1 /* vi:set ts=8 sts=4 sw=4: 2 * 3 * VIM - Vi IMproved by Bram Moolenaar 4 * Multibyte extensions partly by Sung-Hoon Baek 5 * 6 * Do ":help uganda" in Vim to read copying and usage conditions. 7 * Do ":help credits" in Vim to see a list of people who contributed. 8 * See README.txt for an overview of the Vim source code. 9 */ 10 /* 11 * mbyte.c: Code specifically for handling multi-byte characters. 12 * 13 * The encoding used in the core is set with 'encoding'. When 'encoding' is 14 * changed, the following four variables are set (for speed). 15 * Currently these types of character encodings are supported: 16 * 17 * "enc_dbcs" When non-zero it tells the type of double byte character 18 * encoding (Chinese, Korean, Japanese, etc.). 19 * The cell width on the display is equal to the number of 20 * bytes. (exception: DBCS_JPNU with first byte 0x8e) 21 * Recognizing the first or second byte is difficult, it 22 * requires checking a byte sequence from the start. 23 * "enc_utf8" When TRUE use Unicode characters in UTF-8 encoding. 24 * The cell width on the display needs to be determined from 25 * the character value. 26 * Recognizing bytes is easy: 0xxx.xxxx is a single-byte 27 * char, 10xx.xxxx is a trailing byte, 11xx.xxxx is a leading 28 * byte of a multi-byte character. 29 * To make things complicated, up to 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(WIN3264) || 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 #endif 110 111 #if defined(FEAT_GUI_GTK) && defined(FEAT_XIM) 112 # if GTK_CHECK_VERSION(3,0,0) 113 # include <gdk/gdkkeysyms-compat.h> 114 # else 115 # include <gdk/gdkkeysyms.h> 116 # endif 117 # ifdef WIN3264 118 # include <gdk/gdkwin32.h> 119 # else 120 # include <gdk/gdkx.h> 121 # endif 122 #endif 123 124 #ifdef HAVE_WCHAR_H 125 # include <wchar.h> 126 #endif 127 128 #if 0 129 /* This has been disabled, because several people reported problems with the 130 * wcwidth() and iswprint() library functions, esp. for Hebrew. */ 131 # ifdef __STDC_ISO_10646__ 132 # define USE_WCHAR_FUNCTIONS 133 # endif 134 #endif 135 136 #if defined(FEAT_MBYTE) || defined(PROTO) 137 138 static int enc_canon_search(char_u *name); 139 static int dbcs_char2len(int c); 140 static int dbcs_char2bytes(int c, char_u *buf); 141 static int dbcs_ptr2len(char_u *p); 142 static int dbcs_ptr2len_len(char_u *p, int size); 143 static int utf_ptr2cells_len(char_u *p, int size); 144 static int dbcs_char2cells(int c); 145 static int dbcs_ptr2cells_len(char_u *p, int size); 146 static int dbcs_ptr2char(char_u *p); 147 static int utf_safe_read_char_adv(char_u **s, size_t *n); 148 149 /* 150 * Lookup table to quickly get the length in bytes of a UTF-8 character from 151 * the first byte of a UTF-8 string. 152 * Bytes which are illegal when used as the first byte have a 1. 153 * The NUL byte has length 1. 154 */ 155 static char utf8len_tab[256] = 156 { 157 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 158 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 159 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 160 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 161 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 162 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 163 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, 164 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, 165 }; 166 167 /* 168 * Like utf8len_tab above, but using a zero for illegal lead bytes. 169 */ 170 static char utf8len_tab_zero[256] = 171 { 172 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 173 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 174 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 175 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 176 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, 177 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, 178 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, 179 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, 180 }; 181 182 /* 183 * XIM often causes trouble. Define XIM_DEBUG to get a log of XIM callbacks 184 * in the "xim.log" file. 185 */ 186 /* #define XIM_DEBUG */ 187 #ifdef XIM_DEBUG 188 static void 189 xim_log(char *s, ...) 190 { 191 va_list arglist; 192 static FILE *fd = NULL; 193 194 if (fd == (FILE *)-1) 195 return; 196 if (fd == NULL) 197 { 198 fd = mch_fopen("xim.log", "w"); 199 if (fd == NULL) 200 { 201 EMSG("Cannot open xim.log"); 202 fd = (FILE *)-1; 203 return; 204 } 205 } 206 207 va_start(arglist, s); 208 vfprintf(fd, s, arglist); 209 va_end(arglist); 210 } 211 #endif 212 213 #endif 214 215 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 216 /* 217 * Canonical encoding names and their properties. 218 * "iso-8859-n" is handled by enc_canonize() directly. 219 */ 220 static struct 221 { char *name; int prop; int codepage;} 222 enc_canon_table[] = 223 { 224 #define IDX_LATIN_1 0 225 {"latin1", ENC_8BIT + ENC_LATIN1, 1252}, 226 #define IDX_ISO_2 1 227 {"iso-8859-2", ENC_8BIT, 0}, 228 #define IDX_ISO_3 2 229 {"iso-8859-3", ENC_8BIT, 0}, 230 #define IDX_ISO_4 3 231 {"iso-8859-4", ENC_8BIT, 0}, 232 #define IDX_ISO_5 4 233 {"iso-8859-5", ENC_8BIT, 0}, 234 #define IDX_ISO_6 5 235 {"iso-8859-6", ENC_8BIT, 0}, 236 #define IDX_ISO_7 6 237 {"iso-8859-7", ENC_8BIT, 0}, 238 #define IDX_ISO_8 7 239 {"iso-8859-8", ENC_8BIT, 0}, 240 #define IDX_ISO_9 8 241 {"iso-8859-9", ENC_8BIT, 0}, 242 #define IDX_ISO_10 9 243 {"iso-8859-10", ENC_8BIT, 0}, 244 #define IDX_ISO_11 10 245 {"iso-8859-11", ENC_8BIT, 0}, 246 #define IDX_ISO_13 11 247 {"iso-8859-13", ENC_8BIT, 0}, 248 #define IDX_ISO_14 12 249 {"iso-8859-14", ENC_8BIT, 0}, 250 #define IDX_ISO_15 13 251 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0}, 252 #define IDX_KOI8_R 14 253 {"koi8-r", ENC_8BIT, 0}, 254 #define IDX_KOI8_U 15 255 {"koi8-u", ENC_8BIT, 0}, 256 #define IDX_UTF8 16 257 {"utf-8", ENC_UNICODE, 0}, 258 #define IDX_UCS2 17 259 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0}, 260 #define IDX_UCS2LE 18 261 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0}, 262 #define IDX_UTF16 19 263 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0}, 264 #define IDX_UTF16LE 20 265 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0}, 266 #define IDX_UCS4 21 267 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0}, 268 #define IDX_UCS4LE 22 269 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0}, 270 271 /* For debugging DBCS encoding on Unix. */ 272 #define IDX_DEBUG 23 273 {"debug", ENC_DBCS, DBCS_DEBUG}, 274 #define IDX_EUC_JP 24 275 {"euc-jp", ENC_DBCS, DBCS_JPNU}, 276 #define IDX_SJIS 25 277 {"sjis", ENC_DBCS, DBCS_JPN}, 278 #define IDX_EUC_KR 26 279 {"euc-kr", ENC_DBCS, DBCS_KORU}, 280 #define IDX_EUC_CN 27 281 {"euc-cn", ENC_DBCS, DBCS_CHSU}, 282 #define IDX_EUC_TW 28 283 {"euc-tw", ENC_DBCS, DBCS_CHTU}, 284 #define IDX_BIG5 29 285 {"big5", ENC_DBCS, DBCS_CHT}, 286 287 /* MS-DOS and MS-Windows codepages are included here, so that they can be 288 * used on Unix too. Most of them are similar to ISO-8859 encodings, but 289 * not exactly the same. */ 290 #define IDX_CP437 30 291 {"cp437", ENC_8BIT, 437}, /* like iso-8859-1 */ 292 #define IDX_CP737 31 293 {"cp737", ENC_8BIT, 737}, /* like iso-8859-7 */ 294 #define IDX_CP775 32 295 {"cp775", ENC_8BIT, 775}, /* Baltic */ 296 #define IDX_CP850 33 297 {"cp850", ENC_8BIT, 850}, /* like iso-8859-4 */ 298 #define IDX_CP852 34 299 {"cp852", ENC_8BIT, 852}, /* like iso-8859-1 */ 300 #define IDX_CP855 35 301 {"cp855", ENC_8BIT, 855}, /* like iso-8859-2 */ 302 #define IDX_CP857 36 303 {"cp857", ENC_8BIT, 857}, /* like iso-8859-5 */ 304 #define IDX_CP860 37 305 {"cp860", ENC_8BIT, 860}, /* like iso-8859-9 */ 306 #define IDX_CP861 38 307 {"cp861", ENC_8BIT, 861}, /* like iso-8859-1 */ 308 #define IDX_CP862 39 309 {"cp862", ENC_8BIT, 862}, /* like iso-8859-1 */ 310 #define IDX_CP863 40 311 {"cp863", ENC_8BIT, 863}, /* like iso-8859-8 */ 312 #define IDX_CP865 41 313 {"cp865", ENC_8BIT, 865}, /* like iso-8859-1 */ 314 #define IDX_CP866 42 315 {"cp866", ENC_8BIT, 866}, /* like iso-8859-5 */ 316 #define IDX_CP869 43 317 {"cp869", ENC_8BIT, 869}, /* like iso-8859-7 */ 318 #define IDX_CP874 44 319 {"cp874", ENC_8BIT, 874}, /* Thai */ 320 #define IDX_CP932 45 321 {"cp932", ENC_DBCS, DBCS_JPN}, 322 #define IDX_CP936 46 323 {"cp936", ENC_DBCS, DBCS_CHS}, 324 #define IDX_CP949 47 325 {"cp949", ENC_DBCS, DBCS_KOR}, 326 #define IDX_CP950 48 327 {"cp950", ENC_DBCS, DBCS_CHT}, 328 #define IDX_CP1250 49 329 {"cp1250", ENC_8BIT, 1250}, /* Czech, Polish, etc. */ 330 #define IDX_CP1251 50 331 {"cp1251", ENC_8BIT, 1251}, /* Cyrillic */ 332 /* cp1252 is considered to be equal to latin1 */ 333 #define IDX_CP1253 51 334 {"cp1253", ENC_8BIT, 1253}, /* Greek */ 335 #define IDX_CP1254 52 336 {"cp1254", ENC_8BIT, 1254}, /* Turkish */ 337 #define IDX_CP1255 53 338 {"cp1255", ENC_8BIT, 1255}, /* Hebrew */ 339 #define IDX_CP1256 54 340 {"cp1256", ENC_8BIT, 1256}, /* Arabic */ 341 #define IDX_CP1257 55 342 {"cp1257", ENC_8BIT, 1257}, /* Baltic */ 343 #define IDX_CP1258 56 344 {"cp1258", ENC_8BIT, 1258}, /* Vietnamese */ 345 346 #define IDX_MACROMAN 57 347 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, /* Mac OS */ 348 #define IDX_DECMCS 58 349 {"dec-mcs", ENC_8BIT, 0}, /* DEC MCS */ 350 #define IDX_HPROMAN8 59 351 {"hp-roman8", ENC_8BIT, 0}, /* HP Roman8 */ 352 #define IDX_COUNT 60 353 }; 354 355 /* 356 * Aliases for encoding names. 357 */ 358 static struct 359 { char *name; int canon;} 360 enc_alias_table[] = 361 { 362 {"ansi", IDX_LATIN_1}, 363 {"iso-8859-1", IDX_LATIN_1}, 364 {"latin2", IDX_ISO_2}, 365 {"latin3", IDX_ISO_3}, 366 {"latin4", IDX_ISO_4}, 367 {"cyrillic", IDX_ISO_5}, 368 {"arabic", IDX_ISO_6}, 369 {"greek", IDX_ISO_7}, 370 #ifdef WIN3264 371 {"hebrew", IDX_CP1255}, 372 #else 373 {"hebrew", IDX_ISO_8}, 374 #endif 375 {"latin5", IDX_ISO_9}, 376 {"turkish", IDX_ISO_9}, /* ? */ 377 {"latin6", IDX_ISO_10}, 378 {"nordic", IDX_ISO_10}, /* ? */ 379 {"thai", IDX_ISO_11}, /* ? */ 380 {"latin7", IDX_ISO_13}, 381 {"latin8", IDX_ISO_14}, 382 {"latin9", IDX_ISO_15}, 383 {"utf8", IDX_UTF8}, 384 {"unicode", IDX_UCS2}, 385 {"ucs2", IDX_UCS2}, 386 {"ucs2be", IDX_UCS2}, 387 {"ucs-2be", IDX_UCS2}, 388 {"ucs2le", IDX_UCS2LE}, 389 {"utf16", IDX_UTF16}, 390 {"utf16be", IDX_UTF16}, 391 {"utf-16be", IDX_UTF16}, 392 {"utf16le", IDX_UTF16LE}, 393 {"ucs4", IDX_UCS4}, 394 {"ucs4be", IDX_UCS4}, 395 {"ucs-4be", IDX_UCS4}, 396 {"ucs4le", IDX_UCS4LE}, 397 {"utf32", IDX_UCS4}, 398 {"utf-32", IDX_UCS4}, 399 {"utf32be", IDX_UCS4}, 400 {"utf-32be", IDX_UCS4}, 401 {"utf32le", IDX_UCS4LE}, 402 {"utf-32le", IDX_UCS4LE}, 403 {"932", IDX_CP932}, 404 {"949", IDX_CP949}, 405 {"936", IDX_CP936}, 406 {"gbk", IDX_CP936}, 407 {"950", IDX_CP950}, 408 {"eucjp", IDX_EUC_JP}, 409 {"unix-jis", IDX_EUC_JP}, 410 {"ujis", IDX_EUC_JP}, 411 {"shift-jis", IDX_SJIS}, 412 {"pck", IDX_SJIS}, /* Sun: PCK */ 413 {"euckr", IDX_EUC_KR}, 414 {"5601", IDX_EUC_KR}, /* Sun: KS C 5601 */ 415 {"euccn", IDX_EUC_CN}, 416 {"gb2312", IDX_EUC_CN}, 417 {"euctw", IDX_EUC_TW}, 418 #if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS) 419 {"japan", IDX_CP932}, 420 {"korea", IDX_CP949}, 421 {"prc", IDX_CP936}, 422 {"chinese", IDX_CP936}, 423 {"taiwan", IDX_CP950}, 424 {"big5", IDX_CP950}, 425 #else 426 {"japan", IDX_EUC_JP}, 427 {"korea", IDX_EUC_KR}, 428 {"prc", IDX_EUC_CN}, 429 {"chinese", IDX_EUC_CN}, 430 {"taiwan", IDX_EUC_TW}, 431 {"cp950", IDX_BIG5}, 432 {"950", IDX_BIG5}, 433 #endif 434 {"mac", IDX_MACROMAN}, 435 {"mac-roman", IDX_MACROMAN}, 436 {NULL, 0} 437 }; 438 439 #ifndef CP_UTF8 440 # define CP_UTF8 65001 /* magic number from winnls.h */ 441 #endif 442 443 /* 444 * Find encoding "name" in the list of canonical encoding names. 445 * Returns -1 if not found. 446 */ 447 static int 448 enc_canon_search(char_u *name) 449 { 450 int i; 451 452 for (i = 0; i < IDX_COUNT; ++i) 453 if (STRCMP(name, enc_canon_table[i].name) == 0) 454 return i; 455 return -1; 456 } 457 458 #endif 459 460 #if defined(FEAT_MBYTE) || defined(PROTO) 461 462 /* 463 * Find canonical encoding "name" in the list and return its properties. 464 * Returns 0 if not found. 465 */ 466 int 467 enc_canon_props(char_u *name) 468 { 469 int i; 470 471 i = enc_canon_search(name); 472 if (i >= 0) 473 return enc_canon_table[i].prop; 474 #ifdef WIN3264 475 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2])) 476 { 477 CPINFO cpinfo; 478 479 /* Get info on this codepage to find out what it is. */ 480 if (GetCPInfo(atoi((char *)name + 2), &cpinfo) != 0) 481 { 482 if (cpinfo.MaxCharSize == 1) /* some single-byte encoding */ 483 return ENC_8BIT; 484 if (cpinfo.MaxCharSize == 2 485 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 486 /* must be a DBCS encoding */ 487 return ENC_DBCS; 488 } 489 return 0; 490 } 491 #endif 492 if (STRNCMP(name, "2byte-", 6) == 0) 493 return ENC_DBCS; 494 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0) 495 return ENC_8BIT; 496 return 0; 497 } 498 499 /* 500 * Set up for using multi-byte characters. 501 * Called in three cases: 502 * - by main() to initialize (p_enc == NULL) 503 * - by set_init_1() after 'encoding' was set to its default. 504 * - by do_set() when 'encoding' has been set. 505 * p_enc must have been passed through enc_canonize() already. 506 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags. 507 * Fills mb_bytelen_tab[] and returns NULL when there are no problems. 508 * When there is something wrong: Returns an error message and doesn't change 509 * anything. 510 */ 511 char_u * 512 mb_init(void) 513 { 514 int i; 515 int idx; 516 int n; 517 int enc_dbcs_new = 0; 518 #if defined(USE_ICONV) && !defined(WIN3264) && !defined(WIN32UNIX) \ 519 && !defined(MACOS) 520 # define LEN_FROM_CONV 521 vimconv_T vimconv; 522 char_u *p; 523 #endif 524 525 if (p_enc == NULL) 526 { 527 /* Just starting up: set the whole table to one's. */ 528 for (i = 0; i < 256; ++i) 529 mb_bytelen_tab[i] = 1; 530 input_conv.vc_type = CONV_NONE; 531 input_conv.vc_factor = 1; 532 output_conv.vc_type = CONV_NONE; 533 return NULL; 534 } 535 536 #ifdef WIN3264 537 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2])) 538 { 539 CPINFO cpinfo; 540 541 /* Get info on this codepage to find out what it is. */ 542 if (GetCPInfo(atoi((char *)p_enc + 2), &cpinfo) != 0) 543 { 544 if (cpinfo.MaxCharSize == 1) 545 { 546 /* some single-byte encoding */ 547 enc_unicode = 0; 548 enc_utf8 = FALSE; 549 } 550 else if (cpinfo.MaxCharSize == 2 551 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 552 { 553 /* must be a DBCS encoding, check below */ 554 enc_dbcs_new = atoi((char *)p_enc + 2); 555 } 556 else 557 goto codepage_invalid; 558 } 559 else if (GetLastError() == ERROR_INVALID_PARAMETER) 560 { 561 codepage_invalid: 562 return (char_u *)N_("E543: Not a valid codepage"); 563 } 564 } 565 #endif 566 else if (STRNCMP(p_enc, "8bit-", 5) == 0 567 || STRNCMP(p_enc, "iso-8859-", 9) == 0) 568 { 569 /* Accept any "8bit-" or "iso-8859-" name. */ 570 enc_unicode = 0; 571 enc_utf8 = FALSE; 572 } 573 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 574 { 575 #ifdef WIN3264 576 /* Windows: accept only valid codepage numbers, check below. */ 577 if (p_enc[6] != 'c' || p_enc[7] != 'p' 578 || (enc_dbcs_new = atoi((char *)p_enc + 8)) == 0) 579 return e_invarg; 580 #else 581 /* Unix: accept any "2byte-" name, assume current locale. */ 582 enc_dbcs_new = DBCS_2BYTE; 583 #endif 584 } 585 else if ((idx = enc_canon_search(p_enc)) >= 0) 586 { 587 i = enc_canon_table[idx].prop; 588 if (i & ENC_UNICODE) 589 { 590 /* Unicode */ 591 enc_utf8 = TRUE; 592 if (i & (ENC_2BYTE | ENC_2WORD)) 593 enc_unicode = 2; 594 else if (i & ENC_4BYTE) 595 enc_unicode = 4; 596 else 597 enc_unicode = 0; 598 } 599 else if (i & ENC_DBCS) 600 { 601 /* 2byte, handle below */ 602 enc_dbcs_new = enc_canon_table[idx].codepage; 603 } 604 else 605 { 606 /* Must be 8-bit. */ 607 enc_unicode = 0; 608 enc_utf8 = FALSE; 609 } 610 } 611 else /* Don't know what encoding this is, reject it. */ 612 return e_invarg; 613 614 if (enc_dbcs_new != 0) 615 { 616 #ifdef WIN3264 617 /* Check if the DBCS code page is OK. */ 618 if (!IsValidCodePage(enc_dbcs_new)) 619 goto codepage_invalid; 620 #endif 621 enc_unicode = 0; 622 enc_utf8 = FALSE; 623 } 624 enc_dbcs = enc_dbcs_new; 625 has_mbyte = (enc_dbcs != 0 || enc_utf8); 626 627 #if defined(WIN3264) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 628 enc_codepage = encname2codepage(p_enc); 629 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0); 630 #endif 631 632 /* Detect an encoding that uses latin1 characters. */ 633 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0 634 || STRCMP(p_enc, "iso-8859-15") == 0); 635 636 /* 637 * Set the function pointers. 638 */ 639 if (enc_utf8) 640 { 641 mb_ptr2len = utfc_ptr2len; 642 mb_ptr2len_len = utfc_ptr2len_len; 643 mb_char2len = utf_char2len; 644 mb_char2bytes = utf_char2bytes; 645 mb_ptr2cells = utf_ptr2cells; 646 mb_ptr2cells_len = utf_ptr2cells_len; 647 mb_char2cells = utf_char2cells; 648 mb_off2cells = utf_off2cells; 649 mb_ptr2char = utf_ptr2char; 650 mb_head_off = utf_head_off; 651 } 652 else if (enc_dbcs != 0) 653 { 654 mb_ptr2len = dbcs_ptr2len; 655 mb_ptr2len_len = dbcs_ptr2len_len; 656 mb_char2len = dbcs_char2len; 657 mb_char2bytes = dbcs_char2bytes; 658 mb_ptr2cells = dbcs_ptr2cells; 659 mb_ptr2cells_len = dbcs_ptr2cells_len; 660 mb_char2cells = dbcs_char2cells; 661 mb_off2cells = dbcs_off2cells; 662 mb_ptr2char = dbcs_ptr2char; 663 mb_head_off = dbcs_head_off; 664 } 665 else 666 { 667 mb_ptr2len = latin_ptr2len; 668 mb_ptr2len_len = latin_ptr2len_len; 669 mb_char2len = latin_char2len; 670 mb_char2bytes = latin_char2bytes; 671 mb_ptr2cells = latin_ptr2cells; 672 mb_ptr2cells_len = latin_ptr2cells_len; 673 mb_char2cells = latin_char2cells; 674 mb_off2cells = latin_off2cells; 675 mb_ptr2char = latin_ptr2char; 676 mb_head_off = latin_head_off; 677 } 678 679 /* 680 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN(). 681 */ 682 #ifdef LEN_FROM_CONV 683 /* When 'encoding' is different from the current locale mblen() won't 684 * work. Use conversion to "utf-8" instead. */ 685 vimconv.vc_type = CONV_NONE; 686 if (enc_dbcs) 687 { 688 p = enc_locale(); 689 if (p == NULL || STRCMP(p, p_enc) != 0) 690 { 691 convert_setup(&vimconv, p_enc, (char_u *)"utf-8"); 692 vimconv.vc_fail = TRUE; 693 } 694 vim_free(p); 695 } 696 #endif 697 698 for (i = 0; i < 256; ++i) 699 { 700 /* Our own function to reliably check the length of UTF-8 characters, 701 * independent of mblen(). */ 702 if (enc_utf8) 703 n = utf8len_tab[i]; 704 else if (enc_dbcs == 0) 705 n = 1; 706 else 707 { 708 #if defined(WIN3264) || defined(WIN32UNIX) 709 /* enc_dbcs is set by setting 'fileencoding'. It becomes a Windows 710 * CodePage identifier, which we can pass directly in to Windows 711 * API */ 712 n = IsDBCSLeadByteEx(enc_dbcs, (WINBYTE)i) ? 2 : 1; 713 #else 714 # if defined(MACOS) || defined(__amigaos4__) || defined(__ANDROID__) 715 /* 716 * if mblen() is not available, character which MSB is turned on 717 * are treated as leading byte character. (note : This assumption 718 * is not always true.) 719 */ 720 n = (i & 0x80) ? 2 : 1; 721 # else 722 char buf[MB_MAXBYTES + 1]; 723 # ifdef X_LOCALE 724 # ifndef mblen 725 # define mblen _Xmblen 726 # endif 727 # endif 728 if (i == NUL) /* just in case mblen() can't handle "" */ 729 n = 1; 730 else 731 { 732 buf[0] = i; 733 buf[1] = 0; 734 #ifdef LEN_FROM_CONV 735 if (vimconv.vc_type != CONV_NONE) 736 { 737 /* 738 * string_convert() should fail when converting the first 739 * byte of a double-byte character. 740 */ 741 p = string_convert(&vimconv, (char_u *)buf, NULL); 742 if (p != NULL) 743 { 744 vim_free(p); 745 n = 1; 746 } 747 else 748 n = 2; 749 } 750 else 751 #endif 752 { 753 /* 754 * mblen() should return -1 for invalid (means the leading 755 * multibyte) character. However there are some platforms 756 * where mblen() returns 0 for invalid character. 757 * Therefore, following condition includes 0. 758 */ 759 ignored = mblen(NULL, 0); /* First reset the state. */ 760 if (mblen(buf, (size_t)1) <= 0) 761 n = 2; 762 else 763 n = 1; 764 } 765 } 766 # endif 767 #endif 768 } 769 770 mb_bytelen_tab[i] = n; 771 } 772 773 #ifdef LEN_FROM_CONV 774 convert_setup(&vimconv, NULL, NULL); 775 #endif 776 777 /* The cell width depends on the type of multi-byte characters. */ 778 (void)init_chartab(); 779 780 /* When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] */ 781 screenalloc(FALSE); 782 783 /* When using Unicode, set default for 'fileencodings'. */ 784 if (enc_utf8 && !option_was_set((char_u *)"fencs")) 785 set_string_option_direct((char_u *)"fencs", -1, 786 (char_u *)"ucs-bom,utf-8,default,latin1", OPT_FREE, 0); 787 788 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT) 789 /* GNU gettext 0.10.37 supports this feature: set the codeset used for 790 * translated messages independently from the current locale. */ 791 (void)bind_textdomain_codeset(VIMPACKAGE, 792 enc_utf8 ? "utf-8" : (char *)p_enc); 793 #endif 794 795 #ifdef WIN32 796 /* When changing 'encoding' while starting up, then convert the command 797 * line arguments from the active codepage to 'encoding'. */ 798 if (starting != 0) 799 fix_arg_enc(); 800 #endif 801 802 #ifdef FEAT_AUTOCMD 803 /* Fire an autocommand to let people do custom font setup. This must be 804 * after Vim has been setup for the new encoding. */ 805 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); 806 #endif 807 808 #ifdef FEAT_SPELL 809 /* Need to reload spell dictionaries */ 810 spell_reload(); 811 #endif 812 813 return NULL; 814 } 815 816 /* 817 * Return the size of the BOM for the current buffer: 818 * 0 - no BOM 819 * 2 - UCS-2 or UTF-16 BOM 820 * 4 - UCS-4 BOM 821 * 3 - UTF-8 BOM 822 */ 823 int 824 bomb_size(void) 825 { 826 int n = 0; 827 828 if (curbuf->b_p_bomb && !curbuf->b_p_bin) 829 { 830 if (*curbuf->b_p_fenc == NUL) 831 { 832 if (enc_utf8) 833 { 834 if (enc_unicode != 0) 835 n = enc_unicode; 836 else 837 n = 3; 838 } 839 } 840 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) 841 n = 3; 842 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 843 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) 844 n = 2; 845 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) 846 n = 4; 847 } 848 return n; 849 } 850 851 /* 852 * Remove all BOM from "s" by moving remaining text. 853 */ 854 void 855 remove_bom(char_u *s) 856 { 857 if (enc_utf8) 858 { 859 char_u *p = s; 860 861 while ((p = vim_strbyte(p, 0xef)) != NULL) 862 { 863 if (p[1] == 0xbb && p[2] == 0xbf) 864 STRMOVE(p, p + 3); 865 else 866 ++p; 867 } 868 } 869 } 870 871 /* 872 * Get class of pointer: 873 * 0 for blank or NUL 874 * 1 for punctuation 875 * 2 for an (ASCII) word character 876 * >2 for other word characters 877 */ 878 int 879 mb_get_class(char_u *p) 880 { 881 return mb_get_class_buf(p, curbuf); 882 } 883 884 int 885 mb_get_class_buf(char_u *p, buf_T *buf) 886 { 887 if (MB_BYTE2LEN(p[0]) == 1) 888 { 889 if (p[0] == NUL || vim_iswhite(p[0])) 890 return 0; 891 if (vim_iswordc_buf(p[0], buf)) 892 return 2; 893 return 1; 894 } 895 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) 896 return dbcs_class(p[0], p[1]); 897 if (enc_utf8) 898 return utf_class(utf_ptr2char(p)); 899 return 0; 900 } 901 902 /* 903 * Get class of a double-byte character. This always returns 3 or bigger. 904 * TODO: Should return 1 for punctuation. 905 */ 906 int 907 dbcs_class(unsigned lead, unsigned trail) 908 { 909 switch (enc_dbcs) 910 { 911 /* please add classify routine for your language in here */ 912 913 case DBCS_JPNU: /* ? */ 914 case DBCS_JPN: 915 { 916 /* JIS code classification */ 917 unsigned char lb = lead; 918 unsigned char tb = trail; 919 920 /* convert process code to JIS */ 921 # if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS) 922 /* process code is SJIS */ 923 if (lb <= 0x9f) 924 lb = (lb - 0x81) * 2 + 0x21; 925 else 926 lb = (lb - 0xc1) * 2 + 0x21; 927 if (tb <= 0x7e) 928 tb -= 0x1f; 929 else if (tb <= 0x9e) 930 tb -= 0x20; 931 else 932 { 933 tb -= 0x7e; 934 lb += 1; 935 } 936 # else 937 /* 938 * XXX: Code page identification can not use with all 939 * system! So, some other encoding information 940 * will be needed. 941 * In japanese: SJIS,EUC,UNICODE,(JIS) 942 * Note that JIS-code system don't use as 943 * process code in most system because it uses 944 * escape sequences(JIS is context depend encoding). 945 */ 946 /* assume process code is JAPANESE-EUC */ 947 lb &= 0x7f; 948 tb &= 0x7f; 949 # endif 950 /* exceptions */ 951 switch (lb << 8 | tb) 952 { 953 case 0x2121: /* ZENKAKU space */ 954 return 0; 955 case 0x2122: /* TOU-TEN (Japanese comma) */ 956 case 0x2123: /* KU-TEN (Japanese period) */ 957 case 0x2124: /* ZENKAKU comma */ 958 case 0x2125: /* ZENKAKU period */ 959 return 1; 960 case 0x213c: /* prolongedsound handled as KATAKANA */ 961 return 13; 962 } 963 /* sieved by KU code */ 964 switch (lb) 965 { 966 case 0x21: 967 case 0x22: 968 /* special symbols */ 969 return 10; 970 case 0x23: 971 /* alpha-numeric */ 972 return 11; 973 case 0x24: 974 /* hiragana */ 975 return 12; 976 case 0x25: 977 /* katakana */ 978 return 13; 979 case 0x26: 980 /* greek */ 981 return 14; 982 case 0x27: 983 /* russian */ 984 return 15; 985 case 0x28: 986 /* lines */ 987 return 16; 988 default: 989 /* kanji */ 990 return 17; 991 } 992 } 993 994 case DBCS_KORU: /* ? */ 995 case DBCS_KOR: 996 { 997 /* KS code classification */ 998 unsigned char c1 = lead; 999 unsigned char c2 = trail; 1000 1001 /* 1002 * 20 : Hangul 1003 * 21 : Hanja 1004 * 22 : Symbols 1005 * 23 : Alpha-numeric/Roman Letter (Full width) 1006 * 24 : Hangul Letter(Alphabet) 1007 * 25 : Roman Numeral/Greek Letter 1008 * 26 : Box Drawings 1009 * 27 : Unit Symbols 1010 * 28 : Circled/Parenthesized Letter 1011 * 29 : Hiragana/Katakana 1012 * 30 : Cyrillic Letter 1013 */ 1014 1015 if (c1 >= 0xB0 && c1 <= 0xC8) 1016 /* Hangul */ 1017 return 20; 1018 #if defined(WIN3264) || defined(WIN32UNIX) 1019 else if (c1 <= 0xA0 || c2 <= 0xA0) 1020 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */ 1021 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE 1022 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 1023 */ 1024 return 20; 1025 #endif 1026 1027 else if (c1 >= 0xCA && c1 <= 0xFD) 1028 /* Hanja */ 1029 return 21; 1030 else switch (c1) 1031 { 1032 case 0xA1: 1033 case 0xA2: 1034 /* Symbols */ 1035 return 22; 1036 case 0xA3: 1037 /* Alpha-numeric */ 1038 return 23; 1039 case 0xA4: 1040 /* Hangul Letter(Alphabet) */ 1041 return 24; 1042 case 0xA5: 1043 /* Roman Numeral/Greek Letter */ 1044 return 25; 1045 case 0xA6: 1046 /* Box Drawings */ 1047 return 26; 1048 case 0xA7: 1049 /* Unit Symbols */ 1050 return 27; 1051 case 0xA8: 1052 case 0xA9: 1053 if (c2 <= 0xAF) 1054 return 25; /* Roman Letter */ 1055 else if (c2 >= 0xF6) 1056 return 22; /* Symbols */ 1057 else 1058 /* Circled/Parenthesized Letter */ 1059 return 28; 1060 case 0xAA: 1061 case 0xAB: 1062 /* Hiragana/Katakana */ 1063 return 29; 1064 case 0xAC: 1065 /* Cyrillic Letter */ 1066 return 30; 1067 } 1068 } 1069 default: 1070 break; 1071 } 1072 return 3; 1073 } 1074 1075 /* 1076 * mb_char2len() function pointer. 1077 * Return length in bytes of character "c". 1078 * Returns 1 for a single-byte character. 1079 */ 1080 int 1081 latin_char2len(int c UNUSED) 1082 { 1083 return 1; 1084 } 1085 1086 static int 1087 dbcs_char2len( 1088 int c) 1089 { 1090 if (c >= 0x100) 1091 return 2; 1092 return 1; 1093 } 1094 1095 /* 1096 * mb_char2bytes() function pointer. 1097 * Convert a character to its bytes. 1098 * Returns the length in bytes. 1099 */ 1100 int 1101 latin_char2bytes(int c, char_u *buf) 1102 { 1103 buf[0] = c; 1104 return 1; 1105 } 1106 1107 static int 1108 dbcs_char2bytes(int c, char_u *buf) 1109 { 1110 if (c >= 0x100) 1111 { 1112 buf[0] = (unsigned)c >> 8; 1113 buf[1] = c; 1114 /* Never use a NUL byte, it causes lots of trouble. It's an invalid 1115 * character anyway. */ 1116 if (buf[1] == NUL) 1117 buf[1] = '\n'; 1118 return 2; 1119 } 1120 buf[0] = c; 1121 return 1; 1122 } 1123 1124 /* 1125 * mb_ptr2len() function pointer. 1126 * Get byte length of character at "*p" but stop at a NUL. 1127 * For UTF-8 this includes following composing characters. 1128 * Returns 0 when *p is NUL. 1129 */ 1130 int 1131 latin_ptr2len(char_u *p) 1132 { 1133 return MB_BYTE2LEN(*p); 1134 } 1135 1136 static int 1137 dbcs_ptr2len( 1138 char_u *p) 1139 { 1140 int len; 1141 1142 /* Check if second byte is not missing. */ 1143 len = MB_BYTE2LEN(*p); 1144 if (len == 2 && p[1] == NUL) 1145 len = 1; 1146 return len; 1147 } 1148 1149 /* 1150 * mb_ptr2len_len() function pointer. 1151 * Like mb_ptr2len(), but limit to read "size" bytes. 1152 * Returns 0 for an empty string. 1153 * Returns 1 for an illegal char or an incomplete byte sequence. 1154 */ 1155 int 1156 latin_ptr2len_len(char_u *p, int size) 1157 { 1158 if (size < 1 || *p == NUL) 1159 return 0; 1160 return 1; 1161 } 1162 1163 static int 1164 dbcs_ptr2len_len(char_u *p, int size) 1165 { 1166 int len; 1167 1168 if (size < 1 || *p == NUL) 1169 return 0; 1170 if (size == 1) 1171 return 1; 1172 /* Check that second byte is not missing. */ 1173 len = MB_BYTE2LEN(*p); 1174 if (len == 2 && p[1] == NUL) 1175 len = 1; 1176 return len; 1177 } 1178 1179 struct interval 1180 { 1181 long first; 1182 long last; 1183 }; 1184 1185 /* 1186 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]". 1187 */ 1188 static int 1189 intable(struct interval *table, size_t size, int c) 1190 { 1191 int mid, bot, top; 1192 1193 /* first quick check for Latin1 etc. characters */ 1194 if (c < table[0].first) 1195 return FALSE; 1196 1197 /* binary search in table */ 1198 bot = 0; 1199 top = (int)(size / sizeof(struct interval) - 1); 1200 while (top >= bot) 1201 { 1202 mid = (bot + top) / 2; 1203 if (table[mid].last < c) 1204 bot = mid + 1; 1205 else if (table[mid].first > c) 1206 top = mid - 1; 1207 else 1208 return TRUE; 1209 } 1210 return FALSE; 1211 } 1212 1213 /* 1214 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1215 * Returns 4 or 6 for an unprintable character. 1216 * Is only correct for characters >= 0x80. 1217 * When p_ambw is "double", return 2 for a character with East Asian Width 1218 * class 'A'(mbiguous). 1219 */ 1220 int 1221 utf_char2cells(int c) 1222 { 1223 /* Sorted list of non-overlapping intervals of East Asian double width 1224 * characters, generated with ../runtime/tools/unicode.vim. */ 1225 static struct interval doublewidth[] = 1226 { 1227 {0x1100, 0x115f}, 1228 {0x2329, 0x232a}, 1229 {0x2e80, 0x2e99}, 1230 {0x2e9b, 0x2ef3}, 1231 {0x2f00, 0x2fd5}, 1232 {0x2ff0, 0x2ffb}, 1233 {0x3000, 0x303e}, 1234 {0x3041, 0x3096}, 1235 {0x3099, 0x30ff}, 1236 {0x3105, 0x312d}, 1237 {0x3131, 0x318e}, 1238 {0x3190, 0x31ba}, 1239 {0x31c0, 0x31e3}, 1240 {0x31f0, 0x321e}, 1241 {0x3220, 0x3247}, 1242 {0x3250, 0x32fe}, 1243 {0x3300, 0x4dbf}, 1244 {0x4e00, 0xa48c}, 1245 {0xa490, 0xa4c6}, 1246 {0xa960, 0xa97c}, 1247 {0xac00, 0xd7a3}, 1248 {0xf900, 0xfaff}, 1249 {0xfe10, 0xfe19}, 1250 {0xfe30, 0xfe52}, 1251 {0xfe54, 0xfe66}, 1252 {0xfe68, 0xfe6b}, 1253 {0xff01, 0xff60}, 1254 {0xffe0, 0xffe6}, 1255 {0x1b000, 0x1b001}, 1256 {0x1f200, 0x1f202}, 1257 {0x1f210, 0x1f23a}, 1258 {0x1f240, 0x1f248}, 1259 {0x1f250, 0x1f251}, 1260 {0x20000, 0x2fffd}, 1261 {0x30000, 0x3fffd} 1262 }; 1263 1264 /* Sorted list of non-overlapping intervals of East Asian Ambiguous 1265 * characters, generated with ../runtime/tools/unicode.vim. */ 1266 static struct interval ambiguous[] = 1267 { 1268 {0x00a1, 0x00a1}, 1269 {0x00a4, 0x00a4}, 1270 {0x00a7, 0x00a8}, 1271 {0x00aa, 0x00aa}, 1272 {0x00ad, 0x00ae}, 1273 {0x00b0, 0x00b4}, 1274 {0x00b6, 0x00ba}, 1275 {0x00bc, 0x00bf}, 1276 {0x00c6, 0x00c6}, 1277 {0x00d0, 0x00d0}, 1278 {0x00d7, 0x00d8}, 1279 {0x00de, 0x00e1}, 1280 {0x00e6, 0x00e6}, 1281 {0x00e8, 0x00ea}, 1282 {0x00ec, 0x00ed}, 1283 {0x00f0, 0x00f0}, 1284 {0x00f2, 0x00f3}, 1285 {0x00f7, 0x00fa}, 1286 {0x00fc, 0x00fc}, 1287 {0x00fe, 0x00fe}, 1288 {0x0101, 0x0101}, 1289 {0x0111, 0x0111}, 1290 {0x0113, 0x0113}, 1291 {0x011b, 0x011b}, 1292 {0x0126, 0x0127}, 1293 {0x012b, 0x012b}, 1294 {0x0131, 0x0133}, 1295 {0x0138, 0x0138}, 1296 {0x013f, 0x0142}, 1297 {0x0144, 0x0144}, 1298 {0x0148, 0x014b}, 1299 {0x014d, 0x014d}, 1300 {0x0152, 0x0153}, 1301 {0x0166, 0x0167}, 1302 {0x016b, 0x016b}, 1303 {0x01ce, 0x01ce}, 1304 {0x01d0, 0x01d0}, 1305 {0x01d2, 0x01d2}, 1306 {0x01d4, 0x01d4}, 1307 {0x01d6, 0x01d6}, 1308 {0x01d8, 0x01d8}, 1309 {0x01da, 0x01da}, 1310 {0x01dc, 0x01dc}, 1311 {0x0251, 0x0251}, 1312 {0x0261, 0x0261}, 1313 {0x02c4, 0x02c4}, 1314 {0x02c7, 0x02c7}, 1315 {0x02c9, 0x02cb}, 1316 {0x02cd, 0x02cd}, 1317 {0x02d0, 0x02d0}, 1318 {0x02d8, 0x02db}, 1319 {0x02dd, 0x02dd}, 1320 {0x02df, 0x02df}, 1321 {0x0300, 0x036f}, 1322 {0x0391, 0x03a1}, 1323 {0x03a3, 0x03a9}, 1324 {0x03b1, 0x03c1}, 1325 {0x03c3, 0x03c9}, 1326 {0x0401, 0x0401}, 1327 {0x0410, 0x044f}, 1328 {0x0451, 0x0451}, 1329 {0x2010, 0x2010}, 1330 {0x2013, 0x2016}, 1331 {0x2018, 0x2019}, 1332 {0x201c, 0x201d}, 1333 {0x2020, 0x2022}, 1334 {0x2024, 0x2027}, 1335 {0x2030, 0x2030}, 1336 {0x2032, 0x2033}, 1337 {0x2035, 0x2035}, 1338 {0x203b, 0x203b}, 1339 {0x203e, 0x203e}, 1340 {0x2074, 0x2074}, 1341 {0x207f, 0x207f}, 1342 {0x2081, 0x2084}, 1343 {0x20ac, 0x20ac}, 1344 {0x2103, 0x2103}, 1345 {0x2105, 0x2105}, 1346 {0x2109, 0x2109}, 1347 {0x2113, 0x2113}, 1348 {0x2116, 0x2116}, 1349 {0x2121, 0x2122}, 1350 {0x2126, 0x2126}, 1351 {0x212b, 0x212b}, 1352 {0x2153, 0x2154}, 1353 {0x215b, 0x215e}, 1354 {0x2160, 0x216b}, 1355 {0x2170, 0x2179}, 1356 {0x2189, 0x2189}, 1357 {0x2190, 0x2199}, 1358 {0x21b8, 0x21b9}, 1359 {0x21d2, 0x21d2}, 1360 {0x21d4, 0x21d4}, 1361 {0x21e7, 0x21e7}, 1362 {0x2200, 0x2200}, 1363 {0x2202, 0x2203}, 1364 {0x2207, 0x2208}, 1365 {0x220b, 0x220b}, 1366 {0x220f, 0x220f}, 1367 {0x2211, 0x2211}, 1368 {0x2215, 0x2215}, 1369 {0x221a, 0x221a}, 1370 {0x221d, 0x2220}, 1371 {0x2223, 0x2223}, 1372 {0x2225, 0x2225}, 1373 {0x2227, 0x222c}, 1374 {0x222e, 0x222e}, 1375 {0x2234, 0x2237}, 1376 {0x223c, 0x223d}, 1377 {0x2248, 0x2248}, 1378 {0x224c, 0x224c}, 1379 {0x2252, 0x2252}, 1380 {0x2260, 0x2261}, 1381 {0x2264, 0x2267}, 1382 {0x226a, 0x226b}, 1383 {0x226e, 0x226f}, 1384 {0x2282, 0x2283}, 1385 {0x2286, 0x2287}, 1386 {0x2295, 0x2295}, 1387 {0x2299, 0x2299}, 1388 {0x22a5, 0x22a5}, 1389 {0x22bf, 0x22bf}, 1390 {0x2312, 0x2312}, 1391 {0x2460, 0x24e9}, 1392 {0x24eb, 0x254b}, 1393 {0x2550, 0x2573}, 1394 {0x2580, 0x258f}, 1395 {0x2592, 0x2595}, 1396 {0x25a0, 0x25a1}, 1397 {0x25a3, 0x25a9}, 1398 {0x25b2, 0x25b3}, 1399 {0x25b6, 0x25b7}, 1400 {0x25bc, 0x25bd}, 1401 {0x25c0, 0x25c1}, 1402 {0x25c6, 0x25c8}, 1403 {0x25cb, 0x25cb}, 1404 {0x25ce, 0x25d1}, 1405 {0x25e2, 0x25e5}, 1406 {0x25ef, 0x25ef}, 1407 {0x2605, 0x2606}, 1408 {0x2609, 0x2609}, 1409 {0x260e, 0x260f}, 1410 {0x2614, 0x2615}, 1411 {0x261c, 0x261c}, 1412 {0x261e, 0x261e}, 1413 {0x2640, 0x2640}, 1414 {0x2642, 0x2642}, 1415 {0x2660, 0x2661}, 1416 {0x2663, 0x2665}, 1417 {0x2667, 0x266a}, 1418 {0x266c, 0x266d}, 1419 {0x266f, 0x266f}, 1420 {0x269e, 0x269f}, 1421 {0x26be, 0x26bf}, 1422 {0x26c4, 0x26cd}, 1423 {0x26cf, 0x26e1}, 1424 {0x26e3, 0x26e3}, 1425 {0x26e8, 0x26ff}, 1426 {0x273d, 0x273d}, 1427 {0x2757, 0x2757}, 1428 {0x2776, 0x277f}, 1429 {0x2b55, 0x2b59}, 1430 {0x3248, 0x324f}, 1431 {0xe000, 0xf8ff}, 1432 {0xfe00, 0xfe0f}, 1433 {0xfffd, 0xfffd}, 1434 {0x1f100, 0x1f10a}, 1435 {0x1f110, 0x1f12d}, 1436 {0x1f130, 0x1f169}, 1437 {0x1f170, 0x1f19a}, 1438 {0xe0100, 0xe01ef}, 1439 {0xf0000, 0xffffd}, 1440 {0x100000, 0x10fffd} 1441 }; 1442 1443 /* Sorted list of non-overlapping intervals of Emoji characters that don't 1444 * have ambiguous or double width, 1445 * based on http://unicode.org/emoji/charts/emoji-list.html */ 1446 static struct interval emoji_width[] = 1447 { 1448 {0x1f004, 0x1f004}, 1449 {0x1f0cf, 0x1f0cf}, 1450 {0x1f1e6, 0x1f1ff}, 1451 {0x1f300, 0x1f320}, 1452 {0x1f330, 0x1f335}, 1453 {0x1f337, 0x1f37c}, 1454 {0x1f380, 0x1f393}, 1455 {0x1f3a0, 0x1f3c4}, 1456 {0x1f3c6, 0x1f3ca}, 1457 {0x1f3e0, 0x1f3f0}, 1458 {0x1f400, 0x1f43e}, 1459 {0x1f440, 0x1f440}, 1460 {0x1f442, 0x1f4f7}, 1461 {0x1f4f9, 0x1f4fc}, 1462 {0x1f500, 0x1f53d}, 1463 {0x1f550, 0x1f567}, 1464 {0x1f5fb, 0x1f640}, 1465 {0x1f645, 0x1f64f}, 1466 {0x1f680, 0x1f6c5} 1467 }; 1468 1469 if (c >= 0x100) 1470 { 1471 #ifdef USE_WCHAR_FUNCTIONS 1472 /* 1473 * Assume the library function wcwidth() works better than our own 1474 * stuff. It should return 1 for ambiguous width chars! 1475 */ 1476 int n = wcwidth(c); 1477 1478 if (n < 0) 1479 return 6; /* unprintable, displays <xxxx> */ 1480 if (n > 1) 1481 return n; 1482 #else 1483 if (!utf_printable(c)) 1484 return 6; /* unprintable, displays <xxxx> */ 1485 if (intable(doublewidth, sizeof(doublewidth), c)) 1486 return 2; 1487 #endif 1488 if (p_emoji && intable(emoji_width, sizeof(emoji_width), c)) 1489 return 2; 1490 } 1491 1492 /* Characters below 0x100 are influenced by 'isprint' option */ 1493 else if (c >= 0x80 && !vim_isprintc(c)) 1494 return 4; /* unprintable, displays <xx> */ 1495 1496 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c)) 1497 return 2; 1498 1499 return 1; 1500 } 1501 1502 /* 1503 * mb_ptr2cells() function pointer. 1504 * Return the number of display cells character at "*p" occupies. 1505 * This doesn't take care of unprintable characters, use ptr2cells() for that. 1506 */ 1507 int 1508 latin_ptr2cells(char_u *p UNUSED) 1509 { 1510 return 1; 1511 } 1512 1513 int 1514 utf_ptr2cells( 1515 char_u *p) 1516 { 1517 int c; 1518 1519 /* Need to convert to a wide character. */ 1520 if (*p >= 0x80) 1521 { 1522 c = utf_ptr2char(p); 1523 /* An illegal byte is displayed as <xx>. */ 1524 if (utf_ptr2len(p) == 1 || c == NUL) 1525 return 4; 1526 /* If the char is ASCII it must be an overlong sequence. */ 1527 if (c < 0x80) 1528 return char2cells(c); 1529 return utf_char2cells(c); 1530 } 1531 return 1; 1532 } 1533 1534 int 1535 dbcs_ptr2cells(char_u *p) 1536 { 1537 /* Number of cells is equal to number of bytes, except for euc-jp when 1538 * the first byte is 0x8e. */ 1539 if (enc_dbcs == DBCS_JPNU && *p == 0x8e) 1540 return 1; 1541 return MB_BYTE2LEN(*p); 1542 } 1543 1544 /* 1545 * mb_ptr2cells_len() function pointer. 1546 * Like mb_ptr2cells(), but limit string length to "size". 1547 * For an empty string or truncated character returns 1. 1548 */ 1549 int 1550 latin_ptr2cells_len(char_u *p UNUSED, int size UNUSED) 1551 { 1552 return 1; 1553 } 1554 1555 static int 1556 utf_ptr2cells_len(char_u *p, int size) 1557 { 1558 int c; 1559 1560 /* Need to convert to a wide character. */ 1561 if (size > 0 && *p >= 0x80) 1562 { 1563 if (utf_ptr2len_len(p, size) < utf8len_tab[*p]) 1564 return 1; /* truncated */ 1565 c = utf_ptr2char(p); 1566 /* An illegal byte is displayed as <xx>. */ 1567 if (utf_ptr2len(p) == 1 || c == NUL) 1568 return 4; 1569 /* If the char is ASCII it must be an overlong sequence. */ 1570 if (c < 0x80) 1571 return char2cells(c); 1572 return utf_char2cells(c); 1573 } 1574 return 1; 1575 } 1576 1577 static int 1578 dbcs_ptr2cells_len(char_u *p, int size) 1579 { 1580 /* Number of cells is equal to number of bytes, except for euc-jp when 1581 * the first byte is 0x8e. */ 1582 if (size <= 1 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)) 1583 return 1; 1584 return MB_BYTE2LEN(*p); 1585 } 1586 1587 /* 1588 * mb_char2cells() function pointer. 1589 * Return the number of display cells character "c" occupies. 1590 * Only takes care of multi-byte chars, not "^C" and such. 1591 */ 1592 int 1593 latin_char2cells(int c UNUSED) 1594 { 1595 return 1; 1596 } 1597 1598 static int 1599 dbcs_char2cells(int c) 1600 { 1601 /* Number of cells is equal to number of bytes, except for euc-jp when 1602 * the first byte is 0x8e. */ 1603 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e) 1604 return 1; 1605 /* use the first byte */ 1606 return MB_BYTE2LEN((unsigned)c >> 8); 1607 } 1608 1609 /* 1610 * Return the number of cells occupied by string "p". 1611 * Stop at a NUL character. When "len" >= 0 stop at character "p[len]". 1612 */ 1613 int 1614 mb_string2cells(char_u *p, int len) 1615 { 1616 int i; 1617 int clen = 0; 1618 1619 for (i = 0; (len < 0 || i < len) && p[i] != NUL; i += (*mb_ptr2len)(p + i)) 1620 clen += (*mb_ptr2cells)(p + i); 1621 return clen; 1622 } 1623 1624 /* 1625 * mb_off2cells() function pointer. 1626 * Return number of display cells for char at ScreenLines[off]. 1627 * We make sure that the offset used is less than "max_off". 1628 */ 1629 int 1630 latin_off2cells(unsigned off UNUSED, unsigned max_off UNUSED) 1631 { 1632 return 1; 1633 } 1634 1635 int 1636 dbcs_off2cells(unsigned off, unsigned max_off) 1637 { 1638 /* never check beyond end of the line */ 1639 if (off >= max_off) 1640 return 1; 1641 1642 /* Number of cells is equal to number of bytes, except for euc-jp when 1643 * the first byte is 0x8e. */ 1644 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e) 1645 return 1; 1646 return MB_BYTE2LEN(ScreenLines[off]); 1647 } 1648 1649 int 1650 utf_off2cells(unsigned off, unsigned max_off) 1651 { 1652 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1; 1653 } 1654 1655 /* 1656 * mb_ptr2char() function pointer. 1657 * Convert a byte sequence into a character. 1658 */ 1659 int 1660 latin_ptr2char(char_u *p) 1661 { 1662 return *p; 1663 } 1664 1665 static int 1666 dbcs_ptr2char(char_u *p) 1667 { 1668 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL) 1669 return (p[0] << 8) + p[1]; 1670 return *p; 1671 } 1672 1673 /* 1674 * Convert a UTF-8 byte sequence to a wide character. 1675 * If the sequence is illegal or truncated by a NUL the first byte is 1676 * returned. 1677 * Does not include composing characters, of course. 1678 */ 1679 int 1680 utf_ptr2char(char_u *p) 1681 { 1682 int len; 1683 1684 if (p[0] < 0x80) /* be quick for ASCII */ 1685 return p[0]; 1686 1687 len = utf8len_tab_zero[p[0]]; 1688 if (len > 1 && (p[1] & 0xc0) == 0x80) 1689 { 1690 if (len == 2) 1691 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); 1692 if ((p[2] & 0xc0) == 0x80) 1693 { 1694 if (len == 3) 1695 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) 1696 + (p[2] & 0x3f); 1697 if ((p[3] & 0xc0) == 0x80) 1698 { 1699 if (len == 4) 1700 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) 1701 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); 1702 if ((p[4] & 0xc0) == 0x80) 1703 { 1704 if (len == 5) 1705 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) 1706 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) 1707 + (p[4] & 0x3f); 1708 if ((p[5] & 0xc0) == 0x80 && len == 6) 1709 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) 1710 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) 1711 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); 1712 } 1713 } 1714 } 1715 } 1716 /* Illegal value, just return the first byte */ 1717 return p[0]; 1718 } 1719 1720 /* 1721 * Convert a UTF-8 byte sequence to a wide character. 1722 * String is assumed to be terminated by NUL or after "n" bytes, whichever 1723 * comes first. 1724 * The function is safe in the sense that it never accesses memory beyond the 1725 * first "n" bytes of "s". 1726 * 1727 * On success, returns decoded codepoint, advances "s" to the beginning of 1728 * next character and decreases "n" accordingly. 1729 * 1730 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past 1731 * NUL byte. 1732 * 1733 * If byte sequence is illegal or incomplete, returns -1 and does not advance 1734 * "s". 1735 */ 1736 static int 1737 utf_safe_read_char_adv(char_u **s, size_t *n) 1738 { 1739 int c, k; 1740 1741 if (*n == 0) /* end of buffer */ 1742 return 0; 1743 1744 k = utf8len_tab_zero[**s]; 1745 1746 if (k == 1) 1747 { 1748 /* ASCII character or NUL */ 1749 (*n)--; 1750 return *(*s)++; 1751 } 1752 1753 if ((size_t)k <= *n) 1754 { 1755 /* We have a multibyte sequence and it isn't truncated by buffer 1756 * limits so utf_ptr2char() is safe to use. Or the first byte is 1757 * illegal (k=0), and it's also safe to use utf_ptr2char(). */ 1758 c = utf_ptr2char(*s); 1759 1760 /* On failure, utf_ptr2char() returns the first byte, so here we 1761 * check equality with the first byte. The only non-ASCII character 1762 * which equals the first byte of its own UTF-8 representation is 1763 * U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too. 1764 * It's safe even if n=1, else we would have k=2 > n. */ 1765 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83)) 1766 { 1767 /* byte sequence was successfully decoded */ 1768 *s += k; 1769 *n -= k; 1770 return c; 1771 } 1772 } 1773 1774 /* byte sequence is incomplete or illegal */ 1775 return -1; 1776 } 1777 1778 /* 1779 * Get character at **pp and advance *pp to the next character. 1780 * Note: composing characters are skipped! 1781 */ 1782 int 1783 mb_ptr2char_adv(char_u **pp) 1784 { 1785 int c; 1786 1787 c = (*mb_ptr2char)(*pp); 1788 *pp += (*mb_ptr2len)(*pp); 1789 return c; 1790 } 1791 1792 /* 1793 * Get character at **pp and advance *pp to the next character. 1794 * Note: composing characters are returned as separate characters. 1795 */ 1796 int 1797 mb_cptr2char_adv(char_u **pp) 1798 { 1799 int c; 1800 1801 c = (*mb_ptr2char)(*pp); 1802 if (enc_utf8) 1803 *pp += utf_ptr2len(*pp); 1804 else 1805 *pp += (*mb_ptr2len)(*pp); 1806 return c; 1807 } 1808 1809 #if defined(FEAT_ARABIC) || defined(PROTO) 1810 /* 1811 * Check whether we are dealing with Arabic combining characters. 1812 * Note: these are NOT really composing characters! 1813 */ 1814 int 1815 arabic_combine( 1816 int one, /* first character */ 1817 int two) /* character just after "one" */ 1818 { 1819 if (one == a_LAM) 1820 return arabic_maycombine(two); 1821 return FALSE; 1822 } 1823 1824 /* 1825 * Check whether we are dealing with a character that could be regarded as an 1826 * Arabic combining character, need to check the character before this. 1827 */ 1828 int 1829 arabic_maycombine(int two) 1830 { 1831 if (p_arshape && !p_tbidi) 1832 return (two == a_ALEF_MADDA 1833 || two == a_ALEF_HAMZA_ABOVE 1834 || two == a_ALEF_HAMZA_BELOW 1835 || two == a_ALEF); 1836 return FALSE; 1837 } 1838 1839 /* 1840 * Check if the character pointed to by "p2" is a composing character when it 1841 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which 1842 * behaves like a composing character. 1843 */ 1844 int 1845 utf_composinglike(char_u *p1, char_u *p2) 1846 { 1847 int c2; 1848 1849 c2 = utf_ptr2char(p2); 1850 if (utf_iscomposing(c2)) 1851 return TRUE; 1852 if (!arabic_maycombine(c2)) 1853 return FALSE; 1854 return arabic_combine(utf_ptr2char(p1), c2); 1855 } 1856 #endif 1857 1858 /* 1859 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1860 * composing characters. 1861 */ 1862 int 1863 utfc_ptr2char( 1864 char_u *p, 1865 int *pcc) /* return: composing chars, last one is 0 */ 1866 { 1867 int len; 1868 int c; 1869 int cc; 1870 int i = 0; 1871 1872 c = utf_ptr2char(p); 1873 len = utf_ptr2len(p); 1874 1875 /* Only accept a composing char when the first char isn't illegal. */ 1876 if ((len > 1 || *p < 0x80) 1877 && p[len] >= 0x80 1878 && UTF_COMPOSINGLIKE(p, p + len)) 1879 { 1880 cc = utf_ptr2char(p + len); 1881 for (;;) 1882 { 1883 pcc[i++] = cc; 1884 if (i == MAX_MCO) 1885 break; 1886 len += utf_ptr2len(p + len); 1887 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1888 break; 1889 } 1890 } 1891 1892 if (i < MAX_MCO) /* last composing char must be 0 */ 1893 pcc[i] = 0; 1894 1895 return c; 1896 } 1897 1898 /* 1899 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1900 * composing characters. Use no more than p[maxlen]. 1901 */ 1902 int 1903 utfc_ptr2char_len( 1904 char_u *p, 1905 int *pcc, /* return: composing chars, last one is 0 */ 1906 int maxlen) 1907 { 1908 int len; 1909 int c; 1910 int cc; 1911 int i = 0; 1912 1913 c = utf_ptr2char(p); 1914 len = utf_ptr2len_len(p, maxlen); 1915 /* Only accept a composing char when the first char isn't illegal. */ 1916 if ((len > 1 || *p < 0x80) 1917 && len < maxlen 1918 && p[len] >= 0x80 1919 && UTF_COMPOSINGLIKE(p, p + len)) 1920 { 1921 cc = utf_ptr2char(p + len); 1922 for (;;) 1923 { 1924 pcc[i++] = cc; 1925 if (i == MAX_MCO) 1926 break; 1927 len += utf_ptr2len_len(p + len, maxlen - len); 1928 if (len >= maxlen 1929 || p[len] < 0x80 1930 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1931 break; 1932 } 1933 } 1934 1935 if (i < MAX_MCO) /* last composing char must be 0 */ 1936 pcc[i] = 0; 1937 1938 return c; 1939 } 1940 1941 /* 1942 * Convert the character at screen position "off" to a sequence of bytes. 1943 * Includes the composing characters. 1944 * "buf" must at least have the length MB_MAXBYTES + 1. 1945 * Only to be used when ScreenLinesUC[off] != 0. 1946 * Returns the produced number of bytes. 1947 */ 1948 int 1949 utfc_char2bytes(int off, char_u *buf) 1950 { 1951 int len; 1952 int i; 1953 1954 len = utf_char2bytes(ScreenLinesUC[off], buf); 1955 for (i = 0; i < Screen_mco; ++i) 1956 { 1957 if (ScreenLinesC[i][off] == 0) 1958 break; 1959 len += utf_char2bytes(ScreenLinesC[i][off], buf + len); 1960 } 1961 return len; 1962 } 1963 1964 /* 1965 * Get the length of a UTF-8 byte sequence, not including any following 1966 * composing characters. 1967 * Returns 0 for "". 1968 * Returns 1 for an illegal byte sequence. 1969 */ 1970 int 1971 utf_ptr2len(char_u *p) 1972 { 1973 int len; 1974 int i; 1975 1976 if (*p == NUL) 1977 return 0; 1978 len = utf8len_tab[*p]; 1979 for (i = 1; i < len; ++i) 1980 if ((p[i] & 0xc0) != 0x80) 1981 return 1; 1982 return len; 1983 } 1984 1985 /* 1986 * Return length of UTF-8 character, obtained from the first byte. 1987 * "b" must be between 0 and 255! 1988 * Returns 1 for an invalid first byte value. 1989 */ 1990 int 1991 utf_byte2len(int b) 1992 { 1993 return utf8len_tab[b]; 1994 } 1995 1996 /* 1997 * Get the length of UTF-8 byte sequence "p[size]". Does not include any 1998 * following composing characters. 1999 * Returns 1 for "". 2000 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.). 2001 * Returns number > "size" for an incomplete byte sequence. 2002 * Never returns zero. 2003 */ 2004 int 2005 utf_ptr2len_len(char_u *p, int size) 2006 { 2007 int len; 2008 int i; 2009 int m; 2010 2011 len = utf8len_tab[*p]; 2012 if (len == 1) 2013 return 1; /* NUL, ascii or illegal lead byte */ 2014 if (len > size) 2015 m = size; /* incomplete byte sequence. */ 2016 else 2017 m = len; 2018 for (i = 1; i < m; ++i) 2019 if ((p[i] & 0xc0) != 0x80) 2020 return 1; 2021 return len; 2022 } 2023 2024 /* 2025 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. 2026 * This includes following composing characters. 2027 */ 2028 int 2029 utfc_ptr2len(char_u *p) 2030 { 2031 int len; 2032 int b0 = *p; 2033 #ifdef FEAT_ARABIC 2034 int prevlen; 2035 #endif 2036 2037 if (b0 == NUL) 2038 return 0; 2039 if (b0 < 0x80 && p[1] < 0x80) /* be quick for ASCII */ 2040 return 1; 2041 2042 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2043 len = utf_ptr2len(p); 2044 2045 /* Check for illegal byte. */ 2046 if (len == 1 && b0 >= 0x80) 2047 return 1; 2048 2049 /* 2050 * Check for composing characters. We can handle only the first six, but 2051 * skip all of them (otherwise the cursor would get stuck). 2052 */ 2053 #ifdef FEAT_ARABIC 2054 prevlen = 0; 2055 #endif 2056 for (;;) 2057 { 2058 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2059 return len; 2060 2061 /* Skip over composing char */ 2062 #ifdef FEAT_ARABIC 2063 prevlen = len; 2064 #endif 2065 len += utf_ptr2len(p + len); 2066 } 2067 } 2068 2069 /* 2070 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" 2071 * takes. This includes following composing characters. 2072 * Returns 0 for an empty string. 2073 * Returns 1 for an illegal char or an incomplete byte sequence. 2074 */ 2075 int 2076 utfc_ptr2len_len(char_u *p, int size) 2077 { 2078 int len; 2079 #ifdef FEAT_ARABIC 2080 int prevlen; 2081 #endif 2082 2083 if (size < 1 || *p == NUL) 2084 return 0; 2085 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) /* be quick for ASCII */ 2086 return 1; 2087 2088 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2089 len = utf_ptr2len_len(p, size); 2090 2091 /* Check for illegal byte and incomplete byte sequence. */ 2092 if ((len == 1 && p[0] >= 0x80) || len > size) 2093 return 1; 2094 2095 /* 2096 * Check for composing characters. We can handle only the first six, but 2097 * skip all of them (otherwise the cursor would get stuck). 2098 */ 2099 #ifdef FEAT_ARABIC 2100 prevlen = 0; 2101 #endif 2102 while (len < size) 2103 { 2104 int len_next_char; 2105 2106 if (p[len] < 0x80) 2107 break; 2108 2109 /* 2110 * Next character length should not go beyond size to ensure that 2111 * UTF_COMPOSINGLIKE(...) does not read beyond size. 2112 */ 2113 len_next_char = utf_ptr2len_len(p + len, size - len); 2114 if (len_next_char > size - len) 2115 break; 2116 2117 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2118 break; 2119 2120 /* Skip over composing char */ 2121 #ifdef FEAT_ARABIC 2122 prevlen = len; 2123 #endif 2124 len += len_next_char; 2125 } 2126 return len; 2127 } 2128 2129 /* 2130 * Return the number of bytes the UTF-8 encoding of character "c" takes. 2131 * This does not include composing characters. 2132 */ 2133 int 2134 utf_char2len(int c) 2135 { 2136 if (c < 0x80) 2137 return 1; 2138 if (c < 0x800) 2139 return 2; 2140 if (c < 0x10000) 2141 return 3; 2142 if (c < 0x200000) 2143 return 4; 2144 if (c < 0x4000000) 2145 return 5; 2146 return 6; 2147 } 2148 2149 /* 2150 * Convert Unicode character "c" to UTF-8 string in "buf[]". 2151 * Returns the number of bytes. 2152 * This does not include composing characters. 2153 */ 2154 int 2155 utf_char2bytes(int c, char_u *buf) 2156 { 2157 if (c < 0x80) /* 7 bits */ 2158 { 2159 buf[0] = c; 2160 return 1; 2161 } 2162 if (c < 0x800) /* 11 bits */ 2163 { 2164 buf[0] = 0xc0 + ((unsigned)c >> 6); 2165 buf[1] = 0x80 + (c & 0x3f); 2166 return 2; 2167 } 2168 if (c < 0x10000) /* 16 bits */ 2169 { 2170 buf[0] = 0xe0 + ((unsigned)c >> 12); 2171 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2172 buf[2] = 0x80 + (c & 0x3f); 2173 return 3; 2174 } 2175 if (c < 0x200000) /* 21 bits */ 2176 { 2177 buf[0] = 0xf0 + ((unsigned)c >> 18); 2178 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2179 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2180 buf[3] = 0x80 + (c & 0x3f); 2181 return 4; 2182 } 2183 if (c < 0x4000000) /* 26 bits */ 2184 { 2185 buf[0] = 0xf8 + ((unsigned)c >> 24); 2186 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2187 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2188 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2189 buf[4] = 0x80 + (c & 0x3f); 2190 return 5; 2191 } 2192 /* 31 bits */ 2193 buf[0] = 0xfc + ((unsigned)c >> 30); 2194 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f); 2195 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2196 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2197 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2198 buf[5] = 0x80 + (c & 0x3f); 2199 return 6; 2200 } 2201 2202 /* 2203 * Return TRUE if "c" is a composing UTF-8 character. This means it will be 2204 * drawn on top of the preceding character. 2205 * Based on code from Markus Kuhn. 2206 */ 2207 int 2208 utf_iscomposing(int c) 2209 { 2210 /* Sorted list of non-overlapping intervals. 2211 * Generated by ../runtime/tools/unicode.vim. */ 2212 static struct interval combining[] = 2213 { 2214 {0x0300, 0x036f}, 2215 {0x0483, 0x0489}, 2216 {0x0591, 0x05bd}, 2217 {0x05bf, 0x05bf}, 2218 {0x05c1, 0x05c2}, 2219 {0x05c4, 0x05c5}, 2220 {0x05c7, 0x05c7}, 2221 {0x0610, 0x061a}, 2222 {0x064b, 0x065f}, 2223 {0x0670, 0x0670}, 2224 {0x06d6, 0x06dc}, 2225 {0x06df, 0x06e4}, 2226 {0x06e7, 0x06e8}, 2227 {0x06ea, 0x06ed}, 2228 {0x0711, 0x0711}, 2229 {0x0730, 0x074a}, 2230 {0x07a6, 0x07b0}, 2231 {0x07eb, 0x07f3}, 2232 {0x0816, 0x0819}, 2233 {0x081b, 0x0823}, 2234 {0x0825, 0x0827}, 2235 {0x0829, 0x082d}, 2236 {0x0859, 0x085b}, 2237 {0x08e3, 0x0903}, 2238 {0x093a, 0x093c}, 2239 {0x093e, 0x094f}, 2240 {0x0951, 0x0957}, 2241 {0x0962, 0x0963}, 2242 {0x0981, 0x0983}, 2243 {0x09bc, 0x09bc}, 2244 {0x09be, 0x09c4}, 2245 {0x09c7, 0x09c8}, 2246 {0x09cb, 0x09cd}, 2247 {0x09d7, 0x09d7}, 2248 {0x09e2, 0x09e3}, 2249 {0x0a01, 0x0a03}, 2250 {0x0a3c, 0x0a3c}, 2251 {0x0a3e, 0x0a42}, 2252 {0x0a47, 0x0a48}, 2253 {0x0a4b, 0x0a4d}, 2254 {0x0a51, 0x0a51}, 2255 {0x0a70, 0x0a71}, 2256 {0x0a75, 0x0a75}, 2257 {0x0a81, 0x0a83}, 2258 {0x0abc, 0x0abc}, 2259 {0x0abe, 0x0ac5}, 2260 {0x0ac7, 0x0ac9}, 2261 {0x0acb, 0x0acd}, 2262 {0x0ae2, 0x0ae3}, 2263 {0x0b01, 0x0b03}, 2264 {0x0b3c, 0x0b3c}, 2265 {0x0b3e, 0x0b44}, 2266 {0x0b47, 0x0b48}, 2267 {0x0b4b, 0x0b4d}, 2268 {0x0b56, 0x0b57}, 2269 {0x0b62, 0x0b63}, 2270 {0x0b82, 0x0b82}, 2271 {0x0bbe, 0x0bc2}, 2272 {0x0bc6, 0x0bc8}, 2273 {0x0bca, 0x0bcd}, 2274 {0x0bd7, 0x0bd7}, 2275 {0x0c00, 0x0c03}, 2276 {0x0c3e, 0x0c44}, 2277 {0x0c46, 0x0c48}, 2278 {0x0c4a, 0x0c4d}, 2279 {0x0c55, 0x0c56}, 2280 {0x0c62, 0x0c63}, 2281 {0x0c81, 0x0c83}, 2282 {0x0cbc, 0x0cbc}, 2283 {0x0cbe, 0x0cc4}, 2284 {0x0cc6, 0x0cc8}, 2285 {0x0cca, 0x0ccd}, 2286 {0x0cd5, 0x0cd6}, 2287 {0x0ce2, 0x0ce3}, 2288 {0x0d01, 0x0d03}, 2289 {0x0d3e, 0x0d44}, 2290 {0x0d46, 0x0d48}, 2291 {0x0d4a, 0x0d4d}, 2292 {0x0d57, 0x0d57}, 2293 {0x0d62, 0x0d63}, 2294 {0x0d82, 0x0d83}, 2295 {0x0dca, 0x0dca}, 2296 {0x0dcf, 0x0dd4}, 2297 {0x0dd6, 0x0dd6}, 2298 {0x0dd8, 0x0ddf}, 2299 {0x0df2, 0x0df3}, 2300 {0x0e31, 0x0e31}, 2301 {0x0e34, 0x0e3a}, 2302 {0x0e47, 0x0e4e}, 2303 {0x0eb1, 0x0eb1}, 2304 {0x0eb4, 0x0eb9}, 2305 {0x0ebb, 0x0ebc}, 2306 {0x0ec8, 0x0ecd}, 2307 {0x0f18, 0x0f19}, 2308 {0x0f35, 0x0f35}, 2309 {0x0f37, 0x0f37}, 2310 {0x0f39, 0x0f39}, 2311 {0x0f3e, 0x0f3f}, 2312 {0x0f71, 0x0f84}, 2313 {0x0f86, 0x0f87}, 2314 {0x0f8d, 0x0f97}, 2315 {0x0f99, 0x0fbc}, 2316 {0x0fc6, 0x0fc6}, 2317 {0x102b, 0x103e}, 2318 {0x1056, 0x1059}, 2319 {0x105e, 0x1060}, 2320 {0x1062, 0x1064}, 2321 {0x1067, 0x106d}, 2322 {0x1071, 0x1074}, 2323 {0x1082, 0x108d}, 2324 {0x108f, 0x108f}, 2325 {0x109a, 0x109d}, 2326 {0x135d, 0x135f}, 2327 {0x1712, 0x1714}, 2328 {0x1732, 0x1734}, 2329 {0x1752, 0x1753}, 2330 {0x1772, 0x1773}, 2331 {0x17b4, 0x17d3}, 2332 {0x17dd, 0x17dd}, 2333 {0x180b, 0x180d}, 2334 {0x18a9, 0x18a9}, 2335 {0x1920, 0x192b}, 2336 {0x1930, 0x193b}, 2337 {0x1a17, 0x1a1b}, 2338 {0x1a55, 0x1a5e}, 2339 {0x1a60, 0x1a7c}, 2340 {0x1a7f, 0x1a7f}, 2341 {0x1ab0, 0x1abe}, 2342 {0x1b00, 0x1b04}, 2343 {0x1b34, 0x1b44}, 2344 {0x1b6b, 0x1b73}, 2345 {0x1b80, 0x1b82}, 2346 {0x1ba1, 0x1bad}, 2347 {0x1be6, 0x1bf3}, 2348 {0x1c24, 0x1c37}, 2349 {0x1cd0, 0x1cd2}, 2350 {0x1cd4, 0x1ce8}, 2351 {0x1ced, 0x1ced}, 2352 {0x1cf2, 0x1cf4}, 2353 {0x1cf8, 0x1cf9}, 2354 {0x1dc0, 0x1df5}, 2355 {0x1dfc, 0x1dff}, 2356 {0x20d0, 0x20f0}, 2357 {0x2cef, 0x2cf1}, 2358 {0x2d7f, 0x2d7f}, 2359 {0x2de0, 0x2dff}, 2360 {0x302a, 0x302f}, 2361 {0x3099, 0x309a}, 2362 {0xa66f, 0xa672}, 2363 {0xa674, 0xa67d}, 2364 {0xa69e, 0xa69f}, 2365 {0xa6f0, 0xa6f1}, 2366 {0xa802, 0xa802}, 2367 {0xa806, 0xa806}, 2368 {0xa80b, 0xa80b}, 2369 {0xa823, 0xa827}, 2370 {0xa880, 0xa881}, 2371 {0xa8b4, 0xa8c4}, 2372 {0xa8e0, 0xa8f1}, 2373 {0xa926, 0xa92d}, 2374 {0xa947, 0xa953}, 2375 {0xa980, 0xa983}, 2376 {0xa9b3, 0xa9c0}, 2377 {0xa9e5, 0xa9e5}, 2378 {0xaa29, 0xaa36}, 2379 {0xaa43, 0xaa43}, 2380 {0xaa4c, 0xaa4d}, 2381 {0xaa7b, 0xaa7d}, 2382 {0xaab0, 0xaab0}, 2383 {0xaab2, 0xaab4}, 2384 {0xaab7, 0xaab8}, 2385 {0xaabe, 0xaabf}, 2386 {0xaac1, 0xaac1}, 2387 {0xaaeb, 0xaaef}, 2388 {0xaaf5, 0xaaf6}, 2389 {0xabe3, 0xabea}, 2390 {0xabec, 0xabed}, 2391 {0xfb1e, 0xfb1e}, 2392 {0xfe00, 0xfe0f}, 2393 {0xfe20, 0xfe2f}, 2394 {0x101fd, 0x101fd}, 2395 {0x102e0, 0x102e0}, 2396 {0x10376, 0x1037a}, 2397 {0x10a01, 0x10a03}, 2398 {0x10a05, 0x10a06}, 2399 {0x10a0c, 0x10a0f}, 2400 {0x10a38, 0x10a3a}, 2401 {0x10a3f, 0x10a3f}, 2402 {0x10ae5, 0x10ae6}, 2403 {0x11000, 0x11002}, 2404 {0x11038, 0x11046}, 2405 {0x1107f, 0x11082}, 2406 {0x110b0, 0x110ba}, 2407 {0x11100, 0x11102}, 2408 {0x11127, 0x11134}, 2409 {0x11173, 0x11173}, 2410 {0x11180, 0x11182}, 2411 {0x111b3, 0x111c0}, 2412 {0x111ca, 0x111cc}, 2413 {0x1122c, 0x11237}, 2414 {0x112df, 0x112ea}, 2415 {0x11300, 0x11303}, 2416 {0x1133c, 0x1133c}, 2417 {0x1133e, 0x11344}, 2418 {0x11347, 0x11348}, 2419 {0x1134b, 0x1134d}, 2420 {0x11357, 0x11357}, 2421 {0x11362, 0x11363}, 2422 {0x11366, 0x1136c}, 2423 {0x11370, 0x11374}, 2424 {0x114b0, 0x114c3}, 2425 {0x115af, 0x115b5}, 2426 {0x115b8, 0x115c0}, 2427 {0x115dc, 0x115dd}, 2428 {0x11630, 0x11640}, 2429 {0x116ab, 0x116b7}, 2430 {0x1171d, 0x1172b}, 2431 {0x16af0, 0x16af4}, 2432 {0x16b30, 0x16b36}, 2433 {0x16f51, 0x16f7e}, 2434 {0x16f8f, 0x16f92}, 2435 {0x1bc9d, 0x1bc9e}, 2436 {0x1d165, 0x1d169}, 2437 {0x1d16d, 0x1d172}, 2438 {0x1d17b, 0x1d182}, 2439 {0x1d185, 0x1d18b}, 2440 {0x1d1aa, 0x1d1ad}, 2441 {0x1d242, 0x1d244}, 2442 {0x1da00, 0x1da36}, 2443 {0x1da3b, 0x1da6c}, 2444 {0x1da75, 0x1da75}, 2445 {0x1da84, 0x1da84}, 2446 {0x1da9b, 0x1da9f}, 2447 {0x1daa1, 0x1daaf}, 2448 {0x1e8d0, 0x1e8d6}, 2449 {0xe0100, 0xe01ef} 2450 }; 2451 2452 return intable(combining, sizeof(combining), c); 2453 } 2454 2455 /* 2456 * Return TRUE for characters that can be displayed in a normal way. 2457 * Only for characters of 0x100 and above! 2458 */ 2459 int 2460 utf_printable(int c) 2461 { 2462 #ifdef USE_WCHAR_FUNCTIONS 2463 /* 2464 * Assume the iswprint() library function works better than our own stuff. 2465 */ 2466 return iswprint(c); 2467 #else 2468 /* Sorted list of non-overlapping intervals. 2469 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */ 2470 static struct interval nonprint[] = 2471 { 2472 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2473 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2474 {0xfffe, 0xffff} 2475 }; 2476 2477 return !intable(nonprint, sizeof(nonprint), c); 2478 #endif 2479 } 2480 2481 /* 2482 * Get class of a Unicode character. 2483 * 0: white space 2484 * 1: punctuation 2485 * 2 or bigger: some class of word character. 2486 */ 2487 int 2488 utf_class(int c) 2489 { 2490 /* sorted list of non-overlapping intervals */ 2491 static struct clinterval 2492 { 2493 unsigned int first; 2494 unsigned int last; 2495 unsigned int class; 2496 } classes[] = 2497 { 2498 {0x037e, 0x037e, 1}, /* Greek question mark */ 2499 {0x0387, 0x0387, 1}, /* Greek ano teleia */ 2500 {0x055a, 0x055f, 1}, /* Armenian punctuation */ 2501 {0x0589, 0x0589, 1}, /* Armenian full stop */ 2502 {0x05be, 0x05be, 1}, 2503 {0x05c0, 0x05c0, 1}, 2504 {0x05c3, 0x05c3, 1}, 2505 {0x05f3, 0x05f4, 1}, 2506 {0x060c, 0x060c, 1}, 2507 {0x061b, 0x061b, 1}, 2508 {0x061f, 0x061f, 1}, 2509 {0x066a, 0x066d, 1}, 2510 {0x06d4, 0x06d4, 1}, 2511 {0x0700, 0x070d, 1}, /* Syriac punctuation */ 2512 {0x0964, 0x0965, 1}, 2513 {0x0970, 0x0970, 1}, 2514 {0x0df4, 0x0df4, 1}, 2515 {0x0e4f, 0x0e4f, 1}, 2516 {0x0e5a, 0x0e5b, 1}, 2517 {0x0f04, 0x0f12, 1}, 2518 {0x0f3a, 0x0f3d, 1}, 2519 {0x0f85, 0x0f85, 1}, 2520 {0x104a, 0x104f, 1}, /* Myanmar punctuation */ 2521 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */ 2522 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */ 2523 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */ 2524 {0x1680, 0x1680, 0}, 2525 {0x169b, 0x169c, 1}, 2526 {0x16eb, 0x16ed, 1}, 2527 {0x1735, 0x1736, 1}, 2528 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */ 2529 {0x1800, 0x180a, 1}, /* Mongolian punctuation */ 2530 {0x2000, 0x200b, 0}, /* spaces */ 2531 {0x200c, 0x2027, 1}, /* punctuation and symbols */ 2532 {0x2028, 0x2029, 0}, 2533 {0x202a, 0x202e, 1}, /* punctuation and symbols */ 2534 {0x202f, 0x202f, 0}, 2535 {0x2030, 0x205e, 1}, /* punctuation and symbols */ 2536 {0x205f, 0x205f, 0}, 2537 {0x2060, 0x27ff, 1}, /* punctuation and symbols */ 2538 {0x2070, 0x207f, 0x2070}, /* superscript */ 2539 {0x2080, 0x2094, 0x2080}, /* subscript */ 2540 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */ 2541 {0x2800, 0x28ff, 0x2800}, /* braille */ 2542 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */ 2543 {0x29d8, 0x29db, 1}, 2544 {0x29fc, 0x29fd, 1}, 2545 {0x2e00, 0x2e7f, 1}, /* supplemental punctuation */ 2546 {0x3000, 0x3000, 0}, /* ideographic space */ 2547 {0x3001, 0x3020, 1}, /* ideographic punctuation */ 2548 {0x3030, 0x3030, 1}, 2549 {0x303d, 0x303d, 1}, 2550 {0x3040, 0x309f, 0x3040}, /* Hiragana */ 2551 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */ 2552 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */ 2553 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */ 2554 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */ 2555 {0xfd3e, 0xfd3f, 1}, 2556 {0xfe30, 0xfe6b, 1}, /* punctuation forms */ 2557 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */ 2558 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */ 2559 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */ 2560 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */ 2561 {0x20000, 0x2a6df, 0x4e00}, /* CJK Ideographs */ 2562 {0x2a700, 0x2b73f, 0x4e00}, /* CJK Ideographs */ 2563 {0x2b740, 0x2b81f, 0x4e00}, /* CJK Ideographs */ 2564 {0x2f800, 0x2fa1f, 0x4e00}, /* CJK Ideographs */ 2565 }; 2566 2567 /* Sorted list of non-overlapping intervals of all Emoji characters, 2568 * based on http://unicode.org/emoji/charts/emoji-list.html */ 2569 static struct interval emoji_all[] = 2570 { 2571 {0x203c, 0x203c}, 2572 {0x2049, 0x2049}, 2573 {0x2122, 0x2122}, 2574 {0x2139, 0x2139}, 2575 {0x2194, 0x2199}, 2576 {0x21a9, 0x21aa}, 2577 {0x231a, 0x231b}, 2578 {0x2328, 0x2328}, 2579 {0x23cf, 0x23cf}, 2580 {0x23e9, 0x23f3}, 2581 {0x24c2, 0x24c2}, 2582 {0x25aa, 0x25ab}, 2583 {0x25b6, 0x25b6}, 2584 {0x25c0, 0x25c0}, 2585 {0x25fb, 0x25fe}, 2586 {0x2600, 0x2604}, 2587 {0x260e, 0x260e}, 2588 {0x2611, 0x2611}, 2589 {0x2614, 0x2615}, 2590 {0x2618, 0x2618}, 2591 {0x261d, 0x261d}, 2592 {0x2620, 0x2620}, 2593 {0x2622, 0x2623}, 2594 {0x2626, 0x2626}, 2595 {0x262a, 0x262a}, 2596 {0x262e, 0x262f}, 2597 {0x2638, 0x263a}, 2598 {0x2648, 0x2653}, 2599 {0x2660, 0x2660}, 2600 {0x2663, 0x2663}, 2601 {0x2665, 0x2666}, 2602 {0x2668, 0x2668}, 2603 {0x267b, 0x267b}, 2604 {0x267f, 0x267f}, 2605 {0x2692, 0x2694}, 2606 {0x2696, 0x2697}, 2607 {0x2699, 0x2699}, 2608 {0x269b, 0x269c}, 2609 {0x26a0, 0x26a1}, 2610 {0x26aa, 0x26ab}, 2611 {0x26b0, 0x26b1}, 2612 {0x26bd, 0x26be}, 2613 {0x26c4, 0x26c5}, 2614 {0x26c8, 0x26c8}, 2615 {0x26ce, 0x26cf}, 2616 {0x26d1, 0x26d1}, 2617 {0x26d3, 0x26d4}, 2618 {0x26e9, 0x26ea}, 2619 {0x26f0, 0x26f5}, 2620 {0x26f7, 0x26fa}, 2621 {0x26fd, 0x26fd}, 2622 {0x2702, 0x2702}, 2623 {0x2705, 0x2705}, 2624 {0x2708, 0x270d}, 2625 {0x270f, 0x270f}, 2626 {0x2712, 0x2712}, 2627 {0x2714, 0x2714}, 2628 {0x2716, 0x2716}, 2629 {0x271d, 0x271d}, 2630 {0x2721, 0x2721}, 2631 {0x2728, 0x2728}, 2632 {0x2733, 0x2734}, 2633 {0x2744, 0x2744}, 2634 {0x2747, 0x2747}, 2635 {0x274c, 0x274c}, 2636 {0x274e, 0x274e}, 2637 {0x2753, 0x2755}, 2638 {0x2757, 0x2757}, 2639 {0x2763, 0x2764}, 2640 {0x2795, 0x2797}, 2641 {0x27a1, 0x27a1}, 2642 {0x27b0, 0x27b0}, 2643 {0x27bf, 0x27bf}, 2644 {0x2934, 0x2935}, 2645 {0x2b05, 0x2b07}, 2646 {0x2b1b, 0x2b1c}, 2647 {0x2b50, 0x2b50}, 2648 {0x2b55, 0x2b55}, 2649 {0x3030, 0x3030}, 2650 {0x303d, 0x303d}, 2651 {0x3297, 0x3297}, 2652 {0x3299, 0x3299}, 2653 {0x1f004, 0x1f004}, 2654 {0x1f0cf, 0x1f0cf}, 2655 {0x1f170, 0x1f171}, 2656 {0x1f17e, 0x1f17f}, 2657 {0x1f18e, 0x1f18e}, 2658 {0x1f191, 0x1f19a}, 2659 {0x1f1e6, 0x1f1ff}, 2660 {0x1f201, 0x1f202}, 2661 {0x1f21a, 0x1f21a}, 2662 {0x1f22f, 0x1f22f}, 2663 {0x1f232, 0x1f23a}, 2664 {0x1f250, 0x1f251}, 2665 {0x1f300, 0x1f320}, 2666 {0x1f330, 0x1f335}, 2667 {0x1f337, 0x1f37c}, 2668 {0x1f380, 0x1f393}, 2669 {0x1f3a0, 0x1f3c4}, 2670 {0x1f3c6, 0x1f3ca}, 2671 {0x1f3e0, 0x1f3f0}, 2672 {0x1f400, 0x1f43e}, 2673 {0x1f440, 0x1f440}, 2674 {0x1f442, 0x1f4f7}, 2675 {0x1f4f9, 0x1f4fc}, 2676 {0x1f500, 0x1f53d}, 2677 {0x1f550, 0x1f567}, 2678 {0x1f5fb, 0x1f640}, 2679 {0x1f645, 0x1f64f}, 2680 {0x1f680, 0x1f6c5} 2681 }; 2682 2683 int bot = 0; 2684 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2685 int mid; 2686 2687 /* First quick check for Latin1 characters, use 'iskeyword'. */ 2688 if (c < 0x100) 2689 { 2690 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2691 return 0; /* blank */ 2692 if (vim_iswordc(c)) 2693 return 2; /* word character */ 2694 return 1; /* punctuation */ 2695 } 2696 2697 /* binary search in table */ 2698 while (top >= bot) 2699 { 2700 mid = (bot + top) / 2; 2701 if (classes[mid].last < (unsigned int)c) 2702 bot = mid + 1; 2703 else if (classes[mid].first > (unsigned int)c) 2704 top = mid - 1; 2705 else 2706 return (int)classes[mid].class; 2707 } 2708 2709 /* emoji */ 2710 if (intable(emoji_all, sizeof(emoji_all), c)) 2711 return 3; 2712 2713 /* most other characters are "word" characters */ 2714 return 2; 2715 } 2716 2717 /* 2718 * Code for Unicode case-dependent operations. Based on notes in 2719 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2720 * This code uses simple case folding, not full case folding. 2721 * Last updated for Unicode 5.2. 2722 */ 2723 2724 /* 2725 * The following tables are built by ../runtime/tools/unicode.vim. 2726 * They must be in numeric order, because we use binary search. 2727 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2728 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2729 * folded/upper/lower by adding 32. 2730 */ 2731 typedef struct 2732 { 2733 int rangeStart; 2734 int rangeEnd; 2735 int step; 2736 int offset; 2737 } convertStruct; 2738 2739 static convertStruct foldCase[] = 2740 { 2741 {0x41,0x5a,1,32}, 2742 {0xb5,0xb5,-1,775}, 2743 {0xc0,0xd6,1,32}, 2744 {0xd8,0xde,1,32}, 2745 {0x100,0x12e,2,1}, 2746 {0x132,0x136,2,1}, 2747 {0x139,0x147,2,1}, 2748 {0x14a,0x176,2,1}, 2749 {0x178,0x178,-1,-121}, 2750 {0x179,0x17d,2,1}, 2751 {0x17f,0x17f,-1,-268}, 2752 {0x181,0x181,-1,210}, 2753 {0x182,0x184,2,1}, 2754 {0x186,0x186,-1,206}, 2755 {0x187,0x187,-1,1}, 2756 {0x189,0x18a,1,205}, 2757 {0x18b,0x18b,-1,1}, 2758 {0x18e,0x18e,-1,79}, 2759 {0x18f,0x18f,-1,202}, 2760 {0x190,0x190,-1,203}, 2761 {0x191,0x191,-1,1}, 2762 {0x193,0x193,-1,205}, 2763 {0x194,0x194,-1,207}, 2764 {0x196,0x196,-1,211}, 2765 {0x197,0x197,-1,209}, 2766 {0x198,0x198,-1,1}, 2767 {0x19c,0x19c,-1,211}, 2768 {0x19d,0x19d,-1,213}, 2769 {0x19f,0x19f,-1,214}, 2770 {0x1a0,0x1a4,2,1}, 2771 {0x1a6,0x1a6,-1,218}, 2772 {0x1a7,0x1a7,-1,1}, 2773 {0x1a9,0x1a9,-1,218}, 2774 {0x1ac,0x1ac,-1,1}, 2775 {0x1ae,0x1ae,-1,218}, 2776 {0x1af,0x1af,-1,1}, 2777 {0x1b1,0x1b2,1,217}, 2778 {0x1b3,0x1b5,2,1}, 2779 {0x1b7,0x1b7,-1,219}, 2780 {0x1b8,0x1bc,4,1}, 2781 {0x1c4,0x1c4,-1,2}, 2782 {0x1c5,0x1c5,-1,1}, 2783 {0x1c7,0x1c7,-1,2}, 2784 {0x1c8,0x1c8,-1,1}, 2785 {0x1ca,0x1ca,-1,2}, 2786 {0x1cb,0x1db,2,1}, 2787 {0x1de,0x1ee,2,1}, 2788 {0x1f1,0x1f1,-1,2}, 2789 {0x1f2,0x1f4,2,1}, 2790 {0x1f6,0x1f6,-1,-97}, 2791 {0x1f7,0x1f7,-1,-56}, 2792 {0x1f8,0x21e,2,1}, 2793 {0x220,0x220,-1,-130}, 2794 {0x222,0x232,2,1}, 2795 {0x23a,0x23a,-1,10795}, 2796 {0x23b,0x23b,-1,1}, 2797 {0x23d,0x23d,-1,-163}, 2798 {0x23e,0x23e,-1,10792}, 2799 {0x241,0x241,-1,1}, 2800 {0x243,0x243,-1,-195}, 2801 {0x244,0x244,-1,69}, 2802 {0x245,0x245,-1,71}, 2803 {0x246,0x24e,2,1}, 2804 {0x345,0x345,-1,116}, 2805 {0x370,0x372,2,1}, 2806 {0x376,0x376,-1,1}, 2807 {0x37f,0x37f,-1,116}, 2808 {0x386,0x386,-1,38}, 2809 {0x388,0x38a,1,37}, 2810 {0x38c,0x38c,-1,64}, 2811 {0x38e,0x38f,1,63}, 2812 {0x391,0x3a1,1,32}, 2813 {0x3a3,0x3ab,1,32}, 2814 {0x3c2,0x3c2,-1,1}, 2815 {0x3cf,0x3cf,-1,8}, 2816 {0x3d0,0x3d0,-1,-30}, 2817 {0x3d1,0x3d1,-1,-25}, 2818 {0x3d5,0x3d5,-1,-15}, 2819 {0x3d6,0x3d6,-1,-22}, 2820 {0x3d8,0x3ee,2,1}, 2821 {0x3f0,0x3f0,-1,-54}, 2822 {0x3f1,0x3f1,-1,-48}, 2823 {0x3f4,0x3f4,-1,-60}, 2824 {0x3f5,0x3f5,-1,-64}, 2825 {0x3f7,0x3f7,-1,1}, 2826 {0x3f9,0x3f9,-1,-7}, 2827 {0x3fa,0x3fa,-1,1}, 2828 {0x3fd,0x3ff,1,-130}, 2829 {0x400,0x40f,1,80}, 2830 {0x410,0x42f,1,32}, 2831 {0x460,0x480,2,1}, 2832 {0x48a,0x4be,2,1}, 2833 {0x4c0,0x4c0,-1,15}, 2834 {0x4c1,0x4cd,2,1}, 2835 {0x4d0,0x52e,2,1}, 2836 {0x531,0x556,1,48}, 2837 {0x10a0,0x10c5,1,7264}, 2838 {0x10c7,0x10cd,6,7264}, 2839 {0x13f8,0x13fd,1,-8}, 2840 {0x1e00,0x1e94,2,1}, 2841 {0x1e9b,0x1e9b,-1,-58}, 2842 {0x1e9e,0x1e9e,-1,-7615}, 2843 {0x1ea0,0x1efe,2,1}, 2844 {0x1f08,0x1f0f,1,-8}, 2845 {0x1f18,0x1f1d,1,-8}, 2846 {0x1f28,0x1f2f,1,-8}, 2847 {0x1f38,0x1f3f,1,-8}, 2848 {0x1f48,0x1f4d,1,-8}, 2849 {0x1f59,0x1f5f,2,-8}, 2850 {0x1f68,0x1f6f,1,-8}, 2851 {0x1f88,0x1f8f,1,-8}, 2852 {0x1f98,0x1f9f,1,-8}, 2853 {0x1fa8,0x1faf,1,-8}, 2854 {0x1fb8,0x1fb9,1,-8}, 2855 {0x1fba,0x1fbb,1,-74}, 2856 {0x1fbc,0x1fbc,-1,-9}, 2857 {0x1fbe,0x1fbe,-1,-7173}, 2858 {0x1fc8,0x1fcb,1,-86}, 2859 {0x1fcc,0x1fcc,-1,-9}, 2860 {0x1fd8,0x1fd9,1,-8}, 2861 {0x1fda,0x1fdb,1,-100}, 2862 {0x1fe8,0x1fe9,1,-8}, 2863 {0x1fea,0x1feb,1,-112}, 2864 {0x1fec,0x1fec,-1,-7}, 2865 {0x1ff8,0x1ff9,1,-128}, 2866 {0x1ffa,0x1ffb,1,-126}, 2867 {0x1ffc,0x1ffc,-1,-9}, 2868 {0x2126,0x2126,-1,-7517}, 2869 {0x212a,0x212a,-1,-8383}, 2870 {0x212b,0x212b,-1,-8262}, 2871 {0x2132,0x2132,-1,28}, 2872 {0x2160,0x216f,1,16}, 2873 {0x2183,0x2183,-1,1}, 2874 {0x24b6,0x24cf,1,26}, 2875 {0x2c00,0x2c2e,1,48}, 2876 {0x2c60,0x2c60,-1,1}, 2877 {0x2c62,0x2c62,-1,-10743}, 2878 {0x2c63,0x2c63,-1,-3814}, 2879 {0x2c64,0x2c64,-1,-10727}, 2880 {0x2c67,0x2c6b,2,1}, 2881 {0x2c6d,0x2c6d,-1,-10780}, 2882 {0x2c6e,0x2c6e,-1,-10749}, 2883 {0x2c6f,0x2c6f,-1,-10783}, 2884 {0x2c70,0x2c70,-1,-10782}, 2885 {0x2c72,0x2c75,3,1}, 2886 {0x2c7e,0x2c7f,1,-10815}, 2887 {0x2c80,0x2ce2,2,1}, 2888 {0x2ceb,0x2ced,2,1}, 2889 {0x2cf2,0xa640,31054,1}, 2890 {0xa642,0xa66c,2,1}, 2891 {0xa680,0xa69a,2,1}, 2892 {0xa722,0xa72e,2,1}, 2893 {0xa732,0xa76e,2,1}, 2894 {0xa779,0xa77b,2,1}, 2895 {0xa77d,0xa77d,-1,-35332}, 2896 {0xa77e,0xa786,2,1}, 2897 {0xa78b,0xa78b,-1,1}, 2898 {0xa78d,0xa78d,-1,-42280}, 2899 {0xa790,0xa792,2,1}, 2900 {0xa796,0xa7a8,2,1}, 2901 {0xa7aa,0xa7aa,-1,-42308}, 2902 {0xa7ab,0xa7ab,-1,-42319}, 2903 {0xa7ac,0xa7ac,-1,-42315}, 2904 {0xa7ad,0xa7ad,-1,-42305}, 2905 {0xa7b0,0xa7b0,-1,-42258}, 2906 {0xa7b1,0xa7b1,-1,-42282}, 2907 {0xa7b2,0xa7b2,-1,-42261}, 2908 {0xa7b3,0xa7b3,-1,928}, 2909 {0xa7b4,0xa7b6,2,1}, 2910 {0xab70,0xabbf,1,-38864}, 2911 {0xff21,0xff3a,1,32}, 2912 {0x10400,0x10427,1,40}, 2913 {0x10c80,0x10cb2,1,64}, 2914 {0x118a0,0x118bf,1,32} 2915 }; 2916 2917 static int utf_convert(int a, convertStruct table[], int tableSize); 2918 static int utf_strnicmp(char_u *s1, char_u *s2, size_t n1, size_t n2); 2919 2920 /* 2921 * Generic conversion function for case operations. 2922 * Return the converted equivalent of "a", which is a UCS-4 character. Use 2923 * the given conversion "table". Uses binary search on "table". 2924 */ 2925 static int 2926 utf_convert( 2927 int a, 2928 convertStruct table[], 2929 int tableSize) 2930 { 2931 int start, mid, end; /* indices into table */ 2932 int entries = tableSize / sizeof(convertStruct); 2933 2934 start = 0; 2935 end = entries; 2936 while (start < end) 2937 { 2938 /* need to search further */ 2939 mid = (end + start) / 2; 2940 if (table[mid].rangeEnd < a) 2941 start = mid + 1; 2942 else 2943 end = mid; 2944 } 2945 if (start < entries 2946 && table[start].rangeStart <= a 2947 && a <= table[start].rangeEnd 2948 && (a - table[start].rangeStart) % table[start].step == 0) 2949 return (a + table[start].offset); 2950 else 2951 return a; 2952 } 2953 2954 /* 2955 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 2956 * simple case folding. 2957 */ 2958 int 2959 utf_fold(int a) 2960 { 2961 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 2962 } 2963 2964 static convertStruct toLower[] = 2965 { 2966 {0x41,0x5a,1,32}, 2967 {0xc0,0xd6,1,32}, 2968 {0xd8,0xde,1,32}, 2969 {0x100,0x12e,2,1}, 2970 {0x130,0x130,-1,-199}, 2971 {0x132,0x136,2,1}, 2972 {0x139,0x147,2,1}, 2973 {0x14a,0x176,2,1}, 2974 {0x178,0x178,-1,-121}, 2975 {0x179,0x17d,2,1}, 2976 {0x181,0x181,-1,210}, 2977 {0x182,0x184,2,1}, 2978 {0x186,0x186,-1,206}, 2979 {0x187,0x187,-1,1}, 2980 {0x189,0x18a,1,205}, 2981 {0x18b,0x18b,-1,1}, 2982 {0x18e,0x18e,-1,79}, 2983 {0x18f,0x18f,-1,202}, 2984 {0x190,0x190,-1,203}, 2985 {0x191,0x191,-1,1}, 2986 {0x193,0x193,-1,205}, 2987 {0x194,0x194,-1,207}, 2988 {0x196,0x196,-1,211}, 2989 {0x197,0x197,-1,209}, 2990 {0x198,0x198,-1,1}, 2991 {0x19c,0x19c,-1,211}, 2992 {0x19d,0x19d,-1,213}, 2993 {0x19f,0x19f,-1,214}, 2994 {0x1a0,0x1a4,2,1}, 2995 {0x1a6,0x1a6,-1,218}, 2996 {0x1a7,0x1a7,-1,1}, 2997 {0x1a9,0x1a9,-1,218}, 2998 {0x1ac,0x1ac,-1,1}, 2999 {0x1ae,0x1ae,-1,218}, 3000 {0x1af,0x1af,-1,1}, 3001 {0x1b1,0x1b2,1,217}, 3002 {0x1b3,0x1b5,2,1}, 3003 {0x1b7,0x1b7,-1,219}, 3004 {0x1b8,0x1bc,4,1}, 3005 {0x1c4,0x1c4,-1,2}, 3006 {0x1c5,0x1c5,-1,1}, 3007 {0x1c7,0x1c7,-1,2}, 3008 {0x1c8,0x1c8,-1,1}, 3009 {0x1ca,0x1ca,-1,2}, 3010 {0x1cb,0x1db,2,1}, 3011 {0x1de,0x1ee,2,1}, 3012 {0x1f1,0x1f1,-1,2}, 3013 {0x1f2,0x1f4,2,1}, 3014 {0x1f6,0x1f6,-1,-97}, 3015 {0x1f7,0x1f7,-1,-56}, 3016 {0x1f8,0x21e,2,1}, 3017 {0x220,0x220,-1,-130}, 3018 {0x222,0x232,2,1}, 3019 {0x23a,0x23a,-1,10795}, 3020 {0x23b,0x23b,-1,1}, 3021 {0x23d,0x23d,-1,-163}, 3022 {0x23e,0x23e,-1,10792}, 3023 {0x241,0x241,-1,1}, 3024 {0x243,0x243,-1,-195}, 3025 {0x244,0x244,-1,69}, 3026 {0x245,0x245,-1,71}, 3027 {0x246,0x24e,2,1}, 3028 {0x370,0x372,2,1}, 3029 {0x376,0x376,-1,1}, 3030 {0x37f,0x37f,-1,116}, 3031 {0x386,0x386,-1,38}, 3032 {0x388,0x38a,1,37}, 3033 {0x38c,0x38c,-1,64}, 3034 {0x38e,0x38f,1,63}, 3035 {0x391,0x3a1,1,32}, 3036 {0x3a3,0x3ab,1,32}, 3037 {0x3cf,0x3cf,-1,8}, 3038 {0x3d8,0x3ee,2,1}, 3039 {0x3f4,0x3f4,-1,-60}, 3040 {0x3f7,0x3f7,-1,1}, 3041 {0x3f9,0x3f9,-1,-7}, 3042 {0x3fa,0x3fa,-1,1}, 3043 {0x3fd,0x3ff,1,-130}, 3044 {0x400,0x40f,1,80}, 3045 {0x410,0x42f,1,32}, 3046 {0x460,0x480,2,1}, 3047 {0x48a,0x4be,2,1}, 3048 {0x4c0,0x4c0,-1,15}, 3049 {0x4c1,0x4cd,2,1}, 3050 {0x4d0,0x52e,2,1}, 3051 {0x531,0x556,1,48}, 3052 {0x10a0,0x10c5,1,7264}, 3053 {0x10c7,0x10cd,6,7264}, 3054 {0x13a0,0x13ef,1,38864}, 3055 {0x13f0,0x13f5,1,8}, 3056 {0x1e00,0x1e94,2,1}, 3057 {0x1e9e,0x1e9e,-1,-7615}, 3058 {0x1ea0,0x1efe,2,1}, 3059 {0x1f08,0x1f0f,1,-8}, 3060 {0x1f18,0x1f1d,1,-8}, 3061 {0x1f28,0x1f2f,1,-8}, 3062 {0x1f38,0x1f3f,1,-8}, 3063 {0x1f48,0x1f4d,1,-8}, 3064 {0x1f59,0x1f5f,2,-8}, 3065 {0x1f68,0x1f6f,1,-8}, 3066 {0x1f88,0x1f8f,1,-8}, 3067 {0x1f98,0x1f9f,1,-8}, 3068 {0x1fa8,0x1faf,1,-8}, 3069 {0x1fb8,0x1fb9,1,-8}, 3070 {0x1fba,0x1fbb,1,-74}, 3071 {0x1fbc,0x1fbc,-1,-9}, 3072 {0x1fc8,0x1fcb,1,-86}, 3073 {0x1fcc,0x1fcc,-1,-9}, 3074 {0x1fd8,0x1fd9,1,-8}, 3075 {0x1fda,0x1fdb,1,-100}, 3076 {0x1fe8,0x1fe9,1,-8}, 3077 {0x1fea,0x1feb,1,-112}, 3078 {0x1fec,0x1fec,-1,-7}, 3079 {0x1ff8,0x1ff9,1,-128}, 3080 {0x1ffa,0x1ffb,1,-126}, 3081 {0x1ffc,0x1ffc,-1,-9}, 3082 {0x2126,0x2126,-1,-7517}, 3083 {0x212a,0x212a,-1,-8383}, 3084 {0x212b,0x212b,-1,-8262}, 3085 {0x2132,0x2132,-1,28}, 3086 {0x2160,0x216f,1,16}, 3087 {0x2183,0x2183,-1,1}, 3088 {0x24b6,0x24cf,1,26}, 3089 {0x2c00,0x2c2e,1,48}, 3090 {0x2c60,0x2c60,-1,1}, 3091 {0x2c62,0x2c62,-1,-10743}, 3092 {0x2c63,0x2c63,-1,-3814}, 3093 {0x2c64,0x2c64,-1,-10727}, 3094 {0x2c67,0x2c6b,2,1}, 3095 {0x2c6d,0x2c6d,-1,-10780}, 3096 {0x2c6e,0x2c6e,-1,-10749}, 3097 {0x2c6f,0x2c6f,-1,-10783}, 3098 {0x2c70,0x2c70,-1,-10782}, 3099 {0x2c72,0x2c75,3,1}, 3100 {0x2c7e,0x2c7f,1,-10815}, 3101 {0x2c80,0x2ce2,2,1}, 3102 {0x2ceb,0x2ced,2,1}, 3103 {0x2cf2,0xa640,31054,1}, 3104 {0xa642,0xa66c,2,1}, 3105 {0xa680,0xa69a,2,1}, 3106 {0xa722,0xa72e,2,1}, 3107 {0xa732,0xa76e,2,1}, 3108 {0xa779,0xa77b,2,1}, 3109 {0xa77d,0xa77d,-1,-35332}, 3110 {0xa77e,0xa786,2,1}, 3111 {0xa78b,0xa78b,-1,1}, 3112 {0xa78d,0xa78d,-1,-42280}, 3113 {0xa790,0xa792,2,1}, 3114 {0xa796,0xa7a8,2,1}, 3115 {0xa7aa,0xa7aa,-1,-42308}, 3116 {0xa7ab,0xa7ab,-1,-42319}, 3117 {0xa7ac,0xa7ac,-1,-42315}, 3118 {0xa7ad,0xa7ad,-1,-42305}, 3119 {0xa7b0,0xa7b0,-1,-42258}, 3120 {0xa7b1,0xa7b1,-1,-42282}, 3121 {0xa7b2,0xa7b2,-1,-42261}, 3122 {0xa7b3,0xa7b3,-1,928}, 3123 {0xa7b4,0xa7b6,2,1}, 3124 {0xff21,0xff3a,1,32}, 3125 {0x10400,0x10427,1,40}, 3126 {0x10c80,0x10cb2,1,64}, 3127 {0x118a0,0x118bf,1,32} 3128 }; 3129 3130 static convertStruct toUpper[] = 3131 { 3132 {0x61,0x7a,1,-32}, 3133 {0xb5,0xb5,-1,743}, 3134 {0xe0,0xf6,1,-32}, 3135 {0xf8,0xfe,1,-32}, 3136 {0xff,0xff,-1,121}, 3137 {0x101,0x12f,2,-1}, 3138 {0x131,0x131,-1,-232}, 3139 {0x133,0x137,2,-1}, 3140 {0x13a,0x148,2,-1}, 3141 {0x14b,0x177,2,-1}, 3142 {0x17a,0x17e,2,-1}, 3143 {0x17f,0x17f,-1,-300}, 3144 {0x180,0x180,-1,195}, 3145 {0x183,0x185,2,-1}, 3146 {0x188,0x18c,4,-1}, 3147 {0x192,0x192,-1,-1}, 3148 {0x195,0x195,-1,97}, 3149 {0x199,0x199,-1,-1}, 3150 {0x19a,0x19a,-1,163}, 3151 {0x19e,0x19e,-1,130}, 3152 {0x1a1,0x1a5,2,-1}, 3153 {0x1a8,0x1ad,5,-1}, 3154 {0x1b0,0x1b4,4,-1}, 3155 {0x1b6,0x1b9,3,-1}, 3156 {0x1bd,0x1bd,-1,-1}, 3157 {0x1bf,0x1bf,-1,56}, 3158 {0x1c5,0x1c5,-1,-1}, 3159 {0x1c6,0x1c6,-1,-2}, 3160 {0x1c8,0x1c8,-1,-1}, 3161 {0x1c9,0x1c9,-1,-2}, 3162 {0x1cb,0x1cb,-1,-1}, 3163 {0x1cc,0x1cc,-1,-2}, 3164 {0x1ce,0x1dc,2,-1}, 3165 {0x1dd,0x1dd,-1,-79}, 3166 {0x1df,0x1ef,2,-1}, 3167 {0x1f2,0x1f2,-1,-1}, 3168 {0x1f3,0x1f3,-1,-2}, 3169 {0x1f5,0x1f9,4,-1}, 3170 {0x1fb,0x21f,2,-1}, 3171 {0x223,0x233,2,-1}, 3172 {0x23c,0x23c,-1,-1}, 3173 {0x23f,0x240,1,10815}, 3174 {0x242,0x247,5,-1}, 3175 {0x249,0x24f,2,-1}, 3176 {0x250,0x250,-1,10783}, 3177 {0x251,0x251,-1,10780}, 3178 {0x252,0x252,-1,10782}, 3179 {0x253,0x253,-1,-210}, 3180 {0x254,0x254,-1,-206}, 3181 {0x256,0x257,1,-205}, 3182 {0x259,0x259,-1,-202}, 3183 {0x25b,0x25b,-1,-203}, 3184 {0x25c,0x25c,-1,42319}, 3185 {0x260,0x260,-1,-205}, 3186 {0x261,0x261,-1,42315}, 3187 {0x263,0x263,-1,-207}, 3188 {0x265,0x265,-1,42280}, 3189 {0x266,0x266,-1,42308}, 3190 {0x268,0x268,-1,-209}, 3191 {0x269,0x269,-1,-211}, 3192 {0x26b,0x26b,-1,10743}, 3193 {0x26c,0x26c,-1,42305}, 3194 {0x26f,0x26f,-1,-211}, 3195 {0x271,0x271,-1,10749}, 3196 {0x272,0x272,-1,-213}, 3197 {0x275,0x275,-1,-214}, 3198 {0x27d,0x27d,-1,10727}, 3199 {0x280,0x283,3,-218}, 3200 {0x287,0x287,-1,42282}, 3201 {0x288,0x288,-1,-218}, 3202 {0x289,0x289,-1,-69}, 3203 {0x28a,0x28b,1,-217}, 3204 {0x28c,0x28c,-1,-71}, 3205 {0x292,0x292,-1,-219}, 3206 {0x29d,0x29d,-1,42261}, 3207 {0x29e,0x29e,-1,42258}, 3208 {0x345,0x345,-1,84}, 3209 {0x371,0x373,2,-1}, 3210 {0x377,0x377,-1,-1}, 3211 {0x37b,0x37d,1,130}, 3212 {0x3ac,0x3ac,-1,-38}, 3213 {0x3ad,0x3af,1,-37}, 3214 {0x3b1,0x3c1,1,-32}, 3215 {0x3c2,0x3c2,-1,-31}, 3216 {0x3c3,0x3cb,1,-32}, 3217 {0x3cc,0x3cc,-1,-64}, 3218 {0x3cd,0x3ce,1,-63}, 3219 {0x3d0,0x3d0,-1,-62}, 3220 {0x3d1,0x3d1,-1,-57}, 3221 {0x3d5,0x3d5,-1,-47}, 3222 {0x3d6,0x3d6,-1,-54}, 3223 {0x3d7,0x3d7,-1,-8}, 3224 {0x3d9,0x3ef,2,-1}, 3225 {0x3f0,0x3f0,-1,-86}, 3226 {0x3f1,0x3f1,-1,-80}, 3227 {0x3f2,0x3f2,-1,7}, 3228 {0x3f3,0x3f3,-1,-116}, 3229 {0x3f5,0x3f5,-1,-96}, 3230 {0x3f8,0x3fb,3,-1}, 3231 {0x430,0x44f,1,-32}, 3232 {0x450,0x45f,1,-80}, 3233 {0x461,0x481,2,-1}, 3234 {0x48b,0x4bf,2,-1}, 3235 {0x4c2,0x4ce,2,-1}, 3236 {0x4cf,0x4cf,-1,-15}, 3237 {0x4d1,0x52f,2,-1}, 3238 {0x561,0x586,1,-48}, 3239 {0x13f8,0x13fd,1,-8}, 3240 {0x1d79,0x1d79,-1,35332}, 3241 {0x1d7d,0x1d7d,-1,3814}, 3242 {0x1e01,0x1e95,2,-1}, 3243 {0x1e9b,0x1e9b,-1,-59}, 3244 {0x1ea1,0x1eff,2,-1}, 3245 {0x1f00,0x1f07,1,8}, 3246 {0x1f10,0x1f15,1,8}, 3247 {0x1f20,0x1f27,1,8}, 3248 {0x1f30,0x1f37,1,8}, 3249 {0x1f40,0x1f45,1,8}, 3250 {0x1f51,0x1f57,2,8}, 3251 {0x1f60,0x1f67,1,8}, 3252 {0x1f70,0x1f71,1,74}, 3253 {0x1f72,0x1f75,1,86}, 3254 {0x1f76,0x1f77,1,100}, 3255 {0x1f78,0x1f79,1,128}, 3256 {0x1f7a,0x1f7b,1,112}, 3257 {0x1f7c,0x1f7d,1,126}, 3258 {0x1f80,0x1f87,1,8}, 3259 {0x1f90,0x1f97,1,8}, 3260 {0x1fa0,0x1fa7,1,8}, 3261 {0x1fb0,0x1fb1,1,8}, 3262 {0x1fb3,0x1fb3,-1,9}, 3263 {0x1fbe,0x1fbe,-1,-7205}, 3264 {0x1fc3,0x1fc3,-1,9}, 3265 {0x1fd0,0x1fd1,1,8}, 3266 {0x1fe0,0x1fe1,1,8}, 3267 {0x1fe5,0x1fe5,-1,7}, 3268 {0x1ff3,0x1ff3,-1,9}, 3269 {0x214e,0x214e,-1,-28}, 3270 {0x2170,0x217f,1,-16}, 3271 {0x2184,0x2184,-1,-1}, 3272 {0x24d0,0x24e9,1,-26}, 3273 {0x2c30,0x2c5e,1,-48}, 3274 {0x2c61,0x2c61,-1,-1}, 3275 {0x2c65,0x2c65,-1,-10795}, 3276 {0x2c66,0x2c66,-1,-10792}, 3277 {0x2c68,0x2c6c,2,-1}, 3278 {0x2c73,0x2c76,3,-1}, 3279 {0x2c81,0x2ce3,2,-1}, 3280 {0x2cec,0x2cee,2,-1}, 3281 {0x2cf3,0x2cf3,-1,-1}, 3282 {0x2d00,0x2d25,1,-7264}, 3283 {0x2d27,0x2d2d,6,-7264}, 3284 {0xa641,0xa66d,2,-1}, 3285 {0xa681,0xa69b,2,-1}, 3286 {0xa723,0xa72f,2,-1}, 3287 {0xa733,0xa76f,2,-1}, 3288 {0xa77a,0xa77c,2,-1}, 3289 {0xa77f,0xa787,2,-1}, 3290 {0xa78c,0xa791,5,-1}, 3291 {0xa793,0xa797,4,-1}, 3292 {0xa799,0xa7a9,2,-1}, 3293 {0xa7b5,0xa7b7,2,-1}, 3294 {0xab53,0xab53,-1,-928}, 3295 {0xab70,0xabbf,1,-38864}, 3296 {0xff41,0xff5a,1,-32}, 3297 {0x10428,0x1044f,1,-40}, 3298 {0x10cc0,0x10cf2,1,-64}, 3299 {0x118c0,0x118df,1,-32} 3300 }; 3301 3302 /* 3303 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3304 * simple case folding. 3305 */ 3306 int 3307 utf_toupper(int a) 3308 { 3309 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */ 3310 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3311 return TOUPPER_ASC(a); 3312 3313 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3314 /* If towupper() is available and handles Unicode, use it. */ 3315 if (!(cmp_flags & CMP_INTERNAL)) 3316 return towupper(a); 3317 #endif 3318 3319 /* For characters below 128 use locale sensitive toupper(). */ 3320 if (a < 128) 3321 return TOUPPER_LOC(a); 3322 3323 /* For any other characters use the above mapping table. */ 3324 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3325 } 3326 3327 int 3328 utf_islower(int a) 3329 { 3330 /* German sharp s is lower case but has no upper case equivalent. */ 3331 return (utf_toupper(a) != a) || a == 0xdf; 3332 } 3333 3334 /* 3335 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3336 * simple case folding. 3337 */ 3338 int 3339 utf_tolower(int a) 3340 { 3341 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */ 3342 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3343 return TOLOWER_ASC(a); 3344 3345 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3346 /* If towlower() is available and handles Unicode, use it. */ 3347 if (!(cmp_flags & CMP_INTERNAL)) 3348 return towlower(a); 3349 #endif 3350 3351 /* For characters below 128 use locale sensitive tolower(). */ 3352 if (a < 128) 3353 return TOLOWER_LOC(a); 3354 3355 /* For any other characters use the above mapping table. */ 3356 return utf_convert(a, toLower, (int)sizeof(toLower)); 3357 } 3358 3359 int 3360 utf_isupper(int a) 3361 { 3362 return (utf_tolower(a) != a); 3363 } 3364 3365 static int 3366 utf_strnicmp( 3367 char_u *s1, 3368 char_u *s2, 3369 size_t n1, 3370 size_t n2) 3371 { 3372 int c1, c2, cdiff; 3373 char_u buffer[6]; 3374 3375 for (;;) 3376 { 3377 c1 = utf_safe_read_char_adv(&s1, &n1); 3378 c2 = utf_safe_read_char_adv(&s2, &n2); 3379 3380 if (c1 <= 0 || c2 <= 0) 3381 break; 3382 3383 if (c1 == c2) 3384 continue; 3385 3386 cdiff = utf_fold(c1) - utf_fold(c2); 3387 if (cdiff != 0) 3388 return cdiff; 3389 } 3390 3391 /* some string ended or has an incomplete/illegal character sequence */ 3392 3393 if (c1 == 0 || c2 == 0) 3394 { 3395 /* some string ended. shorter string is smaller */ 3396 if (c1 == 0 && c2 == 0) 3397 return 0; 3398 return c1 == 0 ? -1 : 1; 3399 } 3400 3401 /* Continue with bytewise comparison to produce some result that 3402 * would make comparison operations involving this function transitive. 3403 * 3404 * If only one string had an error, comparison should be made with 3405 * folded version of the other string. In this case it is enough 3406 * to fold just one character to determine the result of comparison. */ 3407 3408 if (c1 != -1 && c2 == -1) 3409 { 3410 n1 = utf_char2bytes(utf_fold(c1), buffer); 3411 s1 = buffer; 3412 } 3413 else if (c2 != -1 && c1 == -1) 3414 { 3415 n2 = utf_char2bytes(utf_fold(c2), buffer); 3416 s2 = buffer; 3417 } 3418 3419 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3420 { 3421 cdiff = (int)(*s1) - (int)(*s2); 3422 if (cdiff != 0) 3423 return cdiff; 3424 3425 s1++; 3426 s2++; 3427 n1--; 3428 n2--; 3429 } 3430 3431 if (n1 > 0 && *s1 == NUL) 3432 n1 = 0; 3433 if (n2 > 0 && *s2 == NUL) 3434 n2 = 0; 3435 3436 if (n1 == 0 && n2 == 0) 3437 return 0; 3438 return n1 == 0 ? -1 : 1; 3439 } 3440 3441 /* 3442 * Version of strnicmp() that handles multi-byte characters. 3443 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3444 * probably use strnicmp(), because there are no ASCII characters in the 3445 * second byte. 3446 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3447 * two characters otherwise. 3448 */ 3449 int 3450 mb_strnicmp(char_u *s1, char_u *s2, size_t nn) 3451 { 3452 int i, l; 3453 int cdiff; 3454 int n = (int)nn; 3455 3456 if (enc_utf8) 3457 { 3458 return utf_strnicmp(s1, s2, nn, nn); 3459 } 3460 else 3461 { 3462 for (i = 0; i < n; i += l) 3463 { 3464 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */ 3465 return 0; 3466 3467 l = (*mb_ptr2len)(s1 + i); 3468 if (l <= 1) 3469 { 3470 /* Single byte: first check normally, then with ignore case. */ 3471 if (s1[i] != s2[i]) 3472 { 3473 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3474 if (cdiff != 0) 3475 return cdiff; 3476 } 3477 } 3478 else 3479 { 3480 /* For non-Unicode multi-byte don't ignore case. */ 3481 if (l > n - i) 3482 l = n - i; 3483 cdiff = STRNCMP(s1 + i, s2 + i, l); 3484 if (cdiff != 0) 3485 return cdiff; 3486 } 3487 } 3488 } 3489 return 0; 3490 } 3491 3492 /* 3493 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3494 * 'encoding' has been set to. 3495 */ 3496 void 3497 show_utf8(void) 3498 { 3499 int len; 3500 int rlen = 0; 3501 char_u *line; 3502 int clen; 3503 int i; 3504 3505 /* Get the byte length of the char under the cursor, including composing 3506 * characters. */ 3507 line = ml_get_cursor(); 3508 len = utfc_ptr2len(line); 3509 if (len == 0) 3510 { 3511 MSG("NUL"); 3512 return; 3513 } 3514 3515 clen = 0; 3516 for (i = 0; i < len; ++i) 3517 { 3518 if (clen == 0) 3519 { 3520 /* start of (composing) character, get its length */ 3521 if (i > 0) 3522 { 3523 STRCPY(IObuff + rlen, "+ "); 3524 rlen += 2; 3525 } 3526 clen = utf_ptr2len(line + i); 3527 } 3528 sprintf((char *)IObuff + rlen, "%02x ", 3529 (line[i] == NL) ? NUL : line[i]); /* NUL is stored as NL */ 3530 --clen; 3531 rlen += (int)STRLEN(IObuff + rlen); 3532 if (rlen > IOSIZE - 20) 3533 break; 3534 } 3535 3536 msg(IObuff); 3537 } 3538 3539 /* 3540 * mb_head_off() function pointer. 3541 * Return offset from "p" to the first byte of the character it points into. 3542 * If "p" points to the NUL at the end of the string return 0. 3543 * Returns 0 when already at the first byte of a character. 3544 */ 3545 int 3546 latin_head_off(char_u *base UNUSED, char_u *p UNUSED) 3547 { 3548 return 0; 3549 } 3550 3551 int 3552 dbcs_head_off(char_u *base, char_u *p) 3553 { 3554 char_u *q; 3555 3556 /* It can't be a trailing byte when not using DBCS, at the start of the 3557 * string or the previous byte can't start a double-byte. */ 3558 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3559 return 0; 3560 3561 /* This is slow: need to start at the base and go forward until the 3562 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */ 3563 q = base; 3564 while (q < p) 3565 q += dbcs_ptr2len(q); 3566 return (q == p) ? 0 : 1; 3567 } 3568 3569 /* 3570 * Special version of dbcs_head_off() that works for ScreenLines[], where 3571 * single-width DBCS_JPNU characters are stored separately. 3572 */ 3573 int 3574 dbcs_screen_head_off(char_u *base, char_u *p) 3575 { 3576 char_u *q; 3577 3578 /* It can't be a trailing byte when not using DBCS, at the start of the 3579 * string or the previous byte can't start a double-byte. 3580 * For euc-jp an 0x8e byte in the previous cell always means we have a 3581 * lead byte in the current cell. */ 3582 if (p <= base 3583 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3584 || MB_BYTE2LEN(p[-1]) == 1 3585 || *p == NUL) 3586 return 0; 3587 3588 /* This is slow: need to start at the base and go forward until the 3589 * byte we are looking for. Return 1 when we went past it, 0 otherwise. 3590 * For DBCS_JPNU look out for 0x8e, which means the second byte is not 3591 * stored as the next byte. */ 3592 q = base; 3593 while (q < p) 3594 { 3595 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3596 ++q; 3597 else 3598 q += dbcs_ptr2len(q); 3599 } 3600 return (q == p) ? 0 : 1; 3601 } 3602 3603 int 3604 utf_head_off(char_u *base, char_u *p) 3605 { 3606 char_u *q; 3607 char_u *s; 3608 int c; 3609 int len; 3610 #ifdef FEAT_ARABIC 3611 char_u *j; 3612 #endif 3613 3614 if (*p < 0x80) /* be quick for ASCII */ 3615 return 0; 3616 3617 /* Skip backwards over trailing bytes: 10xx.xxxx 3618 * Skip backwards again if on a composing char. */ 3619 for (q = p; ; --q) 3620 { 3621 /* Move s to the last byte of this char. */ 3622 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3623 ; 3624 /* Move q to the first byte of this char. */ 3625 while (q > base && (*q & 0xc0) == 0x80) 3626 --q; 3627 /* Check for illegal sequence. Do allow an illegal byte after where we 3628 * started. */ 3629 len = utf8len_tab[*q]; 3630 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3631 return 0; 3632 3633 if (q <= base) 3634 break; 3635 3636 c = utf_ptr2char(q); 3637 if (utf_iscomposing(c)) 3638 continue; 3639 3640 #ifdef FEAT_ARABIC 3641 if (arabic_maycombine(c)) 3642 { 3643 /* Advance to get a sneak-peak at the next char */ 3644 j = q; 3645 --j; 3646 /* Move j to the first byte of this char. */ 3647 while (j > base && (*j & 0xc0) == 0x80) 3648 --j; 3649 if (arabic_combine(utf_ptr2char(j), c)) 3650 continue; 3651 } 3652 #endif 3653 break; 3654 } 3655 3656 return (int)(p - q); 3657 } 3658 3659 /* 3660 * Copy a character from "*fp" to "*tp" and advance the pointers. 3661 */ 3662 void 3663 mb_copy_char(char_u **fp, char_u **tp) 3664 { 3665 int l = (*mb_ptr2len)(*fp); 3666 3667 mch_memmove(*tp, *fp, (size_t)l); 3668 *tp += l; 3669 *fp += l; 3670 } 3671 3672 /* 3673 * Return the offset from "p" to the first byte of a character. When "p" is 3674 * at the start of a character 0 is returned, otherwise the offset to the next 3675 * character. Can start anywhere in a stream of bytes. 3676 */ 3677 int 3678 mb_off_next(char_u *base, char_u *p) 3679 { 3680 int i; 3681 int j; 3682 3683 if (enc_utf8) 3684 { 3685 if (*p < 0x80) /* be quick for ASCII */ 3686 return 0; 3687 3688 /* Find the next character that isn't 10xx.xxxx */ 3689 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 3690 ; 3691 if (i > 0) 3692 { 3693 /* Check for illegal sequence. */ 3694 for (j = 0; p - j > base; ++j) 3695 if ((p[-j] & 0xc0) != 0x80) 3696 break; 3697 if (utf8len_tab[p[-j]] != i + j) 3698 return 0; 3699 } 3700 return i; 3701 } 3702 3703 /* Only need to check if we're on a trail byte, it doesn't matter if we 3704 * want the offset to the next or current character. */ 3705 return (*mb_head_off)(base, p); 3706 } 3707 3708 /* 3709 * Return the offset from "p" to the last byte of the character it points 3710 * into. Can start anywhere in a stream of bytes. 3711 */ 3712 int 3713 mb_tail_off(char_u *base, char_u *p) 3714 { 3715 int i; 3716 int j; 3717 3718 if (*p == NUL) 3719 return 0; 3720 3721 if (enc_utf8) 3722 { 3723 /* Find the last character that is 10xx.xxxx */ 3724 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 3725 ; 3726 /* Check for illegal sequence. */ 3727 for (j = 0; p - j > base; ++j) 3728 if ((p[-j] & 0xc0) != 0x80) 3729 break; 3730 if (utf8len_tab[p[-j]] != i + j + 1) 3731 return 0; 3732 return i; 3733 } 3734 3735 /* It can't be the first byte if a double-byte when not using DBCS, at the 3736 * end of the string or the byte can't start a double-byte. */ 3737 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 3738 return 0; 3739 3740 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3741 return 1 - dbcs_head_off(base, p); 3742 } 3743 3744 /* 3745 * Find the next illegal byte sequence. 3746 */ 3747 void 3748 utf_find_illegal(void) 3749 { 3750 pos_T pos = curwin->w_cursor; 3751 char_u *p; 3752 int len; 3753 vimconv_T vimconv; 3754 char_u *tofree = NULL; 3755 3756 vimconv.vc_type = CONV_NONE; 3757 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 3758 { 3759 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 3760 * possibly a utf-8 file with illegal bytes. Setup for conversion 3761 * from utf-8 to 'fileencoding'. */ 3762 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 3763 } 3764 3765 #ifdef FEAT_VIRTUALEDIT 3766 curwin->w_cursor.coladd = 0; 3767 #endif 3768 for (;;) 3769 { 3770 p = ml_get_cursor(); 3771 if (vimconv.vc_type != CONV_NONE) 3772 { 3773 vim_free(tofree); 3774 tofree = string_convert(&vimconv, p, NULL); 3775 if (tofree == NULL) 3776 break; 3777 p = tofree; 3778 } 3779 3780 while (*p != NUL) 3781 { 3782 /* Illegal means that there are not enough trail bytes (checked by 3783 * utf_ptr2len()) or too many of them (overlong sequence). */ 3784 len = utf_ptr2len(p); 3785 if (*p >= 0x80 && (len == 1 3786 || utf_char2len(utf_ptr2char(p)) != len)) 3787 { 3788 if (vimconv.vc_type == CONV_NONE) 3789 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 3790 else 3791 { 3792 int l; 3793 3794 len = (int)(p - tofree); 3795 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 3796 { 3797 l = utf_ptr2len(p); 3798 curwin->w_cursor.col += l; 3799 } 3800 } 3801 goto theend; 3802 } 3803 p += len; 3804 } 3805 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 3806 break; 3807 ++curwin->w_cursor.lnum; 3808 curwin->w_cursor.col = 0; 3809 } 3810 3811 /* didn't find it: don't move and beep */ 3812 curwin->w_cursor = pos; 3813 beep_flush(); 3814 3815 theend: 3816 vim_free(tofree); 3817 convert_setup(&vimconv, NULL, NULL); 3818 } 3819 3820 #if defined(FEAT_GUI_GTK) || defined(PROTO) 3821 /* 3822 * Return TRUE if string "s" is a valid utf-8 string. 3823 * When "end" is NULL stop at the first NUL. 3824 * When "end" is positive stop there. 3825 */ 3826 int 3827 utf_valid_string(char_u *s, char_u *end) 3828 { 3829 int l; 3830 char_u *p = s; 3831 3832 while (end == NULL ? *p != NUL : p < end) 3833 { 3834 l = utf8len_tab_zero[*p]; 3835 if (l == 0) 3836 return FALSE; /* invalid lead byte */ 3837 if (end != NULL && p + l > end) 3838 return FALSE; /* incomplete byte sequence */ 3839 ++p; 3840 while (--l > 0) 3841 if ((*p++ & 0xc0) != 0x80) 3842 return FALSE; /* invalid trail byte */ 3843 } 3844 return TRUE; 3845 } 3846 #endif 3847 3848 #if defined(FEAT_GUI) || defined(PROTO) 3849 /* 3850 * Special version of mb_tail_off() for use in ScreenLines[]. 3851 */ 3852 int 3853 dbcs_screen_tail_off(char_u *base, char_u *p) 3854 { 3855 /* It can't be the first byte if a double-byte when not using DBCS, at the 3856 * end of the string or the byte can't start a double-byte. 3857 * For euc-jp an 0x8e byte always means we have a lead byte in the current 3858 * cell. */ 3859 if (*p == NUL || p[1] == NUL 3860 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 3861 || MB_BYTE2LEN(*p) == 1) 3862 return 0; 3863 3864 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3865 return 1 - dbcs_screen_head_off(base, p); 3866 } 3867 #endif 3868 3869 /* 3870 * If the cursor moves on an trail byte, set the cursor on the lead byte. 3871 * Thus it moves left if necessary. 3872 * Return TRUE when the cursor was adjusted. 3873 */ 3874 void 3875 mb_adjust_cursor(void) 3876 { 3877 mb_adjustpos(curbuf, &curwin->w_cursor); 3878 } 3879 3880 /* 3881 * Adjust position "*lp" to point to the first byte of a multi-byte character. 3882 * If it points to a tail byte it's moved backwards to the head byte. 3883 */ 3884 void 3885 mb_adjustpos(buf_T *buf, pos_T *lp) 3886 { 3887 char_u *p; 3888 3889 if (lp->col > 0 3890 #ifdef FEAT_VIRTUALEDIT 3891 || lp->coladd > 1 3892 #endif 3893 ) 3894 { 3895 p = ml_get_buf(buf, lp->lnum, FALSE); 3896 lp->col -= (*mb_head_off)(p, p + lp->col); 3897 #ifdef FEAT_VIRTUALEDIT 3898 /* Reset "coladd" when the cursor would be on the right half of a 3899 * double-wide character. */ 3900 if (lp->coladd == 1 3901 && p[lp->col] != TAB 3902 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 3903 && ptr2cells(p + lp->col) > 1) 3904 lp->coladd = 0; 3905 #endif 3906 } 3907 } 3908 3909 /* 3910 * Return a pointer to the character before "*p", if there is one. 3911 */ 3912 char_u * 3913 mb_prevptr( 3914 char_u *line, /* start of the string */ 3915 char_u *p) 3916 { 3917 if (p > line) 3918 mb_ptr_back(line, p); 3919 return p; 3920 } 3921 3922 /* 3923 * Return the character length of "str". Each multi-byte character (with 3924 * following composing characters) counts as one. 3925 */ 3926 int 3927 mb_charlen(char_u *str) 3928 { 3929 char_u *p = str; 3930 int count; 3931 3932 if (p == NULL) 3933 return 0; 3934 3935 for (count = 0; *p != NUL; count++) 3936 p += (*mb_ptr2len)(p); 3937 3938 return count; 3939 } 3940 3941 #if defined(FEAT_SPELL) || defined(PROTO) 3942 /* 3943 * Like mb_charlen() but for a string with specified length. 3944 */ 3945 int 3946 mb_charlen_len(char_u *str, int len) 3947 { 3948 char_u *p = str; 3949 int count; 3950 3951 for (count = 0; *p != NUL && p < str + len; count++) 3952 p += (*mb_ptr2len)(p); 3953 3954 return count; 3955 } 3956 #endif 3957 3958 /* 3959 * Try to un-escape a multi-byte character. 3960 * Used for the "to" and "from" part of a mapping. 3961 * Return the un-escaped string if it is a multi-byte character, and advance 3962 * "pp" to just after the bytes that formed it. 3963 * Return NULL if no multi-byte char was found. 3964 */ 3965 char_u * 3966 mb_unescape(char_u **pp) 3967 { 3968 static char_u buf[6]; 3969 int n; 3970 int m = 0; 3971 char_u *str = *pp; 3972 3973 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 3974 * KS_EXTRA KE_CSI to CSI. 3975 * Maximum length of a utf-8 character is 4 bytes. */ 3976 for (n = 0; str[n] != NUL && m < 4; ++n) 3977 { 3978 if (str[n] == K_SPECIAL 3979 && str[n + 1] == KS_SPECIAL 3980 && str[n + 2] == KE_FILLER) 3981 { 3982 buf[m++] = K_SPECIAL; 3983 n += 2; 3984 } 3985 else if ((str[n] == K_SPECIAL 3986 # ifdef FEAT_GUI 3987 || str[n] == CSI 3988 # endif 3989 ) 3990 && str[n + 1] == KS_EXTRA 3991 && str[n + 2] == (int)KE_CSI) 3992 { 3993 buf[m++] = CSI; 3994 n += 2; 3995 } 3996 else if (str[n] == K_SPECIAL 3997 # ifdef FEAT_GUI 3998 || str[n] == CSI 3999 # endif 4000 ) 4001 break; /* a special key can't be a multibyte char */ 4002 else 4003 buf[m++] = str[n]; 4004 buf[m] = NUL; 4005 4006 /* Return a multi-byte character if it's found. An illegal sequence 4007 * will result in a 1 here. */ 4008 if ((*mb_ptr2len)(buf) > 1) 4009 { 4010 *pp = str + n + 1; 4011 return buf; 4012 } 4013 4014 /* Bail out quickly for ASCII. */ 4015 if (buf[0] < 128) 4016 break; 4017 } 4018 return NULL; 4019 } 4020 4021 /* 4022 * Return TRUE if the character at "row"/"col" on the screen is the left side 4023 * of a double-width character. 4024 * Caller must make sure "row" and "col" are not invalid! 4025 */ 4026 int 4027 mb_lefthalve(int row, int col) 4028 { 4029 #ifdef FEAT_HANGULIN 4030 if (composing_hangul) 4031 return TRUE; 4032 #endif 4033 return (*mb_off2cells)(LineOffset[row] + col, 4034 LineOffset[row] + screen_Columns) > 1; 4035 } 4036 4037 /* 4038 * Correct a position on the screen, if it's the right half of a double-wide 4039 * char move it to the left half. Returns the corrected column. 4040 */ 4041 int 4042 mb_fix_col(int col, int row) 4043 { 4044 col = check_col(col); 4045 row = check_row(row); 4046 if (has_mbyte && ScreenLines != NULL && col > 0 4047 && ((enc_dbcs 4048 && ScreenLines[LineOffset[row] + col] != NUL 4049 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 4050 ScreenLines + LineOffset[row] + col)) 4051 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0))) 4052 return col - 1; 4053 return col; 4054 } 4055 #endif 4056 4057 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 4058 static int enc_alias_search(char_u *name); 4059 4060 /* 4061 * Skip the Vim specific head of a 'encoding' name. 4062 */ 4063 char_u * 4064 enc_skip(char_u *p) 4065 { 4066 if (STRNCMP(p, "2byte-", 6) == 0) 4067 return p + 6; 4068 if (STRNCMP(p, "8bit-", 5) == 0) 4069 return p + 5; 4070 return p; 4071 } 4072 4073 /* 4074 * Find the canonical name for encoding "enc". 4075 * When the name isn't recognized, returns "enc" itself, but with all lower 4076 * case characters and '_' replaced with '-'. 4077 * Returns an allocated string. NULL for out-of-memory. 4078 */ 4079 char_u * 4080 enc_canonize(char_u *enc) 4081 { 4082 char_u *r; 4083 char_u *p, *s; 4084 int i; 4085 4086 # ifdef FEAT_MBYTE 4087 if (STRCMP(enc, "default") == 0) 4088 { 4089 /* Use the default encoding as it's found by set_init_1(). */ 4090 r = get_encoding_default(); 4091 if (r == NULL) 4092 r = (char_u *)"latin1"; 4093 return vim_strsave(r); 4094 } 4095 # endif 4096 4097 /* copy "enc" to allocated memory, with room for two '-' */ 4098 r = alloc((unsigned)(STRLEN(enc) + 3)); 4099 if (r != NULL) 4100 { 4101 /* Make it all lower case and replace '_' with '-'. */ 4102 p = r; 4103 for (s = enc; *s != NUL; ++s) 4104 { 4105 if (*s == '_') 4106 *p++ = '-'; 4107 else 4108 *p++ = TOLOWER_ASC(*s); 4109 } 4110 *p = NUL; 4111 4112 /* Skip "2byte-" and "8bit-". */ 4113 p = enc_skip(r); 4114 4115 /* Change "microsoft-cp" to "cp". Used in some spell files. */ 4116 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4117 STRMOVE(p, p + 10); 4118 4119 /* "iso8859" -> "iso-8859" */ 4120 if (STRNCMP(p, "iso8859", 7) == 0) 4121 { 4122 STRMOVE(p + 4, p + 3); 4123 p[3] = '-'; 4124 } 4125 4126 /* "iso-8859n" -> "iso-8859-n" */ 4127 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4128 { 4129 STRMOVE(p + 9, p + 8); 4130 p[8] = '-'; 4131 } 4132 4133 /* "latin-N" -> "latinN" */ 4134 if (STRNCMP(p, "latin-", 6) == 0) 4135 STRMOVE(p + 5, p + 6); 4136 4137 if (enc_canon_search(p) >= 0) 4138 { 4139 /* canonical name can be used unmodified */ 4140 if (p != r) 4141 STRMOVE(r, p); 4142 } 4143 else if ((i = enc_alias_search(p)) >= 0) 4144 { 4145 /* alias recognized, get canonical name */ 4146 vim_free(r); 4147 r = vim_strsave((char_u *)enc_canon_table[i].name); 4148 } 4149 } 4150 return r; 4151 } 4152 4153 /* 4154 * Search for an encoding alias of "name". 4155 * Returns -1 when not found. 4156 */ 4157 static int 4158 enc_alias_search(char_u *name) 4159 { 4160 int i; 4161 4162 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4163 if (STRCMP(name, enc_alias_table[i].name) == 0) 4164 return enc_alias_table[i].canon; 4165 return -1; 4166 } 4167 #endif 4168 4169 #if defined(FEAT_MBYTE) || defined(PROTO) 4170 4171 #ifdef HAVE_LANGINFO_H 4172 # include <langinfo.h> 4173 #endif 4174 4175 /* 4176 * Get the canonicalized encoding of the current locale. 4177 * Returns an allocated string when successful, NULL when not. 4178 */ 4179 char_u * 4180 enc_locale(void) 4181 { 4182 #ifndef WIN3264 4183 char *s; 4184 char *p; 4185 int i; 4186 #endif 4187 char buf[50]; 4188 #ifdef WIN3264 4189 long acp = GetACP(); 4190 4191 if (acp == 1200) 4192 STRCPY(buf, "ucs-2le"); 4193 else if (acp == 1252) /* cp1252 is used as latin1 */ 4194 STRCPY(buf, "latin1"); 4195 else 4196 sprintf(buf, "cp%ld", acp); 4197 #else 4198 # ifdef HAVE_NL_LANGINFO_CODESET 4199 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4200 # endif 4201 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4202 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4203 # endif 4204 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4205 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4206 s = getenv("LANG"); 4207 4208 if (s == NULL || *s == NUL) 4209 return FAIL; 4210 4211 /* The most generic locale format is: 4212 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4213 * If there is a '.' remove the part before it. 4214 * if there is something after the codeset, remove it. 4215 * Make the name lowercase and replace '_' with '-'. 4216 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4217 * "ko_KR.EUC" == "euc-kr" 4218 */ 4219 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4220 { 4221 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4222 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4223 { 4224 /* copy "XY.EUC" to "euc-XY" to buf[10] */ 4225 STRCPY(buf + 10, "euc-"); 4226 buf[14] = p[-2]; 4227 buf[15] = p[-1]; 4228 buf[16] = 0; 4229 s = buf + 10; 4230 } 4231 else 4232 s = p + 1; 4233 } 4234 for (i = 0; s[i] != NUL && i < (int)sizeof(buf) - 1; ++i) 4235 { 4236 if (s[i] == '_' || s[i] == '-') 4237 buf[i] = '-'; 4238 else if (isalnum((int)s[i])) 4239 buf[i] = TOLOWER_ASC(s[i]); 4240 else 4241 break; 4242 } 4243 buf[i] = NUL; 4244 #endif 4245 4246 return enc_canonize((char_u *)buf); 4247 } 4248 4249 #if defined(WIN3264) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4250 /* 4251 * Convert an encoding name to an MS-Windows codepage. 4252 * Returns zero if no codepage can be figured out. 4253 */ 4254 int 4255 encname2codepage(char_u *name) 4256 { 4257 int cp; 4258 char_u *p = name; 4259 int idx; 4260 4261 if (STRNCMP(p, "8bit-", 5) == 0) 4262 p += 5; 4263 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4264 p += 6; 4265 4266 if (p[0] == 'c' && p[1] == 'p') 4267 cp = atoi((char *)p + 2); 4268 else if ((idx = enc_canon_search(p)) >= 0) 4269 cp = enc_canon_table[idx].codepage; 4270 else 4271 return 0; 4272 if (IsValidCodePage(cp)) 4273 return cp; 4274 return 0; 4275 } 4276 #endif 4277 4278 # if defined(USE_ICONV) || defined(PROTO) 4279 4280 static char_u *iconv_string(vimconv_T *vcp, char_u *str, int slen, int *unconvlenp, int *resultlenp); 4281 4282 /* 4283 * Call iconv_open() with a check if iconv() works properly (there are broken 4284 * versions). 4285 * Returns (void *)-1 if failed. 4286 * (should return iconv_t, but that causes problems with prototypes). 4287 */ 4288 void * 4289 my_iconv_open(char_u *to, char_u *from) 4290 { 4291 iconv_t fd; 4292 #define ICONV_TESTLEN 400 4293 char_u tobuf[ICONV_TESTLEN]; 4294 char *p; 4295 size_t tolen; 4296 static int iconv_ok = -1; 4297 4298 if (iconv_ok == FALSE) 4299 return (void *)-1; /* detected a broken iconv() previously */ 4300 4301 #ifdef DYNAMIC_ICONV 4302 /* Check if the iconv.dll can be found. */ 4303 if (!iconv_enabled(TRUE)) 4304 return (void *)-1; 4305 #endif 4306 4307 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4308 4309 if (fd != (iconv_t)-1 && iconv_ok == -1) 4310 { 4311 /* 4312 * Do a dummy iconv() call to check if it actually works. There is a 4313 * version of iconv() on Linux that is broken. We can't ignore it, 4314 * because it's wide-spread. The symptoms are that after outputting 4315 * the initial shift state the "to" pointer is NULL and conversion 4316 * stops for no apparent reason after about 8160 characters. 4317 */ 4318 p = (char *)tobuf; 4319 tolen = ICONV_TESTLEN; 4320 (void)iconv(fd, NULL, NULL, &p, &tolen); 4321 if (p == NULL) 4322 { 4323 iconv_ok = FALSE; 4324 iconv_close(fd); 4325 fd = (iconv_t)-1; 4326 } 4327 else 4328 iconv_ok = TRUE; 4329 } 4330 4331 return (void *)fd; 4332 } 4333 4334 /* 4335 * Convert the string "str[slen]" with iconv(). 4336 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4337 * sequence and set "*unconvlenp" to the length of it. 4338 * Returns the converted string in allocated memory. NULL for an error. 4339 * If resultlenp is not NULL, sets it to the result length in bytes. 4340 */ 4341 static char_u * 4342 iconv_string( 4343 vimconv_T *vcp, 4344 char_u *str, 4345 int slen, 4346 int *unconvlenp, 4347 int *resultlenp) 4348 { 4349 const char *from; 4350 size_t fromlen; 4351 char *to; 4352 size_t tolen; 4353 size_t len = 0; 4354 size_t done = 0; 4355 char_u *result = NULL; 4356 char_u *p; 4357 int l; 4358 4359 from = (char *)str; 4360 fromlen = slen; 4361 for (;;) 4362 { 4363 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4364 { 4365 /* Allocate enough room for most conversions. When re-allocating 4366 * increase the buffer size. */ 4367 len = len + fromlen * 2 + 40; 4368 p = alloc((unsigned)len); 4369 if (p != NULL && done > 0) 4370 mch_memmove(p, result, done); 4371 vim_free(result); 4372 result = p; 4373 if (result == NULL) /* out of memory */ 4374 break; 4375 } 4376 4377 to = (char *)result + done; 4378 tolen = len - done - 2; 4379 /* Avoid a warning for systems with a wrong iconv() prototype by 4380 * casting the second argument to void *. */ 4381 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4382 != (size_t)-1) 4383 { 4384 /* Finished, append a NUL. */ 4385 *to = NUL; 4386 break; 4387 } 4388 4389 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4390 * iconv library may use one of them. */ 4391 if (!vcp->vc_fail && unconvlenp != NULL 4392 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4393 { 4394 /* Handle an incomplete sequence at the end. */ 4395 *to = NUL; 4396 *unconvlenp = (int)fromlen; 4397 break; 4398 } 4399 4400 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4401 * iconv library may use one of them. */ 4402 else if (!vcp->vc_fail 4403 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4404 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4405 { 4406 /* Can't convert: insert a '?' and skip a character. This assumes 4407 * conversion from 'encoding' to something else. In other 4408 * situations we don't know what to skip anyway. */ 4409 *to++ = '?'; 4410 if ((*mb_ptr2cells)((char_u *)from) > 1) 4411 *to++ = '?'; 4412 if (enc_utf8) 4413 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4414 else 4415 { 4416 l = (*mb_ptr2len)((char_u *)from); 4417 if (l > (int)fromlen) 4418 l = (int)fromlen; 4419 } 4420 from += l; 4421 fromlen -= l; 4422 } 4423 else if (ICONV_ERRNO != ICONV_E2BIG) 4424 { 4425 /* conversion failed */ 4426 vim_free(result); 4427 result = NULL; 4428 break; 4429 } 4430 /* Not enough room or skipping illegal sequence. */ 4431 done = to - (char *)result; 4432 } 4433 4434 if (resultlenp != NULL && result != NULL) 4435 *resultlenp = (int)(to - (char *)result); 4436 return result; 4437 } 4438 4439 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4440 /* 4441 * Dynamically load the "iconv.dll" on Win32. 4442 */ 4443 4444 # ifndef DYNAMIC_ICONV /* must be generating prototypes */ 4445 # define HINSTANCE int 4446 # endif 4447 static HINSTANCE hIconvDLL = 0; 4448 static HINSTANCE hMsvcrtDLL = 0; 4449 4450 # ifndef DYNAMIC_ICONV_DLL 4451 # define DYNAMIC_ICONV_DLL "iconv.dll" 4452 # define DYNAMIC_ICONV_DLL_ALT1 "libiconv.dll" 4453 # define DYNAMIC_ICONV_DLL_ALT2 "libiconv2.dll" 4454 # define DYNAMIC_ICONV_DLL_ALT3 "libiconv-2.dll" 4455 # endif 4456 # ifndef DYNAMIC_MSVCRT_DLL 4457 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4458 # endif 4459 4460 /* 4461 * Get the address of 'funcname' which is imported by 'hInst' DLL. 4462 */ 4463 static void * 4464 get_iconv_import_func(HINSTANCE hInst, const char *funcname) 4465 { 4466 PBYTE pImage = (PBYTE)hInst; 4467 PIMAGE_DOS_HEADER pDOS = (PIMAGE_DOS_HEADER)hInst; 4468 PIMAGE_NT_HEADERS pPE; 4469 PIMAGE_IMPORT_DESCRIPTOR pImpDesc; 4470 PIMAGE_THUNK_DATA pIAT; /* Import Address Table */ 4471 PIMAGE_THUNK_DATA pINT; /* Import Name Table */ 4472 PIMAGE_IMPORT_BY_NAME pImpName; 4473 4474 if (pDOS->e_magic != IMAGE_DOS_SIGNATURE) 4475 return NULL; 4476 pPE = (PIMAGE_NT_HEADERS)(pImage + pDOS->e_lfanew); 4477 if (pPE->Signature != IMAGE_NT_SIGNATURE) 4478 return NULL; 4479 pImpDesc = (PIMAGE_IMPORT_DESCRIPTOR)(pImage 4480 + pPE->OptionalHeader.DataDirectory[IMAGE_DIRECTORY_ENTRY_IMPORT] 4481 .VirtualAddress); 4482 for (; pImpDesc->FirstThunk; ++pImpDesc) 4483 { 4484 if (!pImpDesc->OriginalFirstThunk) 4485 continue; 4486 pIAT = (PIMAGE_THUNK_DATA)(pImage + pImpDesc->FirstThunk); 4487 pINT = (PIMAGE_THUNK_DATA)(pImage + pImpDesc->OriginalFirstThunk); 4488 for (; pIAT->u1.Function; ++pIAT, ++pINT) 4489 { 4490 if (IMAGE_SNAP_BY_ORDINAL(pINT->u1.Ordinal)) 4491 continue; 4492 pImpName = (PIMAGE_IMPORT_BY_NAME)(pImage 4493 + (UINT_PTR)(pINT->u1.AddressOfData)); 4494 if (strcmp((char *)pImpName->Name, funcname) == 0) 4495 return (void *)pIAT->u1.Function; 4496 } 4497 } 4498 return NULL; 4499 } 4500 4501 /* 4502 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4503 */ 4504 int 4505 iconv_enabled(int verbose) 4506 { 4507 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4508 return TRUE; 4509 4510 /* The iconv DLL file goes under different names, try them all. 4511 * Do the "2" version first, it's newer. */ 4512 #ifdef DYNAMIC_ICONV_DLL_ALT2 4513 if (hIconvDLL == 0) 4514 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT2); 4515 #endif 4516 #ifdef DYNAMIC_ICONV_DLL_ALT3 4517 if (hIconvDLL == 0) 4518 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT3); 4519 #endif 4520 if (hIconvDLL == 0) 4521 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4522 #ifdef DYNAMIC_ICONV_DLL_ALT1 4523 if (hIconvDLL == 0) 4524 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT1); 4525 #endif 4526 4527 if (hIconvDLL != 0) 4528 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4529 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4530 { 4531 /* Only give the message when 'verbose' is set, otherwise it might be 4532 * done whenever a conversion is attempted. */ 4533 if (verbose && p_verbose > 0) 4534 { 4535 verbose_enter(); 4536 EMSG2(_(e_loadlib), 4537 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4538 verbose_leave(); 4539 } 4540 iconv_end(); 4541 return FALSE; 4542 } 4543 4544 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4545 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4546 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4547 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4548 iconv_errno = get_iconv_import_func(hIconvDLL, "_errno"); 4549 if (iconv_errno == NULL) 4550 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4551 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4552 || iconvctl == NULL || iconv_errno == NULL) 4553 { 4554 iconv_end(); 4555 if (verbose && p_verbose > 0) 4556 { 4557 verbose_enter(); 4558 EMSG2(_(e_loadfunc), "for libiconv"); 4559 verbose_leave(); 4560 } 4561 return FALSE; 4562 } 4563 return TRUE; 4564 } 4565 4566 void 4567 iconv_end(void) 4568 { 4569 /* Don't use iconv() when inputting or outputting characters. */ 4570 if (input_conv.vc_type == CONV_ICONV) 4571 convert_setup(&input_conv, NULL, NULL); 4572 if (output_conv.vc_type == CONV_ICONV) 4573 convert_setup(&output_conv, NULL, NULL); 4574 4575 if (hIconvDLL != 0) 4576 FreeLibrary(hIconvDLL); 4577 if (hMsvcrtDLL != 0) 4578 FreeLibrary(hMsvcrtDLL); 4579 hIconvDLL = 0; 4580 hMsvcrtDLL = 0; 4581 } 4582 # endif /* DYNAMIC_ICONV */ 4583 # endif /* USE_ICONV */ 4584 4585 #endif /* FEAT_MBYTE */ 4586 4587 #if defined(FEAT_XIM) || defined(PROTO) 4588 4589 # if defined(FEAT_GUI_GTK) || defined(PROTO) 4590 static int xim_has_preediting INIT(= FALSE); /* IM current status */ 4591 4592 /* 4593 * Set preedit_start_col to the current cursor position. 4594 */ 4595 static void 4596 init_preedit_start_col(void) 4597 { 4598 if (State & CMDLINE) 4599 preedit_start_col = cmdline_getvcol_cursor(); 4600 else if (curwin != NULL && curwin->w_buffer != NULL) 4601 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); 4602 /* Prevent that preediting marks the buffer as changed. */ 4603 xim_changed_while_preediting = curbuf->b_changed; 4604 } 4605 4606 static int im_is_active = FALSE; /* IM is enabled for current mode */ 4607 static int preedit_is_active = FALSE; 4608 static int im_preedit_cursor = 0; /* cursor offset in characters */ 4609 static int im_preedit_trailing = 0; /* number of characters after cursor */ 4610 4611 static unsigned long im_commit_handler_id = 0; 4612 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; 4613 static unsigned int im_activatekey_state = 0; 4614 4615 void 4616 im_set_active(int active) 4617 { 4618 int was_active; 4619 4620 was_active = !!im_get_status(); 4621 im_is_active = (active && !p_imdisable); 4622 4623 if (im_is_active != was_active) 4624 xim_reset(); 4625 } 4626 4627 void 4628 xim_set_focus(int focus) 4629 { 4630 if (xic != NULL) 4631 { 4632 if (focus) 4633 gtk_im_context_focus_in(xic); 4634 else 4635 gtk_im_context_focus_out(xic); 4636 } 4637 } 4638 4639 void 4640 im_set_position(int row, int col) 4641 { 4642 if (xic != NULL) 4643 { 4644 GdkRectangle area; 4645 4646 area.x = FILL_X(col); 4647 area.y = FILL_Y(row); 4648 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); 4649 area.height = gui.char_height; 4650 4651 gtk_im_context_set_cursor_location(xic, &area); 4652 } 4653 } 4654 4655 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */ 4656 void 4657 xim_set_preedit(void) 4658 { 4659 im_set_position(gui.row, gui.col); 4660 } 4661 # endif 4662 4663 static void 4664 im_add_to_input(char_u *str, int len) 4665 { 4666 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */ 4667 if (input_conv.vc_type != CONV_NONE) 4668 { 4669 str = string_convert(&input_conv, str, &len); 4670 g_return_if_fail(str != NULL); 4671 } 4672 4673 add_to_input_buf_csi(str, len); 4674 4675 if (input_conv.vc_type != CONV_NONE) 4676 vim_free(str); 4677 4678 if (p_mh) /* blank out the pointer if necessary */ 4679 gui_mch_mousehide(TRUE); 4680 } 4681 4682 static void 4683 im_delete_preedit(void) 4684 { 4685 char_u bskey[] = {CSI, 'k', 'b'}; 4686 char_u delkey[] = {CSI, 'k', 'D'}; 4687 4688 if (State & NORMAL) 4689 { 4690 im_preedit_cursor = 0; 4691 return; 4692 } 4693 for (; im_preedit_cursor > 0; --im_preedit_cursor) 4694 add_to_input_buf(bskey, (int)sizeof(bskey)); 4695 4696 for (; im_preedit_trailing > 0; --im_preedit_trailing) 4697 add_to_input_buf(delkey, (int)sizeof(delkey)); 4698 } 4699 4700 /* 4701 * Move the cursor left by "num_move_back" characters. 4702 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for 4703 * these K_LEFT keys. 4704 */ 4705 static void 4706 im_correct_cursor(int num_move_back) 4707 { 4708 char_u backkey[] = {CSI, 'k', 'l'}; 4709 4710 if (State & NORMAL) 4711 return; 4712 # ifdef FEAT_RIGHTLEFT 4713 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) 4714 backkey[2] = 'r'; 4715 # endif 4716 for (; num_move_back > 0; --num_move_back) 4717 add_to_input_buf(backkey, (int)sizeof(backkey)); 4718 } 4719 4720 static int xim_expected_char = NUL; 4721 static int xim_ignored_char = FALSE; 4722 4723 /* 4724 * Update the mode and cursor while in an IM callback. 4725 */ 4726 static void 4727 im_show_info(void) 4728 { 4729 int old_vgetc_busy; 4730 4731 old_vgetc_busy = vgetc_busy; 4732 vgetc_busy = TRUE; 4733 showmode(); 4734 vgetc_busy = old_vgetc_busy; 4735 if ((State & NORMAL) || (State & INSERT)) 4736 setcursor(); 4737 out_flush(); 4738 } 4739 4740 /* 4741 * Callback invoked when the user finished preediting. 4742 * Put the final string into the input buffer. 4743 */ 4744 static void 4745 im_commit_cb(GtkIMContext *context UNUSED, 4746 const gchar *str, 4747 gpointer data UNUSED) 4748 { 4749 int slen = (int)STRLEN(str); 4750 int add_to_input = TRUE; 4751 int clen; 4752 int len = slen; 4753 int commit_with_preedit = TRUE; 4754 char_u *im_str; 4755 4756 #ifdef XIM_DEBUG 4757 xim_log("im_commit_cb(): %s\n", str); 4758 #endif 4759 4760 /* The imhangul module doesn't reset the preedit string before 4761 * committing. Call im_delete_preedit() to work around that. */ 4762 im_delete_preedit(); 4763 4764 /* Indicate that preediting has finished. */ 4765 if (preedit_start_col == MAXCOL) 4766 { 4767 init_preedit_start_col(); 4768 commit_with_preedit = FALSE; 4769 } 4770 4771 /* The thing which setting "preedit_start_col" to MAXCOL means that 4772 * "preedit_start_col" will be set forcedly when calling 4773 * preedit_changed_cb() next time. 4774 * "preedit_start_col" should not reset with MAXCOL on this part. Vim 4775 * is simulating the preediting by using add_to_input_str(). when 4776 * preedit begin immediately before committed, the typebuf is not 4777 * flushed to screen, then it can't get correct "preedit_start_col". 4778 * Thus, it should calculate the cells by adding cells of the committed 4779 * string. */ 4780 if (input_conv.vc_type != CONV_NONE) 4781 { 4782 im_str = string_convert(&input_conv, (char_u *)str, &len); 4783 g_return_if_fail(im_str != NULL); 4784 } 4785 else 4786 im_str = (char_u *)str; 4787 4788 clen = mb_string2cells(im_str, len); 4789 4790 if (input_conv.vc_type != CONV_NONE) 4791 vim_free(im_str); 4792 preedit_start_col += clen; 4793 4794 /* Is this a single character that matches a keypad key that's just 4795 * been pressed? If so, we don't want it to be entered as such - let 4796 * us carry on processing the raw keycode so that it may be used in 4797 * mappings as <kSomething>. */ 4798 if (xim_expected_char != NUL) 4799 { 4800 /* We're currently processing a keypad or other special key */ 4801 if (slen == 1 && str[0] == xim_expected_char) 4802 { 4803 /* It's a match - don't do it here */ 4804 xim_ignored_char = TRUE; 4805 add_to_input = FALSE; 4806 } 4807 else 4808 { 4809 /* Not a match */ 4810 xim_ignored_char = FALSE; 4811 } 4812 } 4813 4814 if (add_to_input) 4815 im_add_to_input((char_u *)str, slen); 4816 4817 /* Inserting chars while "im_is_active" is set does not cause a change of 4818 * buffer. When the chars are committed the buffer must be marked as 4819 * changed. */ 4820 if (!commit_with_preedit) 4821 preedit_start_col = MAXCOL; 4822 4823 /* This flag is used in changed() at next call. */ 4824 xim_changed_while_preediting = TRUE; 4825 4826 if (gtk_main_level() > 0) 4827 gtk_main_quit(); 4828 } 4829 4830 /* 4831 * Callback invoked after start to the preedit. 4832 */ 4833 static void 4834 im_preedit_start_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4835 { 4836 #ifdef XIM_DEBUG 4837 xim_log("im_preedit_start_cb()\n"); 4838 #endif 4839 4840 im_is_active = TRUE; 4841 preedit_is_active = TRUE; 4842 gui_update_cursor(TRUE, FALSE); 4843 im_show_info(); 4844 } 4845 4846 /* 4847 * Callback invoked after end to the preedit. 4848 */ 4849 static void 4850 im_preedit_end_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4851 { 4852 #ifdef XIM_DEBUG 4853 xim_log("im_preedit_end_cb()\n"); 4854 #endif 4855 im_delete_preedit(); 4856 4857 /* Indicate that preediting has finished */ 4858 preedit_start_col = MAXCOL; 4859 xim_has_preediting = FALSE; 4860 4861 #if 0 4862 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was 4863 * switched off unintentionally. We now use preedit_is_active (added by 4864 * SungHyun Nam). */ 4865 im_is_active = FALSE; 4866 #endif 4867 preedit_is_active = FALSE; 4868 gui_update_cursor(TRUE, FALSE); 4869 im_show_info(); 4870 } 4871 4872 /* 4873 * Callback invoked after changes to the preedit string. If the preedit 4874 * string was empty before, remember the preedit start column so we know 4875 * where to apply feedback attributes. Delete the previous preedit string 4876 * if there was one, save the new preedit cursor offset, and put the new 4877 * string into the input buffer. 4878 * 4879 * TODO: The pragmatic "put into input buffer" approach used here has 4880 * several fundamental problems: 4881 * 4882 * - The characters in the preedit string are subject to remapping. 4883 * That's broken, only the finally committed string should be remapped. 4884 * 4885 * - There is a race condition involved: The retrieved value for the 4886 * current cursor position will be wrong if any unprocessed characters 4887 * are still queued in the input buffer. 4888 * 4889 * - Due to the lack of synchronization between the file buffer in memory 4890 * and any typed characters, it's practically impossible to implement the 4891 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM 4892 * modules for languages such as Thai are likely to rely on this feature 4893 * for proper operation. 4894 * 4895 * Conclusions: I think support for preediting needs to be moved to the 4896 * core parts of Vim. Ideally, until it has been committed, the preediting 4897 * string should only be displayed and not affect the buffer content at all. 4898 * The question how to deal with the synchronization issue still remains. 4899 * Circumventing the input buffer is probably not desirable. Anyway, I think 4900 * implementing "retrieve_surrounding" is the only hard problem. 4901 * 4902 * One way to solve all of this in a clean manner would be to queue all key 4903 * press/release events "as is" in the input buffer, and apply the IM filtering 4904 * at the receiving end of the queue. This, however, would have a rather large 4905 * impact on the code base. If there is an easy way to force processing of all 4906 * remaining input from within the "retrieve_surrounding" signal handler, this 4907 * might not be necessary. Gotta ask on vim-dev for opinions. 4908 */ 4909 static void 4910 im_preedit_changed_cb(GtkIMContext *context, gpointer data UNUSED) 4911 { 4912 char *preedit_string = NULL; 4913 int cursor_index = 0; 4914 int num_move_back = 0; 4915 char_u *str; 4916 char_u *p; 4917 int i; 4918 4919 gtk_im_context_get_preedit_string(context, 4920 &preedit_string, NULL, 4921 &cursor_index); 4922 4923 #ifdef XIM_DEBUG 4924 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); 4925 #endif 4926 4927 g_return_if_fail(preedit_string != NULL); /* just in case */ 4928 4929 /* If preedit_start_col is MAXCOL set it to the current cursor position. */ 4930 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') 4931 { 4932 xim_has_preediting = TRUE; 4933 4934 /* Urgh, this breaks if the input buffer isn't empty now */ 4935 init_preedit_start_col(); 4936 } 4937 else if (cursor_index == 0 && preedit_string[0] == '\0') 4938 { 4939 xim_has_preediting = FALSE; 4940 4941 /* If at the start position (after typing backspace) 4942 * preedit_start_col must be reset. */ 4943 preedit_start_col = MAXCOL; 4944 } 4945 4946 im_delete_preedit(); 4947 4948 /* 4949 * Compute the end of the preediting area: "preedit_end_col". 4950 * According to the documentation of gtk_im_context_get_preedit_string(), 4951 * the cursor_pos output argument returns the offset in bytes. This is 4952 * unfortunately not true -- real life shows the offset is in characters, 4953 * and the GTK+ source code agrees with me. Will file a bug later. 4954 */ 4955 if (preedit_start_col != MAXCOL) 4956 preedit_end_col = preedit_start_col; 4957 str = (char_u *)preedit_string; 4958 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) 4959 { 4960 int is_composing; 4961 4962 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); 4963 /* 4964 * These offsets are used as counters when generating <BS> and <Del> 4965 * to delete the preedit string. So don't count composing characters 4966 * unless 'delcombine' is enabled. 4967 */ 4968 if (!is_composing || p_deco) 4969 { 4970 if (i < cursor_index) 4971 ++im_preedit_cursor; 4972 else 4973 ++im_preedit_trailing; 4974 } 4975 if (!is_composing && i >= cursor_index) 4976 { 4977 /* This is essentially the same as im_preedit_trailing, except 4978 * composing characters are not counted even if p_deco is set. */ 4979 ++num_move_back; 4980 } 4981 if (preedit_start_col != MAXCOL) 4982 preedit_end_col += utf_ptr2cells(p); 4983 } 4984 4985 if (p > str) 4986 { 4987 im_add_to_input(str, (int)(p - str)); 4988 im_correct_cursor(num_move_back); 4989 } 4990 4991 g_free(preedit_string); 4992 4993 if (gtk_main_level() > 0) 4994 gtk_main_quit(); 4995 } 4996 4997 /* 4998 * Translate the Pango attributes at iter to Vim highlighting attributes. 4999 * Ignore attributes not supported by Vim highlighting. This shouldn't have 5000 * too much impact -- right now we handle even more attributes than necessary 5001 * for the IM modules I tested with. 5002 */ 5003 static int 5004 translate_pango_attributes(PangoAttrIterator *iter) 5005 { 5006 PangoAttribute *attr; 5007 int char_attr = HL_NORMAL; 5008 5009 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); 5010 if (attr != NULL && ((PangoAttrInt *)attr)->value 5011 != (int)PANGO_UNDERLINE_NONE) 5012 char_attr |= HL_UNDERLINE; 5013 5014 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); 5015 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) 5016 char_attr |= HL_BOLD; 5017 5018 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); 5019 if (attr != NULL && ((PangoAttrInt *)attr)->value 5020 != (int)PANGO_STYLE_NORMAL) 5021 char_attr |= HL_ITALIC; 5022 5023 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); 5024 if (attr != NULL) 5025 { 5026 const PangoColor *color = &((PangoAttrColor *)attr)->color; 5027 5028 /* Assume inverse if black background is requested */ 5029 if ((color->red | color->green | color->blue) == 0) 5030 char_attr |= HL_INVERSE; 5031 } 5032 5033 return char_attr; 5034 } 5035 5036 /* 5037 * Retrieve the highlighting attributes at column col in the preedit string. 5038 * Return -1 if not in preediting mode or if col is out of range. 5039 */ 5040 int 5041 im_get_feedback_attr(int col) 5042 { 5043 char *preedit_string = NULL; 5044 PangoAttrList *attr_list = NULL; 5045 int char_attr = -1; 5046 5047 if (xic == NULL) 5048 return char_attr; 5049 5050 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5051 5052 if (preedit_string != NULL && attr_list != NULL) 5053 { 5054 int idx; 5055 5056 /* Get the byte index as used by PangoAttrIterator */ 5057 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col) 5058 idx += utfc_ptr2len((char_u *)preedit_string + idx); 5059 5060 if (preedit_string[idx] != '\0') 5061 { 5062 PangoAttrIterator *iter; 5063 int start, end; 5064 5065 char_attr = HL_NORMAL; 5066 iter = pango_attr_list_get_iterator(attr_list); 5067 5068 /* Extract all relevant attributes from the list. */ 5069 do 5070 { 5071 pango_attr_iterator_range(iter, &start, &end); 5072 5073 if (idx >= start && idx < end) 5074 char_attr |= translate_pango_attributes(iter); 5075 } 5076 while (pango_attr_iterator_next(iter)); 5077 5078 pango_attr_iterator_destroy(iter); 5079 } 5080 } 5081 5082 if (attr_list != NULL) 5083 pango_attr_list_unref(attr_list); 5084 g_free(preedit_string); 5085 5086 return char_attr; 5087 } 5088 5089 void 5090 xim_init(void) 5091 { 5092 #ifdef XIM_DEBUG 5093 xim_log("xim_init()\n"); 5094 #endif 5095 5096 g_return_if_fail(gui.drawarea != NULL); 5097 #if GTK_CHECK_VERSION(3,0,0) 5098 g_return_if_fail(gtk_widget_get_window(gui.drawarea) != NULL); 5099 #else 5100 g_return_if_fail(gui.drawarea->window != NULL); 5101 #endif 5102 5103 xic = gtk_im_multicontext_new(); 5104 g_object_ref(xic); 5105 5106 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", 5107 G_CALLBACK(&im_commit_cb), NULL); 5108 g_signal_connect(G_OBJECT(xic), "preedit_changed", 5109 G_CALLBACK(&im_preedit_changed_cb), NULL); 5110 g_signal_connect(G_OBJECT(xic), "preedit_start", 5111 G_CALLBACK(&im_preedit_start_cb), NULL); 5112 g_signal_connect(G_OBJECT(xic), "preedit_end", 5113 G_CALLBACK(&im_preedit_end_cb), NULL); 5114 5115 #if GTK_CHECK_VERSION(3,0,0) 5116 gtk_im_context_set_client_window(xic, gtk_widget_get_window(gui.drawarea)); 5117 #else 5118 gtk_im_context_set_client_window(xic, gui.drawarea->window); 5119 #endif 5120 } 5121 5122 void 5123 im_shutdown(void) 5124 { 5125 #ifdef XIM_DEBUG 5126 xim_log("im_shutdown()\n"); 5127 #endif 5128 5129 if (xic != NULL) 5130 { 5131 gtk_im_context_focus_out(xic); 5132 g_object_unref(xic); 5133 xic = NULL; 5134 } 5135 im_is_active = FALSE; 5136 im_commit_handler_id = 0; 5137 preedit_start_col = MAXCOL; 5138 xim_has_preediting = FALSE; 5139 } 5140 5141 /* 5142 * Convert the string argument to keyval and state for GdkEventKey. 5143 * If str is valid return TRUE, otherwise FALSE. 5144 * 5145 * See 'imactivatekey' for documentation of the format. 5146 */ 5147 static int 5148 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) 5149 { 5150 const char *mods_end; 5151 unsigned tmp_keyval; 5152 unsigned tmp_state = 0; 5153 5154 mods_end = strrchr(str, '-'); 5155 mods_end = (mods_end != NULL) ? mods_end + 1 : str; 5156 5157 /* Parse modifier keys */ 5158 while (str < mods_end) 5159 switch (*str++) 5160 { 5161 case '-': break; 5162 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; 5163 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; 5164 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; 5165 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; 5166 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; 5167 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; 5168 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; 5169 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; 5170 default: 5171 return FALSE; 5172 } 5173 5174 tmp_keyval = gdk_keyval_from_name(str); 5175 5176 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) 5177 return FALSE; 5178 5179 if (keyval != NULL) 5180 *keyval = tmp_keyval; 5181 if (state != NULL) 5182 *state = tmp_state; 5183 5184 return TRUE; 5185 } 5186 5187 /* 5188 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an 5189 * empty string is also regarded as valid. 5190 * 5191 * Note: The numerical key value of p_imak is cached if it was valid; thus 5192 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever 5193 * 'imak' changes. This is currently the case but not obvious -- should 5194 * probably rename the function for clarity. 5195 */ 5196 int 5197 im_xim_isvalid_imactivate(void) 5198 { 5199 if (p_imak[0] == NUL) 5200 { 5201 im_activatekey_keyval = GDK_VoidSymbol; 5202 im_activatekey_state = 0; 5203 return TRUE; 5204 } 5205 5206 return im_string_to_keyval((const char *)p_imak, 5207 &im_activatekey_keyval, 5208 &im_activatekey_state); 5209 } 5210 5211 static void 5212 im_synthesize_keypress(unsigned int keyval, unsigned int state) 5213 { 5214 GdkEventKey *event; 5215 5216 # ifdef HAVE_GTK_MULTIHEAD 5217 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); 5218 # if GTK_CHECK_VERSION(3,0,0) 5219 g_object_ref(gtk_widget_get_window(gui.drawarea)); 5220 /* unreffed by gdk_event_free() */ 5221 # else 5222 g_object_ref(gui.drawarea->window); /* unreffed by gdk_event_free() */ 5223 # endif 5224 # else 5225 event = (GdkEventKey *)g_malloc0((gulong)sizeof(GdkEvent)); 5226 event->type = GDK_KEY_PRESS; 5227 # endif 5228 # if GTK_CHECK_VERSION(3,0,0) 5229 event->window = gtk_widget_get_window(gui.drawarea); 5230 # else 5231 event->window = gui.drawarea->window; 5232 # endif 5233 event->send_event = TRUE; 5234 event->time = GDK_CURRENT_TIME; 5235 event->state = state; 5236 event->keyval = keyval; 5237 event->hardware_keycode = /* needed for XIM */ 5238 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); 5239 event->length = 0; 5240 event->string = NULL; 5241 5242 gtk_im_context_filter_keypress(xic, event); 5243 5244 /* For consistency, also send the corresponding release event. */ 5245 event->type = GDK_KEY_RELEASE; 5246 event->send_event = FALSE; 5247 gtk_im_context_filter_keypress(xic, event); 5248 5249 # ifdef HAVE_GTK_MULTIHEAD 5250 gdk_event_free((GdkEvent *)event); 5251 # else 5252 g_free(event); 5253 # endif 5254 } 5255 5256 void 5257 xim_reset(void) 5258 { 5259 if (xic != NULL) 5260 { 5261 gtk_im_context_reset(xic); 5262 5263 if (p_imdisable) 5264 im_shutdown(); 5265 else 5266 { 5267 xim_set_focus(gui.in_focus); 5268 5269 # ifdef FEAT_EVAL 5270 if (p_imaf[0] != NUL) 5271 { 5272 char_u *argv[1]; 5273 5274 if (im_is_active) 5275 argv[0] = (char_u *)"1"; 5276 else 5277 argv[0] = (char_u *)"0"; 5278 (void)call_func_retnr(p_imaf, 1, argv, FALSE); 5279 } 5280 else 5281 # endif 5282 if (im_activatekey_keyval != GDK_VoidSymbol) 5283 { 5284 if (im_is_active) 5285 { 5286 g_signal_handler_block(xic, im_commit_handler_id); 5287 im_synthesize_keypress(im_activatekey_keyval, 5288 im_activatekey_state); 5289 g_signal_handler_unblock(xic, im_commit_handler_id); 5290 } 5291 } 5292 else 5293 { 5294 im_shutdown(); 5295 xim_init(); 5296 xim_set_focus(gui.in_focus); 5297 } 5298 } 5299 } 5300 5301 preedit_start_col = MAXCOL; 5302 xim_has_preediting = FALSE; 5303 } 5304 5305 int 5306 xim_queue_key_press_event(GdkEventKey *event, int down) 5307 { 5308 if (down) 5309 { 5310 /* 5311 * Workaround GTK2 XIM 'feature' that always converts keypad keys to 5312 * chars., even when not part of an IM sequence (ref. feature of 5313 * gdk/gdkkeyuni.c). 5314 * Flag any keypad keys that might represent a single char. 5315 * If this (on its own - i.e., not part of an IM sequence) is 5316 * committed while we're processing one of these keys, we can ignore 5317 * that commit and go ahead & process it ourselves. That way we can 5318 * still distinguish keypad keys for use in mappings. 5319 * Also add GDK_space to make <S-Space> work. 5320 */ 5321 switch (event->keyval) 5322 { 5323 case GDK_KP_Add: xim_expected_char = '+'; break; 5324 case GDK_KP_Subtract: xim_expected_char = '-'; break; 5325 case GDK_KP_Divide: xim_expected_char = '/'; break; 5326 case GDK_KP_Multiply: xim_expected_char = '*'; break; 5327 case GDK_KP_Decimal: xim_expected_char = '.'; break; 5328 case GDK_KP_Equal: xim_expected_char = '='; break; 5329 case GDK_KP_0: xim_expected_char = '0'; break; 5330 case GDK_KP_1: xim_expected_char = '1'; break; 5331 case GDK_KP_2: xim_expected_char = '2'; break; 5332 case GDK_KP_3: xim_expected_char = '3'; break; 5333 case GDK_KP_4: xim_expected_char = '4'; break; 5334 case GDK_KP_5: xim_expected_char = '5'; break; 5335 case GDK_KP_6: xim_expected_char = '6'; break; 5336 case GDK_KP_7: xim_expected_char = '7'; break; 5337 case GDK_KP_8: xim_expected_char = '8'; break; 5338 case GDK_KP_9: xim_expected_char = '9'; break; 5339 case GDK_space: xim_expected_char = ' '; break; 5340 default: xim_expected_char = NUL; 5341 } 5342 xim_ignored_char = FALSE; 5343 } 5344 5345 /* 5346 * When typing fFtT, XIM may be activated. Thus it must pass 5347 * gtk_im_context_filter_keypress() in Normal mode. 5348 * And while doing :sh too. 5349 */ 5350 if (xic != NULL && !p_imdisable 5351 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) 5352 { 5353 /* 5354 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is 5355 * always aware of the current status of IM, and can even emulate 5356 * the activation key for modules that don't support one. 5357 */ 5358 if (event->keyval == im_activatekey_keyval 5359 && (event->state & im_activatekey_state) == im_activatekey_state) 5360 { 5361 unsigned int state_mask; 5362 5363 /* Require the state of the 3 most used modifiers to match exactly. 5364 * Otherwise e.g. <S-C-space> would be unusable for other purposes 5365 * if the IM activate key is <S-space>. */ 5366 state_mask = im_activatekey_state; 5367 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK 5368 | (int)GDK_MOD1_MASK); 5369 5370 if ((event->state & state_mask) != im_activatekey_state) 5371 return FALSE; 5372 5373 /* Don't send it a second time on GDK_KEY_RELEASE. */ 5374 if (event->type != GDK_KEY_PRESS) 5375 return TRUE; 5376 5377 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE)) 5378 { 5379 im_set_active(FALSE); 5380 5381 /* ":lmap" mappings exists, toggle use of mappings. */ 5382 State ^= LANGMAP; 5383 if (State & LANGMAP) 5384 { 5385 curbuf->b_p_iminsert = B_IMODE_NONE; 5386 State &= ~LANGMAP; 5387 } 5388 else 5389 { 5390 curbuf->b_p_iminsert = B_IMODE_LMAP; 5391 State |= LANGMAP; 5392 } 5393 return TRUE; 5394 } 5395 5396 return gtk_im_context_filter_keypress(xic, event); 5397 } 5398 5399 /* Don't filter events through the IM context if IM isn't active 5400 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module 5401 * not doing anything before the activation key was sent. */ 5402 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) 5403 { 5404 int imresult = gtk_im_context_filter_keypress(xic, event); 5405 5406 /* Some XIM send following sequence: 5407 * 1. preedited string. 5408 * 2. committed string. 5409 * 3. line changed key. 5410 * 4. preedited string. 5411 * 5. remove preedited string. 5412 * if 3, Vim can't move back the above line for 5. 5413 * thus, this part should not parse the key. */ 5414 if (!imresult && preedit_start_col != MAXCOL 5415 && event->keyval == GDK_Return) 5416 { 5417 im_synthesize_keypress(GDK_Return, 0U); 5418 return FALSE; 5419 } 5420 5421 /* If XIM tried to commit a keypad key as a single char., 5422 * ignore it so we can use the keypad key 'raw', for mappings. */ 5423 if (xim_expected_char != NUL && xim_ignored_char) 5424 /* We had a keypad key, and XIM tried to thieve it */ 5425 return FALSE; 5426 5427 /* This is supposed to fix a problem with iBus, that space 5428 * characters don't work in input mode. */ 5429 xim_expected_char = NUL; 5430 5431 /* Normal processing */ 5432 return imresult; 5433 } 5434 } 5435 5436 return FALSE; 5437 } 5438 5439 int 5440 im_get_status(void) 5441 { 5442 # ifdef FEAT_EVAL 5443 if (p_imsf[0] != NUL) 5444 { 5445 int is_active; 5446 5447 /* FIXME: Don't execute user function in unsafe situation. */ 5448 if (exiting 5449 # ifdef FEAT_AUTOCMD 5450 || is_autocmd_blocked() 5451 # endif 5452 ) 5453 return FALSE; 5454 /* FIXME: :py print 'xxx' is shown duplicate result. 5455 * Use silent to avoid it. */ 5456 ++msg_silent; 5457 is_active = call_func_retnr(p_imsf, 0, NULL, FALSE); 5458 --msg_silent; 5459 return (is_active > 0); 5460 } 5461 # endif 5462 return im_is_active; 5463 } 5464 5465 int 5466 preedit_get_status(void) 5467 { 5468 return preedit_is_active; 5469 } 5470 5471 int 5472 im_is_preediting(void) 5473 { 5474 return xim_has_preediting; 5475 } 5476 5477 # else /* !FEAT_GUI_GTK */ 5478 5479 static int xim_is_active = FALSE; /* XIM should be active in the current 5480 mode */ 5481 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */ 5482 #ifdef FEAT_GUI_X11 5483 static XIMStyle input_style; 5484 static int status_area_enabled = TRUE; 5485 #endif 5486 5487 /* 5488 * Switch using XIM on/off. This is used by the code that changes "State". 5489 */ 5490 void 5491 im_set_active(int active) 5492 { 5493 if (xic == NULL) 5494 return; 5495 5496 /* If 'imdisable' is set, XIM is never active. */ 5497 if (p_imdisable) 5498 active = FALSE; 5499 #if !defined(FEAT_GUI_GTK) 5500 else if (input_style & XIMPreeditPosition) 5501 /* There is a problem in switching XIM off when preediting is used, 5502 * and it is not clear how this can be solved. For now, keep XIM on 5503 * all the time, like it was done in Vim 5.8. */ 5504 active = TRUE; 5505 #endif 5506 5507 /* Remember the active state, it is needed when Vim gets keyboard focus. */ 5508 xim_is_active = active; 5509 xim_set_preedit(); 5510 } 5511 5512 /* 5513 * Adjust using XIM for gaining or losing keyboard focus. Also called when 5514 * "xim_is_active" changes. 5515 */ 5516 void 5517 xim_set_focus(int focus) 5518 { 5519 if (xic == NULL) 5520 return; 5521 5522 /* 5523 * XIM only gets focus when the Vim window has keyboard focus and XIM has 5524 * been set active for the current mode. 5525 */ 5526 if (focus && xim_is_active) 5527 { 5528 if (!xim_has_focus) 5529 { 5530 xim_has_focus = TRUE; 5531 XSetICFocus(xic); 5532 } 5533 } 5534 else 5535 { 5536 if (xim_has_focus) 5537 { 5538 xim_has_focus = FALSE; 5539 XUnsetICFocus(xic); 5540 } 5541 } 5542 } 5543 5544 void 5545 im_set_position(int row UNUSED, int col UNUSED) 5546 { 5547 xim_set_preedit(); 5548 } 5549 5550 /* 5551 * Set the XIM to the current cursor position. 5552 */ 5553 void 5554 xim_set_preedit(void) 5555 { 5556 XVaNestedList attr_list; 5557 XRectangle spot_area; 5558 XPoint over_spot; 5559 int line_space; 5560 5561 if (xic == NULL) 5562 return; 5563 5564 xim_set_focus(TRUE); 5565 5566 if (!xim_has_focus) 5567 { 5568 /* hide XIM cursor */ 5569 over_spot.x = 0; 5570 over_spot.y = -100; /* arbitrary invisible position */ 5571 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5572 XNSpotLocation, 5573 &over_spot, 5574 NULL); 5575 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); 5576 XFree(attr_list); 5577 return; 5578 } 5579 5580 if (input_style & XIMPreeditPosition) 5581 { 5582 if (xim_fg_color == INVALCOLOR) 5583 { 5584 xim_fg_color = gui.def_norm_pixel; 5585 xim_bg_color = gui.def_back_pixel; 5586 } 5587 over_spot.x = TEXT_X(gui.col); 5588 over_spot.y = TEXT_Y(gui.row); 5589 spot_area.x = 0; 5590 spot_area.y = 0; 5591 spot_area.height = gui.char_height * Rows; 5592 spot_area.width = gui.char_width * Columns; 5593 line_space = gui.char_height; 5594 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5595 XNSpotLocation, &over_spot, 5596 XNForeground, (Pixel) xim_fg_color, 5597 XNBackground, (Pixel) xim_bg_color, 5598 XNArea, &spot_area, 5599 XNLineSpace, line_space, 5600 NULL); 5601 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) 5602 EMSG(_("E284: Cannot set IC values")); 5603 XFree(attr_list); 5604 } 5605 } 5606 5607 #if defined(FEAT_GUI_X11) 5608 static char e_xim[] = N_("E285: Failed to create input context"); 5609 #endif 5610 5611 #if defined(FEAT_GUI_X11) || defined(PROTO) 5612 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(sun) 5613 # define USE_X11R6_XIM 5614 # endif 5615 5616 static int xim_real_init(Window x11_window, Display *x11_display); 5617 5618 5619 #ifdef USE_X11R6_XIM 5620 static void xim_instantiate_cb(Display *display, XPointer client_data, XPointer call_data); 5621 static void xim_destroy_cb(XIM im, XPointer client_data, XPointer call_data); 5622 5623 static void 5624 xim_instantiate_cb( 5625 Display *display, 5626 XPointer client_data UNUSED, 5627 XPointer call_data UNUSED) 5628 { 5629 Window x11_window; 5630 Display *x11_display; 5631 5632 #ifdef XIM_DEBUG 5633 xim_log("xim_instantiate_cb()\n"); 5634 #endif 5635 5636 gui_get_x11_windis(&x11_window, &x11_display); 5637 if (display != x11_display) 5638 return; 5639 5640 xim_real_init(x11_window, x11_display); 5641 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5642 if (xic != NULL) 5643 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5644 xim_instantiate_cb, NULL); 5645 } 5646 5647 static void 5648 xim_destroy_cb( 5649 XIM im UNUSED, 5650 XPointer client_data UNUSED, 5651 XPointer call_data UNUSED) 5652 { 5653 Window x11_window; 5654 Display *x11_display; 5655 5656 #ifdef XIM_DEBUG 5657 xim_log("xim_destroy_cb()\n"); 5658 #endif 5659 gui_get_x11_windis(&x11_window, &x11_display); 5660 5661 xic = NULL; 5662 status_area_enabled = FALSE; 5663 5664 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5665 5666 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5667 xim_instantiate_cb, NULL); 5668 } 5669 #endif 5670 5671 void 5672 xim_init(void) 5673 { 5674 Window x11_window; 5675 Display *x11_display; 5676 5677 #ifdef XIM_DEBUG 5678 xim_log("xim_init()\n"); 5679 #endif 5680 5681 gui_get_x11_windis(&x11_window, &x11_display); 5682 5683 xic = NULL; 5684 5685 if (xim_real_init(x11_window, x11_display)) 5686 return; 5687 5688 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5689 5690 #ifdef USE_X11R6_XIM 5691 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5692 xim_instantiate_cb, NULL); 5693 #endif 5694 } 5695 5696 static int 5697 xim_real_init(Window x11_window, Display *x11_display) 5698 { 5699 int i; 5700 char *p, 5701 *s, 5702 *ns, 5703 *end, 5704 tmp[1024]; 5705 #define IMLEN_MAX 40 5706 char buf[IMLEN_MAX + 7]; 5707 XIM xim = NULL; 5708 XIMStyles *xim_styles; 5709 XIMStyle this_input_style = 0; 5710 Boolean found; 5711 XPoint over_spot; 5712 XVaNestedList preedit_list, status_list; 5713 5714 input_style = 0; 5715 status_area_enabled = FALSE; 5716 5717 if (xic != NULL) 5718 return FALSE; 5719 5720 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) 5721 { 5722 strcpy(tmp, gui.rsrc_input_method); 5723 for (ns = s = tmp; ns != NULL && *s != NUL;) 5724 { 5725 s = (char *)skipwhite((char_u *)s); 5726 if (*s == NUL) 5727 break; 5728 if ((ns = end = strchr(s, ',')) == NULL) 5729 end = s + strlen(s); 5730 while (isspace(((char_u *)end)[-1])) 5731 end--; 5732 *end = NUL; 5733 5734 if (strlen(s) <= IMLEN_MAX) 5735 { 5736 strcpy(buf, "@im="); 5737 strcat(buf, s); 5738 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL 5739 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) 5740 != NULL) 5741 break; 5742 } 5743 5744 s = ns + 1; 5745 } 5746 } 5747 5748 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) 5749 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5750 5751 /* This is supposed to be useful to obtain characters through 5752 * XmbLookupString() without really using a XIM. */ 5753 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL 5754 && *p != NUL) 5755 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5756 5757 if (xim == NULL) 5758 { 5759 /* Only give this message when verbose is set, because too many people 5760 * got this message when they didn't want to use a XIM. */ 5761 if (p_verbose > 0) 5762 { 5763 verbose_enter(); 5764 EMSG(_("E286: Failed to open input method")); 5765 verbose_leave(); 5766 } 5767 return FALSE; 5768 } 5769 5770 #ifdef USE_X11R6_XIM 5771 { 5772 XIMCallback destroy_cb; 5773 5774 destroy_cb.callback = xim_destroy_cb; 5775 destroy_cb.client_data = NULL; 5776 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) 5777 EMSG(_("E287: Warning: Could not set destroy callback to IM")); 5778 } 5779 #endif 5780 5781 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) 5782 { 5783 EMSG(_("E288: input method doesn't support any style")); 5784 XCloseIM(xim); 5785 return FALSE; 5786 } 5787 5788 found = False; 5789 strcpy(tmp, gui.rsrc_preedit_type_name); 5790 for (s = tmp; s && !found; ) 5791 { 5792 while (*s && isspace((unsigned char)*s)) 5793 s++; 5794 if (!*s) 5795 break; 5796 if ((ns = end = strchr(s, ',')) != 0) 5797 ns++; 5798 else 5799 end = s + strlen(s); 5800 while (isspace((unsigned char)*end)) 5801 end--; 5802 *end = '\0'; 5803 5804 if (!strcmp(s, "OverTheSpot")) 5805 this_input_style = (XIMPreeditPosition | XIMStatusArea); 5806 else if (!strcmp(s, "OffTheSpot")) 5807 this_input_style = (XIMPreeditArea | XIMStatusArea); 5808 else if (!strcmp(s, "Root")) 5809 this_input_style = (XIMPreeditNothing | XIMStatusNothing); 5810 5811 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5812 { 5813 if (this_input_style == xim_styles->supported_styles[i]) 5814 { 5815 found = True; 5816 break; 5817 } 5818 } 5819 if (!found) 5820 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5821 { 5822 if ((xim_styles->supported_styles[i] & this_input_style) 5823 == (this_input_style & ~XIMStatusArea)) 5824 { 5825 this_input_style &= ~XIMStatusArea; 5826 found = True; 5827 break; 5828 } 5829 } 5830 5831 s = ns; 5832 } 5833 XFree(xim_styles); 5834 5835 if (!found) 5836 { 5837 /* Only give this message when verbose is set, because too many people 5838 * got this message when they didn't want to use a XIM. */ 5839 if (p_verbose > 0) 5840 { 5841 verbose_enter(); 5842 EMSG(_("E289: input method doesn't support my preedit type")); 5843 verbose_leave(); 5844 } 5845 XCloseIM(xim); 5846 return FALSE; 5847 } 5848 5849 over_spot.x = TEXT_X(gui.col); 5850 over_spot.y = TEXT_Y(gui.row); 5851 input_style = this_input_style; 5852 5853 /* A crash was reported when trying to pass gui.norm_font as XNFontSet, 5854 * thus that has been removed. Hopefully the default works... */ 5855 #ifdef FEAT_XFONTSET 5856 if (gui.fontset != NOFONTSET) 5857 { 5858 preedit_list = XVaCreateNestedList(0, 5859 XNSpotLocation, &over_spot, 5860 XNForeground, (Pixel)gui.def_norm_pixel, 5861 XNBackground, (Pixel)gui.def_back_pixel, 5862 XNFontSet, (XFontSet)gui.fontset, 5863 NULL); 5864 status_list = XVaCreateNestedList(0, 5865 XNForeground, (Pixel)gui.def_norm_pixel, 5866 XNBackground, (Pixel)gui.def_back_pixel, 5867 XNFontSet, (XFontSet)gui.fontset, 5868 NULL); 5869 } 5870 else 5871 #endif 5872 { 5873 preedit_list = XVaCreateNestedList(0, 5874 XNSpotLocation, &over_spot, 5875 XNForeground, (Pixel)gui.def_norm_pixel, 5876 XNBackground, (Pixel)gui.def_back_pixel, 5877 NULL); 5878 status_list = XVaCreateNestedList(0, 5879 XNForeground, (Pixel)gui.def_norm_pixel, 5880 XNBackground, (Pixel)gui.def_back_pixel, 5881 NULL); 5882 } 5883 5884 xic = XCreateIC(xim, 5885 XNInputStyle, input_style, 5886 XNClientWindow, x11_window, 5887 XNFocusWindow, gui.wid, 5888 XNPreeditAttributes, preedit_list, 5889 XNStatusAttributes, status_list, 5890 NULL); 5891 XFree(status_list); 5892 XFree(preedit_list); 5893 if (xic != NULL) 5894 { 5895 if (input_style & XIMStatusArea) 5896 { 5897 xim_set_status_area(); 5898 status_area_enabled = TRUE; 5899 } 5900 else 5901 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5902 } 5903 else 5904 { 5905 EMSG(_(e_xim)); 5906 XCloseIM(xim); 5907 return FALSE; 5908 } 5909 5910 return TRUE; 5911 } 5912 5913 #endif /* FEAT_GUI_X11 */ 5914 5915 /* 5916 * Get IM status. When IM is on, return TRUE. Else return FALSE. 5917 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is 5918 * active, when not having focus XIM may still be active (e.g., when using a 5919 * tear-off menu item). 5920 */ 5921 int 5922 im_get_status(void) 5923 { 5924 return xim_has_focus; 5925 } 5926 5927 # endif /* !FEAT_GUI_GTK */ 5928 5929 # if !defined(FEAT_GUI_GTK) || defined(PROTO) 5930 /* 5931 * Set up the status area. 5932 * 5933 * This should use a separate Widget, but that seems not possible, because 5934 * preedit_area and status_area should be set to the same window as for the 5935 * text input. Unfortunately this means the status area pollutes the text 5936 * window... 5937 */ 5938 void 5939 xim_set_status_area(void) 5940 { 5941 XVaNestedList preedit_list = 0, status_list = 0, list = 0; 5942 XRectangle pre_area, status_area; 5943 5944 if (xic == NULL) 5945 return; 5946 5947 if (input_style & XIMStatusArea) 5948 { 5949 if (input_style & XIMPreeditArea) 5950 { 5951 XRectangle *needed_rect; 5952 5953 /* to get status_area width */ 5954 status_list = XVaCreateNestedList(0, XNAreaNeeded, 5955 &needed_rect, NULL); 5956 XGetICValues(xic, XNStatusAttributes, status_list, NULL); 5957 XFree(status_list); 5958 5959 status_area.width = needed_rect->width; 5960 } 5961 else 5962 status_area.width = gui.char_width * Columns; 5963 5964 status_area.x = 0; 5965 status_area.y = gui.char_height * Rows + gui.border_offset; 5966 if (gui.which_scrollbars[SBAR_BOTTOM]) 5967 status_area.y += gui.scrollbar_height; 5968 #ifdef FEAT_MENU 5969 if (gui.menu_is_active) 5970 status_area.y += gui.menu_height; 5971 #endif 5972 status_area.height = gui.char_height; 5973 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); 5974 } 5975 else 5976 { 5977 status_area.x = 0; 5978 status_area.y = gui.char_height * Rows + gui.border_offset; 5979 if (gui.which_scrollbars[SBAR_BOTTOM]) 5980 status_area.y += gui.scrollbar_height; 5981 #ifdef FEAT_MENU 5982 if (gui.menu_is_active) 5983 status_area.y += gui.menu_height; 5984 #endif 5985 status_area.width = 0; 5986 status_area.height = gui.char_height; 5987 } 5988 5989 if (input_style & XIMPreeditArea) /* off-the-spot */ 5990 { 5991 pre_area.x = status_area.x + status_area.width; 5992 pre_area.y = gui.char_height * Rows + gui.border_offset; 5993 pre_area.width = gui.char_width * Columns - pre_area.x; 5994 if (gui.which_scrollbars[SBAR_BOTTOM]) 5995 pre_area.y += gui.scrollbar_height; 5996 #ifdef FEAT_MENU 5997 if (gui.menu_is_active) 5998 pre_area.y += gui.menu_height; 5999 #endif 6000 pre_area.height = gui.char_height; 6001 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6002 } 6003 else if (input_style & XIMPreeditPosition) /* over-the-spot */ 6004 { 6005 pre_area.x = 0; 6006 pre_area.y = 0; 6007 pre_area.height = gui.char_height * Rows; 6008 pre_area.width = gui.char_width * Columns; 6009 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6010 } 6011 6012 if (preedit_list && status_list) 6013 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6014 XNStatusAttributes, status_list, NULL); 6015 else if (preedit_list) 6016 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6017 NULL); 6018 else if (status_list) 6019 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, 6020 NULL); 6021 else 6022 list = NULL; 6023 6024 if (list) 6025 { 6026 XSetICValues(xic, XNVaNestedList, list, NULL); 6027 XFree(list); 6028 } 6029 if (status_list) 6030 XFree(status_list); 6031 if (preedit_list) 6032 XFree(preedit_list); 6033 } 6034 6035 int 6036 xim_get_status_area_height(void) 6037 { 6038 if (status_area_enabled) 6039 return gui.char_height; 6040 return 0; 6041 } 6042 # endif 6043 6044 #endif /* FEAT_XIM */ 6045 6046 #if defined(FEAT_MBYTE) || defined(PROTO) 6047 6048 /* 6049 * Setup "vcp" for conversion from "from" to "to". 6050 * The names must have been made canonical with enc_canonize(). 6051 * vcp->vc_type must have been initialized to CONV_NONE. 6052 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 6053 * instead). 6054 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 6055 * Return FAIL when conversion is not supported, OK otherwise. 6056 */ 6057 int 6058 convert_setup(vimconv_T *vcp, char_u *from, char_u *to) 6059 { 6060 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 6061 } 6062 6063 /* 6064 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 6065 * "from" unicode charsets be considered utf-8. Same for "to". 6066 */ 6067 int 6068 convert_setup_ext( 6069 vimconv_T *vcp, 6070 char_u *from, 6071 int from_unicode_is_utf8, 6072 char_u *to, 6073 int to_unicode_is_utf8) 6074 { 6075 int from_prop; 6076 int to_prop; 6077 int from_is_utf8; 6078 int to_is_utf8; 6079 6080 /* Reset to no conversion. */ 6081 # ifdef USE_ICONV 6082 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 6083 iconv_close(vcp->vc_fd); 6084 # endif 6085 vcp->vc_type = CONV_NONE; 6086 vcp->vc_factor = 1; 6087 vcp->vc_fail = FALSE; 6088 6089 /* No conversion when one of the names is empty or they are equal. */ 6090 if (from == NULL || *from == NUL || to == NULL || *to == NUL 6091 || STRCMP(from, to) == 0) 6092 return OK; 6093 6094 from_prop = enc_canon_props(from); 6095 to_prop = enc_canon_props(to); 6096 if (from_unicode_is_utf8) 6097 from_is_utf8 = from_prop & ENC_UNICODE; 6098 else 6099 from_is_utf8 = from_prop == ENC_UNICODE; 6100 if (to_unicode_is_utf8) 6101 to_is_utf8 = to_prop & ENC_UNICODE; 6102 else 6103 to_is_utf8 = to_prop == ENC_UNICODE; 6104 6105 if ((from_prop & ENC_LATIN1) && to_is_utf8) 6106 { 6107 /* Internal latin1 -> utf-8 conversion. */ 6108 vcp->vc_type = CONV_TO_UTF8; 6109 vcp->vc_factor = 2; /* up to twice as long */ 6110 } 6111 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 6112 { 6113 /* Internal latin9 -> utf-8 conversion. */ 6114 vcp->vc_type = CONV_9_TO_UTF8; 6115 vcp->vc_factor = 3; /* up to three as long (euro sign) */ 6116 } 6117 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 6118 { 6119 /* Internal utf-8 -> latin1 conversion. */ 6120 vcp->vc_type = CONV_TO_LATIN1; 6121 } 6122 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 6123 { 6124 /* Internal utf-8 -> latin9 conversion. */ 6125 vcp->vc_type = CONV_TO_LATIN9; 6126 } 6127 #ifdef WIN3264 6128 /* Win32-specific codepage <-> codepage conversion without iconv. */ 6129 else if ((from_is_utf8 || encname2codepage(from) > 0) 6130 && (to_is_utf8 || encname2codepage(to) > 0)) 6131 { 6132 vcp->vc_type = CONV_CODEPAGE; 6133 vcp->vc_factor = 2; /* up to twice as long */ 6134 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 6135 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 6136 } 6137 #endif 6138 #ifdef MACOS_X 6139 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 6140 { 6141 vcp->vc_type = CONV_MAC_LATIN1; 6142 } 6143 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 6144 { 6145 vcp->vc_type = CONV_MAC_UTF8; 6146 vcp->vc_factor = 2; /* up to twice as long */ 6147 } 6148 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 6149 { 6150 vcp->vc_type = CONV_LATIN1_MAC; 6151 } 6152 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 6153 { 6154 vcp->vc_type = CONV_UTF8_MAC; 6155 } 6156 #endif 6157 # ifdef USE_ICONV 6158 else 6159 { 6160 /* Use iconv() for conversion. */ 6161 vcp->vc_fd = (iconv_t)my_iconv_open( 6162 to_is_utf8 ? (char_u *)"utf-8" : to, 6163 from_is_utf8 ? (char_u *)"utf-8" : from); 6164 if (vcp->vc_fd != (iconv_t)-1) 6165 { 6166 vcp->vc_type = CONV_ICONV; 6167 vcp->vc_factor = 4; /* could be longer too... */ 6168 } 6169 } 6170 # endif 6171 if (vcp->vc_type == CONV_NONE) 6172 return FAIL; 6173 6174 return OK; 6175 } 6176 6177 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \ 6178 || defined(PROTO) 6179 /* 6180 * Do conversion on typed input characters in-place. 6181 * The input and output are not NUL terminated! 6182 * Returns the length after conversion. 6183 */ 6184 int 6185 convert_input(char_u *ptr, int len, int maxlen) 6186 { 6187 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 6188 } 6189 #endif 6190 6191 /* 6192 * Like convert_input(), but when there is an incomplete byte sequence at the 6193 * end return that as an allocated string in "restp" and set "*restlenp" to 6194 * the length. If "restp" is NULL it is not used. 6195 */ 6196 int 6197 convert_input_safe( 6198 char_u *ptr, 6199 int len, 6200 int maxlen, 6201 char_u **restp, 6202 int *restlenp) 6203 { 6204 char_u *d; 6205 int dlen = len; 6206 int unconvertlen = 0; 6207 6208 d = string_convert_ext(&input_conv, ptr, &dlen, 6209 restp == NULL ? NULL : &unconvertlen); 6210 if (d != NULL) 6211 { 6212 if (dlen <= maxlen) 6213 { 6214 if (unconvertlen > 0) 6215 { 6216 /* Move the unconverted characters to allocated memory. */ 6217 *restp = alloc(unconvertlen); 6218 if (*restp != NULL) 6219 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 6220 *restlenp = unconvertlen; 6221 } 6222 mch_memmove(ptr, d, dlen); 6223 } 6224 else 6225 /* result is too long, keep the unconverted text (the caller must 6226 * have done something wrong!) */ 6227 dlen = len; 6228 vim_free(d); 6229 } 6230 return dlen; 6231 } 6232 6233 /* 6234 * Convert text "ptr[*lenp]" according to "vcp". 6235 * Returns the result in allocated memory and sets "*lenp". 6236 * When "lenp" is NULL, use NUL terminated strings. 6237 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 6238 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 6239 */ 6240 char_u * 6241 string_convert( 6242 vimconv_T *vcp, 6243 char_u *ptr, 6244 int *lenp) 6245 { 6246 return string_convert_ext(vcp, ptr, lenp, NULL); 6247 } 6248 6249 /* 6250 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 6251 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 6252 * set to the number of remaining bytes. 6253 */ 6254 char_u * 6255 string_convert_ext( 6256 vimconv_T *vcp, 6257 char_u *ptr, 6258 int *lenp, 6259 int *unconvlenp) 6260 { 6261 char_u *retval = NULL; 6262 char_u *d; 6263 int len; 6264 int i; 6265 int l; 6266 int c; 6267 6268 if (lenp == NULL) 6269 len = (int)STRLEN(ptr); 6270 else 6271 len = *lenp; 6272 if (len == 0) 6273 return vim_strsave((char_u *)""); 6274 6275 switch (vcp->vc_type) 6276 { 6277 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */ 6278 retval = alloc(len * 2 + 1); 6279 if (retval == NULL) 6280 break; 6281 d = retval; 6282 for (i = 0; i < len; ++i) 6283 { 6284 c = ptr[i]; 6285 if (c < 0x80) 6286 *d++ = c; 6287 else 6288 { 6289 *d++ = 0xc0 + ((unsigned)c >> 6); 6290 *d++ = 0x80 + (c & 0x3f); 6291 } 6292 } 6293 *d = NUL; 6294 if (lenp != NULL) 6295 *lenp = (int)(d - retval); 6296 break; 6297 6298 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */ 6299 retval = alloc(len * 3 + 1); 6300 if (retval == NULL) 6301 break; 6302 d = retval; 6303 for (i = 0; i < len; ++i) 6304 { 6305 c = ptr[i]; 6306 switch (c) 6307 { 6308 case 0xa4: c = 0x20ac; break; /* euro */ 6309 case 0xa6: c = 0x0160; break; /* S hat */ 6310 case 0xa8: c = 0x0161; break; /* S -hat */ 6311 case 0xb4: c = 0x017d; break; /* Z hat */ 6312 case 0xb8: c = 0x017e; break; /* Z -hat */ 6313 case 0xbc: c = 0x0152; break; /* OE */ 6314 case 0xbd: c = 0x0153; break; /* oe */ 6315 case 0xbe: c = 0x0178; break; /* Y */ 6316 } 6317 d += utf_char2bytes(c, d); 6318 } 6319 *d = NUL; 6320 if (lenp != NULL) 6321 *lenp = (int)(d - retval); 6322 break; 6323 6324 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */ 6325 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */ 6326 retval = alloc(len + 1); 6327 if (retval == NULL) 6328 break; 6329 d = retval; 6330 for (i = 0; i < len; ++i) 6331 { 6332 l = utf_ptr2len_len(ptr + i, len - i); 6333 if (l == 0) 6334 *d++ = NUL; 6335 else if (l == 1) 6336 { 6337 int l_w = utf8len_tab_zero[ptr[i]]; 6338 6339 if (l_w == 0) 6340 { 6341 /* Illegal utf-8 byte cannot be converted */ 6342 vim_free(retval); 6343 return NULL; 6344 } 6345 if (unconvlenp != NULL && l_w > len - i) 6346 { 6347 /* Incomplete sequence at the end. */ 6348 *unconvlenp = len - i; 6349 break; 6350 } 6351 *d++ = ptr[i]; 6352 } 6353 else 6354 { 6355 c = utf_ptr2char(ptr + i); 6356 if (vcp->vc_type == CONV_TO_LATIN9) 6357 switch (c) 6358 { 6359 case 0x20ac: c = 0xa4; break; /* euro */ 6360 case 0x0160: c = 0xa6; break; /* S hat */ 6361 case 0x0161: c = 0xa8; break; /* S -hat */ 6362 case 0x017d: c = 0xb4; break; /* Z hat */ 6363 case 0x017e: c = 0xb8; break; /* Z -hat */ 6364 case 0x0152: c = 0xbc; break; /* OE */ 6365 case 0x0153: c = 0xbd; break; /* oe */ 6366 case 0x0178: c = 0xbe; break; /* Y */ 6367 case 0xa4: 6368 case 0xa6: 6369 case 0xa8: 6370 case 0xb4: 6371 case 0xb8: 6372 case 0xbc: 6373 case 0xbd: 6374 case 0xbe: c = 0x100; break; /* not in latin9 */ 6375 } 6376 if (!utf_iscomposing(c)) /* skip composing chars */ 6377 { 6378 if (c < 0x100) 6379 *d++ = c; 6380 else if (vcp->vc_fail) 6381 { 6382 vim_free(retval); 6383 return NULL; 6384 } 6385 else 6386 { 6387 *d++ = 0xbf; 6388 if (utf_char2cells(c) > 1) 6389 *d++ = '?'; 6390 } 6391 } 6392 i += l - 1; 6393 } 6394 } 6395 *d = NUL; 6396 if (lenp != NULL) 6397 *lenp = (int)(d - retval); 6398 break; 6399 6400 # ifdef MACOS_CONVERT 6401 case CONV_MAC_LATIN1: 6402 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6403 'm', 'l', unconvlenp); 6404 break; 6405 6406 case CONV_LATIN1_MAC: 6407 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6408 'l', 'm', unconvlenp); 6409 break; 6410 6411 case CONV_MAC_UTF8: 6412 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6413 'm', 'u', unconvlenp); 6414 break; 6415 6416 case CONV_UTF8_MAC: 6417 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6418 'u', 'm', unconvlenp); 6419 break; 6420 # endif 6421 6422 # ifdef USE_ICONV 6423 case CONV_ICONV: /* conversion with output_conv.vc_fd */ 6424 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 6425 break; 6426 # endif 6427 # ifdef WIN3264 6428 case CONV_CODEPAGE: /* codepage -> codepage */ 6429 { 6430 int retlen; 6431 int tmp_len; 6432 short_u *tmp; 6433 6434 /* 1. codepage/UTF-8 -> ucs-2. */ 6435 if (vcp->vc_cpfrom == 0) 6436 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 6437 else 6438 { 6439 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 6440 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 6441 (char *)ptr, len, 0, 0); 6442 if (tmp_len == 0 6443 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 6444 { 6445 if (lenp != NULL) 6446 *lenp = 0; 6447 if (unconvlenp != NULL) 6448 *unconvlenp = len; 6449 retval = alloc(1); 6450 if (retval) 6451 retval[0] = NUL; 6452 return retval; 6453 } 6454 } 6455 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len); 6456 if (tmp == NULL) 6457 break; 6458 if (vcp->vc_cpfrom == 0) 6459 utf8_to_utf16(ptr, len, tmp, unconvlenp); 6460 else 6461 MultiByteToWideChar(vcp->vc_cpfrom, 0, 6462 (char *)ptr, len, tmp, tmp_len); 6463 6464 /* 2. ucs-2 -> codepage/UTF-8. */ 6465 if (vcp->vc_cpto == 0) 6466 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 6467 else 6468 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 6469 tmp, tmp_len, 0, 0, 0, 0); 6470 retval = alloc(retlen + 1); 6471 if (retval != NULL) 6472 { 6473 if (vcp->vc_cpto == 0) 6474 utf16_to_utf8(tmp, tmp_len, retval); 6475 else 6476 WideCharToMultiByte(vcp->vc_cpto, 0, 6477 tmp, tmp_len, 6478 (char *)retval, retlen, 0, 0); 6479 retval[retlen] = NUL; 6480 if (lenp != NULL) 6481 *lenp = retlen; 6482 } 6483 vim_free(tmp); 6484 break; 6485 } 6486 # endif 6487 } 6488 6489 return retval; 6490 } 6491 #endif 6492