1 /* vi:set ts=8 sts=4 sw=4: 2 * 3 * VIM - Vi IMproved by Bram Moolenaar 4 * Multibyte extensions partly by Sung-Hoon Baek 5 * 6 * Do ":help uganda" in Vim to read copying and usage conditions. 7 * Do ":help credits" in Vim to see a list of people who contributed. 8 * See README.txt for an overview of the Vim source code. 9 */ 10 /* 11 * mbyte.c: Code specifically for handling multi-byte characters. 12 * 13 * The encoding used in the core is set with 'encoding'. When 'encoding' is 14 * changed, the following four variables are set (for speed). 15 * Currently these types of character encodings are supported: 16 * 17 * "enc_dbcs" When non-zero it tells the type of double byte character 18 * encoding (Chinese, Korean, Japanese, etc.). 19 * The cell width on the display is equal to the number of 20 * bytes. (exception: DBCS_JPNU with first byte 0x8e) 21 * Recognizing the first or second byte is difficult, it 22 * requires checking a byte sequence from the start. 23 * "enc_utf8" When TRUE use Unicode characters in UTF-8 encoding. 24 * The cell width on the display needs to be determined from 25 * the character value. 26 * Recognizing bytes is easy: 0xxx.xxxx is a single-byte 27 * char, 10xx.xxxx is a trailing byte, 11xx.xxxx is a leading 28 * byte of a multi-byte character. 29 * To make things complicated, up to six composing characters 30 * are allowed. These are drawn on top of the first char. 31 * For most editing the sequence of bytes with composing 32 * characters included is considered to be one character. 33 * "enc_unicode" When 2 use 16-bit Unicode characters (or UTF-16). 34 * When 4 use 32-but Unicode characters. 35 * Internally characters are stored in UTF-8 encoding to 36 * avoid NUL bytes. Conversion happens when doing I/O. 37 * "enc_utf8" will also be TRUE. 38 * 39 * "has_mbyte" is set when "enc_dbcs" or "enc_utf8" is non-zero. 40 * 41 * If none of these is TRUE, 8-bit bytes are used for a character. The 42 * encoding isn't currently specified (TODO). 43 * 44 * 'encoding' specifies the encoding used in the core. This is in registers, 45 * text manipulation, buffers, etc. Conversion has to be done when characters 46 * in another encoding are received or send: 47 * 48 * clipboard 49 * ^ 50 * | (2) 51 * V 52 * +---------------+ 53 * (1) | | (3) 54 * keyboard ----->| core |-----> display 55 * | | 56 * +---------------+ 57 * ^ 58 * | (4) 59 * V 60 * file 61 * 62 * (1) Typed characters arrive in the current locale. Conversion is to be 63 * done when 'encoding' is different from 'termencoding'. 64 * (2) Text will be made available with the encoding specified with 65 * 'encoding'. If this is not sufficient, system-specific conversion 66 * might be required. 67 * (3) For the GUI the correct font must be selected, no conversion done. 68 * Otherwise, conversion is to be done when 'encoding' differs from 69 * 'termencoding'. (Different in the GTK+ 2 port -- 'termencoding' 70 * is always used for both input and output and must always be set to 71 * "utf-8". gui_mch_init() does this automatically.) 72 * (4) The encoding of the file is specified with 'fileencoding'. Conversion 73 * is to be done when it's different from 'encoding'. 74 * 75 * The viminfo file is a special case: Only text is converted, not file names. 76 * Vim scripts may contain an ":encoding" command. This has an effect for 77 * some commands, like ":menutrans" 78 */ 79 80 #include "vim.h" 81 82 #ifdef WIN32UNIX 83 # ifndef WIN32_LEAN_AND_MEAN 84 # define WIN32_LEAN_AND_MEAN 85 # endif 86 # if defined(FEAT_GUI) || defined(FEAT_XCLIPBOARD) 87 # include <X11/Xwindows.h> 88 # define WINBYTE wBYTE 89 # else 90 # include <windows.h> 91 # define WINBYTE BYTE 92 # endif 93 # ifdef WIN32 94 # undef WIN32 /* Some windows.h define WIN32, we don't want that here. */ 95 # endif 96 #else 97 # define WINBYTE BYTE 98 #endif 99 100 #if (defined(WIN3264) || defined(WIN32UNIX)) && !defined(__MINGW32__) 101 # include <winnls.h> 102 #endif 103 104 #ifdef FEAT_GUI_X11 105 # include <X11/Intrinsic.h> 106 #endif 107 #ifdef X_LOCALE 108 #include <X11/Xlocale.h> 109 #endif 110 111 #if defined(FEAT_GUI_GTK) && defined(FEAT_XIM) 112 # include <gdk/gdkkeysyms.h> 113 # ifdef WIN3264 114 # include <gdk/gdkwin32.h> 115 # else 116 # include <gdk/gdkx.h> 117 # endif 118 #endif 119 120 #ifdef HAVE_WCHAR_H 121 # include <wchar.h> 122 #endif 123 124 #if 0 125 /* This has been disabled, because several people reported problems with the 126 * wcwidth() and iswprint() library functions, esp. for Hebrew. */ 127 # ifdef __STDC_ISO_10646__ 128 # define USE_WCHAR_FUNCTIONS 129 # endif 130 #endif 131 132 #if defined(FEAT_MBYTE) || defined(PROTO) 133 134 static int enc_canon_search __ARGS((char_u *name)); 135 static int dbcs_char2len __ARGS((int c)); 136 static int dbcs_char2bytes __ARGS((int c, char_u *buf)); 137 static int dbcs_ptr2len __ARGS((char_u *p)); 138 static int dbcs_ptr2len_len __ARGS((char_u *p, int size)); 139 static int utf_ptr2cells_len __ARGS((char_u *p, int size)); 140 static int dbcs_char2cells __ARGS((int c)); 141 static int dbcs_ptr2cells_len __ARGS((char_u *p, int size)); 142 static int dbcs_ptr2char __ARGS((char_u *p)); 143 static int utf_safe_read_char_adv __ARGS((char_u **s, size_t *n)); 144 145 /* 146 * Lookup table to quickly get the length in bytes of a UTF-8 character from 147 * the first byte of a UTF-8 string. 148 * Bytes which are illegal when used as the first byte have a 1. 149 * The NUL byte has length 1. 150 */ 151 static char utf8len_tab[256] = 152 { 153 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 154 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 155 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 156 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 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 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, 160 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, 161 }; 162 163 /* 164 * Like utf8len_tab above, but using a zero for illegal lead bytes. 165 */ 166 static char utf8len_tab_zero[256] = 167 { 168 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 169 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 170 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 171 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 172 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, 173 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, 174 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, 175 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, 176 }; 177 178 /* 179 * XIM often causes trouble. Define XIM_DEBUG to get a log of XIM callbacks 180 * in the "xim.log" file. 181 */ 182 /* #define XIM_DEBUG */ 183 #ifdef XIM_DEBUG 184 static void 185 xim_log(char *s, ...) 186 { 187 va_list arglist; 188 static FILE *fd = NULL; 189 190 if (fd == (FILE *)-1) 191 return; 192 if (fd == NULL) 193 { 194 fd = mch_fopen("xim.log", "w"); 195 if (fd == NULL) 196 { 197 EMSG("Cannot open xim.log"); 198 fd = (FILE *)-1; 199 return; 200 } 201 } 202 203 va_start(arglist, s); 204 vfprintf(fd, s, arglist); 205 va_end(arglist); 206 } 207 #endif 208 209 #endif 210 211 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 212 /* 213 * Canonical encoding names and their properties. 214 * "iso-8859-n" is handled by enc_canonize() directly. 215 */ 216 static struct 217 { char *name; int prop; int codepage;} 218 enc_canon_table[] = 219 { 220 #define IDX_LATIN_1 0 221 {"latin1", ENC_8BIT + ENC_LATIN1, 1252}, 222 #define IDX_ISO_2 1 223 {"iso-8859-2", ENC_8BIT, 0}, 224 #define IDX_ISO_3 2 225 {"iso-8859-3", ENC_8BIT, 0}, 226 #define IDX_ISO_4 3 227 {"iso-8859-4", ENC_8BIT, 0}, 228 #define IDX_ISO_5 4 229 {"iso-8859-5", ENC_8BIT, 0}, 230 #define IDX_ISO_6 5 231 {"iso-8859-6", ENC_8BIT, 0}, 232 #define IDX_ISO_7 6 233 {"iso-8859-7", ENC_8BIT, 0}, 234 #define IDX_ISO_8 7 235 {"iso-8859-8", ENC_8BIT, 0}, 236 #define IDX_ISO_9 8 237 {"iso-8859-9", ENC_8BIT, 0}, 238 #define IDX_ISO_10 9 239 {"iso-8859-10", ENC_8BIT, 0}, 240 #define IDX_ISO_11 10 241 {"iso-8859-11", ENC_8BIT, 0}, 242 #define IDX_ISO_13 11 243 {"iso-8859-13", ENC_8BIT, 0}, 244 #define IDX_ISO_14 12 245 {"iso-8859-14", ENC_8BIT, 0}, 246 #define IDX_ISO_15 13 247 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0}, 248 #define IDX_KOI8_R 14 249 {"koi8-r", ENC_8BIT, 0}, 250 #define IDX_KOI8_U 15 251 {"koi8-u", ENC_8BIT, 0}, 252 #define IDX_UTF8 16 253 {"utf-8", ENC_UNICODE, 0}, 254 #define IDX_UCS2 17 255 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0}, 256 #define IDX_UCS2LE 18 257 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0}, 258 #define IDX_UTF16 19 259 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0}, 260 #define IDX_UTF16LE 20 261 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0}, 262 #define IDX_UCS4 21 263 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0}, 264 #define IDX_UCS4LE 22 265 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0}, 266 267 /* For debugging DBCS encoding on Unix. */ 268 #define IDX_DEBUG 23 269 {"debug", ENC_DBCS, DBCS_DEBUG}, 270 #define IDX_EUC_JP 24 271 {"euc-jp", ENC_DBCS, DBCS_JPNU}, 272 #define IDX_SJIS 25 273 {"sjis", ENC_DBCS, DBCS_JPN}, 274 #define IDX_EUC_KR 26 275 {"euc-kr", ENC_DBCS, DBCS_KORU}, 276 #define IDX_EUC_CN 27 277 {"euc-cn", ENC_DBCS, DBCS_CHSU}, 278 #define IDX_EUC_TW 28 279 {"euc-tw", ENC_DBCS, DBCS_CHTU}, 280 #define IDX_BIG5 29 281 {"big5", ENC_DBCS, DBCS_CHT}, 282 283 /* MS-DOS and MS-Windows codepages are included here, so that they can be 284 * used on Unix too. Most of them are similar to ISO-8859 encodings, but 285 * not exactly the same. */ 286 #define IDX_CP437 30 287 {"cp437", ENC_8BIT, 437}, /* like iso-8859-1 */ 288 #define IDX_CP737 31 289 {"cp737", ENC_8BIT, 737}, /* like iso-8859-7 */ 290 #define IDX_CP775 32 291 {"cp775", ENC_8BIT, 775}, /* Baltic */ 292 #define IDX_CP850 33 293 {"cp850", ENC_8BIT, 850}, /* like iso-8859-4 */ 294 #define IDX_CP852 34 295 {"cp852", ENC_8BIT, 852}, /* like iso-8859-1 */ 296 #define IDX_CP855 35 297 {"cp855", ENC_8BIT, 855}, /* like iso-8859-2 */ 298 #define IDX_CP857 36 299 {"cp857", ENC_8BIT, 857}, /* like iso-8859-5 */ 300 #define IDX_CP860 37 301 {"cp860", ENC_8BIT, 860}, /* like iso-8859-9 */ 302 #define IDX_CP861 38 303 {"cp861", ENC_8BIT, 861}, /* like iso-8859-1 */ 304 #define IDX_CP862 39 305 {"cp862", ENC_8BIT, 862}, /* like iso-8859-1 */ 306 #define IDX_CP863 40 307 {"cp863", ENC_8BIT, 863}, /* like iso-8859-8 */ 308 #define IDX_CP865 41 309 {"cp865", ENC_8BIT, 865}, /* like iso-8859-1 */ 310 #define IDX_CP866 42 311 {"cp866", ENC_8BIT, 866}, /* like iso-8859-5 */ 312 #define IDX_CP869 43 313 {"cp869", ENC_8BIT, 869}, /* like iso-8859-7 */ 314 #define IDX_CP874 44 315 {"cp874", ENC_8BIT, 874}, /* Thai */ 316 #define IDX_CP932 45 317 {"cp932", ENC_DBCS, DBCS_JPN}, 318 #define IDX_CP936 46 319 {"cp936", ENC_DBCS, DBCS_CHS}, 320 #define IDX_CP949 47 321 {"cp949", ENC_DBCS, DBCS_KOR}, 322 #define IDX_CP950 48 323 {"cp950", ENC_DBCS, DBCS_CHT}, 324 #define IDX_CP1250 49 325 {"cp1250", ENC_8BIT, 1250}, /* Czech, Polish, etc. */ 326 #define IDX_CP1251 50 327 {"cp1251", ENC_8BIT, 1251}, /* Cyrillic */ 328 /* cp1252 is considered to be equal to latin1 */ 329 #define IDX_CP1253 51 330 {"cp1253", ENC_8BIT, 1253}, /* Greek */ 331 #define IDX_CP1254 52 332 {"cp1254", ENC_8BIT, 1254}, /* Turkish */ 333 #define IDX_CP1255 53 334 {"cp1255", ENC_8BIT, 1255}, /* Hebrew */ 335 #define IDX_CP1256 54 336 {"cp1256", ENC_8BIT, 1256}, /* Arabic */ 337 #define IDX_CP1257 55 338 {"cp1257", ENC_8BIT, 1257}, /* Baltic */ 339 #define IDX_CP1258 56 340 {"cp1258", ENC_8BIT, 1258}, /* Vietnamese */ 341 342 #define IDX_MACROMAN 57 343 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, /* Mac OS */ 344 #define IDX_DECMCS 58 345 {"dec-mcs", ENC_8BIT, 0}, /* DEC MCS */ 346 #define IDX_HPROMAN8 59 347 {"hp-roman8", ENC_8BIT, 0}, /* HP Roman8 */ 348 #define IDX_COUNT 60 349 }; 350 351 /* 352 * Aliases for encoding names. 353 */ 354 static struct 355 { char *name; int canon;} 356 enc_alias_table[] = 357 { 358 {"ansi", IDX_LATIN_1}, 359 {"iso-8859-1", IDX_LATIN_1}, 360 {"latin2", IDX_ISO_2}, 361 {"latin3", IDX_ISO_3}, 362 {"latin4", IDX_ISO_4}, 363 {"cyrillic", IDX_ISO_5}, 364 {"arabic", IDX_ISO_6}, 365 {"greek", IDX_ISO_7}, 366 #ifdef WIN3264 367 {"hebrew", IDX_CP1255}, 368 #else 369 {"hebrew", IDX_ISO_8}, 370 #endif 371 {"latin5", IDX_ISO_9}, 372 {"turkish", IDX_ISO_9}, /* ? */ 373 {"latin6", IDX_ISO_10}, 374 {"nordic", IDX_ISO_10}, /* ? */ 375 {"thai", IDX_ISO_11}, /* ? */ 376 {"latin7", IDX_ISO_13}, 377 {"latin8", IDX_ISO_14}, 378 {"latin9", IDX_ISO_15}, 379 {"utf8", IDX_UTF8}, 380 {"unicode", IDX_UCS2}, 381 {"ucs2", IDX_UCS2}, 382 {"ucs2be", IDX_UCS2}, 383 {"ucs-2be", IDX_UCS2}, 384 {"ucs2le", IDX_UCS2LE}, 385 {"utf16", IDX_UTF16}, 386 {"utf16be", IDX_UTF16}, 387 {"utf-16be", IDX_UTF16}, 388 {"utf16le", IDX_UTF16LE}, 389 {"ucs4", IDX_UCS4}, 390 {"ucs4be", IDX_UCS4}, 391 {"ucs-4be", IDX_UCS4}, 392 {"ucs4le", IDX_UCS4LE}, 393 {"utf32", IDX_UCS4}, 394 {"utf-32", IDX_UCS4}, 395 {"utf32be", IDX_UCS4}, 396 {"utf-32be", IDX_UCS4}, 397 {"utf32le", IDX_UCS4LE}, 398 {"utf-32le", IDX_UCS4LE}, 399 {"932", IDX_CP932}, 400 {"949", IDX_CP949}, 401 {"936", IDX_CP936}, 402 {"gbk", IDX_CP936}, 403 {"950", IDX_CP950}, 404 {"eucjp", IDX_EUC_JP}, 405 {"unix-jis", IDX_EUC_JP}, 406 {"ujis", IDX_EUC_JP}, 407 {"shift-jis", IDX_SJIS}, 408 {"pck", IDX_SJIS}, /* Sun: PCK */ 409 {"euckr", IDX_EUC_KR}, 410 {"5601", IDX_EUC_KR}, /* Sun: KS C 5601 */ 411 {"euccn", IDX_EUC_CN}, 412 {"gb2312", IDX_EUC_CN}, 413 {"euctw", IDX_EUC_TW}, 414 #if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS) 415 {"japan", IDX_CP932}, 416 {"korea", IDX_CP949}, 417 {"prc", IDX_CP936}, 418 {"chinese", IDX_CP936}, 419 {"taiwan", IDX_CP950}, 420 {"big5", IDX_CP950}, 421 #else 422 {"japan", IDX_EUC_JP}, 423 {"korea", IDX_EUC_KR}, 424 {"prc", IDX_EUC_CN}, 425 {"chinese", IDX_EUC_CN}, 426 {"taiwan", IDX_EUC_TW}, 427 {"cp950", IDX_BIG5}, 428 {"950", IDX_BIG5}, 429 #endif 430 {"mac", IDX_MACROMAN}, 431 {"mac-roman", IDX_MACROMAN}, 432 {NULL, 0} 433 }; 434 435 #ifndef CP_UTF8 436 # define CP_UTF8 65001 /* magic number from winnls.h */ 437 #endif 438 439 /* 440 * Find encoding "name" in the list of canonical encoding names. 441 * Returns -1 if not found. 442 */ 443 static int 444 enc_canon_search(name) 445 char_u *name; 446 { 447 int i; 448 449 for (i = 0; i < IDX_COUNT; ++i) 450 if (STRCMP(name, enc_canon_table[i].name) == 0) 451 return i; 452 return -1; 453 } 454 455 #endif 456 457 #if defined(FEAT_MBYTE) || defined(PROTO) 458 459 /* 460 * Find canonical encoding "name" in the list and return its properties. 461 * Returns 0 if not found. 462 */ 463 int 464 enc_canon_props(name) 465 char_u *name; 466 { 467 int i; 468 469 i = enc_canon_search(name); 470 if (i >= 0) 471 return enc_canon_table[i].prop; 472 #ifdef WIN3264 473 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2])) 474 { 475 CPINFO cpinfo; 476 477 /* Get info on this codepage to find out what it is. */ 478 if (GetCPInfo(atoi(name + 2), &cpinfo) != 0) 479 { 480 if (cpinfo.MaxCharSize == 1) /* some single-byte encoding */ 481 return ENC_8BIT; 482 if (cpinfo.MaxCharSize == 2 483 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 484 /* must be a DBCS encoding */ 485 return ENC_DBCS; 486 } 487 return 0; 488 } 489 #endif 490 if (STRNCMP(name, "2byte-", 6) == 0) 491 return ENC_DBCS; 492 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0) 493 return ENC_8BIT; 494 return 0; 495 } 496 497 /* 498 * Set up for using multi-byte characters. 499 * Called in three cases: 500 * - by main() to initialize (p_enc == NULL) 501 * - by set_init_1() after 'encoding' was set to its default. 502 * - by do_set() when 'encoding' has been set. 503 * p_enc must have been passed through enc_canonize() already. 504 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags. 505 * Fills mb_bytelen_tab[] and returns NULL when there are no problems. 506 * When there is something wrong: Returns an error message and doesn't change 507 * anything. 508 */ 509 char_u * 510 mb_init() 511 { 512 int i; 513 int idx; 514 int n; 515 int enc_dbcs_new = 0; 516 #if defined(USE_ICONV) && !defined(WIN3264) && !defined(WIN32UNIX) \ 517 && !defined(MACOS) 518 # define LEN_FROM_CONV 519 vimconv_T vimconv; 520 char_u *p; 521 #endif 522 523 if (p_enc == NULL) 524 { 525 /* Just starting up: set the whole table to one's. */ 526 for (i = 0; i < 256; ++i) 527 mb_bytelen_tab[i] = 1; 528 input_conv.vc_type = CONV_NONE; 529 input_conv.vc_factor = 1; 530 output_conv.vc_type = CONV_NONE; 531 return NULL; 532 } 533 534 #ifdef WIN3264 535 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2])) 536 { 537 CPINFO cpinfo; 538 539 /* Get info on this codepage to find out what it is. */ 540 if (GetCPInfo(atoi(p_enc + 2), &cpinfo) != 0) 541 { 542 if (cpinfo.MaxCharSize == 1) 543 { 544 /* some single-byte encoding */ 545 enc_unicode = 0; 546 enc_utf8 = FALSE; 547 } 548 else if (cpinfo.MaxCharSize == 2 549 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 550 { 551 /* must be a DBCS encoding, check below */ 552 enc_dbcs_new = atoi(p_enc + 2); 553 } 554 else 555 goto codepage_invalid; 556 } 557 else if (GetLastError() == ERROR_INVALID_PARAMETER) 558 { 559 codepage_invalid: 560 return (char_u *)N_("E543: Not a valid codepage"); 561 } 562 } 563 #endif 564 else if (STRNCMP(p_enc, "8bit-", 5) == 0 565 || STRNCMP(p_enc, "iso-8859-", 9) == 0) 566 { 567 /* Accept any "8bit-" or "iso-8859-" name. */ 568 enc_unicode = 0; 569 enc_utf8 = FALSE; 570 } 571 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 572 { 573 #ifdef WIN3264 574 /* Windows: accept only valid codepage numbers, check below. */ 575 if (p_enc[6] != 'c' || p_enc[7] != 'p' 576 || (enc_dbcs_new = atoi(p_enc + 8)) == 0) 577 return e_invarg; 578 #else 579 /* Unix: accept any "2byte-" name, assume current locale. */ 580 enc_dbcs_new = DBCS_2BYTE; 581 #endif 582 } 583 else if ((idx = enc_canon_search(p_enc)) >= 0) 584 { 585 i = enc_canon_table[idx].prop; 586 if (i & ENC_UNICODE) 587 { 588 /* Unicode */ 589 enc_utf8 = TRUE; 590 if (i & (ENC_2BYTE | ENC_2WORD)) 591 enc_unicode = 2; 592 else if (i & ENC_4BYTE) 593 enc_unicode = 4; 594 else 595 enc_unicode = 0; 596 } 597 else if (i & ENC_DBCS) 598 { 599 /* 2byte, handle below */ 600 enc_dbcs_new = enc_canon_table[idx].codepage; 601 } 602 else 603 { 604 /* Must be 8-bit. */ 605 enc_unicode = 0; 606 enc_utf8 = FALSE; 607 } 608 } 609 else /* Don't know what encoding this is, reject it. */ 610 return e_invarg; 611 612 if (enc_dbcs_new != 0) 613 { 614 #ifdef WIN3264 615 /* Check if the DBCS code page is OK. */ 616 if (!IsValidCodePage(enc_dbcs_new)) 617 goto codepage_invalid; 618 #endif 619 enc_unicode = 0; 620 enc_utf8 = FALSE; 621 } 622 enc_dbcs = enc_dbcs_new; 623 has_mbyte = (enc_dbcs != 0 || enc_utf8); 624 625 #if defined(WIN3264) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 626 enc_codepage = encname2codepage(p_enc); 627 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0); 628 #endif 629 630 /* Detect an encoding that uses latin1 characters. */ 631 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0 632 || STRCMP(p_enc, "iso-8859-15") == 0); 633 634 /* 635 * Set the function pointers. 636 */ 637 if (enc_utf8) 638 { 639 mb_ptr2len = utfc_ptr2len; 640 mb_ptr2len_len = utfc_ptr2len_len; 641 mb_char2len = utf_char2len; 642 mb_char2bytes = utf_char2bytes; 643 mb_ptr2cells = utf_ptr2cells; 644 mb_ptr2cells_len = utf_ptr2cells_len; 645 mb_char2cells = utf_char2cells; 646 mb_off2cells = utf_off2cells; 647 mb_ptr2char = utf_ptr2char; 648 mb_head_off = utf_head_off; 649 } 650 else if (enc_dbcs != 0) 651 { 652 mb_ptr2len = dbcs_ptr2len; 653 mb_ptr2len_len = dbcs_ptr2len_len; 654 mb_char2len = dbcs_char2len; 655 mb_char2bytes = dbcs_char2bytes; 656 mb_ptr2cells = dbcs_ptr2cells; 657 mb_ptr2cells_len = dbcs_ptr2cells_len; 658 mb_char2cells = dbcs_char2cells; 659 mb_off2cells = dbcs_off2cells; 660 mb_ptr2char = dbcs_ptr2char; 661 mb_head_off = dbcs_head_off; 662 } 663 else 664 { 665 mb_ptr2len = latin_ptr2len; 666 mb_ptr2len_len = latin_ptr2len_len; 667 mb_char2len = latin_char2len; 668 mb_char2bytes = latin_char2bytes; 669 mb_ptr2cells = latin_ptr2cells; 670 mb_ptr2cells_len = latin_ptr2cells_len; 671 mb_char2cells = latin_char2cells; 672 mb_off2cells = latin_off2cells; 673 mb_ptr2char = latin_ptr2char; 674 mb_head_off = latin_head_off; 675 } 676 677 /* 678 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN(). 679 */ 680 #ifdef LEN_FROM_CONV 681 /* When 'encoding' is different from the current locale mblen() won't 682 * work. Use conversion to "utf-8" instead. */ 683 vimconv.vc_type = CONV_NONE; 684 if (enc_dbcs) 685 { 686 p = enc_locale(); 687 if (p == NULL || STRCMP(p, p_enc) != 0) 688 { 689 convert_setup(&vimconv, p_enc, (char_u *)"utf-8"); 690 vimconv.vc_fail = TRUE; 691 } 692 vim_free(p); 693 } 694 #endif 695 696 for (i = 0; i < 256; ++i) 697 { 698 /* Our own function to reliably check the length of UTF-8 characters, 699 * independent of mblen(). */ 700 if (enc_utf8) 701 n = utf8len_tab[i]; 702 else if (enc_dbcs == 0) 703 n = 1; 704 else 705 { 706 #if defined(WIN3264) || defined(WIN32UNIX) 707 /* enc_dbcs is set by setting 'fileencoding'. It becomes a Windows 708 * CodePage identifier, which we can pass directly in to Windows 709 * API */ 710 n = IsDBCSLeadByteEx(enc_dbcs, (WINBYTE)i) ? 2 : 1; 711 #else 712 # if defined(MACOS) || defined(__amigaos4__) || defined(__ANDROID__) 713 /* 714 * if mblen() is not available, character which MSB is turned on 715 * are treated as leading byte character. (note : This assumption 716 * is not always true.) 717 */ 718 n = (i & 0x80) ? 2 : 1; 719 # else 720 char buf[MB_MAXBYTES + 1]; 721 # ifdef X_LOCALE 722 # ifndef mblen 723 # define mblen _Xmblen 724 # endif 725 # endif 726 if (i == NUL) /* just in case mblen() can't handle "" */ 727 n = 1; 728 else 729 { 730 buf[0] = i; 731 buf[1] = 0; 732 #ifdef LEN_FROM_CONV 733 if (vimconv.vc_type != CONV_NONE) 734 { 735 /* 736 * string_convert() should fail when converting the first 737 * byte of a double-byte character. 738 */ 739 p = string_convert(&vimconv, (char_u *)buf, NULL); 740 if (p != NULL) 741 { 742 vim_free(p); 743 n = 1; 744 } 745 else 746 n = 2; 747 } 748 else 749 #endif 750 { 751 /* 752 * mblen() should return -1 for invalid (means the leading 753 * multibyte) character. However there are some platforms 754 * where mblen() returns 0 for invalid character. 755 * Therefore, following condition includes 0. 756 */ 757 ignored = mblen(NULL, 0); /* First reset the state. */ 758 if (mblen(buf, (size_t)1) <= 0) 759 n = 2; 760 else 761 n = 1; 762 } 763 } 764 # endif 765 #endif 766 } 767 768 mb_bytelen_tab[i] = n; 769 } 770 771 #ifdef LEN_FROM_CONV 772 convert_setup(&vimconv, NULL, NULL); 773 #endif 774 775 /* The cell width depends on the type of multi-byte characters. */ 776 (void)init_chartab(); 777 778 /* When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] */ 779 screenalloc(FALSE); 780 781 /* When using Unicode, set default for 'fileencodings'. */ 782 if (enc_utf8 && !option_was_set((char_u *)"fencs")) 783 set_string_option_direct((char_u *)"fencs", -1, 784 (char_u *)"ucs-bom,utf-8,default,latin1", OPT_FREE, 0); 785 786 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT) 787 /* GNU gettext 0.10.37 supports this feature: set the codeset used for 788 * translated messages independently from the current locale. */ 789 (void)bind_textdomain_codeset(VIMPACKAGE, 790 enc_utf8 ? "utf-8" : (char *)p_enc); 791 #endif 792 793 #ifdef WIN32 794 /* When changing 'encoding' while starting up, then convert the command 795 * line arguments from the active codepage to 'encoding'. */ 796 if (starting != 0) 797 fix_arg_enc(); 798 #endif 799 800 #ifdef FEAT_AUTOCMD 801 /* Fire an autocommand to let people do custom font setup. This must be 802 * after Vim has been setup for the new encoding. */ 803 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); 804 #endif 805 806 #ifdef FEAT_SPELL 807 /* Need to reload spell dictionaries */ 808 spell_reload(); 809 #endif 810 811 return NULL; 812 } 813 814 /* 815 * Return the size of the BOM for the current buffer: 816 * 0 - no BOM 817 * 2 - UCS-2 or UTF-16 BOM 818 * 4 - UCS-4 BOM 819 * 3 - UTF-8 BOM 820 */ 821 int 822 bomb_size() 823 { 824 int n = 0; 825 826 if (curbuf->b_p_bomb && !curbuf->b_p_bin) 827 { 828 if (*curbuf->b_p_fenc == NUL) 829 { 830 if (enc_utf8) 831 { 832 if (enc_unicode != 0) 833 n = enc_unicode; 834 else 835 n = 3; 836 } 837 } 838 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) 839 n = 3; 840 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 841 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) 842 n = 2; 843 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) 844 n = 4; 845 } 846 return n; 847 } 848 849 /* 850 * Remove all BOM from "s" by moving remaining text. 851 */ 852 void 853 remove_bom(s) 854 char_u *s; 855 { 856 if (enc_utf8) 857 { 858 char_u *p = s; 859 860 while ((p = vim_strbyte(p, 0xef)) != NULL) 861 { 862 if (p[1] == 0xbb && p[2] == 0xbf) 863 STRMOVE(p, p + 3); 864 else 865 ++p; 866 } 867 } 868 } 869 870 /* 871 * Get class of pointer: 872 * 0 for blank or NUL 873 * 1 for punctuation 874 * 2 for an (ASCII) word character 875 * >2 for other word characters 876 */ 877 int 878 mb_get_class(p) 879 char_u *p; 880 { 881 return mb_get_class_buf(p, curbuf); 882 } 883 884 int 885 mb_get_class_buf(p, buf) 886 char_u *p; 887 buf_T *buf; 888 { 889 if (MB_BYTE2LEN(p[0]) == 1) 890 { 891 if (p[0] == NUL || vim_iswhite(p[0])) 892 return 0; 893 if (vim_iswordc_buf(p[0], buf)) 894 return 2; 895 return 1; 896 } 897 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) 898 return dbcs_class(p[0], p[1]); 899 if (enc_utf8) 900 return utf_class(utf_ptr2char(p)); 901 return 0; 902 } 903 904 /* 905 * Get class of a double-byte character. This always returns 3 or bigger. 906 * TODO: Should return 1 for punctuation. 907 */ 908 int 909 dbcs_class(lead, trail) 910 unsigned lead; 911 unsigned trail; 912 { 913 switch (enc_dbcs) 914 { 915 /* please add classify routine for your language in here */ 916 917 case DBCS_JPNU: /* ? */ 918 case DBCS_JPN: 919 { 920 /* JIS code classification */ 921 unsigned char lb = lead; 922 unsigned char tb = trail; 923 924 /* convert process code to JIS */ 925 # if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS) 926 /* process code is SJIS */ 927 if (lb <= 0x9f) 928 lb = (lb - 0x81) * 2 + 0x21; 929 else 930 lb = (lb - 0xc1) * 2 + 0x21; 931 if (tb <= 0x7e) 932 tb -= 0x1f; 933 else if (tb <= 0x9e) 934 tb -= 0x20; 935 else 936 { 937 tb -= 0x7e; 938 lb += 1; 939 } 940 # else 941 /* 942 * XXX: Code page identification can not use with all 943 * system! So, some other encoding information 944 * will be needed. 945 * In japanese: SJIS,EUC,UNICODE,(JIS) 946 * Note that JIS-code system don't use as 947 * process code in most system because it uses 948 * escape sequences(JIS is context depend encoding). 949 */ 950 /* assume process code is JAPANESE-EUC */ 951 lb &= 0x7f; 952 tb &= 0x7f; 953 # endif 954 /* exceptions */ 955 switch (lb << 8 | tb) 956 { 957 case 0x2121: /* ZENKAKU space */ 958 return 0; 959 case 0x2122: /* TOU-TEN (Japanese comma) */ 960 case 0x2123: /* KU-TEN (Japanese period) */ 961 case 0x2124: /* ZENKAKU comma */ 962 case 0x2125: /* ZENKAKU period */ 963 return 1; 964 case 0x213c: /* prolongedsound handled as KATAKANA */ 965 return 13; 966 } 967 /* sieved by KU code */ 968 switch (lb) 969 { 970 case 0x21: 971 case 0x22: 972 /* special symbols */ 973 return 10; 974 case 0x23: 975 /* alpha-numeric */ 976 return 11; 977 case 0x24: 978 /* hiragana */ 979 return 12; 980 case 0x25: 981 /* katakana */ 982 return 13; 983 case 0x26: 984 /* greek */ 985 return 14; 986 case 0x27: 987 /* russian */ 988 return 15; 989 case 0x28: 990 /* lines */ 991 return 16; 992 default: 993 /* kanji */ 994 return 17; 995 } 996 } 997 998 case DBCS_KORU: /* ? */ 999 case DBCS_KOR: 1000 { 1001 /* KS code classification */ 1002 unsigned char c1 = lead; 1003 unsigned char c2 = trail; 1004 1005 /* 1006 * 20 : Hangul 1007 * 21 : Hanja 1008 * 22 : Symbols 1009 * 23 : Alpha-numeric/Roman Letter (Full width) 1010 * 24 : Hangul Letter(Alphabet) 1011 * 25 : Roman Numeral/Greek Letter 1012 * 26 : Box Drawings 1013 * 27 : Unit Symbols 1014 * 28 : Circled/Parenthesized Letter 1015 * 29 : Hiragana/Katakana 1016 * 30 : Cyrillic Letter 1017 */ 1018 1019 if (c1 >= 0xB0 && c1 <= 0xC8) 1020 /* Hangul */ 1021 return 20; 1022 #if defined(WIN3264) || defined(WIN32UNIX) 1023 else if (c1 <= 0xA0 || c2 <= 0xA0) 1024 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */ 1025 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE 1026 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 1027 */ 1028 return 20; 1029 #endif 1030 1031 else if (c1 >= 0xCA && c1 <= 0xFD) 1032 /* Hanja */ 1033 return 21; 1034 else switch (c1) 1035 { 1036 case 0xA1: 1037 case 0xA2: 1038 /* Symbols */ 1039 return 22; 1040 case 0xA3: 1041 /* Alpha-numeric */ 1042 return 23; 1043 case 0xA4: 1044 /* Hangul Letter(Alphabet) */ 1045 return 24; 1046 case 0xA5: 1047 /* Roman Numeral/Greek Letter */ 1048 return 25; 1049 case 0xA6: 1050 /* Box Drawings */ 1051 return 26; 1052 case 0xA7: 1053 /* Unit Symbols */ 1054 return 27; 1055 case 0xA8: 1056 case 0xA9: 1057 if (c2 <= 0xAF) 1058 return 25; /* Roman Letter */ 1059 else if (c2 >= 0xF6) 1060 return 22; /* Symbols */ 1061 else 1062 /* Circled/Parenthesized Letter */ 1063 return 28; 1064 case 0xAA: 1065 case 0xAB: 1066 /* Hiragana/Katakana */ 1067 return 29; 1068 case 0xAC: 1069 /* Cyrillic Letter */ 1070 return 30; 1071 } 1072 } 1073 default: 1074 break; 1075 } 1076 return 3; 1077 } 1078 1079 /* 1080 * mb_char2len() function pointer. 1081 * Return length in bytes of character "c". 1082 * Returns 1 for a single-byte character. 1083 */ 1084 int 1085 latin_char2len(c) 1086 int c UNUSED; 1087 { 1088 return 1; 1089 } 1090 1091 static int 1092 dbcs_char2len(c) 1093 int c; 1094 { 1095 if (c >= 0x100) 1096 return 2; 1097 return 1; 1098 } 1099 1100 /* 1101 * mb_char2bytes() function pointer. 1102 * Convert a character to its bytes. 1103 * Returns the length in bytes. 1104 */ 1105 int 1106 latin_char2bytes(c, buf) 1107 int c; 1108 char_u *buf; 1109 { 1110 buf[0] = c; 1111 return 1; 1112 } 1113 1114 static int 1115 dbcs_char2bytes(c, buf) 1116 int c; 1117 char_u *buf; 1118 { 1119 if (c >= 0x100) 1120 { 1121 buf[0] = (unsigned)c >> 8; 1122 buf[1] = c; 1123 /* Never use a NUL byte, it causes lots of trouble. It's an invalid 1124 * character anyway. */ 1125 if (buf[1] == NUL) 1126 buf[1] = '\n'; 1127 return 2; 1128 } 1129 buf[0] = c; 1130 return 1; 1131 } 1132 1133 /* 1134 * mb_ptr2len() function pointer. 1135 * Get byte length of character at "*p" but stop at a NUL. 1136 * For UTF-8 this includes following composing characters. 1137 * Returns 0 when *p is NUL. 1138 */ 1139 int 1140 latin_ptr2len(p) 1141 char_u *p; 1142 { 1143 return MB_BYTE2LEN(*p); 1144 } 1145 1146 static int 1147 dbcs_ptr2len(p) 1148 char_u *p; 1149 { 1150 int len; 1151 1152 /* Check if second byte is not missing. */ 1153 len = MB_BYTE2LEN(*p); 1154 if (len == 2 && p[1] == NUL) 1155 len = 1; 1156 return len; 1157 } 1158 1159 /* 1160 * mb_ptr2len_len() function pointer. 1161 * Like mb_ptr2len(), but limit to read "size" bytes. 1162 * Returns 0 for an empty string. 1163 * Returns 1 for an illegal char or an incomplete byte sequence. 1164 */ 1165 int 1166 latin_ptr2len_len(p, size) 1167 char_u *p; 1168 int size; 1169 { 1170 if (size < 1 || *p == NUL) 1171 return 0; 1172 return 1; 1173 } 1174 1175 static int 1176 dbcs_ptr2len_len(p, size) 1177 char_u *p; 1178 int size; 1179 { 1180 int len; 1181 1182 if (size < 1 || *p == NUL) 1183 return 0; 1184 if (size == 1) 1185 return 1; 1186 /* Check that second byte is not missing. */ 1187 len = MB_BYTE2LEN(*p); 1188 if (len == 2 && p[1] == NUL) 1189 len = 1; 1190 return len; 1191 } 1192 1193 struct interval 1194 { 1195 long first; 1196 long last; 1197 }; 1198 static int intable __ARGS((struct interval *table, size_t size, int c)); 1199 1200 /* 1201 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]". 1202 */ 1203 static int 1204 intable(table, size, c) 1205 struct interval *table; 1206 size_t size; 1207 int c; 1208 { 1209 int mid, bot, top; 1210 1211 /* first quick check for Latin1 etc. characters */ 1212 if (c < table[0].first) 1213 return FALSE; 1214 1215 /* binary search in table */ 1216 bot = 0; 1217 top = (int)(size / sizeof(struct interval) - 1); 1218 while (top >= bot) 1219 { 1220 mid = (bot + top) / 2; 1221 if (table[mid].last < c) 1222 bot = mid + 1; 1223 else if (table[mid].first > c) 1224 top = mid - 1; 1225 else 1226 return TRUE; 1227 } 1228 return FALSE; 1229 } 1230 1231 /* 1232 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1233 * Returns 4 or 6 for an unprintable character. 1234 * Is only correct for characters >= 0x80. 1235 * When p_ambw is "double", return 2 for a character with East Asian Width 1236 * class 'A'(mbiguous). 1237 */ 1238 int 1239 utf_char2cells(c) 1240 int c; 1241 { 1242 /* Sorted list of non-overlapping intervals of East Asian double width 1243 * characters, generated with ../runtime/tools/unicode.vim. */ 1244 static struct interval doublewidth[] = 1245 { 1246 {0x1100, 0x115f}, 1247 {0x2329, 0x232a}, 1248 {0x2e80, 0x2e99}, 1249 {0x2e9b, 0x2ef3}, 1250 {0x2f00, 0x2fd5}, 1251 {0x2ff0, 0x2ffb}, 1252 {0x3000, 0x303e}, 1253 {0x3041, 0x3096}, 1254 {0x3099, 0x30ff}, 1255 {0x3105, 0x312d}, 1256 {0x3131, 0x318e}, 1257 {0x3190, 0x31ba}, 1258 {0x31c0, 0x31e3}, 1259 {0x31f0, 0x321e}, 1260 {0x3220, 0x3247}, 1261 {0x3250, 0x32fe}, 1262 {0x3300, 0x4dbf}, 1263 {0x4e00, 0xa48c}, 1264 {0xa490, 0xa4c6}, 1265 {0xa960, 0xa97c}, 1266 {0xac00, 0xd7a3}, 1267 {0xf900, 0xfaff}, 1268 {0xfe10, 0xfe19}, 1269 {0xfe30, 0xfe52}, 1270 {0xfe54, 0xfe66}, 1271 {0xfe68, 0xfe6b}, 1272 {0xff01, 0xff60}, 1273 {0xffe0, 0xffe6}, 1274 {0x1b000, 0x1b001}, 1275 {0x1f200, 0x1f202}, 1276 {0x1f210, 0x1f23a}, 1277 {0x1f240, 0x1f248}, 1278 {0x1f250, 0x1f251}, 1279 {0x20000, 0x2fffd}, 1280 {0x30000, 0x3fffd} 1281 }; 1282 1283 /* Sorted list of non-overlapping intervals of East Asian Ambiguous 1284 * characters, generated with ../runtime/tools/unicode.vim. */ 1285 static struct interval ambiguous[] = 1286 { 1287 {0x00a1, 0x00a1}, 1288 {0x00a4, 0x00a4}, 1289 {0x00a7, 0x00a8}, 1290 {0x00aa, 0x00aa}, 1291 {0x00ad, 0x00ae}, 1292 {0x00b0, 0x00b4}, 1293 {0x00b6, 0x00ba}, 1294 {0x00bc, 0x00bf}, 1295 {0x00c6, 0x00c6}, 1296 {0x00d0, 0x00d0}, 1297 {0x00d7, 0x00d8}, 1298 {0x00de, 0x00e1}, 1299 {0x00e6, 0x00e6}, 1300 {0x00e8, 0x00ea}, 1301 {0x00ec, 0x00ed}, 1302 {0x00f0, 0x00f0}, 1303 {0x00f2, 0x00f3}, 1304 {0x00f7, 0x00fa}, 1305 {0x00fc, 0x00fc}, 1306 {0x00fe, 0x00fe}, 1307 {0x0101, 0x0101}, 1308 {0x0111, 0x0111}, 1309 {0x0113, 0x0113}, 1310 {0x011b, 0x011b}, 1311 {0x0126, 0x0127}, 1312 {0x012b, 0x012b}, 1313 {0x0131, 0x0133}, 1314 {0x0138, 0x0138}, 1315 {0x013f, 0x0142}, 1316 {0x0144, 0x0144}, 1317 {0x0148, 0x014b}, 1318 {0x014d, 0x014d}, 1319 {0x0152, 0x0153}, 1320 {0x0166, 0x0167}, 1321 {0x016b, 0x016b}, 1322 {0x01ce, 0x01ce}, 1323 {0x01d0, 0x01d0}, 1324 {0x01d2, 0x01d2}, 1325 {0x01d4, 0x01d4}, 1326 {0x01d6, 0x01d6}, 1327 {0x01d8, 0x01d8}, 1328 {0x01da, 0x01da}, 1329 {0x01dc, 0x01dc}, 1330 {0x0251, 0x0251}, 1331 {0x0261, 0x0261}, 1332 {0x02c4, 0x02c4}, 1333 {0x02c7, 0x02c7}, 1334 {0x02c9, 0x02cb}, 1335 {0x02cd, 0x02cd}, 1336 {0x02d0, 0x02d0}, 1337 {0x02d8, 0x02db}, 1338 {0x02dd, 0x02dd}, 1339 {0x02df, 0x02df}, 1340 {0x0300, 0x036f}, 1341 {0x0391, 0x03a1}, 1342 {0x03a3, 0x03a9}, 1343 {0x03b1, 0x03c1}, 1344 {0x03c3, 0x03c9}, 1345 {0x0401, 0x0401}, 1346 {0x0410, 0x044f}, 1347 {0x0451, 0x0451}, 1348 {0x2010, 0x2010}, 1349 {0x2013, 0x2016}, 1350 {0x2018, 0x2019}, 1351 {0x201c, 0x201d}, 1352 {0x2020, 0x2022}, 1353 {0x2024, 0x2027}, 1354 {0x2030, 0x2030}, 1355 {0x2032, 0x2033}, 1356 {0x2035, 0x2035}, 1357 {0x203b, 0x203b}, 1358 {0x203e, 0x203e}, 1359 {0x2074, 0x2074}, 1360 {0x207f, 0x207f}, 1361 {0x2081, 0x2084}, 1362 {0x20ac, 0x20ac}, 1363 {0x2103, 0x2103}, 1364 {0x2105, 0x2105}, 1365 {0x2109, 0x2109}, 1366 {0x2113, 0x2113}, 1367 {0x2116, 0x2116}, 1368 {0x2121, 0x2122}, 1369 {0x2126, 0x2126}, 1370 {0x212b, 0x212b}, 1371 {0x2153, 0x2154}, 1372 {0x215b, 0x215e}, 1373 {0x2160, 0x216b}, 1374 {0x2170, 0x2179}, 1375 {0x2189, 0x2189}, 1376 {0x2190, 0x2199}, 1377 {0x21b8, 0x21b9}, 1378 {0x21d2, 0x21d2}, 1379 {0x21d4, 0x21d4}, 1380 {0x21e7, 0x21e7}, 1381 {0x2200, 0x2200}, 1382 {0x2202, 0x2203}, 1383 {0x2207, 0x2208}, 1384 {0x220b, 0x220b}, 1385 {0x220f, 0x220f}, 1386 {0x2211, 0x2211}, 1387 {0x2215, 0x2215}, 1388 {0x221a, 0x221a}, 1389 {0x221d, 0x2220}, 1390 {0x2223, 0x2223}, 1391 {0x2225, 0x2225}, 1392 {0x2227, 0x222c}, 1393 {0x222e, 0x222e}, 1394 {0x2234, 0x2237}, 1395 {0x223c, 0x223d}, 1396 {0x2248, 0x2248}, 1397 {0x224c, 0x224c}, 1398 {0x2252, 0x2252}, 1399 {0x2260, 0x2261}, 1400 {0x2264, 0x2267}, 1401 {0x226a, 0x226b}, 1402 {0x226e, 0x226f}, 1403 {0x2282, 0x2283}, 1404 {0x2286, 0x2287}, 1405 {0x2295, 0x2295}, 1406 {0x2299, 0x2299}, 1407 {0x22a5, 0x22a5}, 1408 {0x22bf, 0x22bf}, 1409 {0x2312, 0x2312}, 1410 {0x2460, 0x24e9}, 1411 {0x24eb, 0x254b}, 1412 {0x2550, 0x2573}, 1413 {0x2580, 0x258f}, 1414 {0x2592, 0x2595}, 1415 {0x25a0, 0x25a1}, 1416 {0x25a3, 0x25a9}, 1417 {0x25b2, 0x25b3}, 1418 {0x25b6, 0x25b7}, 1419 {0x25bc, 0x25bd}, 1420 {0x25c0, 0x25c1}, 1421 {0x25c6, 0x25c8}, 1422 {0x25cb, 0x25cb}, 1423 {0x25ce, 0x25d1}, 1424 {0x25e2, 0x25e5}, 1425 {0x25ef, 0x25ef}, 1426 {0x2605, 0x2606}, 1427 {0x2609, 0x2609}, 1428 {0x260e, 0x260f}, 1429 {0x2614, 0x2615}, 1430 {0x261c, 0x261c}, 1431 {0x261e, 0x261e}, 1432 {0x2640, 0x2640}, 1433 {0x2642, 0x2642}, 1434 {0x2660, 0x2661}, 1435 {0x2663, 0x2665}, 1436 {0x2667, 0x266a}, 1437 {0x266c, 0x266d}, 1438 {0x266f, 0x266f}, 1439 {0x269e, 0x269f}, 1440 {0x26be, 0x26bf}, 1441 {0x26c4, 0x26cd}, 1442 {0x26cf, 0x26e1}, 1443 {0x26e3, 0x26e3}, 1444 {0x26e8, 0x26ff}, 1445 {0x273d, 0x273d}, 1446 {0x2757, 0x2757}, 1447 {0x2776, 0x277f}, 1448 {0x2b55, 0x2b59}, 1449 {0x3248, 0x324f}, 1450 {0xe000, 0xf8ff}, 1451 {0xfe00, 0xfe0f}, 1452 {0xfffd, 0xfffd}, 1453 {0x1f100, 0x1f10a}, 1454 {0x1f110, 0x1f12d}, 1455 {0x1f130, 0x1f169}, 1456 {0x1f170, 0x1f19a}, 1457 {0xe0100, 0xe01ef}, 1458 {0xf0000, 0xffffd}, 1459 {0x100000, 0x10fffd} 1460 }; 1461 1462 if (c >= 0x100) 1463 { 1464 #ifdef USE_WCHAR_FUNCTIONS 1465 /* 1466 * Assume the library function wcwidth() works better than our own 1467 * stuff. It should return 1 for ambiguous width chars! 1468 */ 1469 int n = wcwidth(c); 1470 1471 if (n < 0) 1472 return 6; /* unprintable, displays <xxxx> */ 1473 if (n > 1) 1474 return n; 1475 #else 1476 if (!utf_printable(c)) 1477 return 6; /* unprintable, displays <xxxx> */ 1478 if (intable(doublewidth, sizeof(doublewidth), c)) 1479 return 2; 1480 #endif 1481 } 1482 1483 /* Characters below 0x100 are influenced by 'isprint' option */ 1484 else if (c >= 0x80 && !vim_isprintc(c)) 1485 return 4; /* unprintable, displays <xx> */ 1486 1487 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c)) 1488 return 2; 1489 1490 return 1; 1491 } 1492 1493 /* 1494 * mb_ptr2cells() function pointer. 1495 * Return the number of display cells character at "*p" occupies. 1496 * This doesn't take care of unprintable characters, use ptr2cells() for that. 1497 */ 1498 int 1499 latin_ptr2cells(p) 1500 char_u *p UNUSED; 1501 { 1502 return 1; 1503 } 1504 1505 int 1506 utf_ptr2cells(p) 1507 char_u *p; 1508 { 1509 int c; 1510 1511 /* Need to convert to a wide character. */ 1512 if (*p >= 0x80) 1513 { 1514 c = utf_ptr2char(p); 1515 /* An illegal byte is displayed as <xx>. */ 1516 if (utf_ptr2len(p) == 1 || c == NUL) 1517 return 4; 1518 /* If the char is ASCII it must be an overlong sequence. */ 1519 if (c < 0x80) 1520 return char2cells(c); 1521 return utf_char2cells(c); 1522 } 1523 return 1; 1524 } 1525 1526 int 1527 dbcs_ptr2cells(p) 1528 char_u *p; 1529 { 1530 /* Number of cells is equal to number of bytes, except for euc-jp when 1531 * the first byte is 0x8e. */ 1532 if (enc_dbcs == DBCS_JPNU && *p == 0x8e) 1533 return 1; 1534 return MB_BYTE2LEN(*p); 1535 } 1536 1537 /* 1538 * mb_ptr2cells_len() function pointer. 1539 * Like mb_ptr2cells(), but limit string length to "size". 1540 * For an empty string or truncated character returns 1. 1541 */ 1542 int 1543 latin_ptr2cells_len(p, size) 1544 char_u *p UNUSED; 1545 int size UNUSED; 1546 { 1547 return 1; 1548 } 1549 1550 static int 1551 utf_ptr2cells_len(p, size) 1552 char_u *p; 1553 int size; 1554 { 1555 int c; 1556 1557 /* Need to convert to a wide character. */ 1558 if (size > 0 && *p >= 0x80) 1559 { 1560 if (utf_ptr2len_len(p, size) < utf8len_tab[*p]) 1561 return 1; /* truncated */ 1562 c = utf_ptr2char(p); 1563 /* An illegal byte is displayed as <xx>. */ 1564 if (utf_ptr2len(p) == 1 || c == NUL) 1565 return 4; 1566 /* If the char is ASCII it must be an overlong sequence. */ 1567 if (c < 0x80) 1568 return char2cells(c); 1569 return utf_char2cells(c); 1570 } 1571 return 1; 1572 } 1573 1574 static int 1575 dbcs_ptr2cells_len(p, size) 1576 char_u *p; 1577 int size; 1578 { 1579 /* Number of cells is equal to number of bytes, except for euc-jp when 1580 * the first byte is 0x8e. */ 1581 if (size <= 1 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)) 1582 return 1; 1583 return MB_BYTE2LEN(*p); 1584 } 1585 1586 /* 1587 * mb_char2cells() function pointer. 1588 * Return the number of display cells character "c" occupies. 1589 * Only takes care of multi-byte chars, not "^C" and such. 1590 */ 1591 int 1592 latin_char2cells(c) 1593 int c UNUSED; 1594 { 1595 return 1; 1596 } 1597 1598 static int 1599 dbcs_char2cells(c) 1600 int c; 1601 { 1602 /* Number of cells is equal to number of bytes, except for euc-jp when 1603 * the first byte is 0x8e. */ 1604 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e) 1605 return 1; 1606 /* use the first byte */ 1607 return MB_BYTE2LEN((unsigned)c >> 8); 1608 } 1609 1610 /* 1611 * Return the number of cells occupied by string "p". 1612 * Stop at a NUL character. When "len" >= 0 stop at character "p[len]". 1613 */ 1614 int 1615 mb_string2cells(p, len) 1616 char_u *p; 1617 int len; 1618 { 1619 int i; 1620 int clen = 0; 1621 1622 for (i = 0; (len < 0 || i < len) && p[i] != NUL; i += (*mb_ptr2len)(p + i)) 1623 clen += (*mb_ptr2cells)(p + i); 1624 return clen; 1625 } 1626 1627 /* 1628 * mb_off2cells() function pointer. 1629 * Return number of display cells for char at ScreenLines[off]. 1630 * We make sure that the offset used is less than "max_off". 1631 */ 1632 int 1633 latin_off2cells(off, max_off) 1634 unsigned off UNUSED; 1635 unsigned max_off UNUSED; 1636 { 1637 return 1; 1638 } 1639 1640 int 1641 dbcs_off2cells(off, max_off) 1642 unsigned off; 1643 unsigned max_off; 1644 { 1645 /* never check beyond end of the line */ 1646 if (off >= max_off) 1647 return 1; 1648 1649 /* Number of cells is equal to number of bytes, except for euc-jp when 1650 * the first byte is 0x8e. */ 1651 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e) 1652 return 1; 1653 return MB_BYTE2LEN(ScreenLines[off]); 1654 } 1655 1656 int 1657 utf_off2cells(off, max_off) 1658 unsigned off; 1659 unsigned max_off; 1660 { 1661 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1; 1662 } 1663 1664 /* 1665 * mb_ptr2char() function pointer. 1666 * Convert a byte sequence into a character. 1667 */ 1668 int 1669 latin_ptr2char(p) 1670 char_u *p; 1671 { 1672 return *p; 1673 } 1674 1675 static int 1676 dbcs_ptr2char(p) 1677 char_u *p; 1678 { 1679 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL) 1680 return (p[0] << 8) + p[1]; 1681 return *p; 1682 } 1683 1684 /* 1685 * Convert a UTF-8 byte sequence to a wide character. 1686 * If the sequence is illegal or truncated by a NUL the first byte is 1687 * returned. 1688 * Does not include composing characters, of course. 1689 */ 1690 int 1691 utf_ptr2char(p) 1692 char_u *p; 1693 { 1694 int len; 1695 1696 if (p[0] < 0x80) /* be quick for ASCII */ 1697 return p[0]; 1698 1699 len = utf8len_tab_zero[p[0]]; 1700 if (len > 1 && (p[1] & 0xc0) == 0x80) 1701 { 1702 if (len == 2) 1703 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); 1704 if ((p[2] & 0xc0) == 0x80) 1705 { 1706 if (len == 3) 1707 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) 1708 + (p[2] & 0x3f); 1709 if ((p[3] & 0xc0) == 0x80) 1710 { 1711 if (len == 4) 1712 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) 1713 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); 1714 if ((p[4] & 0xc0) == 0x80) 1715 { 1716 if (len == 5) 1717 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) 1718 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) 1719 + (p[4] & 0x3f); 1720 if ((p[5] & 0xc0) == 0x80 && len == 6) 1721 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) 1722 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) 1723 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); 1724 } 1725 } 1726 } 1727 } 1728 /* Illegal value, just return the first byte */ 1729 return p[0]; 1730 } 1731 1732 /* 1733 * Convert a UTF-8 byte sequence to a wide character. 1734 * String is assumed to be terminated by NUL or after "n" bytes, whichever 1735 * comes first. 1736 * The function is safe in the sense that it never accesses memory beyond the 1737 * first "n" bytes of "s". 1738 * 1739 * On success, returns decoded codepoint, advances "s" to the beginning of 1740 * next character and decreases "n" accordingly. 1741 * 1742 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past 1743 * NUL byte. 1744 * 1745 * If byte sequence is illegal or incomplete, returns -1 and does not advance 1746 * "s". 1747 */ 1748 static int 1749 utf_safe_read_char_adv(s, n) 1750 char_u **s; 1751 size_t *n; 1752 { 1753 int c, k; 1754 1755 if (*n == 0) /* end of buffer */ 1756 return 0; 1757 1758 k = utf8len_tab_zero[**s]; 1759 1760 if (k == 1) 1761 { 1762 /* ASCII character or NUL */ 1763 (*n)--; 1764 return *(*s)++; 1765 } 1766 1767 if ((size_t)k <= *n) 1768 { 1769 /* We have a multibyte sequence and it isn't truncated by buffer 1770 * limits so utf_ptr2char() is safe to use. Or the first byte is 1771 * illegal (k=0), and it's also safe to use utf_ptr2char(). */ 1772 c = utf_ptr2char(*s); 1773 1774 /* On failure, utf_ptr2char() returns the first byte, so here we 1775 * check equality with the first byte. The only non-ASCII character 1776 * which equals the first byte of its own UTF-8 representation is 1777 * U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too. 1778 * It's safe even if n=1, else we would have k=2 > n. */ 1779 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83)) 1780 { 1781 /* byte sequence was successfully decoded */ 1782 *s += k; 1783 *n -= k; 1784 return c; 1785 } 1786 } 1787 1788 /* byte sequence is incomplete or illegal */ 1789 return -1; 1790 } 1791 1792 /* 1793 * Get character at **pp and advance *pp to the next character. 1794 * Note: composing characters are skipped! 1795 */ 1796 int 1797 mb_ptr2char_adv(pp) 1798 char_u **pp; 1799 { 1800 int c; 1801 1802 c = (*mb_ptr2char)(*pp); 1803 *pp += (*mb_ptr2len)(*pp); 1804 return c; 1805 } 1806 1807 /* 1808 * Get character at **pp and advance *pp to the next character. 1809 * Note: composing characters are returned as separate characters. 1810 */ 1811 int 1812 mb_cptr2char_adv(pp) 1813 char_u **pp; 1814 { 1815 int c; 1816 1817 c = (*mb_ptr2char)(*pp); 1818 if (enc_utf8) 1819 *pp += utf_ptr2len(*pp); 1820 else 1821 *pp += (*mb_ptr2len)(*pp); 1822 return c; 1823 } 1824 1825 #if defined(FEAT_ARABIC) || defined(PROTO) 1826 /* 1827 * Check whether we are dealing with Arabic combining characters. 1828 * Note: these are NOT really composing characters! 1829 */ 1830 int 1831 arabic_combine(one, two) 1832 int one; /* first character */ 1833 int two; /* character just after "one" */ 1834 { 1835 if (one == a_LAM) 1836 return arabic_maycombine(two); 1837 return FALSE; 1838 } 1839 1840 /* 1841 * Check whether we are dealing with a character that could be regarded as an 1842 * Arabic combining character, need to check the character before this. 1843 */ 1844 int 1845 arabic_maycombine(two) 1846 int two; 1847 { 1848 if (p_arshape && !p_tbidi) 1849 return (two == a_ALEF_MADDA 1850 || two == a_ALEF_HAMZA_ABOVE 1851 || two == a_ALEF_HAMZA_BELOW 1852 || two == a_ALEF); 1853 return FALSE; 1854 } 1855 1856 /* 1857 * Check if the character pointed to by "p2" is a composing character when it 1858 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which 1859 * behaves like a composing character. 1860 */ 1861 int 1862 utf_composinglike(p1, p2) 1863 char_u *p1; 1864 char_u *p2; 1865 { 1866 int c2; 1867 1868 c2 = utf_ptr2char(p2); 1869 if (utf_iscomposing(c2)) 1870 return TRUE; 1871 if (!arabic_maycombine(c2)) 1872 return FALSE; 1873 return arabic_combine(utf_ptr2char(p1), c2); 1874 } 1875 #endif 1876 1877 /* 1878 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1879 * composing characters. 1880 */ 1881 int 1882 utfc_ptr2char(p, pcc) 1883 char_u *p; 1884 int *pcc; /* return: composing chars, last one is 0 */ 1885 { 1886 int len; 1887 int c; 1888 int cc; 1889 int i = 0; 1890 1891 c = utf_ptr2char(p); 1892 len = utf_ptr2len(p); 1893 1894 /* Only accept a composing char when the first char isn't illegal. */ 1895 if ((len > 1 || *p < 0x80) 1896 && p[len] >= 0x80 1897 && UTF_COMPOSINGLIKE(p, p + len)) 1898 { 1899 cc = utf_ptr2char(p + len); 1900 for (;;) 1901 { 1902 pcc[i++] = cc; 1903 if (i == MAX_MCO) 1904 break; 1905 len += utf_ptr2len(p + len); 1906 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1907 break; 1908 } 1909 } 1910 1911 if (i < MAX_MCO) /* last composing char must be 0 */ 1912 pcc[i] = 0; 1913 1914 return c; 1915 } 1916 1917 /* 1918 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1919 * composing characters. Use no more than p[maxlen]. 1920 */ 1921 int 1922 utfc_ptr2char_len(p, pcc, maxlen) 1923 char_u *p; 1924 int *pcc; /* return: composing chars, last one is 0 */ 1925 int maxlen; 1926 { 1927 int len; 1928 int c; 1929 int cc; 1930 int i = 0; 1931 1932 c = utf_ptr2char(p); 1933 len = utf_ptr2len_len(p, maxlen); 1934 /* Only accept a composing char when the first char isn't illegal. */ 1935 if ((len > 1 || *p < 0x80) 1936 && len < maxlen 1937 && p[len] >= 0x80 1938 && UTF_COMPOSINGLIKE(p, p + len)) 1939 { 1940 cc = utf_ptr2char(p + len); 1941 for (;;) 1942 { 1943 pcc[i++] = cc; 1944 if (i == MAX_MCO) 1945 break; 1946 len += utf_ptr2len_len(p + len, maxlen - len); 1947 if (len >= maxlen 1948 || p[len] < 0x80 1949 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1950 break; 1951 } 1952 } 1953 1954 if (i < MAX_MCO) /* last composing char must be 0 */ 1955 pcc[i] = 0; 1956 1957 return c; 1958 } 1959 1960 /* 1961 * Convert the character at screen position "off" to a sequence of bytes. 1962 * Includes the composing characters. 1963 * "buf" must at least have the length MB_MAXBYTES + 1. 1964 * Only to be used when ScreenLinesUC[off] != 0. 1965 * Returns the produced number of bytes. 1966 */ 1967 int 1968 utfc_char2bytes(off, buf) 1969 int off; 1970 char_u *buf; 1971 { 1972 int len; 1973 int i; 1974 1975 len = utf_char2bytes(ScreenLinesUC[off], buf); 1976 for (i = 0; i < Screen_mco; ++i) 1977 { 1978 if (ScreenLinesC[i][off] == 0) 1979 break; 1980 len += utf_char2bytes(ScreenLinesC[i][off], buf + len); 1981 } 1982 return len; 1983 } 1984 1985 /* 1986 * Get the length of a UTF-8 byte sequence, not including any following 1987 * composing characters. 1988 * Returns 0 for "". 1989 * Returns 1 for an illegal byte sequence. 1990 */ 1991 int 1992 utf_ptr2len(p) 1993 char_u *p; 1994 { 1995 int len; 1996 int i; 1997 1998 if (*p == NUL) 1999 return 0; 2000 len = utf8len_tab[*p]; 2001 for (i = 1; i < len; ++i) 2002 if ((p[i] & 0xc0) != 0x80) 2003 return 1; 2004 return len; 2005 } 2006 2007 /* 2008 * Return length of UTF-8 character, obtained from the first byte. 2009 * "b" must be between 0 and 255! 2010 * Returns 1 for an invalid first byte value. 2011 */ 2012 int 2013 utf_byte2len(b) 2014 int b; 2015 { 2016 return utf8len_tab[b]; 2017 } 2018 2019 /* 2020 * Get the length of UTF-8 byte sequence "p[size]". Does not include any 2021 * following composing characters. 2022 * Returns 1 for "". 2023 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.). 2024 * Returns number > "size" for an incomplete byte sequence. 2025 * Never returns zero. 2026 */ 2027 int 2028 utf_ptr2len_len(p, size) 2029 char_u *p; 2030 int size; 2031 { 2032 int len; 2033 int i; 2034 int m; 2035 2036 len = utf8len_tab[*p]; 2037 if (len == 1) 2038 return 1; /* NUL, ascii or illegal lead byte */ 2039 if (len > size) 2040 m = size; /* incomplete byte sequence. */ 2041 else 2042 m = len; 2043 for (i = 1; i < m; ++i) 2044 if ((p[i] & 0xc0) != 0x80) 2045 return 1; 2046 return len; 2047 } 2048 2049 /* 2050 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. 2051 * This includes following composing characters. 2052 */ 2053 int 2054 utfc_ptr2len(p) 2055 char_u *p; 2056 { 2057 int len; 2058 int b0 = *p; 2059 #ifdef FEAT_ARABIC 2060 int prevlen; 2061 #endif 2062 2063 if (b0 == NUL) 2064 return 0; 2065 if (b0 < 0x80 && p[1] < 0x80) /* be quick for ASCII */ 2066 return 1; 2067 2068 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2069 len = utf_ptr2len(p); 2070 2071 /* Check for illegal byte. */ 2072 if (len == 1 && b0 >= 0x80) 2073 return 1; 2074 2075 /* 2076 * Check for composing characters. We can handle only the first six, but 2077 * skip all of them (otherwise the cursor would get stuck). 2078 */ 2079 #ifdef FEAT_ARABIC 2080 prevlen = 0; 2081 #endif 2082 for (;;) 2083 { 2084 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2085 return len; 2086 2087 /* Skip over composing char */ 2088 #ifdef FEAT_ARABIC 2089 prevlen = len; 2090 #endif 2091 len += utf_ptr2len(p + len); 2092 } 2093 } 2094 2095 /* 2096 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" 2097 * takes. This includes following composing characters. 2098 * Returns 0 for an empty string. 2099 * Returns 1 for an illegal char or an incomplete byte sequence. 2100 */ 2101 int 2102 utfc_ptr2len_len(p, size) 2103 char_u *p; 2104 int size; 2105 { 2106 int len; 2107 #ifdef FEAT_ARABIC 2108 int prevlen; 2109 #endif 2110 2111 if (size < 1 || *p == NUL) 2112 return 0; 2113 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) /* be quick for ASCII */ 2114 return 1; 2115 2116 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2117 len = utf_ptr2len_len(p, size); 2118 2119 /* Check for illegal byte and incomplete byte sequence. */ 2120 if ((len == 1 && p[0] >= 0x80) || len > size) 2121 return 1; 2122 2123 /* 2124 * Check for composing characters. We can handle only the first six, but 2125 * skip all of them (otherwise the cursor would get stuck). 2126 */ 2127 #ifdef FEAT_ARABIC 2128 prevlen = 0; 2129 #endif 2130 while (len < size) 2131 { 2132 int len_next_char; 2133 2134 if (p[len] < 0x80) 2135 break; 2136 2137 /* 2138 * Next character length should not go beyond size to ensure that 2139 * UTF_COMPOSINGLIKE(...) does not read beyond size. 2140 */ 2141 len_next_char = utf_ptr2len_len(p + len, size - len); 2142 if (len_next_char > size - len) 2143 break; 2144 2145 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2146 break; 2147 2148 /* Skip over composing char */ 2149 #ifdef FEAT_ARABIC 2150 prevlen = len; 2151 #endif 2152 len += len_next_char; 2153 } 2154 return len; 2155 } 2156 2157 /* 2158 * Return the number of bytes the UTF-8 encoding of character "c" takes. 2159 * This does not include composing characters. 2160 */ 2161 int 2162 utf_char2len(c) 2163 int c; 2164 { 2165 if (c < 0x80) 2166 return 1; 2167 if (c < 0x800) 2168 return 2; 2169 if (c < 0x10000) 2170 return 3; 2171 if (c < 0x200000) 2172 return 4; 2173 if (c < 0x4000000) 2174 return 5; 2175 return 6; 2176 } 2177 2178 /* 2179 * Convert Unicode character "c" to UTF-8 string in "buf[]". 2180 * Returns the number of bytes. 2181 * This does not include composing characters. 2182 */ 2183 int 2184 utf_char2bytes(c, buf) 2185 int c; 2186 char_u *buf; 2187 { 2188 if (c < 0x80) /* 7 bits */ 2189 { 2190 buf[0] = c; 2191 return 1; 2192 } 2193 if (c < 0x800) /* 11 bits */ 2194 { 2195 buf[0] = 0xc0 + ((unsigned)c >> 6); 2196 buf[1] = 0x80 + (c & 0x3f); 2197 return 2; 2198 } 2199 if (c < 0x10000) /* 16 bits */ 2200 { 2201 buf[0] = 0xe0 + ((unsigned)c >> 12); 2202 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2203 buf[2] = 0x80 + (c & 0x3f); 2204 return 3; 2205 } 2206 if (c < 0x200000) /* 21 bits */ 2207 { 2208 buf[0] = 0xf0 + ((unsigned)c >> 18); 2209 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2210 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2211 buf[3] = 0x80 + (c & 0x3f); 2212 return 4; 2213 } 2214 if (c < 0x4000000) /* 26 bits */ 2215 { 2216 buf[0] = 0xf8 + ((unsigned)c >> 24); 2217 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2218 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2219 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2220 buf[4] = 0x80 + (c & 0x3f); 2221 return 5; 2222 } 2223 /* 31 bits */ 2224 buf[0] = 0xfc + ((unsigned)c >> 30); 2225 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f); 2226 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2227 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2228 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2229 buf[5] = 0x80 + (c & 0x3f); 2230 return 6; 2231 } 2232 2233 /* 2234 * Return TRUE if "c" is a composing UTF-8 character. This means it will be 2235 * drawn on top of the preceding character. 2236 * Based on code from Markus Kuhn. 2237 */ 2238 int 2239 utf_iscomposing(c) 2240 int c; 2241 { 2242 /* Sorted list of non-overlapping intervals. 2243 * Generated by ../runtime/tools/unicode.vim. */ 2244 static struct interval combining[] = 2245 { 2246 {0x0300, 0x036f}, 2247 {0x0483, 0x0489}, 2248 {0x0591, 0x05bd}, 2249 {0x05bf, 0x05bf}, 2250 {0x05c1, 0x05c2}, 2251 {0x05c4, 0x05c5}, 2252 {0x05c7, 0x05c7}, 2253 {0x0610, 0x061a}, 2254 {0x064b, 0x065f}, 2255 {0x0670, 0x0670}, 2256 {0x06d6, 0x06dc}, 2257 {0x06df, 0x06e4}, 2258 {0x06e7, 0x06e8}, 2259 {0x06ea, 0x06ed}, 2260 {0x0711, 0x0711}, 2261 {0x0730, 0x074a}, 2262 {0x07a6, 0x07b0}, 2263 {0x07eb, 0x07f3}, 2264 {0x0816, 0x0819}, 2265 {0x081b, 0x0823}, 2266 {0x0825, 0x0827}, 2267 {0x0829, 0x082d}, 2268 {0x0859, 0x085b}, 2269 {0x08e4, 0x0903}, 2270 {0x093a, 0x093c}, 2271 {0x093e, 0x094f}, 2272 {0x0951, 0x0957}, 2273 {0x0962, 0x0963}, 2274 {0x0981, 0x0983}, 2275 {0x09bc, 0x09bc}, 2276 {0x09be, 0x09c4}, 2277 {0x09c7, 0x09c8}, 2278 {0x09cb, 0x09cd}, 2279 {0x09d7, 0x09d7}, 2280 {0x09e2, 0x09e3}, 2281 {0x0a01, 0x0a03}, 2282 {0x0a3c, 0x0a3c}, 2283 {0x0a3e, 0x0a42}, 2284 {0x0a47, 0x0a48}, 2285 {0x0a4b, 0x0a4d}, 2286 {0x0a51, 0x0a51}, 2287 {0x0a70, 0x0a71}, 2288 {0x0a75, 0x0a75}, 2289 {0x0a81, 0x0a83}, 2290 {0x0abc, 0x0abc}, 2291 {0x0abe, 0x0ac5}, 2292 {0x0ac7, 0x0ac9}, 2293 {0x0acb, 0x0acd}, 2294 {0x0ae2, 0x0ae3}, 2295 {0x0b01, 0x0b03}, 2296 {0x0b3c, 0x0b3c}, 2297 {0x0b3e, 0x0b44}, 2298 {0x0b47, 0x0b48}, 2299 {0x0b4b, 0x0b4d}, 2300 {0x0b56, 0x0b57}, 2301 {0x0b62, 0x0b63}, 2302 {0x0b82, 0x0b82}, 2303 {0x0bbe, 0x0bc2}, 2304 {0x0bc6, 0x0bc8}, 2305 {0x0bca, 0x0bcd}, 2306 {0x0bd7, 0x0bd7}, 2307 {0x0c00, 0x0c03}, 2308 {0x0c3e, 0x0c44}, 2309 {0x0c46, 0x0c48}, 2310 {0x0c4a, 0x0c4d}, 2311 {0x0c55, 0x0c56}, 2312 {0x0c62, 0x0c63}, 2313 {0x0c81, 0x0c83}, 2314 {0x0cbc, 0x0cbc}, 2315 {0x0cbe, 0x0cc4}, 2316 {0x0cc6, 0x0cc8}, 2317 {0x0cca, 0x0ccd}, 2318 {0x0cd5, 0x0cd6}, 2319 {0x0ce2, 0x0ce3}, 2320 {0x0d01, 0x0d03}, 2321 {0x0d3e, 0x0d44}, 2322 {0x0d46, 0x0d48}, 2323 {0x0d4a, 0x0d4d}, 2324 {0x0d57, 0x0d57}, 2325 {0x0d62, 0x0d63}, 2326 {0x0d82, 0x0d83}, 2327 {0x0dca, 0x0dca}, 2328 {0x0dcf, 0x0dd4}, 2329 {0x0dd6, 0x0dd6}, 2330 {0x0dd8, 0x0ddf}, 2331 {0x0df2, 0x0df3}, 2332 {0x0e31, 0x0e31}, 2333 {0x0e34, 0x0e3a}, 2334 {0x0e47, 0x0e4e}, 2335 {0x0eb1, 0x0eb1}, 2336 {0x0eb4, 0x0eb9}, 2337 {0x0ebb, 0x0ebc}, 2338 {0x0ec8, 0x0ecd}, 2339 {0x0f18, 0x0f19}, 2340 {0x0f35, 0x0f35}, 2341 {0x0f37, 0x0f37}, 2342 {0x0f39, 0x0f39}, 2343 {0x0f3e, 0x0f3f}, 2344 {0x0f71, 0x0f84}, 2345 {0x0f86, 0x0f87}, 2346 {0x0f8d, 0x0f97}, 2347 {0x0f99, 0x0fbc}, 2348 {0x0fc6, 0x0fc6}, 2349 {0x102b, 0x103e}, 2350 {0x1056, 0x1059}, 2351 {0x105e, 0x1060}, 2352 {0x1062, 0x1064}, 2353 {0x1067, 0x106d}, 2354 {0x1071, 0x1074}, 2355 {0x1082, 0x108d}, 2356 {0x108f, 0x108f}, 2357 {0x109a, 0x109d}, 2358 {0x135d, 0x135f}, 2359 {0x1712, 0x1714}, 2360 {0x1732, 0x1734}, 2361 {0x1752, 0x1753}, 2362 {0x1772, 0x1773}, 2363 {0x17b4, 0x17d3}, 2364 {0x17dd, 0x17dd}, 2365 {0x180b, 0x180d}, 2366 {0x18a9, 0x18a9}, 2367 {0x1920, 0x192b}, 2368 {0x1930, 0x193b}, 2369 {0x19b0, 0x19c0}, 2370 {0x19c8, 0x19c9}, 2371 {0x1a17, 0x1a1b}, 2372 {0x1a55, 0x1a5e}, 2373 {0x1a60, 0x1a7c}, 2374 {0x1a7f, 0x1a7f}, 2375 {0x1ab0, 0x1abe}, 2376 {0x1b00, 0x1b04}, 2377 {0x1b34, 0x1b44}, 2378 {0x1b6b, 0x1b73}, 2379 {0x1b80, 0x1b82}, 2380 {0x1ba1, 0x1bad}, 2381 {0x1be6, 0x1bf3}, 2382 {0x1c24, 0x1c37}, 2383 {0x1cd0, 0x1cd2}, 2384 {0x1cd4, 0x1ce8}, 2385 {0x1ced, 0x1ced}, 2386 {0x1cf2, 0x1cf4}, 2387 {0x1cf8, 0x1cf9}, 2388 {0x1dc0, 0x1df5}, 2389 {0x1dfc, 0x1dff}, 2390 {0x20d0, 0x20f0}, 2391 {0x2cef, 0x2cf1}, 2392 {0x2d7f, 0x2d7f}, 2393 {0x2de0, 0x2dff}, 2394 {0x302a, 0x302f}, 2395 {0x3099, 0x309a}, 2396 {0xa66f, 0xa672}, 2397 {0xa674, 0xa67d}, 2398 {0xa69f, 0xa69f}, 2399 {0xa6f0, 0xa6f1}, 2400 {0xa802, 0xa802}, 2401 {0xa806, 0xa806}, 2402 {0xa80b, 0xa80b}, 2403 {0xa823, 0xa827}, 2404 {0xa880, 0xa881}, 2405 {0xa8b4, 0xa8c4}, 2406 {0xa8e0, 0xa8f1}, 2407 {0xa926, 0xa92d}, 2408 {0xa947, 0xa953}, 2409 {0xa980, 0xa983}, 2410 {0xa9b3, 0xa9c0}, 2411 {0xa9e5, 0xa9e5}, 2412 {0xaa29, 0xaa36}, 2413 {0xaa43, 0xaa43}, 2414 {0xaa4c, 0xaa4d}, 2415 {0xaa7b, 0xaa7d}, 2416 {0xaab0, 0xaab0}, 2417 {0xaab2, 0xaab4}, 2418 {0xaab7, 0xaab8}, 2419 {0xaabe, 0xaabf}, 2420 {0xaac1, 0xaac1}, 2421 {0xaaeb, 0xaaef}, 2422 {0xaaf5, 0xaaf6}, 2423 {0xabe3, 0xabea}, 2424 {0xabec, 0xabed}, 2425 {0xfb1e, 0xfb1e}, 2426 {0xfe00, 0xfe0f}, 2427 {0xfe20, 0xfe2d}, 2428 {0x101fd, 0x101fd}, 2429 {0x102e0, 0x102e0}, 2430 {0x10376, 0x1037a}, 2431 {0x10a01, 0x10a03}, 2432 {0x10a05, 0x10a06}, 2433 {0x10a0c, 0x10a0f}, 2434 {0x10a38, 0x10a3a}, 2435 {0x10a3f, 0x10a3f}, 2436 {0x10ae5, 0x10ae6}, 2437 {0x11000, 0x11002}, 2438 {0x11038, 0x11046}, 2439 {0x1107f, 0x11082}, 2440 {0x110b0, 0x110ba}, 2441 {0x11100, 0x11102}, 2442 {0x11127, 0x11134}, 2443 {0x11173, 0x11173}, 2444 {0x11180, 0x11182}, 2445 {0x111b3, 0x111c0}, 2446 {0x1122c, 0x11237}, 2447 {0x112df, 0x112ea}, 2448 {0x11301, 0x11303}, 2449 {0x1133c, 0x1133c}, 2450 {0x1133e, 0x11344}, 2451 {0x11347, 0x11348}, 2452 {0x1134b, 0x1134d}, 2453 {0x11357, 0x11357}, 2454 {0x11362, 0x11363}, 2455 {0x11366, 0x1136c}, 2456 {0x11370, 0x11374}, 2457 {0x114b0, 0x114c3}, 2458 {0x115af, 0x115b5}, 2459 {0x115b8, 0x115c0}, 2460 {0x11630, 0x11640}, 2461 {0x116ab, 0x116b7}, 2462 {0x16af0, 0x16af4}, 2463 {0x16b30, 0x16b36}, 2464 {0x16f51, 0x16f7e}, 2465 {0x16f8f, 0x16f92}, 2466 {0x1bc9d, 0x1bc9e}, 2467 {0x1d165, 0x1d169}, 2468 {0x1d16d, 0x1d172}, 2469 {0x1d17b, 0x1d182}, 2470 {0x1d185, 0x1d18b}, 2471 {0x1d1aa, 0x1d1ad}, 2472 {0x1d242, 0x1d244}, 2473 {0x1e8d0, 0x1e8d6}, 2474 {0xe0100, 0xe01ef} 2475 }; 2476 2477 return intable(combining, sizeof(combining), c); 2478 } 2479 2480 /* 2481 * Return TRUE for characters that can be displayed in a normal way. 2482 * Only for characters of 0x100 and above! 2483 */ 2484 int 2485 utf_printable(c) 2486 int c; 2487 { 2488 #ifdef USE_WCHAR_FUNCTIONS 2489 /* 2490 * Assume the iswprint() library function works better than our own stuff. 2491 */ 2492 return iswprint(c); 2493 #else 2494 /* Sorted list of non-overlapping intervals. 2495 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */ 2496 static struct interval nonprint[] = 2497 { 2498 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2499 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2500 {0xfffe, 0xffff} 2501 }; 2502 2503 return !intable(nonprint, sizeof(nonprint), c); 2504 #endif 2505 } 2506 2507 /* 2508 * Get class of a Unicode character. 2509 * 0: white space 2510 * 1: punctuation 2511 * 2 or bigger: some class of word character. 2512 */ 2513 int 2514 utf_class(c) 2515 int c; 2516 { 2517 /* sorted list of non-overlapping intervals */ 2518 static struct clinterval 2519 { 2520 unsigned int first; 2521 unsigned int last; 2522 unsigned int class; 2523 } classes[] = 2524 { 2525 {0x037e, 0x037e, 1}, /* Greek question mark */ 2526 {0x0387, 0x0387, 1}, /* Greek ano teleia */ 2527 {0x055a, 0x055f, 1}, /* Armenian punctuation */ 2528 {0x0589, 0x0589, 1}, /* Armenian full stop */ 2529 {0x05be, 0x05be, 1}, 2530 {0x05c0, 0x05c0, 1}, 2531 {0x05c3, 0x05c3, 1}, 2532 {0x05f3, 0x05f4, 1}, 2533 {0x060c, 0x060c, 1}, 2534 {0x061b, 0x061b, 1}, 2535 {0x061f, 0x061f, 1}, 2536 {0x066a, 0x066d, 1}, 2537 {0x06d4, 0x06d4, 1}, 2538 {0x0700, 0x070d, 1}, /* Syriac punctuation */ 2539 {0x0964, 0x0965, 1}, 2540 {0x0970, 0x0970, 1}, 2541 {0x0df4, 0x0df4, 1}, 2542 {0x0e4f, 0x0e4f, 1}, 2543 {0x0e5a, 0x0e5b, 1}, 2544 {0x0f04, 0x0f12, 1}, 2545 {0x0f3a, 0x0f3d, 1}, 2546 {0x0f85, 0x0f85, 1}, 2547 {0x104a, 0x104f, 1}, /* Myanmar punctuation */ 2548 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */ 2549 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */ 2550 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */ 2551 {0x1680, 0x1680, 0}, 2552 {0x169b, 0x169c, 1}, 2553 {0x16eb, 0x16ed, 1}, 2554 {0x1735, 0x1736, 1}, 2555 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */ 2556 {0x1800, 0x180a, 1}, /* Mongolian punctuation */ 2557 {0x2000, 0x200b, 0}, /* spaces */ 2558 {0x200c, 0x2027, 1}, /* punctuation and symbols */ 2559 {0x2028, 0x2029, 0}, 2560 {0x202a, 0x202e, 1}, /* punctuation and symbols */ 2561 {0x202f, 0x202f, 0}, 2562 {0x2030, 0x205e, 1}, /* punctuation and symbols */ 2563 {0x205f, 0x205f, 0}, 2564 {0x2060, 0x27ff, 1}, /* punctuation and symbols */ 2565 {0x2070, 0x207f, 0x2070}, /* superscript */ 2566 {0x2080, 0x2094, 0x2080}, /* subscript */ 2567 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */ 2568 {0x2800, 0x28ff, 0x2800}, /* braille */ 2569 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */ 2570 {0x29d8, 0x29db, 1}, 2571 {0x29fc, 0x29fd, 1}, 2572 {0x2e00, 0x2e7f, 1}, /* supplemental punctuation */ 2573 {0x3000, 0x3000, 0}, /* ideographic space */ 2574 {0x3001, 0x3020, 1}, /* ideographic punctuation */ 2575 {0x3030, 0x3030, 1}, 2576 {0x303d, 0x303d, 1}, 2577 {0x3040, 0x309f, 0x3040}, /* Hiragana */ 2578 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */ 2579 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */ 2580 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */ 2581 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */ 2582 {0xfd3e, 0xfd3f, 1}, 2583 {0xfe30, 0xfe6b, 1}, /* punctuation forms */ 2584 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */ 2585 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */ 2586 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */ 2587 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */ 2588 {0x20000, 0x2a6df, 0x4e00}, /* CJK Ideographs */ 2589 {0x2a700, 0x2b73f, 0x4e00}, /* CJK Ideographs */ 2590 {0x2b740, 0x2b81f, 0x4e00}, /* CJK Ideographs */ 2591 {0x2f800, 0x2fa1f, 0x4e00}, /* CJK Ideographs */ 2592 }; 2593 int bot = 0; 2594 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2595 int mid; 2596 2597 /* First quick check for Latin1 characters, use 'iskeyword'. */ 2598 if (c < 0x100) 2599 { 2600 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2601 return 0; /* blank */ 2602 if (vim_iswordc(c)) 2603 return 2; /* word character */ 2604 return 1; /* punctuation */ 2605 } 2606 2607 /* binary search in table */ 2608 while (top >= bot) 2609 { 2610 mid = (bot + top) / 2; 2611 if (classes[mid].last < (unsigned int)c) 2612 bot = mid + 1; 2613 else if (classes[mid].first > (unsigned int)c) 2614 top = mid - 1; 2615 else 2616 return (int)classes[mid].class; 2617 } 2618 2619 /* most other characters are "word" characters */ 2620 return 2; 2621 } 2622 2623 /* 2624 * Code for Unicode case-dependent operations. Based on notes in 2625 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2626 * This code uses simple case folding, not full case folding. 2627 * Last updated for Unicode 5.2. 2628 */ 2629 2630 /* 2631 * The following tables are built by ../runtime/tools/unicode.vim. 2632 * They must be in numeric order, because we use binary search. 2633 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2634 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2635 * folded/upper/lower by adding 32. 2636 */ 2637 typedef struct 2638 { 2639 int rangeStart; 2640 int rangeEnd; 2641 int step; 2642 int offset; 2643 } convertStruct; 2644 2645 static convertStruct foldCase[] = 2646 { 2647 {0x41,0x5a,1,32}, 2648 {0xb5,0xb5,-1,775}, 2649 {0xc0,0xd6,1,32}, 2650 {0xd8,0xde,1,32}, 2651 {0x100,0x12e,2,1}, 2652 {0x132,0x136,2,1}, 2653 {0x139,0x147,2,1}, 2654 {0x14a,0x176,2,1}, 2655 {0x178,0x178,-1,-121}, 2656 {0x179,0x17d,2,1}, 2657 {0x17f,0x17f,-1,-268}, 2658 {0x181,0x181,-1,210}, 2659 {0x182,0x184,2,1}, 2660 {0x186,0x186,-1,206}, 2661 {0x187,0x187,-1,1}, 2662 {0x189,0x18a,1,205}, 2663 {0x18b,0x18b,-1,1}, 2664 {0x18e,0x18e,-1,79}, 2665 {0x18f,0x18f,-1,202}, 2666 {0x190,0x190,-1,203}, 2667 {0x191,0x191,-1,1}, 2668 {0x193,0x193,-1,205}, 2669 {0x194,0x194,-1,207}, 2670 {0x196,0x196,-1,211}, 2671 {0x197,0x197,-1,209}, 2672 {0x198,0x198,-1,1}, 2673 {0x19c,0x19c,-1,211}, 2674 {0x19d,0x19d,-1,213}, 2675 {0x19f,0x19f,-1,214}, 2676 {0x1a0,0x1a4,2,1}, 2677 {0x1a6,0x1a6,-1,218}, 2678 {0x1a7,0x1a7,-1,1}, 2679 {0x1a9,0x1a9,-1,218}, 2680 {0x1ac,0x1ac,-1,1}, 2681 {0x1ae,0x1ae,-1,218}, 2682 {0x1af,0x1af,-1,1}, 2683 {0x1b1,0x1b2,1,217}, 2684 {0x1b3,0x1b5,2,1}, 2685 {0x1b7,0x1b7,-1,219}, 2686 {0x1b8,0x1bc,4,1}, 2687 {0x1c4,0x1c4,-1,2}, 2688 {0x1c5,0x1c5,-1,1}, 2689 {0x1c7,0x1c7,-1,2}, 2690 {0x1c8,0x1c8,-1,1}, 2691 {0x1ca,0x1ca,-1,2}, 2692 {0x1cb,0x1db,2,1}, 2693 {0x1de,0x1ee,2,1}, 2694 {0x1f1,0x1f1,-1,2}, 2695 {0x1f2,0x1f4,2,1}, 2696 {0x1f6,0x1f6,-1,-97}, 2697 {0x1f7,0x1f7,-1,-56}, 2698 {0x1f8,0x21e,2,1}, 2699 {0x220,0x220,-1,-130}, 2700 {0x222,0x232,2,1}, 2701 {0x23a,0x23a,-1,10795}, 2702 {0x23b,0x23b,-1,1}, 2703 {0x23d,0x23d,-1,-163}, 2704 {0x23e,0x23e,-1,10792}, 2705 {0x241,0x241,-1,1}, 2706 {0x243,0x243,-1,-195}, 2707 {0x244,0x244,-1,69}, 2708 {0x245,0x245,-1,71}, 2709 {0x246,0x24e,2,1}, 2710 {0x345,0x345,-1,116}, 2711 {0x370,0x372,2,1}, 2712 {0x376,0x376,-1,1}, 2713 {0x37f,0x37f,-1,116}, 2714 {0x386,0x386,-1,38}, 2715 {0x388,0x38a,1,37}, 2716 {0x38c,0x38c,-1,64}, 2717 {0x38e,0x38f,1,63}, 2718 {0x391,0x3a1,1,32}, 2719 {0x3a3,0x3ab,1,32}, 2720 {0x3c2,0x3c2,-1,1}, 2721 {0x3cf,0x3cf,-1,8}, 2722 {0x3d0,0x3d0,-1,-30}, 2723 {0x3d1,0x3d1,-1,-25}, 2724 {0x3d5,0x3d5,-1,-15}, 2725 {0x3d6,0x3d6,-1,-22}, 2726 {0x3d8,0x3ee,2,1}, 2727 {0x3f0,0x3f0,-1,-54}, 2728 {0x3f1,0x3f1,-1,-48}, 2729 {0x3f4,0x3f4,-1,-60}, 2730 {0x3f5,0x3f5,-1,-64}, 2731 {0x3f7,0x3f7,-1,1}, 2732 {0x3f9,0x3f9,-1,-7}, 2733 {0x3fa,0x3fa,-1,1}, 2734 {0x3fd,0x3ff,1,-130}, 2735 {0x400,0x40f,1,80}, 2736 {0x410,0x42f,1,32}, 2737 {0x460,0x480,2,1}, 2738 {0x48a,0x4be,2,1}, 2739 {0x4c0,0x4c0,-1,15}, 2740 {0x4c1,0x4cd,2,1}, 2741 {0x4d0,0x52e,2,1}, 2742 {0x531,0x556,1,48}, 2743 {0x10a0,0x10c5,1,7264}, 2744 {0x10c7,0x10cd,6,7264}, 2745 {0x1e00,0x1e94,2,1}, 2746 {0x1e9b,0x1e9b,-1,-58}, 2747 {0x1e9e,0x1e9e,-1,-7615}, 2748 {0x1ea0,0x1efe,2,1}, 2749 {0x1f08,0x1f0f,1,-8}, 2750 {0x1f18,0x1f1d,1,-8}, 2751 {0x1f28,0x1f2f,1,-8}, 2752 {0x1f38,0x1f3f,1,-8}, 2753 {0x1f48,0x1f4d,1,-8}, 2754 {0x1f59,0x1f5f,2,-8}, 2755 {0x1f68,0x1f6f,1,-8}, 2756 {0x1f88,0x1f8f,1,-8}, 2757 {0x1f98,0x1f9f,1,-8}, 2758 {0x1fa8,0x1faf,1,-8}, 2759 {0x1fb8,0x1fb9,1,-8}, 2760 {0x1fba,0x1fbb,1,-74}, 2761 {0x1fbc,0x1fbc,-1,-9}, 2762 {0x1fbe,0x1fbe,-1,-7173}, 2763 {0x1fc8,0x1fcb,1,-86}, 2764 {0x1fcc,0x1fcc,-1,-9}, 2765 {0x1fd8,0x1fd9,1,-8}, 2766 {0x1fda,0x1fdb,1,-100}, 2767 {0x1fe8,0x1fe9,1,-8}, 2768 {0x1fea,0x1feb,1,-112}, 2769 {0x1fec,0x1fec,-1,-7}, 2770 {0x1ff8,0x1ff9,1,-128}, 2771 {0x1ffa,0x1ffb,1,-126}, 2772 {0x1ffc,0x1ffc,-1,-9}, 2773 {0x2126,0x2126,-1,-7517}, 2774 {0x212a,0x212a,-1,-8383}, 2775 {0x212b,0x212b,-1,-8262}, 2776 {0x2132,0x2132,-1,28}, 2777 {0x2160,0x216f,1,16}, 2778 {0x2183,0x2183,-1,1}, 2779 {0x24b6,0x24cf,1,26}, 2780 {0x2c00,0x2c2e,1,48}, 2781 {0x2c60,0x2c60,-1,1}, 2782 {0x2c62,0x2c62,-1,-10743}, 2783 {0x2c63,0x2c63,-1,-3814}, 2784 {0x2c64,0x2c64,-1,-10727}, 2785 {0x2c67,0x2c6b,2,1}, 2786 {0x2c6d,0x2c6d,-1,-10780}, 2787 {0x2c6e,0x2c6e,-1,-10749}, 2788 {0x2c6f,0x2c6f,-1,-10783}, 2789 {0x2c70,0x2c70,-1,-10782}, 2790 {0x2c72,0x2c75,3,1}, 2791 {0x2c7e,0x2c7f,1,-10815}, 2792 {0x2c80,0x2ce2,2,1}, 2793 {0x2ceb,0x2ced,2,1}, 2794 {0x2cf2,0xa640,31054,1}, 2795 {0xa642,0xa66c,2,1}, 2796 {0xa680,0xa69a,2,1}, 2797 {0xa722,0xa72e,2,1}, 2798 {0xa732,0xa76e,2,1}, 2799 {0xa779,0xa77b,2,1}, 2800 {0xa77d,0xa77d,-1,-35332}, 2801 {0xa77e,0xa786,2,1}, 2802 {0xa78b,0xa78b,-1,1}, 2803 {0xa78d,0xa78d,-1,-42280}, 2804 {0xa790,0xa792,2,1}, 2805 {0xa796,0xa7a8,2,1}, 2806 {0xa7aa,0xa7aa,-1,-42308}, 2807 {0xa7ab,0xa7ab,-1,-42319}, 2808 {0xa7ac,0xa7ac,-1,-42315}, 2809 {0xa7ad,0xa7ad,-1,-42305}, 2810 {0xa7b0,0xa7b0,-1,-42258}, 2811 {0xa7b1,0xa7b1,-1,-42282}, 2812 {0xff21,0xff3a,1,32}, 2813 {0x10400,0x10427,1,40}, 2814 {0x118a0,0x118bf,1,32} 2815 }; 2816 2817 static int utf_convert __ARGS((int a, convertStruct table[], int tableSize)); 2818 static int utf_strnicmp __ARGS((char_u *s1, char_u *s2, size_t n1, size_t n2)); 2819 2820 /* 2821 * Generic conversion function for case operations. 2822 * Return the converted equivalent of "a", which is a UCS-4 character. Use 2823 * the given conversion "table". Uses binary search on "table". 2824 */ 2825 static int 2826 utf_convert(a, table, tableSize) 2827 int a; 2828 convertStruct table[]; 2829 int tableSize; 2830 { 2831 int start, mid, end; /* indices into table */ 2832 int entries = tableSize / sizeof(convertStruct); 2833 2834 start = 0; 2835 end = entries; 2836 while (start < end) 2837 { 2838 /* need to search further */ 2839 mid = (end + start) / 2; 2840 if (table[mid].rangeEnd < a) 2841 start = mid + 1; 2842 else 2843 end = mid; 2844 } 2845 if (start < entries 2846 && table[start].rangeStart <= a 2847 && a <= table[start].rangeEnd 2848 && (a - table[start].rangeStart) % table[start].step == 0) 2849 return (a + table[start].offset); 2850 else 2851 return a; 2852 } 2853 2854 /* 2855 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 2856 * simple case folding. 2857 */ 2858 int 2859 utf_fold(a) 2860 int a; 2861 { 2862 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 2863 } 2864 2865 static convertStruct toLower[] = 2866 { 2867 {0x41,0x5a,1,32}, 2868 {0xc0,0xd6,1,32}, 2869 {0xd8,0xde,1,32}, 2870 {0x100,0x12e,2,1}, 2871 {0x130,0x130,-1,-199}, 2872 {0x132,0x136,2,1}, 2873 {0x139,0x147,2,1}, 2874 {0x14a,0x176,2,1}, 2875 {0x178,0x178,-1,-121}, 2876 {0x179,0x17d,2,1}, 2877 {0x181,0x181,-1,210}, 2878 {0x182,0x184,2,1}, 2879 {0x186,0x186,-1,206}, 2880 {0x187,0x187,-1,1}, 2881 {0x189,0x18a,1,205}, 2882 {0x18b,0x18b,-1,1}, 2883 {0x18e,0x18e,-1,79}, 2884 {0x18f,0x18f,-1,202}, 2885 {0x190,0x190,-1,203}, 2886 {0x191,0x191,-1,1}, 2887 {0x193,0x193,-1,205}, 2888 {0x194,0x194,-1,207}, 2889 {0x196,0x196,-1,211}, 2890 {0x197,0x197,-1,209}, 2891 {0x198,0x198,-1,1}, 2892 {0x19c,0x19c,-1,211}, 2893 {0x19d,0x19d,-1,213}, 2894 {0x19f,0x19f,-1,214}, 2895 {0x1a0,0x1a4,2,1}, 2896 {0x1a6,0x1a6,-1,218}, 2897 {0x1a7,0x1a7,-1,1}, 2898 {0x1a9,0x1a9,-1,218}, 2899 {0x1ac,0x1ac,-1,1}, 2900 {0x1ae,0x1ae,-1,218}, 2901 {0x1af,0x1af,-1,1}, 2902 {0x1b1,0x1b2,1,217}, 2903 {0x1b3,0x1b5,2,1}, 2904 {0x1b7,0x1b7,-1,219}, 2905 {0x1b8,0x1bc,4,1}, 2906 {0x1c4,0x1c4,-1,2}, 2907 {0x1c5,0x1c5,-1,1}, 2908 {0x1c7,0x1c7,-1,2}, 2909 {0x1c8,0x1c8,-1,1}, 2910 {0x1ca,0x1ca,-1,2}, 2911 {0x1cb,0x1db,2,1}, 2912 {0x1de,0x1ee,2,1}, 2913 {0x1f1,0x1f1,-1,2}, 2914 {0x1f2,0x1f4,2,1}, 2915 {0x1f6,0x1f6,-1,-97}, 2916 {0x1f7,0x1f7,-1,-56}, 2917 {0x1f8,0x21e,2,1}, 2918 {0x220,0x220,-1,-130}, 2919 {0x222,0x232,2,1}, 2920 {0x23a,0x23a,-1,10795}, 2921 {0x23b,0x23b,-1,1}, 2922 {0x23d,0x23d,-1,-163}, 2923 {0x23e,0x23e,-1,10792}, 2924 {0x241,0x241,-1,1}, 2925 {0x243,0x243,-1,-195}, 2926 {0x244,0x244,-1,69}, 2927 {0x245,0x245,-1,71}, 2928 {0x246,0x24e,2,1}, 2929 {0x370,0x372,2,1}, 2930 {0x376,0x376,-1,1}, 2931 {0x37f,0x37f,-1,116}, 2932 {0x386,0x386,-1,38}, 2933 {0x388,0x38a,1,37}, 2934 {0x38c,0x38c,-1,64}, 2935 {0x38e,0x38f,1,63}, 2936 {0x391,0x3a1,1,32}, 2937 {0x3a3,0x3ab,1,32}, 2938 {0x3cf,0x3cf,-1,8}, 2939 {0x3d8,0x3ee,2,1}, 2940 {0x3f4,0x3f4,-1,-60}, 2941 {0x3f7,0x3f7,-1,1}, 2942 {0x3f9,0x3f9,-1,-7}, 2943 {0x3fa,0x3fa,-1,1}, 2944 {0x3fd,0x3ff,1,-130}, 2945 {0x400,0x40f,1,80}, 2946 {0x410,0x42f,1,32}, 2947 {0x460,0x480,2,1}, 2948 {0x48a,0x4be,2,1}, 2949 {0x4c0,0x4c0,-1,15}, 2950 {0x4c1,0x4cd,2,1}, 2951 {0x4d0,0x52e,2,1}, 2952 {0x531,0x556,1,48}, 2953 {0x10a0,0x10c5,1,7264}, 2954 {0x10c7,0x10cd,6,7264}, 2955 {0x1e00,0x1e94,2,1}, 2956 {0x1e9e,0x1e9e,-1,-7615}, 2957 {0x1ea0,0x1efe,2,1}, 2958 {0x1f08,0x1f0f,1,-8}, 2959 {0x1f18,0x1f1d,1,-8}, 2960 {0x1f28,0x1f2f,1,-8}, 2961 {0x1f38,0x1f3f,1,-8}, 2962 {0x1f48,0x1f4d,1,-8}, 2963 {0x1f59,0x1f5f,2,-8}, 2964 {0x1f68,0x1f6f,1,-8}, 2965 {0x1f88,0x1f8f,1,-8}, 2966 {0x1f98,0x1f9f,1,-8}, 2967 {0x1fa8,0x1faf,1,-8}, 2968 {0x1fb8,0x1fb9,1,-8}, 2969 {0x1fba,0x1fbb,1,-74}, 2970 {0x1fbc,0x1fbc,-1,-9}, 2971 {0x1fc8,0x1fcb,1,-86}, 2972 {0x1fcc,0x1fcc,-1,-9}, 2973 {0x1fd8,0x1fd9,1,-8}, 2974 {0x1fda,0x1fdb,1,-100}, 2975 {0x1fe8,0x1fe9,1,-8}, 2976 {0x1fea,0x1feb,1,-112}, 2977 {0x1fec,0x1fec,-1,-7}, 2978 {0x1ff8,0x1ff9,1,-128}, 2979 {0x1ffa,0x1ffb,1,-126}, 2980 {0x1ffc,0x1ffc,-1,-9}, 2981 {0x2126,0x2126,-1,-7517}, 2982 {0x212a,0x212a,-1,-8383}, 2983 {0x212b,0x212b,-1,-8262}, 2984 {0x2132,0x2132,-1,28}, 2985 {0x2160,0x216f,1,16}, 2986 {0x2183,0x2183,-1,1}, 2987 {0x24b6,0x24cf,1,26}, 2988 {0x2c00,0x2c2e,1,48}, 2989 {0x2c60,0x2c60,-1,1}, 2990 {0x2c62,0x2c62,-1,-10743}, 2991 {0x2c63,0x2c63,-1,-3814}, 2992 {0x2c64,0x2c64,-1,-10727}, 2993 {0x2c67,0x2c6b,2,1}, 2994 {0x2c6d,0x2c6d,-1,-10780}, 2995 {0x2c6e,0x2c6e,-1,-10749}, 2996 {0x2c6f,0x2c6f,-1,-10783}, 2997 {0x2c70,0x2c70,-1,-10782}, 2998 {0x2c72,0x2c75,3,1}, 2999 {0x2c7e,0x2c7f,1,-10815}, 3000 {0x2c80,0x2ce2,2,1}, 3001 {0x2ceb,0x2ced,2,1}, 3002 {0x2cf2,0xa640,31054,1}, 3003 {0xa642,0xa66c,2,1}, 3004 {0xa680,0xa69a,2,1}, 3005 {0xa722,0xa72e,2,1}, 3006 {0xa732,0xa76e,2,1}, 3007 {0xa779,0xa77b,2,1}, 3008 {0xa77d,0xa77d,-1,-35332}, 3009 {0xa77e,0xa786,2,1}, 3010 {0xa78b,0xa78b,-1,1}, 3011 {0xa78d,0xa78d,-1,-42280}, 3012 {0xa790,0xa792,2,1}, 3013 {0xa796,0xa7a8,2,1}, 3014 {0xa7aa,0xa7aa,-1,-42308}, 3015 {0xa7ab,0xa7ab,-1,-42319}, 3016 {0xa7ac,0xa7ac,-1,-42315}, 3017 {0xa7ad,0xa7ad,-1,-42305}, 3018 {0xa7b0,0xa7b0,-1,-42258}, 3019 {0xa7b1,0xa7b1,-1,-42282}, 3020 {0xff21,0xff3a,1,32}, 3021 {0x10400,0x10427,1,40}, 3022 {0x118a0,0x118bf,1,32} 3023 }; 3024 3025 static convertStruct toUpper[] = 3026 { 3027 {0x61,0x7a,1,-32}, 3028 {0xb5,0xb5,-1,743}, 3029 {0xe0,0xf6,1,-32}, 3030 {0xf8,0xfe,1,-32}, 3031 {0xff,0xff,-1,121}, 3032 {0x101,0x12f,2,-1}, 3033 {0x131,0x131,-1,-232}, 3034 {0x133,0x137,2,-1}, 3035 {0x13a,0x148,2,-1}, 3036 {0x14b,0x177,2,-1}, 3037 {0x17a,0x17e,2,-1}, 3038 {0x17f,0x17f,-1,-300}, 3039 {0x180,0x180,-1,195}, 3040 {0x183,0x185,2,-1}, 3041 {0x188,0x18c,4,-1}, 3042 {0x192,0x192,-1,-1}, 3043 {0x195,0x195,-1,97}, 3044 {0x199,0x199,-1,-1}, 3045 {0x19a,0x19a,-1,163}, 3046 {0x19e,0x19e,-1,130}, 3047 {0x1a1,0x1a5,2,-1}, 3048 {0x1a8,0x1ad,5,-1}, 3049 {0x1b0,0x1b4,4,-1}, 3050 {0x1b6,0x1b9,3,-1}, 3051 {0x1bd,0x1bd,-1,-1}, 3052 {0x1bf,0x1bf,-1,56}, 3053 {0x1c5,0x1c5,-1,-1}, 3054 {0x1c6,0x1c6,-1,-2}, 3055 {0x1c8,0x1c8,-1,-1}, 3056 {0x1c9,0x1c9,-1,-2}, 3057 {0x1cb,0x1cb,-1,-1}, 3058 {0x1cc,0x1cc,-1,-2}, 3059 {0x1ce,0x1dc,2,-1}, 3060 {0x1dd,0x1dd,-1,-79}, 3061 {0x1df,0x1ef,2,-1}, 3062 {0x1f2,0x1f2,-1,-1}, 3063 {0x1f3,0x1f3,-1,-2}, 3064 {0x1f5,0x1f9,4,-1}, 3065 {0x1fb,0x21f,2,-1}, 3066 {0x223,0x233,2,-1}, 3067 {0x23c,0x23c,-1,-1}, 3068 {0x23f,0x240,1,10815}, 3069 {0x242,0x247,5,-1}, 3070 {0x249,0x24f,2,-1}, 3071 {0x250,0x250,-1,10783}, 3072 {0x251,0x251,-1,10780}, 3073 {0x252,0x252,-1,10782}, 3074 {0x253,0x253,-1,-210}, 3075 {0x254,0x254,-1,-206}, 3076 {0x256,0x257,1,-205}, 3077 {0x259,0x259,-1,-202}, 3078 {0x25b,0x25b,-1,-203}, 3079 {0x25c,0x25c,-1,42319}, 3080 {0x260,0x260,-1,-205}, 3081 {0x261,0x261,-1,42315}, 3082 {0x263,0x263,-1,-207}, 3083 {0x265,0x265,-1,42280}, 3084 {0x266,0x266,-1,42308}, 3085 {0x268,0x268,-1,-209}, 3086 {0x269,0x269,-1,-211}, 3087 {0x26b,0x26b,-1,10743}, 3088 {0x26c,0x26c,-1,42305}, 3089 {0x26f,0x26f,-1,-211}, 3090 {0x271,0x271,-1,10749}, 3091 {0x272,0x272,-1,-213}, 3092 {0x275,0x275,-1,-214}, 3093 {0x27d,0x27d,-1,10727}, 3094 {0x280,0x283,3,-218}, 3095 {0x287,0x287,-1,42282}, 3096 {0x288,0x288,-1,-218}, 3097 {0x289,0x289,-1,-69}, 3098 {0x28a,0x28b,1,-217}, 3099 {0x28c,0x28c,-1,-71}, 3100 {0x292,0x292,-1,-219}, 3101 {0x29e,0x29e,-1,42258}, 3102 {0x345,0x345,-1,84}, 3103 {0x371,0x373,2,-1}, 3104 {0x377,0x377,-1,-1}, 3105 {0x37b,0x37d,1,130}, 3106 {0x3ac,0x3ac,-1,-38}, 3107 {0x3ad,0x3af,1,-37}, 3108 {0x3b1,0x3c1,1,-32}, 3109 {0x3c2,0x3c2,-1,-31}, 3110 {0x3c3,0x3cb,1,-32}, 3111 {0x3cc,0x3cc,-1,-64}, 3112 {0x3cd,0x3ce,1,-63}, 3113 {0x3d0,0x3d0,-1,-62}, 3114 {0x3d1,0x3d1,-1,-57}, 3115 {0x3d5,0x3d5,-1,-47}, 3116 {0x3d6,0x3d6,-1,-54}, 3117 {0x3d7,0x3d7,-1,-8}, 3118 {0x3d9,0x3ef,2,-1}, 3119 {0x3f0,0x3f0,-1,-86}, 3120 {0x3f1,0x3f1,-1,-80}, 3121 {0x3f2,0x3f2,-1,7}, 3122 {0x3f3,0x3f3,-1,-116}, 3123 {0x3f5,0x3f5,-1,-96}, 3124 {0x3f8,0x3fb,3,-1}, 3125 {0x430,0x44f,1,-32}, 3126 {0x450,0x45f,1,-80}, 3127 {0x461,0x481,2,-1}, 3128 {0x48b,0x4bf,2,-1}, 3129 {0x4c2,0x4ce,2,-1}, 3130 {0x4cf,0x4cf,-1,-15}, 3131 {0x4d1,0x52f,2,-1}, 3132 {0x561,0x586,1,-48}, 3133 {0x1d79,0x1d79,-1,35332}, 3134 {0x1d7d,0x1d7d,-1,3814}, 3135 {0x1e01,0x1e95,2,-1}, 3136 {0x1e9b,0x1e9b,-1,-59}, 3137 {0x1ea1,0x1eff,2,-1}, 3138 {0x1f00,0x1f07,1,8}, 3139 {0x1f10,0x1f15,1,8}, 3140 {0x1f20,0x1f27,1,8}, 3141 {0x1f30,0x1f37,1,8}, 3142 {0x1f40,0x1f45,1,8}, 3143 {0x1f51,0x1f57,2,8}, 3144 {0x1f60,0x1f67,1,8}, 3145 {0x1f70,0x1f71,1,74}, 3146 {0x1f72,0x1f75,1,86}, 3147 {0x1f76,0x1f77,1,100}, 3148 {0x1f78,0x1f79,1,128}, 3149 {0x1f7a,0x1f7b,1,112}, 3150 {0x1f7c,0x1f7d,1,126}, 3151 {0x1f80,0x1f87,1,8}, 3152 {0x1f90,0x1f97,1,8}, 3153 {0x1fa0,0x1fa7,1,8}, 3154 {0x1fb0,0x1fb1,1,8}, 3155 {0x1fb3,0x1fb3,-1,9}, 3156 {0x1fbe,0x1fbe,-1,-7205}, 3157 {0x1fc3,0x1fc3,-1,9}, 3158 {0x1fd0,0x1fd1,1,8}, 3159 {0x1fe0,0x1fe1,1,8}, 3160 {0x1fe5,0x1fe5,-1,7}, 3161 {0x1ff3,0x1ff3,-1,9}, 3162 {0x214e,0x214e,-1,-28}, 3163 {0x2170,0x217f,1,-16}, 3164 {0x2184,0x2184,-1,-1}, 3165 {0x24d0,0x24e9,1,-26}, 3166 {0x2c30,0x2c5e,1,-48}, 3167 {0x2c61,0x2c61,-1,-1}, 3168 {0x2c65,0x2c65,-1,-10795}, 3169 {0x2c66,0x2c66,-1,-10792}, 3170 {0x2c68,0x2c6c,2,-1}, 3171 {0x2c73,0x2c76,3,-1}, 3172 {0x2c81,0x2ce3,2,-1}, 3173 {0x2cec,0x2cee,2,-1}, 3174 {0x2cf3,0x2cf3,-1,-1}, 3175 {0x2d00,0x2d25,1,-7264}, 3176 {0x2d27,0x2d2d,6,-7264}, 3177 {0xa641,0xa66d,2,-1}, 3178 {0xa681,0xa69b,2,-1}, 3179 {0xa723,0xa72f,2,-1}, 3180 {0xa733,0xa76f,2,-1}, 3181 {0xa77a,0xa77c,2,-1}, 3182 {0xa77f,0xa787,2,-1}, 3183 {0xa78c,0xa791,5,-1}, 3184 {0xa793,0xa797,4,-1}, 3185 {0xa799,0xa7a9,2,-1}, 3186 {0xff41,0xff5a,1,-32}, 3187 {0x10428,0x1044f,1,-40}, 3188 {0x118c0,0x118df,1,-32} 3189 }; 3190 /* 3191 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3192 * simple case folding. 3193 */ 3194 int 3195 utf_toupper(a) 3196 int a; 3197 { 3198 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */ 3199 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3200 return TOUPPER_ASC(a); 3201 3202 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3203 /* If towupper() is available and handles Unicode, use it. */ 3204 if (!(cmp_flags & CMP_INTERNAL)) 3205 return towupper(a); 3206 #endif 3207 3208 /* For characters below 128 use locale sensitive toupper(). */ 3209 if (a < 128) 3210 return TOUPPER_LOC(a); 3211 3212 /* For any other characters use the above mapping table. */ 3213 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3214 } 3215 3216 int 3217 utf_islower(a) 3218 int a; 3219 { 3220 /* German sharp s is lower case but has no upper case equivalent. */ 3221 return (utf_toupper(a) != a) || a == 0xdf; 3222 } 3223 3224 /* 3225 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3226 * simple case folding. 3227 */ 3228 int 3229 utf_tolower(a) 3230 int a; 3231 { 3232 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */ 3233 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3234 return TOLOWER_ASC(a); 3235 3236 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3237 /* If towlower() is available and handles Unicode, use it. */ 3238 if (!(cmp_flags & CMP_INTERNAL)) 3239 return towlower(a); 3240 #endif 3241 3242 /* For characters below 128 use locale sensitive tolower(). */ 3243 if (a < 128) 3244 return TOLOWER_LOC(a); 3245 3246 /* For any other characters use the above mapping table. */ 3247 return utf_convert(a, toLower, (int)sizeof(toLower)); 3248 } 3249 3250 int 3251 utf_isupper(a) 3252 int a; 3253 { 3254 return (utf_tolower(a) != a); 3255 } 3256 3257 static int 3258 utf_strnicmp(s1, s2, n1, n2) 3259 char_u *s1, *s2; 3260 size_t n1, n2; 3261 { 3262 int c1, c2, cdiff; 3263 char_u buffer[6]; 3264 3265 for (;;) 3266 { 3267 c1 = utf_safe_read_char_adv(&s1, &n1); 3268 c2 = utf_safe_read_char_adv(&s2, &n2); 3269 3270 if (c1 <= 0 || c2 <= 0) 3271 break; 3272 3273 if (c1 == c2) 3274 continue; 3275 3276 cdiff = utf_fold(c1) - utf_fold(c2); 3277 if (cdiff != 0) 3278 return cdiff; 3279 } 3280 3281 /* some string ended or has an incomplete/illegal character sequence */ 3282 3283 if (c1 == 0 || c2 == 0) 3284 { 3285 /* some string ended. shorter string is smaller */ 3286 if (c1 == 0 && c2 == 0) 3287 return 0; 3288 return c1 == 0 ? -1 : 1; 3289 } 3290 3291 /* Continue with bytewise comparison to produce some result that 3292 * would make comparison operations involving this function transitive. 3293 * 3294 * If only one string had an error, comparison should be made with 3295 * folded version of the other string. In this case it is enough 3296 * to fold just one character to determine the result of comparison. */ 3297 3298 if (c1 != -1 && c2 == -1) 3299 { 3300 n1 = utf_char2bytes(utf_fold(c1), buffer); 3301 s1 = buffer; 3302 } 3303 else if (c2 != -1 && c1 == -1) 3304 { 3305 n2 = utf_char2bytes(utf_fold(c2), buffer); 3306 s2 = buffer; 3307 } 3308 3309 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3310 { 3311 cdiff = (int)(*s1) - (int)(*s2); 3312 if (cdiff != 0) 3313 return cdiff; 3314 3315 s1++; 3316 s2++; 3317 n1--; 3318 n2--; 3319 } 3320 3321 if (n1 > 0 && *s1 == NUL) 3322 n1 = 0; 3323 if (n2 > 0 && *s2 == NUL) 3324 n2 = 0; 3325 3326 if (n1 == 0 && n2 == 0) 3327 return 0; 3328 return n1 == 0 ? -1 : 1; 3329 } 3330 3331 /* 3332 * Version of strnicmp() that handles multi-byte characters. 3333 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3334 * probably use strnicmp(), because there are no ASCII characters in the 3335 * second byte. 3336 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3337 * two characters otherwise. 3338 */ 3339 int 3340 mb_strnicmp(s1, s2, nn) 3341 char_u *s1, *s2; 3342 size_t nn; 3343 { 3344 int i, l; 3345 int cdiff; 3346 int n = (int)nn; 3347 3348 if (enc_utf8) 3349 { 3350 return utf_strnicmp(s1, s2, nn, nn); 3351 } 3352 else 3353 { 3354 for (i = 0; i < n; i += l) 3355 { 3356 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */ 3357 return 0; 3358 3359 l = (*mb_ptr2len)(s1 + i); 3360 if (l <= 1) 3361 { 3362 /* Single byte: first check normally, then with ignore case. */ 3363 if (s1[i] != s2[i]) 3364 { 3365 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3366 if (cdiff != 0) 3367 return cdiff; 3368 } 3369 } 3370 else 3371 { 3372 /* For non-Unicode multi-byte don't ignore case. */ 3373 if (l > n - i) 3374 l = n - i; 3375 cdiff = STRNCMP(s1 + i, s2 + i, l); 3376 if (cdiff != 0) 3377 return cdiff; 3378 } 3379 } 3380 } 3381 return 0; 3382 } 3383 3384 /* 3385 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3386 * 'encoding' has been set to. 3387 */ 3388 void 3389 show_utf8() 3390 { 3391 int len; 3392 int rlen = 0; 3393 char_u *line; 3394 int clen; 3395 int i; 3396 3397 /* Get the byte length of the char under the cursor, including composing 3398 * characters. */ 3399 line = ml_get_cursor(); 3400 len = utfc_ptr2len(line); 3401 if (len == 0) 3402 { 3403 MSG("NUL"); 3404 return; 3405 } 3406 3407 clen = 0; 3408 for (i = 0; i < len; ++i) 3409 { 3410 if (clen == 0) 3411 { 3412 /* start of (composing) character, get its length */ 3413 if (i > 0) 3414 { 3415 STRCPY(IObuff + rlen, "+ "); 3416 rlen += 2; 3417 } 3418 clen = utf_ptr2len(line + i); 3419 } 3420 sprintf((char *)IObuff + rlen, "%02x ", 3421 (line[i] == NL) ? NUL : line[i]); /* NUL is stored as NL */ 3422 --clen; 3423 rlen += (int)STRLEN(IObuff + rlen); 3424 if (rlen > IOSIZE - 20) 3425 break; 3426 } 3427 3428 msg(IObuff); 3429 } 3430 3431 /* 3432 * mb_head_off() function pointer. 3433 * Return offset from "p" to the first byte of the character it points into. 3434 * If "p" points to the NUL at the end of the string return 0. 3435 * Returns 0 when already at the first byte of a character. 3436 */ 3437 int 3438 latin_head_off(base, p) 3439 char_u *base UNUSED; 3440 char_u *p UNUSED; 3441 { 3442 return 0; 3443 } 3444 3445 int 3446 dbcs_head_off(base, p) 3447 char_u *base; 3448 char_u *p; 3449 { 3450 char_u *q; 3451 3452 /* It can't be a trailing byte when not using DBCS, at the start of the 3453 * string or the previous byte can't start a double-byte. */ 3454 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3455 return 0; 3456 3457 /* This is slow: need to start at the base and go forward until the 3458 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */ 3459 q = base; 3460 while (q < p) 3461 q += dbcs_ptr2len(q); 3462 return (q == p) ? 0 : 1; 3463 } 3464 3465 /* 3466 * Special version of dbcs_head_off() that works for ScreenLines[], where 3467 * single-width DBCS_JPNU characters are stored separately. 3468 */ 3469 int 3470 dbcs_screen_head_off(base, p) 3471 char_u *base; 3472 char_u *p; 3473 { 3474 char_u *q; 3475 3476 /* It can't be a trailing byte when not using DBCS, at the start of the 3477 * string or the previous byte can't start a double-byte. 3478 * For euc-jp an 0x8e byte in the previous cell always means we have a 3479 * lead byte in the current cell. */ 3480 if (p <= base 3481 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3482 || MB_BYTE2LEN(p[-1]) == 1 3483 || *p == NUL) 3484 return 0; 3485 3486 /* This is slow: need to start at the base and go forward until the 3487 * byte we are looking for. Return 1 when we went past it, 0 otherwise. 3488 * For DBCS_JPNU look out for 0x8e, which means the second byte is not 3489 * stored as the next byte. */ 3490 q = base; 3491 while (q < p) 3492 { 3493 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3494 ++q; 3495 else 3496 q += dbcs_ptr2len(q); 3497 } 3498 return (q == p) ? 0 : 1; 3499 } 3500 3501 int 3502 utf_head_off(base, p) 3503 char_u *base; 3504 char_u *p; 3505 { 3506 char_u *q; 3507 char_u *s; 3508 int c; 3509 int len; 3510 #ifdef FEAT_ARABIC 3511 char_u *j; 3512 #endif 3513 3514 if (*p < 0x80) /* be quick for ASCII */ 3515 return 0; 3516 3517 /* Skip backwards over trailing bytes: 10xx.xxxx 3518 * Skip backwards again if on a composing char. */ 3519 for (q = p; ; --q) 3520 { 3521 /* Move s to the last byte of this char. */ 3522 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3523 ; 3524 /* Move q to the first byte of this char. */ 3525 while (q > base && (*q & 0xc0) == 0x80) 3526 --q; 3527 /* Check for illegal sequence. Do allow an illegal byte after where we 3528 * started. */ 3529 len = utf8len_tab[*q]; 3530 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3531 return 0; 3532 3533 if (q <= base) 3534 break; 3535 3536 c = utf_ptr2char(q); 3537 if (utf_iscomposing(c)) 3538 continue; 3539 3540 #ifdef FEAT_ARABIC 3541 if (arabic_maycombine(c)) 3542 { 3543 /* Advance to get a sneak-peak at the next char */ 3544 j = q; 3545 --j; 3546 /* Move j to the first byte of this char. */ 3547 while (j > base && (*j & 0xc0) == 0x80) 3548 --j; 3549 if (arabic_combine(utf_ptr2char(j), c)) 3550 continue; 3551 } 3552 #endif 3553 break; 3554 } 3555 3556 return (int)(p - q); 3557 } 3558 3559 /* 3560 * Copy a character from "*fp" to "*tp" and advance the pointers. 3561 */ 3562 void 3563 mb_copy_char(fp, tp) 3564 char_u **fp; 3565 char_u **tp; 3566 { 3567 int l = (*mb_ptr2len)(*fp); 3568 3569 mch_memmove(*tp, *fp, (size_t)l); 3570 *tp += l; 3571 *fp += l; 3572 } 3573 3574 /* 3575 * Return the offset from "p" to the first byte of a character. When "p" is 3576 * at the start of a character 0 is returned, otherwise the offset to the next 3577 * character. Can start anywhere in a stream of bytes. 3578 */ 3579 int 3580 mb_off_next(base, p) 3581 char_u *base; 3582 char_u *p; 3583 { 3584 int i; 3585 int j; 3586 3587 if (enc_utf8) 3588 { 3589 if (*p < 0x80) /* be quick for ASCII */ 3590 return 0; 3591 3592 /* Find the next character that isn't 10xx.xxxx */ 3593 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 3594 ; 3595 if (i > 0) 3596 { 3597 /* Check for illegal sequence. */ 3598 for (j = 0; p - j > base; ++j) 3599 if ((p[-j] & 0xc0) != 0x80) 3600 break; 3601 if (utf8len_tab[p[-j]] != i + j) 3602 return 0; 3603 } 3604 return i; 3605 } 3606 3607 /* Only need to check if we're on a trail byte, it doesn't matter if we 3608 * want the offset to the next or current character. */ 3609 return (*mb_head_off)(base, p); 3610 } 3611 3612 /* 3613 * Return the offset from "p" to the last byte of the character it points 3614 * into. Can start anywhere in a stream of bytes. 3615 */ 3616 int 3617 mb_tail_off(base, p) 3618 char_u *base; 3619 char_u *p; 3620 { 3621 int i; 3622 int j; 3623 3624 if (*p == NUL) 3625 return 0; 3626 3627 if (enc_utf8) 3628 { 3629 /* Find the last character that is 10xx.xxxx */ 3630 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 3631 ; 3632 /* Check for illegal sequence. */ 3633 for (j = 0; p - j > base; ++j) 3634 if ((p[-j] & 0xc0) != 0x80) 3635 break; 3636 if (utf8len_tab[p[-j]] != i + j + 1) 3637 return 0; 3638 return i; 3639 } 3640 3641 /* It can't be the first byte if a double-byte when not using DBCS, at the 3642 * end of the string or the byte can't start a double-byte. */ 3643 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 3644 return 0; 3645 3646 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3647 return 1 - dbcs_head_off(base, p); 3648 } 3649 3650 /* 3651 * Find the next illegal byte sequence. 3652 */ 3653 void 3654 utf_find_illegal() 3655 { 3656 pos_T pos = curwin->w_cursor; 3657 char_u *p; 3658 int len; 3659 vimconv_T vimconv; 3660 char_u *tofree = NULL; 3661 3662 vimconv.vc_type = CONV_NONE; 3663 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 3664 { 3665 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 3666 * possibly a utf-8 file with illegal bytes. Setup for conversion 3667 * from utf-8 to 'fileencoding'. */ 3668 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 3669 } 3670 3671 #ifdef FEAT_VIRTUALEDIT 3672 curwin->w_cursor.coladd = 0; 3673 #endif 3674 for (;;) 3675 { 3676 p = ml_get_cursor(); 3677 if (vimconv.vc_type != CONV_NONE) 3678 { 3679 vim_free(tofree); 3680 tofree = string_convert(&vimconv, p, NULL); 3681 if (tofree == NULL) 3682 break; 3683 p = tofree; 3684 } 3685 3686 while (*p != NUL) 3687 { 3688 /* Illegal means that there are not enough trail bytes (checked by 3689 * utf_ptr2len()) or too many of them (overlong sequence). */ 3690 len = utf_ptr2len(p); 3691 if (*p >= 0x80 && (len == 1 3692 || utf_char2len(utf_ptr2char(p)) != len)) 3693 { 3694 if (vimconv.vc_type == CONV_NONE) 3695 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 3696 else 3697 { 3698 int l; 3699 3700 len = (int)(p - tofree); 3701 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 3702 { 3703 l = utf_ptr2len(p); 3704 curwin->w_cursor.col += l; 3705 } 3706 } 3707 goto theend; 3708 } 3709 p += len; 3710 } 3711 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 3712 break; 3713 ++curwin->w_cursor.lnum; 3714 curwin->w_cursor.col = 0; 3715 } 3716 3717 /* didn't find it: don't move and beep */ 3718 curwin->w_cursor = pos; 3719 beep_flush(); 3720 3721 theend: 3722 vim_free(tofree); 3723 convert_setup(&vimconv, NULL, NULL); 3724 } 3725 3726 #if defined(FEAT_GUI_GTK) || defined(PROTO) 3727 /* 3728 * Return TRUE if string "s" is a valid utf-8 string. 3729 * When "end" is NULL stop at the first NUL. 3730 * When "end" is positive stop there. 3731 */ 3732 int 3733 utf_valid_string(s, end) 3734 char_u *s; 3735 char_u *end; 3736 { 3737 int l; 3738 char_u *p = s; 3739 3740 while (end == NULL ? *p != NUL : p < end) 3741 { 3742 l = utf8len_tab_zero[*p]; 3743 if (l == 0) 3744 return FALSE; /* invalid lead byte */ 3745 if (end != NULL && p + l > end) 3746 return FALSE; /* incomplete byte sequence */ 3747 ++p; 3748 while (--l > 0) 3749 if ((*p++ & 0xc0) != 0x80) 3750 return FALSE; /* invalid trail byte */ 3751 } 3752 return TRUE; 3753 } 3754 #endif 3755 3756 #if defined(FEAT_GUI) || defined(PROTO) 3757 /* 3758 * Special version of mb_tail_off() for use in ScreenLines[]. 3759 */ 3760 int 3761 dbcs_screen_tail_off(base, p) 3762 char_u *base; 3763 char_u *p; 3764 { 3765 /* It can't be the first byte if a double-byte when not using DBCS, at the 3766 * end of the string or the byte can't start a double-byte. 3767 * For euc-jp an 0x8e byte always means we have a lead byte in the current 3768 * cell. */ 3769 if (*p == NUL || p[1] == NUL 3770 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 3771 || MB_BYTE2LEN(*p) == 1) 3772 return 0; 3773 3774 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3775 return 1 - dbcs_screen_head_off(base, p); 3776 } 3777 #endif 3778 3779 /* 3780 * If the cursor moves on an trail byte, set the cursor on the lead byte. 3781 * Thus it moves left if necessary. 3782 * Return TRUE when the cursor was adjusted. 3783 */ 3784 void 3785 mb_adjust_cursor() 3786 { 3787 mb_adjustpos(curbuf, &curwin->w_cursor); 3788 } 3789 3790 /* 3791 * Adjust position "*lp" to point to the first byte of a multi-byte character. 3792 * If it points to a tail byte it's moved backwards to the head byte. 3793 */ 3794 void 3795 mb_adjustpos(buf, lp) 3796 buf_T *buf; 3797 pos_T *lp; 3798 { 3799 char_u *p; 3800 3801 if (lp->col > 0 3802 #ifdef FEAT_VIRTUALEDIT 3803 || lp->coladd > 1 3804 #endif 3805 ) 3806 { 3807 p = ml_get_buf(buf, lp->lnum, FALSE); 3808 lp->col -= (*mb_head_off)(p, p + lp->col); 3809 #ifdef FEAT_VIRTUALEDIT 3810 /* Reset "coladd" when the cursor would be on the right half of a 3811 * double-wide character. */ 3812 if (lp->coladd == 1 3813 && p[lp->col] != TAB 3814 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 3815 && ptr2cells(p + lp->col) > 1) 3816 lp->coladd = 0; 3817 #endif 3818 } 3819 } 3820 3821 /* 3822 * Return a pointer to the character before "*p", if there is one. 3823 */ 3824 char_u * 3825 mb_prevptr(line, p) 3826 char_u *line; /* start of the string */ 3827 char_u *p; 3828 { 3829 if (p > line) 3830 mb_ptr_back(line, p); 3831 return p; 3832 } 3833 3834 /* 3835 * Return the character length of "str". Each multi-byte character (with 3836 * following composing characters) counts as one. 3837 */ 3838 int 3839 mb_charlen(str) 3840 char_u *str; 3841 { 3842 char_u *p = str; 3843 int count; 3844 3845 if (p == NULL) 3846 return 0; 3847 3848 for (count = 0; *p != NUL; count++) 3849 p += (*mb_ptr2len)(p); 3850 3851 return count; 3852 } 3853 3854 #if defined(FEAT_SPELL) || defined(PROTO) 3855 /* 3856 * Like mb_charlen() but for a string with specified length. 3857 */ 3858 int 3859 mb_charlen_len(str, len) 3860 char_u *str; 3861 int len; 3862 { 3863 char_u *p = str; 3864 int count; 3865 3866 for (count = 0; *p != NUL && p < str + len; count++) 3867 p += (*mb_ptr2len)(p); 3868 3869 return count; 3870 } 3871 #endif 3872 3873 /* 3874 * Try to un-escape a multi-byte character. 3875 * Used for the "to" and "from" part of a mapping. 3876 * Return the un-escaped string if it is a multi-byte character, and advance 3877 * "pp" to just after the bytes that formed it. 3878 * Return NULL if no multi-byte char was found. 3879 */ 3880 char_u * 3881 mb_unescape(pp) 3882 char_u **pp; 3883 { 3884 static char_u buf[6]; 3885 int n; 3886 int m = 0; 3887 char_u *str = *pp; 3888 3889 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 3890 * KS_EXTRA KE_CSI to CSI. 3891 * Maximum length of a utf-8 character is 4 bytes. */ 3892 for (n = 0; str[n] != NUL && m < 4; ++n) 3893 { 3894 if (str[n] == K_SPECIAL 3895 && str[n + 1] == KS_SPECIAL 3896 && str[n + 2] == KE_FILLER) 3897 { 3898 buf[m++] = K_SPECIAL; 3899 n += 2; 3900 } 3901 else if ((str[n] == K_SPECIAL 3902 # ifdef FEAT_GUI 3903 || str[n] == CSI 3904 # endif 3905 ) 3906 && str[n + 1] == KS_EXTRA 3907 && str[n + 2] == (int)KE_CSI) 3908 { 3909 buf[m++] = CSI; 3910 n += 2; 3911 } 3912 else if (str[n] == K_SPECIAL 3913 # ifdef FEAT_GUI 3914 || str[n] == CSI 3915 # endif 3916 ) 3917 break; /* a special key can't be a multibyte char */ 3918 else 3919 buf[m++] = str[n]; 3920 buf[m] = NUL; 3921 3922 /* Return a multi-byte character if it's found. An illegal sequence 3923 * will result in a 1 here. */ 3924 if ((*mb_ptr2len)(buf) > 1) 3925 { 3926 *pp = str + n + 1; 3927 return buf; 3928 } 3929 3930 /* Bail out quickly for ASCII. */ 3931 if (buf[0] < 128) 3932 break; 3933 } 3934 return NULL; 3935 } 3936 3937 /* 3938 * Return TRUE if the character at "row"/"col" on the screen is the left side 3939 * of a double-width character. 3940 * Caller must make sure "row" and "col" are not invalid! 3941 */ 3942 int 3943 mb_lefthalve(row, col) 3944 int row; 3945 int col; 3946 { 3947 #ifdef FEAT_HANGULIN 3948 if (composing_hangul) 3949 return TRUE; 3950 #endif 3951 return (*mb_off2cells)(LineOffset[row] + col, 3952 LineOffset[row] + screen_Columns) > 1; 3953 } 3954 3955 /* 3956 * Correct a position on the screen, if it's the right half of a double-wide 3957 * char move it to the left half. Returns the corrected column. 3958 */ 3959 int 3960 mb_fix_col(col, row) 3961 int col; 3962 int row; 3963 { 3964 col = check_col(col); 3965 row = check_row(row); 3966 if (has_mbyte && ScreenLines != NULL && col > 0 3967 && ((enc_dbcs 3968 && ScreenLines[LineOffset[row] + col] != NUL 3969 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 3970 ScreenLines + LineOffset[row] + col)) 3971 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0))) 3972 return col - 1; 3973 return col; 3974 } 3975 #endif 3976 3977 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 3978 static int enc_alias_search __ARGS((char_u *name)); 3979 3980 /* 3981 * Skip the Vim specific head of a 'encoding' name. 3982 */ 3983 char_u * 3984 enc_skip(p) 3985 char_u *p; 3986 { 3987 if (STRNCMP(p, "2byte-", 6) == 0) 3988 return p + 6; 3989 if (STRNCMP(p, "8bit-", 5) == 0) 3990 return p + 5; 3991 return p; 3992 } 3993 3994 /* 3995 * Find the canonical name for encoding "enc". 3996 * When the name isn't recognized, returns "enc" itself, but with all lower 3997 * case characters and '_' replaced with '-'. 3998 * Returns an allocated string. NULL for out-of-memory. 3999 */ 4000 char_u * 4001 enc_canonize(enc) 4002 char_u *enc; 4003 { 4004 char_u *r; 4005 char_u *p, *s; 4006 int i; 4007 4008 # ifdef FEAT_MBYTE 4009 if (STRCMP(enc, "default") == 0) 4010 { 4011 /* Use the default encoding as it's found by set_init_1(). */ 4012 r = get_encoding_default(); 4013 if (r == NULL) 4014 r = (char_u *)"latin1"; 4015 return vim_strsave(r); 4016 } 4017 # endif 4018 4019 /* copy "enc" to allocated memory, with room for two '-' */ 4020 r = alloc((unsigned)(STRLEN(enc) + 3)); 4021 if (r != NULL) 4022 { 4023 /* Make it all lower case and replace '_' with '-'. */ 4024 p = r; 4025 for (s = enc; *s != NUL; ++s) 4026 { 4027 if (*s == '_') 4028 *p++ = '-'; 4029 else 4030 *p++ = TOLOWER_ASC(*s); 4031 } 4032 *p = NUL; 4033 4034 /* Skip "2byte-" and "8bit-". */ 4035 p = enc_skip(r); 4036 4037 /* Change "microsoft-cp" to "cp". Used in some spell files. */ 4038 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4039 STRMOVE(p, p + 10); 4040 4041 /* "iso8859" -> "iso-8859" */ 4042 if (STRNCMP(p, "iso8859", 7) == 0) 4043 { 4044 STRMOVE(p + 4, p + 3); 4045 p[3] = '-'; 4046 } 4047 4048 /* "iso-8859n" -> "iso-8859-n" */ 4049 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4050 { 4051 STRMOVE(p + 9, p + 8); 4052 p[8] = '-'; 4053 } 4054 4055 /* "latin-N" -> "latinN" */ 4056 if (STRNCMP(p, "latin-", 6) == 0) 4057 STRMOVE(p + 5, p + 6); 4058 4059 if (enc_canon_search(p) >= 0) 4060 { 4061 /* canonical name can be used unmodified */ 4062 if (p != r) 4063 STRMOVE(r, p); 4064 } 4065 else if ((i = enc_alias_search(p)) >= 0) 4066 { 4067 /* alias recognized, get canonical name */ 4068 vim_free(r); 4069 r = vim_strsave((char_u *)enc_canon_table[i].name); 4070 } 4071 } 4072 return r; 4073 } 4074 4075 /* 4076 * Search for an encoding alias of "name". 4077 * Returns -1 when not found. 4078 */ 4079 static int 4080 enc_alias_search(name) 4081 char_u *name; 4082 { 4083 int i; 4084 4085 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4086 if (STRCMP(name, enc_alias_table[i].name) == 0) 4087 return enc_alias_table[i].canon; 4088 return -1; 4089 } 4090 #endif 4091 4092 #if defined(FEAT_MBYTE) || defined(PROTO) 4093 4094 #ifdef HAVE_LANGINFO_H 4095 # include <langinfo.h> 4096 #endif 4097 4098 /* 4099 * Get the canonicalized encoding of the current locale. 4100 * Returns an allocated string when successful, NULL when not. 4101 */ 4102 char_u * 4103 enc_locale() 4104 { 4105 #ifndef WIN3264 4106 char *s; 4107 char *p; 4108 int i; 4109 #endif 4110 char buf[50]; 4111 #ifdef WIN3264 4112 long acp = GetACP(); 4113 4114 if (acp == 1200) 4115 STRCPY(buf, "ucs-2le"); 4116 else if (acp == 1252) /* cp1252 is used as latin1 */ 4117 STRCPY(buf, "latin1"); 4118 else 4119 sprintf(buf, "cp%ld", acp); 4120 #else 4121 # ifdef HAVE_NL_LANGINFO_CODESET 4122 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4123 # endif 4124 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4125 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4126 # endif 4127 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4128 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4129 s = getenv("LANG"); 4130 4131 if (s == NULL || *s == NUL) 4132 return FAIL; 4133 4134 /* The most generic locale format is: 4135 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4136 * If there is a '.' remove the part before it. 4137 * if there is something after the codeset, remove it. 4138 * Make the name lowercase and replace '_' with '-'. 4139 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4140 * "ko_KR.EUC" == "euc-kr" 4141 */ 4142 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4143 { 4144 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4145 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4146 { 4147 /* copy "XY.EUC" to "euc-XY" to buf[10] */ 4148 STRCPY(buf + 10, "euc-"); 4149 buf[14] = p[-2]; 4150 buf[15] = p[-1]; 4151 buf[16] = 0; 4152 s = buf + 10; 4153 } 4154 else 4155 s = p + 1; 4156 } 4157 for (i = 0; s[i] != NUL && i < (int)sizeof(buf) - 1; ++i) 4158 { 4159 if (s[i] == '_' || s[i] == '-') 4160 buf[i] = '-'; 4161 else if (isalnum((int)s[i])) 4162 buf[i] = TOLOWER_ASC(s[i]); 4163 else 4164 break; 4165 } 4166 buf[i] = NUL; 4167 #endif 4168 4169 return enc_canonize((char_u *)buf); 4170 } 4171 4172 #if defined(WIN3264) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4173 /* 4174 * Convert an encoding name to an MS-Windows codepage. 4175 * Returns zero if no codepage can be figured out. 4176 */ 4177 int 4178 encname2codepage(name) 4179 char_u *name; 4180 { 4181 int cp; 4182 char_u *p = name; 4183 int idx; 4184 4185 if (STRNCMP(p, "8bit-", 5) == 0) 4186 p += 5; 4187 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4188 p += 6; 4189 4190 if (p[0] == 'c' && p[1] == 'p') 4191 cp = atoi((char *)p + 2); 4192 else if ((idx = enc_canon_search(p)) >= 0) 4193 cp = enc_canon_table[idx].codepage; 4194 else 4195 return 0; 4196 if (IsValidCodePage(cp)) 4197 return cp; 4198 return 0; 4199 } 4200 #endif 4201 4202 # if defined(USE_ICONV) || defined(PROTO) 4203 4204 static char_u *iconv_string __ARGS((vimconv_T *vcp, char_u *str, int slen, int *unconvlenp, int *resultlenp)); 4205 4206 /* 4207 * Call iconv_open() with a check if iconv() works properly (there are broken 4208 * versions). 4209 * Returns (void *)-1 if failed. 4210 * (should return iconv_t, but that causes problems with prototypes). 4211 */ 4212 void * 4213 my_iconv_open(to, from) 4214 char_u *to; 4215 char_u *from; 4216 { 4217 iconv_t fd; 4218 #define ICONV_TESTLEN 400 4219 char_u tobuf[ICONV_TESTLEN]; 4220 char *p; 4221 size_t tolen; 4222 static int iconv_ok = -1; 4223 4224 if (iconv_ok == FALSE) 4225 return (void *)-1; /* detected a broken iconv() previously */ 4226 4227 #ifdef DYNAMIC_ICONV 4228 /* Check if the iconv.dll can be found. */ 4229 if (!iconv_enabled(TRUE)) 4230 return (void *)-1; 4231 #endif 4232 4233 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4234 4235 if (fd != (iconv_t)-1 && iconv_ok == -1) 4236 { 4237 /* 4238 * Do a dummy iconv() call to check if it actually works. There is a 4239 * version of iconv() on Linux that is broken. We can't ignore it, 4240 * because it's wide-spread. The symptoms are that after outputting 4241 * the initial shift state the "to" pointer is NULL and conversion 4242 * stops for no apparent reason after about 8160 characters. 4243 */ 4244 p = (char *)tobuf; 4245 tolen = ICONV_TESTLEN; 4246 (void)iconv(fd, NULL, NULL, &p, &tolen); 4247 if (p == NULL) 4248 { 4249 iconv_ok = FALSE; 4250 iconv_close(fd); 4251 fd = (iconv_t)-1; 4252 } 4253 else 4254 iconv_ok = TRUE; 4255 } 4256 4257 return (void *)fd; 4258 } 4259 4260 /* 4261 * Convert the string "str[slen]" with iconv(). 4262 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4263 * sequence and set "*unconvlenp" to the length of it. 4264 * Returns the converted string in allocated memory. NULL for an error. 4265 * If resultlenp is not NULL, sets it to the result length in bytes. 4266 */ 4267 static char_u * 4268 iconv_string(vcp, str, slen, unconvlenp, resultlenp) 4269 vimconv_T *vcp; 4270 char_u *str; 4271 int slen; 4272 int *unconvlenp; 4273 int *resultlenp; 4274 { 4275 const char *from; 4276 size_t fromlen; 4277 char *to; 4278 size_t tolen; 4279 size_t len = 0; 4280 size_t done = 0; 4281 char_u *result = NULL; 4282 char_u *p; 4283 int l; 4284 4285 from = (char *)str; 4286 fromlen = slen; 4287 for (;;) 4288 { 4289 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4290 { 4291 /* Allocate enough room for most conversions. When re-allocating 4292 * increase the buffer size. */ 4293 len = len + fromlen * 2 + 40; 4294 p = alloc((unsigned)len); 4295 if (p != NULL && done > 0) 4296 mch_memmove(p, result, done); 4297 vim_free(result); 4298 result = p; 4299 if (result == NULL) /* out of memory */ 4300 break; 4301 } 4302 4303 to = (char *)result + done; 4304 tolen = len - done - 2; 4305 /* Avoid a warning for systems with a wrong iconv() prototype by 4306 * casting the second argument to void *. */ 4307 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4308 != (size_t)-1) 4309 { 4310 /* Finished, append a NUL. */ 4311 *to = NUL; 4312 break; 4313 } 4314 4315 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4316 * iconv library may use one of them. */ 4317 if (!vcp->vc_fail && unconvlenp != NULL 4318 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4319 { 4320 /* Handle an incomplete sequence at the end. */ 4321 *to = NUL; 4322 *unconvlenp = (int)fromlen; 4323 break; 4324 } 4325 4326 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4327 * iconv library may use one of them. */ 4328 else if (!vcp->vc_fail 4329 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4330 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4331 { 4332 /* Can't convert: insert a '?' and skip a character. This assumes 4333 * conversion from 'encoding' to something else. In other 4334 * situations we don't know what to skip anyway. */ 4335 *to++ = '?'; 4336 if ((*mb_ptr2cells)((char_u *)from) > 1) 4337 *to++ = '?'; 4338 if (enc_utf8) 4339 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4340 else 4341 { 4342 l = (*mb_ptr2len)((char_u *)from); 4343 if (l > (int)fromlen) 4344 l = (int)fromlen; 4345 } 4346 from += l; 4347 fromlen -= l; 4348 } 4349 else if (ICONV_ERRNO != ICONV_E2BIG) 4350 { 4351 /* conversion failed */ 4352 vim_free(result); 4353 result = NULL; 4354 break; 4355 } 4356 /* Not enough room or skipping illegal sequence. */ 4357 done = to - (char *)result; 4358 } 4359 4360 if (resultlenp != NULL && result != NULL) 4361 *resultlenp = (int)(to - (char *)result); 4362 return result; 4363 } 4364 4365 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4366 /* 4367 * Dynamically load the "iconv.dll" on Win32. 4368 */ 4369 4370 #ifndef DYNAMIC_ICONV /* just generating prototypes */ 4371 # define HINSTANCE int 4372 #endif 4373 static HINSTANCE hIconvDLL = 0; 4374 static HINSTANCE hMsvcrtDLL = 0; 4375 4376 # ifndef DYNAMIC_ICONV_DLL 4377 # define DYNAMIC_ICONV_DLL "iconv.dll" 4378 # define DYNAMIC_ICONV_DLL_ALT "libiconv.dll" 4379 # endif 4380 # ifndef DYNAMIC_MSVCRT_DLL 4381 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4382 # endif 4383 4384 /* 4385 * Get the address of 'funcname' which is imported by 'hInst' DLL. 4386 */ 4387 static void * 4388 get_iconv_import_func(HINSTANCE hInst, const char *funcname) 4389 { 4390 PBYTE pImage = (PBYTE)hInst; 4391 PIMAGE_DOS_HEADER pDOS = (PIMAGE_DOS_HEADER)hInst; 4392 PIMAGE_NT_HEADERS pPE; 4393 PIMAGE_IMPORT_DESCRIPTOR pImpDesc; 4394 PIMAGE_THUNK_DATA pIAT; /* Import Address Table */ 4395 PIMAGE_THUNK_DATA pINT; /* Import Name Table */ 4396 PIMAGE_IMPORT_BY_NAME pImpName; 4397 4398 if (pDOS->e_magic != IMAGE_DOS_SIGNATURE) 4399 return NULL; 4400 pPE = (PIMAGE_NT_HEADERS)(pImage + pDOS->e_lfanew); 4401 if (pPE->Signature != IMAGE_NT_SIGNATURE) 4402 return NULL; 4403 pImpDesc = (PIMAGE_IMPORT_DESCRIPTOR)(pImage 4404 + pPE->OptionalHeader.DataDirectory[IMAGE_DIRECTORY_ENTRY_IMPORT] 4405 .VirtualAddress); 4406 for (; pImpDesc->FirstThunk; ++pImpDesc) 4407 { 4408 if (!pImpDesc->OriginalFirstThunk) 4409 continue; 4410 pIAT = (PIMAGE_THUNK_DATA)(pImage + pImpDesc->FirstThunk); 4411 pINT = (PIMAGE_THUNK_DATA)(pImage + pImpDesc->OriginalFirstThunk); 4412 for (; pIAT->u1.Function; ++pIAT, ++pINT) 4413 { 4414 if (IMAGE_SNAP_BY_ORDINAL(pINT->u1.Ordinal)) 4415 continue; 4416 pImpName = (PIMAGE_IMPORT_BY_NAME)(pImage 4417 + (UINT_PTR)(pINT->u1.AddressOfData)); 4418 if (strcmp(pImpName->Name, funcname) == 0) 4419 return (void *)pIAT->u1.Function; 4420 } 4421 } 4422 return NULL; 4423 } 4424 4425 /* 4426 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4427 */ 4428 int 4429 iconv_enabled(verbose) 4430 int verbose; 4431 { 4432 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4433 return TRUE; 4434 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4435 if (hIconvDLL == 0) /* sometimes it's called libiconv.dll */ 4436 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT); 4437 if (hIconvDLL != 0) 4438 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4439 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4440 { 4441 /* Only give the message when 'verbose' is set, otherwise it might be 4442 * done whenever a conversion is attempted. */ 4443 if (verbose && p_verbose > 0) 4444 { 4445 verbose_enter(); 4446 EMSG2(_(e_loadlib), 4447 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4448 verbose_leave(); 4449 } 4450 iconv_end(); 4451 return FALSE; 4452 } 4453 4454 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4455 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4456 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4457 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4458 iconv_errno = get_iconv_import_func(hIconvDLL, "_errno"); 4459 if (iconv_errno == NULL) 4460 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4461 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4462 || iconvctl == NULL || iconv_errno == NULL) 4463 { 4464 iconv_end(); 4465 if (verbose && p_verbose > 0) 4466 { 4467 verbose_enter(); 4468 EMSG2(_(e_loadfunc), "for libiconv"); 4469 verbose_leave(); 4470 } 4471 return FALSE; 4472 } 4473 return TRUE; 4474 } 4475 4476 void 4477 iconv_end() 4478 { 4479 /* Don't use iconv() when inputting or outputting characters. */ 4480 if (input_conv.vc_type == CONV_ICONV) 4481 convert_setup(&input_conv, NULL, NULL); 4482 if (output_conv.vc_type == CONV_ICONV) 4483 convert_setup(&output_conv, NULL, NULL); 4484 4485 if (hIconvDLL != 0) 4486 FreeLibrary(hIconvDLL); 4487 if (hMsvcrtDLL != 0) 4488 FreeLibrary(hMsvcrtDLL); 4489 hIconvDLL = 0; 4490 hMsvcrtDLL = 0; 4491 } 4492 # endif /* DYNAMIC_ICONV */ 4493 # endif /* USE_ICONV */ 4494 4495 #endif /* FEAT_MBYTE */ 4496 4497 #if defined(FEAT_XIM) || defined(PROTO) 4498 4499 # if defined(FEAT_GUI_GTK) || defined(PROTO) 4500 static int xim_has_preediting INIT(= FALSE); /* IM current status */ 4501 4502 /* 4503 * Set preedit_start_col to the current cursor position. 4504 */ 4505 static void 4506 init_preedit_start_col(void) 4507 { 4508 if (State & CMDLINE) 4509 preedit_start_col = cmdline_getvcol_cursor(); 4510 else if (curwin != NULL) 4511 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); 4512 /* Prevent that preediting marks the buffer as changed. */ 4513 xim_changed_while_preediting = curbuf->b_changed; 4514 } 4515 4516 static int im_is_active = FALSE; /* IM is enabled for current mode */ 4517 static int preedit_is_active = FALSE; 4518 static int im_preedit_cursor = 0; /* cursor offset in characters */ 4519 static int im_preedit_trailing = 0; /* number of characters after cursor */ 4520 4521 static unsigned long im_commit_handler_id = 0; 4522 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; 4523 static unsigned int im_activatekey_state = 0; 4524 4525 void 4526 im_set_active(int active) 4527 { 4528 int was_active; 4529 4530 was_active = !!im_get_status(); 4531 im_is_active = (active && !p_imdisable); 4532 4533 if (im_is_active != was_active) 4534 xim_reset(); 4535 } 4536 4537 void 4538 xim_set_focus(int focus) 4539 { 4540 if (xic != NULL) 4541 { 4542 if (focus) 4543 gtk_im_context_focus_in(xic); 4544 else 4545 gtk_im_context_focus_out(xic); 4546 } 4547 } 4548 4549 void 4550 im_set_position(int row, int col) 4551 { 4552 if (xic != NULL) 4553 { 4554 GdkRectangle area; 4555 4556 area.x = FILL_X(col); 4557 area.y = FILL_Y(row); 4558 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); 4559 area.height = gui.char_height; 4560 4561 gtk_im_context_set_cursor_location(xic, &area); 4562 } 4563 } 4564 4565 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */ 4566 void 4567 xim_set_preedit(void) 4568 { 4569 im_set_position(gui.row, gui.col); 4570 } 4571 # endif 4572 4573 static void 4574 im_add_to_input(char_u *str, int len) 4575 { 4576 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */ 4577 if (input_conv.vc_type != CONV_NONE) 4578 { 4579 str = string_convert(&input_conv, str, &len); 4580 g_return_if_fail(str != NULL); 4581 } 4582 4583 add_to_input_buf_csi(str, len); 4584 4585 if (input_conv.vc_type != CONV_NONE) 4586 vim_free(str); 4587 4588 if (p_mh) /* blank out the pointer if necessary */ 4589 gui_mch_mousehide(TRUE); 4590 } 4591 4592 static void 4593 im_delete_preedit(void) 4594 { 4595 char_u bskey[] = {CSI, 'k', 'b'}; 4596 char_u delkey[] = {CSI, 'k', 'D'}; 4597 4598 if (State & NORMAL) 4599 { 4600 im_preedit_cursor = 0; 4601 return; 4602 } 4603 for (; im_preedit_cursor > 0; --im_preedit_cursor) 4604 add_to_input_buf(bskey, (int)sizeof(bskey)); 4605 4606 for (; im_preedit_trailing > 0; --im_preedit_trailing) 4607 add_to_input_buf(delkey, (int)sizeof(delkey)); 4608 } 4609 4610 /* 4611 * Move the cursor left by "num_move_back" characters. 4612 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for 4613 * these K_LEFT keys. 4614 */ 4615 static void 4616 im_correct_cursor(int num_move_back) 4617 { 4618 char_u backkey[] = {CSI, 'k', 'l'}; 4619 4620 if (State & NORMAL) 4621 return; 4622 # ifdef FEAT_RIGHTLEFT 4623 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) 4624 backkey[2] = 'r'; 4625 # endif 4626 for (; num_move_back > 0; --num_move_back) 4627 add_to_input_buf(backkey, (int)sizeof(backkey)); 4628 } 4629 4630 static int xim_expected_char = NUL; 4631 static int xim_ignored_char = FALSE; 4632 4633 /* 4634 * Update the mode and cursor while in an IM callback. 4635 */ 4636 static void 4637 im_show_info(void) 4638 { 4639 int old_vgetc_busy; 4640 4641 old_vgetc_busy = vgetc_busy; 4642 vgetc_busy = TRUE; 4643 showmode(); 4644 vgetc_busy = old_vgetc_busy; 4645 if ((State & NORMAL) || (State & INSERT)) 4646 setcursor(); 4647 out_flush(); 4648 } 4649 4650 /* 4651 * Callback invoked when the user finished preediting. 4652 * Put the final string into the input buffer. 4653 */ 4654 static void 4655 im_commit_cb(GtkIMContext *context UNUSED, 4656 const gchar *str, 4657 gpointer data UNUSED) 4658 { 4659 int slen = (int)STRLEN(str); 4660 int add_to_input = TRUE; 4661 int clen; 4662 int len = slen; 4663 int commit_with_preedit = TRUE; 4664 char_u *im_str; 4665 4666 #ifdef XIM_DEBUG 4667 xim_log("im_commit_cb(): %s\n", str); 4668 #endif 4669 4670 /* The imhangul module doesn't reset the preedit string before 4671 * committing. Call im_delete_preedit() to work around that. */ 4672 im_delete_preedit(); 4673 4674 /* Indicate that preediting has finished. */ 4675 if (preedit_start_col == MAXCOL) 4676 { 4677 init_preedit_start_col(); 4678 commit_with_preedit = FALSE; 4679 } 4680 4681 /* The thing which setting "preedit_start_col" to MAXCOL means that 4682 * "preedit_start_col" will be set forcedly when calling 4683 * preedit_changed_cb() next time. 4684 * "preedit_start_col" should not reset with MAXCOL on this part. Vim 4685 * is simulating the preediting by using add_to_input_str(). when 4686 * preedit begin immediately before committed, the typebuf is not 4687 * flushed to screen, then it can't get correct "preedit_start_col". 4688 * Thus, it should calculate the cells by adding cells of the committed 4689 * string. */ 4690 if (input_conv.vc_type != CONV_NONE) 4691 { 4692 im_str = string_convert(&input_conv, (char_u *)str, &len); 4693 g_return_if_fail(im_str != NULL); 4694 } 4695 else 4696 im_str = (char_u *)str; 4697 4698 clen = mb_string2cells(im_str, len); 4699 4700 if (input_conv.vc_type != CONV_NONE) 4701 vim_free(im_str); 4702 preedit_start_col += clen; 4703 4704 /* Is this a single character that matches a keypad key that's just 4705 * been pressed? If so, we don't want it to be entered as such - let 4706 * us carry on processing the raw keycode so that it may be used in 4707 * mappings as <kSomething>. */ 4708 if (xim_expected_char != NUL) 4709 { 4710 /* We're currently processing a keypad or other special key */ 4711 if (slen == 1 && str[0] == xim_expected_char) 4712 { 4713 /* It's a match - don't do it here */ 4714 xim_ignored_char = TRUE; 4715 add_to_input = FALSE; 4716 } 4717 else 4718 { 4719 /* Not a match */ 4720 xim_ignored_char = FALSE; 4721 } 4722 } 4723 4724 if (add_to_input) 4725 im_add_to_input((char_u *)str, slen); 4726 4727 /* Inserting chars while "im_is_active" is set does not cause a change of 4728 * buffer. When the chars are committed the buffer must be marked as 4729 * changed. */ 4730 if (!commit_with_preedit) 4731 preedit_start_col = MAXCOL; 4732 4733 /* This flag is used in changed() at next call. */ 4734 xim_changed_while_preediting = TRUE; 4735 4736 if (gtk_main_level() > 0) 4737 gtk_main_quit(); 4738 } 4739 4740 /* 4741 * Callback invoked after start to the preedit. 4742 */ 4743 static void 4744 im_preedit_start_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4745 { 4746 #ifdef XIM_DEBUG 4747 xim_log("im_preedit_start_cb()\n"); 4748 #endif 4749 4750 im_is_active = TRUE; 4751 preedit_is_active = TRUE; 4752 gui_update_cursor(TRUE, FALSE); 4753 im_show_info(); 4754 } 4755 4756 /* 4757 * Callback invoked after end to the preedit. 4758 */ 4759 static void 4760 im_preedit_end_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4761 { 4762 #ifdef XIM_DEBUG 4763 xim_log("im_preedit_end_cb()\n"); 4764 #endif 4765 im_delete_preedit(); 4766 4767 /* Indicate that preediting has finished */ 4768 preedit_start_col = MAXCOL; 4769 xim_has_preediting = FALSE; 4770 4771 #if 0 4772 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was 4773 * switched off unintentionally. We now use preedit_is_active (added by 4774 * SungHyun Nam). */ 4775 im_is_active = FALSE; 4776 #endif 4777 preedit_is_active = FALSE; 4778 gui_update_cursor(TRUE, FALSE); 4779 im_show_info(); 4780 } 4781 4782 /* 4783 * Callback invoked after changes to the preedit string. If the preedit 4784 * string was empty before, remember the preedit start column so we know 4785 * where to apply feedback attributes. Delete the previous preedit string 4786 * if there was one, save the new preedit cursor offset, and put the new 4787 * string into the input buffer. 4788 * 4789 * TODO: The pragmatic "put into input buffer" approach used here has 4790 * several fundamental problems: 4791 * 4792 * - The characters in the preedit string are subject to remapping. 4793 * That's broken, only the finally committed string should be remapped. 4794 * 4795 * - There is a race condition involved: The retrieved value for the 4796 * current cursor position will be wrong if any unprocessed characters 4797 * are still queued in the input buffer. 4798 * 4799 * - Due to the lack of synchronization between the file buffer in memory 4800 * and any typed characters, it's practically impossible to implement the 4801 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM 4802 * modules for languages such as Thai are likely to rely on this feature 4803 * for proper operation. 4804 * 4805 * Conclusions: I think support for preediting needs to be moved to the 4806 * core parts of Vim. Ideally, until it has been committed, the preediting 4807 * string should only be displayed and not affect the buffer content at all. 4808 * The question how to deal with the synchronization issue still remains. 4809 * Circumventing the input buffer is probably not desirable. Anyway, I think 4810 * implementing "retrieve_surrounding" is the only hard problem. 4811 * 4812 * One way to solve all of this in a clean manner would be to queue all key 4813 * press/release events "as is" in the input buffer, and apply the IM filtering 4814 * at the receiving end of the queue. This, however, would have a rather large 4815 * impact on the code base. If there is an easy way to force processing of all 4816 * remaining input from within the "retrieve_surrounding" signal handler, this 4817 * might not be necessary. Gotta ask on vim-dev for opinions. 4818 */ 4819 static void 4820 im_preedit_changed_cb(GtkIMContext *context, gpointer data UNUSED) 4821 { 4822 char *preedit_string = NULL; 4823 int cursor_index = 0; 4824 int num_move_back = 0; 4825 char_u *str; 4826 char_u *p; 4827 int i; 4828 4829 gtk_im_context_get_preedit_string(context, 4830 &preedit_string, NULL, 4831 &cursor_index); 4832 4833 #ifdef XIM_DEBUG 4834 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); 4835 #endif 4836 4837 g_return_if_fail(preedit_string != NULL); /* just in case */ 4838 4839 /* If preedit_start_col is MAXCOL set it to the current cursor position. */ 4840 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') 4841 { 4842 xim_has_preediting = TRUE; 4843 4844 /* Urgh, this breaks if the input buffer isn't empty now */ 4845 init_preedit_start_col(); 4846 } 4847 else if (cursor_index == 0 && preedit_string[0] == '\0') 4848 { 4849 xim_has_preediting = FALSE; 4850 4851 /* If at the start position (after typing backspace) 4852 * preedit_start_col must be reset. */ 4853 preedit_start_col = MAXCOL; 4854 } 4855 4856 im_delete_preedit(); 4857 4858 /* 4859 * Compute the end of the preediting area: "preedit_end_col". 4860 * According to the documentation of gtk_im_context_get_preedit_string(), 4861 * the cursor_pos output argument returns the offset in bytes. This is 4862 * unfortunately not true -- real life shows the offset is in characters, 4863 * and the GTK+ source code agrees with me. Will file a bug later. 4864 */ 4865 if (preedit_start_col != MAXCOL) 4866 preedit_end_col = preedit_start_col; 4867 str = (char_u *)preedit_string; 4868 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) 4869 { 4870 int is_composing; 4871 4872 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); 4873 /* 4874 * These offsets are used as counters when generating <BS> and <Del> 4875 * to delete the preedit string. So don't count composing characters 4876 * unless 'delcombine' is enabled. 4877 */ 4878 if (!is_composing || p_deco) 4879 { 4880 if (i < cursor_index) 4881 ++im_preedit_cursor; 4882 else 4883 ++im_preedit_trailing; 4884 } 4885 if (!is_composing && i >= cursor_index) 4886 { 4887 /* This is essentially the same as im_preedit_trailing, except 4888 * composing characters are not counted even if p_deco is set. */ 4889 ++num_move_back; 4890 } 4891 if (preedit_start_col != MAXCOL) 4892 preedit_end_col += utf_ptr2cells(p); 4893 } 4894 4895 if (p > str) 4896 { 4897 im_add_to_input(str, (int)(p - str)); 4898 im_correct_cursor(num_move_back); 4899 } 4900 4901 g_free(preedit_string); 4902 4903 if (gtk_main_level() > 0) 4904 gtk_main_quit(); 4905 } 4906 4907 /* 4908 * Translate the Pango attributes at iter to Vim highlighting attributes. 4909 * Ignore attributes not supported by Vim highlighting. This shouldn't have 4910 * too much impact -- right now we handle even more attributes than necessary 4911 * for the IM modules I tested with. 4912 */ 4913 static int 4914 translate_pango_attributes(PangoAttrIterator *iter) 4915 { 4916 PangoAttribute *attr; 4917 int char_attr = HL_NORMAL; 4918 4919 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); 4920 if (attr != NULL && ((PangoAttrInt *)attr)->value 4921 != (int)PANGO_UNDERLINE_NONE) 4922 char_attr |= HL_UNDERLINE; 4923 4924 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); 4925 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) 4926 char_attr |= HL_BOLD; 4927 4928 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); 4929 if (attr != NULL && ((PangoAttrInt *)attr)->value 4930 != (int)PANGO_STYLE_NORMAL) 4931 char_attr |= HL_ITALIC; 4932 4933 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); 4934 if (attr != NULL) 4935 { 4936 const PangoColor *color = &((PangoAttrColor *)attr)->color; 4937 4938 /* Assume inverse if black background is requested */ 4939 if ((color->red | color->green | color->blue) == 0) 4940 char_attr |= HL_INVERSE; 4941 } 4942 4943 return char_attr; 4944 } 4945 4946 /* 4947 * Retrieve the highlighting attributes at column col in the preedit string. 4948 * Return -1 if not in preediting mode or if col is out of range. 4949 */ 4950 int 4951 im_get_feedback_attr(int col) 4952 { 4953 char *preedit_string = NULL; 4954 PangoAttrList *attr_list = NULL; 4955 int char_attr = -1; 4956 4957 if (xic == NULL) 4958 return char_attr; 4959 4960 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 4961 4962 if (preedit_string != NULL && attr_list != NULL) 4963 { 4964 int idx; 4965 4966 /* Get the byte index as used by PangoAttrIterator */ 4967 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col) 4968 idx += utfc_ptr2len((char_u *)preedit_string + idx); 4969 4970 if (preedit_string[idx] != '\0') 4971 { 4972 PangoAttrIterator *iter; 4973 int start, end; 4974 4975 char_attr = HL_NORMAL; 4976 iter = pango_attr_list_get_iterator(attr_list); 4977 4978 /* Extract all relevant attributes from the list. */ 4979 do 4980 { 4981 pango_attr_iterator_range(iter, &start, &end); 4982 4983 if (idx >= start && idx < end) 4984 char_attr |= translate_pango_attributes(iter); 4985 } 4986 while (pango_attr_iterator_next(iter)); 4987 4988 pango_attr_iterator_destroy(iter); 4989 } 4990 } 4991 4992 if (attr_list != NULL) 4993 pango_attr_list_unref(attr_list); 4994 g_free(preedit_string); 4995 4996 return char_attr; 4997 } 4998 4999 void 5000 xim_init(void) 5001 { 5002 #ifdef XIM_DEBUG 5003 xim_log("xim_init()\n"); 5004 #endif 5005 5006 g_return_if_fail(gui.drawarea != NULL); 5007 g_return_if_fail(gui.drawarea->window != NULL); 5008 5009 xic = gtk_im_multicontext_new(); 5010 g_object_ref(xic); 5011 5012 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", 5013 G_CALLBACK(&im_commit_cb), NULL); 5014 g_signal_connect(G_OBJECT(xic), "preedit_changed", 5015 G_CALLBACK(&im_preedit_changed_cb), NULL); 5016 g_signal_connect(G_OBJECT(xic), "preedit_start", 5017 G_CALLBACK(&im_preedit_start_cb), NULL); 5018 g_signal_connect(G_OBJECT(xic), "preedit_end", 5019 G_CALLBACK(&im_preedit_end_cb), NULL); 5020 5021 gtk_im_context_set_client_window(xic, gui.drawarea->window); 5022 } 5023 5024 void 5025 im_shutdown(void) 5026 { 5027 #ifdef XIM_DEBUG 5028 xim_log("im_shutdown()\n"); 5029 #endif 5030 5031 if (xic != NULL) 5032 { 5033 gtk_im_context_focus_out(xic); 5034 g_object_unref(xic); 5035 xic = NULL; 5036 } 5037 im_is_active = FALSE; 5038 im_commit_handler_id = 0; 5039 preedit_start_col = MAXCOL; 5040 xim_has_preediting = FALSE; 5041 } 5042 5043 /* 5044 * Convert the string argument to keyval and state for GdkEventKey. 5045 * If str is valid return TRUE, otherwise FALSE. 5046 * 5047 * See 'imactivatekey' for documentation of the format. 5048 */ 5049 static int 5050 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) 5051 { 5052 const char *mods_end; 5053 unsigned tmp_keyval; 5054 unsigned tmp_state = 0; 5055 5056 mods_end = strrchr(str, '-'); 5057 mods_end = (mods_end != NULL) ? mods_end + 1 : str; 5058 5059 /* Parse modifier keys */ 5060 while (str < mods_end) 5061 switch (*str++) 5062 { 5063 case '-': break; 5064 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; 5065 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; 5066 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; 5067 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; 5068 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; 5069 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; 5070 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; 5071 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; 5072 default: 5073 return FALSE; 5074 } 5075 5076 tmp_keyval = gdk_keyval_from_name(str); 5077 5078 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) 5079 return FALSE; 5080 5081 if (keyval != NULL) 5082 *keyval = tmp_keyval; 5083 if (state != NULL) 5084 *state = tmp_state; 5085 5086 return TRUE; 5087 } 5088 5089 /* 5090 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an 5091 * empty string is also regarded as valid. 5092 * 5093 * Note: The numerical key value of p_imak is cached if it was valid; thus 5094 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever 5095 * 'imak' changes. This is currently the case but not obvious -- should 5096 * probably rename the function for clarity. 5097 */ 5098 int 5099 im_xim_isvalid_imactivate(void) 5100 { 5101 if (p_imak[0] == NUL) 5102 { 5103 im_activatekey_keyval = GDK_VoidSymbol; 5104 im_activatekey_state = 0; 5105 return TRUE; 5106 } 5107 5108 return im_string_to_keyval((const char *)p_imak, 5109 &im_activatekey_keyval, 5110 &im_activatekey_state); 5111 } 5112 5113 static void 5114 im_synthesize_keypress(unsigned int keyval, unsigned int state) 5115 { 5116 GdkEventKey *event; 5117 5118 # ifdef HAVE_GTK_MULTIHEAD 5119 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); 5120 g_object_ref(gui.drawarea->window); /* unreffed by gdk_event_free() */ 5121 # else 5122 event = (GdkEventKey *)g_malloc0((gulong)sizeof(GdkEvent)); 5123 event->type = GDK_KEY_PRESS; 5124 # endif 5125 event->window = gui.drawarea->window; 5126 event->send_event = TRUE; 5127 event->time = GDK_CURRENT_TIME; 5128 event->state = state; 5129 event->keyval = keyval; 5130 event->hardware_keycode = /* needed for XIM */ 5131 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); 5132 event->length = 0; 5133 event->string = NULL; 5134 5135 gtk_im_context_filter_keypress(xic, event); 5136 5137 /* For consistency, also send the corresponding release event. */ 5138 event->type = GDK_KEY_RELEASE; 5139 event->send_event = FALSE; 5140 gtk_im_context_filter_keypress(xic, event); 5141 5142 # ifdef HAVE_GTK_MULTIHEAD 5143 gdk_event_free((GdkEvent *)event); 5144 # else 5145 g_free(event); 5146 # endif 5147 } 5148 5149 void 5150 xim_reset(void) 5151 { 5152 if (xic != NULL) 5153 { 5154 gtk_im_context_reset(xic); 5155 5156 if (p_imdisable) 5157 im_shutdown(); 5158 else 5159 { 5160 xim_set_focus(gui.in_focus); 5161 5162 # ifdef FEAT_EVAL 5163 if (p_imaf[0] != NUL) 5164 { 5165 char_u *argv[1]; 5166 5167 if (im_is_active) 5168 argv[0] = (char_u *)"1"; 5169 else 5170 argv[0] = (char_u *)"0"; 5171 (void)call_func_retnr(p_imaf, 1, argv, FALSE); 5172 } 5173 else 5174 # endif 5175 if (im_activatekey_keyval != GDK_VoidSymbol) 5176 { 5177 if (im_is_active) 5178 { 5179 g_signal_handler_block(xic, im_commit_handler_id); 5180 im_synthesize_keypress(im_activatekey_keyval, 5181 im_activatekey_state); 5182 g_signal_handler_unblock(xic, im_commit_handler_id); 5183 } 5184 } 5185 else 5186 { 5187 im_shutdown(); 5188 xim_init(); 5189 xim_set_focus(gui.in_focus); 5190 } 5191 } 5192 } 5193 5194 preedit_start_col = MAXCOL; 5195 xim_has_preediting = FALSE; 5196 } 5197 5198 int 5199 xim_queue_key_press_event(GdkEventKey *event, int down) 5200 { 5201 if (down) 5202 { 5203 /* 5204 * Workaround GTK2 XIM 'feature' that always converts keypad keys to 5205 * chars., even when not part of an IM sequence (ref. feature of 5206 * gdk/gdkkeyuni.c). 5207 * Flag any keypad keys that might represent a single char. 5208 * If this (on its own - i.e., not part of an IM sequence) is 5209 * committed while we're processing one of these keys, we can ignore 5210 * that commit and go ahead & process it ourselves. That way we can 5211 * still distinguish keypad keys for use in mappings. 5212 * Also add GDK_space to make <S-Space> work. 5213 */ 5214 switch (event->keyval) 5215 { 5216 case GDK_KP_Add: xim_expected_char = '+'; break; 5217 case GDK_KP_Subtract: xim_expected_char = '-'; break; 5218 case GDK_KP_Divide: xim_expected_char = '/'; break; 5219 case GDK_KP_Multiply: xim_expected_char = '*'; break; 5220 case GDK_KP_Decimal: xim_expected_char = '.'; break; 5221 case GDK_KP_Equal: xim_expected_char = '='; break; 5222 case GDK_KP_0: xim_expected_char = '0'; break; 5223 case GDK_KP_1: xim_expected_char = '1'; break; 5224 case GDK_KP_2: xim_expected_char = '2'; break; 5225 case GDK_KP_3: xim_expected_char = '3'; break; 5226 case GDK_KP_4: xim_expected_char = '4'; break; 5227 case GDK_KP_5: xim_expected_char = '5'; break; 5228 case GDK_KP_6: xim_expected_char = '6'; break; 5229 case GDK_KP_7: xim_expected_char = '7'; break; 5230 case GDK_KP_8: xim_expected_char = '8'; break; 5231 case GDK_KP_9: xim_expected_char = '9'; break; 5232 case GDK_space: xim_expected_char = ' '; break; 5233 default: xim_expected_char = NUL; 5234 } 5235 xim_ignored_char = FALSE; 5236 } 5237 5238 /* 5239 * When typing fFtT, XIM may be activated. Thus it must pass 5240 * gtk_im_context_filter_keypress() in Normal mode. 5241 * And while doing :sh too. 5242 */ 5243 if (xic != NULL && !p_imdisable 5244 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) 5245 { 5246 /* 5247 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is 5248 * always aware of the current status of IM, and can even emulate 5249 * the activation key for modules that don't support one. 5250 */ 5251 if (event->keyval == im_activatekey_keyval 5252 && (event->state & im_activatekey_state) == im_activatekey_state) 5253 { 5254 unsigned int state_mask; 5255 5256 /* Require the state of the 3 most used modifiers to match exactly. 5257 * Otherwise e.g. <S-C-space> would be unusable for other purposes 5258 * if the IM activate key is <S-space>. */ 5259 state_mask = im_activatekey_state; 5260 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK 5261 | (int)GDK_MOD1_MASK); 5262 5263 if ((event->state & state_mask) != im_activatekey_state) 5264 return FALSE; 5265 5266 /* Don't send it a second time on GDK_KEY_RELEASE. */ 5267 if (event->type != GDK_KEY_PRESS) 5268 return TRUE; 5269 5270 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE)) 5271 { 5272 im_set_active(FALSE); 5273 5274 /* ":lmap" mappings exists, toggle use of mappings. */ 5275 State ^= LANGMAP; 5276 if (State & LANGMAP) 5277 { 5278 curbuf->b_p_iminsert = B_IMODE_NONE; 5279 State &= ~LANGMAP; 5280 } 5281 else 5282 { 5283 curbuf->b_p_iminsert = B_IMODE_LMAP; 5284 State |= LANGMAP; 5285 } 5286 return TRUE; 5287 } 5288 5289 return gtk_im_context_filter_keypress(xic, event); 5290 } 5291 5292 /* Don't filter events through the IM context if IM isn't active 5293 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module 5294 * not doing anything before the activation key was sent. */ 5295 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) 5296 { 5297 int imresult = gtk_im_context_filter_keypress(xic, event); 5298 5299 /* Some XIM send following sequence: 5300 * 1. preedited string. 5301 * 2. committed string. 5302 * 3. line changed key. 5303 * 4. preedited string. 5304 * 5. remove preedited string. 5305 * if 3, Vim can't move back the above line for 5. 5306 * thus, this part should not parse the key. */ 5307 if (!imresult && preedit_start_col != MAXCOL 5308 && event->keyval == GDK_Return) 5309 { 5310 im_synthesize_keypress(GDK_Return, 0U); 5311 return FALSE; 5312 } 5313 5314 /* If XIM tried to commit a keypad key as a single char., 5315 * ignore it so we can use the keypad key 'raw', for mappings. */ 5316 if (xim_expected_char != NUL && xim_ignored_char) 5317 /* We had a keypad key, and XIM tried to thieve it */ 5318 return FALSE; 5319 5320 /* This is supposed to fix a problem with iBus, that space 5321 * characters don't work in input mode. */ 5322 xim_expected_char = NUL; 5323 5324 /* Normal processing */ 5325 return imresult; 5326 } 5327 } 5328 5329 return FALSE; 5330 } 5331 5332 int 5333 im_get_status(void) 5334 { 5335 # ifdef FEAT_EVAL 5336 if (p_imsf[0] != NUL) 5337 { 5338 int is_active; 5339 5340 /* FIXME: Don't execute user function in unsafe situation. */ 5341 if (exiting 5342 # ifdef FEAT_AUTOCMD 5343 || is_autocmd_blocked() 5344 # endif 5345 ) 5346 return FALSE; 5347 /* FIXME: :py print 'xxx' is shown duplicate result. 5348 * Use silent to avoid it. */ 5349 ++msg_silent; 5350 is_active = call_func_retnr(p_imsf, 0, NULL, FALSE); 5351 --msg_silent; 5352 return (is_active > 0); 5353 } 5354 # endif 5355 return im_is_active; 5356 } 5357 5358 int 5359 preedit_get_status(void) 5360 { 5361 return preedit_is_active; 5362 } 5363 5364 int 5365 im_is_preediting() 5366 { 5367 return xim_has_preediting; 5368 } 5369 5370 # else /* !FEAT_GUI_GTK */ 5371 5372 static int xim_is_active = FALSE; /* XIM should be active in the current 5373 mode */ 5374 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */ 5375 #ifdef FEAT_GUI_X11 5376 static XIMStyle input_style; 5377 static int status_area_enabled = TRUE; 5378 #endif 5379 5380 /* 5381 * Switch using XIM on/off. This is used by the code that changes "State". 5382 */ 5383 void 5384 im_set_active(active) 5385 int active; 5386 { 5387 if (xic == NULL) 5388 return; 5389 5390 /* If 'imdisable' is set, XIM is never active. */ 5391 if (p_imdisable) 5392 active = FALSE; 5393 #if !defined(FEAT_GUI_GTK) 5394 else if (input_style & XIMPreeditPosition) 5395 /* There is a problem in switching XIM off when preediting is used, 5396 * and it is not clear how this can be solved. For now, keep XIM on 5397 * all the time, like it was done in Vim 5.8. */ 5398 active = TRUE; 5399 #endif 5400 5401 /* Remember the active state, it is needed when Vim gets keyboard focus. */ 5402 xim_is_active = active; 5403 xim_set_preedit(); 5404 } 5405 5406 /* 5407 * Adjust using XIM for gaining or losing keyboard focus. Also called when 5408 * "xim_is_active" changes. 5409 */ 5410 void 5411 xim_set_focus(focus) 5412 int focus; 5413 { 5414 if (xic == NULL) 5415 return; 5416 5417 /* 5418 * XIM only gets focus when the Vim window has keyboard focus and XIM has 5419 * been set active for the current mode. 5420 */ 5421 if (focus && xim_is_active) 5422 { 5423 if (!xim_has_focus) 5424 { 5425 xim_has_focus = TRUE; 5426 XSetICFocus(xic); 5427 } 5428 } 5429 else 5430 { 5431 if (xim_has_focus) 5432 { 5433 xim_has_focus = FALSE; 5434 XUnsetICFocus(xic); 5435 } 5436 } 5437 } 5438 5439 void 5440 im_set_position(row, col) 5441 int row UNUSED; 5442 int col UNUSED; 5443 { 5444 xim_set_preedit(); 5445 } 5446 5447 /* 5448 * Set the XIM to the current cursor position. 5449 */ 5450 void 5451 xim_set_preedit() 5452 { 5453 XVaNestedList attr_list; 5454 XRectangle spot_area; 5455 XPoint over_spot; 5456 int line_space; 5457 5458 if (xic == NULL) 5459 return; 5460 5461 xim_set_focus(TRUE); 5462 5463 if (!xim_has_focus) 5464 { 5465 /* hide XIM cursor */ 5466 over_spot.x = 0; 5467 over_spot.y = -100; /* arbitrary invisible position */ 5468 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5469 XNSpotLocation, 5470 &over_spot, 5471 NULL); 5472 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); 5473 XFree(attr_list); 5474 return; 5475 } 5476 5477 if (input_style & XIMPreeditPosition) 5478 { 5479 if (xim_fg_color == INVALCOLOR) 5480 { 5481 xim_fg_color = gui.def_norm_pixel; 5482 xim_bg_color = gui.def_back_pixel; 5483 } 5484 over_spot.x = TEXT_X(gui.col); 5485 over_spot.y = TEXT_Y(gui.row); 5486 spot_area.x = 0; 5487 spot_area.y = 0; 5488 spot_area.height = gui.char_height * Rows; 5489 spot_area.width = gui.char_width * Columns; 5490 line_space = gui.char_height; 5491 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5492 XNSpotLocation, &over_spot, 5493 XNForeground, (Pixel) xim_fg_color, 5494 XNBackground, (Pixel) xim_bg_color, 5495 XNArea, &spot_area, 5496 XNLineSpace, line_space, 5497 NULL); 5498 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) 5499 EMSG(_("E284: Cannot set IC values")); 5500 XFree(attr_list); 5501 } 5502 } 5503 5504 #if defined(FEAT_GUI_X11) 5505 static char e_xim[] = N_("E285: Failed to create input context"); 5506 #endif 5507 5508 #if defined(FEAT_GUI_X11) || defined(PROTO) 5509 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(sun) 5510 # define USE_X11R6_XIM 5511 # endif 5512 5513 static int xim_real_init __ARGS((Window x11_window, Display *x11_display)); 5514 5515 5516 #ifdef USE_X11R6_XIM 5517 static void xim_instantiate_cb __ARGS((Display *display, XPointer client_data, XPointer call_data)); 5518 static void xim_destroy_cb __ARGS((XIM im, XPointer client_data, XPointer call_data)); 5519 5520 static void 5521 xim_instantiate_cb(display, client_data, call_data) 5522 Display *display; 5523 XPointer client_data UNUSED; 5524 XPointer call_data UNUSED; 5525 { 5526 Window x11_window; 5527 Display *x11_display; 5528 5529 #ifdef XIM_DEBUG 5530 xim_log("xim_instantiate_cb()\n"); 5531 #endif 5532 5533 gui_get_x11_windis(&x11_window, &x11_display); 5534 if (display != x11_display) 5535 return; 5536 5537 xim_real_init(x11_window, x11_display); 5538 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5539 if (xic != NULL) 5540 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5541 xim_instantiate_cb, NULL); 5542 } 5543 5544 static void 5545 xim_destroy_cb(im, client_data, call_data) 5546 XIM im UNUSED; 5547 XPointer client_data UNUSED; 5548 XPointer call_data UNUSED; 5549 { 5550 Window x11_window; 5551 Display *x11_display; 5552 5553 #ifdef XIM_DEBUG 5554 xim_log("xim_destroy_cb()\n"); 5555 #endif 5556 gui_get_x11_windis(&x11_window, &x11_display); 5557 5558 xic = NULL; 5559 status_area_enabled = FALSE; 5560 5561 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5562 5563 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5564 xim_instantiate_cb, NULL); 5565 } 5566 #endif 5567 5568 void 5569 xim_init() 5570 { 5571 Window x11_window; 5572 Display *x11_display; 5573 5574 #ifdef XIM_DEBUG 5575 xim_log("xim_init()\n"); 5576 #endif 5577 5578 gui_get_x11_windis(&x11_window, &x11_display); 5579 5580 xic = NULL; 5581 5582 if (xim_real_init(x11_window, x11_display)) 5583 return; 5584 5585 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5586 5587 #ifdef USE_X11R6_XIM 5588 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5589 xim_instantiate_cb, NULL); 5590 #endif 5591 } 5592 5593 static int 5594 xim_real_init(x11_window, x11_display) 5595 Window x11_window; 5596 Display *x11_display; 5597 { 5598 int i; 5599 char *p, 5600 *s, 5601 *ns, 5602 *end, 5603 tmp[1024]; 5604 #define IMLEN_MAX 40 5605 char buf[IMLEN_MAX + 7]; 5606 XIM xim = NULL; 5607 XIMStyles *xim_styles; 5608 XIMStyle this_input_style = 0; 5609 Boolean found; 5610 XPoint over_spot; 5611 XVaNestedList preedit_list, status_list; 5612 5613 input_style = 0; 5614 status_area_enabled = FALSE; 5615 5616 if (xic != NULL) 5617 return FALSE; 5618 5619 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) 5620 { 5621 strcpy(tmp, gui.rsrc_input_method); 5622 for (ns = s = tmp; ns != NULL && *s != NUL;) 5623 { 5624 s = (char *)skipwhite((char_u *)s); 5625 if (*s == NUL) 5626 break; 5627 if ((ns = end = strchr(s, ',')) == NULL) 5628 end = s + strlen(s); 5629 while (isspace(((char_u *)end)[-1])) 5630 end--; 5631 *end = NUL; 5632 5633 if (strlen(s) <= IMLEN_MAX) 5634 { 5635 strcpy(buf, "@im="); 5636 strcat(buf, s); 5637 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL 5638 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) 5639 != NULL) 5640 break; 5641 } 5642 5643 s = ns + 1; 5644 } 5645 } 5646 5647 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) 5648 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5649 5650 /* This is supposed to be useful to obtain characters through 5651 * XmbLookupString() without really using a XIM. */ 5652 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL 5653 && *p != NUL) 5654 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5655 5656 if (xim == NULL) 5657 { 5658 /* Only give this message when verbose is set, because too many people 5659 * got this message when they didn't want to use a XIM. */ 5660 if (p_verbose > 0) 5661 { 5662 verbose_enter(); 5663 EMSG(_("E286: Failed to open input method")); 5664 verbose_leave(); 5665 } 5666 return FALSE; 5667 } 5668 5669 #ifdef USE_X11R6_XIM 5670 { 5671 XIMCallback destroy_cb; 5672 5673 destroy_cb.callback = xim_destroy_cb; 5674 destroy_cb.client_data = NULL; 5675 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) 5676 EMSG(_("E287: Warning: Could not set destroy callback to IM")); 5677 } 5678 #endif 5679 5680 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) 5681 { 5682 EMSG(_("E288: input method doesn't support any style")); 5683 XCloseIM(xim); 5684 return FALSE; 5685 } 5686 5687 found = False; 5688 strcpy(tmp, gui.rsrc_preedit_type_name); 5689 for (s = tmp; s && !found; ) 5690 { 5691 while (*s && isspace((unsigned char)*s)) 5692 s++; 5693 if (!*s) 5694 break; 5695 if ((ns = end = strchr(s, ',')) != 0) 5696 ns++; 5697 else 5698 end = s + strlen(s); 5699 while (isspace((unsigned char)*end)) 5700 end--; 5701 *end = '\0'; 5702 5703 if (!strcmp(s, "OverTheSpot")) 5704 this_input_style = (XIMPreeditPosition | XIMStatusArea); 5705 else if (!strcmp(s, "OffTheSpot")) 5706 this_input_style = (XIMPreeditArea | XIMStatusArea); 5707 else if (!strcmp(s, "Root")) 5708 this_input_style = (XIMPreeditNothing | XIMStatusNothing); 5709 5710 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5711 { 5712 if (this_input_style == xim_styles->supported_styles[i]) 5713 { 5714 found = True; 5715 break; 5716 } 5717 } 5718 if (!found) 5719 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5720 { 5721 if ((xim_styles->supported_styles[i] & this_input_style) 5722 == (this_input_style & ~XIMStatusArea)) 5723 { 5724 this_input_style &= ~XIMStatusArea; 5725 found = True; 5726 break; 5727 } 5728 } 5729 5730 s = ns; 5731 } 5732 XFree(xim_styles); 5733 5734 if (!found) 5735 { 5736 /* Only give this message when verbose is set, because too many people 5737 * got this message when they didn't want to use a XIM. */ 5738 if (p_verbose > 0) 5739 { 5740 verbose_enter(); 5741 EMSG(_("E289: input method doesn't support my preedit type")); 5742 verbose_leave(); 5743 } 5744 XCloseIM(xim); 5745 return FALSE; 5746 } 5747 5748 over_spot.x = TEXT_X(gui.col); 5749 over_spot.y = TEXT_Y(gui.row); 5750 input_style = this_input_style; 5751 5752 /* A crash was reported when trying to pass gui.norm_font as XNFontSet, 5753 * thus that has been removed. Hopefully the default works... */ 5754 #ifdef FEAT_XFONTSET 5755 if (gui.fontset != NOFONTSET) 5756 { 5757 preedit_list = XVaCreateNestedList(0, 5758 XNSpotLocation, &over_spot, 5759 XNForeground, (Pixel)gui.def_norm_pixel, 5760 XNBackground, (Pixel)gui.def_back_pixel, 5761 XNFontSet, (XFontSet)gui.fontset, 5762 NULL); 5763 status_list = XVaCreateNestedList(0, 5764 XNForeground, (Pixel)gui.def_norm_pixel, 5765 XNBackground, (Pixel)gui.def_back_pixel, 5766 XNFontSet, (XFontSet)gui.fontset, 5767 NULL); 5768 } 5769 else 5770 #endif 5771 { 5772 preedit_list = XVaCreateNestedList(0, 5773 XNSpotLocation, &over_spot, 5774 XNForeground, (Pixel)gui.def_norm_pixel, 5775 XNBackground, (Pixel)gui.def_back_pixel, 5776 NULL); 5777 status_list = XVaCreateNestedList(0, 5778 XNForeground, (Pixel)gui.def_norm_pixel, 5779 XNBackground, (Pixel)gui.def_back_pixel, 5780 NULL); 5781 } 5782 5783 xic = XCreateIC(xim, 5784 XNInputStyle, input_style, 5785 XNClientWindow, x11_window, 5786 XNFocusWindow, gui.wid, 5787 XNPreeditAttributes, preedit_list, 5788 XNStatusAttributes, status_list, 5789 NULL); 5790 XFree(status_list); 5791 XFree(preedit_list); 5792 if (xic != NULL) 5793 { 5794 if (input_style & XIMStatusArea) 5795 { 5796 xim_set_status_area(); 5797 status_area_enabled = TRUE; 5798 } 5799 else 5800 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5801 } 5802 else 5803 { 5804 EMSG(_(e_xim)); 5805 XCloseIM(xim); 5806 return FALSE; 5807 } 5808 5809 return TRUE; 5810 } 5811 5812 #endif /* FEAT_GUI_X11 */ 5813 5814 /* 5815 * Get IM status. When IM is on, return TRUE. Else return FALSE. 5816 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is 5817 * active, when not having focus XIM may still be active (e.g., when using a 5818 * tear-off menu item). 5819 */ 5820 int 5821 im_get_status() 5822 { 5823 return xim_has_focus; 5824 } 5825 5826 # endif /* !FEAT_GUI_GTK */ 5827 5828 # if !defined(FEAT_GUI_GTK) || defined(PROTO) 5829 /* 5830 * Set up the status area. 5831 * 5832 * This should use a separate Widget, but that seems not possible, because 5833 * preedit_area and status_area should be set to the same window as for the 5834 * text input. Unfortunately this means the status area pollutes the text 5835 * window... 5836 */ 5837 void 5838 xim_set_status_area() 5839 { 5840 XVaNestedList preedit_list = 0, status_list = 0, list = 0; 5841 XRectangle pre_area, status_area; 5842 5843 if (xic == NULL) 5844 return; 5845 5846 if (input_style & XIMStatusArea) 5847 { 5848 if (input_style & XIMPreeditArea) 5849 { 5850 XRectangle *needed_rect; 5851 5852 /* to get status_area width */ 5853 status_list = XVaCreateNestedList(0, XNAreaNeeded, 5854 &needed_rect, NULL); 5855 XGetICValues(xic, XNStatusAttributes, status_list, NULL); 5856 XFree(status_list); 5857 5858 status_area.width = needed_rect->width; 5859 } 5860 else 5861 status_area.width = gui.char_width * Columns; 5862 5863 status_area.x = 0; 5864 status_area.y = gui.char_height * Rows + gui.border_offset; 5865 if (gui.which_scrollbars[SBAR_BOTTOM]) 5866 status_area.y += gui.scrollbar_height; 5867 #ifdef FEAT_MENU 5868 if (gui.menu_is_active) 5869 status_area.y += gui.menu_height; 5870 #endif 5871 status_area.height = gui.char_height; 5872 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); 5873 } 5874 else 5875 { 5876 status_area.x = 0; 5877 status_area.y = gui.char_height * Rows + gui.border_offset; 5878 if (gui.which_scrollbars[SBAR_BOTTOM]) 5879 status_area.y += gui.scrollbar_height; 5880 #ifdef FEAT_MENU 5881 if (gui.menu_is_active) 5882 status_area.y += gui.menu_height; 5883 #endif 5884 status_area.width = 0; 5885 status_area.height = gui.char_height; 5886 } 5887 5888 if (input_style & XIMPreeditArea) /* off-the-spot */ 5889 { 5890 pre_area.x = status_area.x + status_area.width; 5891 pre_area.y = gui.char_height * Rows + gui.border_offset; 5892 pre_area.width = gui.char_width * Columns - pre_area.x; 5893 if (gui.which_scrollbars[SBAR_BOTTOM]) 5894 pre_area.y += gui.scrollbar_height; 5895 #ifdef FEAT_MENU 5896 if (gui.menu_is_active) 5897 pre_area.y += gui.menu_height; 5898 #endif 5899 pre_area.height = gui.char_height; 5900 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 5901 } 5902 else if (input_style & XIMPreeditPosition) /* over-the-spot */ 5903 { 5904 pre_area.x = 0; 5905 pre_area.y = 0; 5906 pre_area.height = gui.char_height * Rows; 5907 pre_area.width = gui.char_width * Columns; 5908 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 5909 } 5910 5911 if (preedit_list && status_list) 5912 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 5913 XNStatusAttributes, status_list, NULL); 5914 else if (preedit_list) 5915 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 5916 NULL); 5917 else if (status_list) 5918 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, 5919 NULL); 5920 else 5921 list = NULL; 5922 5923 if (list) 5924 { 5925 XSetICValues(xic, XNVaNestedList, list, NULL); 5926 XFree(list); 5927 } 5928 if (status_list) 5929 XFree(status_list); 5930 if (preedit_list) 5931 XFree(preedit_list); 5932 } 5933 5934 int 5935 xim_get_status_area_height() 5936 { 5937 if (status_area_enabled) 5938 return gui.char_height; 5939 return 0; 5940 } 5941 # endif 5942 5943 #endif /* FEAT_XIM */ 5944 5945 #if defined(FEAT_MBYTE) || defined(PROTO) 5946 5947 /* 5948 * Setup "vcp" for conversion from "from" to "to". 5949 * The names must have been made canonical with enc_canonize(). 5950 * vcp->vc_type must have been initialized to CONV_NONE. 5951 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 5952 * instead). 5953 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 5954 * Return FAIL when conversion is not supported, OK otherwise. 5955 */ 5956 int 5957 convert_setup(vcp, from, to) 5958 vimconv_T *vcp; 5959 char_u *from; 5960 char_u *to; 5961 { 5962 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 5963 } 5964 5965 /* 5966 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 5967 * "from" unicode charsets be considered utf-8. Same for "to". 5968 */ 5969 int 5970 convert_setup_ext(vcp, from, from_unicode_is_utf8, to, to_unicode_is_utf8) 5971 vimconv_T *vcp; 5972 char_u *from; 5973 int from_unicode_is_utf8; 5974 char_u *to; 5975 int to_unicode_is_utf8; 5976 { 5977 int from_prop; 5978 int to_prop; 5979 int from_is_utf8; 5980 int to_is_utf8; 5981 5982 /* Reset to no conversion. */ 5983 # ifdef USE_ICONV 5984 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 5985 iconv_close(vcp->vc_fd); 5986 # endif 5987 vcp->vc_type = CONV_NONE; 5988 vcp->vc_factor = 1; 5989 vcp->vc_fail = FALSE; 5990 5991 /* No conversion when one of the names is empty or they are equal. */ 5992 if (from == NULL || *from == NUL || to == NULL || *to == NUL 5993 || STRCMP(from, to) == 0) 5994 return OK; 5995 5996 from_prop = enc_canon_props(from); 5997 to_prop = enc_canon_props(to); 5998 if (from_unicode_is_utf8) 5999 from_is_utf8 = from_prop & ENC_UNICODE; 6000 else 6001 from_is_utf8 = from_prop == ENC_UNICODE; 6002 if (to_unicode_is_utf8) 6003 to_is_utf8 = to_prop & ENC_UNICODE; 6004 else 6005 to_is_utf8 = to_prop == ENC_UNICODE; 6006 6007 if ((from_prop & ENC_LATIN1) && to_is_utf8) 6008 { 6009 /* Internal latin1 -> utf-8 conversion. */ 6010 vcp->vc_type = CONV_TO_UTF8; 6011 vcp->vc_factor = 2; /* up to twice as long */ 6012 } 6013 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 6014 { 6015 /* Internal latin9 -> utf-8 conversion. */ 6016 vcp->vc_type = CONV_9_TO_UTF8; 6017 vcp->vc_factor = 3; /* up to three as long (euro sign) */ 6018 } 6019 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 6020 { 6021 /* Internal utf-8 -> latin1 conversion. */ 6022 vcp->vc_type = CONV_TO_LATIN1; 6023 } 6024 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 6025 { 6026 /* Internal utf-8 -> latin9 conversion. */ 6027 vcp->vc_type = CONV_TO_LATIN9; 6028 } 6029 #ifdef WIN3264 6030 /* Win32-specific codepage <-> codepage conversion without iconv. */ 6031 else if ((from_is_utf8 || encname2codepage(from) > 0) 6032 && (to_is_utf8 || encname2codepage(to) > 0)) 6033 { 6034 vcp->vc_type = CONV_CODEPAGE; 6035 vcp->vc_factor = 2; /* up to twice as long */ 6036 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 6037 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 6038 } 6039 #endif 6040 #ifdef MACOS_X 6041 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 6042 { 6043 vcp->vc_type = CONV_MAC_LATIN1; 6044 } 6045 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 6046 { 6047 vcp->vc_type = CONV_MAC_UTF8; 6048 vcp->vc_factor = 2; /* up to twice as long */ 6049 } 6050 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 6051 { 6052 vcp->vc_type = CONV_LATIN1_MAC; 6053 } 6054 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 6055 { 6056 vcp->vc_type = CONV_UTF8_MAC; 6057 } 6058 #endif 6059 # ifdef USE_ICONV 6060 else 6061 { 6062 /* Use iconv() for conversion. */ 6063 vcp->vc_fd = (iconv_t)my_iconv_open( 6064 to_is_utf8 ? (char_u *)"utf-8" : to, 6065 from_is_utf8 ? (char_u *)"utf-8" : from); 6066 if (vcp->vc_fd != (iconv_t)-1) 6067 { 6068 vcp->vc_type = CONV_ICONV; 6069 vcp->vc_factor = 4; /* could be longer too... */ 6070 } 6071 } 6072 # endif 6073 if (vcp->vc_type == CONV_NONE) 6074 return FAIL; 6075 6076 return OK; 6077 } 6078 6079 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \ 6080 || defined(MSDOS) || defined(PROTO) 6081 /* 6082 * Do conversion on typed input characters in-place. 6083 * The input and output are not NUL terminated! 6084 * Returns the length after conversion. 6085 */ 6086 int 6087 convert_input(ptr, len, maxlen) 6088 char_u *ptr; 6089 int len; 6090 int maxlen; 6091 { 6092 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 6093 } 6094 #endif 6095 6096 /* 6097 * Like convert_input(), but when there is an incomplete byte sequence at the 6098 * end return that as an allocated string in "restp" and set "*restlenp" to 6099 * the length. If "restp" is NULL it is not used. 6100 */ 6101 int 6102 convert_input_safe(ptr, len, maxlen, restp, restlenp) 6103 char_u *ptr; 6104 int len; 6105 int maxlen; 6106 char_u **restp; 6107 int *restlenp; 6108 { 6109 char_u *d; 6110 int dlen = len; 6111 int unconvertlen = 0; 6112 6113 d = string_convert_ext(&input_conv, ptr, &dlen, 6114 restp == NULL ? NULL : &unconvertlen); 6115 if (d != NULL) 6116 { 6117 if (dlen <= maxlen) 6118 { 6119 if (unconvertlen > 0) 6120 { 6121 /* Move the unconverted characters to allocated memory. */ 6122 *restp = alloc(unconvertlen); 6123 if (*restp != NULL) 6124 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 6125 *restlenp = unconvertlen; 6126 } 6127 mch_memmove(ptr, d, dlen); 6128 } 6129 else 6130 /* result is too long, keep the unconverted text (the caller must 6131 * have done something wrong!) */ 6132 dlen = len; 6133 vim_free(d); 6134 } 6135 return dlen; 6136 } 6137 6138 /* 6139 * Convert text "ptr[*lenp]" according to "vcp". 6140 * Returns the result in allocated memory and sets "*lenp". 6141 * When "lenp" is NULL, use NUL terminated strings. 6142 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 6143 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 6144 */ 6145 char_u * 6146 string_convert(vcp, ptr, lenp) 6147 vimconv_T *vcp; 6148 char_u *ptr; 6149 int *lenp; 6150 { 6151 return string_convert_ext(vcp, ptr, lenp, NULL); 6152 } 6153 6154 /* 6155 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 6156 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 6157 * set to the number of remaining bytes. 6158 */ 6159 char_u * 6160 string_convert_ext(vcp, ptr, lenp, unconvlenp) 6161 vimconv_T *vcp; 6162 char_u *ptr; 6163 int *lenp; 6164 int *unconvlenp; 6165 { 6166 char_u *retval = NULL; 6167 char_u *d; 6168 int len; 6169 int i; 6170 int l; 6171 int c; 6172 6173 if (lenp == NULL) 6174 len = (int)STRLEN(ptr); 6175 else 6176 len = *lenp; 6177 if (len == 0) 6178 return vim_strsave((char_u *)""); 6179 6180 switch (vcp->vc_type) 6181 { 6182 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */ 6183 retval = alloc(len * 2 + 1); 6184 if (retval == NULL) 6185 break; 6186 d = retval; 6187 for (i = 0; i < len; ++i) 6188 { 6189 c = ptr[i]; 6190 if (c < 0x80) 6191 *d++ = c; 6192 else 6193 { 6194 *d++ = 0xc0 + ((unsigned)c >> 6); 6195 *d++ = 0x80 + (c & 0x3f); 6196 } 6197 } 6198 *d = NUL; 6199 if (lenp != NULL) 6200 *lenp = (int)(d - retval); 6201 break; 6202 6203 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */ 6204 retval = alloc(len * 3 + 1); 6205 if (retval == NULL) 6206 break; 6207 d = retval; 6208 for (i = 0; i < len; ++i) 6209 { 6210 c = ptr[i]; 6211 switch (c) 6212 { 6213 case 0xa4: c = 0x20ac; break; /* euro */ 6214 case 0xa6: c = 0x0160; break; /* S hat */ 6215 case 0xa8: c = 0x0161; break; /* S -hat */ 6216 case 0xb4: c = 0x017d; break; /* Z hat */ 6217 case 0xb8: c = 0x017e; break; /* Z -hat */ 6218 case 0xbc: c = 0x0152; break; /* OE */ 6219 case 0xbd: c = 0x0153; break; /* oe */ 6220 case 0xbe: c = 0x0178; break; /* Y */ 6221 } 6222 d += utf_char2bytes(c, d); 6223 } 6224 *d = NUL; 6225 if (lenp != NULL) 6226 *lenp = (int)(d - retval); 6227 break; 6228 6229 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */ 6230 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */ 6231 retval = alloc(len + 1); 6232 if (retval == NULL) 6233 break; 6234 d = retval; 6235 for (i = 0; i < len; ++i) 6236 { 6237 l = utf_ptr2len_len(ptr + i, len - i); 6238 if (l == 0) 6239 *d++ = NUL; 6240 else if (l == 1) 6241 { 6242 int l_w = utf8len_tab_zero[ptr[i]]; 6243 6244 if (l_w == 0) 6245 { 6246 /* Illegal utf-8 byte cannot be converted */ 6247 vim_free(retval); 6248 return NULL; 6249 } 6250 if (unconvlenp != NULL && l_w > len - i) 6251 { 6252 /* Incomplete sequence at the end. */ 6253 *unconvlenp = len - i; 6254 break; 6255 } 6256 *d++ = ptr[i]; 6257 } 6258 else 6259 { 6260 c = utf_ptr2char(ptr + i); 6261 if (vcp->vc_type == CONV_TO_LATIN9) 6262 switch (c) 6263 { 6264 case 0x20ac: c = 0xa4; break; /* euro */ 6265 case 0x0160: c = 0xa6; break; /* S hat */ 6266 case 0x0161: c = 0xa8; break; /* S -hat */ 6267 case 0x017d: c = 0xb4; break; /* Z hat */ 6268 case 0x017e: c = 0xb8; break; /* Z -hat */ 6269 case 0x0152: c = 0xbc; break; /* OE */ 6270 case 0x0153: c = 0xbd; break; /* oe */ 6271 case 0x0178: c = 0xbe; break; /* Y */ 6272 case 0xa4: 6273 case 0xa6: 6274 case 0xa8: 6275 case 0xb4: 6276 case 0xb8: 6277 case 0xbc: 6278 case 0xbd: 6279 case 0xbe: c = 0x100; break; /* not in latin9 */ 6280 } 6281 if (!utf_iscomposing(c)) /* skip composing chars */ 6282 { 6283 if (c < 0x100) 6284 *d++ = c; 6285 else if (vcp->vc_fail) 6286 { 6287 vim_free(retval); 6288 return NULL; 6289 } 6290 else 6291 { 6292 *d++ = 0xbf; 6293 if (utf_char2cells(c) > 1) 6294 *d++ = '?'; 6295 } 6296 } 6297 i += l - 1; 6298 } 6299 } 6300 *d = NUL; 6301 if (lenp != NULL) 6302 *lenp = (int)(d - retval); 6303 break; 6304 6305 # ifdef MACOS_CONVERT 6306 case CONV_MAC_LATIN1: 6307 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6308 'm', 'l', unconvlenp); 6309 break; 6310 6311 case CONV_LATIN1_MAC: 6312 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6313 'l', 'm', unconvlenp); 6314 break; 6315 6316 case CONV_MAC_UTF8: 6317 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6318 'm', 'u', unconvlenp); 6319 break; 6320 6321 case CONV_UTF8_MAC: 6322 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6323 'u', 'm', unconvlenp); 6324 break; 6325 # endif 6326 6327 # ifdef USE_ICONV 6328 case CONV_ICONV: /* conversion with output_conv.vc_fd */ 6329 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 6330 break; 6331 # endif 6332 # ifdef WIN3264 6333 case CONV_CODEPAGE: /* codepage -> codepage */ 6334 { 6335 int retlen; 6336 int tmp_len; 6337 short_u *tmp; 6338 6339 /* 1. codepage/UTF-8 -> ucs-2. */ 6340 if (vcp->vc_cpfrom == 0) 6341 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 6342 else 6343 { 6344 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 6345 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 6346 ptr, len, 0, 0); 6347 if (tmp_len == 0 6348 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 6349 { 6350 if (lenp != NULL) 6351 *lenp = 0; 6352 if (unconvlenp != NULL) 6353 *unconvlenp = len; 6354 retval = alloc(1); 6355 if (retval) 6356 retval[0] = NUL; 6357 return retval; 6358 } 6359 } 6360 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len); 6361 if (tmp == NULL) 6362 break; 6363 if (vcp->vc_cpfrom == 0) 6364 utf8_to_utf16(ptr, len, tmp, unconvlenp); 6365 else 6366 MultiByteToWideChar(vcp->vc_cpfrom, 0, ptr, len, tmp, tmp_len); 6367 6368 /* 2. ucs-2 -> codepage/UTF-8. */ 6369 if (vcp->vc_cpto == 0) 6370 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 6371 else 6372 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 6373 tmp, tmp_len, 0, 0, 0, 0); 6374 retval = alloc(retlen + 1); 6375 if (retval != NULL) 6376 { 6377 if (vcp->vc_cpto == 0) 6378 utf16_to_utf8(tmp, tmp_len, retval); 6379 else 6380 WideCharToMultiByte(vcp->vc_cpto, 0, 6381 tmp, tmp_len, retval, retlen, 0, 0); 6382 retval[retlen] = NUL; 6383 if (lenp != NULL) 6384 *lenp = retlen; 6385 } 6386 vim_free(tmp); 6387 break; 6388 } 6389 # endif 6390 } 6391 6392 return retval; 6393 } 6394 #endif 6395