
|
3321 else
|
|
3322 gtk_im_context_focus_out(xic);
|
|
3323 }
|
|
3324 }
|
|
3325
|
|
3326 void
|
|
3327 im_set_position(int row, int col)
|
|
3328 {
|
|
3329 if (xic != NULL)
|
|
3330 {
|
|
3331 GdkRectangle area;
|
|
3332
|
|
3333 area.x = FILL_X(col);
|
|
3334 area.y = FILL_Y(row);
|
|
3335 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1);
|
|
3336 area.height = gui.char_height;
|
|
3337
|
|
3338 gtk_im_context_set_cursor_location(xic, &area);
|
|
3339 }
|
|
3340 }
|
|
3341
|
|
3342 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */
|
|
3343 void
|
|
3344 xim_set_preedit(void)
|
|
3345 {
|
|
3346 im_set_position(gui.row, gui.col);
|
|
3347 }
|
|
3348 # endif
|
|
3349
|
|
3350 static void
|
|
3351 im_add_to_input(char_u *str, int len)
|
|
3352 {
|
|
3353 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */
|
|
3354 if (input_conv.vc_type != CONV_NONE)
|
|
3355 {
|
|
3356 str = string_convert(&input_conv, str, &len);
|
|
3357 g_return_if_fail(str != NULL);
|
|
3358 }
|
|
3359
|
|
3360 add_to_input_buf_csi(str, len);
|
|
3361
|
|
3362 if (input_conv.vc_type != CONV_NONE)
|
|
3363 vim_free(str);
|
|
3364
|
|
3365 if (p_mh) /* blank out the pointer if necessary */
|
|
3366 gui_mch_mousehide(TRUE);
|
|
3367 }
|
|
3368
|
|
3369 static void
|
|
3370 im_delete_preedit(void)
|
|
3371 {
|
|
3372 char_u bskey[] = {CSI, 'k', 'b'};
|
|
3373 char_u delkey[] = {CSI, 'k', 'D'};
|
|
3374
|
|
3375 if (State & NORMAL)
|
|
3376 {
|
|
3377 im_preedit_cursor = 0;
|
|
3378 return;
|
|
3379 }
|
|
3380 for (; im_preedit_cursor > 0; --im_preedit_cursor)
|
|
3381 add_to_input_buf(bskey, (int)sizeof(bskey));
|
|
3382
|
|
3383 for (; im_preedit_trailing > 0; --im_preedit_trailing)
|
|
3384 add_to_input_buf(delkey, (int)sizeof(delkey));
|
|
3385 }
|
|
3386
|
|
3387 static void
|
|
3388 im_correct_cursor(int num_move_back)
|
|
3389 {
|
|
3390 char_u backkey[] = {CSI, 'k', 'l'};
|
|
3391
|
|
3392 if (State & NORMAL)
|
|
3393 return;
|
|
3394 # ifdef FEAT_RIGHTLEFT
|
|
3395 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl)
|
|
3396 backkey[2] = 'r';
|
|
3397 # endif
|
|
3398 for (; num_move_back > 0; --num_move_back)
|
|
3399 add_to_input_buf(backkey, (int)sizeof(backkey));
|
|
3400 }
|
|
3401
|
|
3402 static int xim_expected_char = NUL;
|
|
3403 static int xim_ignored_char = FALSE;
|
|
3404
|
|
3405 /*
|
|
3406 * Update the mode and cursor while in an IM callback.
|
|
3407 */
|
|
3408 static void
|
|
3409 im_show_info(void)
|
|
3410 {
|
|
3411 int old_vgetc_busy;
|
|
3412 old_vgetc_busy = vgetc_busy;
|
|
3413 vgetc_busy = TRUE;
|
|
3414 showmode();
|
|
3415 vgetc_busy = old_vgetc_busy;
|
|
3416 setcursor();
|
|
3417 out_flush();
|
|
3418 }
|
|
3419
|
|
3420 /*
|
|
3421 * Callback invoked when the user finished preediting.
|
|
3422 * Put the final string into the input buffer.
|
|
3423 */
|
|
3424 /*ARGSUSED0*/
|
|
3425 static void
|
|
3426 im_commit_cb(GtkIMContext *context, const gchar *str, gpointer data)
|
|
3427 {
|
|
3428 int slen = (int)STRLEN(str);
|
|
3429 int add_to_input = TRUE;
|
|
3430 int clen;
|
|
3431 int len = slen;
|
|
3432 int commit_with_preedit = TRUE;
|
|
3433 char_u *im_str, *p;
|
|
3434
|
|
3435 #ifdef XIM_DEBUG
|
|
3436 xim_log("im_commit_cb(): %s\n", str);
|
|
3437 #endif
|
|
3438
|
|
3439 /* The imhangul module doesn't reset the preedit string before
|
|
3440 * committing. Call im_delete_preedit() to work around that. */
|
|
3441 im_delete_preedit();
|
|
3442
|
|
3443 /* Indicate that preediting has finished. */
|
|
3444 if (preedit_start_col == MAXCOL)
|
|
3445 {
|
|
3446 init_preedit_start_col();
|
|
3447 commit_with_preedit = FALSE;
|
|
3448 }
|
|
3449
|
|
3450 /* The thing which setting "preedit_start_col" to MAXCOL means that
|
|
3451 * "preedit_start_col" will be set forcely when calling
|
|
3452 * preedit_changed_cb() next time.
|
|
3453 * "preedit_start_col" should not reset with MAXCOL on this part. Vim
|
|
3454 * is simulating the preediting by using add_to_input_str(). when
|
|
3455 * preedit begin immediately before committed, the typebuf is not
|
|
3456 * flushed to screen, then it can't get correct "preedit_start_col".
|
|
3457 * Thus, it should calculate the cells by adding cells of the committed
|
|
3458 * string. */
|
|
3459 if (input_conv.vc_type != CONV_NONE)
|
|
3460 {
|
|
3461 im_str = string_convert(&input_conv, (char_u *)str, &len);
|
|
3462 g_return_if_fail(im_str != NULL);
|
|
3463 }
|
|
3464 else
|
|
3465 im_str = (char_u *)str;
|
|
3466 clen = 0;
|
|
3467 for (p = im_str; p < im_str + len; p += (*mb_ptr2len_check)(p))
|
|
3468 clen += (*mb_ptr2cells)(p);
|
|
3469 if (input_conv.vc_type != CONV_NONE)
|
|
3470 vim_free(im_str);
|
|
3471 preedit_start_col += clen;
|
|
3472
|
|
3473 /* Is this a single character that matches a keypad key that's just
|
|
3474 * been pressed? If so, we don't want it to be entered as such - let
|
|
3475 * us carry on processing the raw keycode so that it may be used in
|
|
3476 * mappings as <kSomething>. */
|
|
3477 if (xim_expected_char != NUL)
|
|
3478 {
|
|
3479 /* We're currently processing a keypad or other special key */
|
|
3480 if (slen == 1 && str[0] == xim_expected_char)
|
|
3481 {
|
|
3482 /* It's a match - don't do it here */
|
|
3483 xim_ignored_char = TRUE;
|
|
3484 add_to_input = FALSE;
|
|
3485 }
|
|
3486 else
|
|
3487 {
|
|
3488 /* Not a match */
|
|
3489 xim_ignored_char = FALSE;
|
|
3490 }
|
|
3491 }
|
|
3492
|
|
3493 if (add_to_input)
|
|
3494 im_add_to_input((char_u *)str, slen);
|
|
3495
|
|
3496 /* Inserting chars while "im_is_active" is set does not cause a change of
|
|
3497 * buffer. When the chars are committed the buffer must be marked as
|
|
3498 * changed. */
|
|
3499 if (!commit_with_preedit)
|
|
3500 preedit_start_col = MAXCOL;
|
|
3501
|
|
3502 /* This flag is used in changed() at next call. */
|
|
3503 xim_changed_while_preediting = TRUE;
|
|
3504
|
|
3505 if (gtk_main_level() > 0)
|
|
3506 gtk_main_quit();
|
|
3507 }
|
|
3508
|
|
3509 /*
|
|
3510 * Callback invoked after start to the preedit.
|
|
3511 */
|
|
3512 /*ARGSUSED*/
|
|
3513 static void
|
|
3514 im_preedit_start_cb(GtkIMContext *context, gpointer data)
|
|
3515 {
|
|
3516 #ifdef XIM_DEBUG
|
|
3517 xim_log("im_preedit_start_cb()\n");
|
|
3518 #endif
|
|
3519
|
|
3520 im_is_active = TRUE;
|
|
3521 gui_update_cursor(TRUE, FALSE);
|
|
3522 }
|
|
3523
|
|
3524 /*
|
|
3525 * Callback invoked after end to the preedit.
|
|
3526 */
|
|
3527 /*ARGSUSED*/
|
|
3528 static void
|
|
3529 im_preedit_end_cb(GtkIMContext *context, gpointer data)
|
|
3530 {
|
|
3531 #ifdef XIM_DEBUG
|
|
3532 xim_log("im_preedit_end_cb()\n");
|
|
3533 #endif
|
|
3534 im_delete_preedit();
|
|
3535
|
|
3536 /* Indicate that preediting has finished */
|
|
3537 preedit_start_col = MAXCOL;
|
|
3538 xim_has_preediting = FALSE;
|
|
3539
|
|
3540 im_is_active = FALSE;
|
|
3541 gui_update_cursor(TRUE, FALSE);
|
|
3542 im_show_info();
|
|
3543 }
|
|
3544
|
|
3545 /*
|
|
3546 * Callback invoked after changes to the preedit string. If the preedit
|
|
3547 * string was empty before, remember the preedit start column so we know
|
|
3548 * where to apply feedback attributes. Delete the previous preedit string
|
|
3549 * if there was one, save the new preedit cursor offset, and put the new
|
|
3550 * string into the input buffer.
|
|
3551 *
|
|
3552 * TODO: The pragmatic "put into input buffer" approach used here has
|
|
3553 * several fundamental problems:
|
|
3554 *
|
|
3555 * - The characters in the preedit string are subject to remapping.
|
|
3556 * That's broken, only the finally committed string should be remapped.
|
|
3557 *
|
|
3558 * - There is a race condition involved: The retrieved value for the
|
|
3559 * current cursor position will be wrong if any unprocessed characters
|
|
3560 * are still queued in the input buffer.
|
|
3561 *
|
|
3562 * - Due to the lack of synchronization between the file buffer in memory
|
|
3563 * and any typed characters, it's practically impossible to implement the
|
|
3564 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM
|
|
3565 * modules for languages such as Thai are likely to rely on this feature
|
|
3566 * for proper operation.
|
|
3567 *
|
|
3568 * Conclusions: I think support for preediting needs to be moved to the
|
|
3569 * core parts of Vim. Ideally, until it has been committed, the preediting
|
|
3570 * string should only be displayed and not affect the buffer content at all.
|
|
3571 * The question how to deal with the synchronization issue still remains.
|
|
3572 * Circumventing the input buffer is probably not desirable. Anyway, I think
|
|
3573 * implementing "retrieve_surrounding" is the only hard problem.
|
|
3574 *
|
|
3575 * One way to solve all of this in a clean manner would be to queue all key
|
|
3576 * press/release events "as is" in the input buffer, and apply the IM filtering
|
|
3577 * at the receiving end of the queue. This, however, would have a rather large
|
|
3578 * impact on the code base. If there is an easy way to force processing of all
|
|
3579 * remaining input from within the "retrieve_surrounding" signal handler, this
|
|
3580 * might not be necessary. Gotta ask on vim-dev for opinions.
|
|
3581 */
|
|
3582 /*ARGSUSED1*/
|
|
3583 static void
|
|
3584 im_preedit_changed_cb(GtkIMContext *context, gpointer data)
|
|
3585 {
|
|
3586 char *preedit_string = NULL;
|
|
3587 int cursor_index = 0;
|
|
3588 int num_move_back = 0;
|
|
3589 char_u *str;
|
|
3590 char_u *p;
|
|
3591 int i;
|
|
3592
|
|
3593 gtk_im_context_get_preedit_string(context,
|
|
3594 &preedit_string, NULL,
|
|
3595 &cursor_index);
|
|
3596
|
|
3597 #ifdef XIM_DEBUG
|
|
3598 xim_log("im_preedit_changed_cb(): %s\n", preedit_string);
|
|
3599 #endif
|
|
3600
|
|
3601 g_return_if_fail(preedit_string != NULL); /* just in case */
|
|
3602
|
|
3603 /* If preedit_start_col is MAXCOL set it to the current cursor position. */
|
|
3604 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0')
|
|
3605 {
|
|
3606 xim_has_preediting = TRUE;
|
|
3607
|
|
3608 /* Urgh, this breaks if the input buffer isn't empty now */
|
|
3609 init_preedit_start_col();
|
|
3610 }
|
|
3611 else if (cursor_index == 0 && preedit_string[0] == '\0')
|
|
3612 {
|
|
3613 if (preedit_start_col == MAXCOL)
|
|
3614 xim_has_preediting = FALSE;
|
|
3615
|
|
3616 /* If at the start position (after typing backspace)
|
|
3617 * preedit_start_col must be reset. */
|
|
3618 preedit_start_col = MAXCOL;
|
|
3619 }
|
|
3620
|
|
3621 im_delete_preedit();
|
|
3622
|
|
3623 /*
|
|
3624 * Compute the end of the preediting area: "preedit_end_col".
|
|
3625 * According to the documentation of gtk_im_context_get_preedit_string(),
|
|
3626 * the cursor_pos output argument returns the offset in bytes. This is
|
|
3627 * unfortunately not true -- real life shows the offset is in characters,
|
|
3628 * and the GTK+ source code agrees with me. Will file a bug later.
|
|
3629 */
|
|
3630 if (preedit_start_col != MAXCOL)
|
|
3631 preedit_end_col = preedit_start_col;
|
|
3632 str = (char_u *)preedit_string;
|
|
3633 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i)
|
|
3634 {
|
|
3635 int is_composing;
|
|
3636
|
|
3637 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p)));
|
|
3638 /*
|
|
3639 * These offsets are used as counters when generating <BS> and <Del>
|
|
3640 * to delete the preedit string. So don't count composing characters
|
|
3641 * unless 'delcombine' is enabled.
|
|
3642 */
|
|
3643 if (!is_composing || p_deco)
|
|
3644 {
|
|
3645 if (i < cursor_index)
|
|
3646 ++im_preedit_cursor;
|
|
3647 else
|
|
3648 ++im_preedit_trailing;
|
|
3649 }
|
|
3650 if (!is_composing && i >= cursor_index)
|
|
3651 {
|
|
3652 /* This is essentially the same as im_preedit_trailing, except
|
|
3653 * composing characters are not counted even if p_deco is set. */
|
|
3654 ++num_move_back;
|
|
3655 }
|
|
3656 if (preedit_start_col != MAXCOL)
|
|
3657 preedit_end_col += utf_ptr2cells(p);
|
|
3658 }
|
|
3659
|
|
3660 if (p > str)
|
|
3661 {
|
|
3662 im_add_to_input(str, (int)(p - str));
|
|
3663 im_correct_cursor(num_move_back);
|
|
3664 }
|
|
3665
|
|
3666 g_free(preedit_string);
|
|
3667
|
|
3668 if (gtk_main_level() > 0)
|
|
3669 gtk_main_quit();
|
|
3670 }
|
|
3671
|
|
3672 /*
|
|
3673 * Translate the Pango attributes at iter to Vim highlighting attributes.
|
|
3674 * Ignore attributes not supported by Vim highlighting. This shouldn't have
|
|
3675 * too much impact -- right now we handle even more attributes than necessary
|
|
3676 * for the IM modules I tested with.
|
|
3677 */
|
|
3678 static int
|
|
3679 translate_pango_attributes(PangoAttrIterator *iter)
|
|
3680 {
|
|
3681 PangoAttribute *attr;
|
|
3682 int char_attr = HL_NORMAL;
|
|
3683
|
|
3684 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE);
|
|
3685 if (attr != NULL && ((PangoAttrInt *)attr)->value
|
|
3686 != (int)PANGO_UNDERLINE_NONE)
|
|
3687 char_attr |= HL_UNDERLINE;
|
|
3688
|
|
3689 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT);
|
|
3690 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD)
|
|
3691 char_attr |= HL_BOLD;
|
|
3692
|
|
3693 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE);
|
|
3694 if (attr != NULL && ((PangoAttrInt *)attr)->value
|
|
3695 != (int)PANGO_STYLE_NORMAL)
|
|
3696 char_attr |= HL_ITALIC;
|
|
3697
|
|
3698 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND);
|
|
3699 if (attr != NULL)
|
|
3700 {
|
|
3701 const PangoColor *color = &((PangoAttrColor *)attr)->color;
|
|
3702
|
|
3703 /* Assume inverse if black background is requested */
|
|
3704 if ((color->red | color->green | color->blue) == 0)
|
|
3705 char_attr |= HL_INVERSE;
|
|
3706 }
|
|
3707
|
|
3708 return char_attr;
|
|
3709 }
|
|
3710
|
|
3711 /*
|
|
3712 * Retrieve the highlighting attributes at column col in the preedit string.
|
|
3713 * Return -1 if not in preediting mode or if col is out of range.
|
|
3714 */
|
|
3715 int
|
|
3716 im_get_feedback_attr(int col)
|
|
3717 {
|
|
3718 char *preedit_string = NULL;
|
|
3719 PangoAttrList *attr_list = NULL;
|
|
3720 int char_attr = -1;
|
|
3721
|
|
3722 if (xic == NULL)
|
|
3723 return char_attr;
|
|
3724
|
|
3725 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL);
|
|
3726
|
|
3727 if (preedit_string != NULL && attr_list != NULL)
|
|
3728 {
|
|
3729 int index;
|
|
3730
|
|
3731 /* Get the byte index as used by PangoAttrIterator */
|
|
3732 for (index = 0; col > 0 && preedit_string[index] != '\0'; --col)
|
|
3733 index += utfc_ptr2len_check((char_u *)preedit_string + index);
|
|
3734
|
|
3735 if (preedit_string[index] != '\0')
|
|
3736 {
|
|
3737 PangoAttrIterator *iter;
|
|
3738 int start, end;
|
|
3739
|
|
3740 char_attr = HL_NORMAL;
|
|
3741 iter = pango_attr_list_get_iterator(attr_list);
|
|
3742
|
|
3743 /* Extract all relevant attributes from the list. */
|
|
3744 do
|
|
3745 {
|
|
3746 pango_attr_iterator_range(iter, &start, &end);
|
|
3747
|
|
3748 if (index >= start && index < end)
|
|
3749 char_attr |= translate_pango_attributes(iter);
|
|
3750 }
|
|
3751 while (pango_attr_iterator_next(iter));
|
|
3752
|
|
3753 pango_attr_iterator_destroy(iter);
|
|
3754 }
|
|
3755 }
|
|
3756
|
|
3757 if (attr_list != NULL)
|
|
3758 pango_attr_list_unref(attr_list);
|
|
3759 g_free(preedit_string);
|
|
3760
|
|
3761 return char_attr;
|
|
3762 }
|
|
3763
|
|
3764 void
|
|
3765 xim_init(void)
|
|
3766 {
|
|
3767 #ifdef XIM_DEBUG
|
|
3768 xim_log("xim_init()\n");
|
|
3769 #endif
|
|
3770
|
|
3771 g_return_if_fail(gui.drawarea != NULL);
|
|
3772 g_return_if_fail(gui.drawarea->window != NULL);
|
|
3773
|
|
3774 xic = gtk_im_multicontext_new();
|
|
3775 g_object_ref(xic);
|
|
3776
|
|
3777 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit",
|
|
3778 G_CALLBACK(&im_commit_cb), NULL);
|
|
3779 g_signal_connect(G_OBJECT(xic), "preedit_changed",
|
|
3780 G_CALLBACK(&im_preedit_changed_cb), NULL);
|
|
3781 g_signal_connect(G_OBJECT(xic), "preedit_start",
|
|
3782 G_CALLBACK(&im_preedit_start_cb), NULL);
|
|
3783 g_signal_connect(G_OBJECT(xic), "preedit_end",
|
|
3784 G_CALLBACK(&im_preedit_end_cb), NULL);
|
|
3785
|
|
3786 gtk_im_context_set_client_window(xic, gui.drawarea->window);
|
|
3787 }
|
|
3788
|
|
3789 void
|
|
3790 im_shutdown(void)
|
|
3791 {
|
|
3792 #ifdef XIM_DEBUG
|
|
3793 xim_log("im_shutdown()\n");
|
|
3794 #endif
|
|
3795
|
|
3796 if (xic != NULL)
|
|
3797 {
|
|
3798 gtk_im_context_focus_out(xic);
|
|
3799 g_object_unref(xic);
|
|
3800 xic = NULL;
|
|
3801 }
|
|
3802 im_is_active = FALSE;
|
|
3803 im_commit_handler_id = 0;
|
|
3804 preedit_start_col = MAXCOL;
|
|
3805 xim_has_preediting = FALSE;
|
|
3806 }
|
|
3807
|
|
3808 /*
|
|
3809 * Convert the string argument to keyval and state for GdkEventKey.
|
|
3810 * If str is valid return TRUE, otherwise FALSE.
|
|
3811 *
|
|
3812 * See 'imactivatekey' for documentation of the format.
|
|
3813 */
|
|
3814 static int
|
|
3815 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state)
|
|
3816 {
|
|
3817 const char *mods_end;
|
|
3818 unsigned tmp_keyval;
|
|
3819 unsigned tmp_state = 0;
|
|
3820
|
|
3821 mods_end = strrchr(str, '-');
|
|
3822 mods_end = (mods_end != NULL) ? mods_end + 1 : str;
|
|
3823
|
|
3824 /* Parse modifier keys */
|
|
3825 while (str < mods_end)
|
|
3826 switch (*str++)
|
|
3827 {
|
|
3828 case '-': break;
|
|
3829 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break;
|
|
3830 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break;
|
|
3831 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break;
|
|
3832 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break;
|
|
3833 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break;
|
|
3834 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break;
|
|
3835 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break;
|
|
3836 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break;
|
|
3837 default:
|
|
3838 return FALSE;
|
|
3839 }
|
|
3840
|
|
3841 tmp_keyval = gdk_keyval_from_name(str);
|
|
3842
|
|
3843 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol)
|
|
3844 return FALSE;
|
|
3845
|
|
3846 if (keyval != NULL)
|
|
3847 *keyval = tmp_keyval;
|
|
3848 if (state != NULL)
|
|
3849 *state = tmp_state;
|
|
3850
|
|
3851 return TRUE;
|
|
3852 }
|
|
3853
|
|
3854 /*
|
|
3855 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an
|
|
3856 * empty string is also regarded as valid.
|
|
3857 *
|
|
3858 * Note: The numerical key value of p_imak is cached if it was valid; thus
|
|
3859 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever
|
|
3860 * 'imak' changes. This is currently the case but not obvious -- should
|
|
3861 * probably rename the function for clarity.
|
|
3862 */
|
|
3863 int
|
|
3864 im_xim_isvalid_imactivate(void)
|
|
3865 {
|
|
3866 if (p_imak[0] == NUL)
|
|
3867 {
|
|
3868 im_activatekey_keyval = GDK_VoidSymbol;
|
|
3869 im_activatekey_state = 0;
|
|
3870 return TRUE;
|
|
3871 }
|
|
3872
|
|
3873 return im_string_to_keyval((const char *)p_imak,
|
|
3874 &im_activatekey_keyval,
|
|
3875 &im_activatekey_state);
|
|
3876 }
|
|
3877
|
|
3878 static void
|
|
3879 im_synthesize_keypress(unsigned int keyval, unsigned int state)
|
|
3880 {
|
|
3881 GdkEventKey *event;
|
|
3882
|
|
3883 # ifdef HAVE_GTK_MULTIHEAD
|
|
3884 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS);
|
|
3885 g_object_ref(gui.drawarea->window); /* unreffed by gdk_event_free() */
|
|
3886 # else
|
|
3887 event = (GdkEventKey *)g_malloc0((gulong)sizeof(GdkEvent));
|
|
3888 event->type = GDK_KEY_PRESS;
|
|
3889 # endif
|
|
3890 event->window = gui.drawarea->window;
|
|
3891 event->send_event = TRUE;
|
|
3892 event->time = GDK_CURRENT_TIME;
|
|
3893 event->state = state;
|
|
3894 event->keyval = keyval;
|
|
3895 event->hardware_keycode = /* needed for XIM */
|
|
3896 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval);
|
|
3897 event->length = 0;
|
|
3898 event->string = NULL;
|
|
3899
|
|
3900 gtk_im_context_filter_keypress(xic, event);
|
|
3901
|
|
3902 /* For consistency, also send the corresponding release event. */
|
|
3903 event->type = GDK_KEY_RELEASE;
|
|
3904 event->send_event = FALSE;
|
|
3905 gtk_im_context_filter_keypress(xic, event);
|
|
3906
|
|
3907 # ifdef HAVE_GTK_MULTIHEAD
|
|
3908 gdk_event_free((GdkEvent *)event);
|
|
3909 # else
|
|
3910 g_free(event);
|
|
3911 # endif
|
|
3912 }
|
|
3913
|
|
3914 void
|
|
3915 xim_reset(void)
|
|
3916 {
|
|
3917 if (xic != NULL)
|
|
3918 {
|
|
3919 /*
|
|
3920 * The third-party imhangul module (and maybe others too) ignores
|
|
3921 * gtk_im_context_reset() or at least doesn't reset the active state.
|
|
3922 * Thus sending imactivatekey would turn it off if it was on before,
|
|
3923 * which is clearly not what we want. Fortunately we can work around
|
|
3924 * that for imhangul by sending GDK_Escape, but I don't know if it
|
|
3925 * works with all IM modules that support an activation key :/
|
|
3926 *
|
|
3927 * An alternative approach would be to destroy the IM context and
|
|
3928 * recreate it. But that means loading/unloading the IM module on
|
|
3929 * every mode switch, which causes a quite noticable delay even on
|
|
3930 * my rather fast box...
|
|
3931 * *
|
|
3932 * Moreover, there are some XIM which cannot respond to
|
|
3933 * im_synthesize_keypress(). we hope that they reset by
|
|
3934 * xim_shutdown().
|
|
3935 */
|
|
3936 if (im_activatekey_keyval != GDK_VoidSymbol && im_is_active)
|
|
3937 im_synthesize_keypress(GDK_Escape, 0U);
|
|
3938
|
|
3939 gtk_im_context_reset(xic);
|
|
3940
|
|
3941 /*
|
|
3942 * HACK for Ami: This sequence of function calls makes Ami handle
|
|
3943 * the IM reset gratiously, without breaking loads of other stuff.
|
|
3944 * It seems to force English mode as well, which is exactly what we
|
|
3945 * want because it makes the Ami status display work reliably.
|
|
3946 */
|
|
3947 gtk_im_context_set_use_preedit(xic, FALSE);
|
|
3948
|
|
3949 if (p_imdisable)
|
|
3950 im_shutdown();
|
|
3951 else
|
|
3952 {
|
|
3953 gtk_im_context_set_use_preedit(xic, TRUE);
|
|
3954 xim_set_focus(gui.in_focus);
|
|
3955
|
|
3956 if (im_activatekey_keyval != GDK_VoidSymbol)
|
|
3957 {
|
|
3958 if (im_is_active)
|
|
3959 {
|
|
3960 g_signal_handler_block(xic, im_commit_handler_id);
|
|
3961 im_synthesize_keypress(im_activatekey_keyval,
|
|
3962 im_activatekey_state);
|
|
3963 g_signal_handler_unblock(xic, im_commit_handler_id);
|
|
3964 }
|
|
3965 }
|
|
3966 else
|
|
3967 {
|
|
3968 im_shutdown();
|
|
3969 xim_init();
|
|
3970 xim_set_focus(gui.in_focus);
|
|
3971 }
|
|
3972 }
|
|
3973 }
|
|
3974
|
|
3975 preedit_start_col = MAXCOL;
|
|
3976 xim_has_preediting = FALSE;
|
|
3977 }
|
|
3978
|
|
3979 int
|
|
3980 xim_queue_key_press_event(GdkEventKey *event, int down)
|
|
3981 {
|
|
3982 if (down)
|
|
3983 {
|
|
3984 /*
|
|
3985 * Workaround GTK2 XIM 'feature' that always converts keypad keys to
|
|
3986 * chars., even when not part of an IM sequence (ref. feature of
|
|
3987 * gdk/gdkkeyuni.c).
|
|
3988 * Flag any keypad keys that might represent a single char.
|
|
3989 * If this (on its own - i.e., not part of an IM sequence) is
|
|
3990 * committed while we're processing one of these keys, we can ignore
|
|
3991 * that commit and go ahead & process it ourselves. That way we can
|
|
3992 * still distinguish keypad keys for use in mappings.
|
|
3993 */
|
|
3994 switch (event->keyval)
|
|
3995 {
|
|
3996 case GDK_KP_Add: xim_expected_char = '+'; break;
|
|
3997 case GDK_KP_Subtract: xim_expected_char = '-'; break;
|
|
3998 case GDK_KP_Divide: xim_expected_char = '/'; break;
|
|
3999 case GDK_KP_Multiply: xim_expected_char = '*'; break;
|
|
4000 case GDK_KP_Decimal: xim_expected_char = '.'; break;
|
|
4001 case GDK_KP_Equal: xim_expected_char = '='; break;
|
|
4002 case GDK_KP_0: xim_expected_char = '0'; break;
|
|
4003 case GDK_KP_1: xim_expected_char = '1'; break;
|
|
4004 case GDK_KP_2: xim_expected_char = '2'; break;
|
|
4005 case GDK_KP_3: xim_expected_char = '3'; break;
|
|
4006 case GDK_KP_4: xim_expected_char = '4'; break;
|
|
4007 case GDK_KP_5: xim_expected_char = '5'; break;
|
|
4008 case GDK_KP_6: xim_expected_char = '6'; break;
|
|
4009 case GDK_KP_7: xim_expected_char = '7'; break;
|
|
4010 case GDK_KP_8: xim_expected_char = '8'; break;
|
|
4011 case GDK_KP_9: xim_expected_char = '9'; break;
|
|
4012 default: xim_expected_char = NUL;
|
|
4013 }
|
|
4014 xim_ignored_char = FALSE;
|
|
4015 }
|
|
4016
|
|
4017 /*
|
|
4018 * When typing fFtT, XIM may be activated. Thus it must pass
|
|
4019 * gtk_im_context_filter_keypress() in Normal mode.
|
|
4020 * And while doing :sh too.
|
|
4021 */
|
|
4022 if (xic != NULL && !p_imdisable
|
|
4023 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0)
|
|
4024 {
|
|
4025 /*
|
|
4026 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is
|
|
4027 * always aware of the current status of IM, and can even emulate
|
|
4028 * the activation key for modules that don't support one.
|
|
4029 */
|
|
4030 if (event->keyval == im_activatekey_keyval
|
|
4031 && (event->state & im_activatekey_state) == im_activatekey_state)
|
|
4032 {
|
|
4033 unsigned int state_mask;
|
|
4034
|
|
4035 /* Require the state of the 3 most used modifiers to match exactly.
|
|
4036 * Otherwise e.g. <S-C-space> would be unusable for other purposes
|
|
4037 * if the IM activate key is <S-space>. */
|
|
4038 state_mask = im_activatekey_state;
|
|
4039 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK
|
|
4040 | (int)GDK_MOD1_MASK);
|
|
4041
|
|
4042 if ((event->state & state_mask) != im_activatekey_state)
|
|
4043 return FALSE;
|
|
4044
|
|
4045 /* Don't send it a second time on GDK_KEY_RELEASE. */
|
|
4046 if (event->type != GDK_KEY_PRESS)
|
|
4047 return TRUE;
|
|
4048
|
|
4049 if (map_to_exists_mode((char_u *)"", LANGMAP))
|
|
4050 {
|
|
4051 im_set_active(FALSE);
|
|
4052
|
|
4053 /* ":lmap" mappings exists, toggle use of mappings. */
|
|
4054 State ^= LANGMAP;
|
|
4055 if (State & LANGMAP)
|
|
4056 {
|
|
4057 curbuf->b_p_iminsert = B_IMODE_NONE;
|
|
4058 State &= ~LANGMAP;
|
|
4059 }
|
|
4060 else
|
|
4061 {
|
|
4062 curbuf->b_p_iminsert = B_IMODE_LMAP;
|
|
4063 State |= LANGMAP;
|
|
4064 }
|
|
4065 return TRUE;
|
|
4066 }
|
|
4067
|
|
4068 return gtk_im_context_filter_keypress(xic, event);
|
|
4069 }
|
|
4070
|
|
4071 /* Don't filter events through the IM context if IM isn't active
|
|
4072 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module
|
|
4073 * not doing anything before the activation key was sent. */
|
|
4074 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active)
|
|
4075 {
|
|
4076 int imresult = gtk_im_context_filter_keypress(xic, event);
|
|
4077
|
|
4078 /* Some XIM send following sequence:
|
|
4079 * 1. preedited string.
|
|
4080 * 2. committed string.
|
|
4081 * 3. line changed key.
|
|
4082 * 4. preedited string.
|
|
4083 * 5. remove preedited string.
|
|
4084 * if 3, Vim can't move back the above line for 5.
|
|
4085 * thus, this part should not parse the key. */
|
|
4086 if (!imresult && preedit_start_col != MAXCOL
|
|
4087 && event->keyval == GDK_Return)
|
|
4088 {
|
|
4089 im_synthesize_keypress(GDK_Return, 0U);
|
|
4090 return FALSE;
|
|
4091 }
|
|
4092
|
|
4093 /* If XIM tried to commit a keypad key as a single char.,
|
|
4094 * ignore it so we can use the keypad key 'raw', for mappings. */
|
|
4095 if (xim_expected_char != NUL && xim_ignored_char)
|
|
4096 /* We had a keypad key, and XIM tried to thieve it */
|
|
4097 return FALSE;
|
|
4098
|
|
4099 /* Normal processing */
|
|
4100 return imresult;
|
|
4101 }
|
|
4102 }
|
|
4103
|
|
4104 return FALSE;
|
|
4105 }
|
|
4106
|
|
4107 int
|
|
4108 im_get_status(void)
|
|
4109 {
|
|
4110 return im_is_active;
|
|
4111 }
|
|
4112
|
|
4113 # else /* !HAVE_GTK2 */
|
|
4114
|
|
4115 static int xim_is_active = FALSE; /* XIM should be active in the current
|
|
4116 mode */
|
|
4117 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */
|
|
4118 #ifdef FEAT_GUI_X11
|
|
4119 static XIMStyle input_style;
|
|
4120 static int status_area_enabled = TRUE;
|
|
4121 #endif
|
|
4122
|
|
4123 #ifdef FEAT_GUI_GTK
|
|
4124 # ifdef WIN3264
|
|
4125 # include <gdk/gdkwin32.h>
|
|
4126 # else
|
|
4127 # include <gdk/gdkx.h>
|
|
4128 # endif
|
|
4129 #else
|
|
4130 # ifdef PROTO
|
|
4131 /* Define a few things to be able to generate prototypes while not configured
|
|
4132 * for GTK. */
|
|
4133 # define GSList int
|
|
4134 # define gboolean int
|
|
4135 typedef int GdkEvent;
|
|
4136 typedef int GdkEventKey;
|
|
4137 # define GdkIC int
|
|
4138 # endif
|
|
4139 #endif
|
|
4140
|
11
|
4141 #if defined(FEAT_GUI_GTK) || defined(PROTO) || defined(FEAT_GUI_KDE)
|
7
|
4142 static int preedit_buf_len = 0;
|
|
4143 static int xim_can_preediting INIT(= FALSE); /* XIM in showmode() */
|
|
4144 static int xim_input_style;
|
|
4145 #ifndef FEAT_GUI_GTK
|
|
4146 # define gboolean int
|
|
4147 #endif
|
|
4148 static gboolean use_status_area = 0;
|
|
4149
|
|
4150 static int im_xim_str2keycode __ARGS((unsigned int *code, unsigned int *state));
|
|
4151 static void im_xim_send_event_imactivate __ARGS((void));
|
|
4152
|
|
4153 /*
|
|
4154 * Convert string to keycode and state for XKeyEvent.
|
|
4155 * When string is valid return OK, when invalid return FAIL.
|
|
4156 *
|
|
4157 * See 'imactivatekey' documentation for the format.
|
|
4158 */
|
|
4159 static int
|
|
4160 im_xim_str2keycode(code, state)
|
|
4161 unsigned int *code;
|
|
4162 unsigned int *state;
|
|
4163 {
|
|
4164 int retval = OK;
|
|
4165 int len;
|
|
4166 unsigned keycode = 0, keystate = 0;
|
|
4167 Window window;
|
|
4168 Display *display;
|
|
4169 char_u *flag_end;
|
|
4170 char_u *str;
|
|
4171
|
|
4172 if (*p_imak != NUL)
|
|
4173 {
|
|
4174 len = STRLEN(p_imak);
|
|
4175 for (flag_end = p_imak + len - 1;
|
|
4176 flag_end > p_imak && *flag_end != '-'; --flag_end)
|
|
4177 ;
|
|
4178
|
|
4179 /* Parse modifier keys */
|
|
4180 for (str = p_imak; str < flag_end; ++str)
|
|
4181 {
|
|
4182 switch (*str)
|
|
4183 {
|
|
4184 case 's': case 'S':
|
|
4185 keystate |= ShiftMask;
|
|
4186 break;
|
|
4187 case 'l': case 'L':
|
|
4188 keystate |= LockMask;
|
|
4189 break;
|
|
4190 case 'c': case 'C':
|
|
4191 keystate |= ControlMask;
|
|
4192 break;
|
|
4193 case '1':
|
|
4194 keystate |= Mod1Mask;
|
|
4195 break;
|
|
4196 case '2':
|
|
4197 keystate |= Mod2Mask;
|
|
4198 break;
|
|
4199 case '3':
|
|
4200 keystate |= Mod3Mask;
|
|
4201 break;
|
|
4202 case '4':
|
|
4203 keystate |= Mod4Mask;
|
|
4204 break;
|
|
4205 case '5':
|
|
4206 keystate |= Mod5Mask;
|
|
4207 break;
|
|
4208 case '-':
|
|
4209 break;
|
|
4210 default:
|
|
4211 retval = FAIL;
|
|
4212 }
|
|
4213 }
|
|
4214 if (*str == '-')
|
|
4215 ++str;
|
|
4216
|
|
4217 /* Get keycode from string. */
|
|
4218 gui_get_x11_windis(&window, &display);
|
|
4219 if (display)
|
|
4220 keycode = XKeysymToKeycode(display, XStringToKeysym((char *)str));
|
|
4221 if (keycode == 0)
|
|
4222 retval = FAIL;
|
|
4223
|
|
4224 if (code != NULL)
|
|
4225 *code = keycode;
|
|
4226 if (state != NULL)
|
|
4227 *state = keystate;
|
|
4228 }
|
|
4229 return retval;
|
|
4230 }
|
|
4231
|
|
4232 static void
|
|
4233 im_xim_send_event_imactivate()
|
|
4234 {
|
|
4235 /* Force turn on preedit state by symulate keypress event.
|
|
4236 * Keycode and state is specified by 'imactivatekey'.
|
|
4237 */
|
|
4238 XKeyEvent ev;
|
|
4239
|
|
4240 gui_get_x11_windis(&ev.window, &ev.display);
|
|
4241 ev.root = RootWindow(ev.display, DefaultScreen(ev.display));
|
|
4242 ev.subwindow = None;
|
|
4243 ev.time = CurrentTime;
|
|
4244 ev.x = 1;
|
|
4245 ev.y = 1;
|
|
4246 ev.x_root = 1;
|
|
4247 ev.y_root = 1;
|
|
4248 ev.same_screen = 1;
|
|
4249 ev.type = KeyPress;
|
|
4250 if (im_xim_str2keycode(&ev.keycode, &ev.state) == OK)
|
|
4251 XSendEvent(ev.display, ev.window, 1, KeyPressMask, (XEvent*)&ev);
|
|
4252 }
|
|
4253
|
|
4254 /*
|
|
4255 * Return TRUE if 'imactivatekey' has a valid value.
|
|
4256 */
|
|
4257 int
|
|
4258 im_xim_isvalid_imactivate()
|
|
4259 {
|
|
4260 return im_xim_str2keycode(NULL, NULL) == OK;
|
|
4261 }
|
|
4262 #endif /* FEAT_GUI_GTK */
|
|
4263
|
|
4264 /*
|
|
4265 * Switch using XIM on/off. This is used by the code that changes "State".
|
|
4266 */
|
|
4267 void
|
|
4268 im_set_active(active)
|
|
4269 int active;
|
|
4270 {
|
|
4271 if (xic == NULL)
|
|
4272 return;
|
|
4273
|
|
4274 /* If 'imdisable' is set, XIM is never active. */
|
|
4275 if (p_imdisable)
|
|
4276 active = FALSE;
|
11
|
4277 #if !defined (FEAT_GUI_GTK) && !defined (FEAT_GUI_KDE)
|
7
|
4278 else if (input_style & XIMPreeditPosition)
|
|
4279 /* There is a problem in switching XIM off when preediting is used,
|
|
4280 * and it is not clear how this can be solved. For now, keep XIM on
|
|
4281 * all the time, like it was done in Vim 5.8. */
|
|
4282 active = TRUE;
|
|
4283 #endif
|
|
4284
|
|
4285 /* Remember the active state, it is needed when Vim gets keyboard focus. */
|
|
4286 xim_is_active = active;
|
|
4287
|
|
4288 #ifdef FEAT_GUI_GTK
|
|
4289 /* When 'imactivatekey' has valid key-string, try to control XIM preedit
|
|
4290 * state. When 'imactivatekey' has no or invalid string, try old XIM
|
|
4291 * focus control.
|
|
4292 */
|
|
4293 if (*p_imak != NUL)
|
|
4294 {
|
|
4295 /* BASIC STRATEGY:
|
|
4296 * Destroy old Input Context (XIC), and create new one. New XIC
|
|
4297 * would have a state of preedit that is off. When argument:active
|
|
4298 * is false, that's all. Else argument:active is true, send a key
|
|
4299 * event specified by 'imactivatekey' to activate XIM preedit state.
|
|
4300 */
|
|
4301
|
|
4302 xim_is_active = TRUE; /* Disable old XIM focus control */
|
|
4303 /* If we can monitor preedit state with preedit callback functions,
|
|
4304 * try least creation of new XIC.
|
|
4305 */
|
|
4306 if (xim_input_style & (int)GDK_IM_PREEDIT_CALLBACKS)
|
|
4307 {
|
|
4308 if (xim_can_preediting && !active)
|
|
4309 {
|
|
4310 /* Force turn off preedit state. With some IM
|
|
4311 * implementations, we cannot turn off preedit state by
|
|
4312 * symulate keypress event. It is why using such a method
|
|
4313 * that destroy old IC (input context), and create new one.
|
|
4314 * When create new IC, its preedit state is usually off.
|
|
4315 */
|
|
4316 xim_reset();
|
|
4317 xim_set_focus(FALSE);
|
|
4318 gdk_ic_destroy(xic);
|
|
4319 xim_init();
|
|
4320 xim_can_preediting = FALSE;
|
|
4321 }
|
|
4322 else if (!xim_can_preediting && active)
|
|
4323 im_xim_send_event_imactivate();
|
|
4324 }
|
|
4325 else
|
|
4326 {
|
|
4327 /* First, force destroy old IC, and create new one. It
|
|
4328 * symulates "turning off preedit state".
|
|
4329 */
|
|
4330 xim_set_focus(FALSE);
|
|
4331 gdk_ic_destroy(xic);
|
|
4332 xim_init();
|
|
4333 xim_can_preediting = FALSE;
|
|
4334
|
|
4335 /* 2nd, when requested to activate IM, symulate this by sending
|
|
4336 * the event.
|
|
4337 */
|
|
4338 if (active)
|
|
4339 {
|
|
4340 im_xim_send_event_imactivate();
|
|
4341 xim_can_preediting = TRUE;
|
|
4342 }
|
|
4343 }
|
|
4344 }
|
|
4345 else
|
|
4346 {
|
|
4347 # ifndef XIMPreeditUnKnown
|
|
4348 /* X11R5 doesn't have these, it looks safe enough to define here. */
|
|
4349 typedef unsigned long XIMPreeditState;
|
|
4350 # define XIMPreeditUnKnown 0L
|
|
4351 # define XIMPreeditEnable 1L
|
|
4352 # define XIMPreeditDisable (1L<<1)
|
|
4353 # define XNPreeditState "preeditState"
|
|
4354 # endif
|
|
4355 XIMPreeditState preedit_state = XIMPreeditUnKnown;
|
|
4356 XVaNestedList preedit_attr;
|
|
4357 XIC pxic;
|
|
4358
|
|
4359 preedit_attr = XVaCreateNestedList(0,
|
|
4360 XNPreeditState, &preedit_state,
|
|
4361 NULL);
|
|
4362 pxic = ((GdkICPrivate *)xic)->xic;
|
|
4363
|
|
4364 if (!XGetICValues(pxic, XNPreeditAttributes, preedit_attr, NULL))
|
|
4365 {
|
|
4366 XFree(preedit_attr);
|
|
4367 preedit_attr = XVaCreateNestedList(0,
|
|
4368 XNPreeditState,
|
|
4369 active ? XIMPreeditEnable : XIMPreeditDisable,
|
|
4370 NULL);
|
|
4371 XSetICValues(pxic, XNPreeditAttributes, preedit_attr, NULL);
|
|
4372 xim_can_preediting = active;
|
|
4373 xim_is_active = active;
|
|
4374 }
|
|
4375 XFree(preedit_attr);
|
|
4376 }
|
|
4377 if (xim_input_style & XIMPreeditCallbacks)
|
|
4378 {
|
|
4379 preedit_buf_len = 0;
|
|
4380 init_preedit_start_col();
|
|
4381 }
|
|
4382 #else
|
|
4383 # if 0
|
|
4384 /* When had tested kinput2 + canna + Athena GUI version with
|
|
4385 * 'imactivatekey' is "s-space", im_xim_send_event_imactivate() did not
|
|
4386 * work correctly. It just inserted one space. I don't know why we
|
|
4387 * couldn't switch state of XIM preediting. This is reason why these
|
|
4388 * codes are commented out.
|
|
4389 */
|
|
4390 /* First, force destroy old IC, and create new one. It symulates
|
|
4391 * "turning off preedit state".
|
|
4392 */
|
|
4393 xim_set_focus(FALSE);
|
|
4394 XDestroyIC(xic);
|
|
4395 xic = NULL;
|
|
4396 xim_init();
|
|
4397
|
|
4398 /* 2nd, when requested to activate IM, symulate this by sending the
|
|
4399 * event.
|
|
4400 */
|
|
4401 if (active)
|
|
4402 im_xim_send_event_imactivate();
|
|
4403 # endif
|
|
4404 #endif
|
|
4405 xim_set_preedit();
|
|
4406 }
|
|
4407
|
|
4408 /*
|
|
4409 * Adjust using XIM for gaining or losing keyboard focus. Also called when
|
|
4410 * "xim_is_active" changes.
|
|
4411 */
|
|
4412 void
|
|
4413 xim_set_focus(focus)
|
|
4414 int focus;
|
|
4415 {
|
|
4416 if (xic == NULL)
|
|
4417 return;
|
|
4418
|
|
4419 /*
|
|
4420 * XIM only gets focus when the Vim window has keyboard focus and XIM has
|
|
4421 * been set active for the current mode.
|
|
4422 */
|
|
4423 if (focus && xim_is_active)
|
|
4424 {
|
|
4425 if (!xim_has_focus)
|
|
4426 {
|
|
4427 xim_has_focus = TRUE;
|
|
4428 #ifdef FEAT_GUI_GTK
|
|
4429 gdk_im_begin(xic, gui.drawarea->window);
|
|
4430 #else
|
|
4431 XSetICFocus(xic);
|
|
4432 #endif
|
|
4433 }
|
|
4434 }
|
|
4435 else
|
|
4436 {
|
|
4437 if (xim_has_focus)
|
|
4438 {
|
|
4439 xim_has_focus = FALSE;
|
|
4440 #ifdef FEAT_GUI_GTK
|
|
4441 gdk_im_end();
|
|
4442 #else
|
|
4443 XUnsetICFocus(xic);
|
|
4444 #endif
|
|
4445 }
|
|
4446 }
|
|
4447 }
|
|
4448
|
42
|
4449 #ifndef FEAT_GUI_KDE
|
7
|
4450 /*ARGSUSED*/
|
|
4451 void
|
|
4452 im_set_position(row, col)
|
|
4453 int row;
|
|
4454 int col;
|
|
4455 {
|
|
4456 xim_set_preedit();
|
|
4457 }
|
42
|
4458 #endif
|
7
|
4459
|
|
4460 /*
|
|
4461 * Set the XIM to the current cursor position.
|
|
4462 */
|
|
4463 void
|
|
4464 xim_set_preedit()
|
|
4465 {
|
|
4466 if (xic == NULL)
|
|
4467 return;
|
|
4468
|
|
4469 xim_set_focus(TRUE);
|
|
4470
|
|
4471 #ifdef FEAT_GUI_GTK
|
|
4472 if (gdk_im_ready())
|
|
4473 {
|
|
4474 int attrmask;
|
|
4475 GdkICAttr *attr;
|
|
4476
|
|
4477 if (!xic_attr)
|
|
4478 return;
|
|
4479
|
|
4480 attr = xic_attr;
|
|
4481 attrmask = 0;
|
|
4482
|
|
4483 # ifdef FEAT_XFONTSET
|
|
4484 if ((xim_input_style & (int)GDK_IM_PREEDIT_POSITION)
|
|
4485 && gui.fontset != NOFONTSET
|
|
4486 && gui.fontset->type == GDK_FONT_FONTSET)
|
|
4487 {
|
|
4488 if (!xim_has_focus)
|
|
4489 {
|
|
4490 if (attr->spot_location.y >= 0)
|
|
4491 {
|
|
4492 attr->spot_location.x = 0;
|
|
4493 attr->spot_location.y = -100;
|
|
4494 attrmask |= (int)GDK_IC_SPOT_LOCATION;
|
|
4495 }
|
|
4496 }
|
|
4497 else
|
|
4498 {
|
|
4499 gint width, height;
|
|
4500
|
|
4501 if (attr->spot_location.x != TEXT_X(gui.col)
|
|
4502 || attr->spot_location.y != TEXT_Y(gui.row))
|
|
4503 {
|
|
4504 attr->spot_location.x = TEXT_X(gui.col);
|
|
4505 attr->spot_location.y = TEXT_Y(gui.row);
|
|
4506 attrmask |= (int)GDK_IC_SPOT_LOCATION;
|
|
4507 }
|
|
4508
|
|
4509 gdk_window_get_size(gui.drawarea->window, &width, &height);
|
|
4510 width -= 2 * gui.border_offset;
|
|
4511 height -= 2 * gui.border_offset;
|
|
4512 if (xim_input_style & (int)GDK_IM_STATUS_AREA)
|
|
4513 height -= gui.char_height;
|
|
4514 if (attr->preedit_area.width != width
|
|
4515 || attr->preedit_area.height != height)
|
|
4516 {
|
|
4517 attr->preedit_area.x = gui.border_offset;
|
|
4518 attr->preedit_area.y = gui.border_offset;
|
|
4519 attr->preedit_area.width = width;
|
|
4520 attr->preedit_area.height = height;
|
|
4521 attrmask |= (int)GDK_IC_PREEDIT_AREA;
|
|
4522 }
|
|
4523
|
|
4524 if (attr->preedit_fontset != gui.current_font)
|
|
4525 {
|
|
4526 attr->preedit_fontset = gui.current_font;
|
|
4527 attrmask |= (int)GDK_IC_PREEDIT_FONTSET;
|
|
4528 }
|
|
4529 }
|
|
4530 }
|
|
4531 # endif /* FEAT_XFONTSET */
|
|
4532
|
|
4533 if (xim_fg_color == INVALCOLOR)
|
|
4534 {
|
|
4535 xim_fg_color = gui.def_norm_pixel;
|
|
4536 xim_bg_color = gui.def_back_pixel;
|
|
4537 }
|
|
4538 if (attr->preedit_foreground.pixel != xim_fg_color)
|
|
4539 {
|
|
4540 attr->preedit_foreground.pixel = xim_fg_color;
|
|
4541 attrmask |= (int)GDK_IC_PREEDIT_FOREGROUND;
|
|
4542 }
|
|
4543 if (attr->preedit_background.pixel != xim_bg_color)
|
|
4544 {
|
|
4545 attr->preedit_background.pixel = xim_bg_color;
|
|
4546 attrmask |= (int)GDK_IC_PREEDIT_BACKGROUND;
|
|
4547 }
|
|
4548
|
|
4549 if (attrmask != 0)
|
|
4550 gdk_ic_set_attr(xic, attr, (GdkICAttributesType)attrmask);
|
|
4551 }
|
|
4552 #else /* FEAT_GUI_GTK */
|
11
|
4553 # ifdef FEAT_GUI_KDE
|
|
4554 # else
|
7
|
4555 {
|
|
4556 XVaNestedList attr_list;
|
|
4557 XRectangle spot_area;
|
|
4558 XPoint over_spot;
|
|
4559 int line_space;
|
|
4560
|
|
4561 if (!xim_has_focus)
|
|
4562 {
|
|
4563 /* hide XIM cursor */
|
|
4564 over_spot.x = 0;
|
|
4565 over_spot.y = -100; /* arbitrary invisible position */
|
|
4566 attr_list = (XVaNestedList) XVaCreateNestedList(0,
|
|
4567 XNSpotLocation,
|
|
4568 &over_spot,
|
|
4569 NULL);
|
|
4570 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL);
|
|
4571 XFree(attr_list);
|
|
4572 return;
|
|
4573 }
|
|
4574
|
|
4575 if (input_style & XIMPreeditPosition)
|
|
4576 {
|
|
4577 if (xim_fg_color == INVALCOLOR)
|
|
4578 {
|
|
4579 xim_fg_color = gui.def_norm_pixel;
|
|
4580 xim_bg_color = gui.def_back_pixel;
|
|
4581 }
|
|
4582 over_spot.x = TEXT_X(gui.col);
|
|
4583 over_spot.y = TEXT_Y(gui.row);
|
|
4584 spot_area.x = 0;
|
|
4585 spot_area.y = 0;
|
|
4586 spot_area.height = gui.char_height * Rows;
|
|
4587 spot_area.width = gui.char_width * Columns;
|
|
4588 line_space = gui.char_height;
|
|
4589 attr_list = (XVaNestedList) XVaCreateNestedList(0,
|
|
4590 XNSpotLocation, &over_spot,
|
|
4591 XNForeground, (Pixel) xim_fg_color,
|
|
4592 XNBackground, (Pixel) xim_bg_color,
|
|
4593 XNArea, &spot_area,
|
|
4594 XNLineSpace, line_space,
|
|
4595 NULL);
|
|
4596 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL))
|
|
4597 EMSG(_("E284: Cannot set IC values"));
|
|
4598 XFree(attr_list);
|
|
4599 }
|
|
4600 }
|
11
|
4601 # endif /* FEAT_GUI_KDE */
|
7
|
4602 #endif /* FEAT_GUI_GTK */
|
|
4603 }
|
|
4604
|
|
4605 /*
|
|
4606 * Set up the status area.
|
|
4607 *
|
|
4608 * This should use a separate Widget, but that seems not possible, because
|
|
4609 * preedit_area and status_area should be set to the same window as for the
|
|
4610 * text input. Unfortunately this means the status area pollutes the text
|
|
4611 * window...
|
|
4612 */
|
|
4613 void
|
|
4614 xim_set_status_area()
|
|
4615 {
|
|
4616 if (xic == NULL)
|
|
4617 return;
|
|
4618
|
|
4619 #ifdef FEAT_GUI_GTK
|
|
4620 # if defined(FEAT_XFONTSET)
|
|
4621 if (use_status_area)
|
|
4622 {
|
|
4623 GdkICAttr *attr;
|
|
4624 int style;
|
|
4625 gint width, height;
|
|
4626 GtkWidget *widget;
|
|
4627 int attrmask;
|
|
4628
|
|
4629 if (!xic_attr)
|
|
4630 return;
|
|
4631
|
|
4632 attr = xic_attr;
|
|
4633 attrmask = 0;
|
|
4634 style = (int)gdk_ic_get_style(xic);
|
|
4635 if ((style & (int)GDK_IM_STATUS_MASK) == (int)GDK_IM_STATUS_AREA)
|
|
4636 {
|
|
4637 if (gui.fontset != NOFONTSET
|
|
4638 && gui.fontset->type == GDK_FONT_FONTSET)
|
|
4639 {
|
|
4640 widget = gui.mainwin;
|
|
4641 gdk_window_get_size(widget->window, &width, &height);
|
|
4642
|
|
4643 attrmask |= (int)GDK_IC_STATUS_AREA;
|
|
4644 attr->status_area.x = 0;
|
|
4645 attr->status_area.y = height - gui.char_height - 1;
|
|
4646 attr->status_area.width = width;
|
|
4647 attr->status_area.height = gui.char_height;
|
|
4648 }
|
|
4649 }
|
|
4650 if (attrmask != 0)
|
|
4651 gdk_ic_set_attr(xic, attr, (GdkICAttributesType)attrmask);
|
|
4652 }
|
|
4653 # endif
|
|
4654 #else
|
11
|
4655 # ifdef FEAT_GUI_KDE
|
|
4656 # else
|
7
|
4657 {
|
|
4658 XVaNestedList preedit_list = 0, status_list = 0, list = 0;
|
|
4659 XRectangle pre_area, status_area;
|
|
4660
|
|
4661 if (input_style & XIMStatusArea)
|
|
4662 {
|
|
4663 if (input_style & XIMPreeditArea)
|
|
4664 {
|
|
4665 XRectangle *needed_rect;
|
|
4666
|
|
4667 /* to get status_area width */
|
|
4668 status_list = XVaCreateNestedList(0, XNAreaNeeded,
|
|
4669 &needed_rect, NULL);
|
|
4670 XGetICValues(xic, XNStatusAttributes, status_list, NULL);
|
|
4671 XFree(status_list);
|
|
4672
|
|
4673 status_area.width = needed_rect->width;
|
|
4674 }
|
|
4675 else
|
|
4676 status_area.width = gui.char_width * Columns;
|
|
4677
|
|
4678 status_area.x = 0;
|
|
4679 status_area.y = gui.char_height * Rows + gui.border_offset;
|
|
4680 if (gui.which_scrollbars[SBAR_BOTTOM])
|
|
4681 status_area.y += gui.scrollbar_height;
|
|
4682 #ifdef FEAT_MENU
|
|
4683 if (gui.menu_is_active)
|
|
4684 status_area.y += gui.menu_height;
|
|
4685 #endif
|
|
4686 status_area.height = gui.char_height;
|
|
4687 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL);
|
|
4688 }
|
|
4689 else
|
|
4690 {
|
|
4691 status_area.x = 0;
|
|
4692 status_area.y = gui.char_height * Rows + gui.border_offset;
|
|
4693 if (gui.which_scrollbars[SBAR_BOTTOM])
|
|
4694 status_area.y += gui.scrollbar_height;
|
|
4695 #ifdef FEAT_MENU
|
|
4696 if (gui.menu_is_active)
|
|
4697 status_area.y += gui.menu_height;
|
|
4698 #endif
|
|
4699 status_area.width = 0;
|
|
4700 status_area.height = gui.char_height;
|
|
4701 }
|
|
4702
|
|
4703 if (input_style & XIMPreeditArea) /* off-the-spot */
|
|
4704 {
|
|
4705 pre_area.x = status_area.x + status_area.width;
|
|
4706 pre_area.y = gui.char_height * Rows + gui.border_offset;
|
|
4707 pre_area.width = gui.char_width * Columns - pre_area.x;
|
|
4708 if (gui.which_scrollbars[SBAR_BOTTOM])
|
|
4709 pre_area.y += gui.scrollbar_height;
|
|
4710 #ifdef FEAT_MENU
|
|
4711 if (gui.menu_is_active)
|
|
4712 pre_area.y += gui.menu_height;
|
|
4713 #endif
|
|
4714 pre_area.height = gui.char_height;
|
|
4715 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL);
|
|
4716 }
|
|
4717 else if (input_style & XIMPreeditPosition) /* over-the-spot */
|
|
4718 {
|
|
4719 pre_area.x = 0;
|
|
4720 pre_area.y = 0;
|
|
4721 pre_area.height = gui.char_height * Rows;
|
|
4722 pre_area.width = gui.char_width * Columns;
|
|
4723 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL);
|
|
4724 }
|
|
4725
|
|
4726 if (preedit_list && status_list)
|
|
4727 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list,
|
|
4728 XNStatusAttributes, status_list, NULL);
|
|
4729 else if (preedit_list)
|
|
4730 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list,
|
|
4731 NULL);
|
|
4732 else if (status_list)
|
|
4733 list = XVaCreateNestedList(0, XNStatusAttributes, status_list,
|
|
4734 NULL);
|
|
4735 else
|
|
4736 list = NULL;
|
|
4737
|
|
4738 if (list)
|
|
4739 {
|
|
4740 XSetICValues(xic, XNVaNestedList, list, NULL);
|
|
4741 XFree(list);
|
|
4742 }
|
|
4743 if (status_list)
|
|
4744 XFree(status_list);
|
|
4745 if (preedit_list)
|
|
4746 XFree(preedit_list);
|
|
4747 }
|
11
|
4748 # endif /* FEAT_GUI_KDE */
|
7
|
4749 #endif
|
|
4750 }
|
|
4751
|
42
|
4752 #if defined(FEAT_GUI_X11) || defined(FEAT_GUI_GTK) || defined(FEAT_GUI_KDE)
|
7
|
4753 static char e_xim[] = N_("E285: Failed to create input context");
|
|
4754 #endif
|
|
4755
|
|
4756 #if defined(FEAT_GUI_X11) || defined(PROTO)
|
|
4757 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(sun)
|
|
4758 # define USE_X11R6_XIM
|
|
4759 # endif
|
|
4760
|
|
4761 static int xim_real_init __ARGS((Window x11_window, Display *x11_display));
|
|
4762
|
|
4763
|
|
4764 #ifdef USE_X11R6_XIM
|
|
4765 static void xim_instantiate_cb __ARGS((Display *display, XPointer client_data, XPointer call_data));
|
|
4766 static void xim_destroy_cb __ARGS((XIM im, XPointer client_data, XPointer call_data));
|
|
4767
|
|
4768 /*ARGSUSED*/
|
|
4769 static void
|
|
4770 xim_instantiate_cb(display, client_data, call_data)
|
|
4771 Display *display;
|
|
4772 XPointer client_data;
|
|
4773 XPointer call_data;
|
|
4774 {
|
|
4775 Window x11_window;
|
|
4776 Display *x11_display;
|
|
4777
|
|
4778 #ifdef XIM_DEBUG
|
|
4779 xim_log("xim_instantiate_cb()\n");
|
|
4780 #endif
|
|
4781
|
|
4782 gui_get_x11_windis(&x11_window, &x11_display);
|
|
4783 if (display != x11_display)
|
|
4784 return;
|
|
4785
|
|
4786 xim_real_init(x11_window, x11_display);
|
|
4787 gui_set_shellsize(FALSE, FALSE);
|
|
4788 if (xic != NULL)
|
|
4789 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL,
|
|
4790 xim_instantiate_cb, NULL);
|
|
4791 }
|
|
4792
|
|
4793 /*ARGSUSED*/
|
|
4794 static void
|
|
4795 xim_destroy_cb(im, client_data, call_data)
|
|
4796 XIM im;
|
|
4797 XPointer client_data;
|
|
4798 XPointer call_data;
|
|
4799 {
|
|
4800 Window x11_window;
|
|
4801 Display *x11_display;
|
|
4802
|
|
4803 #ifdef XIM_DEBUG
|
|
4804 xim_log("xim_destroy_cb()\n");
|
|
4805 #endif
|
|
4806 gui_get_x11_windis(&x11_window, &x11_display);
|
|
4807
|
|
4808 xic = NULL;
|
|
4809 status_area_enabled = FALSE;
|
|
4810
|
|
4811 gui_set_shellsize(FALSE, FALSE);
|
|
4812
|
|
4813 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL,
|
|
4814 xim_instantiate_cb, NULL);
|
|
4815 }
|
|
4816 #endif
|
|
4817
|
|
4818 void
|
|
4819 xim_init()
|
|
4820 {
|
|
4821 Window x11_window;
|
|
4822 Display *x11_display;
|
|
4823
|
|
4824 #ifdef XIM_DEBUG
|
|
4825 xim_log("xim_init()\n");
|
|
4826 #endif
|
|
4827
|
|
4828 gui_get_x11_windis(&x11_window, &x11_display);
|
|
4829
|
|
4830 xic = NULL;
|
|
4831
|
|
4832 if (xim_real_init(x11_window, x11_display))
|
|
4833 return;
|
|
4834
|
|
4835 gui_set_shellsize(FALSE, FALSE);
|
|
4836
|
|
4837 #ifdef USE_X11R6_XIM
|
|
4838 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL,
|
|
4839 xim_instantiate_cb, NULL);
|
|
4840 #endif
|
|
4841 }
|
|
4842
|
|
4843 static int
|
|
4844 xim_real_init(x11_window, x11_display)
|
|
4845 Window x11_window;
|
|
4846 Display *x11_display;
|
|
4847 {
|
|
4848 int i;
|
|
4849 char *p,
|
|
4850 *s,
|
|
4851 *ns,
|
|
4852 *end,
|
|
4853 tmp[1024];
|
|
4854 #define IMLEN_MAX 40
|
|
4855 char buf[IMLEN_MAX + 7];
|
|
4856 XIM xim = NULL;
|
|
4857 XIMStyles *xim_styles;
|
|
4858 XIMStyle this_input_style = 0;
|
|
4859 Boolean found;
|
|
4860 XPoint over_spot;
|
|
4861 XVaNestedList preedit_list, status_list;
|
|
4862
|
|
4863 input_style = 0;
|
|
4864 status_area_enabled = FALSE;
|
|
4865
|
|
4866 if (xic != NULL)
|
|
4867 return FALSE;
|
|
4868
|
|
4869 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL)
|
|
4870 {
|
|
4871 strcpy(tmp, gui.rsrc_input_method);
|
|
4872 for (ns = s = tmp; ns != NULL && *s != NUL;)
|
|
4873 {
|
|
4874 s = (char *)skipwhite((char_u *)s);
|
|
4875 if (*s == NUL)
|
|
4876 break;
|
|
4877 if ((ns = end = strchr(s, ',')) == NULL)
|
|
4878 end = s + strlen(s);
|
|
4879 while (isspace(((char_u *)end)[-1]))
|
|
4880 end--;
|
|
4881 *end = NUL;
|
|
4882
|
|
4883 if (strlen(s) <= IMLEN_MAX)
|
|
4884 {
|
|
4885 strcpy(buf, "@im=");
|
|
4886 strcat(buf, s);
|
|
4887 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL
|
|
4888 && (xim = XOpenIM(x11_display, NULL, NULL, NULL))
|
|
4889 != NULL)
|
|
4890 break;
|
|
4891 }
|
|
4892
|
|
4893 s = ns + 1;
|
|
4894 }
|
|
4895 }
|
|
4896
|
|
4897 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL)
|
|
4898 xim = XOpenIM(x11_display, NULL, NULL, NULL);
|
|
4899
|
|
4900 /* This is supposed to be useful to obtain characters through
|
|
4901 * XmbLookupString() without really using a XIM. */
|
|
4902 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL
|
|
4903 && *p != NUL)
|
|
4904 xim = XOpenIM(x11_display, NULL, NULL, NULL);
|
|
4905
|
|
4906 if (xim == NULL)
|
|
4907 {
|
|
4908 /* Only give this message when verbose is set, because too many people
|
|
4909 * got this message when they didn't want to use a XIM. */
|
|
4910 if (p_verbose > 0)
|
292
|
4911 {
|
|
4912 verbose_enter();
|
7
|
4913 EMSG(_("E286: Failed to open input method"));
|
292
|
4914 verbose_leave();
|
|
4915 }
|
7
|
4916 return FALSE;
|
|
4917 }
|
|
4918
|
|
4919 #ifdef USE_X11R6_XIM
|
|
4920 {
|
|
4921 XIMCallback destroy_cb;
|
|
4922
|
|
4923 destroy_cb.callback = xim_destroy_cb;
|
|
4924 destroy_cb.client_data = NULL;
|
|
4925 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL))
|
|
4926 EMSG(_("E287: Warning: Could not set destroy callback to IM"));
|
|
4927 }
|
|
4928 #endif
|
|
4929
|
|
4930 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles)
|
|
4931 {
|
|
4932 EMSG(_("E288: input method doesn't support any style"));
|
|
4933 XCloseIM(xim);
|
|
4934 return FALSE;
|
|
4935 }
|
|
4936
|
|
4937 found = False;
|
|
4938 strcpy(tmp, gui.rsrc_preedit_type_name);
|
|
4939 for (s = tmp; s && !found; )
|
|
4940 {
|
|
4941 while (*s && isspace((unsigned char)*s))
|
|
4942 s++;
|
|
4943 if (!*s)
|
|
4944 break;
|
|
4945 if ((ns = end = strchr(s, ',')) != 0)
|
|
4946 ns++;
|
|
4947 else
|
|
4948 end = s + strlen(s);
|
|
4949 while (isspace((unsigned char)*end))
|
|
4950 end--;
|
|
4951 *end = '\0';
|
|
4952
|
|
4953 if (!strcmp(s, "OverTheSpot"))
|
|
4954 this_input_style = (XIMPreeditPosition | XIMStatusArea);
|
|
4955 else if (!strcmp(s, "OffTheSpot"))
|
|
4956 this_input_style = (XIMPreeditArea | XIMStatusArea);
|
|
4957 else if (!strcmp(s, "Root"))
|
|
4958 this_input_style = (XIMPreeditNothing | XIMStatusNothing);
|
|
4959
|
|
4960 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++)
|
|
4961 {
|
|
4962 if (this_input_style == xim_styles->supported_styles[i])
|
|
4963 {
|
|
4964 found = True;
|
|
4965 break;
|
|
4966 }
|
|
4967 }
|
|
4968 if (!found)
|
|
4969 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++)
|
|
4970 {
|
|
4971 if ((xim_styles->supported_styles[i] & this_input_style)
|
|
4972 == (this_input_style & ~XIMStatusArea))
|
|
4973 {
|
|
4974 this_input_style &= ~XIMStatusArea;
|
|
4975 found = True;
|
|
4976 break;
|
|
4977 }
|
|
4978 }
|
|
4979
|
|
4980 s = ns;
|
|
4981 }
|
|
4982 XFree(xim_styles);
|
|
4983
|
|
4984 if (!found)
|
|
4985 {
|
|
4986 /* Only give this message when verbose is set, because too many people
|
|
4987 * got this message when they didn't want to use a XIM. */
|
|
4988 if (p_verbose > 0)
|
292
|
4989 {
|
|
4990 verbose_enter();
|
7
|
4991 EMSG(_("E289: input method doesn't support my preedit type"));
|
292
|
4992 verbose_leave();
|
|
4993 }
|
7
|
4994 XCloseIM(xim);
|
|
4995 return FALSE;
|
|
4996 }
|
|
4997
|
|
4998 over_spot.x = TEXT_X(gui.col);
|
|
4999 over_spot.y = TEXT_Y(gui.row);
|
|
5000 input_style = this_input_style;
|
|
5001
|
|
5002 /* A crash was reported when trying to pass gui.norm_font as XNFontSet,
|
|
5003 * thus that has been removed. Hopefully the default works... */
|
|
5004 #ifdef FEAT_XFONTSET
|
|
5005 if (gui.fontset != NOFONTSET)
|
|
5006 {
|
|
5007 preedit_list = XVaCreateNestedList(0,
|
|
5008 XNSpotLocation, &over_spot,
|
|
5009 XNForeground, (Pixel)gui.def_norm_pixel,
|
|
5010 XNBackground, (Pixel)gui.def_back_pixel,
|
|
5011 XNFontSet, (XFontSet)gui.fontset,
|
|
5012 NULL);
|
|
5013 status_list = XVaCreateNestedList(0,
|
|
5014 XNForeground, (Pixel)gui.def_norm_pixel,
|
|
5015 XNBackground, (Pixel)gui.def_back_pixel,
|
|
5016 XNFontSet, (XFontSet)gui.fontset,
|
|
5017 NULL);
|
|
5018 }
|
|
5019 else
|
|
5020 #endif
|
|
5021 {
|
|
5022 preedit_list = XVaCreateNestedList(0,
|
|
5023 XNSpotLocation, &over_spot,
|
|
5024 XNForeground, (Pixel)gui.def_norm_pixel,
|
|
5025 XNBackground, (Pixel)gui.def_back_pixel,
|
|
5026 NULL);
|
|
5027 status_list = XVaCreateNestedList(0,
|
|
5028 XNForeground, (Pixel)gui.def_norm_pixel,
|
|
5029 XNBackground, (Pixel)gui.def_back_pixel,
|
|
5030 NULL);
|
|
5031 }
|
|
5032
|
|
5033 xic = XCreateIC(xim,
|
|
5034 XNInputStyle, input_style,
|
|
5035 XNClientWindow, x11_window,
|
|
5036 XNFocusWindow, gui.wid,
|
|
5037 XNPreeditAttributes, preedit_list,
|
|
5038 XNStatusAttributes, status_list,
|
|
5039 NULL);
|
|
5040 XFree(status_list);
|
|
5041 XFree(preedit_list);
|
|
5042 if (xic != NULL)
|
|
5043 {
|
|
5044 if (input_style & XIMStatusArea)
|
|
5045 {
|
|
5046 xim_set_status_area();
|
|
5047 status_area_enabled = TRUE;
|
|
5048 }
|
|
5049 else
|
|
5050 gui_set_shellsize(FALSE, FALSE);
|
|
5051 }
|
|
5052 else
|
|
5053 {
|
|
5054 EMSG(_(e_xim));
|
|
5055 XCloseIM(xim);
|
|
5056 return FALSE;
|
|
5057 }
|
|
5058
|
|
5059 return TRUE;
|
|
5060 }
|
|
5061
|
|
5062 #endif /* FEAT_GUI_X11 */
|
|
5063
|
|
5064 #if defined(FEAT_GUI_GTK) || defined(PROTO)
|
|
5065
|
|
5066 # ifdef FEAT_XFONTSET
|
|
5067 static char e_overthespot[] = N_("E290: over-the-spot style requires fontset");
|
|
5068 # endif
|
|
5069
|
|
5070 # ifdef PROTO
|
|
5071 typedef int GdkIC;
|
|
5072 # endif
|
|
5073
|
|
5074 void
|
|
5075 xim_decide_input_style()
|
|
5076 {
|
|
5077 /* GDK_IM_STATUS_CALLBACKS was disabled, enabled it to allow Japanese
|
|
5078 * OverTheSpot. */
|
|
5079 int supported_style = (int)GDK_IM_PREEDIT_NONE |
|
|
5080 (int)GDK_IM_PREEDIT_NOTHING |
|
|
5081 (int)GDK_IM_PREEDIT_POSITION |
|
|
5082 (int)GDK_IM_PREEDIT_CALLBACKS |
|
|
5083 (int)GDK_IM_STATUS_CALLBACKS |
|
|
5084 (int)GDK_IM_STATUS_AREA |
|
|
5085 (int)GDK_IM_STATUS_NONE |
|
|
5086 (int)GDK_IM_STATUS_NOTHING;
|
|
5087
|
|
5088 #ifdef XIM_DEBUG
|
|
5089 xim_log("xim_decide_input_style()\n");
|
|
5090 #endif
|
|
5091
|
|
5092 if (!gdk_im_ready())
|
|
5093 xim_input_style = 0;
|
|
5094 else
|
|
5095 {
|
|
5096 if (gtk_major_version > 1
|
|
5097 || (gtk_major_version == 1
|
|
5098 && (gtk_minor_version > 2
|
|
5099 || (gtk_minor_version == 2 && gtk_micro_version >= 3))))
|
|
5100 use_status_area = TRUE;
|
|
5101 else
|
|
5102 {
|
|
5103 EMSG(_("E291: Your GTK+ is older than 1.2.3. Status area disabled"));
|
|
5104 use_status_area = FALSE;
|
|
5105 }
|
|
5106 #ifdef FEAT_XFONTSET
|
|
5107 if (gui.fontset == NOFONTSET || gui.fontset->type != GDK_FONT_FONTSET)
|
|
5108 #endif
|
|
5109 supported_style &= ~((int)GDK_IM_PREEDIT_POSITION
|
|
5110 | (int)GDK_IM_STATUS_AREA);
|
|
5111 if (!use_status_area)
|
|
5112 supported_style &= ~(int)GDK_IM_STATUS_AREA;
|
|
5113 xim_input_style = (int)gdk_im_decide_style((GdkIMStyle)supported_style);
|
|
5114 }
|
|
5115 }
|
|
5116
|
|
5117 /*ARGSUSED*/
|
|
5118 static void
|
|
5119 preedit_start_cbproc(XIC xic, XPointer client_data, XPointer call_data)
|
|
5120 {
|
|
5121 #ifdef XIM_DEBUG
|
|
5122 xim_log("xim_decide_input_style()\n");
|
|
5123 #endif
|
|
5124
|
|
5125 draw_feedback = NULL;
|
|
5126 xim_can_preediting = TRUE;
|
|
5127 xim_has_preediting = TRUE;
|
|
5128 gui_update_cursor(TRUE, FALSE);
|
|
5129 if (showmode() > 0)
|
|
5130 {
|
|
5131 setcursor();
|
|
5132 out_flush();
|
|
5133 }
|
|
5134 }
|
|
5135
|
|
5136 static void
|
|
5137 xim_back_delete(int n)
|
|
5138 {
|
|
5139 char_u str[3];
|
|
5140
|
|
5141 str[0] = CSI;
|
|
5142 str[1] = 'k';
|
|
5143 str[2] = 'b';
|
|
5144 while (n-- > 0)
|
|
5145 add_to_input_buf(str, 3);
|
|
5146 }
|
|
5147
|
|
5148 static GSList *key_press_event_queue = NULL;
|
|
5149 static gboolean processing_queued_event = FALSE;
|
|
5150
|
|
5151 /*ARGSUSED*/
|
|
5152 static void
|
|
5153 preedit_draw_cbproc(XIC xic, XPointer client_data, XPointer call_data)
|
|
5154 {
|
|
5155 XIMPreeditDrawCallbackStruct *draw_data;
|
|
5156 XIMText *text;
|
|
5157 char *src;
|
|
5158 GSList *event_queue;
|
|
5159
|
|
5160 #ifdef XIM_DEBUG
|
|
5161 xim_log("preedit_draw_cbproc()\n");
|
|
5162 #endif
|
|
5163
|
|
5164 draw_data = (XIMPreeditDrawCallbackStruct *) call_data;
|
|
5165 text = (XIMText *) draw_data->text;
|
|
5166
|
|
5167 if ((text == NULL && draw_data->chg_length == preedit_buf_len)
|
|
5168 || preedit_buf_len == 0)
|
|
5169 {
|
|
5170 init_preedit_start_col();
|
|
5171 vim_free(draw_feedback);
|
|
5172 draw_feedback = NULL;
|
|
5173 }
|
|
5174 if (draw_data->chg_length > 0)
|
|
5175 {
|
|
5176 int bs_cnt;
|
|
5177
|
|
5178 if (draw_data->chg_length > preedit_buf_len)
|
|
5179 bs_cnt = preedit_buf_len;
|
|
5180 else
|
|
5181 bs_cnt = draw_data->chg_length;
|
|
5182 xim_back_delete(bs_cnt);
|
|
5183 preedit_buf_len -= bs_cnt;
|
|
5184 }
|
|
5185 if (text != NULL)
|
|
5186 {
|
|
5187 int len;
|
|
5188 #ifdef FEAT_MBYTE
|
|
5189 char_u *buf = NULL;
|
|
5190 unsigned int nfeedback = 0;
|
|
5191 #endif
|
|
5192 char_u *ptr;
|
|
5193
|
|
5194 src = text->string.multi_byte;
|
|
5195 if (src != NULL && !text->encoding_is_wchar)
|
|
5196 {
|
|
5197 len = strlen(src);
|
|
5198 ptr = (char_u *)src;
|
|
5199 /* Avoid the enter for decision */
|
|
5200 if (*ptr == '\n')
|
|
5201 return;
|
|
5202
|
|
5203 #ifdef FEAT_MBYTE
|
|
5204 if (input_conv.vc_type != CONV_NONE
|
|
5205 && (buf = string_convert(&input_conv,
|
|
5206 (char_u *)src, &len)) != NULL)
|
|
5207 {
|
|
5208 /* Converted from 'termencoding' to 'encoding'. */
|
|
5209 add_to_input_buf_csi(buf, len);
|
|
5210 ptr = buf;
|
|
5211 }
|
|
5212 else
|
|
5213 #endif
|
|
5214 add_to_input_buf_csi((char_u *)src, len);
|
|
5215 /* Add count of character to preedit_buf_len */
|
|
5216 while (*ptr != NUL)
|
|
5217 {
|
|
5218 #ifdef FEAT_MBYTE
|
|
5219 if (draw_data->text->feedback != NULL)
|
|
5220 {
|
|
5221 if (draw_feedback == NULL)
|
|
5222 draw_feedback = (char *)alloc(draw_data->chg_first
|
|
5223 + text->length);
|
|
5224 else
|
|
5225 draw_feedback = realloc(draw_feedback,
|
|
5226 draw_data->chg_first + text->length);
|
|
5227 if (draw_feedback != NULL)
|
|
5228 {
|
|
5229 draw_feedback[nfeedback + draw_data->chg_first]
|
|
5230 = draw_data->text->feedback[nfeedback];
|
|
5231 nfeedback++;
|
|
5232 }
|
|
5233 }
|
|
5234 if (has_mbyte)
|
|
5235 ptr += mb_ptr2len_check(ptr);
|
|
5236 else
|
|
5237 #endif
|
|
5238 ptr++;
|
|
5239 preedit_buf_len++;
|
|
5240 }
|
|
5241 #ifdef FEAT_MBYTE
|
|
5242 vim_free(buf);
|
|
5243 #endif
|
|
5244 preedit_end_col = MAXCOL;
|
|
5245 }
|
|
5246 }
|
|
5247 if (text != NULL || draw_data->chg_length > 0)
|
|
5248 {
|
|
5249 event_queue = key_press_event_queue;
|
|
5250 processing_queued_event = TRUE;
|
|
5251 while (event_queue != NULL && processing_queued_event)
|
|
5252 {
|
|
5253 GdkEvent *ev = event_queue->data;
|
|
5254
|
|
5255 gboolean *ret;
|
|
5256 gtk_signal_emit_by_name((GtkObject*)gui.mainwin, "key_press_event",
|
|
5257 ev, &ret);
|
|
5258 gdk_event_free(ev);
|
|
5259 event_queue = event_queue->next;
|
|
5260 }
|
|
5261 processing_queued_event = FALSE;
|
|
5262 if (key_press_event_queue)
|
|
5263 {
|
|
5264 g_slist_free(key_press_event_queue);
|
|
5265 key_press_event_queue = NULL;
|
|
5266 }
|
|
5267 }
|
|
5268 if (gtk_main_level() > 0)
|
|
5269 gtk_main_quit();
|
|
5270 }
|
|
5271
|
|
5272 /*
|
|
5273 * Retrieve the highlighting attributes at column col in the preedit string.
|
|
5274 * Return -1 if not in preediting mode or if col is out of range.
|
|
5275 */
|
|
5276 int
|
|
5277 im_get_feedback_attr(int col)
|
|
5278 {
|
|
5279 if (draw_feedback != NULL && col < preedit_buf_len)
|
|
5280 {
|
|
5281 if (draw_feedback[col] & XIMReverse)
|
|
5282 return HL_INVERSE;
|
|
5283 else if (draw_feedback[col] & XIMUnderline)
|
|
5284 return HL_UNDERLINE;
|
|
5285 else
|
|
5286 return hl_attr(HLF_V);
|
|
5287 }
|
|
5288
|
|
5289 return -1;
|
|
5290 }
|
|
5291
|
|
5292 /*ARGSUSED*/
|
|
5293 static void
|
|
5294 preedit_caret_cbproc(XIC xic, XPointer client_data, XPointer call_data)
|
|
5295 {
|
|
5296 #ifdef XIM_DEBUG
|
|
5297 xim_log("preedit_caret_cbproc()\n");
|
|
5298 #endif
|
|
5299 }
|
|
5300
|
|
5301 /*ARGSUSED*/
|
|
5302 static void
|
|
5303 preedit_done_cbproc(XIC xic, XPointer client_data, XPointer call_data)
|
|
5304 {
|
|
5305 #ifdef XIM_DEBUG
|
|
5306 xim_log("preedit_done_cbproc()\n");
|
|
5307 #endif
|
|
5308
|
|
5309 vim_free(draw_feedback);
|
|
5310 draw_feedback = NULL;
|
|
5311 xim_can_preediting = FALSE;
|
|
5312 xim_has_preediting = FALSE;
|
|
5313 gui_update_cursor(TRUE, FALSE);
|
|
5314 if (showmode() > 0)
|
|
5315 {
|
|
5316 setcursor();
|
|
5317 out_flush();
|
|
5318 }
|
|
5319 }
|
|
5320
|
|
5321 void
|
|
5322 xim_reset(void)
|
|
5323 {
|
|
5324 char *text;
|
|
5325
|
|
5326 #ifdef XIM_DEBUG
|
|
5327 xim_log("xim_reset()\n");
|
|
5328 #endif
|
|
5329
|
|
5330 if (xic != NULL)
|
|
5331 {
|
|
5332 text = XmbResetIC(((GdkICPrivate *)xic)->xic);
|
|
5333 if (text != NULL && !(xim_input_style & (int)GDK_IM_PREEDIT_CALLBACKS))
|
|
5334 add_to_input_buf_csi((char_u *)text, strlen(text));
|
|
5335 else
|
|
5336 preedit_buf_len = 0;
|
|
5337 if (text != NULL)
|
|
5338 XFree(text);
|
|
5339 }
|
|
5340 }
|
|
5341
|
|
5342 /*ARGSUSED*/
|
|
5343 int
|
|
5344 xim_queue_key_press_event(GdkEventKey *event, int down)
|
|
5345 {
|
|
5346 #ifdef XIM_DEBUG
|
|
5347 xim_log("xim_queue_key_press_event()\n");
|
|
5348 #endif
|
|
5349
|
|
5350 if (preedit_buf_len <= 0)
|
|
5351 return FALSE;
|
|
5352 if (processing_queued_event)
|
|
5353 processing_queued_event = FALSE;
|
|
5354
|
|
5355 key_press_event_queue = g_slist_append(key_press_event_queue,
|
|
5356 gdk_event_copy((GdkEvent *)event));
|
|
5357 return TRUE;
|
|
5358 }
|
|
5359
|
|
5360 /*ARGSUSED*/
|
|
5361 static void
|
|
5362 preedit_callback_setup(GdkIC *ic)
|
|
5363 {
|
|
5364 XIC xxic;
|
|
5365 XVaNestedList preedit_attr;
|
|
5366 XIMCallback preedit_start_cb;
|
|
5367 XIMCallback preedit_draw_cb;
|
|
5368 XIMCallback preedit_caret_cb;
|
|
5369 XIMCallback preedit_done_cb;
|
|
5370
|
|
5371 xxic = ((GdkICPrivate*)xic)->xic;
|
|
5372 preedit_start_cb.callback = (XIMProc)preedit_start_cbproc;
|
|
5373 preedit_draw_cb.callback = (XIMProc)preedit_draw_cbproc;
|
|
5374 preedit_caret_cb.callback = (XIMProc)preedit_caret_cbproc;
|
|
5375 preedit_done_cb.callback = (XIMProc)preedit_done_cbproc;
|
|
5376 preedit_attr
|
|
5377 = XVaCreateNestedList (0,
|
|
5378 XNPreeditStartCallback, &preedit_start_cb,
|
|
5379 XNPreeditDrawCallback, &preedit_draw_cb,
|
|
5380 XNPreeditCaretCallback, &preedit_caret_cb,
|
|
5381 XNPreeditDoneCallback, &preedit_done_cb,
|
|
5382 0);
|
|
5383 XSetICValues (xxic, XNPreeditAttributes, preedit_attr, 0);
|
|
5384 XFree(preedit_attr);
|
|
5385 }
|
|
5386
|
|
5387 /*ARGSUSED*/
|
|
5388 static void
|
|
5389 reset_state_setup(GdkIC *ic)
|
|
5390 {
|
|
5391 #ifdef USE_X11R6_XIM
|
|
5392 /* don't change the input context when we call reset */
|
|
5393 XSetICValues(((GdkICPrivate*)ic)->xic, XNResetState, XIMPreserveState, 0);
|
|
5394 #endif
|
|
5395 }
|
|
5396
|
|
5397 void
|
|
5398 xim_init(void)
|
|
5399 {
|
|
5400 #ifdef XIM_DEBUG
|
|
5401 xim_log("xim_init()\n");
|
|
5402 #endif
|
|
5403
|
|
5404 xic = NULL;
|
|
5405 xic_attr = NULL;
|
|
5406
|
|
5407 if (!gdk_im_ready())
|
|
5408 {
|
|
5409 if (p_verbose > 0)
|
292
|
5410 {
|
|
5411 verbose_enter();
|
7
|
5412 EMSG(_("E292: Input Method Server is not running"));
|
292
|
5413 verbose_leave();
|
|
5414 }
|
7
|
5415 return;
|
|
5416 }
|
|
5417 if ((xic_attr = gdk_ic_attr_new()) != NULL)
|
|
5418 {
|
|
5419 #ifdef FEAT_XFONTSET
|
|
5420 gint width, height;
|
|
5421 #endif
|
|
5422 int mask;
|
|
5423 GdkColormap *colormap;
|
|
5424 GdkICAttr *attr = xic_attr;
|
|
5425 int attrmask = (int)GDK_IC_ALL_REQ;
|
|
5426 GtkWidget *widget = gui.drawarea;
|
|
5427
|
|
5428 attr->style = (GdkIMStyle)xim_input_style;
|
|
5429 attr->client_window = gui.mainwin->window;
|
|
5430
|
|
5431 if ((colormap = gtk_widget_get_colormap(widget)) !=
|
|
5432 gtk_widget_get_default_colormap())
|
|
5433 {
|
|
5434 attrmask |= (int)GDK_IC_PREEDIT_COLORMAP;
|
|
5435 attr->preedit_colormap = colormap;
|
|
5436 }
|
|
5437 attrmask |= (int)GDK_IC_PREEDIT_FOREGROUND;
|
|
5438 attrmask |= (int)GDK_IC_PREEDIT_BACKGROUND;
|
|
5439 attr->preedit_foreground = widget->style->fg[GTK_STATE_NORMAL];
|
|
5440 attr->preedit_background = widget->style->base[GTK_STATE_NORMAL];
|
|
5441
|
|
5442 #ifdef FEAT_XFONTSET
|
|
5443 if ((xim_input_style & (int)GDK_IM_PREEDIT_MASK)
|
|
5444 == (int)GDK_IM_PREEDIT_POSITION)
|
|
5445 {
|
|
5446 if (gui.fontset == NOFONTSET
|
|
5447 || gui.fontset->type != GDK_FONT_FONTSET)
|
|
5448 {
|
|
5449 EMSG(_(e_overthespot));
|
|
5450 }
|
|
5451 else
|
|
5452 {
|
|
5453 gdk_window_get_size(widget->window, &width, &height);
|
|
5454
|
|
5455 attrmask |= (int)GDK_IC_PREEDIT_POSITION_REQ;
|
|
5456 attr->spot_location.x = TEXT_X(0);
|
|
5457 attr->spot_location.y = TEXT_Y(0);
|
|
5458 attr->preedit_area.x = gui.border_offset;
|
|
5459 attr->preedit_area.y = gui.border_offset;
|
|
5460 attr->preedit_area.width = width - 2*gui.border_offset;
|
|
5461 attr->preedit_area.height = height - 2*gui.border_offset;
|
|
5462 attr->preedit_fontset = gui.fontset;
|
|
5463 }
|
|
5464 }
|
|
5465
|
|
5466 if ((xim_input_style & (int)GDK_IM_STATUS_MASK)
|
|
5467 == (int)GDK_IM_STATUS_AREA)
|
|
5468 {
|
|
5469 if (gui.fontset == NOFONTSET
|
|
5470 || gui.fontset->type != GDK_FONT_FONTSET)
|
|
5471 {
|
|
5472 EMSG(_(e_overthespot));
|
|
5473 }
|
|
5474 else
|
|
5475 {
|
|
5476 gdk_window_get_size(gui.mainwin->window, &width, &height);
|
|
5477 attrmask |= (int)GDK_IC_STATUS_AREA_REQ;
|
|
5478 attr->status_area.x = 0;
|
|
5479 attr->status_area.y = height - gui.char_height - 1;
|
|
5480 attr->status_area.width = width;
|
|
5481 attr->status_area.height = gui.char_height;
|
|
5482 attr->status_fontset = gui.fontset;
|
|
5483 }
|
|
5484 }
|
|
5485 else if ((xim_input_style & (int)GDK_IM_STATUS_MASK)
|
|
5486 == (int)GDK_IM_STATUS_CALLBACKS)
|
|
5487 {
|
|
5488 /* FIXME */
|
|
5489 }
|
|
5490 #endif
|
|
5491
|
|
5492 xic = gdk_ic_new(attr, (GdkICAttributesType)attrmask);
|
|
5493
|
|
5494 if (xic == NULL)
|
|
5495 EMSG(_(e_xim));
|
|
5496 else
|
|
5497 {
|
|
5498 mask = (int)gdk_window_get_events(widget->window);
|
|
5499 mask |= (int)gdk_ic_get_events(xic);
|
|
5500 gdk_window_set_events(widget->window, (GdkEventMask)mask);
|
|
5501 if (xim_input_style & (int)GDK_IM_PREEDIT_CALLBACKS)
|
|
5502 preedit_callback_setup(xic);
|
|
5503 reset_state_setup(xic);
|
|
5504 }
|
|
5505 }
|
|
5506 }
|
|
5507
|
|
5508 void
|
|
5509 im_shutdown(void)
|
|
5510 {
|
|
5511 #ifdef XIM_DEBUG
|
|
5512 xim_log("im_shutdown()\n");
|
|
5513 #endif
|
|
5514
|
|
5515 if (xic != NULL)
|
|
5516 {
|
|
5517 gdk_im_end();
|
|
5518 gdk_ic_destroy(xic);
|
|
5519 xic = NULL;
|
|
5520 }
|
|
5521 xim_is_active = FALSE;
|
|
5522 xim_can_preediting = FALSE;
|
|
5523 preedit_start_col = MAXCOL;
|
|
5524 xim_has_preediting = FALSE;
|
|
5525 }
|
|
5526
|
|
5527 #endif /* FEAT_GUI_GTK */
|
|
5528
|
|
5529 int
|
|
5530 xim_get_status_area_height()
|
|
5531 {
|
|
5532 #ifdef FEAT_GUI_GTK
|
|
5533 if (xim_input_style & (int)GDK_IM_STATUS_AREA)
|
|
5534 return gui.char_height;
|
129
|
5535 #else
|
|
5536 # if defined FEAT_GUI_KDE
|
42
|
5537 /* always return zero? */
|
129
|
5538 # else
|
7
|
5539 if (status_area_enabled)
|
|
5540 return gui.char_height;
|
129
|
5541 # endif
|
7
|
5542 #endif
|
|
5543 return 0;
|
|
5544 }
|
|
5545
|
|
5546 /*
|
|
5547 * Get IM status. When IM is on, return TRUE. Else return FALSE.
|
|
5548 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is
|
|
5549 * active, when not having focus XIM may still be active (e.g., when using a
|
|
5550 * tear-off menu item).
|
|
5551 */
|
|
5552 int
|
|
5553 im_get_status()
|
|
5554 {
|
|
5555 # ifdef FEAT_GUI_GTK
|
|
5556 if (xim_input_style & (int)GDK_IM_PREEDIT_CALLBACKS)
|
|
5557 return xim_can_preediting;
|
|
5558 # endif
|
42
|
5559 # ifdef FEAT_GUI_KDE
|
|
5560 if (preedit_start_col != MAXCOL)
|
|
5561 return TRUE;
|
|
5562 # endif
|
7
|
5563 return xim_has_focus;
|
|
5564 }
|
|
5565
|
|
5566 # endif /* !HAVE_GTK2 */
|
|
5567
|
|
5568 # if defined(FEAT_GUI_GTK) || defined(PROTO)
|
|
5569 int
|
|
5570 im_is_preediting()
|
|
5571 {
|
|
5572 return xim_has_preediting;
|
|
5573 }
|
|
5574 # endif
|
|
5575 #endif /* FEAT_XIM */
|
|
5576
|
|
5577 #if defined(FEAT_MBYTE) || defined(PROTO)
|
|
5578
|
|
5579 /*
|
|
5580 * Setup "vcp" for conversion from "from" to "to".
|
|
5581 * The names must have been made canonical with enc_canonize().
|
|
5582 * vcp->vc_type must have been initialized to CONV_NONE.
|
|
5583 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8
|
|
5584 * instead).
|
|
5585 * Afterwards invoke with "from" and "to" equal to NULL to cleanup.
|
|
5586 * Return FAIL when conversion is not supported, OK otherwise.
|
|
5587 */
|
|
5588 int
|
|
5589 convert_setup(vcp, from, to)
|
|
5590 vimconv_T *vcp;
|
|
5591 char_u *from;
|
|
5592 char_u *to;
|
|
5593 {
|
|
5594 int from_prop;
|
|
5595 int to_prop;
|
|
5596
|
|
5597 /* Reset to no conversion. */
|
|
5598 # ifdef USE_ICONV
|
|
5599 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1)
|
|
5600 iconv_close(vcp->vc_fd);
|
|
5601 # endif
|
|
5602 vcp->vc_type = CONV_NONE;
|
|
5603 vcp->vc_factor = 1;
|
|
5604 vcp->vc_fail = FALSE;
|
|
5605
|
|
5606 /* No conversion when one of the names is empty or they are equal. */
|
|
5607 if (from == NULL || *from == NUL || to == NULL || *to == NUL
|
|
5608 || STRCMP(from, to) == 0)
|
|
5609 return OK;
|
|
5610
|
|
5611 from_prop = enc_canon_props(from);
|
|
5612 to_prop = enc_canon_props(to);
|
|
5613 if ((from_prop & ENC_LATIN1) && (to_prop & ENC_UNICODE))
|
|
5614 {
|
|
5615 /* Internal latin1 -> utf-8 conversion. */
|
|
5616 vcp->vc_type = CONV_TO_UTF8;
|
|
5617 vcp->vc_factor = 2; /* up to twice as long */
|
|
5618 }
|
26
|
5619 else if ((from_prop & ENC_LATIN9) && (to_prop & ENC_UNICODE))
|
|
5620 {
|
|
5621 /* Internal latin9 -> utf-8 conversion. */
|
|
5622 vcp->vc_type = CONV_9_TO_UTF8;
|
|
5623 vcp->vc_factor = 3; /* up to three as long (euro sign) */
|
|
5624 }
|
7
|
5625 else if ((from_prop & ENC_UNICODE) && (to_prop & ENC_LATIN1))
|
|
5626 {
|
|
5627 /* Internal utf-8 -> latin1 conversion. */
|
|
5628 vcp->vc_type = CONV_TO_LATIN1;
|
|
5629 }
|
26
|
5630 else if ((from_prop & ENC_UNICODE) && (to_prop & ENC_LATIN9))
|
|
5631 {
|
|
5632 /* Internal utf-8 -> latin9 conversion. */
|
|
5633 vcp->vc_type = CONV_TO_LATIN9;
|
|
5634 }
|
7
|
5635 #ifdef WIN3264
|
|
5636 /* Win32-specific codepage <-> codepage conversion without iconv. */
|
|
5637 else if (((from_prop & ENC_UNICODE) || encname2codepage(from) > 0)
|
|
5638 && ((to_prop & ENC_UNICODE) || encname2codepage(to) > 0))
|
|
5639 {
|
|
5640 vcp->vc_type = CONV_CODEPAGE;
|
|
5641 vcp->vc_factor = 2; /* up to twice as long */
|
|
5642 vcp->vc_cpfrom = (from_prop & ENC_UNICODE) ? 0 : encname2codepage(from);
|
|
5643 vcp->vc_cpto = (to_prop & ENC_UNICODE) ? 0 : encname2codepage(to);
|
|
5644 }
|
|
5645 #endif
|
|
5646 #ifdef MACOS_X
|
|
5647 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1))
|
|
5648 {
|
|
5649 vcp->vc_type = CONV_MAC_LATIN1;
|
|
5650 }
|
|
5651 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_UNICODE))
|
|
5652 {
|
|
5653 vcp->vc_type = CONV_MAC_UTF8;
|
|
5654 vcp->vc_factor = 2; /* up to twice as long */
|
|
5655 }
|
|
5656 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN))
|
|
5657 {
|
|
5658 vcp->vc_type = CONV_LATIN1_MAC;
|
|
5659 }
|
|
5660 else if ((from_prop & ENC_UNICODE) && (to_prop & ENC_MACROMAN))
|
|
5661 {
|
|
5662 vcp->vc_type = CONV_UTF8_MAC;
|
|
5663 }
|
|
5664 #endif
|
|
5665 # ifdef USE_ICONV
|
|
5666 else
|
|
5667 {
|
|
5668 /* Use iconv() for conversion. */
|
|
5669 vcp->vc_fd = (iconv_t)my_iconv_open(
|
|
5670 (to_prop & ENC_UNICODE) ? (char_u *)"utf-8" : to,
|
|
5671 (from_prop & ENC_UNICODE) ? (char_u *)"utf-8" : from);
|
|
5672 if (vcp->vc_fd != (iconv_t)-1)
|
|
5673 {
|
|
5674 vcp->vc_type = CONV_ICONV;
|
|
5675 vcp->vc_factor = 4; /* could be longer too... */
|
|
5676 }
|
|
5677 }
|
|
5678 # endif
|
|
5679 if (vcp->vc_type == CONV_NONE)
|
|
5680 return FAIL;
|
167
|
5681
|
7
|
5682 return OK;
|
|
5683 }
|
|
5684
|
|
5685 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \
|
|
5686 || defined(MSDOS) || defined(PROTO)
|
|
5687 /*
|
|
5688 * Do conversion on typed input characters in-place.
|
|
5689 * The input and output are not NUL terminated!
|
|
5690 * Returns the length after conversion.
|
|
5691 */
|
|
5692 int
|
|
5693 convert_input(ptr, len, maxlen)
|
|
5694 char_u *ptr;
|
|
5695 int len;
|
|
5696 int maxlen;
|
|
5697 {
|
|
5698 return convert_input_safe(ptr, len, maxlen, NULL, NULL);
|
|
5699 }
|
|
5700 #endif
|
|
5701
|
|
5702 /*
|
|
5703 * Like convert_input(), but when there is an incomplete byte sequence at the
|
|
5704 * end return that as an allocated string in "restp" and set "*restlenp" to
|
|
5705 * the length. If "restp" is NULL it is not used.
|
|
5706 */
|
|
5707 int
|
|
5708 convert_input_safe(ptr, len, maxlen, restp, restlenp)
|
|
5709 char_u *ptr;
|
|
5710 int len;
|
|
5711 int maxlen;
|
|
5712 char_u **restp;
|
|
5713 int *restlenp;
|
|
5714 {
|
|
5715 char_u *d;
|
|
5716 int dlen = len;
|
|
5717 int unconvertlen = 0;
|
|
5718
|
|
5719 d = string_convert_ext(&input_conv, ptr, &dlen,
|
|
5720 restp == NULL ? NULL : &unconvertlen);
|
|
5721 if (d != NULL)
|
|
5722 {
|
|
5723 if (dlen <= maxlen)
|
|
5724 {
|
|
5725 if (unconvertlen > 0)
|
|
5726 {
|
|
5727 /* Move the unconverted characters to allocated memory. */
|
|
5728 *restp = alloc(unconvertlen);
|
|
5729 if (*restp != NULL)
|
|
5730 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen);
|
|
5731 *restlenp = unconvertlen;
|
|
5732 }
|
|
5733 mch_memmove(ptr, d, dlen);
|
|
5734 }
|
|
5735 else
|
|
5736 /* result is too long, keep the unconverted text (the caller must
|
|
5737 * have done something wrong!) */
|
|
5738 dlen = len;
|
|
5739 vim_free(d);
|
|
5740 }
|
|
5741 return dlen;
|
|
5742 }
|
|
5743
|
|
5744 #if defined(MACOS_X)
|
18
|
5745 /* This is in os_mac_conv.c. */
|
|
5746 extern char_u *mac_string_convert __ARGS((char_u *ptr, int len, int *lenp, int fail_on_error, int from, int to, int *unconvlenp));
|
7
|
5747 #endif
|
|
5748
|
|
5749 /*
|
|
5750 * Convert text "ptr[*lenp]" according to "vcp".
|
|
5751 * Returns the result in allocated memory and sets "*lenp".
|
|
5752 * When "lenp" is NULL, use NUL terminated strings.
|
|
5753 * Illegal chars are often changed to "?", unless vcp->vc_fail is set.
|
|
5754 * When something goes wrong, NULL is returned and "*lenp" is unchanged.
|
|
5755 */
|
|
5756 char_u *
|
|
5757 string_convert(vcp, ptr, lenp)
|
|
5758 vimconv_T *vcp;
|
|
5759 char_u *ptr;
|
|
5760 int *lenp;
|
|
5761 {
|
|
5762 return string_convert_ext(vcp, ptr, lenp, NULL);
|
|
5763 }
|
|
5764
|
|
5765 /*
|
|
5766 * Like string_convert(), but when "unconvlenp" is not NULL and there are is
|
|
5767 * an incomplete sequence at the end it is not converted and "*unconvlenp" is
|
|
5768 * set to the number of remaining bytes.
|
|
5769 */
|
|
5770 char_u *
|
|
5771 string_convert_ext(vcp, ptr, lenp, unconvlenp)
|
|
5772 vimconv_T *vcp;
|
|
5773 char_u *ptr;
|
|
5774 int *lenp;
|
|
5775 int *unconvlenp;
|
|
5776 {
|
|
5777 char_u *retval = NULL;
|
|
5778 char_u *d;
|
|
5779 int len;
|
|
5780 int i;
|
|
5781 int l;
|
|
5782 int c;
|
|
5783
|
|
5784 if (lenp == NULL)
|
|
5785 len = (int)STRLEN(ptr);
|
|
5786 else
|
|
5787 len = *lenp;
|
|
5788 if (len == 0)
|
|
5789 return vim_strsave((char_u *)"");
|
|
5790
|
|
5791 switch (vcp->vc_type)
|
|
5792 {
|
|
5793 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */
|
|
5794 retval = alloc(len * 2 + 1);
|
|
5795 if (retval == NULL)
|
|
5796 break;
|
|
5797 d = retval;
|
|
5798 for (i = 0; i < len; ++i)
|
|
5799 {
|
26
|
5800 c = ptr[i];
|
|
5801 if (c < 0x80)
|
|
5802 *d++ = c;
|
7
|
5803 else
|
|
5804 {
|
26
|
5805 *d++ = 0xc0 + ((unsigned)c >> 6);
|
|
5806 *d++ = 0x80 + (c & 0x3f);
|
7
|
5807 }
|
|
5808 }
|
|
5809 *d = NUL;
|
|
5810 if (lenp != NULL)
|
|
5811 *lenp = (int)(d - retval);
|
|
5812 break;
|
|
5813
|
26
|
5814 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */
|
|
5815 retval = alloc(len * 3 + 1);
|
|
5816 if (retval == NULL)
|
|
5817 break;
|
|
5818 d = retval;
|
|
5819 for (i = 0; i < len; ++i)
|
|
5820 {
|
|
5821 c = ptr[i];
|
|
5822 switch (c)
|
|
5823 {
|
|
5824 case 0xa4: c = 0x20ac; break; /* euro */
|
|
5825 case 0xa6: c = 0x0160; break; /* S hat */
|
|
5826 case 0xa8: c = 0x0161; break; /* S -hat */
|
|
5827 case 0xb4: c = 0x017d; break; /* Z hat */
|
|
5828 case 0xb8: c = 0x017e; break; /* Z -hat */
|
|
5829 case 0xbc: c = 0x0152; break; /* OE */
|
|
5830 case 0xbd: c = 0x0153; break; /* oe */
|
|
5831 case 0xbe: c = 0x0178; break; /* Y */
|
|
5832 }
|
|
5833 d += utf_char2bytes(c, d);
|
|
5834 }
|
|
5835 *d = NUL;
|
|
5836 if (lenp != NULL)
|
|
5837 *lenp = (int)(d - retval);
|
|
5838 break;
|
|
5839
|
7
|
5840 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */
|
26
|
5841 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */
|
7
|
5842 retval = alloc(len + 1);
|
|
5843 if (retval == NULL)
|
|
5844 break;
|
|
5845 d = retval;
|
|
5846 for (i = 0; i < len; ++i)
|
|
5847 {
|
|
5848 l = utf_ptr2len_check(ptr + i);
|
|
5849 if (l == 0)
|
|
5850 *d++ = NUL;
|
|
5851 else if (l == 1)
|
|
5852 {
|
|
5853 if (unconvlenp != NULL && utf8len_tab[ptr[i]] > len - i)
|
|
5854 {
|
|
5855 /* Incomplete sequence at the end. */
|
|
5856 *unconvlenp = len - i;
|
|
5857 break;
|
|
5858 }
|
|
5859 *d++ = ptr[i];
|
|
5860 }
|
|
5861 else
|
|
5862 {
|
|
5863 c = utf_ptr2char(ptr + i);
|
26
|
5864 if (vcp->vc_type == CONV_TO_LATIN9)
|
|
5865 switch (c)
|
|
5866 {
|
|
5867 case 0x20ac: c = 0xa4; break; /* euro */
|
|
5868 case 0x0160: c = 0xa6; break; /* S hat */
|
|
5869 case 0x0161: c = 0xa8; break; /* S -hat */
|
|
5870 case 0x017d: c = 0xb4; break; /* Z hat */
|
|
5871 case 0x017e: c = 0xb8; break; /* Z -hat */
|
|
5872 case 0x0152: c = 0xbc; break; /* OE */
|
|
5873 case 0x0153: c = 0xbd; break; /* oe */
|
|
5874 case 0x0178: c = 0xbe; break; /* Y */
|
|
5875 case 0xa4:
|
|
5876 case 0xa6:
|
|
5877 case 0xa8:
|
|
5878 case 0xb4:
|
|
5879 case 0xb8:
|
|
5880 case 0xbc:
|
|
5881 case 0xbd:
|
|
5882 case 0xbe: c = 0x100; break; /* not in latin9 */
|
|
5883 }
|
7
|
5884 if (!utf_iscomposing(c)) /* skip composing chars */
|
|
5885 {
|
|
5886 if (c < 0x100)
|
|
5887 *d++ = c;
|
|
5888 else if (vcp->vc_fail)
|
|
5889 {
|
|
5890 vim_free(retval);
|
|
5891 return NULL;
|
|
5892 }
|
|
5893 else
|
|
5894 {
|
|
5895 *d++ = 0xbf;
|
|
5896 if (utf_char2cells(c) > 1)
|
|
5897 *d++ = '?';
|
|
5898 }
|
|
5899 }
|
|
5900 i += l - 1;
|
|
5901 }
|
|
5902 }
|
|
5903 *d = NUL;
|
|
5904 if (lenp != NULL)
|
|
5905 *lenp = (int)(d - retval);
|
|
5906 break;
|
|
5907
|
|
5908 # ifdef MACOS_X
|
|
5909 case CONV_MAC_LATIN1:
|
|
5910 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail,
|
18
|
5911 'm', 'l', unconvlenp);
|
7
|
5912 break;
|
|
5913
|
|
5914 case CONV_LATIN1_MAC:
|
|
5915 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail,
|
18
|
5916 'l', 'm', unconvlenp);
|
7
|
5917 break;
|
|
5918
|
|
5919 case CONV_MAC_UTF8:
|
|
5920 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail,
|
18
|
5921 'm', 'u', unconvlenp);
|
7
|
5922 break;
|
|
5923
|
|
5924 case CONV_UTF8_MAC:
|
|
5925 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail,
|
18
|
5926 'u', 'm', unconvlenp);
|
7
|
5927 break;
|
|
5928 # endif
|
|
5929
|
|
5930 # ifdef USE_ICONV
|
|
5931 case CONV_ICONV: /* conversion with output_conv.vc_fd */
|
|
5932 retval = iconv_string(vcp, ptr, len, unconvlenp);
|
|
5933 if (retval != NULL && lenp != NULL)
|
|
5934 *lenp = (int)STRLEN(retval);
|
|
5935 break;
|
|
5936 # endif
|
|
5937 # ifdef WIN3264
|
|
5938 case CONV_CODEPAGE: /* codepage -> codepage */
|
|
5939 {
|
|
5940 int retlen;
|
|
5941 int tmp_len;
|
|
5942 short_u *tmp;
|
|
5943
|
|
5944 /* 1. codepage/UTF-8 -> ucs-2. */
|
|
5945 if (vcp->vc_cpfrom == 0)
|
|
5946 tmp_len = utf8_to_ucs2(ptr, len, NULL, NULL);
|
|
5947 else
|
|
5948 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 0,
|
|
5949 ptr, len, 0, 0);
|
|
5950 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len);
|
|
5951 if (tmp == NULL)
|
|
5952 break;
|
|
5953 if (vcp->vc_cpfrom == 0)
|
|
5954 utf8_to_ucs2(ptr, len, tmp, unconvlenp);
|
|
5955 else
|
|
5956 MultiByteToWideChar(vcp->vc_cpfrom, 0, ptr, len, tmp, tmp_len);
|
|
5957
|
|
5958 /* 2. ucs-2 -> codepage/UTF-8. */
|
|
5959 if (vcp->vc_cpto == 0)
|
|
5960 retlen = ucs2_to_utf8(tmp, tmp_len, NULL);
|
|
5961 else
|
|
5962 retlen = WideCharToMultiByte(vcp->vc_cpto, 0,
|
|
5963 tmp, tmp_len, 0, 0, 0, 0);
|
|
5964 retval = alloc(retlen + 1);
|
|
5965 if (retval != NULL)
|
|
5966 {
|
|
5967 if (vcp->vc_cpto == 0)
|
|
5968 ucs2_to_utf8(tmp, tmp_len, retval);
|
|
5969 else
|
|
5970 WideCharToMultiByte(vcp->vc_cpto, 0,
|
|
5971 tmp, tmp_len, retval, retlen, 0, 0);
|
|
5972 retval[retlen] = NUL;
|
|
5973 if (lenp != NULL)
|
|
5974 *lenp = retlen;
|
|
5975 }
|
|
5976 vim_free(tmp);
|
|
5977 break;
|
|
5978 }
|
|
5979 # endif
|
|
5980 }
|
|
5981
|
|
5982 return retval;
|
|
5983 }
|
|
5984 #endif
|