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 /* Sorted list of non-overlapping intervals of East Asian Ambiguous 1214 * characters, generated with ../runtime/tools/unicode.vim. */ 1215 static struct interval ambiguous[] = 1216 { 1217 {0x00a1, 0x00a1}, 1218 {0x00a4, 0x00a4}, 1219 {0x00a7, 0x00a8}, 1220 {0x00aa, 0x00aa}, 1221 {0x00ad, 0x00ae}, 1222 {0x00b0, 0x00b4}, 1223 {0x00b6, 0x00ba}, 1224 {0x00bc, 0x00bf}, 1225 {0x00c6, 0x00c6}, 1226 {0x00d0, 0x00d0}, 1227 {0x00d7, 0x00d8}, 1228 {0x00de, 0x00e1}, 1229 {0x00e6, 0x00e6}, 1230 {0x00e8, 0x00ea}, 1231 {0x00ec, 0x00ed}, 1232 {0x00f0, 0x00f0}, 1233 {0x00f2, 0x00f3}, 1234 {0x00f7, 0x00fa}, 1235 {0x00fc, 0x00fc}, 1236 {0x00fe, 0x00fe}, 1237 {0x0101, 0x0101}, 1238 {0x0111, 0x0111}, 1239 {0x0113, 0x0113}, 1240 {0x011b, 0x011b}, 1241 {0x0126, 0x0127}, 1242 {0x012b, 0x012b}, 1243 {0x0131, 0x0133}, 1244 {0x0138, 0x0138}, 1245 {0x013f, 0x0142}, 1246 {0x0144, 0x0144}, 1247 {0x0148, 0x014b}, 1248 {0x014d, 0x014d}, 1249 {0x0152, 0x0153}, 1250 {0x0166, 0x0167}, 1251 {0x016b, 0x016b}, 1252 {0x01ce, 0x01ce}, 1253 {0x01d0, 0x01d0}, 1254 {0x01d2, 0x01d2}, 1255 {0x01d4, 0x01d4}, 1256 {0x01d6, 0x01d6}, 1257 {0x01d8, 0x01d8}, 1258 {0x01da, 0x01da}, 1259 {0x01dc, 0x01dc}, 1260 {0x0251, 0x0251}, 1261 {0x0261, 0x0261}, 1262 {0x02c4, 0x02c4}, 1263 {0x02c7, 0x02c7}, 1264 {0x02c9, 0x02cb}, 1265 {0x02cd, 0x02cd}, 1266 {0x02d0, 0x02d0}, 1267 {0x02d8, 0x02db}, 1268 {0x02dd, 0x02dd}, 1269 {0x02df, 0x02df}, 1270 {0x0300, 0x036f}, 1271 {0x0391, 0x03a1}, 1272 {0x03a3, 0x03a9}, 1273 {0x03b1, 0x03c1}, 1274 {0x03c3, 0x03c9}, 1275 {0x0401, 0x0401}, 1276 {0x0410, 0x044f}, 1277 {0x0451, 0x0451}, 1278 {0x2010, 0x2010}, 1279 {0x2013, 0x2016}, 1280 {0x2018, 0x2019}, 1281 {0x201c, 0x201d}, 1282 {0x2020, 0x2022}, 1283 {0x2024, 0x2027}, 1284 {0x2030, 0x2030}, 1285 {0x2032, 0x2033}, 1286 {0x2035, 0x2035}, 1287 {0x203b, 0x203b}, 1288 {0x203e, 0x203e}, 1289 {0x2074, 0x2074}, 1290 {0x207f, 0x207f}, 1291 {0x2081, 0x2084}, 1292 {0x20ac, 0x20ac}, 1293 {0x2103, 0x2103}, 1294 {0x2105, 0x2105}, 1295 {0x2109, 0x2109}, 1296 {0x2113, 0x2113}, 1297 {0x2116, 0x2116}, 1298 {0x2121, 0x2122}, 1299 {0x2126, 0x2126}, 1300 {0x212b, 0x212b}, 1301 {0x2153, 0x2154}, 1302 {0x215b, 0x215e}, 1303 {0x2160, 0x216b}, 1304 {0x2170, 0x2179}, 1305 {0x2189, 0x2189}, 1306 {0x2190, 0x2199}, 1307 {0x21b8, 0x21b9}, 1308 {0x21d2, 0x21d2}, 1309 {0x21d4, 0x21d4}, 1310 {0x21e7, 0x21e7}, 1311 {0x2200, 0x2200}, 1312 {0x2202, 0x2203}, 1313 {0x2207, 0x2208}, 1314 {0x220b, 0x220b}, 1315 {0x220f, 0x220f}, 1316 {0x2211, 0x2211}, 1317 {0x2215, 0x2215}, 1318 {0x221a, 0x221a}, 1319 {0x221d, 0x2220}, 1320 {0x2223, 0x2223}, 1321 {0x2225, 0x2225}, 1322 {0x2227, 0x222c}, 1323 {0x222e, 0x222e}, 1324 {0x2234, 0x2237}, 1325 {0x223c, 0x223d}, 1326 {0x2248, 0x2248}, 1327 {0x224c, 0x224c}, 1328 {0x2252, 0x2252}, 1329 {0x2260, 0x2261}, 1330 {0x2264, 0x2267}, 1331 {0x226a, 0x226b}, 1332 {0x226e, 0x226f}, 1333 {0x2282, 0x2283}, 1334 {0x2286, 0x2287}, 1335 {0x2295, 0x2295}, 1336 {0x2299, 0x2299}, 1337 {0x22a5, 0x22a5}, 1338 {0x22bf, 0x22bf}, 1339 {0x2312, 0x2312}, 1340 {0x2460, 0x24e9}, 1341 {0x24eb, 0x254b}, 1342 {0x2550, 0x2573}, 1343 {0x2580, 0x258f}, 1344 {0x2592, 0x2595}, 1345 {0x25a0, 0x25a1}, 1346 {0x25a3, 0x25a9}, 1347 {0x25b2, 0x25b3}, 1348 {0x25b6, 0x25b7}, 1349 {0x25bc, 0x25bd}, 1350 {0x25c0, 0x25c1}, 1351 {0x25c6, 0x25c8}, 1352 {0x25cb, 0x25cb}, 1353 {0x25ce, 0x25d1}, 1354 {0x25e2, 0x25e5}, 1355 {0x25ef, 0x25ef}, 1356 {0x2605, 0x2606}, 1357 {0x2609, 0x2609}, 1358 {0x260e, 0x260f}, 1359 {0x2614, 0x2615}, 1360 {0x261c, 0x261c}, 1361 {0x261e, 0x261e}, 1362 {0x2640, 0x2640}, 1363 {0x2642, 0x2642}, 1364 {0x2660, 0x2661}, 1365 {0x2663, 0x2665}, 1366 {0x2667, 0x266a}, 1367 {0x266c, 0x266d}, 1368 {0x266f, 0x266f}, 1369 {0x269e, 0x269f}, 1370 {0x26be, 0x26bf}, 1371 {0x26c4, 0x26cd}, 1372 {0x26cf, 0x26e1}, 1373 {0x26e3, 0x26e3}, 1374 {0x26e8, 0x26ff}, 1375 {0x273d, 0x273d}, 1376 {0x2757, 0x2757}, 1377 {0x2776, 0x277f}, 1378 {0x2b55, 0x2b59}, 1379 {0x3248, 0x324f}, 1380 {0xe000, 0xf8ff}, 1381 {0xfe00, 0xfe0f}, 1382 {0xfffd, 0xfffd}, 1383 {0x1f100, 0x1f10a}, 1384 {0x1f110, 0x1f12d}, 1385 {0x1f130, 0x1f169}, 1386 {0x1f170, 0x1f19a}, 1387 {0xe0100, 0xe01ef}, 1388 {0xf0000, 0xffffd}, 1389 {0x100000, 0x10fffd} 1390 }; 1391 1392 /* 1393 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1394 * Returns 4 or 6 for an unprintable character. 1395 * Is only correct for characters >= 0x80. 1396 * When p_ambw is "double", return 2 for a character with East Asian Width 1397 * class 'A'(mbiguous). 1398 */ 1399 int 1400 utf_char2cells(int c) 1401 { 1402 /* Sorted list of non-overlapping intervals of East Asian double width 1403 * characters, generated with ../runtime/tools/unicode.vim. */ 1404 static struct interval doublewidth[] = 1405 { 1406 {0x1100, 0x115f}, 1407 {0x2329, 0x232a}, 1408 {0x2e80, 0x2e99}, 1409 {0x2e9b, 0x2ef3}, 1410 {0x2f00, 0x2fd5}, 1411 {0x2ff0, 0x2ffb}, 1412 {0x3000, 0x303e}, 1413 {0x3041, 0x3096}, 1414 {0x3099, 0x30ff}, 1415 {0x3105, 0x312d}, 1416 {0x3131, 0x318e}, 1417 {0x3190, 0x31ba}, 1418 {0x31c0, 0x31e3}, 1419 {0x31f0, 0x321e}, 1420 {0x3220, 0x3247}, 1421 {0x3250, 0x32fe}, 1422 {0x3300, 0x4dbf}, 1423 {0x4e00, 0xa48c}, 1424 {0xa490, 0xa4c6}, 1425 {0xa960, 0xa97c}, 1426 {0xac00, 0xd7a3}, 1427 {0xf900, 0xfaff}, 1428 {0xfe10, 0xfe19}, 1429 {0xfe30, 0xfe52}, 1430 {0xfe54, 0xfe66}, 1431 {0xfe68, 0xfe6b}, 1432 {0xff01, 0xff60}, 1433 {0xffe0, 0xffe6}, 1434 {0x1b000, 0x1b001}, 1435 {0x1f200, 0x1f202}, 1436 {0x1f210, 0x1f23a}, 1437 {0x1f240, 0x1f248}, 1438 {0x1f250, 0x1f251}, 1439 {0x20000, 0x2fffd}, 1440 {0x30000, 0x3fffd} 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 /* Sorted list of non-overlapping intervals of all Emoji characters, 2482 * based on http://unicode.org/emoji/charts/emoji-list.html */ 2483 static struct interval emoji_all[] = 2484 { 2485 {0x203c, 0x203c}, 2486 {0x2049, 0x2049}, 2487 {0x2122, 0x2122}, 2488 {0x2139, 0x2139}, 2489 {0x2194, 0x2199}, 2490 {0x21a9, 0x21aa}, 2491 {0x231a, 0x231b}, 2492 {0x2328, 0x2328}, 2493 {0x23cf, 0x23cf}, 2494 {0x23e9, 0x23f3}, 2495 {0x24c2, 0x24c2}, 2496 {0x25aa, 0x25ab}, 2497 {0x25b6, 0x25b6}, 2498 {0x25c0, 0x25c0}, 2499 {0x25fb, 0x25fe}, 2500 {0x2600, 0x2604}, 2501 {0x260e, 0x260e}, 2502 {0x2611, 0x2611}, 2503 {0x2614, 0x2615}, 2504 {0x2618, 0x2618}, 2505 {0x261d, 0x261d}, 2506 {0x2620, 0x2620}, 2507 {0x2622, 0x2623}, 2508 {0x2626, 0x2626}, 2509 {0x262a, 0x262a}, 2510 {0x262e, 0x262f}, 2511 {0x2638, 0x263a}, 2512 {0x2648, 0x2653}, 2513 {0x2660, 0x2660}, 2514 {0x2663, 0x2663}, 2515 {0x2665, 0x2666}, 2516 {0x2668, 0x2668}, 2517 {0x267b, 0x267b}, 2518 {0x267f, 0x267f}, 2519 {0x2692, 0x2694}, 2520 {0x2696, 0x2697}, 2521 {0x2699, 0x2699}, 2522 {0x269b, 0x269c}, 2523 {0x26a0, 0x26a1}, 2524 {0x26aa, 0x26ab}, 2525 {0x26b0, 0x26b1}, 2526 {0x26bd, 0x26be}, 2527 {0x26c4, 0x26c5}, 2528 {0x26c8, 0x26c8}, 2529 {0x26ce, 0x26cf}, 2530 {0x26d1, 0x26d1}, 2531 {0x26d3, 0x26d4}, 2532 {0x26e9, 0x26ea}, 2533 {0x26f0, 0x26f5}, 2534 {0x26f7, 0x26fa}, 2535 {0x26fd, 0x26fd}, 2536 {0x2702, 0x2702}, 2537 {0x2705, 0x2705}, 2538 {0x2708, 0x270d}, 2539 {0x270f, 0x270f}, 2540 {0x2712, 0x2712}, 2541 {0x2714, 0x2714}, 2542 {0x2716, 0x2716}, 2543 {0x271d, 0x271d}, 2544 {0x2721, 0x2721}, 2545 {0x2728, 0x2728}, 2546 {0x2733, 0x2734}, 2547 {0x2744, 0x2744}, 2548 {0x2747, 0x2747}, 2549 {0x274c, 0x274c}, 2550 {0x274e, 0x274e}, 2551 {0x2753, 0x2755}, 2552 {0x2757, 0x2757}, 2553 {0x2763, 0x2764}, 2554 {0x2795, 0x2797}, 2555 {0x27a1, 0x27a1}, 2556 {0x27b0, 0x27b0}, 2557 {0x27bf, 0x27bf}, 2558 {0x2934, 0x2935}, 2559 {0x2b05, 0x2b07}, 2560 {0x2b1b, 0x2b1c}, 2561 {0x2b50, 0x2b50}, 2562 {0x2b55, 0x2b55}, 2563 {0x3030, 0x3030}, 2564 {0x303d, 0x303d}, 2565 {0x3297, 0x3297}, 2566 {0x3299, 0x3299}, 2567 {0x1f004, 0x1f004}, 2568 {0x1f0cf, 0x1f0cf}, 2569 {0x1f170, 0x1f171}, 2570 {0x1f17e, 0x1f17f}, 2571 {0x1f18e, 0x1f18e}, 2572 {0x1f191, 0x1f19a}, 2573 {0x1f1e6, 0x1f1ff}, 2574 {0x1f201, 0x1f202}, 2575 {0x1f21a, 0x1f21a}, 2576 {0x1f22f, 0x1f22f}, 2577 {0x1f232, 0x1f23a}, 2578 {0x1f250, 0x1f251}, 2579 {0x1f300, 0x1f320}, 2580 {0x1f330, 0x1f335}, 2581 {0x1f337, 0x1f37c}, 2582 {0x1f380, 0x1f393}, 2583 {0x1f3a0, 0x1f3c4}, 2584 {0x1f3c6, 0x1f3ca}, 2585 {0x1f3e0, 0x1f3f0}, 2586 {0x1f400, 0x1f43e}, 2587 {0x1f440, 0x1f440}, 2588 {0x1f442, 0x1f4f7}, 2589 {0x1f4f9, 0x1f4fc}, 2590 {0x1f500, 0x1f53d}, 2591 {0x1f550, 0x1f567}, 2592 {0x1f5fb, 0x1f640}, 2593 {0x1f645, 0x1f64f}, 2594 {0x1f680, 0x1f6c5} 2595 }; 2596 2597 /* 2598 * Get class of a Unicode character. 2599 * 0: white space 2600 * 1: punctuation 2601 * 2 or bigger: some class of word character. 2602 */ 2603 int 2604 utf_class(int c) 2605 { 2606 /* sorted list of non-overlapping intervals */ 2607 static struct clinterval 2608 { 2609 unsigned int first; 2610 unsigned int last; 2611 unsigned int class; 2612 } classes[] = 2613 { 2614 {0x037e, 0x037e, 1}, /* Greek question mark */ 2615 {0x0387, 0x0387, 1}, /* Greek ano teleia */ 2616 {0x055a, 0x055f, 1}, /* Armenian punctuation */ 2617 {0x0589, 0x0589, 1}, /* Armenian full stop */ 2618 {0x05be, 0x05be, 1}, 2619 {0x05c0, 0x05c0, 1}, 2620 {0x05c3, 0x05c3, 1}, 2621 {0x05f3, 0x05f4, 1}, 2622 {0x060c, 0x060c, 1}, 2623 {0x061b, 0x061b, 1}, 2624 {0x061f, 0x061f, 1}, 2625 {0x066a, 0x066d, 1}, 2626 {0x06d4, 0x06d4, 1}, 2627 {0x0700, 0x070d, 1}, /* Syriac punctuation */ 2628 {0x0964, 0x0965, 1}, 2629 {0x0970, 0x0970, 1}, 2630 {0x0df4, 0x0df4, 1}, 2631 {0x0e4f, 0x0e4f, 1}, 2632 {0x0e5a, 0x0e5b, 1}, 2633 {0x0f04, 0x0f12, 1}, 2634 {0x0f3a, 0x0f3d, 1}, 2635 {0x0f85, 0x0f85, 1}, 2636 {0x104a, 0x104f, 1}, /* Myanmar punctuation */ 2637 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */ 2638 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */ 2639 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */ 2640 {0x1680, 0x1680, 0}, 2641 {0x169b, 0x169c, 1}, 2642 {0x16eb, 0x16ed, 1}, 2643 {0x1735, 0x1736, 1}, 2644 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */ 2645 {0x1800, 0x180a, 1}, /* Mongolian punctuation */ 2646 {0x2000, 0x200b, 0}, /* spaces */ 2647 {0x200c, 0x2027, 1}, /* punctuation and symbols */ 2648 {0x2028, 0x2029, 0}, 2649 {0x202a, 0x202e, 1}, /* punctuation and symbols */ 2650 {0x202f, 0x202f, 0}, 2651 {0x2030, 0x205e, 1}, /* punctuation and symbols */ 2652 {0x205f, 0x205f, 0}, 2653 {0x2060, 0x27ff, 1}, /* punctuation and symbols */ 2654 {0x2070, 0x207f, 0x2070}, /* superscript */ 2655 {0x2080, 0x2094, 0x2080}, /* subscript */ 2656 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */ 2657 {0x2800, 0x28ff, 0x2800}, /* braille */ 2658 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */ 2659 {0x29d8, 0x29db, 1}, 2660 {0x29fc, 0x29fd, 1}, 2661 {0x2e00, 0x2e7f, 1}, /* supplemental punctuation */ 2662 {0x3000, 0x3000, 0}, /* ideographic space */ 2663 {0x3001, 0x3020, 1}, /* ideographic punctuation */ 2664 {0x3030, 0x3030, 1}, 2665 {0x303d, 0x303d, 1}, 2666 {0x3040, 0x309f, 0x3040}, /* Hiragana */ 2667 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */ 2668 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */ 2669 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */ 2670 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */ 2671 {0xfd3e, 0xfd3f, 1}, 2672 {0xfe30, 0xfe6b, 1}, /* punctuation forms */ 2673 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */ 2674 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */ 2675 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */ 2676 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */ 2677 {0x20000, 0x2a6df, 0x4e00}, /* CJK Ideographs */ 2678 {0x2a700, 0x2b73f, 0x4e00}, /* CJK Ideographs */ 2679 {0x2b740, 0x2b81f, 0x4e00}, /* CJK Ideographs */ 2680 {0x2f800, 0x2fa1f, 0x4e00}, /* CJK Ideographs */ 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 int 2718 utf_ambiguous_width(int c) 2719 { 2720 return c >= 0x80 && (intable(ambiguous, sizeof(ambiguous), c) 2721 || intable(emoji_all, sizeof(emoji_all), c)); 2722 } 2723 2724 /* 2725 * Code for Unicode case-dependent operations. Based on notes in 2726 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2727 * This code uses simple case folding, not full case folding. 2728 * Last updated for Unicode 5.2. 2729 */ 2730 2731 /* 2732 * The following tables are built by ../runtime/tools/unicode.vim. 2733 * They must be in numeric order, because we use binary search. 2734 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2735 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2736 * folded/upper/lower by adding 32. 2737 */ 2738 typedef struct 2739 { 2740 int rangeStart; 2741 int rangeEnd; 2742 int step; 2743 int offset; 2744 } convertStruct; 2745 2746 static convertStruct foldCase[] = 2747 { 2748 {0x41,0x5a,1,32}, 2749 {0xb5,0xb5,-1,775}, 2750 {0xc0,0xd6,1,32}, 2751 {0xd8,0xde,1,32}, 2752 {0x100,0x12e,2,1}, 2753 {0x132,0x136,2,1}, 2754 {0x139,0x147,2,1}, 2755 {0x14a,0x176,2,1}, 2756 {0x178,0x178,-1,-121}, 2757 {0x179,0x17d,2,1}, 2758 {0x17f,0x17f,-1,-268}, 2759 {0x181,0x181,-1,210}, 2760 {0x182,0x184,2,1}, 2761 {0x186,0x186,-1,206}, 2762 {0x187,0x187,-1,1}, 2763 {0x189,0x18a,1,205}, 2764 {0x18b,0x18b,-1,1}, 2765 {0x18e,0x18e,-1,79}, 2766 {0x18f,0x18f,-1,202}, 2767 {0x190,0x190,-1,203}, 2768 {0x191,0x191,-1,1}, 2769 {0x193,0x193,-1,205}, 2770 {0x194,0x194,-1,207}, 2771 {0x196,0x196,-1,211}, 2772 {0x197,0x197,-1,209}, 2773 {0x198,0x198,-1,1}, 2774 {0x19c,0x19c,-1,211}, 2775 {0x19d,0x19d,-1,213}, 2776 {0x19f,0x19f,-1,214}, 2777 {0x1a0,0x1a4,2,1}, 2778 {0x1a6,0x1a6,-1,218}, 2779 {0x1a7,0x1a7,-1,1}, 2780 {0x1a9,0x1a9,-1,218}, 2781 {0x1ac,0x1ac,-1,1}, 2782 {0x1ae,0x1ae,-1,218}, 2783 {0x1af,0x1af,-1,1}, 2784 {0x1b1,0x1b2,1,217}, 2785 {0x1b3,0x1b5,2,1}, 2786 {0x1b7,0x1b7,-1,219}, 2787 {0x1b8,0x1bc,4,1}, 2788 {0x1c4,0x1c4,-1,2}, 2789 {0x1c5,0x1c5,-1,1}, 2790 {0x1c7,0x1c7,-1,2}, 2791 {0x1c8,0x1c8,-1,1}, 2792 {0x1ca,0x1ca,-1,2}, 2793 {0x1cb,0x1db,2,1}, 2794 {0x1de,0x1ee,2,1}, 2795 {0x1f1,0x1f1,-1,2}, 2796 {0x1f2,0x1f4,2,1}, 2797 {0x1f6,0x1f6,-1,-97}, 2798 {0x1f7,0x1f7,-1,-56}, 2799 {0x1f8,0x21e,2,1}, 2800 {0x220,0x220,-1,-130}, 2801 {0x222,0x232,2,1}, 2802 {0x23a,0x23a,-1,10795}, 2803 {0x23b,0x23b,-1,1}, 2804 {0x23d,0x23d,-1,-163}, 2805 {0x23e,0x23e,-1,10792}, 2806 {0x241,0x241,-1,1}, 2807 {0x243,0x243,-1,-195}, 2808 {0x244,0x244,-1,69}, 2809 {0x245,0x245,-1,71}, 2810 {0x246,0x24e,2,1}, 2811 {0x345,0x345,-1,116}, 2812 {0x370,0x372,2,1}, 2813 {0x376,0x376,-1,1}, 2814 {0x37f,0x37f,-1,116}, 2815 {0x386,0x386,-1,38}, 2816 {0x388,0x38a,1,37}, 2817 {0x38c,0x38c,-1,64}, 2818 {0x38e,0x38f,1,63}, 2819 {0x391,0x3a1,1,32}, 2820 {0x3a3,0x3ab,1,32}, 2821 {0x3c2,0x3c2,-1,1}, 2822 {0x3cf,0x3cf,-1,8}, 2823 {0x3d0,0x3d0,-1,-30}, 2824 {0x3d1,0x3d1,-1,-25}, 2825 {0x3d5,0x3d5,-1,-15}, 2826 {0x3d6,0x3d6,-1,-22}, 2827 {0x3d8,0x3ee,2,1}, 2828 {0x3f0,0x3f0,-1,-54}, 2829 {0x3f1,0x3f1,-1,-48}, 2830 {0x3f4,0x3f4,-1,-60}, 2831 {0x3f5,0x3f5,-1,-64}, 2832 {0x3f7,0x3f7,-1,1}, 2833 {0x3f9,0x3f9,-1,-7}, 2834 {0x3fa,0x3fa,-1,1}, 2835 {0x3fd,0x3ff,1,-130}, 2836 {0x400,0x40f,1,80}, 2837 {0x410,0x42f,1,32}, 2838 {0x460,0x480,2,1}, 2839 {0x48a,0x4be,2,1}, 2840 {0x4c0,0x4c0,-1,15}, 2841 {0x4c1,0x4cd,2,1}, 2842 {0x4d0,0x52e,2,1}, 2843 {0x531,0x556,1,48}, 2844 {0x10a0,0x10c5,1,7264}, 2845 {0x10c7,0x10cd,6,7264}, 2846 {0x13f8,0x13fd,1,-8}, 2847 {0x1e00,0x1e94,2,1}, 2848 {0x1e9b,0x1e9b,-1,-58}, 2849 {0x1e9e,0x1e9e,-1,-7615}, 2850 {0x1ea0,0x1efe,2,1}, 2851 {0x1f08,0x1f0f,1,-8}, 2852 {0x1f18,0x1f1d,1,-8}, 2853 {0x1f28,0x1f2f,1,-8}, 2854 {0x1f38,0x1f3f,1,-8}, 2855 {0x1f48,0x1f4d,1,-8}, 2856 {0x1f59,0x1f5f,2,-8}, 2857 {0x1f68,0x1f6f,1,-8}, 2858 {0x1f88,0x1f8f,1,-8}, 2859 {0x1f98,0x1f9f,1,-8}, 2860 {0x1fa8,0x1faf,1,-8}, 2861 {0x1fb8,0x1fb9,1,-8}, 2862 {0x1fba,0x1fbb,1,-74}, 2863 {0x1fbc,0x1fbc,-1,-9}, 2864 {0x1fbe,0x1fbe,-1,-7173}, 2865 {0x1fc8,0x1fcb,1,-86}, 2866 {0x1fcc,0x1fcc,-1,-9}, 2867 {0x1fd8,0x1fd9,1,-8}, 2868 {0x1fda,0x1fdb,1,-100}, 2869 {0x1fe8,0x1fe9,1,-8}, 2870 {0x1fea,0x1feb,1,-112}, 2871 {0x1fec,0x1fec,-1,-7}, 2872 {0x1ff8,0x1ff9,1,-128}, 2873 {0x1ffa,0x1ffb,1,-126}, 2874 {0x1ffc,0x1ffc,-1,-9}, 2875 {0x2126,0x2126,-1,-7517}, 2876 {0x212a,0x212a,-1,-8383}, 2877 {0x212b,0x212b,-1,-8262}, 2878 {0x2132,0x2132,-1,28}, 2879 {0x2160,0x216f,1,16}, 2880 {0x2183,0x2183,-1,1}, 2881 {0x24b6,0x24cf,1,26}, 2882 {0x2c00,0x2c2e,1,48}, 2883 {0x2c60,0x2c60,-1,1}, 2884 {0x2c62,0x2c62,-1,-10743}, 2885 {0x2c63,0x2c63,-1,-3814}, 2886 {0x2c64,0x2c64,-1,-10727}, 2887 {0x2c67,0x2c6b,2,1}, 2888 {0x2c6d,0x2c6d,-1,-10780}, 2889 {0x2c6e,0x2c6e,-1,-10749}, 2890 {0x2c6f,0x2c6f,-1,-10783}, 2891 {0x2c70,0x2c70,-1,-10782}, 2892 {0x2c72,0x2c75,3,1}, 2893 {0x2c7e,0x2c7f,1,-10815}, 2894 {0x2c80,0x2ce2,2,1}, 2895 {0x2ceb,0x2ced,2,1}, 2896 {0x2cf2,0xa640,31054,1}, 2897 {0xa642,0xa66c,2,1}, 2898 {0xa680,0xa69a,2,1}, 2899 {0xa722,0xa72e,2,1}, 2900 {0xa732,0xa76e,2,1}, 2901 {0xa779,0xa77b,2,1}, 2902 {0xa77d,0xa77d,-1,-35332}, 2903 {0xa77e,0xa786,2,1}, 2904 {0xa78b,0xa78b,-1,1}, 2905 {0xa78d,0xa78d,-1,-42280}, 2906 {0xa790,0xa792,2,1}, 2907 {0xa796,0xa7a8,2,1}, 2908 {0xa7aa,0xa7aa,-1,-42308}, 2909 {0xa7ab,0xa7ab,-1,-42319}, 2910 {0xa7ac,0xa7ac,-1,-42315}, 2911 {0xa7ad,0xa7ad,-1,-42305}, 2912 {0xa7b0,0xa7b0,-1,-42258}, 2913 {0xa7b1,0xa7b1,-1,-42282}, 2914 {0xa7b2,0xa7b2,-1,-42261}, 2915 {0xa7b3,0xa7b3,-1,928}, 2916 {0xa7b4,0xa7b6,2,1}, 2917 {0xab70,0xabbf,1,-38864}, 2918 {0xff21,0xff3a,1,32}, 2919 {0x10400,0x10427,1,40}, 2920 {0x10c80,0x10cb2,1,64}, 2921 {0x118a0,0x118bf,1,32} 2922 }; 2923 2924 static int utf_convert(int a, convertStruct table[], int tableSize); 2925 static int utf_strnicmp(char_u *s1, char_u *s2, size_t n1, size_t n2); 2926 2927 /* 2928 * Generic conversion function for case operations. 2929 * Return the converted equivalent of "a", which is a UCS-4 character. Use 2930 * the given conversion "table". Uses binary search on "table". 2931 */ 2932 static int 2933 utf_convert( 2934 int a, 2935 convertStruct table[], 2936 int tableSize) 2937 { 2938 int start, mid, end; /* indices into table */ 2939 int entries = tableSize / sizeof(convertStruct); 2940 2941 start = 0; 2942 end = entries; 2943 while (start < end) 2944 { 2945 /* need to search further */ 2946 mid = (end + start) / 2; 2947 if (table[mid].rangeEnd < a) 2948 start = mid + 1; 2949 else 2950 end = mid; 2951 } 2952 if (start < entries 2953 && table[start].rangeStart <= a 2954 && a <= table[start].rangeEnd 2955 && (a - table[start].rangeStart) % table[start].step == 0) 2956 return (a + table[start].offset); 2957 else 2958 return a; 2959 } 2960 2961 /* 2962 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 2963 * simple case folding. 2964 */ 2965 int 2966 utf_fold(int a) 2967 { 2968 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 2969 } 2970 2971 static convertStruct toLower[] = 2972 { 2973 {0x41,0x5a,1,32}, 2974 {0xc0,0xd6,1,32}, 2975 {0xd8,0xde,1,32}, 2976 {0x100,0x12e,2,1}, 2977 {0x130,0x130,-1,-199}, 2978 {0x132,0x136,2,1}, 2979 {0x139,0x147,2,1}, 2980 {0x14a,0x176,2,1}, 2981 {0x178,0x178,-1,-121}, 2982 {0x179,0x17d,2,1}, 2983 {0x181,0x181,-1,210}, 2984 {0x182,0x184,2,1}, 2985 {0x186,0x186,-1,206}, 2986 {0x187,0x187,-1,1}, 2987 {0x189,0x18a,1,205}, 2988 {0x18b,0x18b,-1,1}, 2989 {0x18e,0x18e,-1,79}, 2990 {0x18f,0x18f,-1,202}, 2991 {0x190,0x190,-1,203}, 2992 {0x191,0x191,-1,1}, 2993 {0x193,0x193,-1,205}, 2994 {0x194,0x194,-1,207}, 2995 {0x196,0x196,-1,211}, 2996 {0x197,0x197,-1,209}, 2997 {0x198,0x198,-1,1}, 2998 {0x19c,0x19c,-1,211}, 2999 {0x19d,0x19d,-1,213}, 3000 {0x19f,0x19f,-1,214}, 3001 {0x1a0,0x1a4,2,1}, 3002 {0x1a6,0x1a6,-1,218}, 3003 {0x1a7,0x1a7,-1,1}, 3004 {0x1a9,0x1a9,-1,218}, 3005 {0x1ac,0x1ac,-1,1}, 3006 {0x1ae,0x1ae,-1,218}, 3007 {0x1af,0x1af,-1,1}, 3008 {0x1b1,0x1b2,1,217}, 3009 {0x1b3,0x1b5,2,1}, 3010 {0x1b7,0x1b7,-1,219}, 3011 {0x1b8,0x1bc,4,1}, 3012 {0x1c4,0x1c4,-1,2}, 3013 {0x1c5,0x1c5,-1,1}, 3014 {0x1c7,0x1c7,-1,2}, 3015 {0x1c8,0x1c8,-1,1}, 3016 {0x1ca,0x1ca,-1,2}, 3017 {0x1cb,0x1db,2,1}, 3018 {0x1de,0x1ee,2,1}, 3019 {0x1f1,0x1f1,-1,2}, 3020 {0x1f2,0x1f4,2,1}, 3021 {0x1f6,0x1f6,-1,-97}, 3022 {0x1f7,0x1f7,-1,-56}, 3023 {0x1f8,0x21e,2,1}, 3024 {0x220,0x220,-1,-130}, 3025 {0x222,0x232,2,1}, 3026 {0x23a,0x23a,-1,10795}, 3027 {0x23b,0x23b,-1,1}, 3028 {0x23d,0x23d,-1,-163}, 3029 {0x23e,0x23e,-1,10792}, 3030 {0x241,0x241,-1,1}, 3031 {0x243,0x243,-1,-195}, 3032 {0x244,0x244,-1,69}, 3033 {0x245,0x245,-1,71}, 3034 {0x246,0x24e,2,1}, 3035 {0x370,0x372,2,1}, 3036 {0x376,0x376,-1,1}, 3037 {0x37f,0x37f,-1,116}, 3038 {0x386,0x386,-1,38}, 3039 {0x388,0x38a,1,37}, 3040 {0x38c,0x38c,-1,64}, 3041 {0x38e,0x38f,1,63}, 3042 {0x391,0x3a1,1,32}, 3043 {0x3a3,0x3ab,1,32}, 3044 {0x3cf,0x3cf,-1,8}, 3045 {0x3d8,0x3ee,2,1}, 3046 {0x3f4,0x3f4,-1,-60}, 3047 {0x3f7,0x3f7,-1,1}, 3048 {0x3f9,0x3f9,-1,-7}, 3049 {0x3fa,0x3fa,-1,1}, 3050 {0x3fd,0x3ff,1,-130}, 3051 {0x400,0x40f,1,80}, 3052 {0x410,0x42f,1,32}, 3053 {0x460,0x480,2,1}, 3054 {0x48a,0x4be,2,1}, 3055 {0x4c0,0x4c0,-1,15}, 3056 {0x4c1,0x4cd,2,1}, 3057 {0x4d0,0x52e,2,1}, 3058 {0x531,0x556,1,48}, 3059 {0x10a0,0x10c5,1,7264}, 3060 {0x10c7,0x10cd,6,7264}, 3061 {0x13a0,0x13ef,1,38864}, 3062 {0x13f0,0x13f5,1,8}, 3063 {0x1e00,0x1e94,2,1}, 3064 {0x1e9e,0x1e9e,-1,-7615}, 3065 {0x1ea0,0x1efe,2,1}, 3066 {0x1f08,0x1f0f,1,-8}, 3067 {0x1f18,0x1f1d,1,-8}, 3068 {0x1f28,0x1f2f,1,-8}, 3069 {0x1f38,0x1f3f,1,-8}, 3070 {0x1f48,0x1f4d,1,-8}, 3071 {0x1f59,0x1f5f,2,-8}, 3072 {0x1f68,0x1f6f,1,-8}, 3073 {0x1f88,0x1f8f,1,-8}, 3074 {0x1f98,0x1f9f,1,-8}, 3075 {0x1fa8,0x1faf,1,-8}, 3076 {0x1fb8,0x1fb9,1,-8}, 3077 {0x1fba,0x1fbb,1,-74}, 3078 {0x1fbc,0x1fbc,-1,-9}, 3079 {0x1fc8,0x1fcb,1,-86}, 3080 {0x1fcc,0x1fcc,-1,-9}, 3081 {0x1fd8,0x1fd9,1,-8}, 3082 {0x1fda,0x1fdb,1,-100}, 3083 {0x1fe8,0x1fe9,1,-8}, 3084 {0x1fea,0x1feb,1,-112}, 3085 {0x1fec,0x1fec,-1,-7}, 3086 {0x1ff8,0x1ff9,1,-128}, 3087 {0x1ffa,0x1ffb,1,-126}, 3088 {0x1ffc,0x1ffc,-1,-9}, 3089 {0x2126,0x2126,-1,-7517}, 3090 {0x212a,0x212a,-1,-8383}, 3091 {0x212b,0x212b,-1,-8262}, 3092 {0x2132,0x2132,-1,28}, 3093 {0x2160,0x216f,1,16}, 3094 {0x2183,0x2183,-1,1}, 3095 {0x24b6,0x24cf,1,26}, 3096 {0x2c00,0x2c2e,1,48}, 3097 {0x2c60,0x2c60,-1,1}, 3098 {0x2c62,0x2c62,-1,-10743}, 3099 {0x2c63,0x2c63,-1,-3814}, 3100 {0x2c64,0x2c64,-1,-10727}, 3101 {0x2c67,0x2c6b,2,1}, 3102 {0x2c6d,0x2c6d,-1,-10780}, 3103 {0x2c6e,0x2c6e,-1,-10749}, 3104 {0x2c6f,0x2c6f,-1,-10783}, 3105 {0x2c70,0x2c70,-1,-10782}, 3106 {0x2c72,0x2c75,3,1}, 3107 {0x2c7e,0x2c7f,1,-10815}, 3108 {0x2c80,0x2ce2,2,1}, 3109 {0x2ceb,0x2ced,2,1}, 3110 {0x2cf2,0xa640,31054,1}, 3111 {0xa642,0xa66c,2,1}, 3112 {0xa680,0xa69a,2,1}, 3113 {0xa722,0xa72e,2,1}, 3114 {0xa732,0xa76e,2,1}, 3115 {0xa779,0xa77b,2,1}, 3116 {0xa77d,0xa77d,-1,-35332}, 3117 {0xa77e,0xa786,2,1}, 3118 {0xa78b,0xa78b,-1,1}, 3119 {0xa78d,0xa78d,-1,-42280}, 3120 {0xa790,0xa792,2,1}, 3121 {0xa796,0xa7a8,2,1}, 3122 {0xa7aa,0xa7aa,-1,-42308}, 3123 {0xa7ab,0xa7ab,-1,-42319}, 3124 {0xa7ac,0xa7ac,-1,-42315}, 3125 {0xa7ad,0xa7ad,-1,-42305}, 3126 {0xa7b0,0xa7b0,-1,-42258}, 3127 {0xa7b1,0xa7b1,-1,-42282}, 3128 {0xa7b2,0xa7b2,-1,-42261}, 3129 {0xa7b3,0xa7b3,-1,928}, 3130 {0xa7b4,0xa7b6,2,1}, 3131 {0xff21,0xff3a,1,32}, 3132 {0x10400,0x10427,1,40}, 3133 {0x10c80,0x10cb2,1,64}, 3134 {0x118a0,0x118bf,1,32} 3135 }; 3136 3137 static convertStruct toUpper[] = 3138 { 3139 {0x61,0x7a,1,-32}, 3140 {0xb5,0xb5,-1,743}, 3141 {0xe0,0xf6,1,-32}, 3142 {0xf8,0xfe,1,-32}, 3143 {0xff,0xff,-1,121}, 3144 {0x101,0x12f,2,-1}, 3145 {0x131,0x131,-1,-232}, 3146 {0x133,0x137,2,-1}, 3147 {0x13a,0x148,2,-1}, 3148 {0x14b,0x177,2,-1}, 3149 {0x17a,0x17e,2,-1}, 3150 {0x17f,0x17f,-1,-300}, 3151 {0x180,0x180,-1,195}, 3152 {0x183,0x185,2,-1}, 3153 {0x188,0x18c,4,-1}, 3154 {0x192,0x192,-1,-1}, 3155 {0x195,0x195,-1,97}, 3156 {0x199,0x199,-1,-1}, 3157 {0x19a,0x19a,-1,163}, 3158 {0x19e,0x19e,-1,130}, 3159 {0x1a1,0x1a5,2,-1}, 3160 {0x1a8,0x1ad,5,-1}, 3161 {0x1b0,0x1b4,4,-1}, 3162 {0x1b6,0x1b9,3,-1}, 3163 {0x1bd,0x1bd,-1,-1}, 3164 {0x1bf,0x1bf,-1,56}, 3165 {0x1c5,0x1c5,-1,-1}, 3166 {0x1c6,0x1c6,-1,-2}, 3167 {0x1c8,0x1c8,-1,-1}, 3168 {0x1c9,0x1c9,-1,-2}, 3169 {0x1cb,0x1cb,-1,-1}, 3170 {0x1cc,0x1cc,-1,-2}, 3171 {0x1ce,0x1dc,2,-1}, 3172 {0x1dd,0x1dd,-1,-79}, 3173 {0x1df,0x1ef,2,-1}, 3174 {0x1f2,0x1f2,-1,-1}, 3175 {0x1f3,0x1f3,-1,-2}, 3176 {0x1f5,0x1f9,4,-1}, 3177 {0x1fb,0x21f,2,-1}, 3178 {0x223,0x233,2,-1}, 3179 {0x23c,0x23c,-1,-1}, 3180 {0x23f,0x240,1,10815}, 3181 {0x242,0x247,5,-1}, 3182 {0x249,0x24f,2,-1}, 3183 {0x250,0x250,-1,10783}, 3184 {0x251,0x251,-1,10780}, 3185 {0x252,0x252,-1,10782}, 3186 {0x253,0x253,-1,-210}, 3187 {0x254,0x254,-1,-206}, 3188 {0x256,0x257,1,-205}, 3189 {0x259,0x259,-1,-202}, 3190 {0x25b,0x25b,-1,-203}, 3191 {0x25c,0x25c,-1,42319}, 3192 {0x260,0x260,-1,-205}, 3193 {0x261,0x261,-1,42315}, 3194 {0x263,0x263,-1,-207}, 3195 {0x265,0x265,-1,42280}, 3196 {0x266,0x266,-1,42308}, 3197 {0x268,0x268,-1,-209}, 3198 {0x269,0x269,-1,-211}, 3199 {0x26b,0x26b,-1,10743}, 3200 {0x26c,0x26c,-1,42305}, 3201 {0x26f,0x26f,-1,-211}, 3202 {0x271,0x271,-1,10749}, 3203 {0x272,0x272,-1,-213}, 3204 {0x275,0x275,-1,-214}, 3205 {0x27d,0x27d,-1,10727}, 3206 {0x280,0x283,3,-218}, 3207 {0x287,0x287,-1,42282}, 3208 {0x288,0x288,-1,-218}, 3209 {0x289,0x289,-1,-69}, 3210 {0x28a,0x28b,1,-217}, 3211 {0x28c,0x28c,-1,-71}, 3212 {0x292,0x292,-1,-219}, 3213 {0x29d,0x29d,-1,42261}, 3214 {0x29e,0x29e,-1,42258}, 3215 {0x345,0x345,-1,84}, 3216 {0x371,0x373,2,-1}, 3217 {0x377,0x377,-1,-1}, 3218 {0x37b,0x37d,1,130}, 3219 {0x3ac,0x3ac,-1,-38}, 3220 {0x3ad,0x3af,1,-37}, 3221 {0x3b1,0x3c1,1,-32}, 3222 {0x3c2,0x3c2,-1,-31}, 3223 {0x3c3,0x3cb,1,-32}, 3224 {0x3cc,0x3cc,-1,-64}, 3225 {0x3cd,0x3ce,1,-63}, 3226 {0x3d0,0x3d0,-1,-62}, 3227 {0x3d1,0x3d1,-1,-57}, 3228 {0x3d5,0x3d5,-1,-47}, 3229 {0x3d6,0x3d6,-1,-54}, 3230 {0x3d7,0x3d7,-1,-8}, 3231 {0x3d9,0x3ef,2,-1}, 3232 {0x3f0,0x3f0,-1,-86}, 3233 {0x3f1,0x3f1,-1,-80}, 3234 {0x3f2,0x3f2,-1,7}, 3235 {0x3f3,0x3f3,-1,-116}, 3236 {0x3f5,0x3f5,-1,-96}, 3237 {0x3f8,0x3fb,3,-1}, 3238 {0x430,0x44f,1,-32}, 3239 {0x450,0x45f,1,-80}, 3240 {0x461,0x481,2,-1}, 3241 {0x48b,0x4bf,2,-1}, 3242 {0x4c2,0x4ce,2,-1}, 3243 {0x4cf,0x4cf,-1,-15}, 3244 {0x4d1,0x52f,2,-1}, 3245 {0x561,0x586,1,-48}, 3246 {0x13f8,0x13fd,1,-8}, 3247 {0x1d79,0x1d79,-1,35332}, 3248 {0x1d7d,0x1d7d,-1,3814}, 3249 {0x1e01,0x1e95,2,-1}, 3250 {0x1e9b,0x1e9b,-1,-59}, 3251 {0x1ea1,0x1eff,2,-1}, 3252 {0x1f00,0x1f07,1,8}, 3253 {0x1f10,0x1f15,1,8}, 3254 {0x1f20,0x1f27,1,8}, 3255 {0x1f30,0x1f37,1,8}, 3256 {0x1f40,0x1f45,1,8}, 3257 {0x1f51,0x1f57,2,8}, 3258 {0x1f60,0x1f67,1,8}, 3259 {0x1f70,0x1f71,1,74}, 3260 {0x1f72,0x1f75,1,86}, 3261 {0x1f76,0x1f77,1,100}, 3262 {0x1f78,0x1f79,1,128}, 3263 {0x1f7a,0x1f7b,1,112}, 3264 {0x1f7c,0x1f7d,1,126}, 3265 {0x1f80,0x1f87,1,8}, 3266 {0x1f90,0x1f97,1,8}, 3267 {0x1fa0,0x1fa7,1,8}, 3268 {0x1fb0,0x1fb1,1,8}, 3269 {0x1fb3,0x1fb3,-1,9}, 3270 {0x1fbe,0x1fbe,-1,-7205}, 3271 {0x1fc3,0x1fc3,-1,9}, 3272 {0x1fd0,0x1fd1,1,8}, 3273 {0x1fe0,0x1fe1,1,8}, 3274 {0x1fe5,0x1fe5,-1,7}, 3275 {0x1ff3,0x1ff3,-1,9}, 3276 {0x214e,0x214e,-1,-28}, 3277 {0x2170,0x217f,1,-16}, 3278 {0x2184,0x2184,-1,-1}, 3279 {0x24d0,0x24e9,1,-26}, 3280 {0x2c30,0x2c5e,1,-48}, 3281 {0x2c61,0x2c61,-1,-1}, 3282 {0x2c65,0x2c65,-1,-10795}, 3283 {0x2c66,0x2c66,-1,-10792}, 3284 {0x2c68,0x2c6c,2,-1}, 3285 {0x2c73,0x2c76,3,-1}, 3286 {0x2c81,0x2ce3,2,-1}, 3287 {0x2cec,0x2cee,2,-1}, 3288 {0x2cf3,0x2cf3,-1,-1}, 3289 {0x2d00,0x2d25,1,-7264}, 3290 {0x2d27,0x2d2d,6,-7264}, 3291 {0xa641,0xa66d,2,-1}, 3292 {0xa681,0xa69b,2,-1}, 3293 {0xa723,0xa72f,2,-1}, 3294 {0xa733,0xa76f,2,-1}, 3295 {0xa77a,0xa77c,2,-1}, 3296 {0xa77f,0xa787,2,-1}, 3297 {0xa78c,0xa791,5,-1}, 3298 {0xa793,0xa797,4,-1}, 3299 {0xa799,0xa7a9,2,-1}, 3300 {0xa7b5,0xa7b7,2,-1}, 3301 {0xab53,0xab53,-1,-928}, 3302 {0xab70,0xabbf,1,-38864}, 3303 {0xff41,0xff5a,1,-32}, 3304 {0x10428,0x1044f,1,-40}, 3305 {0x10cc0,0x10cf2,1,-64}, 3306 {0x118c0,0x118df,1,-32} 3307 }; 3308 3309 /* 3310 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3311 * simple case folding. 3312 */ 3313 int 3314 utf_toupper(int a) 3315 { 3316 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */ 3317 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3318 return TOUPPER_ASC(a); 3319 3320 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3321 /* If towupper() is available and handles Unicode, use it. */ 3322 if (!(cmp_flags & CMP_INTERNAL)) 3323 return towupper(a); 3324 #endif 3325 3326 /* For characters below 128 use locale sensitive toupper(). */ 3327 if (a < 128) 3328 return TOUPPER_LOC(a); 3329 3330 /* For any other characters use the above mapping table. */ 3331 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3332 } 3333 3334 int 3335 utf_islower(int a) 3336 { 3337 /* German sharp s is lower case but has no upper case equivalent. */ 3338 return (utf_toupper(a) != a) || a == 0xdf; 3339 } 3340 3341 /* 3342 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3343 * simple case folding. 3344 */ 3345 int 3346 utf_tolower(int a) 3347 { 3348 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */ 3349 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3350 return TOLOWER_ASC(a); 3351 3352 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3353 /* If towlower() is available and handles Unicode, use it. */ 3354 if (!(cmp_flags & CMP_INTERNAL)) 3355 return towlower(a); 3356 #endif 3357 3358 /* For characters below 128 use locale sensitive tolower(). */ 3359 if (a < 128) 3360 return TOLOWER_LOC(a); 3361 3362 /* For any other characters use the above mapping table. */ 3363 return utf_convert(a, toLower, (int)sizeof(toLower)); 3364 } 3365 3366 int 3367 utf_isupper(int a) 3368 { 3369 return (utf_tolower(a) != a); 3370 } 3371 3372 static int 3373 utf_strnicmp( 3374 char_u *s1, 3375 char_u *s2, 3376 size_t n1, 3377 size_t n2) 3378 { 3379 int c1, c2, cdiff; 3380 char_u buffer[6]; 3381 3382 for (;;) 3383 { 3384 c1 = utf_safe_read_char_adv(&s1, &n1); 3385 c2 = utf_safe_read_char_adv(&s2, &n2); 3386 3387 if (c1 <= 0 || c2 <= 0) 3388 break; 3389 3390 if (c1 == c2) 3391 continue; 3392 3393 cdiff = utf_fold(c1) - utf_fold(c2); 3394 if (cdiff != 0) 3395 return cdiff; 3396 } 3397 3398 /* some string ended or has an incomplete/illegal character sequence */ 3399 3400 if (c1 == 0 || c2 == 0) 3401 { 3402 /* some string ended. shorter string is smaller */ 3403 if (c1 == 0 && c2 == 0) 3404 return 0; 3405 return c1 == 0 ? -1 : 1; 3406 } 3407 3408 /* Continue with bytewise comparison to produce some result that 3409 * would make comparison operations involving this function transitive. 3410 * 3411 * If only one string had an error, comparison should be made with 3412 * folded version of the other string. In this case it is enough 3413 * to fold just one character to determine the result of comparison. */ 3414 3415 if (c1 != -1 && c2 == -1) 3416 { 3417 n1 = utf_char2bytes(utf_fold(c1), buffer); 3418 s1 = buffer; 3419 } 3420 else if (c2 != -1 && c1 == -1) 3421 { 3422 n2 = utf_char2bytes(utf_fold(c2), buffer); 3423 s2 = buffer; 3424 } 3425 3426 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3427 { 3428 cdiff = (int)(*s1) - (int)(*s2); 3429 if (cdiff != 0) 3430 return cdiff; 3431 3432 s1++; 3433 s2++; 3434 n1--; 3435 n2--; 3436 } 3437 3438 if (n1 > 0 && *s1 == NUL) 3439 n1 = 0; 3440 if (n2 > 0 && *s2 == NUL) 3441 n2 = 0; 3442 3443 if (n1 == 0 && n2 == 0) 3444 return 0; 3445 return n1 == 0 ? -1 : 1; 3446 } 3447 3448 /* 3449 * Version of strnicmp() that handles multi-byte characters. 3450 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3451 * probably use strnicmp(), because there are no ASCII characters in the 3452 * second byte. 3453 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3454 * two characters otherwise. 3455 */ 3456 int 3457 mb_strnicmp(char_u *s1, char_u *s2, size_t nn) 3458 { 3459 int i, l; 3460 int cdiff; 3461 int n = (int)nn; 3462 3463 if (enc_utf8) 3464 { 3465 return utf_strnicmp(s1, s2, nn, nn); 3466 } 3467 else 3468 { 3469 for (i = 0; i < n; i += l) 3470 { 3471 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */ 3472 return 0; 3473 3474 l = (*mb_ptr2len)(s1 + i); 3475 if (l <= 1) 3476 { 3477 /* Single byte: first check normally, then with ignore case. */ 3478 if (s1[i] != s2[i]) 3479 { 3480 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3481 if (cdiff != 0) 3482 return cdiff; 3483 } 3484 } 3485 else 3486 { 3487 /* For non-Unicode multi-byte don't ignore case. */ 3488 if (l > n - i) 3489 l = n - i; 3490 cdiff = STRNCMP(s1 + i, s2 + i, l); 3491 if (cdiff != 0) 3492 return cdiff; 3493 } 3494 } 3495 } 3496 return 0; 3497 } 3498 3499 /* 3500 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3501 * 'encoding' has been set to. 3502 */ 3503 void 3504 show_utf8(void) 3505 { 3506 int len; 3507 int rlen = 0; 3508 char_u *line; 3509 int clen; 3510 int i; 3511 3512 /* Get the byte length of the char under the cursor, including composing 3513 * characters. */ 3514 line = ml_get_cursor(); 3515 len = utfc_ptr2len(line); 3516 if (len == 0) 3517 { 3518 MSG("NUL"); 3519 return; 3520 } 3521 3522 clen = 0; 3523 for (i = 0; i < len; ++i) 3524 { 3525 if (clen == 0) 3526 { 3527 /* start of (composing) character, get its length */ 3528 if (i > 0) 3529 { 3530 STRCPY(IObuff + rlen, "+ "); 3531 rlen += 2; 3532 } 3533 clen = utf_ptr2len(line + i); 3534 } 3535 sprintf((char *)IObuff + rlen, "%02x ", 3536 (line[i] == NL) ? NUL : line[i]); /* NUL is stored as NL */ 3537 --clen; 3538 rlen += (int)STRLEN(IObuff + rlen); 3539 if (rlen > IOSIZE - 20) 3540 break; 3541 } 3542 3543 msg(IObuff); 3544 } 3545 3546 /* 3547 * mb_head_off() function pointer. 3548 * Return offset from "p" to the first byte of the character it points into. 3549 * If "p" points to the NUL at the end of the string return 0. 3550 * Returns 0 when already at the first byte of a character. 3551 */ 3552 int 3553 latin_head_off(char_u *base UNUSED, char_u *p UNUSED) 3554 { 3555 return 0; 3556 } 3557 3558 int 3559 dbcs_head_off(char_u *base, char_u *p) 3560 { 3561 char_u *q; 3562 3563 /* It can't be a trailing byte when not using DBCS, at the start of the 3564 * string or the previous byte can't start a double-byte. */ 3565 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3566 return 0; 3567 3568 /* This is slow: need to start at the base and go forward until the 3569 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */ 3570 q = base; 3571 while (q < p) 3572 q += dbcs_ptr2len(q); 3573 return (q == p) ? 0 : 1; 3574 } 3575 3576 /* 3577 * Special version of dbcs_head_off() that works for ScreenLines[], where 3578 * single-width DBCS_JPNU characters are stored separately. 3579 */ 3580 int 3581 dbcs_screen_head_off(char_u *base, char_u *p) 3582 { 3583 char_u *q; 3584 3585 /* It can't be a trailing byte when not using DBCS, at the start of the 3586 * string or the previous byte can't start a double-byte. 3587 * For euc-jp an 0x8e byte in the previous cell always means we have a 3588 * lead byte in the current cell. */ 3589 if (p <= base 3590 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3591 || MB_BYTE2LEN(p[-1]) == 1 3592 || *p == NUL) 3593 return 0; 3594 3595 /* This is slow: need to start at the base and go forward until the 3596 * byte we are looking for. Return 1 when we went past it, 0 otherwise. 3597 * For DBCS_JPNU look out for 0x8e, which means the second byte is not 3598 * stored as the next byte. */ 3599 q = base; 3600 while (q < p) 3601 { 3602 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3603 ++q; 3604 else 3605 q += dbcs_ptr2len(q); 3606 } 3607 return (q == p) ? 0 : 1; 3608 } 3609 3610 int 3611 utf_head_off(char_u *base, char_u *p) 3612 { 3613 char_u *q; 3614 char_u *s; 3615 int c; 3616 int len; 3617 #ifdef FEAT_ARABIC 3618 char_u *j; 3619 #endif 3620 3621 if (*p < 0x80) /* be quick for ASCII */ 3622 return 0; 3623 3624 /* Skip backwards over trailing bytes: 10xx.xxxx 3625 * Skip backwards again if on a composing char. */ 3626 for (q = p; ; --q) 3627 { 3628 /* Move s to the last byte of this char. */ 3629 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3630 ; 3631 /* Move q to the first byte of this char. */ 3632 while (q > base && (*q & 0xc0) == 0x80) 3633 --q; 3634 /* Check for illegal sequence. Do allow an illegal byte after where we 3635 * started. */ 3636 len = utf8len_tab[*q]; 3637 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3638 return 0; 3639 3640 if (q <= base) 3641 break; 3642 3643 c = utf_ptr2char(q); 3644 if (utf_iscomposing(c)) 3645 continue; 3646 3647 #ifdef FEAT_ARABIC 3648 if (arabic_maycombine(c)) 3649 { 3650 /* Advance to get a sneak-peak at the next char */ 3651 j = q; 3652 --j; 3653 /* Move j to the first byte of this char. */ 3654 while (j > base && (*j & 0xc0) == 0x80) 3655 --j; 3656 if (arabic_combine(utf_ptr2char(j), c)) 3657 continue; 3658 } 3659 #endif 3660 break; 3661 } 3662 3663 return (int)(p - q); 3664 } 3665 3666 /* 3667 * Copy a character from "*fp" to "*tp" and advance the pointers. 3668 */ 3669 void 3670 mb_copy_char(char_u **fp, char_u **tp) 3671 { 3672 int l = (*mb_ptr2len)(*fp); 3673 3674 mch_memmove(*tp, *fp, (size_t)l); 3675 *tp += l; 3676 *fp += l; 3677 } 3678 3679 /* 3680 * Return the offset from "p" to the first byte of a character. When "p" is 3681 * at the start of a character 0 is returned, otherwise the offset to the next 3682 * character. Can start anywhere in a stream of bytes. 3683 */ 3684 int 3685 mb_off_next(char_u *base, char_u *p) 3686 { 3687 int i; 3688 int j; 3689 3690 if (enc_utf8) 3691 { 3692 if (*p < 0x80) /* be quick for ASCII */ 3693 return 0; 3694 3695 /* Find the next character that isn't 10xx.xxxx */ 3696 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 3697 ; 3698 if (i > 0) 3699 { 3700 /* Check for illegal sequence. */ 3701 for (j = 0; p - j > base; ++j) 3702 if ((p[-j] & 0xc0) != 0x80) 3703 break; 3704 if (utf8len_tab[p[-j]] != i + j) 3705 return 0; 3706 } 3707 return i; 3708 } 3709 3710 /* Only need to check if we're on a trail byte, it doesn't matter if we 3711 * want the offset to the next or current character. */ 3712 return (*mb_head_off)(base, p); 3713 } 3714 3715 /* 3716 * Return the offset from "p" to the last byte of the character it points 3717 * into. Can start anywhere in a stream of bytes. 3718 */ 3719 int 3720 mb_tail_off(char_u *base, char_u *p) 3721 { 3722 int i; 3723 int j; 3724 3725 if (*p == NUL) 3726 return 0; 3727 3728 if (enc_utf8) 3729 { 3730 /* Find the last character that is 10xx.xxxx */ 3731 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 3732 ; 3733 /* Check for illegal sequence. */ 3734 for (j = 0; p - j > base; ++j) 3735 if ((p[-j] & 0xc0) != 0x80) 3736 break; 3737 if (utf8len_tab[p[-j]] != i + j + 1) 3738 return 0; 3739 return i; 3740 } 3741 3742 /* It can't be the first byte if a double-byte when not using DBCS, at the 3743 * end of the string or the byte can't start a double-byte. */ 3744 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 3745 return 0; 3746 3747 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3748 return 1 - dbcs_head_off(base, p); 3749 } 3750 3751 /* 3752 * Find the next illegal byte sequence. 3753 */ 3754 void 3755 utf_find_illegal(void) 3756 { 3757 pos_T pos = curwin->w_cursor; 3758 char_u *p; 3759 int len; 3760 vimconv_T vimconv; 3761 char_u *tofree = NULL; 3762 3763 vimconv.vc_type = CONV_NONE; 3764 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 3765 { 3766 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 3767 * possibly a utf-8 file with illegal bytes. Setup for conversion 3768 * from utf-8 to 'fileencoding'. */ 3769 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 3770 } 3771 3772 #ifdef FEAT_VIRTUALEDIT 3773 curwin->w_cursor.coladd = 0; 3774 #endif 3775 for (;;) 3776 { 3777 p = ml_get_cursor(); 3778 if (vimconv.vc_type != CONV_NONE) 3779 { 3780 vim_free(tofree); 3781 tofree = string_convert(&vimconv, p, NULL); 3782 if (tofree == NULL) 3783 break; 3784 p = tofree; 3785 } 3786 3787 while (*p != NUL) 3788 { 3789 /* Illegal means that there are not enough trail bytes (checked by 3790 * utf_ptr2len()) or too many of them (overlong sequence). */ 3791 len = utf_ptr2len(p); 3792 if (*p >= 0x80 && (len == 1 3793 || utf_char2len(utf_ptr2char(p)) != len)) 3794 { 3795 if (vimconv.vc_type == CONV_NONE) 3796 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 3797 else 3798 { 3799 int l; 3800 3801 len = (int)(p - tofree); 3802 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 3803 { 3804 l = utf_ptr2len(p); 3805 curwin->w_cursor.col += l; 3806 } 3807 } 3808 goto theend; 3809 } 3810 p += len; 3811 } 3812 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 3813 break; 3814 ++curwin->w_cursor.lnum; 3815 curwin->w_cursor.col = 0; 3816 } 3817 3818 /* didn't find it: don't move and beep */ 3819 curwin->w_cursor = pos; 3820 beep_flush(); 3821 3822 theend: 3823 vim_free(tofree); 3824 convert_setup(&vimconv, NULL, NULL); 3825 } 3826 3827 #if defined(FEAT_GUI_GTK) || defined(PROTO) 3828 /* 3829 * Return TRUE if string "s" is a valid utf-8 string. 3830 * When "end" is NULL stop at the first NUL. 3831 * When "end" is positive stop there. 3832 */ 3833 int 3834 utf_valid_string(char_u *s, char_u *end) 3835 { 3836 int l; 3837 char_u *p = s; 3838 3839 while (end == NULL ? *p != NUL : p < end) 3840 { 3841 l = utf8len_tab_zero[*p]; 3842 if (l == 0) 3843 return FALSE; /* invalid lead byte */ 3844 if (end != NULL && p + l > end) 3845 return FALSE; /* incomplete byte sequence */ 3846 ++p; 3847 while (--l > 0) 3848 if ((*p++ & 0xc0) != 0x80) 3849 return FALSE; /* invalid trail byte */ 3850 } 3851 return TRUE; 3852 } 3853 #endif 3854 3855 #if defined(FEAT_GUI) || defined(PROTO) 3856 /* 3857 * Special version of mb_tail_off() for use in ScreenLines[]. 3858 */ 3859 int 3860 dbcs_screen_tail_off(char_u *base, char_u *p) 3861 { 3862 /* It can't be the first byte if a double-byte when not using DBCS, at the 3863 * end of the string or the byte can't start a double-byte. 3864 * For euc-jp an 0x8e byte always means we have a lead byte in the current 3865 * cell. */ 3866 if (*p == NUL || p[1] == NUL 3867 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 3868 || MB_BYTE2LEN(*p) == 1) 3869 return 0; 3870 3871 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3872 return 1 - dbcs_screen_head_off(base, p); 3873 } 3874 #endif 3875 3876 /* 3877 * If the cursor moves on an trail byte, set the cursor on the lead byte. 3878 * Thus it moves left if necessary. 3879 * Return TRUE when the cursor was adjusted. 3880 */ 3881 void 3882 mb_adjust_cursor(void) 3883 { 3884 mb_adjustpos(curbuf, &curwin->w_cursor); 3885 } 3886 3887 /* 3888 * Adjust position "*lp" to point to the first byte of a multi-byte character. 3889 * If it points to a tail byte it's moved backwards to the head byte. 3890 */ 3891 void 3892 mb_adjustpos(buf_T *buf, pos_T *lp) 3893 { 3894 char_u *p; 3895 3896 if (lp->col > 0 3897 #ifdef FEAT_VIRTUALEDIT 3898 || lp->coladd > 1 3899 #endif 3900 ) 3901 { 3902 p = ml_get_buf(buf, lp->lnum, FALSE); 3903 lp->col -= (*mb_head_off)(p, p + lp->col); 3904 #ifdef FEAT_VIRTUALEDIT 3905 /* Reset "coladd" when the cursor would be on the right half of a 3906 * double-wide character. */ 3907 if (lp->coladd == 1 3908 && p[lp->col] != TAB 3909 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 3910 && ptr2cells(p + lp->col) > 1) 3911 lp->coladd = 0; 3912 #endif 3913 } 3914 } 3915 3916 /* 3917 * Return a pointer to the character before "*p", if there is one. 3918 */ 3919 char_u * 3920 mb_prevptr( 3921 char_u *line, /* start of the string */ 3922 char_u *p) 3923 { 3924 if (p > line) 3925 mb_ptr_back(line, p); 3926 return p; 3927 } 3928 3929 /* 3930 * Return the character length of "str". Each multi-byte character (with 3931 * following composing characters) counts as one. 3932 */ 3933 int 3934 mb_charlen(char_u *str) 3935 { 3936 char_u *p = str; 3937 int count; 3938 3939 if (p == NULL) 3940 return 0; 3941 3942 for (count = 0; *p != NUL; count++) 3943 p += (*mb_ptr2len)(p); 3944 3945 return count; 3946 } 3947 3948 #if defined(FEAT_SPELL) || defined(PROTO) 3949 /* 3950 * Like mb_charlen() but for a string with specified length. 3951 */ 3952 int 3953 mb_charlen_len(char_u *str, int len) 3954 { 3955 char_u *p = str; 3956 int count; 3957 3958 for (count = 0; *p != NUL && p < str + len; count++) 3959 p += (*mb_ptr2len)(p); 3960 3961 return count; 3962 } 3963 #endif 3964 3965 /* 3966 * Try to un-escape a multi-byte character. 3967 * Used for the "to" and "from" part of a mapping. 3968 * Return the un-escaped string if it is a multi-byte character, and advance 3969 * "pp" to just after the bytes that formed it. 3970 * Return NULL if no multi-byte char was found. 3971 */ 3972 char_u * 3973 mb_unescape(char_u **pp) 3974 { 3975 static char_u buf[6]; 3976 int n; 3977 int m = 0; 3978 char_u *str = *pp; 3979 3980 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 3981 * KS_EXTRA KE_CSI to CSI. 3982 * Maximum length of a utf-8 character is 4 bytes. */ 3983 for (n = 0; str[n] != NUL && m < 4; ++n) 3984 { 3985 if (str[n] == K_SPECIAL 3986 && str[n + 1] == KS_SPECIAL 3987 && str[n + 2] == KE_FILLER) 3988 { 3989 buf[m++] = K_SPECIAL; 3990 n += 2; 3991 } 3992 else if ((str[n] == K_SPECIAL 3993 # ifdef FEAT_GUI 3994 || str[n] == CSI 3995 # endif 3996 ) 3997 && str[n + 1] == KS_EXTRA 3998 && str[n + 2] == (int)KE_CSI) 3999 { 4000 buf[m++] = CSI; 4001 n += 2; 4002 } 4003 else if (str[n] == K_SPECIAL 4004 # ifdef FEAT_GUI 4005 || str[n] == CSI 4006 # endif 4007 ) 4008 break; /* a special key can't be a multibyte char */ 4009 else 4010 buf[m++] = str[n]; 4011 buf[m] = NUL; 4012 4013 /* Return a multi-byte character if it's found. An illegal sequence 4014 * will result in a 1 here. */ 4015 if ((*mb_ptr2len)(buf) > 1) 4016 { 4017 *pp = str + n + 1; 4018 return buf; 4019 } 4020 4021 /* Bail out quickly for ASCII. */ 4022 if (buf[0] < 128) 4023 break; 4024 } 4025 return NULL; 4026 } 4027 4028 /* 4029 * Return TRUE if the character at "row"/"col" on the screen is the left side 4030 * of a double-width character. 4031 * Caller must make sure "row" and "col" are not invalid! 4032 */ 4033 int 4034 mb_lefthalve(int row, int col) 4035 { 4036 #ifdef FEAT_HANGULIN 4037 if (composing_hangul) 4038 return TRUE; 4039 #endif 4040 return (*mb_off2cells)(LineOffset[row] + col, 4041 LineOffset[row] + screen_Columns) > 1; 4042 } 4043 4044 /* 4045 * Correct a position on the screen, if it's the right half of a double-wide 4046 * char move it to the left half. Returns the corrected column. 4047 */ 4048 int 4049 mb_fix_col(int col, int row) 4050 { 4051 col = check_col(col); 4052 row = check_row(row); 4053 if (has_mbyte && ScreenLines != NULL && col > 0 4054 && ((enc_dbcs 4055 && ScreenLines[LineOffset[row] + col] != NUL 4056 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 4057 ScreenLines + LineOffset[row] + col)) 4058 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0))) 4059 return col - 1; 4060 return col; 4061 } 4062 #endif 4063 4064 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 4065 static int enc_alias_search(char_u *name); 4066 4067 /* 4068 * Skip the Vim specific head of a 'encoding' name. 4069 */ 4070 char_u * 4071 enc_skip(char_u *p) 4072 { 4073 if (STRNCMP(p, "2byte-", 6) == 0) 4074 return p + 6; 4075 if (STRNCMP(p, "8bit-", 5) == 0) 4076 return p + 5; 4077 return p; 4078 } 4079 4080 /* 4081 * Find the canonical name for encoding "enc". 4082 * When the name isn't recognized, returns "enc" itself, but with all lower 4083 * case characters and '_' replaced with '-'. 4084 * Returns an allocated string. NULL for out-of-memory. 4085 */ 4086 char_u * 4087 enc_canonize(char_u *enc) 4088 { 4089 char_u *r; 4090 char_u *p, *s; 4091 int i; 4092 4093 # ifdef FEAT_MBYTE 4094 if (STRCMP(enc, "default") == 0) 4095 { 4096 /* Use the default encoding as it's found by set_init_1(). */ 4097 r = get_encoding_default(); 4098 if (r == NULL) 4099 r = (char_u *)"latin1"; 4100 return vim_strsave(r); 4101 } 4102 # endif 4103 4104 /* copy "enc" to allocated memory, with room for two '-' */ 4105 r = alloc((unsigned)(STRLEN(enc) + 3)); 4106 if (r != NULL) 4107 { 4108 /* Make it all lower case and replace '_' with '-'. */ 4109 p = r; 4110 for (s = enc; *s != NUL; ++s) 4111 { 4112 if (*s == '_') 4113 *p++ = '-'; 4114 else 4115 *p++ = TOLOWER_ASC(*s); 4116 } 4117 *p = NUL; 4118 4119 /* Skip "2byte-" and "8bit-". */ 4120 p = enc_skip(r); 4121 4122 /* Change "microsoft-cp" to "cp". Used in some spell files. */ 4123 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4124 STRMOVE(p, p + 10); 4125 4126 /* "iso8859" -> "iso-8859" */ 4127 if (STRNCMP(p, "iso8859", 7) == 0) 4128 { 4129 STRMOVE(p + 4, p + 3); 4130 p[3] = '-'; 4131 } 4132 4133 /* "iso-8859n" -> "iso-8859-n" */ 4134 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4135 { 4136 STRMOVE(p + 9, p + 8); 4137 p[8] = '-'; 4138 } 4139 4140 /* "latin-N" -> "latinN" */ 4141 if (STRNCMP(p, "latin-", 6) == 0) 4142 STRMOVE(p + 5, p + 6); 4143 4144 if (enc_canon_search(p) >= 0) 4145 { 4146 /* canonical name can be used unmodified */ 4147 if (p != r) 4148 STRMOVE(r, p); 4149 } 4150 else if ((i = enc_alias_search(p)) >= 0) 4151 { 4152 /* alias recognized, get canonical name */ 4153 vim_free(r); 4154 r = vim_strsave((char_u *)enc_canon_table[i].name); 4155 } 4156 } 4157 return r; 4158 } 4159 4160 /* 4161 * Search for an encoding alias of "name". 4162 * Returns -1 when not found. 4163 */ 4164 static int 4165 enc_alias_search(char_u *name) 4166 { 4167 int i; 4168 4169 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4170 if (STRCMP(name, enc_alias_table[i].name) == 0) 4171 return enc_alias_table[i].canon; 4172 return -1; 4173 } 4174 #endif 4175 4176 #if defined(FEAT_MBYTE) || defined(PROTO) 4177 4178 #ifdef HAVE_LANGINFO_H 4179 # include <langinfo.h> 4180 #endif 4181 4182 /* 4183 * Get the canonicalized encoding of the current locale. 4184 * Returns an allocated string when successful, NULL when not. 4185 */ 4186 char_u * 4187 enc_locale(void) 4188 { 4189 #ifndef WIN3264 4190 char *s; 4191 char *p; 4192 int i; 4193 #endif 4194 char buf[50]; 4195 #ifdef WIN3264 4196 long acp = GetACP(); 4197 4198 if (acp == 1200) 4199 STRCPY(buf, "ucs-2le"); 4200 else if (acp == 1252) /* cp1252 is used as latin1 */ 4201 STRCPY(buf, "latin1"); 4202 else 4203 sprintf(buf, "cp%ld", acp); 4204 #else 4205 # ifdef HAVE_NL_LANGINFO_CODESET 4206 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4207 # endif 4208 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4209 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4210 # endif 4211 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4212 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4213 s = getenv("LANG"); 4214 4215 if (s == NULL || *s == NUL) 4216 return FAIL; 4217 4218 /* The most generic locale format is: 4219 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4220 * If there is a '.' remove the part before it. 4221 * if there is something after the codeset, remove it. 4222 * Make the name lowercase and replace '_' with '-'. 4223 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4224 * "ko_KR.EUC" == "euc-kr" 4225 */ 4226 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4227 { 4228 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4229 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4230 { 4231 /* copy "XY.EUC" to "euc-XY" to buf[10] */ 4232 STRCPY(buf + 10, "euc-"); 4233 buf[14] = p[-2]; 4234 buf[15] = p[-1]; 4235 buf[16] = 0; 4236 s = buf + 10; 4237 } 4238 else 4239 s = p + 1; 4240 } 4241 for (i = 0; s[i] != NUL && i < (int)sizeof(buf) - 1; ++i) 4242 { 4243 if (s[i] == '_' || s[i] == '-') 4244 buf[i] = '-'; 4245 else if (isalnum((int)s[i])) 4246 buf[i] = TOLOWER_ASC(s[i]); 4247 else 4248 break; 4249 } 4250 buf[i] = NUL; 4251 #endif 4252 4253 return enc_canonize((char_u *)buf); 4254 } 4255 4256 #if defined(WIN3264) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4257 /* 4258 * Convert an encoding name to an MS-Windows codepage. 4259 * Returns zero if no codepage can be figured out. 4260 */ 4261 int 4262 encname2codepage(char_u *name) 4263 { 4264 int cp; 4265 char_u *p = name; 4266 int idx; 4267 4268 if (STRNCMP(p, "8bit-", 5) == 0) 4269 p += 5; 4270 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4271 p += 6; 4272 4273 if (p[0] == 'c' && p[1] == 'p') 4274 cp = atoi((char *)p + 2); 4275 else if ((idx = enc_canon_search(p)) >= 0) 4276 cp = enc_canon_table[idx].codepage; 4277 else 4278 return 0; 4279 if (IsValidCodePage(cp)) 4280 return cp; 4281 return 0; 4282 } 4283 #endif 4284 4285 # if defined(USE_ICONV) || defined(PROTO) 4286 4287 static char_u *iconv_string(vimconv_T *vcp, char_u *str, int slen, int *unconvlenp, int *resultlenp); 4288 4289 /* 4290 * Call iconv_open() with a check if iconv() works properly (there are broken 4291 * versions). 4292 * Returns (void *)-1 if failed. 4293 * (should return iconv_t, but that causes problems with prototypes). 4294 */ 4295 void * 4296 my_iconv_open(char_u *to, char_u *from) 4297 { 4298 iconv_t fd; 4299 #define ICONV_TESTLEN 400 4300 char_u tobuf[ICONV_TESTLEN]; 4301 char *p; 4302 size_t tolen; 4303 static int iconv_ok = -1; 4304 4305 if (iconv_ok == FALSE) 4306 return (void *)-1; /* detected a broken iconv() previously */ 4307 4308 #ifdef DYNAMIC_ICONV 4309 /* Check if the iconv.dll can be found. */ 4310 if (!iconv_enabled(TRUE)) 4311 return (void *)-1; 4312 #endif 4313 4314 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4315 4316 if (fd != (iconv_t)-1 && iconv_ok == -1) 4317 { 4318 /* 4319 * Do a dummy iconv() call to check if it actually works. There is a 4320 * version of iconv() on Linux that is broken. We can't ignore it, 4321 * because it's wide-spread. The symptoms are that after outputting 4322 * the initial shift state the "to" pointer is NULL and conversion 4323 * stops for no apparent reason after about 8160 characters. 4324 */ 4325 p = (char *)tobuf; 4326 tolen = ICONV_TESTLEN; 4327 (void)iconv(fd, NULL, NULL, &p, &tolen); 4328 if (p == NULL) 4329 { 4330 iconv_ok = FALSE; 4331 iconv_close(fd); 4332 fd = (iconv_t)-1; 4333 } 4334 else 4335 iconv_ok = TRUE; 4336 } 4337 4338 return (void *)fd; 4339 } 4340 4341 /* 4342 * Convert the string "str[slen]" with iconv(). 4343 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4344 * sequence and set "*unconvlenp" to the length of it. 4345 * Returns the converted string in allocated memory. NULL for an error. 4346 * If resultlenp is not NULL, sets it to the result length in bytes. 4347 */ 4348 static char_u * 4349 iconv_string( 4350 vimconv_T *vcp, 4351 char_u *str, 4352 int slen, 4353 int *unconvlenp, 4354 int *resultlenp) 4355 { 4356 const char *from; 4357 size_t fromlen; 4358 char *to; 4359 size_t tolen; 4360 size_t len = 0; 4361 size_t done = 0; 4362 char_u *result = NULL; 4363 char_u *p; 4364 int l; 4365 4366 from = (char *)str; 4367 fromlen = slen; 4368 for (;;) 4369 { 4370 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4371 { 4372 /* Allocate enough room for most conversions. When re-allocating 4373 * increase the buffer size. */ 4374 len = len + fromlen * 2 + 40; 4375 p = alloc((unsigned)len); 4376 if (p != NULL && done > 0) 4377 mch_memmove(p, result, done); 4378 vim_free(result); 4379 result = p; 4380 if (result == NULL) /* out of memory */ 4381 break; 4382 } 4383 4384 to = (char *)result + done; 4385 tolen = len - done - 2; 4386 /* Avoid a warning for systems with a wrong iconv() prototype by 4387 * casting the second argument to void *. */ 4388 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4389 != (size_t)-1) 4390 { 4391 /* Finished, append a NUL. */ 4392 *to = NUL; 4393 break; 4394 } 4395 4396 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4397 * iconv library may use one of them. */ 4398 if (!vcp->vc_fail && unconvlenp != NULL 4399 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4400 { 4401 /* Handle an incomplete sequence at the end. */ 4402 *to = NUL; 4403 *unconvlenp = (int)fromlen; 4404 break; 4405 } 4406 4407 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4408 * iconv library may use one of them. */ 4409 else if (!vcp->vc_fail 4410 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4411 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4412 { 4413 /* Can't convert: insert a '?' and skip a character. This assumes 4414 * conversion from 'encoding' to something else. In other 4415 * situations we don't know what to skip anyway. */ 4416 *to++ = '?'; 4417 if ((*mb_ptr2cells)((char_u *)from) > 1) 4418 *to++ = '?'; 4419 if (enc_utf8) 4420 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4421 else 4422 { 4423 l = (*mb_ptr2len)((char_u *)from); 4424 if (l > (int)fromlen) 4425 l = (int)fromlen; 4426 } 4427 from += l; 4428 fromlen -= l; 4429 } 4430 else if (ICONV_ERRNO != ICONV_E2BIG) 4431 { 4432 /* conversion failed */ 4433 vim_free(result); 4434 result = NULL; 4435 break; 4436 } 4437 /* Not enough room or skipping illegal sequence. */ 4438 done = to - (char *)result; 4439 } 4440 4441 if (resultlenp != NULL && result != NULL) 4442 *resultlenp = (int)(to - (char *)result); 4443 return result; 4444 } 4445 4446 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4447 /* 4448 * Dynamically load the "iconv.dll" on Win32. 4449 */ 4450 4451 # ifndef DYNAMIC_ICONV /* must be generating prototypes */ 4452 # define HINSTANCE int 4453 # endif 4454 static HINSTANCE hIconvDLL = 0; 4455 static HINSTANCE hMsvcrtDLL = 0; 4456 4457 # ifndef DYNAMIC_ICONV_DLL 4458 # define DYNAMIC_ICONV_DLL "iconv.dll" 4459 # define DYNAMIC_ICONV_DLL_ALT1 "libiconv.dll" 4460 # define DYNAMIC_ICONV_DLL_ALT2 "libiconv2.dll" 4461 # define DYNAMIC_ICONV_DLL_ALT3 "libiconv-2.dll" 4462 # endif 4463 # ifndef DYNAMIC_MSVCRT_DLL 4464 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4465 # endif 4466 4467 /* 4468 * Get the address of 'funcname' which is imported by 'hInst' DLL. 4469 */ 4470 static void * 4471 get_iconv_import_func(HINSTANCE hInst, const char *funcname) 4472 { 4473 PBYTE pImage = (PBYTE)hInst; 4474 PIMAGE_DOS_HEADER pDOS = (PIMAGE_DOS_HEADER)hInst; 4475 PIMAGE_NT_HEADERS pPE; 4476 PIMAGE_IMPORT_DESCRIPTOR pImpDesc; 4477 PIMAGE_THUNK_DATA pIAT; /* Import Address Table */ 4478 PIMAGE_THUNK_DATA pINT; /* Import Name Table */ 4479 PIMAGE_IMPORT_BY_NAME pImpName; 4480 4481 if (pDOS->e_magic != IMAGE_DOS_SIGNATURE) 4482 return NULL; 4483 pPE = (PIMAGE_NT_HEADERS)(pImage + pDOS->e_lfanew); 4484 if (pPE->Signature != IMAGE_NT_SIGNATURE) 4485 return NULL; 4486 pImpDesc = (PIMAGE_IMPORT_DESCRIPTOR)(pImage 4487 + pPE->OptionalHeader.DataDirectory[IMAGE_DIRECTORY_ENTRY_IMPORT] 4488 .VirtualAddress); 4489 for (; pImpDesc->FirstThunk; ++pImpDesc) 4490 { 4491 if (!pImpDesc->OriginalFirstThunk) 4492 continue; 4493 pIAT = (PIMAGE_THUNK_DATA)(pImage + pImpDesc->FirstThunk); 4494 pINT = (PIMAGE_THUNK_DATA)(pImage + pImpDesc->OriginalFirstThunk); 4495 for (; pIAT->u1.Function; ++pIAT, ++pINT) 4496 { 4497 if (IMAGE_SNAP_BY_ORDINAL(pINT->u1.Ordinal)) 4498 continue; 4499 pImpName = (PIMAGE_IMPORT_BY_NAME)(pImage 4500 + (UINT_PTR)(pINT->u1.AddressOfData)); 4501 if (strcmp((char *)pImpName->Name, funcname) == 0) 4502 return (void *)pIAT->u1.Function; 4503 } 4504 } 4505 return NULL; 4506 } 4507 4508 /* 4509 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4510 */ 4511 int 4512 iconv_enabled(int verbose) 4513 { 4514 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4515 return TRUE; 4516 4517 /* The iconv DLL file goes under different names, try them all. 4518 * Do the "2" version first, it's newer. */ 4519 #ifdef DYNAMIC_ICONV_DLL_ALT2 4520 if (hIconvDLL == 0) 4521 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT2); 4522 #endif 4523 #ifdef DYNAMIC_ICONV_DLL_ALT3 4524 if (hIconvDLL == 0) 4525 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT3); 4526 #endif 4527 if (hIconvDLL == 0) 4528 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4529 #ifdef DYNAMIC_ICONV_DLL_ALT1 4530 if (hIconvDLL == 0) 4531 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT1); 4532 #endif 4533 4534 if (hIconvDLL != 0) 4535 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4536 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4537 { 4538 /* Only give the message when 'verbose' is set, otherwise it might be 4539 * done whenever a conversion is attempted. */ 4540 if (verbose && p_verbose > 0) 4541 { 4542 verbose_enter(); 4543 EMSG2(_(e_loadlib), 4544 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4545 verbose_leave(); 4546 } 4547 iconv_end(); 4548 return FALSE; 4549 } 4550 4551 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4552 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4553 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4554 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4555 iconv_errno = get_iconv_import_func(hIconvDLL, "_errno"); 4556 if (iconv_errno == NULL) 4557 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4558 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4559 || iconvctl == NULL || iconv_errno == NULL) 4560 { 4561 iconv_end(); 4562 if (verbose && p_verbose > 0) 4563 { 4564 verbose_enter(); 4565 EMSG2(_(e_loadfunc), "for libiconv"); 4566 verbose_leave(); 4567 } 4568 return FALSE; 4569 } 4570 return TRUE; 4571 } 4572 4573 void 4574 iconv_end(void) 4575 { 4576 /* Don't use iconv() when inputting or outputting characters. */ 4577 if (input_conv.vc_type == CONV_ICONV) 4578 convert_setup(&input_conv, NULL, NULL); 4579 if (output_conv.vc_type == CONV_ICONV) 4580 convert_setup(&output_conv, NULL, NULL); 4581 4582 if (hIconvDLL != 0) 4583 FreeLibrary(hIconvDLL); 4584 if (hMsvcrtDLL != 0) 4585 FreeLibrary(hMsvcrtDLL); 4586 hIconvDLL = 0; 4587 hMsvcrtDLL = 0; 4588 } 4589 # endif /* DYNAMIC_ICONV */ 4590 # endif /* USE_ICONV */ 4591 4592 #endif /* FEAT_MBYTE */ 4593 4594 #if defined(FEAT_XIM) || defined(PROTO) 4595 4596 # if defined(FEAT_GUI_GTK) || defined(PROTO) 4597 static int xim_has_preediting INIT(= FALSE); /* IM current status */ 4598 4599 /* 4600 * Set preedit_start_col to the current cursor position. 4601 */ 4602 static void 4603 init_preedit_start_col(void) 4604 { 4605 if (State & CMDLINE) 4606 preedit_start_col = cmdline_getvcol_cursor(); 4607 else if (curwin != NULL && curwin->w_buffer != NULL) 4608 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); 4609 /* Prevent that preediting marks the buffer as changed. */ 4610 xim_changed_while_preediting = curbuf->b_changed; 4611 } 4612 4613 static int im_is_active = FALSE; /* IM is enabled for current mode */ 4614 static int preedit_is_active = FALSE; 4615 static int im_preedit_cursor = 0; /* cursor offset in characters */ 4616 static int im_preedit_trailing = 0; /* number of characters after cursor */ 4617 4618 static unsigned long im_commit_handler_id = 0; 4619 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; 4620 static unsigned int im_activatekey_state = 0; 4621 4622 void 4623 im_set_active(int active) 4624 { 4625 int was_active; 4626 4627 was_active = !!im_get_status(); 4628 im_is_active = (active && !p_imdisable); 4629 4630 if (im_is_active != was_active) 4631 xim_reset(); 4632 } 4633 4634 void 4635 xim_set_focus(int focus) 4636 { 4637 if (xic != NULL) 4638 { 4639 if (focus) 4640 gtk_im_context_focus_in(xic); 4641 else 4642 gtk_im_context_focus_out(xic); 4643 } 4644 } 4645 4646 void 4647 im_set_position(int row, int col) 4648 { 4649 if (xic != NULL) 4650 { 4651 GdkRectangle area; 4652 4653 area.x = FILL_X(col); 4654 area.y = FILL_Y(row); 4655 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); 4656 area.height = gui.char_height; 4657 4658 gtk_im_context_set_cursor_location(xic, &area); 4659 } 4660 } 4661 4662 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */ 4663 void 4664 xim_set_preedit(void) 4665 { 4666 im_set_position(gui.row, gui.col); 4667 } 4668 # endif 4669 4670 static void 4671 im_add_to_input(char_u *str, int len) 4672 { 4673 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */ 4674 if (input_conv.vc_type != CONV_NONE) 4675 { 4676 str = string_convert(&input_conv, str, &len); 4677 g_return_if_fail(str != NULL); 4678 } 4679 4680 add_to_input_buf_csi(str, len); 4681 4682 if (input_conv.vc_type != CONV_NONE) 4683 vim_free(str); 4684 4685 if (p_mh) /* blank out the pointer if necessary */ 4686 gui_mch_mousehide(TRUE); 4687 } 4688 4689 static void 4690 im_delete_preedit(void) 4691 { 4692 char_u bskey[] = {CSI, 'k', 'b'}; 4693 char_u delkey[] = {CSI, 'k', 'D'}; 4694 4695 if (State & NORMAL) 4696 { 4697 im_preedit_cursor = 0; 4698 return; 4699 } 4700 for (; im_preedit_cursor > 0; --im_preedit_cursor) 4701 add_to_input_buf(bskey, (int)sizeof(bskey)); 4702 4703 for (; im_preedit_trailing > 0; --im_preedit_trailing) 4704 add_to_input_buf(delkey, (int)sizeof(delkey)); 4705 } 4706 4707 /* 4708 * Move the cursor left by "num_move_back" characters. 4709 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for 4710 * these K_LEFT keys. 4711 */ 4712 static void 4713 im_correct_cursor(int num_move_back) 4714 { 4715 char_u backkey[] = {CSI, 'k', 'l'}; 4716 4717 if (State & NORMAL) 4718 return; 4719 # ifdef FEAT_RIGHTLEFT 4720 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) 4721 backkey[2] = 'r'; 4722 # endif 4723 for (; num_move_back > 0; --num_move_back) 4724 add_to_input_buf(backkey, (int)sizeof(backkey)); 4725 } 4726 4727 static int xim_expected_char = NUL; 4728 static int xim_ignored_char = FALSE; 4729 4730 /* 4731 * Update the mode and cursor while in an IM callback. 4732 */ 4733 static void 4734 im_show_info(void) 4735 { 4736 int old_vgetc_busy; 4737 4738 old_vgetc_busy = vgetc_busy; 4739 vgetc_busy = TRUE; 4740 showmode(); 4741 vgetc_busy = old_vgetc_busy; 4742 if ((State & NORMAL) || (State & INSERT)) 4743 setcursor(); 4744 out_flush(); 4745 } 4746 4747 /* 4748 * Callback invoked when the user finished preediting. 4749 * Put the final string into the input buffer. 4750 */ 4751 static void 4752 im_commit_cb(GtkIMContext *context UNUSED, 4753 const gchar *str, 4754 gpointer data UNUSED) 4755 { 4756 int slen = (int)STRLEN(str); 4757 int add_to_input = TRUE; 4758 int clen; 4759 int len = slen; 4760 int commit_with_preedit = TRUE; 4761 char_u *im_str; 4762 4763 #ifdef XIM_DEBUG 4764 xim_log("im_commit_cb(): %s\n", str); 4765 #endif 4766 4767 /* The imhangul module doesn't reset the preedit string before 4768 * committing. Call im_delete_preedit() to work around that. */ 4769 im_delete_preedit(); 4770 4771 /* Indicate that preediting has finished. */ 4772 if (preedit_start_col == MAXCOL) 4773 { 4774 init_preedit_start_col(); 4775 commit_with_preedit = FALSE; 4776 } 4777 4778 /* The thing which setting "preedit_start_col" to MAXCOL means that 4779 * "preedit_start_col" will be set forcedly when calling 4780 * preedit_changed_cb() next time. 4781 * "preedit_start_col" should not reset with MAXCOL on this part. Vim 4782 * is simulating the preediting by using add_to_input_str(). when 4783 * preedit begin immediately before committed, the typebuf is not 4784 * flushed to screen, then it can't get correct "preedit_start_col". 4785 * Thus, it should calculate the cells by adding cells of the committed 4786 * string. */ 4787 if (input_conv.vc_type != CONV_NONE) 4788 { 4789 im_str = string_convert(&input_conv, (char_u *)str, &len); 4790 g_return_if_fail(im_str != NULL); 4791 } 4792 else 4793 im_str = (char_u *)str; 4794 4795 clen = mb_string2cells(im_str, len); 4796 4797 if (input_conv.vc_type != CONV_NONE) 4798 vim_free(im_str); 4799 preedit_start_col += clen; 4800 4801 /* Is this a single character that matches a keypad key that's just 4802 * been pressed? If so, we don't want it to be entered as such - let 4803 * us carry on processing the raw keycode so that it may be used in 4804 * mappings as <kSomething>. */ 4805 if (xim_expected_char != NUL) 4806 { 4807 /* We're currently processing a keypad or other special key */ 4808 if (slen == 1 && str[0] == xim_expected_char) 4809 { 4810 /* It's a match - don't do it here */ 4811 xim_ignored_char = TRUE; 4812 add_to_input = FALSE; 4813 } 4814 else 4815 { 4816 /* Not a match */ 4817 xim_ignored_char = FALSE; 4818 } 4819 } 4820 4821 if (add_to_input) 4822 im_add_to_input((char_u *)str, slen); 4823 4824 /* Inserting chars while "im_is_active" is set does not cause a change of 4825 * buffer. When the chars are committed the buffer must be marked as 4826 * changed. */ 4827 if (!commit_with_preedit) 4828 preedit_start_col = MAXCOL; 4829 4830 /* This flag is used in changed() at next call. */ 4831 xim_changed_while_preediting = TRUE; 4832 4833 if (gtk_main_level() > 0) 4834 gtk_main_quit(); 4835 } 4836 4837 /* 4838 * Callback invoked after start to the preedit. 4839 */ 4840 static void 4841 im_preedit_start_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4842 { 4843 #ifdef XIM_DEBUG 4844 xim_log("im_preedit_start_cb()\n"); 4845 #endif 4846 4847 im_is_active = TRUE; 4848 preedit_is_active = TRUE; 4849 gui_update_cursor(TRUE, FALSE); 4850 im_show_info(); 4851 } 4852 4853 /* 4854 * Callback invoked after end to the preedit. 4855 */ 4856 static void 4857 im_preedit_end_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4858 { 4859 #ifdef XIM_DEBUG 4860 xim_log("im_preedit_end_cb()\n"); 4861 #endif 4862 im_delete_preedit(); 4863 4864 /* Indicate that preediting has finished */ 4865 preedit_start_col = MAXCOL; 4866 xim_has_preediting = FALSE; 4867 4868 #if 0 4869 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was 4870 * switched off unintentionally. We now use preedit_is_active (added by 4871 * SungHyun Nam). */ 4872 im_is_active = FALSE; 4873 #endif 4874 preedit_is_active = FALSE; 4875 gui_update_cursor(TRUE, FALSE); 4876 im_show_info(); 4877 } 4878 4879 /* 4880 * Callback invoked after changes to the preedit string. If the preedit 4881 * string was empty before, remember the preedit start column so we know 4882 * where to apply feedback attributes. Delete the previous preedit string 4883 * if there was one, save the new preedit cursor offset, and put the new 4884 * string into the input buffer. 4885 * 4886 * TODO: The pragmatic "put into input buffer" approach used here has 4887 * several fundamental problems: 4888 * 4889 * - The characters in the preedit string are subject to remapping. 4890 * That's broken, only the finally committed string should be remapped. 4891 * 4892 * - There is a race condition involved: The retrieved value for the 4893 * current cursor position will be wrong if any unprocessed characters 4894 * are still queued in the input buffer. 4895 * 4896 * - Due to the lack of synchronization between the file buffer in memory 4897 * and any typed characters, it's practically impossible to implement the 4898 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM 4899 * modules for languages such as Thai are likely to rely on this feature 4900 * for proper operation. 4901 * 4902 * Conclusions: I think support for preediting needs to be moved to the 4903 * core parts of Vim. Ideally, until it has been committed, the preediting 4904 * string should only be displayed and not affect the buffer content at all. 4905 * The question how to deal with the synchronization issue still remains. 4906 * Circumventing the input buffer is probably not desirable. Anyway, I think 4907 * implementing "retrieve_surrounding" is the only hard problem. 4908 * 4909 * One way to solve all of this in a clean manner would be to queue all key 4910 * press/release events "as is" in the input buffer, and apply the IM filtering 4911 * at the receiving end of the queue. This, however, would have a rather large 4912 * impact on the code base. If there is an easy way to force processing of all 4913 * remaining input from within the "retrieve_surrounding" signal handler, this 4914 * might not be necessary. Gotta ask on vim-dev for opinions. 4915 */ 4916 static void 4917 im_preedit_changed_cb(GtkIMContext *context, gpointer data UNUSED) 4918 { 4919 char *preedit_string = NULL; 4920 int cursor_index = 0; 4921 int num_move_back = 0; 4922 char_u *str; 4923 char_u *p; 4924 int i; 4925 4926 gtk_im_context_get_preedit_string(context, 4927 &preedit_string, NULL, 4928 &cursor_index); 4929 4930 #ifdef XIM_DEBUG 4931 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); 4932 #endif 4933 4934 g_return_if_fail(preedit_string != NULL); /* just in case */ 4935 4936 /* If preedit_start_col is MAXCOL set it to the current cursor position. */ 4937 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') 4938 { 4939 xim_has_preediting = TRUE; 4940 4941 /* Urgh, this breaks if the input buffer isn't empty now */ 4942 init_preedit_start_col(); 4943 } 4944 else if (cursor_index == 0 && preedit_string[0] == '\0') 4945 { 4946 xim_has_preediting = FALSE; 4947 4948 /* If at the start position (after typing backspace) 4949 * preedit_start_col must be reset. */ 4950 preedit_start_col = MAXCOL; 4951 } 4952 4953 im_delete_preedit(); 4954 4955 /* 4956 * Compute the end of the preediting area: "preedit_end_col". 4957 * According to the documentation of gtk_im_context_get_preedit_string(), 4958 * the cursor_pos output argument returns the offset in bytes. This is 4959 * unfortunately not true -- real life shows the offset is in characters, 4960 * and the GTK+ source code agrees with me. Will file a bug later. 4961 */ 4962 if (preedit_start_col != MAXCOL) 4963 preedit_end_col = preedit_start_col; 4964 str = (char_u *)preedit_string; 4965 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) 4966 { 4967 int is_composing; 4968 4969 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); 4970 /* 4971 * These offsets are used as counters when generating <BS> and <Del> 4972 * to delete the preedit string. So don't count composing characters 4973 * unless 'delcombine' is enabled. 4974 */ 4975 if (!is_composing || p_deco) 4976 { 4977 if (i < cursor_index) 4978 ++im_preedit_cursor; 4979 else 4980 ++im_preedit_trailing; 4981 } 4982 if (!is_composing && i >= cursor_index) 4983 { 4984 /* This is essentially the same as im_preedit_trailing, except 4985 * composing characters are not counted even if p_deco is set. */ 4986 ++num_move_back; 4987 } 4988 if (preedit_start_col != MAXCOL) 4989 preedit_end_col += utf_ptr2cells(p); 4990 } 4991 4992 if (p > str) 4993 { 4994 im_add_to_input(str, (int)(p - str)); 4995 im_correct_cursor(num_move_back); 4996 } 4997 4998 g_free(preedit_string); 4999 5000 if (gtk_main_level() > 0) 5001 gtk_main_quit(); 5002 } 5003 5004 /* 5005 * Translate the Pango attributes at iter to Vim highlighting attributes. 5006 * Ignore attributes not supported by Vim highlighting. This shouldn't have 5007 * too much impact -- right now we handle even more attributes than necessary 5008 * for the IM modules I tested with. 5009 */ 5010 static int 5011 translate_pango_attributes(PangoAttrIterator *iter) 5012 { 5013 PangoAttribute *attr; 5014 int char_attr = HL_NORMAL; 5015 5016 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); 5017 if (attr != NULL && ((PangoAttrInt *)attr)->value 5018 != (int)PANGO_UNDERLINE_NONE) 5019 char_attr |= HL_UNDERLINE; 5020 5021 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); 5022 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) 5023 char_attr |= HL_BOLD; 5024 5025 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); 5026 if (attr != NULL && ((PangoAttrInt *)attr)->value 5027 != (int)PANGO_STYLE_NORMAL) 5028 char_attr |= HL_ITALIC; 5029 5030 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); 5031 if (attr != NULL) 5032 { 5033 const PangoColor *color = &((PangoAttrColor *)attr)->color; 5034 5035 /* Assume inverse if black background is requested */ 5036 if ((color->red | color->green | color->blue) == 0) 5037 char_attr |= HL_INVERSE; 5038 } 5039 5040 return char_attr; 5041 } 5042 5043 /* 5044 * Retrieve the highlighting attributes at column col in the preedit string. 5045 * Return -1 if not in preediting mode or if col is out of range. 5046 */ 5047 int 5048 im_get_feedback_attr(int col) 5049 { 5050 char *preedit_string = NULL; 5051 PangoAttrList *attr_list = NULL; 5052 int char_attr = -1; 5053 5054 if (xic == NULL) 5055 return char_attr; 5056 5057 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5058 5059 if (preedit_string != NULL && attr_list != NULL) 5060 { 5061 int idx; 5062 5063 /* Get the byte index as used by PangoAttrIterator */ 5064 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col) 5065 idx += utfc_ptr2len((char_u *)preedit_string + idx); 5066 5067 if (preedit_string[idx] != '\0') 5068 { 5069 PangoAttrIterator *iter; 5070 int start, end; 5071 5072 char_attr = HL_NORMAL; 5073 iter = pango_attr_list_get_iterator(attr_list); 5074 5075 /* Extract all relevant attributes from the list. */ 5076 do 5077 { 5078 pango_attr_iterator_range(iter, &start, &end); 5079 5080 if (idx >= start && idx < end) 5081 char_attr |= translate_pango_attributes(iter); 5082 } 5083 while (pango_attr_iterator_next(iter)); 5084 5085 pango_attr_iterator_destroy(iter); 5086 } 5087 } 5088 5089 if (attr_list != NULL) 5090 pango_attr_list_unref(attr_list); 5091 g_free(preedit_string); 5092 5093 return char_attr; 5094 } 5095 5096 void 5097 xim_init(void) 5098 { 5099 #ifdef XIM_DEBUG 5100 xim_log("xim_init()\n"); 5101 #endif 5102 5103 g_return_if_fail(gui.drawarea != NULL); 5104 #if GTK_CHECK_VERSION(3,0,0) 5105 g_return_if_fail(gtk_widget_get_window(gui.drawarea) != NULL); 5106 #else 5107 g_return_if_fail(gui.drawarea->window != NULL); 5108 #endif 5109 5110 xic = gtk_im_multicontext_new(); 5111 g_object_ref(xic); 5112 5113 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", 5114 G_CALLBACK(&im_commit_cb), NULL); 5115 g_signal_connect(G_OBJECT(xic), "preedit_changed", 5116 G_CALLBACK(&im_preedit_changed_cb), NULL); 5117 g_signal_connect(G_OBJECT(xic), "preedit_start", 5118 G_CALLBACK(&im_preedit_start_cb), NULL); 5119 g_signal_connect(G_OBJECT(xic), "preedit_end", 5120 G_CALLBACK(&im_preedit_end_cb), NULL); 5121 5122 #if GTK_CHECK_VERSION(3,0,0) 5123 gtk_im_context_set_client_window(xic, gtk_widget_get_window(gui.drawarea)); 5124 #else 5125 gtk_im_context_set_client_window(xic, gui.drawarea->window); 5126 #endif 5127 } 5128 5129 void 5130 im_shutdown(void) 5131 { 5132 #ifdef XIM_DEBUG 5133 xim_log("im_shutdown()\n"); 5134 #endif 5135 5136 if (xic != NULL) 5137 { 5138 gtk_im_context_focus_out(xic); 5139 g_object_unref(xic); 5140 xic = NULL; 5141 } 5142 im_is_active = FALSE; 5143 im_commit_handler_id = 0; 5144 preedit_start_col = MAXCOL; 5145 xim_has_preediting = FALSE; 5146 } 5147 5148 /* 5149 * Convert the string argument to keyval and state for GdkEventKey. 5150 * If str is valid return TRUE, otherwise FALSE. 5151 * 5152 * See 'imactivatekey' for documentation of the format. 5153 */ 5154 static int 5155 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) 5156 { 5157 const char *mods_end; 5158 unsigned tmp_keyval; 5159 unsigned tmp_state = 0; 5160 5161 mods_end = strrchr(str, '-'); 5162 mods_end = (mods_end != NULL) ? mods_end + 1 : str; 5163 5164 /* Parse modifier keys */ 5165 while (str < mods_end) 5166 switch (*str++) 5167 { 5168 case '-': break; 5169 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; 5170 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; 5171 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; 5172 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; 5173 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; 5174 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; 5175 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; 5176 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; 5177 default: 5178 return FALSE; 5179 } 5180 5181 tmp_keyval = gdk_keyval_from_name(str); 5182 5183 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) 5184 return FALSE; 5185 5186 if (keyval != NULL) 5187 *keyval = tmp_keyval; 5188 if (state != NULL) 5189 *state = tmp_state; 5190 5191 return TRUE; 5192 } 5193 5194 /* 5195 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an 5196 * empty string is also regarded as valid. 5197 * 5198 * Note: The numerical key value of p_imak is cached if it was valid; thus 5199 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever 5200 * 'imak' changes. This is currently the case but not obvious -- should 5201 * probably rename the function for clarity. 5202 */ 5203 int 5204 im_xim_isvalid_imactivate(void) 5205 { 5206 if (p_imak[0] == NUL) 5207 { 5208 im_activatekey_keyval = GDK_VoidSymbol; 5209 im_activatekey_state = 0; 5210 return TRUE; 5211 } 5212 5213 return im_string_to_keyval((const char *)p_imak, 5214 &im_activatekey_keyval, 5215 &im_activatekey_state); 5216 } 5217 5218 static void 5219 im_synthesize_keypress(unsigned int keyval, unsigned int state) 5220 { 5221 GdkEventKey *event; 5222 5223 # ifdef HAVE_GTK_MULTIHEAD 5224 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); 5225 # if GTK_CHECK_VERSION(3,0,0) 5226 g_object_ref(gtk_widget_get_window(gui.drawarea)); 5227 /* unreffed by gdk_event_free() */ 5228 # else 5229 g_object_ref(gui.drawarea->window); /* unreffed by gdk_event_free() */ 5230 # endif 5231 # else 5232 event = (GdkEventKey *)g_malloc0((gulong)sizeof(GdkEvent)); 5233 event->type = GDK_KEY_PRESS; 5234 # endif 5235 # if GTK_CHECK_VERSION(3,0,0) 5236 event->window = gtk_widget_get_window(gui.drawarea); 5237 # else 5238 event->window = gui.drawarea->window; 5239 # endif 5240 event->send_event = TRUE; 5241 event->time = GDK_CURRENT_TIME; 5242 event->state = state; 5243 event->keyval = keyval; 5244 event->hardware_keycode = /* needed for XIM */ 5245 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); 5246 event->length = 0; 5247 event->string = NULL; 5248 5249 gtk_im_context_filter_keypress(xic, event); 5250 5251 /* For consistency, also send the corresponding release event. */ 5252 event->type = GDK_KEY_RELEASE; 5253 event->send_event = FALSE; 5254 gtk_im_context_filter_keypress(xic, event); 5255 5256 # ifdef HAVE_GTK_MULTIHEAD 5257 gdk_event_free((GdkEvent *)event); 5258 # else 5259 g_free(event); 5260 # endif 5261 } 5262 5263 void 5264 xim_reset(void) 5265 { 5266 if (xic != NULL) 5267 { 5268 gtk_im_context_reset(xic); 5269 5270 if (p_imdisable) 5271 im_shutdown(); 5272 else 5273 { 5274 xim_set_focus(gui.in_focus); 5275 5276 # ifdef FEAT_EVAL 5277 if (p_imaf[0] != NUL) 5278 { 5279 char_u *argv[1]; 5280 5281 if (im_is_active) 5282 argv[0] = (char_u *)"1"; 5283 else 5284 argv[0] = (char_u *)"0"; 5285 (void)call_func_retnr(p_imaf, 1, argv, FALSE); 5286 } 5287 else 5288 # endif 5289 if (im_activatekey_keyval != GDK_VoidSymbol) 5290 { 5291 if (im_is_active) 5292 { 5293 g_signal_handler_block(xic, im_commit_handler_id); 5294 im_synthesize_keypress(im_activatekey_keyval, 5295 im_activatekey_state); 5296 g_signal_handler_unblock(xic, im_commit_handler_id); 5297 } 5298 } 5299 else 5300 { 5301 im_shutdown(); 5302 xim_init(); 5303 xim_set_focus(gui.in_focus); 5304 } 5305 } 5306 } 5307 5308 preedit_start_col = MAXCOL; 5309 xim_has_preediting = FALSE; 5310 } 5311 5312 int 5313 xim_queue_key_press_event(GdkEventKey *event, int down) 5314 { 5315 if (down) 5316 { 5317 /* 5318 * Workaround GTK2 XIM 'feature' that always converts keypad keys to 5319 * chars., even when not part of an IM sequence (ref. feature of 5320 * gdk/gdkkeyuni.c). 5321 * Flag any keypad keys that might represent a single char. 5322 * If this (on its own - i.e., not part of an IM sequence) is 5323 * committed while we're processing one of these keys, we can ignore 5324 * that commit and go ahead & process it ourselves. That way we can 5325 * still distinguish keypad keys for use in mappings. 5326 * Also add GDK_space to make <S-Space> work. 5327 */ 5328 switch (event->keyval) 5329 { 5330 case GDK_KP_Add: xim_expected_char = '+'; break; 5331 case GDK_KP_Subtract: xim_expected_char = '-'; break; 5332 case GDK_KP_Divide: xim_expected_char = '/'; break; 5333 case GDK_KP_Multiply: xim_expected_char = '*'; break; 5334 case GDK_KP_Decimal: xim_expected_char = '.'; break; 5335 case GDK_KP_Equal: xim_expected_char = '='; break; 5336 case GDK_KP_0: xim_expected_char = '0'; break; 5337 case GDK_KP_1: xim_expected_char = '1'; break; 5338 case GDK_KP_2: xim_expected_char = '2'; break; 5339 case GDK_KP_3: xim_expected_char = '3'; break; 5340 case GDK_KP_4: xim_expected_char = '4'; break; 5341 case GDK_KP_5: xim_expected_char = '5'; break; 5342 case GDK_KP_6: xim_expected_char = '6'; break; 5343 case GDK_KP_7: xim_expected_char = '7'; break; 5344 case GDK_KP_8: xim_expected_char = '8'; break; 5345 case GDK_KP_9: xim_expected_char = '9'; break; 5346 case GDK_space: xim_expected_char = ' '; break; 5347 default: xim_expected_char = NUL; 5348 } 5349 xim_ignored_char = FALSE; 5350 } 5351 5352 /* 5353 * When typing fFtT, XIM may be activated. Thus it must pass 5354 * gtk_im_context_filter_keypress() in Normal mode. 5355 * And while doing :sh too. 5356 */ 5357 if (xic != NULL && !p_imdisable 5358 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) 5359 { 5360 /* 5361 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is 5362 * always aware of the current status of IM, and can even emulate 5363 * the activation key for modules that don't support one. 5364 */ 5365 if (event->keyval == im_activatekey_keyval 5366 && (event->state & im_activatekey_state) == im_activatekey_state) 5367 { 5368 unsigned int state_mask; 5369 5370 /* Require the state of the 3 most used modifiers to match exactly. 5371 * Otherwise e.g. <S-C-space> would be unusable for other purposes 5372 * if the IM activate key is <S-space>. */ 5373 state_mask = im_activatekey_state; 5374 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK 5375 | (int)GDK_MOD1_MASK); 5376 5377 if ((event->state & state_mask) != im_activatekey_state) 5378 return FALSE; 5379 5380 /* Don't send it a second time on GDK_KEY_RELEASE. */ 5381 if (event->type != GDK_KEY_PRESS) 5382 return TRUE; 5383 5384 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE)) 5385 { 5386 im_set_active(FALSE); 5387 5388 /* ":lmap" mappings exists, toggle use of mappings. */ 5389 State ^= LANGMAP; 5390 if (State & LANGMAP) 5391 { 5392 curbuf->b_p_iminsert = B_IMODE_NONE; 5393 State &= ~LANGMAP; 5394 } 5395 else 5396 { 5397 curbuf->b_p_iminsert = B_IMODE_LMAP; 5398 State |= LANGMAP; 5399 } 5400 return TRUE; 5401 } 5402 5403 return gtk_im_context_filter_keypress(xic, event); 5404 } 5405 5406 /* Don't filter events through the IM context if IM isn't active 5407 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module 5408 * not doing anything before the activation key was sent. */ 5409 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) 5410 { 5411 int imresult = gtk_im_context_filter_keypress(xic, event); 5412 5413 /* Some XIM send following sequence: 5414 * 1. preedited string. 5415 * 2. committed string. 5416 * 3. line changed key. 5417 * 4. preedited string. 5418 * 5. remove preedited string. 5419 * if 3, Vim can't move back the above line for 5. 5420 * thus, this part should not parse the key. */ 5421 if (!imresult && preedit_start_col != MAXCOL 5422 && event->keyval == GDK_Return) 5423 { 5424 im_synthesize_keypress(GDK_Return, 0U); 5425 return FALSE; 5426 } 5427 5428 /* If XIM tried to commit a keypad key as a single char., 5429 * ignore it so we can use the keypad key 'raw', for mappings. */ 5430 if (xim_expected_char != NUL && xim_ignored_char) 5431 /* We had a keypad key, and XIM tried to thieve it */ 5432 return FALSE; 5433 5434 /* This is supposed to fix a problem with iBus, that space 5435 * characters don't work in input mode. */ 5436 xim_expected_char = NUL; 5437 5438 /* Normal processing */ 5439 return imresult; 5440 } 5441 } 5442 5443 return FALSE; 5444 } 5445 5446 int 5447 im_get_status(void) 5448 { 5449 # ifdef FEAT_EVAL 5450 if (p_imsf[0] != NUL) 5451 { 5452 int is_active; 5453 5454 /* FIXME: Don't execute user function in unsafe situation. */ 5455 if (exiting 5456 # ifdef FEAT_AUTOCMD 5457 || is_autocmd_blocked() 5458 # endif 5459 ) 5460 return FALSE; 5461 /* FIXME: :py print 'xxx' is shown duplicate result. 5462 * Use silent to avoid it. */ 5463 ++msg_silent; 5464 is_active = call_func_retnr(p_imsf, 0, NULL, FALSE); 5465 --msg_silent; 5466 return (is_active > 0); 5467 } 5468 # endif 5469 return im_is_active; 5470 } 5471 5472 int 5473 preedit_get_status(void) 5474 { 5475 return preedit_is_active; 5476 } 5477 5478 int 5479 im_is_preediting(void) 5480 { 5481 return xim_has_preediting; 5482 } 5483 5484 # else /* !FEAT_GUI_GTK */ 5485 5486 static int xim_is_active = FALSE; /* XIM should be active in the current 5487 mode */ 5488 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */ 5489 #ifdef FEAT_GUI_X11 5490 static XIMStyle input_style; 5491 static int status_area_enabled = TRUE; 5492 #endif 5493 5494 /* 5495 * Switch using XIM on/off. This is used by the code that changes "State". 5496 */ 5497 void 5498 im_set_active(int active) 5499 { 5500 if (xic == NULL) 5501 return; 5502 5503 /* If 'imdisable' is set, XIM is never active. */ 5504 if (p_imdisable) 5505 active = FALSE; 5506 #if !defined(FEAT_GUI_GTK) 5507 else if (input_style & XIMPreeditPosition) 5508 /* There is a problem in switching XIM off when preediting is used, 5509 * and it is not clear how this can be solved. For now, keep XIM on 5510 * all the time, like it was done in Vim 5.8. */ 5511 active = TRUE; 5512 #endif 5513 5514 /* Remember the active state, it is needed when Vim gets keyboard focus. */ 5515 xim_is_active = active; 5516 xim_set_preedit(); 5517 } 5518 5519 /* 5520 * Adjust using XIM for gaining or losing keyboard focus. Also called when 5521 * "xim_is_active" changes. 5522 */ 5523 void 5524 xim_set_focus(int focus) 5525 { 5526 if (xic == NULL) 5527 return; 5528 5529 /* 5530 * XIM only gets focus when the Vim window has keyboard focus and XIM has 5531 * been set active for the current mode. 5532 */ 5533 if (focus && xim_is_active) 5534 { 5535 if (!xim_has_focus) 5536 { 5537 xim_has_focus = TRUE; 5538 XSetICFocus(xic); 5539 } 5540 } 5541 else 5542 { 5543 if (xim_has_focus) 5544 { 5545 xim_has_focus = FALSE; 5546 XUnsetICFocus(xic); 5547 } 5548 } 5549 } 5550 5551 void 5552 im_set_position(int row UNUSED, int col UNUSED) 5553 { 5554 xim_set_preedit(); 5555 } 5556 5557 /* 5558 * Set the XIM to the current cursor position. 5559 */ 5560 void 5561 xim_set_preedit(void) 5562 { 5563 XVaNestedList attr_list; 5564 XRectangle spot_area; 5565 XPoint over_spot; 5566 int line_space; 5567 5568 if (xic == NULL) 5569 return; 5570 5571 xim_set_focus(TRUE); 5572 5573 if (!xim_has_focus) 5574 { 5575 /* hide XIM cursor */ 5576 over_spot.x = 0; 5577 over_spot.y = -100; /* arbitrary invisible position */ 5578 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5579 XNSpotLocation, 5580 &over_spot, 5581 NULL); 5582 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); 5583 XFree(attr_list); 5584 return; 5585 } 5586 5587 if (input_style & XIMPreeditPosition) 5588 { 5589 if (xim_fg_color == INVALCOLOR) 5590 { 5591 xim_fg_color = gui.def_norm_pixel; 5592 xim_bg_color = gui.def_back_pixel; 5593 } 5594 over_spot.x = TEXT_X(gui.col); 5595 over_spot.y = TEXT_Y(gui.row); 5596 spot_area.x = 0; 5597 spot_area.y = 0; 5598 spot_area.height = gui.char_height * Rows; 5599 spot_area.width = gui.char_width * Columns; 5600 line_space = gui.char_height; 5601 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5602 XNSpotLocation, &over_spot, 5603 XNForeground, (Pixel) xim_fg_color, 5604 XNBackground, (Pixel) xim_bg_color, 5605 XNArea, &spot_area, 5606 XNLineSpace, line_space, 5607 NULL); 5608 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) 5609 EMSG(_("E284: Cannot set IC values")); 5610 XFree(attr_list); 5611 } 5612 } 5613 5614 #if defined(FEAT_GUI_X11) 5615 static char e_xim[] = N_("E285: Failed to create input context"); 5616 #endif 5617 5618 #if defined(FEAT_GUI_X11) || defined(PROTO) 5619 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(sun) 5620 # define USE_X11R6_XIM 5621 # endif 5622 5623 static int xim_real_init(Window x11_window, Display *x11_display); 5624 5625 5626 #ifdef USE_X11R6_XIM 5627 static void xim_instantiate_cb(Display *display, XPointer client_data, XPointer call_data); 5628 static void xim_destroy_cb(XIM im, XPointer client_data, XPointer call_data); 5629 5630 static void 5631 xim_instantiate_cb( 5632 Display *display, 5633 XPointer client_data UNUSED, 5634 XPointer call_data UNUSED) 5635 { 5636 Window x11_window; 5637 Display *x11_display; 5638 5639 #ifdef XIM_DEBUG 5640 xim_log("xim_instantiate_cb()\n"); 5641 #endif 5642 5643 gui_get_x11_windis(&x11_window, &x11_display); 5644 if (display != x11_display) 5645 return; 5646 5647 xim_real_init(x11_window, x11_display); 5648 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5649 if (xic != NULL) 5650 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5651 xim_instantiate_cb, NULL); 5652 } 5653 5654 static void 5655 xim_destroy_cb( 5656 XIM im UNUSED, 5657 XPointer client_data UNUSED, 5658 XPointer call_data UNUSED) 5659 { 5660 Window x11_window; 5661 Display *x11_display; 5662 5663 #ifdef XIM_DEBUG 5664 xim_log("xim_destroy_cb()\n"); 5665 #endif 5666 gui_get_x11_windis(&x11_window, &x11_display); 5667 5668 xic = NULL; 5669 status_area_enabled = FALSE; 5670 5671 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5672 5673 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5674 xim_instantiate_cb, NULL); 5675 } 5676 #endif 5677 5678 void 5679 xim_init(void) 5680 { 5681 Window x11_window; 5682 Display *x11_display; 5683 5684 #ifdef XIM_DEBUG 5685 xim_log("xim_init()\n"); 5686 #endif 5687 5688 gui_get_x11_windis(&x11_window, &x11_display); 5689 5690 xic = NULL; 5691 5692 if (xim_real_init(x11_window, x11_display)) 5693 return; 5694 5695 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5696 5697 #ifdef USE_X11R6_XIM 5698 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5699 xim_instantiate_cb, NULL); 5700 #endif 5701 } 5702 5703 static int 5704 xim_real_init(Window x11_window, Display *x11_display) 5705 { 5706 int i; 5707 char *p, 5708 *s, 5709 *ns, 5710 *end, 5711 tmp[1024]; 5712 #define IMLEN_MAX 40 5713 char buf[IMLEN_MAX + 7]; 5714 XIM xim = NULL; 5715 XIMStyles *xim_styles; 5716 XIMStyle this_input_style = 0; 5717 Boolean found; 5718 XPoint over_spot; 5719 XVaNestedList preedit_list, status_list; 5720 5721 input_style = 0; 5722 status_area_enabled = FALSE; 5723 5724 if (xic != NULL) 5725 return FALSE; 5726 5727 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) 5728 { 5729 strcpy(tmp, gui.rsrc_input_method); 5730 for (ns = s = tmp; ns != NULL && *s != NUL;) 5731 { 5732 s = (char *)skipwhite((char_u *)s); 5733 if (*s == NUL) 5734 break; 5735 if ((ns = end = strchr(s, ',')) == NULL) 5736 end = s + strlen(s); 5737 while (isspace(((char_u *)end)[-1])) 5738 end--; 5739 *end = NUL; 5740 5741 if (strlen(s) <= IMLEN_MAX) 5742 { 5743 strcpy(buf, "@im="); 5744 strcat(buf, s); 5745 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL 5746 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) 5747 != NULL) 5748 break; 5749 } 5750 5751 s = ns + 1; 5752 } 5753 } 5754 5755 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) 5756 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5757 5758 /* This is supposed to be useful to obtain characters through 5759 * XmbLookupString() without really using a XIM. */ 5760 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL 5761 && *p != NUL) 5762 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5763 5764 if (xim == NULL) 5765 { 5766 /* Only give this message when verbose is set, because too many people 5767 * got this message when they didn't want to use a XIM. */ 5768 if (p_verbose > 0) 5769 { 5770 verbose_enter(); 5771 EMSG(_("E286: Failed to open input method")); 5772 verbose_leave(); 5773 } 5774 return FALSE; 5775 } 5776 5777 #ifdef USE_X11R6_XIM 5778 { 5779 XIMCallback destroy_cb; 5780 5781 destroy_cb.callback = xim_destroy_cb; 5782 destroy_cb.client_data = NULL; 5783 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) 5784 EMSG(_("E287: Warning: Could not set destroy callback to IM")); 5785 } 5786 #endif 5787 5788 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) 5789 { 5790 EMSG(_("E288: input method doesn't support any style")); 5791 XCloseIM(xim); 5792 return FALSE; 5793 } 5794 5795 found = False; 5796 strcpy(tmp, gui.rsrc_preedit_type_name); 5797 for (s = tmp; s && !found; ) 5798 { 5799 while (*s && isspace((unsigned char)*s)) 5800 s++; 5801 if (!*s) 5802 break; 5803 if ((ns = end = strchr(s, ',')) != 0) 5804 ns++; 5805 else 5806 end = s + strlen(s); 5807 while (isspace((unsigned char)*end)) 5808 end--; 5809 *end = '\0'; 5810 5811 if (!strcmp(s, "OverTheSpot")) 5812 this_input_style = (XIMPreeditPosition | XIMStatusArea); 5813 else if (!strcmp(s, "OffTheSpot")) 5814 this_input_style = (XIMPreeditArea | XIMStatusArea); 5815 else if (!strcmp(s, "Root")) 5816 this_input_style = (XIMPreeditNothing | XIMStatusNothing); 5817 5818 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5819 { 5820 if (this_input_style == xim_styles->supported_styles[i]) 5821 { 5822 found = True; 5823 break; 5824 } 5825 } 5826 if (!found) 5827 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5828 { 5829 if ((xim_styles->supported_styles[i] & this_input_style) 5830 == (this_input_style & ~XIMStatusArea)) 5831 { 5832 this_input_style &= ~XIMStatusArea; 5833 found = True; 5834 break; 5835 } 5836 } 5837 5838 s = ns; 5839 } 5840 XFree(xim_styles); 5841 5842 if (!found) 5843 { 5844 /* Only give this message when verbose is set, because too many people 5845 * got this message when they didn't want to use a XIM. */ 5846 if (p_verbose > 0) 5847 { 5848 verbose_enter(); 5849 EMSG(_("E289: input method doesn't support my preedit type")); 5850 verbose_leave(); 5851 } 5852 XCloseIM(xim); 5853 return FALSE; 5854 } 5855 5856 over_spot.x = TEXT_X(gui.col); 5857 over_spot.y = TEXT_Y(gui.row); 5858 input_style = this_input_style; 5859 5860 /* A crash was reported when trying to pass gui.norm_font as XNFontSet, 5861 * thus that has been removed. Hopefully the default works... */ 5862 #ifdef FEAT_XFONTSET 5863 if (gui.fontset != NOFONTSET) 5864 { 5865 preedit_list = XVaCreateNestedList(0, 5866 XNSpotLocation, &over_spot, 5867 XNForeground, (Pixel)gui.def_norm_pixel, 5868 XNBackground, (Pixel)gui.def_back_pixel, 5869 XNFontSet, (XFontSet)gui.fontset, 5870 NULL); 5871 status_list = XVaCreateNestedList(0, 5872 XNForeground, (Pixel)gui.def_norm_pixel, 5873 XNBackground, (Pixel)gui.def_back_pixel, 5874 XNFontSet, (XFontSet)gui.fontset, 5875 NULL); 5876 } 5877 else 5878 #endif 5879 { 5880 preedit_list = XVaCreateNestedList(0, 5881 XNSpotLocation, &over_spot, 5882 XNForeground, (Pixel)gui.def_norm_pixel, 5883 XNBackground, (Pixel)gui.def_back_pixel, 5884 NULL); 5885 status_list = XVaCreateNestedList(0, 5886 XNForeground, (Pixel)gui.def_norm_pixel, 5887 XNBackground, (Pixel)gui.def_back_pixel, 5888 NULL); 5889 } 5890 5891 xic = XCreateIC(xim, 5892 XNInputStyle, input_style, 5893 XNClientWindow, x11_window, 5894 XNFocusWindow, gui.wid, 5895 XNPreeditAttributes, preedit_list, 5896 XNStatusAttributes, status_list, 5897 NULL); 5898 XFree(status_list); 5899 XFree(preedit_list); 5900 if (xic != NULL) 5901 { 5902 if (input_style & XIMStatusArea) 5903 { 5904 xim_set_status_area(); 5905 status_area_enabled = TRUE; 5906 } 5907 else 5908 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5909 } 5910 else 5911 { 5912 EMSG(_(e_xim)); 5913 XCloseIM(xim); 5914 return FALSE; 5915 } 5916 5917 return TRUE; 5918 } 5919 5920 #endif /* FEAT_GUI_X11 */ 5921 5922 /* 5923 * Get IM status. When IM is on, return TRUE. Else return FALSE. 5924 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is 5925 * active, when not having focus XIM may still be active (e.g., when using a 5926 * tear-off menu item). 5927 */ 5928 int 5929 im_get_status(void) 5930 { 5931 return xim_has_focus; 5932 } 5933 5934 # endif /* !FEAT_GUI_GTK */ 5935 5936 # if !defined(FEAT_GUI_GTK) || defined(PROTO) 5937 /* 5938 * Set up the status area. 5939 * 5940 * This should use a separate Widget, but that seems not possible, because 5941 * preedit_area and status_area should be set to the same window as for the 5942 * text input. Unfortunately this means the status area pollutes the text 5943 * window... 5944 */ 5945 void 5946 xim_set_status_area(void) 5947 { 5948 XVaNestedList preedit_list = 0, status_list = 0, list = 0; 5949 XRectangle pre_area, status_area; 5950 5951 if (xic == NULL) 5952 return; 5953 5954 if (input_style & XIMStatusArea) 5955 { 5956 if (input_style & XIMPreeditArea) 5957 { 5958 XRectangle *needed_rect; 5959 5960 /* to get status_area width */ 5961 status_list = XVaCreateNestedList(0, XNAreaNeeded, 5962 &needed_rect, NULL); 5963 XGetICValues(xic, XNStatusAttributes, status_list, NULL); 5964 XFree(status_list); 5965 5966 status_area.width = needed_rect->width; 5967 } 5968 else 5969 status_area.width = gui.char_width * Columns; 5970 5971 status_area.x = 0; 5972 status_area.y = gui.char_height * Rows + gui.border_offset; 5973 if (gui.which_scrollbars[SBAR_BOTTOM]) 5974 status_area.y += gui.scrollbar_height; 5975 #ifdef FEAT_MENU 5976 if (gui.menu_is_active) 5977 status_area.y += gui.menu_height; 5978 #endif 5979 status_area.height = gui.char_height; 5980 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); 5981 } 5982 else 5983 { 5984 status_area.x = 0; 5985 status_area.y = gui.char_height * Rows + gui.border_offset; 5986 if (gui.which_scrollbars[SBAR_BOTTOM]) 5987 status_area.y += gui.scrollbar_height; 5988 #ifdef FEAT_MENU 5989 if (gui.menu_is_active) 5990 status_area.y += gui.menu_height; 5991 #endif 5992 status_area.width = 0; 5993 status_area.height = gui.char_height; 5994 } 5995 5996 if (input_style & XIMPreeditArea) /* off-the-spot */ 5997 { 5998 pre_area.x = status_area.x + status_area.width; 5999 pre_area.y = gui.char_height * Rows + gui.border_offset; 6000 pre_area.width = gui.char_width * Columns - pre_area.x; 6001 if (gui.which_scrollbars[SBAR_BOTTOM]) 6002 pre_area.y += gui.scrollbar_height; 6003 #ifdef FEAT_MENU 6004 if (gui.menu_is_active) 6005 pre_area.y += gui.menu_height; 6006 #endif 6007 pre_area.height = gui.char_height; 6008 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6009 } 6010 else if (input_style & XIMPreeditPosition) /* over-the-spot */ 6011 { 6012 pre_area.x = 0; 6013 pre_area.y = 0; 6014 pre_area.height = gui.char_height * Rows; 6015 pre_area.width = gui.char_width * Columns; 6016 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6017 } 6018 6019 if (preedit_list && status_list) 6020 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6021 XNStatusAttributes, status_list, NULL); 6022 else if (preedit_list) 6023 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6024 NULL); 6025 else if (status_list) 6026 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, 6027 NULL); 6028 else 6029 list = NULL; 6030 6031 if (list) 6032 { 6033 XSetICValues(xic, XNVaNestedList, list, NULL); 6034 XFree(list); 6035 } 6036 if (status_list) 6037 XFree(status_list); 6038 if (preedit_list) 6039 XFree(preedit_list); 6040 } 6041 6042 int 6043 xim_get_status_area_height(void) 6044 { 6045 if (status_area_enabled) 6046 return gui.char_height; 6047 return 0; 6048 } 6049 # endif 6050 6051 #endif /* FEAT_XIM */ 6052 6053 #if defined(FEAT_MBYTE) || defined(PROTO) 6054 6055 /* 6056 * Setup "vcp" for conversion from "from" to "to". 6057 * The names must have been made canonical with enc_canonize(). 6058 * vcp->vc_type must have been initialized to CONV_NONE. 6059 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 6060 * instead). 6061 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 6062 * Return FAIL when conversion is not supported, OK otherwise. 6063 */ 6064 int 6065 convert_setup(vimconv_T *vcp, char_u *from, char_u *to) 6066 { 6067 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 6068 } 6069 6070 /* 6071 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 6072 * "from" unicode charsets be considered utf-8. Same for "to". 6073 */ 6074 int 6075 convert_setup_ext( 6076 vimconv_T *vcp, 6077 char_u *from, 6078 int from_unicode_is_utf8, 6079 char_u *to, 6080 int to_unicode_is_utf8) 6081 { 6082 int from_prop; 6083 int to_prop; 6084 int from_is_utf8; 6085 int to_is_utf8; 6086 6087 /* Reset to no conversion. */ 6088 # ifdef USE_ICONV 6089 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 6090 iconv_close(vcp->vc_fd); 6091 # endif 6092 vcp->vc_type = CONV_NONE; 6093 vcp->vc_factor = 1; 6094 vcp->vc_fail = FALSE; 6095 6096 /* No conversion when one of the names is empty or they are equal. */ 6097 if (from == NULL || *from == NUL || to == NULL || *to == NUL 6098 || STRCMP(from, to) == 0) 6099 return OK; 6100 6101 from_prop = enc_canon_props(from); 6102 to_prop = enc_canon_props(to); 6103 if (from_unicode_is_utf8) 6104 from_is_utf8 = from_prop & ENC_UNICODE; 6105 else 6106 from_is_utf8 = from_prop == ENC_UNICODE; 6107 if (to_unicode_is_utf8) 6108 to_is_utf8 = to_prop & ENC_UNICODE; 6109 else 6110 to_is_utf8 = to_prop == ENC_UNICODE; 6111 6112 if ((from_prop & ENC_LATIN1) && to_is_utf8) 6113 { 6114 /* Internal latin1 -> utf-8 conversion. */ 6115 vcp->vc_type = CONV_TO_UTF8; 6116 vcp->vc_factor = 2; /* up to twice as long */ 6117 } 6118 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 6119 { 6120 /* Internal latin9 -> utf-8 conversion. */ 6121 vcp->vc_type = CONV_9_TO_UTF8; 6122 vcp->vc_factor = 3; /* up to three as long (euro sign) */ 6123 } 6124 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 6125 { 6126 /* Internal utf-8 -> latin1 conversion. */ 6127 vcp->vc_type = CONV_TO_LATIN1; 6128 } 6129 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 6130 { 6131 /* Internal utf-8 -> latin9 conversion. */ 6132 vcp->vc_type = CONV_TO_LATIN9; 6133 } 6134 #ifdef WIN3264 6135 /* Win32-specific codepage <-> codepage conversion without iconv. */ 6136 else if ((from_is_utf8 || encname2codepage(from) > 0) 6137 && (to_is_utf8 || encname2codepage(to) > 0)) 6138 { 6139 vcp->vc_type = CONV_CODEPAGE; 6140 vcp->vc_factor = 2; /* up to twice as long */ 6141 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 6142 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 6143 } 6144 #endif 6145 #ifdef MACOS_X 6146 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 6147 { 6148 vcp->vc_type = CONV_MAC_LATIN1; 6149 } 6150 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 6151 { 6152 vcp->vc_type = CONV_MAC_UTF8; 6153 vcp->vc_factor = 2; /* up to twice as long */ 6154 } 6155 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 6156 { 6157 vcp->vc_type = CONV_LATIN1_MAC; 6158 } 6159 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 6160 { 6161 vcp->vc_type = CONV_UTF8_MAC; 6162 } 6163 #endif 6164 # ifdef USE_ICONV 6165 else 6166 { 6167 /* Use iconv() for conversion. */ 6168 vcp->vc_fd = (iconv_t)my_iconv_open( 6169 to_is_utf8 ? (char_u *)"utf-8" : to, 6170 from_is_utf8 ? (char_u *)"utf-8" : from); 6171 if (vcp->vc_fd != (iconv_t)-1) 6172 { 6173 vcp->vc_type = CONV_ICONV; 6174 vcp->vc_factor = 4; /* could be longer too... */ 6175 } 6176 } 6177 # endif 6178 if (vcp->vc_type == CONV_NONE) 6179 return FAIL; 6180 6181 return OK; 6182 } 6183 6184 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \ 6185 || defined(PROTO) 6186 /* 6187 * Do conversion on typed input characters in-place. 6188 * The input and output are not NUL terminated! 6189 * Returns the length after conversion. 6190 */ 6191 int 6192 convert_input(char_u *ptr, int len, int maxlen) 6193 { 6194 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 6195 } 6196 #endif 6197 6198 /* 6199 * Like convert_input(), but when there is an incomplete byte sequence at the 6200 * end return that as an allocated string in "restp" and set "*restlenp" to 6201 * the length. If "restp" is NULL it is not used. 6202 */ 6203 int 6204 convert_input_safe( 6205 char_u *ptr, 6206 int len, 6207 int maxlen, 6208 char_u **restp, 6209 int *restlenp) 6210 { 6211 char_u *d; 6212 int dlen = len; 6213 int unconvertlen = 0; 6214 6215 d = string_convert_ext(&input_conv, ptr, &dlen, 6216 restp == NULL ? NULL : &unconvertlen); 6217 if (d != NULL) 6218 { 6219 if (dlen <= maxlen) 6220 { 6221 if (unconvertlen > 0) 6222 { 6223 /* Move the unconverted characters to allocated memory. */ 6224 *restp = alloc(unconvertlen); 6225 if (*restp != NULL) 6226 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 6227 *restlenp = unconvertlen; 6228 } 6229 mch_memmove(ptr, d, dlen); 6230 } 6231 else 6232 /* result is too long, keep the unconverted text (the caller must 6233 * have done something wrong!) */ 6234 dlen = len; 6235 vim_free(d); 6236 } 6237 return dlen; 6238 } 6239 6240 /* 6241 * Convert text "ptr[*lenp]" according to "vcp". 6242 * Returns the result in allocated memory and sets "*lenp". 6243 * When "lenp" is NULL, use NUL terminated strings. 6244 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 6245 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 6246 */ 6247 char_u * 6248 string_convert( 6249 vimconv_T *vcp, 6250 char_u *ptr, 6251 int *lenp) 6252 { 6253 return string_convert_ext(vcp, ptr, lenp, NULL); 6254 } 6255 6256 /* 6257 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 6258 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 6259 * set to the number of remaining bytes. 6260 */ 6261 char_u * 6262 string_convert_ext( 6263 vimconv_T *vcp, 6264 char_u *ptr, 6265 int *lenp, 6266 int *unconvlenp) 6267 { 6268 char_u *retval = NULL; 6269 char_u *d; 6270 int len; 6271 int i; 6272 int l; 6273 int c; 6274 6275 if (lenp == NULL) 6276 len = (int)STRLEN(ptr); 6277 else 6278 len = *lenp; 6279 if (len == 0) 6280 return vim_strsave((char_u *)""); 6281 6282 switch (vcp->vc_type) 6283 { 6284 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */ 6285 retval = alloc(len * 2 + 1); 6286 if (retval == NULL) 6287 break; 6288 d = retval; 6289 for (i = 0; i < len; ++i) 6290 { 6291 c = ptr[i]; 6292 if (c < 0x80) 6293 *d++ = c; 6294 else 6295 { 6296 *d++ = 0xc0 + ((unsigned)c >> 6); 6297 *d++ = 0x80 + (c & 0x3f); 6298 } 6299 } 6300 *d = NUL; 6301 if (lenp != NULL) 6302 *lenp = (int)(d - retval); 6303 break; 6304 6305 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */ 6306 retval = alloc(len * 3 + 1); 6307 if (retval == NULL) 6308 break; 6309 d = retval; 6310 for (i = 0; i < len; ++i) 6311 { 6312 c = ptr[i]; 6313 switch (c) 6314 { 6315 case 0xa4: c = 0x20ac; break; /* euro */ 6316 case 0xa6: c = 0x0160; break; /* S hat */ 6317 case 0xa8: c = 0x0161; break; /* S -hat */ 6318 case 0xb4: c = 0x017d; break; /* Z hat */ 6319 case 0xb8: c = 0x017e; break; /* Z -hat */ 6320 case 0xbc: c = 0x0152; break; /* OE */ 6321 case 0xbd: c = 0x0153; break; /* oe */ 6322 case 0xbe: c = 0x0178; break; /* Y */ 6323 } 6324 d += utf_char2bytes(c, d); 6325 } 6326 *d = NUL; 6327 if (lenp != NULL) 6328 *lenp = (int)(d - retval); 6329 break; 6330 6331 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */ 6332 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */ 6333 retval = alloc(len + 1); 6334 if (retval == NULL) 6335 break; 6336 d = retval; 6337 for (i = 0; i < len; ++i) 6338 { 6339 l = utf_ptr2len_len(ptr + i, len - i); 6340 if (l == 0) 6341 *d++ = NUL; 6342 else if (l == 1) 6343 { 6344 int l_w = utf8len_tab_zero[ptr[i]]; 6345 6346 if (l_w == 0) 6347 { 6348 /* Illegal utf-8 byte cannot be converted */ 6349 vim_free(retval); 6350 return NULL; 6351 } 6352 if (unconvlenp != NULL && l_w > len - i) 6353 { 6354 /* Incomplete sequence at the end. */ 6355 *unconvlenp = len - i; 6356 break; 6357 } 6358 *d++ = ptr[i]; 6359 } 6360 else 6361 { 6362 c = utf_ptr2char(ptr + i); 6363 if (vcp->vc_type == CONV_TO_LATIN9) 6364 switch (c) 6365 { 6366 case 0x20ac: c = 0xa4; break; /* euro */ 6367 case 0x0160: c = 0xa6; break; /* S hat */ 6368 case 0x0161: c = 0xa8; break; /* S -hat */ 6369 case 0x017d: c = 0xb4; break; /* Z hat */ 6370 case 0x017e: c = 0xb8; break; /* Z -hat */ 6371 case 0x0152: c = 0xbc; break; /* OE */ 6372 case 0x0153: c = 0xbd; break; /* oe */ 6373 case 0x0178: c = 0xbe; break; /* Y */ 6374 case 0xa4: 6375 case 0xa6: 6376 case 0xa8: 6377 case 0xb4: 6378 case 0xb8: 6379 case 0xbc: 6380 case 0xbd: 6381 case 0xbe: c = 0x100; break; /* not in latin9 */ 6382 } 6383 if (!utf_iscomposing(c)) /* skip composing chars */ 6384 { 6385 if (c < 0x100) 6386 *d++ = c; 6387 else if (vcp->vc_fail) 6388 { 6389 vim_free(retval); 6390 return NULL; 6391 } 6392 else 6393 { 6394 *d++ = 0xbf; 6395 if (utf_char2cells(c) > 1) 6396 *d++ = '?'; 6397 } 6398 } 6399 i += l - 1; 6400 } 6401 } 6402 *d = NUL; 6403 if (lenp != NULL) 6404 *lenp = (int)(d - retval); 6405 break; 6406 6407 # ifdef MACOS_CONVERT 6408 case CONV_MAC_LATIN1: 6409 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6410 'm', 'l', unconvlenp); 6411 break; 6412 6413 case CONV_LATIN1_MAC: 6414 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6415 'l', 'm', unconvlenp); 6416 break; 6417 6418 case CONV_MAC_UTF8: 6419 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6420 'm', 'u', unconvlenp); 6421 break; 6422 6423 case CONV_UTF8_MAC: 6424 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6425 'u', 'm', unconvlenp); 6426 break; 6427 # endif 6428 6429 # ifdef USE_ICONV 6430 case CONV_ICONV: /* conversion with output_conv.vc_fd */ 6431 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 6432 break; 6433 # endif 6434 # ifdef WIN3264 6435 case CONV_CODEPAGE: /* codepage -> codepage */ 6436 { 6437 int retlen; 6438 int tmp_len; 6439 short_u *tmp; 6440 6441 /* 1. codepage/UTF-8 -> ucs-2. */ 6442 if (vcp->vc_cpfrom == 0) 6443 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 6444 else 6445 { 6446 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 6447 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 6448 (char *)ptr, len, 0, 0); 6449 if (tmp_len == 0 6450 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 6451 { 6452 if (lenp != NULL) 6453 *lenp = 0; 6454 if (unconvlenp != NULL) 6455 *unconvlenp = len; 6456 retval = alloc(1); 6457 if (retval) 6458 retval[0] = NUL; 6459 return retval; 6460 } 6461 } 6462 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len); 6463 if (tmp == NULL) 6464 break; 6465 if (vcp->vc_cpfrom == 0) 6466 utf8_to_utf16(ptr, len, tmp, unconvlenp); 6467 else 6468 MultiByteToWideChar(vcp->vc_cpfrom, 0, 6469 (char *)ptr, len, tmp, tmp_len); 6470 6471 /* 2. ucs-2 -> codepage/UTF-8. */ 6472 if (vcp->vc_cpto == 0) 6473 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 6474 else 6475 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 6476 tmp, tmp_len, 0, 0, 0, 0); 6477 retval = alloc(retlen + 1); 6478 if (retval != NULL) 6479 { 6480 if (vcp->vc_cpto == 0) 6481 utf16_to_utf8(tmp, tmp_len, retval); 6482 else 6483 WideCharToMultiByte(vcp->vc_cpto, 0, 6484 tmp, tmp_len, 6485 (char *)retval, retlen, 0, 0); 6486 retval[retlen] = NUL; 6487 if (lenp != NULL) 6488 *lenp = retlen; 6489 } 6490 vim_free(tmp); 6491 break; 6492 } 6493 # endif 6494 } 6495 6496 return retval; 6497 } 6498 #endif 6499