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_char2cells
__ARGS((int c
));
131 static int dbcs_ptr2char
__ARGS((char_u
*p
));
133 /* Lookup table to quickly get the length in bytes of a UTF-8 character from
134 * the first byte of a UTF-8 string. Bytes which are illegal when used as the
135 * first byte have a one, because these will be used separately. */
136 static char utf8len_tab
[256] =
138 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
139 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
140 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
141 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
142 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /*bogus*/
143 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /*bogus*/
144 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
145 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6,1,1,
149 * XIM often causes trouble. Define XIM_DEBUG to get a log of XIM callbacks
150 * in the "xim.log" file.
152 /* #define XIM_DEBUG */
155 xim_log(char *s
, ...)
158 static FILE *fd
= NULL
;
160 if (fd
== (FILE *)-1)
164 fd
= mch_fopen("xim.log", "w");
167 EMSG("Cannot open xim.log");
173 va_start(arglist
, s
);
174 vfprintf(fd
, s
, arglist
);
181 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO)
183 * Canonical encoding names and their properties.
184 * "iso-8859-n" is handled by enc_canonize() directly.
187 { char *name
; int prop
; int codepage
;}
190 #define IDX_LATIN_1 0
191 {"latin1", ENC_8BIT
+ ENC_LATIN1
, 1252},
193 {"iso-8859-2", ENC_8BIT
, 0},
195 {"iso-8859-3", ENC_8BIT
, 0},
197 {"iso-8859-4", ENC_8BIT
, 0},
199 {"iso-8859-5", ENC_8BIT
, 0},
201 {"iso-8859-6", ENC_8BIT
, 0},
203 {"iso-8859-7", ENC_8BIT
, 0},
205 {"iso-8859-8", ENC_8BIT
, 0},
207 {"iso-8859-9", ENC_8BIT
, 0},
209 {"iso-8859-10", ENC_8BIT
, 0},
210 #define IDX_ISO_11 10
211 {"iso-8859-11", ENC_8BIT
, 0},
212 #define IDX_ISO_13 11
213 {"iso-8859-13", ENC_8BIT
, 0},
214 #define IDX_ISO_14 12
215 {"iso-8859-14", ENC_8BIT
, 0},
216 #define IDX_ISO_15 13
217 {"iso-8859-15", ENC_8BIT
+ ENC_LATIN9
, 0},
218 #define IDX_KOI8_R 14
219 {"koi8-r", ENC_8BIT
, 0},
220 #define IDX_KOI8_U 15
221 {"koi8-u", ENC_8BIT
, 0},
223 {"utf-8", ENC_UNICODE
, 0},
225 {"ucs-2", ENC_UNICODE
+ ENC_ENDIAN_B
+ ENC_2BYTE
, 0},
226 #define IDX_UCS2LE 18
227 {"ucs-2le", ENC_UNICODE
+ ENC_ENDIAN_L
+ ENC_2BYTE
, 0},
229 {"utf-16", ENC_UNICODE
+ ENC_ENDIAN_B
+ ENC_2WORD
, 0},
230 #define IDX_UTF16LE 20
231 {"utf-16le", ENC_UNICODE
+ ENC_ENDIAN_L
+ ENC_2WORD
, 0},
233 {"ucs-4", ENC_UNICODE
+ ENC_ENDIAN_B
+ ENC_4BYTE
, 0},
234 #define IDX_UCS4LE 22
235 {"ucs-4le", ENC_UNICODE
+ ENC_ENDIAN_L
+ ENC_4BYTE
, 0},
237 /* For debugging DBCS encoding on Unix. */
239 {"debug", ENC_DBCS
, DBCS_DEBUG
},
240 #define IDX_EUC_JP 24
241 {"euc-jp", ENC_DBCS
, DBCS_JPNU
},
243 {"sjis", ENC_DBCS
, DBCS_JPN
},
244 #define IDX_EUC_KR 26
245 {"euc-kr", ENC_DBCS
, DBCS_KORU
},
246 #define IDX_EUC_CN 27
247 {"euc-cn", ENC_DBCS
, DBCS_CHSU
},
248 #define IDX_EUC_TW 28
249 {"euc-tw", ENC_DBCS
, DBCS_CHTU
},
251 {"big5", ENC_DBCS
, DBCS_CHT
},
253 /* MS-DOS and MS-Windows codepages are included here, so that they can be
254 * used on Unix too. Most of them are similar to ISO-8859 encodings, but
255 * not exactly the same. */
257 {"cp437", ENC_8BIT
, 437}, /* like iso-8859-1 */
259 {"cp737", ENC_8BIT
, 737}, /* like iso-8859-7 */
261 {"cp775", ENC_8BIT
, 775}, /* Baltic */
263 {"cp850", ENC_8BIT
, 850}, /* like iso-8859-4 */
265 {"cp852", ENC_8BIT
, 852}, /* like iso-8859-1 */
267 {"cp855", ENC_8BIT
, 855}, /* like iso-8859-2 */
269 {"cp857", ENC_8BIT
, 857}, /* like iso-8859-5 */
271 {"cp860", ENC_8BIT
, 860}, /* like iso-8859-9 */
273 {"cp861", ENC_8BIT
, 861}, /* like iso-8859-1 */
275 {"cp862", ENC_8BIT
, 862}, /* like iso-8859-1 */
277 {"cp863", ENC_8BIT
, 863}, /* like iso-8859-8 */
279 {"cp865", ENC_8BIT
, 865}, /* like iso-8859-1 */
281 {"cp866", ENC_8BIT
, 866}, /* like iso-8859-5 */
283 {"cp869", ENC_8BIT
, 869}, /* like iso-8859-7 */
285 {"cp874", ENC_8BIT
, 874}, /* Thai */
287 {"cp932", ENC_DBCS
, DBCS_JPN
},
289 {"cp936", ENC_DBCS
, DBCS_CHS
},
291 {"cp949", ENC_DBCS
, DBCS_KOR
},
293 {"cp950", ENC_DBCS
, DBCS_CHT
},
294 #define IDX_CP1250 49
295 {"cp1250", ENC_8BIT
, 1250}, /* Czech, Polish, etc. */
296 #define IDX_CP1251 50
297 {"cp1251", ENC_8BIT
, 1251}, /* Cyrillic */
298 /* cp1252 is considered to be equal to latin1 */
299 #define IDX_CP1253 51
300 {"cp1253", ENC_8BIT
, 1253}, /* Greek */
301 #define IDX_CP1254 52
302 {"cp1254", ENC_8BIT
, 1254}, /* Turkish */
303 #define IDX_CP1255 53
304 {"cp1255", ENC_8BIT
, 1255}, /* Hebrew */
305 #define IDX_CP1256 54
306 {"cp1256", ENC_8BIT
, 1256}, /* Arabic */
307 #define IDX_CP1257 55
308 {"cp1257", ENC_8BIT
, 1257}, /* Baltic */
309 #define IDX_CP1258 56
310 {"cp1258", ENC_8BIT
, 1258}, /* Vietnamese */
312 #define IDX_MACROMAN 57
313 {"macroman", ENC_8BIT
+ ENC_MACROMAN
, 0}, /* Mac OS */
314 #define IDX_DECMCS 58
315 {"dec-mcs", ENC_8BIT
, 0}, /* DEC MCS */
316 #define IDX_HPROMAN8 59
317 {"hp-roman8", ENC_8BIT
, 0}, /* HP Roman8 */
322 * Aliases for encoding names.
325 { char *name
; int canon
;}
328 {"ansi", IDX_LATIN_1
},
329 {"iso-8859-1", IDX_LATIN_1
},
330 {"latin2", IDX_ISO_2
},
331 {"latin3", IDX_ISO_3
},
332 {"latin4", IDX_ISO_4
},
333 {"cyrillic", IDX_ISO_5
},
334 {"arabic", IDX_ISO_6
},
335 {"greek", IDX_ISO_7
},
337 {"hebrew", IDX_CP1255
},
339 {"hebrew", IDX_ISO_8
},
341 {"latin5", IDX_ISO_9
},
342 {"turkish", IDX_ISO_9
}, /* ? */
343 {"latin6", IDX_ISO_10
},
344 {"nordic", IDX_ISO_10
}, /* ? */
345 {"thai", IDX_ISO_11
}, /* ? */
346 {"latin7", IDX_ISO_13
},
347 {"latin8", IDX_ISO_14
},
348 {"latin9", IDX_ISO_15
},
350 {"unicode", IDX_UCS2
},
352 {"ucs2be", IDX_UCS2
},
353 {"ucs-2be", IDX_UCS2
},
354 {"ucs2le", IDX_UCS2LE
},
355 {"utf16", IDX_UTF16
},
356 {"utf16be", IDX_UTF16
},
357 {"utf-16be", IDX_UTF16
},
358 {"utf16le", IDX_UTF16LE
},
360 {"ucs4be", IDX_UCS4
},
361 {"ucs-4be", IDX_UCS4
},
362 {"ucs4le", IDX_UCS4LE
},
364 {"utf-32", IDX_UCS4
},
365 {"utf32be", IDX_UCS4
},
366 {"utf-32be", IDX_UCS4
},
367 {"utf32le", IDX_UCS4LE
},
368 {"utf-32le", IDX_UCS4LE
},
374 {"eucjp", IDX_EUC_JP
},
375 {"unix-jis", IDX_EUC_JP
},
376 {"ujis", IDX_EUC_JP
},
377 {"shift-jis", IDX_SJIS
},
378 {"euckr", IDX_EUC_KR
},
379 {"5601", IDX_EUC_KR
}, /* Sun: KS C 5601 */
380 {"euccn", IDX_EUC_CN
},
381 {"gb2312", IDX_EUC_CN
},
382 {"euctw", IDX_EUC_TW
},
383 #if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS)
384 {"japan", IDX_CP932
},
385 {"korea", IDX_CP949
},
387 {"chinese", IDX_CP936
},
388 {"taiwan", IDX_CP950
},
391 {"japan", IDX_EUC_JP
},
392 {"korea", IDX_EUC_KR
},
394 {"chinese", IDX_EUC_CN
},
395 {"taiwan", IDX_EUC_TW
},
399 {"mac", IDX_MACROMAN
},
400 {"mac-roman", IDX_MACROMAN
},
405 # define CP_UTF8 65001 /* magic number from winnls.h */
409 * Find encoding "name" in the list of canonical encoding names.
410 * Returns -1 if not found.
413 enc_canon_search(name
)
418 for (i
= 0; i
< IDX_COUNT
; ++i
)
419 if (STRCMP(name
, enc_canon_table
[i
].name
) == 0)
426 #if defined(FEAT_MBYTE) || defined(PROTO)
429 * Find canonical encoding "name" in the list and return its properties.
430 * Returns 0 if not found.
433 enc_canon_props(name
)
438 i
= enc_canon_search(name
);
440 return enc_canon_table
[i
].prop
;
442 if (name
[0] == 'c' && name
[1] == 'p' && VIM_ISDIGIT(name
[2]))
446 /* Get info on this codepage to find out what it is. */
447 if (GetCPInfo(atoi(name
+ 2), &cpinfo
) != 0)
449 if (cpinfo
.MaxCharSize
== 1) /* some single-byte encoding */
451 if (cpinfo
.MaxCharSize
== 2
452 && (cpinfo
.LeadByte
[0] != 0 || cpinfo
.LeadByte
[1] != 0))
453 /* must be a DBCS encoding */
459 if (STRNCMP(name
, "2byte-", 6) == 0)
461 if (STRNCMP(name
, "8bit-", 5) == 0 || STRNCMP(name
, "iso-8859-", 9) == 0)
467 * Set up for using multi-byte characters.
468 * Called in three cases:
469 * - by main() to initialize (p_enc == NULL)
470 * - by set_init_1() after 'encoding' was set to its default.
471 * - by do_set() when 'encoding' has been set.
472 * p_enc must have been passed through enc_canonize() already.
473 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags.
474 * Fills mb_bytelen_tab[] and returns NULL when there are no problems.
475 * When there is something wrong: Returns an error message and doesn't change
484 int enc_dbcs_new
= 0;
485 #if defined(USE_ICONV) && !defined(WIN3264) && !defined(WIN32UNIX) \
487 # define LEN_FROM_CONV
494 /* Just starting up: set the whole table to one's. */
495 for (i
= 0; i
< 256; ++i
)
496 mb_bytelen_tab
[i
] = 1;
497 input_conv
.vc_type
= CONV_NONE
;
498 input_conv
.vc_factor
= 1;
499 output_conv
.vc_type
= CONV_NONE
;
504 if (p_enc
[0] == 'c' && p_enc
[1] == 'p' && VIM_ISDIGIT(p_enc
[2]))
508 /* Get info on this codepage to find out what it is. */
509 if (GetCPInfo(atoi(p_enc
+ 2), &cpinfo
) != 0)
511 if (cpinfo
.MaxCharSize
== 1)
513 /* some single-byte encoding */
517 else if (cpinfo
.MaxCharSize
== 2
518 && (cpinfo
.LeadByte
[0] != 0 || cpinfo
.LeadByte
[1] != 0))
520 /* must be a DBCS encoding, check below */
521 enc_dbcs_new
= atoi(p_enc
+ 2);
524 goto codepage_invalid
;
526 else if (GetLastError() == ERROR_INVALID_PARAMETER
)
529 return (char_u
*)N_("E543: Not a valid codepage");
533 else if (STRNCMP(p_enc
, "8bit-", 5) == 0
534 || STRNCMP(p_enc
, "iso-8859-", 9) == 0)
536 /* Accept any "8bit-" or "iso-8859-" name. */
540 else if (STRNCMP(p_enc
, "2byte-", 6) == 0)
543 /* Windows: accept only valid codepage numbers, check below. */
544 if (p_enc
[6] != 'c' || p_enc
[7] != 'p'
545 || (enc_dbcs_new
= atoi(p_enc
+ 8)) == 0)
548 /* Unix: accept any "2byte-" name, assume current locale. */
549 enc_dbcs_new
= DBCS_2BYTE
;
552 else if ((idx
= enc_canon_search(p_enc
)) >= 0)
554 i
= enc_canon_table
[idx
].prop
;
559 if (i
& (ENC_2BYTE
| ENC_2WORD
))
561 else if (i
& ENC_4BYTE
)
566 else if (i
& ENC_DBCS
)
568 /* 2byte, handle below */
569 enc_dbcs_new
= enc_canon_table
[idx
].codepage
;
578 else /* Don't know what encoding this is, reject it. */
581 if (enc_dbcs_new
!= 0)
584 /* Check if the DBCS code page is OK. */
585 if (!IsValidCodePage(enc_dbcs_new
))
586 goto codepage_invalid
;
591 enc_dbcs
= enc_dbcs_new
;
592 has_mbyte
= (enc_dbcs
!= 0 || enc_utf8
);
595 enc_codepage
= encname2codepage(p_enc
);
596 enc_latin9
= (STRCMP(p_enc
, "iso-8859-15") == 0);
599 /* Detect an encoding that uses latin1 characters. */
600 enc_latin1like
= (enc_utf8
|| STRCMP(p_enc
, "latin1") == 0
601 || STRCMP(p_enc
, "iso-8859-15") == 0);
604 * Set the function pointers.
608 mb_ptr2len
= utfc_ptr2len
;
609 mb_char2len
= utf_char2len
;
610 mb_char2bytes
= utf_char2bytes
;
611 mb_ptr2cells
= utf_ptr2cells
;
612 mb_char2cells
= utf_char2cells
;
613 mb_off2cells
= utf_off2cells
;
614 mb_ptr2char
= utf_ptr2char
;
615 mb_head_off
= utf_head_off
;
617 else if (enc_dbcs
!= 0)
619 mb_ptr2len
= dbcs_ptr2len
;
620 mb_char2len
= dbcs_char2len
;
621 mb_char2bytes
= dbcs_char2bytes
;
622 mb_ptr2cells
= dbcs_ptr2cells
;
623 mb_char2cells
= dbcs_char2cells
;
624 mb_off2cells
= dbcs_off2cells
;
625 mb_ptr2char
= dbcs_ptr2char
;
626 mb_head_off
= dbcs_head_off
;
630 mb_ptr2len
= latin_ptr2len
;
631 mb_char2len
= latin_char2len
;
632 mb_char2bytes
= latin_char2bytes
;
633 mb_ptr2cells
= latin_ptr2cells
;
634 mb_char2cells
= latin_char2cells
;
635 mb_off2cells
= latin_off2cells
;
636 mb_ptr2char
= latin_ptr2char
;
637 mb_head_off
= latin_head_off
;
641 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN().
644 /* When 'encoding' is different from the current locale mblen() won't
645 * work. Use conversion to "utf-8" instead. */
646 vimconv
.vc_type
= CONV_NONE
;
650 if (p
== NULL
|| STRCMP(p
, p_enc
) != 0)
652 convert_setup(&vimconv
, p_enc
, (char_u
*)"utf-8");
653 vimconv
.vc_fail
= TRUE
;
659 for (i
= 0; i
< 256; ++i
)
661 /* Our own function to reliably check the length of UTF-8 characters,
662 * independent of mblen(). */
665 else if (enc_dbcs
== 0)
669 #if defined(WIN3264) || defined(WIN32UNIX)
670 /* enc_dbcs is set by setting 'fileencoding'. It becomes a Windows
671 * CodePage identifier, which we can pass directly in to Windows
673 n
= IsDBCSLeadByteEx(enc_dbcs
, (BYTE
)i
) ? 2 : 1;
675 # if defined(MACOS) || defined(__amigaos4__)
677 * if mblen() is not available, character which MSB is turned on
678 * are treated as leading byte character. (note : This assumption
679 * is not always true.)
681 n
= (i
& 0x80) ? 2 : 1;
683 char buf
[MB_MAXBYTES
];
686 # define mblen _Xmblen
689 if (i
== NUL
) /* just in case mblen() can't handle "" */
696 if (vimconv
.vc_type
!= CONV_NONE
)
699 * string_convert() should fail when converting the first
700 * byte of a double-byte character.
702 p
= string_convert(&vimconv
, (char_u
*)buf
, NULL
);
715 * mblen() should return -1 for invalid (means the leading
716 * multibyte) character. However there are some platforms
717 * where mblen() returns 0 for invalid character.
718 * Therefore, following condition includes 0.
720 ignored
= mblen(NULL
, 0); /* First reset the state. */
721 if (mblen(buf
, (size_t)1) <= 0)
731 mb_bytelen_tab
[i
] = n
;
735 convert_setup(&vimconv
, NULL
, NULL
);
738 /* The cell width depends on the type of multi-byte characters. */
739 (void)init_chartab();
741 /* When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] */
744 /* When using Unicode, set default for 'fileencodings'. */
745 if (enc_utf8
&& !option_was_set((char_u
*)"fencs"))
746 set_string_option_direct((char_u
*)"fencs", -1,
747 (char_u
*)"ucs-bom,utf-8,default,latin1", OPT_FREE
, 0);
749 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT)
750 /* GNU gettext 0.10.37 supports this feature: set the codeset used for
751 * translated messages independently from the current locale. */
752 (void)bind_textdomain_codeset(VIMPACKAGE
,
753 enc_utf8
? "utf-8" : (char *)p_enc
);
757 /* When changing 'encoding' while starting up, then convert the command
758 * line arguments from the active codepage to 'encoding'. */
764 /* Fire an autocommand to let people do custom font setup. This must be
765 * after Vim has been setup for the new encoding. */
766 apply_autocmds(EVENT_ENCODINGCHANGED
, NULL
, (char_u
*)"", FALSE
, curbuf
);
770 /* Need to reload spell dictionaries */
778 * Return the size of the BOM for the current buffer:
780 * 2 - UCS-2 or UTF-16 BOM
789 if (curbuf
->b_p_bomb
&& !curbuf
->b_p_bin
)
791 if (*curbuf
->b_p_fenc
== NUL
)
795 if (enc_unicode
!= 0)
801 else if (STRCMP(curbuf
->b_p_fenc
, "utf-8") == 0)
803 else if (STRNCMP(curbuf
->b_p_fenc
, "ucs-2", 5) == 0
804 || STRNCMP(curbuf
->b_p_fenc
, "utf-16", 6) == 0)
806 else if (STRNCMP(curbuf
->b_p_fenc
, "ucs-4", 5) == 0)
813 * Get class of pointer:
816 * 2 for an (ASCII) word character
817 * >2 for other word characters
823 if (MB_BYTE2LEN(p
[0]) == 1)
825 if (p
[0] == NUL
|| vim_iswhite(p
[0]))
827 if (vim_iswordc(p
[0]))
831 if (enc_dbcs
!= 0 && p
[0] != NUL
&& p
[1] != NUL
)
832 return dbcs_class(p
[0], p
[1]);
834 return utf_class(utf_ptr2char(p
));
839 * Get class of a double-byte character. This always returns 3 or bigger.
840 * TODO: Should return 1 for punctuation.
843 dbcs_class(lead
, trail
)
849 /* please add classfy routine for your language in here */
851 case DBCS_JPNU
: /* ? */
854 /* JIS code classification */
855 unsigned char lb
= lead
;
856 unsigned char tb
= trail
;
858 /* convert process code to JIS */
859 # if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS)
860 /* process code is SJIS */
862 lb
= (lb
- 0x81) * 2 + 0x21;
864 lb
= (lb
- 0xc1) * 2 + 0x21;
876 * XXX: Code page identification can not use with all
877 * system! So, some other encoding information
879 * In japanese: SJIS,EUC,UNICODE,(JIS)
880 * Note that JIS-code system don't use as
881 * process code in most system because it uses
882 * escape sequences(JIS is context depend encoding).
884 /* assume process code is JAPANESE-EUC */
889 switch (lb
<< 8 | tb
)
891 case 0x2121: /* ZENKAKU space */
893 case 0x2122: /* KU-TEN (Japanese comma) */
894 case 0x2123: /* TOU-TEN (Japanese period) */
895 case 0x2124: /* ZENKAKU comma */
896 case 0x2125: /* ZENKAKU period */
898 case 0x213c: /* prolongedsound handled as KATAKANA */
901 /* sieved by KU code */
906 /* special symbols */
932 case DBCS_KORU
: /* ? */
935 /* KS code classification */
936 unsigned char c1
= lead
;
937 unsigned char c2
= trail
;
943 * 23 : Alpha-numeric/Roman Letter (Full width)
944 * 24 : Hangul Letter(Alphabet)
945 * 25 : Roman Numeral/Greek Letter
948 * 28 : Circled/Parenthesized Letter
949 * 29 : Hirigana/Katakana
950 * 30 : Cyrillic Letter
953 if (c1
>= 0xB0 && c1
<= 0xC8)
956 #if defined(WIN3264) || defined(WIN32UNIX)
957 else if (c1
<= 0xA0 || c2
<= 0xA0)
958 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */
959 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE
960 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0
965 else if (c1
>= 0xCA && c1
<= 0xFD)
978 /* Hangul Letter(Alphabet) */
981 /* Roman Numeral/Greek Letter */
992 return 25; /* Roman Letter */
994 return 22; /* Symbols */
996 /* Circled/Parenthesized Letter */
1000 /* Hirigana/Katakana */
1003 /* Cyrillic Letter */
1014 * mb_char2len() function pointer.
1015 * Return length in bytes of character "c".
1016 * Returns 1 for a single-byte character.
1035 * mb_char2bytes() function pointer.
1036 * Convert a character to its bytes.
1037 * Returns the length in bytes.
1040 latin_char2bytes(c
, buf
)
1049 dbcs_char2bytes(c
, buf
)
1055 buf
[0] = (unsigned)c
>> 8;
1057 /* Never use a NUL byte, it causes lots of trouble. It's an invalid
1058 * character anyway. */
1068 * mb_ptr2len() function pointer.
1069 * Get byte length of character at "*p" but stop at a NUL.
1070 * For UTF-8 this includes following composing characters.
1071 * Returns 0 when *p is NUL.
1078 return MB_BYTE2LEN(*p
);
1087 /* Check if second byte is not missing. */
1088 len
= MB_BYTE2LEN(*p
);
1089 if (len
== 2 && p
[1] == NUL
)
1096 unsigned short first
;
1097 unsigned short last
;
1099 static int intable
__ARGS((struct interval
*table
, size_t size
, int c
));
1102 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]".
1105 intable(table
, size
, c
)
1106 struct interval
*table
;
1112 /* first quick check for Latin1 etc. characters */
1113 if (c
< table
[0].first
)
1116 /* binary search in table */
1118 top
= (int)(size
/ sizeof(struct interval
) - 1);
1121 mid
= (bot
+ top
) / 2;
1122 if (table
[mid
].last
< c
)
1124 else if (table
[mid
].first
> c
)
1133 * For UTF-8 character "c" return 2 for a double-width character, 1 for others.
1134 * Returns 4 or 6 for an unprintable character.
1135 * Is only correct for characters >= 0x80.
1136 * When p_ambw is "double", return 2 for a character with East Asian Width
1137 * class 'A'(mbiguous).
1143 /* sorted list of non-overlapping intervals of East Asian Ambiguous
1144 * characters, generated with:
1145 * "uniset +WIDTH-A -cat=Me -cat=Mn -cat=Cf c" */
1146 static struct interval ambiguous
[] = {
1147 {0x00A1, 0x00A1}, {0x00A4, 0x00A4}, {0x00A7, 0x00A8},
1148 {0x00AA, 0x00AA}, {0x00AE, 0x00AE}, {0x00B0, 0x00B4},
1149 {0x00B6, 0x00BA}, {0x00BC, 0x00BF}, {0x00C6, 0x00C6},
1150 {0x00D0, 0x00D0}, {0x00D7, 0x00D8}, {0x00DE, 0x00E1},
1151 {0x00E6, 0x00E6}, {0x00E8, 0x00EA}, {0x00EC, 0x00ED},
1152 {0x00F0, 0x00F0}, {0x00F2, 0x00F3}, {0x00F7, 0x00FA},
1153 {0x00FC, 0x00FC}, {0x00FE, 0x00FE}, {0x0101, 0x0101},
1154 {0x0111, 0x0111}, {0x0113, 0x0113}, {0x011B, 0x011B},
1155 {0x0126, 0x0127}, {0x012B, 0x012B}, {0x0131, 0x0133},
1156 {0x0138, 0x0138}, {0x013F, 0x0142}, {0x0144, 0x0144},
1157 {0x0148, 0x014B}, {0x014D, 0x014D}, {0x0152, 0x0153},
1158 {0x0166, 0x0167}, {0x016B, 0x016B}, {0x01CE, 0x01CE},
1159 {0x01D0, 0x01D0}, {0x01D2, 0x01D2}, {0x01D4, 0x01D4},
1160 {0x01D6, 0x01D6}, {0x01D8, 0x01D8}, {0x01DA, 0x01DA},
1161 {0x01DC, 0x01DC}, {0x0251, 0x0251}, {0x0261, 0x0261},
1162 {0x02C4, 0x02C4}, {0x02C7, 0x02C7}, {0x02C9, 0x02CB},
1163 {0x02CD, 0x02CD}, {0x02D0, 0x02D0}, {0x02D8, 0x02DB},
1164 {0x02DD, 0x02DD}, {0x02DF, 0x02DF}, {0x0391, 0x03A1},
1165 {0x03A3, 0x03A9}, {0x03B1, 0x03C1}, {0x03C3, 0x03C9},
1166 {0x0401, 0x0401}, {0x0410, 0x044F}, {0x0451, 0x0451},
1167 {0x2010, 0x2010}, {0x2013, 0x2016}, {0x2018, 0x2019},
1168 {0x201C, 0x201D}, {0x2020, 0x2022}, {0x2024, 0x2027},
1169 {0x2030, 0x2030}, {0x2032, 0x2033}, {0x2035, 0x2035},
1170 {0x203B, 0x203B}, {0x203E, 0x203E}, {0x2074, 0x2074},
1171 {0x207F, 0x207F}, {0x2081, 0x2084}, {0x20AC, 0x20AC},
1172 {0x2103, 0x2103}, {0x2105, 0x2105}, {0x2109, 0x2109},
1173 {0x2113, 0x2113}, {0x2116, 0x2116}, {0x2121, 0x2122},
1174 {0x2126, 0x2126}, {0x212B, 0x212B}, {0x2153, 0x2154},
1175 {0x215B, 0x215E}, {0x2160, 0x216B}, {0x2170, 0x2179},
1176 {0x2190, 0x2199}, {0x21B8, 0x21B9}, {0x21D2, 0x21D2},
1177 {0x21D4, 0x21D4}, {0x21E7, 0x21E7}, {0x2200, 0x2200},
1178 {0x2202, 0x2203}, {0x2207, 0x2208}, {0x220B, 0x220B},
1179 {0x220F, 0x220F}, {0x2211, 0x2211}, {0x2215, 0x2215},
1180 {0x221A, 0x221A}, {0x221D, 0x2220}, {0x2223, 0x2223},
1181 {0x2225, 0x2225}, {0x2227, 0x222C}, {0x222E, 0x222E},
1182 {0x2234, 0x2237}, {0x223C, 0x223D}, {0x2248, 0x2248},
1183 {0x224C, 0x224C}, {0x2252, 0x2252}, {0x2260, 0x2261},
1184 {0x2264, 0x2267}, {0x226A, 0x226B}, {0x226E, 0x226F},
1185 {0x2282, 0x2283}, {0x2286, 0x2287}, {0x2295, 0x2295},
1186 {0x2299, 0x2299}, {0x22A5, 0x22A5}, {0x22BF, 0x22BF},
1187 {0x2312, 0x2312}, {0x2460, 0x24E9}, {0x24EB, 0x254B},
1188 {0x2550, 0x2573}, {0x2580, 0x258F}, {0x2592, 0x2595},
1189 {0x25A0, 0x25A1}, {0x25A3, 0x25A9}, {0x25B2, 0x25B3},
1190 {0x25B6, 0x25B7}, {0x25BC, 0x25BD}, {0x25C0, 0x25C1},
1191 {0x25C6, 0x25C8}, {0x25CB, 0x25CB}, {0x25CE, 0x25D1},
1192 {0x25E2, 0x25E5}, {0x25EF, 0x25EF}, {0x2605, 0x2606},
1193 {0x2609, 0x2609}, {0x260E, 0x260F}, {0x2614, 0x2615},
1194 {0x261C, 0x261C}, {0x261E, 0x261E}, {0x2640, 0x2640},
1195 {0x2642, 0x2642}, {0x2660, 0x2661}, {0x2663, 0x2665},
1196 {0x2667, 0x266A}, {0x266C, 0x266D}, {0x266F, 0x266F},
1197 {0x273D, 0x273D}, {0x2776, 0x277F}, {0xE000, 0xF8FF},
1198 {0xFFFD, 0xFFFD}, /* {0xF0000, 0xFFFFD}, {0x100000, 0x10FFFD} */
1203 #ifdef USE_WCHAR_FUNCTIONS
1205 * Assume the library function wcwidth() works better than our own
1206 * stuff. It should return 1 for ambiguous width chars!
1211 return 6; /* unprintable, displays <xxxx> */
1215 if (!utf_printable(c
))
1216 return 6; /* unprintable, displays <xxxx> */
1218 && (c
<= 0x115f /* Hangul Jamo */
1221 || (c
>= 0x2e80 && c
<= 0xa4cf
1222 && c
!= 0x303f) /* CJK ... Yi */
1223 || (c
>= 0xac00 && c
<= 0xd7a3) /* Hangul Syllables */
1224 || (c
>= 0xf900 && c
<= 0xfaff) /* CJK Compatibility
1226 || (c
>= 0xfe30 && c
<= 0xfe6f) /* CJK Compatibility Forms */
1227 || (c
>= 0xff00 && c
<= 0xff60) /* Fullwidth Forms */
1228 || (c
>= 0xffe0 && c
<= 0xffe6)
1229 || (c
>= 0x20000 && c
<= 0x2fffd)
1230 || (c
>= 0x30000 && c
<= 0x3fffd)))
1235 /* Characters below 0x100 are influenced by 'isprint' option */
1236 else if (c
>= 0x80 && !vim_isprintc(c
))
1237 return 4; /* unprintable, displays <xx> */
1239 if (c
>= 0x80 && *p_ambw
== 'd' && intable(ambiguous
, sizeof(ambiguous
), c
))
1246 * mb_ptr2cells() function pointer.
1247 * Return the number of display cells character at "*p" occupies.
1248 * This doesn't take care of unprintable characters, use ptr2cells() for that.
1263 /* Need to convert to a wide character. */
1266 c
= utf_ptr2char(p
);
1267 /* An illegal byte is displayed as <xx>. */
1268 if (utf_ptr2len(p
) == 1 || c
== NUL
)
1270 /* If the char is ASCII it must be an overlong sequence. */
1272 return char2cells(c
);
1273 return utf_char2cells(c
);
1282 /* Number of cells is equal to number of bytes, except for euc-jp when
1283 * the first byte is 0x8e. */
1284 if (enc_dbcs
== DBCS_JPNU
&& *p
== 0x8e)
1286 return MB_BYTE2LEN(*p
);
1290 * mb_char2cells() function pointer.
1291 * Return the number of display cells character "c" occupies.
1292 * Only takes care of multi-byte chars, not "^C" and such.
1305 /* Number of cells is equal to number of bytes, except for euc-jp when
1306 * the first byte is 0x8e. */
1307 if (enc_dbcs
== DBCS_JPNU
&& ((unsigned)c
>> 8) == 0x8e)
1309 /* use the first byte */
1310 return MB_BYTE2LEN((unsigned)c
>> 8);
1314 * mb_off2cells() function pointer.
1315 * Return number of display cells for char at ScreenLines[off].
1316 * We make sure that the offset used is less than "max_off".
1319 latin_off2cells(off
, max_off
)
1320 unsigned off UNUSED
;
1321 unsigned max_off UNUSED
;
1327 dbcs_off2cells(off
, max_off
)
1331 /* never check beyond end of the line */
1335 /* Number of cells is equal to number of bytes, except for euc-jp when
1336 * the first byte is 0x8e. */
1337 if (enc_dbcs
== DBCS_JPNU
&& ScreenLines
[off
] == 0x8e)
1339 return MB_BYTE2LEN(ScreenLines
[off
]);
1343 utf_off2cells(off
, max_off
)
1347 return (off
+ 1 < max_off
&& ScreenLines
[off
+ 1] == 0) ? 2 : 1;
1351 * mb_ptr2char() function pointer.
1352 * Convert a byte sequence into a character.
1365 if (MB_BYTE2LEN(*p
) > 1 && p
[1] != NUL
)
1366 return (p
[0] << 8) + p
[1];
1371 * Convert a UTF-8 byte sequence to a wide character.
1372 * If the sequence is illegal or truncated by a NUL the first byte is
1374 * Does not include composing characters, of course.
1382 if (p
[0] < 0x80) /* be quick for ASCII */
1385 len
= utf8len_tab
[p
[0]];
1386 if (len
> 1 && (p
[1] & 0xc0) == 0x80)
1389 return ((p
[0] & 0x1f) << 6) + (p
[1] & 0x3f);
1390 if ((p
[2] & 0xc0) == 0x80)
1393 return ((p
[0] & 0x0f) << 12) + ((p
[1] & 0x3f) << 6)
1395 if ((p
[3] & 0xc0) == 0x80)
1398 return ((p
[0] & 0x07) << 18) + ((p
[1] & 0x3f) << 12)
1399 + ((p
[2] & 0x3f) << 6) + (p
[3] & 0x3f);
1400 if ((p
[4] & 0xc0) == 0x80)
1403 return ((p
[0] & 0x03) << 24) + ((p
[1] & 0x3f) << 18)
1404 + ((p
[2] & 0x3f) << 12) + ((p
[3] & 0x3f) << 6)
1406 if ((p
[5] & 0xc0) == 0x80 && len
== 6)
1407 return ((p
[0] & 0x01) << 30) + ((p
[1] & 0x3f) << 24)
1408 + ((p
[2] & 0x3f) << 18) + ((p
[3] & 0x3f) << 12)
1409 + ((p
[4] & 0x3f) << 6) + (p
[5] & 0x3f);
1414 /* Illegal value, just return the first byte */
1419 * Get character at **pp and advance *pp to the next character.
1420 * Note: composing characters are skipped!
1428 c
= (*mb_ptr2char
)(*pp
);
1429 *pp
+= (*mb_ptr2len
)(*pp
);
1434 * Get character at **pp and advance *pp to the next character.
1435 * Note: composing characters are returned as separate characters.
1438 mb_cptr2char_adv(pp
)
1443 c
= (*mb_ptr2char
)(*pp
);
1445 *pp
+= utf_ptr2len(*pp
);
1447 *pp
+= (*mb_ptr2len
)(*pp
);
1451 #if defined(FEAT_ARABIC) || defined(PROTO)
1453 * Check whether we are dealing with Arabic combining characters.
1454 * Note: these are NOT really composing characters!
1457 arabic_combine(one
, two
)
1458 int one
; /* first character */
1459 int two
; /* character just after "one" */
1462 return arabic_maycombine(two
);
1467 * Check whether we are dealing with a character that could be regarded as an
1468 * Arabic combining character, need to check the character before this.
1471 arabic_maycombine(two
)
1474 if (p_arshape
&& !p_tbidi
)
1475 return (two
== a_ALEF_MADDA
1476 || two
== a_ALEF_HAMZA_ABOVE
1477 || two
== a_ALEF_HAMZA_BELOW
1483 * Check if the character pointed to by "p2" is a composing character when it
1484 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which
1485 * behaves like a composing character.
1488 utf_composinglike(p1
, p2
)
1494 c2
= utf_ptr2char(p2
);
1495 if (utf_iscomposing(c2
))
1497 if (!arabic_maycombine(c2
))
1499 return arabic_combine(utf_ptr2char(p1
), c2
);
1504 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO
1505 * composing characters.
1508 utfc_ptr2char(p
, pcc
)
1510 int *pcc
; /* return: composing chars, last one is 0 */
1517 c
= utf_ptr2char(p
);
1518 len
= utf_ptr2len(p
);
1520 /* Only accept a composing char when the first char isn't illegal. */
1521 if ((len
> 1 || *p
< 0x80)
1523 && UTF_COMPOSINGLIKE(p
, p
+ len
))
1525 cc
= utf_ptr2char(p
+ len
);
1531 len
+= utf_ptr2len(p
+ len
);
1532 if (p
[len
] < 0x80 || !utf_iscomposing(cc
= utf_ptr2char(p
+ len
)))
1537 if (i
< MAX_MCO
) /* last composing char must be 0 */
1544 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO
1545 * composing characters. Use no more than p[maxlen].
1548 utfc_ptr2char_len(p
, pcc
, maxlen
)
1550 int *pcc
; /* return: composing chars, last one is 0 */
1558 c
= utf_ptr2char(p
);
1559 len
= utf_ptr2len_len(p
, maxlen
);
1560 /* Only accept a composing char when the first char isn't illegal. */
1561 if ((len
> 1 || *p
< 0x80)
1564 && UTF_COMPOSINGLIKE(p
, p
+ len
))
1566 cc
= utf_ptr2char(p
+ len
);
1572 len
+= utf_ptr2len_len(p
+ len
, maxlen
- len
);
1575 || !utf_iscomposing(cc
= utf_ptr2char(p
+ len
)))
1580 if (i
< MAX_MCO
) /* last composing char must be 0 */
1587 * Convert the character at screen position "off" to a sequence of bytes.
1588 * Includes the composing characters.
1589 * "buf" must at least have the length MB_MAXBYTES.
1590 * Returns the produced number of bytes.
1593 utfc_char2bytes(off
, buf
)
1600 len
= utf_char2bytes(ScreenLinesUC
[off
], buf
);
1601 for (i
= 0; i
< Screen_mco
; ++i
)
1603 if (ScreenLinesC
[i
][off
] == 0)
1605 len
+= utf_char2bytes(ScreenLinesC
[i
][off
], buf
+ len
);
1611 * Get the length of a UTF-8 byte sequence, not including any following
1612 * composing characters.
1614 * Returns 1 for an illegal byte sequence.
1625 len
= utf8len_tab
[*p
];
1626 for (i
= 1; i
< len
; ++i
)
1627 if ((p
[i
] & 0xc0) != 0x80)
1633 * Return length of UTF-8 character, obtained from the first byte.
1634 * "b" must be between 0 and 255!
1640 return utf8len_tab
[b
];
1644 * Get the length of UTF-8 byte sequence "p[size]". Does not include any
1645 * following composing characters.
1647 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.).
1648 * Returns number > "size" for an incomplete byte sequence.
1651 utf_ptr2len_len(p
, size
)
1661 m
= len
= utf8len_tab
[*p
];
1663 m
= size
; /* incomplete byte sequence. */
1664 for (i
= 1; i
< m
; ++i
)
1665 if ((p
[i
] & 0xc0) != 0x80)
1671 * Return the number of bytes the UTF-8 encoding of the character at "p" takes.
1672 * This includes following composing characters.
1686 if (b0
< 0x80 && p
[1] < 0x80) /* be quick for ASCII */
1689 /* Skip over first UTF-8 char, stopping at a NUL byte. */
1690 len
= utf_ptr2len(p
);
1692 /* Check for illegal byte. */
1693 if (len
== 1 && b0
>= 0x80)
1697 * Check for composing characters. We can handle only the first two, but
1698 * skip all of them (otherwise the cursor would get stuck).
1705 if (p
[len
] < 0x80 || !UTF_COMPOSINGLIKE(p
+ prevlen
, p
+ len
))
1708 /* Skip over composing char */
1712 len
+= utf_ptr2len(p
+ len
);
1717 * Return the number of bytes the UTF-8 encoding of the character at "p[size]"
1718 * takes. This includes following composing characters.
1719 * Returns 1 for an illegal char or an incomplete byte sequence.
1722 utfc_ptr2len_len(p
, size
)
1733 if (p
[0] < 0x80 && (size
== 1 || p
[1] < 0x80)) /* be quick for ASCII */
1736 /* Skip over first UTF-8 char, stopping at a NUL byte. */
1737 len
= utf_ptr2len_len(p
, size
);
1739 /* Check for illegal byte and incomplete byte sequence. */
1740 if ((len
== 1 && p
[0] >= 0x80) || len
> size
)
1744 * Check for composing characters. We can handle only the first two, but
1745 * skip all of them (otherwise the cursor would get stuck).
1758 * Next character length should not go beyond size to ensure that
1759 * UTF_COMPOSINGLIKE(...) does not read beyond size.
1761 len_next_char
= utf_ptr2len_len(p
+ len
, size
- len
);
1762 if (len_next_char
> size
- len
)
1765 if (!UTF_COMPOSINGLIKE(p
+ prevlen
, p
+ len
))
1768 /* Skip over composing char */
1772 len
+= len_next_char
;
1778 * Return the number of bytes the UTF-8 encoding of character "c" takes.
1779 * This does not include composing characters.
1799 * Convert Unicode character "c" to UTF-8 string in "buf[]".
1800 * Returns the number of bytes.
1801 * This does not include composing characters.
1804 utf_char2bytes(c
, buf
)
1808 if (c
< 0x80) /* 7 bits */
1813 if (c
< 0x800) /* 11 bits */
1815 buf
[0] = 0xc0 + ((unsigned)c
>> 6);
1816 buf
[1] = 0x80 + (c
& 0x3f);
1819 if (c
< 0x10000) /* 16 bits */
1821 buf
[0] = 0xe0 + ((unsigned)c
>> 12);
1822 buf
[1] = 0x80 + (((unsigned)c
>> 6) & 0x3f);
1823 buf
[2] = 0x80 + (c
& 0x3f);
1826 if (c
< 0x200000) /* 21 bits */
1828 buf
[0] = 0xf0 + ((unsigned)c
>> 18);
1829 buf
[1] = 0x80 + (((unsigned)c
>> 12) & 0x3f);
1830 buf
[2] = 0x80 + (((unsigned)c
>> 6) & 0x3f);
1831 buf
[3] = 0x80 + (c
& 0x3f);
1834 if (c
< 0x4000000) /* 26 bits */
1836 buf
[0] = 0xf8 + ((unsigned)c
>> 24);
1837 buf
[1] = 0x80 + (((unsigned)c
>> 18) & 0x3f);
1838 buf
[2] = 0x80 + (((unsigned)c
>> 12) & 0x3f);
1839 buf
[3] = 0x80 + (((unsigned)c
>> 6) & 0x3f);
1840 buf
[4] = 0x80 + (c
& 0x3f);
1844 buf
[0] = 0xfc + ((unsigned)c
>> 30);
1845 buf
[1] = 0x80 + (((unsigned)c
>> 24) & 0x3f);
1846 buf
[2] = 0x80 + (((unsigned)c
>> 18) & 0x3f);
1847 buf
[3] = 0x80 + (((unsigned)c
>> 12) & 0x3f);
1848 buf
[4] = 0x80 + (((unsigned)c
>> 6) & 0x3f);
1849 buf
[5] = 0x80 + (c
& 0x3f);
1854 * Return TRUE if "c" is a composing UTF-8 character. This means it will be
1855 * drawn on top of the preceding character.
1856 * Based on code from Markus Kuhn.
1862 /* sorted list of non-overlapping intervals */
1863 static struct interval combining
[] =
1865 {0x0300, 0x034f}, {0x0360, 0x036f}, {0x0483, 0x0486}, {0x0488, 0x0489},
1866 {0x0591, 0x05a1}, {0x05a3, 0x05b9}, {0x05bb, 0x05bd}, {0x05bf, 0x05bf},
1867 {0x05c1, 0x05c2}, {0x05c4, 0x05c4}, {0x0610, 0x0615}, {0x064b, 0x0658},
1868 {0x0670, 0x0670}, {0x06d6, 0x06dc}, {0x06de, 0x06e4}, {0x06e7, 0x06e8},
1869 {0x06ea, 0x06ed}, {0x0711, 0x0711}, {0x0730, 0x074a}, {0x07a6, 0x07b0},
1870 {0x0901, 0x0903}, {0x093c, 0x093c}, {0x093e, 0x094d}, {0x0951, 0x0954},
1871 {0x0962, 0x0963}, {0x0981, 0x0983}, {0x09bc, 0x09bc}, {0x09be, 0x09c4},
1872 {0x09c7, 0x09c8}, {0x09cb, 0x09cd}, {0x09d7, 0x09d7}, {0x09e2, 0x09e3},
1873 {0x0a01, 0x0a03}, {0x0a3c, 0x0a3c}, {0x0a3e, 0x0a42}, {0x0a47, 0x0a48},
1874 {0x0a4b, 0x0a4d}, {0x0a70, 0x0a71}, {0x0a81, 0x0a83}, {0x0abc, 0x0abc},
1875 {0x0abe, 0x0ac5}, {0x0ac7, 0x0ac9}, {0x0acb, 0x0acd}, {0x0ae2, 0x0ae3},
1876 {0x0b01, 0x0b03}, {0x0b3c, 0x0b3c}, {0x0b3e, 0x0b43}, {0x0b47, 0x0b48},
1877 {0x0b4b, 0x0b4d}, {0x0b56, 0x0b57}, {0x0b82, 0x0b82}, {0x0bbe, 0x0bc2},
1878 {0x0bc6, 0x0bc8}, {0x0bca, 0x0bcd}, {0x0bd7, 0x0bd7}, {0x0c01, 0x0c03},
1879 {0x0c3e, 0x0c44}, {0x0c46, 0x0c48}, {0x0c4a, 0x0c4d}, {0x0c55, 0x0c56},
1880 {0x0c82, 0x0c83}, {0x0cbc, 0x0cbc}, {0x0cbe, 0x0cc4}, {0x0cc6, 0x0cc8},
1881 {0x0cca, 0x0ccd}, {0x0cd5, 0x0cd6}, {0x0d02, 0x0d03}, {0x0d3e, 0x0d43},
1882 {0x0d46, 0x0d48}, {0x0d4a, 0x0d4d}, {0x0d57, 0x0d57}, {0x0d82, 0x0d83},
1883 {0x0dca, 0x0dca}, {0x0dcf, 0x0dd4}, {0x0dd6, 0x0dd6}, {0x0dd8, 0x0ddf},
1884 {0x0df2, 0x0df3}, {0x0e31, 0x0e31}, {0x0e34, 0x0e3a}, {0x0e47, 0x0e4e},
1885 {0x0eb1, 0x0eb1}, {0x0eb4, 0x0eb9}, {0x0ebb, 0x0ebc}, {0x0ec8, 0x0ecd},
1886 {0x0f18, 0x0f19}, {0x0f35, 0x0f35}, {0x0f37, 0x0f37}, {0x0f39, 0x0f39},
1887 {0x0f3e, 0x0f3f}, {0x0f71, 0x0f84}, {0x0f86, 0x0f87}, {0x0f90, 0x0f97},
1888 {0x0f99, 0x0fbc}, {0x0fc6, 0x0fc6}, {0x102c, 0x1032}, {0x1036, 0x1039},
1889 {0x1056, 0x1059}, {0x1712, 0x1714}, {0x1732, 0x1734}, {0x1752, 0x1753},
1890 {0x1772, 0x1773}, {0x17b6, 0x17d3}, {0x17dd, 0x17dd}, {0x180b, 0x180d},
1891 {0x18a9, 0x18a9}, {0x1920, 0x192b}, {0x1930, 0x193b}, {0x20d0, 0x20ea},
1892 {0x302a, 0x302f}, {0x3099, 0x309a}, {0xfb1e, 0xfb1e}, {0xfe00, 0xfe0f},
1896 return intable(combining
, sizeof(combining
), c
);
1900 * Return TRUE for characters that can be displayed in a normal way.
1901 * Only for characters of 0x100 and above!
1907 #ifdef USE_WCHAR_FUNCTIONS
1909 * Assume the iswprint() library function works better than our own stuff.
1913 /* Sorted list of non-overlapping intervals.
1914 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */
1915 static struct interval nonprint
[] =
1917 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e},
1918 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb},
1922 return !intable(nonprint
, sizeof(nonprint
), c
);
1927 * Get class of a Unicode character.
1930 * 2 or bigger: some class of word character.
1936 /* sorted list of non-overlapping intervals */
1937 static struct clinterval
1939 unsigned short first
;
1940 unsigned short last
;
1941 unsigned short class;
1944 {0x037e, 0x037e, 1}, /* Greek question mark */
1945 {0x0387, 0x0387, 1}, /* Greek ano teleia */
1946 {0x055a, 0x055f, 1}, /* Armenian punctuation */
1947 {0x0589, 0x0589, 1}, /* Armenian full stop */
1948 {0x05be, 0x05be, 1},
1949 {0x05c0, 0x05c0, 1},
1950 {0x05c3, 0x05c3, 1},
1951 {0x05f3, 0x05f4, 1},
1952 {0x060c, 0x060c, 1},
1953 {0x061b, 0x061b, 1},
1954 {0x061f, 0x061f, 1},
1955 {0x066a, 0x066d, 1},
1956 {0x06d4, 0x06d4, 1},
1957 {0x0700, 0x070d, 1}, /* Syriac punctuation */
1958 {0x0964, 0x0965, 1},
1959 {0x0970, 0x0970, 1},
1960 {0x0df4, 0x0df4, 1},
1961 {0x0e4f, 0x0e4f, 1},
1962 {0x0e5a, 0x0e5b, 1},
1963 {0x0f04, 0x0f12, 1},
1964 {0x0f3a, 0x0f3d, 1},
1965 {0x0f85, 0x0f85, 1},
1966 {0x104a, 0x104f, 1}, /* Myanmar punctuation */
1967 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */
1968 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */
1969 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */
1970 {0x1680, 0x1680, 0},
1971 {0x169b, 0x169c, 1},
1972 {0x16eb, 0x16ed, 1},
1973 {0x1735, 0x1736, 1},
1974 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */
1975 {0x1800, 0x180a, 1}, /* Mongolian punctuation */
1976 {0x2000, 0x200b, 0}, /* spaces */
1977 {0x200c, 0x2027, 1}, /* punctuation and symbols */
1978 {0x2028, 0x2029, 0},
1979 {0x202a, 0x202e, 1}, /* punctuation and symbols */
1980 {0x202f, 0x202f, 0},
1981 {0x2030, 0x205e, 1}, /* punctuation and symbols */
1982 {0x205f, 0x205f, 0},
1983 {0x2060, 0x27ff, 1}, /* punctuation and symbols */
1984 {0x2070, 0x207f, 0x2070}, /* superscript */
1985 {0x2080, 0x2094, 0x2080}, /* subscript */
1986 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */
1987 {0x2800, 0x28ff, 0x2800}, /* braille */
1988 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */
1989 {0x29d8, 0x29db, 1},
1990 {0x29fc, 0x29fd, 1},
1991 {0x3000, 0x3000, 0}, /* ideographic space */
1992 {0x3001, 0x3020, 1}, /* ideographic punctuation */
1993 {0x3030, 0x3030, 1},
1994 {0x303d, 0x303d, 1},
1995 {0x3040, 0x309f, 0x3040}, /* Hiragana */
1996 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */
1997 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */
1998 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */
1999 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */
2000 {0xfd3e, 0xfd3f, 1},
2001 {0xfe30, 0xfe6b, 1}, /* punctuation forms */
2002 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */
2003 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */
2004 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */
2005 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */
2008 int top
= sizeof(classes
) / sizeof(struct clinterval
) - 1;
2011 /* First quick check for Latin1 characters, use 'iskeyword'. */
2014 if (c
== ' ' || c
== '\t' || c
== NUL
|| c
== 0xa0)
2015 return 0; /* blank */
2017 return 2; /* word character */
2018 return 1; /* punctuation */
2021 /* binary search in table */
2024 mid
= (bot
+ top
) / 2;
2025 if (classes
[mid
].last
< c
)
2027 else if (classes
[mid
].first
> c
)
2030 return (int)classes
[mid
].class;
2033 /* most other characters are "word" characters */
2038 * Code for Unicode case-dependent operations. Based on notes in
2039 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt
2040 * This code uses simple case folding, not full case folding.
2044 * The following table is built by foldExtract.pl < CaseFolding.txt .
2045 * It must be in numeric order, because we use binary search on it.
2046 * An entry such as {0x41,0x5a,1,32} means that UCS-4 characters in the range
2047 * from 0x41 to 0x5a inclusive, stepping by 1, are folded by adding 32.
2058 static convertStruct foldCase
[] =
2060 {0x41,0x5a,1,32}, {0xc0,0xd6,1,32}, {0xd8,0xde,1,32},
2061 {0x100,0x12e,2,1}, {0x130,0x130,-1,-199}, {0x132,0x136,2,1},
2062 {0x139,0x147,2,1}, {0x14a,0x176,2,1}, {0x178,0x178,-1,-121},
2063 {0x179,0x17d,2,1}, {0x181,0x181,-1,210}, {0x182,0x184,2,1},
2064 {0x186,0x186,-1,206}, {0x187,0x187,-1,1}, {0x189,0x18a,1,205},
2065 {0x18b,0x18b,-1,1}, {0x18e,0x18e,-1,79}, {0x18f,0x18f,-1,202},
2066 {0x190,0x190,-1,203}, {0x191,0x191,-1,1}, {0x193,0x193,-1,205},
2067 {0x194,0x194,-1,207}, {0x196,0x196,-1,211}, {0x197,0x197,-1,209},
2068 {0x198,0x198,-1,1}, {0x19c,0x19c,-1,211}, {0x19d,0x19d,-1,213},
2069 {0x19f,0x19f,-1,214}, {0x1a0,0x1a4,2,1}, {0x1a6,0x1a6,-1,218},
2070 {0x1a7,0x1a7,-1,1}, {0x1a9,0x1a9,-1,218}, {0x1ac,0x1ac,-1,1},
2071 {0x1ae,0x1ae,-1,218}, {0x1af,0x1af,-1,1}, {0x1b1,0x1b2,1,217},
2072 {0x1b3,0x1b5,2,1}, {0x1b7,0x1b7,-1,219}, {0x1b8,0x1bc,4,1},
2073 {0x1c4,0x1c4,-1,2}, {0x1c5,0x1c5,-1,1}, {0x1c7,0x1c7,-1,2},
2074 {0x1c8,0x1c8,-1,1}, {0x1ca,0x1ca,-1,2}, {0x1cb,0x1db,2,1},
2075 {0x1de,0x1ee,2,1}, {0x1f1,0x1f1,-1,2}, {0x1f2,0x1f4,2,1},
2076 {0x1f6,0x1f6,-1,-97}, {0x1f7,0x1f7,-1,-56}, {0x1f8,0x21e,2,1},
2077 {0x220,0x220,-1,-130}, {0x222,0x232,2,1}, {0x386,0x386,-1,38},
2078 {0x388,0x38a,1,37}, {0x38c,0x38c,-1,64}, {0x38e,0x38f,1,63},
2079 {0x391,0x3a1,1,32}, {0x3a3,0x3ab,1,32}, {0x3d8,0x3ee,2,1},
2080 {0x3f4,0x3f4,-1,-60}, {0x3f7,0x3f7,-1,1}, {0x3f9,0x3f9,-1,-7},
2081 {0x3fa,0x3fa,-1,1}, {0x400,0x40f,1,80}, {0x410,0x42f,1,32},
2082 {0x460,0x480,2,1}, {0x48a,0x4be,2,1}, {0x4c1,0x4cd,2,1},
2083 {0x4d0,0x4f4,2,1}, {0x4f8,0x500,8,1}, {0x502,0x50e,2,1},
2084 {0x531,0x556,1,48}, {0x1e00,0x1e94,2,1}, {0x1ea0,0x1ef8,2,1},
2085 {0x1f08,0x1f0f,1,-8}, {0x1f18,0x1f1d,1,-8}, {0x1f28,0x1f2f,1,-8},
2086 {0x1f38,0x1f3f,1,-8}, {0x1f48,0x1f4d,1,-8}, {0x1f59,0x1f5f,2,-8},
2087 {0x1f68,0x1f6f,1,-8}, {0x1f88,0x1f8f,1,-8}, {0x1f98,0x1f9f,1,-8},
2088 {0x1fa8,0x1faf,1,-8}, {0x1fb8,0x1fb9,1,-8}, {0x1fba,0x1fbb,1,-74},
2089 {0x1fbc,0x1fbc,-1,-9}, {0x1fc8,0x1fcb,1,-86}, {0x1fcc,0x1fcc,-1,-9},
2090 {0x1fd8,0x1fd9,1,-8}, {0x1fda,0x1fdb,1,-100}, {0x1fe8,0x1fe9,1,-8},
2091 {0x1fea,0x1feb,1,-112}, {0x1fec,0x1fec,-1,-7}, {0x1ff8,0x1ff9,1,-128},
2092 {0x1ffa,0x1ffb,1,-126}, {0x1ffc,0x1ffc,-1,-9}, {0x2126,0x2126,-1,-7517},
2093 {0x212a,0x212a,-1,-8383}, {0x212b,0x212b,-1,-8262},
2094 {0x2160,0x216f,1,16}, {0x24b6,0x24cf,1,26}, {0xff21,0xff3a,1,32},
2095 {0x10400,0x10427,1,40}
2098 static int utf_convert(int a
, convertStruct table
[], int tableSize
);
2101 * Generic conversion function for case operations.
2102 * Return the converted equivalent of "a", which is a UCS-4 character. Use
2103 * the given conversion "table". Uses binary search on "table".
2106 utf_convert(a
, table
, tableSize
)
2108 convertStruct table
[];
2111 int start
, mid
, end
; /* indices into table */
2114 end
= tableSize
/ sizeof(convertStruct
);
2117 /* need to search further */
2118 mid
= (end
+ start
) /2;
2119 if (table
[mid
].rangeEnd
< a
)
2124 if (table
[start
].rangeStart
<= a
&& a
<= table
[start
].rangeEnd
2125 && (a
- table
[start
].rangeStart
) % table
[start
].step
== 0)
2126 return (a
+ table
[start
].offset
);
2132 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses
2133 * simple case folding.
2139 return utf_convert(a
, foldCase
, sizeof(foldCase
));
2143 * The following tables are built by upperLowerExtract.pl < UnicodeData.txt .
2144 * They must be in numeric order, because we use binary search on them.
2145 * An entry such as {0x41,0x5a,1,32} means that UCS-4 characters in the range
2146 * from 0x41 to 0x5a inclusive, stepping by 1, are switched to lower (for
2147 * example) by adding 32.
2149 static convertStruct toLower
[] =
2151 {0x41,0x5a,1,32}, {0xc0,0xd6,1,32}, {0xd8,0xde,1,32},
2152 {0x100,0x12e,2,1}, {0x130,0x130,-1,-199}, {0x132,0x136,2,1},
2153 {0x139,0x147,2,1}, {0x14a,0x176,2,1}, {0x178,0x178,-1,-121},
2154 {0x179,0x17d,2,1}, {0x181,0x181,-1,210}, {0x182,0x184,2,1},
2155 {0x186,0x186,-1,206}, {0x187,0x187,-1,1}, {0x189,0x18a,1,205},
2156 {0x18b,0x18b,-1,1}, {0x18e,0x18e,-1,79}, {0x18f,0x18f,-1,202},
2157 {0x190,0x190,-1,203}, {0x191,0x191,-1,1}, {0x193,0x193,-1,205},
2158 {0x194,0x194,-1,207}, {0x196,0x196,-1,211}, {0x197,0x197,-1,209},
2159 {0x198,0x198,-1,1}, {0x19c,0x19c,-1,211}, {0x19d,0x19d,-1,213},
2160 {0x19f,0x19f,-1,214}, {0x1a0,0x1a4,2,1}, {0x1a6,0x1a6,-1,218},
2161 {0x1a7,0x1a7,-1,1}, {0x1a9,0x1a9,-1,218}, {0x1ac,0x1ac,-1,1},
2162 {0x1ae,0x1ae,-1,218}, {0x1af,0x1af,-1,1}, {0x1b1,0x1b2,1,217},
2163 {0x1b3,0x1b5,2,1}, {0x1b7,0x1b7,-1,219}, {0x1b8,0x1bc,4,1},
2164 {0x1c4,0x1ca,3,2}, {0x1cd,0x1db,2,1}, {0x1de,0x1ee,2,1},
2165 {0x1f1,0x1f1,-1,2}, {0x1f4,0x1f4,-1,1}, {0x1f6,0x1f6,-1,-97},
2166 {0x1f7,0x1f7,-1,-56}, {0x1f8,0x21e,2,1}, {0x220,0x220,-1,-130},
2167 {0x222,0x232,2,1}, {0x386,0x386,-1,38}, {0x388,0x38a,1,37},
2168 {0x38c,0x38c,-1,64}, {0x38e,0x38f,1,63}, {0x391,0x3a1,1,32},
2169 {0x3a3,0x3ab,1,32}, {0x3d8,0x3ee,2,1}, {0x3f4,0x3f4,-1,-60},
2170 {0x3f7,0x3f7,-1,1}, {0x3f9,0x3f9,-1,-7}, {0x3fa,0x3fa,-1,1},
2171 {0x400,0x40f,1,80}, {0x410,0x42f,1,32}, {0x460,0x480,2,1},
2172 {0x48a,0x4be,2,1}, {0x4c1,0x4cd,2,1}, {0x4d0,0x4f4,2,1},
2173 {0x4f8,0x500,8,1}, {0x502,0x50e,2,1}, {0x531,0x556,1,48},
2174 {0x1e00,0x1e94,2,1}, {0x1ea0,0x1ef8,2,1}, {0x1f08,0x1f0f,1,-8},
2175 {0x1f18,0x1f1d,1,-8}, {0x1f28,0x1f2f,1,-8}, {0x1f38,0x1f3f,1,-8},
2176 {0x1f48,0x1f4d,1,-8}, {0x1f59,0x1f5f,2,-8}, {0x1f68,0x1f6f,1,-8},
2177 {0x1fb8,0x1fb9,1,-8}, {0x1fba,0x1fbb,1,-74}, {0x1fc8,0x1fcb,1,-86},
2178 {0x1fd8,0x1fd9,1,-8}, {0x1fda,0x1fdb,1,-100}, {0x1fe8,0x1fe9,1,-8},
2179 {0x1fea,0x1feb,1,-112}, {0x1fec,0x1fec,-1,-7}, {0x1ff8,0x1ff9,1,-128},
2180 {0x1ffa,0x1ffb,1,-126}, {0x2126,0x2126,-1,-7517}, {0x212a,0x212a,-1,-8383},
2181 {0x212b,0x212b,-1,-8262}, {0xff21,0xff3a,1,32}, {0x10400,0x10427,1,40}
2184 static convertStruct toUpper
[] =
2186 {0x61,0x7a,1,-32}, {0xb5,0xb5,-1,743}, {0xe0,0xf6,1,-32},
2187 {0xf8,0xfe,1,-32}, {0xff,0xff,-1,121}, {0x101,0x12f,2,-1},
2188 {0x131,0x131,-1,-232}, {0x133,0x137,2,-1}, {0x13a,0x148,2,-1},
2189 {0x14b,0x177,2,-1}, {0x17a,0x17e,2,-1}, {0x17f,0x17f,-1,-300},
2190 {0x183,0x185,2,-1}, {0x188,0x18c,4,-1}, {0x192,0x192,-1,-1},
2191 {0x195,0x195,-1,97}, {0x199,0x199,-1,-1}, {0x19e,0x19e,-1,130},
2192 {0x1a1,0x1a5,2,-1}, {0x1a8,0x1ad,5,-1}, {0x1b0,0x1b4,4,-1},
2193 {0x1b6,0x1b9,3,-1}, {0x1bd,0x1bd,-1,-1}, {0x1bf,0x1bf,-1,56},
2194 {0x1c5,0x1c6,1,-1}, {0x1c8,0x1c9,1,-1}, {0x1cb,0x1cc,1,-1},
2195 {0x1ce,0x1dc,2,-1}, {0x1dd,0x1dd,-1,-79}, {0x1df,0x1ef,2,-1},
2196 {0x1f2,0x1f3,1,-1}, {0x1f5,0x1f9,4,-1}, {0x1fb,0x21f,2,-1},
2197 {0x223,0x233,2,-1}, {0x253,0x253,-1,-210}, {0x254,0x254,-1,-206},
2198 {0x256,0x257,1,-205}, {0x259,0x259,-1,-202}, {0x25b,0x25b,-1,-203},
2199 {0x260,0x260,-1,-205}, {0x263,0x263,-1,-207}, {0x268,0x268,-1,-209},
2200 {0x269,0x26f,6,-211}, {0x272,0x272,-1,-213}, {0x275,0x275,-1,-214},
2201 {0x280,0x283,3,-218}, {0x288,0x288,-1,-218}, {0x28a,0x28b,1,-217},
2202 {0x292,0x292,-1,-219}, {0x3ac,0x3ac,-1,-38}, {0x3ad,0x3af,1,-37},
2203 {0x3b1,0x3c1,1,-32}, {0x3c2,0x3c2,-1,-31}, {0x3c3,0x3cb,1,-32},
2204 {0x3cc,0x3cc,-1,-64}, {0x3cd,0x3ce,1,-63}, {0x3d0,0x3d0,-1,-62},
2205 {0x3d1,0x3d1,-1,-57}, {0x3d5,0x3d5,-1,-47}, {0x3d6,0x3d6,-1,-54},
2206 {0x3d9,0x3ef,2,-1}, {0x3f0,0x3f0,-1,-86}, {0x3f1,0x3f1,-1,-80},
2207 {0x3f2,0x3f2,-1,7}, {0x3f5,0x3f5,-1,-96}, {0x3f8,0x3fb,3,-1},
2208 {0x430,0x44f,1,-32}, {0x450,0x45f,1,-80}, {0x461,0x481,2,-1},
2209 {0x48b,0x4bf,2,-1}, {0x4c2,0x4ce,2,-1}, {0x4d1,0x4f5,2,-1},
2210 {0x4f9,0x501,8,-1}, {0x503,0x50f,2,-1}, {0x561,0x586,1,-48},
2211 {0x1e01,0x1e95,2,-1}, {0x1e9b,0x1e9b,-1,-59}, {0x1ea1,0x1ef9,2,-1},
2212 {0x1f00,0x1f07,1,8}, {0x1f10,0x1f15,1,8}, {0x1f20,0x1f27,1,8},
2213 {0x1f30,0x1f37,1,8}, {0x1f40,0x1f45,1,8}, {0x1f51,0x1f57,2,8},
2214 {0x1f60,0x1f67,1,8}, {0x1f70,0x1f71,1,74}, {0x1f72,0x1f75,1,86},
2215 {0x1f76,0x1f77,1,100}, {0x1f78,0x1f79,1,128}, {0x1f7a,0x1f7b,1,112},
2216 {0x1f7c,0x1f7d,1,126}, {0x1f80,0x1f87,1,8}, {0x1f90,0x1f97,1,8},
2217 {0x1fa0,0x1fa7,1,8}, {0x1fb0,0x1fb1,1,8}, {0x1fb3,0x1fb3,-1,9},
2218 {0x1fbe,0x1fbe,-1,-7205}, {0x1fc3,0x1fc3,-1,9}, {0x1fd0,0x1fd1,1,8},
2219 {0x1fe0,0x1fe1,1,8}, {0x1fe5,0x1fe5,-1,7}, {0x1ff3,0x1ff3,-1,9},
2220 {0xff41,0xff5a,1,-32}, {0x10428,0x1044f,1,-40}
2224 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use
2225 * simple case folding.
2231 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */
2232 if (a
< 128 && (cmp_flags
& CMP_KEEPASCII
))
2233 return TOUPPER_ASC(a
);
2235 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__)
2236 /* If towupper() is available and handles Unicode, use it. */
2237 if (!(cmp_flags
& CMP_INTERNAL
))
2241 /* For characters below 128 use locale sensitive toupper(). */
2243 return TOUPPER_LOC(a
);
2245 /* For any other characters use the above mapping table. */
2246 return utf_convert(a
, toUpper
, sizeof(toUpper
));
2253 return (utf_toupper(a
) != a
);
2257 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use
2258 * simple case folding.
2264 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */
2265 if (a
< 128 && (cmp_flags
& CMP_KEEPASCII
))
2266 return TOLOWER_ASC(a
);
2268 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__)
2269 /* If towlower() is available and handles Unicode, use it. */
2270 if (!(cmp_flags
& CMP_INTERNAL
))
2274 /* For characters below 128 use locale sensitive tolower(). */
2276 return TOLOWER_LOC(a
);
2278 /* For any other characters use the above mapping table. */
2279 return utf_convert(a
, toLower
, sizeof(toLower
));
2286 return (utf_tolower(a
) != a
);
2290 * Version of strnicmp() that handles multi-byte characters.
2291 * Needed for Big5, Sjift-JIS and UTF-8 encoding. Other DBCS encodings can
2292 * probably use strnicmp(), because there are no ASCII characters in the
2294 * Returns zero if s1 and s2 are equal (ignoring case), the difference between
2295 * two characters otherwise.
2298 mb_strnicmp(s1
, s2
, nn
)
2304 int incomplete
= FALSE
;
2307 for (i
= 0; i
< n
; i
+= l
)
2309 if (s1
[i
] == NUL
&& s2
[i
] == NUL
) /* both strings end */
2313 l
= utf_byte2len(s1
[i
]);
2316 l
= n
- i
; /* incomplete character */
2319 /* Check directly first, it's faster. */
2320 for (j
= 0; j
< l
; ++j
)
2322 if (s1
[i
+ j
] != s2
[i
+ j
])
2325 /* Both stings have the same bytes but are incomplete or
2326 * have illegal bytes, accept them as equal. */
2331 /* If one of the two characters is incomplete return -1. */
2332 if (incomplete
|| i
+ utf_byte2len(s2
[i
]) > n
)
2334 cdiff
= utf_fold(utf_ptr2char(s1
+ i
))
2335 - utf_fold(utf_ptr2char(s2
+ i
));
2342 l
= (*mb_ptr2len
)(s1
+ i
);
2345 /* Single byte: first check normally, then with ignore case. */
2348 cdiff
= MB_TOLOWER(s1
[i
]) - MB_TOLOWER(s2
[i
]);
2355 /* For non-Unicode multi-byte don't ignore case. */
2358 cdiff
= STRNCMP(s1
+ i
, s2
+ i
, l
);
2368 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what
2369 * 'encoding' has been set to.
2380 /* Get the byte length of the char under the cursor, including composing
2382 line
= ml_get_cursor();
2383 len
= utfc_ptr2len(line
);
2391 for (i
= 0; i
< len
; ++i
)
2395 /* start of (composing) character, get its length */
2398 STRCPY(IObuff
+ rlen
, "+ ");
2401 clen
= utf_ptr2len(line
+ i
);
2403 sprintf((char *)IObuff
+ rlen
, "%02x ", line
[i
]);
2405 rlen
+= (int)STRLEN(IObuff
+ rlen
);
2406 if (rlen
> IOSIZE
- 20)
2414 * mb_head_off() function pointer.
2415 * Return offset from "p" to the first byte of the character it points into.
2416 * Returns 0 when already at the first byte of a character.
2419 latin_head_off(base
, p
)
2420 char_u
*base UNUSED
;
2427 dbcs_head_off(base
, p
)
2433 /* It can't be a trailing byte when not using DBCS, at the start of the
2434 * string or the previous byte can't start a double-byte. */
2435 if (p
<= base
|| MB_BYTE2LEN(p
[-1]) == 1)
2438 /* This is slow: need to start at the base and go forward until the
2439 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */
2442 q
+= dbcs_ptr2len(q
);
2443 return (q
== p
) ? 0 : 1;
2447 * Special version of dbcs_head_off() that works for ScreenLines[], where
2448 * single-width DBCS_JPNU characters are stored separately.
2451 dbcs_screen_head_off(base
, p
)
2457 /* It can't be a trailing byte when not using DBCS, at the start of the
2458 * string or the previous byte can't start a double-byte.
2459 * For euc-jp an 0x8e byte in the previous cell always means we have a
2460 * lead byte in the current cell. */
2462 || (enc_dbcs
== DBCS_JPNU
&& p
[-1] == 0x8e)
2463 || MB_BYTE2LEN(p
[-1]) == 1)
2466 /* This is slow: need to start at the base and go forward until the
2467 * byte we are looking for. Return 1 when we went past it, 0 otherwise.
2468 * For DBCS_JPNU look out for 0x8e, which means the second byte is not
2469 * stored as the next byte. */
2473 if (enc_dbcs
== DBCS_JPNU
&& *q
== 0x8e)
2476 q
+= dbcs_ptr2len(q
);
2478 return (q
== p
) ? 0 : 1;
2482 utf_head_off(base
, p
)
2493 if (*p
< 0x80) /* be quick for ASCII */
2496 /* Skip backwards over trailing bytes: 10xx.xxxx
2497 * Skip backwards again if on a composing char. */
2500 /* Move s to the last byte of this char. */
2501 for (s
= q
; (s
[1] & 0xc0) == 0x80; ++s
)
2503 /* Move q to the first byte of this char. */
2504 while (q
> base
&& (*q
& 0xc0) == 0x80)
2506 /* Check for illegal sequence. Do allow an illegal byte after where we
2508 if (utf8len_tab
[*q
] != (int)(s
- q
+ 1)
2509 && utf8len_tab
[*q
] != (int)(p
- q
+ 1))
2515 c
= utf_ptr2char(q
);
2516 if (utf_iscomposing(c
))
2520 if (arabic_maycombine(c
))
2522 /* Advance to get a sneak-peak at the next char */
2525 /* Move j to the first byte of this char. */
2526 while (j
> base
&& (*j
& 0xc0) == 0x80)
2528 if (arabic_combine(utf_ptr2char(j
), c
))
2535 return (int)(p
- q
);
2539 * Copy a character from "*fp" to "*tp" and advance the pointers.
2542 mb_copy_char(fp
, tp
)
2546 int l
= (*mb_ptr2len
)(*fp
);
2548 mch_memmove(*tp
, *fp
, (size_t)l
);
2554 * Return the offset from "p" to the first byte of a character. When "p" is
2555 * at the start of a character 0 is returned, otherwise the offset to the next
2556 * character. Can start anywhere in a stream of bytes.
2559 mb_off_next(base
, p
)
2568 if (*p
< 0x80) /* be quick for ASCII */
2571 /* Find the next character that isn't 10xx.xxxx */
2572 for (i
= 0; (p
[i
] & 0xc0) == 0x80; ++i
)
2576 /* Check for illegal sequence. */
2577 for (j
= 0; p
- j
> base
; ++j
)
2578 if ((p
[-j
] & 0xc0) != 0x80)
2580 if (utf8len_tab
[p
[-j
]] != i
+ j
)
2586 /* Only need to check if we're on a trail byte, it doesn't matter if we
2587 * want the offset to the next or current character. */
2588 return (*mb_head_off
)(base
, p
);
2592 * Return the offset from "p" to the last byte of the character it points
2593 * into. Can start anywhere in a stream of bytes.
2596 mb_tail_off(base
, p
)
2608 /* Find the last character that is 10xx.xxxx */
2609 for (i
= 0; (p
[i
+ 1] & 0xc0) == 0x80; ++i
)
2611 /* Check for illegal sequence. */
2612 for (j
= 0; p
- j
> base
; ++j
)
2613 if ((p
[-j
] & 0xc0) != 0x80)
2615 if (utf8len_tab
[p
[-j
]] != i
+ j
+ 1)
2620 /* It can't be the first byte if a double-byte when not using DBCS, at the
2621 * end of the string or the byte can't start a double-byte. */
2622 if (enc_dbcs
== 0 || p
[1] == NUL
|| MB_BYTE2LEN(*p
) == 1)
2625 /* Return 1 when on the lead byte, 0 when on the tail byte. */
2626 return 1 - dbcs_head_off(base
, p
);
2630 * Find the next illegal byte sequence.
2635 pos_T pos
= curwin
->w_cursor
;
2639 char_u
*tofree
= NULL
;
2641 vimconv
.vc_type
= CONV_NONE
;
2642 if (enc_utf8
&& (enc_canon_props(curbuf
->b_p_fenc
) & ENC_8BIT
))
2644 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file,
2645 * possibly a utf-8 file with illegal bytes. Setup for conversion
2646 * from utf-8 to 'fileencoding'. */
2647 convert_setup(&vimconv
, p_enc
, curbuf
->b_p_fenc
);
2650 #ifdef FEAT_VIRTUALEDIT
2651 curwin
->w_cursor
.coladd
= 0;
2655 p
= ml_get_cursor();
2656 if (vimconv
.vc_type
!= CONV_NONE
)
2659 tofree
= string_convert(&vimconv
, p
, NULL
);
2667 /* Illegal means that there are not enough trail bytes (checked by
2668 * utf_ptr2len()) or too many of them (overlong sequence). */
2669 len
= utf_ptr2len(p
);
2670 if (*p
>= 0x80 && (len
== 1
2671 || utf_char2len(utf_ptr2char(p
)) != len
))
2673 if (vimconv
.vc_type
== CONV_NONE
)
2674 curwin
->w_cursor
.col
+= (colnr_T
)(p
- ml_get_cursor());
2679 len
= (int)(p
- tofree
);
2680 for (p
= ml_get_cursor(); *p
!= NUL
&& len
-- > 0; p
+= l
)
2683 curwin
->w_cursor
.col
+= l
;
2690 if (curwin
->w_cursor
.lnum
== curbuf
->b_ml
.ml_line_count
)
2692 ++curwin
->w_cursor
.lnum
;
2693 curwin
->w_cursor
.col
= 0;
2696 /* didn't find it: don't move and beep */
2697 curwin
->w_cursor
= pos
;
2702 convert_setup(&vimconv
, NULL
, NULL
);
2705 #if defined(HAVE_GTK2) || defined(PROTO)
2707 * Return TRUE if string "s" is a valid utf-8 string.
2708 * When "end" is NULL stop at the first NUL.
2709 * When "end" is positive stop there.
2712 utf_valid_string(s
, end
)
2719 while (end
== NULL
? *p
!= NUL
: p
< end
)
2721 if ((*p
& 0xc0) == 0x80)
2722 return FALSE
; /* invalid lead byte */
2723 l
= utf8len_tab
[*p
];
2724 if (end
!= NULL
&& p
+ l
> end
)
2725 return FALSE
; /* incomplete byte sequence */
2728 if ((*p
++ & 0xc0) != 0x80)
2729 return FALSE
; /* invalid trail byte */
2735 #if defined(FEAT_GUI) || defined(PROTO)
2737 * Special version of mb_tail_off() for use in ScreenLines[].
2740 dbcs_screen_tail_off(base
, p
)
2744 /* It can't be the first byte if a double-byte when not using DBCS, at the
2745 * end of the string or the byte can't start a double-byte.
2746 * For euc-jp an 0x8e byte always means we have a lead byte in the current
2748 if (*p
== NUL
|| p
[1] == NUL
2749 || (enc_dbcs
== DBCS_JPNU
&& *p
== 0x8e)
2750 || MB_BYTE2LEN(*p
) == 1)
2753 /* Return 1 when on the lead byte, 0 when on the tail byte. */
2754 return 1 - dbcs_screen_head_off(base
, p
);
2759 * If the cursor moves on an trail byte, set the cursor on the lead byte.
2760 * Thus it moves left if necessary.
2761 * Return TRUE when the cursor was adjusted.
2766 mb_adjustpos(&curwin
->w_cursor
);
2770 * Adjust position "*lp" to point to the first byte of a multi-byte character.
2771 * If it points to a tail byte it's moved backwards to the head byte.
2780 #ifdef FEAT_VIRTUALEDIT
2785 p
= ml_get(lp
->lnum
);
2786 lp
->col
-= (*mb_head_off
)(p
, p
+ lp
->col
);
2787 #ifdef FEAT_VIRTUALEDIT
2788 /* Reset "coladd" when the cursor would be on the right half of a
2789 * double-wide character. */
2791 && p
[lp
->col
] != TAB
2792 && vim_isprintc((*mb_ptr2char
)(p
+ lp
->col
))
2793 && ptr2cells(p
+ lp
->col
) > 1)
2800 * Return a pointer to the character before "*p", if there is one.
2804 char_u
*line
; /* start of the string */
2808 mb_ptr_back(line
, p
);
2813 * Return the character length of "str". Each multi-byte character (with
2814 * following composing characters) counts as one.
2826 for (count
= 0; *p
!= NUL
; count
++)
2827 p
+= (*mb_ptr2len
)(p
);
2832 #if defined(FEAT_SPELL) || defined(PROTO)
2834 * Like mb_charlen() but for a string with specified length.
2837 mb_charlen_len(str
, len
)
2844 for (count
= 0; *p
!= NUL
&& p
< str
+ len
; count
++)
2845 p
+= (*mb_ptr2len
)(p
);
2852 * Try to un-escape a multi-byte character.
2853 * Used for the "to" and "from" part of a mapping.
2854 * Return the un-escaped string if it is a multi-byte character, and advance
2855 * "pp" to just after the bytes that formed it.
2856 * Return NULL if no multi-byte char was found.
2862 static char_u buf
[MB_MAXBYTES
+ 1];
2866 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI
2867 * KS_EXTRA KE_CSI to CSI. */
2868 for (n
= 0; str
[n
] != NUL
&& m
<= MB_MAXBYTES
; ++n
)
2870 if (str
[n
] == K_SPECIAL
2871 && str
[n
+ 1] == KS_SPECIAL
2872 && str
[n
+ 2] == KE_FILLER
)
2874 buf
[m
++] = K_SPECIAL
;
2877 else if ((str
[n
] == K_SPECIAL
2882 && str
[n
+ 1] == KS_EXTRA
2883 && str
[n
+ 2] == (int)KE_CSI
)
2888 else if (str
[n
] == K_SPECIAL
2893 break; /* a special key can't be a multibyte char */
2898 /* Return a multi-byte character if it's found. An illegal sequence
2899 * will result in a 1 here. */
2900 if ((*mb_ptr2len
)(buf
) > 1)
2910 * Return TRUE if the character at "row"/"col" on the screen is the left side
2911 * of a double-width character.
2912 * Caller must make sure "row" and "col" are not invalid!
2915 mb_lefthalve(row
, col
)
2919 #ifdef FEAT_HANGULIN
2920 if (composing_hangul
)
2923 return (*mb_off2cells
)(LineOffset
[row
] + col
,
2924 LineOffset
[row
] + screen_Columns
) > 1;
2928 * Correct a position on the screen, if it's the right half of a double-wide
2929 * char move it to the left half. Returns the corrected column.
2932 mb_fix_col(col
, row
)
2936 col
= check_col(col
);
2937 row
= check_row(row
);
2938 if (has_mbyte
&& ScreenLines
!= NULL
&& col
> 0
2940 && ScreenLines
[LineOffset
[row
] + col
] != NUL
2941 && dbcs_screen_head_off(ScreenLines
+ LineOffset
[row
],
2942 ScreenLines
+ LineOffset
[row
] + col
))
2943 || (enc_utf8
&& ScreenLines
[LineOffset
[row
] + col
] == 0)))
2949 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO)
2950 static int enc_alias_search
__ARGS((char_u
*name
));
2953 * Skip the Vim specific head of a 'encoding' name.
2959 if (STRNCMP(p
, "2byte-", 6) == 0)
2961 if (STRNCMP(p
, "8bit-", 5) == 0)
2967 * Find the canonical name for encoding "enc".
2968 * When the name isn't recognized, returns "enc" itself, but with all lower
2969 * case characters and '_' replaced with '-'.
2970 * Returns an allocated string. NULL for out-of-memory.
2981 if (STRCMP(enc
, "default") == 0)
2983 /* Use the default encoding as it's found by set_init_1(). */
2984 r
= get_encoding_default();
2986 r
= (char_u
*)"latin1";
2987 return vim_strsave(r
);
2991 /* copy "enc" to allocated memory, with room for two '-' */
2992 r
= alloc((unsigned)(STRLEN(enc
) + 3));
2995 /* Make it all lower case and replace '_' with '-'. */
2997 for (s
= enc
; *s
!= NUL
; ++s
)
3002 *p
++ = TOLOWER_ASC(*s
);
3006 /* Skip "2byte-" and "8bit-". */
3009 /* Change "microsoft-cp" to "cp". Used in some spell files. */
3010 if (STRNCMP(p
, "microsoft-cp", 12) == 0)
3013 /* "iso8859" -> "iso-8859" */
3014 if (STRNCMP(p
, "iso8859", 7) == 0)
3016 STRMOVE(p
+ 4, p
+ 3);
3020 /* "iso-8859n" -> "iso-8859-n" */
3021 if (STRNCMP(p
, "iso-8859", 8) == 0 && p
[8] != '-')
3023 STRMOVE(p
+ 9, p
+ 8);
3027 /* "latin-N" -> "latinN" */
3028 if (STRNCMP(p
, "latin-", 6) == 0)
3029 STRMOVE(p
+ 5, p
+ 6);
3031 if (enc_canon_search(p
) >= 0)
3033 /* canonical name can be used unmodified */
3037 else if ((i
= enc_alias_search(p
)) >= 0)
3039 /* alias recognized, get canonical name */
3041 r
= vim_strsave((char_u
*)enc_canon_table
[i
].name
);
3048 * Search for an encoding alias of "name".
3049 * Returns -1 when not found.
3052 enc_alias_search(name
)
3057 for (i
= 0; enc_alias_table
[i
].name
!= NULL
; ++i
)
3058 if (STRCMP(name
, enc_alias_table
[i
].name
) == 0)
3059 return enc_alias_table
[i
].canon
;
3064 #if defined(FEAT_MBYTE) || defined(PROTO)
3066 #ifdef HAVE_LANGINFO_H
3067 # include <langinfo.h>
3071 * Get the canonicalized encoding of the current locale.
3072 * Returns an allocated string when successful, NULL when not.
3084 long acp
= GetACP();
3087 STRCPY(buf
, "ucs-2le");
3088 else if (acp
== 1252) /* cp1252 is used as latin1 */
3089 STRCPY(buf
, "latin1");
3091 sprintf(buf
, "cp%ld", acp
);
3093 # ifdef HAVE_NL_LANGINFO_CODESET
3094 if ((s
= nl_langinfo(CODESET
)) == NULL
|| *s
== NUL
)
3096 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE)
3097 if ((s
= setlocale(LC_CTYPE
, NULL
)) == NULL
|| *s
== NUL
)
3099 if ((s
= getenv("LC_ALL")) == NULL
|| *s
== NUL
)
3100 if ((s
= getenv("LC_CTYPE")) == NULL
|| *s
== NUL
)
3103 if (s
== NULL
|| *s
== NUL
)
3106 /* The most generic locale format is:
3107 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]]
3108 * If there is a '.' remove the part before it.
3109 * if there is something after the codeset, remove it.
3110 * Make the name lowercase and replace '_' with '-'.
3111 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn",
3112 * "ko_KR.EUC" == "euc-kr"
3114 if ((p
= (char *)vim_strchr((char_u
*)s
, '.')) != NULL
)
3116 if (p
> s
+ 2 && STRNICMP(p
+ 1, "EUC", 3) == 0
3117 && !isalnum((int)p
[4]) && p
[4] != '-' && p
[-3] == '_')
3119 /* copy "XY.EUC" to "euc-XY" to buf[10] */
3120 STRCPY(buf
+ 10, "euc-");
3129 for (i
= 0; s
[i
] != NUL
&& i
< (int)sizeof(buf
) - 1; ++i
)
3131 if (s
[i
] == '_' || s
[i
] == '-')
3133 else if (isalnum((int)s
[i
]))
3134 buf
[i
] = TOLOWER_ASC(s
[i
]);
3141 return enc_canonize((char_u
*)buf
);
3144 #if defined(WIN3264) || defined(PROTO)
3146 * Convert an encoding name to an MS-Windows codepage.
3147 * Returns zero if no codepage can be figured out.
3150 encname2codepage(name
)
3157 if (STRNCMP(p
, "8bit-", 5) == 0)
3159 else if (STRNCMP(p_enc
, "2byte-", 6) == 0)
3162 if (p
[0] == 'c' && p
[1] == 'p')
3164 else if ((idx
= enc_canon_search(p
)) >= 0)
3165 cp
= enc_canon_table
[idx
].codepage
;
3168 if (IsValidCodePage(cp
))
3174 # if defined(USE_ICONV) || defined(PROTO)
3176 static char_u
*iconv_string
__ARGS((vimconv_T
*vcp
, char_u
*str
, int slen
, int *unconvlenp
));
3179 * Call iconv_open() with a check if iconv() works properly (there are broken
3181 * Returns (void *)-1 if failed.
3182 * (should return iconv_t, but that causes problems with prototypes).
3185 my_iconv_open(to
, from
)
3190 #define ICONV_TESTLEN 400
3191 char_u tobuf
[ICONV_TESTLEN
];
3194 static int iconv_ok
= -1;
3196 if (iconv_ok
== FALSE
)
3197 return (void *)-1; /* detected a broken iconv() previously */
3199 #ifdef DYNAMIC_ICONV
3200 /* Check if the iconv.dll can be found. */
3201 if (!iconv_enabled(TRUE
))
3205 fd
= iconv_open((char *)enc_skip(to
), (char *)enc_skip(from
));
3207 if (fd
!= (iconv_t
)-1 && iconv_ok
== -1)
3210 * Do a dummy iconv() call to check if it actually works. There is a
3211 * version of iconv() on Linux that is broken. We can't ignore it,
3212 * because it's wide-spread. The symptoms are that after outputting
3213 * the initial shift state the "to" pointer is NULL and conversion
3214 * stops for no apparent reason after about 8160 characters.
3217 tolen
= ICONV_TESTLEN
;
3218 (void)iconv(fd
, NULL
, NULL
, &p
, &tolen
);
3233 * Convert the string "str[slen]" with iconv().
3234 * If "unconvlenp" is not NULL handle the string ending in an incomplete
3235 * sequence and set "*unconvlenp" to the length of it.
3236 * Returns the converted string in allocated memory. NULL for an error.
3239 iconv_string(vcp
, str
, slen
, unconvlenp
)
3251 char_u
*result
= NULL
;
3259 if (len
== 0 || ICONV_ERRNO
== ICONV_E2BIG
)
3261 /* Allocate enough room for most conversions. When re-allocating
3262 * increase the buffer size. */
3263 len
= len
+ fromlen
* 2 + 40;
3264 p
= alloc((unsigned)len
);
3265 if (p
!= NULL
&& done
> 0)
3266 mch_memmove(p
, result
, done
);
3269 if (result
== NULL
) /* out of memory */
3273 to
= (char *)result
+ done
;
3274 tolen
= len
- done
- 2;
3275 /* Avoid a warning for systems with a wrong iconv() prototype by
3276 * casting the second argument to void *. */
3277 if (iconv(vcp
->vc_fd
, (void *)&from
, &fromlen
, &to
, &tolen
)
3280 /* Finished, append a NUL. */
3285 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded
3286 * iconv library may use one of them. */
3287 if (!vcp
->vc_fail
&& unconvlenp
!= NULL
3288 && (ICONV_ERRNO
== ICONV_EINVAL
|| ICONV_ERRNO
== EINVAL
))
3290 /* Handle an incomplete sequence at the end. */
3292 *unconvlenp
= (int)fromlen
;
3296 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded
3297 * iconv library may use one of them. */
3298 else if (!vcp
->vc_fail
3299 && (ICONV_ERRNO
== ICONV_EILSEQ
|| ICONV_ERRNO
== EILSEQ
3300 || ICONV_ERRNO
== ICONV_EINVAL
|| ICONV_ERRNO
== EINVAL
))
3302 /* Can't convert: insert a '?' and skip a character. This assumes
3303 * conversion from 'encoding' to something else. In other
3304 * situations we don't know what to skip anyway. */
3306 if ((*mb_ptr2cells
)((char_u
*)from
) > 1)
3309 l
= utfc_ptr2len_len((char_u
*)from
, (int)fromlen
);
3312 l
= (*mb_ptr2len
)((char_u
*)from
);
3313 if (l
> (int)fromlen
)
3319 else if (ICONV_ERRNO
!= ICONV_E2BIG
)
3321 /* conversion failed */
3326 /* Not enough room or skipping illegal sequence. */
3327 done
= to
- (char *)result
;
3332 # if defined(DYNAMIC_ICONV) || defined(PROTO)
3334 * Dynamically load the "iconv.dll" on Win32.
3337 #ifndef DYNAMIC_ICONV /* just generating prototypes */
3338 # define HINSTANCE int
3340 static HINSTANCE hIconvDLL
= 0;
3341 static HINSTANCE hMsvcrtDLL
= 0;
3343 # ifndef DYNAMIC_ICONV_DLL
3344 # define DYNAMIC_ICONV_DLL "iconv.dll"
3345 # define DYNAMIC_ICONV_DLL_ALT "libiconv.dll"
3347 # ifndef DYNAMIC_MSVCRT_DLL
3348 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll"
3352 * Try opening the iconv.dll and return TRUE if iconv() can be used.
3355 iconv_enabled(verbose
)
3358 if (hIconvDLL
!= 0 && hMsvcrtDLL
!= 0)
3360 hIconvDLL
= LoadLibrary(DYNAMIC_ICONV_DLL
);
3361 if (hIconvDLL
== 0) /* sometimes it's called libiconv.dll */
3362 hIconvDLL
= LoadLibrary(DYNAMIC_ICONV_DLL_ALT
);
3364 hMsvcrtDLL
= LoadLibrary(DYNAMIC_MSVCRT_DLL
);
3365 if (hIconvDLL
== 0 || hMsvcrtDLL
== 0)
3367 /* Only give the message when 'verbose' is set, otherwise it might be
3368 * done whenever a conversion is attempted. */
3369 if (verbose
&& p_verbose
> 0)
3373 hIconvDLL
== 0 ? DYNAMIC_ICONV_DLL
: DYNAMIC_MSVCRT_DLL
);
3380 iconv
= (void *)GetProcAddress(hIconvDLL
, "libiconv");
3381 iconv_open
= (void *)GetProcAddress(hIconvDLL
, "libiconv_open");
3382 iconv_close
= (void *)GetProcAddress(hIconvDLL
, "libiconv_close");
3383 iconvctl
= (void *)GetProcAddress(hIconvDLL
, "libiconvctl");
3384 iconv_errno
= (void *)GetProcAddress(hMsvcrtDLL
, "_errno");
3385 if (iconv
== NULL
|| iconv_open
== NULL
|| iconv_close
== NULL
3386 || iconvctl
== NULL
|| iconv_errno
== NULL
)
3389 if (verbose
&& p_verbose
> 0)
3392 EMSG2(_(e_loadfunc
), "for libiconv");
3403 /* Don't use iconv() when inputting or outputting characters. */
3404 if (input_conv
.vc_type
== CONV_ICONV
)
3405 convert_setup(&input_conv
, NULL
, NULL
);
3406 if (output_conv
.vc_type
== CONV_ICONV
)
3407 convert_setup(&output_conv
, NULL
, NULL
);
3410 FreeLibrary(hIconvDLL
);
3411 if (hMsvcrtDLL
!= 0)
3412 FreeLibrary(hMsvcrtDLL
);
3416 # endif /* DYNAMIC_ICONV */
3417 # endif /* USE_ICONV */
3419 #endif /* FEAT_MBYTE */
3421 #if defined(FEAT_XIM) || defined(PROTO)
3423 # ifdef FEAT_GUI_GTK
3424 static int xim_has_preediting
INIT(= FALSE
); /* IM current status */
3427 * Set preedit_start_col to the current cursor position.
3430 init_preedit_start_col(void)
3432 if (State
& CMDLINE
)
3433 preedit_start_col
= cmdline_getvcol_cursor();
3434 else if (curwin
!= NULL
)
3435 getvcol(curwin
, &curwin
->w_cursor
, &preedit_start_col
, NULL
, NULL
);
3436 /* Prevent that preediting marks the buffer as changed. */
3437 xim_changed_while_preediting
= curbuf
->b_changed
;
3441 # if defined(HAVE_GTK2) && !defined(PROTO)
3443 static int im_is_active
= FALSE
; /* IM is enabled for current mode */
3444 static int preedit_is_active
= FALSE
;
3445 static int im_preedit_cursor
= 0; /* cursor offset in characters */
3446 static int im_preedit_trailing
= 0; /* number of characters after cursor */
3448 static unsigned long im_commit_handler_id
= 0;
3449 static unsigned int im_activatekey_keyval
= GDK_VoidSymbol
;
3450 static unsigned int im_activatekey_state
= 0;
3453 im_set_active(int active
)
3457 was_active
= !!im_is_active
;
3458 im_is_active
= (active
&& !p_imdisable
);
3460 if (im_is_active
!= was_active
)
3465 xim_set_focus(int focus
)
3470 gtk_im_context_focus_in(xic
);
3472 gtk_im_context_focus_out(xic
);
3477 im_set_position(int row
, int col
)
3483 area
.x
= FILL_X(col
);
3484 area
.y
= FILL_Y(row
);
3485 area
.width
= gui
.char_width
* (mb_lefthalve(row
, col
) ? 2 : 1);
3486 area
.height
= gui
.char_height
;
3488 gtk_im_context_set_cursor_location(xic
, &area
);
3492 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */
3494 xim_set_preedit(void)
3496 im_set_position(gui
.row
, gui
.col
);
3501 im_add_to_input(char_u
*str
, int len
)
3503 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */
3504 if (input_conv
.vc_type
!= CONV_NONE
)
3506 str
= string_convert(&input_conv
, str
, &len
);
3507 g_return_if_fail(str
!= NULL
);
3510 add_to_input_buf_csi(str
, len
);
3512 if (input_conv
.vc_type
!= CONV_NONE
)
3515 if (p_mh
) /* blank out the pointer if necessary */
3516 gui_mch_mousehide(TRUE
);
3520 im_delete_preedit(void)
3522 char_u bskey
[] = {CSI
, 'k', 'b'};
3523 char_u delkey
[] = {CSI
, 'k', 'D'};
3527 im_preedit_cursor
= 0;
3530 for (; im_preedit_cursor
> 0; --im_preedit_cursor
)
3531 add_to_input_buf(bskey
, (int)sizeof(bskey
));
3533 for (; im_preedit_trailing
> 0; --im_preedit_trailing
)
3534 add_to_input_buf(delkey
, (int)sizeof(delkey
));
3538 * Move the cursor left by "num_move_back" characters.
3539 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for
3540 * these K_LEFT keys.
3543 im_correct_cursor(int num_move_back
)
3545 char_u backkey
[] = {CSI
, 'k', 'l'};
3549 # ifdef FEAT_RIGHTLEFT
3550 if ((State
& CMDLINE
) == 0 && curwin
!= NULL
&& curwin
->w_p_rl
)
3553 for (; num_move_back
> 0; --num_move_back
)
3554 add_to_input_buf(backkey
, (int)sizeof(backkey
));
3557 static int xim_expected_char
= NUL
;
3558 static int xim_ignored_char
= FALSE
;
3561 * Update the mode and cursor while in an IM callback.
3568 old_vgetc_busy
= vgetc_busy
;
3571 vgetc_busy
= old_vgetc_busy
;
3577 * Callback invoked when the user finished preediting.
3578 * Put the final string into the input buffer.
3581 im_commit_cb(GtkIMContext
*context UNUSED
,
3583 gpointer data UNUSED
)
3585 int slen
= (int)STRLEN(str
);
3586 int add_to_input
= TRUE
;
3589 int commit_with_preedit
= TRUE
;
3593 xim_log("im_commit_cb(): %s\n", str
);
3596 /* The imhangul module doesn't reset the preedit string before
3597 * committing. Call im_delete_preedit() to work around that. */
3598 im_delete_preedit();
3600 /* Indicate that preediting has finished. */
3601 if (preedit_start_col
== MAXCOL
)
3603 init_preedit_start_col();
3604 commit_with_preedit
= FALSE
;
3607 /* The thing which setting "preedit_start_col" to MAXCOL means that
3608 * "preedit_start_col" will be set forcely when calling
3609 * preedit_changed_cb() next time.
3610 * "preedit_start_col" should not reset with MAXCOL on this part. Vim
3611 * is simulating the preediting by using add_to_input_str(). when
3612 * preedit begin immediately before committed, the typebuf is not
3613 * flushed to screen, then it can't get correct "preedit_start_col".
3614 * Thus, it should calculate the cells by adding cells of the committed
3616 if (input_conv
.vc_type
!= CONV_NONE
)
3618 im_str
= string_convert(&input_conv
, (char_u
*)str
, &len
);
3619 g_return_if_fail(im_str
!= NULL
);
3622 im_str
= (char_u
*)str
;
3624 for (p
= im_str
; p
< im_str
+ len
; p
+= (*mb_ptr2len
)(p
))
3625 clen
+= (*mb_ptr2cells
)(p
);
3626 if (input_conv
.vc_type
!= CONV_NONE
)
3628 preedit_start_col
+= clen
;
3630 /* Is this a single character that matches a keypad key that's just
3631 * been pressed? If so, we don't want it to be entered as such - let
3632 * us carry on processing the raw keycode so that it may be used in
3633 * mappings as <kSomething>. */
3634 if (xim_expected_char
!= NUL
)
3636 /* We're currently processing a keypad or other special key */
3637 if (slen
== 1 && str
[0] == xim_expected_char
)
3639 /* It's a match - don't do it here */
3640 xim_ignored_char
= TRUE
;
3641 add_to_input
= FALSE
;
3646 xim_ignored_char
= FALSE
;
3651 im_add_to_input((char_u
*)str
, slen
);
3653 /* Inserting chars while "im_is_active" is set does not cause a change of
3654 * buffer. When the chars are committed the buffer must be marked as
3656 if (!commit_with_preedit
)
3657 preedit_start_col
= MAXCOL
;
3659 /* This flag is used in changed() at next call. */
3660 xim_changed_while_preediting
= TRUE
;
3662 if (gtk_main_level() > 0)
3667 * Callback invoked after start to the preedit.
3670 im_preedit_start_cb(GtkIMContext
*context UNUSED
, gpointer data UNUSED
)
3673 xim_log("im_preedit_start_cb()\n");
3676 im_is_active
= TRUE
;
3677 preedit_is_active
= TRUE
;
3678 gui_update_cursor(TRUE
, FALSE
);
3683 * Callback invoked after end to the preedit.
3686 im_preedit_end_cb(GtkIMContext
*context UNUSED
, gpointer data UNUSED
)
3689 xim_log("im_preedit_end_cb()\n");
3691 im_delete_preedit();
3693 /* Indicate that preediting has finished */
3694 preedit_start_col
= MAXCOL
;
3695 xim_has_preediting
= FALSE
;
3698 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was
3699 * switched off unintentionally. We now use preedit_is_active (added by
3701 im_is_active
= FALSE
;
3703 preedit_is_active
= FALSE
;
3704 gui_update_cursor(TRUE
, FALSE
);
3709 * Callback invoked after changes to the preedit string. If the preedit
3710 * string was empty before, remember the preedit start column so we know
3711 * where to apply feedback attributes. Delete the previous preedit string
3712 * if there was one, save the new preedit cursor offset, and put the new
3713 * string into the input buffer.
3715 * TODO: The pragmatic "put into input buffer" approach used here has
3716 * several fundamental problems:
3718 * - The characters in the preedit string are subject to remapping.
3719 * That's broken, only the finally committed string should be remapped.
3721 * - There is a race condition involved: The retrieved value for the
3722 * current cursor position will be wrong if any unprocessed characters
3723 * are still queued in the input buffer.
3725 * - Due to the lack of synchronization between the file buffer in memory
3726 * and any typed characters, it's practically impossible to implement the
3727 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM
3728 * modules for languages such as Thai are likely to rely on this feature
3729 * for proper operation.
3731 * Conclusions: I think support for preediting needs to be moved to the
3732 * core parts of Vim. Ideally, until it has been committed, the preediting
3733 * string should only be displayed and not affect the buffer content at all.
3734 * The question how to deal with the synchronization issue still remains.
3735 * Circumventing the input buffer is probably not desirable. Anyway, I think
3736 * implementing "retrieve_surrounding" is the only hard problem.
3738 * One way to solve all of this in a clean manner would be to queue all key
3739 * press/release events "as is" in the input buffer, and apply the IM filtering
3740 * at the receiving end of the queue. This, however, would have a rather large
3741 * impact on the code base. If there is an easy way to force processing of all
3742 * remaining input from within the "retrieve_surrounding" signal handler, this
3743 * might not be necessary. Gotta ask on vim-dev for opinions.
3746 im_preedit_changed_cb(GtkIMContext
*context
, gpointer data UNUSED
)
3748 char *preedit_string
= NULL
;
3749 int cursor_index
= 0;
3750 int num_move_back
= 0;
3755 gtk_im_context_get_preedit_string(context
,
3756 &preedit_string
, NULL
,
3760 xim_log("im_preedit_changed_cb(): %s\n", preedit_string
);
3763 g_return_if_fail(preedit_string
!= NULL
); /* just in case */
3765 /* If preedit_start_col is MAXCOL set it to the current cursor position. */
3766 if (preedit_start_col
== MAXCOL
&& preedit_string
[0] != '\0')
3768 xim_has_preediting
= TRUE
;
3770 /* Urgh, this breaks if the input buffer isn't empty now */
3771 init_preedit_start_col();
3773 else if (cursor_index
== 0 && preedit_string
[0] == '\0')
3775 xim_has_preediting
= FALSE
;
3777 /* If at the start position (after typing backspace)
3778 * preedit_start_col must be reset. */
3779 preedit_start_col
= MAXCOL
;
3782 im_delete_preedit();
3785 * Compute the end of the preediting area: "preedit_end_col".
3786 * According to the documentation of gtk_im_context_get_preedit_string(),
3787 * the cursor_pos output argument returns the offset in bytes. This is
3788 * unfortunately not true -- real life shows the offset is in characters,
3789 * and the GTK+ source code agrees with me. Will file a bug later.
3791 if (preedit_start_col
!= MAXCOL
)
3792 preedit_end_col
= preedit_start_col
;
3793 str
= (char_u
*)preedit_string
;
3794 for (p
= str
, i
= 0; *p
!= NUL
; p
+= utf_byte2len(*p
), ++i
)
3798 is_composing
= ((*p
& 0x80) != 0 && utf_iscomposing(utf_ptr2char(p
)));
3800 * These offsets are used as counters when generating <BS> and <Del>
3801 * to delete the preedit string. So don't count composing characters
3802 * unless 'delcombine' is enabled.
3804 if (!is_composing
|| p_deco
)
3806 if (i
< cursor_index
)
3807 ++im_preedit_cursor
;
3809 ++im_preedit_trailing
;
3811 if (!is_composing
&& i
>= cursor_index
)
3813 /* This is essentially the same as im_preedit_trailing, except
3814 * composing characters are not counted even if p_deco is set. */
3817 if (preedit_start_col
!= MAXCOL
)
3818 preedit_end_col
+= utf_ptr2cells(p
);
3823 im_add_to_input(str
, (int)(p
- str
));
3824 im_correct_cursor(num_move_back
);
3827 g_free(preedit_string
);
3829 if (gtk_main_level() > 0)
3834 * Translate the Pango attributes at iter to Vim highlighting attributes.
3835 * Ignore attributes not supported by Vim highlighting. This shouldn't have
3836 * too much impact -- right now we handle even more attributes than necessary
3837 * for the IM modules I tested with.
3840 translate_pango_attributes(PangoAttrIterator
*iter
)
3842 PangoAttribute
*attr
;
3843 int char_attr
= HL_NORMAL
;
3845 attr
= pango_attr_iterator_get(iter
, PANGO_ATTR_UNDERLINE
);
3846 if (attr
!= NULL
&& ((PangoAttrInt
*)attr
)->value
3847 != (int)PANGO_UNDERLINE_NONE
)
3848 char_attr
|= HL_UNDERLINE
;
3850 attr
= pango_attr_iterator_get(iter
, PANGO_ATTR_WEIGHT
);
3851 if (attr
!= NULL
&& ((PangoAttrInt
*)attr
)->value
>= (int)PANGO_WEIGHT_BOLD
)
3852 char_attr
|= HL_BOLD
;
3854 attr
= pango_attr_iterator_get(iter
, PANGO_ATTR_STYLE
);
3855 if (attr
!= NULL
&& ((PangoAttrInt
*)attr
)->value
3856 != (int)PANGO_STYLE_NORMAL
)
3857 char_attr
|= HL_ITALIC
;
3859 attr
= pango_attr_iterator_get(iter
, PANGO_ATTR_BACKGROUND
);
3862 const PangoColor
*color
= &((PangoAttrColor
*)attr
)->color
;
3864 /* Assume inverse if black background is requested */
3865 if ((color
->red
| color
->green
| color
->blue
) == 0)
3866 char_attr
|= HL_INVERSE
;
3873 * Retrieve the highlighting attributes at column col in the preedit string.
3874 * Return -1 if not in preediting mode or if col is out of range.
3877 im_get_feedback_attr(int col
)
3879 char *preedit_string
= NULL
;
3880 PangoAttrList
*attr_list
= NULL
;
3886 gtk_im_context_get_preedit_string(xic
, &preedit_string
, &attr_list
, NULL
);
3888 if (preedit_string
!= NULL
&& attr_list
!= NULL
)
3892 /* Get the byte index as used by PangoAttrIterator */
3893 for (idx
= 0; col
> 0 && preedit_string
[idx
] != '\0'; --col
)
3894 idx
+= utfc_ptr2len((char_u
*)preedit_string
+ idx
);
3896 if (preedit_string
[idx
] != '\0')
3898 PangoAttrIterator
*iter
;
3901 char_attr
= HL_NORMAL
;
3902 iter
= pango_attr_list_get_iterator(attr_list
);
3904 /* Extract all relevant attributes from the list. */
3907 pango_attr_iterator_range(iter
, &start
, &end
);
3909 if (idx
>= start
&& idx
< end
)
3910 char_attr
|= translate_pango_attributes(iter
);
3912 while (pango_attr_iterator_next(iter
));
3914 pango_attr_iterator_destroy(iter
);
3918 if (attr_list
!= NULL
)
3919 pango_attr_list_unref(attr_list
);
3920 g_free(preedit_string
);
3929 xim_log("xim_init()\n");
3932 g_return_if_fail(gui
.drawarea
!= NULL
);
3933 g_return_if_fail(gui
.drawarea
->window
!= NULL
);
3935 xic
= gtk_im_multicontext_new();
3938 im_commit_handler_id
= g_signal_connect(G_OBJECT(xic
), "commit",
3939 G_CALLBACK(&im_commit_cb
), NULL
);
3940 g_signal_connect(G_OBJECT(xic
), "preedit_changed",
3941 G_CALLBACK(&im_preedit_changed_cb
), NULL
);
3942 g_signal_connect(G_OBJECT(xic
), "preedit_start",
3943 G_CALLBACK(&im_preedit_start_cb
), NULL
);
3944 g_signal_connect(G_OBJECT(xic
), "preedit_end",
3945 G_CALLBACK(&im_preedit_end_cb
), NULL
);
3947 gtk_im_context_set_client_window(xic
, gui
.drawarea
->window
);
3954 xim_log("im_shutdown()\n");
3959 gtk_im_context_focus_out(xic
);
3960 g_object_unref(xic
);
3963 im_is_active
= FALSE
;
3964 im_commit_handler_id
= 0;
3965 preedit_start_col
= MAXCOL
;
3966 xim_has_preediting
= FALSE
;
3970 * Convert the string argument to keyval and state for GdkEventKey.
3971 * If str is valid return TRUE, otherwise FALSE.
3973 * See 'imactivatekey' for documentation of the format.
3976 im_string_to_keyval(const char *str
, unsigned int *keyval
, unsigned int *state
)
3978 const char *mods_end
;
3979 unsigned tmp_keyval
;
3980 unsigned tmp_state
= 0;
3982 mods_end
= strrchr(str
, '-');
3983 mods_end
= (mods_end
!= NULL
) ? mods_end
+ 1 : str
;
3985 /* Parse modifier keys */
3986 while (str
< mods_end
)
3990 case 'S': case 's': tmp_state
|= (unsigned)GDK_SHIFT_MASK
; break;
3991 case 'L': case 'l': tmp_state
|= (unsigned)GDK_LOCK_MASK
; break;
3992 case 'C': case 'c': tmp_state
|= (unsigned)GDK_CONTROL_MASK
;break;
3993 case '1': tmp_state
|= (unsigned)GDK_MOD1_MASK
; break;
3994 case '2': tmp_state
|= (unsigned)GDK_MOD2_MASK
; break;
3995 case '3': tmp_state
|= (unsigned)GDK_MOD3_MASK
; break;
3996 case '4': tmp_state
|= (unsigned)GDK_MOD4_MASK
; break;
3997 case '5': tmp_state
|= (unsigned)GDK_MOD5_MASK
; break;
4002 tmp_keyval
= gdk_keyval_from_name(str
);
4004 if (tmp_keyval
== 0 || tmp_keyval
== GDK_VoidSymbol
)
4008 *keyval
= tmp_keyval
;
4016 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an
4017 * empty string is also regarded as valid.
4019 * Note: The numerical key value of p_imak is cached if it was valid; thus
4020 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever
4021 * 'imak' changes. This is currently the case but not obvious -- should
4022 * probably rename the function for clarity.
4025 im_xim_isvalid_imactivate(void)
4027 if (p_imak
[0] == NUL
)
4029 im_activatekey_keyval
= GDK_VoidSymbol
;
4030 im_activatekey_state
= 0;
4034 return im_string_to_keyval((const char *)p_imak
,
4035 &im_activatekey_keyval
,
4036 &im_activatekey_state
);
4040 im_synthesize_keypress(unsigned int keyval
, unsigned int state
)
4044 # ifdef HAVE_GTK_MULTIHEAD
4045 event
= (GdkEventKey
*)gdk_event_new(GDK_KEY_PRESS
);
4046 g_object_ref(gui
.drawarea
->window
); /* unreffed by gdk_event_free() */
4048 event
= (GdkEventKey
*)g_malloc0((gulong
)sizeof(GdkEvent
));
4049 event
->type
= GDK_KEY_PRESS
;
4051 event
->window
= gui
.drawarea
->window
;
4052 event
->send_event
= TRUE
;
4053 event
->time
= GDK_CURRENT_TIME
;
4054 event
->state
= state
;
4055 event
->keyval
= keyval
;
4056 event
->hardware_keycode
= /* needed for XIM */
4057 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event
->window
), (KeySym
)keyval
);
4059 event
->string
= NULL
;
4061 gtk_im_context_filter_keypress(xic
, event
);
4063 /* For consistency, also send the corresponding release event. */
4064 event
->type
= GDK_KEY_RELEASE
;
4065 event
->send_event
= FALSE
;
4066 gtk_im_context_filter_keypress(xic
, event
);
4068 # ifdef HAVE_GTK_MULTIHEAD
4069 gdk_event_free((GdkEvent
*)event
);
4081 * The third-party imhangul module (and maybe others too) ignores
4082 * gtk_im_context_reset() or at least doesn't reset the active state.
4083 * Thus sending imactivatekey would turn it off if it was on before,
4084 * which is clearly not what we want. Fortunately we can work around
4085 * that for imhangul by sending GDK_Escape, but I don't know if it
4086 * works with all IM modules that support an activation key :/
4088 * An alternative approach would be to destroy the IM context and
4089 * recreate it. But that means loading/unloading the IM module on
4090 * every mode switch, which causes a quite noticable delay even on
4091 * my rather fast box...
4093 * Moreover, there are some XIM which cannot respond to
4094 * im_synthesize_keypress(). we hope that they reset by
4097 if (im_activatekey_keyval
!= GDK_VoidSymbol
&& im_is_active
)
4098 im_synthesize_keypress(GDK_Escape
, 0U);
4100 gtk_im_context_reset(xic
);
4103 * HACK for Ami: This sequence of function calls makes Ami handle
4104 * the IM reset graciously, without breaking loads of other stuff.
4105 * It seems to force English mode as well, which is exactly what we
4106 * want because it makes the Ami status display work reliably.
4108 gtk_im_context_set_use_preedit(xic
, FALSE
);
4114 gtk_im_context_set_use_preedit(xic
, TRUE
);
4115 xim_set_focus(gui
.in_focus
);
4117 if (im_activatekey_keyval
!= GDK_VoidSymbol
)
4121 g_signal_handler_block(xic
, im_commit_handler_id
);
4122 im_synthesize_keypress(im_activatekey_keyval
,
4123 im_activatekey_state
);
4124 g_signal_handler_unblock(xic
, im_commit_handler_id
);
4131 xim_set_focus(gui
.in_focus
);
4136 preedit_start_col
= MAXCOL
;
4137 xim_has_preediting
= FALSE
;
4141 xim_queue_key_press_event(GdkEventKey
*event
, int down
)
4146 * Workaround GTK2 XIM 'feature' that always converts keypad keys to
4147 * chars., even when not part of an IM sequence (ref. feature of
4149 * Flag any keypad keys that might represent a single char.
4150 * If this (on its own - i.e., not part of an IM sequence) is
4151 * committed while we're processing one of these keys, we can ignore
4152 * that commit and go ahead & process it ourselves. That way we can
4153 * still distinguish keypad keys for use in mappings.
4154 * Also add GDK_space to make <S-Space> work.
4156 switch (event
->keyval
)
4158 case GDK_KP_Add
: xim_expected_char
= '+'; break;
4159 case GDK_KP_Subtract
: xim_expected_char
= '-'; break;
4160 case GDK_KP_Divide
: xim_expected_char
= '/'; break;
4161 case GDK_KP_Multiply
: xim_expected_char
= '*'; break;
4162 case GDK_KP_Decimal
: xim_expected_char
= '.'; break;
4163 case GDK_KP_Equal
: xim_expected_char
= '='; break;
4164 case GDK_KP_0
: xim_expected_char
= '0'; break;
4165 case GDK_KP_1
: xim_expected_char
= '1'; break;
4166 case GDK_KP_2
: xim_expected_char
= '2'; break;
4167 case GDK_KP_3
: xim_expected_char
= '3'; break;
4168 case GDK_KP_4
: xim_expected_char
= '4'; break;
4169 case GDK_KP_5
: xim_expected_char
= '5'; break;
4170 case GDK_KP_6
: xim_expected_char
= '6'; break;
4171 case GDK_KP_7
: xim_expected_char
= '7'; break;
4172 case GDK_KP_8
: xim_expected_char
= '8'; break;
4173 case GDK_KP_9
: xim_expected_char
= '9'; break;
4174 case GDK_space
: xim_expected_char
= ' '; break;
4175 default: xim_expected_char
= NUL
;
4177 xim_ignored_char
= FALSE
;
4181 * When typing fFtT, XIM may be activated. Thus it must pass
4182 * gtk_im_context_filter_keypress() in Normal mode.
4183 * And while doing :sh too.
4185 if (xic
!= NULL
&& !p_imdisable
4186 && (State
& (INSERT
| CMDLINE
| NORMAL
| EXTERNCMD
)) != 0)
4189 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is
4190 * always aware of the current status of IM, and can even emulate
4191 * the activation key for modules that don't support one.
4193 if (event
->keyval
== im_activatekey_keyval
4194 && (event
->state
& im_activatekey_state
) == im_activatekey_state
)
4196 unsigned int state_mask
;
4198 /* Require the state of the 3 most used modifiers to match exactly.
4199 * Otherwise e.g. <S-C-space> would be unusable for other purposes
4200 * if the IM activate key is <S-space>. */
4201 state_mask
= im_activatekey_state
;
4202 state_mask
|= ((int)GDK_SHIFT_MASK
| (int)GDK_CONTROL_MASK
4203 | (int)GDK_MOD1_MASK
);
4205 if ((event
->state
& state_mask
) != im_activatekey_state
)
4208 /* Don't send it a second time on GDK_KEY_RELEASE. */
4209 if (event
->type
!= GDK_KEY_PRESS
)
4212 if (map_to_exists_mode((char_u
*)"", LANGMAP
, FALSE
))
4214 im_set_active(FALSE
);
4216 /* ":lmap" mappings exists, toggle use of mappings. */
4218 if (State
& LANGMAP
)
4220 curbuf
->b_p_iminsert
= B_IMODE_NONE
;
4225 curbuf
->b_p_iminsert
= B_IMODE_LMAP
;
4231 return gtk_im_context_filter_keypress(xic
, event
);
4234 /* Don't filter events through the IM context if IM isn't active
4235 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module
4236 * not doing anything before the activation key was sent. */
4237 if (im_activatekey_keyval
== GDK_VoidSymbol
|| im_is_active
)
4239 int imresult
= gtk_im_context_filter_keypress(xic
, event
);
4241 /* Some XIM send following sequence:
4242 * 1. preedited string.
4243 * 2. committed string.
4244 * 3. line changed key.
4245 * 4. preedited string.
4246 * 5. remove preedited string.
4247 * if 3, Vim can't move back the above line for 5.
4248 * thus, this part should not parse the key. */
4249 if (!imresult
&& preedit_start_col
!= MAXCOL
4250 && event
->keyval
== GDK_Return
)
4252 im_synthesize_keypress(GDK_Return
, 0U);
4256 /* If XIM tried to commit a keypad key as a single char.,
4257 * ignore it so we can use the keypad key 'raw', for mappings. */
4258 if (xim_expected_char
!= NUL
&& xim_ignored_char
)
4259 /* We had a keypad key, and XIM tried to thieve it */
4262 /* Normal processing */
4273 return im_is_active
;
4276 # else /* !HAVE_GTK2 */
4278 static int xim_is_active
= FALSE
; /* XIM should be active in the current
4280 static int xim_has_focus
= FALSE
; /* XIM is really being used for Vim */
4282 static XIMStyle input_style
;
4283 static int status_area_enabled
= TRUE
;
4288 # include <gdk/gdkwin32.h>
4290 # include <gdk/gdkx.h>
4294 /* Define a few things to be able to generate prototypes while not configured
4297 # define gboolean int
4298 typedef int GdkEvent
;
4299 typedef int GdkEventKey
;
4304 #if defined(FEAT_GUI_GTK) || defined(PROTO)
4305 static int preedit_buf_len
= 0;
4306 static int xim_can_preediting
INIT(= FALSE
); /* XIM in showmode() */
4307 static int xim_input_style
;
4308 #ifndef FEAT_GUI_GTK
4309 # define gboolean int
4311 static gboolean use_status_area
= 0;
4313 static int im_xim_str2keycode
__ARGS((unsigned int *code
, unsigned int *state
));
4314 static void im_xim_send_event_imactivate
__ARGS((void));
4317 * Convert string to keycode and state for XKeyEvent.
4318 * When string is valid return OK, when invalid return FAIL.
4320 * See 'imactivatekey' documentation for the format.
4323 im_xim_str2keycode(code
, state
)
4325 unsigned int *state
;
4329 unsigned keycode
= 0, keystate
= 0;
4337 len
= STRLEN(p_imak
);
4338 for (flag_end
= p_imak
+ len
- 1;
4339 flag_end
> p_imak
&& *flag_end
!= '-'; --flag_end
)
4342 /* Parse modifier keys */
4343 for (str
= p_imak
; str
< flag_end
; ++str
)
4348 keystate
|= ShiftMask
;
4351 keystate
|= LockMask
;
4354 keystate
|= ControlMask
;
4357 keystate
|= Mod1Mask
;
4360 keystate
|= Mod2Mask
;
4363 keystate
|= Mod3Mask
;
4366 keystate
|= Mod4Mask
;
4369 keystate
|= Mod5Mask
;
4380 /* Get keycode from string. */
4381 gui_get_x11_windis(&window
, &display
);
4383 keycode
= XKeysymToKeycode(display
, XStringToKeysym((char *)str
));
4396 im_xim_send_event_imactivate()
4398 /* Force turn on preedit state by symulate keypress event.
4399 * Keycode and state is specified by 'imactivatekey'.
4403 gui_get_x11_windis(&ev
.window
, &ev
.display
);
4404 ev
.root
= RootWindow(ev
.display
, DefaultScreen(ev
.display
));
4405 ev
.subwindow
= None
;
4406 ev
.time
= CurrentTime
;
4413 if (im_xim_str2keycode(&ev
.keycode
, &ev
.state
) == OK
)
4414 XSendEvent(ev
.display
, ev
.window
, 1, KeyPressMask
, (XEvent
*)&ev
);
4418 * Return TRUE if 'imactivatekey' has a valid value.
4421 im_xim_isvalid_imactivate()
4423 return im_xim_str2keycode(NULL
, NULL
) == OK
;
4425 #endif /* FEAT_GUI_GTK */
4428 * Switch using XIM on/off. This is used by the code that changes "State".
4431 im_set_active(active
)
4437 /* If 'imdisable' is set, XIM is never active. */
4440 #if !defined (FEAT_GUI_GTK)
4441 else if (input_style
& XIMPreeditPosition
)
4442 /* There is a problem in switching XIM off when preediting is used,
4443 * and it is not clear how this can be solved. For now, keep XIM on
4444 * all the time, like it was done in Vim 5.8. */
4448 /* Remember the active state, it is needed when Vim gets keyboard focus. */
4449 xim_is_active
= active
;
4452 /* When 'imactivatekey' has valid key-string, try to control XIM preedit
4453 * state. When 'imactivatekey' has no or invalid string, try old XIM
4459 * Destroy old Input Context (XIC), and create new one. New XIC
4460 * would have a state of preedit that is off. When argument:active
4461 * is false, that's all. Else argument:active is true, send a key
4462 * event specified by 'imactivatekey' to activate XIM preedit state.
4465 xim_is_active
= TRUE
; /* Disable old XIM focus control */
4466 /* If we can monitor preedit state with preedit callback functions,
4467 * try least creation of new XIC.
4469 if (xim_input_style
& (int)GDK_IM_PREEDIT_CALLBACKS
)
4471 if (xim_can_preediting
&& !active
)
4473 /* Force turn off preedit state. With some IM
4474 * implementations, we cannot turn off preedit state by
4475 * symulate keypress event. It is why using such a method
4476 * that destroy old IC (input context), and create new one.
4477 * When create new IC, its preedit state is usually off.
4480 xim_set_focus(FALSE
);
4481 gdk_ic_destroy(xic
);
4483 xim_can_preediting
= FALSE
;
4485 else if (!xim_can_preediting
&& active
)
4486 im_xim_send_event_imactivate();
4490 /* First, force destroy old IC, and create new one. It
4491 * symulates "turning off preedit state".
4493 xim_set_focus(FALSE
);
4494 gdk_ic_destroy(xic
);
4496 xim_can_preediting
= FALSE
;
4498 /* 2nd, when requested to activate IM, symulate this by sending
4503 im_xim_send_event_imactivate();
4504 xim_can_preediting
= TRUE
;
4510 # ifndef XIMPreeditUnKnown
4511 /* X11R5 doesn't have these, it looks safe enough to define here. */
4512 typedef unsigned long XIMPreeditState
;
4513 # define XIMPreeditUnKnown 0L
4514 # define XIMPreeditEnable 1L
4515 # define XIMPreeditDisable (1L<<1)
4516 # define XNPreeditState "preeditState"
4518 XIMPreeditState preedit_state
= XIMPreeditUnKnown
;
4519 XVaNestedList preedit_attr
;
4522 preedit_attr
= XVaCreateNestedList(0,
4523 XNPreeditState
, &preedit_state
,
4525 pxic
= ((GdkICPrivate
*)xic
)->xic
;
4527 if (!XGetICValues(pxic
, XNPreeditAttributes
, preedit_attr
, NULL
))
4529 XFree(preedit_attr
);
4530 preedit_attr
= XVaCreateNestedList(0,
4532 active
? XIMPreeditEnable
: XIMPreeditDisable
,
4534 XSetICValues(pxic
, XNPreeditAttributes
, preedit_attr
, NULL
);
4535 xim_can_preediting
= active
;
4536 xim_is_active
= active
;
4538 XFree(preedit_attr
);
4540 if (xim_input_style
& XIMPreeditCallbacks
)
4542 preedit_buf_len
= 0;
4543 init_preedit_start_col();
4547 /* When had tested kinput2 + canna + Athena GUI version with
4548 * 'imactivatekey' is "s-space", im_xim_send_event_imactivate() did not
4549 * work correctly. It just inserted one space. I don't know why we
4550 * couldn't switch state of XIM preediting. This is reason why these
4551 * codes are commented out.
4553 /* First, force destroy old IC, and create new one. It symulates
4554 * "turning off preedit state".
4556 xim_set_focus(FALSE
);
4561 /* 2nd, when requested to activate IM, symulate this by sending the
4565 im_xim_send_event_imactivate();
4572 * Adjust using XIM for gaining or losing keyboard focus. Also called when
4573 * "xim_is_active" changes.
4576 xim_set_focus(focus
)
4583 * XIM only gets focus when the Vim window has keyboard focus and XIM has
4584 * been set active for the current mode.
4586 if (focus
&& xim_is_active
)
4590 xim_has_focus
= TRUE
;
4592 gdk_im_begin(xic
, gui
.drawarea
->window
);
4602 xim_has_focus
= FALSE
;
4613 im_set_position(row
, col
)
4621 * Set the XIM to the current cursor position.
4629 xim_set_focus(TRUE
);
4643 # ifdef FEAT_XFONTSET
4644 if ((xim_input_style
& (int)GDK_IM_PREEDIT_POSITION
)
4645 && gui
.fontset
!= NOFONTSET
4646 && gui
.fontset
->type
== GDK_FONT_FONTSET
)
4650 if (attr
->spot_location
.y
>= 0)
4652 attr
->spot_location
.x
= 0;
4653 attr
->spot_location
.y
= -100;
4654 attrmask
|= (int)GDK_IC_SPOT_LOCATION
;
4661 if (attr
->spot_location
.x
!= TEXT_X(gui
.col
)
4662 || attr
->spot_location
.y
!= TEXT_Y(gui
.row
))
4664 attr
->spot_location
.x
= TEXT_X(gui
.col
);
4665 attr
->spot_location
.y
= TEXT_Y(gui
.row
);
4666 attrmask
|= (int)GDK_IC_SPOT_LOCATION
;
4669 gdk_window_get_size(gui
.drawarea
->window
, &width
, &height
);
4670 width
-= 2 * gui
.border_offset
;
4671 height
-= 2 * gui
.border_offset
;
4672 if (xim_input_style
& (int)GDK_IM_STATUS_AREA
)
4673 height
-= gui
.char_height
;
4674 if (attr
->preedit_area
.width
!= width
4675 || attr
->preedit_area
.height
!= height
)
4677 attr
->preedit_area
.x
= gui
.border_offset
;
4678 attr
->preedit_area
.y
= gui
.border_offset
;
4679 attr
->preedit_area
.width
= width
;
4680 attr
->preedit_area
.height
= height
;
4681 attrmask
|= (int)GDK_IC_PREEDIT_AREA
;
4684 if (attr
->preedit_fontset
!= gui
.current_font
)
4686 attr
->preedit_fontset
= gui
.current_font
;
4687 attrmask
|= (int)GDK_IC_PREEDIT_FONTSET
;
4691 # endif /* FEAT_XFONTSET */
4693 if (xim_fg_color
== INVALCOLOR
)
4695 xim_fg_color
= gui
.def_norm_pixel
;
4696 xim_bg_color
= gui
.def_back_pixel
;
4698 if (attr
->preedit_foreground
.pixel
!= xim_fg_color
)
4700 attr
->preedit_foreground
.pixel
= xim_fg_color
;
4701 attrmask
|= (int)GDK_IC_PREEDIT_FOREGROUND
;
4703 if (attr
->preedit_background
.pixel
!= xim_bg_color
)
4705 attr
->preedit_background
.pixel
= xim_bg_color
;
4706 attrmask
|= (int)GDK_IC_PREEDIT_BACKGROUND
;
4710 gdk_ic_set_attr(xic
, attr
, (GdkICAttributesType
)attrmask
);
4712 #else /* FEAT_GUI_GTK */
4714 XVaNestedList attr_list
;
4715 XRectangle spot_area
;
4721 /* hide XIM cursor */
4723 over_spot
.y
= -100; /* arbitrary invisible position */
4724 attr_list
= (XVaNestedList
) XVaCreateNestedList(0,
4728 XSetICValues(xic
, XNPreeditAttributes
, attr_list
, NULL
);
4733 if (input_style
& XIMPreeditPosition
)
4735 if (xim_fg_color
== INVALCOLOR
)
4737 xim_fg_color
= gui
.def_norm_pixel
;
4738 xim_bg_color
= gui
.def_back_pixel
;
4740 over_spot
.x
= TEXT_X(gui
.col
);
4741 over_spot
.y
= TEXT_Y(gui
.row
);
4744 spot_area
.height
= gui
.char_height
* Rows
;
4745 spot_area
.width
= gui
.char_width
* Columns
;
4746 line_space
= gui
.char_height
;
4747 attr_list
= (XVaNestedList
) XVaCreateNestedList(0,
4748 XNSpotLocation
, &over_spot
,
4749 XNForeground
, (Pixel
) xim_fg_color
,
4750 XNBackground
, (Pixel
) xim_bg_color
,
4752 XNLineSpace
, line_space
,
4754 if (XSetICValues(xic
, XNPreeditAttributes
, attr_list
, NULL
))
4755 EMSG(_("E284: Cannot set IC values"));
4759 #endif /* FEAT_GUI_GTK */
4763 * Set up the status area.
4765 * This should use a separate Widget, but that seems not possible, because
4766 * preedit_area and status_area should be set to the same window as for the
4767 * text input. Unfortunately this means the status area pollutes the text
4771 xim_set_status_area()
4777 # if defined(FEAT_XFONTSET)
4778 if (use_status_area
)
4791 style
= (int)gdk_ic_get_style(xic
);
4792 if ((style
& (int)GDK_IM_STATUS_MASK
) == (int)GDK_IM_STATUS_AREA
)
4794 if (gui
.fontset
!= NOFONTSET
4795 && gui
.fontset
->type
== GDK_FONT_FONTSET
)
4797 widget
= gui
.mainwin
;
4798 gdk_window_get_size(widget
->window
, &width
, &height
);
4800 attrmask
|= (int)GDK_IC_STATUS_AREA
;
4801 attr
->status_area
.x
= 0;
4802 attr
->status_area
.y
= height
- gui
.char_height
- 1;
4803 attr
->status_area
.width
= width
;
4804 attr
->status_area
.height
= gui
.char_height
;
4808 gdk_ic_set_attr(xic
, attr
, (GdkICAttributesType
)attrmask
);
4813 XVaNestedList preedit_list
= 0, status_list
= 0, list
= 0;
4814 XRectangle pre_area
, status_area
;
4816 if (input_style
& XIMStatusArea
)
4818 if (input_style
& XIMPreeditArea
)
4820 XRectangle
*needed_rect
;
4822 /* to get status_area width */
4823 status_list
= XVaCreateNestedList(0, XNAreaNeeded
,
4824 &needed_rect
, NULL
);
4825 XGetICValues(xic
, XNStatusAttributes
, status_list
, NULL
);
4828 status_area
.width
= needed_rect
->width
;
4831 status_area
.width
= gui
.char_width
* Columns
;
4834 status_area
.y
= gui
.char_height
* Rows
+ gui
.border_offset
;
4835 if (gui
.which_scrollbars
[SBAR_BOTTOM
])
4836 status_area
.y
+= gui
.scrollbar_height
;
4838 if (gui
.menu_is_active
)
4839 status_area
.y
+= gui
.menu_height
;
4841 status_area
.height
= gui
.char_height
;
4842 status_list
= XVaCreateNestedList(0, XNArea
, &status_area
, NULL
);
4847 status_area
.y
= gui
.char_height
* Rows
+ gui
.border_offset
;
4848 if (gui
.which_scrollbars
[SBAR_BOTTOM
])
4849 status_area
.y
+= gui
.scrollbar_height
;
4851 if (gui
.menu_is_active
)
4852 status_area
.y
+= gui
.menu_height
;
4854 status_area
.width
= 0;
4855 status_area
.height
= gui
.char_height
;
4858 if (input_style
& XIMPreeditArea
) /* off-the-spot */
4860 pre_area
.x
= status_area
.x
+ status_area
.width
;
4861 pre_area
.y
= gui
.char_height
* Rows
+ gui
.border_offset
;
4862 pre_area
.width
= gui
.char_width
* Columns
- pre_area
.x
;
4863 if (gui
.which_scrollbars
[SBAR_BOTTOM
])
4864 pre_area
.y
+= gui
.scrollbar_height
;
4866 if (gui
.menu_is_active
)
4867 pre_area
.y
+= gui
.menu_height
;
4869 pre_area
.height
= gui
.char_height
;
4870 preedit_list
= XVaCreateNestedList(0, XNArea
, &pre_area
, NULL
);
4872 else if (input_style
& XIMPreeditPosition
) /* over-the-spot */
4876 pre_area
.height
= gui
.char_height
* Rows
;
4877 pre_area
.width
= gui
.char_width
* Columns
;
4878 preedit_list
= XVaCreateNestedList(0, XNArea
, &pre_area
, NULL
);
4881 if (preedit_list
&& status_list
)
4882 list
= XVaCreateNestedList(0, XNPreeditAttributes
, preedit_list
,
4883 XNStatusAttributes
, status_list
, NULL
);
4884 else if (preedit_list
)
4885 list
= XVaCreateNestedList(0, XNPreeditAttributes
, preedit_list
,
4887 else if (status_list
)
4888 list
= XVaCreateNestedList(0, XNStatusAttributes
, status_list
,
4895 XSetICValues(xic
, XNVaNestedList
, list
, NULL
);
4901 XFree(preedit_list
);
4906 #if defined(FEAT_GUI_X11) || defined(FEAT_GUI_GTK)
4907 static char e_xim
[] = N_("E285: Failed to create input context");
4910 #if defined(FEAT_GUI_X11) || defined(PROTO)
4911 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(sun)
4912 # define USE_X11R6_XIM
4915 static int xim_real_init
__ARGS((Window x11_window
, Display
*x11_display
));
4918 #ifdef USE_X11R6_XIM
4919 static void xim_instantiate_cb
__ARGS((Display
*display
, XPointer client_data
, XPointer call_data
));
4920 static void xim_destroy_cb
__ARGS((XIM im
, XPointer client_data
, XPointer call_data
));
4923 xim_instantiate_cb(display
, client_data
, call_data
)
4925 XPointer client_data UNUSED
;
4926 XPointer call_data UNUSED
;
4929 Display
*x11_display
;
4932 xim_log("xim_instantiate_cb()\n");
4935 gui_get_x11_windis(&x11_window
, &x11_display
);
4936 if (display
!= x11_display
)
4939 xim_real_init(x11_window
, x11_display
);
4940 gui_set_shellsize(FALSE
, FALSE
, RESIZE_BOTH
);
4942 XUnregisterIMInstantiateCallback(x11_display
, NULL
, NULL
, NULL
,
4943 xim_instantiate_cb
, NULL
);
4947 xim_destroy_cb(im
, client_data
, call_data
)
4949 XPointer client_data UNUSED
;
4950 XPointer call_data UNUSED
;
4953 Display
*x11_display
;
4956 xim_log("xim_destroy_cb()\n");
4958 gui_get_x11_windis(&x11_window
, &x11_display
);
4961 status_area_enabled
= FALSE
;
4963 gui_set_shellsize(FALSE
, FALSE
, RESIZE_BOTH
);
4965 XRegisterIMInstantiateCallback(x11_display
, NULL
, NULL
, NULL
,
4966 xim_instantiate_cb
, NULL
);
4974 Display
*x11_display
;
4977 xim_log("xim_init()\n");
4980 gui_get_x11_windis(&x11_window
, &x11_display
);
4984 if (xim_real_init(x11_window
, x11_display
))
4987 gui_set_shellsize(FALSE
, FALSE
, RESIZE_BOTH
);
4989 #ifdef USE_X11R6_XIM
4990 XRegisterIMInstantiateCallback(x11_display
, NULL
, NULL
, NULL
,
4991 xim_instantiate_cb
, NULL
);
4996 xim_real_init(x11_window
, x11_display
)
4998 Display
*x11_display
;
5006 #define IMLEN_MAX 40
5007 char buf
[IMLEN_MAX
+ 7];
5009 XIMStyles
*xim_styles
;
5010 XIMStyle this_input_style
= 0;
5013 XVaNestedList preedit_list
, status_list
;
5016 status_area_enabled
= FALSE
;
5021 if (gui
.rsrc_input_method
!= NULL
&& *gui
.rsrc_input_method
!= NUL
)
5023 strcpy(tmp
, gui
.rsrc_input_method
);
5024 for (ns
= s
= tmp
; ns
!= NULL
&& *s
!= NUL
;)
5026 s
= (char *)skipwhite((char_u
*)s
);
5029 if ((ns
= end
= strchr(s
, ',')) == NULL
)
5030 end
= s
+ strlen(s
);
5031 while (isspace(((char_u
*)end
)[-1]))
5035 if (strlen(s
) <= IMLEN_MAX
)
5037 strcpy(buf
, "@im=");
5039 if ((p
= XSetLocaleModifiers(buf
)) != NULL
&& *p
!= NUL
5040 && (xim
= XOpenIM(x11_display
, NULL
, NULL
, NULL
))
5049 if (xim
== NULL
&& (p
= XSetLocaleModifiers("")) != NULL
&& *p
!= NUL
)
5050 xim
= XOpenIM(x11_display
, NULL
, NULL
, NULL
);
5052 /* This is supposed to be useful to obtain characters through
5053 * XmbLookupString() without really using a XIM. */
5054 if (xim
== NULL
&& (p
= XSetLocaleModifiers("@im=none")) != NULL
5056 xim
= XOpenIM(x11_display
, NULL
, NULL
, NULL
);
5060 /* Only give this message when verbose is set, because too many people
5061 * got this message when they didn't want to use a XIM. */
5065 EMSG(_("E286: Failed to open input method"));
5071 #ifdef USE_X11R6_XIM
5073 XIMCallback destroy_cb
;
5075 destroy_cb
.callback
= xim_destroy_cb
;
5076 destroy_cb
.client_data
= NULL
;
5077 if (XSetIMValues(xim
, XNDestroyCallback
, &destroy_cb
, NULL
))
5078 EMSG(_("E287: Warning: Could not set destroy callback to IM"));
5082 if (XGetIMValues(xim
, XNQueryInputStyle
, &xim_styles
, NULL
) || !xim_styles
)
5084 EMSG(_("E288: input method doesn't support any style"));
5090 strcpy(tmp
, gui
.rsrc_preedit_type_name
);
5091 for (s
= tmp
; s
&& !found
; )
5093 while (*s
&& isspace((unsigned char)*s
))
5097 if ((ns
= end
= strchr(s
, ',')) != 0)
5100 end
= s
+ strlen(s
);
5101 while (isspace((unsigned char)*end
))
5105 if (!strcmp(s
, "OverTheSpot"))
5106 this_input_style
= (XIMPreeditPosition
| XIMStatusArea
);
5107 else if (!strcmp(s
, "OffTheSpot"))
5108 this_input_style
= (XIMPreeditArea
| XIMStatusArea
);
5109 else if (!strcmp(s
, "Root"))
5110 this_input_style
= (XIMPreeditNothing
| XIMStatusNothing
);
5112 for (i
= 0; (unsigned short)i
< xim_styles
->count_styles
; i
++)
5114 if (this_input_style
== xim_styles
->supported_styles
[i
])
5121 for (i
= 0; (unsigned short)i
< xim_styles
->count_styles
; i
++)
5123 if ((xim_styles
->supported_styles
[i
] & this_input_style
)
5124 == (this_input_style
& ~XIMStatusArea
))
5126 this_input_style
&= ~XIMStatusArea
;
5138 /* Only give this message when verbose is set, because too many people
5139 * got this message when they didn't want to use a XIM. */
5143 EMSG(_("E289: input method doesn't support my preedit type"));
5150 over_spot
.x
= TEXT_X(gui
.col
);
5151 over_spot
.y
= TEXT_Y(gui
.row
);
5152 input_style
= this_input_style
;
5154 /* A crash was reported when trying to pass gui.norm_font as XNFontSet,
5155 * thus that has been removed. Hopefully the default works... */
5156 #ifdef FEAT_XFONTSET
5157 if (gui
.fontset
!= NOFONTSET
)
5159 preedit_list
= XVaCreateNestedList(0,
5160 XNSpotLocation
, &over_spot
,
5161 XNForeground
, (Pixel
)gui
.def_norm_pixel
,
5162 XNBackground
, (Pixel
)gui
.def_back_pixel
,
5163 XNFontSet
, (XFontSet
)gui
.fontset
,
5165 status_list
= XVaCreateNestedList(0,
5166 XNForeground
, (Pixel
)gui
.def_norm_pixel
,
5167 XNBackground
, (Pixel
)gui
.def_back_pixel
,
5168 XNFontSet
, (XFontSet
)gui
.fontset
,
5174 preedit_list
= XVaCreateNestedList(0,
5175 XNSpotLocation
, &over_spot
,
5176 XNForeground
, (Pixel
)gui
.def_norm_pixel
,
5177 XNBackground
, (Pixel
)gui
.def_back_pixel
,
5179 status_list
= XVaCreateNestedList(0,
5180 XNForeground
, (Pixel
)gui
.def_norm_pixel
,
5181 XNBackground
, (Pixel
)gui
.def_back_pixel
,
5185 xic
= XCreateIC(xim
,
5186 XNInputStyle
, input_style
,
5187 XNClientWindow
, x11_window
,
5188 XNFocusWindow
, gui
.wid
,
5189 XNPreeditAttributes
, preedit_list
,
5190 XNStatusAttributes
, status_list
,
5193 XFree(preedit_list
);
5196 if (input_style
& XIMStatusArea
)
5198 xim_set_status_area();
5199 status_area_enabled
= TRUE
;
5202 gui_set_shellsize(FALSE
, FALSE
, RESIZE_BOTH
);
5214 #endif /* FEAT_GUI_X11 */
5216 #if defined(FEAT_GUI_GTK) || defined(PROTO)
5218 # ifdef FEAT_XFONTSET
5219 static char e_overthespot
[] = N_("E290: over-the-spot style requires fontset");
5227 xim_decide_input_style()
5229 /* GDK_IM_STATUS_CALLBACKS was disabled, enabled it to allow Japanese
5231 int supported_style
= (int)GDK_IM_PREEDIT_NONE
|
5232 (int)GDK_IM_PREEDIT_NOTHING
|
5233 (int)GDK_IM_PREEDIT_POSITION
|
5234 (int)GDK_IM_PREEDIT_CALLBACKS
|
5235 (int)GDK_IM_STATUS_CALLBACKS
|
5236 (int)GDK_IM_STATUS_AREA
|
5237 (int)GDK_IM_STATUS_NONE
|
5238 (int)GDK_IM_STATUS_NOTHING
;
5241 xim_log("xim_decide_input_style()\n");
5244 if (!gdk_im_ready())
5245 xim_input_style
= 0;
5248 if (gtk_major_version
> 1
5249 || (gtk_major_version
== 1
5250 && (gtk_minor_version
> 2
5251 || (gtk_minor_version
== 2 && gtk_micro_version
>= 3))))
5252 use_status_area
= TRUE
;
5255 EMSG(_("E291: Your GTK+ is older than 1.2.3. Status area disabled"));
5256 use_status_area
= FALSE
;
5258 #ifdef FEAT_XFONTSET
5259 if (gui
.fontset
== NOFONTSET
|| gui
.fontset
->type
!= GDK_FONT_FONTSET
)
5261 supported_style
&= ~((int)GDK_IM_PREEDIT_POSITION
5262 | (int)GDK_IM_STATUS_AREA
);
5263 if (!use_status_area
)
5264 supported_style
&= ~(int)GDK_IM_STATUS_AREA
;
5265 xim_input_style
= (int)gdk_im_decide_style((GdkIMStyle
)supported_style
);
5270 preedit_start_cbproc(XIC thexic UNUSED
,
5271 XPointer client_data UNUSED
,
5272 XPointer call_data UNUSED
)
5275 xim_log("xim_decide_input_style()\n");
5278 draw_feedback
= NULL
;
5279 xim_can_preediting
= TRUE
;
5280 xim_has_preediting
= TRUE
;
5281 gui_update_cursor(TRUE
, FALSE
);
5290 xim_back_delete(int n
)
5298 add_to_input_buf(str
, 3);
5301 static GSList
*key_press_event_queue
= NULL
;
5302 static gboolean processing_queued_event
= FALSE
;
5305 preedit_draw_cbproc(XIC thexic UNUSED
,
5306 XPointer client_data UNUSED
,
5309 XIMPreeditDrawCallbackStruct
*draw_data
;
5312 GSList
*event_queue
;
5315 xim_log("preedit_draw_cbproc()\n");
5318 draw_data
= (XIMPreeditDrawCallbackStruct
*) call_data
;
5319 text
= (XIMText
*) draw_data
->text
;
5321 if ((text
== NULL
&& draw_data
->chg_length
== preedit_buf_len
)
5322 || preedit_buf_len
== 0)
5324 init_preedit_start_col();
5325 vim_free(draw_feedback
);
5326 draw_feedback
= NULL
;
5328 if (draw_data
->chg_length
> 0)
5332 if (draw_data
->chg_length
> preedit_buf_len
)
5333 bs_cnt
= preedit_buf_len
;
5335 bs_cnt
= draw_data
->chg_length
;
5336 xim_back_delete(bs_cnt
);
5337 preedit_buf_len
-= bs_cnt
;
5344 unsigned int nfeedback
= 0;
5348 src
= text
->string
.multi_byte
;
5349 if (src
!= NULL
&& !text
->encoding_is_wchar
)
5352 ptr
= (char_u
*)src
;
5353 /* Avoid the enter for decision */
5358 if (input_conv
.vc_type
!= CONV_NONE
5359 && (buf
= string_convert(&input_conv
,
5360 (char_u
*)src
, &len
)) != NULL
)
5362 /* Converted from 'termencoding' to 'encoding'. */
5363 add_to_input_buf_csi(buf
, len
);
5368 add_to_input_buf_csi((char_u
*)src
, len
);
5369 /* Add count of character to preedit_buf_len */
5373 if (draw_data
->text
->feedback
!= NULL
)
5375 if (draw_feedback
== NULL
)
5376 draw_feedback
= (char *)alloc(draw_data
->chg_first
5379 draw_feedback
= vim_realloc(draw_feedback
,
5380 draw_data
->chg_first
+ text
->length
);
5381 if (draw_feedback
!= NULL
)
5383 draw_feedback
[nfeedback
+ draw_data
->chg_first
]
5384 = draw_data
->text
->feedback
[nfeedback
];
5389 ptr
+= (*mb_ptr2len
)(ptr
);
5398 preedit_end_col
= MAXCOL
;
5401 if (text
!= NULL
|| draw_data
->chg_length
> 0)
5403 event_queue
= key_press_event_queue
;
5404 processing_queued_event
= TRUE
;
5405 while (event_queue
!= NULL
&& processing_queued_event
)
5407 GdkEvent
*ev
= event_queue
->data
;
5410 gtk_signal_emit_by_name((GtkObject
*)gui
.mainwin
, "key_press_event",
5413 event_queue
= event_queue
->next
;
5415 processing_queued_event
= FALSE
;
5416 if (key_press_event_queue
)
5418 g_slist_free(key_press_event_queue
);
5419 key_press_event_queue
= NULL
;
5422 if (gtk_main_level() > 0)
5427 * Retrieve the highlighting attributes at column col in the preedit string.
5428 * Return -1 if not in preediting mode or if col is out of range.
5431 im_get_feedback_attr(int col
)
5433 if (draw_feedback
!= NULL
&& col
< preedit_buf_len
)
5435 if (draw_feedback
[col
] & XIMReverse
)
5437 else if (draw_feedback
[col
] & XIMUnderline
)
5438 return HL_UNDERLINE
;
5440 return hl_attr(HLF_V
);
5447 preedit_caret_cbproc(XIC thexic UNUSED
,
5448 XPointer client_data UNUSED
,
5449 XPointer call_data UNUSED
)
5452 xim_log("preedit_caret_cbproc()\n");
5457 preedit_done_cbproc(XIC thexic UNUSED
,
5458 XPointer client_data UNUSED
,
5459 XPointer call_data UNUSED
)
5462 xim_log("preedit_done_cbproc()\n");
5465 vim_free(draw_feedback
);
5466 draw_feedback
= NULL
;
5467 xim_can_preediting
= FALSE
;
5468 xim_has_preediting
= FALSE
;
5469 gui_update_cursor(TRUE
, FALSE
);
5483 xim_log("xim_reset()\n");
5488 text
= XmbResetIC(((GdkICPrivate
*)xic
)->xic
);
5489 if (text
!= NULL
&& !(xim_input_style
& (int)GDK_IM_PREEDIT_CALLBACKS
))
5490 add_to_input_buf_csi((char_u
*)text
, strlen(text
));
5492 preedit_buf_len
= 0;
5499 xim_queue_key_press_event(GdkEventKey
*event
, int down UNUSED
)
5502 xim_log("xim_queue_key_press_event()\n");
5505 if (preedit_buf_len
<= 0)
5507 if (processing_queued_event
)
5508 processing_queued_event
= FALSE
;
5510 key_press_event_queue
= g_slist_append(key_press_event_queue
,
5511 gdk_event_copy((GdkEvent
*)event
));
5516 preedit_callback_setup(GdkIC
*ic UNUSED
)
5519 XVaNestedList preedit_attr
;
5520 XIMCallback preedit_start_cb
;
5521 XIMCallback preedit_draw_cb
;
5522 XIMCallback preedit_caret_cb
;
5523 XIMCallback preedit_done_cb
;
5525 xxic
= ((GdkICPrivate
*)xic
)->xic
;
5526 preedit_start_cb
.callback
= (XIMProc
)preedit_start_cbproc
;
5527 preedit_draw_cb
.callback
= (XIMProc
)preedit_draw_cbproc
;
5528 preedit_caret_cb
.callback
= (XIMProc
)preedit_caret_cbproc
;
5529 preedit_done_cb
.callback
= (XIMProc
)preedit_done_cbproc
;
5531 = XVaCreateNestedList(0,
5532 XNPreeditStartCallback
, &preedit_start_cb
,
5533 XNPreeditDrawCallback
, &preedit_draw_cb
,
5534 XNPreeditCaretCallback
, &preedit_caret_cb
,
5535 XNPreeditDoneCallback
, &preedit_done_cb
,
5537 XSetICValues(xxic
, XNPreeditAttributes
, preedit_attr
, NULL
);
5538 XFree(preedit_attr
);
5542 reset_state_setup(GdkIC
*ic UNUSED
)
5544 #ifdef USE_X11R6_XIM
5545 /* don't change the input context when we call reset */
5546 XSetICValues(((GdkICPrivate
*)ic
)->xic
, XNResetState
, XIMPreserveState
,
5555 xim_log("xim_init()\n");
5561 if (!gdk_im_ready())
5566 EMSG(_("E292: Input Method Server is not running"));
5571 if ((xic_attr
= gdk_ic_attr_new()) != NULL
)
5573 #ifdef FEAT_XFONTSET
5577 GdkColormap
*colormap
;
5578 GdkICAttr
*attr
= xic_attr
;
5579 int attrmask
= (int)GDK_IC_ALL_REQ
;
5580 GtkWidget
*widget
= gui
.drawarea
;
5582 attr
->style
= (GdkIMStyle
)xim_input_style
;
5583 attr
->client_window
= gui
.mainwin
->window
;
5585 if ((colormap
= gtk_widget_get_colormap(widget
)) !=
5586 gtk_widget_get_default_colormap())
5588 attrmask
|= (int)GDK_IC_PREEDIT_COLORMAP
;
5589 attr
->preedit_colormap
= colormap
;
5591 attrmask
|= (int)GDK_IC_PREEDIT_FOREGROUND
;
5592 attrmask
|= (int)GDK_IC_PREEDIT_BACKGROUND
;
5593 attr
->preedit_foreground
= widget
->style
->fg
[GTK_STATE_NORMAL
];
5594 attr
->preedit_background
= widget
->style
->base
[GTK_STATE_NORMAL
];
5596 #ifdef FEAT_XFONTSET
5597 if ((xim_input_style
& (int)GDK_IM_PREEDIT_MASK
)
5598 == (int)GDK_IM_PREEDIT_POSITION
)
5600 if (gui
.fontset
== NOFONTSET
5601 || gui
.fontset
->type
!= GDK_FONT_FONTSET
)
5603 EMSG(_(e_overthespot
));
5607 gdk_window_get_size(widget
->window
, &width
, &height
);
5609 attrmask
|= (int)GDK_IC_PREEDIT_POSITION_REQ
;
5610 attr
->spot_location
.x
= TEXT_X(0);
5611 attr
->spot_location
.y
= TEXT_Y(0);
5612 attr
->preedit_area
.x
= gui
.border_offset
;
5613 attr
->preedit_area
.y
= gui
.border_offset
;
5614 attr
->preedit_area
.width
= width
- 2*gui
.border_offset
;
5615 attr
->preedit_area
.height
= height
- 2*gui
.border_offset
;
5616 attr
->preedit_fontset
= gui
.fontset
;
5620 if ((xim_input_style
& (int)GDK_IM_STATUS_MASK
)
5621 == (int)GDK_IM_STATUS_AREA
)
5623 if (gui
.fontset
== NOFONTSET
5624 || gui
.fontset
->type
!= GDK_FONT_FONTSET
)
5626 EMSG(_(e_overthespot
));
5630 gdk_window_get_size(gui
.mainwin
->window
, &width
, &height
);
5631 attrmask
|= (int)GDK_IC_STATUS_AREA_REQ
;
5632 attr
->status_area
.x
= 0;
5633 attr
->status_area
.y
= height
- gui
.char_height
- 1;
5634 attr
->status_area
.width
= width
;
5635 attr
->status_area
.height
= gui
.char_height
;
5636 attr
->status_fontset
= gui
.fontset
;
5639 else if ((xim_input_style
& (int)GDK_IM_STATUS_MASK
)
5640 == (int)GDK_IM_STATUS_CALLBACKS
)
5646 xic
= gdk_ic_new(attr
, (GdkICAttributesType
)attrmask
);
5652 mask
= (int)gdk_window_get_events(widget
->window
);
5653 mask
|= (int)gdk_ic_get_events(xic
);
5654 gdk_window_set_events(widget
->window
, (GdkEventMask
)mask
);
5655 if (xim_input_style
& (int)GDK_IM_PREEDIT_CALLBACKS
)
5656 preedit_callback_setup(xic
);
5657 reset_state_setup(xic
);
5666 xim_log("im_shutdown()\n");
5672 gdk_ic_destroy(xic
);
5675 xim_is_active
= FALSE
;
5676 xim_can_preediting
= FALSE
;
5677 preedit_start_col
= MAXCOL
;
5678 xim_has_preediting
= FALSE
;
5681 #endif /* FEAT_GUI_GTK */
5684 xim_get_status_area_height()
5687 if (xim_input_style
& (int)GDK_IM_STATUS_AREA
)
5688 return gui
.char_height
;
5690 if (status_area_enabled
)
5691 return gui
.char_height
;
5697 * Get IM status. When IM is on, return TRUE. Else return FALSE.
5698 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is
5699 * active, when not having focus XIM may still be active (e.g., when using a
5700 * tear-off menu item).
5705 # ifdef FEAT_GUI_GTK
5706 if (xim_input_style
& (int)GDK_IM_PREEDIT_CALLBACKS
)
5707 return xim_can_preediting
;
5709 return xim_has_focus
;
5712 # endif /* !HAVE_GTK2 */
5714 # if defined(HAVE_GTK2) || defined(PROTO)
5716 preedit_get_status(void)
5718 return preedit_is_active
;
5722 # if defined(FEAT_GUI_GTK) || defined(PROTO)
5726 return xim_has_preediting
;
5729 #endif /* FEAT_XIM */
5731 #if defined(FEAT_MBYTE) || defined(PROTO)
5734 * Setup "vcp" for conversion from "from" to "to".
5735 * The names must have been made canonical with enc_canonize().
5736 * vcp->vc_type must have been initialized to CONV_NONE.
5737 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8
5739 * Afterwards invoke with "from" and "to" equal to NULL to cleanup.
5740 * Return FAIL when conversion is not supported, OK otherwise.
5743 convert_setup(vcp
, from
, to
)
5751 /* Reset to no conversion. */
5753 if (vcp
->vc_type
== CONV_ICONV
&& vcp
->vc_fd
!= (iconv_t
)-1)
5754 iconv_close(vcp
->vc_fd
);
5756 vcp
->vc_type
= CONV_NONE
;
5758 vcp
->vc_fail
= FALSE
;
5760 /* No conversion when one of the names is empty or they are equal. */
5761 if (from
== NULL
|| *from
== NUL
|| to
== NULL
|| *to
== NUL
5762 || STRCMP(from
, to
) == 0)
5765 from_prop
= enc_canon_props(from
);
5766 to_prop
= enc_canon_props(to
);
5767 if ((from_prop
& ENC_LATIN1
) && (to_prop
& ENC_UNICODE
))
5769 /* Internal latin1 -> utf-8 conversion. */
5770 vcp
->vc_type
= CONV_TO_UTF8
;
5771 vcp
->vc_factor
= 2; /* up to twice as long */
5773 else if ((from_prop
& ENC_LATIN9
) && (to_prop
& ENC_UNICODE
))
5775 /* Internal latin9 -> utf-8 conversion. */
5776 vcp
->vc_type
= CONV_9_TO_UTF8
;
5777 vcp
->vc_factor
= 3; /* up to three as long (euro sign) */
5779 else if ((from_prop
& ENC_UNICODE
) && (to_prop
& ENC_LATIN1
))
5781 /* Internal utf-8 -> latin1 conversion. */
5782 vcp
->vc_type
= CONV_TO_LATIN1
;
5784 else if ((from_prop
& ENC_UNICODE
) && (to_prop
& ENC_LATIN9
))
5786 /* Internal utf-8 -> latin9 conversion. */
5787 vcp
->vc_type
= CONV_TO_LATIN9
;
5790 /* Win32-specific codepage <-> codepage conversion without iconv. */
5791 else if (((from_prop
& ENC_UNICODE
) || encname2codepage(from
) > 0)
5792 && ((to_prop
& ENC_UNICODE
) || encname2codepage(to
) > 0))
5794 vcp
->vc_type
= CONV_CODEPAGE
;
5795 vcp
->vc_factor
= 2; /* up to twice as long */
5796 vcp
->vc_cpfrom
= (from_prop
& ENC_UNICODE
) ? 0 : encname2codepage(from
);
5797 vcp
->vc_cpto
= (to_prop
& ENC_UNICODE
) ? 0 : encname2codepage(to
);
5801 else if ((from_prop
& ENC_MACROMAN
) && (to_prop
& ENC_LATIN1
))
5803 vcp
->vc_type
= CONV_MAC_LATIN1
;
5805 else if ((from_prop
& ENC_MACROMAN
) && (to_prop
& ENC_UNICODE
))
5807 vcp
->vc_type
= CONV_MAC_UTF8
;
5808 vcp
->vc_factor
= 2; /* up to twice as long */
5810 else if ((from_prop
& ENC_LATIN1
) && (to_prop
& ENC_MACROMAN
))
5812 vcp
->vc_type
= CONV_LATIN1_MAC
;
5814 else if ((from_prop
& ENC_UNICODE
) && (to_prop
& ENC_MACROMAN
))
5816 vcp
->vc_type
= CONV_UTF8_MAC
;
5822 /* Use iconv() for conversion. */
5823 vcp
->vc_fd
= (iconv_t
)my_iconv_open(
5824 (to_prop
& ENC_UNICODE
) ? (char_u
*)"utf-8" : to
,
5825 (from_prop
& ENC_UNICODE
) ? (char_u
*)"utf-8" : from
);
5826 if (vcp
->vc_fd
!= (iconv_t
)-1)
5828 vcp
->vc_type
= CONV_ICONV
;
5829 vcp
->vc_factor
= 4; /* could be longer too... */
5833 if (vcp
->vc_type
== CONV_NONE
)
5839 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \
5840 || defined(MSDOS) || defined(PROTO)
5842 * Do conversion on typed input characters in-place.
5843 * The input and output are not NUL terminated!
5844 * Returns the length after conversion.
5847 convert_input(ptr
, len
, maxlen
)
5852 return convert_input_safe(ptr
, len
, maxlen
, NULL
, NULL
);
5857 * Like convert_input(), but when there is an incomplete byte sequence at the
5858 * end return that as an allocated string in "restp" and set "*restlenp" to
5859 * the length. If "restp" is NULL it is not used.
5862 convert_input_safe(ptr
, len
, maxlen
, restp
, restlenp
)
5871 int unconvertlen
= 0;
5873 d
= string_convert_ext(&input_conv
, ptr
, &dlen
,
5874 restp
== NULL
? NULL
: &unconvertlen
);
5879 if (unconvertlen
> 0)
5881 /* Move the unconverted characters to allocated memory. */
5882 *restp
= alloc(unconvertlen
);
5884 mch_memmove(*restp
, ptr
+ len
- unconvertlen
, unconvertlen
);
5885 *restlenp
= unconvertlen
;
5887 mch_memmove(ptr
, d
, dlen
);
5890 /* result is too long, keep the unconverted text (the caller must
5891 * have done something wrong!) */
5899 * Convert text "ptr[*lenp]" according to "vcp".
5900 * Returns the result in allocated memory and sets "*lenp".
5901 * When "lenp" is NULL, use NUL terminated strings.
5902 * Illegal chars are often changed to "?", unless vcp->vc_fail is set.
5903 * When something goes wrong, NULL is returned and "*lenp" is unchanged.
5906 string_convert(vcp
, ptr
, lenp
)
5911 return string_convert_ext(vcp
, ptr
, lenp
, NULL
);
5915 * Like string_convert(), but when "unconvlenp" is not NULL and there are is
5916 * an incomplete sequence at the end it is not converted and "*unconvlenp" is
5917 * set to the number of remaining bytes.
5920 string_convert_ext(vcp
, ptr
, lenp
, unconvlenp
)
5926 char_u
*retval
= NULL
;
5934 len
= (int)STRLEN(ptr
);
5938 return vim_strsave((char_u
*)"");
5940 switch (vcp
->vc_type
)
5942 case CONV_TO_UTF8
: /* latin1 to utf-8 conversion */
5943 retval
= alloc(len
* 2 + 1);
5947 for (i
= 0; i
< len
; ++i
)
5954 *d
++ = 0xc0 + ((unsigned)c
>> 6);
5955 *d
++ = 0x80 + (c
& 0x3f);
5960 *lenp
= (int)(d
- retval
);
5963 case CONV_9_TO_UTF8
: /* latin9 to utf-8 conversion */
5964 retval
= alloc(len
* 3 + 1);
5968 for (i
= 0; i
< len
; ++i
)
5973 case 0xa4: c
= 0x20ac; break; /* euro */
5974 case 0xa6: c
= 0x0160; break; /* S hat */
5975 case 0xa8: c
= 0x0161; break; /* S -hat */
5976 case 0xb4: c
= 0x017d; break; /* Z hat */
5977 case 0xb8: c
= 0x017e; break; /* Z -hat */
5978 case 0xbc: c
= 0x0152; break; /* OE */
5979 case 0xbd: c
= 0x0153; break; /* oe */
5980 case 0xbe: c
= 0x0178; break; /* Y */
5982 d
+= utf_char2bytes(c
, d
);
5986 *lenp
= (int)(d
- retval
);
5989 case CONV_TO_LATIN1
: /* utf-8 to latin1 conversion */
5990 case CONV_TO_LATIN9
: /* utf-8 to latin9 conversion */
5991 retval
= alloc(len
+ 1);
5995 for (i
= 0; i
< len
; ++i
)
5997 l
= utf_ptr2len(ptr
+ i
);
6002 if (unconvlenp
!= NULL
&& utf8len_tab
[ptr
[i
]] > len
- i
)
6004 /* Incomplete sequence at the end. */
6005 *unconvlenp
= len
- i
;
6012 c
= utf_ptr2char(ptr
+ i
);
6013 if (vcp
->vc_type
== CONV_TO_LATIN9
)
6016 case 0x20ac: c
= 0xa4; break; /* euro */
6017 case 0x0160: c
= 0xa6; break; /* S hat */
6018 case 0x0161: c
= 0xa8; break; /* S -hat */
6019 case 0x017d: c
= 0xb4; break; /* Z hat */
6020 case 0x017e: c
= 0xb8; break; /* Z -hat */
6021 case 0x0152: c
= 0xbc; break; /* OE */
6022 case 0x0153: c
= 0xbd; break; /* oe */
6023 case 0x0178: c
= 0xbe; break; /* Y */
6031 case 0xbe: c
= 0x100; break; /* not in latin9 */
6033 if (!utf_iscomposing(c
)) /* skip composing chars */
6037 else if (vcp
->vc_fail
)
6045 if (utf_char2cells(c
) > 1)
6054 *lenp
= (int)(d
- retval
);
6057 # ifdef MACOS_CONVERT
6058 case CONV_MAC_LATIN1
:
6059 retval
= mac_string_convert(ptr
, len
, lenp
, vcp
->vc_fail
,
6060 'm', 'l', unconvlenp
);
6063 case CONV_LATIN1_MAC
:
6064 retval
= mac_string_convert(ptr
, len
, lenp
, vcp
->vc_fail
,
6065 'l', 'm', unconvlenp
);
6069 retval
= mac_string_convert(ptr
, len
, lenp
, vcp
->vc_fail
,
6070 'm', 'u', unconvlenp
);
6074 retval
= mac_string_convert(ptr
, len
, lenp
, vcp
->vc_fail
,
6075 'u', 'm', unconvlenp
);
6080 case CONV_ICONV
: /* conversion with output_conv.vc_fd */
6081 retval
= iconv_string(vcp
, ptr
, len
, unconvlenp
);
6082 if (retval
!= NULL
&& lenp
!= NULL
)
6083 *lenp
= (int)STRLEN(retval
);
6087 case CONV_CODEPAGE
: /* codepage -> codepage */
6093 /* 1. codepage/UTF-8 -> ucs-2. */
6094 if (vcp
->vc_cpfrom
== 0)
6095 tmp_len
= utf8_to_utf16(ptr
, len
, NULL
, NULL
);
6097 tmp_len
= MultiByteToWideChar(vcp
->vc_cpfrom
, 0,
6099 tmp
= (short_u
*)alloc(sizeof(short_u
) * tmp_len
);
6102 if (vcp
->vc_cpfrom
== 0)
6103 utf8_to_utf16(ptr
, len
, tmp
, unconvlenp
);
6105 MultiByteToWideChar(vcp
->vc_cpfrom
, 0, ptr
, len
, tmp
, tmp_len
);
6107 /* 2. ucs-2 -> codepage/UTF-8. */
6108 if (vcp
->vc_cpto
== 0)
6109 retlen
= utf16_to_utf8(tmp
, tmp_len
, NULL
);
6111 retlen
= WideCharToMultiByte(vcp
->vc_cpto
, 0,
6112 tmp
, tmp_len
, 0, 0, 0, 0);
6113 retval
= alloc(retlen
+ 1);
6116 if (vcp
->vc_cpto
== 0)
6117 utf16_to_utf8(tmp
, tmp_len
, retval
);
6119 WideCharToMultiByte(vcp
->vc_cpto
, 0,
6120 tmp
, tmp_len
, retval
, retlen
, 0, 0);
6121 retval
[retlen
] = NUL
;