1 /* vi:set ts=8 sts=4 sw=4 noet:
2 *
3 * VIM - Vi IMproved by Bram Moolenaar
4 * Multibyte extensions partly by Sung-Hoon Baek
5 *
6 * Do ":help uganda" in Vim to read copying and usage conditions.
7 * Do ":help credits" in Vim to see a list of people who contributed.
8 * See README.txt for an overview of the Vim source code.
9 */
10 /*
11 * mbyte.c: Code specifically for handling multi-byte characters.
12 *
13 * The encoding used in the core is set with 'encoding'. When 'encoding' is
14 * changed, the following four variables are set (for speed).
15 * Currently these types of character encodings are supported:
16 *
17 * "enc_dbcs" When non-zero it tells the type of double byte character
18 * encoding (Chinese, Korean, Japanese, etc.).
19 * The cell width on the display is equal to the number of
20 * bytes. (exception: DBCS_JPNU with first byte 0x8e)
21 * Recognizing the first or second byte is difficult, it
22 * requires checking a byte sequence from the start.
23 * "enc_utf8" When TRUE use Unicode characters in UTF-8 encoding.
24 * The cell width on the display needs to be determined from
25 * the character value.
26 * Recognizing bytes is easy: 0xxx.xxxx is a single-byte
27 * char, 10xx.xxxx is a trailing byte, 11xx.xxxx is a leading
28 * byte of a multi-byte character.
29 * To make things complicated, up to six composing characters
30 * are allowed. These are drawn on top of the first char.
31 * For most editing the sequence of bytes with composing
32 * characters included is considered to be one character.
33 * "enc_unicode" When 2 use 16-bit Unicode characters (or UTF-16).
34 * When 4 use 32-but Unicode characters.
35 * Internally characters are stored in UTF-8 encoding to
36 * avoid NUL bytes. Conversion happens when doing I/O.
37 * "enc_utf8" will also be TRUE.
38 *
39 * "has_mbyte" is set when "enc_dbcs" or "enc_utf8" is non-zero.
40 *
41 * If none of these is TRUE, 8-bit bytes are used for a character. The
42 * encoding isn't currently specified (TODO).
43 *
44 * 'encoding' specifies the encoding used in the core. This is in registers,
45 * text manipulation, buffers, etc. Conversion has to be done when characters
46 * in another encoding are received or send:
47 *
48 * clipboard
49 * ^
50 * | (2)
51 * V
52 * +---------------+
53 * (1) | | (3)
54 * keyboard ----->| core |-----> display
55 * | |
56 * +---------------+
57 * ^
58 * | (4)
59 * V
60 * file
61 *
62 * (1) Typed characters arrive in the current locale. Conversion is to be
63 * done when 'encoding' is different from 'termencoding'.
64 * (2) Text will be made available with the encoding specified with
65 * 'encoding'. If this is not sufficient, system-specific conversion
66 * might be required.
67 * (3) For the GUI the correct font must be selected, no conversion done.
68 * Otherwise, conversion is to be done when 'encoding' differs from
69 * 'termencoding'. (Different in the GTK+ 2 port -- 'termencoding'
70 * is always used for both input and output and must always be set to
71 * "utf-8". gui_mch_init() does this automatically.)
72 * (4) The encoding of the file is specified with 'fileencoding'. Conversion
73 * is to be done when it's different from 'encoding'.
74 *
75 * The viminfo file is a special case: Only text is converted, not file names.
76 * Vim scripts may contain an ":encoding" command. This has an effect for
77 * some commands, like ":menutrans"
78 */
79
80 #include "vim.h"
81
82 #ifdef WIN32UNIX
83 # ifndef WIN32_LEAN_AND_MEAN
84 # define WIN32_LEAN_AND_MEAN
85 # endif
86 # if defined(FEAT_GUI) || defined(FEAT_XCLIPBOARD)
87 # include <X11/Xwindows.h>
88 # define WINBYTE wBYTE
89 # else
90 # include <windows.h>
91 # define WINBYTE BYTE
92 # endif
93 # ifdef WIN32
94 # undef WIN32 // Some windows.h define WIN32, we don't want that here.
95 # endif
96 #else
97 # define WINBYTE BYTE
98 #endif
99
100 #if (defined(MSWIN) || defined(WIN32UNIX)) && !defined(__MINGW32__)
101 # include <winnls.h>
102 #endif
103
104 #ifdef FEAT_GUI_X11
105 # include <X11/Intrinsic.h>
106 #endif
107 #ifdef X_LOCALE
108 # include <X11/Xlocale.h>
109 # if !defined(HAVE_MBLEN) && !defined(mblen)
110 # define mblen _Xmblen
111 # endif
112 #endif
113
114 #ifdef HAVE_WCHAR_H
115 # include <wchar.h>
116 #endif
117
118 #if 0
119 // This has been disabled, because several people reported problems with the
120 // wcwidth() and iswprint() library functions, esp. for Hebrew.
121 # ifdef __STDC_ISO_10646__
122 # define USE_WCHAR_FUNCTIONS
123 # endif
124 #endif
125
126 static int dbcs_char2len(int c);
127 static int dbcs_char2bytes(int c, char_u *buf);
128 static int dbcs_ptr2len(char_u *p);
129 static int dbcs_ptr2len_len(char_u *p, int size);
130 static int utf_ptr2cells_len(char_u *p, int size);
131 static int dbcs_char2cells(int c);
132 static int dbcs_ptr2cells_len(char_u *p, int size);
133 static int dbcs_ptr2char(char_u *p);
134 static int dbcs_head_off(char_u *base, char_u *p);
135 #ifdef FEAT_EVAL
136 static int cw_value(int c);
137 #endif
138
139 /*
140 * Lookup table to quickly get the length in bytes of a UTF-8 character from
141 * the first byte of a UTF-8 string.
142 * Bytes which are illegal when used as the first byte have a 1.
143 * The NUL byte has length 1.
144 */
145 static char utf8len_tab[256] =
146 {
147 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
148 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
149 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
150 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
151 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
152 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
153 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,
154 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6,1,1,
155 };
156
157 /*
158 * Like utf8len_tab above, but using a zero for illegal lead bytes.
159 */
160 static char utf8len_tab_zero[256] =
161 {
162 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
163 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
164 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
165 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
166 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
167 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
168 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,
169 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6,0,0,
170 };
171
172
173 /*
174 * Canonical encoding names and their properties.
175 * "iso-8859-n" is handled by enc_canonize() directly.
176 */
177 static struct
178 { char *name; int prop; int codepage;}
179 enc_canon_table[] =
180 {
181 #define IDX_LATIN_1 0
182 {"latin1", ENC_8BIT + ENC_LATIN1, 1252},
183 #define IDX_ISO_2 1
184 {"iso-8859-2", ENC_8BIT, 0},
185 #define IDX_ISO_3 2
186 {"iso-8859-3", ENC_8BIT, 0},
187 #define IDX_ISO_4 3
188 {"iso-8859-4", ENC_8BIT, 0},
189 #define IDX_ISO_5 4
190 {"iso-8859-5", ENC_8BIT, 0},
191 #define IDX_ISO_6 5
192 {"iso-8859-6", ENC_8BIT, 0},
193 #define IDX_ISO_7 6
194 {"iso-8859-7", ENC_8BIT, 0},
195 #define IDX_ISO_8 7
196 {"iso-8859-8", ENC_8BIT, 0},
197 #define IDX_ISO_9 8
198 {"iso-8859-9", ENC_8BIT, 0},
199 #define IDX_ISO_10 9
200 {"iso-8859-10", ENC_8BIT, 0},
201 #define IDX_ISO_11 10
202 {"iso-8859-11", ENC_8BIT, 0},
203 #define IDX_ISO_13 11
204 {"iso-8859-13", ENC_8BIT, 0},
205 #define IDX_ISO_14 12
206 {"iso-8859-14", ENC_8BIT, 0},
207 #define IDX_ISO_15 13
208 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0},
209 #define IDX_KOI8_R 14
210 {"koi8-r", ENC_8BIT, 0},
211 #define IDX_KOI8_U 15
212 {"koi8-u", ENC_8BIT, 0},
213 #define IDX_UTF8 16
214 {"utf-8", ENC_UNICODE, 0},
215 #define IDX_UCS2 17
216 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0},
217 #define IDX_UCS2LE 18
218 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0},
219 #define IDX_UTF16 19
220 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0},
221 #define IDX_UTF16LE 20
222 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0},
223 #define IDX_UCS4 21
224 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0},
225 #define IDX_UCS4LE 22
226 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0},
227
228 // For debugging DBCS encoding on Unix.
229 #define IDX_DEBUG 23
230 {"debug", ENC_DBCS, DBCS_DEBUG},
231 #define IDX_EUC_JP 24
232 {"euc-jp", ENC_DBCS, DBCS_JPNU},
233 #define IDX_SJIS 25
234 {"sjis", ENC_DBCS, DBCS_JPN},
235 #define IDX_EUC_KR 26
236 {"euc-kr", ENC_DBCS, DBCS_KORU},
237 #define IDX_EUC_CN 27
238 {"euc-cn", ENC_DBCS, DBCS_CHSU},
239 #define IDX_EUC_TW 28
240 {"euc-tw", ENC_DBCS, DBCS_CHTU},
241 #define IDX_BIG5 29
242 {"big5", ENC_DBCS, DBCS_CHT},
243
244 // MS-DOS and MS-Windows codepages are included here, so that they can be
245 // used on Unix too. Most of them are similar to ISO-8859 encodings, but
246 // not exactly the same.
247 #define IDX_CP437 30
248 {"cp437", ENC_8BIT, 437}, // like iso-8859-1
249 #define IDX_CP737 31
250 {"cp737", ENC_8BIT, 737}, // like iso-8859-7
251 #define IDX_CP775 32
252 {"cp775", ENC_8BIT, 775}, // Baltic
253 #define IDX_CP850 33
254 {"cp850", ENC_8BIT, 850}, // like iso-8859-4
255 #define IDX_CP852 34
256 {"cp852", ENC_8BIT, 852}, // like iso-8859-1
257 #define IDX_CP855 35
258 {"cp855", ENC_8BIT, 855}, // like iso-8859-2
259 #define IDX_CP857 36
260 {"cp857", ENC_8BIT, 857}, // like iso-8859-5
261 #define IDX_CP860 37
262 {"cp860", ENC_8BIT, 860}, // like iso-8859-9
263 #define IDX_CP861 38
264 {"cp861", ENC_8BIT, 861}, // like iso-8859-1
265 #define IDX_CP862 39
266 {"cp862", ENC_8BIT, 862}, // like iso-8859-1
267 #define IDX_CP863 40
268 {"cp863", ENC_8BIT, 863}, // like iso-8859-8
269 #define IDX_CP865 41
270 {"cp865", ENC_8BIT, 865}, // like iso-8859-1
271 #define IDX_CP866 42
272 {"cp866", ENC_8BIT, 866}, // like iso-8859-5
273 #define IDX_CP869 43
274 {"cp869", ENC_8BIT, 869}, // like iso-8859-7
275 #define IDX_CP874 44
276 {"cp874", ENC_8BIT, 874}, // Thai
277 #define IDX_CP932 45
278 {"cp932", ENC_DBCS, DBCS_JPN},
279 #define IDX_CP936 46
280 {"cp936", ENC_DBCS, DBCS_CHS},
281 #define IDX_CP949 47
282 {"cp949", ENC_DBCS, DBCS_KOR},
283 #define IDX_CP950 48
284 {"cp950", ENC_DBCS, DBCS_CHT},
285 #define IDX_CP1250 49
286 {"cp1250", ENC_8BIT, 1250}, // Czech, Polish, etc.
287 #define IDX_CP1251 50
288 {"cp1251", ENC_8BIT, 1251}, // Cyrillic
289 // cp1252 is considered to be equal to latin1
290 #define IDX_CP1253 51
291 {"cp1253", ENC_8BIT, 1253}, // Greek
292 #define IDX_CP1254 52
293 {"cp1254", ENC_8BIT, 1254}, // Turkish
294 #define IDX_CP1255 53
295 {"cp1255", ENC_8BIT, 1255}, // Hebrew
296 #define IDX_CP1256 54
297 {"cp1256", ENC_8BIT, 1256}, // Arabic
298 #define IDX_CP1257 55
299 {"cp1257", ENC_8BIT, 1257}, // Baltic
300 #define IDX_CP1258 56
301 {"cp1258", ENC_8BIT, 1258}, // Vietnamese
302
303 #define IDX_MACROMAN 57
304 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, // Mac OS
305 #define IDX_DECMCS 58
306 {"dec-mcs", ENC_8BIT, 0}, // DEC MCS
307 #define IDX_HPROMAN8 59
308 {"hp-roman8", ENC_8BIT, 0}, // HP Roman8
309 #define IDX_COUNT 60
310 };
311
312 /*
313 * Aliases for encoding names.
314 */
315 static struct
316 { char *name; int canon;}
317 enc_alias_table[] =
318 {
319 {"ansi", IDX_LATIN_1},
320 {"iso-8859-1", IDX_LATIN_1},
321 {"latin2", IDX_ISO_2},
322 {"latin3", IDX_ISO_3},
323 {"latin4", IDX_ISO_4},
324 {"cyrillic", IDX_ISO_5},
325 {"arabic", IDX_ISO_6},
326 {"greek", IDX_ISO_7},
327 #ifdef MSWIN
328 {"hebrew", IDX_CP1255},
329 #else
330 {"hebrew", IDX_ISO_8},
331 #endif
332 {"latin5", IDX_ISO_9},
333 {"turkish", IDX_ISO_9}, // ?
334 {"latin6", IDX_ISO_10},
335 {"nordic", IDX_ISO_10}, // ?
336 {"thai", IDX_ISO_11}, // ?
337 {"latin7", IDX_ISO_13},
338 {"latin8", IDX_ISO_14},
339 {"latin9", IDX_ISO_15},
340 {"utf8", IDX_UTF8},
341 {"unicode", IDX_UCS2},
342 {"ucs2", IDX_UCS2},
343 {"ucs2be", IDX_UCS2},
344 {"ucs-2be", IDX_UCS2},
345 {"ucs2le", IDX_UCS2LE},
346 {"utf16", IDX_UTF16},
347 {"utf16be", IDX_UTF16},
348 {"utf-16be", IDX_UTF16},
349 {"utf16le", IDX_UTF16LE},
350 {"ucs4", IDX_UCS4},
351 {"ucs4be", IDX_UCS4},
352 {"ucs-4be", IDX_UCS4},
353 {"ucs4le", IDX_UCS4LE},
354 {"utf32", IDX_UCS4},
355 {"utf-32", IDX_UCS4},
356 {"utf32be", IDX_UCS4},
357 {"utf-32be", IDX_UCS4},
358 {"utf32le", IDX_UCS4LE},
359 {"utf-32le", IDX_UCS4LE},
360 {"932", IDX_CP932},
361 {"949", IDX_CP949},
362 {"936", IDX_CP936},
363 {"gbk", IDX_CP936},
364 {"950", IDX_CP950},
365 {"eucjp", IDX_EUC_JP},
366 {"unix-jis", IDX_EUC_JP},
367 {"ujis", IDX_EUC_JP},
368 {"shift-jis", IDX_SJIS},
369 {"pck", IDX_SJIS}, // Sun: PCK
370 {"euckr", IDX_EUC_KR},
371 {"5601", IDX_EUC_KR}, // Sun: KS C 5601
372 {"euccn", IDX_EUC_CN},
373 {"gb2312", IDX_EUC_CN},
374 {"euctw", IDX_EUC_TW},
375 #if defined(MSWIN) || defined(WIN32UNIX) || defined(MACOS_X)
376 {"japan", IDX_CP932},
377 {"korea", IDX_CP949},
378 {"prc", IDX_CP936},
379 {"chinese", IDX_CP936},
380 {"taiwan", IDX_CP950},
381 {"big5", IDX_CP950},
382 #else
383 {"japan", IDX_EUC_JP},
384 {"korea", IDX_EUC_KR},
385 {"prc", IDX_EUC_CN},
386 {"chinese", IDX_EUC_CN},
387 {"taiwan", IDX_EUC_TW},
388 {"cp950", IDX_BIG5},
389 {"950", IDX_BIG5},
390 #endif
391 {"mac", IDX_MACROMAN},
392 {"mac-roman", IDX_MACROMAN},
393 {NULL, 0}
394 };
395
396 #ifndef CP_UTF8
397 # define CP_UTF8 65001 // magic number from winnls.h
398 #endif
399
400 /*
401 * Find encoding "name" in the list of canonical encoding names.
402 * Returns -1 if not found.
403 */
404 static int
enc_canon_search(char_u * name)405 enc_canon_search(char_u *name)
406 {
407 int i;
408
409 for (i = 0; i < IDX_COUNT; ++i)
410 if (STRCMP(name, enc_canon_table[i].name) == 0)
411 return i;
412 return -1;
413 }
414
415
416 /*
417 * Find canonical encoding "name" in the list and return its properties.
418 * Returns 0 if not found.
419 */
420 int
enc_canon_props(char_u * name)421 enc_canon_props(char_u *name)
422 {
423 int i;
424
425 i = enc_canon_search(name);
426 if (i >= 0)
427 return enc_canon_table[i].prop;
428 #ifdef MSWIN
429 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2]))
430 {
431 CPINFO cpinfo;
432
433 // Get info on this codepage to find out what it is.
434 if (GetCPInfo(atoi((char *)name + 2), &cpinfo) != 0)
435 {
436 if (cpinfo.MaxCharSize == 1) // some single-byte encoding
437 return ENC_8BIT;
438 if (cpinfo.MaxCharSize == 2
439 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0))
440 // must be a DBCS encoding
441 return ENC_DBCS;
442 }
443 return 0;
444 }
445 #endif
446 if (STRNCMP(name, "2byte-", 6) == 0)
447 return ENC_DBCS;
448 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0)
449 return ENC_8BIT;
450 return 0;
451 }
452
453 /*
454 * Set up for using multi-byte characters.
455 * Called in three cases:
456 * - by main() to initialize (p_enc == NULL)
457 * - by set_init_1() after 'encoding' was set to its default.
458 * - by do_set() when 'encoding' has been set.
459 * p_enc must have been passed through enc_canonize() already.
460 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags.
461 * Fills mb_bytelen_tab[] and returns NULL when there are no problems.
462 * When there is something wrong: Returns an error message and doesn't change
463 * anything.
464 */
465 char *
mb_init(void)466 mb_init(void)
467 {
468 int i;
469 int idx;
470 int n;
471 int enc_dbcs_new = 0;
472 #if defined(USE_ICONV) && !defined(MSWIN) && !defined(WIN32UNIX) \
473 && !defined(MACOS_CONVERT)
474 # define LEN_FROM_CONV
475 vimconv_T vimconv;
476 char_u *p;
477 #endif
478
479 if (p_enc == NULL)
480 {
481 // Just starting up: set the whole table to one's.
482 for (i = 0; i < 256; ++i)
483 mb_bytelen_tab[i] = 1;
484 input_conv.vc_type = CONV_NONE;
485 input_conv.vc_factor = 1;
486 output_conv.vc_type = CONV_NONE;
487 return NULL;
488 }
489
490 #ifdef MSWIN
491 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2]))
492 {
493 CPINFO cpinfo;
494
495 // Get info on this codepage to find out what it is.
496 if (GetCPInfo(atoi((char *)p_enc + 2), &cpinfo) != 0)
497 {
498 if (cpinfo.MaxCharSize == 1)
499 {
500 // some single-byte encoding
501 enc_unicode = 0;
502 enc_utf8 = FALSE;
503 }
504 else if (cpinfo.MaxCharSize == 2
505 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0))
506 {
507 // must be a DBCS encoding, check below
508 enc_dbcs_new = atoi((char *)p_enc + 2);
509 }
510 else
511 goto codepage_invalid;
512 }
513 else if (GetLastError() == ERROR_INVALID_PARAMETER)
514 {
515 codepage_invalid:
516 return N_("E543: Not a valid codepage");
517 }
518 }
519 #endif
520 else if (STRNCMP(p_enc, "8bit-", 5) == 0
521 || STRNCMP(p_enc, "iso-8859-", 9) == 0)
522 {
523 // Accept any "8bit-" or "iso-8859-" name.
524 enc_unicode = 0;
525 enc_utf8 = FALSE;
526 }
527 else if (STRNCMP(p_enc, "2byte-", 6) == 0)
528 {
529 #ifdef MSWIN
530 // Windows: accept only valid codepage numbers, check below.
531 if (p_enc[6] != 'c' || p_enc[7] != 'p'
532 || (enc_dbcs_new = atoi((char *)p_enc + 8)) == 0)
533 return e_invarg;
534 #else
535 // Unix: accept any "2byte-" name, assume current locale.
536 enc_dbcs_new = DBCS_2BYTE;
537 #endif
538 }
539 else if ((idx = enc_canon_search(p_enc)) >= 0)
540 {
541 i = enc_canon_table[idx].prop;
542 if (i & ENC_UNICODE)
543 {
544 // Unicode
545 enc_utf8 = TRUE;
546 if (i & (ENC_2BYTE | ENC_2WORD))
547 enc_unicode = 2;
548 else if (i & ENC_4BYTE)
549 enc_unicode = 4;
550 else
551 enc_unicode = 0;
552 }
553 else if (i & ENC_DBCS)
554 {
555 // 2byte, handle below
556 enc_dbcs_new = enc_canon_table[idx].codepage;
557 }
558 else
559 {
560 // Must be 8-bit.
561 enc_unicode = 0;
562 enc_utf8 = FALSE;
563 }
564 }
565 else // Don't know what encoding this is, reject it.
566 return e_invarg;
567
568 if (enc_dbcs_new != 0)
569 {
570 #ifdef MSWIN
571 // Check if the DBCS code page is OK.
572 if (!IsValidCodePage(enc_dbcs_new))
573 goto codepage_invalid;
574 #endif
575 enc_unicode = 0;
576 enc_utf8 = FALSE;
577 }
578 enc_dbcs = enc_dbcs_new;
579 has_mbyte = (enc_dbcs != 0 || enc_utf8);
580
581 #if defined(MSWIN) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD)
582 enc_codepage = encname2codepage(p_enc);
583 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0);
584 #endif
585
586 // Detect an encoding that uses latin1 characters.
587 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0
588 || STRCMP(p_enc, "iso-8859-15") == 0);
589
590 /*
591 * Set the function pointers.
592 */
593 if (enc_utf8)
594 {
595 mb_ptr2len = utfc_ptr2len;
596 mb_ptr2len_len = utfc_ptr2len_len;
597 mb_char2len = utf_char2len;
598 mb_char2bytes = utf_char2bytes;
599 mb_ptr2cells = utf_ptr2cells;
600 mb_ptr2cells_len = utf_ptr2cells_len;
601 mb_char2cells = utf_char2cells;
602 mb_off2cells = utf_off2cells;
603 mb_ptr2char = utf_ptr2char;
604 mb_head_off = utf_head_off;
605 }
606 else if (enc_dbcs != 0)
607 {
608 mb_ptr2len = dbcs_ptr2len;
609 mb_ptr2len_len = dbcs_ptr2len_len;
610 mb_char2len = dbcs_char2len;
611 mb_char2bytes = dbcs_char2bytes;
612 mb_ptr2cells = dbcs_ptr2cells;
613 mb_ptr2cells_len = dbcs_ptr2cells_len;
614 mb_char2cells = dbcs_char2cells;
615 mb_off2cells = dbcs_off2cells;
616 mb_ptr2char = dbcs_ptr2char;
617 mb_head_off = dbcs_head_off;
618 }
619 else
620 {
621 mb_ptr2len = latin_ptr2len;
622 mb_ptr2len_len = latin_ptr2len_len;
623 mb_char2len = latin_char2len;
624 mb_char2bytes = latin_char2bytes;
625 mb_ptr2cells = latin_ptr2cells;
626 mb_ptr2cells_len = latin_ptr2cells_len;
627 mb_char2cells = latin_char2cells;
628 mb_off2cells = latin_off2cells;
629 mb_ptr2char = latin_ptr2char;
630 mb_head_off = latin_head_off;
631 }
632
633 /*
634 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN().
635 */
636 #ifdef LEN_FROM_CONV
637 // When 'encoding' is different from the current locale mblen() won't
638 // work. Use conversion to "utf-8" instead.
639 vimconv.vc_type = CONV_NONE;
640 if (enc_dbcs)
641 {
642 p = enc_locale();
643 if (p == NULL || STRCMP(p, p_enc) != 0)
644 {
645 convert_setup(&vimconv, p_enc, (char_u *)"utf-8");
646 vimconv.vc_fail = TRUE;
647 }
648 vim_free(p);
649 }
650 #endif
651
652 for (i = 0; i < 256; ++i)
653 {
654 // Our own function to reliably check the length of UTF-8 characters,
655 // independent of mblen().
656 if (enc_utf8)
657 n = utf8len_tab[i];
658 else if (enc_dbcs == 0)
659 n = 1;
660 else
661 {
662 #if defined(MSWIN) || defined(WIN32UNIX)
663 // enc_dbcs is set by setting 'fileencoding'. It becomes a Windows
664 // CodePage identifier, which we can pass directly in to Windows
665 // API
666 n = IsDBCSLeadByteEx(enc_dbcs, (WINBYTE)i) ? 2 : 1;
667 #else
668 # if defined(__amigaos4__) || defined(__ANDROID__) || \
669 !(defined(HAVE_MBLEN) || defined(X_LOCALE))
670 /*
671 * if mblen() is not available, character which MSB is turned on
672 * are treated as leading byte character. (note : This assumption
673 * is not always true.)
674 */
675 n = (i & 0x80) ? 2 : 1;
676 # else
677 char buf[MB_MAXBYTES + 1];
678
679 if (i == NUL) // just in case mblen() can't handle ""
680 n = 1;
681 else
682 {
683 buf[0] = i;
684 buf[1] = 0;
685 # ifdef LEN_FROM_CONV
686 if (vimconv.vc_type != CONV_NONE)
687 {
688 /*
689 * string_convert() should fail when converting the first
690 * byte of a double-byte character.
691 */
692 p = string_convert(&vimconv, (char_u *)buf, NULL);
693 if (p != NULL)
694 {
695 vim_free(p);
696 n = 1;
697 }
698 else
699 n = 2;
700 }
701 else
702 # endif
703 {
704 /*
705 * mblen() should return -1 for invalid (means the leading
706 * multibyte) character. However there are some platforms
707 * where mblen() returns 0 for invalid character.
708 * Therefore, following condition includes 0.
709 */
710 vim_ignored = mblen(NULL, 0); // First reset the state.
711 if (mblen(buf, (size_t)1) <= 0)
712 n = 2;
713 else
714 n = 1;
715 }
716 }
717 # endif
718 #endif
719 }
720
721 mb_bytelen_tab[i] = n;
722 }
723
724 #ifdef LEN_FROM_CONV
725 convert_setup(&vimconv, NULL, NULL);
726 #endif
727
728 // The cell width depends on the type of multi-byte characters.
729 (void)init_chartab();
730
731 // When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[]
732 screenalloc(FALSE);
733
734 // When using Unicode, set default for 'fileencodings'.
735 if (enc_utf8 && !option_was_set((char_u *)"fencs"))
736 set_fencs_unicode();
737
738 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT)
739 // GNU gettext 0.10.37 supports this feature: set the codeset used for
740 // translated messages independently from the current locale.
741 (void)bind_textdomain_codeset(VIMPACKAGE,
742 enc_utf8 ? "utf-8" : (char *)p_enc);
743 #endif
744
745 #ifdef MSWIN
746 // When changing 'encoding' while starting up, then convert the command
747 // line arguments from the active codepage to 'encoding'.
748 if (starting != 0)
749 fix_arg_enc();
750 #endif
751
752 // Fire an autocommand to let people do custom font setup. This must be
753 // after Vim has been setup for the new encoding.
754 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf);
755
756 #ifdef FEAT_SPELL
757 // Need to reload spell dictionaries
758 spell_reload();
759 #endif
760
761 return NULL;
762 }
763
764 /*
765 * Return the size of the BOM for the current buffer:
766 * 0 - no BOM
767 * 2 - UCS-2 or UTF-16 BOM
768 * 4 - UCS-4 BOM
769 * 3 - UTF-8 BOM
770 */
771 int
bomb_size(void)772 bomb_size(void)
773 {
774 int n = 0;
775
776 if (curbuf->b_p_bomb && !curbuf->b_p_bin)
777 {
778 if (*curbuf->b_p_fenc == NUL)
779 {
780 if (enc_utf8)
781 {
782 if (enc_unicode != 0)
783 n = enc_unicode;
784 else
785 n = 3;
786 }
787 }
788 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0)
789 n = 3;
790 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0
791 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0)
792 n = 2;
793 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0)
794 n = 4;
795 }
796 return n;
797 }
798
799 #if defined(FEAT_QUICKFIX) || defined(PROTO)
800 /*
801 * Remove all BOM from "s" by moving remaining text.
802 */
803 void
remove_bom(char_u * s)804 remove_bom(char_u *s)
805 {
806 if (enc_utf8)
807 {
808 char_u *p = s;
809
810 while ((p = vim_strbyte(p, 0xef)) != NULL)
811 {
812 if (p[1] == 0xbb && p[2] == 0xbf)
813 STRMOVE(p, p + 3);
814 else
815 ++p;
816 }
817 }
818 }
819 #endif
820
821 /*
822 * Get class of pointer:
823 * 0 for blank or NUL
824 * 1 for punctuation
825 * 2 for an (ASCII) word character
826 * >2 for other word characters
827 */
828 int
mb_get_class(char_u * p)829 mb_get_class(char_u *p)
830 {
831 return mb_get_class_buf(p, curbuf);
832 }
833
834 int
mb_get_class_buf(char_u * p,buf_T * buf)835 mb_get_class_buf(char_u *p, buf_T *buf)
836 {
837 if (MB_BYTE2LEN(p[0]) == 1)
838 {
839 if (p[0] == NUL || VIM_ISWHITE(p[0]))
840 return 0;
841 if (vim_iswordc_buf(p[0], buf))
842 return 2;
843 return 1;
844 }
845 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL)
846 return dbcs_class(p[0], p[1]);
847 if (enc_utf8)
848 return utf_class_buf(utf_ptr2char(p), buf);
849 return 0;
850 }
851
852 /*
853 * Get class of a double-byte character. This always returns 3 or bigger.
854 * TODO: Should return 1 for punctuation.
855 */
856 int
dbcs_class(unsigned lead,unsigned trail)857 dbcs_class(unsigned lead, unsigned trail)
858 {
859 switch (enc_dbcs)
860 {
861 // please add classify routine for your language in here
862
863 case DBCS_JPNU: // ?
864 case DBCS_JPN:
865 {
866 // JIS code classification
867 unsigned char lb = lead;
868 unsigned char tb = trail;
869
870 // convert process code to JIS
871 # if defined(MSWIN) || defined(WIN32UNIX) || defined(MACOS_X)
872 // process code is SJIS
873 if (lb <= 0x9f)
874 lb = (lb - 0x81) * 2 + 0x21;
875 else
876 lb = (lb - 0xc1) * 2 + 0x21;
877 if (tb <= 0x7e)
878 tb -= 0x1f;
879 else if (tb <= 0x9e)
880 tb -= 0x20;
881 else
882 {
883 tb -= 0x7e;
884 lb += 1;
885 }
886 # else
887 /*
888 * XXX: Code page identification can not use with all
889 * system! So, some other encoding information
890 * will be needed.
891 * In japanese: SJIS,EUC,UNICODE,(JIS)
892 * Note that JIS-code system don't use as
893 * process code in most system because it uses
894 * escape sequences(JIS is context depend encoding).
895 */
896 // assume process code is JAPANESE-EUC
897 lb &= 0x7f;
898 tb &= 0x7f;
899 # endif
900 // exceptions
901 switch (lb << 8 | tb)
902 {
903 case 0x2121: // ZENKAKU space
904 return 0;
905 case 0x2122: // TOU-TEN (Japanese comma)
906 case 0x2123: // KU-TEN (Japanese period)
907 case 0x2124: // ZENKAKU comma
908 case 0x2125: // ZENKAKU period
909 return 1;
910 case 0x213c: // prolongedsound handled as KATAKANA
911 return 13;
912 }
913 // sieved by KU code
914 switch (lb)
915 {
916 case 0x21:
917 case 0x22:
918 // special symbols
919 return 10;
920 case 0x23:
921 // alphanumeric
922 return 11;
923 case 0x24:
924 // hiragana
925 return 12;
926 case 0x25:
927 // katakana
928 return 13;
929 case 0x26:
930 // greek
931 return 14;
932 case 0x27:
933 // russian
934 return 15;
935 case 0x28:
936 // lines
937 return 16;
938 default:
939 // kanji
940 return 17;
941 }
942 }
943
944 case DBCS_KORU: // ?
945 case DBCS_KOR:
946 {
947 // KS code classification
948 unsigned char c1 = lead;
949 unsigned char c2 = trail;
950
951 /*
952 * 20 : Hangul
953 * 21 : Hanja
954 * 22 : Symbols
955 * 23 : Alphanumeric/Roman Letter (Full width)
956 * 24 : Hangul Letter(Alphabet)
957 * 25 : Roman Numeral/Greek Letter
958 * 26 : Box Drawings
959 * 27 : Unit Symbols
960 * 28 : Circled/Parenthesized Letter
961 * 29 : Hiragana/Katakana
962 * 30 : Cyrillic Letter
963 */
964
965 if (c1 >= 0xB0 && c1 <= 0xC8)
966 // Hangul
967 return 20;
968 #if defined(MSWIN) || defined(WIN32UNIX)
969 else if (c1 <= 0xA0 || c2 <= 0xA0)
970 // Extended Hangul Region : MS UHC(Unified Hangul Code)
971 // c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE
972 // c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0
973 return 20;
974 #endif
975
976 else if (c1 >= 0xCA && c1 <= 0xFD)
977 // Hanja
978 return 21;
979 else switch (c1)
980 {
981 case 0xA1:
982 case 0xA2:
983 // Symbols
984 return 22;
985 case 0xA3:
986 // Alphanumeric
987 return 23;
988 case 0xA4:
989 // Hangul Letter(Alphabet)
990 return 24;
991 case 0xA5:
992 // Roman Numeral/Greek Letter
993 return 25;
994 case 0xA6:
995 // Box Drawings
996 return 26;
997 case 0xA7:
998 // Unit Symbols
999 return 27;
1000 case 0xA8:
1001 case 0xA9:
1002 if (c2 <= 0xAF)
1003 return 25; // Roman Letter
1004 else if (c2 >= 0xF6)
1005 return 22; // Symbols
1006 else
1007 // Circled/Parenthesized Letter
1008 return 28;
1009 case 0xAA:
1010 case 0xAB:
1011 // Hiragana/Katakana
1012 return 29;
1013 case 0xAC:
1014 // Cyrillic Letter
1015 return 30;
1016 }
1017 }
1018 default:
1019 break;
1020 }
1021 return 3;
1022 }
1023
1024 /*
1025 * mb_char2len() function pointer.
1026 * Return length in bytes of character "c".
1027 * Returns 1 for a single-byte character.
1028 */
1029 int
latin_char2len(int c UNUSED)1030 latin_char2len(int c UNUSED)
1031 {
1032 return 1;
1033 }
1034
1035 static int
dbcs_char2len(int c)1036 dbcs_char2len(
1037 int c)
1038 {
1039 if (c >= 0x100)
1040 return 2;
1041 return 1;
1042 }
1043
1044 /*
1045 * mb_char2bytes() function pointer.
1046 * Convert a character to its bytes.
1047 * Returns the length in bytes.
1048 */
1049 int
latin_char2bytes(int c,char_u * buf)1050 latin_char2bytes(int c, char_u *buf)
1051 {
1052 buf[0] = c;
1053 return 1;
1054 }
1055
1056 static int
dbcs_char2bytes(int c,char_u * buf)1057 dbcs_char2bytes(int c, char_u *buf)
1058 {
1059 if (c >= 0x100)
1060 {
1061 buf[0] = (unsigned)c >> 8;
1062 buf[1] = c;
1063 // Never use a NUL byte, it causes lots of trouble. It's an invalid
1064 // character anyway.
1065 if (buf[1] == NUL)
1066 buf[1] = '\n';
1067 return 2;
1068 }
1069 buf[0] = c;
1070 return 1;
1071 }
1072
1073 /*
1074 * mb_ptr2len() function pointer.
1075 * Get byte length of character at "*p" but stop at a NUL.
1076 * For UTF-8 this includes following composing characters.
1077 * Returns 0 when *p is NUL.
1078 */
1079 int
latin_ptr2len(char_u * p)1080 latin_ptr2len(char_u *p)
1081 {
1082 return MB_BYTE2LEN(*p);
1083 }
1084
1085 static int
dbcs_ptr2len(char_u * p)1086 dbcs_ptr2len(
1087 char_u *p)
1088 {
1089 int len;
1090
1091 // Check if second byte is not missing.
1092 len = MB_BYTE2LEN(*p);
1093 if (len == 2 && p[1] == NUL)
1094 len = 1;
1095 return len;
1096 }
1097
1098 /*
1099 * mb_ptr2len_len() function pointer.
1100 * Like mb_ptr2len(), but limit to read "size" bytes.
1101 * Returns 0 for an empty string.
1102 * Returns 1 for an illegal char or an incomplete byte sequence.
1103 */
1104 int
latin_ptr2len_len(char_u * p,int size)1105 latin_ptr2len_len(char_u *p, int size)
1106 {
1107 if (size < 1 || *p == NUL)
1108 return 0;
1109 return 1;
1110 }
1111
1112 static int
dbcs_ptr2len_len(char_u * p,int size)1113 dbcs_ptr2len_len(char_u *p, int size)
1114 {
1115 int len;
1116
1117 if (size < 1 || *p == NUL)
1118 return 0;
1119 if (size == 1)
1120 return 1;
1121 // Check that second byte is not missing.
1122 len = MB_BYTE2LEN(*p);
1123 if (len == 2 && p[1] == NUL)
1124 len = 1;
1125 return len;
1126 }
1127
1128 struct interval
1129 {
1130 long first;
1131 long last;
1132 };
1133
1134 /*
1135 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]".
1136 */
1137 static int
intable(struct interval * table,size_t size,int c)1138 intable(struct interval *table, size_t size, int c)
1139 {
1140 int mid, bot, top;
1141
1142 // first quick check for Latin1 etc. characters
1143 if (c < table[0].first)
1144 return FALSE;
1145
1146 // binary search in table
1147 bot = 0;
1148 top = (int)(size / sizeof(struct interval) - 1);
1149 while (top >= bot)
1150 {
1151 mid = (bot + top) / 2;
1152 if (table[mid].last < c)
1153 bot = mid + 1;
1154 else if (table[mid].first > c)
1155 top = mid - 1;
1156 else
1157 return TRUE;
1158 }
1159 return FALSE;
1160 }
1161
1162 // Sorted list of non-overlapping intervals of East Asian Ambiguous
1163 // characters, generated with ../runtime/tools/unicode.vim.
1164 static struct interval ambiguous[] =
1165 {
1166 {0x00a1, 0x00a1},
1167 {0x00a4, 0x00a4},
1168 {0x00a7, 0x00a8},
1169 {0x00aa, 0x00aa},
1170 {0x00ad, 0x00ae},
1171 {0x00b0, 0x00b4},
1172 {0x00b6, 0x00ba},
1173 {0x00bc, 0x00bf},
1174 {0x00c6, 0x00c6},
1175 {0x00d0, 0x00d0},
1176 {0x00d7, 0x00d8},
1177 {0x00de, 0x00e1},
1178 {0x00e6, 0x00e6},
1179 {0x00e8, 0x00ea},
1180 {0x00ec, 0x00ed},
1181 {0x00f0, 0x00f0},
1182 {0x00f2, 0x00f3},
1183 {0x00f7, 0x00fa},
1184 {0x00fc, 0x00fc},
1185 {0x00fe, 0x00fe},
1186 {0x0101, 0x0101},
1187 {0x0111, 0x0111},
1188 {0x0113, 0x0113},
1189 {0x011b, 0x011b},
1190 {0x0126, 0x0127},
1191 {0x012b, 0x012b},
1192 {0x0131, 0x0133},
1193 {0x0138, 0x0138},
1194 {0x013f, 0x0142},
1195 {0x0144, 0x0144},
1196 {0x0148, 0x014b},
1197 {0x014d, 0x014d},
1198 {0x0152, 0x0153},
1199 {0x0166, 0x0167},
1200 {0x016b, 0x016b},
1201 {0x01ce, 0x01ce},
1202 {0x01d0, 0x01d0},
1203 {0x01d2, 0x01d2},
1204 {0x01d4, 0x01d4},
1205 {0x01d6, 0x01d6},
1206 {0x01d8, 0x01d8},
1207 {0x01da, 0x01da},
1208 {0x01dc, 0x01dc},
1209 {0x0251, 0x0251},
1210 {0x0261, 0x0261},
1211 {0x02c4, 0x02c4},
1212 {0x02c7, 0x02c7},
1213 {0x02c9, 0x02cb},
1214 {0x02cd, 0x02cd},
1215 {0x02d0, 0x02d0},
1216 {0x02d8, 0x02db},
1217 {0x02dd, 0x02dd},
1218 {0x02df, 0x02df},
1219 {0x0300, 0x036f},
1220 {0x0391, 0x03a1},
1221 {0x03a3, 0x03a9},
1222 {0x03b1, 0x03c1},
1223 {0x03c3, 0x03c9},
1224 {0x0401, 0x0401},
1225 {0x0410, 0x044f},
1226 {0x0451, 0x0451},
1227 {0x2010, 0x2010},
1228 {0x2013, 0x2016},
1229 {0x2018, 0x2019},
1230 {0x201c, 0x201d},
1231 {0x2020, 0x2022},
1232 {0x2024, 0x2027},
1233 {0x2030, 0x2030},
1234 {0x2032, 0x2033},
1235 {0x2035, 0x2035},
1236 {0x203b, 0x203b},
1237 {0x203e, 0x203e},
1238 {0x2074, 0x2074},
1239 {0x207f, 0x207f},
1240 {0x2081, 0x2084},
1241 {0x20ac, 0x20ac},
1242 {0x2103, 0x2103},
1243 {0x2105, 0x2105},
1244 {0x2109, 0x2109},
1245 {0x2113, 0x2113},
1246 {0x2116, 0x2116},
1247 {0x2121, 0x2122},
1248 {0x2126, 0x2126},
1249 {0x212b, 0x212b},
1250 {0x2153, 0x2154},
1251 {0x215b, 0x215e},
1252 {0x2160, 0x216b},
1253 {0x2170, 0x2179},
1254 {0x2189, 0x2189},
1255 {0x2190, 0x2199},
1256 {0x21b8, 0x21b9},
1257 {0x21d2, 0x21d2},
1258 {0x21d4, 0x21d4},
1259 {0x21e7, 0x21e7},
1260 {0x2200, 0x2200},
1261 {0x2202, 0x2203},
1262 {0x2207, 0x2208},
1263 {0x220b, 0x220b},
1264 {0x220f, 0x220f},
1265 {0x2211, 0x2211},
1266 {0x2215, 0x2215},
1267 {0x221a, 0x221a},
1268 {0x221d, 0x2220},
1269 {0x2223, 0x2223},
1270 {0x2225, 0x2225},
1271 {0x2227, 0x222c},
1272 {0x222e, 0x222e},
1273 {0x2234, 0x2237},
1274 {0x223c, 0x223d},
1275 {0x2248, 0x2248},
1276 {0x224c, 0x224c},
1277 {0x2252, 0x2252},
1278 {0x2260, 0x2261},
1279 {0x2264, 0x2267},
1280 {0x226a, 0x226b},
1281 {0x226e, 0x226f},
1282 {0x2282, 0x2283},
1283 {0x2286, 0x2287},
1284 {0x2295, 0x2295},
1285 {0x2299, 0x2299},
1286 {0x22a5, 0x22a5},
1287 {0x22bf, 0x22bf},
1288 {0x2312, 0x2312},
1289 {0x2460, 0x24e9},
1290 {0x24eb, 0x254b},
1291 {0x2550, 0x2573},
1292 {0x2580, 0x258f},
1293 {0x2592, 0x2595},
1294 {0x25a0, 0x25a1},
1295 {0x25a3, 0x25a9},
1296 {0x25b2, 0x25b3},
1297 {0x25b6, 0x25b7},
1298 {0x25bc, 0x25bd},
1299 {0x25c0, 0x25c1},
1300 {0x25c6, 0x25c8},
1301 {0x25cb, 0x25cb},
1302 {0x25ce, 0x25d1},
1303 {0x25e2, 0x25e5},
1304 {0x25ef, 0x25ef},
1305 {0x2605, 0x2606},
1306 {0x2609, 0x2609},
1307 {0x260e, 0x260f},
1308 {0x261c, 0x261c},
1309 {0x261e, 0x261e},
1310 {0x2640, 0x2640},
1311 {0x2642, 0x2642},
1312 {0x2660, 0x2661},
1313 {0x2663, 0x2665},
1314 {0x2667, 0x266a},
1315 {0x266c, 0x266d},
1316 {0x266f, 0x266f},
1317 {0x269e, 0x269f},
1318 {0x26bf, 0x26bf},
1319 {0x26c6, 0x26cd},
1320 {0x26cf, 0x26d3},
1321 {0x26d5, 0x26e1},
1322 {0x26e3, 0x26e3},
1323 {0x26e8, 0x26e9},
1324 {0x26eb, 0x26f1},
1325 {0x26f4, 0x26f4},
1326 {0x26f6, 0x26f9},
1327 {0x26fb, 0x26fc},
1328 {0x26fe, 0x26ff},
1329 {0x273d, 0x273d},
1330 {0x2776, 0x277f},
1331 {0x2b56, 0x2b59},
1332 {0x3248, 0x324f},
1333 {0xe000, 0xf8ff},
1334 {0xfe00, 0xfe0f},
1335 {0xfffd, 0xfffd},
1336 {0x1f100, 0x1f10a},
1337 {0x1f110, 0x1f12d},
1338 {0x1f130, 0x1f169},
1339 {0x1f170, 0x1f18d},
1340 {0x1f18f, 0x1f190},
1341 {0x1f19b, 0x1f1ac},
1342 {0xe0100, 0xe01ef},
1343 {0xf0000, 0xffffd},
1344 {0x100000, 0x10fffd}
1345 };
1346
1347 #if defined(FEAT_TERMINAL) || defined(PROTO)
1348 /*
1349 * utf_char2cells() with different argument type for libvterm.
1350 */
1351 int
utf_uint2cells(UINT32_T c)1352 utf_uint2cells(UINT32_T c)
1353 {
1354 if (c >= 0x100 && utf_iscomposing((int)c))
1355 return 0;
1356 return utf_char2cells((int)c);
1357 }
1358 #endif
1359
1360 /*
1361 * For UTF-8 character "c" return 2 for a double-width character, 1 for others.
1362 * Returns 4 or 6 for an unprintable character.
1363 * Is only correct for characters >= 0x80.
1364 * When p_ambw is "double", return 2 for a character with East Asian Width
1365 * class 'A'(mbiguous).
1366 */
1367 int
utf_char2cells(int c)1368 utf_char2cells(int c)
1369 {
1370 // Sorted list of non-overlapping intervals of East Asian double width
1371 // characters, generated with ../runtime/tools/unicode.vim.
1372 static struct interval doublewidth[] =
1373 {
1374 {0x1100, 0x115f},
1375 {0x231a, 0x231b},
1376 {0x2329, 0x232a},
1377 {0x23e9, 0x23ec},
1378 {0x23f0, 0x23f0},
1379 {0x23f3, 0x23f3},
1380 {0x25fd, 0x25fe},
1381 {0x2614, 0x2615},
1382 {0x2648, 0x2653},
1383 {0x267f, 0x267f},
1384 {0x2693, 0x2693},
1385 {0x26a1, 0x26a1},
1386 {0x26aa, 0x26ab},
1387 {0x26bd, 0x26be},
1388 {0x26c4, 0x26c5},
1389 {0x26ce, 0x26ce},
1390 {0x26d4, 0x26d4},
1391 {0x26ea, 0x26ea},
1392 {0x26f2, 0x26f3},
1393 {0x26f5, 0x26f5},
1394 {0x26fa, 0x26fa},
1395 {0x26fd, 0x26fd},
1396 {0x2705, 0x2705},
1397 {0x270a, 0x270b},
1398 {0x2728, 0x2728},
1399 {0x274c, 0x274c},
1400 {0x274e, 0x274e},
1401 {0x2753, 0x2755},
1402 {0x2757, 0x2757},
1403 {0x2795, 0x2797},
1404 {0x27b0, 0x27b0},
1405 {0x27bf, 0x27bf},
1406 {0x2b1b, 0x2b1c},
1407 {0x2b50, 0x2b50},
1408 {0x2b55, 0x2b55},
1409 {0x2e80, 0x2e99},
1410 {0x2e9b, 0x2ef3},
1411 {0x2f00, 0x2fd5},
1412 {0x2ff0, 0x2ffb},
1413 {0x3000, 0x303e},
1414 {0x3041, 0x3096},
1415 {0x3099, 0x30ff},
1416 {0x3105, 0x312f},
1417 {0x3131, 0x318e},
1418 {0x3190, 0x31e3},
1419 {0x31f0, 0x321e},
1420 {0x3220, 0x3247},
1421 {0x3250, 0x4dbf},
1422 {0x4e00, 0xa48c},
1423 {0xa490, 0xa4c6},
1424 {0xa960, 0xa97c},
1425 {0xac00, 0xd7a3},
1426 {0xf900, 0xfaff},
1427 {0xfe10, 0xfe19},
1428 {0xfe30, 0xfe52},
1429 {0xfe54, 0xfe66},
1430 {0xfe68, 0xfe6b},
1431 {0xff01, 0xff60},
1432 {0xffe0, 0xffe6},
1433 {0x16fe0, 0x16fe3},
1434 {0x16ff0, 0x16ff1},
1435 {0x17000, 0x187f7},
1436 {0x18800, 0x18cd5},
1437 {0x18d00, 0x18d08},
1438 {0x1b000, 0x1b11e},
1439 {0x1b150, 0x1b152},
1440 {0x1b164, 0x1b167},
1441 {0x1b170, 0x1b2fb},
1442 {0x1f004, 0x1f004},
1443 {0x1f0cf, 0x1f0cf},
1444 {0x1f18e, 0x1f18e},
1445 {0x1f191, 0x1f19a},
1446 {0x1f200, 0x1f202},
1447 {0x1f210, 0x1f23b},
1448 {0x1f240, 0x1f248},
1449 {0x1f250, 0x1f251},
1450 {0x1f260, 0x1f265},
1451 {0x1f300, 0x1f320},
1452 {0x1f32d, 0x1f335},
1453 {0x1f337, 0x1f37c},
1454 {0x1f37e, 0x1f393},
1455 {0x1f3a0, 0x1f3ca},
1456 {0x1f3cf, 0x1f3d3},
1457 {0x1f3e0, 0x1f3f0},
1458 {0x1f3f4, 0x1f3f4},
1459 {0x1f3f8, 0x1f43e},
1460 {0x1f440, 0x1f440},
1461 {0x1f442, 0x1f4fc},
1462 {0x1f4ff, 0x1f53d},
1463 {0x1f54b, 0x1f54e},
1464 {0x1f550, 0x1f567},
1465 {0x1f57a, 0x1f57a},
1466 {0x1f595, 0x1f596},
1467 {0x1f5a4, 0x1f5a4},
1468 {0x1f5fb, 0x1f64f},
1469 {0x1f680, 0x1f6c5},
1470 {0x1f6cc, 0x1f6cc},
1471 {0x1f6d0, 0x1f6d2},
1472 {0x1f6d5, 0x1f6d7},
1473 {0x1f6eb, 0x1f6ec},
1474 {0x1f6f4, 0x1f6fc},
1475 {0x1f7e0, 0x1f7eb},
1476 {0x1f90c, 0x1f93a},
1477 {0x1f93c, 0x1f945},
1478 {0x1f947, 0x1f978},
1479 {0x1f97a, 0x1f9cb},
1480 {0x1f9cd, 0x1f9ff},
1481 {0x1fa70, 0x1fa74},
1482 {0x1fa78, 0x1fa7a},
1483 {0x1fa80, 0x1fa86},
1484 {0x1fa90, 0x1faa8},
1485 {0x1fab0, 0x1fab6},
1486 {0x1fac0, 0x1fac2},
1487 {0x1fad0, 0x1fad6},
1488 {0x20000, 0x2fffd},
1489 {0x30000, 0x3fffd}
1490 };
1491
1492 // Sorted list of non-overlapping intervals of Emoji characters that don't
1493 // have ambiguous or double width,
1494 // based on http://unicode.org/emoji/charts/emoji-list.html
1495 static struct interval emoji_wide[] =
1496 {
1497 {0x23ed, 0x23ef},
1498 {0x23f1, 0x23f2},
1499 {0x23f8, 0x23fa},
1500 {0x24c2, 0x24c2},
1501 {0x261d, 0x261d},
1502 {0x26c8, 0x26c8},
1503 {0x26cf, 0x26cf},
1504 {0x26d1, 0x26d1},
1505 {0x26d3, 0x26d3},
1506 {0x26e9, 0x26e9},
1507 {0x26f0, 0x26f1},
1508 {0x26f7, 0x26f9},
1509 {0x270c, 0x270d},
1510 {0x2934, 0x2935},
1511 {0x1f170, 0x1f189},
1512 {0x1f1e6, 0x1f1ff},
1513 {0x1f321, 0x1f321},
1514 {0x1f324, 0x1f32c},
1515 {0x1f336, 0x1f336},
1516 {0x1f37d, 0x1f37d},
1517 {0x1f396, 0x1f397},
1518 {0x1f399, 0x1f39b},
1519 {0x1f39e, 0x1f39f},
1520 {0x1f3cb, 0x1f3ce},
1521 {0x1f3d4, 0x1f3df},
1522 {0x1f3f3, 0x1f3f5},
1523 {0x1f3f7, 0x1f3f7},
1524 {0x1f43f, 0x1f43f},
1525 {0x1f441, 0x1f441},
1526 {0x1f4fd, 0x1f4fd},
1527 {0x1f549, 0x1f54a},
1528 {0x1f56f, 0x1f570},
1529 {0x1f573, 0x1f579},
1530 {0x1f587, 0x1f587},
1531 {0x1f58a, 0x1f58d},
1532 {0x1f590, 0x1f590},
1533 {0x1f5a5, 0x1f5a5},
1534 {0x1f5a8, 0x1f5a8},
1535 {0x1f5b1, 0x1f5b2},
1536 {0x1f5bc, 0x1f5bc},
1537 {0x1f5c2, 0x1f5c4},
1538 {0x1f5d1, 0x1f5d3},
1539 {0x1f5dc, 0x1f5de},
1540 {0x1f5e1, 0x1f5e1},
1541 {0x1f5e3, 0x1f5e3},
1542 {0x1f5e8, 0x1f5e8},
1543 {0x1f5ef, 0x1f5ef},
1544 {0x1f5f3, 0x1f5f3},
1545 {0x1f5fa, 0x1f5fa},
1546 {0x1f6cb, 0x1f6cf},
1547 {0x1f6e0, 0x1f6e5},
1548 {0x1f6e9, 0x1f6e9},
1549 {0x1f6f0, 0x1f6f0},
1550 {0x1f6f3, 0x1f6f3}
1551
1552 #ifdef MACOS_X
1553 // Include SF Symbols characters, which should be rendered as
1554 // double-width. All of them are in the Supplementary Private Use
1555 // Area-B range. The exact range was determined by downloading the "SF
1556 // Symbols" app from Apple, and then selecting all symbols, copying
1557 // them out, and inspecting the unicode values of them.
1558 , {0x100000, 0x100d7f}
1559 #endif
1560 };
1561
1562 if (c >= 0x100)
1563 {
1564 #if defined(FEAT_EVAL) || defined(USE_WCHAR_FUNCTIONS)
1565 int n;
1566 #endif
1567
1568 #ifdef FEAT_EVAL
1569 n = cw_value(c);
1570 if (n != 0)
1571 return n;
1572 #endif
1573
1574 #ifdef USE_WCHAR_FUNCTIONS
1575 /*
1576 * Assume the library function wcwidth() works better than our own
1577 * stuff. It should return 1 for ambiguous width chars!
1578 */
1579 n = wcwidth(c);
1580
1581 if (n < 0)
1582 return 6; // unprintable, displays <xxxx>
1583 if (n > 1)
1584 return n;
1585 #else
1586 if (!utf_printable(c))
1587 return 6; // unprintable, displays <xxxx>
1588 if (intable(doublewidth, sizeof(doublewidth), c))
1589 return 2;
1590 #endif
1591 if (p_emoji && intable(emoji_wide, sizeof(emoji_wide), c))
1592 return 2;
1593 }
1594
1595 // Characters below 0x100 are influenced by 'isprint' option
1596 else if (c >= 0x80 && !vim_isprintc(c))
1597 return 4; // unprintable, displays <xx>
1598
1599 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c))
1600 return 2;
1601
1602 return 1;
1603 }
1604
1605 /*
1606 * mb_ptr2cells() function pointer.
1607 * Return the number of display cells character at "*p" occupies.
1608 * This doesn't take care of unprintable characters, use ptr2cells() for that.
1609 */
1610 int
latin_ptr2cells(char_u * p UNUSED)1611 latin_ptr2cells(char_u *p UNUSED)
1612 {
1613 return 1;
1614 }
1615
1616 int
utf_ptr2cells(char_u * p)1617 utf_ptr2cells(
1618 char_u *p)
1619 {
1620 int c;
1621
1622 // Need to convert to a character number.
1623 if (*p >= 0x80)
1624 {
1625 c = utf_ptr2char(p);
1626 // An illegal byte is displayed as <xx>.
1627 if (utf_ptr2len(p) == 1 || c == NUL)
1628 return 4;
1629 // If the char is ASCII it must be an overlong sequence.
1630 if (c < 0x80)
1631 return char2cells(c);
1632 return utf_char2cells(c);
1633 }
1634 return 1;
1635 }
1636
1637 int
dbcs_ptr2cells(char_u * p)1638 dbcs_ptr2cells(char_u *p)
1639 {
1640 // Number of cells is equal to number of bytes, except for euc-jp when
1641 // the first byte is 0x8e.
1642 if (enc_dbcs == DBCS_JPNU && *p == 0x8e)
1643 return 1;
1644 return MB_BYTE2LEN(*p);
1645 }
1646
1647 /*
1648 * mb_ptr2cells_len() function pointer.
1649 * Like mb_ptr2cells(), but limit string length to "size".
1650 * For an empty string or truncated character returns 1.
1651 */
1652 int
latin_ptr2cells_len(char_u * p UNUSED,int size UNUSED)1653 latin_ptr2cells_len(char_u *p UNUSED, int size UNUSED)
1654 {
1655 return 1;
1656 }
1657
1658 static int
utf_ptr2cells_len(char_u * p,int size)1659 utf_ptr2cells_len(char_u *p, int size)
1660 {
1661 int c;
1662
1663 // Need to convert to a wide character.
1664 if (size > 0 && *p >= 0x80)
1665 {
1666 if (utf_ptr2len_len(p, size) < utf8len_tab[*p])
1667 return 1; // truncated
1668 c = utf_ptr2char(p);
1669 // An illegal byte is displayed as <xx>.
1670 if (utf_ptr2len(p) == 1 || c == NUL)
1671 return 4;
1672 // If the char is ASCII it must be an overlong sequence.
1673 if (c < 0x80)
1674 return char2cells(c);
1675 return utf_char2cells(c);
1676 }
1677 return 1;
1678 }
1679
1680 static int
dbcs_ptr2cells_len(char_u * p,int size)1681 dbcs_ptr2cells_len(char_u *p, int size)
1682 {
1683 // Number of cells is equal to number of bytes, except for euc-jp when
1684 // the first byte is 0x8e.
1685 if (size <= 1 || (enc_dbcs == DBCS_JPNU && *p == 0x8e))
1686 return 1;
1687 return MB_BYTE2LEN(*p);
1688 }
1689
1690 /*
1691 * mb_char2cells() function pointer.
1692 * Return the number of display cells character "c" occupies.
1693 * Only takes care of multi-byte chars, not "^C" and such.
1694 */
1695 int
latin_char2cells(int c UNUSED)1696 latin_char2cells(int c UNUSED)
1697 {
1698 return 1;
1699 }
1700
1701 static int
dbcs_char2cells(int c)1702 dbcs_char2cells(int c)
1703 {
1704 // Number of cells is equal to number of bytes, except for euc-jp when
1705 // the first byte is 0x8e.
1706 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e)
1707 return 1;
1708 // use the first byte
1709 return MB_BYTE2LEN((unsigned)c >> 8);
1710 }
1711
1712 /*
1713 * Return the number of cells occupied by string "p".
1714 * Stop at a NUL character. When "len" >= 0 stop at character "p[len]".
1715 */
1716 int
mb_string2cells(char_u * p,int len)1717 mb_string2cells(char_u *p, int len)
1718 {
1719 int i;
1720 int clen = 0;
1721
1722 for (i = 0; (len < 0 || i < len) && p[i] != NUL; i += (*mb_ptr2len)(p + i))
1723 clen += (*mb_ptr2cells)(p + i);
1724 return clen;
1725 }
1726
1727 /*
1728 * mb_off2cells() function pointer.
1729 * Return number of display cells for char at ScreenLines[off].
1730 * We make sure that the offset used is less than "max_off".
1731 */
1732 int
latin_off2cells(unsigned off UNUSED,unsigned max_off UNUSED)1733 latin_off2cells(unsigned off UNUSED, unsigned max_off UNUSED)
1734 {
1735 return 1;
1736 }
1737
1738 int
dbcs_off2cells(unsigned off,unsigned max_off)1739 dbcs_off2cells(unsigned off, unsigned max_off)
1740 {
1741 // never check beyond end of the line
1742 if (off >= max_off)
1743 return 1;
1744
1745 // Number of cells is equal to number of bytes, except for euc-jp when
1746 // the first byte is 0x8e.
1747 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e)
1748 return 1;
1749 return MB_BYTE2LEN(ScreenLines[off]);
1750 }
1751
1752 int
utf_off2cells(unsigned off,unsigned max_off)1753 utf_off2cells(unsigned off, unsigned max_off)
1754 {
1755 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1;
1756 }
1757
1758 /*
1759 * mb_ptr2char() function pointer.
1760 * Convert a byte sequence into a character.
1761 */
1762 int
latin_ptr2char(char_u * p)1763 latin_ptr2char(char_u *p)
1764 {
1765 return *p;
1766 }
1767
1768 static int
dbcs_ptr2char(char_u * p)1769 dbcs_ptr2char(char_u *p)
1770 {
1771 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL)
1772 return (p[0] << 8) + p[1];
1773 return *p;
1774 }
1775
1776 /*
1777 * Convert a UTF-8 byte sequence to a character number.
1778 * If the sequence is illegal or truncated by a NUL the first byte is
1779 * returned.
1780 * For an overlong sequence this may return zero.
1781 * Does not include composing characters, of course.
1782 */
1783 int
utf_ptr2char(char_u * p)1784 utf_ptr2char(char_u *p)
1785 {
1786 int len;
1787
1788 if (p[0] < 0x80) // be quick for ASCII
1789 return p[0];
1790
1791 len = utf8len_tab_zero[p[0]];
1792 if (len > 1 && (p[1] & 0xc0) == 0x80)
1793 {
1794 if (len == 2)
1795 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f);
1796 if ((p[2] & 0xc0) == 0x80)
1797 {
1798 if (len == 3)
1799 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6)
1800 + (p[2] & 0x3f);
1801 if ((p[3] & 0xc0) == 0x80)
1802 {
1803 if (len == 4)
1804 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12)
1805 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f);
1806 if ((p[4] & 0xc0) == 0x80)
1807 {
1808 if (len == 5)
1809 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18)
1810 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6)
1811 + (p[4] & 0x3f);
1812 if ((p[5] & 0xc0) == 0x80 && len == 6)
1813 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24)
1814 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12)
1815 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f);
1816 }
1817 }
1818 }
1819 }
1820 // Illegal value, just return the first byte
1821 return p[0];
1822 }
1823
1824 /*
1825 * Convert a UTF-8 byte sequence to a wide character.
1826 * String is assumed to be terminated by NUL or after "n" bytes, whichever
1827 * comes first.
1828 * The function is safe in the sense that it never accesses memory beyond the
1829 * first "n" bytes of "s".
1830 *
1831 * On success, returns decoded codepoint, advances "s" to the beginning of
1832 * next character and decreases "n" accordingly.
1833 *
1834 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past
1835 * NUL byte.
1836 *
1837 * If byte sequence is illegal or incomplete, returns -1 and does not advance
1838 * "s".
1839 */
1840 static int
utf_safe_read_char_adv(char_u ** s,size_t * n)1841 utf_safe_read_char_adv(char_u **s, size_t *n)
1842 {
1843 int c, k;
1844
1845 if (*n == 0) // end of buffer
1846 return 0;
1847
1848 k = utf8len_tab_zero[**s];
1849
1850 if (k == 1)
1851 {
1852 // ASCII character or NUL
1853 (*n)--;
1854 return *(*s)++;
1855 }
1856
1857 if ((size_t)k <= *n)
1858 {
1859 // We have a multibyte sequence and it isn't truncated by buffer
1860 // limits so utf_ptr2char() is safe to use. Or the first byte is
1861 // illegal (k=0), and it's also safe to use utf_ptr2char().
1862 c = utf_ptr2char(*s);
1863
1864 // On failure, utf_ptr2char() returns the first byte, so here we
1865 // check equality with the first byte. The only non-ASCII character
1866 // which equals the first byte of its own UTF-8 representation is
1867 // U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too.
1868 // It's safe even if n=1, else we would have k=2 > n.
1869 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83))
1870 {
1871 // byte sequence was successfully decoded
1872 *s += k;
1873 *n -= k;
1874 return c;
1875 }
1876 }
1877
1878 // byte sequence is incomplete or illegal
1879 return -1;
1880 }
1881
1882 /*
1883 * Get character at **pp and advance *pp to the next character.
1884 * Note: composing characters are skipped!
1885 */
1886 int
mb_ptr2char_adv(char_u ** pp)1887 mb_ptr2char_adv(char_u **pp)
1888 {
1889 int c;
1890
1891 c = (*mb_ptr2char)(*pp);
1892 *pp += (*mb_ptr2len)(*pp);
1893 return c;
1894 }
1895
1896 /*
1897 * Get character at **pp and advance *pp to the next character.
1898 * Note: composing characters are returned as separate characters.
1899 */
1900 int
mb_cptr2char_adv(char_u ** pp)1901 mb_cptr2char_adv(char_u **pp)
1902 {
1903 int c;
1904
1905 c = (*mb_ptr2char)(*pp);
1906 if (enc_utf8)
1907 *pp += utf_ptr2len(*pp);
1908 else
1909 *pp += (*mb_ptr2len)(*pp);
1910 return c;
1911 }
1912
1913 #if defined(FEAT_ARABIC) || defined(PROTO)
1914 /*
1915 * Check if the character pointed to by "p2" is a composing character when it
1916 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which
1917 * behaves like a composing character.
1918 */
1919 int
utf_composinglike(char_u * p1,char_u * p2)1920 utf_composinglike(char_u *p1, char_u *p2)
1921 {
1922 int c2;
1923
1924 c2 = utf_ptr2char(p2);
1925 if (utf_iscomposing(c2))
1926 return TRUE;
1927 if (!arabic_maycombine(c2))
1928 return FALSE;
1929 return arabic_combine(utf_ptr2char(p1), c2);
1930 }
1931 #endif
1932
1933 /*
1934 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO
1935 * composing characters.
1936 */
1937 int
utfc_ptr2char(char_u * p,int * pcc)1938 utfc_ptr2char(
1939 char_u *p,
1940 int *pcc) // return: composing chars, last one is 0
1941 {
1942 int len;
1943 int c;
1944 int cc;
1945 int i = 0;
1946
1947 c = utf_ptr2char(p);
1948 len = utf_ptr2len(p);
1949
1950 // Only accept a composing char when the first char isn't illegal.
1951 if ((len > 1 || *p < 0x80)
1952 && p[len] >= 0x80
1953 && UTF_COMPOSINGLIKE(p, p + len))
1954 {
1955 cc = utf_ptr2char(p + len);
1956 for (;;)
1957 {
1958 pcc[i++] = cc;
1959 if (i == MAX_MCO)
1960 break;
1961 len += utf_ptr2len(p + len);
1962 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len)))
1963 break;
1964 }
1965 }
1966
1967 if (i < MAX_MCO) // last composing char must be 0
1968 pcc[i] = 0;
1969
1970 return c;
1971 }
1972
1973 /*
1974 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO
1975 * composing characters. Use no more than p[maxlen].
1976 */
1977 int
utfc_ptr2char_len(char_u * p,int * pcc,int maxlen)1978 utfc_ptr2char_len(
1979 char_u *p,
1980 int *pcc, // return: composing chars, last one is 0
1981 int maxlen)
1982 {
1983 int len;
1984 int c;
1985 int cc;
1986 int i = 0;
1987
1988 c = utf_ptr2char(p);
1989 len = utf_ptr2len_len(p, maxlen);
1990 // Only accept a composing char when the first char isn't illegal.
1991 if ((len > 1 || *p < 0x80)
1992 && len < maxlen
1993 && p[len] >= 0x80
1994 && UTF_COMPOSINGLIKE(p, p + len))
1995 {
1996 cc = utf_ptr2char(p + len);
1997 for (;;)
1998 {
1999 pcc[i++] = cc;
2000 if (i == MAX_MCO)
2001 break;
2002 len += utf_ptr2len_len(p + len, maxlen - len);
2003 if (len >= maxlen
2004 || p[len] < 0x80
2005 || !utf_iscomposing(cc = utf_ptr2char(p + len)))
2006 break;
2007 }
2008 }
2009
2010 if (i < MAX_MCO) // last composing char must be 0
2011 pcc[i] = 0;
2012
2013 return c;
2014 }
2015
2016 /*
2017 * Convert the character at screen position "off" to a sequence of bytes.
2018 * Includes the composing characters.
2019 * "buf" must at least have the length MB_MAXBYTES + 1.
2020 * Only to be used when ScreenLinesUC[off] != 0.
2021 * Returns the produced number of bytes.
2022 */
2023 int
utfc_char2bytes(int off,char_u * buf)2024 utfc_char2bytes(int off, char_u *buf)
2025 {
2026 int len;
2027 int i;
2028
2029 len = utf_char2bytes(ScreenLinesUC[off], buf);
2030 for (i = 0; i < Screen_mco; ++i)
2031 {
2032 if (ScreenLinesC[i][off] == 0)
2033 break;
2034 len += utf_char2bytes(ScreenLinesC[i][off], buf + len);
2035 }
2036 return len;
2037 }
2038
2039 /*
2040 * Get the length of a UTF-8 byte sequence, not including any following
2041 * composing characters.
2042 * Returns 0 for "".
2043 * Returns 1 for an illegal byte sequence.
2044 */
2045 int
utf_ptr2len(char_u * p)2046 utf_ptr2len(char_u *p)
2047 {
2048 int len;
2049 int i;
2050
2051 if (*p == NUL)
2052 return 0;
2053 len = utf8len_tab[*p];
2054 for (i = 1; i < len; ++i)
2055 if ((p[i] & 0xc0) != 0x80)
2056 return 1;
2057 return len;
2058 }
2059
2060 /*
2061 * Return length of UTF-8 character, obtained from the first byte.
2062 * "b" must be between 0 and 255!
2063 * Returns 1 for an invalid first byte value.
2064 */
2065 int
utf_byte2len(int b)2066 utf_byte2len(int b)
2067 {
2068 return utf8len_tab[b];
2069 }
2070
2071 /*
2072 * Get the length of UTF-8 byte sequence "p[size]". Does not include any
2073 * following composing characters.
2074 * Returns 1 for "".
2075 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.).
2076 * Returns number > "size" for an incomplete byte sequence.
2077 * Never returns zero.
2078 */
2079 int
utf_ptr2len_len(char_u * p,int size)2080 utf_ptr2len_len(char_u *p, int size)
2081 {
2082 int len;
2083 int i;
2084 int m;
2085
2086 len = utf8len_tab[*p];
2087 if (len == 1)
2088 return 1; // NUL, ascii or illegal lead byte
2089 if (len > size)
2090 m = size; // incomplete byte sequence.
2091 else
2092 m = len;
2093 for (i = 1; i < m; ++i)
2094 if ((p[i] & 0xc0) != 0x80)
2095 return 1;
2096 return len;
2097 }
2098
2099 /*
2100 * Return the number of bytes the UTF-8 encoding of the character at "p" takes.
2101 * This includes following composing characters.
2102 */
2103 int
utfc_ptr2len(char_u * p)2104 utfc_ptr2len(char_u *p)
2105 {
2106 int len;
2107 int b0 = *p;
2108 #ifdef FEAT_ARABIC
2109 int prevlen;
2110 #endif
2111
2112 if (b0 == NUL)
2113 return 0;
2114 if (b0 < 0x80 && p[1] < 0x80) // be quick for ASCII
2115 return 1;
2116
2117 // Skip over first UTF-8 char, stopping at a NUL byte.
2118 len = utf_ptr2len(p);
2119
2120 // Check for illegal byte.
2121 if (len == 1 && b0 >= 0x80)
2122 return 1;
2123
2124 /*
2125 * Check for composing characters. We can handle only the first six, but
2126 * skip all of them (otherwise the cursor would get stuck).
2127 */
2128 #ifdef FEAT_ARABIC
2129 prevlen = 0;
2130 #endif
2131 for (;;)
2132 {
2133 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len))
2134 return len;
2135
2136 // Skip over composing char
2137 #ifdef FEAT_ARABIC
2138 prevlen = len;
2139 #endif
2140 len += utf_ptr2len(p + len);
2141 }
2142 }
2143
2144 /*
2145 * Return the number of bytes the UTF-8 encoding of the character at "p[size]"
2146 * takes. This includes following composing characters.
2147 * Returns 0 for an empty string.
2148 * Returns 1 for an illegal char or an incomplete byte sequence.
2149 */
2150 int
utfc_ptr2len_len(char_u * p,int size)2151 utfc_ptr2len_len(char_u *p, int size)
2152 {
2153 int len;
2154 #ifdef FEAT_ARABIC
2155 int prevlen;
2156 #endif
2157
2158 if (size < 1 || *p == NUL)
2159 return 0;
2160 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) // be quick for ASCII
2161 return 1;
2162
2163 // Skip over first UTF-8 char, stopping at a NUL byte.
2164 len = utf_ptr2len_len(p, size);
2165
2166 // Check for illegal byte and incomplete byte sequence.
2167 if ((len == 1 && p[0] >= 0x80) || len > size)
2168 return 1;
2169
2170 /*
2171 * Check for composing characters. We can handle only the first six, but
2172 * skip all of them (otherwise the cursor would get stuck).
2173 */
2174 #ifdef FEAT_ARABIC
2175 prevlen = 0;
2176 #endif
2177 while (len < size)
2178 {
2179 int len_next_char;
2180
2181 if (p[len] < 0x80)
2182 break;
2183
2184 /*
2185 * Next character length should not go beyond size to ensure that
2186 * UTF_COMPOSINGLIKE(...) does not read beyond size.
2187 */
2188 len_next_char = utf_ptr2len_len(p + len, size - len);
2189 if (len_next_char > size - len)
2190 break;
2191
2192 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len))
2193 break;
2194
2195 // Skip over composing char
2196 #ifdef FEAT_ARABIC
2197 prevlen = len;
2198 #endif
2199 len += len_next_char;
2200 }
2201 return len;
2202 }
2203
2204 /*
2205 * Return the number of bytes the UTF-8 encoding of character "c" takes.
2206 * This does not include composing characters.
2207 */
2208 int
utf_char2len(int c)2209 utf_char2len(int c)
2210 {
2211 if (c < 0x80)
2212 return 1;
2213 if (c < 0x800)
2214 return 2;
2215 if (c < 0x10000)
2216 return 3;
2217 if (c < 0x200000)
2218 return 4;
2219 if (c < 0x4000000)
2220 return 5;
2221 return 6;
2222 }
2223
2224 /*
2225 * Convert Unicode character "c" to UTF-8 string in "buf[]".
2226 * Returns the number of bytes.
2227 */
2228 int
utf_char2bytes(int c,char_u * buf)2229 utf_char2bytes(int c, char_u *buf)
2230 {
2231 if (c < 0x80) // 7 bits
2232 {
2233 buf[0] = c;
2234 return 1;
2235 }
2236 if (c < 0x800) // 11 bits
2237 {
2238 buf[0] = 0xc0 + ((unsigned)c >> 6);
2239 buf[1] = 0x80 + (c & 0x3f);
2240 return 2;
2241 }
2242 if (c < 0x10000) // 16 bits
2243 {
2244 buf[0] = 0xe0 + ((unsigned)c >> 12);
2245 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f);
2246 buf[2] = 0x80 + (c & 0x3f);
2247 return 3;
2248 }
2249 if (c < 0x200000) // 21 bits
2250 {
2251 buf[0] = 0xf0 + ((unsigned)c >> 18);
2252 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f);
2253 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f);
2254 buf[3] = 0x80 + (c & 0x3f);
2255 return 4;
2256 }
2257 if (c < 0x4000000) // 26 bits
2258 {
2259 buf[0] = 0xf8 + ((unsigned)c >> 24);
2260 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f);
2261 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f);
2262 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f);
2263 buf[4] = 0x80 + (c & 0x3f);
2264 return 5;
2265 }
2266 // 31 bits
2267 buf[0] = 0xfc + ((unsigned)c >> 30);
2268 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f);
2269 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f);
2270 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f);
2271 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f);
2272 buf[5] = 0x80 + (c & 0x3f);
2273 return 6;
2274 }
2275
2276 #if defined(FEAT_TERMINAL) || defined(PROTO)
2277 /*
2278 * utf_iscomposing() with different argument type for libvterm.
2279 */
2280 int
utf_iscomposing_uint(UINT32_T c)2281 utf_iscomposing_uint(UINT32_T c)
2282 {
2283 return utf_iscomposing((int)c);
2284 }
2285 #endif
2286
2287 /*
2288 * Return TRUE if "c" is a composing UTF-8 character. This means it will be
2289 * drawn on top of the preceding character.
2290 * Based on code from Markus Kuhn.
2291 */
2292 int
utf_iscomposing(int c)2293 utf_iscomposing(int c)
2294 {
2295 // Sorted list of non-overlapping intervals.
2296 // Generated by ../runtime/tools/unicode.vim.
2297 static struct interval combining[] =
2298 {
2299 {0x0300, 0x036f},
2300 {0x0483, 0x0489},
2301 {0x0591, 0x05bd},
2302 {0x05bf, 0x05bf},
2303 {0x05c1, 0x05c2},
2304 {0x05c4, 0x05c5},
2305 {0x05c7, 0x05c7},
2306 {0x0610, 0x061a},
2307 {0x064b, 0x065f},
2308 {0x0670, 0x0670},
2309 {0x06d6, 0x06dc},
2310 {0x06df, 0x06e4},
2311 {0x06e7, 0x06e8},
2312 {0x06ea, 0x06ed},
2313 {0x0711, 0x0711},
2314 {0x0730, 0x074a},
2315 {0x07a6, 0x07b0},
2316 {0x07eb, 0x07f3},
2317 {0x07fd, 0x07fd},
2318 {0x0816, 0x0819},
2319 {0x081b, 0x0823},
2320 {0x0825, 0x0827},
2321 {0x0829, 0x082d},
2322 {0x0859, 0x085b},
2323 {0x08d3, 0x08e1},
2324 {0x08e3, 0x0903},
2325 {0x093a, 0x093c},
2326 {0x093e, 0x094f},
2327 {0x0951, 0x0957},
2328 {0x0962, 0x0963},
2329 {0x0981, 0x0983},
2330 {0x09bc, 0x09bc},
2331 {0x09be, 0x09c4},
2332 {0x09c7, 0x09c8},
2333 {0x09cb, 0x09cd},
2334 {0x09d7, 0x09d7},
2335 {0x09e2, 0x09e3},
2336 {0x09fe, 0x09fe},
2337 {0x0a01, 0x0a03},
2338 {0x0a3c, 0x0a3c},
2339 {0x0a3e, 0x0a42},
2340 {0x0a47, 0x0a48},
2341 {0x0a4b, 0x0a4d},
2342 {0x0a51, 0x0a51},
2343 {0x0a70, 0x0a71},
2344 {0x0a75, 0x0a75},
2345 {0x0a81, 0x0a83},
2346 {0x0abc, 0x0abc},
2347 {0x0abe, 0x0ac5},
2348 {0x0ac7, 0x0ac9},
2349 {0x0acb, 0x0acd},
2350 {0x0ae2, 0x0ae3},
2351 {0x0afa, 0x0aff},
2352 {0x0b01, 0x0b03},
2353 {0x0b3c, 0x0b3c},
2354 {0x0b3e, 0x0b44},
2355 {0x0b47, 0x0b48},
2356 {0x0b4b, 0x0b4d},
2357 {0x0b55, 0x0b57},
2358 {0x0b62, 0x0b63},
2359 {0x0b82, 0x0b82},
2360 {0x0bbe, 0x0bc2},
2361 {0x0bc6, 0x0bc8},
2362 {0x0bca, 0x0bcd},
2363 {0x0bd7, 0x0bd7},
2364 {0x0c00, 0x0c04},
2365 {0x0c3e, 0x0c44},
2366 {0x0c46, 0x0c48},
2367 {0x0c4a, 0x0c4d},
2368 {0x0c55, 0x0c56},
2369 {0x0c62, 0x0c63},
2370 {0x0c81, 0x0c83},
2371 {0x0cbc, 0x0cbc},
2372 {0x0cbe, 0x0cc4},
2373 {0x0cc6, 0x0cc8},
2374 {0x0cca, 0x0ccd},
2375 {0x0cd5, 0x0cd6},
2376 {0x0ce2, 0x0ce3},
2377 {0x0d00, 0x0d03},
2378 {0x0d3b, 0x0d3c},
2379 {0x0d3e, 0x0d44},
2380 {0x0d46, 0x0d48},
2381 {0x0d4a, 0x0d4d},
2382 {0x0d57, 0x0d57},
2383 {0x0d62, 0x0d63},
2384 {0x0d81, 0x0d83},
2385 {0x0dca, 0x0dca},
2386 {0x0dcf, 0x0dd4},
2387 {0x0dd6, 0x0dd6},
2388 {0x0dd8, 0x0ddf},
2389 {0x0df2, 0x0df3},
2390 {0x0e31, 0x0e31},
2391 {0x0e34, 0x0e3a},
2392 {0x0e47, 0x0e4e},
2393 {0x0eb1, 0x0eb1},
2394 {0x0eb4, 0x0ebc},
2395 {0x0ec8, 0x0ecd},
2396 {0x0f18, 0x0f19},
2397 {0x0f35, 0x0f35},
2398 {0x0f37, 0x0f37},
2399 {0x0f39, 0x0f39},
2400 {0x0f3e, 0x0f3f},
2401 {0x0f71, 0x0f84},
2402 {0x0f86, 0x0f87},
2403 {0x0f8d, 0x0f97},
2404 {0x0f99, 0x0fbc},
2405 {0x0fc6, 0x0fc6},
2406 {0x102b, 0x103e},
2407 {0x1056, 0x1059},
2408 {0x105e, 0x1060},
2409 {0x1062, 0x1064},
2410 {0x1067, 0x106d},
2411 {0x1071, 0x1074},
2412 {0x1082, 0x108d},
2413 {0x108f, 0x108f},
2414 {0x109a, 0x109d},
2415 {0x135d, 0x135f},
2416 {0x1712, 0x1714},
2417 {0x1732, 0x1734},
2418 {0x1752, 0x1753},
2419 {0x1772, 0x1773},
2420 {0x17b4, 0x17d3},
2421 {0x17dd, 0x17dd},
2422 {0x180b, 0x180d},
2423 {0x1885, 0x1886},
2424 {0x18a9, 0x18a9},
2425 {0x1920, 0x192b},
2426 {0x1930, 0x193b},
2427 {0x1a17, 0x1a1b},
2428 {0x1a55, 0x1a5e},
2429 {0x1a60, 0x1a7c},
2430 {0x1a7f, 0x1a7f},
2431 {0x1ab0, 0x1ac0},
2432 {0x1b00, 0x1b04},
2433 {0x1b34, 0x1b44},
2434 {0x1b6b, 0x1b73},
2435 {0x1b80, 0x1b82},
2436 {0x1ba1, 0x1bad},
2437 {0x1be6, 0x1bf3},
2438 {0x1c24, 0x1c37},
2439 {0x1cd0, 0x1cd2},
2440 {0x1cd4, 0x1ce8},
2441 {0x1ced, 0x1ced},
2442 {0x1cf4, 0x1cf4},
2443 {0x1cf7, 0x1cf9},
2444 {0x1dc0, 0x1df9},
2445 {0x1dfb, 0x1dff},
2446 {0x20d0, 0x20f0},
2447 {0x2cef, 0x2cf1},
2448 {0x2d7f, 0x2d7f},
2449 {0x2de0, 0x2dff},
2450 {0x302a, 0x302f},
2451 {0x3099, 0x309a},
2452 {0xa66f, 0xa672},
2453 {0xa674, 0xa67d},
2454 {0xa69e, 0xa69f},
2455 {0xa6f0, 0xa6f1},
2456 {0xa802, 0xa802},
2457 {0xa806, 0xa806},
2458 {0xa80b, 0xa80b},
2459 {0xa823, 0xa827},
2460 {0xa82c, 0xa82c},
2461 {0xa880, 0xa881},
2462 {0xa8b4, 0xa8c5},
2463 {0xa8e0, 0xa8f1},
2464 {0xa8ff, 0xa8ff},
2465 {0xa926, 0xa92d},
2466 {0xa947, 0xa953},
2467 {0xa980, 0xa983},
2468 {0xa9b3, 0xa9c0},
2469 {0xa9e5, 0xa9e5},
2470 {0xaa29, 0xaa36},
2471 {0xaa43, 0xaa43},
2472 {0xaa4c, 0xaa4d},
2473 {0xaa7b, 0xaa7d},
2474 {0xaab0, 0xaab0},
2475 {0xaab2, 0xaab4},
2476 {0xaab7, 0xaab8},
2477 {0xaabe, 0xaabf},
2478 {0xaac1, 0xaac1},
2479 {0xaaeb, 0xaaef},
2480 {0xaaf5, 0xaaf6},
2481 {0xabe3, 0xabea},
2482 {0xabec, 0xabed},
2483 {0xfb1e, 0xfb1e},
2484 {0xfe00, 0xfe0f},
2485 {0xfe20, 0xfe2f},
2486 {0x101fd, 0x101fd},
2487 {0x102e0, 0x102e0},
2488 {0x10376, 0x1037a},
2489 {0x10a01, 0x10a03},
2490 {0x10a05, 0x10a06},
2491 {0x10a0c, 0x10a0f},
2492 {0x10a38, 0x10a3a},
2493 {0x10a3f, 0x10a3f},
2494 {0x10ae5, 0x10ae6},
2495 {0x10d24, 0x10d27},
2496 {0x10eab, 0x10eac},
2497 {0x10f46, 0x10f50},
2498 {0x11000, 0x11002},
2499 {0x11038, 0x11046},
2500 {0x1107f, 0x11082},
2501 {0x110b0, 0x110ba},
2502 {0x11100, 0x11102},
2503 {0x11127, 0x11134},
2504 {0x11145, 0x11146},
2505 {0x11173, 0x11173},
2506 {0x11180, 0x11182},
2507 {0x111b3, 0x111c0},
2508 {0x111c9, 0x111cc},
2509 {0x111ce, 0x111cf},
2510 {0x1122c, 0x11237},
2511 {0x1123e, 0x1123e},
2512 {0x112df, 0x112ea},
2513 {0x11300, 0x11303},
2514 {0x1133b, 0x1133c},
2515 {0x1133e, 0x11344},
2516 {0x11347, 0x11348},
2517 {0x1134b, 0x1134d},
2518 {0x11357, 0x11357},
2519 {0x11362, 0x11363},
2520 {0x11366, 0x1136c},
2521 {0x11370, 0x11374},
2522 {0x11435, 0x11446},
2523 {0x1145e, 0x1145e},
2524 {0x114b0, 0x114c3},
2525 {0x115af, 0x115b5},
2526 {0x115b8, 0x115c0},
2527 {0x115dc, 0x115dd},
2528 {0x11630, 0x11640},
2529 {0x116ab, 0x116b7},
2530 {0x1171d, 0x1172b},
2531 {0x1182c, 0x1183a},
2532 {0x11930, 0x11935},
2533 {0x11937, 0x11938},
2534 {0x1193b, 0x1193e},
2535 {0x11940, 0x11940},
2536 {0x11942, 0x11943},
2537 {0x119d1, 0x119d7},
2538 {0x119da, 0x119e0},
2539 {0x119e4, 0x119e4},
2540 {0x11a01, 0x11a0a},
2541 {0x11a33, 0x11a39},
2542 {0x11a3b, 0x11a3e},
2543 {0x11a47, 0x11a47},
2544 {0x11a51, 0x11a5b},
2545 {0x11a8a, 0x11a99},
2546 {0x11c2f, 0x11c36},
2547 {0x11c38, 0x11c3f},
2548 {0x11c92, 0x11ca7},
2549 {0x11ca9, 0x11cb6},
2550 {0x11d31, 0x11d36},
2551 {0x11d3a, 0x11d3a},
2552 {0x11d3c, 0x11d3d},
2553 {0x11d3f, 0x11d45},
2554 {0x11d47, 0x11d47},
2555 {0x11d8a, 0x11d8e},
2556 {0x11d90, 0x11d91},
2557 {0x11d93, 0x11d97},
2558 {0x11ef3, 0x11ef6},
2559 {0x16af0, 0x16af4},
2560 {0x16b30, 0x16b36},
2561 {0x16f4f, 0x16f4f},
2562 {0x16f51, 0x16f87},
2563 {0x16f8f, 0x16f92},
2564 {0x16fe4, 0x16fe4},
2565 {0x16ff0, 0x16ff1},
2566 {0x1bc9d, 0x1bc9e},
2567 {0x1d165, 0x1d169},
2568 {0x1d16d, 0x1d172},
2569 {0x1d17b, 0x1d182},
2570 {0x1d185, 0x1d18b},
2571 {0x1d1aa, 0x1d1ad},
2572 {0x1d242, 0x1d244},
2573 {0x1da00, 0x1da36},
2574 {0x1da3b, 0x1da6c},
2575 {0x1da75, 0x1da75},
2576 {0x1da84, 0x1da84},
2577 {0x1da9b, 0x1da9f},
2578 {0x1daa1, 0x1daaf},
2579 {0x1e000, 0x1e006},
2580 {0x1e008, 0x1e018},
2581 {0x1e01b, 0x1e021},
2582 {0x1e023, 0x1e024},
2583 {0x1e026, 0x1e02a},
2584 {0x1e130, 0x1e136},
2585 {0x1e2ec, 0x1e2ef},
2586 {0x1e8d0, 0x1e8d6},
2587 {0x1e944, 0x1e94a},
2588 {0xe0100, 0xe01ef}
2589 };
2590
2591 return intable(combining, sizeof(combining), c);
2592 }
2593
2594 /*
2595 * Return TRUE for characters that can be displayed in a normal way.
2596 * Only for characters of 0x100 and above!
2597 */
2598 int
utf_printable(int c)2599 utf_printable(int c)
2600 {
2601 #ifdef USE_WCHAR_FUNCTIONS
2602 /*
2603 * Assume the iswprint() library function works better than our own stuff.
2604 */
2605 return iswprint(c);
2606 #else
2607 // Sorted list of non-overlapping intervals.
2608 // 0xd800-0xdfff is reserved for UTF-16, actually illegal.
2609 static struct interval nonprint[] =
2610 {
2611 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e},
2612 {0x2060, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb},
2613 {0xfffe, 0xffff}
2614 };
2615
2616 return !intable(nonprint, sizeof(nonprint), c);
2617 #endif
2618 }
2619
2620 // Sorted list of non-overlapping intervals of all Emoji characters,
2621 // based on http://unicode.org/emoji/charts/emoji-list.html
2622 // Generated by ../runtime/tools/unicode.vim.
2623 // Excludes 0x00a9 and 0x00ae because they are considered latin1.
2624 static struct interval emoji_all[] =
2625 {
2626 {0x203c, 0x203c},
2627 {0x2049, 0x2049},
2628 {0x2122, 0x2122},
2629 {0x2139, 0x2139},
2630 {0x2194, 0x2199},
2631 {0x21a9, 0x21aa},
2632 {0x231a, 0x231b},
2633 {0x2328, 0x2328},
2634 {0x23cf, 0x23cf},
2635 {0x23e9, 0x23f3},
2636 {0x23f8, 0x23fa},
2637 {0x24c2, 0x24c2},
2638 {0x25aa, 0x25ab},
2639 {0x25b6, 0x25b6},
2640 {0x25c0, 0x25c0},
2641 {0x25fb, 0x25fe},
2642 {0x2600, 0x2604},
2643 {0x260e, 0x260e},
2644 {0x2611, 0x2611},
2645 {0x2614, 0x2615},
2646 {0x2618, 0x2618},
2647 {0x261d, 0x261d},
2648 {0x2620, 0x2620},
2649 {0x2622, 0x2623},
2650 {0x2626, 0x2626},
2651 {0x262a, 0x262a},
2652 {0x262e, 0x262f},
2653 {0x2638, 0x263a},
2654 {0x2640, 0x2640},
2655 {0x2642, 0x2642},
2656 {0x2648, 0x2653},
2657 {0x265f, 0x2660},
2658 {0x2663, 0x2663},
2659 {0x2665, 0x2666},
2660 {0x2668, 0x2668},
2661 {0x267b, 0x267b},
2662 {0x267e, 0x267f},
2663 {0x2692, 0x2697},
2664 {0x2699, 0x2699},
2665 {0x269b, 0x269c},
2666 {0x26a0, 0x26a1},
2667 {0x26a7, 0x26a7},
2668 {0x26aa, 0x26ab},
2669 {0x26b0, 0x26b1},
2670 {0x26bd, 0x26be},
2671 {0x26c4, 0x26c5},
2672 {0x26c8, 0x26c8},
2673 {0x26ce, 0x26cf},
2674 {0x26d1, 0x26d1},
2675 {0x26d3, 0x26d4},
2676 {0x26e9, 0x26ea},
2677 {0x26f0, 0x26f5},
2678 {0x26f7, 0x26fa},
2679 {0x26fd, 0x26fd},
2680 {0x2702, 0x2702},
2681 {0x2705, 0x2705},
2682 {0x2708, 0x270d},
2683 {0x270f, 0x270f},
2684 {0x2712, 0x2712},
2685 {0x2714, 0x2714},
2686 {0x2716, 0x2716},
2687 {0x271d, 0x271d},
2688 {0x2721, 0x2721},
2689 {0x2728, 0x2728},
2690 {0x2733, 0x2734},
2691 {0x2744, 0x2744},
2692 {0x2747, 0x2747},
2693 {0x274c, 0x274c},
2694 {0x274e, 0x274e},
2695 {0x2753, 0x2755},
2696 {0x2757, 0x2757},
2697 {0x2763, 0x2764},
2698 {0x2795, 0x2797},
2699 {0x27a1, 0x27a1},
2700 {0x27b0, 0x27b0},
2701 {0x27bf, 0x27bf},
2702 {0x2934, 0x2935},
2703 {0x2b05, 0x2b07},
2704 {0x2b1b, 0x2b1c},
2705 {0x2b50, 0x2b50},
2706 {0x2b55, 0x2b55},
2707 {0x3030, 0x3030},
2708 {0x303d, 0x303d},
2709 {0x3297, 0x3297},
2710 {0x3299, 0x3299},
2711 {0x1f004, 0x1f004},
2712 {0x1f0cf, 0x1f0cf},
2713 {0x1f170, 0x1f171},
2714 {0x1f17e, 0x1f17f},
2715 {0x1f18e, 0x1f18e},
2716 {0x1f191, 0x1f19a},
2717 {0x1f1e6, 0x1f1ff},
2718 {0x1f201, 0x1f202},
2719 {0x1f21a, 0x1f21a},
2720 {0x1f22f, 0x1f22f},
2721 {0x1f232, 0x1f23a},
2722 {0x1f250, 0x1f251},
2723 {0x1f300, 0x1f321},
2724 {0x1f324, 0x1f393},
2725 {0x1f396, 0x1f397},
2726 {0x1f399, 0x1f39b},
2727 {0x1f39e, 0x1f3f0},
2728 {0x1f3f3, 0x1f3f5},
2729 {0x1f3f7, 0x1f4fd},
2730 {0x1f4ff, 0x1f53d},
2731 {0x1f549, 0x1f54e},
2732 {0x1f550, 0x1f567},
2733 {0x1f56f, 0x1f570},
2734 {0x1f573, 0x1f57a},
2735 {0x1f587, 0x1f587},
2736 {0x1f58a, 0x1f58d},
2737 {0x1f590, 0x1f590},
2738 {0x1f595, 0x1f596},
2739 {0x1f5a4, 0x1f5a5},
2740 {0x1f5a8, 0x1f5a8},
2741 {0x1f5b1, 0x1f5b2},
2742 {0x1f5bc, 0x1f5bc},
2743 {0x1f5c2, 0x1f5c4},
2744 {0x1f5d1, 0x1f5d3},
2745 {0x1f5dc, 0x1f5de},
2746 {0x1f5e1, 0x1f5e1},
2747 {0x1f5e3, 0x1f5e3},
2748 {0x1f5e8, 0x1f5e8},
2749 {0x1f5ef, 0x1f5ef},
2750 {0x1f5f3, 0x1f5f3},
2751 {0x1f5fa, 0x1f64f},
2752 {0x1f680, 0x1f6c5},
2753 {0x1f6cb, 0x1f6d2},
2754 {0x1f6d5, 0x1f6d7},
2755 {0x1f6e0, 0x1f6e5},
2756 {0x1f6e9, 0x1f6e9},
2757 {0x1f6eb, 0x1f6ec},
2758 {0x1f6f0, 0x1f6f0},
2759 {0x1f6f3, 0x1f6fc},
2760 {0x1f7e0, 0x1f7eb},
2761 {0x1f90c, 0x1f93a},
2762 {0x1f93c, 0x1f945},
2763 {0x1f947, 0x1f978},
2764 {0x1f97a, 0x1f9cb},
2765 {0x1f9cd, 0x1f9ff},
2766 {0x1fa70, 0x1fa74},
2767 {0x1fa78, 0x1fa7a},
2768 {0x1fa80, 0x1fa86},
2769 {0x1fa90, 0x1faa8},
2770 {0x1fab0, 0x1fab6},
2771 {0x1fac0, 0x1fac2},
2772 {0x1fad0, 0x1fad6}
2773 };
2774
2775 /*
2776 * Get class of a Unicode character.
2777 * 0: white space
2778 * 1: punctuation
2779 * 2 or bigger: some class of word character.
2780 */
2781 int
utf_class(int c)2782 utf_class(int c)
2783 {
2784 return utf_class_buf(c, curbuf);
2785 }
2786
2787 int
utf_class_buf(int c,buf_T * buf)2788 utf_class_buf(int c, buf_T *buf)
2789 {
2790 // sorted list of non-overlapping intervals
2791 static struct clinterval
2792 {
2793 unsigned int first;
2794 unsigned int last;
2795 unsigned int class;
2796 } classes[] =
2797 {
2798 {0x037e, 0x037e, 1}, // Greek question mark
2799 {0x0387, 0x0387, 1}, // Greek ano teleia
2800 {0x055a, 0x055f, 1}, // Armenian punctuation
2801 {0x0589, 0x0589, 1}, // Armenian full stop
2802 {0x05be, 0x05be, 1},
2803 {0x05c0, 0x05c0, 1},
2804 {0x05c3, 0x05c3, 1},
2805 {0x05f3, 0x05f4, 1},
2806 {0x060c, 0x060c, 1},
2807 {0x061b, 0x061b, 1},
2808 {0x061f, 0x061f, 1},
2809 {0x066a, 0x066d, 1},
2810 {0x06d4, 0x06d4, 1},
2811 {0x0700, 0x070d, 1}, // Syriac punctuation
2812 {0x0964, 0x0965, 1},
2813 {0x0970, 0x0970, 1},
2814 {0x0df4, 0x0df4, 1},
2815 {0x0e4f, 0x0e4f, 1},
2816 {0x0e5a, 0x0e5b, 1},
2817 {0x0f04, 0x0f12, 1},
2818 {0x0f3a, 0x0f3d, 1},
2819 {0x0f85, 0x0f85, 1},
2820 {0x104a, 0x104f, 1}, // Myanmar punctuation
2821 {0x10fb, 0x10fb, 1}, // Georgian punctuation
2822 {0x1361, 0x1368, 1}, // Ethiopic punctuation
2823 {0x166d, 0x166e, 1}, // Canadian Syl. punctuation
2824 {0x1680, 0x1680, 0},
2825 {0x169b, 0x169c, 1},
2826 {0x16eb, 0x16ed, 1},
2827 {0x1735, 0x1736, 1},
2828 {0x17d4, 0x17dc, 1}, // Khmer punctuation
2829 {0x1800, 0x180a, 1}, // Mongolian punctuation
2830 {0x2000, 0x200b, 0}, // spaces
2831 {0x200c, 0x2027, 1}, // punctuation and symbols
2832 {0x2028, 0x2029, 0},
2833 {0x202a, 0x202e, 1}, // punctuation and symbols
2834 {0x202f, 0x202f, 0},
2835 {0x2030, 0x205e, 1}, // punctuation and symbols
2836 {0x205f, 0x205f, 0},
2837 {0x2060, 0x27ff, 1}, // punctuation and symbols
2838 {0x2070, 0x207f, 0x2070}, // superscript
2839 {0x2080, 0x2094, 0x2080}, // subscript
2840 {0x20a0, 0x27ff, 1}, // all kinds of symbols
2841 {0x2800, 0x28ff, 0x2800}, // braille
2842 {0x2900, 0x2998, 1}, // arrows, brackets, etc.
2843 {0x29d8, 0x29db, 1},
2844 {0x29fc, 0x29fd, 1},
2845 {0x2e00, 0x2e7f, 1}, // supplemental punctuation
2846 {0x3000, 0x3000, 0}, // ideographic space
2847 {0x3001, 0x3020, 1}, // ideographic punctuation
2848 {0x3030, 0x3030, 1},
2849 {0x303d, 0x303d, 1},
2850 {0x3040, 0x309f, 0x3040}, // Hiragana
2851 {0x30a0, 0x30ff, 0x30a0}, // Katakana
2852 {0x3300, 0x9fff, 0x4e00}, // CJK Ideographs
2853 {0xac00, 0xd7a3, 0xac00}, // Hangul Syllables
2854 {0xf900, 0xfaff, 0x4e00}, // CJK Ideographs
2855 {0xfd3e, 0xfd3f, 1},
2856 {0xfe30, 0xfe6b, 1}, // punctuation forms
2857 {0xff00, 0xff0f, 1}, // half/fullwidth ASCII
2858 {0xff1a, 0xff20, 1}, // half/fullwidth ASCII
2859 {0xff3b, 0xff40, 1}, // half/fullwidth ASCII
2860 {0xff5b, 0xff65, 1}, // half/fullwidth ASCII
2861 {0x1d000, 0x1d24f, 1}, // Musical notation
2862 {0x1d400, 0x1d7ff, 1}, // Mathematical Alphanumeric Symbols
2863 {0x1f000, 0x1f2ff, 1}, // Game pieces; enclosed characters
2864 {0x1f300, 0x1f9ff, 1}, // Many symbol blocks
2865 {0x20000, 0x2a6df, 0x4e00}, // CJK Ideographs
2866 {0x2a700, 0x2b73f, 0x4e00}, // CJK Ideographs
2867 {0x2b740, 0x2b81f, 0x4e00}, // CJK Ideographs
2868 {0x2f800, 0x2fa1f, 0x4e00}, // CJK Ideographs
2869 };
2870
2871 int bot = 0;
2872 int top = ARRAY_LENGTH(classes) - 1;
2873 int mid;
2874
2875 // First quick check for Latin1 characters, use 'iskeyword'.
2876 if (c < 0x100)
2877 {
2878 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0)
2879 return 0; // blank
2880 if (vim_iswordc_buf(c, buf))
2881 return 2; // word character
2882 return 1; // punctuation
2883 }
2884
2885 // emoji
2886 if (intable(emoji_all, sizeof(emoji_all), c))
2887 return 3;
2888
2889 // binary search in table
2890 while (top >= bot)
2891 {
2892 mid = (bot + top) / 2;
2893 if (classes[mid].last < (unsigned int)c)
2894 bot = mid + 1;
2895 else if (classes[mid].first > (unsigned int)c)
2896 top = mid - 1;
2897 else
2898 return (int)classes[mid].class;
2899 }
2900
2901 // most other characters are "word" characters
2902 return 2;
2903 }
2904
2905 int
utf_ambiguous_width(int c)2906 utf_ambiguous_width(int c)
2907 {
2908 return c >= 0x80 && (intable(ambiguous, sizeof(ambiguous), c)
2909 || intable(emoji_all, sizeof(emoji_all), c));
2910 }
2911
2912 /*
2913 * Code for Unicode case-dependent operations. Based on notes in
2914 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt
2915 * This code uses simple case folding, not full case folding.
2916 * Last updated for Unicode 5.2.
2917 */
2918
2919 /*
2920 * The following tables are built by ../runtime/tools/unicode.vim.
2921 * They must be in numeric order, because we use binary search.
2922 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the
2923 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to
2924 * folded/upper/lower by adding 32.
2925 */
2926 typedef struct
2927 {
2928 int rangeStart;
2929 int rangeEnd;
2930 int step;
2931 int offset;
2932 } convertStruct;
2933
2934 static convertStruct foldCase[] =
2935 {
2936 {0x41,0x5a,1,32},
2937 {0xb5,0xb5,-1,775},
2938 {0xc0,0xd6,1,32},
2939 {0xd8,0xde,1,32},
2940 {0x100,0x12e,2,1},
2941 {0x132,0x136,2,1},
2942 {0x139,0x147,2,1},
2943 {0x14a,0x176,2,1},
2944 {0x178,0x178,-1,-121},
2945 {0x179,0x17d,2,1},
2946 {0x17f,0x17f,-1,-268},
2947 {0x181,0x181,-1,210},
2948 {0x182,0x184,2,1},
2949 {0x186,0x186,-1,206},
2950 {0x187,0x187,-1,1},
2951 {0x189,0x18a,1,205},
2952 {0x18b,0x18b,-1,1},
2953 {0x18e,0x18e,-1,79},
2954 {0x18f,0x18f,-1,202},
2955 {0x190,0x190,-1,203},
2956 {0x191,0x191,-1,1},
2957 {0x193,0x193,-1,205},
2958 {0x194,0x194,-1,207},
2959 {0x196,0x196,-1,211},
2960 {0x197,0x197,-1,209},
2961 {0x198,0x198,-1,1},
2962 {0x19c,0x19c,-1,211},
2963 {0x19d,0x19d,-1,213},
2964 {0x19f,0x19f,-1,214},
2965 {0x1a0,0x1a4,2,1},
2966 {0x1a6,0x1a6,-1,218},
2967 {0x1a7,0x1a7,-1,1},
2968 {0x1a9,0x1a9,-1,218},
2969 {0x1ac,0x1ac,-1,1},
2970 {0x1ae,0x1ae,-1,218},
2971 {0x1af,0x1af,-1,1},
2972 {0x1b1,0x1b2,1,217},
2973 {0x1b3,0x1b5,2,1},
2974 {0x1b7,0x1b7,-1,219},
2975 {0x1b8,0x1bc,4,1},
2976 {0x1c4,0x1c4,-1,2},
2977 {0x1c5,0x1c5,-1,1},
2978 {0x1c7,0x1c7,-1,2},
2979 {0x1c8,0x1c8,-1,1},
2980 {0x1ca,0x1ca,-1,2},
2981 {0x1cb,0x1db,2,1},
2982 {0x1de,0x1ee,2,1},
2983 {0x1f1,0x1f1,-1,2},
2984 {0x1f2,0x1f4,2,1},
2985 {0x1f6,0x1f6,-1,-97},
2986 {0x1f7,0x1f7,-1,-56},
2987 {0x1f8,0x21e,2,1},
2988 {0x220,0x220,-1,-130},
2989 {0x222,0x232,2,1},
2990 {0x23a,0x23a,-1,10795},
2991 {0x23b,0x23b,-1,1},
2992 {0x23d,0x23d,-1,-163},
2993 {0x23e,0x23e,-1,10792},
2994 {0x241,0x241,-1,1},
2995 {0x243,0x243,-1,-195},
2996 {0x244,0x244,-1,69},
2997 {0x245,0x245,-1,71},
2998 {0x246,0x24e,2,1},
2999 {0x345,0x345,-1,116},
3000 {0x370,0x372,2,1},
3001 {0x376,0x376,-1,1},
3002 {0x37f,0x37f,-1,116},
3003 {0x386,0x386,-1,38},
3004 {0x388,0x38a,1,37},
3005 {0x38c,0x38c,-1,64},
3006 {0x38e,0x38f,1,63},
3007 {0x391,0x3a1,1,32},
3008 {0x3a3,0x3ab,1,32},
3009 {0x3c2,0x3c2,-1,1},
3010 {0x3cf,0x3cf,-1,8},
3011 {0x3d0,0x3d0,-1,-30},
3012 {0x3d1,0x3d1,-1,-25},
3013 {0x3d5,0x3d5,-1,-15},
3014 {0x3d6,0x3d6,-1,-22},
3015 {0x3d8,0x3ee,2,1},
3016 {0x3f0,0x3f0,-1,-54},
3017 {0x3f1,0x3f1,-1,-48},
3018 {0x3f4,0x3f4,-1,-60},
3019 {0x3f5,0x3f5,-1,-64},
3020 {0x3f7,0x3f7,-1,1},
3021 {0x3f9,0x3f9,-1,-7},
3022 {0x3fa,0x3fa,-1,1},
3023 {0x3fd,0x3ff,1,-130},
3024 {0x400,0x40f,1,80},
3025 {0x410,0x42f,1,32},
3026 {0x460,0x480,2,1},
3027 {0x48a,0x4be,2,1},
3028 {0x4c0,0x4c0,-1,15},
3029 {0x4c1,0x4cd,2,1},
3030 {0x4d0,0x52e,2,1},
3031 {0x531,0x556,1,48},
3032 {0x10a0,0x10c5,1,7264},
3033 {0x10c7,0x10cd,6,7264},
3034 {0x13f8,0x13fd,1,-8},
3035 {0x1c80,0x1c80,-1,-6222},
3036 {0x1c81,0x1c81,-1,-6221},
3037 {0x1c82,0x1c82,-1,-6212},
3038 {0x1c83,0x1c84,1,-6210},
3039 {0x1c85,0x1c85,-1,-6211},
3040 {0x1c86,0x1c86,-1,-6204},
3041 {0x1c87,0x1c87,-1,-6180},
3042 {0x1c88,0x1c88,-1,35267},
3043 {0x1c90,0x1cba,1,-3008},
3044 {0x1cbd,0x1cbf,1,-3008},
3045 {0x1e00,0x1e94,2,1},
3046 {0x1e9b,0x1e9b,-1,-58},
3047 {0x1e9e,0x1e9e,-1,-7615},
3048 {0x1ea0,0x1efe,2,1},
3049 {0x1f08,0x1f0f,1,-8},
3050 {0x1f18,0x1f1d,1,-8},
3051 {0x1f28,0x1f2f,1,-8},
3052 {0x1f38,0x1f3f,1,-8},
3053 {0x1f48,0x1f4d,1,-8},
3054 {0x1f59,0x1f5f,2,-8},
3055 {0x1f68,0x1f6f,1,-8},
3056 {0x1f88,0x1f8f,1,-8},
3057 {0x1f98,0x1f9f,1,-8},
3058 {0x1fa8,0x1faf,1,-8},
3059 {0x1fb8,0x1fb9,1,-8},
3060 {0x1fba,0x1fbb,1,-74},
3061 {0x1fbc,0x1fbc,-1,-9},
3062 {0x1fbe,0x1fbe,-1,-7173},
3063 {0x1fc8,0x1fcb,1,-86},
3064 {0x1fcc,0x1fcc,-1,-9},
3065 {0x1fd8,0x1fd9,1,-8},
3066 {0x1fda,0x1fdb,1,-100},
3067 {0x1fe8,0x1fe9,1,-8},
3068 {0x1fea,0x1feb,1,-112},
3069 {0x1fec,0x1fec,-1,-7},
3070 {0x1ff8,0x1ff9,1,-128},
3071 {0x1ffa,0x1ffb,1,-126},
3072 {0x1ffc,0x1ffc,-1,-9},
3073 {0x2126,0x2126,-1,-7517},
3074 {0x212a,0x212a,-1,-8383},
3075 {0x212b,0x212b,-1,-8262},
3076 {0x2132,0x2132,-1,28},
3077 {0x2160,0x216f,1,16},
3078 {0x2183,0x2183,-1,1},
3079 {0x24b6,0x24cf,1,26},
3080 {0x2c00,0x2c2e,1,48},
3081 {0x2c60,0x2c60,-1,1},
3082 {0x2c62,0x2c62,-1,-10743},
3083 {0x2c63,0x2c63,-1,-3814},
3084 {0x2c64,0x2c64,-1,-10727},
3085 {0x2c67,0x2c6b,2,1},
3086 {0x2c6d,0x2c6d,-1,-10780},
3087 {0x2c6e,0x2c6e,-1,-10749},
3088 {0x2c6f,0x2c6f,-1,-10783},
3089 {0x2c70,0x2c70,-1,-10782},
3090 {0x2c72,0x2c75,3,1},
3091 {0x2c7e,0x2c7f,1,-10815},
3092 {0x2c80,0x2ce2,2,1},
3093 {0x2ceb,0x2ced,2,1},
3094 {0x2cf2,0xa640,31054,1},
3095 {0xa642,0xa66c,2,1},
3096 {0xa680,0xa69a,2,1},
3097 {0xa722,0xa72e,2,1},
3098 {0xa732,0xa76e,2,1},
3099 {0xa779,0xa77b,2,1},
3100 {0xa77d,0xa77d,-1,-35332},
3101 {0xa77e,0xa786,2,1},
3102 {0xa78b,0xa78b,-1,1},
3103 {0xa78d,0xa78d,-1,-42280},
3104 {0xa790,0xa792,2,1},
3105 {0xa796,0xa7a8,2,1},
3106 {0xa7aa,0xa7aa,-1,-42308},
3107 {0xa7ab,0xa7ab,-1,-42319},
3108 {0xa7ac,0xa7ac,-1,-42315},
3109 {0xa7ad,0xa7ad,-1,-42305},
3110 {0xa7ae,0xa7ae,-1,-42308},
3111 {0xa7b0,0xa7b0,-1,-42258},
3112 {0xa7b1,0xa7b1,-1,-42282},
3113 {0xa7b2,0xa7b2,-1,-42261},
3114 {0xa7b3,0xa7b3,-1,928},
3115 {0xa7b4,0xa7be,2,1},
3116 {0xa7c2,0xa7c2,-1,1},
3117 {0xa7c4,0xa7c4,-1,-48},
3118 {0xa7c5,0xa7c5,-1,-42307},
3119 {0xa7c6,0xa7c6,-1,-35384},
3120 {0xa7c7,0xa7c9,2,1},
3121 {0xa7f5,0xa7f5,-1,1},
3122 {0xab70,0xabbf,1,-38864},
3123 {0xff21,0xff3a,1,32},
3124 {0x10400,0x10427,1,40},
3125 {0x104b0,0x104d3,1,40},
3126 {0x10c80,0x10cb2,1,64},
3127 {0x118a0,0x118bf,1,32},
3128 {0x16e40,0x16e5f,1,32},
3129 {0x1e900,0x1e921,1,34}
3130 };
3131
3132 /*
3133 * Generic conversion function for case operations.
3134 * Return the converted equivalent of "a", which is a UCS-4 character. Use
3135 * the given conversion "table". Uses binary search on "table".
3136 */
3137 static int
utf_convert(int a,convertStruct table[],int tableSize)3138 utf_convert(
3139 int a,
3140 convertStruct table[],
3141 int tableSize)
3142 {
3143 int start, mid, end; // indices into table
3144 int entries = tableSize / sizeof(convertStruct);
3145
3146 start = 0;
3147 end = entries;
3148 while (start < end)
3149 {
3150 // need to search further
3151 mid = (end + start) / 2;
3152 if (table[mid].rangeEnd < a)
3153 start = mid + 1;
3154 else
3155 end = mid;
3156 }
3157 if (start < entries
3158 && table[start].rangeStart <= a
3159 && a <= table[start].rangeEnd
3160 && (a - table[start].rangeStart) % table[start].step == 0)
3161 return (a + table[start].offset);
3162 else
3163 return a;
3164 }
3165
3166 /*
3167 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses
3168 * simple case folding.
3169 */
3170 int
utf_fold(int a)3171 utf_fold(int a)
3172 {
3173 if (a < 0x80)
3174 // be fast for ASCII
3175 return a >= 0x41 && a <= 0x5a ? a + 32 : a;
3176 return utf_convert(a, foldCase, (int)sizeof(foldCase));
3177 }
3178
3179 static convertStruct toLower[] =
3180 {
3181 {0x41,0x5a,1,32},
3182 {0xc0,0xd6,1,32},
3183 {0xd8,0xde,1,32},
3184 {0x100,0x12e,2,1},
3185 {0x130,0x130,-1,-199},
3186 {0x132,0x136,2,1},
3187 {0x139,0x147,2,1},
3188 {0x14a,0x176,2,1},
3189 {0x178,0x178,-1,-121},
3190 {0x179,0x17d,2,1},
3191 {0x181,0x181,-1,210},
3192 {0x182,0x184,2,1},
3193 {0x186,0x186,-1,206},
3194 {0x187,0x187,-1,1},
3195 {0x189,0x18a,1,205},
3196 {0x18b,0x18b,-1,1},
3197 {0x18e,0x18e,-1,79},
3198 {0x18f,0x18f,-1,202},
3199 {0x190,0x190,-1,203},
3200 {0x191,0x191,-1,1},
3201 {0x193,0x193,-1,205},
3202 {0x194,0x194,-1,207},
3203 {0x196,0x196,-1,211},
3204 {0x197,0x197,-1,209},
3205 {0x198,0x198,-1,1},
3206 {0x19c,0x19c,-1,211},
3207 {0x19d,0x19d,-1,213},
3208 {0x19f,0x19f,-1,214},
3209 {0x1a0,0x1a4,2,1},
3210 {0x1a6,0x1a6,-1,218},
3211 {0x1a7,0x1a7,-1,1},
3212 {0x1a9,0x1a9,-1,218},
3213 {0x1ac,0x1ac,-1,1},
3214 {0x1ae,0x1ae,-1,218},
3215 {0x1af,0x1af,-1,1},
3216 {0x1b1,0x1b2,1,217},
3217 {0x1b3,0x1b5,2,1},
3218 {0x1b7,0x1b7,-1,219},
3219 {0x1b8,0x1bc,4,1},
3220 {0x1c4,0x1c4,-1,2},
3221 {0x1c5,0x1c5,-1,1},
3222 {0x1c7,0x1c7,-1,2},
3223 {0x1c8,0x1c8,-1,1},
3224 {0x1ca,0x1ca,-1,2},
3225 {0x1cb,0x1db,2,1},
3226 {0x1de,0x1ee,2,1},
3227 {0x1f1,0x1f1,-1,2},
3228 {0x1f2,0x1f4,2,1},
3229 {0x1f6,0x1f6,-1,-97},
3230 {0x1f7,0x1f7,-1,-56},
3231 {0x1f8,0x21e,2,1},
3232 {0x220,0x220,-1,-130},
3233 {0x222,0x232,2,1},
3234 {0x23a,0x23a,-1,10795},
3235 {0x23b,0x23b,-1,1},
3236 {0x23d,0x23d,-1,-163},
3237 {0x23e,0x23e,-1,10792},
3238 {0x241,0x241,-1,1},
3239 {0x243,0x243,-1,-195},
3240 {0x244,0x244,-1,69},
3241 {0x245,0x245,-1,71},
3242 {0x246,0x24e,2,1},
3243 {0x370,0x372,2,1},
3244 {0x376,0x376,-1,1},
3245 {0x37f,0x37f,-1,116},
3246 {0x386,0x386,-1,38},
3247 {0x388,0x38a,1,37},
3248 {0x38c,0x38c,-1,64},
3249 {0x38e,0x38f,1,63},
3250 {0x391,0x3a1,1,32},
3251 {0x3a3,0x3ab,1,32},
3252 {0x3cf,0x3cf,-1,8},
3253 {0x3d8,0x3ee,2,1},
3254 {0x3f4,0x3f4,-1,-60},
3255 {0x3f7,0x3f7,-1,1},
3256 {0x3f9,0x3f9,-1,-7},
3257 {0x3fa,0x3fa,-1,1},
3258 {0x3fd,0x3ff,1,-130},
3259 {0x400,0x40f,1,80},
3260 {0x410,0x42f,1,32},
3261 {0x460,0x480,2,1},
3262 {0x48a,0x4be,2,1},
3263 {0x4c0,0x4c0,-1,15},
3264 {0x4c1,0x4cd,2,1},
3265 {0x4d0,0x52e,2,1},
3266 {0x531,0x556,1,48},
3267 {0x10a0,0x10c5,1,7264},
3268 {0x10c7,0x10cd,6,7264},
3269 {0x13a0,0x13ef,1,38864},
3270 {0x13f0,0x13f5,1,8},
3271 {0x1c90,0x1cba,1,-3008},
3272 {0x1cbd,0x1cbf,1,-3008},
3273 {0x1e00,0x1e94,2,1},
3274 {0x1e9e,0x1e9e,-1,-7615},
3275 {0x1ea0,0x1efe,2,1},
3276 {0x1f08,0x1f0f,1,-8},
3277 {0x1f18,0x1f1d,1,-8},
3278 {0x1f28,0x1f2f,1,-8},
3279 {0x1f38,0x1f3f,1,-8},
3280 {0x1f48,0x1f4d,1,-8},
3281 {0x1f59,0x1f5f,2,-8},
3282 {0x1f68,0x1f6f,1,-8},
3283 {0x1f88,0x1f8f,1,-8},
3284 {0x1f98,0x1f9f,1,-8},
3285 {0x1fa8,0x1faf,1,-8},
3286 {0x1fb8,0x1fb9,1,-8},
3287 {0x1fba,0x1fbb,1,-74},
3288 {0x1fbc,0x1fbc,-1,-9},
3289 {0x1fc8,0x1fcb,1,-86},
3290 {0x1fcc,0x1fcc,-1,-9},
3291 {0x1fd8,0x1fd9,1,-8},
3292 {0x1fda,0x1fdb,1,-100},
3293 {0x1fe8,0x1fe9,1,-8},
3294 {0x1fea,0x1feb,1,-112},
3295 {0x1fec,0x1fec,-1,-7},
3296 {0x1ff8,0x1ff9,1,-128},
3297 {0x1ffa,0x1ffb,1,-126},
3298 {0x1ffc,0x1ffc,-1,-9},
3299 {0x2126,0x2126,-1,-7517},
3300 {0x212a,0x212a,-1,-8383},
3301 {0x212b,0x212b,-1,-8262},
3302 {0x2132,0x2132,-1,28},
3303 {0x2160,0x216f,1,16},
3304 {0x2183,0x2183,-1,1},
3305 {0x24b6,0x24cf,1,26},
3306 {0x2c00,0x2c2e,1,48},
3307 {0x2c60,0x2c60,-1,1},
3308 {0x2c62,0x2c62,-1,-10743},
3309 {0x2c63,0x2c63,-1,-3814},
3310 {0x2c64,0x2c64,-1,-10727},
3311 {0x2c67,0x2c6b,2,1},
3312 {0x2c6d,0x2c6d,-1,-10780},
3313 {0x2c6e,0x2c6e,-1,-10749},
3314 {0x2c6f,0x2c6f,-1,-10783},
3315 {0x2c70,0x2c70,-1,-10782},
3316 {0x2c72,0x2c75,3,1},
3317 {0x2c7e,0x2c7f,1,-10815},
3318 {0x2c80,0x2ce2,2,1},
3319 {0x2ceb,0x2ced,2,1},
3320 {0x2cf2,0xa640,31054,1},
3321 {0xa642,0xa66c,2,1},
3322 {0xa680,0xa69a,2,1},
3323 {0xa722,0xa72e,2,1},
3324 {0xa732,0xa76e,2,1},
3325 {0xa779,0xa77b,2,1},
3326 {0xa77d,0xa77d,-1,-35332},
3327 {0xa77e,0xa786,2,1},
3328 {0xa78b,0xa78b,-1,1},
3329 {0xa78d,0xa78d,-1,-42280},
3330 {0xa790,0xa792,2,1},
3331 {0xa796,0xa7a8,2,1},
3332 {0xa7aa,0xa7aa,-1,-42308},
3333 {0xa7ab,0xa7ab,-1,-42319},
3334 {0xa7ac,0xa7ac,-1,-42315},
3335 {0xa7ad,0xa7ad,-1,-42305},
3336 {0xa7ae,0xa7ae,-1,-42308},
3337 {0xa7b0,0xa7b0,-1,-42258},
3338 {0xa7b1,0xa7b1,-1,-42282},
3339 {0xa7b2,0xa7b2,-1,-42261},
3340 {0xa7b3,0xa7b3,-1,928},
3341 {0xa7b4,0xa7be,2,1},
3342 {0xa7c2,0xa7c2,-1,1},
3343 {0xa7c4,0xa7c4,-1,-48},
3344 {0xa7c5,0xa7c5,-1,-42307},
3345 {0xa7c6,0xa7c6,-1,-35384},
3346 {0xa7c7,0xa7c9,2,1},
3347 {0xa7f5,0xa7f5,-1,1},
3348 {0xff21,0xff3a,1,32},
3349 {0x10400,0x10427,1,40},
3350 {0x104b0,0x104d3,1,40},
3351 {0x10c80,0x10cb2,1,64},
3352 {0x118a0,0x118bf,1,32},
3353 {0x16e40,0x16e5f,1,32},
3354 {0x1e900,0x1e921,1,34}
3355 };
3356
3357 static convertStruct toUpper[] =
3358 {
3359 {0x61,0x7a,1,-32},
3360 {0xb5,0xb5,-1,743},
3361 {0xe0,0xf6,1,-32},
3362 {0xf8,0xfe,1,-32},
3363 {0xff,0xff,-1,121},
3364 {0x101,0x12f,2,-1},
3365 {0x131,0x131,-1,-232},
3366 {0x133,0x137,2,-1},
3367 {0x13a,0x148,2,-1},
3368 {0x14b,0x177,2,-1},
3369 {0x17a,0x17e,2,-1},
3370 {0x17f,0x17f,-1,-300},
3371 {0x180,0x180,-1,195},
3372 {0x183,0x185,2,-1},
3373 {0x188,0x18c,4,-1},
3374 {0x192,0x192,-1,-1},
3375 {0x195,0x195,-1,97},
3376 {0x199,0x199,-1,-1},
3377 {0x19a,0x19a,-1,163},
3378 {0x19e,0x19e,-1,130},
3379 {0x1a1,0x1a5,2,-1},
3380 {0x1a8,0x1ad,5,-1},
3381 {0x1b0,0x1b4,4,-1},
3382 {0x1b6,0x1b9,3,-1},
3383 {0x1bd,0x1bd,-1,-1},
3384 {0x1bf,0x1bf,-1,56},
3385 {0x1c5,0x1c5,-1,-1},
3386 {0x1c6,0x1c6,-1,-2},
3387 {0x1c8,0x1c8,-1,-1},
3388 {0x1c9,0x1c9,-1,-2},
3389 {0x1cb,0x1cb,-1,-1},
3390 {0x1cc,0x1cc,-1,-2},
3391 {0x1ce,0x1dc,2,-1},
3392 {0x1dd,0x1dd,-1,-79},
3393 {0x1df,0x1ef,2,-1},
3394 {0x1f2,0x1f2,-1,-1},
3395 {0x1f3,0x1f3,-1,-2},
3396 {0x1f5,0x1f9,4,-1},
3397 {0x1fb,0x21f,2,-1},
3398 {0x223,0x233,2,-1},
3399 {0x23c,0x23c,-1,-1},
3400 {0x23f,0x240,1,10815},
3401 {0x242,0x247,5,-1},
3402 {0x249,0x24f,2,-1},
3403 {0x250,0x250,-1,10783},
3404 {0x251,0x251,-1,10780},
3405 {0x252,0x252,-1,10782},
3406 {0x253,0x253,-1,-210},
3407 {0x254,0x254,-1,-206},
3408 {0x256,0x257,1,-205},
3409 {0x259,0x259,-1,-202},
3410 {0x25b,0x25b,-1,-203},
3411 {0x25c,0x25c,-1,42319},
3412 {0x260,0x260,-1,-205},
3413 {0x261,0x261,-1,42315},
3414 {0x263,0x263,-1,-207},
3415 {0x265,0x265,-1,42280},
3416 {0x266,0x266,-1,42308},
3417 {0x268,0x268,-1,-209},
3418 {0x269,0x269,-1,-211},
3419 {0x26a,0x26a,-1,42308},
3420 {0x26b,0x26b,-1,10743},
3421 {0x26c,0x26c,-1,42305},
3422 {0x26f,0x26f,-1,-211},
3423 {0x271,0x271,-1,10749},
3424 {0x272,0x272,-1,-213},
3425 {0x275,0x275,-1,-214},
3426 {0x27d,0x27d,-1,10727},
3427 {0x280,0x280,-1,-218},
3428 {0x282,0x282,-1,42307},
3429 {0x283,0x283,-1,-218},
3430 {0x287,0x287,-1,42282},
3431 {0x288,0x288,-1,-218},
3432 {0x289,0x289,-1,-69},
3433 {0x28a,0x28b,1,-217},
3434 {0x28c,0x28c,-1,-71},
3435 {0x292,0x292,-1,-219},
3436 {0x29d,0x29d,-1,42261},
3437 {0x29e,0x29e,-1,42258},
3438 {0x345,0x345,-1,84},
3439 {0x371,0x373,2,-1},
3440 {0x377,0x377,-1,-1},
3441 {0x37b,0x37d,1,130},
3442 {0x3ac,0x3ac,-1,-38},
3443 {0x3ad,0x3af,1,-37},
3444 {0x3b1,0x3c1,1,-32},
3445 {0x3c2,0x3c2,-1,-31},
3446 {0x3c3,0x3cb,1,-32},
3447 {0x3cc,0x3cc,-1,-64},
3448 {0x3cd,0x3ce,1,-63},
3449 {0x3d0,0x3d0,-1,-62},
3450 {0x3d1,0x3d1,-1,-57},
3451 {0x3d5,0x3d5,-1,-47},
3452 {0x3d6,0x3d6,-1,-54},
3453 {0x3d7,0x3d7,-1,-8},
3454 {0x3d9,0x3ef,2,-1},
3455 {0x3f0,0x3f0,-1,-86},
3456 {0x3f1,0x3f1,-1,-80},
3457 {0x3f2,0x3f2,-1,7},
3458 {0x3f3,0x3f3,-1,-116},
3459 {0x3f5,0x3f5,-1,-96},
3460 {0x3f8,0x3fb,3,-1},
3461 {0x430,0x44f,1,-32},
3462 {0x450,0x45f,1,-80},
3463 {0x461,0x481,2,-1},
3464 {0x48b,0x4bf,2,-1},
3465 {0x4c2,0x4ce,2,-1},
3466 {0x4cf,0x4cf,-1,-15},
3467 {0x4d1,0x52f,2,-1},
3468 {0x561,0x586,1,-48},
3469 {0x10d0,0x10fa,1,3008},
3470 {0x10fd,0x10ff,1,3008},
3471 {0x13f8,0x13fd,1,-8},
3472 {0x1c80,0x1c80,-1,-6254},
3473 {0x1c81,0x1c81,-1,-6253},
3474 {0x1c82,0x1c82,-1,-6244},
3475 {0x1c83,0x1c84,1,-6242},
3476 {0x1c85,0x1c85,-1,-6243},
3477 {0x1c86,0x1c86,-1,-6236},
3478 {0x1c87,0x1c87,-1,-6181},
3479 {0x1c88,0x1c88,-1,35266},
3480 {0x1d79,0x1d79,-1,35332},
3481 {0x1d7d,0x1d7d,-1,3814},
3482 {0x1d8e,0x1d8e,-1,35384},
3483 {0x1e01,0x1e95,2,-1},
3484 {0x1e9b,0x1e9b,-1,-59},
3485 {0x1ea1,0x1eff,2,-1},
3486 {0x1f00,0x1f07,1,8},
3487 {0x1f10,0x1f15,1,8},
3488 {0x1f20,0x1f27,1,8},
3489 {0x1f30,0x1f37,1,8},
3490 {0x1f40,0x1f45,1,8},
3491 {0x1f51,0x1f57,2,8},
3492 {0x1f60,0x1f67,1,8},
3493 {0x1f70,0x1f71,1,74},
3494 {0x1f72,0x1f75,1,86},
3495 {0x1f76,0x1f77,1,100},
3496 {0x1f78,0x1f79,1,128},
3497 {0x1f7a,0x1f7b,1,112},
3498 {0x1f7c,0x1f7d,1,126},
3499 {0x1f80,0x1f87,1,8},
3500 {0x1f90,0x1f97,1,8},
3501 {0x1fa0,0x1fa7,1,8},
3502 {0x1fb0,0x1fb1,1,8},
3503 {0x1fb3,0x1fb3,-1,9},
3504 {0x1fbe,0x1fbe,-1,-7205},
3505 {0x1fc3,0x1fc3,-1,9},
3506 {0x1fd0,0x1fd1,1,8},
3507 {0x1fe0,0x1fe1,1,8},
3508 {0x1fe5,0x1fe5,-1,7},
3509 {0x1ff3,0x1ff3,-1,9},
3510 {0x214e,0x214e,-1,-28},
3511 {0x2170,0x217f,1,-16},
3512 {0x2184,0x2184,-1,-1},
3513 {0x24d0,0x24e9,1,-26},
3514 {0x2c30,0x2c5e,1,-48},
3515 {0x2c61,0x2c61,-1,-1},
3516 {0x2c65,0x2c65,-1,-10795},
3517 {0x2c66,0x2c66,-1,-10792},
3518 {0x2c68,0x2c6c,2,-1},
3519 {0x2c73,0x2c76,3,-1},
3520 {0x2c81,0x2ce3,2,-1},
3521 {0x2cec,0x2cee,2,-1},
3522 {0x2cf3,0x2cf3,-1,-1},
3523 {0x2d00,0x2d25,1,-7264},
3524 {0x2d27,0x2d2d,6,-7264},
3525 {0xa641,0xa66d,2,-1},
3526 {0xa681,0xa69b,2,-1},
3527 {0xa723,0xa72f,2,-1},
3528 {0xa733,0xa76f,2,-1},
3529 {0xa77a,0xa77c,2,-1},
3530 {0xa77f,0xa787,2,-1},
3531 {0xa78c,0xa791,5,-1},
3532 {0xa793,0xa793,-1,-1},
3533 {0xa794,0xa794,-1,48},
3534 {0xa797,0xa7a9,2,-1},
3535 {0xa7b5,0xa7bf,2,-1},
3536 {0xa7c3,0xa7c8,5,-1},
3537 {0xa7ca,0xa7f6,44,-1},
3538 {0xab53,0xab53,-1,-928},
3539 {0xab70,0xabbf,1,-38864},
3540 {0xff41,0xff5a,1,-32},
3541 {0x10428,0x1044f,1,-40},
3542 {0x104d8,0x104fb,1,-40},
3543 {0x10cc0,0x10cf2,1,-64},
3544 {0x118c0,0x118df,1,-32},
3545 {0x16e60,0x16e7f,1,-32},
3546 {0x1e922,0x1e943,1,-34}
3547 };
3548
3549 /*
3550 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use
3551 * simple case folding.
3552 */
3553 int
utf_toupper(int a)3554 utf_toupper(int a)
3555 {
3556 // If 'casemap' contains "keepascii" use ASCII style toupper().
3557 if (a < 128 && (cmp_flags & CMP_KEEPASCII))
3558 return TOUPPER_ASC(a);
3559
3560 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__)
3561 // If towupper() is available and handles Unicode, use it.
3562 if (!(cmp_flags & CMP_INTERNAL))
3563 return towupper(a);
3564 #endif
3565
3566 // For characters below 128 use locale sensitive toupper().
3567 if (a < 128)
3568 return TOUPPER_LOC(a);
3569
3570 // For any other characters use the above mapping table.
3571 return utf_convert(a, toUpper, (int)sizeof(toUpper));
3572 }
3573
3574 int
utf_islower(int a)3575 utf_islower(int a)
3576 {
3577 // German sharp s is lower case but has no upper case equivalent.
3578 return (utf_toupper(a) != a) || a == 0xdf;
3579 }
3580
3581 /*
3582 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use
3583 * simple case folding.
3584 */
3585 int
utf_tolower(int a)3586 utf_tolower(int a)
3587 {
3588 // If 'casemap' contains "keepascii" use ASCII style tolower().
3589 if (a < 128 && (cmp_flags & CMP_KEEPASCII))
3590 return TOLOWER_ASC(a);
3591
3592 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__)
3593 // If towlower() is available and handles Unicode, use it.
3594 if (!(cmp_flags & CMP_INTERNAL))
3595 return towlower(a);
3596 #endif
3597
3598 // For characters below 128 use locale sensitive tolower().
3599 if (a < 128)
3600 return TOLOWER_LOC(a);
3601
3602 // For any other characters use the above mapping table.
3603 return utf_convert(a, toLower, (int)sizeof(toLower));
3604 }
3605
3606 int
utf_isupper(int a)3607 utf_isupper(int a)
3608 {
3609 return (utf_tolower(a) != a);
3610 }
3611
3612 static int
utf_strnicmp(char_u * s1,char_u * s2,size_t n1,size_t n2)3613 utf_strnicmp(
3614 char_u *s1,
3615 char_u *s2,
3616 size_t n1,
3617 size_t n2)
3618 {
3619 int c1, c2, cdiff;
3620 char_u buffer[6];
3621
3622 for (;;)
3623 {
3624 c1 = utf_safe_read_char_adv(&s1, &n1);
3625 c2 = utf_safe_read_char_adv(&s2, &n2);
3626
3627 if (c1 <= 0 || c2 <= 0)
3628 break;
3629
3630 if (c1 == c2)
3631 continue;
3632
3633 cdiff = utf_fold(c1) - utf_fold(c2);
3634 if (cdiff != 0)
3635 return cdiff;
3636 }
3637
3638 // some string ended or has an incomplete/illegal character sequence
3639
3640 if (c1 == 0 || c2 == 0)
3641 {
3642 // some string ended. shorter string is smaller
3643 if (c1 == 0 && c2 == 0)
3644 return 0;
3645 return c1 == 0 ? -1 : 1;
3646 }
3647
3648 // Continue with bytewise comparison to produce some result that
3649 // would make comparison operations involving this function transitive.
3650 //
3651 // If only one string had an error, comparison should be made with
3652 // folded version of the other string. In this case it is enough
3653 // to fold just one character to determine the result of comparison.
3654
3655 if (c1 != -1 && c2 == -1)
3656 {
3657 n1 = utf_char2bytes(utf_fold(c1), buffer);
3658 s1 = buffer;
3659 }
3660 else if (c2 != -1 && c1 == -1)
3661 {
3662 n2 = utf_char2bytes(utf_fold(c2), buffer);
3663 s2 = buffer;
3664 }
3665
3666 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL)
3667 {
3668 cdiff = (int)(*s1) - (int)(*s2);
3669 if (cdiff != 0)
3670 return cdiff;
3671
3672 s1++;
3673 s2++;
3674 n1--;
3675 n2--;
3676 }
3677
3678 if (n1 > 0 && *s1 == NUL)
3679 n1 = 0;
3680 if (n2 > 0 && *s2 == NUL)
3681 n2 = 0;
3682
3683 if (n1 == 0 && n2 == 0)
3684 return 0;
3685 return n1 == 0 ? -1 : 1;
3686 }
3687
3688 /*
3689 * Version of strnicmp() that handles multi-byte characters.
3690 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can
3691 * probably use strnicmp(), because there are no ASCII characters in the
3692 * second byte.
3693 * Returns zero if s1 and s2 are equal (ignoring case), the difference between
3694 * two characters otherwise.
3695 */
3696 int
mb_strnicmp(char_u * s1,char_u * s2,size_t nn)3697 mb_strnicmp(char_u *s1, char_u *s2, size_t nn)
3698 {
3699 int i, l;
3700 int cdiff;
3701 int n = (int)nn;
3702
3703 if (enc_utf8)
3704 {
3705 return utf_strnicmp(s1, s2, nn, nn);
3706 }
3707 else
3708 {
3709 for (i = 0; i < n; i += l)
3710 {
3711 if (s1[i] == NUL && s2[i] == NUL) // both strings end
3712 return 0;
3713
3714 l = (*mb_ptr2len)(s1 + i);
3715 if (l <= 1)
3716 {
3717 // Single byte: first check normally, then with ignore case.
3718 if (s1[i] != s2[i])
3719 {
3720 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]);
3721 if (cdiff != 0)
3722 return cdiff;
3723 }
3724 }
3725 else
3726 {
3727 // For non-Unicode multi-byte don't ignore case.
3728 if (l > n - i)
3729 l = n - i;
3730 cdiff = STRNCMP(s1 + i, s2 + i, l);
3731 if (cdiff != 0)
3732 return cdiff;
3733 }
3734 }
3735 }
3736 return 0;
3737 }
3738
3739 /*
3740 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what
3741 * 'encoding' has been set to.
3742 */
3743 void
show_utf8(void)3744 show_utf8(void)
3745 {
3746 int len;
3747 int rlen = 0;
3748 char_u *line;
3749 int clen;
3750 int i;
3751
3752 // Get the byte length of the char under the cursor, including composing
3753 // characters.
3754 line = ml_get_cursor();
3755 len = utfc_ptr2len(line);
3756 if (len == 0)
3757 {
3758 msg("NUL");
3759 return;
3760 }
3761
3762 clen = 0;
3763 for (i = 0; i < len; ++i)
3764 {
3765 if (clen == 0)
3766 {
3767 // start of (composing) character, get its length
3768 if (i > 0)
3769 {
3770 STRCPY(IObuff + rlen, "+ ");
3771 rlen += 2;
3772 }
3773 clen = utf_ptr2len(line + i);
3774 }
3775 sprintf((char *)IObuff + rlen, "%02x ",
3776 (line[i] == NL) ? NUL : line[i]); // NUL is stored as NL
3777 --clen;
3778 rlen += (int)STRLEN(IObuff + rlen);
3779 if (rlen > IOSIZE - 20)
3780 break;
3781 }
3782
3783 msg((char *)IObuff);
3784 }
3785
3786 /*
3787 * mb_head_off() function pointer.
3788 * Return offset from "p" to the first byte of the character it points into.
3789 * If "p" points to the NUL at the end of the string return 0.
3790 * Returns 0 when already at the first byte of a character.
3791 */
3792 int
latin_head_off(char_u * base UNUSED,char_u * p UNUSED)3793 latin_head_off(char_u *base UNUSED, char_u *p UNUSED)
3794 {
3795 return 0;
3796 }
3797
3798 static int
dbcs_head_off(char_u * base,char_u * p)3799 dbcs_head_off(char_u *base, char_u *p)
3800 {
3801 char_u *q;
3802
3803 // It can't be a trailing byte when not using DBCS, at the start of the
3804 // string or the previous byte can't start a double-byte.
3805 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL)
3806 return 0;
3807
3808 // This is slow: need to start at the base and go forward until the
3809 // byte we are looking for. Return 1 when we went past it, 0 otherwise.
3810 q = base;
3811 while (q < p)
3812 q += dbcs_ptr2len(q);
3813 return (q == p) ? 0 : 1;
3814 }
3815
3816 /*
3817 * Special version of dbcs_head_off() that works for ScreenLines[], where
3818 * single-width DBCS_JPNU characters are stored separately.
3819 */
3820 int
dbcs_screen_head_off(char_u * base,char_u * p)3821 dbcs_screen_head_off(char_u *base, char_u *p)
3822 {
3823 char_u *q;
3824
3825 // It can't be a trailing byte when not using DBCS, at the start of the
3826 // string or the previous byte can't start a double-byte.
3827 // For euc-jp an 0x8e byte in the previous cell always means we have a
3828 // lead byte in the current cell.
3829 if (p <= base
3830 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e)
3831 || MB_BYTE2LEN(p[-1]) == 1
3832 || *p == NUL)
3833 return 0;
3834
3835 // This is slow: need to start at the base and go forward until the
3836 // byte we are looking for. Return 1 when we went past it, 0 otherwise.
3837 // For DBCS_JPNU look out for 0x8e, which means the second byte is not
3838 // stored as the next byte.
3839 q = base;
3840 while (q < p)
3841 {
3842 if (enc_dbcs == DBCS_JPNU && *q == 0x8e)
3843 ++q;
3844 else
3845 q += dbcs_ptr2len(q);
3846 }
3847 return (q == p) ? 0 : 1;
3848 }
3849
3850 int
utf_head_off(char_u * base,char_u * p)3851 utf_head_off(char_u *base, char_u *p)
3852 {
3853 char_u *q;
3854 char_u *s;
3855 int c;
3856 int len;
3857 #ifdef FEAT_ARABIC
3858 char_u *j;
3859 #endif
3860
3861 if (*p < 0x80) // be quick for ASCII
3862 return 0;
3863
3864 // Skip backwards over trailing bytes: 10xx.xxxx
3865 // Skip backwards again if on a composing char.
3866 for (q = p; ; --q)
3867 {
3868 // Move s to the last byte of this char.
3869 for (s = q; (s[1] & 0xc0) == 0x80; ++s)
3870 ;
3871 // Move q to the first byte of this char.
3872 while (q > base && (*q & 0xc0) == 0x80)
3873 --q;
3874 // Check for illegal sequence. Do allow an illegal byte after where we
3875 // started.
3876 len = utf8len_tab[*q];
3877 if (len != (int)(s - q + 1) && len != (int)(p - q + 1))
3878 return 0;
3879
3880 if (q <= base)
3881 break;
3882
3883 c = utf_ptr2char(q);
3884 if (utf_iscomposing(c))
3885 continue;
3886
3887 #ifdef FEAT_ARABIC
3888 if (arabic_maycombine(c))
3889 {
3890 // Advance to get a sneak-peak at the next char
3891 j = q;
3892 --j;
3893 // Move j to the first byte of this char.
3894 while (j > base && (*j & 0xc0) == 0x80)
3895 --j;
3896 if (arabic_combine(utf_ptr2char(j), c))
3897 continue;
3898 }
3899 #endif
3900 break;
3901 }
3902
3903 return (int)(p - q);
3904 }
3905
3906 /*
3907 * Whether space is NOT allowed before/after 'c'.
3908 */
3909 int
utf_eat_space(int cc)3910 utf_eat_space(int cc)
3911 {
3912 return ((cc >= 0x2000 && cc <= 0x206F) // General punctuations
3913 || (cc >= 0x2e00 && cc <= 0x2e7f) // Supplemental punctuations
3914 || (cc >= 0x3000 && cc <= 0x303f) // CJK symbols and punctuations
3915 || (cc >= 0xff01 && cc <= 0xff0f) // Full width ASCII punctuations
3916 || (cc >= 0xff1a && cc <= 0xff20) // ..
3917 || (cc >= 0xff3b && cc <= 0xff40) // ..
3918 || (cc >= 0xff5b && cc <= 0xff65)); // ..
3919 }
3920
3921 /*
3922 * Whether line break is allowed before "cc".
3923 */
3924 int
utf_allow_break_before(int cc)3925 utf_allow_break_before(int cc)
3926 {
3927 static const int BOL_prohibition_punct[] =
3928 {
3929 '!',
3930 '%',
3931 ')',
3932 ',',
3933 ':',
3934 ';',
3935 '>',
3936 '?',
3937 ']',
3938 '}',
3939 0x2019, // ’ right single quotation mark
3940 0x201d, // ” right double quotation mark
3941 0x2020, // † dagger
3942 0x2021, // ‡ double dagger
3943 0x2026, // … horizontal ellipsis
3944 0x2030, // ‰ per mille sign
3945 0x2031, // ‱ per then thousand sign
3946 0x203c, // ‼ double exclamation mark
3947 0x2047, // ⁇ double question mark
3948 0x2048, // ⁈ question exclamation mark
3949 0x2049, // ⁉ exclamation question mark
3950 0x2103, // ℃ degree celsius
3951 0x2109, // ℉ degree fahrenheit
3952 0x3001, // 、 ideographic comma
3953 0x3002, // 。 ideographic full stop
3954 0x3009, // 〉 right angle bracket
3955 0x300b, // 》 right double angle bracket
3956 0x300d, // 」 right corner bracket
3957 0x300f, // 』 right white corner bracket
3958 0x3011, // 】 right black lenticular bracket
3959 0x3015, // 〕 right tortoise shell bracket
3960 0x3017, // 〗 right white lenticular bracket
3961 0x3019, // 〙 right white tortoise shell bracket
3962 0x301b, // 〛 right white square bracket
3963 0xff01, // ！ fullwidth exclamation mark
3964 0xff09, // ） fullwidth right parenthesis
3965 0xff0c, // ， fullwidth comma
3966 0xff0e, // ． fullwidth full stop
3967 0xff1a, // ： fullwidth colon
3968 0xff1b, // ； fullwidth semicolon
3969 0xff1f, // ？ fullwidth question mark
3970 0xff3d, // ］ fullwidth right square bracket
3971 0xff5d, // ｝ fullwidth right curly bracket
3972 };
3973
3974 int first = 0;
3975 int last = ARRAY_LENGTH(BOL_prohibition_punct) - 1;
3976 int mid = 0;
3977
3978 while (first < last)
3979 {
3980 mid = (first + last)/2;
3981
3982 if (cc == BOL_prohibition_punct[mid])
3983 return FALSE;
3984 else if (cc > BOL_prohibition_punct[mid])
3985 first = mid + 1;
3986 else
3987 last = mid - 1;
3988 }
3989
3990 return cc != BOL_prohibition_punct[first];
3991 }
3992
3993 /*
3994 * Whether line break is allowed after "cc".
3995 */
3996 static int
utf_allow_break_after(int cc)3997 utf_allow_break_after(int cc)
3998 {
3999 static const int EOL_prohibition_punct[] =
4000 {
4001 '(',
4002 '<',
4003 '[',
4004 '`',
4005 '{',
4006 //0x2014, // — em dash
4007 0x2018, // ‘ left single quotation mark
4008 0x201c, // “ left double quotation mark
4009 //0x2053, // ～ swung dash
4010 0x3008, // 〈 left angle bracket
4011 0x300a, // 《 left double angle bracket
4012 0x300c, // 「 left corner bracket
4013 0x300e, // 『 left white corner bracket
4014 0x3010, // 【 left black lenticular bracket
4015 0x3014, // 〔 left tortoise shell bracket
4016 0x3016, // 〖 left white lenticular bracket
4017 0x3018, // 〘 left white tortoise shell bracket
4018 0x301a, // 〚 left white square bracket
4019 0xff08, // （ fullwidth left parenthesis
4020 0xff3b, // ［ fullwidth left square bracket
4021 0xff5b, // ｛ fullwidth left curly bracket
4022 };
4023
4024 int first = 0;
4025 int last = ARRAY_LENGTH(EOL_prohibition_punct) - 1;
4026 int mid = 0;
4027
4028 while (first < last)
4029 {
4030 mid = (first + last)/2;
4031
4032 if (cc == EOL_prohibition_punct[mid])
4033 return FALSE;
4034 else if (cc > EOL_prohibition_punct[mid])
4035 first = mid + 1;
4036 else
4037 last = mid - 1;
4038 }
4039
4040 return cc != EOL_prohibition_punct[first];
4041 }
4042
4043 /*
4044 * Whether line break is allowed between "cc" and "ncc".
4045 */
4046 int
utf_allow_break(int cc,int ncc)4047 utf_allow_break(int cc, int ncc)
4048 {
4049 // don't break between two-letter punctuations
4050 if (cc == ncc
4051 && (cc == 0x2014 // em dash
4052 || cc == 0x2026)) // horizontal ellipsis
4053 return FALSE;
4054
4055 return utf_allow_break_after(cc) && utf_allow_break_before(ncc);
4056 }
4057
4058 /*
4059 * Copy a character from "*fp" to "*tp" and advance the pointers.
4060 */
4061 void
mb_copy_char(char_u ** fp,char_u ** tp)4062 mb_copy_char(char_u **fp, char_u **tp)
4063 {
4064 int l = (*mb_ptr2len)(*fp);
4065
4066 mch_memmove(*tp, *fp, (size_t)l);
4067 *tp += l;
4068 *fp += l;
4069 }
4070
4071 /*
4072 * Return the offset from "p" to the first byte of a character. When "p" is
4073 * at the start of a character 0 is returned, otherwise the offset to the next
4074 * character. Can start anywhere in a stream of bytes.
4075 */
4076 int
mb_off_next(char_u * base,char_u * p)4077 mb_off_next(char_u *base, char_u *p)
4078 {
4079 int i;
4080 int j;
4081
4082 if (enc_utf8)
4083 {
4084 if (*p < 0x80) // be quick for ASCII
4085 return 0;
4086
4087 // Find the next character that isn't 10xx.xxxx
4088 for (i = 0; (p[i] & 0xc0) == 0x80; ++i)
4089 ;
4090 if (i > 0)
4091 {
4092 // Check for illegal sequence.
4093 for (j = 0; p - j > base; ++j)
4094 if ((p[-j] & 0xc0) != 0x80)
4095 break;
4096 if (utf8len_tab[p[-j]] != i + j)
4097 return 0;
4098 }
4099 return i;
4100 }
4101
4102 // Only need to check if we're on a trail byte, it doesn't matter if we
4103 // want the offset to the next or current character.
4104 return (*mb_head_off)(base, p);
4105 }
4106
4107 /*
4108 * Return the offset from "p" to the last byte of the character it points
4109 * into. Can start anywhere in a stream of bytes.
4110 */
4111 int
mb_tail_off(char_u * base,char_u * p)4112 mb_tail_off(char_u *base, char_u *p)
4113 {
4114 int i;
4115 int j;
4116
4117 if (*p == NUL)
4118 return 0;
4119
4120 if (enc_utf8)
4121 {
4122 // Find the last character that is 10xx.xxxx
4123 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i)
4124 ;
4125 // Check for illegal sequence.
4126 for (j = 0; p - j > base; ++j)
4127 if ((p[-j] & 0xc0) != 0x80)
4128 break;
4129 if (utf8len_tab[p[-j]] != i + j + 1)
4130 return 0;
4131 return i;
4132 }
4133
4134 // It can't be the first byte if a double-byte when not using DBCS, at the
4135 // end of the string or the byte can't start a double-byte.
4136 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1)
4137 return 0;
4138
4139 // Return 1 when on the lead byte, 0 when on the tail byte.
4140 return 1 - dbcs_head_off(base, p);
4141 }
4142
4143 /*
4144 * Find the next illegal byte sequence.
4145 */
4146 void
utf_find_illegal(void)4147 utf_find_illegal(void)
4148 {
4149 pos_T pos = curwin->w_cursor;
4150 char_u *p;
4151 int len;
4152 vimconv_T vimconv;
4153 char_u *tofree = NULL;
4154
4155 vimconv.vc_type = CONV_NONE;
4156 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT))
4157 {
4158 // 'encoding' is "utf-8" but we are editing a 8-bit encoded file,
4159 // possibly a utf-8 file with illegal bytes. Setup for conversion
4160 // from utf-8 to 'fileencoding'.
4161 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc);
4162 }
4163
4164 curwin->w_cursor.coladd = 0;
4165 for (;;)
4166 {
4167 p = ml_get_cursor();
4168 if (vimconv.vc_type != CONV_NONE)
4169 {
4170 vim_free(tofree);
4171 tofree = string_convert(&vimconv, p, NULL);
4172 if (tofree == NULL)
4173 break;
4174 p = tofree;
4175 }
4176
4177 while (*p != NUL)
4178 {
4179 // Illegal means that there are not enough trail bytes (checked by
4180 // utf_ptr2len()) or too many of them (overlong sequence).
4181 len = utf_ptr2len(p);
4182 if (*p >= 0x80 && (len == 1
4183 || utf_char2len(utf_ptr2char(p)) != len))
4184 {
4185 if (vimconv.vc_type == CONV_NONE)
4186 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor());
4187 else
4188 {
4189 int l;
4190
4191 len = (int)(p - tofree);
4192 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l)
4193 {
4194 l = utf_ptr2len(p);
4195 curwin->w_cursor.col += l;
4196 }
4197 }
4198 goto theend;
4199 }
4200 p += len;
4201 }
4202 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count)
4203 break;
4204 ++curwin->w_cursor.lnum;
4205 curwin->w_cursor.col = 0;
4206 }
4207
4208 // didn't find it: don't move and beep
4209 curwin->w_cursor = pos;
4210 beep_flush();
4211
4212 theend:
4213 vim_free(tofree);
4214 convert_setup(&vimconv, NULL, NULL);
4215 }
4216
4217 #if defined(FEAT_GUI_GTK) || defined(PROTO)
4218 /*
4219 * Return TRUE if string "s" is a valid utf-8 string.
4220 * When "end" is NULL stop at the first NUL.
4221 * When "end" is positive stop there.
4222 */
4223 int
utf_valid_string(char_u * s,char_u * end)4224 utf_valid_string(char_u *s, char_u *end)
4225 {
4226 int l;
4227 char_u *p = s;
4228
4229 while (end == NULL ? *p != NUL : p < end)
4230 {
4231 l = utf8len_tab_zero[*p];
4232 if (l == 0)
4233 return FALSE; // invalid lead byte
4234 if (end != NULL && p + l > end)
4235 return FALSE; // incomplete byte sequence
4236 ++p;
4237 while (--l > 0)
4238 if ((*p++ & 0xc0) != 0x80)
4239 return FALSE; // invalid trail byte
4240 }
4241 return TRUE;
4242 }
4243 #endif
4244
4245 #if defined(FEAT_GUI) || defined(PROTO)
4246 /*
4247 * Special version of mb_tail_off() for use in ScreenLines[].
4248 */
4249 int
dbcs_screen_tail_off(char_u * base,char_u * p)4250 dbcs_screen_tail_off(char_u *base, char_u *p)
4251 {
4252 // It can't be the first byte if a double-byte when not using DBCS, at the
4253 // end of the string or the byte can't start a double-byte.
4254 // For euc-jp an 0x8e byte always means we have a lead byte in the current
4255 // cell.
4256 if (*p == NUL || p[1] == NUL
4257 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)
4258 || MB_BYTE2LEN(*p) == 1)
4259 return 0;
4260
4261 // Return 1 when on the lead byte, 0 when on the tail byte.
4262 return 1 - dbcs_screen_head_off(base, p);
4263 }
4264 #endif
4265
4266 /*
4267 * If the cursor moves on an trail byte, set the cursor on the lead byte.
4268 * Thus it moves left if necessary.
4269 * Return TRUE when the cursor was adjusted.
4270 */
4271 void
mb_adjust_cursor(void)4272 mb_adjust_cursor(void)
4273 {
4274 mb_adjustpos(curbuf, &curwin->w_cursor);
4275 }
4276
4277 /*
4278 * Adjust position "*lp" to point to the first byte of a multi-byte character.
4279 * If it points to a tail byte it's moved backwards to the head byte.
4280 */
4281 void
mb_adjustpos(buf_T * buf,pos_T * lp)4282 mb_adjustpos(buf_T *buf, pos_T *lp)
4283 {
4284 char_u *p;
4285
4286 if (lp->col > 0 || lp->coladd > 1)
4287 {
4288 p = ml_get_buf(buf, lp->lnum, FALSE);
4289 if (*p == NUL || (int)STRLEN(p) < lp->col)
4290 lp->col = 0;
4291 else
4292 lp->col -= (*mb_head_off)(p, p + lp->col);
4293 // Reset "coladd" when the cursor would be on the right half of a
4294 // double-wide character.
4295 if (lp->coladd == 1
4296 && p[lp->col] != TAB
4297 && vim_isprintc((*mb_ptr2char)(p + lp->col))
4298 && ptr2cells(p + lp->col) > 1)
4299 lp->coladd = 0;
4300 }
4301 }
4302
4303 /*
4304 * Return a pointer to the character before "*p", if there is one.
4305 */
4306 char_u *
mb_prevptr(char_u * line,char_u * p)4307 mb_prevptr(
4308 char_u *line, // start of the string
4309 char_u *p)
4310 {
4311 if (p > line)
4312 MB_PTR_BACK(line, p);
4313 return p;
4314 }
4315
4316 /*
4317 * Return the character length of "str". Each multi-byte character (with
4318 * following composing characters) counts as one.
4319 */
4320 int
mb_charlen(char_u * str)4321 mb_charlen(char_u *str)
4322 {
4323 char_u *p = str;
4324 int count;
4325
4326 if (p == NULL)
4327 return 0;
4328
4329 for (count = 0; *p != NUL; count++)
4330 p += (*mb_ptr2len)(p);
4331
4332 return count;
4333 }
4334
4335 /*
4336 * Like mb_charlen() but for a string with specified length.
4337 */
4338 int
mb_charlen_len(char_u * str,int len)4339 mb_charlen_len(char_u *str, int len)
4340 {
4341 char_u *p = str;
4342 int count;
4343
4344 for (count = 0; *p != NUL && p < str + len; count++)
4345 p += (*mb_ptr2len)(p);
4346
4347 return count;
4348 }
4349
4350 /*
4351 * Try to un-escape a multi-byte character.
4352 * Used for the "to" and "from" part of a mapping.
4353 * Return the un-escaped string if it is a multi-byte character, and advance
4354 * "pp" to just after the bytes that formed it.
4355 * Return NULL if no multi-byte char was found.
4356 */
4357 char_u *
mb_unescape(char_u ** pp)4358 mb_unescape(char_u **pp)
4359 {
4360 static char_u buf[6];
4361 int n;
4362 int m = 0;
4363 char_u *str = *pp;
4364
4365 // Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI
4366 // KS_EXTRA KE_CSI to CSI.
4367 // Maximum length of a utf-8 character is 4 bytes.
4368 for (n = 0; str[n] != NUL && m < 4; ++n)
4369 {
4370 if (str[n] == K_SPECIAL
4371 && str[n + 1] == KS_SPECIAL
4372 && str[n + 2] == KE_FILLER)
4373 {
4374 buf[m++] = K_SPECIAL;
4375 n += 2;
4376 }
4377 else if ((str[n] == K_SPECIAL
4378 # ifdef FEAT_GUI
4379 || str[n] == CSI
4380 # endif
4381 )
4382 && str[n + 1] == KS_EXTRA
4383 && str[n + 2] == (int)KE_CSI)
4384 {
4385 buf[m++] = CSI;
4386 n += 2;
4387 }
4388 else if (str[n] == K_SPECIAL
4389 # ifdef FEAT_GUI
4390 || str[n] == CSI
4391 # endif
4392 )
4393 break; // a special key can't be a multibyte char
4394 else
4395 buf[m++] = str[n];
4396 buf[m] = NUL;
4397
4398 // Return a multi-byte character if it's found. An illegal sequence
4399 // will result in a 1 here.
4400 if ((*mb_ptr2len)(buf) > 1)
4401 {
4402 *pp = str + n + 1;
4403 return buf;
4404 }
4405
4406 // Bail out quickly for ASCII.
4407 if (buf[0] < 128)
4408 break;
4409 }
4410 return NULL;
4411 }
4412
4413 /*
4414 * Return TRUE if the character at "row"/"col" on the screen is the left side
4415 * of a double-width character.
4416 * Caller must make sure "row" and "col" are not invalid!
4417 */
4418 int
mb_lefthalve(int row,int col)4419 mb_lefthalve(int row, int col)
4420 {
4421 return (*mb_off2cells)(LineOffset[row] + col,
4422 LineOffset[row] + screen_Columns) > 1;
4423 }
4424
4425 /*
4426 * Correct a position on the screen, if it's the right half of a double-wide
4427 * char move it to the left half. Returns the corrected column.
4428 */
4429 int
mb_fix_col(int col,int row)4430 mb_fix_col(int col, int row)
4431 {
4432 int off;
4433
4434 col = check_col(col);
4435 row = check_row(row);
4436 off = LineOffset[row] + col;
4437 if (has_mbyte && ScreenLines != NULL && col > 0
4438 && ((enc_dbcs
4439 && ScreenLines[off] != NUL
4440 && dbcs_screen_head_off(ScreenLines + LineOffset[row],
4441 ScreenLines + off))
4442 || (enc_utf8 && ScreenLines[off] == 0
4443 && ScreenLinesUC[off] == 0)))
4444 return col - 1;
4445 return col;
4446 }
4447
4448 static int enc_alias_search(char_u *name);
4449
4450 /*
4451 * Skip the Vim specific head of a 'encoding' name.
4452 */
4453 char_u *
enc_skip(char_u * p)4454 enc_skip(char_u *p)
4455 {
4456 if (STRNCMP(p, "2byte-", 6) == 0)
4457 return p + 6;
4458 if (STRNCMP(p, "8bit-", 5) == 0)
4459 return p + 5;
4460 return p;
4461 }
4462
4463 /*
4464 * Find the canonical name for encoding "enc".
4465 * When the name isn't recognized, returns "enc" itself, but with all lower
4466 * case characters and '_' replaced with '-'.
4467 * Returns an allocated string. NULL for out-of-memory.
4468 */
4469 char_u *
enc_canonize(char_u * enc)4470 enc_canonize(char_u *enc)
4471 {
4472 char_u *r;
4473 char_u *p, *s;
4474 int i;
4475
4476 if (STRCMP(enc, "default") == 0)
4477 {
4478 #ifdef MSWIN
4479 // Use the system encoding, the default is always utf-8.
4480 r = enc_locale();
4481 #else
4482 // Use the default encoding as it's found by set_init_1().
4483 r = get_encoding_default();
4484 #endif
4485 if (r == NULL)
4486 r = (char_u *)ENC_DFLT;
4487 return vim_strsave(r);
4488 }
4489
4490 // copy "enc" to allocated memory, with room for two '-'
4491 r = alloc(STRLEN(enc) + 3);
4492 if (r != NULL)
4493 {
4494 // Make it all lower case and replace '_' with '-'.
4495 p = r;
4496 for (s = enc; *s != NUL; ++s)
4497 {
4498 if (*s == '_')
4499 *p++ = '-';
4500 else
4501 *p++ = TOLOWER_ASC(*s);
4502 }
4503 *p = NUL;
4504
4505 // Skip "2byte-" and "8bit-".
4506 p = enc_skip(r);
4507
4508 // Change "microsoft-cp" to "cp". Used in some spell files.
4509 if (STRNCMP(p, "microsoft-cp", 12) == 0)
4510 STRMOVE(p, p + 10);
4511
4512 // "iso8859" -> "iso-8859"
4513 if (STRNCMP(p, "iso8859", 7) == 0)
4514 {
4515 STRMOVE(p + 4, p + 3);
4516 p[3] = '-';
4517 }
4518
4519 // "iso-8859n" -> "iso-8859-n"
4520 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-')
4521 {
4522 STRMOVE(p + 9, p + 8);
4523 p[8] = '-';
4524 }
4525
4526 // "latin-N" -> "latinN"
4527 if (STRNCMP(p, "latin-", 6) == 0)
4528 STRMOVE(p + 5, p + 6);
4529
4530 if (enc_canon_search(p) >= 0)
4531 {
4532 // canonical name can be used unmodified
4533 if (p != r)
4534 STRMOVE(r, p);
4535 }
4536 else if ((i = enc_alias_search(p)) >= 0)
4537 {
4538 // alias recognized, get canonical name
4539 vim_free(r);
4540 r = vim_strsave((char_u *)enc_canon_table[i].name);
4541 }
4542 }
4543 return r;
4544 }
4545
4546 /*
4547 * Search for an encoding alias of "name".
4548 * Returns -1 when not found.
4549 */
4550 static int
enc_alias_search(char_u * name)4551 enc_alias_search(char_u *name)
4552 {
4553 int i;
4554
4555 for (i = 0; enc_alias_table[i].name != NULL; ++i)
4556 if (STRCMP(name, enc_alias_table[i].name) == 0)
4557 return enc_alias_table[i].canon;
4558 return -1;
4559 }
4560
4561
4562 #ifdef HAVE_LANGINFO_H
4563 # include <langinfo.h>
4564 #endif
4565
4566 #if !defined(FEAT_GUI_MSWIN) || defined(VIMDLL)
4567 /*
4568 * Get the canonicalized encoding from the specified locale string "locale"
4569 * or from the environment variables LC_ALL, LC_CTYPE and LANG.
4570 * Returns an allocated string when successful, NULL when not.
4571 */
4572 char_u *
enc_locale_env(char * locale)4573 enc_locale_env(char *locale)
4574 {
4575 char *s = locale;
4576 char *p;
4577 int i;
4578 char buf[50];
4579
4580 if (s == NULL || *s == NUL)
4581 if ((s = getenv("LC_ALL")) == NULL || *s == NUL)
4582 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL)
4583 s = getenv("LANG");
4584
4585 if (s == NULL || *s == NUL)
4586 return NULL;
4587
4588 // The most generic locale format is:
4589 // language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]]
4590 // If there is a '.' remove the part before it.
4591 // if there is something after the codeset, remove it.
4592 // Make the name lowercase and replace '_' with '-'.
4593 // Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn",
4594 // "ko_KR.EUC" == "euc-kr"
4595 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL)
4596 {
4597 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0
4598 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_')
4599 {
4600 // copy "XY.EUC" to "euc-XY" to buf[10]
4601 STRCPY(buf + 10, "euc-");
4602 buf[14] = p[-2];
4603 buf[15] = p[-1];
4604 buf[16] = 0;
4605 s = buf + 10;
4606 }
4607 else
4608 s = p + 1;
4609 }
4610 for (i = 0; i < (int)sizeof(buf) - 1 && s[i] != NUL; ++i)
4611 {
4612 if (s[i] == '_' || s[i] == '-')
4613 buf[i] = '-';
4614 else if (isalnum((int)s[i]))
4615 buf[i] = TOLOWER_ASC(s[i]);
4616 else
4617 break;
4618 }
4619 buf[i] = NUL;
4620
4621 return enc_canonize((char_u *)buf);
4622 }
4623 #endif
4624
4625 /*
4626 * Get the canonicalized encoding of the current locale.
4627 * Returns an allocated string when successful, NULL when not.
4628 */
4629 char_u *
enc_locale(void)4630 enc_locale(void)
4631 {
4632 #ifdef MSWIN
4633 char buf[50];
4634 long acp = GetACP();
4635
4636 if (acp == 1200)
4637 STRCPY(buf, "ucs-2le");
4638 else if (acp == 1252) // cp1252 is used as latin1
4639 STRCPY(buf, "latin1");
4640 else if (acp == 65001)
4641 STRCPY(buf, "utf-8");
4642 else
4643 sprintf(buf, "cp%ld", acp);
4644
4645 return enc_canonize((char_u *)buf);
4646 #else
4647 char *s;
4648
4649 # ifdef HAVE_NL_LANGINFO_CODESET
4650 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL)
4651 # endif
4652 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE)
4653 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL)
4654 # endif
4655 s = NULL;
4656
4657 return enc_locale_env(s);
4658 #endif
4659 }
4660
4661 # if defined(MSWIN) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD)
4662 /*
4663 * Convert an encoding name to an MS-Windows codepage.
4664 * Returns zero if no codepage can be figured out.
4665 */
4666 int
encname2codepage(char_u * name)4667 encname2codepage(char_u *name)
4668 {
4669 int cp;
4670 char_u *p = name;
4671 int idx;
4672
4673 if (STRNCMP(p, "8bit-", 5) == 0)
4674 p += 5;
4675 else if (STRNCMP(p_enc, "2byte-", 6) == 0)
4676 p += 6;
4677
4678 if (p[0] == 'c' && p[1] == 'p')
4679 cp = atoi((char *)p + 2);
4680 else if ((idx = enc_canon_search(p)) >= 0)
4681 cp = enc_canon_table[idx].codepage;
4682 else
4683 return 0;
4684 if (IsValidCodePage(cp))
4685 return cp;
4686 return 0;
4687 }
4688 # endif
4689
4690 # if defined(USE_ICONV) || defined(PROTO)
4691
4692 /*
4693 * Call iconv_open() with a check if iconv() works properly (there are broken
4694 * versions).
4695 * Returns (void *)-1 if failed.
4696 * (should return iconv_t, but that causes problems with prototypes).
4697 */
4698 void *
my_iconv_open(char_u * to,char_u * from)4699 my_iconv_open(char_u *to, char_u *from)
4700 {
4701 iconv_t fd;
4702 #define ICONV_TESTLEN 400
4703 char_u tobuf[ICONV_TESTLEN];
4704 char *p;
4705 size_t tolen;
4706 static int iconv_ok = -1;
4707
4708 if (iconv_ok == FALSE)
4709 return (void *)-1; // detected a broken iconv() previously
4710
4711 #ifdef DYNAMIC_ICONV
4712 // Check if the iconv.dll can be found.
4713 if (!iconv_enabled(TRUE))
4714 return (void *)-1;
4715 #endif
4716
4717 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from));
4718
4719 if (fd != (iconv_t)-1 && iconv_ok == -1)
4720 {
4721 /*
4722 * Do a dummy iconv() call to check if it actually works. There is a
4723 * version of iconv() on Linux that is broken. We can't ignore it,
4724 * because it's wide-spread. The symptoms are that after outputting
4725 * the initial shift state the "to" pointer is NULL and conversion
4726 * stops for no apparent reason after about 8160 characters.
4727 */
4728 p = (char *)tobuf;
4729 tolen = ICONV_TESTLEN;
4730 (void)iconv(fd, NULL, NULL, &p, &tolen);
4731 if (p == NULL)
4732 {
4733 iconv_ok = FALSE;
4734 iconv_close(fd);
4735 fd = (iconv_t)-1;
4736 }
4737 else
4738 iconv_ok = TRUE;
4739 }
4740
4741 return (void *)fd;
4742 }
4743
4744 /*
4745 * Convert the string "str[slen]" with iconv().
4746 * If "unconvlenp" is not NULL handle the string ending in an incomplete
4747 * sequence and set "*unconvlenp" to the length of it.
4748 * Returns the converted string in allocated memory. NULL for an error.
4749 * If resultlenp is not NULL, sets it to the result length in bytes.
4750 */
4751 static char_u *
iconv_string(vimconv_T * vcp,char_u * str,int slen,int * unconvlenp,int * resultlenp)4752 iconv_string(
4753 vimconv_T *vcp,
4754 char_u *str,
4755 int slen,
4756 int *unconvlenp,
4757 int *resultlenp)
4758 {
4759 const char *from;
4760 size_t fromlen;
4761 char *to;
4762 size_t tolen;
4763 size_t len = 0;
4764 size_t done = 0;
4765 char_u *result = NULL;
4766 char_u *p;
4767 int l;
4768
4769 from = (char *)str;
4770 fromlen = slen;
4771 for (;;)
4772 {
4773 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG)
4774 {
4775 // Allocate enough room for most conversions. When re-allocating
4776 // increase the buffer size.
4777 len = len + fromlen * 2 + 40;
4778 p = alloc(len);
4779 if (p != NULL && done > 0)
4780 mch_memmove(p, result, done);
4781 vim_free(result);
4782 result = p;
4783 if (result == NULL) // out of memory
4784 break;
4785 }
4786
4787 to = (char *)result + done;
4788 tolen = len - done - 2;
4789 // Avoid a warning for systems with a wrong iconv() prototype by
4790 // casting the second argument to void *.
4791 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen)
4792 != (size_t)-1)
4793 {
4794 // Finished, append a NUL.
4795 *to = NUL;
4796 break;
4797 }
4798
4799 // Check both ICONV_EINVAL and EINVAL, because the dynamically loaded
4800 // iconv library may use one of them.
4801 if (!vcp->vc_fail && unconvlenp != NULL
4802 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL))
4803 {
4804 // Handle an incomplete sequence at the end.
4805 *to = NUL;
4806 *unconvlenp = (int)fromlen;
4807 break;
4808 }
4809
4810 // Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded
4811 // iconv library may use one of them.
4812 else if (!vcp->vc_fail
4813 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ
4814 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL))
4815 {
4816 // Can't convert: insert a '?' and skip a character. This assumes
4817 // conversion from 'encoding' to something else. In other
4818 // situations we don't know what to skip anyway.
4819 *to++ = '?';
4820 if ((*mb_ptr2cells)((char_u *)from) > 1)
4821 *to++ = '?';
4822 if (enc_utf8)
4823 l = utfc_ptr2len_len((char_u *)from, (int)fromlen);
4824 else
4825 {
4826 l = (*mb_ptr2len)((char_u *)from);
4827 if (l > (int)fromlen)
4828 l = (int)fromlen;
4829 }
4830 from += l;
4831 fromlen -= l;
4832 }
4833 else if (ICONV_ERRNO != ICONV_E2BIG)
4834 {
4835 // conversion failed
4836 VIM_CLEAR(result);
4837 break;
4838 }
4839 // Not enough room or skipping illegal sequence.
4840 done = to - (char *)result;
4841 }
4842
4843 if (resultlenp != NULL && result != NULL)
4844 *resultlenp = (int)(to - (char *)result);
4845 return result;
4846 }
4847
4848 # if defined(DYNAMIC_ICONV) || defined(PROTO)
4849 /*
4850 * Dynamically load the "iconv.dll" on Win32.
4851 */
4852
4853 # ifndef DYNAMIC_ICONV // must be generating prototypes
4854 # define HINSTANCE int
4855 # endif
4856 static HINSTANCE hIconvDLL = 0;
4857 static HINSTANCE hMsvcrtDLL = 0;
4858
4859 # ifndef DYNAMIC_ICONV_DLL
4860 # define DYNAMIC_ICONV_DLL "iconv.dll"
4861 # define DYNAMIC_ICONV_DLL_ALT1 "libiconv.dll"
4862 # define DYNAMIC_ICONV_DLL_ALT2 "libiconv2.dll"
4863 # define DYNAMIC_ICONV_DLL_ALT3 "libiconv-2.dll"
4864 # endif
4865 # ifndef DYNAMIC_MSVCRT_DLL
4866 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll"
4867 # endif
4868
4869 /*
4870 * Try opening the iconv.dll and return TRUE if iconv() can be used.
4871 */
4872 int
iconv_enabled(int verbose)4873 iconv_enabled(int verbose)
4874 {
4875 if (hIconvDLL != 0 && hMsvcrtDLL != 0)
4876 return TRUE;
4877
4878 // The iconv DLL file goes under different names, try them all.
4879 // Do the "2" version first, it's newer.
4880 #ifdef DYNAMIC_ICONV_DLL_ALT2
4881 if (hIconvDLL == 0)
4882 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT2);
4883 #endif
4884 #ifdef DYNAMIC_ICONV_DLL_ALT3
4885 if (hIconvDLL == 0)
4886 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT3);
4887 #endif
4888 if (hIconvDLL == 0)
4889 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL);
4890 #ifdef DYNAMIC_ICONV_DLL_ALT1
4891 if (hIconvDLL == 0)
4892 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT1);
4893 #endif
4894
4895 if (hIconvDLL != 0)
4896 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL);
4897 if (hIconvDLL == 0 || hMsvcrtDLL == 0)
4898 {
4899 // Only give the message when 'verbose' is set, otherwise it might be
4900 // done whenever a conversion is attempted.
4901 if (verbose && p_verbose > 0)
4902 {
4903 verbose_enter();
4904 semsg(_(e_loadlib),
4905 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL,
4906 GetWin32Error());
4907 verbose_leave();
4908 }
4909 iconv_end();
4910 return FALSE;
4911 }
4912
4913 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv");
4914 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open");
4915 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close");
4916 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl");
4917 iconv_errno = get_dll_import_func(hIconvDLL, "_errno");
4918 if (iconv_errno == NULL)
4919 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno");
4920 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL
4921 || iconvctl == NULL || iconv_errno == NULL)
4922 {
4923 iconv_end();
4924 if (verbose && p_verbose > 0)
4925 {
4926 verbose_enter();
4927 semsg(_(e_loadfunc), "for libiconv");
4928 verbose_leave();
4929 }
4930 return FALSE;
4931 }
4932 return TRUE;
4933 }
4934
4935 void
iconv_end(void)4936 iconv_end(void)
4937 {
4938 // Don't use iconv() when inputting or outputting characters.
4939 if (input_conv.vc_type == CONV_ICONV)
4940 convert_setup(&input_conv, NULL, NULL);
4941 if (output_conv.vc_type == CONV_ICONV)
4942 convert_setup(&output_conv, NULL, NULL);
4943
4944 if (hIconvDLL != 0)
4945 FreeLibrary(hIconvDLL);
4946 if (hMsvcrtDLL != 0)
4947 FreeLibrary(hMsvcrtDLL);
4948 hIconvDLL = 0;
4949 hMsvcrtDLL = 0;
4950 }
4951 # endif // DYNAMIC_ICONV
4952 # endif // USE_ICONV
4953
4954 #if defined(FEAT_EVAL) || defined(PROTO)
4955 /*
4956 * "getimstatus()" function
4957 */
4958 void
f_getimstatus(typval_T * argvars UNUSED,typval_T * rettv)4959 f_getimstatus(typval_T *argvars UNUSED, typval_T *rettv)
4960 {
4961 # if defined(HAVE_INPUT_METHOD)
4962 rettv->vval.v_number = im_get_status();
4963 # endif
4964 }
4965
4966 /*
4967 * iconv() function
4968 */
4969 void
f_iconv(typval_T * argvars UNUSED,typval_T * rettv)4970 f_iconv(typval_T *argvars UNUSED, typval_T *rettv)
4971 {
4972 char_u buf1[NUMBUFLEN];
4973 char_u buf2[NUMBUFLEN];
4974 char_u *from, *to, *str;
4975 vimconv_T vimconv;
4976
4977 rettv->v_type = VAR_STRING;
4978 rettv->vval.v_string = NULL;
4979
4980 if (in_vim9script()
4981 && (check_for_string_arg(argvars, 0) == FAIL
4982 || check_for_string_arg(argvars, 1) == FAIL
4983 || check_for_string_arg(argvars, 2) == FAIL))
4984 return;
4985
4986 str = tv_get_string(&argvars[0]);
4987 from = enc_canonize(enc_skip(tv_get_string_buf(&argvars[1], buf1)));
4988 to = enc_canonize(enc_skip(tv_get_string_buf(&argvars[2], buf2)));
4989 vimconv.vc_type = CONV_NONE;
4990 convert_setup(&vimconv, from, to);
4991
4992 // If the encodings are equal, no conversion needed.
4993 if (vimconv.vc_type == CONV_NONE)
4994 rettv->vval.v_string = vim_strsave(str);
4995 else
4996 rettv->vval.v_string = string_convert(&vimconv, str, NULL);
4997
4998 convert_setup(&vimconv, NULL, NULL);
4999 vim_free(from);
5000 vim_free(to);
5001 }
5002 #endif
5003
5004 /*
5005 * Setup "vcp" for conversion from "from" to "to".
5006 * The names must have been made canonical with enc_canonize().
5007 * vcp->vc_type must have been initialized to CONV_NONE.
5008 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8
5009 * instead).
5010 * Afterwards invoke with "from" and "to" equal to NULL to cleanup.
5011 * Return FAIL when conversion is not supported, OK otherwise.
5012 */
5013 int
convert_setup(vimconv_T * vcp,char_u * from,char_u * to)5014 convert_setup(vimconv_T *vcp, char_u *from, char_u *to)
5015 {
5016 return convert_setup_ext(vcp, from, TRUE, to, TRUE);
5017 }
5018
5019 /*
5020 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all
5021 * "from" unicode charsets be considered utf-8. Same for "to".
5022 */
5023 int
convert_setup_ext(vimconv_T * vcp,char_u * from,int from_unicode_is_utf8,char_u * to,int to_unicode_is_utf8)5024 convert_setup_ext(
5025 vimconv_T *vcp,
5026 char_u *from,
5027 int from_unicode_is_utf8,
5028 char_u *to,
5029 int to_unicode_is_utf8)
5030 {
5031 int from_prop;
5032 int to_prop;
5033 int from_is_utf8;
5034 int to_is_utf8;
5035
5036 // Reset to no conversion.
5037 #ifdef USE_ICONV
5038 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1)
5039 iconv_close(vcp->vc_fd);
5040 #endif
5041 vcp->vc_type = CONV_NONE;
5042 vcp->vc_factor = 1;
5043 vcp->vc_fail = FALSE;
5044
5045 // No conversion when one of the names is empty or they are equal.
5046 if (from == NULL || *from == NUL || to == NULL || *to == NUL
5047 || STRCMP(from, to) == 0)
5048 return OK;
5049
5050 from_prop = enc_canon_props(from);
5051 to_prop = enc_canon_props(to);
5052 if (from_unicode_is_utf8)
5053 from_is_utf8 = from_prop & ENC_UNICODE;
5054 else
5055 from_is_utf8 = from_prop == ENC_UNICODE;
5056 if (to_unicode_is_utf8)
5057 to_is_utf8 = to_prop & ENC_UNICODE;
5058 else
5059 to_is_utf8 = to_prop == ENC_UNICODE;
5060
5061 if ((from_prop & ENC_LATIN1) && to_is_utf8)
5062 {
5063 // Internal latin1 -> utf-8 conversion.
5064 vcp->vc_type = CONV_TO_UTF8;
5065 vcp->vc_factor = 2; // up to twice as long
5066 }
5067 else if ((from_prop & ENC_LATIN9) && to_is_utf8)
5068 {
5069 // Internal latin9 -> utf-8 conversion.
5070 vcp->vc_type = CONV_9_TO_UTF8;
5071 vcp->vc_factor = 3; // up to three as long (euro sign)
5072 }
5073 else if (from_is_utf8 && (to_prop & ENC_LATIN1))
5074 {
5075 // Internal utf-8 -> latin1 conversion.
5076 vcp->vc_type = CONV_TO_LATIN1;
5077 }
5078 else if (from_is_utf8 && (to_prop & ENC_LATIN9))
5079 {
5080 // Internal utf-8 -> latin9 conversion.
5081 vcp->vc_type = CONV_TO_LATIN9;
5082 }
5083 #ifdef MSWIN
5084 // Win32-specific codepage <-> codepage conversion without iconv.
5085 else if ((from_is_utf8 || encname2codepage(from) > 0)
5086 && (to_is_utf8 || encname2codepage(to) > 0))
5087 {
5088 vcp->vc_type = CONV_CODEPAGE;
5089 vcp->vc_factor = 2; // up to twice as long
5090 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from);
5091 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to);
5092 }
5093 #endif
5094 #ifdef MACOS_CONVERT
5095 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1))
5096 {
5097 vcp->vc_type = CONV_MAC_LATIN1;
5098 }
5099 else if ((from_prop & ENC_MACROMAN) && to_is_utf8)
5100 {
5101 vcp->vc_type = CONV_MAC_UTF8;
5102 vcp->vc_factor = 2; // up to twice as long
5103 }
5104 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN))
5105 {
5106 vcp->vc_type = CONV_LATIN1_MAC;
5107 }
5108 else if (from_is_utf8 && (to_prop & ENC_MACROMAN))
5109 {
5110 vcp->vc_type = CONV_UTF8_MAC;
5111 }
5112 #endif
5113 #ifdef USE_ICONV
5114 else
5115 {
5116 // Use iconv() for conversion.
5117 vcp->vc_fd = (iconv_t)my_iconv_open(
5118 to_is_utf8 ? (char_u *)"utf-8" : to,
5119 from_is_utf8 ? (char_u *)"utf-8" : from);
5120 if (vcp->vc_fd != (iconv_t)-1)
5121 {
5122 vcp->vc_type = CONV_ICONV;
5123 vcp->vc_factor = 4; // could be longer too...
5124 }
5125 }
5126 #endif
5127 if (vcp->vc_type == CONV_NONE)
5128 return FAIL;
5129
5130 return OK;
5131 }
5132
5133 #if defined(FEAT_GUI) || defined(AMIGA) || defined(MSWIN) \
5134 || defined(PROTO)
5135 /*
5136 * Do conversion on typed input characters in-place.
5137 * The input and output are not NUL terminated!
5138 * Returns the length after conversion.
5139 */
5140 int
convert_input(char_u * ptr,int len,int maxlen)5141 convert_input(char_u *ptr, int len, int maxlen)
5142 {
5143 return convert_input_safe(ptr, len, maxlen, NULL, NULL);
5144 }
5145 #endif
5146
5147 /*
5148 * Like convert_input(), but when there is an incomplete byte sequence at the
5149 * end return that as an allocated string in "restp" and set "*restlenp" to
5150 * the length. If "restp" is NULL it is not used.
5151 */
5152 int
convert_input_safe(char_u * ptr,int len,int maxlen,char_u ** restp,int * restlenp)5153 convert_input_safe(
5154 char_u *ptr,
5155 int len,
5156 int maxlen,
5157 char_u **restp,
5158 int *restlenp)
5159 {
5160 char_u *d;
5161 int dlen = len;
5162 int unconvertlen = 0;
5163
5164 d = string_convert_ext(&input_conv, ptr, &dlen,
5165 restp == NULL ? NULL : &unconvertlen);
5166 if (d != NULL)
5167 {
5168 if (dlen <= maxlen)
5169 {
5170 if (unconvertlen > 0)
5171 {
5172 // Move the unconverted characters to allocated memory.
5173 *restp = alloc(unconvertlen);
5174 if (*restp != NULL)
5175 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen);
5176 *restlenp = unconvertlen;
5177 }
5178 mch_memmove(ptr, d, dlen);
5179 }
5180 else
5181 // result is too long, keep the unconverted text (the caller must
5182 // have done something wrong!)
5183 dlen = len;
5184 vim_free(d);
5185 }
5186 return dlen;
5187 }
5188
5189 /*
5190 * Convert text "ptr[*lenp]" according to "vcp".
5191 * Returns the result in allocated memory and sets "*lenp".
5192 * When "lenp" is NULL, use NUL terminated strings.
5193 * Illegal chars are often changed to "?", unless vcp->vc_fail is set.
5194 * When something goes wrong, NULL is returned and "*lenp" is unchanged.
5195 */
5196 char_u *
string_convert(vimconv_T * vcp,char_u * ptr,int * lenp)5197 string_convert(
5198 vimconv_T *vcp,
5199 char_u *ptr,
5200 int *lenp)
5201 {
5202 return string_convert_ext(vcp, ptr, lenp, NULL);
5203 }
5204
5205 /*
5206 * Like string_convert(), but when "unconvlenp" is not NULL and there are is
5207 * an incomplete sequence at the end it is not converted and "*unconvlenp" is
5208 * set to the number of remaining bytes.
5209 */
5210 char_u *
string_convert_ext(vimconv_T * vcp,char_u * ptr,int * lenp,int * unconvlenp)5211 string_convert_ext(
5212 vimconv_T *vcp,
5213 char_u *ptr,
5214 int *lenp,
5215 int *unconvlenp)
5216 {
5217 char_u *retval = NULL;
5218 char_u *d;
5219 int len;
5220 int i;
5221 int l;
5222 int c;
5223
5224 if (lenp == NULL)
5225 len = (int)STRLEN(ptr);
5226 else
5227 len = *lenp;
5228 if (len == 0)
5229 return vim_strsave((char_u *)"");
5230
5231 switch (vcp->vc_type)
5232 {
5233 case CONV_TO_UTF8: // latin1 to utf-8 conversion
5234 retval = alloc(len * 2 + 1);
5235 if (retval == NULL)
5236 break;
5237 d = retval;
5238 for (i = 0; i < len; ++i)
5239 {
5240 c = ptr[i];
5241 if (c < 0x80)
5242 *d++ = c;
5243 else
5244 {
5245 *d++ = 0xc0 + ((unsigned)c >> 6);
5246 *d++ = 0x80 + (c & 0x3f);
5247 }
5248 }
5249 *d = NUL;
5250 if (lenp != NULL)
5251 *lenp = (int)(d - retval);
5252 break;
5253
5254 case CONV_9_TO_UTF8: // latin9 to utf-8 conversion
5255 retval = alloc(len * 3 + 1);
5256 if (retval == NULL)
5257 break;
5258 d = retval;
5259 for (i = 0; i < len; ++i)
5260 {
5261 c = ptr[i];
5262 switch (c)
5263 {
5264 case 0xa4: c = 0x20ac; break; // euro
5265 case 0xa6: c = 0x0160; break; // S hat
5266 case 0xa8: c = 0x0161; break; // S -hat
5267 case 0xb4: c = 0x017d; break; // Z hat
5268 case 0xb8: c = 0x017e; break; // Z -hat
5269 case 0xbc: c = 0x0152; break; // OE
5270 case 0xbd: c = 0x0153; break; // oe
5271 case 0xbe: c = 0x0178; break; // Y
5272 }
5273 d += utf_char2bytes(c, d);
5274 }
5275 *d = NUL;
5276 if (lenp != NULL)
5277 *lenp = (int)(d - retval);
5278 break;
5279
5280 case CONV_TO_LATIN1: // utf-8 to latin1 conversion
5281 case CONV_TO_LATIN9: // utf-8 to latin9 conversion
5282 retval = alloc(len + 1);
5283 if (retval == NULL)
5284 break;
5285 d = retval;
5286 for (i = 0; i < len; ++i)
5287 {
5288 l = utf_ptr2len_len(ptr + i, len - i);
5289 if (l == 0)
5290 *d++ = NUL;
5291 else if (l == 1)
5292 {
5293 int l_w = utf8len_tab_zero[ptr[i]];
5294
5295 if (l_w == 0)
5296 {
5297 // Illegal utf-8 byte cannot be converted
5298 vim_free(retval);
5299 return NULL;
5300 }
5301 if (unconvlenp != NULL && l_w > len - i)
5302 {
5303 // Incomplete sequence at the end.
5304 *unconvlenp = len - i;
5305 break;
5306 }
5307 *d++ = ptr[i];
5308 }
5309 else
5310 {
5311 c = utf_ptr2char(ptr + i);
5312 if (vcp->vc_type == CONV_TO_LATIN9)
5313 switch (c)
5314 {
5315 case 0x20ac: c = 0xa4; break; // euro
5316 case 0x0160: c = 0xa6; break; // S hat
5317 case 0x0161: c = 0xa8; break; // S -hat
5318 case 0x017d: c = 0xb4; break; // Z hat
5319 case 0x017e: c = 0xb8; break; // Z -hat
5320 case 0x0152: c = 0xbc; break; // OE
5321 case 0x0153: c = 0xbd; break; // oe
5322 case 0x0178: c = 0xbe; break; // Y
5323 case 0xa4:
5324 case 0xa6:
5325 case 0xa8:
5326 case 0xb4:
5327 case 0xb8:
5328 case 0xbc:
5329 case 0xbd:
5330 case 0xbe: c = 0x100; break; // not in latin9
5331 }
5332 if (!utf_iscomposing(c)) // skip composing chars
5333 {
5334 if (c < 0x100)
5335 *d++ = c;
5336 else if (vcp->vc_fail)
5337 {
5338 vim_free(retval);
5339 return NULL;
5340 }
5341 else
5342 {
5343 *d++ = 0xbf;
5344 if (utf_char2cells(c) > 1)
5345 *d++ = '?';
5346 }
5347 }
5348 i += l - 1;
5349 }
5350 }
5351 *d = NUL;
5352 if (lenp != NULL)
5353 *lenp = (int)(d - retval);
5354 break;
5355
5356 # ifdef MACOS_CONVERT
5357 case CONV_MAC_LATIN1:
5358 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail,
5359 'm', 'l', unconvlenp);
5360 break;
5361
5362 case CONV_LATIN1_MAC:
5363 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail,
5364 'l', 'm', unconvlenp);
5365 break;
5366
5367 case CONV_MAC_UTF8:
5368 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail,
5369 'm', 'u', unconvlenp);
5370 break;
5371
5372 case CONV_UTF8_MAC:
5373 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail,
5374 'u', 'm', unconvlenp);
5375 break;
5376 # endif
5377
5378 # ifdef USE_ICONV
5379 case CONV_ICONV: // conversion with output_conv.vc_fd
5380 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp);
5381 break;
5382 # endif
5383 # ifdef MSWIN
5384 case CONV_CODEPAGE: // codepage -> codepage
5385 {
5386 int retlen;
5387 int tmp_len;
5388 short_u *tmp;
5389
5390 // 1. codepage/UTF-8 -> ucs-2.
5391 if (vcp->vc_cpfrom == 0)
5392 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL);
5393 else
5394 {
5395 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom,
5396 unconvlenp ? MB_ERR_INVALID_CHARS : 0,
5397 (char *)ptr, len, 0, 0);
5398 if (tmp_len == 0
5399 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION)
5400 {
5401 if (lenp != NULL)
5402 *lenp = 0;
5403 if (unconvlenp != NULL)
5404 *unconvlenp = len;
5405 retval = alloc(1);
5406 if (retval)
5407 retval[0] = NUL;
5408 return retval;
5409 }
5410 }
5411 tmp = ALLOC_MULT(short_u, tmp_len);
5412 if (tmp == NULL)
5413 break;
5414 if (vcp->vc_cpfrom == 0)
5415 utf8_to_utf16(ptr, len, tmp, unconvlenp);
5416 else
5417 MultiByteToWideChar(vcp->vc_cpfrom, 0,
5418 (char *)ptr, len, tmp, tmp_len);
5419
5420 // 2. ucs-2 -> codepage/UTF-8.
5421 if (vcp->vc_cpto == 0)
5422 retlen = utf16_to_utf8(tmp, tmp_len, NULL);
5423 else
5424 retlen = WideCharToMultiByte(vcp->vc_cpto, 0,
5425 tmp, tmp_len, 0, 0, 0, 0);
5426 retval = alloc(retlen + 1);
5427 if (retval != NULL)
5428 {
5429 if (vcp->vc_cpto == 0)
5430 utf16_to_utf8(tmp, tmp_len, retval);
5431 else
5432 WideCharToMultiByte(vcp->vc_cpto, 0,
5433 tmp, tmp_len,
5434 (char *)retval, retlen, 0, 0);
5435 retval[retlen] = NUL;
5436 if (lenp != NULL)
5437 *lenp = retlen;
5438 }
5439 vim_free(tmp);
5440 break;
5441 }
5442 # endif
5443 }
5444
5445 return retval;
5446 }
5447
5448 #if defined(FEAT_EVAL) || defined(PROTO)
5449
5450 /*
5451 * Table set by setcellwidths().
5452 */
5453 typedef struct
5454 {
5455 long first;
5456 long last;
5457 char width;
5458 } cw_interval_T;
5459
5460 static cw_interval_T *cw_table = NULL;
5461 static size_t cw_table_size = 0;
5462
5463 /*
5464 * Return 1 or 2 when "c" is in the cellwidth table.
5465 * Return 0 if not.
5466 */
5467 static int
cw_value(int c)5468 cw_value(int c)
5469 {
5470 int mid, bot, top;
5471
5472 if (cw_table == NULL)
5473 return 0;
5474
5475 // first quick check for Latin1 etc. characters
5476 if (c < cw_table[0].first)
5477 return 0;
5478
5479 // binary search in table
5480 bot = 0;
5481 top = (int)cw_table_size - 1;
5482 while (top >= bot)
5483 {
5484 mid = (bot + top) / 2;
5485 if (cw_table[mid].last < c)
5486 bot = mid + 1;
5487 else if (cw_table[mid].first > c)
5488 top = mid - 1;
5489 else
5490 return cw_table[mid].width;
5491 }
5492 return 0;
5493 }
5494
5495 static int
tv_nr_compare(const void * a1,const void * a2)5496 tv_nr_compare(const void *a1, const void *a2)
5497 {
5498 listitem_T *li1 = *(listitem_T **)a1;
5499 listitem_T *li2 = *(listitem_T **)a2;
5500
5501 return li1->li_tv.vval.v_number - li2->li_tv.vval.v_number;
5502 }
5503
5504 void
f_setcellwidths(typval_T * argvars,typval_T * rettv UNUSED)5505 f_setcellwidths(typval_T *argvars, typval_T *rettv UNUSED)
5506 {
5507 list_T *l;
5508 listitem_T *li;
5509 int item;
5510 int i;
5511 listitem_T **ptrs;
5512 cw_interval_T *table;
5513 cw_interval_T *cw_table_save;
5514 size_t cw_table_size_save;
5515
5516 if (in_vim9script() && check_for_list_arg(argvars, 0) == FAIL)
5517 return;
5518
5519 if (argvars[0].v_type != VAR_LIST || argvars[0].vval.v_list == NULL)
5520 {
5521 emsg(_(e_listreq));
5522 return;
5523 }
5524 l = argvars[0].vval.v_list;
5525 if (l->lv_len == 0)
5526 {
5527 // Clearing the table.
5528 vim_free(cw_table);
5529 cw_table = NULL;
5530 cw_table_size = 0;
5531 return;
5532 }
5533
5534 ptrs = ALLOC_MULT(listitem_T *, l->lv_len);
5535 if (ptrs == NULL)
5536 return;
5537
5538 // Check that all entries are a list with three numbers, the range is
5539 // valid and the cell width is valid.
5540 item = 0;
5541 for (li = l->lv_first; li != NULL; li = li->li_next)
5542 {
5543 listitem_T *lili;
5544 varnumber_T n1;
5545
5546 if (li->li_tv.v_type != VAR_LIST || li->li_tv.vval.v_list == NULL)
5547 {
5548 semsg(_(e_list_item_nr_is_not_list), item);
5549 vim_free(ptrs);
5550 return;
5551 }
5552
5553 lili = li->li_tv.vval.v_list->lv_first;
5554 ptrs[item] = lili;
5555 for (i = 0; lili != NULL; lili = lili->li_next, ++i)
5556 {
5557 if (lili->li_tv.v_type != VAR_NUMBER)
5558 break;
5559 if (i == 0)
5560 {
5561 n1 = lili->li_tv.vval.v_number;
5562 if (n1 < 0x100)
5563 {
5564 emsg(_(e_only_values_of_0x100_and_higher_supported));
5565 vim_free(ptrs);
5566 return;
5567 }
5568 }
5569 else if (i == 1 && lili->li_tv.vval.v_number < n1)
5570 {
5571 semsg(_(e_list_item_nr_range_invalid), item);
5572 vim_free(ptrs);
5573 return;
5574 }
5575 else if (i == 2 && (lili->li_tv.vval.v_number < 1
5576 || lili->li_tv.vval.v_number > 2))
5577 {
5578 semsg(_(e_list_item_nr_cell_width_invalid), item);
5579 vim_free(ptrs);
5580 return;
5581 }
5582 }
5583 if (i != 3)
5584 {
5585 semsg(_(e_list_item_nr_does_not_contain_3_numbers), item);
5586 vim_free(ptrs);
5587 return;
5588 }
5589 ++item;
5590 }
5591
5592 // Sort the list on the first number.
5593 qsort((void *)ptrs, (size_t)l->lv_len, sizeof(listitem_T *), tv_nr_compare);
5594
5595 table = ALLOC_MULT(cw_interval_T, l->lv_len);
5596 if (table == NULL)
5597 {
5598 vim_free(ptrs);
5599 return;
5600 }
5601
5602 // Store the items in the new table.
5603 item = 0;
5604 for (item = 0; item < l->lv_len; ++item)
5605 {
5606 listitem_T *lili = ptrs[item];
5607 varnumber_T n1;
5608
5609 n1 = lili->li_tv.vval.v_number;
5610 if (item > 0 && n1 <= table[item - 1].last)
5611 {
5612 semsg(_(e_overlapping_ranges_for_nr), (long)n1);
5613 vim_free(ptrs);
5614 vim_free(table);
5615 return;
5616 }
5617 table[item].first = n1;
5618 lili = lili->li_next;
5619 table[item].last = lili->li_tv.vval.v_number;
5620 lili = lili->li_next;
5621 table[item].width = lili->li_tv.vval.v_number;
5622 }
5623
5624 vim_free(ptrs);
5625
5626 cw_table_save = cw_table;
5627 cw_table_size_save = cw_table_size;
5628 cw_table = table;
5629 cw_table_size = l->lv_len;
5630
5631 // Check that the new value does not conflict with 'fillchars' or
5632 // 'listchars'.
5633 if (set_chars_option(curwin, &p_fcs) != NULL)
5634 {
5635 emsg(_(e_conflicts_with_value_of_fillchars));
5636 cw_table = cw_table_save;
5637 cw_table_size = cw_table_size_save;
5638 vim_free(table);
5639 return;
5640 }
5641 else
5642 {
5643 tabpage_T *tp;
5644 win_T *wp;
5645
5646 FOR_ALL_TAB_WINDOWS(tp, wp)
5647 {
5648 if (set_chars_option(wp, &wp->w_p_lcs) != NULL)
5649 {
5650 emsg((e_conflicts_with_value_of_listchars));
5651 cw_table = cw_table_save;
5652 cw_table_size = cw_table_size_save;
5653 vim_free(table);
5654 return;
5655 }
5656 }
5657 }
5658
5659 vim_free(cw_table_save);
5660 }
5661
5662 void
f_charclass(typval_T * argvars,typval_T * rettv UNUSED)5663 f_charclass(typval_T *argvars, typval_T *rettv UNUSED)
5664 {
5665 if (check_for_string_arg(argvars, 0) == FAIL
5666 || argvars[0].vval.v_string == NULL)
5667 return;
5668 rettv->vval.v_number = mb_get_class(argvars[0].vval.v_string);
5669 }
5670 #endif
5671