1 /* vi:set ts=8 sts=4 sw=4 noet: 2 * 3 * VIM - Vi IMproved by Bram Moolenaar 4 * Multibyte extensions partly by Sung-Hoon Baek 5 * 6 * Do ":help uganda" in Vim to read copying and usage conditions. 7 * Do ":help credits" in Vim to see a list of people who contributed. 8 * See README.txt for an overview of the Vim source code. 9 */ 10 /* 11 * mbyte.c: Code specifically for handling multi-byte characters. 12 * 13 * The encoding used in the core is set with 'encoding'. When 'encoding' is 14 * changed, the following four variables are set (for speed). 15 * Currently these types of character encodings are supported: 16 * 17 * "enc_dbcs" When non-zero it tells the type of double byte character 18 * encoding (Chinese, Korean, Japanese, etc.). 19 * The cell width on the display is equal to the number of 20 * bytes. (exception: DBCS_JPNU with first byte 0x8e) 21 * Recognizing the first or second byte is difficult, it 22 * requires checking a byte sequence from the start. 23 * "enc_utf8" When TRUE use Unicode characters in UTF-8 encoding. 24 * The cell width on the display needs to be determined from 25 * the character value. 26 * Recognizing bytes is easy: 0xxx.xxxx is a single-byte 27 * char, 10xx.xxxx is a trailing byte, 11xx.xxxx is a leading 28 * byte of a multi-byte character. 29 * To make things complicated, up to six composing characters 30 * are allowed. These are drawn on top of the first char. 31 * For most editing the sequence of bytes with composing 32 * characters included is considered to be one character. 33 * "enc_unicode" When 2 use 16-bit Unicode characters (or UTF-16). 34 * When 4 use 32-but Unicode characters. 35 * Internally characters are stored in UTF-8 encoding to 36 * avoid NUL bytes. Conversion happens when doing I/O. 37 * "enc_utf8" will also be TRUE. 38 * 39 * "has_mbyte" is set when "enc_dbcs" or "enc_utf8" is non-zero. 40 * 41 * If none of these is TRUE, 8-bit bytes are used for a character. The 42 * encoding isn't currently specified (TODO). 43 * 44 * 'encoding' specifies the encoding used in the core. This is in registers, 45 * text manipulation, buffers, etc. Conversion has to be done when characters 46 * in another encoding are received or send: 47 * 48 * clipboard 49 * ^ 50 * | (2) 51 * V 52 * +---------------+ 53 * (1) | | (3) 54 * keyboard ----->| core |-----> display 55 * | | 56 * +---------------+ 57 * ^ 58 * | (4) 59 * V 60 * file 61 * 62 * (1) Typed characters arrive in the current locale. Conversion is to be 63 * done when 'encoding' is different from 'termencoding'. 64 * (2) Text will be made available with the encoding specified with 65 * 'encoding'. If this is not sufficient, system-specific conversion 66 * might be required. 67 * (3) For the GUI the correct font must be selected, no conversion done. 68 * Otherwise, conversion is to be done when 'encoding' differs from 69 * 'termencoding'. (Different in the GTK+ 2 port -- 'termencoding' 70 * is always used for both input and output and must always be set to 71 * "utf-8". gui_mch_init() does this automatically.) 72 * (4) The encoding of the file is specified with 'fileencoding'. Conversion 73 * is to be done when it's different from 'encoding'. 74 * 75 * The viminfo file is a special case: Only text is converted, not file names. 76 * Vim scripts may contain an ":encoding" command. This has an effect for 77 * some commands, like ":menutrans" 78 */ 79 80 #include "vim.h" 81 82 #ifdef WIN32UNIX 83 # ifndef WIN32_LEAN_AND_MEAN 84 # define WIN32_LEAN_AND_MEAN 85 # endif 86 # if defined(FEAT_GUI) || defined(FEAT_XCLIPBOARD) 87 # include <X11/Xwindows.h> 88 # define WINBYTE wBYTE 89 # else 90 # include <windows.h> 91 # define WINBYTE BYTE 92 # endif 93 # ifdef WIN32 94 # undef WIN32 /* Some windows.h define WIN32, we don't want that here. */ 95 # endif 96 #else 97 # define WINBYTE BYTE 98 #endif 99 100 #if (defined(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_buf(utf_ptr2char(p), buf); 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 {0x261c, 0x261c}, 1360 {0x261e, 0x261e}, 1361 {0x2640, 0x2640}, 1362 {0x2642, 0x2642}, 1363 {0x2660, 0x2661}, 1364 {0x2663, 0x2665}, 1365 {0x2667, 0x266a}, 1366 {0x266c, 0x266d}, 1367 {0x266f, 0x266f}, 1368 {0x269e, 0x269f}, 1369 {0x26bf, 0x26bf}, 1370 {0x26c6, 0x26cd}, 1371 {0x26cf, 0x26d3}, 1372 {0x26d5, 0x26e1}, 1373 {0x26e3, 0x26e3}, 1374 {0x26e8, 0x26e9}, 1375 {0x26eb, 0x26f1}, 1376 {0x26f4, 0x26f4}, 1377 {0x26f6, 0x26f9}, 1378 {0x26fb, 0x26fc}, 1379 {0x26fe, 0x26ff}, 1380 {0x273d, 0x273d}, 1381 {0x2776, 0x277f}, 1382 {0x2b56, 0x2b59}, 1383 {0x3248, 0x324f}, 1384 {0xe000, 0xf8ff}, 1385 {0xfe00, 0xfe0f}, 1386 {0xfffd, 0xfffd}, 1387 {0x1f100, 0x1f10a}, 1388 {0x1f110, 0x1f12d}, 1389 {0x1f130, 0x1f169}, 1390 {0x1f170, 0x1f18d}, 1391 {0x1f18f, 0x1f190}, 1392 {0x1f19b, 0x1f1ac}, 1393 {0xe0100, 0xe01ef}, 1394 {0xf0000, 0xffffd}, 1395 {0x100000, 0x10fffd} 1396 }; 1397 1398 /* 1399 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1400 * Returns 4 or 6 for an unprintable character. 1401 * Is only correct for characters >= 0x80. 1402 * When p_ambw is "double", return 2 for a character with East Asian Width 1403 * class 'A'(mbiguous). 1404 */ 1405 int 1406 utf_char2cells(int c) 1407 { 1408 /* Sorted list of non-overlapping intervals of East Asian double width 1409 * characters, generated with ../runtime/tools/unicode.vim. */ 1410 static struct interval doublewidth[] = 1411 { 1412 {0x1100, 0x115f}, 1413 {0x231a, 0x231b}, 1414 {0x2329, 0x232a}, 1415 {0x23e9, 0x23ec}, 1416 {0x23f0, 0x23f0}, 1417 {0x23f3, 0x23f3}, 1418 {0x25fd, 0x25fe}, 1419 {0x2614, 0x2615}, 1420 {0x2648, 0x2653}, 1421 {0x267f, 0x267f}, 1422 {0x2693, 0x2693}, 1423 {0x26a1, 0x26a1}, 1424 {0x26aa, 0x26ab}, 1425 {0x26bd, 0x26be}, 1426 {0x26c4, 0x26c5}, 1427 {0x26ce, 0x26ce}, 1428 {0x26d4, 0x26d4}, 1429 {0x26ea, 0x26ea}, 1430 {0x26f2, 0x26f3}, 1431 {0x26f5, 0x26f5}, 1432 {0x26fa, 0x26fa}, 1433 {0x26fd, 0x26fd}, 1434 {0x2705, 0x2705}, 1435 {0x270a, 0x270b}, 1436 {0x2728, 0x2728}, 1437 {0x274c, 0x274c}, 1438 {0x274e, 0x274e}, 1439 {0x2753, 0x2755}, 1440 {0x2757, 0x2757}, 1441 {0x2795, 0x2797}, 1442 {0x27b0, 0x27b0}, 1443 {0x27bf, 0x27bf}, 1444 {0x2b1b, 0x2b1c}, 1445 {0x2b50, 0x2b50}, 1446 {0x2b55, 0x2b55}, 1447 {0x2e80, 0x2e99}, 1448 {0x2e9b, 0x2ef3}, 1449 {0x2f00, 0x2fd5}, 1450 {0x2ff0, 0x2ffb}, 1451 {0x3000, 0x303e}, 1452 {0x3041, 0x3096}, 1453 {0x3099, 0x30ff}, 1454 {0x3105, 0x312d}, 1455 {0x3131, 0x318e}, 1456 {0x3190, 0x31ba}, 1457 {0x31c0, 0x31e3}, 1458 {0x31f0, 0x321e}, 1459 {0x3220, 0x3247}, 1460 {0x3250, 0x32fe}, 1461 {0x3300, 0x4dbf}, 1462 {0x4e00, 0xa48c}, 1463 {0xa490, 0xa4c6}, 1464 {0xa960, 0xa97c}, 1465 {0xac00, 0xd7a3}, 1466 {0xf900, 0xfaff}, 1467 {0xfe10, 0xfe19}, 1468 {0xfe30, 0xfe52}, 1469 {0xfe54, 0xfe66}, 1470 {0xfe68, 0xfe6b}, 1471 {0xff01, 0xff60}, 1472 {0xffe0, 0xffe6}, 1473 {0x16fe0, 0x16fe0}, 1474 {0x17000, 0x187ec}, 1475 {0x18800, 0x18af2}, 1476 {0x1b000, 0x1b001}, 1477 {0x1f004, 0x1f004}, 1478 {0x1f0cf, 0x1f0cf}, 1479 {0x1f18e, 0x1f18e}, 1480 {0x1f191, 0x1f19a}, 1481 {0x1f200, 0x1f202}, 1482 {0x1f210, 0x1f23b}, 1483 {0x1f240, 0x1f248}, 1484 {0x1f250, 0x1f251}, 1485 {0x1f300, 0x1f320}, 1486 {0x1f32d, 0x1f335}, 1487 {0x1f337, 0x1f37c}, 1488 {0x1f37e, 0x1f393}, 1489 {0x1f3a0, 0x1f3ca}, 1490 {0x1f3cf, 0x1f3d3}, 1491 {0x1f3e0, 0x1f3f0}, 1492 {0x1f3f4, 0x1f3f4}, 1493 {0x1f3f8, 0x1f43e}, 1494 {0x1f440, 0x1f440}, 1495 {0x1f442, 0x1f4fc}, 1496 {0x1f4ff, 0x1f53d}, 1497 {0x1f54b, 0x1f54e}, 1498 {0x1f550, 0x1f567}, 1499 {0x1f57a, 0x1f57a}, 1500 {0x1f595, 0x1f596}, 1501 {0x1f5a4, 0x1f5a4}, 1502 {0x1f5fb, 0x1f64f}, 1503 {0x1f680, 0x1f6c5}, 1504 {0x1f6cc, 0x1f6cc}, 1505 {0x1f6d0, 0x1f6d2}, 1506 {0x1f6eb, 0x1f6ec}, 1507 {0x1f6f4, 0x1f6f6}, 1508 {0x1f910, 0x1f91e}, 1509 {0x1f920, 0x1f927}, 1510 {0x1f930, 0x1f930}, 1511 {0x1f933, 0x1f93e}, 1512 {0x1f940, 0x1f94b}, 1513 {0x1f950, 0x1f95e}, 1514 {0x1f980, 0x1f991}, 1515 {0x1f9c0, 0x1f9c0}, 1516 {0x20000, 0x2fffd}, 1517 {0x30000, 0x3fffd} 1518 }; 1519 1520 /* Sorted list of non-overlapping intervals of Emoji characters that don't 1521 * have ambiguous or double width, 1522 * based on http://unicode.org/emoji/charts/emoji-list.html */ 1523 static struct interval emoji_width[] = 1524 { 1525 {0x1f004, 0x1f004}, 1526 {0x1f0cf, 0x1f0cf}, 1527 {0x1f1e6, 0x1f1ff}, 1528 {0x1f300, 0x1f320}, 1529 {0x1f330, 0x1f335}, 1530 {0x1f337, 0x1f37c}, 1531 {0x1f380, 0x1f393}, 1532 {0x1f3a0, 0x1f3c4}, 1533 {0x1f3c6, 0x1f3ca}, 1534 {0x1f3e0, 0x1f3f0}, 1535 {0x1f400, 0x1f43e}, 1536 {0x1f440, 0x1f440}, 1537 {0x1f442, 0x1f4f7}, 1538 {0x1f4f9, 0x1f4fc}, 1539 {0x1f500, 0x1f53d}, 1540 {0x1f550, 0x1f567}, 1541 {0x1f5fb, 0x1f640}, 1542 {0x1f645, 0x1f64f}, 1543 {0x1f680, 0x1f6c5} 1544 }; 1545 1546 if (c >= 0x100) 1547 { 1548 #ifdef USE_WCHAR_FUNCTIONS 1549 /* 1550 * Assume the library function wcwidth() works better than our own 1551 * stuff. It should return 1 for ambiguous width chars! 1552 */ 1553 int n = wcwidth(c); 1554 1555 if (n < 0) 1556 return 6; /* unprintable, displays <xxxx> */ 1557 if (n > 1) 1558 return n; 1559 #else 1560 if (!utf_printable(c)) 1561 return 6; /* unprintable, displays <xxxx> */ 1562 if (intable(doublewidth, sizeof(doublewidth), c)) 1563 return 2; 1564 #endif 1565 if (p_emoji && intable(emoji_width, sizeof(emoji_width), c)) 1566 return 2; 1567 } 1568 1569 /* Characters below 0x100 are influenced by 'isprint' option */ 1570 else if (c >= 0x80 && !vim_isprintc(c)) 1571 return 4; /* unprintable, displays <xx> */ 1572 1573 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c)) 1574 return 2; 1575 1576 return 1; 1577 } 1578 1579 /* 1580 * mb_ptr2cells() function pointer. 1581 * Return the number of display cells character at "*p" occupies. 1582 * This doesn't take care of unprintable characters, use ptr2cells() for that. 1583 */ 1584 int 1585 latin_ptr2cells(char_u *p UNUSED) 1586 { 1587 return 1; 1588 } 1589 1590 int 1591 utf_ptr2cells( 1592 char_u *p) 1593 { 1594 int c; 1595 1596 /* Need to convert to a wide character. */ 1597 if (*p >= 0x80) 1598 { 1599 c = utf_ptr2char(p); 1600 /* An illegal byte is displayed as <xx>. */ 1601 if (utf_ptr2len(p) == 1 || c == NUL) 1602 return 4; 1603 /* If the char is ASCII it must be an overlong sequence. */ 1604 if (c < 0x80) 1605 return char2cells(c); 1606 return utf_char2cells(c); 1607 } 1608 return 1; 1609 } 1610 1611 int 1612 dbcs_ptr2cells(char_u *p) 1613 { 1614 /* Number of cells is equal to number of bytes, except for euc-jp when 1615 * the first byte is 0x8e. */ 1616 if (enc_dbcs == DBCS_JPNU && *p == 0x8e) 1617 return 1; 1618 return MB_BYTE2LEN(*p); 1619 } 1620 1621 /* 1622 * mb_ptr2cells_len() function pointer. 1623 * Like mb_ptr2cells(), but limit string length to "size". 1624 * For an empty string or truncated character returns 1. 1625 */ 1626 int 1627 latin_ptr2cells_len(char_u *p UNUSED, int size UNUSED) 1628 { 1629 return 1; 1630 } 1631 1632 static int 1633 utf_ptr2cells_len(char_u *p, int size) 1634 { 1635 int c; 1636 1637 /* Need to convert to a wide character. */ 1638 if (size > 0 && *p >= 0x80) 1639 { 1640 if (utf_ptr2len_len(p, size) < utf8len_tab[*p]) 1641 return 1; /* truncated */ 1642 c = utf_ptr2char(p); 1643 /* An illegal byte is displayed as <xx>. */ 1644 if (utf_ptr2len(p) == 1 || c == NUL) 1645 return 4; 1646 /* If the char is ASCII it must be an overlong sequence. */ 1647 if (c < 0x80) 1648 return char2cells(c); 1649 return utf_char2cells(c); 1650 } 1651 return 1; 1652 } 1653 1654 static int 1655 dbcs_ptr2cells_len(char_u *p, int size) 1656 { 1657 /* Number of cells is equal to number of bytes, except for euc-jp when 1658 * the first byte is 0x8e. */ 1659 if (size <= 1 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)) 1660 return 1; 1661 return MB_BYTE2LEN(*p); 1662 } 1663 1664 /* 1665 * mb_char2cells() function pointer. 1666 * Return the number of display cells character "c" occupies. 1667 * Only takes care of multi-byte chars, not "^C" and such. 1668 */ 1669 int 1670 latin_char2cells(int c UNUSED) 1671 { 1672 return 1; 1673 } 1674 1675 static int 1676 dbcs_char2cells(int c) 1677 { 1678 /* Number of cells is equal to number of bytes, except for euc-jp when 1679 * the first byte is 0x8e. */ 1680 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e) 1681 return 1; 1682 /* use the first byte */ 1683 return MB_BYTE2LEN((unsigned)c >> 8); 1684 } 1685 1686 /* 1687 * Return the number of cells occupied by string "p". 1688 * Stop at a NUL character. When "len" >= 0 stop at character "p[len]". 1689 */ 1690 int 1691 mb_string2cells(char_u *p, int len) 1692 { 1693 int i; 1694 int clen = 0; 1695 1696 for (i = 0; (len < 0 || i < len) && p[i] != NUL; i += (*mb_ptr2len)(p + i)) 1697 clen += (*mb_ptr2cells)(p + i); 1698 return clen; 1699 } 1700 1701 /* 1702 * mb_off2cells() function pointer. 1703 * Return number of display cells for char at ScreenLines[off]. 1704 * We make sure that the offset used is less than "max_off". 1705 */ 1706 int 1707 latin_off2cells(unsigned off UNUSED, unsigned max_off UNUSED) 1708 { 1709 return 1; 1710 } 1711 1712 int 1713 dbcs_off2cells(unsigned off, unsigned max_off) 1714 { 1715 /* never check beyond end of the line */ 1716 if (off >= max_off) 1717 return 1; 1718 1719 /* Number of cells is equal to number of bytes, except for euc-jp when 1720 * the first byte is 0x8e. */ 1721 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e) 1722 return 1; 1723 return MB_BYTE2LEN(ScreenLines[off]); 1724 } 1725 1726 int 1727 utf_off2cells(unsigned off, unsigned max_off) 1728 { 1729 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1; 1730 } 1731 1732 /* 1733 * mb_ptr2char() function pointer. 1734 * Convert a byte sequence into a character. 1735 */ 1736 int 1737 latin_ptr2char(char_u *p) 1738 { 1739 return *p; 1740 } 1741 1742 static int 1743 dbcs_ptr2char(char_u *p) 1744 { 1745 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL) 1746 return (p[0] << 8) + p[1]; 1747 return *p; 1748 } 1749 1750 /* 1751 * Convert a UTF-8 byte sequence to a wide character. 1752 * If the sequence is illegal or truncated by a NUL the first byte is 1753 * returned. 1754 * Does not include composing characters, of course. 1755 */ 1756 int 1757 utf_ptr2char(char_u *p) 1758 { 1759 int len; 1760 1761 if (p[0] < 0x80) /* be quick for ASCII */ 1762 return p[0]; 1763 1764 len = utf8len_tab_zero[p[0]]; 1765 if (len > 1 && (p[1] & 0xc0) == 0x80) 1766 { 1767 if (len == 2) 1768 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); 1769 if ((p[2] & 0xc0) == 0x80) 1770 { 1771 if (len == 3) 1772 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) 1773 + (p[2] & 0x3f); 1774 if ((p[3] & 0xc0) == 0x80) 1775 { 1776 if (len == 4) 1777 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) 1778 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); 1779 if ((p[4] & 0xc0) == 0x80) 1780 { 1781 if (len == 5) 1782 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) 1783 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) 1784 + (p[4] & 0x3f); 1785 if ((p[5] & 0xc0) == 0x80 && len == 6) 1786 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) 1787 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) 1788 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); 1789 } 1790 } 1791 } 1792 } 1793 /* Illegal value, just return the first byte */ 1794 return p[0]; 1795 } 1796 1797 /* 1798 * Convert a UTF-8 byte sequence to a wide character. 1799 * String is assumed to be terminated by NUL or after "n" bytes, whichever 1800 * comes first. 1801 * The function is safe in the sense that it never accesses memory beyond the 1802 * first "n" bytes of "s". 1803 * 1804 * On success, returns decoded codepoint, advances "s" to the beginning of 1805 * next character and decreases "n" accordingly. 1806 * 1807 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past 1808 * NUL byte. 1809 * 1810 * If byte sequence is illegal or incomplete, returns -1 and does not advance 1811 * "s". 1812 */ 1813 static int 1814 utf_safe_read_char_adv(char_u **s, size_t *n) 1815 { 1816 int c, k; 1817 1818 if (*n == 0) /* end of buffer */ 1819 return 0; 1820 1821 k = utf8len_tab_zero[**s]; 1822 1823 if (k == 1) 1824 { 1825 /* ASCII character or NUL */ 1826 (*n)--; 1827 return *(*s)++; 1828 } 1829 1830 if ((size_t)k <= *n) 1831 { 1832 /* We have a multibyte sequence and it isn't truncated by buffer 1833 * limits so utf_ptr2char() is safe to use. Or the first byte is 1834 * illegal (k=0), and it's also safe to use utf_ptr2char(). */ 1835 c = utf_ptr2char(*s); 1836 1837 /* On failure, utf_ptr2char() returns the first byte, so here we 1838 * check equality with the first byte. The only non-ASCII character 1839 * which equals the first byte of its own UTF-8 representation is 1840 * U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too. 1841 * It's safe even if n=1, else we would have k=2 > n. */ 1842 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83)) 1843 { 1844 /* byte sequence was successfully decoded */ 1845 *s += k; 1846 *n -= k; 1847 return c; 1848 } 1849 } 1850 1851 /* byte sequence is incomplete or illegal */ 1852 return -1; 1853 } 1854 1855 /* 1856 * Get character at **pp and advance *pp to the next character. 1857 * Note: composing characters are skipped! 1858 */ 1859 int 1860 mb_ptr2char_adv(char_u **pp) 1861 { 1862 int c; 1863 1864 c = (*mb_ptr2char)(*pp); 1865 *pp += (*mb_ptr2len)(*pp); 1866 return c; 1867 } 1868 1869 /* 1870 * Get character at **pp and advance *pp to the next character. 1871 * Note: composing characters are returned as separate characters. 1872 */ 1873 int 1874 mb_cptr2char_adv(char_u **pp) 1875 { 1876 int c; 1877 1878 c = (*mb_ptr2char)(*pp); 1879 if (enc_utf8) 1880 *pp += utf_ptr2len(*pp); 1881 else 1882 *pp += (*mb_ptr2len)(*pp); 1883 return c; 1884 } 1885 1886 #if defined(FEAT_ARABIC) || defined(PROTO) 1887 /* 1888 * Check whether we are dealing with Arabic combining characters. 1889 * Note: these are NOT really composing characters! 1890 */ 1891 int 1892 arabic_combine( 1893 int one, /* first character */ 1894 int two) /* character just after "one" */ 1895 { 1896 if (one == a_LAM) 1897 return arabic_maycombine(two); 1898 return FALSE; 1899 } 1900 1901 /* 1902 * Check whether we are dealing with a character that could be regarded as an 1903 * Arabic combining character, need to check the character before this. 1904 */ 1905 int 1906 arabic_maycombine(int two) 1907 { 1908 if (p_arshape && !p_tbidi) 1909 return (two == a_ALEF_MADDA 1910 || two == a_ALEF_HAMZA_ABOVE 1911 || two == a_ALEF_HAMZA_BELOW 1912 || two == a_ALEF); 1913 return FALSE; 1914 } 1915 1916 /* 1917 * Check if the character pointed to by "p2" is a composing character when it 1918 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which 1919 * behaves like a composing character. 1920 */ 1921 int 1922 utf_composinglike(char_u *p1, char_u *p2) 1923 { 1924 int c2; 1925 1926 c2 = utf_ptr2char(p2); 1927 if (utf_iscomposing(c2)) 1928 return TRUE; 1929 if (!arabic_maycombine(c2)) 1930 return FALSE; 1931 return arabic_combine(utf_ptr2char(p1), c2); 1932 } 1933 #endif 1934 1935 /* 1936 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1937 * composing characters. 1938 */ 1939 int 1940 utfc_ptr2char( 1941 char_u *p, 1942 int *pcc) /* return: composing chars, last one is 0 */ 1943 { 1944 int len; 1945 int c; 1946 int cc; 1947 int i = 0; 1948 1949 c = utf_ptr2char(p); 1950 len = utf_ptr2len(p); 1951 1952 /* Only accept a composing char when the first char isn't illegal. */ 1953 if ((len > 1 || *p < 0x80) 1954 && p[len] >= 0x80 1955 && UTF_COMPOSINGLIKE(p, p + len)) 1956 { 1957 cc = utf_ptr2char(p + len); 1958 for (;;) 1959 { 1960 pcc[i++] = cc; 1961 if (i == MAX_MCO) 1962 break; 1963 len += utf_ptr2len(p + len); 1964 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1965 break; 1966 } 1967 } 1968 1969 if (i < MAX_MCO) /* last composing char must be 0 */ 1970 pcc[i] = 0; 1971 1972 return c; 1973 } 1974 1975 /* 1976 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1977 * composing characters. Use no more than p[maxlen]. 1978 */ 1979 int 1980 utfc_ptr2char_len( 1981 char_u *p, 1982 int *pcc, /* return: composing chars, last one is 0 */ 1983 int maxlen) 1984 { 1985 int len; 1986 int c; 1987 int cc; 1988 int i = 0; 1989 1990 c = utf_ptr2char(p); 1991 len = utf_ptr2len_len(p, maxlen); 1992 /* Only accept a composing char when the first char isn't illegal. */ 1993 if ((len > 1 || *p < 0x80) 1994 && len < maxlen 1995 && p[len] >= 0x80 1996 && UTF_COMPOSINGLIKE(p, p + len)) 1997 { 1998 cc = utf_ptr2char(p + len); 1999 for (;;) 2000 { 2001 pcc[i++] = cc; 2002 if (i == MAX_MCO) 2003 break; 2004 len += utf_ptr2len_len(p + len, maxlen - len); 2005 if (len >= maxlen 2006 || p[len] < 0x80 2007 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 2008 break; 2009 } 2010 } 2011 2012 if (i < MAX_MCO) /* last composing char must be 0 */ 2013 pcc[i] = 0; 2014 2015 return c; 2016 } 2017 2018 /* 2019 * Convert the character at screen position "off" to a sequence of bytes. 2020 * Includes the composing characters. 2021 * "buf" must at least have the length MB_MAXBYTES + 1. 2022 * Only to be used when ScreenLinesUC[off] != 0. 2023 * Returns the produced number of bytes. 2024 */ 2025 int 2026 utfc_char2bytes(int off, char_u *buf) 2027 { 2028 int len; 2029 int i; 2030 2031 len = utf_char2bytes(ScreenLinesUC[off], buf); 2032 for (i = 0; i < Screen_mco; ++i) 2033 { 2034 if (ScreenLinesC[i][off] == 0) 2035 break; 2036 len += utf_char2bytes(ScreenLinesC[i][off], buf + len); 2037 } 2038 return len; 2039 } 2040 2041 /* 2042 * Get the length of a UTF-8 byte sequence, not including any following 2043 * composing characters. 2044 * Returns 0 for "". 2045 * Returns 1 for an illegal byte sequence. 2046 */ 2047 int 2048 utf_ptr2len(char_u *p) 2049 { 2050 int len; 2051 int i; 2052 2053 if (*p == NUL) 2054 return 0; 2055 len = utf8len_tab[*p]; 2056 for (i = 1; i < len; ++i) 2057 if ((p[i] & 0xc0) != 0x80) 2058 return 1; 2059 return len; 2060 } 2061 2062 /* 2063 * Return length of UTF-8 character, obtained from the first byte. 2064 * "b" must be between 0 and 255! 2065 * Returns 1 for an invalid first byte value. 2066 */ 2067 int 2068 utf_byte2len(int b) 2069 { 2070 return utf8len_tab[b]; 2071 } 2072 2073 /* 2074 * Get the length of UTF-8 byte sequence "p[size]". Does not include any 2075 * following composing characters. 2076 * Returns 1 for "". 2077 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.). 2078 * Returns number > "size" for an incomplete byte sequence. 2079 * Never returns zero. 2080 */ 2081 int 2082 utf_ptr2len_len(char_u *p, int size) 2083 { 2084 int len; 2085 int i; 2086 int m; 2087 2088 len = utf8len_tab[*p]; 2089 if (len == 1) 2090 return 1; /* NUL, ascii or illegal lead byte */ 2091 if (len > size) 2092 m = size; /* incomplete byte sequence. */ 2093 else 2094 m = len; 2095 for (i = 1; i < m; ++i) 2096 if ((p[i] & 0xc0) != 0x80) 2097 return 1; 2098 return len; 2099 } 2100 2101 /* 2102 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. 2103 * This includes following composing characters. 2104 */ 2105 int 2106 utfc_ptr2len(char_u *p) 2107 { 2108 int len; 2109 int b0 = *p; 2110 #ifdef FEAT_ARABIC 2111 int prevlen; 2112 #endif 2113 2114 if (b0 == NUL) 2115 return 0; 2116 if (b0 < 0x80 && p[1] < 0x80) /* be quick for ASCII */ 2117 return 1; 2118 2119 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2120 len = utf_ptr2len(p); 2121 2122 /* Check for illegal byte. */ 2123 if (len == 1 && b0 >= 0x80) 2124 return 1; 2125 2126 /* 2127 * Check for composing characters. We can handle only the first six, but 2128 * skip all of them (otherwise the cursor would get stuck). 2129 */ 2130 #ifdef FEAT_ARABIC 2131 prevlen = 0; 2132 #endif 2133 for (;;) 2134 { 2135 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2136 return len; 2137 2138 /* Skip over composing char */ 2139 #ifdef FEAT_ARABIC 2140 prevlen = len; 2141 #endif 2142 len += utf_ptr2len(p + len); 2143 } 2144 } 2145 2146 /* 2147 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" 2148 * takes. This includes following composing characters. 2149 * Returns 0 for an empty string. 2150 * Returns 1 for an illegal char or an incomplete byte sequence. 2151 */ 2152 int 2153 utfc_ptr2len_len(char_u *p, int size) 2154 { 2155 int len; 2156 #ifdef FEAT_ARABIC 2157 int prevlen; 2158 #endif 2159 2160 if (size < 1 || *p == NUL) 2161 return 0; 2162 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) /* be quick for ASCII */ 2163 return 1; 2164 2165 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2166 len = utf_ptr2len_len(p, size); 2167 2168 /* Check for illegal byte and incomplete byte sequence. */ 2169 if ((len == 1 && p[0] >= 0x80) || len > size) 2170 return 1; 2171 2172 /* 2173 * Check for composing characters. We can handle only the first six, but 2174 * skip all of them (otherwise the cursor would get stuck). 2175 */ 2176 #ifdef FEAT_ARABIC 2177 prevlen = 0; 2178 #endif 2179 while (len < size) 2180 { 2181 int len_next_char; 2182 2183 if (p[len] < 0x80) 2184 break; 2185 2186 /* 2187 * Next character length should not go beyond size to ensure that 2188 * UTF_COMPOSINGLIKE(...) does not read beyond size. 2189 */ 2190 len_next_char = utf_ptr2len_len(p + len, size - len); 2191 if (len_next_char > size - len) 2192 break; 2193 2194 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2195 break; 2196 2197 /* Skip over composing char */ 2198 #ifdef FEAT_ARABIC 2199 prevlen = len; 2200 #endif 2201 len += len_next_char; 2202 } 2203 return len; 2204 } 2205 2206 /* 2207 * Return the number of bytes the UTF-8 encoding of character "c" takes. 2208 * This does not include composing characters. 2209 */ 2210 int 2211 utf_char2len(int c) 2212 { 2213 if (c < 0x80) 2214 return 1; 2215 if (c < 0x800) 2216 return 2; 2217 if (c < 0x10000) 2218 return 3; 2219 if (c < 0x200000) 2220 return 4; 2221 if (c < 0x4000000) 2222 return 5; 2223 return 6; 2224 } 2225 2226 /* 2227 * Convert Unicode character "c" to UTF-8 string in "buf[]". 2228 * Returns the number of bytes. 2229 * This does not include composing characters. 2230 */ 2231 int 2232 utf_char2bytes(int c, char_u *buf) 2233 { 2234 if (c < 0x80) /* 7 bits */ 2235 { 2236 buf[0] = c; 2237 return 1; 2238 } 2239 if (c < 0x800) /* 11 bits */ 2240 { 2241 buf[0] = 0xc0 + ((unsigned)c >> 6); 2242 buf[1] = 0x80 + (c & 0x3f); 2243 return 2; 2244 } 2245 if (c < 0x10000) /* 16 bits */ 2246 { 2247 buf[0] = 0xe0 + ((unsigned)c >> 12); 2248 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2249 buf[2] = 0x80 + (c & 0x3f); 2250 return 3; 2251 } 2252 if (c < 0x200000) /* 21 bits */ 2253 { 2254 buf[0] = 0xf0 + ((unsigned)c >> 18); 2255 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2256 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2257 buf[3] = 0x80 + (c & 0x3f); 2258 return 4; 2259 } 2260 if (c < 0x4000000) /* 26 bits */ 2261 { 2262 buf[0] = 0xf8 + ((unsigned)c >> 24); 2263 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2264 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2265 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2266 buf[4] = 0x80 + (c & 0x3f); 2267 return 5; 2268 } 2269 /* 31 bits */ 2270 buf[0] = 0xfc + ((unsigned)c >> 30); 2271 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f); 2272 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2273 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2274 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2275 buf[5] = 0x80 + (c & 0x3f); 2276 return 6; 2277 } 2278 2279 /* 2280 * Return TRUE if "c" is a composing UTF-8 character. This means it will be 2281 * drawn on top of the preceding character. 2282 * Based on code from Markus Kuhn. 2283 */ 2284 int 2285 utf_iscomposing(int c) 2286 { 2287 /* Sorted list of non-overlapping intervals. 2288 * Generated by ../runtime/tools/unicode.vim. */ 2289 static struct interval combining[] = 2290 { 2291 {0x0300, 0x036f}, 2292 {0x0483, 0x0489}, 2293 {0x0591, 0x05bd}, 2294 {0x05bf, 0x05bf}, 2295 {0x05c1, 0x05c2}, 2296 {0x05c4, 0x05c5}, 2297 {0x05c7, 0x05c7}, 2298 {0x0610, 0x061a}, 2299 {0x064b, 0x065f}, 2300 {0x0670, 0x0670}, 2301 {0x06d6, 0x06dc}, 2302 {0x06df, 0x06e4}, 2303 {0x06e7, 0x06e8}, 2304 {0x06ea, 0x06ed}, 2305 {0x0711, 0x0711}, 2306 {0x0730, 0x074a}, 2307 {0x07a6, 0x07b0}, 2308 {0x07eb, 0x07f3}, 2309 {0x0816, 0x0819}, 2310 {0x081b, 0x0823}, 2311 {0x0825, 0x0827}, 2312 {0x0829, 0x082d}, 2313 {0x0859, 0x085b}, 2314 {0x08d4, 0x08e1}, 2315 {0x08e3, 0x0903}, 2316 {0x093a, 0x093c}, 2317 {0x093e, 0x094f}, 2318 {0x0951, 0x0957}, 2319 {0x0962, 0x0963}, 2320 {0x0981, 0x0983}, 2321 {0x09bc, 0x09bc}, 2322 {0x09be, 0x09c4}, 2323 {0x09c7, 0x09c8}, 2324 {0x09cb, 0x09cd}, 2325 {0x09d7, 0x09d7}, 2326 {0x09e2, 0x09e3}, 2327 {0x0a01, 0x0a03}, 2328 {0x0a3c, 0x0a3c}, 2329 {0x0a3e, 0x0a42}, 2330 {0x0a47, 0x0a48}, 2331 {0x0a4b, 0x0a4d}, 2332 {0x0a51, 0x0a51}, 2333 {0x0a70, 0x0a71}, 2334 {0x0a75, 0x0a75}, 2335 {0x0a81, 0x0a83}, 2336 {0x0abc, 0x0abc}, 2337 {0x0abe, 0x0ac5}, 2338 {0x0ac7, 0x0ac9}, 2339 {0x0acb, 0x0acd}, 2340 {0x0ae2, 0x0ae3}, 2341 {0x0b01, 0x0b03}, 2342 {0x0b3c, 0x0b3c}, 2343 {0x0b3e, 0x0b44}, 2344 {0x0b47, 0x0b48}, 2345 {0x0b4b, 0x0b4d}, 2346 {0x0b56, 0x0b57}, 2347 {0x0b62, 0x0b63}, 2348 {0x0b82, 0x0b82}, 2349 {0x0bbe, 0x0bc2}, 2350 {0x0bc6, 0x0bc8}, 2351 {0x0bca, 0x0bcd}, 2352 {0x0bd7, 0x0bd7}, 2353 {0x0c00, 0x0c03}, 2354 {0x0c3e, 0x0c44}, 2355 {0x0c46, 0x0c48}, 2356 {0x0c4a, 0x0c4d}, 2357 {0x0c55, 0x0c56}, 2358 {0x0c62, 0x0c63}, 2359 {0x0c81, 0x0c83}, 2360 {0x0cbc, 0x0cbc}, 2361 {0x0cbe, 0x0cc4}, 2362 {0x0cc6, 0x0cc8}, 2363 {0x0cca, 0x0ccd}, 2364 {0x0cd5, 0x0cd6}, 2365 {0x0ce2, 0x0ce3}, 2366 {0x0d01, 0x0d03}, 2367 {0x0d3e, 0x0d44}, 2368 {0x0d46, 0x0d48}, 2369 {0x0d4a, 0x0d4d}, 2370 {0x0d57, 0x0d57}, 2371 {0x0d62, 0x0d63}, 2372 {0x0d82, 0x0d83}, 2373 {0x0dca, 0x0dca}, 2374 {0x0dcf, 0x0dd4}, 2375 {0x0dd6, 0x0dd6}, 2376 {0x0dd8, 0x0ddf}, 2377 {0x0df2, 0x0df3}, 2378 {0x0e31, 0x0e31}, 2379 {0x0e34, 0x0e3a}, 2380 {0x0e47, 0x0e4e}, 2381 {0x0eb1, 0x0eb1}, 2382 {0x0eb4, 0x0eb9}, 2383 {0x0ebb, 0x0ebc}, 2384 {0x0ec8, 0x0ecd}, 2385 {0x0f18, 0x0f19}, 2386 {0x0f35, 0x0f35}, 2387 {0x0f37, 0x0f37}, 2388 {0x0f39, 0x0f39}, 2389 {0x0f3e, 0x0f3f}, 2390 {0x0f71, 0x0f84}, 2391 {0x0f86, 0x0f87}, 2392 {0x0f8d, 0x0f97}, 2393 {0x0f99, 0x0fbc}, 2394 {0x0fc6, 0x0fc6}, 2395 {0x102b, 0x103e}, 2396 {0x1056, 0x1059}, 2397 {0x105e, 0x1060}, 2398 {0x1062, 0x1064}, 2399 {0x1067, 0x106d}, 2400 {0x1071, 0x1074}, 2401 {0x1082, 0x108d}, 2402 {0x108f, 0x108f}, 2403 {0x109a, 0x109d}, 2404 {0x135d, 0x135f}, 2405 {0x1712, 0x1714}, 2406 {0x1732, 0x1734}, 2407 {0x1752, 0x1753}, 2408 {0x1772, 0x1773}, 2409 {0x17b4, 0x17d3}, 2410 {0x17dd, 0x17dd}, 2411 {0x180b, 0x180d}, 2412 {0x1885, 0x1886}, 2413 {0x18a9, 0x18a9}, 2414 {0x1920, 0x192b}, 2415 {0x1930, 0x193b}, 2416 {0x1a17, 0x1a1b}, 2417 {0x1a55, 0x1a5e}, 2418 {0x1a60, 0x1a7c}, 2419 {0x1a7f, 0x1a7f}, 2420 {0x1ab0, 0x1abe}, 2421 {0x1b00, 0x1b04}, 2422 {0x1b34, 0x1b44}, 2423 {0x1b6b, 0x1b73}, 2424 {0x1b80, 0x1b82}, 2425 {0x1ba1, 0x1bad}, 2426 {0x1be6, 0x1bf3}, 2427 {0x1c24, 0x1c37}, 2428 {0x1cd0, 0x1cd2}, 2429 {0x1cd4, 0x1ce8}, 2430 {0x1ced, 0x1ced}, 2431 {0x1cf2, 0x1cf4}, 2432 {0x1cf8, 0x1cf9}, 2433 {0x1dc0, 0x1df5}, 2434 {0x1dfb, 0x1dff}, 2435 {0x20d0, 0x20f0}, 2436 {0x2cef, 0x2cf1}, 2437 {0x2d7f, 0x2d7f}, 2438 {0x2de0, 0x2dff}, 2439 {0x302a, 0x302f}, 2440 {0x3099, 0x309a}, 2441 {0xa66f, 0xa672}, 2442 {0xa674, 0xa67d}, 2443 {0xa69e, 0xa69f}, 2444 {0xa6f0, 0xa6f1}, 2445 {0xa802, 0xa802}, 2446 {0xa806, 0xa806}, 2447 {0xa80b, 0xa80b}, 2448 {0xa823, 0xa827}, 2449 {0xa880, 0xa881}, 2450 {0xa8b4, 0xa8c5}, 2451 {0xa8e0, 0xa8f1}, 2452 {0xa926, 0xa92d}, 2453 {0xa947, 0xa953}, 2454 {0xa980, 0xa983}, 2455 {0xa9b3, 0xa9c0}, 2456 {0xa9e5, 0xa9e5}, 2457 {0xaa29, 0xaa36}, 2458 {0xaa43, 0xaa43}, 2459 {0xaa4c, 0xaa4d}, 2460 {0xaa7b, 0xaa7d}, 2461 {0xaab0, 0xaab0}, 2462 {0xaab2, 0xaab4}, 2463 {0xaab7, 0xaab8}, 2464 {0xaabe, 0xaabf}, 2465 {0xaac1, 0xaac1}, 2466 {0xaaeb, 0xaaef}, 2467 {0xaaf5, 0xaaf6}, 2468 {0xabe3, 0xabea}, 2469 {0xabec, 0xabed}, 2470 {0xfb1e, 0xfb1e}, 2471 {0xfe00, 0xfe0f}, 2472 {0xfe20, 0xfe2f}, 2473 {0x101fd, 0x101fd}, 2474 {0x102e0, 0x102e0}, 2475 {0x10376, 0x1037a}, 2476 {0x10a01, 0x10a03}, 2477 {0x10a05, 0x10a06}, 2478 {0x10a0c, 0x10a0f}, 2479 {0x10a38, 0x10a3a}, 2480 {0x10a3f, 0x10a3f}, 2481 {0x10ae5, 0x10ae6}, 2482 {0x11000, 0x11002}, 2483 {0x11038, 0x11046}, 2484 {0x1107f, 0x11082}, 2485 {0x110b0, 0x110ba}, 2486 {0x11100, 0x11102}, 2487 {0x11127, 0x11134}, 2488 {0x11173, 0x11173}, 2489 {0x11180, 0x11182}, 2490 {0x111b3, 0x111c0}, 2491 {0x111ca, 0x111cc}, 2492 {0x1122c, 0x11237}, 2493 {0x1123e, 0x1123e}, 2494 {0x112df, 0x112ea}, 2495 {0x11300, 0x11303}, 2496 {0x1133c, 0x1133c}, 2497 {0x1133e, 0x11344}, 2498 {0x11347, 0x11348}, 2499 {0x1134b, 0x1134d}, 2500 {0x11357, 0x11357}, 2501 {0x11362, 0x11363}, 2502 {0x11366, 0x1136c}, 2503 {0x11370, 0x11374}, 2504 {0x11435, 0x11446}, 2505 {0x114b0, 0x114c3}, 2506 {0x115af, 0x115b5}, 2507 {0x115b8, 0x115c0}, 2508 {0x115dc, 0x115dd}, 2509 {0x11630, 0x11640}, 2510 {0x116ab, 0x116b7}, 2511 {0x1171d, 0x1172b}, 2512 {0x11c2f, 0x11c36}, 2513 {0x11c38, 0x11c3f}, 2514 {0x11c92, 0x11ca7}, 2515 {0x11ca9, 0x11cb6}, 2516 {0x16af0, 0x16af4}, 2517 {0x16b30, 0x16b36}, 2518 {0x16f51, 0x16f7e}, 2519 {0x16f8f, 0x16f92}, 2520 {0x1bc9d, 0x1bc9e}, 2521 {0x1d165, 0x1d169}, 2522 {0x1d16d, 0x1d172}, 2523 {0x1d17b, 0x1d182}, 2524 {0x1d185, 0x1d18b}, 2525 {0x1d1aa, 0x1d1ad}, 2526 {0x1d242, 0x1d244}, 2527 {0x1da00, 0x1da36}, 2528 {0x1da3b, 0x1da6c}, 2529 {0x1da75, 0x1da75}, 2530 {0x1da84, 0x1da84}, 2531 {0x1da9b, 0x1da9f}, 2532 {0x1daa1, 0x1daaf}, 2533 {0x1e000, 0x1e006}, 2534 {0x1e008, 0x1e018}, 2535 {0x1e01b, 0x1e021}, 2536 {0x1e023, 0x1e024}, 2537 {0x1e026, 0x1e02a}, 2538 {0x1e8d0, 0x1e8d6}, 2539 {0x1e944, 0x1e94a}, 2540 {0xe0100, 0xe01ef} 2541 }; 2542 2543 return intable(combining, sizeof(combining), c); 2544 } 2545 2546 /* 2547 * Return TRUE for characters that can be displayed in a normal way. 2548 * Only for characters of 0x100 and above! 2549 */ 2550 int 2551 utf_printable(int c) 2552 { 2553 #ifdef USE_WCHAR_FUNCTIONS 2554 /* 2555 * Assume the iswprint() library function works better than our own stuff. 2556 */ 2557 return iswprint(c); 2558 #else 2559 /* Sorted list of non-overlapping intervals. 2560 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */ 2561 static struct interval nonprint[] = 2562 { 2563 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2564 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2565 {0xfffe, 0xffff} 2566 }; 2567 2568 return !intable(nonprint, sizeof(nonprint), c); 2569 #endif 2570 } 2571 2572 /* Sorted list of non-overlapping intervals of all Emoji characters, 2573 * based on http://unicode.org/emoji/charts/emoji-list.html */ 2574 static struct interval emoji_all[] = 2575 { 2576 {0x203c, 0x203c}, 2577 {0x2049, 0x2049}, 2578 {0x2122, 0x2122}, 2579 {0x2139, 0x2139}, 2580 {0x2194, 0x2199}, 2581 {0x21a9, 0x21aa}, 2582 {0x231a, 0x231b}, 2583 {0x2328, 0x2328}, 2584 {0x23cf, 0x23cf}, 2585 {0x23e9, 0x23f3}, 2586 {0x24c2, 0x24c2}, 2587 {0x25aa, 0x25ab}, 2588 {0x25b6, 0x25b6}, 2589 {0x25c0, 0x25c0}, 2590 {0x25fb, 0x25fe}, 2591 {0x2600, 0x2604}, 2592 {0x260e, 0x260e}, 2593 {0x2611, 0x2611}, 2594 {0x2614, 0x2615}, 2595 {0x2618, 0x2618}, 2596 {0x261d, 0x261d}, 2597 {0x2620, 0x2620}, 2598 {0x2622, 0x2623}, 2599 {0x2626, 0x2626}, 2600 {0x262a, 0x262a}, 2601 {0x262e, 0x262f}, 2602 {0x2638, 0x263a}, 2603 {0x2648, 0x2653}, 2604 {0x2660, 0x2660}, 2605 {0x2663, 0x2663}, 2606 {0x2665, 0x2666}, 2607 {0x2668, 0x2668}, 2608 {0x267b, 0x267b}, 2609 {0x267f, 0x267f}, 2610 {0x2692, 0x2694}, 2611 {0x2696, 0x2697}, 2612 {0x2699, 0x2699}, 2613 {0x269b, 0x269c}, 2614 {0x26a0, 0x26a1}, 2615 {0x26aa, 0x26ab}, 2616 {0x26b0, 0x26b1}, 2617 {0x26bd, 0x26be}, 2618 {0x26c4, 0x26c5}, 2619 {0x26c8, 0x26c8}, 2620 {0x26ce, 0x26cf}, 2621 {0x26d1, 0x26d1}, 2622 {0x26d3, 0x26d4}, 2623 {0x26e9, 0x26ea}, 2624 {0x26f0, 0x26f5}, 2625 {0x26f7, 0x26fa}, 2626 {0x26fd, 0x26fd}, 2627 {0x2702, 0x2702}, 2628 {0x2705, 0x2705}, 2629 {0x2708, 0x270d}, 2630 {0x270f, 0x270f}, 2631 {0x2712, 0x2712}, 2632 {0x2714, 0x2714}, 2633 {0x2716, 0x2716}, 2634 {0x271d, 0x271d}, 2635 {0x2721, 0x2721}, 2636 {0x2728, 0x2728}, 2637 {0x2733, 0x2734}, 2638 {0x2744, 0x2744}, 2639 {0x2747, 0x2747}, 2640 {0x274c, 0x274c}, 2641 {0x274e, 0x274e}, 2642 {0x2753, 0x2755}, 2643 {0x2757, 0x2757}, 2644 {0x2763, 0x2764}, 2645 {0x2795, 0x2797}, 2646 {0x27a1, 0x27a1}, 2647 {0x27b0, 0x27b0}, 2648 {0x27bf, 0x27bf}, 2649 {0x2934, 0x2935}, 2650 {0x2b05, 0x2b07}, 2651 {0x2b1b, 0x2b1c}, 2652 {0x2b50, 0x2b50}, 2653 {0x2b55, 0x2b55}, 2654 {0x3030, 0x3030}, 2655 {0x303d, 0x303d}, 2656 {0x3297, 0x3297}, 2657 {0x3299, 0x3299}, 2658 {0x1f004, 0x1f004}, 2659 {0x1f0cf, 0x1f0cf}, 2660 {0x1f170, 0x1f171}, 2661 {0x1f17e, 0x1f17f}, 2662 {0x1f18e, 0x1f18e}, 2663 {0x1f191, 0x1f19a}, 2664 {0x1f1e6, 0x1f1ff}, 2665 {0x1f201, 0x1f202}, 2666 {0x1f21a, 0x1f21a}, 2667 {0x1f22f, 0x1f22f}, 2668 {0x1f232, 0x1f23a}, 2669 {0x1f250, 0x1f251}, 2670 {0x1f300, 0x1f320}, 2671 {0x1f330, 0x1f335}, 2672 {0x1f337, 0x1f37c}, 2673 {0x1f380, 0x1f393}, 2674 {0x1f3a0, 0x1f3c4}, 2675 {0x1f3c6, 0x1f3ca}, 2676 {0x1f3e0, 0x1f3f0}, 2677 {0x1f400, 0x1f43e}, 2678 {0x1f440, 0x1f440}, 2679 {0x1f442, 0x1f4f7}, 2680 {0x1f4f9, 0x1f4fc}, 2681 {0x1f500, 0x1f53d}, 2682 {0x1f550, 0x1f567}, 2683 {0x1f5fb, 0x1f640}, 2684 {0x1f645, 0x1f64f}, 2685 {0x1f680, 0x1f6c5} 2686 }; 2687 2688 /* 2689 * Get class of a Unicode character. 2690 * 0: white space 2691 * 1: punctuation 2692 * 2 or bigger: some class of word character. 2693 */ 2694 int 2695 utf_class(int c) 2696 { 2697 return utf_class_buf(c, curbuf); 2698 } 2699 2700 int 2701 utf_class_buf(int c, buf_T *buf) 2702 { 2703 /* sorted list of non-overlapping intervals */ 2704 static struct clinterval 2705 { 2706 unsigned int first; 2707 unsigned int last; 2708 unsigned int class; 2709 } classes[] = 2710 { 2711 {0x037e, 0x037e, 1}, /* Greek question mark */ 2712 {0x0387, 0x0387, 1}, /* Greek ano teleia */ 2713 {0x055a, 0x055f, 1}, /* Armenian punctuation */ 2714 {0x0589, 0x0589, 1}, /* Armenian full stop */ 2715 {0x05be, 0x05be, 1}, 2716 {0x05c0, 0x05c0, 1}, 2717 {0x05c3, 0x05c3, 1}, 2718 {0x05f3, 0x05f4, 1}, 2719 {0x060c, 0x060c, 1}, 2720 {0x061b, 0x061b, 1}, 2721 {0x061f, 0x061f, 1}, 2722 {0x066a, 0x066d, 1}, 2723 {0x06d4, 0x06d4, 1}, 2724 {0x0700, 0x070d, 1}, /* Syriac punctuation */ 2725 {0x0964, 0x0965, 1}, 2726 {0x0970, 0x0970, 1}, 2727 {0x0df4, 0x0df4, 1}, 2728 {0x0e4f, 0x0e4f, 1}, 2729 {0x0e5a, 0x0e5b, 1}, 2730 {0x0f04, 0x0f12, 1}, 2731 {0x0f3a, 0x0f3d, 1}, 2732 {0x0f85, 0x0f85, 1}, 2733 {0x104a, 0x104f, 1}, /* Myanmar punctuation */ 2734 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */ 2735 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */ 2736 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */ 2737 {0x1680, 0x1680, 0}, 2738 {0x169b, 0x169c, 1}, 2739 {0x16eb, 0x16ed, 1}, 2740 {0x1735, 0x1736, 1}, 2741 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */ 2742 {0x1800, 0x180a, 1}, /* Mongolian punctuation */ 2743 {0x2000, 0x200b, 0}, /* spaces */ 2744 {0x200c, 0x2027, 1}, /* punctuation and symbols */ 2745 {0x2028, 0x2029, 0}, 2746 {0x202a, 0x202e, 1}, /* punctuation and symbols */ 2747 {0x202f, 0x202f, 0}, 2748 {0x2030, 0x205e, 1}, /* punctuation and symbols */ 2749 {0x205f, 0x205f, 0}, 2750 {0x2060, 0x27ff, 1}, /* punctuation and symbols */ 2751 {0x2070, 0x207f, 0x2070}, /* superscript */ 2752 {0x2080, 0x2094, 0x2080}, /* subscript */ 2753 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */ 2754 {0x2800, 0x28ff, 0x2800}, /* braille */ 2755 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */ 2756 {0x29d8, 0x29db, 1}, 2757 {0x29fc, 0x29fd, 1}, 2758 {0x2e00, 0x2e7f, 1}, /* supplemental punctuation */ 2759 {0x3000, 0x3000, 0}, /* ideographic space */ 2760 {0x3001, 0x3020, 1}, /* ideographic punctuation */ 2761 {0x3030, 0x3030, 1}, 2762 {0x303d, 0x303d, 1}, 2763 {0x3040, 0x309f, 0x3040}, /* Hiragana */ 2764 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */ 2765 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */ 2766 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */ 2767 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */ 2768 {0xfd3e, 0xfd3f, 1}, 2769 {0xfe30, 0xfe6b, 1}, /* punctuation forms */ 2770 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */ 2771 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */ 2772 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */ 2773 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */ 2774 {0x20000, 0x2a6df, 0x4e00}, /* CJK Ideographs */ 2775 {0x2a700, 0x2b73f, 0x4e00}, /* CJK Ideographs */ 2776 {0x2b740, 0x2b81f, 0x4e00}, /* CJK Ideographs */ 2777 {0x2f800, 0x2fa1f, 0x4e00}, /* CJK Ideographs */ 2778 }; 2779 2780 int bot = 0; 2781 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2782 int mid; 2783 2784 /* First quick check for Latin1 characters, use 'iskeyword'. */ 2785 if (c < 0x100) 2786 { 2787 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2788 return 0; /* blank */ 2789 if (vim_iswordc_buf(c, buf)) 2790 return 2; /* word character */ 2791 return 1; /* punctuation */ 2792 } 2793 2794 /* binary search in table */ 2795 while (top >= bot) 2796 { 2797 mid = (bot + top) / 2; 2798 if (classes[mid].last < (unsigned int)c) 2799 bot = mid + 1; 2800 else if (classes[mid].first > (unsigned int)c) 2801 top = mid - 1; 2802 else 2803 return (int)classes[mid].class; 2804 } 2805 2806 /* emoji */ 2807 if (intable(emoji_all, sizeof(emoji_all), c)) 2808 return 3; 2809 2810 /* most other characters are "word" characters */ 2811 return 2; 2812 } 2813 2814 int 2815 utf_ambiguous_width(int c) 2816 { 2817 return c >= 0x80 && (intable(ambiguous, sizeof(ambiguous), c) 2818 || intable(emoji_all, sizeof(emoji_all), c)); 2819 } 2820 2821 /* 2822 * Code for Unicode case-dependent operations. Based on notes in 2823 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2824 * This code uses simple case folding, not full case folding. 2825 * Last updated for Unicode 5.2. 2826 */ 2827 2828 /* 2829 * The following tables are built by ../runtime/tools/unicode.vim. 2830 * They must be in numeric order, because we use binary search. 2831 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2832 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2833 * folded/upper/lower by adding 32. 2834 */ 2835 typedef struct 2836 { 2837 int rangeStart; 2838 int rangeEnd; 2839 int step; 2840 int offset; 2841 } convertStruct; 2842 2843 static convertStruct foldCase[] = 2844 { 2845 {0x41,0x5a,1,32}, 2846 {0xb5,0xb5,-1,775}, 2847 {0xc0,0xd6,1,32}, 2848 {0xd8,0xde,1,32}, 2849 {0x100,0x12e,2,1}, 2850 {0x132,0x136,2,1}, 2851 {0x139,0x147,2,1}, 2852 {0x14a,0x176,2,1}, 2853 {0x178,0x178,-1,-121}, 2854 {0x179,0x17d,2,1}, 2855 {0x17f,0x17f,-1,-268}, 2856 {0x181,0x181,-1,210}, 2857 {0x182,0x184,2,1}, 2858 {0x186,0x186,-1,206}, 2859 {0x187,0x187,-1,1}, 2860 {0x189,0x18a,1,205}, 2861 {0x18b,0x18b,-1,1}, 2862 {0x18e,0x18e,-1,79}, 2863 {0x18f,0x18f,-1,202}, 2864 {0x190,0x190,-1,203}, 2865 {0x191,0x191,-1,1}, 2866 {0x193,0x193,-1,205}, 2867 {0x194,0x194,-1,207}, 2868 {0x196,0x196,-1,211}, 2869 {0x197,0x197,-1,209}, 2870 {0x198,0x198,-1,1}, 2871 {0x19c,0x19c,-1,211}, 2872 {0x19d,0x19d,-1,213}, 2873 {0x19f,0x19f,-1,214}, 2874 {0x1a0,0x1a4,2,1}, 2875 {0x1a6,0x1a6,-1,218}, 2876 {0x1a7,0x1a7,-1,1}, 2877 {0x1a9,0x1a9,-1,218}, 2878 {0x1ac,0x1ac,-1,1}, 2879 {0x1ae,0x1ae,-1,218}, 2880 {0x1af,0x1af,-1,1}, 2881 {0x1b1,0x1b2,1,217}, 2882 {0x1b3,0x1b5,2,1}, 2883 {0x1b7,0x1b7,-1,219}, 2884 {0x1b8,0x1bc,4,1}, 2885 {0x1c4,0x1c4,-1,2}, 2886 {0x1c5,0x1c5,-1,1}, 2887 {0x1c7,0x1c7,-1,2}, 2888 {0x1c8,0x1c8,-1,1}, 2889 {0x1ca,0x1ca,-1,2}, 2890 {0x1cb,0x1db,2,1}, 2891 {0x1de,0x1ee,2,1}, 2892 {0x1f1,0x1f1,-1,2}, 2893 {0x1f2,0x1f4,2,1}, 2894 {0x1f6,0x1f6,-1,-97}, 2895 {0x1f7,0x1f7,-1,-56}, 2896 {0x1f8,0x21e,2,1}, 2897 {0x220,0x220,-1,-130}, 2898 {0x222,0x232,2,1}, 2899 {0x23a,0x23a,-1,10795}, 2900 {0x23b,0x23b,-1,1}, 2901 {0x23d,0x23d,-1,-163}, 2902 {0x23e,0x23e,-1,10792}, 2903 {0x241,0x241,-1,1}, 2904 {0x243,0x243,-1,-195}, 2905 {0x244,0x244,-1,69}, 2906 {0x245,0x245,-1,71}, 2907 {0x246,0x24e,2,1}, 2908 {0x345,0x345,-1,116}, 2909 {0x370,0x372,2,1}, 2910 {0x376,0x376,-1,1}, 2911 {0x37f,0x37f,-1,116}, 2912 {0x386,0x386,-1,38}, 2913 {0x388,0x38a,1,37}, 2914 {0x38c,0x38c,-1,64}, 2915 {0x38e,0x38f,1,63}, 2916 {0x391,0x3a1,1,32}, 2917 {0x3a3,0x3ab,1,32}, 2918 {0x3c2,0x3c2,-1,1}, 2919 {0x3cf,0x3cf,-1,8}, 2920 {0x3d0,0x3d0,-1,-30}, 2921 {0x3d1,0x3d1,-1,-25}, 2922 {0x3d5,0x3d5,-1,-15}, 2923 {0x3d6,0x3d6,-1,-22}, 2924 {0x3d8,0x3ee,2,1}, 2925 {0x3f0,0x3f0,-1,-54}, 2926 {0x3f1,0x3f1,-1,-48}, 2927 {0x3f4,0x3f4,-1,-60}, 2928 {0x3f5,0x3f5,-1,-64}, 2929 {0x3f7,0x3f7,-1,1}, 2930 {0x3f9,0x3f9,-1,-7}, 2931 {0x3fa,0x3fa,-1,1}, 2932 {0x3fd,0x3ff,1,-130}, 2933 {0x400,0x40f,1,80}, 2934 {0x410,0x42f,1,32}, 2935 {0x460,0x480,2,1}, 2936 {0x48a,0x4be,2,1}, 2937 {0x4c0,0x4c0,-1,15}, 2938 {0x4c1,0x4cd,2,1}, 2939 {0x4d0,0x52e,2,1}, 2940 {0x531,0x556,1,48}, 2941 {0x10a0,0x10c5,1,7264}, 2942 {0x10c7,0x10cd,6,7264}, 2943 {0x13f8,0x13fd,1,-8}, 2944 {0x1c80,0x1c80,-1,-6222}, 2945 {0x1c81,0x1c81,-1,-6221}, 2946 {0x1c82,0x1c82,-1,-6212}, 2947 {0x1c83,0x1c84,1,-6210}, 2948 {0x1c85,0x1c85,-1,-6211}, 2949 {0x1c86,0x1c86,-1,-6204}, 2950 {0x1c87,0x1c87,-1,-6180}, 2951 {0x1c88,0x1c88,-1,35267}, 2952 {0x1e00,0x1e94,2,1}, 2953 {0x1e9b,0x1e9b,-1,-58}, 2954 {0x1e9e,0x1e9e,-1,-7615}, 2955 {0x1ea0,0x1efe,2,1}, 2956 {0x1f08,0x1f0f,1,-8}, 2957 {0x1f18,0x1f1d,1,-8}, 2958 {0x1f28,0x1f2f,1,-8}, 2959 {0x1f38,0x1f3f,1,-8}, 2960 {0x1f48,0x1f4d,1,-8}, 2961 {0x1f59,0x1f5f,2,-8}, 2962 {0x1f68,0x1f6f,1,-8}, 2963 {0x1f88,0x1f8f,1,-8}, 2964 {0x1f98,0x1f9f,1,-8}, 2965 {0x1fa8,0x1faf,1,-8}, 2966 {0x1fb8,0x1fb9,1,-8}, 2967 {0x1fba,0x1fbb,1,-74}, 2968 {0x1fbc,0x1fbc,-1,-9}, 2969 {0x1fbe,0x1fbe,-1,-7173}, 2970 {0x1fc8,0x1fcb,1,-86}, 2971 {0x1fcc,0x1fcc,-1,-9}, 2972 {0x1fd8,0x1fd9,1,-8}, 2973 {0x1fda,0x1fdb,1,-100}, 2974 {0x1fe8,0x1fe9,1,-8}, 2975 {0x1fea,0x1feb,1,-112}, 2976 {0x1fec,0x1fec,-1,-7}, 2977 {0x1ff8,0x1ff9,1,-128}, 2978 {0x1ffa,0x1ffb,1,-126}, 2979 {0x1ffc,0x1ffc,-1,-9}, 2980 {0x2126,0x2126,-1,-7517}, 2981 {0x212a,0x212a,-1,-8383}, 2982 {0x212b,0x212b,-1,-8262}, 2983 {0x2132,0x2132,-1,28}, 2984 {0x2160,0x216f,1,16}, 2985 {0x2183,0x2183,-1,1}, 2986 {0x24b6,0x24cf,1,26}, 2987 {0x2c00,0x2c2e,1,48}, 2988 {0x2c60,0x2c60,-1,1}, 2989 {0x2c62,0x2c62,-1,-10743}, 2990 {0x2c63,0x2c63,-1,-3814}, 2991 {0x2c64,0x2c64,-1,-10727}, 2992 {0x2c67,0x2c6b,2,1}, 2993 {0x2c6d,0x2c6d,-1,-10780}, 2994 {0x2c6e,0x2c6e,-1,-10749}, 2995 {0x2c6f,0x2c6f,-1,-10783}, 2996 {0x2c70,0x2c70,-1,-10782}, 2997 {0x2c72,0x2c75,3,1}, 2998 {0x2c7e,0x2c7f,1,-10815}, 2999 {0x2c80,0x2ce2,2,1}, 3000 {0x2ceb,0x2ced,2,1}, 3001 {0x2cf2,0xa640,31054,1}, 3002 {0xa642,0xa66c,2,1}, 3003 {0xa680,0xa69a,2,1}, 3004 {0xa722,0xa72e,2,1}, 3005 {0xa732,0xa76e,2,1}, 3006 {0xa779,0xa77b,2,1}, 3007 {0xa77d,0xa77d,-1,-35332}, 3008 {0xa77e,0xa786,2,1}, 3009 {0xa78b,0xa78b,-1,1}, 3010 {0xa78d,0xa78d,-1,-42280}, 3011 {0xa790,0xa792,2,1}, 3012 {0xa796,0xa7a8,2,1}, 3013 {0xa7aa,0xa7aa,-1,-42308}, 3014 {0xa7ab,0xa7ab,-1,-42319}, 3015 {0xa7ac,0xa7ac,-1,-42315}, 3016 {0xa7ad,0xa7ad,-1,-42305}, 3017 {0xa7ae,0xa7ae,-1,-42308}, 3018 {0xa7b0,0xa7b0,-1,-42258}, 3019 {0xa7b1,0xa7b1,-1,-42282}, 3020 {0xa7b2,0xa7b2,-1,-42261}, 3021 {0xa7b3,0xa7b3,-1,928}, 3022 {0xa7b4,0xa7b6,2,1}, 3023 {0xab70,0xabbf,1,-38864}, 3024 {0xff21,0xff3a,1,32}, 3025 {0x10400,0x10427,1,40}, 3026 {0x104b0,0x104d3,1,40}, 3027 {0x10c80,0x10cb2,1,64}, 3028 {0x118a0,0x118bf,1,32}, 3029 {0x1e900,0x1e921,1,34} 3030 }; 3031 3032 static int utf_convert(int a, convertStruct table[], int tableSize); 3033 static int utf_strnicmp(char_u *s1, char_u *s2, size_t n1, size_t n2); 3034 3035 /* 3036 * Generic conversion function for case operations. 3037 * Return the converted equivalent of "a", which is a UCS-4 character. Use 3038 * the given conversion "table". Uses binary search on "table". 3039 */ 3040 static int 3041 utf_convert( 3042 int a, 3043 convertStruct table[], 3044 int tableSize) 3045 { 3046 int start, mid, end; /* indices into table */ 3047 int entries = tableSize / sizeof(convertStruct); 3048 3049 start = 0; 3050 end = entries; 3051 while (start < end) 3052 { 3053 /* need to search further */ 3054 mid = (end + start) / 2; 3055 if (table[mid].rangeEnd < a) 3056 start = mid + 1; 3057 else 3058 end = mid; 3059 } 3060 if (start < entries 3061 && table[start].rangeStart <= a 3062 && a <= table[start].rangeEnd 3063 && (a - table[start].rangeStart) % table[start].step == 0) 3064 return (a + table[start].offset); 3065 else 3066 return a; 3067 } 3068 3069 /* 3070 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 3071 * simple case folding. 3072 */ 3073 int 3074 utf_fold(int a) 3075 { 3076 if (a < 0x80) 3077 /* be fast for ASCII */ 3078 return a >= 0x41 && a <= 0x5a ? a + 32 : a; 3079 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 3080 } 3081 3082 static convertStruct toLower[] = 3083 { 3084 {0x41,0x5a,1,32}, 3085 {0xc0,0xd6,1,32}, 3086 {0xd8,0xde,1,32}, 3087 {0x100,0x12e,2,1}, 3088 {0x130,0x130,-1,-199}, 3089 {0x132,0x136,2,1}, 3090 {0x139,0x147,2,1}, 3091 {0x14a,0x176,2,1}, 3092 {0x178,0x178,-1,-121}, 3093 {0x179,0x17d,2,1}, 3094 {0x181,0x181,-1,210}, 3095 {0x182,0x184,2,1}, 3096 {0x186,0x186,-1,206}, 3097 {0x187,0x187,-1,1}, 3098 {0x189,0x18a,1,205}, 3099 {0x18b,0x18b,-1,1}, 3100 {0x18e,0x18e,-1,79}, 3101 {0x18f,0x18f,-1,202}, 3102 {0x190,0x190,-1,203}, 3103 {0x191,0x191,-1,1}, 3104 {0x193,0x193,-1,205}, 3105 {0x194,0x194,-1,207}, 3106 {0x196,0x196,-1,211}, 3107 {0x197,0x197,-1,209}, 3108 {0x198,0x198,-1,1}, 3109 {0x19c,0x19c,-1,211}, 3110 {0x19d,0x19d,-1,213}, 3111 {0x19f,0x19f,-1,214}, 3112 {0x1a0,0x1a4,2,1}, 3113 {0x1a6,0x1a6,-1,218}, 3114 {0x1a7,0x1a7,-1,1}, 3115 {0x1a9,0x1a9,-1,218}, 3116 {0x1ac,0x1ac,-1,1}, 3117 {0x1ae,0x1ae,-1,218}, 3118 {0x1af,0x1af,-1,1}, 3119 {0x1b1,0x1b2,1,217}, 3120 {0x1b3,0x1b5,2,1}, 3121 {0x1b7,0x1b7,-1,219}, 3122 {0x1b8,0x1bc,4,1}, 3123 {0x1c4,0x1c4,-1,2}, 3124 {0x1c5,0x1c5,-1,1}, 3125 {0x1c7,0x1c7,-1,2}, 3126 {0x1c8,0x1c8,-1,1}, 3127 {0x1ca,0x1ca,-1,2}, 3128 {0x1cb,0x1db,2,1}, 3129 {0x1de,0x1ee,2,1}, 3130 {0x1f1,0x1f1,-1,2}, 3131 {0x1f2,0x1f4,2,1}, 3132 {0x1f6,0x1f6,-1,-97}, 3133 {0x1f7,0x1f7,-1,-56}, 3134 {0x1f8,0x21e,2,1}, 3135 {0x220,0x220,-1,-130}, 3136 {0x222,0x232,2,1}, 3137 {0x23a,0x23a,-1,10795}, 3138 {0x23b,0x23b,-1,1}, 3139 {0x23d,0x23d,-1,-163}, 3140 {0x23e,0x23e,-1,10792}, 3141 {0x241,0x241,-1,1}, 3142 {0x243,0x243,-1,-195}, 3143 {0x244,0x244,-1,69}, 3144 {0x245,0x245,-1,71}, 3145 {0x246,0x24e,2,1}, 3146 {0x370,0x372,2,1}, 3147 {0x376,0x376,-1,1}, 3148 {0x37f,0x37f,-1,116}, 3149 {0x386,0x386,-1,38}, 3150 {0x388,0x38a,1,37}, 3151 {0x38c,0x38c,-1,64}, 3152 {0x38e,0x38f,1,63}, 3153 {0x391,0x3a1,1,32}, 3154 {0x3a3,0x3ab,1,32}, 3155 {0x3cf,0x3cf,-1,8}, 3156 {0x3d8,0x3ee,2,1}, 3157 {0x3f4,0x3f4,-1,-60}, 3158 {0x3f7,0x3f7,-1,1}, 3159 {0x3f9,0x3f9,-1,-7}, 3160 {0x3fa,0x3fa,-1,1}, 3161 {0x3fd,0x3ff,1,-130}, 3162 {0x400,0x40f,1,80}, 3163 {0x410,0x42f,1,32}, 3164 {0x460,0x480,2,1}, 3165 {0x48a,0x4be,2,1}, 3166 {0x4c0,0x4c0,-1,15}, 3167 {0x4c1,0x4cd,2,1}, 3168 {0x4d0,0x52e,2,1}, 3169 {0x531,0x556,1,48}, 3170 {0x10a0,0x10c5,1,7264}, 3171 {0x10c7,0x10cd,6,7264}, 3172 {0x13a0,0x13ef,1,38864}, 3173 {0x13f0,0x13f5,1,8}, 3174 {0x1e00,0x1e94,2,1}, 3175 {0x1e9e,0x1e9e,-1,-7615}, 3176 {0x1ea0,0x1efe,2,1}, 3177 {0x1f08,0x1f0f,1,-8}, 3178 {0x1f18,0x1f1d,1,-8}, 3179 {0x1f28,0x1f2f,1,-8}, 3180 {0x1f38,0x1f3f,1,-8}, 3181 {0x1f48,0x1f4d,1,-8}, 3182 {0x1f59,0x1f5f,2,-8}, 3183 {0x1f68,0x1f6f,1,-8}, 3184 {0x1f88,0x1f8f,1,-8}, 3185 {0x1f98,0x1f9f,1,-8}, 3186 {0x1fa8,0x1faf,1,-8}, 3187 {0x1fb8,0x1fb9,1,-8}, 3188 {0x1fba,0x1fbb,1,-74}, 3189 {0x1fbc,0x1fbc,-1,-9}, 3190 {0x1fc8,0x1fcb,1,-86}, 3191 {0x1fcc,0x1fcc,-1,-9}, 3192 {0x1fd8,0x1fd9,1,-8}, 3193 {0x1fda,0x1fdb,1,-100}, 3194 {0x1fe8,0x1fe9,1,-8}, 3195 {0x1fea,0x1feb,1,-112}, 3196 {0x1fec,0x1fec,-1,-7}, 3197 {0x1ff8,0x1ff9,1,-128}, 3198 {0x1ffa,0x1ffb,1,-126}, 3199 {0x1ffc,0x1ffc,-1,-9}, 3200 {0x2126,0x2126,-1,-7517}, 3201 {0x212a,0x212a,-1,-8383}, 3202 {0x212b,0x212b,-1,-8262}, 3203 {0x2132,0x2132,-1,28}, 3204 {0x2160,0x216f,1,16}, 3205 {0x2183,0x2183,-1,1}, 3206 {0x24b6,0x24cf,1,26}, 3207 {0x2c00,0x2c2e,1,48}, 3208 {0x2c60,0x2c60,-1,1}, 3209 {0x2c62,0x2c62,-1,-10743}, 3210 {0x2c63,0x2c63,-1,-3814}, 3211 {0x2c64,0x2c64,-1,-10727}, 3212 {0x2c67,0x2c6b,2,1}, 3213 {0x2c6d,0x2c6d,-1,-10780}, 3214 {0x2c6e,0x2c6e,-1,-10749}, 3215 {0x2c6f,0x2c6f,-1,-10783}, 3216 {0x2c70,0x2c70,-1,-10782}, 3217 {0x2c72,0x2c75,3,1}, 3218 {0x2c7e,0x2c7f,1,-10815}, 3219 {0x2c80,0x2ce2,2,1}, 3220 {0x2ceb,0x2ced,2,1}, 3221 {0x2cf2,0xa640,31054,1}, 3222 {0xa642,0xa66c,2,1}, 3223 {0xa680,0xa69a,2,1}, 3224 {0xa722,0xa72e,2,1}, 3225 {0xa732,0xa76e,2,1}, 3226 {0xa779,0xa77b,2,1}, 3227 {0xa77d,0xa77d,-1,-35332}, 3228 {0xa77e,0xa786,2,1}, 3229 {0xa78b,0xa78b,-1,1}, 3230 {0xa78d,0xa78d,-1,-42280}, 3231 {0xa790,0xa792,2,1}, 3232 {0xa796,0xa7a8,2,1}, 3233 {0xa7aa,0xa7aa,-1,-42308}, 3234 {0xa7ab,0xa7ab,-1,-42319}, 3235 {0xa7ac,0xa7ac,-1,-42315}, 3236 {0xa7ad,0xa7ad,-1,-42305}, 3237 {0xa7ae,0xa7ae,-1,-42308}, 3238 {0xa7b0,0xa7b0,-1,-42258}, 3239 {0xa7b1,0xa7b1,-1,-42282}, 3240 {0xa7b2,0xa7b2,-1,-42261}, 3241 {0xa7b3,0xa7b3,-1,928}, 3242 {0xa7b4,0xa7b6,2,1}, 3243 {0xff21,0xff3a,1,32}, 3244 {0x10400,0x10427,1,40}, 3245 {0x104b0,0x104d3,1,40}, 3246 {0x10c80,0x10cb2,1,64}, 3247 {0x118a0,0x118bf,1,32}, 3248 {0x1e900,0x1e921,1,34} 3249 }; 3250 3251 static convertStruct toUpper[] = 3252 { 3253 {0x61,0x7a,1,-32}, 3254 {0xb5,0xb5,-1,743}, 3255 {0xe0,0xf6,1,-32}, 3256 {0xf8,0xfe,1,-32}, 3257 {0xff,0xff,-1,121}, 3258 {0x101,0x12f,2,-1}, 3259 {0x131,0x131,-1,-232}, 3260 {0x133,0x137,2,-1}, 3261 {0x13a,0x148,2,-1}, 3262 {0x14b,0x177,2,-1}, 3263 {0x17a,0x17e,2,-1}, 3264 {0x17f,0x17f,-1,-300}, 3265 {0x180,0x180,-1,195}, 3266 {0x183,0x185,2,-1}, 3267 {0x188,0x18c,4,-1}, 3268 {0x192,0x192,-1,-1}, 3269 {0x195,0x195,-1,97}, 3270 {0x199,0x199,-1,-1}, 3271 {0x19a,0x19a,-1,163}, 3272 {0x19e,0x19e,-1,130}, 3273 {0x1a1,0x1a5,2,-1}, 3274 {0x1a8,0x1ad,5,-1}, 3275 {0x1b0,0x1b4,4,-1}, 3276 {0x1b6,0x1b9,3,-1}, 3277 {0x1bd,0x1bd,-1,-1}, 3278 {0x1bf,0x1bf,-1,56}, 3279 {0x1c5,0x1c5,-1,-1}, 3280 {0x1c6,0x1c6,-1,-2}, 3281 {0x1c8,0x1c8,-1,-1}, 3282 {0x1c9,0x1c9,-1,-2}, 3283 {0x1cb,0x1cb,-1,-1}, 3284 {0x1cc,0x1cc,-1,-2}, 3285 {0x1ce,0x1dc,2,-1}, 3286 {0x1dd,0x1dd,-1,-79}, 3287 {0x1df,0x1ef,2,-1}, 3288 {0x1f2,0x1f2,-1,-1}, 3289 {0x1f3,0x1f3,-1,-2}, 3290 {0x1f5,0x1f9,4,-1}, 3291 {0x1fb,0x21f,2,-1}, 3292 {0x223,0x233,2,-1}, 3293 {0x23c,0x23c,-1,-1}, 3294 {0x23f,0x240,1,10815}, 3295 {0x242,0x247,5,-1}, 3296 {0x249,0x24f,2,-1}, 3297 {0x250,0x250,-1,10783}, 3298 {0x251,0x251,-1,10780}, 3299 {0x252,0x252,-1,10782}, 3300 {0x253,0x253,-1,-210}, 3301 {0x254,0x254,-1,-206}, 3302 {0x256,0x257,1,-205}, 3303 {0x259,0x259,-1,-202}, 3304 {0x25b,0x25b,-1,-203}, 3305 {0x25c,0x25c,-1,42319}, 3306 {0x260,0x260,-1,-205}, 3307 {0x261,0x261,-1,42315}, 3308 {0x263,0x263,-1,-207}, 3309 {0x265,0x265,-1,42280}, 3310 {0x266,0x266,-1,42308}, 3311 {0x268,0x268,-1,-209}, 3312 {0x269,0x269,-1,-211}, 3313 {0x26a,0x26a,-1,42308}, 3314 {0x26b,0x26b,-1,10743}, 3315 {0x26c,0x26c,-1,42305}, 3316 {0x26f,0x26f,-1,-211}, 3317 {0x271,0x271,-1,10749}, 3318 {0x272,0x272,-1,-213}, 3319 {0x275,0x275,-1,-214}, 3320 {0x27d,0x27d,-1,10727}, 3321 {0x280,0x283,3,-218}, 3322 {0x287,0x287,-1,42282}, 3323 {0x288,0x288,-1,-218}, 3324 {0x289,0x289,-1,-69}, 3325 {0x28a,0x28b,1,-217}, 3326 {0x28c,0x28c,-1,-71}, 3327 {0x292,0x292,-1,-219}, 3328 {0x29d,0x29d,-1,42261}, 3329 {0x29e,0x29e,-1,42258}, 3330 {0x345,0x345,-1,84}, 3331 {0x371,0x373,2,-1}, 3332 {0x377,0x377,-1,-1}, 3333 {0x37b,0x37d,1,130}, 3334 {0x3ac,0x3ac,-1,-38}, 3335 {0x3ad,0x3af,1,-37}, 3336 {0x3b1,0x3c1,1,-32}, 3337 {0x3c2,0x3c2,-1,-31}, 3338 {0x3c3,0x3cb,1,-32}, 3339 {0x3cc,0x3cc,-1,-64}, 3340 {0x3cd,0x3ce,1,-63}, 3341 {0x3d0,0x3d0,-1,-62}, 3342 {0x3d1,0x3d1,-1,-57}, 3343 {0x3d5,0x3d5,-1,-47}, 3344 {0x3d6,0x3d6,-1,-54}, 3345 {0x3d7,0x3d7,-1,-8}, 3346 {0x3d9,0x3ef,2,-1}, 3347 {0x3f0,0x3f0,-1,-86}, 3348 {0x3f1,0x3f1,-1,-80}, 3349 {0x3f2,0x3f2,-1,7}, 3350 {0x3f3,0x3f3,-1,-116}, 3351 {0x3f5,0x3f5,-1,-96}, 3352 {0x3f8,0x3fb,3,-1}, 3353 {0x430,0x44f,1,-32}, 3354 {0x450,0x45f,1,-80}, 3355 {0x461,0x481,2,-1}, 3356 {0x48b,0x4bf,2,-1}, 3357 {0x4c2,0x4ce,2,-1}, 3358 {0x4cf,0x4cf,-1,-15}, 3359 {0x4d1,0x52f,2,-1}, 3360 {0x561,0x586,1,-48}, 3361 {0x13f8,0x13fd,1,-8}, 3362 {0x1c80,0x1c80,-1,-6254}, 3363 {0x1c81,0x1c81,-1,-6253}, 3364 {0x1c82,0x1c82,-1,-6244}, 3365 {0x1c83,0x1c84,1,-6242}, 3366 {0x1c85,0x1c85,-1,-6243}, 3367 {0x1c86,0x1c86,-1,-6236}, 3368 {0x1c87,0x1c87,-1,-6181}, 3369 {0x1c88,0x1c88,-1,35266}, 3370 {0x1d79,0x1d79,-1,35332}, 3371 {0x1d7d,0x1d7d,-1,3814}, 3372 {0x1e01,0x1e95,2,-1}, 3373 {0x1e9b,0x1e9b,-1,-59}, 3374 {0x1ea1,0x1eff,2,-1}, 3375 {0x1f00,0x1f07,1,8}, 3376 {0x1f10,0x1f15,1,8}, 3377 {0x1f20,0x1f27,1,8}, 3378 {0x1f30,0x1f37,1,8}, 3379 {0x1f40,0x1f45,1,8}, 3380 {0x1f51,0x1f57,2,8}, 3381 {0x1f60,0x1f67,1,8}, 3382 {0x1f70,0x1f71,1,74}, 3383 {0x1f72,0x1f75,1,86}, 3384 {0x1f76,0x1f77,1,100}, 3385 {0x1f78,0x1f79,1,128}, 3386 {0x1f7a,0x1f7b,1,112}, 3387 {0x1f7c,0x1f7d,1,126}, 3388 {0x1f80,0x1f87,1,8}, 3389 {0x1f90,0x1f97,1,8}, 3390 {0x1fa0,0x1fa7,1,8}, 3391 {0x1fb0,0x1fb1,1,8}, 3392 {0x1fb3,0x1fb3,-1,9}, 3393 {0x1fbe,0x1fbe,-1,-7205}, 3394 {0x1fc3,0x1fc3,-1,9}, 3395 {0x1fd0,0x1fd1,1,8}, 3396 {0x1fe0,0x1fe1,1,8}, 3397 {0x1fe5,0x1fe5,-1,7}, 3398 {0x1ff3,0x1ff3,-1,9}, 3399 {0x214e,0x214e,-1,-28}, 3400 {0x2170,0x217f,1,-16}, 3401 {0x2184,0x2184,-1,-1}, 3402 {0x24d0,0x24e9,1,-26}, 3403 {0x2c30,0x2c5e,1,-48}, 3404 {0x2c61,0x2c61,-1,-1}, 3405 {0x2c65,0x2c65,-1,-10795}, 3406 {0x2c66,0x2c66,-1,-10792}, 3407 {0x2c68,0x2c6c,2,-1}, 3408 {0x2c73,0x2c76,3,-1}, 3409 {0x2c81,0x2ce3,2,-1}, 3410 {0x2cec,0x2cee,2,-1}, 3411 {0x2cf3,0x2cf3,-1,-1}, 3412 {0x2d00,0x2d25,1,-7264}, 3413 {0x2d27,0x2d2d,6,-7264}, 3414 {0xa641,0xa66d,2,-1}, 3415 {0xa681,0xa69b,2,-1}, 3416 {0xa723,0xa72f,2,-1}, 3417 {0xa733,0xa76f,2,-1}, 3418 {0xa77a,0xa77c,2,-1}, 3419 {0xa77f,0xa787,2,-1}, 3420 {0xa78c,0xa791,5,-1}, 3421 {0xa793,0xa797,4,-1}, 3422 {0xa799,0xa7a9,2,-1}, 3423 {0xa7b5,0xa7b7,2,-1}, 3424 {0xab53,0xab53,-1,-928}, 3425 {0xab70,0xabbf,1,-38864}, 3426 {0xff41,0xff5a,1,-32}, 3427 {0x10428,0x1044f,1,-40}, 3428 {0x104d8,0x104fb,1,-40}, 3429 {0x10cc0,0x10cf2,1,-64}, 3430 {0x118c0,0x118df,1,-32}, 3431 {0x1e922,0x1e943,1,-34} 3432 }; 3433 3434 /* 3435 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3436 * simple case folding. 3437 */ 3438 int 3439 utf_toupper(int a) 3440 { 3441 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */ 3442 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3443 return TOUPPER_ASC(a); 3444 3445 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3446 /* If towupper() is available and handles Unicode, use it. */ 3447 if (!(cmp_flags & CMP_INTERNAL)) 3448 return towupper(a); 3449 #endif 3450 3451 /* For characters below 128 use locale sensitive toupper(). */ 3452 if (a < 128) 3453 return TOUPPER_LOC(a); 3454 3455 /* For any other characters use the above mapping table. */ 3456 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3457 } 3458 3459 int 3460 utf_islower(int a) 3461 { 3462 /* German sharp s is lower case but has no upper case equivalent. */ 3463 return (utf_toupper(a) != a) || a == 0xdf; 3464 } 3465 3466 /* 3467 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3468 * simple case folding. 3469 */ 3470 int 3471 utf_tolower(int a) 3472 { 3473 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */ 3474 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3475 return TOLOWER_ASC(a); 3476 3477 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3478 /* If towlower() is available and handles Unicode, use it. */ 3479 if (!(cmp_flags & CMP_INTERNAL)) 3480 return towlower(a); 3481 #endif 3482 3483 /* For characters below 128 use locale sensitive tolower(). */ 3484 if (a < 128) 3485 return TOLOWER_LOC(a); 3486 3487 /* For any other characters use the above mapping table. */ 3488 return utf_convert(a, toLower, (int)sizeof(toLower)); 3489 } 3490 3491 int 3492 utf_isupper(int a) 3493 { 3494 return (utf_tolower(a) != a); 3495 } 3496 3497 static int 3498 utf_strnicmp( 3499 char_u *s1, 3500 char_u *s2, 3501 size_t n1, 3502 size_t n2) 3503 { 3504 int c1, c2, cdiff; 3505 char_u buffer[6]; 3506 3507 for (;;) 3508 { 3509 c1 = utf_safe_read_char_adv(&s1, &n1); 3510 c2 = utf_safe_read_char_adv(&s2, &n2); 3511 3512 if (c1 <= 0 || c2 <= 0) 3513 break; 3514 3515 if (c1 == c2) 3516 continue; 3517 3518 cdiff = utf_fold(c1) - utf_fold(c2); 3519 if (cdiff != 0) 3520 return cdiff; 3521 } 3522 3523 /* some string ended or has an incomplete/illegal character sequence */ 3524 3525 if (c1 == 0 || c2 == 0) 3526 { 3527 /* some string ended. shorter string is smaller */ 3528 if (c1 == 0 && c2 == 0) 3529 return 0; 3530 return c1 == 0 ? -1 : 1; 3531 } 3532 3533 /* Continue with bytewise comparison to produce some result that 3534 * would make comparison operations involving this function transitive. 3535 * 3536 * If only one string had an error, comparison should be made with 3537 * folded version of the other string. In this case it is enough 3538 * to fold just one character to determine the result of comparison. */ 3539 3540 if (c1 != -1 && c2 == -1) 3541 { 3542 n1 = utf_char2bytes(utf_fold(c1), buffer); 3543 s1 = buffer; 3544 } 3545 else if (c2 != -1 && c1 == -1) 3546 { 3547 n2 = utf_char2bytes(utf_fold(c2), buffer); 3548 s2 = buffer; 3549 } 3550 3551 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3552 { 3553 cdiff = (int)(*s1) - (int)(*s2); 3554 if (cdiff != 0) 3555 return cdiff; 3556 3557 s1++; 3558 s2++; 3559 n1--; 3560 n2--; 3561 } 3562 3563 if (n1 > 0 && *s1 == NUL) 3564 n1 = 0; 3565 if (n2 > 0 && *s2 == NUL) 3566 n2 = 0; 3567 3568 if (n1 == 0 && n2 == 0) 3569 return 0; 3570 return n1 == 0 ? -1 : 1; 3571 } 3572 3573 /* 3574 * Version of strnicmp() that handles multi-byte characters. 3575 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3576 * probably use strnicmp(), because there are no ASCII characters in the 3577 * second byte. 3578 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3579 * two characters otherwise. 3580 */ 3581 int 3582 mb_strnicmp(char_u *s1, char_u *s2, size_t nn) 3583 { 3584 int i, l; 3585 int cdiff; 3586 int n = (int)nn; 3587 3588 if (enc_utf8) 3589 { 3590 return utf_strnicmp(s1, s2, nn, nn); 3591 } 3592 else 3593 { 3594 for (i = 0; i < n; i += l) 3595 { 3596 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */ 3597 return 0; 3598 3599 l = (*mb_ptr2len)(s1 + i); 3600 if (l <= 1) 3601 { 3602 /* Single byte: first check normally, then with ignore case. */ 3603 if (s1[i] != s2[i]) 3604 { 3605 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3606 if (cdiff != 0) 3607 return cdiff; 3608 } 3609 } 3610 else 3611 { 3612 /* For non-Unicode multi-byte don't ignore case. */ 3613 if (l > n - i) 3614 l = n - i; 3615 cdiff = STRNCMP(s1 + i, s2 + i, l); 3616 if (cdiff != 0) 3617 return cdiff; 3618 } 3619 } 3620 } 3621 return 0; 3622 } 3623 3624 /* 3625 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3626 * 'encoding' has been set to. 3627 */ 3628 void 3629 show_utf8(void) 3630 { 3631 int len; 3632 int rlen = 0; 3633 char_u *line; 3634 int clen; 3635 int i; 3636 3637 /* Get the byte length of the char under the cursor, including composing 3638 * characters. */ 3639 line = ml_get_cursor(); 3640 len = utfc_ptr2len(line); 3641 if (len == 0) 3642 { 3643 MSG("NUL"); 3644 return; 3645 } 3646 3647 clen = 0; 3648 for (i = 0; i < len; ++i) 3649 { 3650 if (clen == 0) 3651 { 3652 /* start of (composing) character, get its length */ 3653 if (i > 0) 3654 { 3655 STRCPY(IObuff + rlen, "+ "); 3656 rlen += 2; 3657 } 3658 clen = utf_ptr2len(line + i); 3659 } 3660 sprintf((char *)IObuff + rlen, "%02x ", 3661 (line[i] == NL) ? NUL : line[i]); /* NUL is stored as NL */ 3662 --clen; 3663 rlen += (int)STRLEN(IObuff + rlen); 3664 if (rlen > IOSIZE - 20) 3665 break; 3666 } 3667 3668 msg(IObuff); 3669 } 3670 3671 /* 3672 * mb_head_off() function pointer. 3673 * Return offset from "p" to the first byte of the character it points into. 3674 * If "p" points to the NUL at the end of the string return 0. 3675 * Returns 0 when already at the first byte of a character. 3676 */ 3677 int 3678 latin_head_off(char_u *base UNUSED, char_u *p UNUSED) 3679 { 3680 return 0; 3681 } 3682 3683 int 3684 dbcs_head_off(char_u *base, char_u *p) 3685 { 3686 char_u *q; 3687 3688 /* It can't be a trailing byte when not using DBCS, at the start of the 3689 * string or the previous byte can't start a double-byte. */ 3690 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3691 return 0; 3692 3693 /* This is slow: need to start at the base and go forward until the 3694 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */ 3695 q = base; 3696 while (q < p) 3697 q += dbcs_ptr2len(q); 3698 return (q == p) ? 0 : 1; 3699 } 3700 3701 /* 3702 * Special version of dbcs_head_off() that works for ScreenLines[], where 3703 * single-width DBCS_JPNU characters are stored separately. 3704 */ 3705 int 3706 dbcs_screen_head_off(char_u *base, char_u *p) 3707 { 3708 char_u *q; 3709 3710 /* It can't be a trailing byte when not using DBCS, at the start of the 3711 * string or the previous byte can't start a double-byte. 3712 * For euc-jp an 0x8e byte in the previous cell always means we have a 3713 * lead byte in the current cell. */ 3714 if (p <= base 3715 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3716 || MB_BYTE2LEN(p[-1]) == 1 3717 || *p == NUL) 3718 return 0; 3719 3720 /* This is slow: need to start at the base and go forward until the 3721 * byte we are looking for. Return 1 when we went past it, 0 otherwise. 3722 * For DBCS_JPNU look out for 0x8e, which means the second byte is not 3723 * stored as the next byte. */ 3724 q = base; 3725 while (q < p) 3726 { 3727 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3728 ++q; 3729 else 3730 q += dbcs_ptr2len(q); 3731 } 3732 return (q == p) ? 0 : 1; 3733 } 3734 3735 int 3736 utf_head_off(char_u *base, char_u *p) 3737 { 3738 char_u *q; 3739 char_u *s; 3740 int c; 3741 int len; 3742 #ifdef FEAT_ARABIC 3743 char_u *j; 3744 #endif 3745 3746 if (*p < 0x80) /* be quick for ASCII */ 3747 return 0; 3748 3749 /* Skip backwards over trailing bytes: 10xx.xxxx 3750 * Skip backwards again if on a composing char. */ 3751 for (q = p; ; --q) 3752 { 3753 /* Move s to the last byte of this char. */ 3754 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3755 ; 3756 /* Move q to the first byte of this char. */ 3757 while (q > base && (*q & 0xc0) == 0x80) 3758 --q; 3759 /* Check for illegal sequence. Do allow an illegal byte after where we 3760 * started. */ 3761 len = utf8len_tab[*q]; 3762 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3763 return 0; 3764 3765 if (q <= base) 3766 break; 3767 3768 c = utf_ptr2char(q); 3769 if (utf_iscomposing(c)) 3770 continue; 3771 3772 #ifdef FEAT_ARABIC 3773 if (arabic_maycombine(c)) 3774 { 3775 /* Advance to get a sneak-peak at the next char */ 3776 j = q; 3777 --j; 3778 /* Move j to the first byte of this char. */ 3779 while (j > base && (*j & 0xc0) == 0x80) 3780 --j; 3781 if (arabic_combine(utf_ptr2char(j), c)) 3782 continue; 3783 } 3784 #endif 3785 break; 3786 } 3787 3788 return (int)(p - q); 3789 } 3790 3791 /* 3792 * Copy a character from "*fp" to "*tp" and advance the pointers. 3793 */ 3794 void 3795 mb_copy_char(char_u **fp, char_u **tp) 3796 { 3797 int l = (*mb_ptr2len)(*fp); 3798 3799 mch_memmove(*tp, *fp, (size_t)l); 3800 *tp += l; 3801 *fp += l; 3802 } 3803 3804 /* 3805 * Return the offset from "p" to the first byte of a character. When "p" is 3806 * at the start of a character 0 is returned, otherwise the offset to the next 3807 * character. Can start anywhere in a stream of bytes. 3808 */ 3809 int 3810 mb_off_next(char_u *base, char_u *p) 3811 { 3812 int i; 3813 int j; 3814 3815 if (enc_utf8) 3816 { 3817 if (*p < 0x80) /* be quick for ASCII */ 3818 return 0; 3819 3820 /* Find the next character that isn't 10xx.xxxx */ 3821 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 3822 ; 3823 if (i > 0) 3824 { 3825 /* Check for illegal sequence. */ 3826 for (j = 0; p - j > base; ++j) 3827 if ((p[-j] & 0xc0) != 0x80) 3828 break; 3829 if (utf8len_tab[p[-j]] != i + j) 3830 return 0; 3831 } 3832 return i; 3833 } 3834 3835 /* Only need to check if we're on a trail byte, it doesn't matter if we 3836 * want the offset to the next or current character. */ 3837 return (*mb_head_off)(base, p); 3838 } 3839 3840 /* 3841 * Return the offset from "p" to the last byte of the character it points 3842 * into. Can start anywhere in a stream of bytes. 3843 */ 3844 int 3845 mb_tail_off(char_u *base, char_u *p) 3846 { 3847 int i; 3848 int j; 3849 3850 if (*p == NUL) 3851 return 0; 3852 3853 if (enc_utf8) 3854 { 3855 /* Find the last character that is 10xx.xxxx */ 3856 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 3857 ; 3858 /* Check for illegal sequence. */ 3859 for (j = 0; p - j > base; ++j) 3860 if ((p[-j] & 0xc0) != 0x80) 3861 break; 3862 if (utf8len_tab[p[-j]] != i + j + 1) 3863 return 0; 3864 return i; 3865 } 3866 3867 /* It can't be the first byte if a double-byte when not using DBCS, at the 3868 * end of the string or the byte can't start a double-byte. */ 3869 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 3870 return 0; 3871 3872 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3873 return 1 - dbcs_head_off(base, p); 3874 } 3875 3876 /* 3877 * Find the next illegal byte sequence. 3878 */ 3879 void 3880 utf_find_illegal(void) 3881 { 3882 pos_T pos = curwin->w_cursor; 3883 char_u *p; 3884 int len; 3885 vimconv_T vimconv; 3886 char_u *tofree = NULL; 3887 3888 vimconv.vc_type = CONV_NONE; 3889 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 3890 { 3891 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 3892 * possibly a utf-8 file with illegal bytes. Setup for conversion 3893 * from utf-8 to 'fileencoding'. */ 3894 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 3895 } 3896 3897 #ifdef FEAT_VIRTUALEDIT 3898 curwin->w_cursor.coladd = 0; 3899 #endif 3900 for (;;) 3901 { 3902 p = ml_get_cursor(); 3903 if (vimconv.vc_type != CONV_NONE) 3904 { 3905 vim_free(tofree); 3906 tofree = string_convert(&vimconv, p, NULL); 3907 if (tofree == NULL) 3908 break; 3909 p = tofree; 3910 } 3911 3912 while (*p != NUL) 3913 { 3914 /* Illegal means that there are not enough trail bytes (checked by 3915 * utf_ptr2len()) or too many of them (overlong sequence). */ 3916 len = utf_ptr2len(p); 3917 if (*p >= 0x80 && (len == 1 3918 || utf_char2len(utf_ptr2char(p)) != len)) 3919 { 3920 if (vimconv.vc_type == CONV_NONE) 3921 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 3922 else 3923 { 3924 int l; 3925 3926 len = (int)(p - tofree); 3927 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 3928 { 3929 l = utf_ptr2len(p); 3930 curwin->w_cursor.col += l; 3931 } 3932 } 3933 goto theend; 3934 } 3935 p += len; 3936 } 3937 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 3938 break; 3939 ++curwin->w_cursor.lnum; 3940 curwin->w_cursor.col = 0; 3941 } 3942 3943 /* didn't find it: don't move and beep */ 3944 curwin->w_cursor = pos; 3945 beep_flush(); 3946 3947 theend: 3948 vim_free(tofree); 3949 convert_setup(&vimconv, NULL, NULL); 3950 } 3951 3952 #if defined(FEAT_GUI_GTK) || defined(PROTO) 3953 /* 3954 * Return TRUE if string "s" is a valid utf-8 string. 3955 * When "end" is NULL stop at the first NUL. 3956 * When "end" is positive stop there. 3957 */ 3958 int 3959 utf_valid_string(char_u *s, char_u *end) 3960 { 3961 int l; 3962 char_u *p = s; 3963 3964 while (end == NULL ? *p != NUL : p < end) 3965 { 3966 l = utf8len_tab_zero[*p]; 3967 if (l == 0) 3968 return FALSE; /* invalid lead byte */ 3969 if (end != NULL && p + l > end) 3970 return FALSE; /* incomplete byte sequence */ 3971 ++p; 3972 while (--l > 0) 3973 if ((*p++ & 0xc0) != 0x80) 3974 return FALSE; /* invalid trail byte */ 3975 } 3976 return TRUE; 3977 } 3978 #endif 3979 3980 #if defined(FEAT_GUI) || defined(PROTO) 3981 /* 3982 * Special version of mb_tail_off() for use in ScreenLines[]. 3983 */ 3984 int 3985 dbcs_screen_tail_off(char_u *base, char_u *p) 3986 { 3987 /* It can't be the first byte if a double-byte when not using DBCS, at the 3988 * end of the string or the byte can't start a double-byte. 3989 * For euc-jp an 0x8e byte always means we have a lead byte in the current 3990 * cell. */ 3991 if (*p == NUL || p[1] == NUL 3992 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 3993 || MB_BYTE2LEN(*p) == 1) 3994 return 0; 3995 3996 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3997 return 1 - dbcs_screen_head_off(base, p); 3998 } 3999 #endif 4000 4001 /* 4002 * If the cursor moves on an trail byte, set the cursor on the lead byte. 4003 * Thus it moves left if necessary. 4004 * Return TRUE when the cursor was adjusted. 4005 */ 4006 void 4007 mb_adjust_cursor(void) 4008 { 4009 mb_adjustpos(curbuf, &curwin->w_cursor); 4010 } 4011 4012 /* 4013 * Adjust position "*lp" to point to the first byte of a multi-byte character. 4014 * If it points to a tail byte it's moved backwards to the head byte. 4015 */ 4016 void 4017 mb_adjustpos(buf_T *buf, pos_T *lp) 4018 { 4019 char_u *p; 4020 4021 if (lp->col > 0 4022 #ifdef FEAT_VIRTUALEDIT 4023 || lp->coladd > 1 4024 #endif 4025 ) 4026 { 4027 p = ml_get_buf(buf, lp->lnum, FALSE); 4028 lp->col -= (*mb_head_off)(p, p + lp->col); 4029 #ifdef FEAT_VIRTUALEDIT 4030 /* Reset "coladd" when the cursor would be on the right half of a 4031 * double-wide character. */ 4032 if (lp->coladd == 1 4033 && p[lp->col] != TAB 4034 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 4035 && ptr2cells(p + lp->col) > 1) 4036 lp->coladd = 0; 4037 #endif 4038 } 4039 } 4040 4041 /* 4042 * Return a pointer to the character before "*p", if there is one. 4043 */ 4044 char_u * 4045 mb_prevptr( 4046 char_u *line, /* start of the string */ 4047 char_u *p) 4048 { 4049 if (p > line) 4050 MB_PTR_BACK(line, p); 4051 return p; 4052 } 4053 4054 /* 4055 * Return the character length of "str". Each multi-byte character (with 4056 * following composing characters) counts as one. 4057 */ 4058 int 4059 mb_charlen(char_u *str) 4060 { 4061 char_u *p = str; 4062 int count; 4063 4064 if (p == NULL) 4065 return 0; 4066 4067 for (count = 0; *p != NUL; count++) 4068 p += (*mb_ptr2len)(p); 4069 4070 return count; 4071 } 4072 4073 #if defined(FEAT_SPELL) || defined(PROTO) 4074 /* 4075 * Like mb_charlen() but for a string with specified length. 4076 */ 4077 int 4078 mb_charlen_len(char_u *str, int len) 4079 { 4080 char_u *p = str; 4081 int count; 4082 4083 for (count = 0; *p != NUL && p < str + len; count++) 4084 p += (*mb_ptr2len)(p); 4085 4086 return count; 4087 } 4088 #endif 4089 4090 /* 4091 * Try to un-escape a multi-byte character. 4092 * Used for the "to" and "from" part of a mapping. 4093 * Return the un-escaped string if it is a multi-byte character, and advance 4094 * "pp" to just after the bytes that formed it. 4095 * Return NULL if no multi-byte char was found. 4096 */ 4097 char_u * 4098 mb_unescape(char_u **pp) 4099 { 4100 static char_u buf[6]; 4101 int n; 4102 int m = 0; 4103 char_u *str = *pp; 4104 4105 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 4106 * KS_EXTRA KE_CSI to CSI. 4107 * Maximum length of a utf-8 character is 4 bytes. */ 4108 for (n = 0; str[n] != NUL && m < 4; ++n) 4109 { 4110 if (str[n] == K_SPECIAL 4111 && str[n + 1] == KS_SPECIAL 4112 && str[n + 2] == KE_FILLER) 4113 { 4114 buf[m++] = K_SPECIAL; 4115 n += 2; 4116 } 4117 else if ((str[n] == K_SPECIAL 4118 # ifdef FEAT_GUI 4119 || str[n] == CSI 4120 # endif 4121 ) 4122 && str[n + 1] == KS_EXTRA 4123 && str[n + 2] == (int)KE_CSI) 4124 { 4125 buf[m++] = CSI; 4126 n += 2; 4127 } 4128 else if (str[n] == K_SPECIAL 4129 # ifdef FEAT_GUI 4130 || str[n] == CSI 4131 # endif 4132 ) 4133 break; /* a special key can't be a multibyte char */ 4134 else 4135 buf[m++] = str[n]; 4136 buf[m] = NUL; 4137 4138 /* Return a multi-byte character if it's found. An illegal sequence 4139 * will result in a 1 here. */ 4140 if ((*mb_ptr2len)(buf) > 1) 4141 { 4142 *pp = str + n + 1; 4143 return buf; 4144 } 4145 4146 /* Bail out quickly for ASCII. */ 4147 if (buf[0] < 128) 4148 break; 4149 } 4150 return NULL; 4151 } 4152 4153 /* 4154 * Return TRUE if the character at "row"/"col" on the screen is the left side 4155 * of a double-width character. 4156 * Caller must make sure "row" and "col" are not invalid! 4157 */ 4158 int 4159 mb_lefthalve(int row, int col) 4160 { 4161 #ifdef FEAT_HANGULIN 4162 if (composing_hangul) 4163 return TRUE; 4164 #endif 4165 return (*mb_off2cells)(LineOffset[row] + col, 4166 LineOffset[row] + screen_Columns) > 1; 4167 } 4168 4169 /* 4170 * Correct a position on the screen, if it's the right half of a double-wide 4171 * char move it to the left half. Returns the corrected column. 4172 */ 4173 int 4174 mb_fix_col(int col, int row) 4175 { 4176 col = check_col(col); 4177 row = check_row(row); 4178 if (has_mbyte && ScreenLines != NULL && col > 0 4179 && ((enc_dbcs 4180 && ScreenLines[LineOffset[row] + col] != NUL 4181 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 4182 ScreenLines + LineOffset[row] + col)) 4183 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0))) 4184 return col - 1; 4185 return col; 4186 } 4187 #endif 4188 4189 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 4190 static int enc_alias_search(char_u *name); 4191 4192 /* 4193 * Skip the Vim specific head of a 'encoding' name. 4194 */ 4195 char_u * 4196 enc_skip(char_u *p) 4197 { 4198 if (STRNCMP(p, "2byte-", 6) == 0) 4199 return p + 6; 4200 if (STRNCMP(p, "8bit-", 5) == 0) 4201 return p + 5; 4202 return p; 4203 } 4204 4205 /* 4206 * Find the canonical name for encoding "enc". 4207 * When the name isn't recognized, returns "enc" itself, but with all lower 4208 * case characters and '_' replaced with '-'. 4209 * Returns an allocated string. NULL for out-of-memory. 4210 */ 4211 char_u * 4212 enc_canonize(char_u *enc) 4213 { 4214 char_u *r; 4215 char_u *p, *s; 4216 int i; 4217 4218 # ifdef FEAT_MBYTE 4219 if (STRCMP(enc, "default") == 0) 4220 { 4221 /* Use the default encoding as it's found by set_init_1(). */ 4222 r = get_encoding_default(); 4223 if (r == NULL) 4224 r = (char_u *)"latin1"; 4225 return vim_strsave(r); 4226 } 4227 # endif 4228 4229 /* copy "enc" to allocated memory, with room for two '-' */ 4230 r = alloc((unsigned)(STRLEN(enc) + 3)); 4231 if (r != NULL) 4232 { 4233 /* Make it all lower case and replace '_' with '-'. */ 4234 p = r; 4235 for (s = enc; *s != NUL; ++s) 4236 { 4237 if (*s == '_') 4238 *p++ = '-'; 4239 else 4240 *p++ = TOLOWER_ASC(*s); 4241 } 4242 *p = NUL; 4243 4244 /* Skip "2byte-" and "8bit-". */ 4245 p = enc_skip(r); 4246 4247 /* Change "microsoft-cp" to "cp". Used in some spell files. */ 4248 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4249 STRMOVE(p, p + 10); 4250 4251 /* "iso8859" -> "iso-8859" */ 4252 if (STRNCMP(p, "iso8859", 7) == 0) 4253 { 4254 STRMOVE(p + 4, p + 3); 4255 p[3] = '-'; 4256 } 4257 4258 /* "iso-8859n" -> "iso-8859-n" */ 4259 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4260 { 4261 STRMOVE(p + 9, p + 8); 4262 p[8] = '-'; 4263 } 4264 4265 /* "latin-N" -> "latinN" */ 4266 if (STRNCMP(p, "latin-", 6) == 0) 4267 STRMOVE(p + 5, p + 6); 4268 4269 if (enc_canon_search(p) >= 0) 4270 { 4271 /* canonical name can be used unmodified */ 4272 if (p != r) 4273 STRMOVE(r, p); 4274 } 4275 else if ((i = enc_alias_search(p)) >= 0) 4276 { 4277 /* alias recognized, get canonical name */ 4278 vim_free(r); 4279 r = vim_strsave((char_u *)enc_canon_table[i].name); 4280 } 4281 } 4282 return r; 4283 } 4284 4285 /* 4286 * Search for an encoding alias of "name". 4287 * Returns -1 when not found. 4288 */ 4289 static int 4290 enc_alias_search(char_u *name) 4291 { 4292 int i; 4293 4294 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4295 if (STRCMP(name, enc_alias_table[i].name) == 0) 4296 return enc_alias_table[i].canon; 4297 return -1; 4298 } 4299 #endif 4300 4301 #if defined(FEAT_MBYTE) || defined(PROTO) 4302 4303 #ifdef HAVE_LANGINFO_H 4304 # include <langinfo.h> 4305 #endif 4306 4307 /* 4308 * Get the canonicalized encoding of the current locale. 4309 * Returns an allocated string when successful, NULL when not. 4310 */ 4311 char_u * 4312 enc_locale(void) 4313 { 4314 #ifndef WIN3264 4315 char *s; 4316 char *p; 4317 int i; 4318 #endif 4319 char buf[50]; 4320 #ifdef WIN3264 4321 long acp = GetACP(); 4322 4323 if (acp == 1200) 4324 STRCPY(buf, "ucs-2le"); 4325 else if (acp == 1252) /* cp1252 is used as latin1 */ 4326 STRCPY(buf, "latin1"); 4327 else 4328 sprintf(buf, "cp%ld", acp); 4329 #else 4330 # ifdef HAVE_NL_LANGINFO_CODESET 4331 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4332 # endif 4333 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4334 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4335 # endif 4336 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4337 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4338 s = getenv("LANG"); 4339 4340 if (s == NULL || *s == NUL) 4341 return FAIL; 4342 4343 /* The most generic locale format is: 4344 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4345 * If there is a '.' remove the part before it. 4346 * if there is something after the codeset, remove it. 4347 * Make the name lowercase and replace '_' with '-'. 4348 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4349 * "ko_KR.EUC" == "euc-kr" 4350 */ 4351 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4352 { 4353 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4354 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4355 { 4356 /* copy "XY.EUC" to "euc-XY" to buf[10] */ 4357 STRCPY(buf + 10, "euc-"); 4358 buf[14] = p[-2]; 4359 buf[15] = p[-1]; 4360 buf[16] = 0; 4361 s = buf + 10; 4362 } 4363 else 4364 s = p + 1; 4365 } 4366 for (i = 0; i < (int)sizeof(buf) - 1 && s[i] != NUL; ++i) 4367 { 4368 if (s[i] == '_' || s[i] == '-') 4369 buf[i] = '-'; 4370 else if (isalnum((int)s[i])) 4371 buf[i] = TOLOWER_ASC(s[i]); 4372 else 4373 break; 4374 } 4375 buf[i] = NUL; 4376 #endif 4377 4378 return enc_canonize((char_u *)buf); 4379 } 4380 4381 #if defined(WIN3264) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4382 /* 4383 * Convert an encoding name to an MS-Windows codepage. 4384 * Returns zero if no codepage can be figured out. 4385 */ 4386 int 4387 encname2codepage(char_u *name) 4388 { 4389 int cp; 4390 char_u *p = name; 4391 int idx; 4392 4393 if (STRNCMP(p, "8bit-", 5) == 0) 4394 p += 5; 4395 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4396 p += 6; 4397 4398 if (p[0] == 'c' && p[1] == 'p') 4399 cp = atoi((char *)p + 2); 4400 else if ((idx = enc_canon_search(p)) >= 0) 4401 cp = enc_canon_table[idx].codepage; 4402 else 4403 return 0; 4404 if (IsValidCodePage(cp)) 4405 return cp; 4406 return 0; 4407 } 4408 #endif 4409 4410 # if defined(USE_ICONV) || defined(PROTO) 4411 4412 static char_u *iconv_string(vimconv_T *vcp, char_u *str, int slen, int *unconvlenp, int *resultlenp); 4413 4414 /* 4415 * Call iconv_open() with a check if iconv() works properly (there are broken 4416 * versions). 4417 * Returns (void *)-1 if failed. 4418 * (should return iconv_t, but that causes problems with prototypes). 4419 */ 4420 void * 4421 my_iconv_open(char_u *to, char_u *from) 4422 { 4423 iconv_t fd; 4424 #define ICONV_TESTLEN 400 4425 char_u tobuf[ICONV_TESTLEN]; 4426 char *p; 4427 size_t tolen; 4428 static int iconv_ok = -1; 4429 4430 if (iconv_ok == FALSE) 4431 return (void *)-1; /* detected a broken iconv() previously */ 4432 4433 #ifdef DYNAMIC_ICONV 4434 /* Check if the iconv.dll can be found. */ 4435 if (!iconv_enabled(TRUE)) 4436 return (void *)-1; 4437 #endif 4438 4439 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4440 4441 if (fd != (iconv_t)-1 && iconv_ok == -1) 4442 { 4443 /* 4444 * Do a dummy iconv() call to check if it actually works. There is a 4445 * version of iconv() on Linux that is broken. We can't ignore it, 4446 * because it's wide-spread. The symptoms are that after outputting 4447 * the initial shift state the "to" pointer is NULL and conversion 4448 * stops for no apparent reason after about 8160 characters. 4449 */ 4450 p = (char *)tobuf; 4451 tolen = ICONV_TESTLEN; 4452 (void)iconv(fd, NULL, NULL, &p, &tolen); 4453 if (p == NULL) 4454 { 4455 iconv_ok = FALSE; 4456 iconv_close(fd); 4457 fd = (iconv_t)-1; 4458 } 4459 else 4460 iconv_ok = TRUE; 4461 } 4462 4463 return (void *)fd; 4464 } 4465 4466 /* 4467 * Convert the string "str[slen]" with iconv(). 4468 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4469 * sequence and set "*unconvlenp" to the length of it. 4470 * Returns the converted string in allocated memory. NULL for an error. 4471 * If resultlenp is not NULL, sets it to the result length in bytes. 4472 */ 4473 static char_u * 4474 iconv_string( 4475 vimconv_T *vcp, 4476 char_u *str, 4477 int slen, 4478 int *unconvlenp, 4479 int *resultlenp) 4480 { 4481 const char *from; 4482 size_t fromlen; 4483 char *to; 4484 size_t tolen; 4485 size_t len = 0; 4486 size_t done = 0; 4487 char_u *result = NULL; 4488 char_u *p; 4489 int l; 4490 4491 from = (char *)str; 4492 fromlen = slen; 4493 for (;;) 4494 { 4495 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4496 { 4497 /* Allocate enough room for most conversions. When re-allocating 4498 * increase the buffer size. */ 4499 len = len + fromlen * 2 + 40; 4500 p = alloc((unsigned)len); 4501 if (p != NULL && done > 0) 4502 mch_memmove(p, result, done); 4503 vim_free(result); 4504 result = p; 4505 if (result == NULL) /* out of memory */ 4506 break; 4507 } 4508 4509 to = (char *)result + done; 4510 tolen = len - done - 2; 4511 /* Avoid a warning for systems with a wrong iconv() prototype by 4512 * casting the second argument to void *. */ 4513 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4514 != (size_t)-1) 4515 { 4516 /* Finished, append a NUL. */ 4517 *to = NUL; 4518 break; 4519 } 4520 4521 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4522 * iconv library may use one of them. */ 4523 if (!vcp->vc_fail && unconvlenp != NULL 4524 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4525 { 4526 /* Handle an incomplete sequence at the end. */ 4527 *to = NUL; 4528 *unconvlenp = (int)fromlen; 4529 break; 4530 } 4531 4532 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4533 * iconv library may use one of them. */ 4534 else if (!vcp->vc_fail 4535 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4536 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4537 { 4538 /* Can't convert: insert a '?' and skip a character. This assumes 4539 * conversion from 'encoding' to something else. In other 4540 * situations we don't know what to skip anyway. */ 4541 *to++ = '?'; 4542 if ((*mb_ptr2cells)((char_u *)from) > 1) 4543 *to++ = '?'; 4544 if (enc_utf8) 4545 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4546 else 4547 { 4548 l = (*mb_ptr2len)((char_u *)from); 4549 if (l > (int)fromlen) 4550 l = (int)fromlen; 4551 } 4552 from += l; 4553 fromlen -= l; 4554 } 4555 else if (ICONV_ERRNO != ICONV_E2BIG) 4556 { 4557 /* conversion failed */ 4558 vim_free(result); 4559 result = NULL; 4560 break; 4561 } 4562 /* Not enough room or skipping illegal sequence. */ 4563 done = to - (char *)result; 4564 } 4565 4566 if (resultlenp != NULL && result != NULL) 4567 *resultlenp = (int)(to - (char *)result); 4568 return result; 4569 } 4570 4571 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4572 /* 4573 * Dynamically load the "iconv.dll" on Win32. 4574 */ 4575 4576 # ifndef DYNAMIC_ICONV /* must be generating prototypes */ 4577 # define HINSTANCE int 4578 # endif 4579 static HINSTANCE hIconvDLL = 0; 4580 static HINSTANCE hMsvcrtDLL = 0; 4581 4582 # ifndef DYNAMIC_ICONV_DLL 4583 # define DYNAMIC_ICONV_DLL "iconv.dll" 4584 # define DYNAMIC_ICONV_DLL_ALT1 "libiconv.dll" 4585 # define DYNAMIC_ICONV_DLL_ALT2 "libiconv2.dll" 4586 # define DYNAMIC_ICONV_DLL_ALT3 "libiconv-2.dll" 4587 # endif 4588 # ifndef DYNAMIC_MSVCRT_DLL 4589 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4590 # endif 4591 4592 /* 4593 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4594 */ 4595 int 4596 iconv_enabled(int verbose) 4597 { 4598 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4599 return TRUE; 4600 4601 /* The iconv DLL file goes under different names, try them all. 4602 * Do the "2" version first, it's newer. */ 4603 #ifdef DYNAMIC_ICONV_DLL_ALT2 4604 if (hIconvDLL == 0) 4605 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT2); 4606 #endif 4607 #ifdef DYNAMIC_ICONV_DLL_ALT3 4608 if (hIconvDLL == 0) 4609 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT3); 4610 #endif 4611 if (hIconvDLL == 0) 4612 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4613 #ifdef DYNAMIC_ICONV_DLL_ALT1 4614 if (hIconvDLL == 0) 4615 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT1); 4616 #endif 4617 4618 if (hIconvDLL != 0) 4619 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4620 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4621 { 4622 /* Only give the message when 'verbose' is set, otherwise it might be 4623 * done whenever a conversion is attempted. */ 4624 if (verbose && p_verbose > 0) 4625 { 4626 verbose_enter(); 4627 EMSG2(_(e_loadlib), 4628 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4629 verbose_leave(); 4630 } 4631 iconv_end(); 4632 return FALSE; 4633 } 4634 4635 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4636 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4637 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4638 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4639 iconv_errno = get_dll_import_func(hIconvDLL, "_errno"); 4640 if (iconv_errno == NULL) 4641 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4642 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4643 || iconvctl == NULL || iconv_errno == NULL) 4644 { 4645 iconv_end(); 4646 if (verbose && p_verbose > 0) 4647 { 4648 verbose_enter(); 4649 EMSG2(_(e_loadfunc), "for libiconv"); 4650 verbose_leave(); 4651 } 4652 return FALSE; 4653 } 4654 return TRUE; 4655 } 4656 4657 void 4658 iconv_end(void) 4659 { 4660 /* Don't use iconv() when inputting or outputting characters. */ 4661 if (input_conv.vc_type == CONV_ICONV) 4662 convert_setup(&input_conv, NULL, NULL); 4663 if (output_conv.vc_type == CONV_ICONV) 4664 convert_setup(&output_conv, NULL, NULL); 4665 4666 if (hIconvDLL != 0) 4667 FreeLibrary(hIconvDLL); 4668 if (hMsvcrtDLL != 0) 4669 FreeLibrary(hMsvcrtDLL); 4670 hIconvDLL = 0; 4671 hMsvcrtDLL = 0; 4672 } 4673 # endif /* DYNAMIC_ICONV */ 4674 # endif /* USE_ICONV */ 4675 4676 #endif /* FEAT_MBYTE */ 4677 4678 #if defined(FEAT_XIM) || defined(PROTO) 4679 4680 # if defined(FEAT_GUI_GTK) || defined(PROTO) 4681 static int xim_has_preediting INIT(= FALSE); /* IM current status */ 4682 4683 /* 4684 * Set preedit_start_col to the current cursor position. 4685 */ 4686 static void 4687 init_preedit_start_col(void) 4688 { 4689 if (State & CMDLINE) 4690 preedit_start_col = cmdline_getvcol_cursor(); 4691 else if (curwin != NULL && curwin->w_buffer != NULL) 4692 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); 4693 /* Prevent that preediting marks the buffer as changed. */ 4694 xim_changed_while_preediting = curbuf->b_changed; 4695 } 4696 4697 static int im_is_active = FALSE; /* IM is enabled for current mode */ 4698 static int preedit_is_active = FALSE; 4699 static int im_preedit_cursor = 0; /* cursor offset in characters */ 4700 static int im_preedit_trailing = 0; /* number of characters after cursor */ 4701 4702 static unsigned long im_commit_handler_id = 0; 4703 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; 4704 static unsigned int im_activatekey_state = 0; 4705 4706 void 4707 im_set_active(int active) 4708 { 4709 int was_active; 4710 4711 was_active = !!im_get_status(); 4712 im_is_active = (active && !p_imdisable); 4713 4714 if (im_is_active != was_active) 4715 xim_reset(); 4716 } 4717 4718 void 4719 xim_set_focus(int focus) 4720 { 4721 if (xic != NULL) 4722 { 4723 if (focus) 4724 gtk_im_context_focus_in(xic); 4725 else 4726 gtk_im_context_focus_out(xic); 4727 } 4728 } 4729 4730 void 4731 im_set_position(int row, int col) 4732 { 4733 if (xic != NULL) 4734 { 4735 GdkRectangle area; 4736 4737 area.x = FILL_X(col); 4738 area.y = FILL_Y(row); 4739 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); 4740 area.height = gui.char_height; 4741 4742 gtk_im_context_set_cursor_location(xic, &area); 4743 } 4744 } 4745 4746 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */ 4747 void 4748 xim_set_preedit(void) 4749 { 4750 im_set_position(gui.row, gui.col); 4751 } 4752 # endif 4753 4754 static void 4755 im_add_to_input(char_u *str, int len) 4756 { 4757 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */ 4758 if (input_conv.vc_type != CONV_NONE) 4759 { 4760 str = string_convert(&input_conv, str, &len); 4761 g_return_if_fail(str != NULL); 4762 } 4763 4764 add_to_input_buf_csi(str, len); 4765 4766 if (input_conv.vc_type != CONV_NONE) 4767 vim_free(str); 4768 4769 if (p_mh) /* blank out the pointer if necessary */ 4770 gui_mch_mousehide(TRUE); 4771 } 4772 4773 static void 4774 im_delete_preedit(void) 4775 { 4776 char_u bskey[] = {CSI, 'k', 'b'}; 4777 char_u delkey[] = {CSI, 'k', 'D'}; 4778 4779 if (State & NORMAL) 4780 { 4781 im_preedit_cursor = 0; 4782 return; 4783 } 4784 for (; im_preedit_cursor > 0; --im_preedit_cursor) 4785 add_to_input_buf(bskey, (int)sizeof(bskey)); 4786 4787 for (; im_preedit_trailing > 0; --im_preedit_trailing) 4788 add_to_input_buf(delkey, (int)sizeof(delkey)); 4789 } 4790 4791 /* 4792 * Move the cursor left by "num_move_back" characters. 4793 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for 4794 * these K_LEFT keys. 4795 */ 4796 static void 4797 im_correct_cursor(int num_move_back) 4798 { 4799 char_u backkey[] = {CSI, 'k', 'l'}; 4800 4801 if (State & NORMAL) 4802 return; 4803 # ifdef FEAT_RIGHTLEFT 4804 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) 4805 backkey[2] = 'r'; 4806 # endif 4807 for (; num_move_back > 0; --num_move_back) 4808 add_to_input_buf(backkey, (int)sizeof(backkey)); 4809 } 4810 4811 static int xim_expected_char = NUL; 4812 static int xim_ignored_char = FALSE; 4813 4814 /* 4815 * Update the mode and cursor while in an IM callback. 4816 */ 4817 static void 4818 im_show_info(void) 4819 { 4820 int old_vgetc_busy; 4821 4822 old_vgetc_busy = vgetc_busy; 4823 vgetc_busy = TRUE; 4824 showmode(); 4825 vgetc_busy = old_vgetc_busy; 4826 if ((State & NORMAL) || (State & INSERT)) 4827 setcursor(); 4828 out_flush(); 4829 } 4830 4831 /* 4832 * Callback invoked when the user finished preediting. 4833 * Put the final string into the input buffer. 4834 */ 4835 static void 4836 im_commit_cb(GtkIMContext *context UNUSED, 4837 const gchar *str, 4838 gpointer data UNUSED) 4839 { 4840 int slen = (int)STRLEN(str); 4841 int add_to_input = TRUE; 4842 int clen; 4843 int len = slen; 4844 int commit_with_preedit = TRUE; 4845 char_u *im_str; 4846 4847 #ifdef XIM_DEBUG 4848 xim_log("im_commit_cb(): %s\n", str); 4849 #endif 4850 4851 /* The imhangul module doesn't reset the preedit string before 4852 * committing. Call im_delete_preedit() to work around that. */ 4853 im_delete_preedit(); 4854 4855 /* Indicate that preediting has finished. */ 4856 if (preedit_start_col == MAXCOL) 4857 { 4858 init_preedit_start_col(); 4859 commit_with_preedit = FALSE; 4860 } 4861 4862 /* The thing which setting "preedit_start_col" to MAXCOL means that 4863 * "preedit_start_col" will be set forcedly when calling 4864 * preedit_changed_cb() next time. 4865 * "preedit_start_col" should not reset with MAXCOL on this part. Vim 4866 * is simulating the preediting by using add_to_input_str(). when 4867 * preedit begin immediately before committed, the typebuf is not 4868 * flushed to screen, then it can't get correct "preedit_start_col". 4869 * Thus, it should calculate the cells by adding cells of the committed 4870 * string. */ 4871 if (input_conv.vc_type != CONV_NONE) 4872 { 4873 im_str = string_convert(&input_conv, (char_u *)str, &len); 4874 g_return_if_fail(im_str != NULL); 4875 } 4876 else 4877 im_str = (char_u *)str; 4878 4879 clen = mb_string2cells(im_str, len); 4880 4881 if (input_conv.vc_type != CONV_NONE) 4882 vim_free(im_str); 4883 preedit_start_col += clen; 4884 4885 /* Is this a single character that matches a keypad key that's just 4886 * been pressed? If so, we don't want it to be entered as such - let 4887 * us carry on processing the raw keycode so that it may be used in 4888 * mappings as <kSomething>. */ 4889 if (xim_expected_char != NUL) 4890 { 4891 /* We're currently processing a keypad or other special key */ 4892 if (slen == 1 && str[0] == xim_expected_char) 4893 { 4894 /* It's a match - don't do it here */ 4895 xim_ignored_char = TRUE; 4896 add_to_input = FALSE; 4897 } 4898 else 4899 { 4900 /* Not a match */ 4901 xim_ignored_char = FALSE; 4902 } 4903 } 4904 4905 if (add_to_input) 4906 im_add_to_input((char_u *)str, slen); 4907 4908 /* Inserting chars while "im_is_active" is set does not cause a change of 4909 * buffer. When the chars are committed the buffer must be marked as 4910 * changed. */ 4911 if (!commit_with_preedit) 4912 preedit_start_col = MAXCOL; 4913 4914 /* This flag is used in changed() at next call. */ 4915 xim_changed_while_preediting = TRUE; 4916 4917 if (gtk_main_level() > 0) 4918 gtk_main_quit(); 4919 } 4920 4921 /* 4922 * Callback invoked after start to the preedit. 4923 */ 4924 static void 4925 im_preedit_start_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4926 { 4927 #ifdef XIM_DEBUG 4928 xim_log("im_preedit_start_cb()\n"); 4929 #endif 4930 4931 im_is_active = TRUE; 4932 preedit_is_active = TRUE; 4933 gui_update_cursor(TRUE, FALSE); 4934 im_show_info(); 4935 } 4936 4937 /* 4938 * Callback invoked after end to the preedit. 4939 */ 4940 static void 4941 im_preedit_end_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4942 { 4943 #ifdef XIM_DEBUG 4944 xim_log("im_preedit_end_cb()\n"); 4945 #endif 4946 im_delete_preedit(); 4947 4948 /* Indicate that preediting has finished */ 4949 preedit_start_col = MAXCOL; 4950 xim_has_preediting = FALSE; 4951 4952 #if 0 4953 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was 4954 * switched off unintentionally. We now use preedit_is_active (added by 4955 * SungHyun Nam). */ 4956 im_is_active = FALSE; 4957 #endif 4958 preedit_is_active = FALSE; 4959 gui_update_cursor(TRUE, FALSE); 4960 im_show_info(); 4961 } 4962 4963 /* 4964 * Callback invoked after changes to the preedit string. If the preedit 4965 * string was empty before, remember the preedit start column so we know 4966 * where to apply feedback attributes. Delete the previous preedit string 4967 * if there was one, save the new preedit cursor offset, and put the new 4968 * string into the input buffer. 4969 * 4970 * TODO: The pragmatic "put into input buffer" approach used here has 4971 * several fundamental problems: 4972 * 4973 * - The characters in the preedit string are subject to remapping. 4974 * That's broken, only the finally committed string should be remapped. 4975 * 4976 * - There is a race condition involved: The retrieved value for the 4977 * current cursor position will be wrong if any unprocessed characters 4978 * are still queued in the input buffer. 4979 * 4980 * - Due to the lack of synchronization between the file buffer in memory 4981 * and any typed characters, it's practically impossible to implement the 4982 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM 4983 * modules for languages such as Thai are likely to rely on this feature 4984 * for proper operation. 4985 * 4986 * Conclusions: I think support for preediting needs to be moved to the 4987 * core parts of Vim. Ideally, until it has been committed, the preediting 4988 * string should only be displayed and not affect the buffer content at all. 4989 * The question how to deal with the synchronization issue still remains. 4990 * Circumventing the input buffer is probably not desirable. Anyway, I think 4991 * implementing "retrieve_surrounding" is the only hard problem. 4992 * 4993 * One way to solve all of this in a clean manner would be to queue all key 4994 * press/release events "as is" in the input buffer, and apply the IM filtering 4995 * at the receiving end of the queue. This, however, would have a rather large 4996 * impact on the code base. If there is an easy way to force processing of all 4997 * remaining input from within the "retrieve_surrounding" signal handler, this 4998 * might not be necessary. Gotta ask on vim-dev for opinions. 4999 */ 5000 static void 5001 im_preedit_changed_cb(GtkIMContext *context, gpointer data UNUSED) 5002 { 5003 char *preedit_string = NULL; 5004 int cursor_index = 0; 5005 int num_move_back = 0; 5006 char_u *str; 5007 char_u *p; 5008 int i; 5009 5010 gtk_im_context_get_preedit_string(context, 5011 &preedit_string, NULL, 5012 &cursor_index); 5013 5014 #ifdef XIM_DEBUG 5015 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); 5016 #endif 5017 5018 g_return_if_fail(preedit_string != NULL); /* just in case */ 5019 5020 /* If preedit_start_col is MAXCOL set it to the current cursor position. */ 5021 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') 5022 { 5023 xim_has_preediting = TRUE; 5024 5025 /* Urgh, this breaks if the input buffer isn't empty now */ 5026 init_preedit_start_col(); 5027 } 5028 else if (cursor_index == 0 && preedit_string[0] == '\0') 5029 { 5030 xim_has_preediting = FALSE; 5031 5032 /* If at the start position (after typing backspace) 5033 * preedit_start_col must be reset. */ 5034 preedit_start_col = MAXCOL; 5035 } 5036 5037 im_delete_preedit(); 5038 5039 /* 5040 * Compute the end of the preediting area: "preedit_end_col". 5041 * According to the documentation of gtk_im_context_get_preedit_string(), 5042 * the cursor_pos output argument returns the offset in bytes. This is 5043 * unfortunately not true -- real life shows the offset is in characters, 5044 * and the GTK+ source code agrees with me. Will file a bug later. 5045 */ 5046 if (preedit_start_col != MAXCOL) 5047 preedit_end_col = preedit_start_col; 5048 str = (char_u *)preedit_string; 5049 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) 5050 { 5051 int is_composing; 5052 5053 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); 5054 /* 5055 * These offsets are used as counters when generating <BS> and <Del> 5056 * to delete the preedit string. So don't count composing characters 5057 * unless 'delcombine' is enabled. 5058 */ 5059 if (!is_composing || p_deco) 5060 { 5061 if (i < cursor_index) 5062 ++im_preedit_cursor; 5063 else 5064 ++im_preedit_trailing; 5065 } 5066 if (!is_composing && i >= cursor_index) 5067 { 5068 /* This is essentially the same as im_preedit_trailing, except 5069 * composing characters are not counted even if p_deco is set. */ 5070 ++num_move_back; 5071 } 5072 if (preedit_start_col != MAXCOL) 5073 preedit_end_col += utf_ptr2cells(p); 5074 } 5075 5076 if (p > str) 5077 { 5078 im_add_to_input(str, (int)(p - str)); 5079 im_correct_cursor(num_move_back); 5080 } 5081 5082 g_free(preedit_string); 5083 5084 if (gtk_main_level() > 0) 5085 gtk_main_quit(); 5086 } 5087 5088 /* 5089 * Translate the Pango attributes at iter to Vim highlighting attributes. 5090 * Ignore attributes not supported by Vim highlighting. This shouldn't have 5091 * too much impact -- right now we handle even more attributes than necessary 5092 * for the IM modules I tested with. 5093 */ 5094 static int 5095 translate_pango_attributes(PangoAttrIterator *iter) 5096 { 5097 PangoAttribute *attr; 5098 int char_attr = HL_NORMAL; 5099 5100 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); 5101 if (attr != NULL && ((PangoAttrInt *)attr)->value 5102 != (int)PANGO_UNDERLINE_NONE) 5103 char_attr |= HL_UNDERLINE; 5104 5105 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); 5106 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) 5107 char_attr |= HL_BOLD; 5108 5109 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); 5110 if (attr != NULL && ((PangoAttrInt *)attr)->value 5111 != (int)PANGO_STYLE_NORMAL) 5112 char_attr |= HL_ITALIC; 5113 5114 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); 5115 if (attr != NULL) 5116 { 5117 const PangoColor *color = &((PangoAttrColor *)attr)->color; 5118 5119 /* Assume inverse if black background is requested */ 5120 if ((color->red | color->green | color->blue) == 0) 5121 char_attr |= HL_INVERSE; 5122 } 5123 5124 return char_attr; 5125 } 5126 5127 /* 5128 * Retrieve the highlighting attributes at column col in the preedit string. 5129 * Return -1 if not in preediting mode or if col is out of range. 5130 */ 5131 int 5132 im_get_feedback_attr(int col) 5133 { 5134 char *preedit_string = NULL; 5135 PangoAttrList *attr_list = NULL; 5136 int char_attr = -1; 5137 5138 if (xic == NULL) 5139 return char_attr; 5140 5141 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5142 5143 if (preedit_string != NULL && attr_list != NULL) 5144 { 5145 int idx; 5146 5147 /* Get the byte index as used by PangoAttrIterator */ 5148 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col) 5149 idx += utfc_ptr2len((char_u *)preedit_string + idx); 5150 5151 if (preedit_string[idx] != '\0') 5152 { 5153 PangoAttrIterator *iter; 5154 int start, end; 5155 5156 char_attr = HL_NORMAL; 5157 iter = pango_attr_list_get_iterator(attr_list); 5158 5159 /* Extract all relevant attributes from the list. */ 5160 do 5161 { 5162 pango_attr_iterator_range(iter, &start, &end); 5163 5164 if (idx >= start && idx < end) 5165 char_attr |= translate_pango_attributes(iter); 5166 } 5167 while (pango_attr_iterator_next(iter)); 5168 5169 pango_attr_iterator_destroy(iter); 5170 } 5171 } 5172 5173 if (attr_list != NULL) 5174 pango_attr_list_unref(attr_list); 5175 g_free(preedit_string); 5176 5177 return char_attr; 5178 } 5179 5180 void 5181 xim_init(void) 5182 { 5183 #ifdef XIM_DEBUG 5184 xim_log("xim_init()\n"); 5185 #endif 5186 5187 g_return_if_fail(gui.drawarea != NULL); 5188 #if GTK_CHECK_VERSION(3,0,0) 5189 g_return_if_fail(gtk_widget_get_window(gui.drawarea) != NULL); 5190 #else 5191 g_return_if_fail(gui.drawarea->window != NULL); 5192 #endif 5193 5194 xic = gtk_im_multicontext_new(); 5195 g_object_ref(xic); 5196 5197 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", 5198 G_CALLBACK(&im_commit_cb), NULL); 5199 g_signal_connect(G_OBJECT(xic), "preedit_changed", 5200 G_CALLBACK(&im_preedit_changed_cb), NULL); 5201 g_signal_connect(G_OBJECT(xic), "preedit_start", 5202 G_CALLBACK(&im_preedit_start_cb), NULL); 5203 g_signal_connect(G_OBJECT(xic), "preedit_end", 5204 G_CALLBACK(&im_preedit_end_cb), NULL); 5205 5206 #if GTK_CHECK_VERSION(3,0,0) 5207 gtk_im_context_set_client_window(xic, gtk_widget_get_window(gui.drawarea)); 5208 #else 5209 gtk_im_context_set_client_window(xic, gui.drawarea->window); 5210 #endif 5211 } 5212 5213 void 5214 im_shutdown(void) 5215 { 5216 #ifdef XIM_DEBUG 5217 xim_log("im_shutdown()\n"); 5218 #endif 5219 5220 if (xic != NULL) 5221 { 5222 gtk_im_context_focus_out(xic); 5223 g_object_unref(xic); 5224 xic = NULL; 5225 } 5226 im_is_active = FALSE; 5227 im_commit_handler_id = 0; 5228 preedit_start_col = MAXCOL; 5229 xim_has_preediting = FALSE; 5230 } 5231 5232 /* 5233 * Convert the string argument to keyval and state for GdkEventKey. 5234 * If str is valid return TRUE, otherwise FALSE. 5235 * 5236 * See 'imactivatekey' for documentation of the format. 5237 */ 5238 static int 5239 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) 5240 { 5241 const char *mods_end; 5242 unsigned tmp_keyval; 5243 unsigned tmp_state = 0; 5244 5245 mods_end = strrchr(str, '-'); 5246 mods_end = (mods_end != NULL) ? mods_end + 1 : str; 5247 5248 /* Parse modifier keys */ 5249 while (str < mods_end) 5250 switch (*str++) 5251 { 5252 case '-': break; 5253 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; 5254 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; 5255 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; 5256 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; 5257 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; 5258 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; 5259 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; 5260 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; 5261 default: 5262 return FALSE; 5263 } 5264 5265 tmp_keyval = gdk_keyval_from_name(str); 5266 5267 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) 5268 return FALSE; 5269 5270 if (keyval != NULL) 5271 *keyval = tmp_keyval; 5272 if (state != NULL) 5273 *state = tmp_state; 5274 5275 return TRUE; 5276 } 5277 5278 /* 5279 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an 5280 * empty string is also regarded as valid. 5281 * 5282 * Note: The numerical key value of p_imak is cached if it was valid; thus 5283 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever 5284 * 'imak' changes. This is currently the case but not obvious -- should 5285 * probably rename the function for clarity. 5286 */ 5287 int 5288 im_xim_isvalid_imactivate(void) 5289 { 5290 if (p_imak[0] == NUL) 5291 { 5292 im_activatekey_keyval = GDK_VoidSymbol; 5293 im_activatekey_state = 0; 5294 return TRUE; 5295 } 5296 5297 return im_string_to_keyval((const char *)p_imak, 5298 &im_activatekey_keyval, 5299 &im_activatekey_state); 5300 } 5301 5302 static void 5303 im_synthesize_keypress(unsigned int keyval, unsigned int state) 5304 { 5305 GdkEventKey *event; 5306 5307 # ifdef HAVE_GTK_MULTIHEAD 5308 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); 5309 # if GTK_CHECK_VERSION(3,0,0) 5310 g_object_ref(gtk_widget_get_window(gui.drawarea)); 5311 /* unreffed by gdk_event_free() */ 5312 # else 5313 g_object_ref(gui.drawarea->window); /* unreffed by gdk_event_free() */ 5314 # endif 5315 # else 5316 event = (GdkEventKey *)g_malloc0((gulong)sizeof(GdkEvent)); 5317 event->type = GDK_KEY_PRESS; 5318 # endif 5319 # if GTK_CHECK_VERSION(3,0,0) 5320 event->window = gtk_widget_get_window(gui.drawarea); 5321 # else 5322 event->window = gui.drawarea->window; 5323 # endif 5324 event->send_event = TRUE; 5325 event->time = GDK_CURRENT_TIME; 5326 event->state = state; 5327 event->keyval = keyval; 5328 event->hardware_keycode = /* needed for XIM */ 5329 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); 5330 event->length = 0; 5331 event->string = NULL; 5332 5333 gtk_im_context_filter_keypress(xic, event); 5334 5335 /* For consistency, also send the corresponding release event. */ 5336 event->type = GDK_KEY_RELEASE; 5337 event->send_event = FALSE; 5338 gtk_im_context_filter_keypress(xic, event); 5339 5340 # ifdef HAVE_GTK_MULTIHEAD 5341 gdk_event_free((GdkEvent *)event); 5342 # else 5343 g_free(event); 5344 # endif 5345 } 5346 5347 void 5348 xim_reset(void) 5349 { 5350 if (xic != NULL) 5351 { 5352 gtk_im_context_reset(xic); 5353 5354 if (p_imdisable) 5355 im_shutdown(); 5356 else 5357 { 5358 xim_set_focus(gui.in_focus); 5359 5360 # ifdef FEAT_EVAL 5361 if (p_imaf[0] != NUL) 5362 { 5363 char_u *argv[1]; 5364 5365 if (im_is_active) 5366 argv[0] = (char_u *)"1"; 5367 else 5368 argv[0] = (char_u *)"0"; 5369 (void)call_func_retnr(p_imaf, 1, argv, FALSE); 5370 } 5371 else 5372 # endif 5373 if (im_activatekey_keyval != GDK_VoidSymbol) 5374 { 5375 if (im_is_active) 5376 { 5377 g_signal_handler_block(xic, im_commit_handler_id); 5378 im_synthesize_keypress(im_activatekey_keyval, 5379 im_activatekey_state); 5380 g_signal_handler_unblock(xic, im_commit_handler_id); 5381 } 5382 } 5383 else 5384 { 5385 im_shutdown(); 5386 xim_init(); 5387 xim_set_focus(gui.in_focus); 5388 } 5389 } 5390 } 5391 5392 preedit_start_col = MAXCOL; 5393 xim_has_preediting = FALSE; 5394 } 5395 5396 int 5397 xim_queue_key_press_event(GdkEventKey *event, int down) 5398 { 5399 if (down) 5400 { 5401 /* 5402 * Workaround GTK2 XIM 'feature' that always converts keypad keys to 5403 * chars., even when not part of an IM sequence (ref. feature of 5404 * gdk/gdkkeyuni.c). 5405 * Flag any keypad keys that might represent a single char. 5406 * If this (on its own - i.e., not part of an IM sequence) is 5407 * committed while we're processing one of these keys, we can ignore 5408 * that commit and go ahead & process it ourselves. That way we can 5409 * still distinguish keypad keys for use in mappings. 5410 * Also add GDK_space to make <S-Space> work. 5411 */ 5412 switch (event->keyval) 5413 { 5414 case GDK_KP_Add: xim_expected_char = '+'; break; 5415 case GDK_KP_Subtract: xim_expected_char = '-'; break; 5416 case GDK_KP_Divide: xim_expected_char = '/'; break; 5417 case GDK_KP_Multiply: xim_expected_char = '*'; break; 5418 case GDK_KP_Decimal: xim_expected_char = '.'; break; 5419 case GDK_KP_Equal: xim_expected_char = '='; break; 5420 case GDK_KP_0: xim_expected_char = '0'; break; 5421 case GDK_KP_1: xim_expected_char = '1'; break; 5422 case GDK_KP_2: xim_expected_char = '2'; break; 5423 case GDK_KP_3: xim_expected_char = '3'; break; 5424 case GDK_KP_4: xim_expected_char = '4'; break; 5425 case GDK_KP_5: xim_expected_char = '5'; break; 5426 case GDK_KP_6: xim_expected_char = '6'; break; 5427 case GDK_KP_7: xim_expected_char = '7'; break; 5428 case GDK_KP_8: xim_expected_char = '8'; break; 5429 case GDK_KP_9: xim_expected_char = '9'; break; 5430 case GDK_space: xim_expected_char = ' '; break; 5431 default: xim_expected_char = NUL; 5432 } 5433 xim_ignored_char = FALSE; 5434 } 5435 5436 /* 5437 * When typing fFtT, XIM may be activated. Thus it must pass 5438 * gtk_im_context_filter_keypress() in Normal mode. 5439 * And while doing :sh too. 5440 */ 5441 if (xic != NULL && !p_imdisable 5442 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) 5443 { 5444 /* 5445 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is 5446 * always aware of the current status of IM, and can even emulate 5447 * the activation key for modules that don't support one. 5448 */ 5449 if (event->keyval == im_activatekey_keyval 5450 && (event->state & im_activatekey_state) == im_activatekey_state) 5451 { 5452 unsigned int state_mask; 5453 5454 /* Require the state of the 3 most used modifiers to match exactly. 5455 * Otherwise e.g. <S-C-space> would be unusable for other purposes 5456 * if the IM activate key is <S-space>. */ 5457 state_mask = im_activatekey_state; 5458 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK 5459 | (int)GDK_MOD1_MASK); 5460 5461 if ((event->state & state_mask) != im_activatekey_state) 5462 return FALSE; 5463 5464 /* Don't send it a second time on GDK_KEY_RELEASE. */ 5465 if (event->type != GDK_KEY_PRESS) 5466 return TRUE; 5467 5468 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE)) 5469 { 5470 im_set_active(FALSE); 5471 5472 /* ":lmap" mappings exists, toggle use of mappings. */ 5473 State ^= LANGMAP; 5474 if (State & LANGMAP) 5475 { 5476 curbuf->b_p_iminsert = B_IMODE_NONE; 5477 State &= ~LANGMAP; 5478 } 5479 else 5480 { 5481 curbuf->b_p_iminsert = B_IMODE_LMAP; 5482 State |= LANGMAP; 5483 } 5484 return TRUE; 5485 } 5486 5487 return gtk_im_context_filter_keypress(xic, event); 5488 } 5489 5490 /* Don't filter events through the IM context if IM isn't active 5491 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module 5492 * not doing anything before the activation key was sent. */ 5493 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) 5494 { 5495 int imresult = gtk_im_context_filter_keypress(xic, event); 5496 5497 /* Some XIM send following sequence: 5498 * 1. preedited string. 5499 * 2. committed string. 5500 * 3. line changed key. 5501 * 4. preedited string. 5502 * 5. remove preedited string. 5503 * if 3, Vim can't move back the above line for 5. 5504 * thus, this part should not parse the key. */ 5505 if (!imresult && preedit_start_col != MAXCOL 5506 && event->keyval == GDK_Return) 5507 { 5508 im_synthesize_keypress(GDK_Return, 0U); 5509 return FALSE; 5510 } 5511 5512 /* If XIM tried to commit a keypad key as a single char., 5513 * ignore it so we can use the keypad key 'raw', for mappings. */ 5514 if (xim_expected_char != NUL && xim_ignored_char) 5515 /* We had a keypad key, and XIM tried to thieve it */ 5516 return FALSE; 5517 5518 /* This is supposed to fix a problem with iBus, that space 5519 * characters don't work in input mode. */ 5520 xim_expected_char = NUL; 5521 5522 /* Normal processing */ 5523 return imresult; 5524 } 5525 } 5526 5527 return FALSE; 5528 } 5529 5530 int 5531 im_get_status(void) 5532 { 5533 # ifdef FEAT_EVAL 5534 if (p_imsf[0] != NUL) 5535 { 5536 int is_active; 5537 5538 /* FIXME: Don't execute user function in unsafe situation. */ 5539 if (exiting 5540 # ifdef FEAT_AUTOCMD 5541 || is_autocmd_blocked() 5542 # endif 5543 ) 5544 return FALSE; 5545 /* FIXME: :py print 'xxx' is shown duplicate result. 5546 * Use silent to avoid it. */ 5547 ++msg_silent; 5548 is_active = call_func_retnr(p_imsf, 0, NULL, FALSE); 5549 --msg_silent; 5550 return (is_active > 0); 5551 } 5552 # endif 5553 return im_is_active; 5554 } 5555 5556 int 5557 preedit_get_status(void) 5558 { 5559 return preedit_is_active; 5560 } 5561 5562 int 5563 im_is_preediting(void) 5564 { 5565 return xim_has_preediting; 5566 } 5567 5568 # else /* !FEAT_GUI_GTK */ 5569 5570 static int xim_is_active = FALSE; /* XIM should be active in the current 5571 mode */ 5572 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */ 5573 #ifdef FEAT_GUI_X11 5574 static XIMStyle input_style; 5575 static int status_area_enabled = TRUE; 5576 #endif 5577 5578 /* 5579 * Switch using XIM on/off. This is used by the code that changes "State". 5580 */ 5581 void 5582 im_set_active(int active) 5583 { 5584 if (xic == NULL) 5585 return; 5586 5587 /* If 'imdisable' is set, XIM is never active. */ 5588 if (p_imdisable) 5589 active = FALSE; 5590 #if !defined(FEAT_GUI_GTK) 5591 else if (input_style & XIMPreeditPosition) 5592 /* There is a problem in switching XIM off when preediting is used, 5593 * and it is not clear how this can be solved. For now, keep XIM on 5594 * all the time, like it was done in Vim 5.8. */ 5595 active = TRUE; 5596 #endif 5597 5598 /* Remember the active state, it is needed when Vim gets keyboard focus. */ 5599 xim_is_active = active; 5600 xim_set_preedit(); 5601 } 5602 5603 /* 5604 * Adjust using XIM for gaining or losing keyboard focus. Also called when 5605 * "xim_is_active" changes. 5606 */ 5607 void 5608 xim_set_focus(int focus) 5609 { 5610 if (xic == NULL) 5611 return; 5612 5613 /* 5614 * XIM only gets focus when the Vim window has keyboard focus and XIM has 5615 * been set active for the current mode. 5616 */ 5617 if (focus && xim_is_active) 5618 { 5619 if (!xim_has_focus) 5620 { 5621 xim_has_focus = TRUE; 5622 XSetICFocus(xic); 5623 } 5624 } 5625 else 5626 { 5627 if (xim_has_focus) 5628 { 5629 xim_has_focus = FALSE; 5630 XUnsetICFocus(xic); 5631 } 5632 } 5633 } 5634 5635 void 5636 im_set_position(int row UNUSED, int col UNUSED) 5637 { 5638 xim_set_preedit(); 5639 } 5640 5641 /* 5642 * Set the XIM to the current cursor position. 5643 */ 5644 void 5645 xim_set_preedit(void) 5646 { 5647 XVaNestedList attr_list; 5648 XRectangle spot_area; 5649 XPoint over_spot; 5650 int line_space; 5651 5652 if (xic == NULL) 5653 return; 5654 5655 xim_set_focus(TRUE); 5656 5657 if (!xim_has_focus) 5658 { 5659 /* hide XIM cursor */ 5660 over_spot.x = 0; 5661 over_spot.y = -100; /* arbitrary invisible position */ 5662 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5663 XNSpotLocation, 5664 &over_spot, 5665 NULL); 5666 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); 5667 XFree(attr_list); 5668 return; 5669 } 5670 5671 if (input_style & XIMPreeditPosition) 5672 { 5673 if (xim_fg_color == INVALCOLOR) 5674 { 5675 xim_fg_color = gui.def_norm_pixel; 5676 xim_bg_color = gui.def_back_pixel; 5677 } 5678 over_spot.x = TEXT_X(gui.col); 5679 over_spot.y = TEXT_Y(gui.row); 5680 spot_area.x = 0; 5681 spot_area.y = 0; 5682 spot_area.height = gui.char_height * Rows; 5683 spot_area.width = gui.char_width * Columns; 5684 line_space = gui.char_height; 5685 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5686 XNSpotLocation, &over_spot, 5687 XNForeground, (Pixel) xim_fg_color, 5688 XNBackground, (Pixel) xim_bg_color, 5689 XNArea, &spot_area, 5690 XNLineSpace, line_space, 5691 NULL); 5692 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) 5693 EMSG(_("E284: Cannot set IC values")); 5694 XFree(attr_list); 5695 } 5696 } 5697 5698 #if defined(FEAT_GUI_X11) 5699 static char e_xim[] = N_("E285: Failed to create input context"); 5700 #endif 5701 5702 #if defined(FEAT_GUI_X11) || defined(PROTO) 5703 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(SUN_SYSTEM) 5704 # define USE_X11R6_XIM 5705 # endif 5706 5707 static int xim_real_init(Window x11_window, Display *x11_display); 5708 5709 5710 #ifdef USE_X11R6_XIM 5711 static void xim_destroy_cb(XIM im, XPointer client_data, XPointer call_data); 5712 5713 static void 5714 xim_instantiate_cb( 5715 Display *display, 5716 XPointer client_data UNUSED, 5717 XPointer call_data UNUSED) 5718 { 5719 Window x11_window; 5720 Display *x11_display; 5721 5722 #ifdef XIM_DEBUG 5723 xim_log("xim_instantiate_cb()\n"); 5724 #endif 5725 5726 gui_get_x11_windis(&x11_window, &x11_display); 5727 if (display != x11_display) 5728 return; 5729 5730 xim_real_init(x11_window, x11_display); 5731 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5732 if (xic != NULL) 5733 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5734 xim_instantiate_cb, NULL); 5735 } 5736 5737 static void 5738 xim_destroy_cb( 5739 XIM im UNUSED, 5740 XPointer client_data UNUSED, 5741 XPointer call_data UNUSED) 5742 { 5743 Window x11_window; 5744 Display *x11_display; 5745 5746 #ifdef XIM_DEBUG 5747 xim_log("xim_destroy_cb()\n"); 5748 #endif 5749 gui_get_x11_windis(&x11_window, &x11_display); 5750 5751 xic = NULL; 5752 status_area_enabled = FALSE; 5753 5754 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5755 5756 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5757 xim_instantiate_cb, NULL); 5758 } 5759 #endif 5760 5761 void 5762 xim_init(void) 5763 { 5764 Window x11_window; 5765 Display *x11_display; 5766 5767 #ifdef XIM_DEBUG 5768 xim_log("xim_init()\n"); 5769 #endif 5770 5771 gui_get_x11_windis(&x11_window, &x11_display); 5772 5773 xic = NULL; 5774 5775 if (xim_real_init(x11_window, x11_display)) 5776 return; 5777 5778 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5779 5780 #ifdef USE_X11R6_XIM 5781 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5782 xim_instantiate_cb, NULL); 5783 #endif 5784 } 5785 5786 static int 5787 xim_real_init(Window x11_window, Display *x11_display) 5788 { 5789 int i; 5790 char *p, 5791 *s, 5792 *ns, 5793 *end, 5794 tmp[1024]; 5795 #define IMLEN_MAX 40 5796 char buf[IMLEN_MAX + 7]; 5797 XIM xim = NULL; 5798 XIMStyles *xim_styles; 5799 XIMStyle this_input_style = 0; 5800 Boolean found; 5801 XPoint over_spot; 5802 XVaNestedList preedit_list, status_list; 5803 5804 input_style = 0; 5805 status_area_enabled = FALSE; 5806 5807 if (xic != NULL) 5808 return FALSE; 5809 5810 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) 5811 { 5812 strcpy(tmp, gui.rsrc_input_method); 5813 for (ns = s = tmp; ns != NULL && *s != NUL;) 5814 { 5815 s = (char *)skipwhite((char_u *)s); 5816 if (*s == NUL) 5817 break; 5818 if ((ns = end = strchr(s, ',')) == NULL) 5819 end = s + strlen(s); 5820 while (isspace(((char_u *)end)[-1])) 5821 end--; 5822 *end = NUL; 5823 5824 if (strlen(s) <= IMLEN_MAX) 5825 { 5826 strcpy(buf, "@im="); 5827 strcat(buf, s); 5828 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL 5829 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) 5830 != NULL) 5831 break; 5832 } 5833 5834 s = ns + 1; 5835 } 5836 } 5837 5838 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) 5839 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5840 5841 /* This is supposed to be useful to obtain characters through 5842 * XmbLookupString() without really using a XIM. */ 5843 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL 5844 && *p != NUL) 5845 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5846 5847 if (xim == NULL) 5848 { 5849 /* Only give this message when verbose is set, because too many people 5850 * got this message when they didn't want to use a XIM. */ 5851 if (p_verbose > 0) 5852 { 5853 verbose_enter(); 5854 EMSG(_("E286: Failed to open input method")); 5855 verbose_leave(); 5856 } 5857 return FALSE; 5858 } 5859 5860 #ifdef USE_X11R6_XIM 5861 { 5862 XIMCallback destroy_cb; 5863 5864 destroy_cb.callback = xim_destroy_cb; 5865 destroy_cb.client_data = NULL; 5866 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) 5867 EMSG(_("E287: Warning: Could not set destroy callback to IM")); 5868 } 5869 #endif 5870 5871 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) 5872 { 5873 EMSG(_("E288: input method doesn't support any style")); 5874 XCloseIM(xim); 5875 return FALSE; 5876 } 5877 5878 found = False; 5879 strcpy(tmp, gui.rsrc_preedit_type_name); 5880 for (s = tmp; s && !found; ) 5881 { 5882 while (*s && isspace((unsigned char)*s)) 5883 s++; 5884 if (!*s) 5885 break; 5886 if ((ns = end = strchr(s, ',')) != 0) 5887 ns++; 5888 else 5889 end = s + strlen(s); 5890 while (isspace((unsigned char)*end)) 5891 end--; 5892 *end = '\0'; 5893 5894 if (!strcmp(s, "OverTheSpot")) 5895 this_input_style = (XIMPreeditPosition | XIMStatusArea); 5896 else if (!strcmp(s, "OffTheSpot")) 5897 this_input_style = (XIMPreeditArea | XIMStatusArea); 5898 else if (!strcmp(s, "Root")) 5899 this_input_style = (XIMPreeditNothing | XIMStatusNothing); 5900 5901 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5902 { 5903 if (this_input_style == xim_styles->supported_styles[i]) 5904 { 5905 found = True; 5906 break; 5907 } 5908 } 5909 if (!found) 5910 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5911 { 5912 if ((xim_styles->supported_styles[i] & this_input_style) 5913 == (this_input_style & ~XIMStatusArea)) 5914 { 5915 this_input_style &= ~XIMStatusArea; 5916 found = True; 5917 break; 5918 } 5919 } 5920 5921 s = ns; 5922 } 5923 XFree(xim_styles); 5924 5925 if (!found) 5926 { 5927 /* Only give this message when verbose is set, because too many people 5928 * got this message when they didn't want to use a XIM. */ 5929 if (p_verbose > 0) 5930 { 5931 verbose_enter(); 5932 EMSG(_("E289: input method doesn't support my preedit type")); 5933 verbose_leave(); 5934 } 5935 XCloseIM(xim); 5936 return FALSE; 5937 } 5938 5939 over_spot.x = TEXT_X(gui.col); 5940 over_spot.y = TEXT_Y(gui.row); 5941 input_style = this_input_style; 5942 5943 /* A crash was reported when trying to pass gui.norm_font as XNFontSet, 5944 * thus that has been removed. Hopefully the default works... */ 5945 #ifdef FEAT_XFONTSET 5946 if (gui.fontset != NOFONTSET) 5947 { 5948 preedit_list = XVaCreateNestedList(0, 5949 XNSpotLocation, &over_spot, 5950 XNForeground, (Pixel)gui.def_norm_pixel, 5951 XNBackground, (Pixel)gui.def_back_pixel, 5952 XNFontSet, (XFontSet)gui.fontset, 5953 NULL); 5954 status_list = XVaCreateNestedList(0, 5955 XNForeground, (Pixel)gui.def_norm_pixel, 5956 XNBackground, (Pixel)gui.def_back_pixel, 5957 XNFontSet, (XFontSet)gui.fontset, 5958 NULL); 5959 } 5960 else 5961 #endif 5962 { 5963 preedit_list = XVaCreateNestedList(0, 5964 XNSpotLocation, &over_spot, 5965 XNForeground, (Pixel)gui.def_norm_pixel, 5966 XNBackground, (Pixel)gui.def_back_pixel, 5967 NULL); 5968 status_list = XVaCreateNestedList(0, 5969 XNForeground, (Pixel)gui.def_norm_pixel, 5970 XNBackground, (Pixel)gui.def_back_pixel, 5971 NULL); 5972 } 5973 5974 xic = XCreateIC(xim, 5975 XNInputStyle, input_style, 5976 XNClientWindow, x11_window, 5977 XNFocusWindow, gui.wid, 5978 XNPreeditAttributes, preedit_list, 5979 XNStatusAttributes, status_list, 5980 NULL); 5981 XFree(status_list); 5982 XFree(preedit_list); 5983 if (xic != NULL) 5984 { 5985 if (input_style & XIMStatusArea) 5986 { 5987 xim_set_status_area(); 5988 status_area_enabled = TRUE; 5989 } 5990 else 5991 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5992 } 5993 else 5994 { 5995 EMSG(_(e_xim)); 5996 XCloseIM(xim); 5997 return FALSE; 5998 } 5999 6000 return TRUE; 6001 } 6002 6003 #endif /* FEAT_GUI_X11 */ 6004 6005 /* 6006 * Get IM status. When IM is on, return TRUE. Else return FALSE. 6007 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is 6008 * active, when not having focus XIM may still be active (e.g., when using a 6009 * tear-off menu item). 6010 */ 6011 int 6012 im_get_status(void) 6013 { 6014 return xim_has_focus; 6015 } 6016 6017 # endif /* !FEAT_GUI_GTK */ 6018 6019 # if !defined(FEAT_GUI_GTK) || defined(PROTO) 6020 /* 6021 * Set up the status area. 6022 * 6023 * This should use a separate Widget, but that seems not possible, because 6024 * preedit_area and status_area should be set to the same window as for the 6025 * text input. Unfortunately this means the status area pollutes the text 6026 * window... 6027 */ 6028 void 6029 xim_set_status_area(void) 6030 { 6031 XVaNestedList preedit_list = 0, status_list = 0, list = 0; 6032 XRectangle pre_area, status_area; 6033 6034 if (xic == NULL) 6035 return; 6036 6037 if (input_style & XIMStatusArea) 6038 { 6039 if (input_style & XIMPreeditArea) 6040 { 6041 XRectangle *needed_rect; 6042 6043 /* to get status_area width */ 6044 status_list = XVaCreateNestedList(0, XNAreaNeeded, 6045 &needed_rect, NULL); 6046 XGetICValues(xic, XNStatusAttributes, status_list, NULL); 6047 XFree(status_list); 6048 6049 status_area.width = needed_rect->width; 6050 } 6051 else 6052 status_area.width = gui.char_width * Columns; 6053 6054 status_area.x = 0; 6055 status_area.y = gui.char_height * Rows + gui.border_offset; 6056 if (gui.which_scrollbars[SBAR_BOTTOM]) 6057 status_area.y += gui.scrollbar_height; 6058 #ifdef FEAT_MENU 6059 if (gui.menu_is_active) 6060 status_area.y += gui.menu_height; 6061 #endif 6062 status_area.height = gui.char_height; 6063 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); 6064 } 6065 else 6066 { 6067 status_area.x = 0; 6068 status_area.y = gui.char_height * Rows + gui.border_offset; 6069 if (gui.which_scrollbars[SBAR_BOTTOM]) 6070 status_area.y += gui.scrollbar_height; 6071 #ifdef FEAT_MENU 6072 if (gui.menu_is_active) 6073 status_area.y += gui.menu_height; 6074 #endif 6075 status_area.width = 0; 6076 status_area.height = gui.char_height; 6077 } 6078 6079 if (input_style & XIMPreeditArea) /* off-the-spot */ 6080 { 6081 pre_area.x = status_area.x + status_area.width; 6082 pre_area.y = gui.char_height * Rows + gui.border_offset; 6083 pre_area.width = gui.char_width * Columns - pre_area.x; 6084 if (gui.which_scrollbars[SBAR_BOTTOM]) 6085 pre_area.y += gui.scrollbar_height; 6086 #ifdef FEAT_MENU 6087 if (gui.menu_is_active) 6088 pre_area.y += gui.menu_height; 6089 #endif 6090 pre_area.height = gui.char_height; 6091 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6092 } 6093 else if (input_style & XIMPreeditPosition) /* over-the-spot */ 6094 { 6095 pre_area.x = 0; 6096 pre_area.y = 0; 6097 pre_area.height = gui.char_height * Rows; 6098 pre_area.width = gui.char_width * Columns; 6099 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6100 } 6101 6102 if (preedit_list && status_list) 6103 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6104 XNStatusAttributes, status_list, NULL); 6105 else if (preedit_list) 6106 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6107 NULL); 6108 else if (status_list) 6109 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, 6110 NULL); 6111 else 6112 list = NULL; 6113 6114 if (list) 6115 { 6116 XSetICValues(xic, XNVaNestedList, list, NULL); 6117 XFree(list); 6118 } 6119 if (status_list) 6120 XFree(status_list); 6121 if (preedit_list) 6122 XFree(preedit_list); 6123 } 6124 6125 int 6126 xim_get_status_area_height(void) 6127 { 6128 if (status_area_enabled) 6129 return gui.char_height; 6130 return 0; 6131 } 6132 # endif 6133 6134 #endif /* FEAT_XIM */ 6135 6136 #if defined(FEAT_MBYTE) || defined(PROTO) 6137 6138 /* 6139 * Setup "vcp" for conversion from "from" to "to". 6140 * The names must have been made canonical with enc_canonize(). 6141 * vcp->vc_type must have been initialized to CONV_NONE. 6142 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 6143 * instead). 6144 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 6145 * Return FAIL when conversion is not supported, OK otherwise. 6146 */ 6147 int 6148 convert_setup(vimconv_T *vcp, char_u *from, char_u *to) 6149 { 6150 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 6151 } 6152 6153 /* 6154 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 6155 * "from" unicode charsets be considered utf-8. Same for "to". 6156 */ 6157 int 6158 convert_setup_ext( 6159 vimconv_T *vcp, 6160 char_u *from, 6161 int from_unicode_is_utf8, 6162 char_u *to, 6163 int to_unicode_is_utf8) 6164 { 6165 int from_prop; 6166 int to_prop; 6167 int from_is_utf8; 6168 int to_is_utf8; 6169 6170 /* Reset to no conversion. */ 6171 # ifdef USE_ICONV 6172 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 6173 iconv_close(vcp->vc_fd); 6174 # endif 6175 vcp->vc_type = CONV_NONE; 6176 vcp->vc_factor = 1; 6177 vcp->vc_fail = FALSE; 6178 6179 /* No conversion when one of the names is empty or they are equal. */ 6180 if (from == NULL || *from == NUL || to == NULL || *to == NUL 6181 || STRCMP(from, to) == 0) 6182 return OK; 6183 6184 from_prop = enc_canon_props(from); 6185 to_prop = enc_canon_props(to); 6186 if (from_unicode_is_utf8) 6187 from_is_utf8 = from_prop & ENC_UNICODE; 6188 else 6189 from_is_utf8 = from_prop == ENC_UNICODE; 6190 if (to_unicode_is_utf8) 6191 to_is_utf8 = to_prop & ENC_UNICODE; 6192 else 6193 to_is_utf8 = to_prop == ENC_UNICODE; 6194 6195 if ((from_prop & ENC_LATIN1) && to_is_utf8) 6196 { 6197 /* Internal latin1 -> utf-8 conversion. */ 6198 vcp->vc_type = CONV_TO_UTF8; 6199 vcp->vc_factor = 2; /* up to twice as long */ 6200 } 6201 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 6202 { 6203 /* Internal latin9 -> utf-8 conversion. */ 6204 vcp->vc_type = CONV_9_TO_UTF8; 6205 vcp->vc_factor = 3; /* up to three as long (euro sign) */ 6206 } 6207 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 6208 { 6209 /* Internal utf-8 -> latin1 conversion. */ 6210 vcp->vc_type = CONV_TO_LATIN1; 6211 } 6212 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 6213 { 6214 /* Internal utf-8 -> latin9 conversion. */ 6215 vcp->vc_type = CONV_TO_LATIN9; 6216 } 6217 #ifdef WIN3264 6218 /* Win32-specific codepage <-> codepage conversion without iconv. */ 6219 else if ((from_is_utf8 || encname2codepage(from) > 0) 6220 && (to_is_utf8 || encname2codepage(to) > 0)) 6221 { 6222 vcp->vc_type = CONV_CODEPAGE; 6223 vcp->vc_factor = 2; /* up to twice as long */ 6224 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 6225 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 6226 } 6227 #endif 6228 #ifdef MACOS_X 6229 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 6230 { 6231 vcp->vc_type = CONV_MAC_LATIN1; 6232 } 6233 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 6234 { 6235 vcp->vc_type = CONV_MAC_UTF8; 6236 vcp->vc_factor = 2; /* up to twice as long */ 6237 } 6238 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 6239 { 6240 vcp->vc_type = CONV_LATIN1_MAC; 6241 } 6242 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 6243 { 6244 vcp->vc_type = CONV_UTF8_MAC; 6245 } 6246 #endif 6247 # ifdef USE_ICONV 6248 else 6249 { 6250 /* Use iconv() for conversion. */ 6251 vcp->vc_fd = (iconv_t)my_iconv_open( 6252 to_is_utf8 ? (char_u *)"utf-8" : to, 6253 from_is_utf8 ? (char_u *)"utf-8" : from); 6254 if (vcp->vc_fd != (iconv_t)-1) 6255 { 6256 vcp->vc_type = CONV_ICONV; 6257 vcp->vc_factor = 4; /* could be longer too... */ 6258 } 6259 } 6260 # endif 6261 if (vcp->vc_type == CONV_NONE) 6262 return FAIL; 6263 6264 return OK; 6265 } 6266 6267 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \ 6268 || defined(PROTO) 6269 /* 6270 * Do conversion on typed input characters in-place. 6271 * The input and output are not NUL terminated! 6272 * Returns the length after conversion. 6273 */ 6274 int 6275 convert_input(char_u *ptr, int len, int maxlen) 6276 { 6277 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 6278 } 6279 #endif 6280 6281 /* 6282 * Like convert_input(), but when there is an incomplete byte sequence at the 6283 * end return that as an allocated string in "restp" and set "*restlenp" to 6284 * the length. If "restp" is NULL it is not used. 6285 */ 6286 int 6287 convert_input_safe( 6288 char_u *ptr, 6289 int len, 6290 int maxlen, 6291 char_u **restp, 6292 int *restlenp) 6293 { 6294 char_u *d; 6295 int dlen = len; 6296 int unconvertlen = 0; 6297 6298 d = string_convert_ext(&input_conv, ptr, &dlen, 6299 restp == NULL ? NULL : &unconvertlen); 6300 if (d != NULL) 6301 { 6302 if (dlen <= maxlen) 6303 { 6304 if (unconvertlen > 0) 6305 { 6306 /* Move the unconverted characters to allocated memory. */ 6307 *restp = alloc(unconvertlen); 6308 if (*restp != NULL) 6309 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 6310 *restlenp = unconvertlen; 6311 } 6312 mch_memmove(ptr, d, dlen); 6313 } 6314 else 6315 /* result is too long, keep the unconverted text (the caller must 6316 * have done something wrong!) */ 6317 dlen = len; 6318 vim_free(d); 6319 } 6320 return dlen; 6321 } 6322 6323 /* 6324 * Convert text "ptr[*lenp]" according to "vcp". 6325 * Returns the result in allocated memory and sets "*lenp". 6326 * When "lenp" is NULL, use NUL terminated strings. 6327 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 6328 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 6329 */ 6330 char_u * 6331 string_convert( 6332 vimconv_T *vcp, 6333 char_u *ptr, 6334 int *lenp) 6335 { 6336 return string_convert_ext(vcp, ptr, lenp, NULL); 6337 } 6338 6339 /* 6340 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 6341 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 6342 * set to the number of remaining bytes. 6343 */ 6344 char_u * 6345 string_convert_ext( 6346 vimconv_T *vcp, 6347 char_u *ptr, 6348 int *lenp, 6349 int *unconvlenp) 6350 { 6351 char_u *retval = NULL; 6352 char_u *d; 6353 int len; 6354 int i; 6355 int l; 6356 int c; 6357 6358 if (lenp == NULL) 6359 len = (int)STRLEN(ptr); 6360 else 6361 len = *lenp; 6362 if (len == 0) 6363 return vim_strsave((char_u *)""); 6364 6365 switch (vcp->vc_type) 6366 { 6367 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */ 6368 retval = alloc(len * 2 + 1); 6369 if (retval == NULL) 6370 break; 6371 d = retval; 6372 for (i = 0; i < len; ++i) 6373 { 6374 c = ptr[i]; 6375 if (c < 0x80) 6376 *d++ = c; 6377 else 6378 { 6379 *d++ = 0xc0 + ((unsigned)c >> 6); 6380 *d++ = 0x80 + (c & 0x3f); 6381 } 6382 } 6383 *d = NUL; 6384 if (lenp != NULL) 6385 *lenp = (int)(d - retval); 6386 break; 6387 6388 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */ 6389 retval = alloc(len * 3 + 1); 6390 if (retval == NULL) 6391 break; 6392 d = retval; 6393 for (i = 0; i < len; ++i) 6394 { 6395 c = ptr[i]; 6396 switch (c) 6397 { 6398 case 0xa4: c = 0x20ac; break; /* euro */ 6399 case 0xa6: c = 0x0160; break; /* S hat */ 6400 case 0xa8: c = 0x0161; break; /* S -hat */ 6401 case 0xb4: c = 0x017d; break; /* Z hat */ 6402 case 0xb8: c = 0x017e; break; /* Z -hat */ 6403 case 0xbc: c = 0x0152; break; /* OE */ 6404 case 0xbd: c = 0x0153; break; /* oe */ 6405 case 0xbe: c = 0x0178; break; /* Y */ 6406 } 6407 d += utf_char2bytes(c, d); 6408 } 6409 *d = NUL; 6410 if (lenp != NULL) 6411 *lenp = (int)(d - retval); 6412 break; 6413 6414 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */ 6415 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */ 6416 retval = alloc(len + 1); 6417 if (retval == NULL) 6418 break; 6419 d = retval; 6420 for (i = 0; i < len; ++i) 6421 { 6422 l = utf_ptr2len_len(ptr + i, len - i); 6423 if (l == 0) 6424 *d++ = NUL; 6425 else if (l == 1) 6426 { 6427 int l_w = utf8len_tab_zero[ptr[i]]; 6428 6429 if (l_w == 0) 6430 { 6431 /* Illegal utf-8 byte cannot be converted */ 6432 vim_free(retval); 6433 return NULL; 6434 } 6435 if (unconvlenp != NULL && l_w > len - i) 6436 { 6437 /* Incomplete sequence at the end. */ 6438 *unconvlenp = len - i; 6439 break; 6440 } 6441 *d++ = ptr[i]; 6442 } 6443 else 6444 { 6445 c = utf_ptr2char(ptr + i); 6446 if (vcp->vc_type == CONV_TO_LATIN9) 6447 switch (c) 6448 { 6449 case 0x20ac: c = 0xa4; break; /* euro */ 6450 case 0x0160: c = 0xa6; break; /* S hat */ 6451 case 0x0161: c = 0xa8; break; /* S -hat */ 6452 case 0x017d: c = 0xb4; break; /* Z hat */ 6453 case 0x017e: c = 0xb8; break; /* Z -hat */ 6454 case 0x0152: c = 0xbc; break; /* OE */ 6455 case 0x0153: c = 0xbd; break; /* oe */ 6456 case 0x0178: c = 0xbe; break; /* Y */ 6457 case 0xa4: 6458 case 0xa6: 6459 case 0xa8: 6460 case 0xb4: 6461 case 0xb8: 6462 case 0xbc: 6463 case 0xbd: 6464 case 0xbe: c = 0x100; break; /* not in latin9 */ 6465 } 6466 if (!utf_iscomposing(c)) /* skip composing chars */ 6467 { 6468 if (c < 0x100) 6469 *d++ = c; 6470 else if (vcp->vc_fail) 6471 { 6472 vim_free(retval); 6473 return NULL; 6474 } 6475 else 6476 { 6477 *d++ = 0xbf; 6478 if (utf_char2cells(c) > 1) 6479 *d++ = '?'; 6480 } 6481 } 6482 i += l - 1; 6483 } 6484 } 6485 *d = NUL; 6486 if (lenp != NULL) 6487 *lenp = (int)(d - retval); 6488 break; 6489 6490 # ifdef MACOS_CONVERT 6491 case CONV_MAC_LATIN1: 6492 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6493 'm', 'l', unconvlenp); 6494 break; 6495 6496 case CONV_LATIN1_MAC: 6497 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6498 'l', 'm', unconvlenp); 6499 break; 6500 6501 case CONV_MAC_UTF8: 6502 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6503 'm', 'u', unconvlenp); 6504 break; 6505 6506 case CONV_UTF8_MAC: 6507 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6508 'u', 'm', unconvlenp); 6509 break; 6510 # endif 6511 6512 # ifdef USE_ICONV 6513 case CONV_ICONV: /* conversion with output_conv.vc_fd */ 6514 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 6515 break; 6516 # endif 6517 # ifdef WIN3264 6518 case CONV_CODEPAGE: /* codepage -> codepage */ 6519 { 6520 int retlen; 6521 int tmp_len; 6522 short_u *tmp; 6523 6524 /* 1. codepage/UTF-8 -> ucs-2. */ 6525 if (vcp->vc_cpfrom == 0) 6526 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 6527 else 6528 { 6529 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 6530 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 6531 (char *)ptr, len, 0, 0); 6532 if (tmp_len == 0 6533 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 6534 { 6535 if (lenp != NULL) 6536 *lenp = 0; 6537 if (unconvlenp != NULL) 6538 *unconvlenp = len; 6539 retval = alloc(1); 6540 if (retval) 6541 retval[0] = NUL; 6542 return retval; 6543 } 6544 } 6545 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len); 6546 if (tmp == NULL) 6547 break; 6548 if (vcp->vc_cpfrom == 0) 6549 utf8_to_utf16(ptr, len, tmp, unconvlenp); 6550 else 6551 MultiByteToWideChar(vcp->vc_cpfrom, 0, 6552 (char *)ptr, len, tmp, tmp_len); 6553 6554 /* 2. ucs-2 -> codepage/UTF-8. */ 6555 if (vcp->vc_cpto == 0) 6556 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 6557 else 6558 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 6559 tmp, tmp_len, 0, 0, 0, 0); 6560 retval = alloc(retlen + 1); 6561 if (retval != NULL) 6562 { 6563 if (vcp->vc_cpto == 0) 6564 utf16_to_utf8(tmp, tmp_len, retval); 6565 else 6566 WideCharToMultiByte(vcp->vc_cpto, 0, 6567 tmp, tmp_len, 6568 (char *)retval, retlen, 0, 0); 6569 retval[retlen] = NUL; 6570 if (lenp != NULL) 6571 *lenp = retlen; 6572 } 6573 vim_free(tmp); 6574 break; 6575 } 6576 # endif 6577 } 6578 6579 return retval; 6580 } 6581 #endif 6582