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 {0x08e3, 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 {0x1a17, 0x1a1b}, 2370 {0x1a55, 0x1a5e}, 2371 {0x1a60, 0x1a7c}, 2372 {0x1a7f, 0x1a7f}, 2373 {0x1ab0, 0x1abe}, 2374 {0x1b00, 0x1b04}, 2375 {0x1b34, 0x1b44}, 2376 {0x1b6b, 0x1b73}, 2377 {0x1b80, 0x1b82}, 2378 {0x1ba1, 0x1bad}, 2379 {0x1be6, 0x1bf3}, 2380 {0x1c24, 0x1c37}, 2381 {0x1cd0, 0x1cd2}, 2382 {0x1cd4, 0x1ce8}, 2383 {0x1ced, 0x1ced}, 2384 {0x1cf2, 0x1cf4}, 2385 {0x1cf8, 0x1cf9}, 2386 {0x1dc0, 0x1df5}, 2387 {0x1dfc, 0x1dff}, 2388 {0x20d0, 0x20f0}, 2389 {0x2cef, 0x2cf1}, 2390 {0x2d7f, 0x2d7f}, 2391 {0x2de0, 0x2dff}, 2392 {0x302a, 0x302f}, 2393 {0x3099, 0x309a}, 2394 {0xa66f, 0xa672}, 2395 {0xa674, 0xa67d}, 2396 {0xa69e, 0xa69f}, 2397 {0xa6f0, 0xa6f1}, 2398 {0xa802, 0xa802}, 2399 {0xa806, 0xa806}, 2400 {0xa80b, 0xa80b}, 2401 {0xa823, 0xa827}, 2402 {0xa880, 0xa881}, 2403 {0xa8b4, 0xa8c4}, 2404 {0xa8e0, 0xa8f1}, 2405 {0xa926, 0xa92d}, 2406 {0xa947, 0xa953}, 2407 {0xa980, 0xa983}, 2408 {0xa9b3, 0xa9c0}, 2409 {0xa9e5, 0xa9e5}, 2410 {0xaa29, 0xaa36}, 2411 {0xaa43, 0xaa43}, 2412 {0xaa4c, 0xaa4d}, 2413 {0xaa7b, 0xaa7d}, 2414 {0xaab0, 0xaab0}, 2415 {0xaab2, 0xaab4}, 2416 {0xaab7, 0xaab8}, 2417 {0xaabe, 0xaabf}, 2418 {0xaac1, 0xaac1}, 2419 {0xaaeb, 0xaaef}, 2420 {0xaaf5, 0xaaf6}, 2421 {0xabe3, 0xabea}, 2422 {0xabec, 0xabed}, 2423 {0xfb1e, 0xfb1e}, 2424 {0xfe00, 0xfe0f}, 2425 {0xfe20, 0xfe2f}, 2426 {0x101fd, 0x101fd}, 2427 {0x102e0, 0x102e0}, 2428 {0x10376, 0x1037a}, 2429 {0x10a01, 0x10a03}, 2430 {0x10a05, 0x10a06}, 2431 {0x10a0c, 0x10a0f}, 2432 {0x10a38, 0x10a3a}, 2433 {0x10a3f, 0x10a3f}, 2434 {0x10ae5, 0x10ae6}, 2435 {0x11000, 0x11002}, 2436 {0x11038, 0x11046}, 2437 {0x1107f, 0x11082}, 2438 {0x110b0, 0x110ba}, 2439 {0x11100, 0x11102}, 2440 {0x11127, 0x11134}, 2441 {0x11173, 0x11173}, 2442 {0x11180, 0x11182}, 2443 {0x111b3, 0x111c0}, 2444 {0x111ca, 0x111cc}, 2445 {0x1122c, 0x11237}, 2446 {0x112df, 0x112ea}, 2447 {0x11300, 0x11303}, 2448 {0x1133c, 0x1133c}, 2449 {0x1133e, 0x11344}, 2450 {0x11347, 0x11348}, 2451 {0x1134b, 0x1134d}, 2452 {0x11357, 0x11357}, 2453 {0x11362, 0x11363}, 2454 {0x11366, 0x1136c}, 2455 {0x11370, 0x11374}, 2456 {0x114b0, 0x114c3}, 2457 {0x115af, 0x115b5}, 2458 {0x115b8, 0x115c0}, 2459 {0x115dc, 0x115dd}, 2460 {0x11630, 0x11640}, 2461 {0x116ab, 0x116b7}, 2462 {0x1171d, 0x1172b}, 2463 {0x16af0, 0x16af4}, 2464 {0x16b30, 0x16b36}, 2465 {0x16f51, 0x16f7e}, 2466 {0x16f8f, 0x16f92}, 2467 {0x1bc9d, 0x1bc9e}, 2468 {0x1d165, 0x1d169}, 2469 {0x1d16d, 0x1d172}, 2470 {0x1d17b, 0x1d182}, 2471 {0x1d185, 0x1d18b}, 2472 {0x1d1aa, 0x1d1ad}, 2473 {0x1d242, 0x1d244}, 2474 {0x1da00, 0x1da36}, 2475 {0x1da3b, 0x1da6c}, 2476 {0x1da75, 0x1da75}, 2477 {0x1da84, 0x1da84}, 2478 {0x1da9b, 0x1da9f}, 2479 {0x1daa1, 0x1daaf}, 2480 {0x1e8d0, 0x1e8d6}, 2481 {0xe0100, 0xe01ef} 2482 }; 2483 2484 return intable(combining, sizeof(combining), c); 2485 } 2486 2487 /* 2488 * Return TRUE for characters that can be displayed in a normal way. 2489 * Only for characters of 0x100 and above! 2490 */ 2491 int 2492 utf_printable(c) 2493 int c; 2494 { 2495 #ifdef USE_WCHAR_FUNCTIONS 2496 /* 2497 * Assume the iswprint() library function works better than our own stuff. 2498 */ 2499 return iswprint(c); 2500 #else 2501 /* Sorted list of non-overlapping intervals. 2502 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */ 2503 static struct interval nonprint[] = 2504 { 2505 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2506 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2507 {0xfffe, 0xffff} 2508 }; 2509 2510 return !intable(nonprint, sizeof(nonprint), c); 2511 #endif 2512 } 2513 2514 /* 2515 * Get class of a Unicode character. 2516 * 0: white space 2517 * 1: punctuation 2518 * 2 or bigger: some class of word character. 2519 */ 2520 int 2521 utf_class(c) 2522 int c; 2523 { 2524 /* sorted list of non-overlapping intervals */ 2525 static struct clinterval 2526 { 2527 unsigned int first; 2528 unsigned int last; 2529 unsigned int class; 2530 } classes[] = 2531 { 2532 {0x037e, 0x037e, 1}, /* Greek question mark */ 2533 {0x0387, 0x0387, 1}, /* Greek ano teleia */ 2534 {0x055a, 0x055f, 1}, /* Armenian punctuation */ 2535 {0x0589, 0x0589, 1}, /* Armenian full stop */ 2536 {0x05be, 0x05be, 1}, 2537 {0x05c0, 0x05c0, 1}, 2538 {0x05c3, 0x05c3, 1}, 2539 {0x05f3, 0x05f4, 1}, 2540 {0x060c, 0x060c, 1}, 2541 {0x061b, 0x061b, 1}, 2542 {0x061f, 0x061f, 1}, 2543 {0x066a, 0x066d, 1}, 2544 {0x06d4, 0x06d4, 1}, 2545 {0x0700, 0x070d, 1}, /* Syriac punctuation */ 2546 {0x0964, 0x0965, 1}, 2547 {0x0970, 0x0970, 1}, 2548 {0x0df4, 0x0df4, 1}, 2549 {0x0e4f, 0x0e4f, 1}, 2550 {0x0e5a, 0x0e5b, 1}, 2551 {0x0f04, 0x0f12, 1}, 2552 {0x0f3a, 0x0f3d, 1}, 2553 {0x0f85, 0x0f85, 1}, 2554 {0x104a, 0x104f, 1}, /* Myanmar punctuation */ 2555 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */ 2556 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */ 2557 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */ 2558 {0x1680, 0x1680, 0}, 2559 {0x169b, 0x169c, 1}, 2560 {0x16eb, 0x16ed, 1}, 2561 {0x1735, 0x1736, 1}, 2562 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */ 2563 {0x1800, 0x180a, 1}, /* Mongolian punctuation */ 2564 {0x2000, 0x200b, 0}, /* spaces */ 2565 {0x200c, 0x2027, 1}, /* punctuation and symbols */ 2566 {0x2028, 0x2029, 0}, 2567 {0x202a, 0x202e, 1}, /* punctuation and symbols */ 2568 {0x202f, 0x202f, 0}, 2569 {0x2030, 0x205e, 1}, /* punctuation and symbols */ 2570 {0x205f, 0x205f, 0}, 2571 {0x2060, 0x27ff, 1}, /* punctuation and symbols */ 2572 {0x2070, 0x207f, 0x2070}, /* superscript */ 2573 {0x2080, 0x2094, 0x2080}, /* subscript */ 2574 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */ 2575 {0x2800, 0x28ff, 0x2800}, /* braille */ 2576 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */ 2577 {0x29d8, 0x29db, 1}, 2578 {0x29fc, 0x29fd, 1}, 2579 {0x2e00, 0x2e7f, 1}, /* supplemental punctuation */ 2580 {0x3000, 0x3000, 0}, /* ideographic space */ 2581 {0x3001, 0x3020, 1}, /* ideographic punctuation */ 2582 {0x3030, 0x3030, 1}, 2583 {0x303d, 0x303d, 1}, 2584 {0x3040, 0x309f, 0x3040}, /* Hiragana */ 2585 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */ 2586 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */ 2587 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */ 2588 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */ 2589 {0xfd3e, 0xfd3f, 1}, 2590 {0xfe30, 0xfe6b, 1}, /* punctuation forms */ 2591 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */ 2592 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */ 2593 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */ 2594 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */ 2595 {0x20000, 0x2a6df, 0x4e00}, /* CJK Ideographs */ 2596 {0x2a700, 0x2b73f, 0x4e00}, /* CJK Ideographs */ 2597 {0x2b740, 0x2b81f, 0x4e00}, /* CJK Ideographs */ 2598 {0x2f800, 0x2fa1f, 0x4e00}, /* CJK Ideographs */ 2599 }; 2600 int bot = 0; 2601 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2602 int mid; 2603 2604 /* First quick check for Latin1 characters, use 'iskeyword'. */ 2605 if (c < 0x100) 2606 { 2607 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2608 return 0; /* blank */ 2609 if (vim_iswordc(c)) 2610 return 2; /* word character */ 2611 return 1; /* punctuation */ 2612 } 2613 2614 /* binary search in table */ 2615 while (top >= bot) 2616 { 2617 mid = (bot + top) / 2; 2618 if (classes[mid].last < (unsigned int)c) 2619 bot = mid + 1; 2620 else if (classes[mid].first > (unsigned int)c) 2621 top = mid - 1; 2622 else 2623 return (int)classes[mid].class; 2624 } 2625 2626 /* most other characters are "word" characters */ 2627 return 2; 2628 } 2629 2630 /* 2631 * Code for Unicode case-dependent operations. Based on notes in 2632 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2633 * This code uses simple case folding, not full case folding. 2634 * Last updated for Unicode 5.2. 2635 */ 2636 2637 /* 2638 * The following tables are built by ../runtime/tools/unicode.vim. 2639 * They must be in numeric order, because we use binary search. 2640 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2641 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2642 * folded/upper/lower by adding 32. 2643 */ 2644 typedef struct 2645 { 2646 int rangeStart; 2647 int rangeEnd; 2648 int step; 2649 int offset; 2650 } convertStruct; 2651 2652 static convertStruct foldCase[] = 2653 { 2654 {0x41,0x5a,1,32}, 2655 {0xb5,0xb5,-1,775}, 2656 {0xc0,0xd6,1,32}, 2657 {0xd8,0xde,1,32}, 2658 {0x100,0x12e,2,1}, 2659 {0x132,0x136,2,1}, 2660 {0x139,0x147,2,1}, 2661 {0x14a,0x176,2,1}, 2662 {0x178,0x178,-1,-121}, 2663 {0x179,0x17d,2,1}, 2664 {0x17f,0x17f,-1,-268}, 2665 {0x181,0x181,-1,210}, 2666 {0x182,0x184,2,1}, 2667 {0x186,0x186,-1,206}, 2668 {0x187,0x187,-1,1}, 2669 {0x189,0x18a,1,205}, 2670 {0x18b,0x18b,-1,1}, 2671 {0x18e,0x18e,-1,79}, 2672 {0x18f,0x18f,-1,202}, 2673 {0x190,0x190,-1,203}, 2674 {0x191,0x191,-1,1}, 2675 {0x193,0x193,-1,205}, 2676 {0x194,0x194,-1,207}, 2677 {0x196,0x196,-1,211}, 2678 {0x197,0x197,-1,209}, 2679 {0x198,0x198,-1,1}, 2680 {0x19c,0x19c,-1,211}, 2681 {0x19d,0x19d,-1,213}, 2682 {0x19f,0x19f,-1,214}, 2683 {0x1a0,0x1a4,2,1}, 2684 {0x1a6,0x1a6,-1,218}, 2685 {0x1a7,0x1a7,-1,1}, 2686 {0x1a9,0x1a9,-1,218}, 2687 {0x1ac,0x1ac,-1,1}, 2688 {0x1ae,0x1ae,-1,218}, 2689 {0x1af,0x1af,-1,1}, 2690 {0x1b1,0x1b2,1,217}, 2691 {0x1b3,0x1b5,2,1}, 2692 {0x1b7,0x1b7,-1,219}, 2693 {0x1b8,0x1bc,4,1}, 2694 {0x1c4,0x1c4,-1,2}, 2695 {0x1c5,0x1c5,-1,1}, 2696 {0x1c7,0x1c7,-1,2}, 2697 {0x1c8,0x1c8,-1,1}, 2698 {0x1ca,0x1ca,-1,2}, 2699 {0x1cb,0x1db,2,1}, 2700 {0x1de,0x1ee,2,1}, 2701 {0x1f1,0x1f1,-1,2}, 2702 {0x1f2,0x1f4,2,1}, 2703 {0x1f6,0x1f6,-1,-97}, 2704 {0x1f7,0x1f7,-1,-56}, 2705 {0x1f8,0x21e,2,1}, 2706 {0x220,0x220,-1,-130}, 2707 {0x222,0x232,2,1}, 2708 {0x23a,0x23a,-1,10795}, 2709 {0x23b,0x23b,-1,1}, 2710 {0x23d,0x23d,-1,-163}, 2711 {0x23e,0x23e,-1,10792}, 2712 {0x241,0x241,-1,1}, 2713 {0x243,0x243,-1,-195}, 2714 {0x244,0x244,-1,69}, 2715 {0x245,0x245,-1,71}, 2716 {0x246,0x24e,2,1}, 2717 {0x345,0x345,-1,116}, 2718 {0x370,0x372,2,1}, 2719 {0x376,0x376,-1,1}, 2720 {0x37f,0x37f,-1,116}, 2721 {0x386,0x386,-1,38}, 2722 {0x388,0x38a,1,37}, 2723 {0x38c,0x38c,-1,64}, 2724 {0x38e,0x38f,1,63}, 2725 {0x391,0x3a1,1,32}, 2726 {0x3a3,0x3ab,1,32}, 2727 {0x3c2,0x3c2,-1,1}, 2728 {0x3cf,0x3cf,-1,8}, 2729 {0x3d0,0x3d0,-1,-30}, 2730 {0x3d1,0x3d1,-1,-25}, 2731 {0x3d5,0x3d5,-1,-15}, 2732 {0x3d6,0x3d6,-1,-22}, 2733 {0x3d8,0x3ee,2,1}, 2734 {0x3f0,0x3f0,-1,-54}, 2735 {0x3f1,0x3f1,-1,-48}, 2736 {0x3f4,0x3f4,-1,-60}, 2737 {0x3f5,0x3f5,-1,-64}, 2738 {0x3f7,0x3f7,-1,1}, 2739 {0x3f9,0x3f9,-1,-7}, 2740 {0x3fa,0x3fa,-1,1}, 2741 {0x3fd,0x3ff,1,-130}, 2742 {0x400,0x40f,1,80}, 2743 {0x410,0x42f,1,32}, 2744 {0x460,0x480,2,1}, 2745 {0x48a,0x4be,2,1}, 2746 {0x4c0,0x4c0,-1,15}, 2747 {0x4c1,0x4cd,2,1}, 2748 {0x4d0,0x52e,2,1}, 2749 {0x531,0x556,1,48}, 2750 {0x10a0,0x10c5,1,7264}, 2751 {0x10c7,0x10cd,6,7264}, 2752 {0x13f8,0x13fd,1,-8}, 2753 {0x1e00,0x1e94,2,1}, 2754 {0x1e9b,0x1e9b,-1,-58}, 2755 {0x1e9e,0x1e9e,-1,-7615}, 2756 {0x1ea0,0x1efe,2,1}, 2757 {0x1f08,0x1f0f,1,-8}, 2758 {0x1f18,0x1f1d,1,-8}, 2759 {0x1f28,0x1f2f,1,-8}, 2760 {0x1f38,0x1f3f,1,-8}, 2761 {0x1f48,0x1f4d,1,-8}, 2762 {0x1f59,0x1f5f,2,-8}, 2763 {0x1f68,0x1f6f,1,-8}, 2764 {0x1f88,0x1f8f,1,-8}, 2765 {0x1f98,0x1f9f,1,-8}, 2766 {0x1fa8,0x1faf,1,-8}, 2767 {0x1fb8,0x1fb9,1,-8}, 2768 {0x1fba,0x1fbb,1,-74}, 2769 {0x1fbc,0x1fbc,-1,-9}, 2770 {0x1fbe,0x1fbe,-1,-7173}, 2771 {0x1fc8,0x1fcb,1,-86}, 2772 {0x1fcc,0x1fcc,-1,-9}, 2773 {0x1fd8,0x1fd9,1,-8}, 2774 {0x1fda,0x1fdb,1,-100}, 2775 {0x1fe8,0x1fe9,1,-8}, 2776 {0x1fea,0x1feb,1,-112}, 2777 {0x1fec,0x1fec,-1,-7}, 2778 {0x1ff8,0x1ff9,1,-128}, 2779 {0x1ffa,0x1ffb,1,-126}, 2780 {0x1ffc,0x1ffc,-1,-9}, 2781 {0x2126,0x2126,-1,-7517}, 2782 {0x212a,0x212a,-1,-8383}, 2783 {0x212b,0x212b,-1,-8262}, 2784 {0x2132,0x2132,-1,28}, 2785 {0x2160,0x216f,1,16}, 2786 {0x2183,0x2183,-1,1}, 2787 {0x24b6,0x24cf,1,26}, 2788 {0x2c00,0x2c2e,1,48}, 2789 {0x2c60,0x2c60,-1,1}, 2790 {0x2c62,0x2c62,-1,-10743}, 2791 {0x2c63,0x2c63,-1,-3814}, 2792 {0x2c64,0x2c64,-1,-10727}, 2793 {0x2c67,0x2c6b,2,1}, 2794 {0x2c6d,0x2c6d,-1,-10780}, 2795 {0x2c6e,0x2c6e,-1,-10749}, 2796 {0x2c6f,0x2c6f,-1,-10783}, 2797 {0x2c70,0x2c70,-1,-10782}, 2798 {0x2c72,0x2c75,3,1}, 2799 {0x2c7e,0x2c7f,1,-10815}, 2800 {0x2c80,0x2ce2,2,1}, 2801 {0x2ceb,0x2ced,2,1}, 2802 {0x2cf2,0xa640,31054,1}, 2803 {0xa642,0xa66c,2,1}, 2804 {0xa680,0xa69a,2,1}, 2805 {0xa722,0xa72e,2,1}, 2806 {0xa732,0xa76e,2,1}, 2807 {0xa779,0xa77b,2,1}, 2808 {0xa77d,0xa77d,-1,-35332}, 2809 {0xa77e,0xa786,2,1}, 2810 {0xa78b,0xa78b,-1,1}, 2811 {0xa78d,0xa78d,-1,-42280}, 2812 {0xa790,0xa792,2,1}, 2813 {0xa796,0xa7a8,2,1}, 2814 {0xa7aa,0xa7aa,-1,-42308}, 2815 {0xa7ab,0xa7ab,-1,-42319}, 2816 {0xa7ac,0xa7ac,-1,-42315}, 2817 {0xa7ad,0xa7ad,-1,-42305}, 2818 {0xa7b0,0xa7b0,-1,-42258}, 2819 {0xa7b1,0xa7b1,-1,-42282}, 2820 {0xa7b2,0xa7b2,-1,-42261}, 2821 {0xa7b3,0xa7b3,-1,928}, 2822 {0xa7b4,0xa7b6,2,1}, 2823 {0xab70,0xabbf,1,-38864}, 2824 {0xff21,0xff3a,1,32}, 2825 {0x10400,0x10427,1,40}, 2826 {0x10c80,0x10cb2,1,64}, 2827 {0x118a0,0x118bf,1,32} 2828 }; 2829 2830 static int utf_convert __ARGS((int a, convertStruct table[], int tableSize)); 2831 static int utf_strnicmp __ARGS((char_u *s1, char_u *s2, size_t n1, size_t n2)); 2832 2833 /* 2834 * Generic conversion function for case operations. 2835 * Return the converted equivalent of "a", which is a UCS-4 character. Use 2836 * the given conversion "table". Uses binary search on "table". 2837 */ 2838 static int 2839 utf_convert(a, table, tableSize) 2840 int a; 2841 convertStruct table[]; 2842 int tableSize; 2843 { 2844 int start, mid, end; /* indices into table */ 2845 int entries = tableSize / sizeof(convertStruct); 2846 2847 start = 0; 2848 end = entries; 2849 while (start < end) 2850 { 2851 /* need to search further */ 2852 mid = (end + start) / 2; 2853 if (table[mid].rangeEnd < a) 2854 start = mid + 1; 2855 else 2856 end = mid; 2857 } 2858 if (start < entries 2859 && table[start].rangeStart <= a 2860 && a <= table[start].rangeEnd 2861 && (a - table[start].rangeStart) % table[start].step == 0) 2862 return (a + table[start].offset); 2863 else 2864 return a; 2865 } 2866 2867 /* 2868 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 2869 * simple case folding. 2870 */ 2871 int 2872 utf_fold(a) 2873 int a; 2874 { 2875 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 2876 } 2877 2878 static convertStruct toLower[] = 2879 { 2880 {0x41,0x5a,1,32}, 2881 {0xc0,0xd6,1,32}, 2882 {0xd8,0xde,1,32}, 2883 {0x100,0x12e,2,1}, 2884 {0x130,0x130,-1,-199}, 2885 {0x132,0x136,2,1}, 2886 {0x139,0x147,2,1}, 2887 {0x14a,0x176,2,1}, 2888 {0x178,0x178,-1,-121}, 2889 {0x179,0x17d,2,1}, 2890 {0x181,0x181,-1,210}, 2891 {0x182,0x184,2,1}, 2892 {0x186,0x186,-1,206}, 2893 {0x187,0x187,-1,1}, 2894 {0x189,0x18a,1,205}, 2895 {0x18b,0x18b,-1,1}, 2896 {0x18e,0x18e,-1,79}, 2897 {0x18f,0x18f,-1,202}, 2898 {0x190,0x190,-1,203}, 2899 {0x191,0x191,-1,1}, 2900 {0x193,0x193,-1,205}, 2901 {0x194,0x194,-1,207}, 2902 {0x196,0x196,-1,211}, 2903 {0x197,0x197,-1,209}, 2904 {0x198,0x198,-1,1}, 2905 {0x19c,0x19c,-1,211}, 2906 {0x19d,0x19d,-1,213}, 2907 {0x19f,0x19f,-1,214}, 2908 {0x1a0,0x1a4,2,1}, 2909 {0x1a6,0x1a6,-1,218}, 2910 {0x1a7,0x1a7,-1,1}, 2911 {0x1a9,0x1a9,-1,218}, 2912 {0x1ac,0x1ac,-1,1}, 2913 {0x1ae,0x1ae,-1,218}, 2914 {0x1af,0x1af,-1,1}, 2915 {0x1b1,0x1b2,1,217}, 2916 {0x1b3,0x1b5,2,1}, 2917 {0x1b7,0x1b7,-1,219}, 2918 {0x1b8,0x1bc,4,1}, 2919 {0x1c4,0x1c4,-1,2}, 2920 {0x1c5,0x1c5,-1,1}, 2921 {0x1c7,0x1c7,-1,2}, 2922 {0x1c8,0x1c8,-1,1}, 2923 {0x1ca,0x1ca,-1,2}, 2924 {0x1cb,0x1db,2,1}, 2925 {0x1de,0x1ee,2,1}, 2926 {0x1f1,0x1f1,-1,2}, 2927 {0x1f2,0x1f4,2,1}, 2928 {0x1f6,0x1f6,-1,-97}, 2929 {0x1f7,0x1f7,-1,-56}, 2930 {0x1f8,0x21e,2,1}, 2931 {0x220,0x220,-1,-130}, 2932 {0x222,0x232,2,1}, 2933 {0x23a,0x23a,-1,10795}, 2934 {0x23b,0x23b,-1,1}, 2935 {0x23d,0x23d,-1,-163}, 2936 {0x23e,0x23e,-1,10792}, 2937 {0x241,0x241,-1,1}, 2938 {0x243,0x243,-1,-195}, 2939 {0x244,0x244,-1,69}, 2940 {0x245,0x245,-1,71}, 2941 {0x246,0x24e,2,1}, 2942 {0x370,0x372,2,1}, 2943 {0x376,0x376,-1,1}, 2944 {0x37f,0x37f,-1,116}, 2945 {0x386,0x386,-1,38}, 2946 {0x388,0x38a,1,37}, 2947 {0x38c,0x38c,-1,64}, 2948 {0x38e,0x38f,1,63}, 2949 {0x391,0x3a1,1,32}, 2950 {0x3a3,0x3ab,1,32}, 2951 {0x3cf,0x3cf,-1,8}, 2952 {0x3d8,0x3ee,2,1}, 2953 {0x3f4,0x3f4,-1,-60}, 2954 {0x3f7,0x3f7,-1,1}, 2955 {0x3f9,0x3f9,-1,-7}, 2956 {0x3fa,0x3fa,-1,1}, 2957 {0x3fd,0x3ff,1,-130}, 2958 {0x400,0x40f,1,80}, 2959 {0x410,0x42f,1,32}, 2960 {0x460,0x480,2,1}, 2961 {0x48a,0x4be,2,1}, 2962 {0x4c0,0x4c0,-1,15}, 2963 {0x4c1,0x4cd,2,1}, 2964 {0x4d0,0x52e,2,1}, 2965 {0x531,0x556,1,48}, 2966 {0x10a0,0x10c5,1,7264}, 2967 {0x10c7,0x10cd,6,7264}, 2968 {0x13a0,0x13ef,1,38864}, 2969 {0x13f0,0x13f5,1,8}, 2970 {0x1e00,0x1e94,2,1}, 2971 {0x1e9e,0x1e9e,-1,-7615}, 2972 {0x1ea0,0x1efe,2,1}, 2973 {0x1f08,0x1f0f,1,-8}, 2974 {0x1f18,0x1f1d,1,-8}, 2975 {0x1f28,0x1f2f,1,-8}, 2976 {0x1f38,0x1f3f,1,-8}, 2977 {0x1f48,0x1f4d,1,-8}, 2978 {0x1f59,0x1f5f,2,-8}, 2979 {0x1f68,0x1f6f,1,-8}, 2980 {0x1f88,0x1f8f,1,-8}, 2981 {0x1f98,0x1f9f,1,-8}, 2982 {0x1fa8,0x1faf,1,-8}, 2983 {0x1fb8,0x1fb9,1,-8}, 2984 {0x1fba,0x1fbb,1,-74}, 2985 {0x1fbc,0x1fbc,-1,-9}, 2986 {0x1fc8,0x1fcb,1,-86}, 2987 {0x1fcc,0x1fcc,-1,-9}, 2988 {0x1fd8,0x1fd9,1,-8}, 2989 {0x1fda,0x1fdb,1,-100}, 2990 {0x1fe8,0x1fe9,1,-8}, 2991 {0x1fea,0x1feb,1,-112}, 2992 {0x1fec,0x1fec,-1,-7}, 2993 {0x1ff8,0x1ff9,1,-128}, 2994 {0x1ffa,0x1ffb,1,-126}, 2995 {0x1ffc,0x1ffc,-1,-9}, 2996 {0x2126,0x2126,-1,-7517}, 2997 {0x212a,0x212a,-1,-8383}, 2998 {0x212b,0x212b,-1,-8262}, 2999 {0x2132,0x2132,-1,28}, 3000 {0x2160,0x216f,1,16}, 3001 {0x2183,0x2183,-1,1}, 3002 {0x24b6,0x24cf,1,26}, 3003 {0x2c00,0x2c2e,1,48}, 3004 {0x2c60,0x2c60,-1,1}, 3005 {0x2c62,0x2c62,-1,-10743}, 3006 {0x2c63,0x2c63,-1,-3814}, 3007 {0x2c64,0x2c64,-1,-10727}, 3008 {0x2c67,0x2c6b,2,1}, 3009 {0x2c6d,0x2c6d,-1,-10780}, 3010 {0x2c6e,0x2c6e,-1,-10749}, 3011 {0x2c6f,0x2c6f,-1,-10783}, 3012 {0x2c70,0x2c70,-1,-10782}, 3013 {0x2c72,0x2c75,3,1}, 3014 {0x2c7e,0x2c7f,1,-10815}, 3015 {0x2c80,0x2ce2,2,1}, 3016 {0x2ceb,0x2ced,2,1}, 3017 {0x2cf2,0xa640,31054,1}, 3018 {0xa642,0xa66c,2,1}, 3019 {0xa680,0xa69a,2,1}, 3020 {0xa722,0xa72e,2,1}, 3021 {0xa732,0xa76e,2,1}, 3022 {0xa779,0xa77b,2,1}, 3023 {0xa77d,0xa77d,-1,-35332}, 3024 {0xa77e,0xa786,2,1}, 3025 {0xa78b,0xa78b,-1,1}, 3026 {0xa78d,0xa78d,-1,-42280}, 3027 {0xa790,0xa792,2,1}, 3028 {0xa796,0xa7a8,2,1}, 3029 {0xa7aa,0xa7aa,-1,-42308}, 3030 {0xa7ab,0xa7ab,-1,-42319}, 3031 {0xa7ac,0xa7ac,-1,-42315}, 3032 {0xa7ad,0xa7ad,-1,-42305}, 3033 {0xa7b0,0xa7b0,-1,-42258}, 3034 {0xa7b1,0xa7b1,-1,-42282}, 3035 {0xa7b2,0xa7b2,-1,-42261}, 3036 {0xa7b3,0xa7b3,-1,928}, 3037 {0xa7b4,0xa7b6,2,1}, 3038 {0xff21,0xff3a,1,32}, 3039 {0x10400,0x10427,1,40}, 3040 {0x10c80,0x10cb2,1,64}, 3041 {0x118a0,0x118bf,1,32} 3042 }; 3043 3044 static convertStruct toUpper[] = 3045 { 3046 {0x61,0x7a,1,-32}, 3047 {0xb5,0xb5,-1,743}, 3048 {0xe0,0xf6,1,-32}, 3049 {0xf8,0xfe,1,-32}, 3050 {0xff,0xff,-1,121}, 3051 {0x101,0x12f,2,-1}, 3052 {0x131,0x131,-1,-232}, 3053 {0x133,0x137,2,-1}, 3054 {0x13a,0x148,2,-1}, 3055 {0x14b,0x177,2,-1}, 3056 {0x17a,0x17e,2,-1}, 3057 {0x17f,0x17f,-1,-300}, 3058 {0x180,0x180,-1,195}, 3059 {0x183,0x185,2,-1}, 3060 {0x188,0x18c,4,-1}, 3061 {0x192,0x192,-1,-1}, 3062 {0x195,0x195,-1,97}, 3063 {0x199,0x199,-1,-1}, 3064 {0x19a,0x19a,-1,163}, 3065 {0x19e,0x19e,-1,130}, 3066 {0x1a1,0x1a5,2,-1}, 3067 {0x1a8,0x1ad,5,-1}, 3068 {0x1b0,0x1b4,4,-1}, 3069 {0x1b6,0x1b9,3,-1}, 3070 {0x1bd,0x1bd,-1,-1}, 3071 {0x1bf,0x1bf,-1,56}, 3072 {0x1c5,0x1c5,-1,-1}, 3073 {0x1c6,0x1c6,-1,-2}, 3074 {0x1c8,0x1c8,-1,-1}, 3075 {0x1c9,0x1c9,-1,-2}, 3076 {0x1cb,0x1cb,-1,-1}, 3077 {0x1cc,0x1cc,-1,-2}, 3078 {0x1ce,0x1dc,2,-1}, 3079 {0x1dd,0x1dd,-1,-79}, 3080 {0x1df,0x1ef,2,-1}, 3081 {0x1f2,0x1f2,-1,-1}, 3082 {0x1f3,0x1f3,-1,-2}, 3083 {0x1f5,0x1f9,4,-1}, 3084 {0x1fb,0x21f,2,-1}, 3085 {0x223,0x233,2,-1}, 3086 {0x23c,0x23c,-1,-1}, 3087 {0x23f,0x240,1,10815}, 3088 {0x242,0x247,5,-1}, 3089 {0x249,0x24f,2,-1}, 3090 {0x250,0x250,-1,10783}, 3091 {0x251,0x251,-1,10780}, 3092 {0x252,0x252,-1,10782}, 3093 {0x253,0x253,-1,-210}, 3094 {0x254,0x254,-1,-206}, 3095 {0x256,0x257,1,-205}, 3096 {0x259,0x259,-1,-202}, 3097 {0x25b,0x25b,-1,-203}, 3098 {0x25c,0x25c,-1,42319}, 3099 {0x260,0x260,-1,-205}, 3100 {0x261,0x261,-1,42315}, 3101 {0x263,0x263,-1,-207}, 3102 {0x265,0x265,-1,42280}, 3103 {0x266,0x266,-1,42308}, 3104 {0x268,0x268,-1,-209}, 3105 {0x269,0x269,-1,-211}, 3106 {0x26b,0x26b,-1,10743}, 3107 {0x26c,0x26c,-1,42305}, 3108 {0x26f,0x26f,-1,-211}, 3109 {0x271,0x271,-1,10749}, 3110 {0x272,0x272,-1,-213}, 3111 {0x275,0x275,-1,-214}, 3112 {0x27d,0x27d,-1,10727}, 3113 {0x280,0x283,3,-218}, 3114 {0x287,0x287,-1,42282}, 3115 {0x288,0x288,-1,-218}, 3116 {0x289,0x289,-1,-69}, 3117 {0x28a,0x28b,1,-217}, 3118 {0x28c,0x28c,-1,-71}, 3119 {0x292,0x292,-1,-219}, 3120 {0x29d,0x29d,-1,42261}, 3121 {0x29e,0x29e,-1,42258}, 3122 {0x345,0x345,-1,84}, 3123 {0x371,0x373,2,-1}, 3124 {0x377,0x377,-1,-1}, 3125 {0x37b,0x37d,1,130}, 3126 {0x3ac,0x3ac,-1,-38}, 3127 {0x3ad,0x3af,1,-37}, 3128 {0x3b1,0x3c1,1,-32}, 3129 {0x3c2,0x3c2,-1,-31}, 3130 {0x3c3,0x3cb,1,-32}, 3131 {0x3cc,0x3cc,-1,-64}, 3132 {0x3cd,0x3ce,1,-63}, 3133 {0x3d0,0x3d0,-1,-62}, 3134 {0x3d1,0x3d1,-1,-57}, 3135 {0x3d5,0x3d5,-1,-47}, 3136 {0x3d6,0x3d6,-1,-54}, 3137 {0x3d7,0x3d7,-1,-8}, 3138 {0x3d9,0x3ef,2,-1}, 3139 {0x3f0,0x3f0,-1,-86}, 3140 {0x3f1,0x3f1,-1,-80}, 3141 {0x3f2,0x3f2,-1,7}, 3142 {0x3f3,0x3f3,-1,-116}, 3143 {0x3f5,0x3f5,-1,-96}, 3144 {0x3f8,0x3fb,3,-1}, 3145 {0x430,0x44f,1,-32}, 3146 {0x450,0x45f,1,-80}, 3147 {0x461,0x481,2,-1}, 3148 {0x48b,0x4bf,2,-1}, 3149 {0x4c2,0x4ce,2,-1}, 3150 {0x4cf,0x4cf,-1,-15}, 3151 {0x4d1,0x52f,2,-1}, 3152 {0x561,0x586,1,-48}, 3153 {0x13f8,0x13fd,1,-8}, 3154 {0x1d79,0x1d79,-1,35332}, 3155 {0x1d7d,0x1d7d,-1,3814}, 3156 {0x1e01,0x1e95,2,-1}, 3157 {0x1e9b,0x1e9b,-1,-59}, 3158 {0x1ea1,0x1eff,2,-1}, 3159 {0x1f00,0x1f07,1,8}, 3160 {0x1f10,0x1f15,1,8}, 3161 {0x1f20,0x1f27,1,8}, 3162 {0x1f30,0x1f37,1,8}, 3163 {0x1f40,0x1f45,1,8}, 3164 {0x1f51,0x1f57,2,8}, 3165 {0x1f60,0x1f67,1,8}, 3166 {0x1f70,0x1f71,1,74}, 3167 {0x1f72,0x1f75,1,86}, 3168 {0x1f76,0x1f77,1,100}, 3169 {0x1f78,0x1f79,1,128}, 3170 {0x1f7a,0x1f7b,1,112}, 3171 {0x1f7c,0x1f7d,1,126}, 3172 {0x1f80,0x1f87,1,8}, 3173 {0x1f90,0x1f97,1,8}, 3174 {0x1fa0,0x1fa7,1,8}, 3175 {0x1fb0,0x1fb1,1,8}, 3176 {0x1fb3,0x1fb3,-1,9}, 3177 {0x1fbe,0x1fbe,-1,-7205}, 3178 {0x1fc3,0x1fc3,-1,9}, 3179 {0x1fd0,0x1fd1,1,8}, 3180 {0x1fe0,0x1fe1,1,8}, 3181 {0x1fe5,0x1fe5,-1,7}, 3182 {0x1ff3,0x1ff3,-1,9}, 3183 {0x214e,0x214e,-1,-28}, 3184 {0x2170,0x217f,1,-16}, 3185 {0x2184,0x2184,-1,-1}, 3186 {0x24d0,0x24e9,1,-26}, 3187 {0x2c30,0x2c5e,1,-48}, 3188 {0x2c61,0x2c61,-1,-1}, 3189 {0x2c65,0x2c65,-1,-10795}, 3190 {0x2c66,0x2c66,-1,-10792}, 3191 {0x2c68,0x2c6c,2,-1}, 3192 {0x2c73,0x2c76,3,-1}, 3193 {0x2c81,0x2ce3,2,-1}, 3194 {0x2cec,0x2cee,2,-1}, 3195 {0x2cf3,0x2cf3,-1,-1}, 3196 {0x2d00,0x2d25,1,-7264}, 3197 {0x2d27,0x2d2d,6,-7264}, 3198 {0xa641,0xa66d,2,-1}, 3199 {0xa681,0xa69b,2,-1}, 3200 {0xa723,0xa72f,2,-1}, 3201 {0xa733,0xa76f,2,-1}, 3202 {0xa77a,0xa77c,2,-1}, 3203 {0xa77f,0xa787,2,-1}, 3204 {0xa78c,0xa791,5,-1}, 3205 {0xa793,0xa797,4,-1}, 3206 {0xa799,0xa7a9,2,-1}, 3207 {0xa7b5,0xa7b7,2,-1}, 3208 {0xab53,0xab53,-1,-928}, 3209 {0xab70,0xabbf,1,-38864}, 3210 {0xff41,0xff5a,1,-32}, 3211 {0x10428,0x1044f,1,-40}, 3212 {0x10cc0,0x10cf2,1,-64}, 3213 {0x118c0,0x118df,1,-32} 3214 }; 3215 /* 3216 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3217 * simple case folding. 3218 */ 3219 int 3220 utf_toupper(a) 3221 int a; 3222 { 3223 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */ 3224 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3225 return TOUPPER_ASC(a); 3226 3227 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3228 /* If towupper() is available and handles Unicode, use it. */ 3229 if (!(cmp_flags & CMP_INTERNAL)) 3230 return towupper(a); 3231 #endif 3232 3233 /* For characters below 128 use locale sensitive toupper(). */ 3234 if (a < 128) 3235 return TOUPPER_LOC(a); 3236 3237 /* For any other characters use the above mapping table. */ 3238 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3239 } 3240 3241 int 3242 utf_islower(a) 3243 int a; 3244 { 3245 /* German sharp s is lower case but has no upper case equivalent. */ 3246 return (utf_toupper(a) != a) || a == 0xdf; 3247 } 3248 3249 /* 3250 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3251 * simple case folding. 3252 */ 3253 int 3254 utf_tolower(a) 3255 int a; 3256 { 3257 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */ 3258 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3259 return TOLOWER_ASC(a); 3260 3261 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3262 /* If towlower() is available and handles Unicode, use it. */ 3263 if (!(cmp_flags & CMP_INTERNAL)) 3264 return towlower(a); 3265 #endif 3266 3267 /* For characters below 128 use locale sensitive tolower(). */ 3268 if (a < 128) 3269 return TOLOWER_LOC(a); 3270 3271 /* For any other characters use the above mapping table. */ 3272 return utf_convert(a, toLower, (int)sizeof(toLower)); 3273 } 3274 3275 int 3276 utf_isupper(a) 3277 int a; 3278 { 3279 return (utf_tolower(a) != a); 3280 } 3281 3282 static int 3283 utf_strnicmp(s1, s2, n1, n2) 3284 char_u *s1, *s2; 3285 size_t n1, n2; 3286 { 3287 int c1, c2, cdiff; 3288 char_u buffer[6]; 3289 3290 for (;;) 3291 { 3292 c1 = utf_safe_read_char_adv(&s1, &n1); 3293 c2 = utf_safe_read_char_adv(&s2, &n2); 3294 3295 if (c1 <= 0 || c2 <= 0) 3296 break; 3297 3298 if (c1 == c2) 3299 continue; 3300 3301 cdiff = utf_fold(c1) - utf_fold(c2); 3302 if (cdiff != 0) 3303 return cdiff; 3304 } 3305 3306 /* some string ended or has an incomplete/illegal character sequence */ 3307 3308 if (c1 == 0 || c2 == 0) 3309 { 3310 /* some string ended. shorter string is smaller */ 3311 if (c1 == 0 && c2 == 0) 3312 return 0; 3313 return c1 == 0 ? -1 : 1; 3314 } 3315 3316 /* Continue with bytewise comparison to produce some result that 3317 * would make comparison operations involving this function transitive. 3318 * 3319 * If only one string had an error, comparison should be made with 3320 * folded version of the other string. In this case it is enough 3321 * to fold just one character to determine the result of comparison. */ 3322 3323 if (c1 != -1 && c2 == -1) 3324 { 3325 n1 = utf_char2bytes(utf_fold(c1), buffer); 3326 s1 = buffer; 3327 } 3328 else if (c2 != -1 && c1 == -1) 3329 { 3330 n2 = utf_char2bytes(utf_fold(c2), buffer); 3331 s2 = buffer; 3332 } 3333 3334 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3335 { 3336 cdiff = (int)(*s1) - (int)(*s2); 3337 if (cdiff != 0) 3338 return cdiff; 3339 3340 s1++; 3341 s2++; 3342 n1--; 3343 n2--; 3344 } 3345 3346 if (n1 > 0 && *s1 == NUL) 3347 n1 = 0; 3348 if (n2 > 0 && *s2 == NUL) 3349 n2 = 0; 3350 3351 if (n1 == 0 && n2 == 0) 3352 return 0; 3353 return n1 == 0 ? -1 : 1; 3354 } 3355 3356 /* 3357 * Version of strnicmp() that handles multi-byte characters. 3358 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3359 * probably use strnicmp(), because there are no ASCII characters in the 3360 * second byte. 3361 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3362 * two characters otherwise. 3363 */ 3364 int 3365 mb_strnicmp(s1, s2, nn) 3366 char_u *s1, *s2; 3367 size_t nn; 3368 { 3369 int i, l; 3370 int cdiff; 3371 int n = (int)nn; 3372 3373 if (enc_utf8) 3374 { 3375 return utf_strnicmp(s1, s2, nn, nn); 3376 } 3377 else 3378 { 3379 for (i = 0; i < n; i += l) 3380 { 3381 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */ 3382 return 0; 3383 3384 l = (*mb_ptr2len)(s1 + i); 3385 if (l <= 1) 3386 { 3387 /* Single byte: first check normally, then with ignore case. */ 3388 if (s1[i] != s2[i]) 3389 { 3390 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3391 if (cdiff != 0) 3392 return cdiff; 3393 } 3394 } 3395 else 3396 { 3397 /* For non-Unicode multi-byte don't ignore case. */ 3398 if (l > n - i) 3399 l = n - i; 3400 cdiff = STRNCMP(s1 + i, s2 + i, l); 3401 if (cdiff != 0) 3402 return cdiff; 3403 } 3404 } 3405 } 3406 return 0; 3407 } 3408 3409 /* 3410 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3411 * 'encoding' has been set to. 3412 */ 3413 void 3414 show_utf8() 3415 { 3416 int len; 3417 int rlen = 0; 3418 char_u *line; 3419 int clen; 3420 int i; 3421 3422 /* Get the byte length of the char under the cursor, including composing 3423 * characters. */ 3424 line = ml_get_cursor(); 3425 len = utfc_ptr2len(line); 3426 if (len == 0) 3427 { 3428 MSG("NUL"); 3429 return; 3430 } 3431 3432 clen = 0; 3433 for (i = 0; i < len; ++i) 3434 { 3435 if (clen == 0) 3436 { 3437 /* start of (composing) character, get its length */ 3438 if (i > 0) 3439 { 3440 STRCPY(IObuff + rlen, "+ "); 3441 rlen += 2; 3442 } 3443 clen = utf_ptr2len(line + i); 3444 } 3445 sprintf((char *)IObuff + rlen, "%02x ", 3446 (line[i] == NL) ? NUL : line[i]); /* NUL is stored as NL */ 3447 --clen; 3448 rlen += (int)STRLEN(IObuff + rlen); 3449 if (rlen > IOSIZE - 20) 3450 break; 3451 } 3452 3453 msg(IObuff); 3454 } 3455 3456 /* 3457 * mb_head_off() function pointer. 3458 * Return offset from "p" to the first byte of the character it points into. 3459 * If "p" points to the NUL at the end of the string return 0. 3460 * Returns 0 when already at the first byte of a character. 3461 */ 3462 int 3463 latin_head_off(base, p) 3464 char_u *base UNUSED; 3465 char_u *p UNUSED; 3466 { 3467 return 0; 3468 } 3469 3470 int 3471 dbcs_head_off(base, p) 3472 char_u *base; 3473 char_u *p; 3474 { 3475 char_u *q; 3476 3477 /* It can't be a trailing byte when not using DBCS, at the start of the 3478 * string or the previous byte can't start a double-byte. */ 3479 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3480 return 0; 3481 3482 /* This is slow: need to start at the base and go forward until the 3483 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */ 3484 q = base; 3485 while (q < p) 3486 q += dbcs_ptr2len(q); 3487 return (q == p) ? 0 : 1; 3488 } 3489 3490 /* 3491 * Special version of dbcs_head_off() that works for ScreenLines[], where 3492 * single-width DBCS_JPNU characters are stored separately. 3493 */ 3494 int 3495 dbcs_screen_head_off(base, p) 3496 char_u *base; 3497 char_u *p; 3498 { 3499 char_u *q; 3500 3501 /* It can't be a trailing byte when not using DBCS, at the start of the 3502 * string or the previous byte can't start a double-byte. 3503 * For euc-jp an 0x8e byte in the previous cell always means we have a 3504 * lead byte in the current cell. */ 3505 if (p <= base 3506 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3507 || MB_BYTE2LEN(p[-1]) == 1 3508 || *p == NUL) 3509 return 0; 3510 3511 /* This is slow: need to start at the base and go forward until the 3512 * byte we are looking for. Return 1 when we went past it, 0 otherwise. 3513 * For DBCS_JPNU look out for 0x8e, which means the second byte is not 3514 * stored as the next byte. */ 3515 q = base; 3516 while (q < p) 3517 { 3518 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3519 ++q; 3520 else 3521 q += dbcs_ptr2len(q); 3522 } 3523 return (q == p) ? 0 : 1; 3524 } 3525 3526 int 3527 utf_head_off(base, p) 3528 char_u *base; 3529 char_u *p; 3530 { 3531 char_u *q; 3532 char_u *s; 3533 int c; 3534 int len; 3535 #ifdef FEAT_ARABIC 3536 char_u *j; 3537 #endif 3538 3539 if (*p < 0x80) /* be quick for ASCII */ 3540 return 0; 3541 3542 /* Skip backwards over trailing bytes: 10xx.xxxx 3543 * Skip backwards again if on a composing char. */ 3544 for (q = p; ; --q) 3545 { 3546 /* Move s to the last byte of this char. */ 3547 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3548 ; 3549 /* Move q to the first byte of this char. */ 3550 while (q > base && (*q & 0xc0) == 0x80) 3551 --q; 3552 /* Check for illegal sequence. Do allow an illegal byte after where we 3553 * started. */ 3554 len = utf8len_tab[*q]; 3555 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3556 return 0; 3557 3558 if (q <= base) 3559 break; 3560 3561 c = utf_ptr2char(q); 3562 if (utf_iscomposing(c)) 3563 continue; 3564 3565 #ifdef FEAT_ARABIC 3566 if (arabic_maycombine(c)) 3567 { 3568 /* Advance to get a sneak-peak at the next char */ 3569 j = q; 3570 --j; 3571 /* Move j to the first byte of this char. */ 3572 while (j > base && (*j & 0xc0) == 0x80) 3573 --j; 3574 if (arabic_combine(utf_ptr2char(j), c)) 3575 continue; 3576 } 3577 #endif 3578 break; 3579 } 3580 3581 return (int)(p - q); 3582 } 3583 3584 /* 3585 * Copy a character from "*fp" to "*tp" and advance the pointers. 3586 */ 3587 void 3588 mb_copy_char(fp, tp) 3589 char_u **fp; 3590 char_u **tp; 3591 { 3592 int l = (*mb_ptr2len)(*fp); 3593 3594 mch_memmove(*tp, *fp, (size_t)l); 3595 *tp += l; 3596 *fp += l; 3597 } 3598 3599 /* 3600 * Return the offset from "p" to the first byte of a character. When "p" is 3601 * at the start of a character 0 is returned, otherwise the offset to the next 3602 * character. Can start anywhere in a stream of bytes. 3603 */ 3604 int 3605 mb_off_next(base, p) 3606 char_u *base; 3607 char_u *p; 3608 { 3609 int i; 3610 int j; 3611 3612 if (enc_utf8) 3613 { 3614 if (*p < 0x80) /* be quick for ASCII */ 3615 return 0; 3616 3617 /* Find the next character that isn't 10xx.xxxx */ 3618 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 3619 ; 3620 if (i > 0) 3621 { 3622 /* Check for illegal sequence. */ 3623 for (j = 0; p - j > base; ++j) 3624 if ((p[-j] & 0xc0) != 0x80) 3625 break; 3626 if (utf8len_tab[p[-j]] != i + j) 3627 return 0; 3628 } 3629 return i; 3630 } 3631 3632 /* Only need to check if we're on a trail byte, it doesn't matter if we 3633 * want the offset to the next or current character. */ 3634 return (*mb_head_off)(base, p); 3635 } 3636 3637 /* 3638 * Return the offset from "p" to the last byte of the character it points 3639 * into. Can start anywhere in a stream of bytes. 3640 */ 3641 int 3642 mb_tail_off(base, p) 3643 char_u *base; 3644 char_u *p; 3645 { 3646 int i; 3647 int j; 3648 3649 if (*p == NUL) 3650 return 0; 3651 3652 if (enc_utf8) 3653 { 3654 /* Find the last character that is 10xx.xxxx */ 3655 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 3656 ; 3657 /* Check for illegal sequence. */ 3658 for (j = 0; p - j > base; ++j) 3659 if ((p[-j] & 0xc0) != 0x80) 3660 break; 3661 if (utf8len_tab[p[-j]] != i + j + 1) 3662 return 0; 3663 return i; 3664 } 3665 3666 /* It can't be the first byte if a double-byte when not using DBCS, at the 3667 * end of the string or the byte can't start a double-byte. */ 3668 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 3669 return 0; 3670 3671 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3672 return 1 - dbcs_head_off(base, p); 3673 } 3674 3675 /* 3676 * Find the next illegal byte sequence. 3677 */ 3678 void 3679 utf_find_illegal() 3680 { 3681 pos_T pos = curwin->w_cursor; 3682 char_u *p; 3683 int len; 3684 vimconv_T vimconv; 3685 char_u *tofree = NULL; 3686 3687 vimconv.vc_type = CONV_NONE; 3688 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 3689 { 3690 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 3691 * possibly a utf-8 file with illegal bytes. Setup for conversion 3692 * from utf-8 to 'fileencoding'. */ 3693 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 3694 } 3695 3696 #ifdef FEAT_VIRTUALEDIT 3697 curwin->w_cursor.coladd = 0; 3698 #endif 3699 for (;;) 3700 { 3701 p = ml_get_cursor(); 3702 if (vimconv.vc_type != CONV_NONE) 3703 { 3704 vim_free(tofree); 3705 tofree = string_convert(&vimconv, p, NULL); 3706 if (tofree == NULL) 3707 break; 3708 p = tofree; 3709 } 3710 3711 while (*p != NUL) 3712 { 3713 /* Illegal means that there are not enough trail bytes (checked by 3714 * utf_ptr2len()) or too many of them (overlong sequence). */ 3715 len = utf_ptr2len(p); 3716 if (*p >= 0x80 && (len == 1 3717 || utf_char2len(utf_ptr2char(p)) != len)) 3718 { 3719 if (vimconv.vc_type == CONV_NONE) 3720 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 3721 else 3722 { 3723 int l; 3724 3725 len = (int)(p - tofree); 3726 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 3727 { 3728 l = utf_ptr2len(p); 3729 curwin->w_cursor.col += l; 3730 } 3731 } 3732 goto theend; 3733 } 3734 p += len; 3735 } 3736 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 3737 break; 3738 ++curwin->w_cursor.lnum; 3739 curwin->w_cursor.col = 0; 3740 } 3741 3742 /* didn't find it: don't move and beep */ 3743 curwin->w_cursor = pos; 3744 beep_flush(); 3745 3746 theend: 3747 vim_free(tofree); 3748 convert_setup(&vimconv, NULL, NULL); 3749 } 3750 3751 #if defined(FEAT_GUI_GTK) || defined(PROTO) 3752 /* 3753 * Return TRUE if string "s" is a valid utf-8 string. 3754 * When "end" is NULL stop at the first NUL. 3755 * When "end" is positive stop there. 3756 */ 3757 int 3758 utf_valid_string(s, end) 3759 char_u *s; 3760 char_u *end; 3761 { 3762 int l; 3763 char_u *p = s; 3764 3765 while (end == NULL ? *p != NUL : p < end) 3766 { 3767 l = utf8len_tab_zero[*p]; 3768 if (l == 0) 3769 return FALSE; /* invalid lead byte */ 3770 if (end != NULL && p + l > end) 3771 return FALSE; /* incomplete byte sequence */ 3772 ++p; 3773 while (--l > 0) 3774 if ((*p++ & 0xc0) != 0x80) 3775 return FALSE; /* invalid trail byte */ 3776 } 3777 return TRUE; 3778 } 3779 #endif 3780 3781 #if defined(FEAT_GUI) || defined(PROTO) 3782 /* 3783 * Special version of mb_tail_off() for use in ScreenLines[]. 3784 */ 3785 int 3786 dbcs_screen_tail_off(base, p) 3787 char_u *base; 3788 char_u *p; 3789 { 3790 /* It can't be the first byte if a double-byte when not using DBCS, at the 3791 * end of the string or the byte can't start a double-byte. 3792 * For euc-jp an 0x8e byte always means we have a lead byte in the current 3793 * cell. */ 3794 if (*p == NUL || p[1] == NUL 3795 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 3796 || MB_BYTE2LEN(*p) == 1) 3797 return 0; 3798 3799 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3800 return 1 - dbcs_screen_head_off(base, p); 3801 } 3802 #endif 3803 3804 /* 3805 * If the cursor moves on an trail byte, set the cursor on the lead byte. 3806 * Thus it moves left if necessary. 3807 * Return TRUE when the cursor was adjusted. 3808 */ 3809 void 3810 mb_adjust_cursor() 3811 { 3812 mb_adjustpos(curbuf, &curwin->w_cursor); 3813 } 3814 3815 /* 3816 * Adjust position "*lp" to point to the first byte of a multi-byte character. 3817 * If it points to a tail byte it's moved backwards to the head byte. 3818 */ 3819 void 3820 mb_adjustpos(buf, lp) 3821 buf_T *buf; 3822 pos_T *lp; 3823 { 3824 char_u *p; 3825 3826 if (lp->col > 0 3827 #ifdef FEAT_VIRTUALEDIT 3828 || lp->coladd > 1 3829 #endif 3830 ) 3831 { 3832 p = ml_get_buf(buf, lp->lnum, FALSE); 3833 lp->col -= (*mb_head_off)(p, p + lp->col); 3834 #ifdef FEAT_VIRTUALEDIT 3835 /* Reset "coladd" when the cursor would be on the right half of a 3836 * double-wide character. */ 3837 if (lp->coladd == 1 3838 && p[lp->col] != TAB 3839 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 3840 && ptr2cells(p + lp->col) > 1) 3841 lp->coladd = 0; 3842 #endif 3843 } 3844 } 3845 3846 /* 3847 * Return a pointer to the character before "*p", if there is one. 3848 */ 3849 char_u * 3850 mb_prevptr(line, p) 3851 char_u *line; /* start of the string */ 3852 char_u *p; 3853 { 3854 if (p > line) 3855 mb_ptr_back(line, p); 3856 return p; 3857 } 3858 3859 /* 3860 * Return the character length of "str". Each multi-byte character (with 3861 * following composing characters) counts as one. 3862 */ 3863 int 3864 mb_charlen(str) 3865 char_u *str; 3866 { 3867 char_u *p = str; 3868 int count; 3869 3870 if (p == NULL) 3871 return 0; 3872 3873 for (count = 0; *p != NUL; count++) 3874 p += (*mb_ptr2len)(p); 3875 3876 return count; 3877 } 3878 3879 #if defined(FEAT_SPELL) || defined(PROTO) 3880 /* 3881 * Like mb_charlen() but for a string with specified length. 3882 */ 3883 int 3884 mb_charlen_len(str, len) 3885 char_u *str; 3886 int len; 3887 { 3888 char_u *p = str; 3889 int count; 3890 3891 for (count = 0; *p != NUL && p < str + len; count++) 3892 p += (*mb_ptr2len)(p); 3893 3894 return count; 3895 } 3896 #endif 3897 3898 /* 3899 * Try to un-escape a multi-byte character. 3900 * Used for the "to" and "from" part of a mapping. 3901 * Return the un-escaped string if it is a multi-byte character, and advance 3902 * "pp" to just after the bytes that formed it. 3903 * Return NULL if no multi-byte char was found. 3904 */ 3905 char_u * 3906 mb_unescape(pp) 3907 char_u **pp; 3908 { 3909 static char_u buf[6]; 3910 int n; 3911 int m = 0; 3912 char_u *str = *pp; 3913 3914 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 3915 * KS_EXTRA KE_CSI to CSI. 3916 * Maximum length of a utf-8 character is 4 bytes. */ 3917 for (n = 0; str[n] != NUL && m < 4; ++n) 3918 { 3919 if (str[n] == K_SPECIAL 3920 && str[n + 1] == KS_SPECIAL 3921 && str[n + 2] == KE_FILLER) 3922 { 3923 buf[m++] = K_SPECIAL; 3924 n += 2; 3925 } 3926 else if ((str[n] == K_SPECIAL 3927 # ifdef FEAT_GUI 3928 || str[n] == CSI 3929 # endif 3930 ) 3931 && str[n + 1] == KS_EXTRA 3932 && str[n + 2] == (int)KE_CSI) 3933 { 3934 buf[m++] = CSI; 3935 n += 2; 3936 } 3937 else if (str[n] == K_SPECIAL 3938 # ifdef FEAT_GUI 3939 || str[n] == CSI 3940 # endif 3941 ) 3942 break; /* a special key can't be a multibyte char */ 3943 else 3944 buf[m++] = str[n]; 3945 buf[m] = NUL; 3946 3947 /* Return a multi-byte character if it's found. An illegal sequence 3948 * will result in a 1 here. */ 3949 if ((*mb_ptr2len)(buf) > 1) 3950 { 3951 *pp = str + n + 1; 3952 return buf; 3953 } 3954 3955 /* Bail out quickly for ASCII. */ 3956 if (buf[0] < 128) 3957 break; 3958 } 3959 return NULL; 3960 } 3961 3962 /* 3963 * Return TRUE if the character at "row"/"col" on the screen is the left side 3964 * of a double-width character. 3965 * Caller must make sure "row" and "col" are not invalid! 3966 */ 3967 int 3968 mb_lefthalve(row, col) 3969 int row; 3970 int col; 3971 { 3972 #ifdef FEAT_HANGULIN 3973 if (composing_hangul) 3974 return TRUE; 3975 #endif 3976 return (*mb_off2cells)(LineOffset[row] + col, 3977 LineOffset[row] + screen_Columns) > 1; 3978 } 3979 3980 /* 3981 * Correct a position on the screen, if it's the right half of a double-wide 3982 * char move it to the left half. Returns the corrected column. 3983 */ 3984 int 3985 mb_fix_col(col, row) 3986 int col; 3987 int row; 3988 { 3989 col = check_col(col); 3990 row = check_row(row); 3991 if (has_mbyte && ScreenLines != NULL && col > 0 3992 && ((enc_dbcs 3993 && ScreenLines[LineOffset[row] + col] != NUL 3994 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 3995 ScreenLines + LineOffset[row] + col)) 3996 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0))) 3997 return col - 1; 3998 return col; 3999 } 4000 #endif 4001 4002 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 4003 static int enc_alias_search __ARGS((char_u *name)); 4004 4005 /* 4006 * Skip the Vim specific head of a 'encoding' name. 4007 */ 4008 char_u * 4009 enc_skip(p) 4010 char_u *p; 4011 { 4012 if (STRNCMP(p, "2byte-", 6) == 0) 4013 return p + 6; 4014 if (STRNCMP(p, "8bit-", 5) == 0) 4015 return p + 5; 4016 return p; 4017 } 4018 4019 /* 4020 * Find the canonical name for encoding "enc". 4021 * When the name isn't recognized, returns "enc" itself, but with all lower 4022 * case characters and '_' replaced with '-'. 4023 * Returns an allocated string. NULL for out-of-memory. 4024 */ 4025 char_u * 4026 enc_canonize(enc) 4027 char_u *enc; 4028 { 4029 char_u *r; 4030 char_u *p, *s; 4031 int i; 4032 4033 # ifdef FEAT_MBYTE 4034 if (STRCMP(enc, "default") == 0) 4035 { 4036 /* Use the default encoding as it's found by set_init_1(). */ 4037 r = get_encoding_default(); 4038 if (r == NULL) 4039 r = (char_u *)"latin1"; 4040 return vim_strsave(r); 4041 } 4042 # endif 4043 4044 /* copy "enc" to allocated memory, with room for two '-' */ 4045 r = alloc((unsigned)(STRLEN(enc) + 3)); 4046 if (r != NULL) 4047 { 4048 /* Make it all lower case and replace '_' with '-'. */ 4049 p = r; 4050 for (s = enc; *s != NUL; ++s) 4051 { 4052 if (*s == '_') 4053 *p++ = '-'; 4054 else 4055 *p++ = TOLOWER_ASC(*s); 4056 } 4057 *p = NUL; 4058 4059 /* Skip "2byte-" and "8bit-". */ 4060 p = enc_skip(r); 4061 4062 /* Change "microsoft-cp" to "cp". Used in some spell files. */ 4063 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4064 STRMOVE(p, p + 10); 4065 4066 /* "iso8859" -> "iso-8859" */ 4067 if (STRNCMP(p, "iso8859", 7) == 0) 4068 { 4069 STRMOVE(p + 4, p + 3); 4070 p[3] = '-'; 4071 } 4072 4073 /* "iso-8859n" -> "iso-8859-n" */ 4074 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4075 { 4076 STRMOVE(p + 9, p + 8); 4077 p[8] = '-'; 4078 } 4079 4080 /* "latin-N" -> "latinN" */ 4081 if (STRNCMP(p, "latin-", 6) == 0) 4082 STRMOVE(p + 5, p + 6); 4083 4084 if (enc_canon_search(p) >= 0) 4085 { 4086 /* canonical name can be used unmodified */ 4087 if (p != r) 4088 STRMOVE(r, p); 4089 } 4090 else if ((i = enc_alias_search(p)) >= 0) 4091 { 4092 /* alias recognized, get canonical name */ 4093 vim_free(r); 4094 r = vim_strsave((char_u *)enc_canon_table[i].name); 4095 } 4096 } 4097 return r; 4098 } 4099 4100 /* 4101 * Search for an encoding alias of "name". 4102 * Returns -1 when not found. 4103 */ 4104 static int 4105 enc_alias_search(name) 4106 char_u *name; 4107 { 4108 int i; 4109 4110 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4111 if (STRCMP(name, enc_alias_table[i].name) == 0) 4112 return enc_alias_table[i].canon; 4113 return -1; 4114 } 4115 #endif 4116 4117 #if defined(FEAT_MBYTE) || defined(PROTO) 4118 4119 #ifdef HAVE_LANGINFO_H 4120 # include <langinfo.h> 4121 #endif 4122 4123 /* 4124 * Get the canonicalized encoding of the current locale. 4125 * Returns an allocated string when successful, NULL when not. 4126 */ 4127 char_u * 4128 enc_locale() 4129 { 4130 #ifndef WIN3264 4131 char *s; 4132 char *p; 4133 int i; 4134 #endif 4135 char buf[50]; 4136 #ifdef WIN3264 4137 long acp = GetACP(); 4138 4139 if (acp == 1200) 4140 STRCPY(buf, "ucs-2le"); 4141 else if (acp == 1252) /* cp1252 is used as latin1 */ 4142 STRCPY(buf, "latin1"); 4143 else 4144 sprintf(buf, "cp%ld", acp); 4145 #else 4146 # ifdef HAVE_NL_LANGINFO_CODESET 4147 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4148 # endif 4149 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4150 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4151 # endif 4152 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4153 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4154 s = getenv("LANG"); 4155 4156 if (s == NULL || *s == NUL) 4157 return FAIL; 4158 4159 /* The most generic locale format is: 4160 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4161 * If there is a '.' remove the part before it. 4162 * if there is something after the codeset, remove it. 4163 * Make the name lowercase and replace '_' with '-'. 4164 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4165 * "ko_KR.EUC" == "euc-kr" 4166 */ 4167 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4168 { 4169 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4170 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4171 { 4172 /* copy "XY.EUC" to "euc-XY" to buf[10] */ 4173 STRCPY(buf + 10, "euc-"); 4174 buf[14] = p[-2]; 4175 buf[15] = p[-1]; 4176 buf[16] = 0; 4177 s = buf + 10; 4178 } 4179 else 4180 s = p + 1; 4181 } 4182 for (i = 0; s[i] != NUL && i < (int)sizeof(buf) - 1; ++i) 4183 { 4184 if (s[i] == '_' || s[i] == '-') 4185 buf[i] = '-'; 4186 else if (isalnum((int)s[i])) 4187 buf[i] = TOLOWER_ASC(s[i]); 4188 else 4189 break; 4190 } 4191 buf[i] = NUL; 4192 #endif 4193 4194 return enc_canonize((char_u *)buf); 4195 } 4196 4197 #if defined(WIN3264) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4198 /* 4199 * Convert an encoding name to an MS-Windows codepage. 4200 * Returns zero if no codepage can be figured out. 4201 */ 4202 int 4203 encname2codepage(name) 4204 char_u *name; 4205 { 4206 int cp; 4207 char_u *p = name; 4208 int idx; 4209 4210 if (STRNCMP(p, "8bit-", 5) == 0) 4211 p += 5; 4212 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4213 p += 6; 4214 4215 if (p[0] == 'c' && p[1] == 'p') 4216 cp = atoi((char *)p + 2); 4217 else if ((idx = enc_canon_search(p)) >= 0) 4218 cp = enc_canon_table[idx].codepage; 4219 else 4220 return 0; 4221 if (IsValidCodePage(cp)) 4222 return cp; 4223 return 0; 4224 } 4225 #endif 4226 4227 # if defined(USE_ICONV) || defined(PROTO) 4228 4229 static char_u *iconv_string __ARGS((vimconv_T *vcp, char_u *str, int slen, int *unconvlenp, int *resultlenp)); 4230 4231 /* 4232 * Call iconv_open() with a check if iconv() works properly (there are broken 4233 * versions). 4234 * Returns (void *)-1 if failed. 4235 * (should return iconv_t, but that causes problems with prototypes). 4236 */ 4237 void * 4238 my_iconv_open(to, from) 4239 char_u *to; 4240 char_u *from; 4241 { 4242 iconv_t fd; 4243 #define ICONV_TESTLEN 400 4244 char_u tobuf[ICONV_TESTLEN]; 4245 char *p; 4246 size_t tolen; 4247 static int iconv_ok = -1; 4248 4249 if (iconv_ok == FALSE) 4250 return (void *)-1; /* detected a broken iconv() previously */ 4251 4252 #ifdef DYNAMIC_ICONV 4253 /* Check if the iconv.dll can be found. */ 4254 if (!iconv_enabled(TRUE)) 4255 return (void *)-1; 4256 #endif 4257 4258 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4259 4260 if (fd != (iconv_t)-1 && iconv_ok == -1) 4261 { 4262 /* 4263 * Do a dummy iconv() call to check if it actually works. There is a 4264 * version of iconv() on Linux that is broken. We can't ignore it, 4265 * because it's wide-spread. The symptoms are that after outputting 4266 * the initial shift state the "to" pointer is NULL and conversion 4267 * stops for no apparent reason after about 8160 characters. 4268 */ 4269 p = (char *)tobuf; 4270 tolen = ICONV_TESTLEN; 4271 (void)iconv(fd, NULL, NULL, &p, &tolen); 4272 if (p == NULL) 4273 { 4274 iconv_ok = FALSE; 4275 iconv_close(fd); 4276 fd = (iconv_t)-1; 4277 } 4278 else 4279 iconv_ok = TRUE; 4280 } 4281 4282 return (void *)fd; 4283 } 4284 4285 /* 4286 * Convert the string "str[slen]" with iconv(). 4287 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4288 * sequence and set "*unconvlenp" to the length of it. 4289 * Returns the converted string in allocated memory. NULL for an error. 4290 * If resultlenp is not NULL, sets it to the result length in bytes. 4291 */ 4292 static char_u * 4293 iconv_string(vcp, str, slen, unconvlenp, resultlenp) 4294 vimconv_T *vcp; 4295 char_u *str; 4296 int slen; 4297 int *unconvlenp; 4298 int *resultlenp; 4299 { 4300 const char *from; 4301 size_t fromlen; 4302 char *to; 4303 size_t tolen; 4304 size_t len = 0; 4305 size_t done = 0; 4306 char_u *result = NULL; 4307 char_u *p; 4308 int l; 4309 4310 from = (char *)str; 4311 fromlen = slen; 4312 for (;;) 4313 { 4314 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4315 { 4316 /* Allocate enough room for most conversions. When re-allocating 4317 * increase the buffer size. */ 4318 len = len + fromlen * 2 + 40; 4319 p = alloc((unsigned)len); 4320 if (p != NULL && done > 0) 4321 mch_memmove(p, result, done); 4322 vim_free(result); 4323 result = p; 4324 if (result == NULL) /* out of memory */ 4325 break; 4326 } 4327 4328 to = (char *)result + done; 4329 tolen = len - done - 2; 4330 /* Avoid a warning for systems with a wrong iconv() prototype by 4331 * casting the second argument to void *. */ 4332 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4333 != (size_t)-1) 4334 { 4335 /* Finished, append a NUL. */ 4336 *to = NUL; 4337 break; 4338 } 4339 4340 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4341 * iconv library may use one of them. */ 4342 if (!vcp->vc_fail && unconvlenp != NULL 4343 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4344 { 4345 /* Handle an incomplete sequence at the end. */ 4346 *to = NUL; 4347 *unconvlenp = (int)fromlen; 4348 break; 4349 } 4350 4351 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4352 * iconv library may use one of them. */ 4353 else if (!vcp->vc_fail 4354 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4355 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4356 { 4357 /* Can't convert: insert a '?' and skip a character. This assumes 4358 * conversion from 'encoding' to something else. In other 4359 * situations we don't know what to skip anyway. */ 4360 *to++ = '?'; 4361 if ((*mb_ptr2cells)((char_u *)from) > 1) 4362 *to++ = '?'; 4363 if (enc_utf8) 4364 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4365 else 4366 { 4367 l = (*mb_ptr2len)((char_u *)from); 4368 if (l > (int)fromlen) 4369 l = (int)fromlen; 4370 } 4371 from += l; 4372 fromlen -= l; 4373 } 4374 else if (ICONV_ERRNO != ICONV_E2BIG) 4375 { 4376 /* conversion failed */ 4377 vim_free(result); 4378 result = NULL; 4379 break; 4380 } 4381 /* Not enough room or skipping illegal sequence. */ 4382 done = to - (char *)result; 4383 } 4384 4385 if (resultlenp != NULL && result != NULL) 4386 *resultlenp = (int)(to - (char *)result); 4387 return result; 4388 } 4389 4390 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4391 /* 4392 * Dynamically load the "iconv.dll" on Win32. 4393 */ 4394 4395 #ifndef DYNAMIC_ICONV /* just generating prototypes */ 4396 # define HINSTANCE int 4397 #endif 4398 static HINSTANCE hIconvDLL = 0; 4399 static HINSTANCE hMsvcrtDLL = 0; 4400 4401 # ifndef DYNAMIC_ICONV_DLL 4402 # define DYNAMIC_ICONV_DLL "iconv.dll" 4403 # define DYNAMIC_ICONV_DLL_ALT "libiconv.dll" 4404 # endif 4405 # ifndef DYNAMIC_MSVCRT_DLL 4406 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4407 # endif 4408 4409 /* 4410 * Get the address of 'funcname' which is imported by 'hInst' DLL. 4411 */ 4412 static void * 4413 get_iconv_import_func(HINSTANCE hInst, const char *funcname) 4414 { 4415 PBYTE pImage = (PBYTE)hInst; 4416 PIMAGE_DOS_HEADER pDOS = (PIMAGE_DOS_HEADER)hInst; 4417 PIMAGE_NT_HEADERS pPE; 4418 PIMAGE_IMPORT_DESCRIPTOR pImpDesc; 4419 PIMAGE_THUNK_DATA pIAT; /* Import Address Table */ 4420 PIMAGE_THUNK_DATA pINT; /* Import Name Table */ 4421 PIMAGE_IMPORT_BY_NAME pImpName; 4422 4423 if (pDOS->e_magic != IMAGE_DOS_SIGNATURE) 4424 return NULL; 4425 pPE = (PIMAGE_NT_HEADERS)(pImage + pDOS->e_lfanew); 4426 if (pPE->Signature != IMAGE_NT_SIGNATURE) 4427 return NULL; 4428 pImpDesc = (PIMAGE_IMPORT_DESCRIPTOR)(pImage 4429 + pPE->OptionalHeader.DataDirectory[IMAGE_DIRECTORY_ENTRY_IMPORT] 4430 .VirtualAddress); 4431 for (; pImpDesc->FirstThunk; ++pImpDesc) 4432 { 4433 if (!pImpDesc->OriginalFirstThunk) 4434 continue; 4435 pIAT = (PIMAGE_THUNK_DATA)(pImage + pImpDesc->FirstThunk); 4436 pINT = (PIMAGE_THUNK_DATA)(pImage + pImpDesc->OriginalFirstThunk); 4437 for (; pIAT->u1.Function; ++pIAT, ++pINT) 4438 { 4439 if (IMAGE_SNAP_BY_ORDINAL(pINT->u1.Ordinal)) 4440 continue; 4441 pImpName = (PIMAGE_IMPORT_BY_NAME)(pImage 4442 + (UINT_PTR)(pINT->u1.AddressOfData)); 4443 if (strcmp(pImpName->Name, funcname) == 0) 4444 return (void *)pIAT->u1.Function; 4445 } 4446 } 4447 return NULL; 4448 } 4449 4450 /* 4451 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4452 */ 4453 int 4454 iconv_enabled(verbose) 4455 int verbose; 4456 { 4457 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4458 return TRUE; 4459 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4460 if (hIconvDLL == 0) /* sometimes it's called libiconv.dll */ 4461 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT); 4462 if (hIconvDLL != 0) 4463 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4464 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4465 { 4466 /* Only give the message when 'verbose' is set, otherwise it might be 4467 * done whenever a conversion is attempted. */ 4468 if (verbose && p_verbose > 0) 4469 { 4470 verbose_enter(); 4471 EMSG2(_(e_loadlib), 4472 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4473 verbose_leave(); 4474 } 4475 iconv_end(); 4476 return FALSE; 4477 } 4478 4479 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4480 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4481 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4482 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4483 iconv_errno = get_iconv_import_func(hIconvDLL, "_errno"); 4484 if (iconv_errno == NULL) 4485 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4486 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4487 || iconvctl == NULL || iconv_errno == NULL) 4488 { 4489 iconv_end(); 4490 if (verbose && p_verbose > 0) 4491 { 4492 verbose_enter(); 4493 EMSG2(_(e_loadfunc), "for libiconv"); 4494 verbose_leave(); 4495 } 4496 return FALSE; 4497 } 4498 return TRUE; 4499 } 4500 4501 void 4502 iconv_end() 4503 { 4504 /* Don't use iconv() when inputting or outputting characters. */ 4505 if (input_conv.vc_type == CONV_ICONV) 4506 convert_setup(&input_conv, NULL, NULL); 4507 if (output_conv.vc_type == CONV_ICONV) 4508 convert_setup(&output_conv, NULL, NULL); 4509 4510 if (hIconvDLL != 0) 4511 FreeLibrary(hIconvDLL); 4512 if (hMsvcrtDLL != 0) 4513 FreeLibrary(hMsvcrtDLL); 4514 hIconvDLL = 0; 4515 hMsvcrtDLL = 0; 4516 } 4517 # endif /* DYNAMIC_ICONV */ 4518 # endif /* USE_ICONV */ 4519 4520 #endif /* FEAT_MBYTE */ 4521 4522 #if defined(FEAT_XIM) || defined(PROTO) 4523 4524 # if defined(FEAT_GUI_GTK) || defined(PROTO) 4525 static int xim_has_preediting INIT(= FALSE); /* IM current status */ 4526 4527 /* 4528 * Set preedit_start_col to the current cursor position. 4529 */ 4530 static void 4531 init_preedit_start_col(void) 4532 { 4533 if (State & CMDLINE) 4534 preedit_start_col = cmdline_getvcol_cursor(); 4535 else if (curwin != NULL) 4536 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); 4537 /* Prevent that preediting marks the buffer as changed. */ 4538 xim_changed_while_preediting = curbuf->b_changed; 4539 } 4540 4541 static int im_is_active = FALSE; /* IM is enabled for current mode */ 4542 static int preedit_is_active = FALSE; 4543 static int im_preedit_cursor = 0; /* cursor offset in characters */ 4544 static int im_preedit_trailing = 0; /* number of characters after cursor */ 4545 4546 static unsigned long im_commit_handler_id = 0; 4547 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; 4548 static unsigned int im_activatekey_state = 0; 4549 4550 void 4551 im_set_active(int active) 4552 { 4553 int was_active; 4554 4555 was_active = !!im_get_status(); 4556 im_is_active = (active && !p_imdisable); 4557 4558 if (im_is_active != was_active) 4559 xim_reset(); 4560 } 4561 4562 void 4563 xim_set_focus(int focus) 4564 { 4565 if (xic != NULL) 4566 { 4567 if (focus) 4568 gtk_im_context_focus_in(xic); 4569 else 4570 gtk_im_context_focus_out(xic); 4571 } 4572 } 4573 4574 void 4575 im_set_position(int row, int col) 4576 { 4577 if (xic != NULL) 4578 { 4579 GdkRectangle area; 4580 4581 area.x = FILL_X(col); 4582 area.y = FILL_Y(row); 4583 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); 4584 area.height = gui.char_height; 4585 4586 gtk_im_context_set_cursor_location(xic, &area); 4587 } 4588 } 4589 4590 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */ 4591 void 4592 xim_set_preedit(void) 4593 { 4594 im_set_position(gui.row, gui.col); 4595 } 4596 # endif 4597 4598 static void 4599 im_add_to_input(char_u *str, int len) 4600 { 4601 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */ 4602 if (input_conv.vc_type != CONV_NONE) 4603 { 4604 str = string_convert(&input_conv, str, &len); 4605 g_return_if_fail(str != NULL); 4606 } 4607 4608 add_to_input_buf_csi(str, len); 4609 4610 if (input_conv.vc_type != CONV_NONE) 4611 vim_free(str); 4612 4613 if (p_mh) /* blank out the pointer if necessary */ 4614 gui_mch_mousehide(TRUE); 4615 } 4616 4617 static void 4618 im_delete_preedit(void) 4619 { 4620 char_u bskey[] = {CSI, 'k', 'b'}; 4621 char_u delkey[] = {CSI, 'k', 'D'}; 4622 4623 if (State & NORMAL) 4624 { 4625 im_preedit_cursor = 0; 4626 return; 4627 } 4628 for (; im_preedit_cursor > 0; --im_preedit_cursor) 4629 add_to_input_buf(bskey, (int)sizeof(bskey)); 4630 4631 for (; im_preedit_trailing > 0; --im_preedit_trailing) 4632 add_to_input_buf(delkey, (int)sizeof(delkey)); 4633 } 4634 4635 /* 4636 * Move the cursor left by "num_move_back" characters. 4637 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for 4638 * these K_LEFT keys. 4639 */ 4640 static void 4641 im_correct_cursor(int num_move_back) 4642 { 4643 char_u backkey[] = {CSI, 'k', 'l'}; 4644 4645 if (State & NORMAL) 4646 return; 4647 # ifdef FEAT_RIGHTLEFT 4648 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) 4649 backkey[2] = 'r'; 4650 # endif 4651 for (; num_move_back > 0; --num_move_back) 4652 add_to_input_buf(backkey, (int)sizeof(backkey)); 4653 } 4654 4655 static int xim_expected_char = NUL; 4656 static int xim_ignored_char = FALSE; 4657 4658 /* 4659 * Update the mode and cursor while in an IM callback. 4660 */ 4661 static void 4662 im_show_info(void) 4663 { 4664 int old_vgetc_busy; 4665 4666 old_vgetc_busy = vgetc_busy; 4667 vgetc_busy = TRUE; 4668 showmode(); 4669 vgetc_busy = old_vgetc_busy; 4670 if ((State & NORMAL) || (State & INSERT)) 4671 setcursor(); 4672 out_flush(); 4673 } 4674 4675 /* 4676 * Callback invoked when the user finished preediting. 4677 * Put the final string into the input buffer. 4678 */ 4679 static void 4680 im_commit_cb(GtkIMContext *context UNUSED, 4681 const gchar *str, 4682 gpointer data UNUSED) 4683 { 4684 int slen = (int)STRLEN(str); 4685 int add_to_input = TRUE; 4686 int clen; 4687 int len = slen; 4688 int commit_with_preedit = TRUE; 4689 char_u *im_str; 4690 4691 #ifdef XIM_DEBUG 4692 xim_log("im_commit_cb(): %s\n", str); 4693 #endif 4694 4695 /* The imhangul module doesn't reset the preedit string before 4696 * committing. Call im_delete_preedit() to work around that. */ 4697 im_delete_preedit(); 4698 4699 /* Indicate that preediting has finished. */ 4700 if (preedit_start_col == MAXCOL) 4701 { 4702 init_preedit_start_col(); 4703 commit_with_preedit = FALSE; 4704 } 4705 4706 /* The thing which setting "preedit_start_col" to MAXCOL means that 4707 * "preedit_start_col" will be set forcedly when calling 4708 * preedit_changed_cb() next time. 4709 * "preedit_start_col" should not reset with MAXCOL on this part. Vim 4710 * is simulating the preediting by using add_to_input_str(). when 4711 * preedit begin immediately before committed, the typebuf is not 4712 * flushed to screen, then it can't get correct "preedit_start_col". 4713 * Thus, it should calculate the cells by adding cells of the committed 4714 * string. */ 4715 if (input_conv.vc_type != CONV_NONE) 4716 { 4717 im_str = string_convert(&input_conv, (char_u *)str, &len); 4718 g_return_if_fail(im_str != NULL); 4719 } 4720 else 4721 im_str = (char_u *)str; 4722 4723 clen = mb_string2cells(im_str, len); 4724 4725 if (input_conv.vc_type != CONV_NONE) 4726 vim_free(im_str); 4727 preedit_start_col += clen; 4728 4729 /* Is this a single character that matches a keypad key that's just 4730 * been pressed? If so, we don't want it to be entered as such - let 4731 * us carry on processing the raw keycode so that it may be used in 4732 * mappings as <kSomething>. */ 4733 if (xim_expected_char != NUL) 4734 { 4735 /* We're currently processing a keypad or other special key */ 4736 if (slen == 1 && str[0] == xim_expected_char) 4737 { 4738 /* It's a match - don't do it here */ 4739 xim_ignored_char = TRUE; 4740 add_to_input = FALSE; 4741 } 4742 else 4743 { 4744 /* Not a match */ 4745 xim_ignored_char = FALSE; 4746 } 4747 } 4748 4749 if (add_to_input) 4750 im_add_to_input((char_u *)str, slen); 4751 4752 /* Inserting chars while "im_is_active" is set does not cause a change of 4753 * buffer. When the chars are committed the buffer must be marked as 4754 * changed. */ 4755 if (!commit_with_preedit) 4756 preedit_start_col = MAXCOL; 4757 4758 /* This flag is used in changed() at next call. */ 4759 xim_changed_while_preediting = TRUE; 4760 4761 if (gtk_main_level() > 0) 4762 gtk_main_quit(); 4763 } 4764 4765 /* 4766 * Callback invoked after start to the preedit. 4767 */ 4768 static void 4769 im_preedit_start_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4770 { 4771 #ifdef XIM_DEBUG 4772 xim_log("im_preedit_start_cb()\n"); 4773 #endif 4774 4775 im_is_active = TRUE; 4776 preedit_is_active = TRUE; 4777 gui_update_cursor(TRUE, FALSE); 4778 im_show_info(); 4779 } 4780 4781 /* 4782 * Callback invoked after end to the preedit. 4783 */ 4784 static void 4785 im_preedit_end_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4786 { 4787 #ifdef XIM_DEBUG 4788 xim_log("im_preedit_end_cb()\n"); 4789 #endif 4790 im_delete_preedit(); 4791 4792 /* Indicate that preediting has finished */ 4793 preedit_start_col = MAXCOL; 4794 xim_has_preediting = FALSE; 4795 4796 #if 0 4797 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was 4798 * switched off unintentionally. We now use preedit_is_active (added by 4799 * SungHyun Nam). */ 4800 im_is_active = FALSE; 4801 #endif 4802 preedit_is_active = FALSE; 4803 gui_update_cursor(TRUE, FALSE); 4804 im_show_info(); 4805 } 4806 4807 /* 4808 * Callback invoked after changes to the preedit string. If the preedit 4809 * string was empty before, remember the preedit start column so we know 4810 * where to apply feedback attributes. Delete the previous preedit string 4811 * if there was one, save the new preedit cursor offset, and put the new 4812 * string into the input buffer. 4813 * 4814 * TODO: The pragmatic "put into input buffer" approach used here has 4815 * several fundamental problems: 4816 * 4817 * - The characters in the preedit string are subject to remapping. 4818 * That's broken, only the finally committed string should be remapped. 4819 * 4820 * - There is a race condition involved: The retrieved value for the 4821 * current cursor position will be wrong if any unprocessed characters 4822 * are still queued in the input buffer. 4823 * 4824 * - Due to the lack of synchronization between the file buffer in memory 4825 * and any typed characters, it's practically impossible to implement the 4826 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM 4827 * modules for languages such as Thai are likely to rely on this feature 4828 * for proper operation. 4829 * 4830 * Conclusions: I think support for preediting needs to be moved to the 4831 * core parts of Vim. Ideally, until it has been committed, the preediting 4832 * string should only be displayed and not affect the buffer content at all. 4833 * The question how to deal with the synchronization issue still remains. 4834 * Circumventing the input buffer is probably not desirable. Anyway, I think 4835 * implementing "retrieve_surrounding" is the only hard problem. 4836 * 4837 * One way to solve all of this in a clean manner would be to queue all key 4838 * press/release events "as is" in the input buffer, and apply the IM filtering 4839 * at the receiving end of the queue. This, however, would have a rather large 4840 * impact on the code base. If there is an easy way to force processing of all 4841 * remaining input from within the "retrieve_surrounding" signal handler, this 4842 * might not be necessary. Gotta ask on vim-dev for opinions. 4843 */ 4844 static void 4845 im_preedit_changed_cb(GtkIMContext *context, gpointer data UNUSED) 4846 { 4847 char *preedit_string = NULL; 4848 int cursor_index = 0; 4849 int num_move_back = 0; 4850 char_u *str; 4851 char_u *p; 4852 int i; 4853 4854 gtk_im_context_get_preedit_string(context, 4855 &preedit_string, NULL, 4856 &cursor_index); 4857 4858 #ifdef XIM_DEBUG 4859 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); 4860 #endif 4861 4862 g_return_if_fail(preedit_string != NULL); /* just in case */ 4863 4864 /* If preedit_start_col is MAXCOL set it to the current cursor position. */ 4865 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') 4866 { 4867 xim_has_preediting = TRUE; 4868 4869 /* Urgh, this breaks if the input buffer isn't empty now */ 4870 init_preedit_start_col(); 4871 } 4872 else if (cursor_index == 0 && preedit_string[0] == '\0') 4873 { 4874 xim_has_preediting = FALSE; 4875 4876 /* If at the start position (after typing backspace) 4877 * preedit_start_col must be reset. */ 4878 preedit_start_col = MAXCOL; 4879 } 4880 4881 im_delete_preedit(); 4882 4883 /* 4884 * Compute the end of the preediting area: "preedit_end_col". 4885 * According to the documentation of gtk_im_context_get_preedit_string(), 4886 * the cursor_pos output argument returns the offset in bytes. This is 4887 * unfortunately not true -- real life shows the offset is in characters, 4888 * and the GTK+ source code agrees with me. Will file a bug later. 4889 */ 4890 if (preedit_start_col != MAXCOL) 4891 preedit_end_col = preedit_start_col; 4892 str = (char_u *)preedit_string; 4893 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) 4894 { 4895 int is_composing; 4896 4897 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); 4898 /* 4899 * These offsets are used as counters when generating <BS> and <Del> 4900 * to delete the preedit string. So don't count composing characters 4901 * unless 'delcombine' is enabled. 4902 */ 4903 if (!is_composing || p_deco) 4904 { 4905 if (i < cursor_index) 4906 ++im_preedit_cursor; 4907 else 4908 ++im_preedit_trailing; 4909 } 4910 if (!is_composing && i >= cursor_index) 4911 { 4912 /* This is essentially the same as im_preedit_trailing, except 4913 * composing characters are not counted even if p_deco is set. */ 4914 ++num_move_back; 4915 } 4916 if (preedit_start_col != MAXCOL) 4917 preedit_end_col += utf_ptr2cells(p); 4918 } 4919 4920 if (p > str) 4921 { 4922 im_add_to_input(str, (int)(p - str)); 4923 im_correct_cursor(num_move_back); 4924 } 4925 4926 g_free(preedit_string); 4927 4928 if (gtk_main_level() > 0) 4929 gtk_main_quit(); 4930 } 4931 4932 /* 4933 * Translate the Pango attributes at iter to Vim highlighting attributes. 4934 * Ignore attributes not supported by Vim highlighting. This shouldn't have 4935 * too much impact -- right now we handle even more attributes than necessary 4936 * for the IM modules I tested with. 4937 */ 4938 static int 4939 translate_pango_attributes(PangoAttrIterator *iter) 4940 { 4941 PangoAttribute *attr; 4942 int char_attr = HL_NORMAL; 4943 4944 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); 4945 if (attr != NULL && ((PangoAttrInt *)attr)->value 4946 != (int)PANGO_UNDERLINE_NONE) 4947 char_attr |= HL_UNDERLINE; 4948 4949 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); 4950 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) 4951 char_attr |= HL_BOLD; 4952 4953 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); 4954 if (attr != NULL && ((PangoAttrInt *)attr)->value 4955 != (int)PANGO_STYLE_NORMAL) 4956 char_attr |= HL_ITALIC; 4957 4958 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); 4959 if (attr != NULL) 4960 { 4961 const PangoColor *color = &((PangoAttrColor *)attr)->color; 4962 4963 /* Assume inverse if black background is requested */ 4964 if ((color->red | color->green | color->blue) == 0) 4965 char_attr |= HL_INVERSE; 4966 } 4967 4968 return char_attr; 4969 } 4970 4971 /* 4972 * Retrieve the highlighting attributes at column col in the preedit string. 4973 * Return -1 if not in preediting mode or if col is out of range. 4974 */ 4975 int 4976 im_get_feedback_attr(int col) 4977 { 4978 char *preedit_string = NULL; 4979 PangoAttrList *attr_list = NULL; 4980 int char_attr = -1; 4981 4982 if (xic == NULL) 4983 return char_attr; 4984 4985 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 4986 4987 if (preedit_string != NULL && attr_list != NULL) 4988 { 4989 int idx; 4990 4991 /* Get the byte index as used by PangoAttrIterator */ 4992 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col) 4993 idx += utfc_ptr2len((char_u *)preedit_string + idx); 4994 4995 if (preedit_string[idx] != '\0') 4996 { 4997 PangoAttrIterator *iter; 4998 int start, end; 4999 5000 char_attr = HL_NORMAL; 5001 iter = pango_attr_list_get_iterator(attr_list); 5002 5003 /* Extract all relevant attributes from the list. */ 5004 do 5005 { 5006 pango_attr_iterator_range(iter, &start, &end); 5007 5008 if (idx >= start && idx < end) 5009 char_attr |= translate_pango_attributes(iter); 5010 } 5011 while (pango_attr_iterator_next(iter)); 5012 5013 pango_attr_iterator_destroy(iter); 5014 } 5015 } 5016 5017 if (attr_list != NULL) 5018 pango_attr_list_unref(attr_list); 5019 g_free(preedit_string); 5020 5021 return char_attr; 5022 } 5023 5024 void 5025 xim_init(void) 5026 { 5027 #ifdef XIM_DEBUG 5028 xim_log("xim_init()\n"); 5029 #endif 5030 5031 g_return_if_fail(gui.drawarea != NULL); 5032 g_return_if_fail(gui.drawarea->window != NULL); 5033 5034 xic = gtk_im_multicontext_new(); 5035 g_object_ref(xic); 5036 5037 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", 5038 G_CALLBACK(&im_commit_cb), NULL); 5039 g_signal_connect(G_OBJECT(xic), "preedit_changed", 5040 G_CALLBACK(&im_preedit_changed_cb), NULL); 5041 g_signal_connect(G_OBJECT(xic), "preedit_start", 5042 G_CALLBACK(&im_preedit_start_cb), NULL); 5043 g_signal_connect(G_OBJECT(xic), "preedit_end", 5044 G_CALLBACK(&im_preedit_end_cb), NULL); 5045 5046 gtk_im_context_set_client_window(xic, gui.drawarea->window); 5047 } 5048 5049 void 5050 im_shutdown(void) 5051 { 5052 #ifdef XIM_DEBUG 5053 xim_log("im_shutdown()\n"); 5054 #endif 5055 5056 if (xic != NULL) 5057 { 5058 gtk_im_context_focus_out(xic); 5059 g_object_unref(xic); 5060 xic = NULL; 5061 } 5062 im_is_active = FALSE; 5063 im_commit_handler_id = 0; 5064 preedit_start_col = MAXCOL; 5065 xim_has_preediting = FALSE; 5066 } 5067 5068 /* 5069 * Convert the string argument to keyval and state for GdkEventKey. 5070 * If str is valid return TRUE, otherwise FALSE. 5071 * 5072 * See 'imactivatekey' for documentation of the format. 5073 */ 5074 static int 5075 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) 5076 { 5077 const char *mods_end; 5078 unsigned tmp_keyval; 5079 unsigned tmp_state = 0; 5080 5081 mods_end = strrchr(str, '-'); 5082 mods_end = (mods_end != NULL) ? mods_end + 1 : str; 5083 5084 /* Parse modifier keys */ 5085 while (str < mods_end) 5086 switch (*str++) 5087 { 5088 case '-': break; 5089 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; 5090 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; 5091 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; 5092 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; 5093 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; 5094 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; 5095 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; 5096 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; 5097 default: 5098 return FALSE; 5099 } 5100 5101 tmp_keyval = gdk_keyval_from_name(str); 5102 5103 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) 5104 return FALSE; 5105 5106 if (keyval != NULL) 5107 *keyval = tmp_keyval; 5108 if (state != NULL) 5109 *state = tmp_state; 5110 5111 return TRUE; 5112 } 5113 5114 /* 5115 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an 5116 * empty string is also regarded as valid. 5117 * 5118 * Note: The numerical key value of p_imak is cached if it was valid; thus 5119 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever 5120 * 'imak' changes. This is currently the case but not obvious -- should 5121 * probably rename the function for clarity. 5122 */ 5123 int 5124 im_xim_isvalid_imactivate(void) 5125 { 5126 if (p_imak[0] == NUL) 5127 { 5128 im_activatekey_keyval = GDK_VoidSymbol; 5129 im_activatekey_state = 0; 5130 return TRUE; 5131 } 5132 5133 return im_string_to_keyval((const char *)p_imak, 5134 &im_activatekey_keyval, 5135 &im_activatekey_state); 5136 } 5137 5138 static void 5139 im_synthesize_keypress(unsigned int keyval, unsigned int state) 5140 { 5141 GdkEventKey *event; 5142 5143 # ifdef HAVE_GTK_MULTIHEAD 5144 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); 5145 g_object_ref(gui.drawarea->window); /* unreffed by gdk_event_free() */ 5146 # else 5147 event = (GdkEventKey *)g_malloc0((gulong)sizeof(GdkEvent)); 5148 event->type = GDK_KEY_PRESS; 5149 # endif 5150 event->window = gui.drawarea->window; 5151 event->send_event = TRUE; 5152 event->time = GDK_CURRENT_TIME; 5153 event->state = state; 5154 event->keyval = keyval; 5155 event->hardware_keycode = /* needed for XIM */ 5156 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); 5157 event->length = 0; 5158 event->string = NULL; 5159 5160 gtk_im_context_filter_keypress(xic, event); 5161 5162 /* For consistency, also send the corresponding release event. */ 5163 event->type = GDK_KEY_RELEASE; 5164 event->send_event = FALSE; 5165 gtk_im_context_filter_keypress(xic, event); 5166 5167 # ifdef HAVE_GTK_MULTIHEAD 5168 gdk_event_free((GdkEvent *)event); 5169 # else 5170 g_free(event); 5171 # endif 5172 } 5173 5174 void 5175 xim_reset(void) 5176 { 5177 if (xic != NULL) 5178 { 5179 gtk_im_context_reset(xic); 5180 5181 if (p_imdisable) 5182 im_shutdown(); 5183 else 5184 { 5185 xim_set_focus(gui.in_focus); 5186 5187 # ifdef FEAT_EVAL 5188 if (p_imaf[0] != NUL) 5189 { 5190 char_u *argv[1]; 5191 5192 if (im_is_active) 5193 argv[0] = (char_u *)"1"; 5194 else 5195 argv[0] = (char_u *)"0"; 5196 (void)call_func_retnr(p_imaf, 1, argv, FALSE); 5197 } 5198 else 5199 # endif 5200 if (im_activatekey_keyval != GDK_VoidSymbol) 5201 { 5202 if (im_is_active) 5203 { 5204 g_signal_handler_block(xic, im_commit_handler_id); 5205 im_synthesize_keypress(im_activatekey_keyval, 5206 im_activatekey_state); 5207 g_signal_handler_unblock(xic, im_commit_handler_id); 5208 } 5209 } 5210 else 5211 { 5212 im_shutdown(); 5213 xim_init(); 5214 xim_set_focus(gui.in_focus); 5215 } 5216 } 5217 } 5218 5219 preedit_start_col = MAXCOL; 5220 xim_has_preediting = FALSE; 5221 } 5222 5223 int 5224 xim_queue_key_press_event(GdkEventKey *event, int down) 5225 { 5226 if (down) 5227 { 5228 /* 5229 * Workaround GTK2 XIM 'feature' that always converts keypad keys to 5230 * chars., even when not part of an IM sequence (ref. feature of 5231 * gdk/gdkkeyuni.c). 5232 * Flag any keypad keys that might represent a single char. 5233 * If this (on its own - i.e., not part of an IM sequence) is 5234 * committed while we're processing one of these keys, we can ignore 5235 * that commit and go ahead & process it ourselves. That way we can 5236 * still distinguish keypad keys for use in mappings. 5237 * Also add GDK_space to make <S-Space> work. 5238 */ 5239 switch (event->keyval) 5240 { 5241 case GDK_KP_Add: xim_expected_char = '+'; break; 5242 case GDK_KP_Subtract: xim_expected_char = '-'; break; 5243 case GDK_KP_Divide: xim_expected_char = '/'; break; 5244 case GDK_KP_Multiply: xim_expected_char = '*'; break; 5245 case GDK_KP_Decimal: xim_expected_char = '.'; break; 5246 case GDK_KP_Equal: xim_expected_char = '='; break; 5247 case GDK_KP_0: xim_expected_char = '0'; break; 5248 case GDK_KP_1: xim_expected_char = '1'; break; 5249 case GDK_KP_2: xim_expected_char = '2'; break; 5250 case GDK_KP_3: xim_expected_char = '3'; break; 5251 case GDK_KP_4: xim_expected_char = '4'; break; 5252 case GDK_KP_5: xim_expected_char = '5'; break; 5253 case GDK_KP_6: xim_expected_char = '6'; break; 5254 case GDK_KP_7: xim_expected_char = '7'; break; 5255 case GDK_KP_8: xim_expected_char = '8'; break; 5256 case GDK_KP_9: xim_expected_char = '9'; break; 5257 case GDK_space: xim_expected_char = ' '; break; 5258 default: xim_expected_char = NUL; 5259 } 5260 xim_ignored_char = FALSE; 5261 } 5262 5263 /* 5264 * When typing fFtT, XIM may be activated. Thus it must pass 5265 * gtk_im_context_filter_keypress() in Normal mode. 5266 * And while doing :sh too. 5267 */ 5268 if (xic != NULL && !p_imdisable 5269 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) 5270 { 5271 /* 5272 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is 5273 * always aware of the current status of IM, and can even emulate 5274 * the activation key for modules that don't support one. 5275 */ 5276 if (event->keyval == im_activatekey_keyval 5277 && (event->state & im_activatekey_state) == im_activatekey_state) 5278 { 5279 unsigned int state_mask; 5280 5281 /* Require the state of the 3 most used modifiers to match exactly. 5282 * Otherwise e.g. <S-C-space> would be unusable for other purposes 5283 * if the IM activate key is <S-space>. */ 5284 state_mask = im_activatekey_state; 5285 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK 5286 | (int)GDK_MOD1_MASK); 5287 5288 if ((event->state & state_mask) != im_activatekey_state) 5289 return FALSE; 5290 5291 /* Don't send it a second time on GDK_KEY_RELEASE. */ 5292 if (event->type != GDK_KEY_PRESS) 5293 return TRUE; 5294 5295 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE)) 5296 { 5297 im_set_active(FALSE); 5298 5299 /* ":lmap" mappings exists, toggle use of mappings. */ 5300 State ^= LANGMAP; 5301 if (State & LANGMAP) 5302 { 5303 curbuf->b_p_iminsert = B_IMODE_NONE; 5304 State &= ~LANGMAP; 5305 } 5306 else 5307 { 5308 curbuf->b_p_iminsert = B_IMODE_LMAP; 5309 State |= LANGMAP; 5310 } 5311 return TRUE; 5312 } 5313 5314 return gtk_im_context_filter_keypress(xic, event); 5315 } 5316 5317 /* Don't filter events through the IM context if IM isn't active 5318 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module 5319 * not doing anything before the activation key was sent. */ 5320 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) 5321 { 5322 int imresult = gtk_im_context_filter_keypress(xic, event); 5323 5324 /* Some XIM send following sequence: 5325 * 1. preedited string. 5326 * 2. committed string. 5327 * 3. line changed key. 5328 * 4. preedited string. 5329 * 5. remove preedited string. 5330 * if 3, Vim can't move back the above line for 5. 5331 * thus, this part should not parse the key. */ 5332 if (!imresult && preedit_start_col != MAXCOL 5333 && event->keyval == GDK_Return) 5334 { 5335 im_synthesize_keypress(GDK_Return, 0U); 5336 return FALSE; 5337 } 5338 5339 /* If XIM tried to commit a keypad key as a single char., 5340 * ignore it so we can use the keypad key 'raw', for mappings. */ 5341 if (xim_expected_char != NUL && xim_ignored_char) 5342 /* We had a keypad key, and XIM tried to thieve it */ 5343 return FALSE; 5344 5345 /* This is supposed to fix a problem with iBus, that space 5346 * characters don't work in input mode. */ 5347 xim_expected_char = NUL; 5348 5349 /* Normal processing */ 5350 return imresult; 5351 } 5352 } 5353 5354 return FALSE; 5355 } 5356 5357 int 5358 im_get_status(void) 5359 { 5360 # ifdef FEAT_EVAL 5361 if (p_imsf[0] != NUL) 5362 { 5363 int is_active; 5364 5365 /* FIXME: Don't execute user function in unsafe situation. */ 5366 if (exiting 5367 # ifdef FEAT_AUTOCMD 5368 || is_autocmd_blocked() 5369 # endif 5370 ) 5371 return FALSE; 5372 /* FIXME: :py print 'xxx' is shown duplicate result. 5373 * Use silent to avoid it. */ 5374 ++msg_silent; 5375 is_active = call_func_retnr(p_imsf, 0, NULL, FALSE); 5376 --msg_silent; 5377 return (is_active > 0); 5378 } 5379 # endif 5380 return im_is_active; 5381 } 5382 5383 int 5384 preedit_get_status(void) 5385 { 5386 return preedit_is_active; 5387 } 5388 5389 int 5390 im_is_preediting() 5391 { 5392 return xim_has_preediting; 5393 } 5394 5395 # else /* !FEAT_GUI_GTK */ 5396 5397 static int xim_is_active = FALSE; /* XIM should be active in the current 5398 mode */ 5399 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */ 5400 #ifdef FEAT_GUI_X11 5401 static XIMStyle input_style; 5402 static int status_area_enabled = TRUE; 5403 #endif 5404 5405 /* 5406 * Switch using XIM on/off. This is used by the code that changes "State". 5407 */ 5408 void 5409 im_set_active(active) 5410 int active; 5411 { 5412 if (xic == NULL) 5413 return; 5414 5415 /* If 'imdisable' is set, XIM is never active. */ 5416 if (p_imdisable) 5417 active = FALSE; 5418 #if !defined(FEAT_GUI_GTK) 5419 else if (input_style & XIMPreeditPosition) 5420 /* There is a problem in switching XIM off when preediting is used, 5421 * and it is not clear how this can be solved. For now, keep XIM on 5422 * all the time, like it was done in Vim 5.8. */ 5423 active = TRUE; 5424 #endif 5425 5426 /* Remember the active state, it is needed when Vim gets keyboard focus. */ 5427 xim_is_active = active; 5428 xim_set_preedit(); 5429 } 5430 5431 /* 5432 * Adjust using XIM for gaining or losing keyboard focus. Also called when 5433 * "xim_is_active" changes. 5434 */ 5435 void 5436 xim_set_focus(focus) 5437 int focus; 5438 { 5439 if (xic == NULL) 5440 return; 5441 5442 /* 5443 * XIM only gets focus when the Vim window has keyboard focus and XIM has 5444 * been set active for the current mode. 5445 */ 5446 if (focus && xim_is_active) 5447 { 5448 if (!xim_has_focus) 5449 { 5450 xim_has_focus = TRUE; 5451 XSetICFocus(xic); 5452 } 5453 } 5454 else 5455 { 5456 if (xim_has_focus) 5457 { 5458 xim_has_focus = FALSE; 5459 XUnsetICFocus(xic); 5460 } 5461 } 5462 } 5463 5464 void 5465 im_set_position(row, col) 5466 int row UNUSED; 5467 int col UNUSED; 5468 { 5469 xim_set_preedit(); 5470 } 5471 5472 /* 5473 * Set the XIM to the current cursor position. 5474 */ 5475 void 5476 xim_set_preedit() 5477 { 5478 XVaNestedList attr_list; 5479 XRectangle spot_area; 5480 XPoint over_spot; 5481 int line_space; 5482 5483 if (xic == NULL) 5484 return; 5485 5486 xim_set_focus(TRUE); 5487 5488 if (!xim_has_focus) 5489 { 5490 /* hide XIM cursor */ 5491 over_spot.x = 0; 5492 over_spot.y = -100; /* arbitrary invisible position */ 5493 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5494 XNSpotLocation, 5495 &over_spot, 5496 NULL); 5497 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); 5498 XFree(attr_list); 5499 return; 5500 } 5501 5502 if (input_style & XIMPreeditPosition) 5503 { 5504 if (xim_fg_color == INVALCOLOR) 5505 { 5506 xim_fg_color = gui.def_norm_pixel; 5507 xim_bg_color = gui.def_back_pixel; 5508 } 5509 over_spot.x = TEXT_X(gui.col); 5510 over_spot.y = TEXT_Y(gui.row); 5511 spot_area.x = 0; 5512 spot_area.y = 0; 5513 spot_area.height = gui.char_height * Rows; 5514 spot_area.width = gui.char_width * Columns; 5515 line_space = gui.char_height; 5516 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5517 XNSpotLocation, &over_spot, 5518 XNForeground, (Pixel) xim_fg_color, 5519 XNBackground, (Pixel) xim_bg_color, 5520 XNArea, &spot_area, 5521 XNLineSpace, line_space, 5522 NULL); 5523 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) 5524 EMSG(_("E284: Cannot set IC values")); 5525 XFree(attr_list); 5526 } 5527 } 5528 5529 #if defined(FEAT_GUI_X11) 5530 static char e_xim[] = N_("E285: Failed to create input context"); 5531 #endif 5532 5533 #if defined(FEAT_GUI_X11) || defined(PROTO) 5534 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(sun) 5535 # define USE_X11R6_XIM 5536 # endif 5537 5538 static int xim_real_init __ARGS((Window x11_window, Display *x11_display)); 5539 5540 5541 #ifdef USE_X11R6_XIM 5542 static void xim_instantiate_cb __ARGS((Display *display, XPointer client_data, XPointer call_data)); 5543 static void xim_destroy_cb __ARGS((XIM im, XPointer client_data, XPointer call_data)); 5544 5545 static void 5546 xim_instantiate_cb(display, client_data, call_data) 5547 Display *display; 5548 XPointer client_data UNUSED; 5549 XPointer call_data UNUSED; 5550 { 5551 Window x11_window; 5552 Display *x11_display; 5553 5554 #ifdef XIM_DEBUG 5555 xim_log("xim_instantiate_cb()\n"); 5556 #endif 5557 5558 gui_get_x11_windis(&x11_window, &x11_display); 5559 if (display != x11_display) 5560 return; 5561 5562 xim_real_init(x11_window, x11_display); 5563 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5564 if (xic != NULL) 5565 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5566 xim_instantiate_cb, NULL); 5567 } 5568 5569 static void 5570 xim_destroy_cb(im, client_data, call_data) 5571 XIM im UNUSED; 5572 XPointer client_data UNUSED; 5573 XPointer call_data UNUSED; 5574 { 5575 Window x11_window; 5576 Display *x11_display; 5577 5578 #ifdef XIM_DEBUG 5579 xim_log("xim_destroy_cb()\n"); 5580 #endif 5581 gui_get_x11_windis(&x11_window, &x11_display); 5582 5583 xic = NULL; 5584 status_area_enabled = FALSE; 5585 5586 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5587 5588 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5589 xim_instantiate_cb, NULL); 5590 } 5591 #endif 5592 5593 void 5594 xim_init() 5595 { 5596 Window x11_window; 5597 Display *x11_display; 5598 5599 #ifdef XIM_DEBUG 5600 xim_log("xim_init()\n"); 5601 #endif 5602 5603 gui_get_x11_windis(&x11_window, &x11_display); 5604 5605 xic = NULL; 5606 5607 if (xim_real_init(x11_window, x11_display)) 5608 return; 5609 5610 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5611 5612 #ifdef USE_X11R6_XIM 5613 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5614 xim_instantiate_cb, NULL); 5615 #endif 5616 } 5617 5618 static int 5619 xim_real_init(x11_window, x11_display) 5620 Window x11_window; 5621 Display *x11_display; 5622 { 5623 int i; 5624 char *p, 5625 *s, 5626 *ns, 5627 *end, 5628 tmp[1024]; 5629 #define IMLEN_MAX 40 5630 char buf[IMLEN_MAX + 7]; 5631 XIM xim = NULL; 5632 XIMStyles *xim_styles; 5633 XIMStyle this_input_style = 0; 5634 Boolean found; 5635 XPoint over_spot; 5636 XVaNestedList preedit_list, status_list; 5637 5638 input_style = 0; 5639 status_area_enabled = FALSE; 5640 5641 if (xic != NULL) 5642 return FALSE; 5643 5644 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) 5645 { 5646 strcpy(tmp, gui.rsrc_input_method); 5647 for (ns = s = tmp; ns != NULL && *s != NUL;) 5648 { 5649 s = (char *)skipwhite((char_u *)s); 5650 if (*s == NUL) 5651 break; 5652 if ((ns = end = strchr(s, ',')) == NULL) 5653 end = s + strlen(s); 5654 while (isspace(((char_u *)end)[-1])) 5655 end--; 5656 *end = NUL; 5657 5658 if (strlen(s) <= IMLEN_MAX) 5659 { 5660 strcpy(buf, "@im="); 5661 strcat(buf, s); 5662 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL 5663 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) 5664 != NULL) 5665 break; 5666 } 5667 5668 s = ns + 1; 5669 } 5670 } 5671 5672 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) 5673 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5674 5675 /* This is supposed to be useful to obtain characters through 5676 * XmbLookupString() without really using a XIM. */ 5677 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL 5678 && *p != NUL) 5679 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5680 5681 if (xim == NULL) 5682 { 5683 /* Only give this message when verbose is set, because too many people 5684 * got this message when they didn't want to use a XIM. */ 5685 if (p_verbose > 0) 5686 { 5687 verbose_enter(); 5688 EMSG(_("E286: Failed to open input method")); 5689 verbose_leave(); 5690 } 5691 return FALSE; 5692 } 5693 5694 #ifdef USE_X11R6_XIM 5695 { 5696 XIMCallback destroy_cb; 5697 5698 destroy_cb.callback = xim_destroy_cb; 5699 destroy_cb.client_data = NULL; 5700 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) 5701 EMSG(_("E287: Warning: Could not set destroy callback to IM")); 5702 } 5703 #endif 5704 5705 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) 5706 { 5707 EMSG(_("E288: input method doesn't support any style")); 5708 XCloseIM(xim); 5709 return FALSE; 5710 } 5711 5712 found = False; 5713 strcpy(tmp, gui.rsrc_preedit_type_name); 5714 for (s = tmp; s && !found; ) 5715 { 5716 while (*s && isspace((unsigned char)*s)) 5717 s++; 5718 if (!*s) 5719 break; 5720 if ((ns = end = strchr(s, ',')) != 0) 5721 ns++; 5722 else 5723 end = s + strlen(s); 5724 while (isspace((unsigned char)*end)) 5725 end--; 5726 *end = '\0'; 5727 5728 if (!strcmp(s, "OverTheSpot")) 5729 this_input_style = (XIMPreeditPosition | XIMStatusArea); 5730 else if (!strcmp(s, "OffTheSpot")) 5731 this_input_style = (XIMPreeditArea | XIMStatusArea); 5732 else if (!strcmp(s, "Root")) 5733 this_input_style = (XIMPreeditNothing | XIMStatusNothing); 5734 5735 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5736 { 5737 if (this_input_style == xim_styles->supported_styles[i]) 5738 { 5739 found = True; 5740 break; 5741 } 5742 } 5743 if (!found) 5744 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5745 { 5746 if ((xim_styles->supported_styles[i] & this_input_style) 5747 == (this_input_style & ~XIMStatusArea)) 5748 { 5749 this_input_style &= ~XIMStatusArea; 5750 found = True; 5751 break; 5752 } 5753 } 5754 5755 s = ns; 5756 } 5757 XFree(xim_styles); 5758 5759 if (!found) 5760 { 5761 /* Only give this message when verbose is set, because too many people 5762 * got this message when they didn't want to use a XIM. */ 5763 if (p_verbose > 0) 5764 { 5765 verbose_enter(); 5766 EMSG(_("E289: input method doesn't support my preedit type")); 5767 verbose_leave(); 5768 } 5769 XCloseIM(xim); 5770 return FALSE; 5771 } 5772 5773 over_spot.x = TEXT_X(gui.col); 5774 over_spot.y = TEXT_Y(gui.row); 5775 input_style = this_input_style; 5776 5777 /* A crash was reported when trying to pass gui.norm_font as XNFontSet, 5778 * thus that has been removed. Hopefully the default works... */ 5779 #ifdef FEAT_XFONTSET 5780 if (gui.fontset != NOFONTSET) 5781 { 5782 preedit_list = XVaCreateNestedList(0, 5783 XNSpotLocation, &over_spot, 5784 XNForeground, (Pixel)gui.def_norm_pixel, 5785 XNBackground, (Pixel)gui.def_back_pixel, 5786 XNFontSet, (XFontSet)gui.fontset, 5787 NULL); 5788 status_list = XVaCreateNestedList(0, 5789 XNForeground, (Pixel)gui.def_norm_pixel, 5790 XNBackground, (Pixel)gui.def_back_pixel, 5791 XNFontSet, (XFontSet)gui.fontset, 5792 NULL); 5793 } 5794 else 5795 #endif 5796 { 5797 preedit_list = XVaCreateNestedList(0, 5798 XNSpotLocation, &over_spot, 5799 XNForeground, (Pixel)gui.def_norm_pixel, 5800 XNBackground, (Pixel)gui.def_back_pixel, 5801 NULL); 5802 status_list = XVaCreateNestedList(0, 5803 XNForeground, (Pixel)gui.def_norm_pixel, 5804 XNBackground, (Pixel)gui.def_back_pixel, 5805 NULL); 5806 } 5807 5808 xic = XCreateIC(xim, 5809 XNInputStyle, input_style, 5810 XNClientWindow, x11_window, 5811 XNFocusWindow, gui.wid, 5812 XNPreeditAttributes, preedit_list, 5813 XNStatusAttributes, status_list, 5814 NULL); 5815 XFree(status_list); 5816 XFree(preedit_list); 5817 if (xic != NULL) 5818 { 5819 if (input_style & XIMStatusArea) 5820 { 5821 xim_set_status_area(); 5822 status_area_enabled = TRUE; 5823 } 5824 else 5825 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5826 } 5827 else 5828 { 5829 EMSG(_(e_xim)); 5830 XCloseIM(xim); 5831 return FALSE; 5832 } 5833 5834 return TRUE; 5835 } 5836 5837 #endif /* FEAT_GUI_X11 */ 5838 5839 /* 5840 * Get IM status. When IM is on, return TRUE. Else return FALSE. 5841 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is 5842 * active, when not having focus XIM may still be active (e.g., when using a 5843 * tear-off menu item). 5844 */ 5845 int 5846 im_get_status() 5847 { 5848 return xim_has_focus; 5849 } 5850 5851 # endif /* !FEAT_GUI_GTK */ 5852 5853 # if !defined(FEAT_GUI_GTK) || defined(PROTO) 5854 /* 5855 * Set up the status area. 5856 * 5857 * This should use a separate Widget, but that seems not possible, because 5858 * preedit_area and status_area should be set to the same window as for the 5859 * text input. Unfortunately this means the status area pollutes the text 5860 * window... 5861 */ 5862 void 5863 xim_set_status_area() 5864 { 5865 XVaNestedList preedit_list = 0, status_list = 0, list = 0; 5866 XRectangle pre_area, status_area; 5867 5868 if (xic == NULL) 5869 return; 5870 5871 if (input_style & XIMStatusArea) 5872 { 5873 if (input_style & XIMPreeditArea) 5874 { 5875 XRectangle *needed_rect; 5876 5877 /* to get status_area width */ 5878 status_list = XVaCreateNestedList(0, XNAreaNeeded, 5879 &needed_rect, NULL); 5880 XGetICValues(xic, XNStatusAttributes, status_list, NULL); 5881 XFree(status_list); 5882 5883 status_area.width = needed_rect->width; 5884 } 5885 else 5886 status_area.width = gui.char_width * Columns; 5887 5888 status_area.x = 0; 5889 status_area.y = gui.char_height * Rows + gui.border_offset; 5890 if (gui.which_scrollbars[SBAR_BOTTOM]) 5891 status_area.y += gui.scrollbar_height; 5892 #ifdef FEAT_MENU 5893 if (gui.menu_is_active) 5894 status_area.y += gui.menu_height; 5895 #endif 5896 status_area.height = gui.char_height; 5897 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); 5898 } 5899 else 5900 { 5901 status_area.x = 0; 5902 status_area.y = gui.char_height * Rows + gui.border_offset; 5903 if (gui.which_scrollbars[SBAR_BOTTOM]) 5904 status_area.y += gui.scrollbar_height; 5905 #ifdef FEAT_MENU 5906 if (gui.menu_is_active) 5907 status_area.y += gui.menu_height; 5908 #endif 5909 status_area.width = 0; 5910 status_area.height = gui.char_height; 5911 } 5912 5913 if (input_style & XIMPreeditArea) /* off-the-spot */ 5914 { 5915 pre_area.x = status_area.x + status_area.width; 5916 pre_area.y = gui.char_height * Rows + gui.border_offset; 5917 pre_area.width = gui.char_width * Columns - pre_area.x; 5918 if (gui.which_scrollbars[SBAR_BOTTOM]) 5919 pre_area.y += gui.scrollbar_height; 5920 #ifdef FEAT_MENU 5921 if (gui.menu_is_active) 5922 pre_area.y += gui.menu_height; 5923 #endif 5924 pre_area.height = gui.char_height; 5925 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 5926 } 5927 else if (input_style & XIMPreeditPosition) /* over-the-spot */ 5928 { 5929 pre_area.x = 0; 5930 pre_area.y = 0; 5931 pre_area.height = gui.char_height * Rows; 5932 pre_area.width = gui.char_width * Columns; 5933 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 5934 } 5935 5936 if (preedit_list && status_list) 5937 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 5938 XNStatusAttributes, status_list, NULL); 5939 else if (preedit_list) 5940 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 5941 NULL); 5942 else if (status_list) 5943 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, 5944 NULL); 5945 else 5946 list = NULL; 5947 5948 if (list) 5949 { 5950 XSetICValues(xic, XNVaNestedList, list, NULL); 5951 XFree(list); 5952 } 5953 if (status_list) 5954 XFree(status_list); 5955 if (preedit_list) 5956 XFree(preedit_list); 5957 } 5958 5959 int 5960 xim_get_status_area_height() 5961 { 5962 if (status_area_enabled) 5963 return gui.char_height; 5964 return 0; 5965 } 5966 # endif 5967 5968 #endif /* FEAT_XIM */ 5969 5970 #if defined(FEAT_MBYTE) || defined(PROTO) 5971 5972 /* 5973 * Setup "vcp" for conversion from "from" to "to". 5974 * The names must have been made canonical with enc_canonize(). 5975 * vcp->vc_type must have been initialized to CONV_NONE. 5976 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 5977 * instead). 5978 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 5979 * Return FAIL when conversion is not supported, OK otherwise. 5980 */ 5981 int 5982 convert_setup(vcp, from, to) 5983 vimconv_T *vcp; 5984 char_u *from; 5985 char_u *to; 5986 { 5987 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 5988 } 5989 5990 /* 5991 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 5992 * "from" unicode charsets be considered utf-8. Same for "to". 5993 */ 5994 int 5995 convert_setup_ext(vcp, from, from_unicode_is_utf8, to, to_unicode_is_utf8) 5996 vimconv_T *vcp; 5997 char_u *from; 5998 int from_unicode_is_utf8; 5999 char_u *to; 6000 int to_unicode_is_utf8; 6001 { 6002 int from_prop; 6003 int to_prop; 6004 int from_is_utf8; 6005 int to_is_utf8; 6006 6007 /* Reset to no conversion. */ 6008 # ifdef USE_ICONV 6009 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 6010 iconv_close(vcp->vc_fd); 6011 # endif 6012 vcp->vc_type = CONV_NONE; 6013 vcp->vc_factor = 1; 6014 vcp->vc_fail = FALSE; 6015 6016 /* No conversion when one of the names is empty or they are equal. */ 6017 if (from == NULL || *from == NUL || to == NULL || *to == NUL 6018 || STRCMP(from, to) == 0) 6019 return OK; 6020 6021 from_prop = enc_canon_props(from); 6022 to_prop = enc_canon_props(to); 6023 if (from_unicode_is_utf8) 6024 from_is_utf8 = from_prop & ENC_UNICODE; 6025 else 6026 from_is_utf8 = from_prop == ENC_UNICODE; 6027 if (to_unicode_is_utf8) 6028 to_is_utf8 = to_prop & ENC_UNICODE; 6029 else 6030 to_is_utf8 = to_prop == ENC_UNICODE; 6031 6032 if ((from_prop & ENC_LATIN1) && to_is_utf8) 6033 { 6034 /* Internal latin1 -> utf-8 conversion. */ 6035 vcp->vc_type = CONV_TO_UTF8; 6036 vcp->vc_factor = 2; /* up to twice as long */ 6037 } 6038 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 6039 { 6040 /* Internal latin9 -> utf-8 conversion. */ 6041 vcp->vc_type = CONV_9_TO_UTF8; 6042 vcp->vc_factor = 3; /* up to three as long (euro sign) */ 6043 } 6044 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 6045 { 6046 /* Internal utf-8 -> latin1 conversion. */ 6047 vcp->vc_type = CONV_TO_LATIN1; 6048 } 6049 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 6050 { 6051 /* Internal utf-8 -> latin9 conversion. */ 6052 vcp->vc_type = CONV_TO_LATIN9; 6053 } 6054 #ifdef WIN3264 6055 /* Win32-specific codepage <-> codepage conversion without iconv. */ 6056 else if ((from_is_utf8 || encname2codepage(from) > 0) 6057 && (to_is_utf8 || encname2codepage(to) > 0)) 6058 { 6059 vcp->vc_type = CONV_CODEPAGE; 6060 vcp->vc_factor = 2; /* up to twice as long */ 6061 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 6062 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 6063 } 6064 #endif 6065 #ifdef MACOS_X 6066 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 6067 { 6068 vcp->vc_type = CONV_MAC_LATIN1; 6069 } 6070 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 6071 { 6072 vcp->vc_type = CONV_MAC_UTF8; 6073 vcp->vc_factor = 2; /* up to twice as long */ 6074 } 6075 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 6076 { 6077 vcp->vc_type = CONV_LATIN1_MAC; 6078 } 6079 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 6080 { 6081 vcp->vc_type = CONV_UTF8_MAC; 6082 } 6083 #endif 6084 # ifdef USE_ICONV 6085 else 6086 { 6087 /* Use iconv() for conversion. */ 6088 vcp->vc_fd = (iconv_t)my_iconv_open( 6089 to_is_utf8 ? (char_u *)"utf-8" : to, 6090 from_is_utf8 ? (char_u *)"utf-8" : from); 6091 if (vcp->vc_fd != (iconv_t)-1) 6092 { 6093 vcp->vc_type = CONV_ICONV; 6094 vcp->vc_factor = 4; /* could be longer too... */ 6095 } 6096 } 6097 # endif 6098 if (vcp->vc_type == CONV_NONE) 6099 return FAIL; 6100 6101 return OK; 6102 } 6103 6104 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \ 6105 || defined(MSDOS) || defined(PROTO) 6106 /* 6107 * Do conversion on typed input characters in-place. 6108 * The input and output are not NUL terminated! 6109 * Returns the length after conversion. 6110 */ 6111 int 6112 convert_input(ptr, len, maxlen) 6113 char_u *ptr; 6114 int len; 6115 int maxlen; 6116 { 6117 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 6118 } 6119 #endif 6120 6121 /* 6122 * Like convert_input(), but when there is an incomplete byte sequence at the 6123 * end return that as an allocated string in "restp" and set "*restlenp" to 6124 * the length. If "restp" is NULL it is not used. 6125 */ 6126 int 6127 convert_input_safe(ptr, len, maxlen, restp, restlenp) 6128 char_u *ptr; 6129 int len; 6130 int maxlen; 6131 char_u **restp; 6132 int *restlenp; 6133 { 6134 char_u *d; 6135 int dlen = len; 6136 int unconvertlen = 0; 6137 6138 d = string_convert_ext(&input_conv, ptr, &dlen, 6139 restp == NULL ? NULL : &unconvertlen); 6140 if (d != NULL) 6141 { 6142 if (dlen <= maxlen) 6143 { 6144 if (unconvertlen > 0) 6145 { 6146 /* Move the unconverted characters to allocated memory. */ 6147 *restp = alloc(unconvertlen); 6148 if (*restp != NULL) 6149 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 6150 *restlenp = unconvertlen; 6151 } 6152 mch_memmove(ptr, d, dlen); 6153 } 6154 else 6155 /* result is too long, keep the unconverted text (the caller must 6156 * have done something wrong!) */ 6157 dlen = len; 6158 vim_free(d); 6159 } 6160 return dlen; 6161 } 6162 6163 /* 6164 * Convert text "ptr[*lenp]" according to "vcp". 6165 * Returns the result in allocated memory and sets "*lenp". 6166 * When "lenp" is NULL, use NUL terminated strings. 6167 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 6168 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 6169 */ 6170 char_u * 6171 string_convert(vcp, ptr, lenp) 6172 vimconv_T *vcp; 6173 char_u *ptr; 6174 int *lenp; 6175 { 6176 return string_convert_ext(vcp, ptr, lenp, NULL); 6177 } 6178 6179 /* 6180 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 6181 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 6182 * set to the number of remaining bytes. 6183 */ 6184 char_u * 6185 string_convert_ext(vcp, ptr, lenp, unconvlenp) 6186 vimconv_T *vcp; 6187 char_u *ptr; 6188 int *lenp; 6189 int *unconvlenp; 6190 { 6191 char_u *retval = NULL; 6192 char_u *d; 6193 int len; 6194 int i; 6195 int l; 6196 int c; 6197 6198 if (lenp == NULL) 6199 len = (int)STRLEN(ptr); 6200 else 6201 len = *lenp; 6202 if (len == 0) 6203 return vim_strsave((char_u *)""); 6204 6205 switch (vcp->vc_type) 6206 { 6207 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */ 6208 retval = alloc(len * 2 + 1); 6209 if (retval == NULL) 6210 break; 6211 d = retval; 6212 for (i = 0; i < len; ++i) 6213 { 6214 c = ptr[i]; 6215 if (c < 0x80) 6216 *d++ = c; 6217 else 6218 { 6219 *d++ = 0xc0 + ((unsigned)c >> 6); 6220 *d++ = 0x80 + (c & 0x3f); 6221 } 6222 } 6223 *d = NUL; 6224 if (lenp != NULL) 6225 *lenp = (int)(d - retval); 6226 break; 6227 6228 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */ 6229 retval = alloc(len * 3 + 1); 6230 if (retval == NULL) 6231 break; 6232 d = retval; 6233 for (i = 0; i < len; ++i) 6234 { 6235 c = ptr[i]; 6236 switch (c) 6237 { 6238 case 0xa4: c = 0x20ac; break; /* euro */ 6239 case 0xa6: c = 0x0160; break; /* S hat */ 6240 case 0xa8: c = 0x0161; break; /* S -hat */ 6241 case 0xb4: c = 0x017d; break; /* Z hat */ 6242 case 0xb8: c = 0x017e; break; /* Z -hat */ 6243 case 0xbc: c = 0x0152; break; /* OE */ 6244 case 0xbd: c = 0x0153; break; /* oe */ 6245 case 0xbe: c = 0x0178; break; /* Y */ 6246 } 6247 d += utf_char2bytes(c, d); 6248 } 6249 *d = NUL; 6250 if (lenp != NULL) 6251 *lenp = (int)(d - retval); 6252 break; 6253 6254 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */ 6255 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */ 6256 retval = alloc(len + 1); 6257 if (retval == NULL) 6258 break; 6259 d = retval; 6260 for (i = 0; i < len; ++i) 6261 { 6262 l = utf_ptr2len_len(ptr + i, len - i); 6263 if (l == 0) 6264 *d++ = NUL; 6265 else if (l == 1) 6266 { 6267 int l_w = utf8len_tab_zero[ptr[i]]; 6268 6269 if (l_w == 0) 6270 { 6271 /* Illegal utf-8 byte cannot be converted */ 6272 vim_free(retval); 6273 return NULL; 6274 } 6275 if (unconvlenp != NULL && l_w > len - i) 6276 { 6277 /* Incomplete sequence at the end. */ 6278 *unconvlenp = len - i; 6279 break; 6280 } 6281 *d++ = ptr[i]; 6282 } 6283 else 6284 { 6285 c = utf_ptr2char(ptr + i); 6286 if (vcp->vc_type == CONV_TO_LATIN9) 6287 switch (c) 6288 { 6289 case 0x20ac: c = 0xa4; break; /* euro */ 6290 case 0x0160: c = 0xa6; break; /* S hat */ 6291 case 0x0161: c = 0xa8; break; /* S -hat */ 6292 case 0x017d: c = 0xb4; break; /* Z hat */ 6293 case 0x017e: c = 0xb8; break; /* Z -hat */ 6294 case 0x0152: c = 0xbc; break; /* OE */ 6295 case 0x0153: c = 0xbd; break; /* oe */ 6296 case 0x0178: c = 0xbe; break; /* Y */ 6297 case 0xa4: 6298 case 0xa6: 6299 case 0xa8: 6300 case 0xb4: 6301 case 0xb8: 6302 case 0xbc: 6303 case 0xbd: 6304 case 0xbe: c = 0x100; break; /* not in latin9 */ 6305 } 6306 if (!utf_iscomposing(c)) /* skip composing chars */ 6307 { 6308 if (c < 0x100) 6309 *d++ = c; 6310 else if (vcp->vc_fail) 6311 { 6312 vim_free(retval); 6313 return NULL; 6314 } 6315 else 6316 { 6317 *d++ = 0xbf; 6318 if (utf_char2cells(c) > 1) 6319 *d++ = '?'; 6320 } 6321 } 6322 i += l - 1; 6323 } 6324 } 6325 *d = NUL; 6326 if (lenp != NULL) 6327 *lenp = (int)(d - retval); 6328 break; 6329 6330 # ifdef MACOS_CONVERT 6331 case CONV_MAC_LATIN1: 6332 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6333 'm', 'l', unconvlenp); 6334 break; 6335 6336 case CONV_LATIN1_MAC: 6337 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6338 'l', 'm', unconvlenp); 6339 break; 6340 6341 case CONV_MAC_UTF8: 6342 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6343 'm', 'u', unconvlenp); 6344 break; 6345 6346 case CONV_UTF8_MAC: 6347 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6348 'u', 'm', unconvlenp); 6349 break; 6350 # endif 6351 6352 # ifdef USE_ICONV 6353 case CONV_ICONV: /* conversion with output_conv.vc_fd */ 6354 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 6355 break; 6356 # endif 6357 # ifdef WIN3264 6358 case CONV_CODEPAGE: /* codepage -> codepage */ 6359 { 6360 int retlen; 6361 int tmp_len; 6362 short_u *tmp; 6363 6364 /* 1. codepage/UTF-8 -> ucs-2. */ 6365 if (vcp->vc_cpfrom == 0) 6366 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 6367 else 6368 { 6369 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 6370 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 6371 ptr, len, 0, 0); 6372 if (tmp_len == 0 6373 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 6374 { 6375 if (lenp != NULL) 6376 *lenp = 0; 6377 if (unconvlenp != NULL) 6378 *unconvlenp = len; 6379 retval = alloc(1); 6380 if (retval) 6381 retval[0] = NUL; 6382 return retval; 6383 } 6384 } 6385 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len); 6386 if (tmp == NULL) 6387 break; 6388 if (vcp->vc_cpfrom == 0) 6389 utf8_to_utf16(ptr, len, tmp, unconvlenp); 6390 else 6391 MultiByteToWideChar(vcp->vc_cpfrom, 0, ptr, len, tmp, tmp_len); 6392 6393 /* 2. ucs-2 -> codepage/UTF-8. */ 6394 if (vcp->vc_cpto == 0) 6395 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 6396 else 6397 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 6398 tmp, tmp_len, 0, 0, 0, 0); 6399 retval = alloc(retlen + 1); 6400 if (retval != NULL) 6401 { 6402 if (vcp->vc_cpto == 0) 6403 utf16_to_utf8(tmp, tmp_len, retval); 6404 else 6405 WideCharToMultiByte(vcp->vc_cpto, 0, 6406 tmp, tmp_len, retval, retlen, 0, 0); 6407 retval[retlen] = NUL; 6408 if (lenp != NULL) 6409 *lenp = retlen; 6410 } 6411 vim_free(tmp); 6412 break; 6413 } 6414 # endif 6415 } 6416 6417 return retval; 6418 } 6419 #endif 6420