1 /* vi:set ts=8 sts=4 sw=4:
3 * VIM - Vi IMproved by Bram Moolenaar
4 * Multibyte extensions partly by Sung-Hoon Baek
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.
11 * mbyte.c: Code specifically for handling multi-byte characters.
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:
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.
39 * "has_mbyte" is set when "enc_dbcs" or "enc_utf8" is non-zero.
41 * If none of these is TRUE, 8-bit bytes are used for a character. The
42 * encoding isn't currently specified (TODO).
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:
54 * keyboard ----->| core |-----> display
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
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'.
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"
83 # ifndef WIN32_LEAN_AND_MEAN
84 # define WIN32_LEAN_AND_MEAN
88 # undef WIN32 /* Some windows.h define WIN32, we don't want that here. */
92 #if (defined(WIN3264) || defined(WIN32UNIX)) && !defined(__MINGW32__)
97 # include <X11/Intrinsic.h>
100 #include <X11/Xlocale.h>
103 #if defined(FEAT_GUI_GTK) && defined(FEAT_XIM) && defined(HAVE_GTK2)
104 # include <gdk/gdkkeysyms.h>
106 # include <gdk/gdkwin32.h>
108 # include <gdk/gdkx.h>
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
124 #if defined(FEAT_MBYTE) || defined(PROTO)
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
__ARGS((char_u
*p
));
130 static int dbcs_ptr2len_len
__ARGS((char_u
*p
, int size
));
131 static int utf_ptr2cells_len
__ARGS((char_u
*p
, int size
));
132 static int dbcs_char2cells
__ARGS((int c
));
133 static int dbcs_ptr2cells_len
__ARGS((char_u
*p
, int size
));
134 static int dbcs_ptr2char
__ARGS((char_u
*p
));
137 * Lookup table to quickly get the length in bytes of a UTF-8 character from
138 * the first byte of a UTF-8 string.
139 * Bytes which are illegal when used as the first byte have a 1.
140 * The NUL byte has length 1.
142 static char utf8len_tab
[256] =
144 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,
145 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,
146 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,
147 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,
148 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,
149 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,
150 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,
151 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,
155 * Like utf8len_tab above, but using a zero for illegal lead bytes.
157 static char utf8len_tab_zero
[256] =
159 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,
160 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,
161 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,
162 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,
163 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
164 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
165 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,
166 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6,0,0,
170 * XIM often causes trouble. Define XIM_DEBUG to get a log of XIM callbacks
171 * in the "xim.log" file.
173 /* #define XIM_DEBUG */
176 xim_log(char *s
, ...)
179 static FILE *fd
= NULL
;
181 if (fd
== (FILE *)-1)
185 fd
= mch_fopen("xim.log", "w");
188 EMSG("Cannot open xim.log");
194 va_start(arglist
, s
);
195 vfprintf(fd
, s
, arglist
);
202 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO)
204 * Canonical encoding names and their properties.
205 * "iso-8859-n" is handled by enc_canonize() directly.
208 { char *name
; int prop
; int codepage
;}
211 #define IDX_LATIN_1 0
212 {"latin1", ENC_8BIT
+ ENC_LATIN1
, 1252},
214 {"iso-8859-2", ENC_8BIT
, 0},
216 {"iso-8859-3", ENC_8BIT
, 0},
218 {"iso-8859-4", ENC_8BIT
, 0},
220 {"iso-8859-5", ENC_8BIT
, 0},
222 {"iso-8859-6", ENC_8BIT
, 0},
224 {"iso-8859-7", ENC_8BIT
, 0},
226 {"iso-8859-8", ENC_8BIT
, 0},
228 {"iso-8859-9", ENC_8BIT
, 0},
230 {"iso-8859-10", ENC_8BIT
, 0},
231 #define IDX_ISO_11 10
232 {"iso-8859-11", ENC_8BIT
, 0},
233 #define IDX_ISO_13 11
234 {"iso-8859-13", ENC_8BIT
, 0},
235 #define IDX_ISO_14 12
236 {"iso-8859-14", ENC_8BIT
, 0},
237 #define IDX_ISO_15 13
238 {"iso-8859-15", ENC_8BIT
+ ENC_LATIN9
, 0},
239 #define IDX_KOI8_R 14
240 {"koi8-r", ENC_8BIT
, 0},
241 #define IDX_KOI8_U 15
242 {"koi8-u", ENC_8BIT
, 0},
244 {"utf-8", ENC_UNICODE
, 0},
246 {"ucs-2", ENC_UNICODE
+ ENC_ENDIAN_B
+ ENC_2BYTE
, 0},
247 #define IDX_UCS2LE 18
248 {"ucs-2le", ENC_UNICODE
+ ENC_ENDIAN_L
+ ENC_2BYTE
, 0},
250 {"utf-16", ENC_UNICODE
+ ENC_ENDIAN_B
+ ENC_2WORD
, 0},
251 #define IDX_UTF16LE 20
252 {"utf-16le", ENC_UNICODE
+ ENC_ENDIAN_L
+ ENC_2WORD
, 0},
254 {"ucs-4", ENC_UNICODE
+ ENC_ENDIAN_B
+ ENC_4BYTE
, 0},
255 #define IDX_UCS4LE 22
256 {"ucs-4le", ENC_UNICODE
+ ENC_ENDIAN_L
+ ENC_4BYTE
, 0},
258 /* For debugging DBCS encoding on Unix. */
260 {"debug", ENC_DBCS
, DBCS_DEBUG
},
261 #define IDX_EUC_JP 24
262 {"euc-jp", ENC_DBCS
, DBCS_JPNU
},
264 {"sjis", ENC_DBCS
, DBCS_JPN
},
265 #define IDX_EUC_KR 26
266 {"euc-kr", ENC_DBCS
, DBCS_KORU
},
267 #define IDX_EUC_CN 27
268 {"euc-cn", ENC_DBCS
, DBCS_CHSU
},
269 #define IDX_EUC_TW 28
270 {"euc-tw", ENC_DBCS
, DBCS_CHTU
},
272 {"big5", ENC_DBCS
, DBCS_CHT
},
274 /* MS-DOS and MS-Windows codepages are included here, so that they can be
275 * used on Unix too. Most of them are similar to ISO-8859 encodings, but
276 * not exactly the same. */
278 {"cp437", ENC_8BIT
, 437}, /* like iso-8859-1 */
280 {"cp737", ENC_8BIT
, 737}, /* like iso-8859-7 */
282 {"cp775", ENC_8BIT
, 775}, /* Baltic */
284 {"cp850", ENC_8BIT
, 850}, /* like iso-8859-4 */
286 {"cp852", ENC_8BIT
, 852}, /* like iso-8859-1 */
288 {"cp855", ENC_8BIT
, 855}, /* like iso-8859-2 */
290 {"cp857", ENC_8BIT
, 857}, /* like iso-8859-5 */
292 {"cp860", ENC_8BIT
, 860}, /* like iso-8859-9 */
294 {"cp861", ENC_8BIT
, 861}, /* like iso-8859-1 */
296 {"cp862", ENC_8BIT
, 862}, /* like iso-8859-1 */
298 {"cp863", ENC_8BIT
, 863}, /* like iso-8859-8 */
300 {"cp865", ENC_8BIT
, 865}, /* like iso-8859-1 */
302 {"cp866", ENC_8BIT
, 866}, /* like iso-8859-5 */
304 {"cp869", ENC_8BIT
, 869}, /* like iso-8859-7 */
306 {"cp874", ENC_8BIT
, 874}, /* Thai */
308 {"cp932", ENC_DBCS
, DBCS_JPN
},
310 {"cp936", ENC_DBCS
, DBCS_CHS
},
312 {"cp949", ENC_DBCS
, DBCS_KOR
},
314 {"cp950", ENC_DBCS
, DBCS_CHT
},
315 #define IDX_CP1250 49
316 {"cp1250", ENC_8BIT
, 1250}, /* Czech, Polish, etc. */
317 #define IDX_CP1251 50
318 {"cp1251", ENC_8BIT
, 1251}, /* Cyrillic */
319 /* cp1252 is considered to be equal to latin1 */
320 #define IDX_CP1253 51
321 {"cp1253", ENC_8BIT
, 1253}, /* Greek */
322 #define IDX_CP1254 52
323 {"cp1254", ENC_8BIT
, 1254}, /* Turkish */
324 #define IDX_CP1255 53
325 {"cp1255", ENC_8BIT
, 1255}, /* Hebrew */
326 #define IDX_CP1256 54
327 {"cp1256", ENC_8BIT
, 1256}, /* Arabic */
328 #define IDX_CP1257 55
329 {"cp1257", ENC_8BIT
, 1257}, /* Baltic */
330 #define IDX_CP1258 56
331 {"cp1258", ENC_8BIT
, 1258}, /* Vietnamese */
333 #define IDX_MACROMAN 57
334 {"macroman", ENC_8BIT
+ ENC_MACROMAN
, 0}, /* Mac OS */
335 #define IDX_DECMCS 58
336 {"dec-mcs", ENC_8BIT
, 0}, /* DEC MCS */
337 #define IDX_HPROMAN8 59
338 {"hp-roman8", ENC_8BIT
, 0}, /* HP Roman8 */
343 * Aliases for encoding names.
346 { char *name
; int canon
;}
349 {"ansi", IDX_LATIN_1
},
350 {"iso-8859-1", IDX_LATIN_1
},
351 {"latin2", IDX_ISO_2
},
352 {"latin3", IDX_ISO_3
},
353 {"latin4", IDX_ISO_4
},
354 {"cyrillic", IDX_ISO_5
},
355 {"arabic", IDX_ISO_6
},
356 {"greek", IDX_ISO_7
},
358 {"hebrew", IDX_CP1255
},
360 {"hebrew", IDX_ISO_8
},
362 {"latin5", IDX_ISO_9
},
363 {"turkish", IDX_ISO_9
}, /* ? */
364 {"latin6", IDX_ISO_10
},
365 {"nordic", IDX_ISO_10
}, /* ? */
366 {"thai", IDX_ISO_11
}, /* ? */
367 {"latin7", IDX_ISO_13
},
368 {"latin8", IDX_ISO_14
},
369 {"latin9", IDX_ISO_15
},
371 {"unicode", IDX_UCS2
},
373 {"ucs2be", IDX_UCS2
},
374 {"ucs-2be", IDX_UCS2
},
375 {"ucs2le", IDX_UCS2LE
},
376 {"utf16", IDX_UTF16
},
377 {"utf16be", IDX_UTF16
},
378 {"utf-16be", IDX_UTF16
},
379 {"utf16le", IDX_UTF16LE
},
381 {"ucs4be", IDX_UCS4
},
382 {"ucs-4be", IDX_UCS4
},
383 {"ucs4le", IDX_UCS4LE
},
385 {"utf-32", IDX_UCS4
},
386 {"utf32be", IDX_UCS4
},
387 {"utf-32be", IDX_UCS4
},
388 {"utf32le", IDX_UCS4LE
},
389 {"utf-32le", IDX_UCS4LE
},
395 {"eucjp", IDX_EUC_JP
},
396 {"unix-jis", IDX_EUC_JP
},
397 {"ujis", IDX_EUC_JP
},
398 {"shift-jis", IDX_SJIS
},
399 {"euckr", IDX_EUC_KR
},
400 {"5601", IDX_EUC_KR
}, /* Sun: KS C 5601 */
401 {"euccn", IDX_EUC_CN
},
402 {"gb2312", IDX_EUC_CN
},
403 {"euctw", IDX_EUC_TW
},
404 #if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS)
405 {"japan", IDX_CP932
},
406 {"korea", IDX_CP949
},
408 {"chinese", IDX_CP936
},
409 {"taiwan", IDX_CP950
},
412 {"japan", IDX_EUC_JP
},
413 {"korea", IDX_EUC_KR
},
415 {"chinese", IDX_EUC_CN
},
416 {"taiwan", IDX_EUC_TW
},
420 {"mac", IDX_MACROMAN
},
421 {"mac-roman", IDX_MACROMAN
},
426 # define CP_UTF8 65001 /* magic number from winnls.h */
430 * Find encoding "name" in the list of canonical encoding names.
431 * Returns -1 if not found.
434 enc_canon_search(name
)
439 for (i
= 0; i
< IDX_COUNT
; ++i
)
440 if (STRCMP(name
, enc_canon_table
[i
].name
) == 0)
447 #if defined(FEAT_MBYTE) || defined(PROTO)
450 * Find canonical encoding "name" in the list and return its properties.
451 * Returns 0 if not found.
454 enc_canon_props(name
)
459 i
= enc_canon_search(name
);
461 return enc_canon_table
[i
].prop
;
463 if (name
[0] == 'c' && name
[1] == 'p' && VIM_ISDIGIT(name
[2]))
467 /* Get info on this codepage to find out what it is. */
468 if (GetCPInfo(atoi(name
+ 2), &cpinfo
) != 0)
470 if (cpinfo
.MaxCharSize
== 1) /* some single-byte encoding */
472 if (cpinfo
.MaxCharSize
== 2
473 && (cpinfo
.LeadByte
[0] != 0 || cpinfo
.LeadByte
[1] != 0))
474 /* must be a DBCS encoding */
480 if (STRNCMP(name
, "2byte-", 6) == 0)
482 if (STRNCMP(name
, "8bit-", 5) == 0 || STRNCMP(name
, "iso-8859-", 9) == 0)
488 * Set up for using multi-byte characters.
489 * Called in three cases:
490 * - by main() to initialize (p_enc == NULL)
491 * - by set_init_1() after 'encoding' was set to its default.
492 * - by do_set() when 'encoding' has been set.
493 * p_enc must have been passed through enc_canonize() already.
494 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags.
495 * Fills mb_bytelen_tab[] and returns NULL when there are no problems.
496 * When there is something wrong: Returns an error message and doesn't change
505 int enc_dbcs_new
= 0;
506 #if defined(USE_ICONV) && !defined(WIN3264) && !defined(WIN32UNIX) \
508 # define LEN_FROM_CONV
515 /* Just starting up: set the whole table to one's. */
516 for (i
= 0; i
< 256; ++i
)
517 mb_bytelen_tab
[i
] = 1;
518 input_conv
.vc_type
= CONV_NONE
;
519 input_conv
.vc_factor
= 1;
520 output_conv
.vc_type
= CONV_NONE
;
525 if (p_enc
[0] == 'c' && p_enc
[1] == 'p' && VIM_ISDIGIT(p_enc
[2]))
529 /* Get info on this codepage to find out what it is. */
530 if (GetCPInfo(atoi(p_enc
+ 2), &cpinfo
) != 0)
532 if (cpinfo
.MaxCharSize
== 1)
534 /* some single-byte encoding */
538 else if (cpinfo
.MaxCharSize
== 2
539 && (cpinfo
.LeadByte
[0] != 0 || cpinfo
.LeadByte
[1] != 0))
541 /* must be a DBCS encoding, check below */
542 enc_dbcs_new
= atoi(p_enc
+ 2);
545 goto codepage_invalid
;
547 else if (GetLastError() == ERROR_INVALID_PARAMETER
)
550 return (char_u
*)N_("E543: Not a valid codepage");
554 else if (STRNCMP(p_enc
, "8bit-", 5) == 0
555 || STRNCMP(p_enc
, "iso-8859-", 9) == 0)
557 /* Accept any "8bit-" or "iso-8859-" name. */
561 else if (STRNCMP(p_enc
, "2byte-", 6) == 0)
564 /* Windows: accept only valid codepage numbers, check below. */
565 if (p_enc
[6] != 'c' || p_enc
[7] != 'p'
566 || (enc_dbcs_new
= atoi(p_enc
+ 8)) == 0)
569 /* Unix: accept any "2byte-" name, assume current locale. */
570 enc_dbcs_new
= DBCS_2BYTE
;
573 else if ((idx
= enc_canon_search(p_enc
)) >= 0)
575 i
= enc_canon_table
[idx
].prop
;
580 if (i
& (ENC_2BYTE
| ENC_2WORD
))
582 else if (i
& ENC_4BYTE
)
587 else if (i
& ENC_DBCS
)
589 /* 2byte, handle below */
590 enc_dbcs_new
= enc_canon_table
[idx
].codepage
;
599 else /* Don't know what encoding this is, reject it. */
602 if (enc_dbcs_new
!= 0)
605 /* Check if the DBCS code page is OK. */
606 if (!IsValidCodePage(enc_dbcs_new
))
607 goto codepage_invalid
;
612 enc_dbcs
= enc_dbcs_new
;
613 has_mbyte
= (enc_dbcs
!= 0 || enc_utf8
);
616 enc_codepage
= encname2codepage(p_enc
);
617 enc_latin9
= (STRCMP(p_enc
, "iso-8859-15") == 0);
620 /* Detect an encoding that uses latin1 characters. */
621 enc_latin1like
= (enc_utf8
|| STRCMP(p_enc
, "latin1") == 0
622 || STRCMP(p_enc
, "iso-8859-15") == 0);
625 * Set the function pointers.
629 mb_ptr2len
= utfc_ptr2len
;
630 mb_ptr2len_len
= utfc_ptr2len_len
;
631 mb_char2len
= utf_char2len
;
632 mb_char2bytes
= utf_char2bytes
;
633 mb_ptr2cells
= utf_ptr2cells
;
634 mb_ptr2cells_len
= utf_ptr2cells_len
;
635 mb_char2cells
= utf_char2cells
;
636 mb_off2cells
= utf_off2cells
;
637 mb_ptr2char
= utf_ptr2char
;
638 mb_head_off
= utf_head_off
;
640 else if (enc_dbcs
!= 0)
642 mb_ptr2len
= dbcs_ptr2len
;
643 mb_ptr2len_len
= dbcs_ptr2len_len
;
644 mb_char2len
= dbcs_char2len
;
645 mb_char2bytes
= dbcs_char2bytes
;
646 mb_ptr2cells
= dbcs_ptr2cells
;
647 mb_ptr2cells_len
= dbcs_ptr2cells_len
;
648 mb_char2cells
= dbcs_char2cells
;
649 mb_off2cells
= dbcs_off2cells
;
650 mb_ptr2char
= dbcs_ptr2char
;
651 mb_head_off
= dbcs_head_off
;
655 mb_ptr2len
= latin_ptr2len
;
656 mb_ptr2len_len
= latin_ptr2len_len
;
657 mb_char2len
= latin_char2len
;
658 mb_char2bytes
= latin_char2bytes
;
659 mb_ptr2cells
= latin_ptr2cells
;
660 mb_ptr2cells_len
= latin_ptr2cells_len
;
661 mb_char2cells
= latin_char2cells
;
662 mb_off2cells
= latin_off2cells
;
663 mb_ptr2char
= latin_ptr2char
;
664 mb_head_off
= latin_head_off
;
668 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN().
671 /* When 'encoding' is different from the current locale mblen() won't
672 * work. Use conversion to "utf-8" instead. */
673 vimconv
.vc_type
= CONV_NONE
;
677 if (p
== NULL
|| STRCMP(p
, p_enc
) != 0)
679 convert_setup(&vimconv
, p_enc
, (char_u
*)"utf-8");
680 vimconv
.vc_fail
= TRUE
;
686 for (i
= 0; i
< 256; ++i
)
688 /* Our own function to reliably check the length of UTF-8 characters,
689 * independent of mblen(). */
692 else if (enc_dbcs
== 0)
696 #if defined(WIN3264) || defined(WIN32UNIX)
697 /* enc_dbcs is set by setting 'fileencoding'. It becomes a Windows
698 * CodePage identifier, which we can pass directly in to Windows
700 n
= IsDBCSLeadByteEx(enc_dbcs
, (BYTE
)i
) ? 2 : 1;
702 # if defined(MACOS) || defined(__amigaos4__)
704 * if mblen() is not available, character which MSB is turned on
705 * are treated as leading byte character. (note : This assumption
706 * is not always true.)
708 n
= (i
& 0x80) ? 2 : 1;
710 char buf
[MB_MAXBYTES
];
713 # define mblen _Xmblen
716 if (i
== NUL
) /* just in case mblen() can't handle "" */
723 if (vimconv
.vc_type
!= CONV_NONE
)
726 * string_convert() should fail when converting the first
727 * byte of a double-byte character.
729 p
= string_convert(&vimconv
, (char_u
*)buf
, NULL
);
742 * mblen() should return -1 for invalid (means the leading
743 * multibyte) character. However there are some platforms
744 * where mblen() returns 0 for invalid character.
745 * Therefore, following condition includes 0.
747 ignored
= mblen(NULL
, 0); /* First reset the state. */
748 if (mblen(buf
, (size_t)1) <= 0)
758 mb_bytelen_tab
[i
] = n
;
762 convert_setup(&vimconv
, NULL
, NULL
);
765 /* The cell width depends on the type of multi-byte characters. */
766 (void)init_chartab();
768 /* When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] */
771 /* When using Unicode, set default for 'fileencodings'. */
772 if (enc_utf8
&& !option_was_set((char_u
*)"fencs"))
773 set_string_option_direct((char_u
*)"fencs", -1,
774 (char_u
*)"ucs-bom,utf-8,default,latin1", OPT_FREE
, 0);
776 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT)
777 /* GNU gettext 0.10.37 supports this feature: set the codeset used for
778 * translated messages independently from the current locale. */
779 (void)bind_textdomain_codeset(VIMPACKAGE
,
780 enc_utf8
? "utf-8" : (char *)p_enc
);
784 /* When changing 'encoding' while starting up, then convert the command
785 * line arguments from the active codepage to 'encoding'. */
791 /* Fire an autocommand to let people do custom font setup. This must be
792 * after Vim has been setup for the new encoding. */
793 apply_autocmds(EVENT_ENCODINGCHANGED
, NULL
, (char_u
*)"", FALSE
, curbuf
);
797 /* Need to reload spell dictionaries */
805 * Return the size of the BOM for the current buffer:
807 * 2 - UCS-2 or UTF-16 BOM
816 if (curbuf
->b_p_bomb
&& !curbuf
->b_p_bin
)
818 if (*curbuf
->b_p_fenc
== NUL
)
822 if (enc_unicode
!= 0)
828 else if (STRCMP(curbuf
->b_p_fenc
, "utf-8") == 0)
830 else if (STRNCMP(curbuf
->b_p_fenc
, "ucs-2", 5) == 0
831 || STRNCMP(curbuf
->b_p_fenc
, "utf-16", 6) == 0)
833 else if (STRNCMP(curbuf
->b_p_fenc
, "ucs-4", 5) == 0)
840 * Get class of pointer:
843 * 2 for an (ASCII) word character
844 * >2 for other word characters
850 if (MB_BYTE2LEN(p
[0]) == 1)
852 if (p
[0] == NUL
|| vim_iswhite(p
[0]))
854 if (vim_iswordc(p
[0]))
858 if (enc_dbcs
!= 0 && p
[0] != NUL
&& p
[1] != NUL
)
859 return dbcs_class(p
[0], p
[1]);
861 return utf_class(utf_ptr2char(p
));
866 * Get class of a double-byte character. This always returns 3 or bigger.
867 * TODO: Should return 1 for punctuation.
870 dbcs_class(lead
, trail
)
876 /* please add classfy routine for your language in here */
878 case DBCS_JPNU
: /* ? */
881 /* JIS code classification */
882 unsigned char lb
= lead
;
883 unsigned char tb
= trail
;
885 /* convert process code to JIS */
886 # if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS)
887 /* process code is SJIS */
889 lb
= (lb
- 0x81) * 2 + 0x21;
891 lb
= (lb
- 0xc1) * 2 + 0x21;
903 * XXX: Code page identification can not use with all
904 * system! So, some other encoding information
906 * In japanese: SJIS,EUC,UNICODE,(JIS)
907 * Note that JIS-code system don't use as
908 * process code in most system because it uses
909 * escape sequences(JIS is context depend encoding).
911 /* assume process code is JAPANESE-EUC */
916 switch (lb
<< 8 | tb
)
918 case 0x2121: /* ZENKAKU space */
920 case 0x2122: /* KU-TEN (Japanese comma) */
921 case 0x2123: /* TOU-TEN (Japanese period) */
922 case 0x2124: /* ZENKAKU comma */
923 case 0x2125: /* ZENKAKU period */
925 case 0x213c: /* prolongedsound handled as KATAKANA */
928 /* sieved by KU code */
933 /* special symbols */
959 case DBCS_KORU
: /* ? */
962 /* KS code classification */
963 unsigned char c1
= lead
;
964 unsigned char c2
= trail
;
970 * 23 : Alpha-numeric/Roman Letter (Full width)
971 * 24 : Hangul Letter(Alphabet)
972 * 25 : Roman Numeral/Greek Letter
975 * 28 : Circled/Parenthesized Letter
976 * 29 : Hirigana/Katakana
977 * 30 : Cyrillic Letter
980 if (c1
>= 0xB0 && c1
<= 0xC8)
983 #if defined(WIN3264) || defined(WIN32UNIX)
984 else if (c1
<= 0xA0 || c2
<= 0xA0)
985 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */
986 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE
987 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0
992 else if (c1
>= 0xCA && c1
<= 0xFD)
1005 /* Hangul Letter(Alphabet) */
1008 /* Roman Numeral/Greek Letter */
1019 return 25; /* Roman Letter */
1020 else if (c2
>= 0xF6)
1021 return 22; /* Symbols */
1023 /* Circled/Parenthesized Letter */
1027 /* Hirigana/Katakana */
1030 /* Cyrillic Letter */
1041 * mb_char2len() function pointer.
1042 * Return length in bytes of character "c".
1043 * Returns 1 for a single-byte character.
1062 * mb_char2bytes() function pointer.
1063 * Convert a character to its bytes.
1064 * Returns the length in bytes.
1067 latin_char2bytes(c
, buf
)
1076 dbcs_char2bytes(c
, buf
)
1082 buf
[0] = (unsigned)c
>> 8;
1084 /* Never use a NUL byte, it causes lots of trouble. It's an invalid
1085 * character anyway. */
1095 * mb_ptr2len() function pointer.
1096 * Get byte length of character at "*p" but stop at a NUL.
1097 * For UTF-8 this includes following composing characters.
1098 * Returns 0 when *p is NUL.
1104 return MB_BYTE2LEN(*p
);
1113 /* Check if second byte is not missing. */
1114 len
= MB_BYTE2LEN(*p
);
1115 if (len
== 2 && p
[1] == NUL
)
1121 * mb_ptr2len_len() function pointer.
1122 * Like mb_ptr2len(), but limit to read "size" bytes.
1123 * Returns 0 for an empty string.
1124 * Returns 1 for an illegal char or an incomplete byte sequence.
1127 latin_ptr2len_len(p
, size
)
1131 if (size
< 1 || *p
== NUL
)
1137 dbcs_ptr2len_len(p
, size
)
1143 if (size
< 1 || *p
== NUL
)
1147 /* Check that second byte is not missing. */
1148 len
= MB_BYTE2LEN(*p
);
1149 if (len
== 2 && p
[1] == NUL
)
1156 unsigned short first
;
1157 unsigned short last
;
1159 static int intable
__ARGS((struct interval
*table
, size_t size
, int c
));
1162 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]".
1165 intable(table
, size
, c
)
1166 struct interval
*table
;
1172 /* first quick check for Latin1 etc. characters */
1173 if (c
< table
[0].first
)
1176 /* binary search in table */
1178 top
= (int)(size
/ sizeof(struct interval
) - 1);
1181 mid
= (bot
+ top
) / 2;
1182 if (table
[mid
].last
< c
)
1184 else if (table
[mid
].first
> c
)
1193 * For UTF-8 character "c" return 2 for a double-width character, 1 for others.
1194 * Returns 4 or 6 for an unprintable character.
1195 * Is only correct for characters >= 0x80.
1196 * When p_ambw is "double", return 2 for a character with East Asian Width
1197 * class 'A'(mbiguous).
1203 /* sorted list of non-overlapping intervals of East Asian Ambiguous
1204 * characters, generated with:
1205 * "uniset +WIDTH-A -cat=Me -cat=Mn -cat=Cf c" */
1206 static struct interval ambiguous
[] = {
1207 {0x00A1, 0x00A1}, {0x00A4, 0x00A4}, {0x00A7, 0x00A8},
1208 {0x00AA, 0x00AA}, {0x00AE, 0x00AE}, {0x00B0, 0x00B4},
1209 {0x00B6, 0x00BA}, {0x00BC, 0x00BF}, {0x00C6, 0x00C6},
1210 {0x00D0, 0x00D0}, {0x00D7, 0x00D8}, {0x00DE, 0x00E1},
1211 {0x00E6, 0x00E6}, {0x00E8, 0x00EA}, {0x00EC, 0x00ED},
1212 {0x00F0, 0x00F0}, {0x00F2, 0x00F3}, {0x00F7, 0x00FA},
1213 {0x00FC, 0x00FC}, {0x00FE, 0x00FE}, {0x0101, 0x0101},
1214 {0x0111, 0x0111}, {0x0113, 0x0113}, {0x011B, 0x011B},
1215 {0x0126, 0x0127}, {0x012B, 0x012B}, {0x0131, 0x0133},
1216 {0x0138, 0x0138}, {0x013F, 0x0142}, {0x0144, 0x0144},
1217 {0x0148, 0x014B}, {0x014D, 0x014D}, {0x0152, 0x0153},
1218 {0x0166, 0x0167}, {0x016B, 0x016B}, {0x01CE, 0x01CE},
1219 {0x01D0, 0x01D0}, {0x01D2, 0x01D2}, {0x01D4, 0x01D4},
1220 {0x01D6, 0x01D6}, {0x01D8, 0x01D8}, {0x01DA, 0x01DA},
1221 {0x01DC, 0x01DC}, {0x0251, 0x0251}, {0x0261, 0x0261},
1222 {0x02C4, 0x02C4}, {0x02C7, 0x02C7}, {0x02C9, 0x02CB},
1223 {0x02CD, 0x02CD}, {0x02D0, 0x02D0}, {0x02D8, 0x02DB},
1224 {0x02DD, 0x02DD}, {0x02DF, 0x02DF}, {0x0391, 0x03A1},
1225 {0x03A3, 0x03A9}, {0x03B1, 0x03C1}, {0x03C3, 0x03C9},
1226 {0x0401, 0x0401}, {0x0410, 0x044F}, {0x0451, 0x0451},
1227 {0x2010, 0x2010}, {0x2013, 0x2016}, {0x2018, 0x2019},
1228 {0x201C, 0x201D}, {0x2020, 0x2022}, {0x2024, 0x2027},
1229 {0x2030, 0x2030}, {0x2032, 0x2033}, {0x2035, 0x2035},
1230 {0x203B, 0x203B}, {0x203E, 0x203E}, {0x2074, 0x2074},
1231 {0x207F, 0x207F}, {0x2081, 0x2084}, {0x20AC, 0x20AC},
1232 {0x2103, 0x2103}, {0x2105, 0x2105}, {0x2109, 0x2109},
1233 {0x2113, 0x2113}, {0x2116, 0x2116}, {0x2121, 0x2122},
1234 {0x2126, 0x2126}, {0x212B, 0x212B}, {0x2153, 0x2154},
1235 {0x215B, 0x215E}, {0x2160, 0x216B}, {0x2170, 0x2179},
1236 {0x2190, 0x2199}, {0x21B8, 0x21B9}, {0x21D2, 0x21D2},
1237 {0x21D4, 0x21D4}, {0x21E7, 0x21E7}, {0x2200, 0x2200},
1238 {0x2202, 0x2203}, {0x2207, 0x2208}, {0x220B, 0x220B},
1239 {0x220F, 0x220F}, {0x2211, 0x2211}, {0x2215, 0x2215},
1240 {0x221A, 0x221A}, {0x221D, 0x2220}, {0x2223, 0x2223},
1241 {0x2225, 0x2225}, {0x2227, 0x222C}, {0x222E, 0x222E},
1242 {0x2234, 0x2237}, {0x223C, 0x223D}, {0x2248, 0x2248},
1243 {0x224C, 0x224C}, {0x2252, 0x2252}, {0x2260, 0x2261},
1244 {0x2264, 0x2267}, {0x226A, 0x226B}, {0x226E, 0x226F},
1245 {0x2282, 0x2283}, {0x2286, 0x2287}, {0x2295, 0x2295},
1246 {0x2299, 0x2299}, {0x22A5, 0x22A5}, {0x22BF, 0x22BF},
1247 {0x2312, 0x2312}, {0x2460, 0x24E9}, {0x24EB, 0x254B},
1248 {0x2550, 0x2573}, {0x2580, 0x258F}, {0x2592, 0x2595},
1249 {0x25A0, 0x25A1}, {0x25A3, 0x25A9}, {0x25B2, 0x25B3},
1250 {0x25B6, 0x25B7}, {0x25BC, 0x25BD}, {0x25C0, 0x25C1},
1251 {0x25C6, 0x25C8}, {0x25CB, 0x25CB}, {0x25CE, 0x25D1},
1252 {0x25E2, 0x25E5}, {0x25EF, 0x25EF}, {0x2605, 0x2606},
1253 {0x2609, 0x2609}, {0x260E, 0x260F}, {0x2614, 0x2615},
1254 {0x261C, 0x261C}, {0x261E, 0x261E}, {0x2640, 0x2640},
1255 {0x2642, 0x2642}, {0x2660, 0x2661}, {0x2663, 0x2665},
1256 {0x2667, 0x266A}, {0x266C, 0x266D}, {0x266F, 0x266F},
1257 {0x273D, 0x273D}, {0x2776, 0x277F}, {0xE000, 0xF8FF},
1258 {0xFFFD, 0xFFFD}, /* {0xF0000, 0xFFFFD}, {0x100000, 0x10FFFD} */
1263 #ifdef USE_WCHAR_FUNCTIONS
1265 * Assume the library function wcwidth() works better than our own
1266 * stuff. It should return 1 for ambiguous width chars!
1271 return 6; /* unprintable, displays <xxxx> */
1275 if (!utf_printable(c
))
1276 return 6; /* unprintable, displays <xxxx> */
1278 && (c
<= 0x115f /* Hangul Jamo */
1281 || (c
>= 0x2e80 && c
<= 0xa4cf
1282 && c
!= 0x303f) /* CJK ... Yi */
1283 || (c
>= 0xac00 && c
<= 0xd7a3) /* Hangul Syllables */
1284 || (c
>= 0xf900 && c
<= 0xfaff) /* CJK Compatibility
1286 || (c
>= 0xfe30 && c
<= 0xfe6f) /* CJK Compatibility Forms */
1287 || (c
>= 0xff00 && c
<= 0xff60) /* Fullwidth Forms */
1288 || (c
>= 0xffe0 && c
<= 0xffe6)
1289 || (c
>= 0x20000 && c
<= 0x2fffd)
1290 || (c
>= 0x30000 && c
<= 0x3fffd)))
1295 /* Characters below 0x100 are influenced by 'isprint' option */
1296 else if (c
>= 0x80 && !vim_isprintc(c
))
1297 return 4; /* unprintable, displays <xx> */
1299 if (c
>= 0x80 && *p_ambw
== 'd' && intable(ambiguous
, sizeof(ambiguous
), c
))
1306 * mb_ptr2cells() function pointer.
1307 * Return the number of display cells character at "*p" occupies.
1308 * This doesn't take care of unprintable characters, use ptr2cells() for that.
1323 /* Need to convert to a wide character. */
1326 c
= utf_ptr2char(p
);
1327 /* An illegal byte is displayed as <xx>. */
1328 if (utf_ptr2len(p
) == 1 || c
== NUL
)
1330 /* If the char is ASCII it must be an overlong sequence. */
1332 return char2cells(c
);
1333 return utf_char2cells(c
);
1342 /* Number of cells is equal to number of bytes, except for euc-jp when
1343 * the first byte is 0x8e. */
1344 if (enc_dbcs
== DBCS_JPNU
&& *p
== 0x8e)
1346 return MB_BYTE2LEN(*p
);
1350 * mb_ptr2cells_len() function pointer.
1351 * Like mb_ptr2cells(), but limit string length to "size".
1352 * For an empty string or truncated character returns 1.
1355 latin_ptr2cells_len(p
, size
)
1363 utf_ptr2cells_len(p
, size
)
1369 /* Need to convert to a wide character. */
1370 if (size
> 0 && *p
>= 0x80)
1372 if (utf_ptr2len_len(p
, size
) < utf8len_tab
[*p
])
1373 return 1; /* truncated */
1374 c
= utf_ptr2char(p
);
1375 /* An illegal byte is displayed as <xx>. */
1376 if (utf_ptr2len(p
) == 1 || c
== NUL
)
1378 /* If the char is ASCII it must be an overlong sequence. */
1380 return char2cells(c
);
1381 return utf_char2cells(c
);
1387 dbcs_ptr2cells_len(p
, size
)
1391 /* Number of cells is equal to number of bytes, except for euc-jp when
1392 * the first byte is 0x8e. */
1393 if (size
<= 1 || (enc_dbcs
== DBCS_JPNU
&& *p
== 0x8e))
1395 return MB_BYTE2LEN(*p
);
1399 * mb_char2cells() function pointer.
1400 * Return the number of display cells character "c" occupies.
1401 * Only takes care of multi-byte chars, not "^C" and such.
1414 /* Number of cells is equal to number of bytes, except for euc-jp when
1415 * the first byte is 0x8e. */
1416 if (enc_dbcs
== DBCS_JPNU
&& ((unsigned)c
>> 8) == 0x8e)
1418 /* use the first byte */
1419 return MB_BYTE2LEN((unsigned)c
>> 8);
1423 * mb_off2cells() function pointer.
1424 * Return number of display cells for char at ScreenLines[off].
1425 * We make sure that the offset used is less than "max_off".
1428 latin_off2cells(off
, max_off
)
1429 unsigned off UNUSED
;
1430 unsigned max_off UNUSED
;
1436 dbcs_off2cells(off
, max_off
)
1440 /* never check beyond end of the line */
1444 /* Number of cells is equal to number of bytes, except for euc-jp when
1445 * the first byte is 0x8e. */
1446 if (enc_dbcs
== DBCS_JPNU
&& ScreenLines
[off
] == 0x8e)
1448 return MB_BYTE2LEN(ScreenLines
[off
]);
1452 utf_off2cells(off
, max_off
)
1456 return (off
+ 1 < max_off
&& ScreenLines
[off
+ 1] == 0) ? 2 : 1;
1460 * mb_ptr2char() function pointer.
1461 * Convert a byte sequence into a character.
1474 if (MB_BYTE2LEN(*p
) > 1 && p
[1] != NUL
)
1475 return (p
[0] << 8) + p
[1];
1480 * Convert a UTF-8 byte sequence to a wide character.
1481 * If the sequence is illegal or truncated by a NUL the first byte is
1483 * Does not include composing characters, of course.
1491 if (p
[0] < 0x80) /* be quick for ASCII */
1494 len
= utf8len_tab_zero
[p
[0]];
1495 if (len
> 1 && (p
[1] & 0xc0) == 0x80)
1498 return ((p
[0] & 0x1f) << 6) + (p
[1] & 0x3f);
1499 if ((p
[2] & 0xc0) == 0x80)
1502 return ((p
[0] & 0x0f) << 12) + ((p
[1] & 0x3f) << 6)
1504 if ((p
[3] & 0xc0) == 0x80)
1507 return ((p
[0] & 0x07) << 18) + ((p
[1] & 0x3f) << 12)
1508 + ((p
[2] & 0x3f) << 6) + (p
[3] & 0x3f);
1509 if ((p
[4] & 0xc0) == 0x80)
1512 return ((p
[0] & 0x03) << 24) + ((p
[1] & 0x3f) << 18)
1513 + ((p
[2] & 0x3f) << 12) + ((p
[3] & 0x3f) << 6)
1515 if ((p
[5] & 0xc0) == 0x80 && len
== 6)
1516 return ((p
[0] & 0x01) << 30) + ((p
[1] & 0x3f) << 24)
1517 + ((p
[2] & 0x3f) << 18) + ((p
[3] & 0x3f) << 12)
1518 + ((p
[4] & 0x3f) << 6) + (p
[5] & 0x3f);
1523 /* Illegal value, just return the first byte */
1528 * Get character at **pp and advance *pp to the next character.
1529 * Note: composing characters are skipped!
1537 c
= (*mb_ptr2char
)(*pp
);
1538 *pp
+= (*mb_ptr2len
)(*pp
);
1543 * Get character at **pp and advance *pp to the next character.
1544 * Note: composing characters are returned as separate characters.
1547 mb_cptr2char_adv(pp
)
1552 c
= (*mb_ptr2char
)(*pp
);
1554 *pp
+= utf_ptr2len(*pp
);
1556 *pp
+= (*mb_ptr2len
)(*pp
);
1560 #if defined(FEAT_ARABIC) || defined(PROTO)
1562 * Check whether we are dealing with Arabic combining characters.
1563 * Note: these are NOT really composing characters!
1566 arabic_combine(one
, two
)
1567 int one
; /* first character */
1568 int two
; /* character just after "one" */
1571 return arabic_maycombine(two
);
1576 * Check whether we are dealing with a character that could be regarded as an
1577 * Arabic combining character, need to check the character before this.
1580 arabic_maycombine(two
)
1583 if (p_arshape
&& !p_tbidi
)
1584 return (two
== a_ALEF_MADDA
1585 || two
== a_ALEF_HAMZA_ABOVE
1586 || two
== a_ALEF_HAMZA_BELOW
1592 * Check if the character pointed to by "p2" is a composing character when it
1593 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which
1594 * behaves like a composing character.
1597 utf_composinglike(p1
, p2
)
1603 c2
= utf_ptr2char(p2
);
1604 if (utf_iscomposing(c2
))
1606 if (!arabic_maycombine(c2
))
1608 return arabic_combine(utf_ptr2char(p1
), c2
);
1613 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO
1614 * composing characters.
1617 utfc_ptr2char(p
, pcc
)
1619 int *pcc
; /* return: composing chars, last one is 0 */
1626 c
= utf_ptr2char(p
);
1627 len
= utf_ptr2len(p
);
1629 /* Only accept a composing char when the first char isn't illegal. */
1630 if ((len
> 1 || *p
< 0x80)
1632 && UTF_COMPOSINGLIKE(p
, p
+ len
))
1634 cc
= utf_ptr2char(p
+ len
);
1640 len
+= utf_ptr2len(p
+ len
);
1641 if (p
[len
] < 0x80 || !utf_iscomposing(cc
= utf_ptr2char(p
+ len
)))
1646 if (i
< MAX_MCO
) /* last composing char must be 0 */
1653 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO
1654 * composing characters. Use no more than p[maxlen].
1657 utfc_ptr2char_len(p
, pcc
, maxlen
)
1659 int *pcc
; /* return: composing chars, last one is 0 */
1667 c
= utf_ptr2char(p
);
1668 len
= utf_ptr2len_len(p
, maxlen
);
1669 /* Only accept a composing char when the first char isn't illegal. */
1670 if ((len
> 1 || *p
< 0x80)
1673 && UTF_COMPOSINGLIKE(p
, p
+ len
))
1675 cc
= utf_ptr2char(p
+ len
);
1681 len
+= utf_ptr2len_len(p
+ len
, maxlen
- len
);
1684 || !utf_iscomposing(cc
= utf_ptr2char(p
+ len
)))
1689 if (i
< MAX_MCO
) /* last composing char must be 0 */
1696 * Convert the character at screen position "off" to a sequence of bytes.
1697 * Includes the composing characters.
1698 * "buf" must at least have the length MB_MAXBYTES.
1699 * Returns the produced number of bytes.
1702 utfc_char2bytes(off
, buf
)
1709 len
= utf_char2bytes(ScreenLinesUC
[off
], buf
);
1710 for (i
= 0; i
< Screen_mco
; ++i
)
1712 if (ScreenLinesC
[i
][off
] == 0)
1714 len
+= utf_char2bytes(ScreenLinesC
[i
][off
], buf
+ len
);
1720 * Get the length of a UTF-8 byte sequence, not including any following
1721 * composing characters.
1723 * Returns 1 for an illegal byte sequence.
1734 len
= utf8len_tab
[*p
];
1735 for (i
= 1; i
< len
; ++i
)
1736 if ((p
[i
] & 0xc0) != 0x80)
1742 * Return length of UTF-8 character, obtained from the first byte.
1743 * "b" must be between 0 and 255!
1744 * Returns 1 for an invalid first byte value.
1750 return utf8len_tab
[b
];
1754 * Get the length of UTF-8 byte sequence "p[size]". Does not include any
1755 * following composing characters.
1757 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.).
1758 * Returns number > "size" for an incomplete byte sequence.
1759 * Never returns zero.
1762 utf_ptr2len_len(p
, size
)
1770 len
= utf8len_tab
[*p
];
1772 return 1; /* NUL, ascii or illegal lead byte */
1774 m
= size
; /* incomplete byte sequence. */
1777 for (i
= 1; i
< m
; ++i
)
1778 if ((p
[i
] & 0xc0) != 0x80)
1784 * Return the number of bytes the UTF-8 encoding of the character at "p" takes.
1785 * This includes following composing characters.
1799 if (b0
< 0x80 && p
[1] < 0x80) /* be quick for ASCII */
1802 /* Skip over first UTF-8 char, stopping at a NUL byte. */
1803 len
= utf_ptr2len(p
);
1805 /* Check for illegal byte. */
1806 if (len
== 1 && b0
>= 0x80)
1810 * Check for composing characters. We can handle only the first two, but
1811 * skip all of them (otherwise the cursor would get stuck).
1818 if (p
[len
] < 0x80 || !UTF_COMPOSINGLIKE(p
+ prevlen
, p
+ len
))
1821 /* Skip over composing char */
1825 len
+= utf_ptr2len(p
+ len
);
1830 * Return the number of bytes the UTF-8 encoding of the character at "p[size]"
1831 * takes. This includes following composing characters.
1832 * Returns 0 for an empty string.
1833 * Returns 1 for an illegal char or an incomplete byte sequence.
1836 utfc_ptr2len_len(p
, size
)
1845 if (size
< 1 || *p
== NUL
)
1847 if (p
[0] < 0x80 && (size
== 1 || p
[1] < 0x80)) /* be quick for ASCII */
1850 /* Skip over first UTF-8 char, stopping at a NUL byte. */
1851 len
= utf_ptr2len_len(p
, size
);
1853 /* Check for illegal byte and incomplete byte sequence. */
1854 if ((len
== 1 && p
[0] >= 0x80) || len
> size
)
1858 * Check for composing characters. We can handle only the first two, but
1859 * skip all of them (otherwise the cursor would get stuck).
1872 * Next character length should not go beyond size to ensure that
1873 * UTF_COMPOSINGLIKE(...) does not read beyond size.
1875 len_next_char
= utf_ptr2len_len(p
+ len
, size
- len
);
1876 if (len_next_char
> size
- len
)
1879 if (!UTF_COMPOSINGLIKE(p
+ prevlen
, p
+ len
))
1882 /* Skip over composing char */
1886 len
+= len_next_char
;
1892 * Return the number of bytes the UTF-8 encoding of character "c" takes.
1893 * This does not include composing characters.
1913 * Convert Unicode character "c" to UTF-8 string in "buf[]".
1914 * Returns the number of bytes.
1915 * This does not include composing characters.
1918 utf_char2bytes(c
, buf
)
1922 if (c
< 0x80) /* 7 bits */
1927 if (c
< 0x800) /* 11 bits */
1929 buf
[0] = 0xc0 + ((unsigned)c
>> 6);
1930 buf
[1] = 0x80 + (c
& 0x3f);
1933 if (c
< 0x10000) /* 16 bits */
1935 buf
[0] = 0xe0 + ((unsigned)c
>> 12);
1936 buf
[1] = 0x80 + (((unsigned)c
>> 6) & 0x3f);
1937 buf
[2] = 0x80 + (c
& 0x3f);
1940 if (c
< 0x200000) /* 21 bits */
1942 buf
[0] = 0xf0 + ((unsigned)c
>> 18);
1943 buf
[1] = 0x80 + (((unsigned)c
>> 12) & 0x3f);
1944 buf
[2] = 0x80 + (((unsigned)c
>> 6) & 0x3f);
1945 buf
[3] = 0x80 + (c
& 0x3f);
1948 if (c
< 0x4000000) /* 26 bits */
1950 buf
[0] = 0xf8 + ((unsigned)c
>> 24);
1951 buf
[1] = 0x80 + (((unsigned)c
>> 18) & 0x3f);
1952 buf
[2] = 0x80 + (((unsigned)c
>> 12) & 0x3f);
1953 buf
[3] = 0x80 + (((unsigned)c
>> 6) & 0x3f);
1954 buf
[4] = 0x80 + (c
& 0x3f);
1958 buf
[0] = 0xfc + ((unsigned)c
>> 30);
1959 buf
[1] = 0x80 + (((unsigned)c
>> 24) & 0x3f);
1960 buf
[2] = 0x80 + (((unsigned)c
>> 18) & 0x3f);
1961 buf
[3] = 0x80 + (((unsigned)c
>> 12) & 0x3f);
1962 buf
[4] = 0x80 + (((unsigned)c
>> 6) & 0x3f);
1963 buf
[5] = 0x80 + (c
& 0x3f);
1968 * Return TRUE if "c" is a composing UTF-8 character. This means it will be
1969 * drawn on top of the preceding character.
1970 * Based on code from Markus Kuhn.
1976 /* sorted list of non-overlapping intervals */
1977 static struct interval combining
[] =
1979 {0x0300, 0x034f}, {0x0360, 0x036f}, {0x0483, 0x0486}, {0x0488, 0x0489},
1980 {0x0591, 0x05a1}, {0x05a3, 0x05b9}, {0x05bb, 0x05bd}, {0x05bf, 0x05bf},
1981 {0x05c1, 0x05c2}, {0x05c4, 0x05c4}, {0x0610, 0x0615}, {0x064b, 0x0658},
1982 {0x0670, 0x0670}, {0x06d6, 0x06dc}, {0x06de, 0x06e4}, {0x06e7, 0x06e8},
1983 {0x06ea, 0x06ed}, {0x0711, 0x0711}, {0x0730, 0x074a}, {0x07a6, 0x07b0},
1984 {0x0901, 0x0903}, {0x093c, 0x093c}, {0x093e, 0x094d}, {0x0951, 0x0954},
1985 {0x0962, 0x0963}, {0x0981, 0x0983}, {0x09bc, 0x09bc}, {0x09be, 0x09c4},
1986 {0x09c7, 0x09c8}, {0x09cb, 0x09cd}, {0x09d7, 0x09d7}, {0x09e2, 0x09e3},
1987 {0x0a01, 0x0a03}, {0x0a3c, 0x0a3c}, {0x0a3e, 0x0a42}, {0x0a47, 0x0a48},
1988 {0x0a4b, 0x0a4d}, {0x0a70, 0x0a71}, {0x0a81, 0x0a83}, {0x0abc, 0x0abc},
1989 {0x0abe, 0x0ac5}, {0x0ac7, 0x0ac9}, {0x0acb, 0x0acd}, {0x0ae2, 0x0ae3},
1990 {0x0b01, 0x0b03}, {0x0b3c, 0x0b3c}, {0x0b3e, 0x0b43}, {0x0b47, 0x0b48},
1991 {0x0b4b, 0x0b4d}, {0x0b56, 0x0b57}, {0x0b82, 0x0b82}, {0x0bbe, 0x0bc2},
1992 {0x0bc6, 0x0bc8}, {0x0bca, 0x0bcd}, {0x0bd7, 0x0bd7}, {0x0c01, 0x0c03},
1993 {0x0c3e, 0x0c44}, {0x0c46, 0x0c48}, {0x0c4a, 0x0c4d}, {0x0c55, 0x0c56},
1994 {0x0c82, 0x0c83}, {0x0cbc, 0x0cbc}, {0x0cbe, 0x0cc4}, {0x0cc6, 0x0cc8},
1995 {0x0cca, 0x0ccd}, {0x0cd5, 0x0cd6}, {0x0d02, 0x0d03}, {0x0d3e, 0x0d43},
1996 {0x0d46, 0x0d48}, {0x0d4a, 0x0d4d}, {0x0d57, 0x0d57}, {0x0d82, 0x0d83},
1997 {0x0dca, 0x0dca}, {0x0dcf, 0x0dd4}, {0x0dd6, 0x0dd6}, {0x0dd8, 0x0ddf},
1998 {0x0df2, 0x0df3}, {0x0e31, 0x0e31}, {0x0e34, 0x0e3a}, {0x0e47, 0x0e4e},
1999 {0x0eb1, 0x0eb1}, {0x0eb4, 0x0eb9}, {0x0ebb, 0x0ebc}, {0x0ec8, 0x0ecd},
2000 {0x0f18, 0x0f19}, {0x0f35, 0x0f35}, {0x0f37, 0x0f37}, {0x0f39, 0x0f39},
2001 {0x0f3e, 0x0f3f}, {0x0f71, 0x0f84}, {0x0f86, 0x0f87}, {0x0f90, 0x0f97},
2002 {0x0f99, 0x0fbc}, {0x0fc6, 0x0fc6}, {0x102c, 0x1032}, {0x1036, 0x1039},
2003 {0x1056, 0x1059}, {0x1712, 0x1714}, {0x1732, 0x1734}, {0x1752, 0x1753},
2004 {0x1772, 0x1773}, {0x17b6, 0x17d3}, {0x17dd, 0x17dd}, {0x180b, 0x180d},
2005 {0x18a9, 0x18a9}, {0x1920, 0x192b}, {0x1930, 0x193b}, {0x20d0, 0x20ea},
2006 {0x302a, 0x302f}, {0x3099, 0x309a}, {0xfb1e, 0xfb1e}, {0xfe00, 0xfe0f},
2010 return intable(combining
, sizeof(combining
), c
);
2014 * Return TRUE for characters that can be displayed in a normal way.
2015 * Only for characters of 0x100 and above!
2021 #ifdef USE_WCHAR_FUNCTIONS
2023 * Assume the iswprint() library function works better than our own stuff.
2027 /* Sorted list of non-overlapping intervals.
2028 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */
2029 static struct interval nonprint
[] =
2031 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e},
2032 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb},
2036 return !intable(nonprint
, sizeof(nonprint
), c
);
2041 * Get class of a Unicode character.
2044 * 2 or bigger: some class of word character.
2050 /* sorted list of non-overlapping intervals */
2051 static struct clinterval
2053 unsigned short first
;
2054 unsigned short last
;
2055 unsigned short class;
2058 {0x037e, 0x037e, 1}, /* Greek question mark */
2059 {0x0387, 0x0387, 1}, /* Greek ano teleia */
2060 {0x055a, 0x055f, 1}, /* Armenian punctuation */
2061 {0x0589, 0x0589, 1}, /* Armenian full stop */
2062 {0x05be, 0x05be, 1},
2063 {0x05c0, 0x05c0, 1},
2064 {0x05c3, 0x05c3, 1},
2065 {0x05f3, 0x05f4, 1},
2066 {0x060c, 0x060c, 1},
2067 {0x061b, 0x061b, 1},
2068 {0x061f, 0x061f, 1},
2069 {0x066a, 0x066d, 1},
2070 {0x06d4, 0x06d4, 1},
2071 {0x0700, 0x070d, 1}, /* Syriac punctuation */
2072 {0x0964, 0x0965, 1},
2073 {0x0970, 0x0970, 1},
2074 {0x0df4, 0x0df4, 1},
2075 {0x0e4f, 0x0e4f, 1},
2076 {0x0e5a, 0x0e5b, 1},
2077 {0x0f04, 0x0f12, 1},
2078 {0x0f3a, 0x0f3d, 1},
2079 {0x0f85, 0x0f85, 1},
2080 {0x104a, 0x104f, 1}, /* Myanmar punctuation */
2081 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */
2082 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */
2083 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */
2084 {0x1680, 0x1680, 0},
2085 {0x169b, 0x169c, 1},
2086 {0x16eb, 0x16ed, 1},
2087 {0x1735, 0x1736, 1},
2088 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */
2089 {0x1800, 0x180a, 1}, /* Mongolian punctuation */
2090 {0x2000, 0x200b, 0}, /* spaces */
2091 {0x200c, 0x2027, 1}, /* punctuation and symbols */
2092 {0x2028, 0x2029, 0},
2093 {0x202a, 0x202e, 1}, /* punctuation and symbols */
2094 {0x202f, 0x202f, 0},
2095 {0x2030, 0x205e, 1}, /* punctuation and symbols */
2096 {0x205f, 0x205f, 0},
2097 {0x2060, 0x27ff, 1}, /* punctuation and symbols */
2098 {0x2070, 0x207f, 0x2070}, /* superscript */
2099 {0x2080, 0x2094, 0x2080}, /* subscript */
2100 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */
2101 {0x2800, 0x28ff, 0x2800}, /* braille */
2102 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */
2103 {0x29d8, 0x29db, 1},
2104 {0x29fc, 0x29fd, 1},
2105 {0x3000, 0x3000, 0}, /* ideographic space */
2106 {0x3001, 0x3020, 1}, /* ideographic punctuation */
2107 {0x3030, 0x3030, 1},
2108 {0x303d, 0x303d, 1},
2109 {0x3040, 0x309f, 0x3040}, /* Hiragana */
2110 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */
2111 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */
2112 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */
2113 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */
2114 {0xfd3e, 0xfd3f, 1},
2115 {0xfe30, 0xfe6b, 1}, /* punctuation forms */
2116 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */
2117 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */
2118 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */
2119 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */
2122 int top
= sizeof(classes
) / sizeof(struct clinterval
) - 1;
2125 /* First quick check for Latin1 characters, use 'iskeyword'. */
2128 if (c
== ' ' || c
== '\t' || c
== NUL
|| c
== 0xa0)
2129 return 0; /* blank */
2131 return 2; /* word character */
2132 return 1; /* punctuation */
2135 /* binary search in table */
2138 mid
= (bot
+ top
) / 2;
2139 if (classes
[mid
].last
< c
)
2141 else if (classes
[mid
].first
> c
)
2144 return (int)classes
[mid
].class;
2147 /* most other characters are "word" characters */
2152 * Code for Unicode case-dependent operations. Based on notes in
2153 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt
2154 * This code uses simple case folding, not full case folding.
2158 * The following table is built by foldExtract.pl < CaseFolding.txt .
2159 * It must be in numeric order, because we use binary search on it.
2160 * An entry such as {0x41,0x5a,1,32} means that UCS-4 characters in the range
2161 * from 0x41 to 0x5a inclusive, stepping by 1, are folded by adding 32.
2172 static convertStruct foldCase
[] =
2174 {0x41,0x5a,1,32}, {0xc0,0xd6,1,32}, {0xd8,0xde,1,32},
2175 {0x100,0x12e,2,1}, {0x130,0x130,-1,-199}, {0x132,0x136,2,1},
2176 {0x139,0x147,2,1}, {0x14a,0x176,2,1}, {0x178,0x178,-1,-121},
2177 {0x179,0x17d,2,1}, {0x181,0x181,-1,210}, {0x182,0x184,2,1},
2178 {0x186,0x186,-1,206}, {0x187,0x187,-1,1}, {0x189,0x18a,1,205},
2179 {0x18b,0x18b,-1,1}, {0x18e,0x18e,-1,79}, {0x18f,0x18f,-1,202},
2180 {0x190,0x190,-1,203}, {0x191,0x191,-1,1}, {0x193,0x193,-1,205},
2181 {0x194,0x194,-1,207}, {0x196,0x196,-1,211}, {0x197,0x197,-1,209},
2182 {0x198,0x198,-1,1}, {0x19c,0x19c,-1,211}, {0x19d,0x19d,-1,213},
2183 {0x19f,0x19f,-1,214}, {0x1a0,0x1a4,2,1}, {0x1a6,0x1a6,-1,218},
2184 {0x1a7,0x1a7,-1,1}, {0x1a9,0x1a9,-1,218}, {0x1ac,0x1ac,-1,1},
2185 {0x1ae,0x1ae,-1,218}, {0x1af,0x1af,-1,1}, {0x1b1,0x1b2,1,217},
2186 {0x1b3,0x1b5,2,1}, {0x1b7,0x1b7,-1,219}, {0x1b8,0x1bc,4,1},
2187 {0x1c4,0x1c4,-1,2}, {0x1c5,0x1c5,-1,1}, {0x1c7,0x1c7,-1,2},
2188 {0x1c8,0x1c8,-1,1}, {0x1ca,0x1ca,-1,2}, {0x1cb,0x1db,2,1},
2189 {0x1de,0x1ee,2,1}, {0x1f1,0x1f1,-1,2}, {0x1f2,0x1f4,2,1},
2190 {0x1f6,0x1f6,-1,-97}, {0x1f7,0x1f7,-1,-56}, {0x1f8,0x21e,2,1},
2191 {0x220,0x220,-1,-130}, {0x222,0x232,2,1}, {0x386,0x386,-1,38},
2192 {0x388,0x38a,1,37}, {0x38c,0x38c,-1,64}, {0x38e,0x38f,1,63},
2193 {0x391,0x3a1,1,32}, {0x3a3,0x3ab,1,32}, {0x3d8,0x3ee,2,1},
2194 {0x3f4,0x3f4,-1,-60}, {0x3f7,0x3f7,-1,1}, {0x3f9,0x3f9,-1,-7},
2195 {0x3fa,0x3fa,-1,1}, {0x400,0x40f,1,80}, {0x410,0x42f,1,32},
2196 {0x460,0x480,2,1}, {0x48a,0x4be,2,1}, {0x4c1,0x4cd,2,1},
2197 {0x4d0,0x4f4,2,1}, {0x4f8,0x500,8,1}, {0x502,0x50e,2,1},
2198 {0x531,0x556,1,48}, {0x1e00,0x1e94,2,1}, {0x1ea0,0x1ef8,2,1},
2199 {0x1f08,0x1f0f,1,-8}, {0x1f18,0x1f1d,1,-8}, {0x1f28,0x1f2f,1,-8},
2200 {0x1f38,0x1f3f,1,-8}, {0x1f48,0x1f4d,1,-8}, {0x1f59,0x1f5f,2,-8},
2201 {0x1f68,0x1f6f,1,-8}, {0x1f88,0x1f8f,1,-8}, {0x1f98,0x1f9f,1,-8},
2202 {0x1fa8,0x1faf,1,-8}, {0x1fb8,0x1fb9,1,-8}, {0x1fba,0x1fbb,1,-74},
2203 {0x1fbc,0x1fbc,-1,-9}, {0x1fc8,0x1fcb,1,-86}, {0x1fcc,0x1fcc,-1,-9},
2204 {0x1fd8,0x1fd9,1,-8}, {0x1fda,0x1fdb,1,-100}, {0x1fe8,0x1fe9,1,-8},
2205 {0x1fea,0x1feb,1,-112}, {0x1fec,0x1fec,-1,-7}, {0x1ff8,0x1ff9,1,-128},
2206 {0x1ffa,0x1ffb,1,-126}, {0x1ffc,0x1ffc,-1,-9}, {0x2126,0x2126,-1,-7517},
2207 {0x212a,0x212a,-1,-8383}, {0x212b,0x212b,-1,-8262},
2208 {0x2160,0x216f,1,16}, {0x24b6,0x24cf,1,26}, {0xff21,0xff3a,1,32},
2209 {0x10400,0x10427,1,40}
2212 static int utf_convert(int a
, convertStruct table
[], int tableSize
);
2215 * Generic conversion function for case operations.
2216 * Return the converted equivalent of "a", which is a UCS-4 character. Use
2217 * the given conversion "table". Uses binary search on "table".
2220 utf_convert(a
, table
, tableSize
)
2222 convertStruct table
[];
2225 int start
, mid
, end
; /* indices into table */
2228 end
= tableSize
/ sizeof(convertStruct
);
2231 /* need to search further */
2232 mid
= (end
+ start
) /2;
2233 if (table
[mid
].rangeEnd
< a
)
2238 if (table
[start
].rangeStart
<= a
&& a
<= table
[start
].rangeEnd
2239 && (a
- table
[start
].rangeStart
) % table
[start
].step
== 0)
2240 return (a
+ table
[start
].offset
);
2246 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses
2247 * simple case folding.
2253 return utf_convert(a
, foldCase
, sizeof(foldCase
));
2257 * The following tables are built by upperLowerExtract.pl < UnicodeData.txt .
2258 * They must be in numeric order, because we use binary search on them.
2259 * An entry such as {0x41,0x5a,1,32} means that UCS-4 characters in the range
2260 * from 0x41 to 0x5a inclusive, stepping by 1, are switched to lower (for
2261 * example) by adding 32.
2263 static convertStruct toLower
[] =
2265 {0x41,0x5a,1,32}, {0xc0,0xd6,1,32}, {0xd8,0xde,1,32},
2266 {0x100,0x12e,2,1}, {0x130,0x130,-1,-199}, {0x132,0x136,2,1},
2267 {0x139,0x147,2,1}, {0x14a,0x176,2,1}, {0x178,0x178,-1,-121},
2268 {0x179,0x17d,2,1}, {0x181,0x181,-1,210}, {0x182,0x184,2,1},
2269 {0x186,0x186,-1,206}, {0x187,0x187,-1,1}, {0x189,0x18a,1,205},
2270 {0x18b,0x18b,-1,1}, {0x18e,0x18e,-1,79}, {0x18f,0x18f,-1,202},
2271 {0x190,0x190,-1,203}, {0x191,0x191,-1,1}, {0x193,0x193,-1,205},
2272 {0x194,0x194,-1,207}, {0x196,0x196,-1,211}, {0x197,0x197,-1,209},
2273 {0x198,0x198,-1,1}, {0x19c,0x19c,-1,211}, {0x19d,0x19d,-1,213},
2274 {0x19f,0x19f,-1,214}, {0x1a0,0x1a4,2,1}, {0x1a6,0x1a6,-1,218},
2275 {0x1a7,0x1a7,-1,1}, {0x1a9,0x1a9,-1,218}, {0x1ac,0x1ac,-1,1},
2276 {0x1ae,0x1ae,-1,218}, {0x1af,0x1af,-1,1}, {0x1b1,0x1b2,1,217},
2277 {0x1b3,0x1b5,2,1}, {0x1b7,0x1b7,-1,219}, {0x1b8,0x1bc,4,1},
2278 {0x1c4,0x1ca,3,2}, {0x1cd,0x1db,2,1}, {0x1de,0x1ee,2,1},
2279 {0x1f1,0x1f1,-1,2}, {0x1f4,0x1f4,-1,1}, {0x1f6,0x1f6,-1,-97},
2280 {0x1f7,0x1f7,-1,-56}, {0x1f8,0x21e,2,1}, {0x220,0x220,-1,-130},
2281 {0x222,0x232,2,1}, {0x386,0x386,-1,38}, {0x388,0x38a,1,37},
2282 {0x38c,0x38c,-1,64}, {0x38e,0x38f,1,63}, {0x391,0x3a1,1,32},
2283 {0x3a3,0x3ab,1,32}, {0x3d8,0x3ee,2,1}, {0x3f4,0x3f4,-1,-60},
2284 {0x3f7,0x3f7,-1,1}, {0x3f9,0x3f9,-1,-7}, {0x3fa,0x3fa,-1,1},
2285 {0x400,0x40f,1,80}, {0x410,0x42f,1,32}, {0x460,0x480,2,1},
2286 {0x48a,0x4be,2,1}, {0x4c1,0x4cd,2,1}, {0x4d0,0x4f4,2,1},
2287 {0x4f8,0x500,8,1}, {0x502,0x50e,2,1}, {0x531,0x556,1,48},
2288 {0x1e00,0x1e94,2,1}, {0x1ea0,0x1ef8,2,1}, {0x1f08,0x1f0f,1,-8},
2289 {0x1f18,0x1f1d,1,-8}, {0x1f28,0x1f2f,1,-8}, {0x1f38,0x1f3f,1,-8},
2290 {0x1f48,0x1f4d,1,-8}, {0x1f59,0x1f5f,2,-8}, {0x1f68,0x1f6f,1,-8},
2291 {0x1fb8,0x1fb9,1,-8}, {0x1fba,0x1fbb,1,-74}, {0x1fc8,0x1fcb,1,-86},
2292 {0x1fd8,0x1fd9,1,-8}, {0x1fda,0x1fdb,1,-100}, {0x1fe8,0x1fe9,1,-8},
2293 {0x1fea,0x1feb,1,-112}, {0x1fec,0x1fec,-1,-7}, {0x1ff8,0x1ff9,1,-128},
2294 {0x1ffa,0x1ffb,1,-126}, {0x2126,0x2126,-1,-7517}, {0x212a,0x212a,-1,-8383},
2295 {0x212b,0x212b,-1,-8262}, {0xff21,0xff3a,1,32}, {0x10400,0x10427,1,40}
2298 static convertStruct toUpper
[] =
2300 {0x61,0x7a,1,-32}, {0xb5,0xb5,-1,743}, {0xe0,0xf6,1,-32},
2301 {0xf8,0xfe,1,-32}, {0xff,0xff,-1,121}, {0x101,0x12f,2,-1},
2302 {0x131,0x131,-1,-232}, {0x133,0x137,2,-1}, {0x13a,0x148,2,-1},
2303 {0x14b,0x177,2,-1}, {0x17a,0x17e,2,-1}, {0x17f,0x17f,-1,-300},
2304 {0x183,0x185,2,-1}, {0x188,0x18c,4,-1}, {0x192,0x192,-1,-1},
2305 {0x195,0x195,-1,97}, {0x199,0x199,-1,-1}, {0x19e,0x19e,-1,130},
2306 {0x1a1,0x1a5,2,-1}, {0x1a8,0x1ad,5,-1}, {0x1b0,0x1b4,4,-1},
2307 {0x1b6,0x1b9,3,-1}, {0x1bd,0x1bd,-1,-1}, {0x1bf,0x1bf,-1,56},
2308 {0x1c5,0x1c6,1,-1}, {0x1c8,0x1c9,1,-1}, {0x1cb,0x1cc,1,-1},
2309 {0x1ce,0x1dc,2,-1}, {0x1dd,0x1dd,-1,-79}, {0x1df,0x1ef,2,-1},
2310 {0x1f2,0x1f3,1,-1}, {0x1f5,0x1f9,4,-1}, {0x1fb,0x21f,2,-1},
2311 {0x223,0x233,2,-1}, {0x253,0x253,-1,-210}, {0x254,0x254,-1,-206},
2312 {0x256,0x257,1,-205}, {0x259,0x259,-1,-202}, {0x25b,0x25b,-1,-203},
2313 {0x260,0x260,-1,-205}, {0x263,0x263,-1,-207}, {0x268,0x268,-1,-209},
2314 {0x269,0x26f,6,-211}, {0x272,0x272,-1,-213}, {0x275,0x275,-1,-214},
2315 {0x280,0x283,3,-218}, {0x288,0x288,-1,-218}, {0x28a,0x28b,1,-217},
2316 {0x292,0x292,-1,-219}, {0x3ac,0x3ac,-1,-38}, {0x3ad,0x3af,1,-37},
2317 {0x3b1,0x3c1,1,-32}, {0x3c2,0x3c2,-1,-31}, {0x3c3,0x3cb,1,-32},
2318 {0x3cc,0x3cc,-1,-64}, {0x3cd,0x3ce,1,-63}, {0x3d0,0x3d0,-1,-62},
2319 {0x3d1,0x3d1,-1,-57}, {0x3d5,0x3d5,-1,-47}, {0x3d6,0x3d6,-1,-54},
2320 {0x3d9,0x3ef,2,-1}, {0x3f0,0x3f0,-1,-86}, {0x3f1,0x3f1,-1,-80},
2321 {0x3f2,0x3f2,-1,7}, {0x3f5,0x3f5,-1,-96}, {0x3f8,0x3fb,3,-1},
2322 {0x430,0x44f,1,-32}, {0x450,0x45f,1,-80}, {0x461,0x481,2,-1},
2323 {0x48b,0x4bf,2,-1}, {0x4c2,0x4ce,2,-1}, {0x4d1,0x4f5,2,-1},
2324 {0x4f9,0x501,8,-1}, {0x503,0x50f,2,-1}, {0x561,0x586,1,-48},
2325 {0x1e01,0x1e95,2,-1}, {0x1e9b,0x1e9b,-1,-59}, {0x1ea1,0x1ef9,2,-1},
2326 {0x1f00,0x1f07,1,8}, {0x1f10,0x1f15,1,8}, {0x1f20,0x1f27,1,8},
2327 {0x1f30,0x1f37,1,8}, {0x1f40,0x1f45,1,8}, {0x1f51,0x1f57,2,8},
2328 {0x1f60,0x1f67,1,8}, {0x1f70,0x1f71,1,74}, {0x1f72,0x1f75,1,86},
2329 {0x1f76,0x1f77,1,100}, {0x1f78,0x1f79,1,128}, {0x1f7a,0x1f7b,1,112},
2330 {0x1f7c,0x1f7d,1,126}, {0x1f80,0x1f87,1,8}, {0x1f90,0x1f97,1,8},
2331 {0x1fa0,0x1fa7,1,8}, {0x1fb0,0x1fb1,1,8}, {0x1fb3,0x1fb3,-1,9},
2332 {0x1fbe,0x1fbe,-1,-7205}, {0x1fc3,0x1fc3,-1,9}, {0x1fd0,0x1fd1,1,8},
2333 {0x1fe0,0x1fe1,1,8}, {0x1fe5,0x1fe5,-1,7}, {0x1ff3,0x1ff3,-1,9},
2334 {0xff41,0xff5a,1,-32}, {0x10428,0x1044f,1,-40}
2338 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use
2339 * simple case folding.
2345 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */
2346 if (a
< 128 && (cmp_flags
& CMP_KEEPASCII
))
2347 return TOUPPER_ASC(a
);
2349 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__)
2350 /* If towupper() is available and handles Unicode, use it. */
2351 if (!(cmp_flags
& CMP_INTERNAL
))
2355 /* For characters below 128 use locale sensitive toupper(). */
2357 return TOUPPER_LOC(a
);
2359 /* For any other characters use the above mapping table. */
2360 return utf_convert(a
, toUpper
, sizeof(toUpper
));
2367 return (utf_toupper(a
) != a
);
2371 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use
2372 * simple case folding.
2378 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */
2379 if (a
< 128 && (cmp_flags
& CMP_KEEPASCII
))
2380 return TOLOWER_ASC(a
);
2382 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__)
2383 /* If towlower() is available and handles Unicode, use it. */
2384 if (!(cmp_flags
& CMP_INTERNAL
))
2388 /* For characters below 128 use locale sensitive tolower(). */
2390 return TOLOWER_LOC(a
);
2392 /* For any other characters use the above mapping table. */
2393 return utf_convert(a
, toLower
, sizeof(toLower
));
2400 return (utf_tolower(a
) != a
);
2404 * Version of strnicmp() that handles multi-byte characters.
2405 * Needed for Big5, Sjift-JIS and UTF-8 encoding. Other DBCS encodings can
2406 * probably use strnicmp(), because there are no ASCII characters in the
2408 * Returns zero if s1 and s2 are equal (ignoring case), the difference between
2409 * two characters otherwise.
2412 mb_strnicmp(s1
, s2
, nn
)
2418 int incomplete
= FALSE
;
2421 for (i
= 0; i
< n
; i
+= l
)
2423 if (s1
[i
] == NUL
&& s2
[i
] == NUL
) /* both strings end */
2427 l
= utf_byte2len(s1
[i
]);
2430 l
= n
- i
; /* incomplete character */
2433 /* Check directly first, it's faster. */
2434 for (j
= 0; j
< l
; ++j
)
2436 if (s1
[i
+ j
] != s2
[i
+ j
])
2439 /* Both stings have the same bytes but are incomplete or
2440 * have illegal bytes, accept them as equal. */
2445 /* If one of the two characters is incomplete return -1. */
2446 if (incomplete
|| i
+ utf_byte2len(s2
[i
]) > n
)
2448 cdiff
= utf_fold(utf_ptr2char(s1
+ i
))
2449 - utf_fold(utf_ptr2char(s2
+ i
));
2456 l
= (*mb_ptr2len
)(s1
+ i
);
2459 /* Single byte: first check normally, then with ignore case. */
2462 cdiff
= MB_TOLOWER(s1
[i
]) - MB_TOLOWER(s2
[i
]);
2469 /* For non-Unicode multi-byte don't ignore case. */
2472 cdiff
= STRNCMP(s1
+ i
, s2
+ i
, l
);
2482 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what
2483 * 'encoding' has been set to.
2494 /* Get the byte length of the char under the cursor, including composing
2496 line
= ml_get_cursor();
2497 len
= utfc_ptr2len(line
);
2505 for (i
= 0; i
< len
; ++i
)
2509 /* start of (composing) character, get its length */
2512 STRCPY(IObuff
+ rlen
, "+ ");
2515 clen
= utf_ptr2len(line
+ i
);
2517 sprintf((char *)IObuff
+ rlen
, "%02x ", line
[i
]);
2519 rlen
+= (int)STRLEN(IObuff
+ rlen
);
2520 if (rlen
> IOSIZE
- 20)
2528 * mb_head_off() function pointer.
2529 * Return offset from "p" to the first byte of the character it points into.
2530 * If "p" points to the NUL at the end of the string return 0.
2531 * Returns 0 when already at the first byte of a character.
2534 latin_head_off(base
, p
)
2535 char_u
*base UNUSED
;
2542 dbcs_head_off(base
, p
)
2548 /* It can't be a trailing byte when not using DBCS, at the start of the
2549 * string or the previous byte can't start a double-byte. */
2550 if (p
<= base
|| MB_BYTE2LEN(p
[-1]) == 1 || *p
== NUL
)
2553 /* This is slow: need to start at the base and go forward until the
2554 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */
2557 q
+= dbcs_ptr2len(q
);
2558 return (q
== p
) ? 0 : 1;
2562 * Special version of dbcs_head_off() that works for ScreenLines[], where
2563 * single-width DBCS_JPNU characters are stored separately.
2566 dbcs_screen_head_off(base
, p
)
2572 /* It can't be a trailing byte when not using DBCS, at the start of the
2573 * string or the previous byte can't start a double-byte.
2574 * For euc-jp an 0x8e byte in the previous cell always means we have a
2575 * lead byte in the current cell. */
2577 || (enc_dbcs
== DBCS_JPNU
&& p
[-1] == 0x8e)
2578 || MB_BYTE2LEN(p
[-1]) == 1
2582 /* This is slow: need to start at the base and go forward until the
2583 * byte we are looking for. Return 1 when we went past it, 0 otherwise.
2584 * For DBCS_JPNU look out for 0x8e, which means the second byte is not
2585 * stored as the next byte. */
2589 if (enc_dbcs
== DBCS_JPNU
&& *q
== 0x8e)
2592 q
+= dbcs_ptr2len(q
);
2594 return (q
== p
) ? 0 : 1;
2598 utf_head_off(base
, p
)
2610 if (*p
< 0x80) /* be quick for ASCII */
2613 /* Skip backwards over trailing bytes: 10xx.xxxx
2614 * Skip backwards again if on a composing char. */
2617 /* Move s to the last byte of this char. */
2618 for (s
= q
; (s
[1] & 0xc0) == 0x80; ++s
)
2620 /* Move q to the first byte of this char. */
2621 while (q
> base
&& (*q
& 0xc0) == 0x80)
2623 /* Check for illegal sequence. Do allow an illegal byte after where we
2625 len
= utf8len_tab
[*q
];
2626 if (len
!= (int)(s
- q
+ 1) && len
!= (int)(p
- q
+ 1))
2632 c
= utf_ptr2char(q
);
2633 if (utf_iscomposing(c
))
2637 if (arabic_maycombine(c
))
2639 /* Advance to get a sneak-peak at the next char */
2642 /* Move j to the first byte of this char. */
2643 while (j
> base
&& (*j
& 0xc0) == 0x80)
2645 if (arabic_combine(utf_ptr2char(j
), c
))
2652 return (int)(p
- q
);
2656 * Copy a character from "*fp" to "*tp" and advance the pointers.
2659 mb_copy_char(fp
, tp
)
2663 int l
= (*mb_ptr2len
)(*fp
);
2665 mch_memmove(*tp
, *fp
, (size_t)l
);
2671 * Return the offset from "p" to the first byte of a character. When "p" is
2672 * at the start of a character 0 is returned, otherwise the offset to the next
2673 * character. Can start anywhere in a stream of bytes.
2676 mb_off_next(base
, p
)
2685 if (*p
< 0x80) /* be quick for ASCII */
2688 /* Find the next character that isn't 10xx.xxxx */
2689 for (i
= 0; (p
[i
] & 0xc0) == 0x80; ++i
)
2693 /* Check for illegal sequence. */
2694 for (j
= 0; p
- j
> base
; ++j
)
2695 if ((p
[-j
] & 0xc0) != 0x80)
2697 if (utf8len_tab
[p
[-j
]] != i
+ j
)
2703 /* Only need to check if we're on a trail byte, it doesn't matter if we
2704 * want the offset to the next or current character. */
2705 return (*mb_head_off
)(base
, p
);
2709 * Return the offset from "p" to the last byte of the character it points
2710 * into. Can start anywhere in a stream of bytes.
2713 mb_tail_off(base
, p
)
2725 /* Find the last character that is 10xx.xxxx */
2726 for (i
= 0; (p
[i
+ 1] & 0xc0) == 0x80; ++i
)
2728 /* Check for illegal sequence. */
2729 for (j
= 0; p
- j
> base
; ++j
)
2730 if ((p
[-j
] & 0xc0) != 0x80)
2732 if (utf8len_tab
[p
[-j
]] != i
+ j
+ 1)
2737 /* It can't be the first byte if a double-byte when not using DBCS, at the
2738 * end of the string or the byte can't start a double-byte. */
2739 if (enc_dbcs
== 0 || p
[1] == NUL
|| MB_BYTE2LEN(*p
) == 1)
2742 /* Return 1 when on the lead byte, 0 when on the tail byte. */
2743 return 1 - dbcs_head_off(base
, p
);
2747 * Find the next illegal byte sequence.
2752 pos_T pos
= curwin
->w_cursor
;
2756 char_u
*tofree
= NULL
;
2758 vimconv
.vc_type
= CONV_NONE
;
2759 if (enc_utf8
&& (enc_canon_props(curbuf
->b_p_fenc
) & ENC_8BIT
))
2761 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file,
2762 * possibly a utf-8 file with illegal bytes. Setup for conversion
2763 * from utf-8 to 'fileencoding'. */
2764 convert_setup(&vimconv
, p_enc
, curbuf
->b_p_fenc
);
2767 #ifdef FEAT_VIRTUALEDIT
2768 curwin
->w_cursor
.coladd
= 0;
2772 p
= ml_get_cursor();
2773 if (vimconv
.vc_type
!= CONV_NONE
)
2776 tofree
= string_convert(&vimconv
, p
, NULL
);
2784 /* Illegal means that there are not enough trail bytes (checked by
2785 * utf_ptr2len()) or too many of them (overlong sequence). */
2786 len
= utf_ptr2len(p
);
2787 if (*p
>= 0x80 && (len
== 1
2788 || utf_char2len(utf_ptr2char(p
)) != len
))
2790 if (vimconv
.vc_type
== CONV_NONE
)
2791 curwin
->w_cursor
.col
+= (colnr_T
)(p
- ml_get_cursor());
2796 len
= (int)(p
- tofree
);
2797 for (p
= ml_get_cursor(); *p
!= NUL
&& len
-- > 0; p
+= l
)
2800 curwin
->w_cursor
.col
+= l
;
2807 if (curwin
->w_cursor
.lnum
== curbuf
->b_ml
.ml_line_count
)
2809 ++curwin
->w_cursor
.lnum
;
2810 curwin
->w_cursor
.col
= 0;
2813 /* didn't find it: don't move and beep */
2814 curwin
->w_cursor
= pos
;
2819 convert_setup(&vimconv
, NULL
, NULL
);
2822 #if defined(HAVE_GTK2) || defined(PROTO)
2824 * Return TRUE if string "s" is a valid utf-8 string.
2825 * When "end" is NULL stop at the first NUL.
2826 * When "end" is positive stop there.
2829 utf_valid_string(s
, end
)
2836 while (end
== NULL
? *p
!= NUL
: p
< end
)
2838 l
= utf8len_tab_zero
[*p
];
2840 return FALSE
; /* invalid lead byte */
2841 if (end
!= NULL
&& p
+ l
> end
)
2842 return FALSE
; /* incomplete byte sequence */
2845 if ((*p
++ & 0xc0) != 0x80)
2846 return FALSE
; /* invalid trail byte */
2852 #if defined(FEAT_GUI) || defined(PROTO)
2854 * Special version of mb_tail_off() for use in ScreenLines[].
2857 dbcs_screen_tail_off(base
, p
)
2861 /* It can't be the first byte if a double-byte when not using DBCS, at the
2862 * end of the string or the byte can't start a double-byte.
2863 * For euc-jp an 0x8e byte always means we have a lead byte in the current
2865 if (*p
== NUL
|| p
[1] == NUL
2866 || (enc_dbcs
== DBCS_JPNU
&& *p
== 0x8e)
2867 || MB_BYTE2LEN(*p
) == 1)
2870 /* Return 1 when on the lead byte, 0 when on the tail byte. */
2871 return 1 - dbcs_screen_head_off(base
, p
);
2876 * If the cursor moves on an trail byte, set the cursor on the lead byte.
2877 * Thus it moves left if necessary.
2878 * Return TRUE when the cursor was adjusted.
2883 mb_adjustpos(&curwin
->w_cursor
);
2887 * Adjust position "*lp" to point to the first byte of a multi-byte character.
2888 * If it points to a tail byte it's moved backwards to the head byte.
2897 #ifdef FEAT_VIRTUALEDIT
2902 p
= ml_get(lp
->lnum
);
2903 lp
->col
-= (*mb_head_off
)(p
, p
+ lp
->col
);
2904 #ifdef FEAT_VIRTUALEDIT
2905 /* Reset "coladd" when the cursor would be on the right half of a
2906 * double-wide character. */
2908 && p
[lp
->col
] != TAB
2909 && vim_isprintc((*mb_ptr2char
)(p
+ lp
->col
))
2910 && ptr2cells(p
+ lp
->col
) > 1)
2917 * Return a pointer to the character before "*p", if there is one.
2921 char_u
*line
; /* start of the string */
2925 mb_ptr_back(line
, p
);
2930 * Return the character length of "str". Each multi-byte character (with
2931 * following composing characters) counts as one.
2943 for (count
= 0; *p
!= NUL
; count
++)
2944 p
+= (*mb_ptr2len
)(p
);
2949 #if defined(FEAT_SPELL) || defined(PROTO)
2951 * Like mb_charlen() but for a string with specified length.
2954 mb_charlen_len(str
, len
)
2961 for (count
= 0; *p
!= NUL
&& p
< str
+ len
; count
++)
2962 p
+= (*mb_ptr2len
)(p
);
2969 * Try to un-escape a multi-byte character.
2970 * Used for the "to" and "from" part of a mapping.
2971 * Return the un-escaped string if it is a multi-byte character, and advance
2972 * "pp" to just after the bytes that formed it.
2973 * Return NULL if no multi-byte char was found.
2979 static char_u buf
[MB_MAXBYTES
+ 1];
2983 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI
2984 * KS_EXTRA KE_CSI to CSI. */
2985 for (n
= 0; str
[n
] != NUL
&& m
<= MB_MAXBYTES
; ++n
)
2987 if (str
[n
] == K_SPECIAL
2988 && str
[n
+ 1] == KS_SPECIAL
2989 && str
[n
+ 2] == KE_FILLER
)
2991 buf
[m
++] = K_SPECIAL
;
2994 else if ((str
[n
] == K_SPECIAL
2999 && str
[n
+ 1] == KS_EXTRA
3000 && str
[n
+ 2] == (int)KE_CSI
)
3005 else if (str
[n
] == K_SPECIAL
3010 break; /* a special key can't be a multibyte char */
3015 /* Return a multi-byte character if it's found. An illegal sequence
3016 * will result in a 1 here. */
3017 if ((*mb_ptr2len
)(buf
) > 1)
3027 * Return TRUE if the character at "row"/"col" on the screen is the left side
3028 * of a double-width character.
3029 * Caller must make sure "row" and "col" are not invalid!
3032 mb_lefthalve(row
, col
)
3036 #ifdef FEAT_HANGULIN
3037 if (composing_hangul
)
3040 return (*mb_off2cells
)(LineOffset
[row
] + col
,
3041 LineOffset
[row
] + screen_Columns
) > 1;
3045 * Correct a position on the screen, if it's the right half of a double-wide
3046 * char move it to the left half. Returns the corrected column.
3049 mb_fix_col(col
, row
)
3053 col
= check_col(col
);
3054 row
= check_row(row
);
3055 if (has_mbyte
&& ScreenLines
!= NULL
&& col
> 0
3057 && ScreenLines
[LineOffset
[row
] + col
] != NUL
3058 && dbcs_screen_head_off(ScreenLines
+ LineOffset
[row
],
3059 ScreenLines
+ LineOffset
[row
] + col
))
3060 || (enc_utf8
&& ScreenLines
[LineOffset
[row
] + col
] == 0)))
3066 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO)
3067 static int enc_alias_search
__ARGS((char_u
*name
));
3070 * Skip the Vim specific head of a 'encoding' name.
3076 if (STRNCMP(p
, "2byte-", 6) == 0)
3078 if (STRNCMP(p
, "8bit-", 5) == 0)
3084 * Find the canonical name for encoding "enc".
3085 * When the name isn't recognized, returns "enc" itself, but with all lower
3086 * case characters and '_' replaced with '-'.
3087 * Returns an allocated string. NULL for out-of-memory.
3098 if (STRCMP(enc
, "default") == 0)
3100 /* Use the default encoding as it's found by set_init_1(). */
3101 r
= get_encoding_default();
3103 r
= (char_u
*)"latin1";
3104 return vim_strsave(r
);
3108 /* copy "enc" to allocated memory, with room for two '-' */
3109 r
= alloc((unsigned)(STRLEN(enc
) + 3));
3112 /* Make it all lower case and replace '_' with '-'. */
3114 for (s
= enc
; *s
!= NUL
; ++s
)
3119 *p
++ = TOLOWER_ASC(*s
);
3123 /* Skip "2byte-" and "8bit-". */
3126 /* Change "microsoft-cp" to "cp". Used in some spell files. */
3127 if (STRNCMP(p
, "microsoft-cp", 12) == 0)
3130 /* "iso8859" -> "iso-8859" */
3131 if (STRNCMP(p
, "iso8859", 7) == 0)
3133 STRMOVE(p
+ 4, p
+ 3);
3137 /* "iso-8859n" -> "iso-8859-n" */
3138 if (STRNCMP(p
, "iso-8859", 8) == 0 && p
[8] != '-')
3140 STRMOVE(p
+ 9, p
+ 8);
3144 /* "latin-N" -> "latinN" */
3145 if (STRNCMP(p
, "latin-", 6) == 0)
3146 STRMOVE(p
+ 5, p
+ 6);
3148 if (enc_canon_search(p
) >= 0)
3150 /* canonical name can be used unmodified */
3154 else if ((i
= enc_alias_search(p
)) >= 0)
3156 /* alias recognized, get canonical name */
3158 r
= vim_strsave((char_u
*)enc_canon_table
[i
].name
);
3165 * Search for an encoding alias of "name".
3166 * Returns -1 when not found.
3169 enc_alias_search(name
)
3174 for (i
= 0; enc_alias_table
[i
].name
!= NULL
; ++i
)
3175 if (STRCMP(name
, enc_alias_table
[i
].name
) == 0)
3176 return enc_alias_table
[i
].canon
;
3181 #if defined(FEAT_MBYTE) || defined(PROTO)
3183 #ifdef HAVE_LANGINFO_H
3184 # include <langinfo.h>
3188 * Get the canonicalized encoding of the current locale.
3189 * Returns an allocated string when successful, NULL when not.
3201 long acp
= GetACP();
3204 STRCPY(buf
, "ucs-2le");
3205 else if (acp
== 1252) /* cp1252 is used as latin1 */
3206 STRCPY(buf
, "latin1");
3208 sprintf(buf
, "cp%ld", acp
);
3210 # ifdef HAVE_NL_LANGINFO_CODESET
3211 if ((s
= nl_langinfo(CODESET
)) == NULL
|| *s
== NUL
)
3213 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE)
3214 if ((s
= setlocale(LC_CTYPE
, NULL
)) == NULL
|| *s
== NUL
)
3216 if ((s
= getenv("LC_ALL")) == NULL
|| *s
== NUL
)
3217 if ((s
= getenv("LC_CTYPE")) == NULL
|| *s
== NUL
)
3220 if (s
== NULL
|| *s
== NUL
)
3223 /* The most generic locale format is:
3224 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]]
3225 * If there is a '.' remove the part before it.
3226 * if there is something after the codeset, remove it.
3227 * Make the name lowercase and replace '_' with '-'.
3228 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn",
3229 * "ko_KR.EUC" == "euc-kr"
3231 if ((p
= (char *)vim_strchr((char_u
*)s
, '.')) != NULL
)
3233 if (p
> s
+ 2 && STRNICMP(p
+ 1, "EUC", 3) == 0
3234 && !isalnum((int)p
[4]) && p
[4] != '-' && p
[-3] == '_')
3236 /* copy "XY.EUC" to "euc-XY" to buf[10] */
3237 STRCPY(buf
+ 10, "euc-");
3246 for (i
= 0; s
[i
] != NUL
&& i
< (int)sizeof(buf
) - 1; ++i
)
3248 if (s
[i
] == '_' || s
[i
] == '-')
3250 else if (isalnum((int)s
[i
]))
3251 buf
[i
] = TOLOWER_ASC(s
[i
]);
3258 return enc_canonize((char_u
*)buf
);
3261 #if defined(WIN3264) || defined(PROTO)
3263 * Convert an encoding name to an MS-Windows codepage.
3264 * Returns zero if no codepage can be figured out.
3267 encname2codepage(name
)
3274 if (STRNCMP(p
, "8bit-", 5) == 0)
3276 else if (STRNCMP(p_enc
, "2byte-", 6) == 0)
3279 if (p
[0] == 'c' && p
[1] == 'p')
3281 else if ((idx
= enc_canon_search(p
)) >= 0)
3282 cp
= enc_canon_table
[idx
].codepage
;
3285 if (IsValidCodePage(cp
))
3291 # if defined(USE_ICONV) || defined(PROTO)
3293 static char_u
*iconv_string
__ARGS((vimconv_T
*vcp
, char_u
*str
, int slen
, int *unconvlenp
, int *resultlenp
));
3296 * Call iconv_open() with a check if iconv() works properly (there are broken
3298 * Returns (void *)-1 if failed.
3299 * (should return iconv_t, but that causes problems with prototypes).
3302 my_iconv_open(to
, from
)
3307 #define ICONV_TESTLEN 400
3308 char_u tobuf
[ICONV_TESTLEN
];
3311 static int iconv_ok
= -1;
3313 if (iconv_ok
== FALSE
)
3314 return (void *)-1; /* detected a broken iconv() previously */
3316 #ifdef DYNAMIC_ICONV
3317 /* Check if the iconv.dll can be found. */
3318 if (!iconv_enabled(TRUE
))
3322 fd
= iconv_open((char *)enc_skip(to
), (char *)enc_skip(from
));
3324 if (fd
!= (iconv_t
)-1 && iconv_ok
== -1)
3327 * Do a dummy iconv() call to check if it actually works. There is a
3328 * version of iconv() on Linux that is broken. We can't ignore it,
3329 * because it's wide-spread. The symptoms are that after outputting
3330 * the initial shift state the "to" pointer is NULL and conversion
3331 * stops for no apparent reason after about 8160 characters.
3334 tolen
= ICONV_TESTLEN
;
3335 (void)iconv(fd
, NULL
, NULL
, &p
, &tolen
);
3350 * Convert the string "str[slen]" with iconv().
3351 * If "unconvlenp" is not NULL handle the string ending in an incomplete
3352 * sequence and set "*unconvlenp" to the length of it.
3353 * Returns the converted string in allocated memory. NULL for an error.
3354 * If resultlenp is not NULL, sets it to the result length in bytes.
3357 iconv_string(vcp
, str
, slen
, unconvlenp
, resultlenp
)
3370 char_u
*result
= NULL
;
3378 if (len
== 0 || ICONV_ERRNO
== ICONV_E2BIG
)
3380 /* Allocate enough room for most conversions. When re-allocating
3381 * increase the buffer size. */
3382 len
= len
+ fromlen
* 2 + 40;
3383 p
= alloc((unsigned)len
);
3384 if (p
!= NULL
&& done
> 0)
3385 mch_memmove(p
, result
, done
);
3388 if (result
== NULL
) /* out of memory */
3392 to
= (char *)result
+ done
;
3393 tolen
= len
- done
- 2;
3394 /* Avoid a warning for systems with a wrong iconv() prototype by
3395 * casting the second argument to void *. */
3396 if (iconv(vcp
->vc_fd
, (void *)&from
, &fromlen
, &to
, &tolen
)
3399 /* Finished, append a NUL. */
3404 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded
3405 * iconv library may use one of them. */
3406 if (!vcp
->vc_fail
&& unconvlenp
!= NULL
3407 && (ICONV_ERRNO
== ICONV_EINVAL
|| ICONV_ERRNO
== EINVAL
))
3409 /* Handle an incomplete sequence at the end. */
3411 *unconvlenp
= (int)fromlen
;
3415 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded
3416 * iconv library may use one of them. */
3417 else if (!vcp
->vc_fail
3418 && (ICONV_ERRNO
== ICONV_EILSEQ
|| ICONV_ERRNO
== EILSEQ
3419 || ICONV_ERRNO
== ICONV_EINVAL
|| ICONV_ERRNO
== EINVAL
))
3421 /* Can't convert: insert a '?' and skip a character. This assumes
3422 * conversion from 'encoding' to something else. In other
3423 * situations we don't know what to skip anyway. */
3425 if ((*mb_ptr2cells
)((char_u
*)from
) > 1)
3428 l
= utfc_ptr2len_len((char_u
*)from
, (int)fromlen
);
3431 l
= (*mb_ptr2len
)((char_u
*)from
);
3432 if (l
> (int)fromlen
)
3438 else if (ICONV_ERRNO
!= ICONV_E2BIG
)
3440 /* conversion failed */
3445 /* Not enough room or skipping illegal sequence. */
3446 done
= to
- (char *)result
;
3449 if (resultlenp
!= NULL
)
3450 *resultlenp
= (int)(to
- (char *)result
);
3454 # if defined(DYNAMIC_ICONV) || defined(PROTO)
3456 * Dynamically load the "iconv.dll" on Win32.
3459 #ifndef DYNAMIC_ICONV /* just generating prototypes */
3460 # define HINSTANCE int
3462 static HINSTANCE hIconvDLL
= 0;
3463 static HINSTANCE hMsvcrtDLL
= 0;
3465 # ifndef DYNAMIC_ICONV_DLL
3466 # define DYNAMIC_ICONV_DLL "iconv.dll"
3467 # define DYNAMIC_ICONV_DLL_ALT "libiconv.dll"
3469 # ifndef DYNAMIC_MSVCRT_DLL
3470 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll"
3474 * Try opening the iconv.dll and return TRUE if iconv() can be used.
3477 iconv_enabled(verbose
)
3480 if (hIconvDLL
!= 0 && hMsvcrtDLL
!= 0)
3482 hIconvDLL
= LoadLibrary(DYNAMIC_ICONV_DLL
);
3483 if (hIconvDLL
== 0) /* sometimes it's called libiconv.dll */
3484 hIconvDLL
= LoadLibrary(DYNAMIC_ICONV_DLL_ALT
);
3486 hMsvcrtDLL
= LoadLibrary(DYNAMIC_MSVCRT_DLL
);
3487 if (hIconvDLL
== 0 || hMsvcrtDLL
== 0)
3489 /* Only give the message when 'verbose' is set, otherwise it might be
3490 * done whenever a conversion is attempted. */
3491 if (verbose
&& p_verbose
> 0)
3495 hIconvDLL
== 0 ? DYNAMIC_ICONV_DLL
: DYNAMIC_MSVCRT_DLL
);
3502 iconv
= (void *)GetProcAddress(hIconvDLL
, "libiconv");
3503 iconv_open
= (void *)GetProcAddress(hIconvDLL
, "libiconv_open");
3504 iconv_close
= (void *)GetProcAddress(hIconvDLL
, "libiconv_close");
3505 iconvctl
= (void *)GetProcAddress(hIconvDLL
, "libiconvctl");
3506 iconv_errno
= (void *)GetProcAddress(hMsvcrtDLL
, "_errno");
3507 if (iconv
== NULL
|| iconv_open
== NULL
|| iconv_close
== NULL
3508 || iconvctl
== NULL
|| iconv_errno
== NULL
)
3511 if (verbose
&& p_verbose
> 0)
3514 EMSG2(_(e_loadfunc
), "for libiconv");
3525 /* Don't use iconv() when inputting or outputting characters. */
3526 if (input_conv
.vc_type
== CONV_ICONV
)
3527 convert_setup(&input_conv
, NULL
, NULL
);
3528 if (output_conv
.vc_type
== CONV_ICONV
)
3529 convert_setup(&output_conv
, NULL
, NULL
);
3532 FreeLibrary(hIconvDLL
);
3533 if (hMsvcrtDLL
!= 0)
3534 FreeLibrary(hMsvcrtDLL
);
3538 # endif /* DYNAMIC_ICONV */
3539 # endif /* USE_ICONV */
3541 #endif /* FEAT_MBYTE */
3543 #if defined(FEAT_XIM) || defined(PROTO)
3545 # ifdef FEAT_GUI_GTK
3546 static int xim_has_preediting
INIT(= FALSE
); /* IM current status */
3549 * Set preedit_start_col to the current cursor position.
3552 init_preedit_start_col(void)
3554 if (State
& CMDLINE
)
3555 preedit_start_col
= cmdline_getvcol_cursor();
3556 else if (curwin
!= NULL
)
3557 getvcol(curwin
, &curwin
->w_cursor
, &preedit_start_col
, NULL
, NULL
);
3558 /* Prevent that preediting marks the buffer as changed. */
3559 xim_changed_while_preediting
= curbuf
->b_changed
;
3563 # if defined(HAVE_GTK2) && !defined(PROTO)
3565 static int im_is_active
= FALSE
; /* IM is enabled for current mode */
3566 static int preedit_is_active
= FALSE
;
3567 static int im_preedit_cursor
= 0; /* cursor offset in characters */
3568 static int im_preedit_trailing
= 0; /* number of characters after cursor */
3570 static unsigned long im_commit_handler_id
= 0;
3571 static unsigned int im_activatekey_keyval
= GDK_VoidSymbol
;
3572 static unsigned int im_activatekey_state
= 0;
3575 im_set_active(int active
)
3579 was_active
= !!im_is_active
;
3580 im_is_active
= (active
&& !p_imdisable
);
3582 if (im_is_active
!= was_active
)
3587 xim_set_focus(int focus
)
3592 gtk_im_context_focus_in(xic
);
3594 gtk_im_context_focus_out(xic
);
3599 im_set_position(int row
, int col
)
3605 area
.x
= FILL_X(col
);
3606 area
.y
= FILL_Y(row
);
3607 area
.width
= gui
.char_width
* (mb_lefthalve(row
, col
) ? 2 : 1);
3608 area
.height
= gui
.char_height
;
3610 gtk_im_context_set_cursor_location(xic
, &area
);
3614 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */
3616 xim_set_preedit(void)
3618 im_set_position(gui
.row
, gui
.col
);
3623 im_add_to_input(char_u
*str
, int len
)
3625 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */
3626 if (input_conv
.vc_type
!= CONV_NONE
)
3628 str
= string_convert(&input_conv
, str
, &len
);
3629 g_return_if_fail(str
!= NULL
);
3632 add_to_input_buf_csi(str
, len
);
3634 if (input_conv
.vc_type
!= CONV_NONE
)
3637 if (p_mh
) /* blank out the pointer if necessary */
3638 gui_mch_mousehide(TRUE
);
3642 im_delete_preedit(void)
3644 char_u bskey
[] = {CSI
, 'k', 'b'};
3645 char_u delkey
[] = {CSI
, 'k', 'D'};
3649 im_preedit_cursor
= 0;
3652 for (; im_preedit_cursor
> 0; --im_preedit_cursor
)
3653 add_to_input_buf(bskey
, (int)sizeof(bskey
));
3655 for (; im_preedit_trailing
> 0; --im_preedit_trailing
)
3656 add_to_input_buf(delkey
, (int)sizeof(delkey
));
3660 * Move the cursor left by "num_move_back" characters.
3661 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for
3662 * these K_LEFT keys.
3665 im_correct_cursor(int num_move_back
)
3667 char_u backkey
[] = {CSI
, 'k', 'l'};
3671 # ifdef FEAT_RIGHTLEFT
3672 if ((State
& CMDLINE
) == 0 && curwin
!= NULL
&& curwin
->w_p_rl
)
3675 for (; num_move_back
> 0; --num_move_back
)
3676 add_to_input_buf(backkey
, (int)sizeof(backkey
));
3679 static int xim_expected_char
= NUL
;
3680 static int xim_ignored_char
= FALSE
;
3683 * Update the mode and cursor while in an IM callback.
3690 old_vgetc_busy
= vgetc_busy
;
3693 vgetc_busy
= old_vgetc_busy
;
3699 * Callback invoked when the user finished preediting.
3700 * Put the final string into the input buffer.
3703 im_commit_cb(GtkIMContext
*context UNUSED
,
3705 gpointer data UNUSED
)
3707 int slen
= (int)STRLEN(str
);
3708 int add_to_input
= TRUE
;
3711 int commit_with_preedit
= TRUE
;
3715 xim_log("im_commit_cb(): %s\n", str
);
3718 /* The imhangul module doesn't reset the preedit string before
3719 * committing. Call im_delete_preedit() to work around that. */
3720 im_delete_preedit();
3722 /* Indicate that preediting has finished. */
3723 if (preedit_start_col
== MAXCOL
)
3725 init_preedit_start_col();
3726 commit_with_preedit
= FALSE
;
3729 /* The thing which setting "preedit_start_col" to MAXCOL means that
3730 * "preedit_start_col" will be set forcely when calling
3731 * preedit_changed_cb() next time.
3732 * "preedit_start_col" should not reset with MAXCOL on this part. Vim
3733 * is simulating the preediting by using add_to_input_str(). when
3734 * preedit begin immediately before committed, the typebuf is not
3735 * flushed to screen, then it can't get correct "preedit_start_col".
3736 * Thus, it should calculate the cells by adding cells of the committed
3738 if (input_conv
.vc_type
!= CONV_NONE
)
3740 im_str
= string_convert(&input_conv
, (char_u
*)str
, &len
);
3741 g_return_if_fail(im_str
!= NULL
);
3744 im_str
= (char_u
*)str
;
3746 for (p
= im_str
; p
< im_str
+ len
; p
+= (*mb_ptr2len
)(p
))
3747 clen
+= (*mb_ptr2cells
)(p
);
3748 if (input_conv
.vc_type
!= CONV_NONE
)
3750 preedit_start_col
+= clen
;
3752 /* Is this a single character that matches a keypad key that's just
3753 * been pressed? If so, we don't want it to be entered as such - let
3754 * us carry on processing the raw keycode so that it may be used in
3755 * mappings as <kSomething>. */
3756 if (xim_expected_char
!= NUL
)
3758 /* We're currently processing a keypad or other special key */
3759 if (slen
== 1 && str
[0] == xim_expected_char
)
3761 /* It's a match - don't do it here */
3762 xim_ignored_char
= TRUE
;
3763 add_to_input
= FALSE
;
3768 xim_ignored_char
= FALSE
;
3773 im_add_to_input((char_u
*)str
, slen
);
3775 /* Inserting chars while "im_is_active" is set does not cause a change of
3776 * buffer. When the chars are committed the buffer must be marked as
3778 if (!commit_with_preedit
)
3779 preedit_start_col
= MAXCOL
;
3781 /* This flag is used in changed() at next call. */
3782 xim_changed_while_preediting
= TRUE
;
3784 if (gtk_main_level() > 0)
3789 * Callback invoked after start to the preedit.
3792 im_preedit_start_cb(GtkIMContext
*context UNUSED
, gpointer data UNUSED
)
3795 xim_log("im_preedit_start_cb()\n");
3798 im_is_active
= TRUE
;
3799 preedit_is_active
= TRUE
;
3800 gui_update_cursor(TRUE
, FALSE
);
3805 * Callback invoked after end to the preedit.
3808 im_preedit_end_cb(GtkIMContext
*context UNUSED
, gpointer data UNUSED
)
3811 xim_log("im_preedit_end_cb()\n");
3813 im_delete_preedit();
3815 /* Indicate that preediting has finished */
3816 preedit_start_col
= MAXCOL
;
3817 xim_has_preediting
= FALSE
;
3820 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was
3821 * switched off unintentionally. We now use preedit_is_active (added by
3823 im_is_active
= FALSE
;
3825 preedit_is_active
= FALSE
;
3826 gui_update_cursor(TRUE
, FALSE
);
3831 * Callback invoked after changes to the preedit string. If the preedit
3832 * string was empty before, remember the preedit start column so we know
3833 * where to apply feedback attributes. Delete the previous preedit string
3834 * if there was one, save the new preedit cursor offset, and put the new
3835 * string into the input buffer.
3837 * TODO: The pragmatic "put into input buffer" approach used here has
3838 * several fundamental problems:
3840 * - The characters in the preedit string are subject to remapping.
3841 * That's broken, only the finally committed string should be remapped.
3843 * - There is a race condition involved: The retrieved value for the
3844 * current cursor position will be wrong if any unprocessed characters
3845 * are still queued in the input buffer.
3847 * - Due to the lack of synchronization between the file buffer in memory
3848 * and any typed characters, it's practically impossible to implement the
3849 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM
3850 * modules for languages such as Thai are likely to rely on this feature
3851 * for proper operation.
3853 * Conclusions: I think support for preediting needs to be moved to the
3854 * core parts of Vim. Ideally, until it has been committed, the preediting
3855 * string should only be displayed and not affect the buffer content at all.
3856 * The question how to deal with the synchronization issue still remains.
3857 * Circumventing the input buffer is probably not desirable. Anyway, I think
3858 * implementing "retrieve_surrounding" is the only hard problem.
3860 * One way to solve all of this in a clean manner would be to queue all key
3861 * press/release events "as is" in the input buffer, and apply the IM filtering
3862 * at the receiving end of the queue. This, however, would have a rather large
3863 * impact on the code base. If there is an easy way to force processing of all
3864 * remaining input from within the "retrieve_surrounding" signal handler, this
3865 * might not be necessary. Gotta ask on vim-dev for opinions.
3868 im_preedit_changed_cb(GtkIMContext
*context
, gpointer data UNUSED
)
3870 char *preedit_string
= NULL
;
3871 int cursor_index
= 0;
3872 int num_move_back
= 0;
3877 gtk_im_context_get_preedit_string(context
,
3878 &preedit_string
, NULL
,
3882 xim_log("im_preedit_changed_cb(): %s\n", preedit_string
);
3885 g_return_if_fail(preedit_string
!= NULL
); /* just in case */
3887 /* If preedit_start_col is MAXCOL set it to the current cursor position. */
3888 if (preedit_start_col
== MAXCOL
&& preedit_string
[0] != '\0')
3890 xim_has_preediting
= TRUE
;
3892 /* Urgh, this breaks if the input buffer isn't empty now */
3893 init_preedit_start_col();
3895 else if (cursor_index
== 0 && preedit_string
[0] == '\0')
3897 xim_has_preediting
= FALSE
;
3899 /* If at the start position (after typing backspace)
3900 * preedit_start_col must be reset. */
3901 preedit_start_col
= MAXCOL
;
3904 im_delete_preedit();
3907 * Compute the end of the preediting area: "preedit_end_col".
3908 * According to the documentation of gtk_im_context_get_preedit_string(),
3909 * the cursor_pos output argument returns the offset in bytes. This is
3910 * unfortunately not true -- real life shows the offset is in characters,
3911 * and the GTK+ source code agrees with me. Will file a bug later.
3913 if (preedit_start_col
!= MAXCOL
)
3914 preedit_end_col
= preedit_start_col
;
3915 str
= (char_u
*)preedit_string
;
3916 for (p
= str
, i
= 0; *p
!= NUL
; p
+= utf_byte2len(*p
), ++i
)
3920 is_composing
= ((*p
& 0x80) != 0 && utf_iscomposing(utf_ptr2char(p
)));
3922 * These offsets are used as counters when generating <BS> and <Del>
3923 * to delete the preedit string. So don't count composing characters
3924 * unless 'delcombine' is enabled.
3926 if (!is_composing
|| p_deco
)
3928 if (i
< cursor_index
)
3929 ++im_preedit_cursor
;
3931 ++im_preedit_trailing
;
3933 if (!is_composing
&& i
>= cursor_index
)
3935 /* This is essentially the same as im_preedit_trailing, except
3936 * composing characters are not counted even if p_deco is set. */
3939 if (preedit_start_col
!= MAXCOL
)
3940 preedit_end_col
+= utf_ptr2cells(p
);
3945 im_add_to_input(str
, (int)(p
- str
));
3946 im_correct_cursor(num_move_back
);
3949 g_free(preedit_string
);
3951 if (gtk_main_level() > 0)
3956 * Translate the Pango attributes at iter to Vim highlighting attributes.
3957 * Ignore attributes not supported by Vim highlighting. This shouldn't have
3958 * too much impact -- right now we handle even more attributes than necessary
3959 * for the IM modules I tested with.
3962 translate_pango_attributes(PangoAttrIterator
*iter
)
3964 PangoAttribute
*attr
;
3965 int char_attr
= HL_NORMAL
;
3967 attr
= pango_attr_iterator_get(iter
, PANGO_ATTR_UNDERLINE
);
3968 if (attr
!= NULL
&& ((PangoAttrInt
*)attr
)->value
3969 != (int)PANGO_UNDERLINE_NONE
)
3970 char_attr
|= HL_UNDERLINE
;
3972 attr
= pango_attr_iterator_get(iter
, PANGO_ATTR_WEIGHT
);
3973 if (attr
!= NULL
&& ((PangoAttrInt
*)attr
)->value
>= (int)PANGO_WEIGHT_BOLD
)
3974 char_attr
|= HL_BOLD
;
3976 attr
= pango_attr_iterator_get(iter
, PANGO_ATTR_STYLE
);
3977 if (attr
!= NULL
&& ((PangoAttrInt
*)attr
)->value
3978 != (int)PANGO_STYLE_NORMAL
)
3979 char_attr
|= HL_ITALIC
;
3981 attr
= pango_attr_iterator_get(iter
, PANGO_ATTR_BACKGROUND
);
3984 const PangoColor
*color
= &((PangoAttrColor
*)attr
)->color
;
3986 /* Assume inverse if black background is requested */
3987 if ((color
->red
| color
->green
| color
->blue
) == 0)
3988 char_attr
|= HL_INVERSE
;
3995 * Retrieve the highlighting attributes at column col in the preedit string.
3996 * Return -1 if not in preediting mode or if col is out of range.
3999 im_get_feedback_attr(int col
)
4001 char *preedit_string
= NULL
;
4002 PangoAttrList
*attr_list
= NULL
;
4008 gtk_im_context_get_preedit_string(xic
, &preedit_string
, &attr_list
, NULL
);
4010 if (preedit_string
!= NULL
&& attr_list
!= NULL
)
4014 /* Get the byte index as used by PangoAttrIterator */
4015 for (idx
= 0; col
> 0 && preedit_string
[idx
] != '\0'; --col
)
4016 idx
+= utfc_ptr2len((char_u
*)preedit_string
+ idx
);
4018 if (preedit_string
[idx
] != '\0')
4020 PangoAttrIterator
*iter
;
4023 char_attr
= HL_NORMAL
;
4024 iter
= pango_attr_list_get_iterator(attr_list
);
4026 /* Extract all relevant attributes from the list. */
4029 pango_attr_iterator_range(iter
, &start
, &end
);
4031 if (idx
>= start
&& idx
< end
)
4032 char_attr
|= translate_pango_attributes(iter
);
4034 while (pango_attr_iterator_next(iter
));
4036 pango_attr_iterator_destroy(iter
);
4040 if (attr_list
!= NULL
)
4041 pango_attr_list_unref(attr_list
);
4042 g_free(preedit_string
);
4051 xim_log("xim_init()\n");
4054 g_return_if_fail(gui
.drawarea
!= NULL
);
4055 g_return_if_fail(gui
.drawarea
->window
!= NULL
);
4057 xic
= gtk_im_multicontext_new();
4060 im_commit_handler_id
= g_signal_connect(G_OBJECT(xic
), "commit",
4061 G_CALLBACK(&im_commit_cb
), NULL
);
4062 g_signal_connect(G_OBJECT(xic
), "preedit_changed",
4063 G_CALLBACK(&im_preedit_changed_cb
), NULL
);
4064 g_signal_connect(G_OBJECT(xic
), "preedit_start",
4065 G_CALLBACK(&im_preedit_start_cb
), NULL
);
4066 g_signal_connect(G_OBJECT(xic
), "preedit_end",
4067 G_CALLBACK(&im_preedit_end_cb
), NULL
);
4069 gtk_im_context_set_client_window(xic
, gui
.drawarea
->window
);
4076 xim_log("im_shutdown()\n");
4081 gtk_im_context_focus_out(xic
);
4082 g_object_unref(xic
);
4085 im_is_active
= FALSE
;
4086 im_commit_handler_id
= 0;
4087 preedit_start_col
= MAXCOL
;
4088 xim_has_preediting
= FALSE
;
4092 * Convert the string argument to keyval and state for GdkEventKey.
4093 * If str is valid return TRUE, otherwise FALSE.
4095 * See 'imactivatekey' for documentation of the format.
4098 im_string_to_keyval(const char *str
, unsigned int *keyval
, unsigned int *state
)
4100 const char *mods_end
;
4101 unsigned tmp_keyval
;
4102 unsigned tmp_state
= 0;
4104 mods_end
= strrchr(str
, '-');
4105 mods_end
= (mods_end
!= NULL
) ? mods_end
+ 1 : str
;
4107 /* Parse modifier keys */
4108 while (str
< mods_end
)
4112 case 'S': case 's': tmp_state
|= (unsigned)GDK_SHIFT_MASK
; break;
4113 case 'L': case 'l': tmp_state
|= (unsigned)GDK_LOCK_MASK
; break;
4114 case 'C': case 'c': tmp_state
|= (unsigned)GDK_CONTROL_MASK
;break;
4115 case '1': tmp_state
|= (unsigned)GDK_MOD1_MASK
; break;
4116 case '2': tmp_state
|= (unsigned)GDK_MOD2_MASK
; break;
4117 case '3': tmp_state
|= (unsigned)GDK_MOD3_MASK
; break;
4118 case '4': tmp_state
|= (unsigned)GDK_MOD4_MASK
; break;
4119 case '5': tmp_state
|= (unsigned)GDK_MOD5_MASK
; break;
4124 tmp_keyval
= gdk_keyval_from_name(str
);
4126 if (tmp_keyval
== 0 || tmp_keyval
== GDK_VoidSymbol
)
4130 *keyval
= tmp_keyval
;
4138 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an
4139 * empty string is also regarded as valid.
4141 * Note: The numerical key value of p_imak is cached if it was valid; thus
4142 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever
4143 * 'imak' changes. This is currently the case but not obvious -- should
4144 * probably rename the function for clarity.
4147 im_xim_isvalid_imactivate(void)
4149 if (p_imak
[0] == NUL
)
4151 im_activatekey_keyval
= GDK_VoidSymbol
;
4152 im_activatekey_state
= 0;
4156 return im_string_to_keyval((const char *)p_imak
,
4157 &im_activatekey_keyval
,
4158 &im_activatekey_state
);
4162 im_synthesize_keypress(unsigned int keyval
, unsigned int state
)
4166 # ifdef HAVE_GTK_MULTIHEAD
4167 event
= (GdkEventKey
*)gdk_event_new(GDK_KEY_PRESS
);
4168 g_object_ref(gui
.drawarea
->window
); /* unreffed by gdk_event_free() */
4170 event
= (GdkEventKey
*)g_malloc0((gulong
)sizeof(GdkEvent
));
4171 event
->type
= GDK_KEY_PRESS
;
4173 event
->window
= gui
.drawarea
->window
;
4174 event
->send_event
= TRUE
;
4175 event
->time
= GDK_CURRENT_TIME
;
4176 event
->state
= state
;
4177 event
->keyval
= keyval
;
4178 event
->hardware_keycode
= /* needed for XIM */
4179 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event
->window
), (KeySym
)keyval
);
4181 event
->string
= NULL
;
4183 gtk_im_context_filter_keypress(xic
, event
);
4185 /* For consistency, also send the corresponding release event. */
4186 event
->type
= GDK_KEY_RELEASE
;
4187 event
->send_event
= FALSE
;
4188 gtk_im_context_filter_keypress(xic
, event
);
4190 # ifdef HAVE_GTK_MULTIHEAD
4191 gdk_event_free((GdkEvent
*)event
);
4203 * The third-party imhangul module (and maybe others too) ignores
4204 * gtk_im_context_reset() or at least doesn't reset the active state.
4205 * Thus sending imactivatekey would turn it off if it was on before,
4206 * which is clearly not what we want. Fortunately we can work around
4207 * that for imhangul by sending GDK_Escape, but I don't know if it
4208 * works with all IM modules that support an activation key :/
4210 * An alternative approach would be to destroy the IM context and
4211 * recreate it. But that means loading/unloading the IM module on
4212 * every mode switch, which causes a quite noticable delay even on
4213 * my rather fast box...
4215 * Moreover, there are some XIM which cannot respond to
4216 * im_synthesize_keypress(). we hope that they reset by
4219 if (im_activatekey_keyval
!= GDK_VoidSymbol
&& im_is_active
)
4220 im_synthesize_keypress(GDK_Escape
, 0U);
4222 gtk_im_context_reset(xic
);
4225 * HACK for Ami: This sequence of function calls makes Ami handle
4226 * the IM reset graciously, without breaking loads of other stuff.
4227 * It seems to force English mode as well, which is exactly what we
4228 * want because it makes the Ami status display work reliably.
4230 gtk_im_context_set_use_preedit(xic
, FALSE
);
4236 gtk_im_context_set_use_preedit(xic
, TRUE
);
4237 xim_set_focus(gui
.in_focus
);
4239 if (im_activatekey_keyval
!= GDK_VoidSymbol
)
4243 g_signal_handler_block(xic
, im_commit_handler_id
);
4244 im_synthesize_keypress(im_activatekey_keyval
,
4245 im_activatekey_state
);
4246 g_signal_handler_unblock(xic
, im_commit_handler_id
);
4253 xim_set_focus(gui
.in_focus
);
4258 preedit_start_col
= MAXCOL
;
4259 xim_has_preediting
= FALSE
;
4263 xim_queue_key_press_event(GdkEventKey
*event
, int down
)
4268 * Workaround GTK2 XIM 'feature' that always converts keypad keys to
4269 * chars., even when not part of an IM sequence (ref. feature of
4271 * Flag any keypad keys that might represent a single char.
4272 * If this (on its own - i.e., not part of an IM sequence) is
4273 * committed while we're processing one of these keys, we can ignore
4274 * that commit and go ahead & process it ourselves. That way we can
4275 * still distinguish keypad keys for use in mappings.
4276 * Also add GDK_space to make <S-Space> work.
4278 switch (event
->keyval
)
4280 case GDK_KP_Add
: xim_expected_char
= '+'; break;
4281 case GDK_KP_Subtract
: xim_expected_char
= '-'; break;
4282 case GDK_KP_Divide
: xim_expected_char
= '/'; break;
4283 case GDK_KP_Multiply
: xim_expected_char
= '*'; break;
4284 case GDK_KP_Decimal
: xim_expected_char
= '.'; break;
4285 case GDK_KP_Equal
: xim_expected_char
= '='; break;
4286 case GDK_KP_0
: xim_expected_char
= '0'; break;
4287 case GDK_KP_1
: xim_expected_char
= '1'; break;
4288 case GDK_KP_2
: xim_expected_char
= '2'; break;
4289 case GDK_KP_3
: xim_expected_char
= '3'; break;
4290 case GDK_KP_4
: xim_expected_char
= '4'; break;
4291 case GDK_KP_5
: xim_expected_char
= '5'; break;
4292 case GDK_KP_6
: xim_expected_char
= '6'; break;
4293 case GDK_KP_7
: xim_expected_char
= '7'; break;
4294 case GDK_KP_8
: xim_expected_char
= '8'; break;
4295 case GDK_KP_9
: xim_expected_char
= '9'; break;
4296 case GDK_space
: xim_expected_char
= ' '; break;
4297 default: xim_expected_char
= NUL
;
4299 xim_ignored_char
= FALSE
;
4303 * When typing fFtT, XIM may be activated. Thus it must pass
4304 * gtk_im_context_filter_keypress() in Normal mode.
4305 * And while doing :sh too.
4307 if (xic
!= NULL
&& !p_imdisable
4308 && (State
& (INSERT
| CMDLINE
| NORMAL
| EXTERNCMD
)) != 0)
4311 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is
4312 * always aware of the current status of IM, and can even emulate
4313 * the activation key for modules that don't support one.
4315 if (event
->keyval
== im_activatekey_keyval
4316 && (event
->state
& im_activatekey_state
) == im_activatekey_state
)
4318 unsigned int state_mask
;
4320 /* Require the state of the 3 most used modifiers to match exactly.
4321 * Otherwise e.g. <S-C-space> would be unusable for other purposes
4322 * if the IM activate key is <S-space>. */
4323 state_mask
= im_activatekey_state
;
4324 state_mask
|= ((int)GDK_SHIFT_MASK
| (int)GDK_CONTROL_MASK
4325 | (int)GDK_MOD1_MASK
);
4327 if ((event
->state
& state_mask
) != im_activatekey_state
)
4330 /* Don't send it a second time on GDK_KEY_RELEASE. */
4331 if (event
->type
!= GDK_KEY_PRESS
)
4334 if (map_to_exists_mode((char_u
*)"", LANGMAP
, FALSE
))
4336 im_set_active(FALSE
);
4338 /* ":lmap" mappings exists, toggle use of mappings. */
4340 if (State
& LANGMAP
)
4342 curbuf
->b_p_iminsert
= B_IMODE_NONE
;
4347 curbuf
->b_p_iminsert
= B_IMODE_LMAP
;
4353 return gtk_im_context_filter_keypress(xic
, event
);
4356 /* Don't filter events through the IM context if IM isn't active
4357 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module
4358 * not doing anything before the activation key was sent. */
4359 if (im_activatekey_keyval
== GDK_VoidSymbol
|| im_is_active
)
4361 int imresult
= gtk_im_context_filter_keypress(xic
, event
);
4363 /* Some XIM send following sequence:
4364 * 1. preedited string.
4365 * 2. committed string.
4366 * 3. line changed key.
4367 * 4. preedited string.
4368 * 5. remove preedited string.
4369 * if 3, Vim can't move back the above line for 5.
4370 * thus, this part should not parse the key. */
4371 if (!imresult
&& preedit_start_col
!= MAXCOL
4372 && event
->keyval
== GDK_Return
)
4374 im_synthesize_keypress(GDK_Return
, 0U);
4378 /* If XIM tried to commit a keypad key as a single char.,
4379 * ignore it so we can use the keypad key 'raw', for mappings. */
4380 if (xim_expected_char
!= NUL
&& xim_ignored_char
)
4381 /* We had a keypad key, and XIM tried to thieve it */
4384 /* Normal processing */
4395 return im_is_active
;
4398 # else /* !HAVE_GTK2 */
4400 static int xim_is_active
= FALSE
; /* XIM should be active in the current
4402 static int xim_has_focus
= FALSE
; /* XIM is really being used for Vim */
4404 static XIMStyle input_style
;
4405 static int status_area_enabled
= TRUE
;
4410 # include <gdk/gdkwin32.h>
4412 # include <gdk/gdkx.h>
4416 /* Define a few things to be able to generate prototypes while not configured
4419 # define gboolean int
4420 typedef int GdkEvent
;
4421 typedef int GdkEventKey
;
4426 #if defined(FEAT_GUI_GTK) || defined(PROTO)
4427 static int preedit_buf_len
= 0;
4428 static int xim_can_preediting
INIT(= FALSE
); /* XIM in showmode() */
4429 static int xim_input_style
;
4430 #ifndef FEAT_GUI_GTK
4431 # define gboolean int
4433 static gboolean use_status_area
= 0;
4435 static int im_xim_str2keycode
__ARGS((unsigned int *code
, unsigned int *state
));
4436 static void im_xim_send_event_imactivate
__ARGS((void));
4439 * Convert string to keycode and state for XKeyEvent.
4440 * When string is valid return OK, when invalid return FAIL.
4442 * See 'imactivatekey' documentation for the format.
4445 im_xim_str2keycode(code
, state
)
4447 unsigned int *state
;
4451 unsigned keycode
= 0, keystate
= 0;
4459 len
= STRLEN(p_imak
);
4460 for (flag_end
= p_imak
+ len
- 1;
4461 flag_end
> p_imak
&& *flag_end
!= '-'; --flag_end
)
4464 /* Parse modifier keys */
4465 for (str
= p_imak
; str
< flag_end
; ++str
)
4470 keystate
|= ShiftMask
;
4473 keystate
|= LockMask
;
4476 keystate
|= ControlMask
;
4479 keystate
|= Mod1Mask
;
4482 keystate
|= Mod2Mask
;
4485 keystate
|= Mod3Mask
;
4488 keystate
|= Mod4Mask
;
4491 keystate
|= Mod5Mask
;
4502 /* Get keycode from string. */
4503 gui_get_x11_windis(&window
, &display
);
4505 keycode
= XKeysymToKeycode(display
, XStringToKeysym((char *)str
));
4518 im_xim_send_event_imactivate()
4520 /* Force turn on preedit state by symulate keypress event.
4521 * Keycode and state is specified by 'imactivatekey'.
4525 gui_get_x11_windis(&ev
.window
, &ev
.display
);
4526 ev
.root
= RootWindow(ev
.display
, DefaultScreen(ev
.display
));
4527 ev
.subwindow
= None
;
4528 ev
.time
= CurrentTime
;
4535 if (im_xim_str2keycode(&ev
.keycode
, &ev
.state
) == OK
)
4536 XSendEvent(ev
.display
, ev
.window
, 1, KeyPressMask
, (XEvent
*)&ev
);
4540 * Return TRUE if 'imactivatekey' has a valid value.
4543 im_xim_isvalid_imactivate()
4545 return im_xim_str2keycode(NULL
, NULL
) == OK
;
4547 #endif /* FEAT_GUI_GTK */
4550 * Switch using XIM on/off. This is used by the code that changes "State".
4553 im_set_active(active
)
4559 /* If 'imdisable' is set, XIM is never active. */
4562 #if !defined (FEAT_GUI_GTK)
4563 else if (input_style
& XIMPreeditPosition
)
4564 /* There is a problem in switching XIM off when preediting is used,
4565 * and it is not clear how this can be solved. For now, keep XIM on
4566 * all the time, like it was done in Vim 5.8. */
4570 /* Remember the active state, it is needed when Vim gets keyboard focus. */
4571 xim_is_active
= active
;
4574 /* When 'imactivatekey' has valid key-string, try to control XIM preedit
4575 * state. When 'imactivatekey' has no or invalid string, try old XIM
4581 * Destroy old Input Context (XIC), and create new one. New XIC
4582 * would have a state of preedit that is off. When argument:active
4583 * is false, that's all. Else argument:active is true, send a key
4584 * event specified by 'imactivatekey' to activate XIM preedit state.
4587 xim_is_active
= TRUE
; /* Disable old XIM focus control */
4588 /* If we can monitor preedit state with preedit callback functions,
4589 * try least creation of new XIC.
4591 if (xim_input_style
& (int)GDK_IM_PREEDIT_CALLBACKS
)
4593 if (xim_can_preediting
&& !active
)
4595 /* Force turn off preedit state. With some IM
4596 * implementations, we cannot turn off preedit state by
4597 * symulate keypress event. It is why using such a method
4598 * that destroy old IC (input context), and create new one.
4599 * When create new IC, its preedit state is usually off.
4602 xim_set_focus(FALSE
);
4603 gdk_ic_destroy(xic
);
4605 xim_can_preediting
= FALSE
;
4607 else if (!xim_can_preediting
&& active
)
4608 im_xim_send_event_imactivate();
4612 /* First, force destroy old IC, and create new one. It
4613 * symulates "turning off preedit state".
4615 xim_set_focus(FALSE
);
4616 gdk_ic_destroy(xic
);
4618 xim_can_preediting
= FALSE
;
4620 /* 2nd, when requested to activate IM, symulate this by sending
4625 im_xim_send_event_imactivate();
4626 xim_can_preediting
= TRUE
;
4632 # ifndef XIMPreeditUnKnown
4633 /* X11R5 doesn't have these, it looks safe enough to define here. */
4634 typedef unsigned long XIMPreeditState
;
4635 # define XIMPreeditUnKnown 0L
4636 # define XIMPreeditEnable 1L
4637 # define XIMPreeditDisable (1L<<1)
4638 # define XNPreeditState "preeditState"
4640 XIMPreeditState preedit_state
= XIMPreeditUnKnown
;
4641 XVaNestedList preedit_attr
;
4644 preedit_attr
= XVaCreateNestedList(0,
4645 XNPreeditState
, &preedit_state
,
4647 pxic
= ((GdkICPrivate
*)xic
)->xic
;
4649 if (!XGetICValues(pxic
, XNPreeditAttributes
, preedit_attr
, NULL
))
4651 XFree(preedit_attr
);
4652 preedit_attr
= XVaCreateNestedList(0,
4654 active
? XIMPreeditEnable
: XIMPreeditDisable
,
4656 XSetICValues(pxic
, XNPreeditAttributes
, preedit_attr
, NULL
);
4657 xim_can_preediting
= active
;
4658 xim_is_active
= active
;
4660 XFree(preedit_attr
);
4662 if (xim_input_style
& XIMPreeditCallbacks
)
4664 preedit_buf_len
= 0;
4665 init_preedit_start_col();
4669 /* When had tested kinput2 + canna + Athena GUI version with
4670 * 'imactivatekey' is "s-space", im_xim_send_event_imactivate() did not
4671 * work correctly. It just inserted one space. I don't know why we
4672 * couldn't switch state of XIM preediting. This is reason why these
4673 * codes are commented out.
4675 /* First, force destroy old IC, and create new one. It symulates
4676 * "turning off preedit state".
4678 xim_set_focus(FALSE
);
4683 /* 2nd, when requested to activate IM, symulate this by sending the
4687 im_xim_send_event_imactivate();
4694 * Adjust using XIM for gaining or losing keyboard focus. Also called when
4695 * "xim_is_active" changes.
4698 xim_set_focus(focus
)
4705 * XIM only gets focus when the Vim window has keyboard focus and XIM has
4706 * been set active for the current mode.
4708 if (focus
&& xim_is_active
)
4712 xim_has_focus
= TRUE
;
4714 gdk_im_begin(xic
, gui
.drawarea
->window
);
4724 xim_has_focus
= FALSE
;
4735 im_set_position(row
, col
)
4743 * Set the XIM to the current cursor position.
4751 xim_set_focus(TRUE
);
4765 # ifdef FEAT_XFONTSET
4766 if ((xim_input_style
& (int)GDK_IM_PREEDIT_POSITION
)
4767 && gui
.fontset
!= NOFONTSET
4768 && gui
.fontset
->type
== GDK_FONT_FONTSET
)
4772 if (attr
->spot_location
.y
>= 0)
4774 attr
->spot_location
.x
= 0;
4775 attr
->spot_location
.y
= -100;
4776 attrmask
|= (int)GDK_IC_SPOT_LOCATION
;
4783 if (attr
->spot_location
.x
!= TEXT_X(gui
.col
)
4784 || attr
->spot_location
.y
!= TEXT_Y(gui
.row
))
4786 attr
->spot_location
.x
= TEXT_X(gui
.col
);
4787 attr
->spot_location
.y
= TEXT_Y(gui
.row
);
4788 attrmask
|= (int)GDK_IC_SPOT_LOCATION
;
4791 gdk_window_get_size(gui
.drawarea
->window
, &width
, &height
);
4792 width
-= 2 * gui
.border_offset
;
4793 height
-= 2 * gui
.border_offset
;
4794 if (xim_input_style
& (int)GDK_IM_STATUS_AREA
)
4795 height
-= gui
.char_height
;
4796 if (attr
->preedit_area
.width
!= width
4797 || attr
->preedit_area
.height
!= height
)
4799 attr
->preedit_area
.x
= gui
.border_offset
;
4800 attr
->preedit_area
.y
= gui
.border_offset
;
4801 attr
->preedit_area
.width
= width
;
4802 attr
->preedit_area
.height
= height
;
4803 attrmask
|= (int)GDK_IC_PREEDIT_AREA
;
4806 if (attr
->preedit_fontset
!= gui
.current_font
)
4808 attr
->preedit_fontset
= gui
.current_font
;
4809 attrmask
|= (int)GDK_IC_PREEDIT_FONTSET
;
4813 # endif /* FEAT_XFONTSET */
4815 if (xim_fg_color
== INVALCOLOR
)
4817 xim_fg_color
= gui
.def_norm_pixel
;
4818 xim_bg_color
= gui
.def_back_pixel
;
4820 if (attr
->preedit_foreground
.pixel
!= xim_fg_color
)
4822 attr
->preedit_foreground
.pixel
= xim_fg_color
;
4823 attrmask
|= (int)GDK_IC_PREEDIT_FOREGROUND
;
4825 if (attr
->preedit_background
.pixel
!= xim_bg_color
)
4827 attr
->preedit_background
.pixel
= xim_bg_color
;
4828 attrmask
|= (int)GDK_IC_PREEDIT_BACKGROUND
;
4832 gdk_ic_set_attr(xic
, attr
, (GdkICAttributesType
)attrmask
);
4834 #else /* FEAT_GUI_GTK */
4836 XVaNestedList attr_list
;
4837 XRectangle spot_area
;
4843 /* hide XIM cursor */
4845 over_spot
.y
= -100; /* arbitrary invisible position */
4846 attr_list
= (XVaNestedList
) XVaCreateNestedList(0,
4850 XSetICValues(xic
, XNPreeditAttributes
, attr_list
, NULL
);
4855 if (input_style
& XIMPreeditPosition
)
4857 if (xim_fg_color
== INVALCOLOR
)
4859 xim_fg_color
= gui
.def_norm_pixel
;
4860 xim_bg_color
= gui
.def_back_pixel
;
4862 over_spot
.x
= TEXT_X(gui
.col
);
4863 over_spot
.y
= TEXT_Y(gui
.row
);
4866 spot_area
.height
= gui
.char_height
* Rows
;
4867 spot_area
.width
= gui
.char_width
* Columns
;
4868 line_space
= gui
.char_height
;
4869 attr_list
= (XVaNestedList
) XVaCreateNestedList(0,
4870 XNSpotLocation
, &over_spot
,
4871 XNForeground
, (Pixel
) xim_fg_color
,
4872 XNBackground
, (Pixel
) xim_bg_color
,
4874 XNLineSpace
, line_space
,
4876 if (XSetICValues(xic
, XNPreeditAttributes
, attr_list
, NULL
))
4877 EMSG(_("E284: Cannot set IC values"));
4881 #endif /* FEAT_GUI_GTK */
4885 * Set up the status area.
4887 * This should use a separate Widget, but that seems not possible, because
4888 * preedit_area and status_area should be set to the same window as for the
4889 * text input. Unfortunately this means the status area pollutes the text
4893 xim_set_status_area()
4899 # if defined(FEAT_XFONTSET)
4900 if (use_status_area
)
4913 style
= (int)gdk_ic_get_style(xic
);
4914 if ((style
& (int)GDK_IM_STATUS_MASK
) == (int)GDK_IM_STATUS_AREA
)
4916 if (gui
.fontset
!= NOFONTSET
4917 && gui
.fontset
->type
== GDK_FONT_FONTSET
)
4919 widget
= gui
.mainwin
;
4920 gdk_window_get_size(widget
->window
, &width
, &height
);
4922 attrmask
|= (int)GDK_IC_STATUS_AREA
;
4923 attr
->status_area
.x
= 0;
4924 attr
->status_area
.y
= height
- gui
.char_height
- 1;
4925 attr
->status_area
.width
= width
;
4926 attr
->status_area
.height
= gui
.char_height
;
4930 gdk_ic_set_attr(xic
, attr
, (GdkICAttributesType
)attrmask
);
4935 XVaNestedList preedit_list
= 0, status_list
= 0, list
= 0;
4936 XRectangle pre_area
, status_area
;
4938 if (input_style
& XIMStatusArea
)
4940 if (input_style
& XIMPreeditArea
)
4942 XRectangle
*needed_rect
;
4944 /* to get status_area width */
4945 status_list
= XVaCreateNestedList(0, XNAreaNeeded
,
4946 &needed_rect
, NULL
);
4947 XGetICValues(xic
, XNStatusAttributes
, status_list
, NULL
);
4950 status_area
.width
= needed_rect
->width
;
4953 status_area
.width
= gui
.char_width
* Columns
;
4956 status_area
.y
= gui
.char_height
* Rows
+ gui
.border_offset
;
4957 if (gui
.which_scrollbars
[SBAR_BOTTOM
])
4958 status_area
.y
+= gui
.scrollbar_height
;
4960 if (gui
.menu_is_active
)
4961 status_area
.y
+= gui
.menu_height
;
4963 status_area
.height
= gui
.char_height
;
4964 status_list
= XVaCreateNestedList(0, XNArea
, &status_area
, NULL
);
4969 status_area
.y
= gui
.char_height
* Rows
+ gui
.border_offset
;
4970 if (gui
.which_scrollbars
[SBAR_BOTTOM
])
4971 status_area
.y
+= gui
.scrollbar_height
;
4973 if (gui
.menu_is_active
)
4974 status_area
.y
+= gui
.menu_height
;
4976 status_area
.width
= 0;
4977 status_area
.height
= gui
.char_height
;
4980 if (input_style
& XIMPreeditArea
) /* off-the-spot */
4982 pre_area
.x
= status_area
.x
+ status_area
.width
;
4983 pre_area
.y
= gui
.char_height
* Rows
+ gui
.border_offset
;
4984 pre_area
.width
= gui
.char_width
* Columns
- pre_area
.x
;
4985 if (gui
.which_scrollbars
[SBAR_BOTTOM
])
4986 pre_area
.y
+= gui
.scrollbar_height
;
4988 if (gui
.menu_is_active
)
4989 pre_area
.y
+= gui
.menu_height
;
4991 pre_area
.height
= gui
.char_height
;
4992 preedit_list
= XVaCreateNestedList(0, XNArea
, &pre_area
, NULL
);
4994 else if (input_style
& XIMPreeditPosition
) /* over-the-spot */
4998 pre_area
.height
= gui
.char_height
* Rows
;
4999 pre_area
.width
= gui
.char_width
* Columns
;
5000 preedit_list
= XVaCreateNestedList(0, XNArea
, &pre_area
, NULL
);
5003 if (preedit_list
&& status_list
)
5004 list
= XVaCreateNestedList(0, XNPreeditAttributes
, preedit_list
,
5005 XNStatusAttributes
, status_list
, NULL
);
5006 else if (preedit_list
)
5007 list
= XVaCreateNestedList(0, XNPreeditAttributes
, preedit_list
,
5009 else if (status_list
)
5010 list
= XVaCreateNestedList(0, XNStatusAttributes
, status_list
,
5017 XSetICValues(xic
, XNVaNestedList
, list
, NULL
);
5023 XFree(preedit_list
);
5028 #if defined(FEAT_GUI_X11) || defined(FEAT_GUI_GTK)
5029 static char e_xim
[] = N_("E285: Failed to create input context");
5032 #if defined(FEAT_GUI_X11) || defined(PROTO)
5033 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(sun)
5034 # define USE_X11R6_XIM
5037 static int xim_real_init
__ARGS((Window x11_window
, Display
*x11_display
));
5040 #ifdef USE_X11R6_XIM
5041 static void xim_instantiate_cb
__ARGS((Display
*display
, XPointer client_data
, XPointer call_data
));
5042 static void xim_destroy_cb
__ARGS((XIM im
, XPointer client_data
, XPointer call_data
));
5045 xim_instantiate_cb(display
, client_data
, call_data
)
5047 XPointer client_data UNUSED
;
5048 XPointer call_data UNUSED
;
5051 Display
*x11_display
;
5054 xim_log("xim_instantiate_cb()\n");
5057 gui_get_x11_windis(&x11_window
, &x11_display
);
5058 if (display
!= x11_display
)
5061 xim_real_init(x11_window
, x11_display
);
5062 gui_set_shellsize(FALSE
, FALSE
, RESIZE_BOTH
);
5064 XUnregisterIMInstantiateCallback(x11_display
, NULL
, NULL
, NULL
,
5065 xim_instantiate_cb
, NULL
);
5069 xim_destroy_cb(im
, client_data
, call_data
)
5071 XPointer client_data UNUSED
;
5072 XPointer call_data UNUSED
;
5075 Display
*x11_display
;
5078 xim_log("xim_destroy_cb()\n");
5080 gui_get_x11_windis(&x11_window
, &x11_display
);
5083 status_area_enabled
= FALSE
;
5085 gui_set_shellsize(FALSE
, FALSE
, RESIZE_BOTH
);
5087 XRegisterIMInstantiateCallback(x11_display
, NULL
, NULL
, NULL
,
5088 xim_instantiate_cb
, NULL
);
5096 Display
*x11_display
;
5099 xim_log("xim_init()\n");
5102 gui_get_x11_windis(&x11_window
, &x11_display
);
5106 if (xim_real_init(x11_window
, x11_display
))
5109 gui_set_shellsize(FALSE
, FALSE
, RESIZE_BOTH
);
5111 #ifdef USE_X11R6_XIM
5112 XRegisterIMInstantiateCallback(x11_display
, NULL
, NULL
, NULL
,
5113 xim_instantiate_cb
, NULL
);
5118 xim_real_init(x11_window
, x11_display
)
5120 Display
*x11_display
;
5128 #define IMLEN_MAX 40
5129 char buf
[IMLEN_MAX
+ 7];
5131 XIMStyles
*xim_styles
;
5132 XIMStyle this_input_style
= 0;
5135 XVaNestedList preedit_list
, status_list
;
5138 status_area_enabled
= FALSE
;
5143 if (gui
.rsrc_input_method
!= NULL
&& *gui
.rsrc_input_method
!= NUL
)
5145 strcpy(tmp
, gui
.rsrc_input_method
);
5146 for (ns
= s
= tmp
; ns
!= NULL
&& *s
!= NUL
;)
5148 s
= (char *)skipwhite((char_u
*)s
);
5151 if ((ns
= end
= strchr(s
, ',')) == NULL
)
5152 end
= s
+ strlen(s
);
5153 while (isspace(((char_u
*)end
)[-1]))
5157 if (strlen(s
) <= IMLEN_MAX
)
5159 strcpy(buf
, "@im=");
5161 if ((p
= XSetLocaleModifiers(buf
)) != NULL
&& *p
!= NUL
5162 && (xim
= XOpenIM(x11_display
, NULL
, NULL
, NULL
))
5171 if (xim
== NULL
&& (p
= XSetLocaleModifiers("")) != NULL
&& *p
!= NUL
)
5172 xim
= XOpenIM(x11_display
, NULL
, NULL
, NULL
);
5174 /* This is supposed to be useful to obtain characters through
5175 * XmbLookupString() without really using a XIM. */
5176 if (xim
== NULL
&& (p
= XSetLocaleModifiers("@im=none")) != NULL
5178 xim
= XOpenIM(x11_display
, NULL
, NULL
, NULL
);
5182 /* Only give this message when verbose is set, because too many people
5183 * got this message when they didn't want to use a XIM. */
5187 EMSG(_("E286: Failed to open input method"));
5193 #ifdef USE_X11R6_XIM
5195 XIMCallback destroy_cb
;
5197 destroy_cb
.callback
= xim_destroy_cb
;
5198 destroy_cb
.client_data
= NULL
;
5199 if (XSetIMValues(xim
, XNDestroyCallback
, &destroy_cb
, NULL
))
5200 EMSG(_("E287: Warning: Could not set destroy callback to IM"));
5204 if (XGetIMValues(xim
, XNQueryInputStyle
, &xim_styles
, NULL
) || !xim_styles
)
5206 EMSG(_("E288: input method doesn't support any style"));
5212 strcpy(tmp
, gui
.rsrc_preedit_type_name
);
5213 for (s
= tmp
; s
&& !found
; )
5215 while (*s
&& isspace((unsigned char)*s
))
5219 if ((ns
= end
= strchr(s
, ',')) != 0)
5222 end
= s
+ strlen(s
);
5223 while (isspace((unsigned char)*end
))
5227 if (!strcmp(s
, "OverTheSpot"))
5228 this_input_style
= (XIMPreeditPosition
| XIMStatusArea
);
5229 else if (!strcmp(s
, "OffTheSpot"))
5230 this_input_style
= (XIMPreeditArea
| XIMStatusArea
);
5231 else if (!strcmp(s
, "Root"))
5232 this_input_style
= (XIMPreeditNothing
| XIMStatusNothing
);
5234 for (i
= 0; (unsigned short)i
< xim_styles
->count_styles
; i
++)
5236 if (this_input_style
== xim_styles
->supported_styles
[i
])
5243 for (i
= 0; (unsigned short)i
< xim_styles
->count_styles
; i
++)
5245 if ((xim_styles
->supported_styles
[i
] & this_input_style
)
5246 == (this_input_style
& ~XIMStatusArea
))
5248 this_input_style
&= ~XIMStatusArea
;
5260 /* Only give this message when verbose is set, because too many people
5261 * got this message when they didn't want to use a XIM. */
5265 EMSG(_("E289: input method doesn't support my preedit type"));
5272 over_spot
.x
= TEXT_X(gui
.col
);
5273 over_spot
.y
= TEXT_Y(gui
.row
);
5274 input_style
= this_input_style
;
5276 /* A crash was reported when trying to pass gui.norm_font as XNFontSet,
5277 * thus that has been removed. Hopefully the default works... */
5278 #ifdef FEAT_XFONTSET
5279 if (gui
.fontset
!= NOFONTSET
)
5281 preedit_list
= XVaCreateNestedList(0,
5282 XNSpotLocation
, &over_spot
,
5283 XNForeground
, (Pixel
)gui
.def_norm_pixel
,
5284 XNBackground
, (Pixel
)gui
.def_back_pixel
,
5285 XNFontSet
, (XFontSet
)gui
.fontset
,
5287 status_list
= XVaCreateNestedList(0,
5288 XNForeground
, (Pixel
)gui
.def_norm_pixel
,
5289 XNBackground
, (Pixel
)gui
.def_back_pixel
,
5290 XNFontSet
, (XFontSet
)gui
.fontset
,
5296 preedit_list
= XVaCreateNestedList(0,
5297 XNSpotLocation
, &over_spot
,
5298 XNForeground
, (Pixel
)gui
.def_norm_pixel
,
5299 XNBackground
, (Pixel
)gui
.def_back_pixel
,
5301 status_list
= XVaCreateNestedList(0,
5302 XNForeground
, (Pixel
)gui
.def_norm_pixel
,
5303 XNBackground
, (Pixel
)gui
.def_back_pixel
,
5307 xic
= XCreateIC(xim
,
5308 XNInputStyle
, input_style
,
5309 XNClientWindow
, x11_window
,
5310 XNFocusWindow
, gui
.wid
,
5311 XNPreeditAttributes
, preedit_list
,
5312 XNStatusAttributes
, status_list
,
5315 XFree(preedit_list
);
5318 if (input_style
& XIMStatusArea
)
5320 xim_set_status_area();
5321 status_area_enabled
= TRUE
;
5324 gui_set_shellsize(FALSE
, FALSE
, RESIZE_BOTH
);
5336 #endif /* FEAT_GUI_X11 */
5338 #if defined(FEAT_GUI_GTK) || defined(PROTO)
5340 # ifdef FEAT_XFONTSET
5341 static char e_overthespot
[] = N_("E290: over-the-spot style requires fontset");
5349 xim_decide_input_style()
5351 /* GDK_IM_STATUS_CALLBACKS was disabled, enabled it to allow Japanese
5353 int supported_style
= (int)GDK_IM_PREEDIT_NONE
|
5354 (int)GDK_IM_PREEDIT_NOTHING
|
5355 (int)GDK_IM_PREEDIT_POSITION
|
5356 (int)GDK_IM_PREEDIT_CALLBACKS
|
5357 (int)GDK_IM_STATUS_CALLBACKS
|
5358 (int)GDK_IM_STATUS_AREA
|
5359 (int)GDK_IM_STATUS_NONE
|
5360 (int)GDK_IM_STATUS_NOTHING
;
5363 xim_log("xim_decide_input_style()\n");
5366 if (!gdk_im_ready())
5367 xim_input_style
= 0;
5370 if (gtk_major_version
> 1
5371 || (gtk_major_version
== 1
5372 && (gtk_minor_version
> 2
5373 || (gtk_minor_version
== 2 && gtk_micro_version
>= 3))))
5374 use_status_area
= TRUE
;
5377 EMSG(_("E291: Your GTK+ is older than 1.2.3. Status area disabled"));
5378 use_status_area
= FALSE
;
5380 #ifdef FEAT_XFONTSET
5381 if (gui
.fontset
== NOFONTSET
|| gui
.fontset
->type
!= GDK_FONT_FONTSET
)
5383 supported_style
&= ~((int)GDK_IM_PREEDIT_POSITION
5384 | (int)GDK_IM_STATUS_AREA
);
5385 if (!use_status_area
)
5386 supported_style
&= ~(int)GDK_IM_STATUS_AREA
;
5387 xim_input_style
= (int)gdk_im_decide_style((GdkIMStyle
)supported_style
);
5392 preedit_start_cbproc(XIC thexic UNUSED
,
5393 XPointer client_data UNUSED
,
5394 XPointer call_data UNUSED
)
5397 xim_log("xim_decide_input_style()\n");
5400 draw_feedback
= NULL
;
5401 xim_can_preediting
= TRUE
;
5402 xim_has_preediting
= TRUE
;
5403 gui_update_cursor(TRUE
, FALSE
);
5412 xim_back_delete(int n
)
5420 add_to_input_buf(str
, 3);
5423 static GSList
*key_press_event_queue
= NULL
;
5424 static gboolean processing_queued_event
= FALSE
;
5427 preedit_draw_cbproc(XIC thexic UNUSED
,
5428 XPointer client_data UNUSED
,
5431 XIMPreeditDrawCallbackStruct
*draw_data
;
5434 GSList
*event_queue
;
5437 xim_log("preedit_draw_cbproc()\n");
5440 draw_data
= (XIMPreeditDrawCallbackStruct
*) call_data
;
5441 text
= (XIMText
*) draw_data
->text
;
5443 if ((text
== NULL
&& draw_data
->chg_length
== preedit_buf_len
)
5444 || preedit_buf_len
== 0)
5446 init_preedit_start_col();
5447 vim_free(draw_feedback
);
5448 draw_feedback
= NULL
;
5450 if (draw_data
->chg_length
> 0)
5454 if (draw_data
->chg_length
> preedit_buf_len
)
5455 bs_cnt
= preedit_buf_len
;
5457 bs_cnt
= draw_data
->chg_length
;
5458 xim_back_delete(bs_cnt
);
5459 preedit_buf_len
-= bs_cnt
;
5466 unsigned int nfeedback
= 0;
5470 src
= text
->string
.multi_byte
;
5471 if (src
!= NULL
&& !text
->encoding_is_wchar
)
5474 ptr
= (char_u
*)src
;
5475 /* Avoid the enter for decision */
5480 if (input_conv
.vc_type
!= CONV_NONE
5481 && (buf
= string_convert(&input_conv
,
5482 (char_u
*)src
, &len
)) != NULL
)
5484 /* Converted from 'termencoding' to 'encoding'. */
5485 add_to_input_buf_csi(buf
, len
);
5490 add_to_input_buf_csi((char_u
*)src
, len
);
5491 /* Add count of character to preedit_buf_len */
5495 if (draw_data
->text
->feedback
!= NULL
)
5497 if (draw_feedback
== NULL
)
5498 draw_feedback
= (char *)alloc(draw_data
->chg_first
5501 draw_feedback
= vim_realloc(draw_feedback
,
5502 draw_data
->chg_first
+ text
->length
);
5503 if (draw_feedback
!= NULL
)
5505 draw_feedback
[nfeedback
+ draw_data
->chg_first
]
5506 = draw_data
->text
->feedback
[nfeedback
];
5511 ptr
+= (*mb_ptr2len
)(ptr
);
5520 preedit_end_col
= MAXCOL
;
5523 if (text
!= NULL
|| draw_data
->chg_length
> 0)
5525 event_queue
= key_press_event_queue
;
5526 processing_queued_event
= TRUE
;
5527 while (event_queue
!= NULL
&& processing_queued_event
)
5529 GdkEvent
*ev
= event_queue
->data
;
5532 gtk_signal_emit_by_name((GtkObject
*)gui
.mainwin
, "key_press_event",
5535 event_queue
= event_queue
->next
;
5537 processing_queued_event
= FALSE
;
5538 if (key_press_event_queue
)
5540 g_slist_free(key_press_event_queue
);
5541 key_press_event_queue
= NULL
;
5544 if (gtk_main_level() > 0)
5549 * Retrieve the highlighting attributes at column col in the preedit string.
5550 * Return -1 if not in preediting mode or if col is out of range.
5553 im_get_feedback_attr(int col
)
5555 if (draw_feedback
!= NULL
&& col
< preedit_buf_len
)
5557 if (draw_feedback
[col
] & XIMReverse
)
5559 else if (draw_feedback
[col
] & XIMUnderline
)
5560 return HL_UNDERLINE
;
5562 return hl_attr(HLF_V
);
5569 preedit_caret_cbproc(XIC thexic UNUSED
,
5570 XPointer client_data UNUSED
,
5571 XPointer call_data UNUSED
)
5574 xim_log("preedit_caret_cbproc()\n");
5579 preedit_done_cbproc(XIC thexic UNUSED
,
5580 XPointer client_data UNUSED
,
5581 XPointer call_data UNUSED
)
5584 xim_log("preedit_done_cbproc()\n");
5587 vim_free(draw_feedback
);
5588 draw_feedback
= NULL
;
5589 xim_can_preediting
= FALSE
;
5590 xim_has_preediting
= FALSE
;
5591 gui_update_cursor(TRUE
, FALSE
);
5605 xim_log("xim_reset()\n");
5610 text
= XmbResetIC(((GdkICPrivate
*)xic
)->xic
);
5611 if (text
!= NULL
&& !(xim_input_style
& (int)GDK_IM_PREEDIT_CALLBACKS
))
5612 add_to_input_buf_csi((char_u
*)text
, strlen(text
));
5614 preedit_buf_len
= 0;
5621 xim_queue_key_press_event(GdkEventKey
*event
, int down UNUSED
)
5624 xim_log("xim_queue_key_press_event()\n");
5627 if (preedit_buf_len
<= 0)
5629 if (processing_queued_event
)
5630 processing_queued_event
= FALSE
;
5632 key_press_event_queue
= g_slist_append(key_press_event_queue
,
5633 gdk_event_copy((GdkEvent
*)event
));
5638 preedit_callback_setup(GdkIC
*ic UNUSED
)
5641 XVaNestedList preedit_attr
;
5642 XIMCallback preedit_start_cb
;
5643 XIMCallback preedit_draw_cb
;
5644 XIMCallback preedit_caret_cb
;
5645 XIMCallback preedit_done_cb
;
5647 xxic
= ((GdkICPrivate
*)xic
)->xic
;
5648 preedit_start_cb
.callback
= (XIMProc
)preedit_start_cbproc
;
5649 preedit_draw_cb
.callback
= (XIMProc
)preedit_draw_cbproc
;
5650 preedit_caret_cb
.callback
= (XIMProc
)preedit_caret_cbproc
;
5651 preedit_done_cb
.callback
= (XIMProc
)preedit_done_cbproc
;
5653 = XVaCreateNestedList(0,
5654 XNPreeditStartCallback
, &preedit_start_cb
,
5655 XNPreeditDrawCallback
, &preedit_draw_cb
,
5656 XNPreeditCaretCallback
, &preedit_caret_cb
,
5657 XNPreeditDoneCallback
, &preedit_done_cb
,
5659 XSetICValues(xxic
, XNPreeditAttributes
, preedit_attr
, NULL
);
5660 XFree(preedit_attr
);
5664 reset_state_setup(GdkIC
*ic UNUSED
)
5666 #ifdef USE_X11R6_XIM
5667 /* don't change the input context when we call reset */
5668 XSetICValues(((GdkICPrivate
*)ic
)->xic
, XNResetState
, XIMPreserveState
,
5677 xim_log("xim_init()\n");
5683 if (!gdk_im_ready())
5688 EMSG(_("E292: Input Method Server is not running"));
5693 if ((xic_attr
= gdk_ic_attr_new()) != NULL
)
5695 #ifdef FEAT_XFONTSET
5699 GdkColormap
*colormap
;
5700 GdkICAttr
*attr
= xic_attr
;
5701 int attrmask
= (int)GDK_IC_ALL_REQ
;
5702 GtkWidget
*widget
= gui
.drawarea
;
5704 attr
->style
= (GdkIMStyle
)xim_input_style
;
5705 attr
->client_window
= gui
.mainwin
->window
;
5707 if ((colormap
= gtk_widget_get_colormap(widget
)) !=
5708 gtk_widget_get_default_colormap())
5710 attrmask
|= (int)GDK_IC_PREEDIT_COLORMAP
;
5711 attr
->preedit_colormap
= colormap
;
5713 attrmask
|= (int)GDK_IC_PREEDIT_FOREGROUND
;
5714 attrmask
|= (int)GDK_IC_PREEDIT_BACKGROUND
;
5715 attr
->preedit_foreground
= widget
->style
->fg
[GTK_STATE_NORMAL
];
5716 attr
->preedit_background
= widget
->style
->base
[GTK_STATE_NORMAL
];
5718 #ifdef FEAT_XFONTSET
5719 if ((xim_input_style
& (int)GDK_IM_PREEDIT_MASK
)
5720 == (int)GDK_IM_PREEDIT_POSITION
)
5722 if (gui
.fontset
== NOFONTSET
5723 || gui
.fontset
->type
!= GDK_FONT_FONTSET
)
5725 EMSG(_(e_overthespot
));
5729 gdk_window_get_size(widget
->window
, &width
, &height
);
5731 attrmask
|= (int)GDK_IC_PREEDIT_POSITION_REQ
;
5732 attr
->spot_location
.x
= TEXT_X(0);
5733 attr
->spot_location
.y
= TEXT_Y(0);
5734 attr
->preedit_area
.x
= gui
.border_offset
;
5735 attr
->preedit_area
.y
= gui
.border_offset
;
5736 attr
->preedit_area
.width
= width
- 2*gui
.border_offset
;
5737 attr
->preedit_area
.height
= height
- 2*gui
.border_offset
;
5738 attr
->preedit_fontset
= gui
.fontset
;
5742 if ((xim_input_style
& (int)GDK_IM_STATUS_MASK
)
5743 == (int)GDK_IM_STATUS_AREA
)
5745 if (gui
.fontset
== NOFONTSET
5746 || gui
.fontset
->type
!= GDK_FONT_FONTSET
)
5748 EMSG(_(e_overthespot
));
5752 gdk_window_get_size(gui
.mainwin
->window
, &width
, &height
);
5753 attrmask
|= (int)GDK_IC_STATUS_AREA_REQ
;
5754 attr
->status_area
.x
= 0;
5755 attr
->status_area
.y
= height
- gui
.char_height
- 1;
5756 attr
->status_area
.width
= width
;
5757 attr
->status_area
.height
= gui
.char_height
;
5758 attr
->status_fontset
= gui
.fontset
;
5761 else if ((xim_input_style
& (int)GDK_IM_STATUS_MASK
)
5762 == (int)GDK_IM_STATUS_CALLBACKS
)
5768 xic
= gdk_ic_new(attr
, (GdkICAttributesType
)attrmask
);
5774 mask
= (int)gdk_window_get_events(widget
->window
);
5775 mask
|= (int)gdk_ic_get_events(xic
);
5776 gdk_window_set_events(widget
->window
, (GdkEventMask
)mask
);
5777 if (xim_input_style
& (int)GDK_IM_PREEDIT_CALLBACKS
)
5778 preedit_callback_setup(xic
);
5779 reset_state_setup(xic
);
5788 xim_log("im_shutdown()\n");
5794 gdk_ic_destroy(xic
);
5797 xim_is_active
= FALSE
;
5798 xim_can_preediting
= FALSE
;
5799 preedit_start_col
= MAXCOL
;
5800 xim_has_preediting
= FALSE
;
5803 #endif /* FEAT_GUI_GTK */
5806 xim_get_status_area_height()
5809 if (xim_input_style
& (int)GDK_IM_STATUS_AREA
)
5810 return gui
.char_height
;
5812 if (status_area_enabled
)
5813 return gui
.char_height
;
5819 * Get IM status. When IM is on, return TRUE. Else return FALSE.
5820 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is
5821 * active, when not having focus XIM may still be active (e.g., when using a
5822 * tear-off menu item).
5827 # ifdef FEAT_GUI_GTK
5828 if (xim_input_style
& (int)GDK_IM_PREEDIT_CALLBACKS
)
5829 return xim_can_preediting
;
5831 return xim_has_focus
;
5834 # endif /* !HAVE_GTK2 */
5836 # if defined(HAVE_GTK2) || defined(PROTO)
5838 preedit_get_status(void)
5840 return preedit_is_active
;
5844 # if defined(FEAT_GUI_GTK) || defined(PROTO)
5848 return xim_has_preediting
;
5851 #endif /* FEAT_XIM */
5853 #if defined(FEAT_MBYTE) || defined(PROTO)
5856 * Setup "vcp" for conversion from "from" to "to".
5857 * The names must have been made canonical with enc_canonize().
5858 * vcp->vc_type must have been initialized to CONV_NONE.
5859 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8
5861 * Afterwards invoke with "from" and "to" equal to NULL to cleanup.
5862 * Return FAIL when conversion is not supported, OK otherwise.
5865 convert_setup(vcp
, from
, to
)
5870 return convert_setup_ext(vcp
, from
, TRUE
, to
, TRUE
);
5874 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all
5875 * "from" unicode charsets be considered utf-8. Same for "to".
5878 convert_setup_ext(vcp
, from
, from_unicode_is_utf8
, to
, to_unicode_is_utf8
)
5881 int from_unicode_is_utf8
;
5883 int to_unicode_is_utf8
;
5890 /* Reset to no conversion. */
5892 if (vcp
->vc_type
== CONV_ICONV
&& vcp
->vc_fd
!= (iconv_t
)-1)
5893 iconv_close(vcp
->vc_fd
);
5895 vcp
->vc_type
= CONV_NONE
;
5897 vcp
->vc_fail
= FALSE
;
5899 /* No conversion when one of the names is empty or they are equal. */
5900 if (from
== NULL
|| *from
== NUL
|| to
== NULL
|| *to
== NUL
5901 || STRCMP(from
, to
) == 0)
5904 from_prop
= enc_canon_props(from
);
5905 to_prop
= enc_canon_props(to
);
5906 if (from_unicode_is_utf8
)
5907 from_is_utf8
= from_prop
& ENC_UNICODE
;
5909 from_is_utf8
= from_prop
== ENC_UNICODE
;
5910 if (to_unicode_is_utf8
)
5911 to_is_utf8
= to_prop
& ENC_UNICODE
;
5913 to_is_utf8
= to_prop
== ENC_UNICODE
;
5915 if ((from_prop
& ENC_LATIN1
) && to_is_utf8
)
5917 /* Internal latin1 -> utf-8 conversion. */
5918 vcp
->vc_type
= CONV_TO_UTF8
;
5919 vcp
->vc_factor
= 2; /* up to twice as long */
5921 else if ((from_prop
& ENC_LATIN9
) && to_is_utf8
)
5923 /* Internal latin9 -> utf-8 conversion. */
5924 vcp
->vc_type
= CONV_9_TO_UTF8
;
5925 vcp
->vc_factor
= 3; /* up to three as long (euro sign) */
5927 else if (from_is_utf8
&& (to_prop
& ENC_LATIN1
))
5929 /* Internal utf-8 -> latin1 conversion. */
5930 vcp
->vc_type
= CONV_TO_LATIN1
;
5932 else if (from_is_utf8
&& (to_prop
& ENC_LATIN9
))
5934 /* Internal utf-8 -> latin9 conversion. */
5935 vcp
->vc_type
= CONV_TO_LATIN9
;
5938 /* Win32-specific codepage <-> codepage conversion without iconv. */
5939 else if ((from_is_utf8
|| encname2codepage(from
) > 0)
5940 && (to_is_utf8
|| encname2codepage(to
) > 0))
5942 vcp
->vc_type
= CONV_CODEPAGE
;
5943 vcp
->vc_factor
= 2; /* up to twice as long */
5944 vcp
->vc_cpfrom
= from_is_utf8
? 0 : encname2codepage(from
);
5945 vcp
->vc_cpto
= to_is_utf8
? 0 : encname2codepage(to
);
5949 else if ((from_prop
& ENC_MACROMAN
) && (to_prop
& ENC_LATIN1
))
5951 vcp
->vc_type
= CONV_MAC_LATIN1
;
5953 else if ((from_prop
& ENC_MACROMAN
) && to_is_utf8
)
5955 vcp
->vc_type
= CONV_MAC_UTF8
;
5956 vcp
->vc_factor
= 2; /* up to twice as long */
5958 else if ((from_prop
& ENC_LATIN1
) && (to_prop
& ENC_MACROMAN
))
5960 vcp
->vc_type
= CONV_LATIN1_MAC
;
5962 else if (from_is_utf8
&& (to_prop
& ENC_MACROMAN
))
5964 vcp
->vc_type
= CONV_UTF8_MAC
;
5970 /* Use iconv() for conversion. */
5971 vcp
->vc_fd
= (iconv_t
)my_iconv_open(
5972 to_is_utf8
? (char_u
*)"utf-8" : to
,
5973 from_is_utf8
? (char_u
*)"utf-8" : from
);
5974 if (vcp
->vc_fd
!= (iconv_t
)-1)
5976 vcp
->vc_type
= CONV_ICONV
;
5977 vcp
->vc_factor
= 4; /* could be longer too... */
5981 if (vcp
->vc_type
== CONV_NONE
)
5987 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \
5988 || defined(MSDOS) || defined(PROTO)
5990 * Do conversion on typed input characters in-place.
5991 * The input and output are not NUL terminated!
5992 * Returns the length after conversion.
5995 convert_input(ptr
, len
, maxlen
)
6000 return convert_input_safe(ptr
, len
, maxlen
, NULL
, NULL
);
6005 * Like convert_input(), but when there is an incomplete byte sequence at the
6006 * end return that as an allocated string in "restp" and set "*restlenp" to
6007 * the length. If "restp" is NULL it is not used.
6010 convert_input_safe(ptr
, len
, maxlen
, restp
, restlenp
)
6019 int unconvertlen
= 0;
6021 d
= string_convert_ext(&input_conv
, ptr
, &dlen
,
6022 restp
== NULL
? NULL
: &unconvertlen
);
6027 if (unconvertlen
> 0)
6029 /* Move the unconverted characters to allocated memory. */
6030 *restp
= alloc(unconvertlen
);
6032 mch_memmove(*restp
, ptr
+ len
- unconvertlen
, unconvertlen
);
6033 *restlenp
= unconvertlen
;
6035 mch_memmove(ptr
, d
, dlen
);
6038 /* result is too long, keep the unconverted text (the caller must
6039 * have done something wrong!) */
6047 * Convert text "ptr[*lenp]" according to "vcp".
6048 * Returns the result in allocated memory and sets "*lenp".
6049 * When "lenp" is NULL, use NUL terminated strings.
6050 * Illegal chars are often changed to "?", unless vcp->vc_fail is set.
6051 * When something goes wrong, NULL is returned and "*lenp" is unchanged.
6054 string_convert(vcp
, ptr
, lenp
)
6059 return string_convert_ext(vcp
, ptr
, lenp
, NULL
);
6063 * Like string_convert(), but when "unconvlenp" is not NULL and there are is
6064 * an incomplete sequence at the end it is not converted and "*unconvlenp" is
6065 * set to the number of remaining bytes.
6068 string_convert_ext(vcp
, ptr
, lenp
, unconvlenp
)
6074 char_u
*retval
= NULL
;
6082 len
= (int)STRLEN(ptr
);
6086 return vim_strsave((char_u
*)"");
6088 switch (vcp
->vc_type
)
6090 case CONV_TO_UTF8
: /* latin1 to utf-8 conversion */
6091 retval
= alloc(len
* 2 + 1);
6095 for (i
= 0; i
< len
; ++i
)
6102 *d
++ = 0xc0 + ((unsigned)c
>> 6);
6103 *d
++ = 0x80 + (c
& 0x3f);
6108 *lenp
= (int)(d
- retval
);
6111 case CONV_9_TO_UTF8
: /* latin9 to utf-8 conversion */
6112 retval
= alloc(len
* 3 + 1);
6116 for (i
= 0; i
< len
; ++i
)
6121 case 0xa4: c
= 0x20ac; break; /* euro */
6122 case 0xa6: c
= 0x0160; break; /* S hat */
6123 case 0xa8: c
= 0x0161; break; /* S -hat */
6124 case 0xb4: c
= 0x017d; break; /* Z hat */
6125 case 0xb8: c
= 0x017e; break; /* Z -hat */
6126 case 0xbc: c
= 0x0152; break; /* OE */
6127 case 0xbd: c
= 0x0153; break; /* oe */
6128 case 0xbe: c
= 0x0178; break; /* Y */
6130 d
+= utf_char2bytes(c
, d
);
6134 *lenp
= (int)(d
- retval
);
6137 case CONV_TO_LATIN1
: /* utf-8 to latin1 conversion */
6138 case CONV_TO_LATIN9
: /* utf-8 to latin9 conversion */
6139 retval
= alloc(len
+ 1);
6143 for (i
= 0; i
< len
; ++i
)
6145 l
= utf_ptr2len_len(ptr
+ i
, len
- i
);
6150 int l_w
= utf8len_tab_zero
[ptr
[i
]];
6154 /* Illegal utf-8 byte cannot be converted */
6158 if (unconvlenp
!= NULL
&& l_w
> len
- i
)
6160 /* Incomplete sequence at the end. */
6161 *unconvlenp
= len
- i
;
6168 c
= utf_ptr2char(ptr
+ i
);
6169 if (vcp
->vc_type
== CONV_TO_LATIN9
)
6172 case 0x20ac: c
= 0xa4; break; /* euro */
6173 case 0x0160: c
= 0xa6; break; /* S hat */
6174 case 0x0161: c
= 0xa8; break; /* S -hat */
6175 case 0x017d: c
= 0xb4; break; /* Z hat */
6176 case 0x017e: c
= 0xb8; break; /* Z -hat */
6177 case 0x0152: c
= 0xbc; break; /* OE */
6178 case 0x0153: c
= 0xbd; break; /* oe */
6179 case 0x0178: c
= 0xbe; break; /* Y */
6187 case 0xbe: c
= 0x100; break; /* not in latin9 */
6189 if (!utf_iscomposing(c
)) /* skip composing chars */
6193 else if (vcp
->vc_fail
)
6201 if (utf_char2cells(c
) > 1)
6210 *lenp
= (int)(d
- retval
);
6213 # ifdef MACOS_CONVERT
6214 case CONV_MAC_LATIN1
:
6215 retval
= mac_string_convert(ptr
, len
, lenp
, vcp
->vc_fail
,
6216 'm', 'l', unconvlenp
);
6219 case CONV_LATIN1_MAC
:
6220 retval
= mac_string_convert(ptr
, len
, lenp
, vcp
->vc_fail
,
6221 'l', 'm', unconvlenp
);
6225 retval
= mac_string_convert(ptr
, len
, lenp
, vcp
->vc_fail
,
6226 'm', 'u', unconvlenp
);
6230 retval
= mac_string_convert(ptr
, len
, lenp
, vcp
->vc_fail
,
6231 'u', 'm', unconvlenp
);
6236 case CONV_ICONV
: /* conversion with output_conv.vc_fd */
6237 retval
= iconv_string(vcp
, ptr
, len
, unconvlenp
, lenp
);
6241 case CONV_CODEPAGE
: /* codepage -> codepage */
6247 /* 1. codepage/UTF-8 -> ucs-2. */
6248 if (vcp
->vc_cpfrom
== 0)
6249 tmp_len
= utf8_to_utf16(ptr
, len
, NULL
, NULL
);
6251 tmp_len
= MultiByteToWideChar(vcp
->vc_cpfrom
, 0,
6253 tmp
= (short_u
*)alloc(sizeof(short_u
) * tmp_len
);
6256 if (vcp
->vc_cpfrom
== 0)
6257 utf8_to_utf16(ptr
, len
, tmp
, unconvlenp
);
6259 MultiByteToWideChar(vcp
->vc_cpfrom
, 0, ptr
, len
, tmp
, tmp_len
);
6261 /* 2. ucs-2 -> codepage/UTF-8. */
6262 if (vcp
->vc_cpto
== 0)
6263 retlen
= utf16_to_utf8(tmp
, tmp_len
, NULL
);
6265 retlen
= WideCharToMultiByte(vcp
->vc_cpto
, 0,
6266 tmp
, tmp_len
, 0, 0, 0, 0);
6267 retval
= alloc(retlen
+ 1);
6270 if (vcp
->vc_cpto
== 0)
6271 utf16_to_utf8(tmp
, tmp_len
, retval
);
6273 WideCharToMultiByte(vcp
->vc_cpto
, 0,
6274 tmp
, tmp_len
, retval
, retlen
, 0, 0);
6275 retval
[retlen
] = NUL
;