Mercurial > vim
comparison src/mbyte.c @ 7:3fc0f57ecb91 v7.0001
updated for version 7.0001
author | vimboss |
---|---|
date | Sun, 13 Jun 2004 20:20:40 +0000 |
parents | |
children | 4424b47a0797 |
comparison
equal
deleted
inserted
replaced
6:c2daee826b8f | 7:3fc0f57ecb91 |
---|---|
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}, | |
204 #define IDX_CP1255 7 | |
205 {"cp1255", ENC_8BIT, 1255}, /* close to iso-8859-8 */ | |
206 #define IDX_ISO_8 8 | |
207 {"iso-8859-8", ENC_8BIT, 0}, | |
208 #define IDX_ISO_9 9 | |
209 {"iso-8859-9", ENC_8BIT, 0}, | |
210 #define IDX_ISO_10 10 | |
211 {"iso-8859-10", ENC_8BIT, 0}, | |
212 #define IDX_ISO_11 11 | |
213 {"iso-8859-11", ENC_8BIT, 0}, | |
214 #define IDX_ISO_13 12 | |
215 {"iso-8859-13", ENC_8BIT, 0}, | |
216 #define IDX_ISO_14 13 | |
217 {"iso-8859-14", ENC_8BIT, 0}, | |
218 #define IDX_ISO_15 14 | |
219 {"iso-8859-15", ENC_8BIT, 0}, | |
220 #define IDX_KOI8_R 15 | |
221 {"koi8-r", ENC_8BIT, 0}, | |
222 #define IDX_KOI8_U 16 | |
223 {"koi8-u", ENC_8BIT, 0}, | |
224 #define IDX_UTF8 17 | |
225 {"utf-8", ENC_UNICODE, 0}, | |
226 #define IDX_UCS2 18 | |
227 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0}, | |
228 #define IDX_UCS2LE 19 | |
229 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0}, | |
230 #define IDX_UTF16 20 | |
231 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0}, | |
232 #define IDX_UTF16LE 21 | |
233 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0}, | |
234 #define IDX_UCS4 22 | |
235 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0}, | |
236 #define IDX_UCS4LE 23 | |
237 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0}, | |
238 #define IDX_DEBUG 24 | |
239 {"debug", ENC_DBCS, DBCS_DEBUG}, | |
240 #define IDX_CP932 25 | |
241 {"cp932", ENC_DBCS, DBCS_JPN}, | |
242 #define IDX_CP949 26 | |
243 {"cp949", ENC_DBCS, DBCS_KOR}, | |
244 #define IDX_CP936 27 | |
245 {"cp936", ENC_DBCS, DBCS_CHS}, | |
246 #define IDX_CP950 28 | |
247 {"cp950", ENC_DBCS, DBCS_CHT}, | |
248 #define IDX_EUC_JP 29 | |
249 {"euc-jp", ENC_DBCS, DBCS_JPNU}, | |
250 #define IDX_SJIS 30 | |
251 {"sjis", ENC_DBCS, DBCS_JPN}, | |
252 #define IDX_EUC_KR 31 | |
253 {"euc-kr", ENC_DBCS, DBCS_KORU}, | |
254 #define IDX_EUC_CN 32 | |
255 {"euc-cn", ENC_DBCS, DBCS_CHSU}, | |
256 #define IDX_EUC_TW 33 | |
257 {"euc-tw", ENC_DBCS, DBCS_CHTU}, | |
258 #define IDX_BIG5 34 | |
259 {"big5", ENC_DBCS, DBCS_CHT}, | |
260 #define IDX_CP1251 35 | |
261 {"cp1251", ENC_8BIT, 1251}, | |
262 #define IDX_MACROMAN 36 | |
263 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, | |
264 #define IDX_COUNT 37 | |
265 }; | |
266 | |
267 /* | |
268 * Aliases for encoding names. | |
269 */ | |
270 static struct | |
271 { char *name; int canon;} | |
272 enc_alias_table[] = | |
273 { | |
274 {"ansi", IDX_LATIN_1}, | |
275 {"iso-8859-1", IDX_LATIN_1}, | |
276 {"latin2", IDX_ISO_2}, | |
277 {"latin3", IDX_ISO_3}, | |
278 {"latin4", IDX_ISO_4}, | |
279 {"cyrillic", IDX_ISO_5}, | |
280 {"arabic", IDX_ISO_6}, | |
281 {"greek", IDX_ISO_7}, | |
282 #ifdef WIN3264 | |
283 {"hebrew", IDX_CP1255}, | |
284 #else | |
285 {"hebrew", IDX_ISO_8}, | |
286 #endif | |
287 {"latin5", IDX_ISO_9}, | |
288 {"turkish", IDX_ISO_9}, /* ? */ | |
289 {"latin6", IDX_ISO_10}, | |
290 {"nordic", IDX_ISO_10}, /* ? */ | |
291 {"thai", IDX_ISO_11}, /* ? */ | |
292 {"latin7", IDX_ISO_13}, | |
293 {"latin8", IDX_ISO_14}, | |
294 {"latin9", IDX_ISO_15}, | |
295 {"utf8", IDX_UTF8}, | |
296 {"unicode", IDX_UCS2}, | |
297 {"ucs2", IDX_UCS2}, | |
298 {"ucs2be", IDX_UCS2}, | |
299 {"ucs-2be", IDX_UCS2}, | |
300 {"ucs2le", IDX_UCS2LE}, | |
301 {"utf16", IDX_UTF16}, | |
302 {"utf16be", IDX_UTF16}, | |
303 {"utf-16be", IDX_UTF16}, | |
304 {"utf16le", IDX_UTF16LE}, | |
305 {"ucs4", IDX_UCS4}, | |
306 {"ucs4be", IDX_UCS4}, | |
307 {"ucs-4be", IDX_UCS4}, | |
308 {"ucs4le", IDX_UCS4LE}, | |
309 {"932", IDX_CP932}, | |
310 {"949", IDX_CP949}, | |
311 {"936", IDX_CP936}, | |
312 {"950", IDX_CP950}, | |
313 {"eucjp", IDX_EUC_JP}, | |
314 {"unix-jis", IDX_EUC_JP}, | |
315 {"ujis", IDX_EUC_JP}, | |
316 {"shift-jis", IDX_SJIS}, | |
317 {"euckr", IDX_EUC_KR}, | |
318 {"5601", IDX_EUC_KR}, /* Sun: KS C 5601 */ | |
319 {"euccn", IDX_EUC_CN}, | |
320 {"gb2312", IDX_EUC_CN}, | |
321 {"euctw", IDX_EUC_TW}, | |
322 #if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS) | |
323 {"japan", IDX_CP932}, | |
324 {"korea", IDX_CP949}, | |
325 {"prc", IDX_CP936}, | |
326 {"chinese", IDX_CP936}, | |
327 {"taiwan", IDX_CP950}, | |
328 {"big5", IDX_CP950}, | |
329 #else | |
330 {"japan", IDX_EUC_JP}, | |
331 {"korea", IDX_EUC_KR}, | |
332 {"prc", IDX_EUC_CN}, | |
333 {"chinese", IDX_EUC_CN}, | |
334 {"taiwan", IDX_EUC_TW}, | |
335 {"cp950", IDX_BIG5}, | |
336 {"950", IDX_BIG5}, | |
337 #endif | |
338 {"mac", IDX_MACROMAN}, | |
339 {NULL, 0} | |
340 }; | |
341 | |
342 #ifndef CP_UTF8 | |
343 # define CP_UTF8 65001 /* magic number from winnls.h */ | |
344 #endif | |
345 | |
346 /* | |
347 * Find encoding "name" in the list of canonical encoding names. | |
348 * Returns -1 if not found. | |
349 */ | |
350 static int | |
351 enc_canon_search(name) | |
352 char_u *name; | |
353 { | |
354 int i; | |
355 | |
356 for (i = 0; i < IDX_COUNT; ++i) | |
357 if (STRCMP(name, enc_canon_table[i].name) == 0) | |
358 return i; | |
359 return -1; | |
360 } | |
361 | |
362 #endif | |
363 | |
364 #if defined(FEAT_MBYTE) || defined(PROTO) | |
365 | |
366 /* | |
367 * Find canonical encoding "name" in the list and return its properties. | |
368 * Returns 0 if not found. | |
369 */ | |
370 int | |
371 enc_canon_props(name) | |
372 char_u *name; | |
373 { | |
374 int i; | |
375 | |
376 i = enc_canon_search(name); | |
377 if (i >= 0) | |
378 return enc_canon_table[i].prop; | |
379 #ifdef WIN3264 | |
380 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2])) | |
381 { | |
382 CPINFO cpinfo; | |
383 | |
384 /* Get info on this codepage to find out what it is. */ | |
385 if (GetCPInfo(atoi(name + 2), &cpinfo) != 0) | |
386 { | |
387 if (cpinfo.MaxCharSize == 1) /* some single-byte encoding */ | |
388 return ENC_8BIT; | |
389 if (cpinfo.MaxCharSize == 2 | |
390 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) | |
391 /* must be a DBCS encoding */ | |
392 return ENC_DBCS; | |
393 } | |
394 return 0; | |
395 } | |
396 #endif | |
397 if (STRNCMP(name, "2byte-", 6) == 0) | |
398 return ENC_DBCS; | |
399 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0) | |
400 return ENC_8BIT; | |
401 return 0; | |
402 } | |
403 | |
404 /* | |
405 * Set up for using multi-byte characters. | |
406 * Called in three cases: | |
407 * - by main() to initialize (p_enc == NULL) | |
408 * - by set_init_1() after 'encoding' was set to its default. | |
409 * - by do_set() when 'encoding' has been set. | |
410 * p_enc must have been passed through enc_canonize() already. | |
411 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags. | |
412 * Fills mb_bytelen_tab[] and returns NULL when there are no problems. | |
413 * When there is something wrong: Returns an error message and doesn't change | |
414 * anything. | |
415 */ | |
416 char_u * | |
417 mb_init() | |
418 { | |
419 int i; | |
420 int idx; | |
421 int n; | |
422 int enc_dbcs_new = 0; | |
423 #if defined(USE_ICONV) && !defined(WIN3264) && !defined(WIN32UNIX) \ | |
424 && !defined(MACOS) | |
425 # define LEN_FROM_CONV | |
426 vimconv_T vimconv; | |
427 char_u *p; | |
428 #endif | |
429 | |
430 if (p_enc == NULL) | |
431 { | |
432 /* Just starting up: set the whole table to one's. */ | |
433 for (i = 0; i < 256; ++i) | |
434 mb_bytelen_tab[i] = 1; | |
435 input_conv.vc_type = CONV_NONE; | |
436 input_conv.vc_factor = 1; | |
437 output_conv.vc_type = CONV_NONE; | |
438 return NULL; | |
439 } | |
440 | |
441 #ifdef WIN3264 | |
442 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2])) | |
443 { | |
444 CPINFO cpinfo; | |
445 | |
446 /* Get info on this codepage to find out what it is. */ | |
447 if (GetCPInfo(atoi(p_enc + 2), &cpinfo) != 0) | |
448 { | |
449 if (cpinfo.MaxCharSize == 1) | |
450 { | |
451 /* some single-byte encoding */ | |
452 enc_unicode = 0; | |
453 enc_utf8 = FALSE; | |
454 } | |
455 else if (cpinfo.MaxCharSize == 2 | |
456 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) | |
457 { | |
458 /* must be a DBCS encoding, check below */ | |
459 enc_dbcs_new = atoi(p_enc + 2); | |
460 } | |
461 else | |
462 goto codepage_invalid; | |
463 } | |
464 else if (GetLastError() == ERROR_INVALID_PARAMETER) | |
465 { | |
466 codepage_invalid: | |
467 return (char_u *)N_("E543: Not a valid codepage"); | |
468 } | |
469 } | |
470 #endif | |
471 else if (STRNCMP(p_enc, "8bit-", 5) == 0 | |
472 || STRNCMP(p_enc, "iso-8859-", 9) == 0) | |
473 { | |
474 /* Accept any "8bit-" or "iso-8859-" name. */ | |
475 enc_unicode = 0; | |
476 enc_utf8 = FALSE; | |
477 } | |
478 else if (STRNCMP(p_enc, "2byte-", 6) == 0) | |
479 { | |
480 #ifdef WIN3264 | |
481 /* Windows: accept only valid codepage numbers, check below. */ | |
482 if (p_enc[6] != 'c' || p_enc[7] != 'p' | |
483 || (enc_dbcs_new = atoi(p_enc + 8)) == 0) | |
484 return e_invarg; | |
485 #else | |
486 /* Unix: accept any "2byte-" name, assume current locale. */ | |
487 enc_dbcs_new = DBCS_2BYTE; | |
488 #endif | |
489 } | |
490 else if ((idx = enc_canon_search(p_enc)) >= 0) | |
491 { | |
492 i = enc_canon_table[idx].prop; | |
493 if (i & ENC_UNICODE) | |
494 { | |
495 /* Unicode */ | |
496 enc_utf8 = TRUE; | |
497 if (i & (ENC_2BYTE | ENC_2WORD)) | |
498 enc_unicode = 2; | |
499 else if (i & ENC_4BYTE) | |
500 enc_unicode = 4; | |
501 else | |
502 enc_unicode = 0; | |
503 } | |
504 else if (i & ENC_DBCS) | |
505 { | |
506 /* 2byte, handle below */ | |
507 enc_dbcs_new = enc_canon_table[idx].codepage; | |
508 } | |
509 else | |
510 { | |
511 /* Must be 8-bit. */ | |
512 enc_unicode = 0; | |
513 enc_utf8 = FALSE; | |
514 } | |
515 } | |
516 else /* Don't know what encoding this is, reject it. */ | |
517 return e_invarg; | |
518 | |
519 if (enc_dbcs_new != 0) | |
520 { | |
521 #ifdef WIN3264 | |
522 /* Check if the DBCS code page is OK. */ | |
523 if (!IsValidCodePage(enc_dbcs_new)) | |
524 goto codepage_invalid; | |
525 #endif | |
526 enc_unicode = 0; | |
527 enc_utf8 = FALSE; | |
528 } | |
529 enc_dbcs = enc_dbcs_new; | |
530 has_mbyte = (enc_dbcs != 0 || enc_utf8); | |
531 | |
532 #ifdef WIN3264 | |
533 enc_codepage = encname2codepage(p_enc); | |
534 #endif | |
535 | |
536 /* | |
537 * Set the function pointers. | |
538 */ | |
539 if (enc_utf8) | |
540 { | |
541 mb_ptr2len_check = utfc_ptr2len_check; | |
542 mb_char2len = utf_char2len; | |
543 mb_char2bytes = utf_char2bytes; | |
544 mb_ptr2cells = utf_ptr2cells; | |
545 mb_char2cells = utf_char2cells; | |
546 mb_off2cells = utf_off2cells; | |
547 mb_ptr2char = utf_ptr2char; | |
548 mb_head_off = utf_head_off; | |
549 } | |
550 else if (enc_dbcs != 0) | |
551 { | |
552 mb_ptr2len_check = dbcs_ptr2len_check; | |
553 mb_char2len = dbcs_char2len; | |
554 mb_char2bytes = dbcs_char2bytes; | |
555 mb_ptr2cells = dbcs_ptr2cells; | |
556 mb_char2cells = dbcs_char2cells; | |
557 mb_off2cells = dbcs_off2cells; | |
558 mb_ptr2char = dbcs_ptr2char; | |
559 mb_head_off = dbcs_head_off; | |
560 } | |
561 else | |
562 { | |
563 mb_ptr2len_check = latin_ptr2len_check; | |
564 mb_char2len = latin_char2len; | |
565 mb_char2bytes = latin_char2bytes; | |
566 mb_ptr2cells = latin_ptr2cells; | |
567 mb_char2cells = latin_char2cells; | |
568 mb_off2cells = latin_off2cells; | |
569 mb_ptr2char = latin_ptr2char; | |
570 mb_head_off = latin_head_off; | |
571 } | |
572 | |
573 /* | |
574 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN(). | |
575 */ | |
576 #ifdef LEN_FROM_CONV | |
577 /* When 'encoding' is different from the current locale mblen() won't | |
578 * work. Use conversion to "utf-8" instead. */ | |
579 vimconv.vc_type = CONV_NONE; | |
580 if (enc_dbcs) | |
581 { | |
582 p = enc_locale(); | |
583 if (p == NULL || STRCMP(p, p_enc) != 0) | |
584 { | |
585 convert_setup(&vimconv, p_enc, (char_u *)"utf-8"); | |
586 vimconv.vc_fail = TRUE; | |
587 } | |
588 vim_free(p); | |
589 } | |
590 #endif | |
591 | |
592 for (i = 0; i < 256; ++i) | |
593 { | |
594 /* Our own function to reliably check the length of UTF-8 characters, | |
595 * independent of mblen(). */ | |
596 if (enc_utf8) | |
597 n = utf8len_tab[i]; | |
598 else if (enc_dbcs == 0) | |
599 n = 1; | |
600 else | |
601 { | |
602 #if defined(WIN3264) || defined(WIN32UNIX) | |
603 /* enc_dbcs is set by setting 'fileencoding'. It becomes a Windows | |
604 * CodePage identifier, which we can pass directly in to Windows | |
605 * API */ | |
606 n = IsDBCSLeadByteEx(enc_dbcs, (BYTE)i) ? 2 : 1; | |
607 #else | |
608 # ifdef MACOS | |
609 /* | |
610 * if mblen() is not available, character which MSB is turned on | |
611 * are treated as leading byte character. (note : This assumption | |
612 * is not always true.) | |
613 */ | |
614 n = (i & 0x80) ? 2 : 1; | |
615 # else | |
616 char buf[MB_MAXBYTES]; | |
617 # ifdef X_LOCALE | |
618 # ifndef mblen | |
619 # define mblen _Xmblen | |
620 # endif | |
621 # endif | |
622 if (i == NUL) /* just in case mblen() can't handle "" */ | |
623 n = 1; | |
624 else | |
625 { | |
626 buf[0] = i; | |
627 buf[1] = 0; | |
628 #ifdef LEN_FROM_CONV | |
629 if (vimconv.vc_type != CONV_NONE) | |
630 { | |
631 /* | |
632 * string_convert() should fail when converting the first | |
633 * byte of a double-byte character. | |
634 */ | |
635 p = string_convert(&vimconv, (char_u *)buf, NULL); | |
636 if (p != NULL) | |
637 { | |
638 vim_free(p); | |
639 n = 1; | |
640 } | |
641 else | |
642 n = 2; | |
643 } | |
644 else | |
645 #endif | |
646 { | |
647 /* | |
648 * mblen() should return -1 for invalid (means the leading | |
649 * multibyte) character. However there are some platforms | |
650 * where mblen() returns 0 for invalid character. | |
651 * Therefore, following condition includes 0. | |
652 */ | |
653 if (mblen(buf, (size_t)1) <= 0) | |
654 n = 2; | |
655 else | |
656 n = 1; | |
657 } | |
658 } | |
659 # endif | |
660 #endif | |
661 } | |
662 | |
663 mb_bytelen_tab[i] = n; | |
664 } | |
665 | |
666 #ifdef LEN_FROM_CONV | |
667 convert_setup(&vimconv, NULL, NULL); | |
668 #endif | |
669 | |
670 /* The cell width depends on the type of multi-byte characters. */ | |
671 (void)init_chartab(); | |
672 | |
673 /* When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] */ | |
674 screenalloc(FALSE); | |
675 | |
676 /* When using Unicode, set default for 'fileencodings'. */ | |
677 if (enc_utf8 && !option_was_set((char_u *)"fencs")) | |
678 set_string_option_direct((char_u *)"fencs", -1, | |
679 (char_u *)"ucs-bom,utf-8,latin1", OPT_FREE); | |
680 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT) | |
681 /* GNU gettext 0.10.37 supports this feature: set the codeset used for | |
682 * translated messages independently from the current locale. */ | |
683 (void)bind_textdomain_codeset(VIMPACKAGE, | |
684 enc_utf8 ? "utf-8" : (char *)p_enc); | |
685 #endif | |
686 | |
687 #ifdef FEAT_AUTOCMD | |
688 /* Fire an autocommand to let people do custom font setup. This must be | |
689 * after Vim has been setup for the new encoding. */ | |
690 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); | |
691 #endif | |
692 | |
693 return NULL; | |
694 } | |
695 | |
696 /* | |
697 * Return the size of the BOM for the current buffer: | |
698 * 0 - no BOM | |
699 * 2 - UCS-2 or UTF-16 BOM | |
700 * 4 - UCS-4 BOM | |
701 * 3 - UTF-8 BOM | |
702 */ | |
703 int | |
704 bomb_size() | |
705 { | |
706 int n = 0; | |
707 | |
708 if (curbuf->b_p_bomb && !curbuf->b_p_bin) | |
709 { | |
710 if (*curbuf->b_p_fenc == NUL) | |
711 { | |
712 if (enc_utf8) | |
713 { | |
714 if (enc_unicode != 0) | |
715 n = enc_unicode; | |
716 else | |
717 n = 3; | |
718 } | |
719 } | |
720 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) | |
721 n = 3; | |
722 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 | |
723 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) | |
724 n = 2; | |
725 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) | |
726 n = 4; | |
727 } | |
728 return n; | |
729 } | |
730 | |
731 /* | |
732 * Get class of pointer: | |
733 * 0 for blank or NUL | |
734 * 1 for punctuation | |
735 * 2 for an (ASCII) word character | |
736 * >2 for other word characters | |
737 */ | |
738 int | |
739 mb_get_class(p) | |
740 char_u *p; | |
741 { | |
742 if (MB_BYTE2LEN(p[0]) == 1) | |
743 { | |
744 if (p[0] == NUL || vim_iswhite(p[0])) | |
745 return 0; | |
746 if (vim_iswordc(p[0])) | |
747 return 2; | |
748 return 1; | |
749 } | |
750 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) | |
751 return dbcs_class(p[0], p[1]); | |
752 if (enc_utf8) | |
753 return utf_class(utf_ptr2char(p)); | |
754 return 0; | |
755 } | |
756 | |
757 /* | |
758 * Get class of a double-byte character. This always returns 3 or bigger. | |
759 * TODO: Should return 1 for punctuation. | |
760 */ | |
761 int | |
762 dbcs_class(lead, trail) | |
763 unsigned lead; | |
764 unsigned trail; | |
765 { | |
766 switch (enc_dbcs) | |
767 { | |
768 /* please add classfy routine for your language in here */ | |
769 | |
770 case DBCS_JPNU: /* ? */ | |
771 case DBCS_JPN: | |
772 { | |
773 /* JIS code classification */ | |
774 unsigned char lb = lead; | |
775 unsigned char tb = trail; | |
776 | |
777 /* convert process code to JIS */ | |
778 # if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS) | |
779 /* process code is SJIS */ | |
780 if (lb <= 0x9f) | |
781 lb = (lb - 0x81) * 2 + 0x21; | |
782 else | |
783 lb = (lb - 0xc1) * 2 + 0x21; | |
784 if (tb <= 0x7e) | |
785 tb -= 0x1f; | |
786 else if (tb <= 0x9e) | |
787 tb -= 0x20; | |
788 else | |
789 { | |
790 tb -= 0x7e; | |
791 lb += 1; | |
792 } | |
793 # else | |
794 /* | |
795 * XXX: Code page identification can not use with all | |
796 * system! So, some other encoding information | |
797 * will be needed. | |
798 * In japanese: SJIS,EUC,UNICODE,(JIS) | |
799 * Note that JIS-code system don't use as | |
800 * process code in most system because it uses | |
801 * escape sequences(JIS is context depend encoding). | |
802 */ | |
803 /* assume process code is JAPANESE-EUC */ | |
804 lb &= 0x7f; | |
805 tb &= 0x7f; | |
806 # endif | |
807 /* exceptions */ | |
808 switch (lb << 8 | tb) | |
809 { | |
810 case 0x2121: /* ZENKAKU space */ | |
811 return 0; | |
812 case 0x2122: /* KU-TEN (Japanese comma) */ | |
813 case 0x2123: /* TOU-TEN (Japanese period) */ | |
814 case 0x2124: /* ZENKAKU comma */ | |
815 case 0x2125: /* ZENKAKU period */ | |
816 return 1; | |
817 case 0x213c: /* prolongedsound handled as KATAKANA */ | |
818 return 13; | |
819 } | |
820 /* sieved by KU code */ | |
821 switch (lb) | |
822 { | |
823 case 0x21: | |
824 case 0x22: | |
825 /* special symbols */ | |
826 return 10; | |
827 case 0x23: | |
828 /* alpha-numeric */ | |
829 return 11; | |
830 case 0x24: | |
831 /* hiragana */ | |
832 return 12; | |
833 case 0x25: | |
834 /* katakana */ | |
835 return 13; | |
836 case 0x26: | |
837 /* greek */ | |
838 return 14; | |
839 case 0x27: | |
840 /* russian */ | |
841 return 15; | |
842 case 0x28: | |
843 /* lines */ | |
844 return 16; | |
845 default: | |
846 /* kanji */ | |
847 return 17; | |
848 } | |
849 } | |
850 | |
851 case DBCS_KORU: /* ? */ | |
852 case DBCS_KOR: | |
853 { | |
854 /* KS code classification */ | |
855 unsigned char c1 = lead; | |
856 unsigned char c2 = trail; | |
857 | |
858 /* | |
859 * 20 : Hangul | |
860 * 21 : Hanja | |
861 * 22 : Symbols | |
862 * 23 : Alpha-numeric/Roman Letter (Full width) | |
863 * 24 : Hangul Letter(Alphabet) | |
864 * 25 : Roman Numeral/Greek Letter | |
865 * 26 : Box Drawings | |
866 * 27 : Unit Symbols | |
867 * 28 : Circled/Parenthesized Letter | |
868 * 29 : Hirigana/Katakana | |
869 * 30 : Cyrillic Letter | |
870 */ | |
871 | |
872 if (c1 >= 0xB0 && c1 <= 0xC8) | |
873 /* Hangul */ | |
874 return 20; | |
875 #if defined(WIN3264) || defined(WIN32UNIX) | |
876 else if (c1 <= 0xA0 || c2 <= 0xA0) | |
877 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */ | |
878 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE | |
879 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 | |
880 */ | |
881 return 20; | |
882 #endif | |
883 | |
884 else if (c1 >= 0xCA && c1 <= 0xFD) | |
885 /* Hanja */ | |
886 return 21; | |
887 else switch (c1) | |
888 { | |
889 case 0xA1: | |
890 case 0xA2: | |
891 /* Symbols */ | |
892 return 22; | |
893 case 0xA3: | |
894 /* Alpha-numeric */ | |
895 return 23; | |
896 case 0xA4: | |
897 /* Hangul Letter(Alphabet) */ | |
898 return 24; | |
899 case 0xA5: | |
900 /* Roman Numeral/Greek Letter */ | |
901 return 25; | |
902 case 0xA6: | |
903 /* Box Drawings */ | |
904 return 26; | |
905 case 0xA7: | |
906 /* Unit Symbols */ | |
907 return 27; | |
908 case 0xA8: | |
909 case 0xA9: | |
910 if (c2 <= 0xAF) | |
911 return 25; /* Roman Letter */ | |
912 else if (c2 >= 0xF6) | |
913 return 22; /* Symbols */ | |
914 else | |
915 /* Circled/Parenthesized Letter */ | |
916 return 28; | |
917 case 0xAA: | |
918 case 0xAB: | |
919 /* Hirigana/Katakana */ | |
920 return 29; | |
921 case 0xAC: | |
922 /* Cyrillic Letter */ | |
923 return 30; | |
924 } | |
925 } | |
926 default: | |
927 break; | |
928 } | |
929 return 3; | |
930 } | |
931 | |
932 /* | |
933 * mb_char2len() function pointer. | |
934 * Return length in bytes of character "c". | |
935 * Returns 1 for a single-byte character. | |
936 */ | |
937 /* ARGSUSED */ | |
938 int | |
939 latin_char2len(c) | |
940 int c; | |
941 { | |
942 return 1; | |
943 } | |
944 | |
945 static int | |
946 dbcs_char2len(c) | |
947 int c; | |
948 { | |
949 if (c >= 0x100) | |
950 return 2; | |
951 return 1; | |
952 } | |
953 | |
954 /* | |
955 * mb_char2bytes() function pointer. | |
956 * Convert a character to its bytes. | |
957 * Returns the length in bytes. | |
958 */ | |
959 int | |
960 latin_char2bytes(c, buf) | |
961 int c; | |
962 char_u *buf; | |
963 { | |
964 buf[0] = c; | |
965 return 1; | |
966 } | |
967 | |
968 static int | |
969 dbcs_char2bytes(c, buf) | |
970 int c; | |
971 char_u *buf; | |
972 { | |
973 if (c >= 0x100) | |
974 { | |
975 buf[0] = (unsigned)c >> 8; | |
976 buf[1] = c; | |
977 return 2; | |
978 } | |
979 buf[0] = c; | |
980 return 1; | |
981 } | |
982 | |
983 /* | |
984 * mb_ptr2len_check() function pointer. | |
985 * Get byte length of character at "*p" but stop at a NUL. | |
986 * For UTF-8 this includes following composing characters. | |
987 * Returns 0 when *p is NUL. | |
988 * | |
989 */ | |
990 int | |
991 latin_ptr2len_check(p) | |
992 char_u *p; | |
993 { | |
994 return MB_BYTE2LEN(*p); | |
995 } | |
996 | |
997 static int | |
998 dbcs_ptr2len_check(p) | |
999 char_u *p; | |
1000 { | |
1001 int len; | |
1002 | |
1003 /* Check if second byte is not missing. */ | |
1004 len = MB_BYTE2LEN(*p); | |
1005 if (len == 2 && p[1] == NUL) | |
1006 len = 1; | |
1007 return len; | |
1008 } | |
1009 | |
1010 struct interval | |
1011 { | |
1012 unsigned short first; | |
1013 unsigned short last; | |
1014 }; | |
1015 static int intable __ARGS((struct interval *table, size_t size, int c)); | |
1016 | |
1017 /* | |
1018 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]". | |
1019 */ | |
1020 static int | |
1021 intable(table, size, c) | |
1022 struct interval *table; | |
1023 size_t size; | |
1024 int c; | |
1025 { | |
1026 int mid, bot, top; | |
1027 | |
1028 /* first quick check for Latin1 etc. characters */ | |
1029 if (c < table[0].first) | |
1030 return FALSE; | |
1031 | |
1032 /* binary search in table */ | |
1033 bot = 0; | |
1034 top = size / sizeof(struct interval) - 1; | |
1035 while (top >= bot) | |
1036 { | |
1037 mid = (bot + top) / 2; | |
1038 if (table[mid].last < c) | |
1039 bot = mid + 1; | |
1040 else if (table[mid].first > c) | |
1041 top = mid - 1; | |
1042 else | |
1043 return TRUE; | |
1044 } | |
1045 return FALSE; | |
1046 } | |
1047 | |
1048 /* | |
1049 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. | |
1050 * Returns 4 or 6 for an unprintable character. | |
1051 * Is only correct for characters >= 0x80. | |
1052 * When p_ambw is "double", return 2 for a character with East Asian Width | |
1053 * class 'A'(mbiguous). | |
1054 */ | |
1055 int | |
1056 utf_char2cells(c) | |
1057 int c; | |
1058 { | |
1059 /* sorted list of non-overlapping intervals of East Asian Ambiguous | |
1060 * characters, generated with: | |
1061 * "uniset +WIDTH-A -cat=Me -cat=Mn -cat=Cf c" */ | |
1062 static struct interval ambiguous[] = { | |
1063 {0x00A1, 0x00A1}, {0x00A4, 0x00A4}, {0x00A7, 0x00A8}, | |
1064 {0x00AA, 0x00AA}, {0x00AE, 0x00AE}, {0x00B0, 0x00B4}, | |
1065 {0x00B6, 0x00BA}, {0x00BC, 0x00BF}, {0x00C6, 0x00C6}, | |
1066 {0x00D0, 0x00D0}, {0x00D7, 0x00D8}, {0x00DE, 0x00E1}, | |
1067 {0x00E6, 0x00E6}, {0x00E8, 0x00EA}, {0x00EC, 0x00ED}, | |
1068 {0x00F0, 0x00F0}, {0x00F2, 0x00F3}, {0x00F7, 0x00FA}, | |
1069 {0x00FC, 0x00FC}, {0x00FE, 0x00FE}, {0x0101, 0x0101}, | |
1070 {0x0111, 0x0111}, {0x0113, 0x0113}, {0x011B, 0x011B}, | |
1071 {0x0126, 0x0127}, {0x012B, 0x012B}, {0x0131, 0x0133}, | |
1072 {0x0138, 0x0138}, {0x013F, 0x0142}, {0x0144, 0x0144}, | |
1073 {0x0148, 0x014B}, {0x014D, 0x014D}, {0x0152, 0x0153}, | |
1074 {0x0166, 0x0167}, {0x016B, 0x016B}, {0x01CE, 0x01CE}, | |
1075 {0x01D0, 0x01D0}, {0x01D2, 0x01D2}, {0x01D4, 0x01D4}, | |
1076 {0x01D6, 0x01D6}, {0x01D8, 0x01D8}, {0x01DA, 0x01DA}, | |
1077 {0x01DC, 0x01DC}, {0x0251, 0x0251}, {0x0261, 0x0261}, | |
1078 {0x02C4, 0x02C4}, {0x02C7, 0x02C7}, {0x02C9, 0x02CB}, | |
1079 {0x02CD, 0x02CD}, {0x02D0, 0x02D0}, {0x02D8, 0x02DB}, | |
1080 {0x02DD, 0x02DD}, {0x02DF, 0x02DF}, {0x0391, 0x03A1}, | |
1081 {0x03A3, 0x03A9}, {0x03B1, 0x03C1}, {0x03C3, 0x03C9}, | |
1082 {0x0401, 0x0401}, {0x0410, 0x044F}, {0x0451, 0x0451}, | |
1083 {0x2010, 0x2010}, {0x2013, 0x2016}, {0x2018, 0x2019}, | |
1084 {0x201C, 0x201D}, {0x2020, 0x2022}, {0x2024, 0x2027}, | |
1085 {0x2030, 0x2030}, {0x2032, 0x2033}, {0x2035, 0x2035}, | |
1086 {0x203B, 0x203B}, {0x203E, 0x203E}, {0x2074, 0x2074}, | |
1087 {0x207F, 0x207F}, {0x2081, 0x2084}, {0x20AC, 0x20AC}, | |
1088 {0x2103, 0x2103}, {0x2105, 0x2105}, {0x2109, 0x2109}, | |
1089 {0x2113, 0x2113}, {0x2116, 0x2116}, {0x2121, 0x2122}, | |
1090 {0x2126, 0x2126}, {0x212B, 0x212B}, {0x2153, 0x2154}, | |
1091 {0x215B, 0x215E}, {0x2160, 0x216B}, {0x2170, 0x2179}, | |
1092 {0x2190, 0x2199}, {0x21B8, 0x21B9}, {0x21D2, 0x21D2}, | |
1093 {0x21D4, 0x21D4}, {0x21E7, 0x21E7}, {0x2200, 0x2200}, | |
1094 {0x2202, 0x2203}, {0x2207, 0x2208}, {0x220B, 0x220B}, | |
1095 {0x220F, 0x220F}, {0x2211, 0x2211}, {0x2215, 0x2215}, | |
1096 {0x221A, 0x221A}, {0x221D, 0x2220}, {0x2223, 0x2223}, | |
1097 {0x2225, 0x2225}, {0x2227, 0x222C}, {0x222E, 0x222E}, | |
1098 {0x2234, 0x2237}, {0x223C, 0x223D}, {0x2248, 0x2248}, | |
1099 {0x224C, 0x224C}, {0x2252, 0x2252}, {0x2260, 0x2261}, | |
1100 {0x2264, 0x2267}, {0x226A, 0x226B}, {0x226E, 0x226F}, | |
1101 {0x2282, 0x2283}, {0x2286, 0x2287}, {0x2295, 0x2295}, | |
1102 {0x2299, 0x2299}, {0x22A5, 0x22A5}, {0x22BF, 0x22BF}, | |
1103 {0x2312, 0x2312}, {0x2460, 0x24E9}, {0x24EB, 0x254B}, | |
1104 {0x2550, 0x2573}, {0x2580, 0x258F}, {0x2592, 0x2595}, | |
1105 {0x25A0, 0x25A1}, {0x25A3, 0x25A9}, {0x25B2, 0x25B3}, | |
1106 {0x25B6, 0x25B7}, {0x25BC, 0x25BD}, {0x25C0, 0x25C1}, | |
1107 {0x25C6, 0x25C8}, {0x25CB, 0x25CB}, {0x25CE, 0x25D1}, | |
1108 {0x25E2, 0x25E5}, {0x25EF, 0x25EF}, {0x2605, 0x2606}, | |
1109 {0x2609, 0x2609}, {0x260E, 0x260F}, {0x2614, 0x2615}, | |
1110 {0x261C, 0x261C}, {0x261E, 0x261E}, {0x2640, 0x2640}, | |
1111 {0x2642, 0x2642}, {0x2660, 0x2661}, {0x2663, 0x2665}, | |
1112 {0x2667, 0x266A}, {0x266C, 0x266D}, {0x266F, 0x266F}, | |
1113 {0x273D, 0x273D}, {0x2776, 0x277F}, {0xE000, 0xF8FF}, | |
1114 {0xFFFD, 0xFFFD}, /* {0xF0000, 0xFFFFD}, {0x100000, 0x10FFFD} */ | |
1115 }; | |
1116 | |
1117 if (c >= 0x100) | |
1118 { | |
1119 #ifdef USE_WCHAR_FUNCTIONS | |
1120 /* | |
1121 * Assume the library function wcwidth() works better than our own | |
1122 * stuff. It should return 1 for ambiguous width chars! | |
1123 */ | |
1124 int n = wcwidth(c); | |
1125 | |
1126 if (n < 0) | |
1127 return 6; /* unprintable, displays <xxxx> */ | |
1128 if (n > 1) | |
1129 return n; | |
1130 #else | |
1131 if (!utf_printable(c)) | |
1132 return 6; /* unprintable, displays <xxxx> */ | |
1133 if (c >= 0x1100 | |
1134 && (c <= 0x115f /* Hangul Jamo */ | |
1135 || c == 0x2329 | |
1136 || c == 0x232a | |
1137 || (c >= 0x2e80 && c <= 0xa4cf | |
1138 && c != 0x303f) /* CJK ... Yi */ | |
1139 || (c >= 0xac00 && c <= 0xd7a3) /* Hangul Syllables */ | |
1140 || (c >= 0xf900 && c <= 0xfaff) /* CJK Compatibility | |
1141 Ideographs */ | |
1142 || (c >= 0xfe30 && c <= 0xfe6f) /* CJK Compatibility Forms */ | |
1143 || (c >= 0xff00 && c <= 0xff60) /* Fullwidth Forms */ | |
1144 || (c >= 0xffe0 && c <= 0xffe6) | |
1145 || (c >= 0x20000 && c <= 0x2fffd) | |
1146 || (c >= 0x30000 && c <= 0x3fffd))) | |
1147 return 2; | |
1148 #endif | |
1149 } | |
1150 | |
1151 /* Characters below 0x100 are influenced by 'isprint' option */ | |
1152 else if (c >= 0x80 && !vim_isprintc(c)) | |
1153 return 4; /* unprintable, displays <xx> */ | |
1154 | |
1155 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c)) | |
1156 return 2; | |
1157 | |
1158 return 1; | |
1159 } | |
1160 | |
1161 /* | |
1162 * mb_ptr2cells() function pointer. | |
1163 * Return the number of display cells character at "*p" occupies. | |
1164 * This doesn't take care of unprintable characters, use ptr2cells() for that. | |
1165 */ | |
1166 /*ARGSUSED*/ | |
1167 int | |
1168 latin_ptr2cells(p) | |
1169 char_u *p; | |
1170 { | |
1171 return 1; | |
1172 } | |
1173 | |
1174 int | |
1175 utf_ptr2cells(p) | |
1176 char_u *p; | |
1177 { | |
1178 int c; | |
1179 | |
1180 /* Need to convert to a wide character. */ | |
1181 if (*p >= 0x80) | |
1182 { | |
1183 c = utf_ptr2char(p); | |
1184 /* An illegal byte is displayed as <xx>. */ | |
1185 if (utf_ptr2len_check(p) == 1 || c == NUL) | |
1186 return 4; | |
1187 /* If the char is ASCII it must be an overlong sequence. */ | |
1188 if (c < 0x80) | |
1189 return char2cells(c); | |
1190 return utf_char2cells(c); | |
1191 } | |
1192 return 1; | |
1193 } | |
1194 | |
1195 int | |
1196 dbcs_ptr2cells(p) | |
1197 char_u *p; | |
1198 { | |
1199 /* Number of cells is equal to number of bytes, except for euc-jp when | |
1200 * the first byte is 0x8e. */ | |
1201 if (enc_dbcs == DBCS_JPNU && *p == 0x8e) | |
1202 return 1; | |
1203 return MB_BYTE2LEN(*p); | |
1204 } | |
1205 | |
1206 /* | |
1207 * mb_char2cells() function pointer. | |
1208 * Return the number of display cells character "c" occupies. | |
1209 * Only takes care of multi-byte chars, not "^C" and such. | |
1210 */ | |
1211 /*ARGSUSED*/ | |
1212 int | |
1213 latin_char2cells(c) | |
1214 int c; | |
1215 { | |
1216 return 1; | |
1217 } | |
1218 | |
1219 static int | |
1220 dbcs_char2cells(c) | |
1221 int c; | |
1222 { | |
1223 /* Number of cells is equal to number of bytes, except for euc-jp when | |
1224 * the first byte is 0x8e. */ | |
1225 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e) | |
1226 return 1; | |
1227 /* use the first byte */ | |
1228 return MB_BYTE2LEN((unsigned)c >> 8); | |
1229 } | |
1230 | |
1231 /* | |
1232 * mb_off2cells() function pointer. | |
1233 * Return number of display cells for char at ScreenLines[off]. | |
1234 * Caller must make sure "off" and "off + 1" are valid! | |
1235 */ | |
1236 /*ARGSUSED*/ | |
1237 int | |
1238 latin_off2cells(off) | |
1239 unsigned off; | |
1240 { | |
1241 return 1; | |
1242 } | |
1243 | |
1244 int | |
1245 dbcs_off2cells(off) | |
1246 unsigned off; | |
1247 { | |
1248 /* Number of cells is equal to number of bytes, except for euc-jp when | |
1249 * the first byte is 0x8e. */ | |
1250 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e) | |
1251 return 1; | |
1252 return MB_BYTE2LEN(ScreenLines[off]); | |
1253 } | |
1254 | |
1255 int | |
1256 utf_off2cells(off) | |
1257 unsigned off; | |
1258 { | |
1259 return ScreenLines[off + 1] == 0 ? 2 : 1; | |
1260 } | |
1261 | |
1262 /* | |
1263 * mb_ptr2char() function pointer. | |
1264 * Convert a byte sequence into a character. | |
1265 */ | |
1266 int | |
1267 latin_ptr2char(p) | |
1268 char_u *p; | |
1269 { | |
1270 return *p; | |
1271 } | |
1272 | |
1273 static int | |
1274 dbcs_ptr2char(p) | |
1275 char_u *p; | |
1276 { | |
1277 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL) | |
1278 return (p[0] << 8) + p[1]; | |
1279 return *p; | |
1280 } | |
1281 | |
1282 /* | |
1283 * Convert a UTF-8 byte sequence to a wide character. | |
1284 * If the sequence is illegal or truncated by a NUL the first byte is | |
1285 * returned. | |
1286 * Does not include composing characters, of course. | |
1287 */ | |
1288 int | |
1289 utf_ptr2char(p) | |
1290 char_u *p; | |
1291 { | |
1292 int len; | |
1293 | |
1294 if (p[0] < 0x80) /* be quick for ASCII */ | |
1295 return p[0]; | |
1296 | |
1297 len = utf8len_tab[p[0]]; | |
1298 if ((p[1] & 0xc0) == 0x80) | |
1299 { | |
1300 if (len == 2) | |
1301 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); | |
1302 if ((p[2] & 0xc0) == 0x80) | |
1303 { | |
1304 if (len == 3) | |
1305 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) | |
1306 + (p[2] & 0x3f); | |
1307 if ((p[3] & 0xc0) == 0x80) | |
1308 { | |
1309 if (len == 4) | |
1310 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) | |
1311 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); | |
1312 if ((p[4] & 0xc0) == 0x80) | |
1313 { | |
1314 if (len == 5) | |
1315 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) | |
1316 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) | |
1317 + (p[4] & 0x3f); | |
1318 if ((p[5] & 0xc0) == 0x80 && len == 6) | |
1319 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) | |
1320 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) | |
1321 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); | |
1322 } | |
1323 } | |
1324 } | |
1325 } | |
1326 /* Illegal value, just return the first byte */ | |
1327 return p[0]; | |
1328 } | |
1329 | |
1330 /* | |
1331 * Get character at **pp and advance *pp to the next character. | |
1332 * Note: composing characters are skipped! | |
1333 */ | |
1334 int | |
1335 mb_ptr2char_adv(pp) | |
1336 char_u **pp; | |
1337 { | |
1338 int c; | |
1339 | |
1340 c = (*mb_ptr2char)(*pp); | |
1341 *pp += (*mb_ptr2len_check)(*pp); | |
1342 return c; | |
1343 } | |
1344 | |
1345 #if defined(FEAT_ARABIC) || defined(PROTO) | |
1346 /* | |
1347 * Check whether we are dealing with Arabic combining characters. | |
1348 * Note: these are NOT really composing characters! | |
1349 */ | |
1350 int | |
1351 arabic_combine(one, two) | |
1352 int one; /* first character */ | |
1353 int two; /* character just after "one" */ | |
1354 { | |
1355 if (one == a_LAM) | |
1356 return arabic_maycombine(two); | |
1357 return FALSE; | |
1358 } | |
1359 | |
1360 /* | |
1361 * Check whether we are dealing with a character that could be regarded as an | |
1362 * Arabic combining character, need to check the character before this. | |
1363 */ | |
1364 int | |
1365 arabic_maycombine(two) | |
1366 int two; | |
1367 { | |
1368 if (p_arshape && !p_tbidi) | |
1369 return (two == a_ALEF_MADDA | |
1370 || two == a_ALEF_HAMZA_ABOVE | |
1371 || two == a_ALEF_HAMZA_BELOW | |
1372 || two == a_ALEF); | |
1373 return FALSE; | |
1374 } | |
1375 | |
1376 /* | |
1377 * Check if the character pointed to by "p2" is a composing character when it | |
1378 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which | |
1379 * behaves like a composing character. | |
1380 */ | |
1381 int | |
1382 utf_composinglike(p1, p2) | |
1383 char_u *p1; | |
1384 char_u *p2; | |
1385 { | |
1386 int c2; | |
1387 | |
1388 c2 = utf_ptr2char(p2); | |
1389 if (utf_iscomposing(c2)) | |
1390 return TRUE; | |
1391 if (!arabic_maycombine(c2)) | |
1392 return FALSE; | |
1393 return arabic_combine(utf_ptr2char(p1), c2); | |
1394 } | |
1395 #endif | |
1396 | |
1397 /* | |
1398 * Convert a UTF-8 byte string to a wide chararacter. Also get up to two | |
1399 * composing characters. | |
1400 */ | |
1401 int | |
1402 utfc_ptr2char(p, p1, p2) | |
1403 char_u *p; | |
1404 int *p1; /* return: first composing char or 0 */ | |
1405 int *p2; /* return: second composing char or 0 */ | |
1406 { | |
1407 int len; | |
1408 int c; | |
1409 int cc; | |
1410 | |
1411 c = utf_ptr2char(p); | |
1412 len = utf_ptr2len_check(p); | |
1413 /* Only accept a composing char when the first char isn't illegal. */ | |
1414 if ((len > 1 || *p < 0x80) | |
1415 && p[len] >= 0x80 | |
1416 && UTF_COMPOSINGLIKE(p, p + len)) | |
1417 { | |
1418 *p1 = utf_ptr2char(p + len); | |
1419 len += utf_ptr2len_check(p + len); | |
1420 if (p[len] >= 0x80 && utf_iscomposing(cc = utf_ptr2char(p + len))) | |
1421 *p2 = cc; | |
1422 else | |
1423 *p2 = 0; | |
1424 } | |
1425 else | |
1426 { | |
1427 *p1 = 0; | |
1428 *p2 = 0; | |
1429 } | |
1430 return c; | |
1431 } | |
1432 | |
1433 /* | |
1434 * Convert a UTF-8 byte string to a wide chararacter. Also get up to two | |
1435 * composing characters. Use no more than p[maxlen]. | |
1436 */ | |
1437 int | |
1438 utfc_ptr2char_len(p, p1, p2, maxlen) | |
1439 char_u *p; | |
1440 int *p1; /* return: first composing char or 0 */ | |
1441 int *p2; /* return: second composing char or 0 */ | |
1442 int maxlen; | |
1443 { | |
1444 int len; | |
1445 int c; | |
1446 int cc; | |
1447 | |
1448 c = utf_ptr2char(p); | |
1449 len = utf_ptr2len_check_len(p, maxlen); | |
1450 /* Only accept a composing char when the first char isn't illegal. */ | |
1451 if ((len > 1 || *p < 0x80) | |
1452 && len < maxlen | |
1453 && p[len] >= 0x80 | |
1454 && UTF_COMPOSINGLIKE(p, p + len)) | |
1455 { | |
1456 *p1 = utf_ptr2char(p + len); | |
1457 len += utf_ptr2len_check_len(p + len, maxlen - len); | |
1458 if (len < maxlen | |
1459 && p[len] >= 0x80 | |
1460 && utf_iscomposing(cc = utf_ptr2char(p + len))) | |
1461 *p2 = cc; | |
1462 else | |
1463 *p2 = 0; | |
1464 } | |
1465 else | |
1466 { | |
1467 *p1 = 0; | |
1468 *p2 = 0; | |
1469 } | |
1470 return c; | |
1471 } | |
1472 | |
1473 /* | |
1474 * Convert the character at screen position "off" to a sequence of bytes. | |
1475 * Includes the composing characters. | |
1476 * "buf" must at least have the length MB_MAXBYTES. | |
1477 * Returns the produced number of bytes. | |
1478 */ | |
1479 int | |
1480 utfc_char2bytes(off, buf) | |
1481 int off; | |
1482 char_u *buf; | |
1483 { | |
1484 int len; | |
1485 | |
1486 len = utf_char2bytes(ScreenLinesUC[off], buf); | |
1487 if (ScreenLinesC1[off] != 0) | |
1488 { | |
1489 len += utf_char2bytes(ScreenLinesC1[off], buf + len); | |
1490 if (ScreenLinesC2[off] != 0) | |
1491 len += utf_char2bytes(ScreenLinesC2[off], buf + len); | |
1492 } | |
1493 return len; | |
1494 } | |
1495 | |
1496 /* | |
1497 * Get the length of a UTF-8 byte sequence, not including any following | |
1498 * composing characters. | |
1499 * Returns 0 for "". | |
1500 * Returns 1 for an illegal byte sequence. | |
1501 */ | |
1502 int | |
1503 utf_ptr2len_check(p) | |
1504 char_u *p; | |
1505 { | |
1506 int len; | |
1507 int i; | |
1508 | |
1509 if (*p == NUL) | |
1510 return 0; | |
1511 len = utf8len_tab[*p]; | |
1512 for (i = 1; i < len; ++i) | |
1513 if ((p[i] & 0xc0) != 0x80) | |
1514 return 1; | |
1515 return len; | |
1516 } | |
1517 | |
1518 /* | |
1519 * Return length of UTF-8 character, obtained from the first byte. | |
1520 * "b" must be between 0 and 255! | |
1521 */ | |
1522 int | |
1523 utf_byte2len(b) | |
1524 int b; | |
1525 { | |
1526 return utf8len_tab[b]; | |
1527 } | |
1528 | |
1529 /* | |
1530 * Get the length of UTF-8 byte sequence "p[size]". Does not include any | |
1531 * following composing characters. | |
1532 * Returns 1 for "". | |
1533 * Returns 1 for an illegal byte sequence. | |
1534 * Returns number > "size" for an incomplete byte sequence. | |
1535 */ | |
1536 int | |
1537 utf_ptr2len_check_len(p, size) | |
1538 char_u *p; | |
1539 int size; | |
1540 { | |
1541 int len; | |
1542 int i; | |
1543 | |
1544 if (*p == NUL) | |
1545 return 1; | |
1546 len = utf8len_tab[*p]; | |
1547 if (len > size) | |
1548 return len; /* incomplete byte sequence. */ | |
1549 for (i = 1; i < len; ++i) | |
1550 if ((p[i] & 0xc0) != 0x80) | |
1551 return 1; | |
1552 return len; | |
1553 } | |
1554 | |
1555 /* | |
1556 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. | |
1557 * This includes following composing characters. | |
1558 */ | |
1559 int | |
1560 utfc_ptr2len_check(p) | |
1561 char_u *p; | |
1562 { | |
1563 int len; | |
1564 #ifdef FEAT_ARABIC | |
1565 int prevlen; | |
1566 #endif | |
1567 | |
1568 if (*p == NUL) | |
1569 return 0; | |
1570 if (p[0] < 0x80 && p[1] < 0x80) /* be quick for ASCII */ | |
1571 return 1; | |
1572 | |
1573 /* Skip over first UTF-8 char, stopping at a NUL byte. */ | |
1574 len = utf_ptr2len_check(p); | |
1575 | |
1576 /* Check for illegal byte. */ | |
1577 if (len == 1 && p[0] >= 0x80) | |
1578 return 1; | |
1579 | |
1580 /* | |
1581 * Check for composing characters. We can handle only the first two, but | |
1582 * skip all of them (otherwise the cursor would get stuck). | |
1583 */ | |
1584 #ifdef FEAT_ARABIC | |
1585 prevlen = 0; | |
1586 #endif | |
1587 for (;;) | |
1588 { | |
1589 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) | |
1590 return len; | |
1591 | |
1592 /* Skip over composing char */ | |
1593 #ifdef FEAT_ARABIC | |
1594 prevlen = len; | |
1595 #endif | |
1596 len += utf_ptr2len_check(p + len); | |
1597 } | |
1598 } | |
1599 | |
1600 /* | |
1601 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" | |
1602 * takes. This includes following composing characters. | |
1603 * Returns 1 for an illegal char or an incomplete byte sequence. | |
1604 */ | |
1605 int | |
1606 utfc_ptr2len_check_len(p, size) | |
1607 char_u *p; | |
1608 int size; | |
1609 { | |
1610 int len; | |
1611 #ifdef FEAT_ARABIC | |
1612 int prevlen; | |
1613 #endif | |
1614 | |
1615 if (*p == NUL) | |
1616 return 0; | |
1617 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) /* be quick for ASCII */ | |
1618 return 1; | |
1619 | |
1620 /* Skip over first UTF-8 char, stopping at a NUL byte. */ | |
1621 len = utf_ptr2len_check_len(p, size); | |
1622 | |
1623 /* Check for illegal byte and incomplete byte sequence. */ | |
1624 if ((len == 1 && p[0] >= 0x80) || len > size) | |
1625 return 1; | |
1626 | |
1627 /* | |
1628 * Check for composing characters. We can handle only the first two, but | |
1629 * skip all of them (otherwise the cursor would get stuck). | |
1630 */ | |
1631 #ifdef FEAT_ARABIC | |
1632 prevlen = 0; | |
1633 #endif | |
1634 while (len < size) | |
1635 { | |
1636 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) | |
1637 break; | |
1638 | |
1639 /* Skip over composing char */ | |
1640 #ifdef FEAT_ARABIC | |
1641 prevlen = len; | |
1642 #endif | |
1643 len += utf_ptr2len_check_len(p + len, size - len); | |
1644 } | |
1645 return len; | |
1646 } | |
1647 | |
1648 /* | |
1649 * Return the number of bytes the UTF-8 encoding of character "c" takes. | |
1650 * This does not include composing characters. | |
1651 */ | |
1652 int | |
1653 utf_char2len(c) | |
1654 int c; | |
1655 { | |
1656 if (c < 0x80) | |
1657 return 1; | |
1658 if (c < 0x800) | |
1659 return 2; | |
1660 if (c < 0x10000) | |
1661 return 3; | |
1662 if (c < 0x200000) | |
1663 return 4; | |
1664 if (c < 0x4000000) | |
1665 return 5; | |
1666 return 6; | |
1667 } | |
1668 | |
1669 /* | |
1670 * Convert Unicode character "c" to UTF-8 string in "buf[]". | |
1671 * Returns the number of bytes. | |
1672 * This does not include composing characters. | |
1673 */ | |
1674 int | |
1675 utf_char2bytes(c, buf) | |
1676 int c; | |
1677 char_u *buf; | |
1678 { | |
1679 if (c < 0x80) /* 7 bits */ | |
1680 { | |
1681 buf[0] = c; | |
1682 return 1; | |
1683 } | |
1684 if (c < 0x800) /* 11 bits */ | |
1685 { | |
1686 buf[0] = 0xc0 + ((unsigned)c >> 6); | |
1687 buf[1] = 0x80 + (c & 0x3f); | |
1688 return 2; | |
1689 } | |
1690 if (c < 0x10000) /* 16 bits */ | |
1691 { | |
1692 buf[0] = 0xe0 + ((unsigned)c >> 12); | |
1693 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f); | |
1694 buf[2] = 0x80 + (c & 0x3f); | |
1695 return 3; | |
1696 } | |
1697 if (c < 0x200000) /* 21 bits */ | |
1698 { | |
1699 buf[0] = 0xf0 + ((unsigned)c >> 18); | |
1700 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f); | |
1701 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f); | |
1702 buf[3] = 0x80 + (c & 0x3f); | |
1703 return 4; | |
1704 } | |
1705 if (c < 0x4000000) /* 26 bits */ | |
1706 { | |
1707 buf[0] = 0xf8 + ((unsigned)c >> 24); | |
1708 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f); | |
1709 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f); | |
1710 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f); | |
1711 buf[4] = 0x80 + (c & 0x3f); | |
1712 return 5; | |
1713 } | |
1714 /* 31 bits */ | |
1715 buf[0] = 0xfc + ((unsigned)c >> 30); | |
1716 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f); | |
1717 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f); | |
1718 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f); | |
1719 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f); | |
1720 buf[5] = 0x80 + (c & 0x3f); | |
1721 return 6; | |
1722 } | |
1723 | |
1724 /* | |
1725 * Return TRUE if "c" is a composing UTF-8 character. This means it will be | |
1726 * drawn on top of the preceding character. | |
1727 * Based on code from Markus Kuhn. | |
1728 */ | |
1729 int | |
1730 utf_iscomposing(c) | |
1731 int c; | |
1732 { | |
1733 /* sorted list of non-overlapping intervals */ | |
1734 static struct interval combining[] = | |
1735 { | |
1736 {0x0300, 0x034f}, {0x0360, 0x036f}, {0x0483, 0x0486}, {0x0488, 0x0489}, | |
1737 {0x0591, 0x05a1}, {0x05a3, 0x05b9}, {0x05bb, 0x05bd}, {0x05bf, 0x05bf}, | |
1738 {0x05c1, 0x05c2}, {0x05c4, 0x05c4}, {0x0610, 0x0615}, {0x064b, 0x0658}, | |
1739 {0x0670, 0x0670}, {0x06d6, 0x06dc}, {0x06de, 0x06e4}, {0x06e7, 0x06e8}, | |
1740 {0x06ea, 0x06ed}, {0x0711, 0x0711}, {0x0730, 0x074a}, {0x07a6, 0x07b0}, | |
1741 {0x0901, 0x0903}, {0x093c, 0x093c}, {0x093e, 0x094d}, {0x0951, 0x0954}, | |
1742 {0x0962, 0x0963}, {0x0981, 0x0983}, {0x09bc, 0x09bc}, {0x09be, 0x09c4}, | |
1743 {0x09c7, 0x09c8}, {0x09cb, 0x09cd}, {0x09d7, 0x09d7}, {0x09e2, 0x09e3}, | |
1744 {0x0a01, 0x0a03}, {0x0a3c, 0x0a3c}, {0x0a3e, 0x0a42}, {0x0a47, 0x0a48}, | |
1745 {0x0a4b, 0x0a4d}, {0x0a70, 0x0a71}, {0x0a81, 0x0a83}, {0x0abc, 0x0abc}, | |
1746 {0x0abe, 0x0ac5}, {0x0ac7, 0x0ac9}, {0x0acb, 0x0acd}, {0x0ae2, 0x0ae3}, | |
1747 {0x0b01, 0x0b03}, {0x0b3c, 0x0b3c}, {0x0b3e, 0x0b43}, {0x0b47, 0x0b48}, | |
1748 {0x0b4b, 0x0b4d}, {0x0b56, 0x0b57}, {0x0b82, 0x0b82}, {0x0bbe, 0x0bc2}, | |
1749 {0x0bc6, 0x0bc8}, {0x0bca, 0x0bcd}, {0x0bd7, 0x0bd7}, {0x0c01, 0x0c03}, | |
1750 {0x0c3e, 0x0c44}, {0x0c46, 0x0c48}, {0x0c4a, 0x0c4d}, {0x0c55, 0x0c56}, | |
1751 {0x0c82, 0x0c83}, {0x0cbc, 0x0cbc}, {0x0cbe, 0x0cc4}, {0x0cc6, 0x0cc8}, | |
1752 {0x0cca, 0x0ccd}, {0x0cd5, 0x0cd6}, {0x0d02, 0x0d03}, {0x0d3e, 0x0d43}, | |
1753 {0x0d46, 0x0d48}, {0x0d4a, 0x0d4d}, {0x0d57, 0x0d57}, {0x0d82, 0x0d83}, | |
1754 {0x0dca, 0x0dca}, {0x0dcf, 0x0dd4}, {0x0dd6, 0x0dd6}, {0x0dd8, 0x0ddf}, | |
1755 {0x0df2, 0x0df3}, {0x0e31, 0x0e31}, {0x0e34, 0x0e3a}, {0x0e47, 0x0e4e}, | |
1756 {0x0eb1, 0x0eb1}, {0x0eb4, 0x0eb9}, {0x0ebb, 0x0ebc}, {0x0ec8, 0x0ecd}, | |
1757 {0x0f18, 0x0f19}, {0x0f35, 0x0f35}, {0x0f37, 0x0f37}, {0x0f39, 0x0f39}, | |
1758 {0x0f3e, 0x0f3f}, {0x0f71, 0x0f84}, {0x0f86, 0x0f87}, {0x0f90, 0x0f97}, | |
1759 {0x0f99, 0x0fbc}, {0x0fc6, 0x0fc6}, {0x102c, 0x1032}, {0x1036, 0x1039}, | |
1760 {0x1056, 0x1059}, {0x1712, 0x1714}, {0x1732, 0x1734}, {0x1752, 0x1753}, | |
1761 {0x1772, 0x1773}, {0x17b6, 0x17d3}, {0x17dd, 0x17dd}, {0x180b, 0x180d}, | |
1762 {0x18a9, 0x18a9}, {0x1920, 0x192b}, {0x1930, 0x193b}, {0x20d0, 0x20ea}, | |
1763 {0x302a, 0x302f}, {0x3099, 0x309a}, {0xfb1e, 0xfb1e}, {0xfe00, 0xfe0f}, | |
1764 {0xfe20, 0xfe23}, | |
1765 }; | |
1766 | |
1767 return intable(combining, sizeof(combining), c); | |
1768 } | |
1769 | |
1770 /* | |
1771 * Return TRUE for characters that can be displayed in a normal way. | |
1772 * Only for characters of 0x100 and above! | |
1773 */ | |
1774 int | |
1775 utf_printable(c) | |
1776 int c; | |
1777 { | |
1778 #ifdef USE_WCHAR_FUNCTIONS | |
1779 /* | |
1780 * Assume the iswprint() library function works better than our own stuff. | |
1781 */ | |
1782 return iswprint(c); | |
1783 #else | |
1784 /* Sorted list of non-overlapping intervals. | |
1785 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */ | |
1786 static struct interval nonprint[] = | |
1787 { | |
1788 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, | |
1789 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, | |
1790 {0xfffe, 0xffff} | |
1791 }; | |
1792 | |
1793 return !intable(nonprint, sizeof(nonprint), c); | |
1794 #endif | |
1795 } | |
1796 | |
1797 /* | |
1798 * Get class of a Unicode character. | |
1799 * 0: white space | |
1800 * 1: punctuation | |
1801 * 2 or bigger: some class of word character. | |
1802 */ | |
1803 int | |
1804 utf_class(c) | |
1805 int c; | |
1806 { | |
1807 /* sorted list of non-overlapping intervals */ | |
1808 static struct clinterval | |
1809 { | |
1810 unsigned short first; | |
1811 unsigned short last; | |
1812 unsigned short class; | |
1813 } classes[] = | |
1814 { | |
1815 {0x037e, 0x037e, 1}, /* Greek question mark */ | |
1816 {0x0387, 0x0387, 1}, /* Greek ano teleia */ | |
1817 {0x055a, 0x055f, 1}, /* Armenian punctuation */ | |
1818 {0x0589, 0x0589, 1}, /* Armenian full stop */ | |
1819 {0x05be, 0x05be, 1}, | |
1820 {0x05c0, 0x05c0, 1}, | |
1821 {0x05c3, 0x05c3, 1}, | |
1822 {0x05f3, 0x05f4, 1}, | |
1823 {0x060c, 0x060c, 1}, | |
1824 {0x061b, 0x061b, 1}, | |
1825 {0x061f, 0x061f, 1}, | |
1826 {0x066a, 0x066d, 1}, | |
1827 {0x06d4, 0x06d4, 1}, | |
1828 {0x0700, 0x070d, 1}, /* Syriac punctuation */ | |
1829 {0x0964, 0x0965, 1}, | |
1830 {0x0970, 0x0970, 1}, | |
1831 {0x0df4, 0x0df4, 1}, | |
1832 {0x0e4f, 0x0e4f, 1}, | |
1833 {0x0e5a, 0x0e5b, 1}, | |
1834 {0x0f04, 0x0f12, 1}, | |
1835 {0x0f3a, 0x0f3d, 1}, | |
1836 {0x0f85, 0x0f85, 1}, | |
1837 {0x104a, 0x104f, 1}, /* Myanmar punctuation */ | |
1838 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */ | |
1839 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */ | |
1840 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */ | |
1841 {0x1680, 0x1680, 0}, | |
1842 {0x169b, 0x169c, 1}, | |
1843 {0x16eb, 0x16ed, 1}, | |
1844 {0x1735, 0x1736, 1}, | |
1845 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */ | |
1846 {0x1800, 0x180a, 1}, /* Mongolian punctuation */ | |
1847 {0x2000, 0x200b, 0}, /* spaces */ | |
1848 {0x200c, 0x2027, 1}, /* punctuation and symbols */ | |
1849 {0x2028, 0x2029, 0}, | |
1850 {0x202a, 0x202e, 1}, /* punctuation and symbols */ | |
1851 {0x202f, 0x202f, 0}, | |
1852 {0x2030, 0x205e, 1}, /* punctuation and symbols */ | |
1853 {0x205f, 0x205f, 0}, | |
1854 {0x2060, 0x27ff, 1}, /* punctuation and symbols */ | |
1855 {0x2070, 0x207f, 0x2070}, /* superscript */ | |
1856 {0x2080, 0x208f, 0x2080}, /* subscript */ | |
1857 {0x2983, 0x2998, 1}, | |
1858 {0x29d8, 0x29db, 1}, | |
1859 {0x29fc, 0x29fd, 1}, | |
1860 {0x3000, 0x3000, 0}, /* ideographic space */ | |
1861 {0x3001, 0x3020, 1}, /* ideographic punctuation */ | |
1862 {0x3030, 0x3030, 1}, | |
1863 {0x303d, 0x303d, 1}, | |
1864 {0x3040, 0x309f, 0x3040}, /* Hiragana */ | |
1865 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */ | |
1866 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */ | |
1867 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */ | |
1868 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */ | |
1869 {0xfd3e, 0xfd3f, 1}, | |
1870 {0xfe30, 0xfe6b, 1}, /* punctuation forms */ | |
1871 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */ | |
1872 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */ | |
1873 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */ | |
1874 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */ | |
1875 }; | |
1876 int bot = 0; | |
1877 int top = sizeof(classes) / sizeof(struct clinterval) - 1; | |
1878 int mid; | |
1879 | |
1880 /* First quick check for Latin1 characters, use 'iskeyword'. */ | |
1881 if (c < 0x100) | |
1882 { | |
1883 if (c == ' ' || c == '\t' || c == NUL) | |
1884 return 0; /* blank */ | |
1885 if (vim_iswordc(c)) | |
1886 return 2; /* word character */ | |
1887 return 1; /* punctuation */ | |
1888 } | |
1889 | |
1890 /* binary search in table */ | |
1891 while (top >= bot) | |
1892 { | |
1893 mid = (bot + top) / 2; | |
1894 if (classes[mid].last < c) | |
1895 bot = mid + 1; | |
1896 else if (classes[mid].first > c) | |
1897 top = mid - 1; | |
1898 else | |
1899 return (int)classes[mid].class; | |
1900 } | |
1901 | |
1902 /* most other characters are "word" characters */ | |
1903 return 2; | |
1904 } | |
1905 | |
1906 /* | |
1907 * Code for Unicode case-dependent operations. Based on notes in | |
1908 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt | |
1909 * This code uses simple case folding, not full case folding. | |
1910 */ | |
1911 | |
1912 /* | |
1913 * The following table is built by foldExtract.pl < CaseFolding.txt . | |
1914 * It must be in numeric order, because we use binary search on it. | |
1915 * An entry such as {0x41,0x5a,1,32} means that UCS-4 characters in the range | |
1916 * from 0x41 to 0x5a inclusive, stepping by 1, are folded by adding 32. | |
1917 */ | |
1918 | |
1919 typedef struct | |
1920 { | |
1921 int rangeStart; | |
1922 int rangeEnd; | |
1923 int step; | |
1924 int offset; | |
1925 } convertStruct; | |
1926 | |
1927 convertStruct foldCase[] = | |
1928 { | |
1929 {0x41,0x5a,1,32}, {0xc0,0xd6,1,32}, {0xd8,0xde,1,32}, | |
1930 {0x100,0x12e,2,1}, {0x130,0x130,-1,-199}, {0x132,0x136,2,1}, | |
1931 {0x139,0x147,2,1}, {0x14a,0x176,2,1}, {0x178,0x178,-1,-121}, | |
1932 {0x179,0x17d,2,1}, {0x181,0x181,-1,210}, {0x182,0x184,2,1}, | |
1933 {0x186,0x186,-1,206}, {0x187,0x187,-1,1}, {0x189,0x18a,1,205}, | |
1934 {0x18b,0x18b,-1,1}, {0x18e,0x18e,-1,79}, {0x18f,0x18f,-1,202}, | |
1935 {0x190,0x190,-1,203}, {0x191,0x191,-1,1}, {0x193,0x193,-1,205}, | |
1936 {0x194,0x194,-1,207}, {0x196,0x196,-1,211}, {0x197,0x197,-1,209}, | |
1937 {0x198,0x198,-1,1}, {0x19c,0x19c,-1,211}, {0x19d,0x19d,-1,213}, | |
1938 {0x19f,0x19f,-1,214}, {0x1a0,0x1a4,2,1}, {0x1a6,0x1a6,-1,218}, | |
1939 {0x1a7,0x1a7,-1,1}, {0x1a9,0x1a9,-1,218}, {0x1ac,0x1ac,-1,1}, | |
1940 {0x1ae,0x1ae,-1,218}, {0x1af,0x1af,-1,1}, {0x1b1,0x1b2,1,217}, | |
1941 {0x1b3,0x1b5,2,1}, {0x1b7,0x1b7,-1,219}, {0x1b8,0x1bc,4,1}, | |
1942 {0x1c4,0x1c4,-1,2}, {0x1c5,0x1c5,-1,1}, {0x1c7,0x1c7,-1,2}, | |
1943 {0x1c8,0x1c8,-1,1}, {0x1ca,0x1ca,-1,2}, {0x1cb,0x1db,2,1}, | |
1944 {0x1de,0x1ee,2,1}, {0x1f1,0x1f1,-1,2}, {0x1f2,0x1f4,2,1}, | |
1945 {0x1f6,0x1f6,-1,-97}, {0x1f7,0x1f7,-1,-56}, {0x1f8,0x21e,2,1}, | |
1946 {0x220,0x220,-1,-130}, {0x222,0x232,2,1}, {0x386,0x386,-1,38}, | |
1947 {0x388,0x38a,1,37}, {0x38c,0x38c,-1,64}, {0x38e,0x38f,1,63}, | |
1948 {0x391,0x3a1,1,32}, {0x3a3,0x3ab,1,32}, {0x3d8,0x3ee,2,1}, | |
1949 {0x3f4,0x3f4,-1,-60}, {0x3f7,0x3f7,-1,1}, {0x3f9,0x3f9,-1,-7}, | |
1950 {0x3fa,0x3fa,-1,1}, {0x400,0x40f,1,80}, {0x410,0x42f,1,32}, | |
1951 {0x460,0x480,2,1}, {0x48a,0x4be,2,1}, {0x4c1,0x4cd,2,1}, | |
1952 {0x4d0,0x4f4,2,1}, {0x4f8,0x500,8,1}, {0x502,0x50e,2,1}, | |
1953 {0x531,0x556,1,48}, {0x1e00,0x1e94,2,1}, {0x1ea0,0x1ef8,2,1}, | |
1954 {0x1f08,0x1f0f,1,-8}, {0x1f18,0x1f1d,1,-8}, {0x1f28,0x1f2f,1,-8}, | |
1955 {0x1f38,0x1f3f,1,-8}, {0x1f48,0x1f4d,1,-8}, {0x1f59,0x1f5f,2,-8}, | |
1956 {0x1f68,0x1f6f,1,-8}, {0x1f88,0x1f8f,1,-8}, {0x1f98,0x1f9f,1,-8}, | |
1957 {0x1fa8,0x1faf,1,-8}, {0x1fb8,0x1fb9,1,-8}, {0x1fba,0x1fbb,1,-74}, | |
1958 {0x1fbc,0x1fbc,-1,-9}, {0x1fc8,0x1fcb,1,-86}, {0x1fcc,0x1fcc,-1,-9}, | |
1959 {0x1fd8,0x1fd9,1,-8}, {0x1fda,0x1fdb,1,-100}, {0x1fe8,0x1fe9,1,-8}, | |
1960 {0x1fea,0x1feb,1,-112}, {0x1fec,0x1fec,-1,-7}, {0x1ff8,0x1ff9,1,-128}, | |
1961 {0x1ffa,0x1ffb,1,-126}, {0x1ffc,0x1ffc,-1,-9}, {0x2126,0x2126,-1,-7517}, | |
1962 {0x212a,0x212a,-1,-8383}, {0x212b,0x212b,-1,-8262}, | |
1963 {0x2160,0x216f,1,16}, {0x24b6,0x24cf,1,26}, {0xff21,0xff3a,1,32}, | |
1964 {0x10400,0x10427,1,40} | |
1965 }; | |
1966 | |
1967 static int utf_convert(int a, convertStruct table[], int tableSize); | |
1968 | |
1969 /* | |
1970 * Generic conversion function for case operations. | |
1971 * Return the converted equivalent of "a", which is a UCS-4 character. Use | |
1972 * the given conversion "table". Uses binary search on "table". | |
1973 */ | |
1974 static int | |
1975 utf_convert(a, table, tableSize) | |
1976 int a; | |
1977 convertStruct table[]; | |
1978 int tableSize; | |
1979 { | |
1980 int start, mid, end; /* indices into table */ | |
1981 | |
1982 start = 0; | |
1983 end = tableSize / sizeof(convertStruct); | |
1984 while (start < end) | |
1985 { | |
1986 /* need to search further */ | |
1987 mid = (end + start) /2; | |
1988 if (table[mid].rangeEnd < a) | |
1989 start = mid + 1; | |
1990 else | |
1991 end = mid; | |
1992 } | |
1993 if (table[start].rangeStart <= a && a <= table[start].rangeEnd | |
1994 && (a - table[start].rangeStart) % table[start].step == 0) | |
1995 return (a + table[start].offset); | |
1996 else | |
1997 return a; | |
1998 } | |
1999 | |
2000 /* | |
2001 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses | |
2002 * simple case folding. | |
2003 */ | |
2004 int | |
2005 utf_fold(a) | |
2006 int a; | |
2007 { | |
2008 return utf_convert(a, foldCase, sizeof(foldCase)); | |
2009 } | |
2010 | |
2011 /* | |
2012 * The following tables are built by upperLowerExtract.pl < UnicodeData.txt . | |
2013 * They must be in numeric order, because we use binary search on them. | |
2014 * An entry such as {0x41,0x5a,1,32} means that UCS-4 characters in the range | |
2015 * from 0x41 to 0x5a inclusive, stepping by 1, are switched to lower (for | |
2016 * example) by adding 32. | |
2017 */ | |
2018 convertStruct toLower[] = | |
2019 { | |
2020 {0x41,0x5a,1,32}, {0xc0,0xd6,1,32}, {0xd8,0xde,1,32}, | |
2021 {0x100,0x12e,2,1}, {0x130,0x130,-1,-199}, {0x132,0x136,2,1}, | |
2022 {0x139,0x147,2,1}, {0x14a,0x176,2,1}, {0x178,0x178,-1,-121}, | |
2023 {0x179,0x17d,2,1}, {0x181,0x181,-1,210}, {0x182,0x184,2,1}, | |
2024 {0x186,0x186,-1,206}, {0x187,0x187,-1,1}, {0x189,0x18a,1,205}, | |
2025 {0x18b,0x18b,-1,1}, {0x18e,0x18e,-1,79}, {0x18f,0x18f,-1,202}, | |
2026 {0x190,0x190,-1,203}, {0x191,0x191,-1,1}, {0x193,0x193,-1,205}, | |
2027 {0x194,0x194,-1,207}, {0x196,0x196,-1,211}, {0x197,0x197,-1,209}, | |
2028 {0x198,0x198,-1,1}, {0x19c,0x19c,-1,211}, {0x19d,0x19d,-1,213}, | |
2029 {0x19f,0x19f,-1,214}, {0x1a0,0x1a4,2,1}, {0x1a6,0x1a6,-1,218}, | |
2030 {0x1a7,0x1a7,-1,1}, {0x1a9,0x1a9,-1,218}, {0x1ac,0x1ac,-1,1}, | |
2031 {0x1ae,0x1ae,-1,218}, {0x1af,0x1af,-1,1}, {0x1b1,0x1b2,1,217}, | |
2032 {0x1b3,0x1b5,2,1}, {0x1b7,0x1b7,-1,219}, {0x1b8,0x1bc,4,1}, | |
2033 {0x1c4,0x1ca,3,2}, {0x1cd,0x1db,2,1}, {0x1de,0x1ee,2,1}, | |
2034 {0x1f1,0x1f1,-1,2}, {0x1f4,0x1f4,-1,1}, {0x1f6,0x1f6,-1,-97}, | |
2035 {0x1f7,0x1f7,-1,-56}, {0x1f8,0x21e,2,1}, {0x220,0x220,-1,-130}, | |
2036 {0x222,0x232,2,1}, {0x386,0x386,-1,38}, {0x388,0x38a,1,37}, | |
2037 {0x38c,0x38c,-1,64}, {0x38e,0x38f,1,63}, {0x391,0x3a1,1,32}, | |
2038 {0x3a3,0x3ab,1,32}, {0x3d8,0x3ee,2,1}, {0x3f4,0x3f4,-1,-60}, | |
2039 {0x3f7,0x3f7,-1,1}, {0x3f9,0x3f9,-1,-7}, {0x3fa,0x3fa,-1,1}, | |
2040 {0x400,0x40f,1,80}, {0x410,0x42f,1,32}, {0x460,0x480,2,1}, | |
2041 {0x48a,0x4be,2,1}, {0x4c1,0x4cd,2,1}, {0x4d0,0x4f4,2,1}, | |
2042 {0x4f8,0x500,8,1}, {0x502,0x50e,2,1}, {0x531,0x556,1,48}, | |
2043 {0x1e00,0x1e94,2,1}, {0x1ea0,0x1ef8,2,1}, {0x1f08,0x1f0f,1,-8}, | |
2044 {0x1f18,0x1f1d,1,-8}, {0x1f28,0x1f2f,1,-8}, {0x1f38,0x1f3f,1,-8}, | |
2045 {0x1f48,0x1f4d,1,-8}, {0x1f59,0x1f5f,2,-8}, {0x1f68,0x1f6f,1,-8}, | |
2046 {0x1fb8,0x1fb9,1,-8}, {0x1fba,0x1fbb,1,-74}, {0x1fc8,0x1fcb,1,-86}, | |
2047 {0x1fd8,0x1fd9,1,-8}, {0x1fda,0x1fdb,1,-100}, {0x1fe8,0x1fe9,1,-8}, | |
2048 {0x1fea,0x1feb,1,-112}, {0x1fec,0x1fec,-1,-7}, {0x1ff8,0x1ff9,1,-128}, | |
2049 {0x1ffa,0x1ffb,1,-126}, {0x2126,0x2126,-1,-7517}, {0x212a,0x212a,-1,-8383}, | |
2050 {0x212b,0x212b,-1,-8262}, {0xff21,0xff3a,1,32}, {0x10400,0x10427,1,40} | |
2051 }; | |
2052 | |
2053 convertStruct toUpper[] = | |
2054 { | |
2055 {0x61,0x7a,1,-32}, {0xb5,0xb5,-1,743}, {0xe0,0xf6,1,-32}, | |
2056 {0xf8,0xfe,1,-32}, {0xff,0xff,-1,121}, {0x101,0x12f,2,-1}, | |
2057 {0x131,0x131,-1,-232}, {0x133,0x137,2,-1}, {0x13a,0x148,2,-1}, | |
2058 {0x14b,0x177,2,-1}, {0x17a,0x17e,2,-1}, {0x17f,0x17f,-1,-300}, | |
2059 {0x183,0x185,2,-1}, {0x188,0x18c,4,-1}, {0x192,0x192,-1,-1}, | |
2060 {0x195,0x195,-1,97}, {0x199,0x199,-1,-1}, {0x19e,0x19e,-1,130}, | |
2061 {0x1a1,0x1a5,2,-1}, {0x1a8,0x1ad,5,-1}, {0x1b0,0x1b4,4,-1}, | |
2062 {0x1b6,0x1b9,3,-1}, {0x1bd,0x1bd,-1,-1}, {0x1bf,0x1bf,-1,56}, | |
2063 {0x1c5,0x1c6,1,-1}, {0x1c8,0x1c9,1,-1}, {0x1cb,0x1cc,1,-1}, | |
2064 {0x1ce,0x1dc,2,-1}, {0x1dd,0x1dd,-1,-79}, {0x1df,0x1ef,2,-1}, | |
2065 {0x1f2,0x1f3,1,-1}, {0x1f5,0x1f9,4,-1}, {0x1fb,0x21f,2,-1}, | |
2066 {0x223,0x233,2,-1}, {0x253,0x253,-1,-210}, {0x254,0x254,-1,-206}, | |
2067 {0x256,0x257,1,-205}, {0x259,0x259,-1,-202}, {0x25b,0x25b,-1,-203}, | |
2068 {0x260,0x260,-1,-205}, {0x263,0x263,-1,-207}, {0x268,0x268,-1,-209}, | |
2069 {0x269,0x26f,6,-211}, {0x272,0x272,-1,-213}, {0x275,0x275,-1,-214}, | |
2070 {0x280,0x283,3,-218}, {0x288,0x288,-1,-218}, {0x28a,0x28b,1,-217}, | |
2071 {0x292,0x292,-1,-219}, {0x3ac,0x3ac,-1,-38}, {0x3ad,0x3af,1,-37}, | |
2072 {0x3b1,0x3c1,1,-32}, {0x3c2,0x3c2,-1,-31}, {0x3c3,0x3cb,1,-32}, | |
2073 {0x3cc,0x3cc,-1,-64}, {0x3cd,0x3ce,1,-63}, {0x3d0,0x3d0,-1,-62}, | |
2074 {0x3d1,0x3d1,-1,-57}, {0x3d5,0x3d5,-1,-47}, {0x3d6,0x3d6,-1,-54}, | |
2075 {0x3d9,0x3ef,2,-1}, {0x3f0,0x3f0,-1,-86}, {0x3f1,0x3f1,-1,-80}, | |
2076 {0x3f2,0x3f2,-1,7}, {0x3f5,0x3f5,-1,-96}, {0x3f8,0x3fb,3,-1}, | |
2077 {0x430,0x44f,1,-32}, {0x450,0x45f,1,-80}, {0x461,0x481,2,-1}, | |
2078 {0x48b,0x4bf,2,-1}, {0x4c2,0x4ce,2,-1}, {0x4d1,0x4f5,2,-1}, | |
2079 {0x4f9,0x501,8,-1}, {0x503,0x50f,2,-1}, {0x561,0x586,1,-48}, | |
2080 {0x1e01,0x1e95,2,-1}, {0x1e9b,0x1e9b,-1,-59}, {0x1ea1,0x1ef9,2,-1}, | |
2081 {0x1f00,0x1f07,1,8}, {0x1f10,0x1f15,1,8}, {0x1f20,0x1f27,1,8}, | |
2082 {0x1f30,0x1f37,1,8}, {0x1f40,0x1f45,1,8}, {0x1f51,0x1f57,2,8}, | |
2083 {0x1f60,0x1f67,1,8}, {0x1f70,0x1f71,1,74}, {0x1f72,0x1f75,1,86}, | |
2084 {0x1f76,0x1f77,1,100}, {0x1f78,0x1f79,1,128}, {0x1f7a,0x1f7b,1,112}, | |
2085 {0x1f7c,0x1f7d,1,126}, {0x1f80,0x1f87,1,8}, {0x1f90,0x1f97,1,8}, | |
2086 {0x1fa0,0x1fa7,1,8}, {0x1fb0,0x1fb1,1,8}, {0x1fb3,0x1fb3,-1,9}, | |
2087 {0x1fbe,0x1fbe,-1,-7205}, {0x1fc3,0x1fc3,-1,9}, {0x1fd0,0x1fd1,1,8}, | |
2088 {0x1fe0,0x1fe1,1,8}, {0x1fe5,0x1fe5,-1,7}, {0x1ff3,0x1ff3,-1,9}, | |
2089 {0xff41,0xff5a,1,-32}, {0x10428,0x1044f,1,-40} | |
2090 }; | |
2091 | |
2092 /* | |
2093 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use | |
2094 * simple case folding. | |
2095 */ | |
2096 int | |
2097 utf_toupper(a) | |
2098 int a; | |
2099 { | |
2100 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */ | |
2101 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) | |
2102 return TOUPPER_ASC(a); | |
2103 | |
2104 #if defined(HAVE_TOWUPPER) && defined(__STDC__ISO_10646__) | |
2105 /* If towupper() is availble and handles Unicode, use it. */ | |
2106 if (!(cmp_flags & CMP_INTERNAL)) | |
2107 return towupper(a); | |
2108 #endif | |
2109 | |
2110 /* For characters below 128 use locale sensitive toupper(). */ | |
2111 if (a < 128) | |
2112 return TOUPPER_LOC(a); | |
2113 | |
2114 /* For any other characters use the above mapping table. */ | |
2115 return utf_convert(a, toUpper, sizeof(toUpper)); | |
2116 } | |
2117 | |
2118 int | |
2119 utf_islower(a) | |
2120 int a; | |
2121 { | |
2122 return (utf_toupper(a) != a); | |
2123 } | |
2124 | |
2125 /* | |
2126 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use | |
2127 * simple case folding. | |
2128 */ | |
2129 int | |
2130 utf_tolower(a) | |
2131 int a; | |
2132 { | |
2133 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */ | |
2134 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) | |
2135 return TOLOWER_ASC(a); | |
2136 | |
2137 #if defined(HAVE_TOWLOWER) && defined(__STDC__ISO_10646__) | |
2138 /* If towlower() is availble and handles Unicode, use it. */ | |
2139 if (!(cmp_flags & CMP_INTERNAL)) | |
2140 return towlower(a); | |
2141 #endif | |
2142 | |
2143 /* For characters below 128 use locale sensitive tolower(). */ | |
2144 if (a < 128) | |
2145 return TOLOWER_LOC(a); | |
2146 | |
2147 /* For any other characters use the above mapping table. */ | |
2148 return utf_convert(a, toLower, sizeof(toLower)); | |
2149 } | |
2150 | |
2151 int | |
2152 utf_isupper(a) | |
2153 int a; | |
2154 { | |
2155 return (utf_tolower(a) != a); | |
2156 } | |
2157 | |
2158 /* | |
2159 * Version of strnicmp() that handles multi-byte characters. | |
2160 * Needed for Big5, Sjift-JIS and UTF-8 encoding. Other DBCS encodings can | |
2161 * probably use strnicmp(), because there are no ASCII characters in the | |
2162 * second byte. | |
2163 * Returns zero if s1 and s2 are equal (ignoring case), the difference between | |
2164 * two characters otherwise. | |
2165 */ | |
2166 int | |
2167 mb_strnicmp(s1, s2, n) | |
2168 char_u *s1, *s2; | |
2169 int n; | |
2170 { | |
2171 int i, j, l; | |
2172 int cdiff; | |
2173 | |
2174 for (i = 0; i < n; i += l) | |
2175 { | |
2176 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */ | |
2177 return 0; | |
2178 if (enc_utf8) | |
2179 { | |
2180 l = utf_byte2len(s1[i]); | |
2181 if (l > n - i) | |
2182 l = n - i; /* incomplete character */ | |
2183 /* Check directly first, it's faster. */ | |
2184 for (j = 0; j < l; ++j) | |
2185 if (s1[i + j] != s2[i + j]) | |
2186 break; | |
2187 if (j < l) | |
2188 { | |
2189 /* If one of the two characters is incomplete return -1. */ | |
2190 if (i + utf_byte2len(s1[i]) > n || i + utf_byte2len(s2[i]) > n) | |
2191 return -1; | |
2192 cdiff = utf_fold(utf_ptr2char(s1 + i)) | |
2193 - utf_fold(utf_ptr2char(s2 + i)); | |
2194 if (cdiff != 0) | |
2195 return cdiff; | |
2196 } | |
2197 } | |
2198 else | |
2199 { | |
2200 l = (*mb_ptr2len_check)(s1 + i); | |
2201 if (l <= 1) | |
2202 { | |
2203 /* Single byte: first check normally, then with ignore case. */ | |
2204 if (s1[i] != s2[i]) | |
2205 { | |
2206 cdiff = TOLOWER_LOC(s1[i]) - TOLOWER_LOC(s2[i]); | |
2207 if (cdiff != 0) | |
2208 return cdiff; | |
2209 } | |
2210 } | |
2211 else | |
2212 { | |
2213 /* For non-Unicode multi-byte don't ignore case. */ | |
2214 if (l > n - i) | |
2215 l = n - i; | |
2216 cdiff = STRNCMP(s1 + i, s2 + i, l); | |
2217 if (cdiff != 0) | |
2218 return cdiff; | |
2219 } | |
2220 } | |
2221 } | |
2222 return 0; | |
2223 } | |
2224 | |
2225 /* | |
2226 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what | |
2227 * 'encoding' has been set to. | |
2228 */ | |
2229 void | |
2230 show_utf8() | |
2231 { | |
2232 int len; | |
2233 char_u *line; | |
2234 int clen; | |
2235 int i; | |
2236 | |
2237 /* Get the byte length of the char under the cursor, including composing | |
2238 * characters. */ | |
2239 line = ml_get_cursor(); | |
2240 len = utfc_ptr2len_check(line); | |
2241 if (len == 0) | |
2242 { | |
2243 MSG("NUL"); | |
2244 return; | |
2245 } | |
2246 | |
2247 IObuff[0] = NUL; | |
2248 clen = 0; | |
2249 for (i = 0; i < len; ++i) | |
2250 { | |
2251 if (clen == 0) | |
2252 { | |
2253 /* start of (composing) character, get its length */ | |
2254 if (i > 0) | |
2255 STRCAT(IObuff, "+ "); | |
2256 clen = utf_ptr2len_check(line + i); | |
2257 } | |
2258 sprintf((char *)IObuff + STRLEN(IObuff), "%02x ", line[i]); | |
2259 --clen; | |
2260 } | |
2261 | |
2262 msg(IObuff); | |
2263 } | |
2264 | |
2265 /* | |
2266 * mb_head_off() function pointer. | |
2267 * Return offset from "p" to the first byte of the character it points into. | |
2268 * Returns 0 when already at the first byte of a character. | |
2269 */ | |
2270 /*ARGSUSED*/ | |
2271 int | |
2272 latin_head_off(base, p) | |
2273 char_u *base; | |
2274 char_u *p; | |
2275 { | |
2276 return 0; | |
2277 } | |
2278 | |
2279 int | |
2280 dbcs_head_off(base, p) | |
2281 char_u *base; | |
2282 char_u *p; | |
2283 { | |
2284 char_u *q; | |
2285 | |
2286 /* It can't be a trailing byte when not using DBCS, at the start of the | |
2287 * string or the previous byte can't start a double-byte. */ | |
2288 if (p <= base || MB_BYTE2LEN(p[-1]) == 1) | |
2289 return 0; | |
2290 | |
2291 /* This is slow: need to start at the base and go forward until the | |
2292 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */ | |
2293 q = base; | |
2294 while (q < p) | |
2295 q += dbcs_ptr2len_check(q); | |
2296 return (q == p) ? 0 : 1; | |
2297 } | |
2298 | |
2299 #if defined(FEAT_CLIPBOARD) || defined(FEAT_GUI) || defined(FEAT_RIGHTLEFT) \ | |
2300 || defined(PROTO) | |
2301 /* | |
2302 * Special version of dbcs_head_off() that works for ScreenLines[], where | |
2303 * single-width DBCS_JPNU characters are stored separately. | |
2304 */ | |
2305 int | |
2306 dbcs_screen_head_off(base, p) | |
2307 char_u *base; | |
2308 char_u *p; | |
2309 { | |
2310 char_u *q; | |
2311 | |
2312 /* It can't be a trailing byte when not using DBCS, at the start of the | |
2313 * string or the previous byte can't start a double-byte. | |
2314 * For euc-jp an 0x8e byte in the previous cell always means we have a | |
2315 * lead byte in the current cell. */ | |
2316 if (p <= base | |
2317 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) | |
2318 || MB_BYTE2LEN(p[-1]) == 1) | |
2319 return 0; | |
2320 | |
2321 /* This is slow: need to start at the base and go forward until the | |
2322 * byte we are looking for. Return 1 when we went past it, 0 otherwise. | |
2323 * For DBCS_JPNU look out for 0x8e, which means the second byte is not | |
2324 * stored as the next byte. */ | |
2325 q = base; | |
2326 while (q < p) | |
2327 { | |
2328 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) | |
2329 ++q; | |
2330 else | |
2331 q += dbcs_ptr2len_check(q); | |
2332 } | |
2333 return (q == p) ? 0 : 1; | |
2334 } | |
2335 #endif | |
2336 | |
2337 int | |
2338 utf_head_off(base, p) | |
2339 char_u *base; | |
2340 char_u *p; | |
2341 { | |
2342 char_u *q; | |
2343 char_u *s; | |
2344 int c; | |
2345 #ifdef FEAT_ARABIC | |
2346 char_u *j; | |
2347 #endif | |
2348 | |
2349 if (*p < 0x80) /* be quick for ASCII */ | |
2350 return 0; | |
2351 | |
2352 /* Skip backwards over trailing bytes: 10xx.xxxx | |
2353 * Skip backwards again if on a composing char. */ | |
2354 for (q = p; ; --q) | |
2355 { | |
2356 /* Move s to the last byte of this char. */ | |
2357 for (s = q; (s[1] & 0xc0) == 0x80; ++s) | |
2358 ; | |
2359 /* Move q to the first byte of this char. */ | |
2360 while (q > base && (*q & 0xc0) == 0x80) | |
2361 --q; | |
2362 /* Check for illegal sequence. Do allow an illegal byte after where we | |
2363 * started. */ | |
2364 if (utf8len_tab[*q] != (int)(s - q + 1) | |
2365 && utf8len_tab[*q] != (int)(p - q + 1)) | |
2366 return 0; | |
2367 | |
2368 if (q <= base) | |
2369 break; | |
2370 | |
2371 c = utf_ptr2char(q); | |
2372 if (utf_iscomposing(c)) | |
2373 continue; | |
2374 | |
2375 #ifdef FEAT_ARABIC | |
2376 if (arabic_maycombine(c)) | |
2377 { | |
2378 /* Advance to get a sneak-peak at the next char */ | |
2379 j = q; | |
2380 --j; | |
2381 /* Move j to the first byte of this char. */ | |
2382 while (j > base && (*j & 0xc0) == 0x80) | |
2383 --j; | |
2384 if (arabic_combine(utf_ptr2char(j), c)) | |
2385 continue; | |
2386 } | |
2387 #endif | |
2388 break; | |
2389 } | |
2390 | |
2391 return (int)(p - q); | |
2392 } | |
2393 | |
2394 /* | |
2395 * Return the offset from "p" to the first byte of a character. When "p" is | |
2396 * at the start of a character 0 is returned, otherwise the offset to the next | |
2397 * character. Can start anywhere in a stream of bytes. | |
2398 */ | |
2399 int | |
2400 mb_off_next(base, p) | |
2401 char_u *base; | |
2402 char_u *p; | |
2403 { | |
2404 int i; | |
2405 int j; | |
2406 | |
2407 if (enc_utf8) | |
2408 { | |
2409 if (*p < 0x80) /* be quick for ASCII */ | |
2410 return 0; | |
2411 | |
2412 /* Find the next character that isn't 10xx.xxxx */ | |
2413 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) | |
2414 ; | |
2415 if (i > 0) | |
2416 { | |
2417 /* Check for illegal sequence. */ | |
2418 for (j = 0; p - j > base; ++j) | |
2419 if ((p[-j] & 0xc0) != 0x80) | |
2420 break; | |
2421 if (utf8len_tab[p[-j]] != i + j) | |
2422 return 0; | |
2423 } | |
2424 return i; | |
2425 } | |
2426 | |
2427 /* Only need to check if we're on a trail byte, it doesn't matter if we | |
2428 * want the offset to the next or current character. */ | |
2429 return (*mb_head_off)(base, p); | |
2430 } | |
2431 | |
2432 /* | |
2433 * Return the offset from "p" to the last byte of the character it points | |
2434 * into. Can start anywhere in a stream of bytes. | |
2435 */ | |
2436 int | |
2437 mb_tail_off(base, p) | |
2438 char_u *base; | |
2439 char_u *p; | |
2440 { | |
2441 int i; | |
2442 int j; | |
2443 | |
2444 if (*p == NUL) | |
2445 return 0; | |
2446 | |
2447 if (enc_utf8) | |
2448 { | |
2449 /* Find the last character that is 10xx.xxxx */ | |
2450 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) | |
2451 ; | |
2452 /* Check for illegal sequence. */ | |
2453 for (j = 0; p - j > base; ++j) | |
2454 if ((p[-j] & 0xc0) != 0x80) | |
2455 break; | |
2456 if (utf8len_tab[p[-j]] != i + j + 1) | |
2457 return 0; | |
2458 return i; | |
2459 } | |
2460 | |
2461 /* It can't be the first byte if a double-byte when not using DBCS, at the | |
2462 * end of the string or the byte can't start a double-byte. */ | |
2463 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) | |
2464 return 0; | |
2465 | |
2466 /* Return 1 when on the lead byte, 0 when on the tail byte. */ | |
2467 return 1 - dbcs_head_off(base, p); | |
2468 } | |
2469 | |
2470 #if defined(FEAT_GUI) || defined(PROTO) | |
2471 /* | |
2472 * Special version of mb_tail_off() for use in ScreenLines[]. | |
2473 */ | |
2474 int | |
2475 dbcs_screen_tail_off(base, p) | |
2476 char_u *base; | |
2477 char_u *p; | |
2478 { | |
2479 /* It can't be the first byte if a double-byte when not using DBCS, at the | |
2480 * end of the string or the byte can't start a double-byte. | |
2481 * For euc-jp an 0x8e byte always means we have a lead byte in the current | |
2482 * cell. */ | |
2483 if (*p == NUL || p[1] == NUL | |
2484 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) | |
2485 || MB_BYTE2LEN(*p) == 1) | |
2486 return 0; | |
2487 | |
2488 /* Return 1 when on the lead byte, 0 when on the tail byte. */ | |
2489 return 1 - dbcs_screen_head_off(base, p); | |
2490 } | |
2491 #endif | |
2492 | |
2493 /* | |
2494 * If the cursor moves on an trail byte, set the cursor on the lead byte. | |
2495 * Thus it moves left if necessary. | |
2496 * Return TRUE when the cursor was adjusted. | |
2497 */ | |
2498 void | |
2499 mb_adjust_cursor() | |
2500 { | |
2501 mb_adjustpos(&curwin->w_cursor); | |
2502 } | |
2503 | |
2504 /* | |
2505 * Adjust position "*lp" to point to the first byte of a multi-byte character. | |
2506 * If it points to a tail byte it's moved backwards to the head byte. | |
2507 */ | |
2508 void | |
2509 mb_adjustpos(lp) | |
2510 pos_T *lp; | |
2511 { | |
2512 char_u *p; | |
2513 | |
2514 if (lp->col > 0 | |
2515 #ifdef FEAT_VIRTUALEDIT | |
2516 || lp->coladd > 1 | |
2517 #endif | |
2518 ) | |
2519 { | |
2520 p = ml_get(lp->lnum); | |
2521 lp->col -= (*mb_head_off)(p, p + lp->col); | |
2522 #ifdef FEAT_VIRTUALEDIT | |
2523 /* Reset "coladd" when the cursor would be on the right half of a | |
2524 * double-wide character. */ | |
2525 if (lp->coladd == 1 | |
2526 && p[lp->col] != TAB | |
2527 && vim_isprintc((*mb_ptr2char)(p + lp->col)) | |
2528 && ptr2cells(p + lp->col) > 1) | |
2529 lp->coladd = 0; | |
2530 #endif | |
2531 } | |
2532 } | |
2533 | |
2534 /* | |
2535 * Return a pointer to the character before "*p", if there is one. | |
2536 */ | |
2537 char_u * | |
2538 mb_prevptr(line, p) | |
2539 char_u *line; /* start of the string */ | |
2540 char_u *p; | |
2541 { | |
2542 if (p > line) | |
2543 p = p - (*mb_head_off)(line, p - 1) - 1; | |
2544 return p; | |
2545 } | |
2546 | |
2547 /* | |
2548 * Return the character length of "str". multi-byte characters counts as one. | |
2549 */ | |
2550 int | |
2551 mb_charlen(str) | |
2552 char_u *str; | |
2553 { | |
2554 int count; | |
2555 | |
2556 if (str == NULL) | |
2557 return 0; | |
2558 | |
2559 for (count = 0; *str != NUL; count++) | |
2560 str += (*mb_ptr2len_check)(str); | |
2561 | |
2562 return count; | |
2563 } | |
2564 | |
2565 /* | |
2566 * Try to un-escape a multi-byte character. | |
2567 * Used for the "to" and "from" part of a mapping. | |
2568 * Return the un-escaped string if it is a multi-byte character, and advance | |
2569 * "pp" to just after the bytes that formed it. | |
2570 * Return NULL if no multi-byte char was found. | |
2571 */ | |
2572 char_u * | |
2573 mb_unescape(pp) | |
2574 char_u **pp; | |
2575 { | |
2576 static char_u buf[MB_MAXBYTES + 1]; | |
2577 int n, m = 0; | |
2578 char_u *str = *pp; | |
2579 | |
2580 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI | |
2581 * KS_EXTRA KE_CSI to CSI. */ | |
2582 for (n = 0; str[n] != NUL && m <= MB_MAXBYTES; ++n) | |
2583 { | |
2584 if (str[n] == K_SPECIAL | |
2585 && str[n + 1] == KS_SPECIAL | |
2586 && str[n + 2] == KE_FILLER) | |
2587 { | |
2588 buf[m++] = K_SPECIAL; | |
2589 n += 2; | |
2590 } | |
2591 # ifdef FEAT_GUI | |
2592 else if (str[n] == CSI | |
2593 && str[n + 1] == KS_EXTRA | |
2594 && str[n + 2] == (int)KE_CSI) | |
2595 { | |
2596 buf[m++] = CSI; | |
2597 n += 2; | |
2598 } | |
2599 # endif | |
2600 else if (str[n] == K_SPECIAL | |
2601 # ifdef FEAT_GUI | |
2602 || str[n] == CSI | |
2603 # endif | |
2604 ) | |
2605 break; /* a special key can't be a multibyte char */ | |
2606 else | |
2607 buf[m++] = str[n]; | |
2608 buf[m] = NUL; | |
2609 | |
2610 /* Return a multi-byte character if it's found. An illegal sequence | |
2611 * will result in a 1 here. */ | |
2612 if ((*mb_ptr2len_check)(buf) > 1) | |
2613 { | |
2614 *pp = str + n + 1; | |
2615 return buf; | |
2616 } | |
2617 } | |
2618 return NULL; | |
2619 } | |
2620 | |
2621 /* | |
2622 * Return TRUE if the character at "row"/"col" on the screen is the left side | |
2623 * of a double-width character. | |
2624 * Caller must make sure "row" and "col" are not invalid! | |
2625 */ | |
2626 int | |
2627 mb_lefthalve(row, col) | |
2628 int row; | |
2629 int col; | |
2630 { | |
2631 #ifdef FEAT_HANGULIN | |
2632 if (composing_hangul) | |
2633 return TRUE; | |
2634 #endif | |
2635 if (enc_dbcs != 0) | |
2636 return dbcs_off2cells(LineOffset[row] + col) > 1; | |
2637 if (enc_utf8) | |
2638 return (col + 1 < Columns | |
2639 && ScreenLines[LineOffset[row] + col + 1] == 0); | |
2640 return FALSE; | |
2641 } | |
2642 | |
2643 # if defined(FEAT_CLIPBOARD) || defined(FEAT_GUI) || defined(FEAT_RIGHTLEFT) \ | |
2644 || defined(PROTO) | |
2645 /* | |
2646 * Correct a position on the screen, if it's the right halve of a double-wide | |
2647 * char move it to the left halve. Returns the corrected column. | |
2648 */ | |
2649 int | |
2650 mb_fix_col(col, row) | |
2651 int col; | |
2652 int row; | |
2653 { | |
2654 col = check_col(col); | |
2655 row = check_row(row); | |
2656 if (has_mbyte && ScreenLines != NULL && col > 0 | |
2657 && ((enc_dbcs | |
2658 && ScreenLines[LineOffset[row] + col] != NUL | |
2659 && dbcs_screen_head_off(ScreenLines + LineOffset[row], | |
2660 ScreenLines + LineOffset[row] + col)) | |
2661 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0))) | |
2662 --col; | |
2663 return col; | |
2664 } | |
2665 # endif | |
2666 #endif | |
2667 | |
2668 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) | |
2669 static int enc_alias_search __ARGS((char_u *name)); | |
2670 | |
2671 /* | |
2672 * Skip the Vim specific head of a 'encoding' name. | |
2673 */ | |
2674 char_u * | |
2675 enc_skip(p) | |
2676 char_u *p; | |
2677 { | |
2678 if (STRNCMP(p, "2byte-", 6) == 0) | |
2679 return p + 6; | |
2680 if (STRNCMP(p, "8bit-", 5) == 0) | |
2681 return p + 5; | |
2682 return p; | |
2683 } | |
2684 | |
2685 /* | |
2686 * Find the canonical name for encoding "enc". | |
2687 * When the name isn't recognized, returns "enc" itself, but with all lower | |
2688 * case characters and '_' replaced with '-'. | |
2689 * Returns an allocated string. NULL for out-of-memory. | |
2690 */ | |
2691 char_u * | |
2692 enc_canonize(enc) | |
2693 char_u *enc; | |
2694 { | |
2695 char_u *r; | |
2696 char_u *p, *s; | |
2697 int i; | |
2698 | |
2699 /* copy "enc" to allocted memory, with room for two '-' */ | |
2700 r = alloc((unsigned)(STRLEN(enc) + 3)); | |
2701 if (r != NULL) | |
2702 { | |
2703 /* Make it all lower case and replace '_' with '-'. */ | |
2704 p = r; | |
2705 for (s = enc; *s != NUL; ++s) | |
2706 { | |
2707 if (*s == '_') | |
2708 *p++ = '-'; | |
2709 else | |
2710 *p++ = TOLOWER_ASC(*s); | |
2711 } | |
2712 *p = NUL; | |
2713 | |
2714 /* Skip "2byte-" and "8bit-". */ | |
2715 p = enc_skip(r); | |
2716 | |
2717 /* "iso8859" -> "iso-8859" */ | |
2718 if (STRNCMP(p, "iso8859", 7) == 0) | |
2719 { | |
2720 mch_memmove(p + 4, p + 3, STRLEN(p + 2)); | |
2721 p[3] = '-'; | |
2722 } | |
2723 | |
2724 /* "iso-8859n" -> "iso-8859-n" */ | |
2725 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') | |
2726 { | |
2727 mch_memmove(p + 9, p + 8, STRLEN(p + 7)); | |
2728 p[8] = '-'; | |
2729 } | |
2730 | |
2731 /* "latin-N" -> "latinN" */ | |
2732 if (STRNCMP(p, "latin-", 6) == 0) | |
2733 mch_memmove(p + 5, p + 6, STRLEN(p + 5)); | |
2734 | |
2735 if (enc_canon_search(p) >= 0) | |
2736 { | |
2737 /* canonical name can be used unmodified */ | |
2738 if (p != r) | |
2739 mch_memmove(r, p, STRLEN(p) + 1); | |
2740 } | |
2741 else if ((i = enc_alias_search(p)) >= 0) | |
2742 { | |
2743 /* alias recognized, get canonical name */ | |
2744 vim_free(r); | |
2745 r = vim_strsave((char_u *)enc_canon_table[i].name); | |
2746 } | |
2747 } | |
2748 return r; | |
2749 } | |
2750 | |
2751 /* | |
2752 * Search for an encoding alias of "name". | |
2753 * Returns -1 when not found. | |
2754 */ | |
2755 static int | |
2756 enc_alias_search(name) | |
2757 char_u *name; | |
2758 { | |
2759 int i; | |
2760 | |
2761 for (i = 0; enc_alias_table[i].name != NULL; ++i) | |
2762 if (STRCMP(name, enc_alias_table[i].name) == 0) | |
2763 return enc_alias_table[i].canon; | |
2764 return -1; | |
2765 } | |
2766 #endif | |
2767 | |
2768 #if defined(FEAT_MBYTE) || defined(PROTO) | |
2769 | |
2770 #ifdef HAVE_LANGINFO_H | |
2771 # include <langinfo.h> | |
2772 #endif | |
2773 | |
2774 /* | |
2775 * Get the canonicalized encoding of the current locale. | |
2776 * Returns an allocated string when successful, NULL when not. | |
2777 */ | |
2778 char_u * | |
2779 enc_locale() | |
2780 { | |
2781 #ifndef WIN3264 | |
2782 char *s; | |
2783 char *p; | |
2784 int i; | |
2785 #endif | |
2786 char buf[50]; | |
2787 #ifdef WIN3264 | |
2788 long acp = GetACP(); | |
2789 | |
2790 if (acp == 1200) | |
2791 STRCPY(buf, "ucs-2le"); | |
2792 else if (acp == 1252) | |
2793 STRCPY(buf, "latin1"); | |
2794 else | |
2795 sprintf(buf, "cp%ld", acp); | |
2796 #else | |
2797 # ifdef HAVE_NL_LANGINFO_CODESET | |
2798 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) | |
2799 # endif | |
2800 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) | |
2801 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) | |
2802 # endif | |
2803 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) | |
2804 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) | |
2805 s = getenv("LANG"); | |
2806 | |
2807 if (s == NULL || *s == NUL) | |
2808 return FAIL; | |
2809 | |
2810 /* The most generic locale format is: | |
2811 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] | |
2812 * If there is a '.' remove the part before it. | |
2813 * if there is something after the codeset, remove it. | |
2814 * Make the name lowercase and replace '_' with '-'. | |
2815 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", | |
2816 * "ko_KR.EUC" == "euc-kr" | |
2817 */ | |
2818 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) | |
2819 { | |
2820 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 | |
2821 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') | |
2822 { | |
2823 /* copy "XY.EUC" to "euc-XY" to buf[10] */ | |
2824 STRCPY(buf + 10, "euc-"); | |
2825 buf[14] = p[-2]; | |
2826 buf[15] = p[-1]; | |
2827 buf[16] = 0; | |
2828 s = buf + 10; | |
2829 } | |
2830 else | |
2831 s = p + 1; | |
2832 } | |
2833 for (i = 0; s[i] != NUL && i < sizeof(buf) - 1; ++i) | |
2834 { | |
2835 if (s[i] == '_' || s[i] == '-') | |
2836 buf[i] = '-'; | |
2837 else if (isalnum((int)s[i])) | |
2838 buf[i] = TOLOWER_ASC(s[i]); | |
2839 else | |
2840 break; | |
2841 } | |
2842 buf[i] = NUL; | |
2843 #endif | |
2844 | |
2845 return enc_canonize((char_u *)buf); | |
2846 } | |
2847 | |
2848 #if defined(WIN3264) || defined(PROTO) | |
2849 /* | |
2850 * Convert an encoding name to an MS-Windows codepage. | |
2851 * Returns zero if no codepage can be figured out. | |
2852 */ | |
2853 int | |
2854 encname2codepage(name) | |
2855 char_u *name; | |
2856 { | |
2857 int cp; | |
2858 char_u *p = name; | |
2859 int idx; | |
2860 | |
2861 if (STRNCMP(p, "8bit-", 5) == 0) | |
2862 p += 5; | |
2863 else if (STRNCMP(p_enc, "2byte-", 6) == 0) | |
2864 p += 6; | |
2865 | |
2866 if (p[0] == 'c' && p[1] == 'p') | |
2867 cp = atoi(p + 2); | |
2868 else if ((idx = enc_canon_search(p)) >= 0) | |
2869 cp = enc_canon_table[idx].codepage; | |
2870 else | |
2871 return 0; | |
2872 if (IsValidCodePage(cp)) | |
2873 return cp; | |
2874 return 0; | |
2875 } | |
2876 #endif | |
2877 | |
2878 # if defined(USE_ICONV) || defined(PROTO) | |
2879 | |
2880 static char_u *iconv_string __ARGS((vimconv_T *vcp, char_u *str, int slen, int *unconvlenp)); | |
2881 | |
2882 /* | |
2883 * Call iconv_open() with a check if iconv() works properly (there are broken | |
2884 * versions). | |
2885 * Returns (void *)-1 if failed. | |
2886 * (should return iconv_t, but that causes problems with prototypes). | |
2887 */ | |
2888 void * | |
2889 my_iconv_open(to, from) | |
2890 char_u *to; | |
2891 char_u *from; | |
2892 { | |
2893 iconv_t fd; | |
2894 #define ICONV_TESTLEN 400 | |
2895 char_u tobuf[ICONV_TESTLEN]; | |
2896 char *p; | |
2897 size_t tolen; | |
2898 static int iconv_ok = -1; | |
2899 | |
2900 if (iconv_ok == FALSE) | |
2901 return (void *)-1; /* detected a broken iconv() previously */ | |
2902 | |
2903 #ifdef DYNAMIC_ICONV | |
2904 /* Check if the iconv.dll can be found. */ | |
2905 if (!iconv_enabled(TRUE)) | |
2906 return (void *)-1; | |
2907 #endif | |
2908 | |
2909 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); | |
2910 | |
2911 if (fd != (iconv_t)-1 && iconv_ok == -1) | |
2912 { | |
2913 /* | |
2914 * Do a dummy iconv() call to check if it actually works. There is a | |
2915 * version of iconv() on Linux that is broken. We can't ignore it, | |
2916 * because it's wide-spread. The symptoms are that after outputting | |
2917 * the initial shift state the "to" pointer is NULL and conversion | |
2918 * stops for no apparent reason after about 8160 characters. | |
2919 */ | |
2920 p = (char *)tobuf; | |
2921 tolen = ICONV_TESTLEN; | |
2922 (void)iconv(fd, NULL, NULL, &p, &tolen); | |
2923 if (p == NULL) | |
2924 { | |
2925 iconv_ok = FALSE; | |
2926 iconv_close(fd); | |
2927 fd = (iconv_t)-1; | |
2928 } | |
2929 else | |
2930 iconv_ok = TRUE; | |
2931 } | |
2932 | |
2933 return (void *)fd; | |
2934 } | |
2935 | |
2936 /* | |
2937 * Convert the string "str[slen]" with iconv(). | |
2938 * If "unconvlenp" is not NULL handle the string ending in an incomplete | |
2939 * sequence and set "*unconvlenp" to the length of it. | |
2940 * Returns the converted string in allocated memory. NULL for an error. | |
2941 */ | |
2942 static char_u * | |
2943 iconv_string(vcp, str, slen, unconvlenp) | |
2944 vimconv_T *vcp; | |
2945 char_u *str; | |
2946 int slen; | |
2947 int *unconvlenp; | |
2948 { | |
2949 const char *from; | |
2950 size_t fromlen; | |
2951 char *to; | |
2952 size_t tolen; | |
2953 size_t len = 0; | |
2954 size_t done = 0; | |
2955 char_u *result = NULL; | |
2956 char_u *p; | |
2957 int l; | |
2958 | |
2959 from = (char *)str; | |
2960 fromlen = slen; | |
2961 for (;;) | |
2962 { | |
2963 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) | |
2964 { | |
2965 /* Allocate enough room for most conversions. When re-allocating | |
2966 * increase the buffer size. */ | |
2967 len = len + fromlen * 2 + 40; | |
2968 p = alloc((unsigned)len); | |
2969 if (p != NULL && done > 0) | |
2970 mch_memmove(p, result, done); | |
2971 vim_free(result); | |
2972 result = p; | |
2973 if (result == NULL) /* out of memory */ | |
2974 break; | |
2975 } | |
2976 | |
2977 to = (char *)result + done; | |
2978 tolen = len - done - 2; | |
2979 /* Avoid a warning for systems with a wrong iconv() prototype by | |
2980 * casting the second argument to void *. */ | |
2981 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) | |
2982 != (size_t)-1) | |
2983 { | |
2984 /* Finished, append a NUL. */ | |
2985 *to = NUL; | |
2986 break; | |
2987 } | |
2988 | |
2989 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded | |
2990 * iconv library may use one of them. */ | |
2991 if (!vcp->vc_fail && unconvlenp != NULL | |
2992 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) | |
2993 { | |
2994 /* Handle an incomplete sequence at the end. */ | |
2995 *to = NUL; | |
2996 *unconvlenp = fromlen; | |
2997 break; | |
2998 } | |
2999 | |
3000 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded | |
3001 * iconv library may use one of them. */ | |
3002 else if (!vcp->vc_fail | |
3003 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ | |
3004 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) | |
3005 { | |
3006 /* Can't convert: insert a '?' and skip a character. This assumes | |
3007 * conversion from 'encoding' to something else. In other | |
3008 * situations we don't know what to skip anyway. */ | |
3009 *to++ = '?'; | |
3010 if ((*mb_ptr2cells)((char_u *)from) > 1) | |
3011 *to++ = '?'; | |
3012 l = (*mb_ptr2len_check)((char_u *)from); | |
3013 from += l; | |
3014 fromlen -= l; | |
3015 } | |
3016 else if (ICONV_ERRNO != ICONV_E2BIG) | |
3017 { | |
3018 /* conversion failed */ | |
3019 vim_free(result); | |
3020 result = NULL; | |
3021 break; | |
3022 } | |
3023 /* Not enough room or skipping illegal sequence. */ | |
3024 done = to - (char *)result; | |
3025 } | |
3026 return result; | |
3027 } | |
3028 | |
3029 # if defined(DYNAMIC_ICONV) || defined(PROTO) | |
3030 /* | |
3031 * Dynamically load the "iconv.dll" on Win32. | |
3032 */ | |
3033 | |
3034 #ifndef DYNAMIC_ICONV /* just generating prototypes */ | |
3035 # define HINSTANCE int | |
3036 #endif | |
3037 HINSTANCE hIconvDLL = 0; | |
3038 HINSTANCE hMsvcrtDLL = 0; | |
3039 | |
3040 # ifndef DYNAMIC_ICONV_DLL | |
3041 # define DYNAMIC_ICONV_DLL "iconv.dll" | |
3042 # define DYNAMIC_ICONV_DLL_ALT "libiconv.dll" | |
3043 # endif | |
3044 # ifndef DYNAMIC_MSVCRT_DLL | |
3045 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" | |
3046 # endif | |
3047 | |
3048 /* | |
3049 * Try opening the iconv.dll and return TRUE if iconv() can be used. | |
3050 */ | |
3051 int | |
3052 iconv_enabled(verbose) | |
3053 int verbose; | |
3054 { | |
3055 if (hIconvDLL != 0 && hMsvcrtDLL != 0) | |
3056 return TRUE; | |
3057 hIconvDLL = LoadLibrary(DYNAMIC_ICONV_DLL); | |
3058 if (hIconvDLL == 0) /* sometimes it's called libiconv.dll */ | |
3059 hIconvDLL = LoadLibrary(DYNAMIC_ICONV_DLL_ALT); | |
3060 if (hIconvDLL != 0) | |
3061 hMsvcrtDLL = LoadLibrary(DYNAMIC_MSVCRT_DLL); | |
3062 if (hIconvDLL == 0 || hMsvcrtDLL == 0) | |
3063 { | |
3064 /* Only give the message when 'verbose' is set, otherwise it might be | |
3065 * done whenever a conversion is attempted. */ | |
3066 if (verbose && p_verbose > 0) | |
3067 EMSG2(_(e_loadlib), | |
3068 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); | |
3069 iconv_end(); | |
3070 return FALSE; | |
3071 } | |
3072 | |
3073 *((FARPROC*)&iconv) = GetProcAddress(hIconvDLL, "libiconv"); | |
3074 *((FARPROC*)&iconv_open) = GetProcAddress(hIconvDLL, "libiconv_open"); | |
3075 *((FARPROC*)&iconv_close) = GetProcAddress(hIconvDLL, "libiconv_close"); | |
3076 *((FARPROC*)&iconvctl) = GetProcAddress(hIconvDLL, "libiconvctl"); | |
3077 *((FARPROC*)&iconv_errno) = GetProcAddress(hMsvcrtDLL, "_errno"); | |
3078 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL | |
3079 || iconvctl == NULL || iconv_errno == NULL) | |
3080 { | |
3081 iconv_end(); | |
3082 if (verbose && p_verbose > 0) | |
3083 EMSG2(_(e_loadfunc), "for libiconv"); | |
3084 return FALSE; | |
3085 } | |
3086 return TRUE; | |
3087 } | |
3088 | |
3089 void | |
3090 iconv_end() | |
3091 { | |
3092 /* Don't use iconv() when inputting or outputting characters. */ | |
3093 if (input_conv.vc_type == CONV_ICONV) | |
3094 convert_setup(&input_conv, NULL, NULL); | |
3095 if (output_conv.vc_type == CONV_ICONV) | |
3096 convert_setup(&output_conv, NULL, NULL); | |
3097 | |
3098 if (hIconvDLL != 0) | |
3099 FreeLibrary(hIconvDLL); | |
3100 if (hMsvcrtDLL != 0) | |
3101 FreeLibrary(hMsvcrtDLL); | |
3102 hIconvDLL = 0; | |
3103 hMsvcrtDLL = 0; | |
3104 } | |
3105 # endif /* DYNAMIC_ICONV */ | |
3106 # endif /* USE_ICONV */ | |
3107 | |
3108 #endif /* FEAT_MBYTE */ | |
3109 | |
3110 #if defined(FEAT_XIM) || defined(PROTO) | |
3111 | |
3112 # ifdef FEAT_GUI_GTK | |
3113 static int xim_has_preediting INIT(= FALSE); /* IM current status */ | |
3114 | |
3115 /* | |
3116 * Set preedit_start_col to the current cursor position. | |
3117 */ | |
3118 static void | |
3119 init_preedit_start_col(void) | |
3120 { | |
3121 if (State & CMDLINE) | |
3122 preedit_start_col = cmdline_getvcol_cursor(); | |
3123 else if (curwin != NULL) | |
3124 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); | |
3125 /* Prevent that preediting marks the buffer as changed. */ | |
3126 xim_changed_while_preediting = curbuf->b_changed; | |
3127 } | |
3128 # endif | |
3129 | |
3130 # if defined(HAVE_GTK2) && !defined(PROTO) | |
3131 | |
3132 static int im_is_active = FALSE; /* IM is enabled for current mode */ | |
3133 static int im_preedit_cursor = 0; /* cursor offset in characters */ | |
3134 static int im_preedit_trailing = 0; /* number of characters after cursor */ | |
3135 | |
3136 static unsigned long im_commit_handler_id = 0; | |
3137 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; | |
3138 static unsigned int im_activatekey_state = 0; | |
3139 | |
3140 void | |
3141 im_set_active(int active) | |
3142 { | |
3143 int was_active; | |
3144 | |
3145 was_active = !!im_is_active; | |
3146 im_is_active = (active && !p_imdisable); | |
3147 | |
3148 if (im_is_active != was_active) | |
3149 xim_reset(); | |
3150 } | |
3151 | |
3152 void | |
3153 xim_set_focus(int focus) | |
3154 { | |
3155 if (xic != NULL) | |
3156 { | |
3157 if (focus) | |
3158 gtk_im_context_focus_in(xic); | |
3159 else | |
3160 gtk_im_context_focus_out(xic); | |
3161 } | |
3162 } | |
3163 | |
3164 void | |
3165 im_set_position(int row, int col) | |
3166 { | |
3167 if (xic != NULL) | |
3168 { | |
3169 GdkRectangle area; | |
3170 | |
3171 area.x = FILL_X(col); | |
3172 area.y = FILL_Y(row); | |
3173 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); | |
3174 area.height = gui.char_height; | |
3175 | |
3176 gtk_im_context_set_cursor_location(xic, &area); | |
3177 } | |
3178 } | |
3179 | |
3180 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */ | |
3181 void | |
3182 xim_set_preedit(void) | |
3183 { | |
3184 im_set_position(gui.row, gui.col); | |
3185 } | |
3186 # endif | |
3187 | |
3188 static void | |
3189 im_add_to_input(char_u *str, int len) | |
3190 { | |
3191 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */ | |
3192 if (input_conv.vc_type != CONV_NONE) | |
3193 { | |
3194 str = string_convert(&input_conv, str, &len); | |
3195 g_return_if_fail(str != NULL); | |
3196 } | |
3197 | |
3198 add_to_input_buf_csi(str, len); | |
3199 | |
3200 if (input_conv.vc_type != CONV_NONE) | |
3201 vim_free(str); | |
3202 | |
3203 if (p_mh) /* blank out the pointer if necessary */ | |
3204 gui_mch_mousehide(TRUE); | |
3205 } | |
3206 | |
3207 static void | |
3208 im_delete_preedit(void) | |
3209 { | |
3210 char_u bskey[] = {CSI, 'k', 'b'}; | |
3211 char_u delkey[] = {CSI, 'k', 'D'}; | |
3212 | |
3213 if (State & NORMAL) | |
3214 { | |
3215 im_preedit_cursor = 0; | |
3216 return; | |
3217 } | |
3218 for (; im_preedit_cursor > 0; --im_preedit_cursor) | |
3219 add_to_input_buf(bskey, (int)sizeof(bskey)); | |
3220 | |
3221 for (; im_preedit_trailing > 0; --im_preedit_trailing) | |
3222 add_to_input_buf(delkey, (int)sizeof(delkey)); | |
3223 } | |
3224 | |
3225 static void | |
3226 im_correct_cursor(int num_move_back) | |
3227 { | |
3228 char_u backkey[] = {CSI, 'k', 'l'}; | |
3229 | |
3230 if (State & NORMAL) | |
3231 return; | |
3232 # ifdef FEAT_RIGHTLEFT | |
3233 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) | |
3234 backkey[2] = 'r'; | |
3235 # endif | |
3236 for (; num_move_back > 0; --num_move_back) | |
3237 add_to_input_buf(backkey, (int)sizeof(backkey)); | |
3238 } | |
3239 | |
3240 static int xim_expected_char = NUL; | |
3241 static int xim_ignored_char = FALSE; | |
3242 | |
3243 /* | |
3244 * Update the mode and cursor while in an IM callback. | |
3245 */ | |
3246 static void | |
3247 im_show_info(void) | |
3248 { | |
3249 int old_vgetc_busy; | |
3250 old_vgetc_busy = vgetc_busy; | |
3251 vgetc_busy = TRUE; | |
3252 showmode(); | |
3253 vgetc_busy = old_vgetc_busy; | |
3254 setcursor(); | |
3255 out_flush(); | |
3256 } | |
3257 | |
3258 /* | |
3259 * Callback invoked when the user finished preediting. | |
3260 * Put the final string into the input buffer. | |
3261 */ | |
3262 /*ARGSUSED0*/ | |
3263 static void | |
3264 im_commit_cb(GtkIMContext *context, const gchar *str, gpointer data) | |
3265 { | |
3266 int slen = (int)STRLEN(str); | |
3267 int add_to_input = TRUE; | |
3268 int clen; | |
3269 int len = slen; | |
3270 int commit_with_preedit = TRUE; | |
3271 char_u *im_str, *p; | |
3272 | |
3273 #ifdef XIM_DEBUG | |
3274 xim_log("im_commit_cb(): %s\n", str); | |
3275 #endif | |
3276 | |
3277 /* The imhangul module doesn't reset the preedit string before | |
3278 * committing. Call im_delete_preedit() to work around that. */ | |
3279 im_delete_preedit(); | |
3280 | |
3281 /* Indicate that preediting has finished. */ | |
3282 if (preedit_start_col == MAXCOL) | |
3283 { | |
3284 init_preedit_start_col(); | |
3285 commit_with_preedit = FALSE; | |
3286 } | |
3287 | |
3288 /* The thing which setting "preedit_start_col" to MAXCOL means that | |
3289 * "preedit_start_col" will be set forcely when calling | |
3290 * preedit_changed_cb() next time. | |
3291 * "preedit_start_col" should not reset with MAXCOL on this part. Vim | |
3292 * is simulating the preediting by using add_to_input_str(). when | |
3293 * preedit begin immediately before committed, the typebuf is not | |
3294 * flushed to screen, then it can't get correct "preedit_start_col". | |
3295 * Thus, it should calculate the cells by adding cells of the committed | |
3296 * string. */ | |
3297 if (input_conv.vc_type != CONV_NONE) | |
3298 { | |
3299 im_str = string_convert(&input_conv, (char_u *)str, &len); | |
3300 g_return_if_fail(im_str != NULL); | |
3301 } | |
3302 else | |
3303 im_str = (char_u *)str; | |
3304 clen = 0; | |
3305 for (p = im_str; p < im_str + len; p += (*mb_ptr2len_check)(p)) | |
3306 clen += (*mb_ptr2cells)(p); | |
3307 if (input_conv.vc_type != CONV_NONE) | |
3308 vim_free(im_str); | |
3309 preedit_start_col += clen; | |
3310 | |
3311 /* Is this a single character that matches a keypad key that's just | |
3312 * been pressed? If so, we don't want it to be entered as such - let | |
3313 * us carry on processing the raw keycode so that it may be used in | |
3314 * mappings as <kSomething>. */ | |
3315 if (xim_expected_char != NUL) | |
3316 { | |
3317 /* We're currently processing a keypad or other special key */ | |
3318 if (slen == 1 && str[0] == xim_expected_char) | |
3319 { | |
3320 /* It's a match - don't do it here */ | |
3321 xim_ignored_char = TRUE; | |
3322 add_to_input = FALSE; | |
3323 } | |
3324 else | |
3325 { | |
3326 /* Not a match */ | |
3327 xim_ignored_char = FALSE; | |
3328 } | |
3329 } | |
3330 | |
3331 if (add_to_input) | |
3332 im_add_to_input((char_u *)str, slen); | |
3333 | |
3334 /* Inserting chars while "im_is_active" is set does not cause a change of | |
3335 * buffer. When the chars are committed the buffer must be marked as | |
3336 * changed. */ | |
3337 if (!commit_with_preedit) | |
3338 preedit_start_col = MAXCOL; | |
3339 | |
3340 /* This flag is used in changed() at next call. */ | |
3341 xim_changed_while_preediting = TRUE; | |
3342 | |
3343 if (gtk_main_level() > 0) | |
3344 gtk_main_quit(); | |
3345 } | |
3346 | |
3347 /* | |
3348 * Callback invoked after start to the preedit. | |
3349 */ | |
3350 /*ARGSUSED*/ | |
3351 static void | |
3352 im_preedit_start_cb(GtkIMContext *context, gpointer data) | |
3353 { | |
3354 #ifdef XIM_DEBUG | |
3355 xim_log("im_preedit_start_cb()\n"); | |
3356 #endif | |
3357 | |
3358 im_is_active = TRUE; | |
3359 gui_update_cursor(TRUE, FALSE); | |
3360 } | |
3361 | |
3362 /* | |
3363 * Callback invoked after end to the preedit. | |
3364 */ | |
3365 /*ARGSUSED*/ | |
3366 static void | |
3367 im_preedit_end_cb(GtkIMContext *context, gpointer data) | |
3368 { | |
3369 #ifdef XIM_DEBUG | |
3370 xim_log("im_preedit_end_cb()\n"); | |
3371 #endif | |
3372 im_delete_preedit(); | |
3373 | |
3374 /* Indicate that preediting has finished */ | |
3375 preedit_start_col = MAXCOL; | |
3376 xim_has_preediting = FALSE; | |
3377 | |
3378 im_is_active = FALSE; | |
3379 gui_update_cursor(TRUE, FALSE); | |
3380 im_show_info(); | |
3381 } | |
3382 | |
3383 /* | |
3384 * Callback invoked after changes to the preedit string. If the preedit | |
3385 * string was empty before, remember the preedit start column so we know | |
3386 * where to apply feedback attributes. Delete the previous preedit string | |
3387 * if there was one, save the new preedit cursor offset, and put the new | |
3388 * string into the input buffer. | |
3389 * | |
3390 * TODO: The pragmatic "put into input buffer" approach used here has | |
3391 * several fundamental problems: | |
3392 * | |
3393 * - The characters in the preedit string are subject to remapping. | |
3394 * That's broken, only the finally committed string should be remapped. | |
3395 * | |
3396 * - There is a race condition involved: The retrieved value for the | |
3397 * current cursor position will be wrong if any unprocessed characters | |
3398 * are still queued in the input buffer. | |
3399 * | |
3400 * - Due to the lack of synchronization between the file buffer in memory | |
3401 * and any typed characters, it's practically impossible to implement the | |
3402 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM | |
3403 * modules for languages such as Thai are likely to rely on this feature | |
3404 * for proper operation. | |
3405 * | |
3406 * Conclusions: I think support for preediting needs to be moved to the | |
3407 * core parts of Vim. Ideally, until it has been committed, the preediting | |
3408 * string should only be displayed and not affect the buffer content at all. | |
3409 * The question how to deal with the synchronization issue still remains. | |
3410 * Circumventing the input buffer is probably not desirable. Anyway, I think | |
3411 * implementing "retrieve_surrounding" is the only hard problem. | |
3412 * | |
3413 * One way to solve all of this in a clean manner would be to queue all key | |
3414 * press/release events "as is" in the input buffer, and apply the IM filtering | |
3415 * at the receiving end of the queue. This, however, would have a rather large | |
3416 * impact on the code base. If there is an easy way to force processing of all | |
3417 * remaining input from within the "retrieve_surrounding" signal handler, this | |
3418 * might not be necessary. Gotta ask on vim-dev for opinions. | |
3419 */ | |
3420 /*ARGSUSED1*/ | |
3421 static void | |
3422 im_preedit_changed_cb(GtkIMContext *context, gpointer data) | |
3423 { | |
3424 char *preedit_string = NULL; | |
3425 int cursor_index = 0; | |
3426 int num_move_back = 0; | |
3427 char_u *str; | |
3428 char_u *p; | |
3429 int i; | |
3430 | |
3431 gtk_im_context_get_preedit_string(context, | |
3432 &preedit_string, NULL, | |
3433 &cursor_index); | |
3434 | |
3435 #ifdef XIM_DEBUG | |
3436 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); | |
3437 #endif | |
3438 | |
3439 g_return_if_fail(preedit_string != NULL); /* just in case */ | |
3440 | |
3441 /* If preedit_start_col is MAXCOL set it to the current cursor position. */ | |
3442 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') | |
3443 { | |
3444 xim_has_preediting = TRUE; | |
3445 | |
3446 /* Urgh, this breaks if the input buffer isn't empty now */ | |
3447 init_preedit_start_col(); | |
3448 } | |
3449 else if (cursor_index == 0 && preedit_string[0] == '\0') | |
3450 { | |
3451 if (preedit_start_col == MAXCOL) | |
3452 xim_has_preediting = FALSE; | |
3453 | |
3454 /* If at the start position (after typing backspace) | |
3455 * preedit_start_col must be reset. */ | |
3456 preedit_start_col = MAXCOL; | |
3457 } | |
3458 | |
3459 im_delete_preedit(); | |
3460 | |
3461 /* | |
3462 * Compute the end of the preediting area: "preedit_end_col". | |
3463 * According to the documentation of gtk_im_context_get_preedit_string(), | |
3464 * the cursor_pos output argument returns the offset in bytes. This is | |
3465 * unfortunately not true -- real life shows the offset is in characters, | |
3466 * and the GTK+ source code agrees with me. Will file a bug later. | |
3467 */ | |
3468 if (preedit_start_col != MAXCOL) | |
3469 preedit_end_col = preedit_start_col; | |
3470 str = (char_u *)preedit_string; | |
3471 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) | |
3472 { | |
3473 int is_composing; | |
3474 | |
3475 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); | |
3476 /* | |
3477 * These offsets are used as counters when generating <BS> and <Del> | |
3478 * to delete the preedit string. So don't count composing characters | |
3479 * unless 'delcombine' is enabled. | |
3480 */ | |
3481 if (!is_composing || p_deco) | |
3482 { | |
3483 if (i < cursor_index) | |
3484 ++im_preedit_cursor; | |
3485 else | |
3486 ++im_preedit_trailing; | |
3487 } | |
3488 if (!is_composing && i >= cursor_index) | |
3489 { | |
3490 /* This is essentially the same as im_preedit_trailing, except | |
3491 * composing characters are not counted even if p_deco is set. */ | |
3492 ++num_move_back; | |
3493 } | |
3494 if (preedit_start_col != MAXCOL) | |
3495 preedit_end_col += utf_ptr2cells(p); | |
3496 } | |
3497 | |
3498 if (p > str) | |
3499 { | |
3500 im_add_to_input(str, (int)(p - str)); | |
3501 im_correct_cursor(num_move_back); | |
3502 } | |
3503 | |
3504 g_free(preedit_string); | |
3505 | |
3506 if (gtk_main_level() > 0) | |
3507 gtk_main_quit(); | |
3508 } | |
3509 | |
3510 /* | |
3511 * Translate the Pango attributes at iter to Vim highlighting attributes. | |
3512 * Ignore attributes not supported by Vim highlighting. This shouldn't have | |
3513 * too much impact -- right now we handle even more attributes than necessary | |
3514 * for the IM modules I tested with. | |
3515 */ | |
3516 static int | |
3517 translate_pango_attributes(PangoAttrIterator *iter) | |
3518 { | |
3519 PangoAttribute *attr; | |
3520 int char_attr = HL_NORMAL; | |
3521 | |
3522 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); | |
3523 if (attr != NULL && ((PangoAttrInt *)attr)->value | |
3524 != (int)PANGO_UNDERLINE_NONE) | |
3525 char_attr |= HL_UNDERLINE; | |
3526 | |
3527 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); | |
3528 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) | |
3529 char_attr |= HL_BOLD; | |
3530 | |
3531 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); | |
3532 if (attr != NULL && ((PangoAttrInt *)attr)->value | |
3533 != (int)PANGO_STYLE_NORMAL) | |
3534 char_attr |= HL_ITALIC; | |
3535 | |
3536 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); | |
3537 if (attr != NULL) | |
3538 { | |
3539 const PangoColor *color = &((PangoAttrColor *)attr)->color; | |
3540 | |
3541 /* Assume inverse if black background is requested */ | |
3542 if ((color->red | color->green | color->blue) == 0) | |
3543 char_attr |= HL_INVERSE; | |
3544 } | |
3545 | |
3546 return char_attr; | |
3547 } | |
3548 | |
3549 /* | |
3550 * Retrieve the highlighting attributes at column col in the preedit string. | |
3551 * Return -1 if not in preediting mode or if col is out of range. | |
3552 */ | |
3553 int | |
3554 im_get_feedback_attr(int col) | |
3555 { | |
3556 char *preedit_string = NULL; | |
3557 PangoAttrList *attr_list = NULL; | |
3558 int char_attr = -1; | |
3559 | |
3560 if (xic == NULL) | |
3561 return char_attr; | |
3562 | |
3563 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); | |
3564 | |
3565 if (preedit_string != NULL && attr_list != NULL) | |
3566 { | |
3567 int index; | |
3568 | |
3569 /* Get the byte index as used by PangoAttrIterator */ | |
3570 for (index = 0; col > 0 && preedit_string[index] != '\0'; --col) | |
3571 index += utfc_ptr2len_check((char_u *)preedit_string + index); | |
3572 | |
3573 if (preedit_string[index] != '\0') | |
3574 { | |
3575 PangoAttrIterator *iter; | |
3576 int start, end; | |
3577 | |
3578 char_attr = HL_NORMAL; | |
3579 iter = pango_attr_list_get_iterator(attr_list); | |
3580 | |
3581 /* Extract all relevant attributes from the list. */ | |
3582 do | |
3583 { | |
3584 pango_attr_iterator_range(iter, &start, &end); | |
3585 | |
3586 if (index >= start && index < end) | |
3587 char_attr |= translate_pango_attributes(iter); | |
3588 } | |
3589 while (pango_attr_iterator_next(iter)); | |
3590 | |
3591 pango_attr_iterator_destroy(iter); | |
3592 } | |
3593 } | |
3594 | |
3595 if (attr_list != NULL) | |
3596 pango_attr_list_unref(attr_list); | |
3597 g_free(preedit_string); | |
3598 | |
3599 return char_attr; | |
3600 } | |
3601 | |
3602 void | |
3603 xim_init(void) | |
3604 { | |
3605 #ifdef XIM_DEBUG | |
3606 xim_log("xim_init()\n"); | |
3607 #endif | |
3608 | |
3609 g_return_if_fail(gui.drawarea != NULL); | |
3610 g_return_if_fail(gui.drawarea->window != NULL); | |
3611 | |
3612 xic = gtk_im_multicontext_new(); | |
3613 g_object_ref(xic); | |
3614 | |
3615 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", | |
3616 G_CALLBACK(&im_commit_cb), NULL); | |
3617 g_signal_connect(G_OBJECT(xic), "preedit_changed", | |
3618 G_CALLBACK(&im_preedit_changed_cb), NULL); | |
3619 g_signal_connect(G_OBJECT(xic), "preedit_start", | |
3620 G_CALLBACK(&im_preedit_start_cb), NULL); | |
3621 g_signal_connect(G_OBJECT(xic), "preedit_end", | |
3622 G_CALLBACK(&im_preedit_end_cb), NULL); | |
3623 | |
3624 gtk_im_context_set_client_window(xic, gui.drawarea->window); | |
3625 } | |
3626 | |
3627 void | |
3628 im_shutdown(void) | |
3629 { | |
3630 #ifdef XIM_DEBUG | |
3631 xim_log("im_shutdown()\n"); | |
3632 #endif | |
3633 | |
3634 if (xic != NULL) | |
3635 { | |
3636 gtk_im_context_focus_out(xic); | |
3637 g_object_unref(xic); | |
3638 xic = NULL; | |
3639 } | |
3640 im_is_active = FALSE; | |
3641 im_commit_handler_id = 0; | |
3642 preedit_start_col = MAXCOL; | |
3643 xim_has_preediting = FALSE; | |
3644 } | |
3645 | |
3646 /* | |
3647 * Convert the string argument to keyval and state for GdkEventKey. | |
3648 * If str is valid return TRUE, otherwise FALSE. | |
3649 * | |
3650 * See 'imactivatekey' for documentation of the format. | |
3651 */ | |
3652 static int | |
3653 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) | |
3654 { | |
3655 const char *mods_end; | |
3656 unsigned tmp_keyval; | |
3657 unsigned tmp_state = 0; | |
3658 | |
3659 mods_end = strrchr(str, '-'); | |
3660 mods_end = (mods_end != NULL) ? mods_end + 1 : str; | |
3661 | |
3662 /* Parse modifier keys */ | |
3663 while (str < mods_end) | |
3664 switch (*str++) | |
3665 { | |
3666 case '-': break; | |
3667 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; | |
3668 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; | |
3669 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; | |
3670 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; | |
3671 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; | |
3672 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; | |
3673 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; | |
3674 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; | |
3675 default: | |
3676 return FALSE; | |
3677 } | |
3678 | |
3679 tmp_keyval = gdk_keyval_from_name(str); | |
3680 | |
3681 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) | |
3682 return FALSE; | |
3683 | |
3684 if (keyval != NULL) | |
3685 *keyval = tmp_keyval; | |
3686 if (state != NULL) | |
3687 *state = tmp_state; | |
3688 | |
3689 return TRUE; | |
3690 } | |
3691 | |
3692 /* | |
3693 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an | |
3694 * empty string is also regarded as valid. | |
3695 * | |
3696 * Note: The numerical key value of p_imak is cached if it was valid; thus | |
3697 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever | |
3698 * 'imak' changes. This is currently the case but not obvious -- should | |
3699 * probably rename the function for clarity. | |
3700 */ | |
3701 int | |
3702 im_xim_isvalid_imactivate(void) | |
3703 { | |
3704 if (p_imak[0] == NUL) | |
3705 { | |
3706 im_activatekey_keyval = GDK_VoidSymbol; | |
3707 im_activatekey_state = 0; | |
3708 return TRUE; | |
3709 } | |
3710 | |
3711 return im_string_to_keyval((const char *)p_imak, | |
3712 &im_activatekey_keyval, | |
3713 &im_activatekey_state); | |
3714 } | |
3715 | |
3716 static void | |
3717 im_synthesize_keypress(unsigned int keyval, unsigned int state) | |
3718 { | |
3719 GdkEventKey *event; | |
3720 | |
3721 # ifdef HAVE_GTK_MULTIHEAD | |
3722 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); | |
3723 g_object_ref(gui.drawarea->window); /* unreffed by gdk_event_free() */ | |
3724 # else | |
3725 event = (GdkEventKey *)g_malloc0((gulong)sizeof(GdkEvent)); | |
3726 event->type = GDK_KEY_PRESS; | |
3727 # endif | |
3728 event->window = gui.drawarea->window; | |
3729 event->send_event = TRUE; | |
3730 event->time = GDK_CURRENT_TIME; | |
3731 event->state = state; | |
3732 event->keyval = keyval; | |
3733 event->hardware_keycode = /* needed for XIM */ | |
3734 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); | |
3735 event->length = 0; | |
3736 event->string = NULL; | |
3737 | |
3738 gtk_im_context_filter_keypress(xic, event); | |
3739 | |
3740 /* For consistency, also send the corresponding release event. */ | |
3741 event->type = GDK_KEY_RELEASE; | |
3742 event->send_event = FALSE; | |
3743 gtk_im_context_filter_keypress(xic, event); | |
3744 | |
3745 # ifdef HAVE_GTK_MULTIHEAD | |
3746 gdk_event_free((GdkEvent *)event); | |
3747 # else | |
3748 g_free(event); | |
3749 # endif | |
3750 } | |
3751 | |
3752 void | |
3753 xim_reset(void) | |
3754 { | |
3755 if (xic != NULL) | |
3756 { | |
3757 /* | |
3758 * The third-party imhangul module (and maybe others too) ignores | |
3759 * gtk_im_context_reset() or at least doesn't reset the active state. | |
3760 * Thus sending imactivatekey would turn it off if it was on before, | |
3761 * which is clearly not what we want. Fortunately we can work around | |
3762 * that for imhangul by sending GDK_Escape, but I don't know if it | |
3763 * works with all IM modules that support an activation key :/ | |
3764 * | |
3765 * An alternative approach would be to destroy the IM context and | |
3766 * recreate it. But that means loading/unloading the IM module on | |
3767 * every mode switch, which causes a quite noticable delay even on | |
3768 * my rather fast box... | |
3769 * * | |
3770 * Moreover, there are some XIM which cannot respond to | |
3771 * im_synthesize_keypress(). we hope that they reset by | |
3772 * xim_shutdown(). | |
3773 */ | |
3774 if (im_activatekey_keyval != GDK_VoidSymbol && im_is_active) | |
3775 im_synthesize_keypress(GDK_Escape, 0U); | |
3776 | |
3777 gtk_im_context_reset(xic); | |
3778 | |
3779 /* | |
3780 * HACK for Ami: This sequence of function calls makes Ami handle | |
3781 * the IM reset gratiously, without breaking loads of other stuff. | |
3782 * It seems to force English mode as well, which is exactly what we | |
3783 * want because it makes the Ami status display work reliably. | |
3784 */ | |
3785 gtk_im_context_set_use_preedit(xic, FALSE); | |
3786 | |
3787 if (p_imdisable) | |
3788 im_shutdown(); | |
3789 else | |
3790 { | |
3791 gtk_im_context_set_use_preedit(xic, TRUE); | |
3792 xim_set_focus(gui.in_focus); | |
3793 | |
3794 if (im_activatekey_keyval != GDK_VoidSymbol) | |
3795 { | |
3796 if (im_is_active) | |
3797 { | |
3798 g_signal_handler_block(xic, im_commit_handler_id); | |
3799 im_synthesize_keypress(im_activatekey_keyval, | |
3800 im_activatekey_state); | |
3801 g_signal_handler_unblock(xic, im_commit_handler_id); | |
3802 } | |
3803 } | |
3804 else | |
3805 { | |
3806 im_shutdown(); | |
3807 xim_init(); | |
3808 xim_set_focus(gui.in_focus); | |
3809 } | |
3810 } | |
3811 } | |
3812 | |
3813 preedit_start_col = MAXCOL; | |
3814 xim_has_preediting = FALSE; | |
3815 } | |
3816 | |
3817 int | |
3818 xim_queue_key_press_event(GdkEventKey *event, int down) | |
3819 { | |
3820 if (down) | |
3821 { | |
3822 /* | |
3823 * Workaround GTK2 XIM 'feature' that always converts keypad keys to | |
3824 * chars., even when not part of an IM sequence (ref. feature of | |
3825 * gdk/gdkkeyuni.c). | |
3826 * Flag any keypad keys that might represent a single char. | |
3827 * If this (on its own - i.e., not part of an IM sequence) is | |
3828 * committed while we're processing one of these keys, we can ignore | |
3829 * that commit and go ahead & process it ourselves. That way we can | |
3830 * still distinguish keypad keys for use in mappings. | |
3831 */ | |
3832 switch (event->keyval) | |
3833 { | |
3834 case GDK_KP_Add: xim_expected_char = '+'; break; | |
3835 case GDK_KP_Subtract: xim_expected_char = '-'; break; | |
3836 case GDK_KP_Divide: xim_expected_char = '/'; break; | |
3837 case GDK_KP_Multiply: xim_expected_char = '*'; break; | |
3838 case GDK_KP_Decimal: xim_expected_char = '.'; break; | |
3839 case GDK_KP_Equal: xim_expected_char = '='; break; | |
3840 case GDK_KP_0: xim_expected_char = '0'; break; | |
3841 case GDK_KP_1: xim_expected_char = '1'; break; | |
3842 case GDK_KP_2: xim_expected_char = '2'; break; | |
3843 case GDK_KP_3: xim_expected_char = '3'; break; | |
3844 case GDK_KP_4: xim_expected_char = '4'; break; | |
3845 case GDK_KP_5: xim_expected_char = '5'; break; | |
3846 case GDK_KP_6: xim_expected_char = '6'; break; | |
3847 case GDK_KP_7: xim_expected_char = '7'; break; | |
3848 case GDK_KP_8: xim_expected_char = '8'; break; | |
3849 case GDK_KP_9: xim_expected_char = '9'; break; | |
3850 default: xim_expected_char = NUL; | |
3851 } | |
3852 xim_ignored_char = FALSE; | |
3853 } | |
3854 | |
3855 /* | |
3856 * When typing fFtT, XIM may be activated. Thus it must pass | |
3857 * gtk_im_context_filter_keypress() in Normal mode. | |
3858 * And while doing :sh too. | |
3859 */ | |
3860 if (xic != NULL && !p_imdisable | |
3861 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) | |
3862 { | |
3863 /* | |
3864 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is | |
3865 * always aware of the current status of IM, and can even emulate | |
3866 * the activation key for modules that don't support one. | |
3867 */ | |
3868 if (event->keyval == im_activatekey_keyval | |
3869 && (event->state & im_activatekey_state) == im_activatekey_state) | |
3870 { | |
3871 unsigned int state_mask; | |
3872 | |
3873 /* Require the state of the 3 most used modifiers to match exactly. | |
3874 * Otherwise e.g. <S-C-space> would be unusable for other purposes | |
3875 * if the IM activate key is <S-space>. */ | |
3876 state_mask = im_activatekey_state; | |
3877 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK | |
3878 | (int)GDK_MOD1_MASK); | |
3879 | |
3880 if ((event->state & state_mask) != im_activatekey_state) | |
3881 return FALSE; | |
3882 | |
3883 /* Don't send it a second time on GDK_KEY_RELEASE. */ | |
3884 if (event->type != GDK_KEY_PRESS) | |
3885 return TRUE; | |
3886 | |
3887 if (map_to_exists_mode((char_u *)"", LANGMAP)) | |
3888 { | |
3889 im_set_active(FALSE); | |
3890 | |
3891 /* ":lmap" mappings exists, toggle use of mappings. */ | |
3892 State ^= LANGMAP; | |
3893 if (State & LANGMAP) | |
3894 { | |
3895 curbuf->b_p_iminsert = B_IMODE_NONE; | |
3896 State &= ~LANGMAP; | |
3897 } | |
3898 else | |
3899 { | |
3900 curbuf->b_p_iminsert = B_IMODE_LMAP; | |
3901 State |= LANGMAP; | |
3902 } | |
3903 return TRUE; | |
3904 } | |
3905 | |
3906 return gtk_im_context_filter_keypress(xic, event); | |
3907 } | |
3908 | |
3909 /* Don't filter events through the IM context if IM isn't active | |
3910 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module | |
3911 * not doing anything before the activation key was sent. */ | |
3912 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) | |
3913 { | |
3914 int imresult = gtk_im_context_filter_keypress(xic, event); | |
3915 | |
3916 /* Some XIM send following sequence: | |
3917 * 1. preedited string. | |
3918 * 2. committed string. | |
3919 * 3. line changed key. | |
3920 * 4. preedited string. | |
3921 * 5. remove preedited string. | |
3922 * if 3, Vim can't move back the above line for 5. | |
3923 * thus, this part should not parse the key. */ | |
3924 if (!imresult && preedit_start_col != MAXCOL | |
3925 && event->keyval == GDK_Return) | |
3926 { | |
3927 im_synthesize_keypress(GDK_Return, 0U); | |
3928 return FALSE; | |
3929 } | |
3930 | |
3931 /* If XIM tried to commit a keypad key as a single char., | |
3932 * ignore it so we can use the keypad key 'raw', for mappings. */ | |
3933 if (xim_expected_char != NUL && xim_ignored_char) | |
3934 /* We had a keypad key, and XIM tried to thieve it */ | |
3935 return FALSE; | |
3936 | |
3937 /* Normal processing */ | |
3938 return imresult; | |
3939 } | |
3940 } | |
3941 | |
3942 return FALSE; | |
3943 } | |
3944 | |
3945 int | |
3946 im_get_status(void) | |
3947 { | |
3948 return im_is_active; | |
3949 } | |
3950 | |
3951 # else /* !HAVE_GTK2 */ | |
3952 | |
3953 static int xim_is_active = FALSE; /* XIM should be active in the current | |
3954 mode */ | |
3955 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */ | |
3956 #ifdef FEAT_GUI_X11 | |
3957 static XIMStyle input_style; | |
3958 static int status_area_enabled = TRUE; | |
3959 #endif | |
3960 | |
3961 #ifdef FEAT_GUI_GTK | |
3962 # ifdef WIN3264 | |
3963 # include <gdk/gdkwin32.h> | |
3964 # else | |
3965 # include <gdk/gdkx.h> | |
3966 # endif | |
3967 #else | |
3968 # ifdef PROTO | |
3969 /* Define a few things to be able to generate prototypes while not configured | |
3970 * for GTK. */ | |
3971 # define GSList int | |
3972 # define gboolean int | |
3973 typedef int GdkEvent; | |
3974 typedef int GdkEventKey; | |
3975 # define GdkIC int | |
3976 # endif | |
3977 #endif | |
3978 | |
3979 #if defined(FEAT_GUI_GTK) || defined(PROTO) | |
3980 static int preedit_buf_len = 0; | |
3981 static int xim_can_preediting INIT(= FALSE); /* XIM in showmode() */ | |
3982 static int xim_input_style; | |
3983 #ifndef FEAT_GUI_GTK | |
3984 # define gboolean int | |
3985 #endif | |
3986 static gboolean use_status_area = 0; | |
3987 | |
3988 static int im_xim_str2keycode __ARGS((unsigned int *code, unsigned int *state)); | |
3989 static void im_xim_send_event_imactivate __ARGS((void)); | |
3990 | |
3991 /* | |
3992 * Convert string to keycode and state for XKeyEvent. | |
3993 * When string is valid return OK, when invalid return FAIL. | |
3994 * | |
3995 * See 'imactivatekey' documentation for the format. | |
3996 */ | |
3997 static int | |
3998 im_xim_str2keycode(code, state) | |
3999 unsigned int *code; | |
4000 unsigned int *state; | |
4001 { | |
4002 int retval = OK; | |
4003 int len; | |
4004 unsigned keycode = 0, keystate = 0; | |
4005 Window window; | |
4006 Display *display; | |
4007 char_u *flag_end; | |
4008 char_u *str; | |
4009 | |
4010 if (*p_imak != NUL) | |
4011 { | |
4012 len = STRLEN(p_imak); | |
4013 for (flag_end = p_imak + len - 1; | |
4014 flag_end > p_imak && *flag_end != '-'; --flag_end) | |
4015 ; | |
4016 | |
4017 /* Parse modifier keys */ | |
4018 for (str = p_imak; str < flag_end; ++str) | |
4019 { | |
4020 switch (*str) | |
4021 { | |
4022 case 's': case 'S': | |
4023 keystate |= ShiftMask; | |
4024 break; | |
4025 case 'l': case 'L': | |
4026 keystate |= LockMask; | |
4027 break; | |
4028 case 'c': case 'C': | |
4029 keystate |= ControlMask; | |
4030 break; | |
4031 case '1': | |
4032 keystate |= Mod1Mask; | |
4033 break; | |
4034 case '2': | |
4035 keystate |= Mod2Mask; | |
4036 break; | |
4037 case '3': | |
4038 keystate |= Mod3Mask; | |
4039 break; | |
4040 case '4': | |
4041 keystate |= Mod4Mask; | |
4042 break; | |
4043 case '5': | |
4044 keystate |= Mod5Mask; | |
4045 break; | |
4046 case '-': | |
4047 break; | |
4048 default: | |
4049 retval = FAIL; | |
4050 } | |
4051 } | |
4052 if (*str == '-') | |
4053 ++str; | |
4054 | |
4055 /* Get keycode from string. */ | |
4056 gui_get_x11_windis(&window, &display); | |
4057 if (display) | |
4058 keycode = XKeysymToKeycode(display, XStringToKeysym((char *)str)); | |
4059 if (keycode == 0) | |
4060 retval = FAIL; | |
4061 | |
4062 if (code != NULL) | |
4063 *code = keycode; | |
4064 if (state != NULL) | |
4065 *state = keystate; | |
4066 } | |
4067 return retval; | |
4068 } | |
4069 | |
4070 static void | |
4071 im_xim_send_event_imactivate() | |
4072 { | |
4073 /* Force turn on preedit state by symulate keypress event. | |
4074 * Keycode and state is specified by 'imactivatekey'. | |
4075 */ | |
4076 XKeyEvent ev; | |
4077 | |
4078 gui_get_x11_windis(&ev.window, &ev.display); | |
4079 ev.root = RootWindow(ev.display, DefaultScreen(ev.display)); | |
4080 ev.subwindow = None; | |
4081 ev.time = CurrentTime; | |
4082 ev.x = 1; | |
4083 ev.y = 1; | |
4084 ev.x_root = 1; | |
4085 ev.y_root = 1; | |
4086 ev.same_screen = 1; | |
4087 ev.type = KeyPress; | |
4088 if (im_xim_str2keycode(&ev.keycode, &ev.state) == OK) | |
4089 XSendEvent(ev.display, ev.window, 1, KeyPressMask, (XEvent*)&ev); | |
4090 } | |
4091 | |
4092 /* | |
4093 * Return TRUE if 'imactivatekey' has a valid value. | |
4094 */ | |
4095 int | |
4096 im_xim_isvalid_imactivate() | |
4097 { | |
4098 return im_xim_str2keycode(NULL, NULL) == OK; | |
4099 } | |
4100 #endif /* FEAT_GUI_GTK */ | |
4101 | |
4102 /* | |
4103 * Switch using XIM on/off. This is used by the code that changes "State". | |
4104 */ | |
4105 void | |
4106 im_set_active(active) | |
4107 int active; | |
4108 { | |
4109 if (xic == NULL) | |
4110 return; | |
4111 | |
4112 /* If 'imdisable' is set, XIM is never active. */ | |
4113 if (p_imdisable) | |
4114 active = FALSE; | |
4115 #ifndef FEAT_GUI_GTK | |
4116 else if (input_style & XIMPreeditPosition) | |
4117 /* There is a problem in switching XIM off when preediting is used, | |
4118 * and it is not clear how this can be solved. For now, keep XIM on | |
4119 * all the time, like it was done in Vim 5.8. */ | |
4120 active = TRUE; | |
4121 #endif | |
4122 | |
4123 /* Remember the active state, it is needed when Vim gets keyboard focus. */ | |
4124 xim_is_active = active; | |
4125 | |
4126 #ifdef FEAT_GUI_GTK | |
4127 /* When 'imactivatekey' has valid key-string, try to control XIM preedit | |
4128 * state. When 'imactivatekey' has no or invalid string, try old XIM | |
4129 * focus control. | |
4130 */ | |
4131 if (*p_imak != NUL) | |
4132 { | |
4133 /* BASIC STRATEGY: | |
4134 * Destroy old Input Context (XIC), and create new one. New XIC | |
4135 * would have a state of preedit that is off. When argument:active | |
4136 * is false, that's all. Else argument:active is true, send a key | |
4137 * event specified by 'imactivatekey' to activate XIM preedit state. | |
4138 */ | |
4139 | |
4140 xim_is_active = TRUE; /* Disable old XIM focus control */ | |
4141 /* If we can monitor preedit state with preedit callback functions, | |
4142 * try least creation of new XIC. | |
4143 */ | |
4144 if (xim_input_style & (int)GDK_IM_PREEDIT_CALLBACKS) | |
4145 { | |
4146 if (xim_can_preediting && !active) | |
4147 { | |
4148 /* Force turn off preedit state. With some IM | |
4149 * implementations, we cannot turn off preedit state by | |
4150 * symulate keypress event. It is why using such a method | |
4151 * that destroy old IC (input context), and create new one. | |
4152 * When create new IC, its preedit state is usually off. | |
4153 */ | |
4154 xim_reset(); | |
4155 xim_set_focus(FALSE); | |
4156 gdk_ic_destroy(xic); | |
4157 xim_init(); | |
4158 xim_can_preediting = FALSE; | |
4159 } | |
4160 else if (!xim_can_preediting && active) | |
4161 im_xim_send_event_imactivate(); | |
4162 } | |
4163 else | |
4164 { | |
4165 /* First, force destroy old IC, and create new one. It | |
4166 * symulates "turning off preedit state". | |
4167 */ | |
4168 xim_set_focus(FALSE); | |
4169 gdk_ic_destroy(xic); | |
4170 xim_init(); | |
4171 xim_can_preediting = FALSE; | |
4172 | |
4173 /* 2nd, when requested to activate IM, symulate this by sending | |
4174 * the event. | |
4175 */ | |
4176 if (active) | |
4177 { | |
4178 im_xim_send_event_imactivate(); | |
4179 xim_can_preediting = TRUE; | |
4180 } | |
4181 } | |
4182 } | |
4183 else | |
4184 { | |
4185 # ifndef XIMPreeditUnKnown | |
4186 /* X11R5 doesn't have these, it looks safe enough to define here. */ | |
4187 typedef unsigned long XIMPreeditState; | |
4188 # define XIMPreeditUnKnown 0L | |
4189 # define XIMPreeditEnable 1L | |
4190 # define XIMPreeditDisable (1L<<1) | |
4191 # define XNPreeditState "preeditState" | |
4192 # endif | |
4193 XIMPreeditState preedit_state = XIMPreeditUnKnown; | |
4194 XVaNestedList preedit_attr; | |
4195 XIC pxic; | |
4196 | |
4197 preedit_attr = XVaCreateNestedList(0, | |
4198 XNPreeditState, &preedit_state, | |
4199 NULL); | |
4200 pxic = ((GdkICPrivate *)xic)->xic; | |
4201 | |
4202 if (!XGetICValues(pxic, XNPreeditAttributes, preedit_attr, NULL)) | |
4203 { | |
4204 XFree(preedit_attr); | |
4205 preedit_attr = XVaCreateNestedList(0, | |
4206 XNPreeditState, | |
4207 active ? XIMPreeditEnable : XIMPreeditDisable, | |
4208 NULL); | |
4209 XSetICValues(pxic, XNPreeditAttributes, preedit_attr, NULL); | |
4210 xim_can_preediting = active; | |
4211 xim_is_active = active; | |
4212 } | |
4213 XFree(preedit_attr); | |
4214 } | |
4215 if (xim_input_style & XIMPreeditCallbacks) | |
4216 { | |
4217 preedit_buf_len = 0; | |
4218 init_preedit_start_col(); | |
4219 } | |
4220 #else | |
4221 # if 0 | |
4222 /* When had tested kinput2 + canna + Athena GUI version with | |
4223 * 'imactivatekey' is "s-space", im_xim_send_event_imactivate() did not | |
4224 * work correctly. It just inserted one space. I don't know why we | |
4225 * couldn't switch state of XIM preediting. This is reason why these | |
4226 * codes are commented out. | |
4227 */ | |
4228 /* First, force destroy old IC, and create new one. It symulates | |
4229 * "turning off preedit state". | |
4230 */ | |
4231 xim_set_focus(FALSE); | |
4232 XDestroyIC(xic); | |
4233 xic = NULL; | |
4234 xim_init(); | |
4235 | |
4236 /* 2nd, when requested to activate IM, symulate this by sending the | |
4237 * event. | |
4238 */ | |
4239 if (active) | |
4240 im_xim_send_event_imactivate(); | |
4241 # endif | |
4242 #endif | |
4243 xim_set_preedit(); | |
4244 } | |
4245 | |
4246 /* | |
4247 * Adjust using XIM for gaining or losing keyboard focus. Also called when | |
4248 * "xim_is_active" changes. | |
4249 */ | |
4250 void | |
4251 xim_set_focus(focus) | |
4252 int focus; | |
4253 { | |
4254 if (xic == NULL) | |
4255 return; | |
4256 | |
4257 /* | |
4258 * XIM only gets focus when the Vim window has keyboard focus and XIM has | |
4259 * been set active for the current mode. | |
4260 */ | |
4261 if (focus && xim_is_active) | |
4262 { | |
4263 if (!xim_has_focus) | |
4264 { | |
4265 xim_has_focus = TRUE; | |
4266 #ifdef FEAT_GUI_GTK | |
4267 gdk_im_begin(xic, gui.drawarea->window); | |
4268 #else | |
4269 XSetICFocus(xic); | |
4270 #endif | |
4271 } | |
4272 } | |
4273 else | |
4274 { | |
4275 if (xim_has_focus) | |
4276 { | |
4277 xim_has_focus = FALSE; | |
4278 #ifdef FEAT_GUI_GTK | |
4279 gdk_im_end(); | |
4280 #else | |
4281 XUnsetICFocus(xic); | |
4282 #endif | |
4283 } | |
4284 } | |
4285 } | |
4286 | |
4287 /*ARGSUSED*/ | |
4288 void | |
4289 im_set_position(row, col) | |
4290 int row; | |
4291 int col; | |
4292 { | |
4293 xim_set_preedit(); | |
4294 } | |
4295 | |
4296 /* | |
4297 * Set the XIM to the current cursor position. | |
4298 */ | |
4299 void | |
4300 xim_set_preedit() | |
4301 { | |
4302 if (xic == NULL) | |
4303 return; | |
4304 | |
4305 xim_set_focus(TRUE); | |
4306 | |
4307 #ifdef FEAT_GUI_GTK | |
4308 if (gdk_im_ready()) | |
4309 { | |
4310 int attrmask; | |
4311 GdkICAttr *attr; | |
4312 | |
4313 if (!xic_attr) | |
4314 return; | |
4315 | |
4316 attr = xic_attr; | |
4317 attrmask = 0; | |
4318 | |
4319 # ifdef FEAT_XFONTSET | |
4320 if ((xim_input_style & (int)GDK_IM_PREEDIT_POSITION) | |
4321 && gui.fontset != NOFONTSET | |
4322 && gui.fontset->type == GDK_FONT_FONTSET) | |
4323 { | |
4324 if (!xim_has_focus) | |
4325 { | |
4326 if (attr->spot_location.y >= 0) | |
4327 { | |
4328 attr->spot_location.x = 0; | |
4329 attr->spot_location.y = -100; | |
4330 attrmask |= (int)GDK_IC_SPOT_LOCATION; | |
4331 } | |
4332 } | |
4333 else | |
4334 { | |
4335 gint width, height; | |
4336 | |
4337 if (attr->spot_location.x != TEXT_X(gui.col) | |
4338 || attr->spot_location.y != TEXT_Y(gui.row)) | |
4339 { | |
4340 attr->spot_location.x = TEXT_X(gui.col); | |
4341 attr->spot_location.y = TEXT_Y(gui.row); | |
4342 attrmask |= (int)GDK_IC_SPOT_LOCATION; | |
4343 } | |
4344 | |
4345 gdk_window_get_size(gui.drawarea->window, &width, &height); | |
4346 width -= 2 * gui.border_offset; | |
4347 height -= 2 * gui.border_offset; | |
4348 if (xim_input_style & (int)GDK_IM_STATUS_AREA) | |
4349 height -= gui.char_height; | |
4350 if (attr->preedit_area.width != width | |
4351 || attr->preedit_area.height != height) | |
4352 { | |
4353 attr->preedit_area.x = gui.border_offset; | |
4354 attr->preedit_area.y = gui.border_offset; | |
4355 attr->preedit_area.width = width; | |
4356 attr->preedit_area.height = height; | |
4357 attrmask |= (int)GDK_IC_PREEDIT_AREA; | |
4358 } | |
4359 | |
4360 if (attr->preedit_fontset != gui.current_font) | |
4361 { | |
4362 attr->preedit_fontset = gui.current_font; | |
4363 attrmask |= (int)GDK_IC_PREEDIT_FONTSET; | |
4364 } | |
4365 } | |
4366 } | |
4367 # endif /* FEAT_XFONTSET */ | |
4368 | |
4369 if (xim_fg_color == INVALCOLOR) | |
4370 { | |
4371 xim_fg_color = gui.def_norm_pixel; | |
4372 xim_bg_color = gui.def_back_pixel; | |
4373 } | |
4374 if (attr->preedit_foreground.pixel != xim_fg_color) | |
4375 { | |
4376 attr->preedit_foreground.pixel = xim_fg_color; | |
4377 attrmask |= (int)GDK_IC_PREEDIT_FOREGROUND; | |
4378 } | |
4379 if (attr->preedit_background.pixel != xim_bg_color) | |
4380 { | |
4381 attr->preedit_background.pixel = xim_bg_color; | |
4382 attrmask |= (int)GDK_IC_PREEDIT_BACKGROUND; | |
4383 } | |
4384 | |
4385 if (attrmask != 0) | |
4386 gdk_ic_set_attr(xic, attr, (GdkICAttributesType)attrmask); | |
4387 } | |
4388 #else /* FEAT_GUI_GTK */ | |
4389 { | |
4390 XVaNestedList attr_list; | |
4391 XRectangle spot_area; | |
4392 XPoint over_spot; | |
4393 int line_space; | |
4394 | |
4395 if (!xim_has_focus) | |
4396 { | |
4397 /* hide XIM cursor */ | |
4398 over_spot.x = 0; | |
4399 over_spot.y = -100; /* arbitrary invisible position */ | |
4400 attr_list = (XVaNestedList) XVaCreateNestedList(0, | |
4401 XNSpotLocation, | |
4402 &over_spot, | |
4403 NULL); | |
4404 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); | |
4405 XFree(attr_list); | |
4406 return; | |
4407 } | |
4408 | |
4409 if (input_style & XIMPreeditPosition) | |
4410 { | |
4411 if (xim_fg_color == INVALCOLOR) | |
4412 { | |
4413 xim_fg_color = gui.def_norm_pixel; | |
4414 xim_bg_color = gui.def_back_pixel; | |
4415 } | |
4416 over_spot.x = TEXT_X(gui.col); | |
4417 over_spot.y = TEXT_Y(gui.row); | |
4418 spot_area.x = 0; | |
4419 spot_area.y = 0; | |
4420 spot_area.height = gui.char_height * Rows; | |
4421 spot_area.width = gui.char_width * Columns; | |
4422 line_space = gui.char_height; | |
4423 attr_list = (XVaNestedList) XVaCreateNestedList(0, | |
4424 XNSpotLocation, &over_spot, | |
4425 XNForeground, (Pixel) xim_fg_color, | |
4426 XNBackground, (Pixel) xim_bg_color, | |
4427 XNArea, &spot_area, | |
4428 XNLineSpace, line_space, | |
4429 NULL); | |
4430 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) | |
4431 EMSG(_("E284: Cannot set IC values")); | |
4432 XFree(attr_list); | |
4433 } | |
4434 } | |
4435 #endif /* FEAT_GUI_GTK */ | |
4436 } | |
4437 | |
4438 /* | |
4439 * Set up the status area. | |
4440 * | |
4441 * This should use a separate Widget, but that seems not possible, because | |
4442 * preedit_area and status_area should be set to the same window as for the | |
4443 * text input. Unfortunately this means the status area pollutes the text | |
4444 * window... | |
4445 */ | |
4446 void | |
4447 xim_set_status_area() | |
4448 { | |
4449 if (xic == NULL) | |
4450 return; | |
4451 | |
4452 #ifdef FEAT_GUI_GTK | |
4453 # if defined(FEAT_XFONTSET) | |
4454 if (use_status_area) | |
4455 { | |
4456 GdkICAttr *attr; | |
4457 int style; | |
4458 gint width, height; | |
4459 GtkWidget *widget; | |
4460 int attrmask; | |
4461 | |
4462 if (!xic_attr) | |
4463 return; | |
4464 | |
4465 attr = xic_attr; | |
4466 attrmask = 0; | |
4467 style = (int)gdk_ic_get_style(xic); | |
4468 if ((style & (int)GDK_IM_STATUS_MASK) == (int)GDK_IM_STATUS_AREA) | |
4469 { | |
4470 if (gui.fontset != NOFONTSET | |
4471 && gui.fontset->type == GDK_FONT_FONTSET) | |
4472 { | |
4473 widget = gui.mainwin; | |
4474 gdk_window_get_size(widget->window, &width, &height); | |
4475 | |
4476 attrmask |= (int)GDK_IC_STATUS_AREA; | |
4477 attr->status_area.x = 0; | |
4478 attr->status_area.y = height - gui.char_height - 1; | |
4479 attr->status_area.width = width; | |
4480 attr->status_area.height = gui.char_height; | |
4481 } | |
4482 } | |
4483 if (attrmask != 0) | |
4484 gdk_ic_set_attr(xic, attr, (GdkICAttributesType)attrmask); | |
4485 } | |
4486 # endif | |
4487 #else | |
4488 { | |
4489 XVaNestedList preedit_list = 0, status_list = 0, list = 0; | |
4490 XRectangle pre_area, status_area; | |
4491 | |
4492 if (input_style & XIMStatusArea) | |
4493 { | |
4494 if (input_style & XIMPreeditArea) | |
4495 { | |
4496 XRectangle *needed_rect; | |
4497 | |
4498 /* to get status_area width */ | |
4499 status_list = XVaCreateNestedList(0, XNAreaNeeded, | |
4500 &needed_rect, NULL); | |
4501 XGetICValues(xic, XNStatusAttributes, status_list, NULL); | |
4502 XFree(status_list); | |
4503 | |
4504 status_area.width = needed_rect->width; | |
4505 } | |
4506 else | |
4507 status_area.width = gui.char_width * Columns; | |
4508 | |
4509 status_area.x = 0; | |
4510 status_area.y = gui.char_height * Rows + gui.border_offset; | |
4511 if (gui.which_scrollbars[SBAR_BOTTOM]) | |
4512 status_area.y += gui.scrollbar_height; | |
4513 #ifdef FEAT_MENU | |
4514 if (gui.menu_is_active) | |
4515 status_area.y += gui.menu_height; | |
4516 #endif | |
4517 status_area.height = gui.char_height; | |
4518 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); | |
4519 } | |
4520 else | |
4521 { | |
4522 status_area.x = 0; | |
4523 status_area.y = gui.char_height * Rows + gui.border_offset; | |
4524 if (gui.which_scrollbars[SBAR_BOTTOM]) | |
4525 status_area.y += gui.scrollbar_height; | |
4526 #ifdef FEAT_MENU | |
4527 if (gui.menu_is_active) | |
4528 status_area.y += gui.menu_height; | |
4529 #endif | |
4530 status_area.width = 0; | |
4531 status_area.height = gui.char_height; | |
4532 } | |
4533 | |
4534 if (input_style & XIMPreeditArea) /* off-the-spot */ | |
4535 { | |
4536 pre_area.x = status_area.x + status_area.width; | |
4537 pre_area.y = gui.char_height * Rows + gui.border_offset; | |
4538 pre_area.width = gui.char_width * Columns - pre_area.x; | |
4539 if (gui.which_scrollbars[SBAR_BOTTOM]) | |
4540 pre_area.y += gui.scrollbar_height; | |
4541 #ifdef FEAT_MENU | |
4542 if (gui.menu_is_active) | |
4543 pre_area.y += gui.menu_height; | |
4544 #endif | |
4545 pre_area.height = gui.char_height; | |
4546 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); | |
4547 } | |
4548 else if (input_style & XIMPreeditPosition) /* over-the-spot */ | |
4549 { | |
4550 pre_area.x = 0; | |
4551 pre_area.y = 0; | |
4552 pre_area.height = gui.char_height * Rows; | |
4553 pre_area.width = gui.char_width * Columns; | |
4554 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); | |
4555 } | |
4556 | |
4557 if (preedit_list && status_list) | |
4558 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, | |
4559 XNStatusAttributes, status_list, NULL); | |
4560 else if (preedit_list) | |
4561 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, | |
4562 NULL); | |
4563 else if (status_list) | |
4564 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, | |
4565 NULL); | |
4566 else | |
4567 list = NULL; | |
4568 | |
4569 if (list) | |
4570 { | |
4571 XSetICValues(xic, XNVaNestedList, list, NULL); | |
4572 XFree(list); | |
4573 } | |
4574 if (status_list) | |
4575 XFree(status_list); | |
4576 if (preedit_list) | |
4577 XFree(preedit_list); | |
4578 } | |
4579 #endif | |
4580 } | |
4581 | |
4582 #if defined(FEAT_GUI_X11) || defined(FEAT_GUI_GTK) | |
4583 static char e_xim[] = N_("E285: Failed to create input context"); | |
4584 #endif | |
4585 | |
4586 #if defined(FEAT_GUI_X11) || defined(PROTO) | |
4587 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(sun) | |
4588 # define USE_X11R6_XIM | |
4589 # endif | |
4590 | |
4591 static int xim_real_init __ARGS((Window x11_window, Display *x11_display)); | |
4592 | |
4593 | |
4594 #ifdef USE_X11R6_XIM | |
4595 static void xim_instantiate_cb __ARGS((Display *display, XPointer client_data, XPointer call_data)); | |
4596 static void xim_destroy_cb __ARGS((XIM im, XPointer client_data, XPointer call_data)); | |
4597 | |
4598 /*ARGSUSED*/ | |
4599 static void | |
4600 xim_instantiate_cb(display, client_data, call_data) | |
4601 Display *display; | |
4602 XPointer client_data; | |
4603 XPointer call_data; | |
4604 { | |
4605 Window x11_window; | |
4606 Display *x11_display; | |
4607 | |
4608 #ifdef XIM_DEBUG | |
4609 xim_log("xim_instantiate_cb()\n"); | |
4610 #endif | |
4611 | |
4612 gui_get_x11_windis(&x11_window, &x11_display); | |
4613 if (display != x11_display) | |
4614 return; | |
4615 | |
4616 xim_real_init(x11_window, x11_display); | |
4617 gui_set_shellsize(FALSE, FALSE); | |
4618 if (xic != NULL) | |
4619 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, | |
4620 xim_instantiate_cb, NULL); | |
4621 } | |
4622 | |
4623 /*ARGSUSED*/ | |
4624 static void | |
4625 xim_destroy_cb(im, client_data, call_data) | |
4626 XIM im; | |
4627 XPointer client_data; | |
4628 XPointer call_data; | |
4629 { | |
4630 Window x11_window; | |
4631 Display *x11_display; | |
4632 | |
4633 #ifdef XIM_DEBUG | |
4634 xim_log("xim_destroy_cb()\n"); | |
4635 #endif | |
4636 gui_get_x11_windis(&x11_window, &x11_display); | |
4637 | |
4638 xic = NULL; | |
4639 status_area_enabled = FALSE; | |
4640 | |
4641 gui_set_shellsize(FALSE, FALSE); | |
4642 | |
4643 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, | |
4644 xim_instantiate_cb, NULL); | |
4645 } | |
4646 #endif | |
4647 | |
4648 void | |
4649 xim_init() | |
4650 { | |
4651 Window x11_window; | |
4652 Display *x11_display; | |
4653 | |
4654 #ifdef XIM_DEBUG | |
4655 xim_log("xim_init()\n"); | |
4656 #endif | |
4657 | |
4658 gui_get_x11_windis(&x11_window, &x11_display); | |
4659 | |
4660 xic = NULL; | |
4661 | |
4662 if (xim_real_init(x11_window, x11_display)) | |
4663 return; | |
4664 | |
4665 gui_set_shellsize(FALSE, FALSE); | |
4666 | |
4667 #ifdef USE_X11R6_XIM | |
4668 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, | |
4669 xim_instantiate_cb, NULL); | |
4670 #endif | |
4671 } | |
4672 | |
4673 static int | |
4674 xim_real_init(x11_window, x11_display) | |
4675 Window x11_window; | |
4676 Display *x11_display; | |
4677 { | |
4678 int i; | |
4679 char *p, | |
4680 *s, | |
4681 *ns, | |
4682 *end, | |
4683 tmp[1024]; | |
4684 #define IMLEN_MAX 40 | |
4685 char buf[IMLEN_MAX + 7]; | |
4686 XIM xim = NULL; | |
4687 XIMStyles *xim_styles; | |
4688 XIMStyle this_input_style = 0; | |
4689 Boolean found; | |
4690 XPoint over_spot; | |
4691 XVaNestedList preedit_list, status_list; | |
4692 | |
4693 input_style = 0; | |
4694 status_area_enabled = FALSE; | |
4695 | |
4696 if (xic != NULL) | |
4697 return FALSE; | |
4698 | |
4699 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) | |
4700 { | |
4701 strcpy(tmp, gui.rsrc_input_method); | |
4702 for (ns = s = tmp; ns != NULL && *s != NUL;) | |
4703 { | |
4704 s = (char *)skipwhite((char_u *)s); | |
4705 if (*s == NUL) | |
4706 break; | |
4707 if ((ns = end = strchr(s, ',')) == NULL) | |
4708 end = s + strlen(s); | |
4709 while (isspace(((char_u *)end)[-1])) | |
4710 end--; | |
4711 *end = NUL; | |
4712 | |
4713 if (strlen(s) <= IMLEN_MAX) | |
4714 { | |
4715 strcpy(buf, "@im="); | |
4716 strcat(buf, s); | |
4717 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL | |
4718 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) | |
4719 != NULL) | |
4720 break; | |
4721 } | |
4722 | |
4723 s = ns + 1; | |
4724 } | |
4725 } | |
4726 | |
4727 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) | |
4728 xim = XOpenIM(x11_display, NULL, NULL, NULL); | |
4729 | |
4730 /* This is supposed to be useful to obtain characters through | |
4731 * XmbLookupString() without really using a XIM. */ | |
4732 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL | |
4733 && *p != NUL) | |
4734 xim = XOpenIM(x11_display, NULL, NULL, NULL); | |
4735 | |
4736 if (xim == NULL) | |
4737 { | |
4738 /* Only give this message when verbose is set, because too many people | |
4739 * got this message when they didn't want to use a XIM. */ | |
4740 if (p_verbose > 0) | |
4741 EMSG(_("E286: Failed to open input method")); | |
4742 return FALSE; | |
4743 } | |
4744 | |
4745 #ifdef USE_X11R6_XIM | |
4746 { | |
4747 XIMCallback destroy_cb; | |
4748 | |
4749 destroy_cb.callback = xim_destroy_cb; | |
4750 destroy_cb.client_data = NULL; | |
4751 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) | |
4752 EMSG(_("E287: Warning: Could not set destroy callback to IM")); | |
4753 } | |
4754 #endif | |
4755 | |
4756 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) | |
4757 { | |
4758 EMSG(_("E288: input method doesn't support any style")); | |
4759 XCloseIM(xim); | |
4760 return FALSE; | |
4761 } | |
4762 | |
4763 found = False; | |
4764 strcpy(tmp, gui.rsrc_preedit_type_name); | |
4765 for (s = tmp; s && !found; ) | |
4766 { | |
4767 while (*s && isspace((unsigned char)*s)) | |
4768 s++; | |
4769 if (!*s) | |
4770 break; | |
4771 if ((ns = end = strchr(s, ',')) != 0) | |
4772 ns++; | |
4773 else | |
4774 end = s + strlen(s); | |
4775 while (isspace((unsigned char)*end)) | |
4776 end--; | |
4777 *end = '\0'; | |
4778 | |
4779 if (!strcmp(s, "OverTheSpot")) | |
4780 this_input_style = (XIMPreeditPosition | XIMStatusArea); | |
4781 else if (!strcmp(s, "OffTheSpot")) | |
4782 this_input_style = (XIMPreeditArea | XIMStatusArea); | |
4783 else if (!strcmp(s, "Root")) | |
4784 this_input_style = (XIMPreeditNothing | XIMStatusNothing); | |
4785 | |
4786 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) | |
4787 { | |
4788 if (this_input_style == xim_styles->supported_styles[i]) | |
4789 { | |
4790 found = True; | |
4791 break; | |
4792 } | |
4793 } | |
4794 if (!found) | |
4795 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) | |
4796 { | |
4797 if ((xim_styles->supported_styles[i] & this_input_style) | |
4798 == (this_input_style & ~XIMStatusArea)) | |
4799 { | |
4800 this_input_style &= ~XIMStatusArea; | |
4801 found = True; | |
4802 break; | |
4803 } | |
4804 } | |
4805 | |
4806 s = ns; | |
4807 } | |
4808 XFree(xim_styles); | |
4809 | |
4810 if (!found) | |
4811 { | |
4812 /* Only give this message when verbose is set, because too many people | |
4813 * got this message when they didn't want to use a XIM. */ | |
4814 if (p_verbose > 0) | |
4815 EMSG(_("E289: input method doesn't support my preedit type")); | |
4816 XCloseIM(xim); | |
4817 return FALSE; | |
4818 } | |
4819 | |
4820 over_spot.x = TEXT_X(gui.col); | |
4821 over_spot.y = TEXT_Y(gui.row); | |
4822 input_style = this_input_style; | |
4823 | |
4824 /* A crash was reported when trying to pass gui.norm_font as XNFontSet, | |
4825 * thus that has been removed. Hopefully the default works... */ | |
4826 #ifdef FEAT_XFONTSET | |
4827 if (gui.fontset != NOFONTSET) | |
4828 { | |
4829 preedit_list = XVaCreateNestedList(0, | |
4830 XNSpotLocation, &over_spot, | |
4831 XNForeground, (Pixel)gui.def_norm_pixel, | |
4832 XNBackground, (Pixel)gui.def_back_pixel, | |
4833 XNFontSet, (XFontSet)gui.fontset, | |
4834 NULL); | |
4835 status_list = XVaCreateNestedList(0, | |
4836 XNForeground, (Pixel)gui.def_norm_pixel, | |
4837 XNBackground, (Pixel)gui.def_back_pixel, | |
4838 XNFontSet, (XFontSet)gui.fontset, | |
4839 NULL); | |
4840 } | |
4841 else | |
4842 #endif | |
4843 { | |
4844 preedit_list = XVaCreateNestedList(0, | |
4845 XNSpotLocation, &over_spot, | |
4846 XNForeground, (Pixel)gui.def_norm_pixel, | |
4847 XNBackground, (Pixel)gui.def_back_pixel, | |
4848 NULL); | |
4849 status_list = XVaCreateNestedList(0, | |
4850 XNForeground, (Pixel)gui.def_norm_pixel, | |
4851 XNBackground, (Pixel)gui.def_back_pixel, | |
4852 NULL); | |
4853 } | |
4854 | |
4855 xic = XCreateIC(xim, | |
4856 XNInputStyle, input_style, | |
4857 XNClientWindow, x11_window, | |
4858 XNFocusWindow, gui.wid, | |
4859 XNPreeditAttributes, preedit_list, | |
4860 XNStatusAttributes, status_list, | |
4861 NULL); | |
4862 XFree(status_list); | |
4863 XFree(preedit_list); | |
4864 if (xic != NULL) | |
4865 { | |
4866 if (input_style & XIMStatusArea) | |
4867 { | |
4868 xim_set_status_area(); | |
4869 status_area_enabled = TRUE; | |
4870 } | |
4871 else | |
4872 gui_set_shellsize(FALSE, FALSE); | |
4873 } | |
4874 else | |
4875 { | |
4876 EMSG(_(e_xim)); | |
4877 XCloseIM(xim); | |
4878 return FALSE; | |
4879 } | |
4880 | |
4881 return TRUE; | |
4882 } | |
4883 | |
4884 #endif /* FEAT_GUI_X11 */ | |
4885 | |
4886 #if defined(FEAT_GUI_GTK) || defined(PROTO) | |
4887 | |
4888 # ifdef FEAT_XFONTSET | |
4889 static char e_overthespot[] = N_("E290: over-the-spot style requires fontset"); | |
4890 # endif | |
4891 | |
4892 # ifdef PROTO | |
4893 typedef int GdkIC; | |
4894 # endif | |
4895 | |
4896 void | |
4897 xim_decide_input_style() | |
4898 { | |
4899 /* GDK_IM_STATUS_CALLBACKS was disabled, enabled it to allow Japanese | |
4900 * OverTheSpot. */ | |
4901 int supported_style = (int)GDK_IM_PREEDIT_NONE | | |
4902 (int)GDK_IM_PREEDIT_NOTHING | | |
4903 (int)GDK_IM_PREEDIT_POSITION | | |
4904 (int)GDK_IM_PREEDIT_CALLBACKS | | |
4905 (int)GDK_IM_STATUS_CALLBACKS | | |
4906 (int)GDK_IM_STATUS_AREA | | |
4907 (int)GDK_IM_STATUS_NONE | | |
4908 (int)GDK_IM_STATUS_NOTHING; | |
4909 | |
4910 #ifdef XIM_DEBUG | |
4911 xim_log("xim_decide_input_style()\n"); | |
4912 #endif | |
4913 | |
4914 if (!gdk_im_ready()) | |
4915 xim_input_style = 0; | |
4916 else | |
4917 { | |
4918 if (gtk_major_version > 1 | |
4919 || (gtk_major_version == 1 | |
4920 && (gtk_minor_version > 2 | |
4921 || (gtk_minor_version == 2 && gtk_micro_version >= 3)))) | |
4922 use_status_area = TRUE; | |
4923 else | |
4924 { | |
4925 EMSG(_("E291: Your GTK+ is older than 1.2.3. Status area disabled")); | |
4926 use_status_area = FALSE; | |
4927 } | |
4928 #ifdef FEAT_XFONTSET | |
4929 if (gui.fontset == NOFONTSET || gui.fontset->type != GDK_FONT_FONTSET) | |
4930 #endif | |
4931 supported_style &= ~((int)GDK_IM_PREEDIT_POSITION | |
4932 | (int)GDK_IM_STATUS_AREA); | |
4933 if (!use_status_area) | |
4934 supported_style &= ~(int)GDK_IM_STATUS_AREA; | |
4935 xim_input_style = (int)gdk_im_decide_style((GdkIMStyle)supported_style); | |
4936 } | |
4937 } | |
4938 | |
4939 /*ARGSUSED*/ | |
4940 static void | |
4941 preedit_start_cbproc(XIC xic, XPointer client_data, XPointer call_data) | |
4942 { | |
4943 #ifdef XIM_DEBUG | |
4944 xim_log("xim_decide_input_style()\n"); | |
4945 #endif | |
4946 | |
4947 draw_feedback = NULL; | |
4948 xim_can_preediting = TRUE; | |
4949 xim_has_preediting = TRUE; | |
4950 gui_update_cursor(TRUE, FALSE); | |
4951 if (showmode() > 0) | |
4952 { | |
4953 setcursor(); | |
4954 out_flush(); | |
4955 } | |
4956 } | |
4957 | |
4958 static void | |
4959 xim_back_delete(int n) | |
4960 { | |
4961 char_u str[3]; | |
4962 | |
4963 str[0] = CSI; | |
4964 str[1] = 'k'; | |
4965 str[2] = 'b'; | |
4966 while (n-- > 0) | |
4967 add_to_input_buf(str, 3); | |
4968 } | |
4969 | |
4970 static GSList *key_press_event_queue = NULL; | |
4971 static gboolean processing_queued_event = FALSE; | |
4972 | |
4973 /*ARGSUSED*/ | |
4974 static void | |
4975 preedit_draw_cbproc(XIC xic, XPointer client_data, XPointer call_data) | |
4976 { | |
4977 XIMPreeditDrawCallbackStruct *draw_data; | |
4978 XIMText *text; | |
4979 char *src; | |
4980 GSList *event_queue; | |
4981 | |
4982 #ifdef XIM_DEBUG | |
4983 xim_log("preedit_draw_cbproc()\n"); | |
4984 #endif | |
4985 | |
4986 draw_data = (XIMPreeditDrawCallbackStruct *) call_data; | |
4987 text = (XIMText *) draw_data->text; | |
4988 | |
4989 if ((text == NULL && draw_data->chg_length == preedit_buf_len) | |
4990 || preedit_buf_len == 0) | |
4991 { | |
4992 init_preedit_start_col(); | |
4993 vim_free(draw_feedback); | |
4994 draw_feedback = NULL; | |
4995 } | |
4996 if (draw_data->chg_length > 0) | |
4997 { | |
4998 int bs_cnt; | |
4999 | |
5000 if (draw_data->chg_length > preedit_buf_len) | |
5001 bs_cnt = preedit_buf_len; | |
5002 else | |
5003 bs_cnt = draw_data->chg_length; | |
5004 xim_back_delete(bs_cnt); | |
5005 preedit_buf_len -= bs_cnt; | |
5006 } | |
5007 if (text != NULL) | |
5008 { | |
5009 int len; | |
5010 #ifdef FEAT_MBYTE | |
5011 char_u *buf = NULL; | |
5012 unsigned int nfeedback = 0; | |
5013 #endif | |
5014 char_u *ptr; | |
5015 | |
5016 src = text->string.multi_byte; | |
5017 if (src != NULL && !text->encoding_is_wchar) | |
5018 { | |
5019 len = strlen(src); | |
5020 ptr = (char_u *)src; | |
5021 /* Avoid the enter for decision */ | |
5022 if (*ptr == '\n') | |
5023 return; | |
5024 | |
5025 #ifdef FEAT_MBYTE | |
5026 if (input_conv.vc_type != CONV_NONE | |
5027 && (buf = string_convert(&input_conv, | |
5028 (char_u *)src, &len)) != NULL) | |
5029 { | |
5030 /* Converted from 'termencoding' to 'encoding'. */ | |
5031 add_to_input_buf_csi(buf, len); | |
5032 ptr = buf; | |
5033 } | |
5034 else | |
5035 #endif | |
5036 add_to_input_buf_csi((char_u *)src, len); | |
5037 /* Add count of character to preedit_buf_len */ | |
5038 while (*ptr != NUL) | |
5039 { | |
5040 #ifdef FEAT_MBYTE | |
5041 if (draw_data->text->feedback != NULL) | |
5042 { | |
5043 if (draw_feedback == NULL) | |
5044 draw_feedback = (char *)alloc(draw_data->chg_first | |
5045 + text->length); | |
5046 else | |
5047 draw_feedback = realloc(draw_feedback, | |
5048 draw_data->chg_first + text->length); | |
5049 if (draw_feedback != NULL) | |
5050 { | |
5051 draw_feedback[nfeedback + draw_data->chg_first] | |
5052 = draw_data->text->feedback[nfeedback]; | |
5053 nfeedback++; | |
5054 } | |
5055 } | |
5056 if (has_mbyte) | |
5057 ptr += mb_ptr2len_check(ptr); | |
5058 else | |
5059 #endif | |
5060 ptr++; | |
5061 preedit_buf_len++; | |
5062 } | |
5063 #ifdef FEAT_MBYTE | |
5064 vim_free(buf); | |
5065 #endif | |
5066 preedit_end_col = MAXCOL; | |
5067 } | |
5068 } | |
5069 if (text != NULL || draw_data->chg_length > 0) | |
5070 { | |
5071 event_queue = key_press_event_queue; | |
5072 processing_queued_event = TRUE; | |
5073 while (event_queue != NULL && processing_queued_event) | |
5074 { | |
5075 GdkEvent *ev = event_queue->data; | |
5076 | |
5077 gboolean *ret; | |
5078 gtk_signal_emit_by_name((GtkObject*)gui.mainwin, "key_press_event", | |
5079 ev, &ret); | |
5080 gdk_event_free(ev); | |
5081 event_queue = event_queue->next; | |
5082 } | |
5083 processing_queued_event = FALSE; | |
5084 if (key_press_event_queue) | |
5085 { | |
5086 g_slist_free(key_press_event_queue); | |
5087 key_press_event_queue = NULL; | |
5088 } | |
5089 } | |
5090 if (gtk_main_level() > 0) | |
5091 gtk_main_quit(); | |
5092 } | |
5093 | |
5094 /* | |
5095 * Retrieve the highlighting attributes at column col in the preedit string. | |
5096 * Return -1 if not in preediting mode or if col is out of range. | |
5097 */ | |
5098 int | |
5099 im_get_feedback_attr(int col) | |
5100 { | |
5101 if (draw_feedback != NULL && col < preedit_buf_len) | |
5102 { | |
5103 if (draw_feedback[col] & XIMReverse) | |
5104 return HL_INVERSE; | |
5105 else if (draw_feedback[col] & XIMUnderline) | |
5106 return HL_UNDERLINE; | |
5107 else | |
5108 return hl_attr(HLF_V); | |
5109 } | |
5110 | |
5111 return -1; | |
5112 } | |
5113 | |
5114 /*ARGSUSED*/ | |
5115 static void | |
5116 preedit_caret_cbproc(XIC xic, XPointer client_data, XPointer call_data) | |
5117 { | |
5118 #ifdef XIM_DEBUG | |
5119 xim_log("preedit_caret_cbproc()\n"); | |
5120 #endif | |
5121 } | |
5122 | |
5123 /*ARGSUSED*/ | |
5124 static void | |
5125 preedit_done_cbproc(XIC xic, XPointer client_data, XPointer call_data) | |
5126 { | |
5127 #ifdef XIM_DEBUG | |
5128 xim_log("preedit_done_cbproc()\n"); | |
5129 #endif | |
5130 | |
5131 vim_free(draw_feedback); | |
5132 draw_feedback = NULL; | |
5133 xim_can_preediting = FALSE; | |
5134 xim_has_preediting = FALSE; | |
5135 gui_update_cursor(TRUE, FALSE); | |
5136 if (showmode() > 0) | |
5137 { | |
5138 setcursor(); | |
5139 out_flush(); | |
5140 } | |
5141 } | |
5142 | |
5143 void | |
5144 xim_reset(void) | |
5145 { | |
5146 char *text; | |
5147 | |
5148 #ifdef XIM_DEBUG | |
5149 xim_log("xim_reset()\n"); | |
5150 #endif | |
5151 | |
5152 if (xic != NULL) | |
5153 { | |
5154 text = XmbResetIC(((GdkICPrivate *)xic)->xic); | |
5155 if (text != NULL && !(xim_input_style & (int)GDK_IM_PREEDIT_CALLBACKS)) | |
5156 add_to_input_buf_csi((char_u *)text, strlen(text)); | |
5157 else | |
5158 preedit_buf_len = 0; | |
5159 if (text != NULL) | |
5160 XFree(text); | |
5161 } | |
5162 } | |
5163 | |
5164 /*ARGSUSED*/ | |
5165 int | |
5166 xim_queue_key_press_event(GdkEventKey *event, int down) | |
5167 { | |
5168 #ifdef XIM_DEBUG | |
5169 xim_log("xim_queue_key_press_event()\n"); | |
5170 #endif | |
5171 | |
5172 if (preedit_buf_len <= 0) | |
5173 return FALSE; | |
5174 if (processing_queued_event) | |
5175 processing_queued_event = FALSE; | |
5176 | |
5177 key_press_event_queue = g_slist_append(key_press_event_queue, | |
5178 gdk_event_copy((GdkEvent *)event)); | |
5179 return TRUE; | |
5180 } | |
5181 | |
5182 /*ARGSUSED*/ | |
5183 static void | |
5184 preedit_callback_setup(GdkIC *ic) | |
5185 { | |
5186 XIC xxic; | |
5187 XVaNestedList preedit_attr; | |
5188 XIMCallback preedit_start_cb; | |
5189 XIMCallback preedit_draw_cb; | |
5190 XIMCallback preedit_caret_cb; | |
5191 XIMCallback preedit_done_cb; | |
5192 | |
5193 xxic = ((GdkICPrivate*)xic)->xic; | |
5194 preedit_start_cb.callback = (XIMProc)preedit_start_cbproc; | |
5195 preedit_draw_cb.callback = (XIMProc)preedit_draw_cbproc; | |
5196 preedit_caret_cb.callback = (XIMProc)preedit_caret_cbproc; | |
5197 preedit_done_cb.callback = (XIMProc)preedit_done_cbproc; | |
5198 preedit_attr | |
5199 = XVaCreateNestedList (0, | |
5200 XNPreeditStartCallback, &preedit_start_cb, | |
5201 XNPreeditDrawCallback, &preedit_draw_cb, | |
5202 XNPreeditCaretCallback, &preedit_caret_cb, | |
5203 XNPreeditDoneCallback, &preedit_done_cb, | |
5204 0); | |
5205 XSetICValues (xxic, XNPreeditAttributes, preedit_attr, 0); | |
5206 XFree(preedit_attr); | |
5207 } | |
5208 | |
5209 /*ARGSUSED*/ | |
5210 static void | |
5211 reset_state_setup(GdkIC *ic) | |
5212 { | |
5213 #ifdef USE_X11R6_XIM | |
5214 /* don't change the input context when we call reset */ | |
5215 XSetICValues(((GdkICPrivate*)ic)->xic, XNResetState, XIMPreserveState, 0); | |
5216 #endif | |
5217 } | |
5218 | |
5219 void | |
5220 xim_init(void) | |
5221 { | |
5222 #ifdef XIM_DEBUG | |
5223 xim_log("xim_init()\n"); | |
5224 #endif | |
5225 | |
5226 xic = NULL; | |
5227 xic_attr = NULL; | |
5228 | |
5229 if (!gdk_im_ready()) | |
5230 { | |
5231 if (p_verbose > 0) | |
5232 EMSG(_("E292: Input Method Server is not running")); | |
5233 return; | |
5234 } | |
5235 if ((xic_attr = gdk_ic_attr_new()) != NULL) | |
5236 { | |
5237 #ifdef FEAT_XFONTSET | |
5238 gint width, height; | |
5239 #endif | |
5240 int mask; | |
5241 GdkColormap *colormap; | |
5242 GdkICAttr *attr = xic_attr; | |
5243 int attrmask = (int)GDK_IC_ALL_REQ; | |
5244 GtkWidget *widget = gui.drawarea; | |
5245 | |
5246 attr->style = (GdkIMStyle)xim_input_style; | |
5247 attr->client_window = gui.mainwin->window; | |
5248 | |
5249 if ((colormap = gtk_widget_get_colormap(widget)) != | |
5250 gtk_widget_get_default_colormap()) | |
5251 { | |
5252 attrmask |= (int)GDK_IC_PREEDIT_COLORMAP; | |
5253 attr->preedit_colormap = colormap; | |
5254 } | |
5255 attrmask |= (int)GDK_IC_PREEDIT_FOREGROUND; | |
5256 attrmask |= (int)GDK_IC_PREEDIT_BACKGROUND; | |
5257 attr->preedit_foreground = widget->style->fg[GTK_STATE_NORMAL]; | |
5258 attr->preedit_background = widget->style->base[GTK_STATE_NORMAL]; | |
5259 | |
5260 #ifdef FEAT_XFONTSET | |
5261 if ((xim_input_style & (int)GDK_IM_PREEDIT_MASK) | |
5262 == (int)GDK_IM_PREEDIT_POSITION) | |
5263 { | |
5264 if (gui.fontset == NOFONTSET | |
5265 || gui.fontset->type != GDK_FONT_FONTSET) | |
5266 { | |
5267 EMSG(_(e_overthespot)); | |
5268 } | |
5269 else | |
5270 { | |
5271 gdk_window_get_size(widget->window, &width, &height); | |
5272 | |
5273 attrmask |= (int)GDK_IC_PREEDIT_POSITION_REQ; | |
5274 attr->spot_location.x = TEXT_X(0); | |
5275 attr->spot_location.y = TEXT_Y(0); | |
5276 attr->preedit_area.x = gui.border_offset; | |
5277 attr->preedit_area.y = gui.border_offset; | |
5278 attr->preedit_area.width = width - 2*gui.border_offset; | |
5279 attr->preedit_area.height = height - 2*gui.border_offset; | |
5280 attr->preedit_fontset = gui.fontset; | |
5281 } | |
5282 } | |
5283 | |
5284 if ((xim_input_style & (int)GDK_IM_STATUS_MASK) | |
5285 == (int)GDK_IM_STATUS_AREA) | |
5286 { | |
5287 if (gui.fontset == NOFONTSET | |
5288 || gui.fontset->type != GDK_FONT_FONTSET) | |
5289 { | |
5290 EMSG(_(e_overthespot)); | |
5291 } | |
5292 else | |
5293 { | |
5294 gdk_window_get_size(gui.mainwin->window, &width, &height); | |
5295 attrmask |= (int)GDK_IC_STATUS_AREA_REQ; | |
5296 attr->status_area.x = 0; | |
5297 attr->status_area.y = height - gui.char_height - 1; | |
5298 attr->status_area.width = width; | |
5299 attr->status_area.height = gui.char_height; | |
5300 attr->status_fontset = gui.fontset; | |
5301 } | |
5302 } | |
5303 else if ((xim_input_style & (int)GDK_IM_STATUS_MASK) | |
5304 == (int)GDK_IM_STATUS_CALLBACKS) | |
5305 { | |
5306 /* FIXME */ | |
5307 } | |
5308 #endif | |
5309 | |
5310 xic = gdk_ic_new(attr, (GdkICAttributesType)attrmask); | |
5311 | |
5312 if (xic == NULL) | |
5313 EMSG(_(e_xim)); | |
5314 else | |
5315 { | |
5316 mask = (int)gdk_window_get_events(widget->window); | |
5317 mask |= (int)gdk_ic_get_events(xic); | |
5318 gdk_window_set_events(widget->window, (GdkEventMask)mask); | |
5319 if (xim_input_style & (int)GDK_IM_PREEDIT_CALLBACKS) | |
5320 preedit_callback_setup(xic); | |
5321 reset_state_setup(xic); | |
5322 } | |
5323 } | |
5324 } | |
5325 | |
5326 void | |
5327 im_shutdown(void) | |
5328 { | |
5329 #ifdef XIM_DEBUG | |
5330 xim_log("im_shutdown()\n"); | |
5331 #endif | |
5332 | |
5333 if (xic != NULL) | |
5334 { | |
5335 gdk_im_end(); | |
5336 gdk_ic_destroy(xic); | |
5337 xic = NULL; | |
5338 } | |
5339 xim_is_active = FALSE; | |
5340 xim_can_preediting = FALSE; | |
5341 preedit_start_col = MAXCOL; | |
5342 xim_has_preediting = FALSE; | |
5343 } | |
5344 | |
5345 #endif /* FEAT_GUI_GTK */ | |
5346 | |
5347 int | |
5348 xim_get_status_area_height() | |
5349 { | |
5350 #ifdef FEAT_GUI_GTK | |
5351 if (xim_input_style & (int)GDK_IM_STATUS_AREA) | |
5352 return gui.char_height; | |
5353 #else | |
5354 if (status_area_enabled) | |
5355 return gui.char_height; | |
5356 #endif | |
5357 return 0; | |
5358 } | |
5359 | |
5360 /* | |
5361 * Get IM status. When IM is on, return TRUE. Else return FALSE. | |
5362 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is | |
5363 * active, when not having focus XIM may still be active (e.g., when using a | |
5364 * tear-off menu item). | |
5365 */ | |
5366 int | |
5367 im_get_status() | |
5368 { | |
5369 # ifdef FEAT_GUI_GTK | |
5370 if (xim_input_style & (int)GDK_IM_PREEDIT_CALLBACKS) | |
5371 return xim_can_preediting; | |
5372 # endif | |
5373 return xim_has_focus; | |
5374 } | |
5375 | |
5376 # endif /* !HAVE_GTK2 */ | |
5377 | |
5378 # if defined(FEAT_GUI_GTK) || defined(PROTO) | |
5379 int | |
5380 im_is_preediting() | |
5381 { | |
5382 return xim_has_preediting; | |
5383 } | |
5384 # endif | |
5385 #endif /* FEAT_XIM */ | |
5386 | |
5387 #if defined(FEAT_MBYTE) || defined(PROTO) | |
5388 | |
5389 /* | |
5390 * Setup "vcp" for conversion from "from" to "to". | |
5391 * The names must have been made canonical with enc_canonize(). | |
5392 * vcp->vc_type must have been initialized to CONV_NONE. | |
5393 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 | |
5394 * instead). | |
5395 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. | |
5396 * Return FAIL when conversion is not supported, OK otherwise. | |
5397 */ | |
5398 int | |
5399 convert_setup(vcp, from, to) | |
5400 vimconv_T *vcp; | |
5401 char_u *from; | |
5402 char_u *to; | |
5403 { | |
5404 int from_prop; | |
5405 int to_prop; | |
5406 | |
5407 /* Reset to no conversion. */ | |
5408 # ifdef USE_ICONV | |
5409 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) | |
5410 iconv_close(vcp->vc_fd); | |
5411 # endif | |
5412 vcp->vc_type = CONV_NONE; | |
5413 vcp->vc_factor = 1; | |
5414 vcp->vc_fail = FALSE; | |
5415 | |
5416 /* No conversion when one of the names is empty or they are equal. */ | |
5417 if (from == NULL || *from == NUL || to == NULL || *to == NUL | |
5418 || STRCMP(from, to) == 0) | |
5419 return OK; | |
5420 | |
5421 from_prop = enc_canon_props(from); | |
5422 to_prop = enc_canon_props(to); | |
5423 if ((from_prop & ENC_LATIN1) && (to_prop & ENC_UNICODE)) | |
5424 { | |
5425 /* Internal latin1 -> utf-8 conversion. */ | |
5426 vcp->vc_type = CONV_TO_UTF8; | |
5427 vcp->vc_factor = 2; /* up to twice as long */ | |
5428 } | |
5429 else if ((from_prop & ENC_UNICODE) && (to_prop & ENC_LATIN1)) | |
5430 { | |
5431 /* Internal utf-8 -> latin1 conversion. */ | |
5432 vcp->vc_type = CONV_TO_LATIN1; | |
5433 } | |
5434 #ifdef WIN3264 | |
5435 /* Win32-specific codepage <-> codepage conversion without iconv. */ | |
5436 else if (((from_prop & ENC_UNICODE) || encname2codepage(from) > 0) | |
5437 && ((to_prop & ENC_UNICODE) || encname2codepage(to) > 0)) | |
5438 { | |
5439 vcp->vc_type = CONV_CODEPAGE; | |
5440 vcp->vc_factor = 2; /* up to twice as long */ | |
5441 vcp->vc_cpfrom = (from_prop & ENC_UNICODE) ? 0 : encname2codepage(from); | |
5442 vcp->vc_cpto = (to_prop & ENC_UNICODE) ? 0 : encname2codepage(to); | |
5443 } | |
5444 #endif | |
5445 #ifdef MACOS_X | |
5446 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) | |
5447 { | |
5448 vcp->vc_type = CONV_MAC_LATIN1; | |
5449 } | |
5450 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_UNICODE)) | |
5451 { | |
5452 vcp->vc_type = CONV_MAC_UTF8; | |
5453 vcp->vc_factor = 2; /* up to twice as long */ | |
5454 } | |
5455 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) | |
5456 { | |
5457 vcp->vc_type = CONV_LATIN1_MAC; | |
5458 } | |
5459 else if ((from_prop & ENC_UNICODE) && (to_prop & ENC_MACROMAN)) | |
5460 { | |
5461 vcp->vc_type = CONV_UTF8_MAC; | |
5462 } | |
5463 #endif | |
5464 # ifdef USE_ICONV | |
5465 else | |
5466 { | |
5467 /* Use iconv() for conversion. */ | |
5468 vcp->vc_fd = (iconv_t)my_iconv_open( | |
5469 (to_prop & ENC_UNICODE) ? (char_u *)"utf-8" : to, | |
5470 (from_prop & ENC_UNICODE) ? (char_u *)"utf-8" : from); | |
5471 if (vcp->vc_fd != (iconv_t)-1) | |
5472 { | |
5473 vcp->vc_type = CONV_ICONV; | |
5474 vcp->vc_factor = 4; /* could be longer too... */ | |
5475 } | |
5476 } | |
5477 # endif | |
5478 if (vcp->vc_type == CONV_NONE) | |
5479 return FAIL; | |
5480 return OK; | |
5481 } | |
5482 | |
5483 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \ | |
5484 || defined(MSDOS) || defined(PROTO) | |
5485 /* | |
5486 * Do conversion on typed input characters in-place. | |
5487 * The input and output are not NUL terminated! | |
5488 * Returns the length after conversion. | |
5489 */ | |
5490 int | |
5491 convert_input(ptr, len, maxlen) | |
5492 char_u *ptr; | |
5493 int len; | |
5494 int maxlen; | |
5495 { | |
5496 return convert_input_safe(ptr, len, maxlen, NULL, NULL); | |
5497 } | |
5498 #endif | |
5499 | |
5500 /* | |
5501 * Like convert_input(), but when there is an incomplete byte sequence at the | |
5502 * end return that as an allocated string in "restp" and set "*restlenp" to | |
5503 * the length. If "restp" is NULL it is not used. | |
5504 */ | |
5505 int | |
5506 convert_input_safe(ptr, len, maxlen, restp, restlenp) | |
5507 char_u *ptr; | |
5508 int len; | |
5509 int maxlen; | |
5510 char_u **restp; | |
5511 int *restlenp; | |
5512 { | |
5513 char_u *d; | |
5514 int dlen = len; | |
5515 int unconvertlen = 0; | |
5516 | |
5517 d = string_convert_ext(&input_conv, ptr, &dlen, | |
5518 restp == NULL ? NULL : &unconvertlen); | |
5519 if (d != NULL) | |
5520 { | |
5521 if (dlen <= maxlen) | |
5522 { | |
5523 if (unconvertlen > 0) | |
5524 { | |
5525 /* Move the unconverted characters to allocated memory. */ | |
5526 *restp = alloc(unconvertlen); | |
5527 if (*restp != NULL) | |
5528 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); | |
5529 *restlenp = unconvertlen; | |
5530 } | |
5531 mch_memmove(ptr, d, dlen); | |
5532 } | |
5533 else | |
5534 /* result is too long, keep the unconverted text (the caller must | |
5535 * have done something wrong!) */ | |
5536 dlen = len; | |
5537 vim_free(d); | |
5538 } | |
5539 return dlen; | |
5540 } | |
5541 | |
5542 #if defined(MACOS_X) | |
5543 static char_u *mac_string_convert __ARGS((char_u *ptr, int len, int *lenp, int fail_on_error, CFStringEncoding from, CFStringEncoding to, int *unconvlenp)); | |
5544 | |
5545 /* | |
5546 * A Mac version of string_convert_ext() for special cases. | |
5547 */ | |
5548 static char_u * | |
5549 mac_string_convert(ptr, len, lenp, fail_on_error, from, to, unconvlenp) | |
5550 char_u *ptr; | |
5551 int len; | |
5552 int *lenp; | |
5553 int fail_on_error; | |
5554 CFStringEncoding from; | |
5555 CFStringEncoding to; | |
5556 int *unconvlenp; | |
5557 { | |
5558 char_u *retval, *d; | |
5559 CFStringRef cfstr; | |
5560 int buflen, in, out, l, i; | |
5561 | |
5562 if (unconvlenp != NULL) | |
5563 *unconvlenp = 0; | |
5564 cfstr = CFStringCreateWithBytes(NULL, ptr, len, from, 0); | |
5565 /* When conversion failed, try excluding bytes from the end, helps when | |
5566 * there is an incomplete byte sequence. Only do up to 6 bytes to avoid | |
5567 * looping a long time when there really is something unconvertable. */ | |
5568 while (cfstr == NULL && unconvlenp != NULL && len > 1 && *unconvlenp < 6) | |
5569 { | |
5570 --len; | |
5571 ++*unconvlenp; | |
5572 cfstr = CFStringCreateWithBytes(NULL, ptr, len, from, 0); | |
5573 } | |
5574 if (cfstr == NULL) | |
5575 return NULL; | |
5576 if (to == kCFStringEncodingUTF8) | |
5577 buflen = len * 6 + 1; | |
5578 else | |
5579 buflen = len + 1; | |
5580 retval = alloc(buflen); | |
5581 if (retval == NULL) | |
5582 { | |
5583 CFRelease(cfstr); | |
5584 return NULL; | |
5585 } | |
5586 if (!CFStringGetCString(cfstr, retval, buflen, to)) | |
5587 { | |
5588 CFRelease(cfstr); | |
5589 if (fail_on_error) | |
5590 { | |
5591 vim_free(retval); | |
5592 return NULL; | |
5593 } | |
5594 | |
5595 /* conversion failed for the whole string, but maybe it will work | |
5596 * for each character */ | |
5597 for (d = retval, in = 0, out = 0; in < len && out < buflen - 1;) | |
5598 { | |
5599 if (from == kCFStringEncodingUTF8) | |
5600 l = utf_ptr2len_check(ptr + in); | |
5601 else | |
5602 l = 1; | |
5603 cfstr = CFStringCreateWithBytes(NULL, ptr + in, l, from, 0); | |
5604 if (cfstr == NULL) | |
5605 { | |
5606 *d++ = '?'; | |
5607 out++; | |
5608 } | |
5609 else | |
5610 { | |
5611 if (!CFStringGetCString(cfstr, d, buflen - out, to)) | |
5612 { | |
5613 *d++ = '?'; | |
5614 out++; | |
5615 } | |
5616 else | |
5617 { | |
5618 i = strlen(d); | |
5619 d += i; | |
5620 out += i; | |
5621 } | |
5622 CFRelease(cfstr); | |
5623 } | |
5624 in += l; | |
5625 } | |
5626 *d = NUL; | |
5627 if (lenp != NULL) | |
5628 *lenp = out; | |
5629 return retval; | |
5630 } | |
5631 CFRelease(cfstr); | |
5632 if (lenp != NULL) | |
5633 *lenp = strlen(retval); | |
5634 return retval; | |
5635 } | |
5636 #endif | |
5637 | |
5638 /* | |
5639 * Convert text "ptr[*lenp]" according to "vcp". | |
5640 * Returns the result in allocated memory and sets "*lenp". | |
5641 * When "lenp" is NULL, use NUL terminated strings. | |
5642 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. | |
5643 * When something goes wrong, NULL is returned and "*lenp" is unchanged. | |
5644 */ | |
5645 char_u * | |
5646 string_convert(vcp, ptr, lenp) | |
5647 vimconv_T *vcp; | |
5648 char_u *ptr; | |
5649 int *lenp; | |
5650 { | |
5651 return string_convert_ext(vcp, ptr, lenp, NULL); | |
5652 } | |
5653 | |
5654 /* | |
5655 * Like string_convert(), but when "unconvlenp" is not NULL and there are is | |
5656 * an incomplete sequence at the end it is not converted and "*unconvlenp" is | |
5657 * set to the number of remaining bytes. | |
5658 */ | |
5659 char_u * | |
5660 string_convert_ext(vcp, ptr, lenp, unconvlenp) | |
5661 vimconv_T *vcp; | |
5662 char_u *ptr; | |
5663 int *lenp; | |
5664 int *unconvlenp; | |
5665 { | |
5666 char_u *retval = NULL; | |
5667 char_u *d; | |
5668 int len; | |
5669 int i; | |
5670 int l; | |
5671 int c; | |
5672 | |
5673 if (lenp == NULL) | |
5674 len = (int)STRLEN(ptr); | |
5675 else | |
5676 len = *lenp; | |
5677 if (len == 0) | |
5678 return vim_strsave((char_u *)""); | |
5679 | |
5680 switch (vcp->vc_type) | |
5681 { | |
5682 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */ | |
5683 retval = alloc(len * 2 + 1); | |
5684 if (retval == NULL) | |
5685 break; | |
5686 d = retval; | |
5687 for (i = 0; i < len; ++i) | |
5688 { | |
5689 if (ptr[i] < 0x80) | |
5690 *d++ = ptr[i]; | |
5691 else | |
5692 { | |
5693 *d++ = 0xc0 + ((unsigned)ptr[i] >> 6); | |
5694 *d++ = 0x80 + (ptr[i] & 0x3f); | |
5695 } | |
5696 } | |
5697 *d = NUL; | |
5698 if (lenp != NULL) | |
5699 *lenp = (int)(d - retval); | |
5700 break; | |
5701 | |
5702 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */ | |
5703 retval = alloc(len + 1); | |
5704 if (retval == NULL) | |
5705 break; | |
5706 d = retval; | |
5707 for (i = 0; i < len; ++i) | |
5708 { | |
5709 l = utf_ptr2len_check(ptr + i); | |
5710 if (l == 0) | |
5711 *d++ = NUL; | |
5712 else if (l == 1) | |
5713 { | |
5714 if (unconvlenp != NULL && utf8len_tab[ptr[i]] > len - i) | |
5715 { | |
5716 /* Incomplete sequence at the end. */ | |
5717 *unconvlenp = len - i; | |
5718 break; | |
5719 } | |
5720 *d++ = ptr[i]; | |
5721 } | |
5722 else | |
5723 { | |
5724 c = utf_ptr2char(ptr + i); | |
5725 if (!utf_iscomposing(c)) /* skip composing chars */ | |
5726 { | |
5727 if (c < 0x100) | |
5728 *d++ = c; | |
5729 else if (vcp->vc_fail) | |
5730 { | |
5731 vim_free(retval); | |
5732 return NULL; | |
5733 } | |
5734 else | |
5735 { | |
5736 *d++ = 0xbf; | |
5737 if (utf_char2cells(c) > 1) | |
5738 *d++ = '?'; | |
5739 } | |
5740 } | |
5741 i += l - 1; | |
5742 } | |
5743 } | |
5744 *d = NUL; | |
5745 if (lenp != NULL) | |
5746 *lenp = (int)(d - retval); | |
5747 break; | |
5748 | |
5749 # ifdef MACOS_X | |
5750 case CONV_MAC_LATIN1: | |
5751 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, | |
5752 kCFStringEncodingMacRoman, | |
5753 kCFStringEncodingISOLatin1, | |
5754 unconvlenp); | |
5755 break; | |
5756 | |
5757 case CONV_LATIN1_MAC: | |
5758 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, | |
5759 kCFStringEncodingISOLatin1, | |
5760 kCFStringEncodingMacRoman, | |
5761 unconvlenp); | |
5762 break; | |
5763 | |
5764 case CONV_MAC_UTF8: | |
5765 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, | |
5766 kCFStringEncodingMacRoman, | |
5767 kCFStringEncodingUTF8, | |
5768 unconvlenp); | |
5769 break; | |
5770 | |
5771 case CONV_UTF8_MAC: | |
5772 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, | |
5773 kCFStringEncodingUTF8, | |
5774 kCFStringEncodingMacRoman, | |
5775 unconvlenp); | |
5776 break; | |
5777 # endif | |
5778 | |
5779 # ifdef USE_ICONV | |
5780 case CONV_ICONV: /* conversion with output_conv.vc_fd */ | |
5781 retval = iconv_string(vcp, ptr, len, unconvlenp); | |
5782 if (retval != NULL && lenp != NULL) | |
5783 *lenp = (int)STRLEN(retval); | |
5784 break; | |
5785 # endif | |
5786 # ifdef WIN3264 | |
5787 case CONV_CODEPAGE: /* codepage -> codepage */ | |
5788 { | |
5789 int retlen; | |
5790 int tmp_len; | |
5791 short_u *tmp; | |
5792 | |
5793 /* 1. codepage/UTF-8 -> ucs-2. */ | |
5794 if (vcp->vc_cpfrom == 0) | |
5795 tmp_len = utf8_to_ucs2(ptr, len, NULL, NULL); | |
5796 else | |
5797 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 0, | |
5798 ptr, len, 0, 0); | |
5799 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len); | |
5800 if (tmp == NULL) | |
5801 break; | |
5802 if (vcp->vc_cpfrom == 0) | |
5803 utf8_to_ucs2(ptr, len, tmp, unconvlenp); | |
5804 else | |
5805 MultiByteToWideChar(vcp->vc_cpfrom, 0, ptr, len, tmp, tmp_len); | |
5806 | |
5807 /* 2. ucs-2 -> codepage/UTF-8. */ | |
5808 if (vcp->vc_cpto == 0) | |
5809 retlen = ucs2_to_utf8(tmp, tmp_len, NULL); | |
5810 else | |
5811 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, | |
5812 tmp, tmp_len, 0, 0, 0, 0); | |
5813 retval = alloc(retlen + 1); | |
5814 if (retval != NULL) | |
5815 { | |
5816 if (vcp->vc_cpto == 0) | |
5817 ucs2_to_utf8(tmp, tmp_len, retval); | |
5818 else | |
5819 WideCharToMultiByte(vcp->vc_cpto, 0, | |
5820 tmp, tmp_len, retval, retlen, 0, 0); | |
5821 retval[retlen] = NUL; | |
5822 if (lenp != NULL) | |
5823 *lenp = retlen; | |
5824 } | |
5825 vim_free(tmp); | |
5826 break; | |
5827 } | |
5828 # endif | |
5829 } | |
5830 | |
5831 return retval; | |
5832 } | |
5833 #endif |