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 # include <windows.h> 87 # ifdef WIN32 88 # undef WIN32 /* Some windows.h define WIN32, we don't want that here. */ 89 # endif 90 #endif 91 92 #if (defined(WIN3264) || defined(WIN32UNIX)) && !defined(__MINGW32__) 93 # include <winnls.h> 94 #endif 95 96 #ifdef FEAT_GUI_X11 97 # include <X11/Intrinsic.h> 98 #endif 99 #ifdef X_LOCALE 100 #include <X11/Xlocale.h> 101 #endif 102 103 #if defined(FEAT_GUI_GTK) && defined(FEAT_XIM) 104 # include <gdk/gdkkeysyms.h> 105 # ifdef WIN3264 106 # include <gdk/gdkwin32.h> 107 # else 108 # include <gdk/gdkx.h> 109 # endif 110 #endif 111 112 #ifdef HAVE_WCHAR_H 113 # include <wchar.h> 114 #endif 115 116 #if 0 117 /* This has been disabled, because several people reported problems with the 118 * wcwidth() and iswprint() library functions, esp. for Hebrew. */ 119 # ifdef __STDC_ISO_10646__ 120 # define USE_WCHAR_FUNCTIONS 121 # endif 122 #endif 123 124 #if defined(FEAT_MBYTE) || defined(PROTO) 125 126 static int enc_canon_search __ARGS((char_u *name)); 127 static int dbcs_char2len __ARGS((int c)); 128 static int dbcs_char2bytes __ARGS((int c, char_u *buf)); 129 static int dbcs_ptr2len __ARGS((char_u *p)); 130 static int dbcs_ptr2len_len __ARGS((char_u *p, int size)); 131 static int utf_ptr2cells_len __ARGS((char_u *p, int size)); 132 static int dbcs_char2cells __ARGS((int c)); 133 static int dbcs_ptr2cells_len __ARGS((char_u *p, int size)); 134 static int dbcs_ptr2char __ARGS((char_u *p)); 135 static int utf_safe_read_char_adv __ARGS((char_u **s, size_t *n)); 136 137 /* 138 * Lookup table to quickly get the length in bytes of a UTF-8 character from 139 * the first byte of a UTF-8 string. 140 * Bytes which are illegal when used as the first byte have a 1. 141 * The NUL byte has length 1. 142 */ 143 static char utf8len_tab[256] = 144 { 145 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 146 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 147 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 148 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 149 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 150 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 151 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, 152 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, 153 }; 154 155 /* 156 * Like utf8len_tab above, but using a zero for illegal lead bytes. 157 */ 158 static char utf8len_tab_zero[256] = 159 { 160 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 161 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 162 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 163 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 164 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, 165 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, 166 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, 167 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, 168 }; 169 170 /* 171 * XIM often causes trouble. Define XIM_DEBUG to get a log of XIM callbacks 172 * in the "xim.log" file. 173 */ 174 /* #define XIM_DEBUG */ 175 #ifdef XIM_DEBUG 176 static void 177 xim_log(char *s, ...) 178 { 179 va_list arglist; 180 static FILE *fd = NULL; 181 182 if (fd == (FILE *)-1) 183 return; 184 if (fd == NULL) 185 { 186 fd = mch_fopen("xim.log", "w"); 187 if (fd == NULL) 188 { 189 EMSG("Cannot open xim.log"); 190 fd = (FILE *)-1; 191 return; 192 } 193 } 194 195 va_start(arglist, s); 196 vfprintf(fd, s, arglist); 197 va_end(arglist); 198 } 199 #endif 200 201 #endif 202 203 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 204 /* 205 * Canonical encoding names and their properties. 206 * "iso-8859-n" is handled by enc_canonize() directly. 207 */ 208 static struct 209 { char *name; int prop; int codepage;} 210 enc_canon_table[] = 211 { 212 #define IDX_LATIN_1 0 213 {"latin1", ENC_8BIT + ENC_LATIN1, 1252}, 214 #define IDX_ISO_2 1 215 {"iso-8859-2", ENC_8BIT, 0}, 216 #define IDX_ISO_3 2 217 {"iso-8859-3", ENC_8BIT, 0}, 218 #define IDX_ISO_4 3 219 {"iso-8859-4", ENC_8BIT, 0}, 220 #define IDX_ISO_5 4 221 {"iso-8859-5", ENC_8BIT, 0}, 222 #define IDX_ISO_6 5 223 {"iso-8859-6", ENC_8BIT, 0}, 224 #define IDX_ISO_7 6 225 {"iso-8859-7", ENC_8BIT, 0}, 226 #define IDX_ISO_8 7 227 {"iso-8859-8", ENC_8BIT, 0}, 228 #define IDX_ISO_9 8 229 {"iso-8859-9", ENC_8BIT, 0}, 230 #define IDX_ISO_10 9 231 {"iso-8859-10", ENC_8BIT, 0}, 232 #define IDX_ISO_11 10 233 {"iso-8859-11", ENC_8BIT, 0}, 234 #define IDX_ISO_13 11 235 {"iso-8859-13", ENC_8BIT, 0}, 236 #define IDX_ISO_14 12 237 {"iso-8859-14", ENC_8BIT, 0}, 238 #define IDX_ISO_15 13 239 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0}, 240 #define IDX_KOI8_R 14 241 {"koi8-r", ENC_8BIT, 0}, 242 #define IDX_KOI8_U 15 243 {"koi8-u", ENC_8BIT, 0}, 244 #define IDX_UTF8 16 245 {"utf-8", ENC_UNICODE, 0}, 246 #define IDX_UCS2 17 247 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0}, 248 #define IDX_UCS2LE 18 249 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0}, 250 #define IDX_UTF16 19 251 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0}, 252 #define IDX_UTF16LE 20 253 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0}, 254 #define IDX_UCS4 21 255 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0}, 256 #define IDX_UCS4LE 22 257 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0}, 258 259 /* For debugging DBCS encoding on Unix. */ 260 #define IDX_DEBUG 23 261 {"debug", ENC_DBCS, DBCS_DEBUG}, 262 #define IDX_EUC_JP 24 263 {"euc-jp", ENC_DBCS, DBCS_JPNU}, 264 #define IDX_SJIS 25 265 {"sjis", ENC_DBCS, DBCS_JPN}, 266 #define IDX_EUC_KR 26 267 {"euc-kr", ENC_DBCS, DBCS_KORU}, 268 #define IDX_EUC_CN 27 269 {"euc-cn", ENC_DBCS, DBCS_CHSU}, 270 #define IDX_EUC_TW 28 271 {"euc-tw", ENC_DBCS, DBCS_CHTU}, 272 #define IDX_BIG5 29 273 {"big5", ENC_DBCS, DBCS_CHT}, 274 275 /* MS-DOS and MS-Windows codepages are included here, so that they can be 276 * used on Unix too. Most of them are similar to ISO-8859 encodings, but 277 * not exactly the same. */ 278 #define IDX_CP437 30 279 {"cp437", ENC_8BIT, 437}, /* like iso-8859-1 */ 280 #define IDX_CP737 31 281 {"cp737", ENC_8BIT, 737}, /* like iso-8859-7 */ 282 #define IDX_CP775 32 283 {"cp775", ENC_8BIT, 775}, /* Baltic */ 284 #define IDX_CP850 33 285 {"cp850", ENC_8BIT, 850}, /* like iso-8859-4 */ 286 #define IDX_CP852 34 287 {"cp852", ENC_8BIT, 852}, /* like iso-8859-1 */ 288 #define IDX_CP855 35 289 {"cp855", ENC_8BIT, 855}, /* like iso-8859-2 */ 290 #define IDX_CP857 36 291 {"cp857", ENC_8BIT, 857}, /* like iso-8859-5 */ 292 #define IDX_CP860 37 293 {"cp860", ENC_8BIT, 860}, /* like iso-8859-9 */ 294 #define IDX_CP861 38 295 {"cp861", ENC_8BIT, 861}, /* like iso-8859-1 */ 296 #define IDX_CP862 39 297 {"cp862", ENC_8BIT, 862}, /* like iso-8859-1 */ 298 #define IDX_CP863 40 299 {"cp863", ENC_8BIT, 863}, /* like iso-8859-8 */ 300 #define IDX_CP865 41 301 {"cp865", ENC_8BIT, 865}, /* like iso-8859-1 */ 302 #define IDX_CP866 42 303 {"cp866", ENC_8BIT, 866}, /* like iso-8859-5 */ 304 #define IDX_CP869 43 305 {"cp869", ENC_8BIT, 869}, /* like iso-8859-7 */ 306 #define IDX_CP874 44 307 {"cp874", ENC_8BIT, 874}, /* Thai */ 308 #define IDX_CP932 45 309 {"cp932", ENC_DBCS, DBCS_JPN}, 310 #define IDX_CP936 46 311 {"cp936", ENC_DBCS, DBCS_CHS}, 312 #define IDX_CP949 47 313 {"cp949", ENC_DBCS, DBCS_KOR}, 314 #define IDX_CP950 48 315 {"cp950", ENC_DBCS, DBCS_CHT}, 316 #define IDX_CP1250 49 317 {"cp1250", ENC_8BIT, 1250}, /* Czech, Polish, etc. */ 318 #define IDX_CP1251 50 319 {"cp1251", ENC_8BIT, 1251}, /* Cyrillic */ 320 /* cp1252 is considered to be equal to latin1 */ 321 #define IDX_CP1253 51 322 {"cp1253", ENC_8BIT, 1253}, /* Greek */ 323 #define IDX_CP1254 52 324 {"cp1254", ENC_8BIT, 1254}, /* Turkish */ 325 #define IDX_CP1255 53 326 {"cp1255", ENC_8BIT, 1255}, /* Hebrew */ 327 #define IDX_CP1256 54 328 {"cp1256", ENC_8BIT, 1256}, /* Arabic */ 329 #define IDX_CP1257 55 330 {"cp1257", ENC_8BIT, 1257}, /* Baltic */ 331 #define IDX_CP1258 56 332 {"cp1258", ENC_8BIT, 1258}, /* Vietnamese */ 333 334 #define IDX_MACROMAN 57 335 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, /* Mac OS */ 336 #define IDX_DECMCS 58 337 {"dec-mcs", ENC_8BIT, 0}, /* DEC MCS */ 338 #define IDX_HPROMAN8 59 339 {"hp-roman8", ENC_8BIT, 0}, /* HP Roman8 */ 340 #define IDX_COUNT 60 341 }; 342 343 /* 344 * Aliases for encoding names. 345 */ 346 static struct 347 { char *name; int canon;} 348 enc_alias_table[] = 349 { 350 {"ansi", IDX_LATIN_1}, 351 {"iso-8859-1", IDX_LATIN_1}, 352 {"latin2", IDX_ISO_2}, 353 {"latin3", IDX_ISO_3}, 354 {"latin4", IDX_ISO_4}, 355 {"cyrillic", IDX_ISO_5}, 356 {"arabic", IDX_ISO_6}, 357 {"greek", IDX_ISO_7}, 358 #ifdef WIN3264 359 {"hebrew", IDX_CP1255}, 360 #else 361 {"hebrew", IDX_ISO_8}, 362 #endif 363 {"latin5", IDX_ISO_9}, 364 {"turkish", IDX_ISO_9}, /* ? */ 365 {"latin6", IDX_ISO_10}, 366 {"nordic", IDX_ISO_10}, /* ? */ 367 {"thai", IDX_ISO_11}, /* ? */ 368 {"latin7", IDX_ISO_13}, 369 {"latin8", IDX_ISO_14}, 370 {"latin9", IDX_ISO_15}, 371 {"utf8", IDX_UTF8}, 372 {"unicode", IDX_UCS2}, 373 {"ucs2", IDX_UCS2}, 374 {"ucs2be", IDX_UCS2}, 375 {"ucs-2be", IDX_UCS2}, 376 {"ucs2le", IDX_UCS2LE}, 377 {"utf16", IDX_UTF16}, 378 {"utf16be", IDX_UTF16}, 379 {"utf-16be", IDX_UTF16}, 380 {"utf16le", IDX_UTF16LE}, 381 {"ucs4", IDX_UCS4}, 382 {"ucs4be", IDX_UCS4}, 383 {"ucs-4be", IDX_UCS4}, 384 {"ucs4le", IDX_UCS4LE}, 385 {"utf32", IDX_UCS4}, 386 {"utf-32", IDX_UCS4}, 387 {"utf32be", IDX_UCS4}, 388 {"utf-32be", IDX_UCS4}, 389 {"utf32le", IDX_UCS4LE}, 390 {"utf-32le", IDX_UCS4LE}, 391 {"932", IDX_CP932}, 392 {"949", IDX_CP949}, 393 {"936", IDX_CP936}, 394 {"gbk", IDX_CP936}, 395 {"950", IDX_CP950}, 396 {"eucjp", IDX_EUC_JP}, 397 {"unix-jis", IDX_EUC_JP}, 398 {"ujis", IDX_EUC_JP}, 399 {"shift-jis", IDX_SJIS}, 400 {"euckr", IDX_EUC_KR}, 401 {"5601", IDX_EUC_KR}, /* Sun: KS C 5601 */ 402 {"euccn", IDX_EUC_CN}, 403 {"gb2312", IDX_EUC_CN}, 404 {"euctw", IDX_EUC_TW}, 405 #if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS) 406 {"japan", IDX_CP932}, 407 {"korea", IDX_CP949}, 408 {"prc", IDX_CP936}, 409 {"chinese", IDX_CP936}, 410 {"taiwan", IDX_CP950}, 411 {"big5", IDX_CP950}, 412 #else 413 {"japan", IDX_EUC_JP}, 414 {"korea", IDX_EUC_KR}, 415 {"prc", IDX_EUC_CN}, 416 {"chinese", IDX_EUC_CN}, 417 {"taiwan", IDX_EUC_TW}, 418 {"cp950", IDX_BIG5}, 419 {"950", IDX_BIG5}, 420 #endif 421 {"mac", IDX_MACROMAN}, 422 {"mac-roman", IDX_MACROMAN}, 423 {NULL, 0} 424 }; 425 426 #ifndef CP_UTF8 427 # define CP_UTF8 65001 /* magic number from winnls.h */ 428 #endif 429 430 /* 431 * Find encoding "name" in the list of canonical encoding names. 432 * Returns -1 if not found. 433 */ 434 static int 435 enc_canon_search(name) 436 char_u *name; 437 { 438 int i; 439 440 for (i = 0; i < IDX_COUNT; ++i) 441 if (STRCMP(name, enc_canon_table[i].name) == 0) 442 return i; 443 return -1; 444 } 445 446 #endif 447 448 #if defined(FEAT_MBYTE) || defined(PROTO) 449 450 /* 451 * Find canonical encoding "name" in the list and return its properties. 452 * Returns 0 if not found. 453 */ 454 int 455 enc_canon_props(name) 456 char_u *name; 457 { 458 int i; 459 460 i = enc_canon_search(name); 461 if (i >= 0) 462 return enc_canon_table[i].prop; 463 #ifdef WIN3264 464 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2])) 465 { 466 CPINFO cpinfo; 467 468 /* Get info on this codepage to find out what it is. */ 469 if (GetCPInfo(atoi(name + 2), &cpinfo) != 0) 470 { 471 if (cpinfo.MaxCharSize == 1) /* some single-byte encoding */ 472 return ENC_8BIT; 473 if (cpinfo.MaxCharSize == 2 474 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 475 /* must be a DBCS encoding */ 476 return ENC_DBCS; 477 } 478 return 0; 479 } 480 #endif 481 if (STRNCMP(name, "2byte-", 6) == 0) 482 return ENC_DBCS; 483 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0) 484 return ENC_8BIT; 485 return 0; 486 } 487 488 /* 489 * Set up for using multi-byte characters. 490 * Called in three cases: 491 * - by main() to initialize (p_enc == NULL) 492 * - by set_init_1() after 'encoding' was set to its default. 493 * - by do_set() when 'encoding' has been set. 494 * p_enc must have been passed through enc_canonize() already. 495 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags. 496 * Fills mb_bytelen_tab[] and returns NULL when there are no problems. 497 * When there is something wrong: Returns an error message and doesn't change 498 * anything. 499 */ 500 char_u * 501 mb_init() 502 { 503 int i; 504 int idx; 505 int n; 506 int enc_dbcs_new = 0; 507 #if defined(USE_ICONV) && !defined(WIN3264) && !defined(WIN32UNIX) \ 508 && !defined(MACOS) 509 # define LEN_FROM_CONV 510 vimconv_T vimconv; 511 char_u *p; 512 #endif 513 514 if (p_enc == NULL) 515 { 516 /* Just starting up: set the whole table to one's. */ 517 for (i = 0; i < 256; ++i) 518 mb_bytelen_tab[i] = 1; 519 input_conv.vc_type = CONV_NONE; 520 input_conv.vc_factor = 1; 521 output_conv.vc_type = CONV_NONE; 522 return NULL; 523 } 524 525 #ifdef WIN3264 526 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2])) 527 { 528 CPINFO cpinfo; 529 530 /* Get info on this codepage to find out what it is. */ 531 if (GetCPInfo(atoi(p_enc + 2), &cpinfo) != 0) 532 { 533 if (cpinfo.MaxCharSize == 1) 534 { 535 /* some single-byte encoding */ 536 enc_unicode = 0; 537 enc_utf8 = FALSE; 538 } 539 else if (cpinfo.MaxCharSize == 2 540 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 541 { 542 /* must be a DBCS encoding, check below */ 543 enc_dbcs_new = atoi(p_enc + 2); 544 } 545 else 546 goto codepage_invalid; 547 } 548 else if (GetLastError() == ERROR_INVALID_PARAMETER) 549 { 550 codepage_invalid: 551 return (char_u *)N_("E543: Not a valid codepage"); 552 } 553 } 554 #endif 555 else if (STRNCMP(p_enc, "8bit-", 5) == 0 556 || STRNCMP(p_enc, "iso-8859-", 9) == 0) 557 { 558 /* Accept any "8bit-" or "iso-8859-" name. */ 559 enc_unicode = 0; 560 enc_utf8 = FALSE; 561 } 562 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 563 { 564 #ifdef WIN3264 565 /* Windows: accept only valid codepage numbers, check below. */ 566 if (p_enc[6] != 'c' || p_enc[7] != 'p' 567 || (enc_dbcs_new = atoi(p_enc + 8)) == 0) 568 return e_invarg; 569 #else 570 /* Unix: accept any "2byte-" name, assume current locale. */ 571 enc_dbcs_new = DBCS_2BYTE; 572 #endif 573 } 574 else if ((idx = enc_canon_search(p_enc)) >= 0) 575 { 576 i = enc_canon_table[idx].prop; 577 if (i & ENC_UNICODE) 578 { 579 /* Unicode */ 580 enc_utf8 = TRUE; 581 if (i & (ENC_2BYTE | ENC_2WORD)) 582 enc_unicode = 2; 583 else if (i & ENC_4BYTE) 584 enc_unicode = 4; 585 else 586 enc_unicode = 0; 587 } 588 else if (i & ENC_DBCS) 589 { 590 /* 2byte, handle below */ 591 enc_dbcs_new = enc_canon_table[idx].codepage; 592 } 593 else 594 { 595 /* Must be 8-bit. */ 596 enc_unicode = 0; 597 enc_utf8 = FALSE; 598 } 599 } 600 else /* Don't know what encoding this is, reject it. */ 601 return e_invarg; 602 603 if (enc_dbcs_new != 0) 604 { 605 #ifdef WIN3264 606 /* Check if the DBCS code page is OK. */ 607 if (!IsValidCodePage(enc_dbcs_new)) 608 goto codepage_invalid; 609 #endif 610 enc_unicode = 0; 611 enc_utf8 = FALSE; 612 } 613 enc_dbcs = enc_dbcs_new; 614 has_mbyte = (enc_dbcs != 0 || enc_utf8); 615 616 #ifdef WIN3264 617 enc_codepage = encname2codepage(p_enc); 618 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0); 619 #endif 620 621 /* Detect an encoding that uses latin1 characters. */ 622 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0 623 || STRCMP(p_enc, "iso-8859-15") == 0); 624 625 /* 626 * Set the function pointers. 627 */ 628 if (enc_utf8) 629 { 630 mb_ptr2len = utfc_ptr2len; 631 mb_ptr2len_len = utfc_ptr2len_len; 632 mb_char2len = utf_char2len; 633 mb_char2bytes = utf_char2bytes; 634 mb_ptr2cells = utf_ptr2cells; 635 mb_ptr2cells_len = utf_ptr2cells_len; 636 mb_char2cells = utf_char2cells; 637 mb_off2cells = utf_off2cells; 638 mb_ptr2char = utf_ptr2char; 639 mb_head_off = utf_head_off; 640 } 641 else if (enc_dbcs != 0) 642 { 643 mb_ptr2len = dbcs_ptr2len; 644 mb_ptr2len_len = dbcs_ptr2len_len; 645 mb_char2len = dbcs_char2len; 646 mb_char2bytes = dbcs_char2bytes; 647 mb_ptr2cells = dbcs_ptr2cells; 648 mb_ptr2cells_len = dbcs_ptr2cells_len; 649 mb_char2cells = dbcs_char2cells; 650 mb_off2cells = dbcs_off2cells; 651 mb_ptr2char = dbcs_ptr2char; 652 mb_head_off = dbcs_head_off; 653 } 654 else 655 { 656 mb_ptr2len = latin_ptr2len; 657 mb_ptr2len_len = latin_ptr2len_len; 658 mb_char2len = latin_char2len; 659 mb_char2bytes = latin_char2bytes; 660 mb_ptr2cells = latin_ptr2cells; 661 mb_ptr2cells_len = latin_ptr2cells_len; 662 mb_char2cells = latin_char2cells; 663 mb_off2cells = latin_off2cells; 664 mb_ptr2char = latin_ptr2char; 665 mb_head_off = latin_head_off; 666 } 667 668 /* 669 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN(). 670 */ 671 #ifdef LEN_FROM_CONV 672 /* When 'encoding' is different from the current locale mblen() won't 673 * work. Use conversion to "utf-8" instead. */ 674 vimconv.vc_type = CONV_NONE; 675 if (enc_dbcs) 676 { 677 p = enc_locale(); 678 if (p == NULL || STRCMP(p, p_enc) != 0) 679 { 680 convert_setup(&vimconv, p_enc, (char_u *)"utf-8"); 681 vimconv.vc_fail = TRUE; 682 } 683 vim_free(p); 684 } 685 #endif 686 687 for (i = 0; i < 256; ++i) 688 { 689 /* Our own function to reliably check the length of UTF-8 characters, 690 * independent of mblen(). */ 691 if (enc_utf8) 692 n = utf8len_tab[i]; 693 else if (enc_dbcs == 0) 694 n = 1; 695 else 696 { 697 #if defined(WIN3264) || defined(WIN32UNIX) 698 /* enc_dbcs is set by setting 'fileencoding'. It becomes a Windows 699 * CodePage identifier, which we can pass directly in to Windows 700 * API */ 701 n = IsDBCSLeadByteEx(enc_dbcs, (BYTE)i) ? 2 : 1; 702 #else 703 # if defined(MACOS) || defined(__amigaos4__) 704 /* 705 * if mblen() is not available, character which MSB is turned on 706 * are treated as leading byte character. (note : This assumption 707 * is not always true.) 708 */ 709 n = (i & 0x80) ? 2 : 1; 710 # else 711 char buf[MB_MAXBYTES + 1]; 712 # ifdef X_LOCALE 713 # ifndef mblen 714 # define mblen _Xmblen 715 # endif 716 # endif 717 if (i == NUL) /* just in case mblen() can't handle "" */ 718 n = 1; 719 else 720 { 721 buf[0] = i; 722 buf[1] = 0; 723 #ifdef LEN_FROM_CONV 724 if (vimconv.vc_type != CONV_NONE) 725 { 726 /* 727 * string_convert() should fail when converting the first 728 * byte of a double-byte character. 729 */ 730 p = string_convert(&vimconv, (char_u *)buf, NULL); 731 if (p != NULL) 732 { 733 vim_free(p); 734 n = 1; 735 } 736 else 737 n = 2; 738 } 739 else 740 #endif 741 { 742 /* 743 * mblen() should return -1 for invalid (means the leading 744 * multibyte) character. However there are some platforms 745 * where mblen() returns 0 for invalid character. 746 * Therefore, following condition includes 0. 747 */ 748 ignored = mblen(NULL, 0); /* First reset the state. */ 749 if (mblen(buf, (size_t)1) <= 0) 750 n = 2; 751 else 752 n = 1; 753 } 754 } 755 # endif 756 #endif 757 } 758 759 mb_bytelen_tab[i] = n; 760 } 761 762 #ifdef LEN_FROM_CONV 763 convert_setup(&vimconv, NULL, NULL); 764 #endif 765 766 /* The cell width depends on the type of multi-byte characters. */ 767 (void)init_chartab(); 768 769 /* When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] */ 770 screenalloc(FALSE); 771 772 /* When using Unicode, set default for 'fileencodings'. */ 773 if (enc_utf8 && !option_was_set((char_u *)"fencs")) 774 set_string_option_direct((char_u *)"fencs", -1, 775 (char_u *)"ucs-bom,utf-8,default,latin1", OPT_FREE, 0); 776 777 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT) 778 /* GNU gettext 0.10.37 supports this feature: set the codeset used for 779 * translated messages independently from the current locale. */ 780 (void)bind_textdomain_codeset(VIMPACKAGE, 781 enc_utf8 ? "utf-8" : (char *)p_enc); 782 #endif 783 784 #ifdef WIN32 785 /* When changing 'encoding' while starting up, then convert the command 786 * line arguments from the active codepage to 'encoding'. */ 787 if (starting != 0) 788 fix_arg_enc(); 789 #endif 790 791 #ifdef FEAT_AUTOCMD 792 /* Fire an autocommand to let people do custom font setup. This must be 793 * after Vim has been setup for the new encoding. */ 794 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); 795 #endif 796 797 #ifdef FEAT_SPELL 798 /* Need to reload spell dictionaries */ 799 spell_reload(); 800 #endif 801 802 return NULL; 803 } 804 805 /* 806 * Return the size of the BOM for the current buffer: 807 * 0 - no BOM 808 * 2 - UCS-2 or UTF-16 BOM 809 * 4 - UCS-4 BOM 810 * 3 - UTF-8 BOM 811 */ 812 int 813 bomb_size() 814 { 815 int n = 0; 816 817 if (curbuf->b_p_bomb && !curbuf->b_p_bin) 818 { 819 if (*curbuf->b_p_fenc == NUL) 820 { 821 if (enc_utf8) 822 { 823 if (enc_unicode != 0) 824 n = enc_unicode; 825 else 826 n = 3; 827 } 828 } 829 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) 830 n = 3; 831 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 832 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) 833 n = 2; 834 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) 835 n = 4; 836 } 837 return n; 838 } 839 840 /* 841 * Remove all BOM from "s" by moving remaining text. 842 */ 843 void 844 remove_bom(s) 845 char_u *s; 846 { 847 if (enc_utf8) 848 { 849 char_u *p = s; 850 851 while ((p = vim_strbyte(p, 0xef)) != NULL) 852 { 853 if (p[1] == 0xbb && p[2] == 0xbf) 854 STRMOVE(p, p + 3); 855 else 856 ++p; 857 } 858 } 859 } 860 861 /* 862 * Get class of pointer: 863 * 0 for blank or NUL 864 * 1 for punctuation 865 * 2 for an (ASCII) word character 866 * >2 for other word characters 867 */ 868 int 869 mb_get_class(p) 870 char_u *p; 871 { 872 if (MB_BYTE2LEN(p[0]) == 1) 873 { 874 if (p[0] == NUL || vim_iswhite(p[0])) 875 return 0; 876 if (vim_iswordc(p[0])) 877 return 2; 878 return 1; 879 } 880 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) 881 return dbcs_class(p[0], p[1]); 882 if (enc_utf8) 883 return utf_class(utf_ptr2char(p)); 884 return 0; 885 } 886 887 /* 888 * Get class of a double-byte character. This always returns 3 or bigger. 889 * TODO: Should return 1 for punctuation. 890 */ 891 int 892 dbcs_class(lead, trail) 893 unsigned lead; 894 unsigned trail; 895 { 896 switch (enc_dbcs) 897 { 898 /* please add classfy routine for your language in here */ 899 900 case DBCS_JPNU: /* ? */ 901 case DBCS_JPN: 902 { 903 /* JIS code classification */ 904 unsigned char lb = lead; 905 unsigned char tb = trail; 906 907 /* convert process code to JIS */ 908 # if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS) 909 /* process code is SJIS */ 910 if (lb <= 0x9f) 911 lb = (lb - 0x81) * 2 + 0x21; 912 else 913 lb = (lb - 0xc1) * 2 + 0x21; 914 if (tb <= 0x7e) 915 tb -= 0x1f; 916 else if (tb <= 0x9e) 917 tb -= 0x20; 918 else 919 { 920 tb -= 0x7e; 921 lb += 1; 922 } 923 # else 924 /* 925 * XXX: Code page identification can not use with all 926 * system! So, some other encoding information 927 * will be needed. 928 * In japanese: SJIS,EUC,UNICODE,(JIS) 929 * Note that JIS-code system don't use as 930 * process code in most system because it uses 931 * escape sequences(JIS is context depend encoding). 932 */ 933 /* assume process code is JAPANESE-EUC */ 934 lb &= 0x7f; 935 tb &= 0x7f; 936 # endif 937 /* exceptions */ 938 switch (lb << 8 | tb) 939 { 940 case 0x2121: /* ZENKAKU space */ 941 return 0; 942 case 0x2122: /* KU-TEN (Japanese comma) */ 943 case 0x2123: /* TOU-TEN (Japanese period) */ 944 case 0x2124: /* ZENKAKU comma */ 945 case 0x2125: /* ZENKAKU period */ 946 return 1; 947 case 0x213c: /* prolongedsound handled as KATAKANA */ 948 return 13; 949 } 950 /* sieved by KU code */ 951 switch (lb) 952 { 953 case 0x21: 954 case 0x22: 955 /* special symbols */ 956 return 10; 957 case 0x23: 958 /* alpha-numeric */ 959 return 11; 960 case 0x24: 961 /* hiragana */ 962 return 12; 963 case 0x25: 964 /* katakana */ 965 return 13; 966 case 0x26: 967 /* greek */ 968 return 14; 969 case 0x27: 970 /* russian */ 971 return 15; 972 case 0x28: 973 /* lines */ 974 return 16; 975 default: 976 /* kanji */ 977 return 17; 978 } 979 } 980 981 case DBCS_KORU: /* ? */ 982 case DBCS_KOR: 983 { 984 /* KS code classification */ 985 unsigned char c1 = lead; 986 unsigned char c2 = trail; 987 988 /* 989 * 20 : Hangul 990 * 21 : Hanja 991 * 22 : Symbols 992 * 23 : Alpha-numeric/Roman Letter (Full width) 993 * 24 : Hangul Letter(Alphabet) 994 * 25 : Roman Numeral/Greek Letter 995 * 26 : Box Drawings 996 * 27 : Unit Symbols 997 * 28 : Circled/Parenthesized Letter 998 * 29 : Hirigana/Katakana 999 * 30 : Cyrillic Letter 1000 */ 1001 1002 if (c1 >= 0xB0 && c1 <= 0xC8) 1003 /* Hangul */ 1004 return 20; 1005 #if defined(WIN3264) || defined(WIN32UNIX) 1006 else if (c1 <= 0xA0 || c2 <= 0xA0) 1007 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */ 1008 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE 1009 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 1010 */ 1011 return 20; 1012 #endif 1013 1014 else if (c1 >= 0xCA && c1 <= 0xFD) 1015 /* Hanja */ 1016 return 21; 1017 else switch (c1) 1018 { 1019 case 0xA1: 1020 case 0xA2: 1021 /* Symbols */ 1022 return 22; 1023 case 0xA3: 1024 /* Alpha-numeric */ 1025 return 23; 1026 case 0xA4: 1027 /* Hangul Letter(Alphabet) */ 1028 return 24; 1029 case 0xA5: 1030 /* Roman Numeral/Greek Letter */ 1031 return 25; 1032 case 0xA6: 1033 /* Box Drawings */ 1034 return 26; 1035 case 0xA7: 1036 /* Unit Symbols */ 1037 return 27; 1038 case 0xA8: 1039 case 0xA9: 1040 if (c2 <= 0xAF) 1041 return 25; /* Roman Letter */ 1042 else if (c2 >= 0xF6) 1043 return 22; /* Symbols */ 1044 else 1045 /* Circled/Parenthesized Letter */ 1046 return 28; 1047 case 0xAA: 1048 case 0xAB: 1049 /* Hirigana/Katakana */ 1050 return 29; 1051 case 0xAC: 1052 /* Cyrillic Letter */ 1053 return 30; 1054 } 1055 } 1056 default: 1057 break; 1058 } 1059 return 3; 1060 } 1061 1062 /* 1063 * mb_char2len() function pointer. 1064 * Return length in bytes of character "c". 1065 * Returns 1 for a single-byte character. 1066 */ 1067 int 1068 latin_char2len(c) 1069 int c UNUSED; 1070 { 1071 return 1; 1072 } 1073 1074 static int 1075 dbcs_char2len(c) 1076 int c; 1077 { 1078 if (c >= 0x100) 1079 return 2; 1080 return 1; 1081 } 1082 1083 /* 1084 * mb_char2bytes() function pointer. 1085 * Convert a character to its bytes. 1086 * Returns the length in bytes. 1087 */ 1088 int 1089 latin_char2bytes(c, buf) 1090 int c; 1091 char_u *buf; 1092 { 1093 buf[0] = c; 1094 return 1; 1095 } 1096 1097 static int 1098 dbcs_char2bytes(c, buf) 1099 int c; 1100 char_u *buf; 1101 { 1102 if (c >= 0x100) 1103 { 1104 buf[0] = (unsigned)c >> 8; 1105 buf[1] = c; 1106 /* Never use a NUL byte, it causes lots of trouble. It's an invalid 1107 * character anyway. */ 1108 if (buf[1] == NUL) 1109 buf[1] = '\n'; 1110 return 2; 1111 } 1112 buf[0] = c; 1113 return 1; 1114 } 1115 1116 /* 1117 * mb_ptr2len() function pointer. 1118 * Get byte length of character at "*p" but stop at a NUL. 1119 * For UTF-8 this includes following composing characters. 1120 * Returns 0 when *p is NUL. 1121 */ 1122 int 1123 latin_ptr2len(p) 1124 char_u *p; 1125 { 1126 return MB_BYTE2LEN(*p); 1127 } 1128 1129 static int 1130 dbcs_ptr2len(p) 1131 char_u *p; 1132 { 1133 int len; 1134 1135 /* Check if second byte is not missing. */ 1136 len = MB_BYTE2LEN(*p); 1137 if (len == 2 && p[1] == NUL) 1138 len = 1; 1139 return len; 1140 } 1141 1142 /* 1143 * mb_ptr2len_len() function pointer. 1144 * Like mb_ptr2len(), but limit to read "size" bytes. 1145 * Returns 0 for an empty string. 1146 * Returns 1 for an illegal char or an incomplete byte sequence. 1147 */ 1148 int 1149 latin_ptr2len_len(p, size) 1150 char_u *p; 1151 int size; 1152 { 1153 if (size < 1 || *p == NUL) 1154 return 0; 1155 return 1; 1156 } 1157 1158 static int 1159 dbcs_ptr2len_len(p, size) 1160 char_u *p; 1161 int size; 1162 { 1163 int len; 1164 1165 if (size < 1 || *p == NUL) 1166 return 0; 1167 if (size == 1) 1168 return 1; 1169 /* Check that second byte is not missing. */ 1170 len = MB_BYTE2LEN(*p); 1171 if (len == 2 && p[1] == NUL) 1172 len = 1; 1173 return len; 1174 } 1175 1176 struct interval 1177 { 1178 long first; 1179 long last; 1180 }; 1181 static int intable __ARGS((struct interval *table, size_t size, int c)); 1182 1183 /* 1184 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]". 1185 */ 1186 static int 1187 intable(table, size, c) 1188 struct interval *table; 1189 size_t size; 1190 int c; 1191 { 1192 int mid, bot, top; 1193 1194 /* first quick check for Latin1 etc. characters */ 1195 if (c < table[0].first) 1196 return FALSE; 1197 1198 /* binary search in table */ 1199 bot = 0; 1200 top = (int)(size / sizeof(struct interval) - 1); 1201 while (top >= bot) 1202 { 1203 mid = (bot + top) / 2; 1204 if (table[mid].last < c) 1205 bot = mid + 1; 1206 else if (table[mid].first > c) 1207 top = mid - 1; 1208 else 1209 return TRUE; 1210 } 1211 return FALSE; 1212 } 1213 1214 /* 1215 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1216 * Returns 4 or 6 for an unprintable character. 1217 * Is only correct for characters >= 0x80. 1218 * When p_ambw is "double", return 2 for a character with East Asian Width 1219 * class 'A'(mbiguous). 1220 */ 1221 int 1222 utf_char2cells(c) 1223 int c; 1224 { 1225 /* Sorted list of non-overlapping intervals of East Asian double width 1226 * characters, generated with ../runtime/tools/unicode.vim. */ 1227 static struct interval doublewidth[] = 1228 { 1229 {0x1100, 0x115f}, 1230 {0x11a3, 0x11a7}, 1231 {0x11fa, 0x11ff}, 1232 {0x2329, 0x232a}, 1233 {0x2e80, 0x2e99}, 1234 {0x2e9b, 0x2ef3}, 1235 {0x2f00, 0x2fd5}, 1236 {0x2ff0, 0x2ffb}, 1237 {0x3000, 0x3029}, 1238 {0x3030, 0x303e}, 1239 {0x3041, 0x3096}, 1240 {0x309b, 0x30ff}, 1241 {0x3105, 0x312d}, 1242 {0x3131, 0x318e}, 1243 {0x3190, 0x31b7}, 1244 {0x31c0, 0x31e3}, 1245 {0x31f0, 0x321e}, 1246 {0x3220, 0x3247}, 1247 {0x3250, 0x32fe}, 1248 {0x3300, 0x4dbf}, 1249 {0x4e00, 0xa48c}, 1250 {0xa490, 0xa4c6}, 1251 {0xa960, 0xa97c}, 1252 {0xac00, 0xd7a3}, 1253 {0xd7b0, 0xd7c6}, 1254 {0xd7cb, 0xd7fb}, 1255 {0xf900, 0xfaff}, 1256 {0xfe10, 0xfe19}, 1257 {0xfe30, 0xfe52}, 1258 {0xfe54, 0xfe66}, 1259 {0xfe68, 0xfe6b}, 1260 {0xff01, 0xff60}, 1261 {0xffe0, 0xffe6}, 1262 {0x1f200, 0x1f200}, 1263 {0x1f210, 0x1f231}, 1264 {0x1f240, 0x1f248}, 1265 {0x20000, 0x2fffd}, 1266 {0x30000, 0x3fffd} 1267 }; 1268 /* Sorted list of non-overlapping intervals of East Asian Ambiguous 1269 * characters, generated with ../runtime/tools/unicode.vim. */ 1270 static struct interval ambiguous[] = 1271 { 1272 {0x00a1, 0x00a1}, 1273 {0x00a4, 0x00a4}, 1274 {0x00a7, 0x00a8}, 1275 {0x00aa, 0x00aa}, 1276 {0x00ad, 0x00ae}, 1277 {0x00b0, 0x00b4}, 1278 {0x00b6, 0x00ba}, 1279 {0x00bc, 0x00bf}, 1280 {0x00c6, 0x00c6}, 1281 {0x00d0, 0x00d0}, 1282 {0x00d7, 0x00d8}, 1283 {0x00de, 0x00e1}, 1284 {0x00e6, 0x00e6}, 1285 {0x00e8, 0x00ea}, 1286 {0x00ec, 0x00ed}, 1287 {0x00f0, 0x00f0}, 1288 {0x00f2, 0x00f3}, 1289 {0x00f7, 0x00fa}, 1290 {0x00fc, 0x00fc}, 1291 {0x00fe, 0x00fe}, 1292 {0x0101, 0x0101}, 1293 {0x0111, 0x0111}, 1294 {0x0113, 0x0113}, 1295 {0x011b, 0x011b}, 1296 {0x0126, 0x0127}, 1297 {0x012b, 0x012b}, 1298 {0x0131, 0x0133}, 1299 {0x0138, 0x0138}, 1300 {0x013f, 0x0142}, 1301 {0x0144, 0x0144}, 1302 {0x0148, 0x014b}, 1303 {0x014d, 0x014d}, 1304 {0x0152, 0x0153}, 1305 {0x0166, 0x0167}, 1306 {0x016b, 0x016b}, 1307 {0x01ce, 0x01ce}, 1308 {0x01d0, 0x01d0}, 1309 {0x01d2, 0x01d2}, 1310 {0x01d4, 0x01d4}, 1311 {0x01d6, 0x01d6}, 1312 {0x01d8, 0x01d8}, 1313 {0x01da, 0x01da}, 1314 {0x01dc, 0x01dc}, 1315 {0x0251, 0x0251}, 1316 {0x0261, 0x0261}, 1317 {0x02c4, 0x02c4}, 1318 {0x02c7, 0x02c7}, 1319 {0x02c9, 0x02cb}, 1320 {0x02cd, 0x02cd}, 1321 {0x02d0, 0x02d0}, 1322 {0x02d8, 0x02db}, 1323 {0x02dd, 0x02dd}, 1324 {0x02df, 0x02df}, 1325 {0x0391, 0x03a1}, 1326 {0x03a3, 0x03a9}, 1327 {0x03b1, 0x03c1}, 1328 {0x03c3, 0x03c9}, 1329 {0x0401, 0x0401}, 1330 {0x0410, 0x044f}, 1331 {0x0451, 0x0451}, 1332 {0x2010, 0x2010}, 1333 {0x2013, 0x2016}, 1334 {0x2018, 0x2019}, 1335 {0x201c, 0x201d}, 1336 {0x2020, 0x2022}, 1337 {0x2024, 0x2027}, 1338 {0x2030, 0x2030}, 1339 {0x2032, 0x2033}, 1340 {0x2035, 0x2035}, 1341 {0x203b, 0x203b}, 1342 {0x203e, 0x203e}, 1343 {0x2074, 0x2074}, 1344 {0x207f, 0x207f}, 1345 {0x2081, 0x2084}, 1346 {0x20ac, 0x20ac}, 1347 {0x2103, 0x2103}, 1348 {0x2105, 0x2105}, 1349 {0x2109, 0x2109}, 1350 {0x2113, 0x2113}, 1351 {0x2116, 0x2116}, 1352 {0x2121, 0x2122}, 1353 {0x2126, 0x2126}, 1354 {0x212b, 0x212b}, 1355 {0x2153, 0x2154}, 1356 {0x215b, 0x215e}, 1357 {0x2160, 0x216b}, 1358 {0x2170, 0x2179}, 1359 {0x2189, 0x2189}, 1360 {0x2190, 0x2199}, 1361 {0x21b8, 0x21b9}, 1362 {0x21d2, 0x21d2}, 1363 {0x21d4, 0x21d4}, 1364 {0x21e7, 0x21e7}, 1365 {0x2200, 0x2200}, 1366 {0x2202, 0x2203}, 1367 {0x2207, 0x2208}, 1368 {0x220b, 0x220b}, 1369 {0x220f, 0x220f}, 1370 {0x2211, 0x2211}, 1371 {0x2215, 0x2215}, 1372 {0x221a, 0x221a}, 1373 {0x221d, 0x2220}, 1374 {0x2223, 0x2223}, 1375 {0x2225, 0x2225}, 1376 {0x2227, 0x222c}, 1377 {0x222e, 0x222e}, 1378 {0x2234, 0x2237}, 1379 {0x223c, 0x223d}, 1380 {0x2248, 0x2248}, 1381 {0x224c, 0x224c}, 1382 {0x2252, 0x2252}, 1383 {0x2260, 0x2261}, 1384 {0x2264, 0x2267}, 1385 {0x226a, 0x226b}, 1386 {0x226e, 0x226f}, 1387 {0x2282, 0x2283}, 1388 {0x2286, 0x2287}, 1389 {0x2295, 0x2295}, 1390 {0x2299, 0x2299}, 1391 {0x22a5, 0x22a5}, 1392 {0x22bf, 0x22bf}, 1393 {0x2312, 0x2312}, 1394 {0x2460, 0x24e9}, 1395 {0x24eb, 0x254b}, 1396 {0x2550, 0x2573}, 1397 {0x2580, 0x258f}, 1398 {0x2592, 0x2595}, 1399 {0x25a0, 0x25a1}, 1400 {0x25a3, 0x25a9}, 1401 {0x25b2, 0x25b3}, 1402 {0x25b6, 0x25b7}, 1403 {0x25bc, 0x25bd}, 1404 {0x25c0, 0x25c1}, 1405 {0x25c6, 0x25c8}, 1406 {0x25cb, 0x25cb}, 1407 {0x25ce, 0x25d1}, 1408 {0x25e2, 0x25e5}, 1409 {0x25ef, 0x25ef}, 1410 {0x2605, 0x2606}, 1411 {0x2609, 0x2609}, 1412 {0x260e, 0x260f}, 1413 {0x2614, 0x2615}, 1414 {0x261c, 0x261c}, 1415 {0x261e, 0x261e}, 1416 {0x2640, 0x2640}, 1417 {0x2642, 0x2642}, 1418 {0x2660, 0x2661}, 1419 {0x2663, 0x2665}, 1420 {0x2667, 0x266a}, 1421 {0x266c, 0x266d}, 1422 {0x266f, 0x266f}, 1423 {0x269e, 0x269f}, 1424 {0x26be, 0x26bf}, 1425 {0x26c4, 0x26cd}, 1426 {0x26cf, 0x26e1}, 1427 {0x26e3, 0x26e3}, 1428 {0x26e8, 0x26ff}, 1429 {0x273d, 0x273d}, 1430 {0x2757, 0x2757}, 1431 {0x2776, 0x277f}, 1432 {0x2b55, 0x2b59}, 1433 {0x3248, 0x324f}, 1434 {0xe000, 0xf8ff}, 1435 {0xfffd, 0xfffd}, 1436 {0x1f100, 0x1f10a}, 1437 {0x1f110, 0x1f12d}, 1438 {0x1f131, 0x1f131}, 1439 {0x1f13d, 0x1f13d}, 1440 {0x1f13f, 0x1f13f}, 1441 {0x1f142, 0x1f142}, 1442 {0x1f146, 0x1f146}, 1443 {0x1f14a, 0x1f14e}, 1444 {0x1f157, 0x1f157}, 1445 {0x1f15f, 0x1f15f}, 1446 {0x1f179, 0x1f179}, 1447 {0x1f17b, 0x1f17c}, 1448 {0x1f17f, 0x1f17f}, 1449 {0x1f18a, 0x1f18d}, 1450 {0x1f190, 0x1f190}, 1451 {0xf0000, 0xffffd}, 1452 {0x100000, 0x10fffd} 1453 }; 1454 1455 if (c >= 0x100) 1456 { 1457 #ifdef USE_WCHAR_FUNCTIONS 1458 /* 1459 * Assume the library function wcwidth() works better than our own 1460 * stuff. It should return 1 for ambiguous width chars! 1461 */ 1462 int n = wcwidth(c); 1463 1464 if (n < 0) 1465 return 6; /* unprintable, displays <xxxx> */ 1466 if (n > 1) 1467 return n; 1468 #else 1469 if (!utf_printable(c)) 1470 return 6; /* unprintable, displays <xxxx> */ 1471 if (intable(doublewidth, sizeof(doublewidth), c)) 1472 return 2; 1473 #endif 1474 } 1475 1476 /* Characters below 0x100 are influenced by 'isprint' option */ 1477 else if (c >= 0x80 && !vim_isprintc(c)) 1478 return 4; /* unprintable, displays <xx> */ 1479 1480 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c)) 1481 return 2; 1482 1483 return 1; 1484 } 1485 1486 /* 1487 * mb_ptr2cells() function pointer. 1488 * Return the number of display cells character at "*p" occupies. 1489 * This doesn't take care of unprintable characters, use ptr2cells() for that. 1490 */ 1491 int 1492 latin_ptr2cells(p) 1493 char_u *p UNUSED; 1494 { 1495 return 1; 1496 } 1497 1498 int 1499 utf_ptr2cells(p) 1500 char_u *p; 1501 { 1502 int c; 1503 1504 /* Need to convert to a wide character. */ 1505 if (*p >= 0x80) 1506 { 1507 c = utf_ptr2char(p); 1508 /* An illegal byte is displayed as <xx>. */ 1509 if (utf_ptr2len(p) == 1 || c == NUL) 1510 return 4; 1511 /* If the char is ASCII it must be an overlong sequence. */ 1512 if (c < 0x80) 1513 return char2cells(c); 1514 return utf_char2cells(c); 1515 } 1516 return 1; 1517 } 1518 1519 int 1520 dbcs_ptr2cells(p) 1521 char_u *p; 1522 { 1523 /* Number of cells is equal to number of bytes, except for euc-jp when 1524 * the first byte is 0x8e. */ 1525 if (enc_dbcs == DBCS_JPNU && *p == 0x8e) 1526 return 1; 1527 return MB_BYTE2LEN(*p); 1528 } 1529 1530 /* 1531 * mb_ptr2cells_len() function pointer. 1532 * Like mb_ptr2cells(), but limit string length to "size". 1533 * For an empty string or truncated character returns 1. 1534 */ 1535 int 1536 latin_ptr2cells_len(p, size) 1537 char_u *p UNUSED; 1538 int size UNUSED; 1539 { 1540 return 1; 1541 } 1542 1543 static int 1544 utf_ptr2cells_len(p, size) 1545 char_u *p; 1546 int size; 1547 { 1548 int c; 1549 1550 /* Need to convert to a wide character. */ 1551 if (size > 0 && *p >= 0x80) 1552 { 1553 if (utf_ptr2len_len(p, size) < utf8len_tab[*p]) 1554 return 1; /* truncated */ 1555 c = utf_ptr2char(p); 1556 /* An illegal byte is displayed as <xx>. */ 1557 if (utf_ptr2len(p) == 1 || c == NUL) 1558 return 4; 1559 /* If the char is ASCII it must be an overlong sequence. */ 1560 if (c < 0x80) 1561 return char2cells(c); 1562 return utf_char2cells(c); 1563 } 1564 return 1; 1565 } 1566 1567 static int 1568 dbcs_ptr2cells_len(p, size) 1569 char_u *p; 1570 int size; 1571 { 1572 /* Number of cells is equal to number of bytes, except for euc-jp when 1573 * the first byte is 0x8e. */ 1574 if (size <= 1 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)) 1575 return 1; 1576 return MB_BYTE2LEN(*p); 1577 } 1578 1579 /* 1580 * mb_char2cells() function pointer. 1581 * Return the number of display cells character "c" occupies. 1582 * Only takes care of multi-byte chars, not "^C" and such. 1583 */ 1584 int 1585 latin_char2cells(c) 1586 int c UNUSED; 1587 { 1588 return 1; 1589 } 1590 1591 static int 1592 dbcs_char2cells(c) 1593 int c; 1594 { 1595 /* Number of cells is equal to number of bytes, except for euc-jp when 1596 * the first byte is 0x8e. */ 1597 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e) 1598 return 1; 1599 /* use the first byte */ 1600 return MB_BYTE2LEN((unsigned)c >> 8); 1601 } 1602 1603 /* 1604 * Return the number of cells occupied by string "p". 1605 * Stop at a NUL character. When "len" >= 0 stop at character "p[len]". 1606 */ 1607 int 1608 mb_string2cells(p, len) 1609 char_u *p; 1610 int len; 1611 { 1612 int i; 1613 int clen = 0; 1614 1615 for (i = 0; (len < 0 || i < len) && p[i] != NUL; i += (*mb_ptr2len)(p + i)) 1616 clen += (*mb_ptr2cells)(p + i); 1617 return clen; 1618 } 1619 1620 /* 1621 * mb_off2cells() function pointer. 1622 * Return number of display cells for char at ScreenLines[off]. 1623 * We make sure that the offset used is less than "max_off". 1624 */ 1625 int 1626 latin_off2cells(off, max_off) 1627 unsigned off UNUSED; 1628 unsigned max_off UNUSED; 1629 { 1630 return 1; 1631 } 1632 1633 int 1634 dbcs_off2cells(off, max_off) 1635 unsigned off; 1636 unsigned max_off; 1637 { 1638 /* never check beyond end of the line */ 1639 if (off >= max_off) 1640 return 1; 1641 1642 /* Number of cells is equal to number of bytes, except for euc-jp when 1643 * the first byte is 0x8e. */ 1644 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e) 1645 return 1; 1646 return MB_BYTE2LEN(ScreenLines[off]); 1647 } 1648 1649 int 1650 utf_off2cells(off, max_off) 1651 unsigned off; 1652 unsigned max_off; 1653 { 1654 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1; 1655 } 1656 1657 /* 1658 * mb_ptr2char() function pointer. 1659 * Convert a byte sequence into a character. 1660 */ 1661 int 1662 latin_ptr2char(p) 1663 char_u *p; 1664 { 1665 return *p; 1666 } 1667 1668 static int 1669 dbcs_ptr2char(p) 1670 char_u *p; 1671 { 1672 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL) 1673 return (p[0] << 8) + p[1]; 1674 return *p; 1675 } 1676 1677 /* 1678 * Convert a UTF-8 byte sequence to a wide character. 1679 * If the sequence is illegal or truncated by a NUL the first byte is 1680 * returned. 1681 * Does not include composing characters, of course. 1682 */ 1683 int 1684 utf_ptr2char(p) 1685 char_u *p; 1686 { 1687 int len; 1688 1689 if (p[0] < 0x80) /* be quick for ASCII */ 1690 return p[0]; 1691 1692 len = utf8len_tab_zero[p[0]]; 1693 if (len > 1 && (p[1] & 0xc0) == 0x80) 1694 { 1695 if (len == 2) 1696 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); 1697 if ((p[2] & 0xc0) == 0x80) 1698 { 1699 if (len == 3) 1700 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) 1701 + (p[2] & 0x3f); 1702 if ((p[3] & 0xc0) == 0x80) 1703 { 1704 if (len == 4) 1705 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) 1706 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); 1707 if ((p[4] & 0xc0) == 0x80) 1708 { 1709 if (len == 5) 1710 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) 1711 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) 1712 + (p[4] & 0x3f); 1713 if ((p[5] & 0xc0) == 0x80 && len == 6) 1714 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) 1715 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) 1716 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); 1717 } 1718 } 1719 } 1720 } 1721 /* Illegal value, just return the first byte */ 1722 return p[0]; 1723 } 1724 1725 /* 1726 * Convert a UTF-8 byte sequence to a wide character. 1727 * String is assumed to be terminated by NUL or after "n" bytes, whichever 1728 * comes first. 1729 * The function is safe in the sense that it never accesses memory beyond the 1730 * first "n" bytes of "s". 1731 * 1732 * On success, returns decoded codepoint, advances "s" to the beginning of 1733 * next character and decreases "n" accordingly. 1734 * 1735 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past 1736 * NUL byte. 1737 * 1738 * If byte sequence is illegal or incomplete, returns -1 and does not advance 1739 * "s". 1740 */ 1741 static int 1742 utf_safe_read_char_adv(s, n) 1743 char_u **s; 1744 size_t *n; 1745 { 1746 int c, k; 1747 1748 if (*n == 0) /* end of buffer */ 1749 return 0; 1750 1751 k = utf8len_tab_zero[**s]; 1752 1753 if (k == 1) 1754 { 1755 /* ASCII character or NUL */ 1756 (*n)--; 1757 return *(*s)++; 1758 } 1759 1760 if ((size_t)k <= *n) 1761 { 1762 /* We have a multibyte sequence and it isn't truncated by buffer 1763 * limits so utf_ptr2char() is safe to use. Or the first byte is 1764 * illegal (k=0), and it's also safe to use utf_ptr2char(). */ 1765 c = utf_ptr2char(*s); 1766 1767 /* On failure, utf_ptr2char() returns the first byte, so here we 1768 * check equality with the first byte. The only non-ASCII character 1769 * which equals the first byte of its own UTF-8 representation is 1770 * U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too. 1771 * It's safe even if n=1, else we would have k=2 > n. */ 1772 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83)) 1773 { 1774 /* byte sequence was successfully decoded */ 1775 *s += k; 1776 *n -= k; 1777 return c; 1778 } 1779 } 1780 1781 /* byte sequence is incomplete or illegal */ 1782 return -1; 1783 } 1784 1785 /* 1786 * Get character at **pp and advance *pp to the next character. 1787 * Note: composing characters are skipped! 1788 */ 1789 int 1790 mb_ptr2char_adv(pp) 1791 char_u **pp; 1792 { 1793 int c; 1794 1795 c = (*mb_ptr2char)(*pp); 1796 *pp += (*mb_ptr2len)(*pp); 1797 return c; 1798 } 1799 1800 /* 1801 * Get character at **pp and advance *pp to the next character. 1802 * Note: composing characters are returned as separate characters. 1803 */ 1804 int 1805 mb_cptr2char_adv(pp) 1806 char_u **pp; 1807 { 1808 int c; 1809 1810 c = (*mb_ptr2char)(*pp); 1811 if (enc_utf8) 1812 *pp += utf_ptr2len(*pp); 1813 else 1814 *pp += (*mb_ptr2len)(*pp); 1815 return c; 1816 } 1817 1818 #if defined(FEAT_ARABIC) || defined(PROTO) 1819 /* 1820 * Check whether we are dealing with Arabic combining characters. 1821 * Note: these are NOT really composing characters! 1822 */ 1823 int 1824 arabic_combine(one, two) 1825 int one; /* first character */ 1826 int two; /* character just after "one" */ 1827 { 1828 if (one == a_LAM) 1829 return arabic_maycombine(two); 1830 return FALSE; 1831 } 1832 1833 /* 1834 * Check whether we are dealing with a character that could be regarded as an 1835 * Arabic combining character, need to check the character before this. 1836 */ 1837 int 1838 arabic_maycombine(two) 1839 int two; 1840 { 1841 if (p_arshape && !p_tbidi) 1842 return (two == a_ALEF_MADDA 1843 || two == a_ALEF_HAMZA_ABOVE 1844 || two == a_ALEF_HAMZA_BELOW 1845 || two == a_ALEF); 1846 return FALSE; 1847 } 1848 1849 /* 1850 * Check if the character pointed to by "p2" is a composing character when it 1851 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which 1852 * behaves like a composing character. 1853 */ 1854 int 1855 utf_composinglike(p1, p2) 1856 char_u *p1; 1857 char_u *p2; 1858 { 1859 int c2; 1860 1861 c2 = utf_ptr2char(p2); 1862 if (utf_iscomposing(c2)) 1863 return TRUE; 1864 if (!arabic_maycombine(c2)) 1865 return FALSE; 1866 return arabic_combine(utf_ptr2char(p1), c2); 1867 } 1868 #endif 1869 1870 /* 1871 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1872 * composing characters. 1873 */ 1874 int 1875 utfc_ptr2char(p, pcc) 1876 char_u *p; 1877 int *pcc; /* return: composing chars, last one is 0 */ 1878 { 1879 int len; 1880 int c; 1881 int cc; 1882 int i = 0; 1883 1884 c = utf_ptr2char(p); 1885 len = utf_ptr2len(p); 1886 1887 /* Only accept a composing char when the first char isn't illegal. */ 1888 if ((len > 1 || *p < 0x80) 1889 && p[len] >= 0x80 1890 && UTF_COMPOSINGLIKE(p, p + len)) 1891 { 1892 cc = utf_ptr2char(p + len); 1893 for (;;) 1894 { 1895 pcc[i++] = cc; 1896 if (i == MAX_MCO) 1897 break; 1898 len += utf_ptr2len(p + len); 1899 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1900 break; 1901 } 1902 } 1903 1904 if (i < MAX_MCO) /* last composing char must be 0 */ 1905 pcc[i] = 0; 1906 1907 return c; 1908 } 1909 1910 /* 1911 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1912 * composing characters. Use no more than p[maxlen]. 1913 */ 1914 int 1915 utfc_ptr2char_len(p, pcc, maxlen) 1916 char_u *p; 1917 int *pcc; /* return: composing chars, last one is 0 */ 1918 int maxlen; 1919 { 1920 int len; 1921 int c; 1922 int cc; 1923 int i = 0; 1924 1925 c = utf_ptr2char(p); 1926 len = utf_ptr2len_len(p, maxlen); 1927 /* Only accept a composing char when the first char isn't illegal. */ 1928 if ((len > 1 || *p < 0x80) 1929 && len < maxlen 1930 && p[len] >= 0x80 1931 && UTF_COMPOSINGLIKE(p, p + len)) 1932 { 1933 cc = utf_ptr2char(p + len); 1934 for (;;) 1935 { 1936 pcc[i++] = cc; 1937 if (i == MAX_MCO) 1938 break; 1939 len += utf_ptr2len_len(p + len, maxlen - len); 1940 if (len >= maxlen 1941 || p[len] < 0x80 1942 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1943 break; 1944 } 1945 } 1946 1947 if (i < MAX_MCO) /* last composing char must be 0 */ 1948 pcc[i] = 0; 1949 1950 return c; 1951 } 1952 1953 /* 1954 * Convert the character at screen position "off" to a sequence of bytes. 1955 * Includes the composing characters. 1956 * "buf" must at least have the length MB_MAXBYTES + 1. 1957 * Only to be used when ScreenLinesUC[off] != 0. 1958 * Returns the produced number of bytes. 1959 */ 1960 int 1961 utfc_char2bytes(off, buf) 1962 int off; 1963 char_u *buf; 1964 { 1965 int len; 1966 int i; 1967 1968 len = utf_char2bytes(ScreenLinesUC[off], buf); 1969 for (i = 0; i < Screen_mco; ++i) 1970 { 1971 if (ScreenLinesC[i][off] == 0) 1972 break; 1973 len += utf_char2bytes(ScreenLinesC[i][off], buf + len); 1974 } 1975 return len; 1976 } 1977 1978 /* 1979 * Get the length of a UTF-8 byte sequence, not including any following 1980 * composing characters. 1981 * Returns 0 for "". 1982 * Returns 1 for an illegal byte sequence. 1983 */ 1984 int 1985 utf_ptr2len(p) 1986 char_u *p; 1987 { 1988 int len; 1989 int i; 1990 1991 if (*p == NUL) 1992 return 0; 1993 len = utf8len_tab[*p]; 1994 for (i = 1; i < len; ++i) 1995 if ((p[i] & 0xc0) != 0x80) 1996 return 1; 1997 return len; 1998 } 1999 2000 /* 2001 * Return length of UTF-8 character, obtained from the first byte. 2002 * "b" must be between 0 and 255! 2003 * Returns 1 for an invalid first byte value. 2004 */ 2005 int 2006 utf_byte2len(b) 2007 int b; 2008 { 2009 return utf8len_tab[b]; 2010 } 2011 2012 /* 2013 * Get the length of UTF-8 byte sequence "p[size]". Does not include any 2014 * following composing characters. 2015 * Returns 1 for "". 2016 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.). 2017 * Returns number > "size" for an incomplete byte sequence. 2018 * Never returns zero. 2019 */ 2020 int 2021 utf_ptr2len_len(p, size) 2022 char_u *p; 2023 int size; 2024 { 2025 int len; 2026 int i; 2027 int m; 2028 2029 len = utf8len_tab[*p]; 2030 if (len == 1) 2031 return 1; /* NUL, ascii or illegal lead byte */ 2032 if (len > size) 2033 m = size; /* incomplete byte sequence. */ 2034 else 2035 m = len; 2036 for (i = 1; i < m; ++i) 2037 if ((p[i] & 0xc0) != 0x80) 2038 return 1; 2039 return len; 2040 } 2041 2042 /* 2043 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. 2044 * This includes following composing characters. 2045 */ 2046 int 2047 utfc_ptr2len(p) 2048 char_u *p; 2049 { 2050 int len; 2051 int b0 = *p; 2052 #ifdef FEAT_ARABIC 2053 int prevlen; 2054 #endif 2055 2056 if (b0 == NUL) 2057 return 0; 2058 if (b0 < 0x80 && p[1] < 0x80) /* be quick for ASCII */ 2059 return 1; 2060 2061 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2062 len = utf_ptr2len(p); 2063 2064 /* Check for illegal byte. */ 2065 if (len == 1 && b0 >= 0x80) 2066 return 1; 2067 2068 /* 2069 * Check for composing characters. We can handle only the first six, but 2070 * skip all of them (otherwise the cursor would get stuck). 2071 */ 2072 #ifdef FEAT_ARABIC 2073 prevlen = 0; 2074 #endif 2075 for (;;) 2076 { 2077 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2078 return len; 2079 2080 /* Skip over composing char */ 2081 #ifdef FEAT_ARABIC 2082 prevlen = len; 2083 #endif 2084 len += utf_ptr2len(p + len); 2085 } 2086 } 2087 2088 /* 2089 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" 2090 * takes. This includes following composing characters. 2091 * Returns 0 for an empty string. 2092 * Returns 1 for an illegal char or an incomplete byte sequence. 2093 */ 2094 int 2095 utfc_ptr2len_len(p, size) 2096 char_u *p; 2097 int size; 2098 { 2099 int len; 2100 #ifdef FEAT_ARABIC 2101 int prevlen; 2102 #endif 2103 2104 if (size < 1 || *p == NUL) 2105 return 0; 2106 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) /* be quick for ASCII */ 2107 return 1; 2108 2109 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2110 len = utf_ptr2len_len(p, size); 2111 2112 /* Check for illegal byte and incomplete byte sequence. */ 2113 if ((len == 1 && p[0] >= 0x80) || len > size) 2114 return 1; 2115 2116 /* 2117 * Check for composing characters. We can handle only the first six, but 2118 * skip all of them (otherwise the cursor would get stuck). 2119 */ 2120 #ifdef FEAT_ARABIC 2121 prevlen = 0; 2122 #endif 2123 while (len < size) 2124 { 2125 int len_next_char; 2126 2127 if (p[len] < 0x80) 2128 break; 2129 2130 /* 2131 * Next character length should not go beyond size to ensure that 2132 * UTF_COMPOSINGLIKE(...) does not read beyond size. 2133 */ 2134 len_next_char = utf_ptr2len_len(p + len, size - len); 2135 if (len_next_char > size - len) 2136 break; 2137 2138 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2139 break; 2140 2141 /* Skip over composing char */ 2142 #ifdef FEAT_ARABIC 2143 prevlen = len; 2144 #endif 2145 len += len_next_char; 2146 } 2147 return len; 2148 } 2149 2150 /* 2151 * Return the number of bytes the UTF-8 encoding of character "c" takes. 2152 * This does not include composing characters. 2153 */ 2154 int 2155 utf_char2len(c) 2156 int c; 2157 { 2158 if (c < 0x80) 2159 return 1; 2160 if (c < 0x800) 2161 return 2; 2162 if (c < 0x10000) 2163 return 3; 2164 if (c < 0x200000) 2165 return 4; 2166 if (c < 0x4000000) 2167 return 5; 2168 return 6; 2169 } 2170 2171 /* 2172 * Convert Unicode character "c" to UTF-8 string in "buf[]". 2173 * Returns the number of bytes. 2174 * This does not include composing characters. 2175 */ 2176 int 2177 utf_char2bytes(c, buf) 2178 int c; 2179 char_u *buf; 2180 { 2181 if (c < 0x80) /* 7 bits */ 2182 { 2183 buf[0] = c; 2184 return 1; 2185 } 2186 if (c < 0x800) /* 11 bits */ 2187 { 2188 buf[0] = 0xc0 + ((unsigned)c >> 6); 2189 buf[1] = 0x80 + (c & 0x3f); 2190 return 2; 2191 } 2192 if (c < 0x10000) /* 16 bits */ 2193 { 2194 buf[0] = 0xe0 + ((unsigned)c >> 12); 2195 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2196 buf[2] = 0x80 + (c & 0x3f); 2197 return 3; 2198 } 2199 if (c < 0x200000) /* 21 bits */ 2200 { 2201 buf[0] = 0xf0 + ((unsigned)c >> 18); 2202 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2203 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2204 buf[3] = 0x80 + (c & 0x3f); 2205 return 4; 2206 } 2207 if (c < 0x4000000) /* 26 bits */ 2208 { 2209 buf[0] = 0xf8 + ((unsigned)c >> 24); 2210 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2211 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2212 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2213 buf[4] = 0x80 + (c & 0x3f); 2214 return 5; 2215 } 2216 /* 31 bits */ 2217 buf[0] = 0xfc + ((unsigned)c >> 30); 2218 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f); 2219 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2220 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2221 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2222 buf[5] = 0x80 + (c & 0x3f); 2223 return 6; 2224 } 2225 2226 /* 2227 * Return TRUE if "c" is a composing UTF-8 character. This means it will be 2228 * drawn on top of the preceding character. 2229 * Based on code from Markus Kuhn. 2230 */ 2231 int 2232 utf_iscomposing(c) 2233 int c; 2234 { 2235 /* Sorted list of non-overlapping intervals. 2236 * Generated by ../runtime/tools/unicode.vim. */ 2237 static struct interval combining[] = 2238 { 2239 {0x0300, 0x036f}, 2240 {0x0483, 0x0489}, 2241 {0x0591, 0x05bd}, 2242 {0x05bf, 0x05bf}, 2243 {0x05c1, 0x05c2}, 2244 {0x05c4, 0x05c5}, 2245 {0x05c7, 0x05c7}, 2246 {0x0610, 0x061a}, 2247 {0x064b, 0x065e}, 2248 {0x0670, 0x0670}, 2249 {0x06d6, 0x06dc}, 2250 {0x06de, 0x06e4}, 2251 {0x06e7, 0x06e8}, 2252 {0x06ea, 0x06ed}, 2253 {0x0711, 0x0711}, 2254 {0x0730, 0x074a}, 2255 {0x07a6, 0x07b0}, 2256 {0x07eb, 0x07f3}, 2257 {0x0816, 0x0819}, 2258 {0x081b, 0x0823}, 2259 {0x0825, 0x0827}, 2260 {0x0829, 0x082d}, 2261 {0x0900, 0x0903}, 2262 {0x093c, 0x093c}, 2263 {0x093e, 0x094e}, 2264 {0x0951, 0x0955}, 2265 {0x0962, 0x0963}, 2266 {0x0981, 0x0983}, 2267 {0x09bc, 0x09bc}, 2268 {0x09be, 0x09c4}, 2269 {0x09c7, 0x09c8}, 2270 {0x09cb, 0x09cd}, 2271 {0x09d7, 0x09d7}, 2272 {0x09e2, 0x09e3}, 2273 {0x0a01, 0x0a03}, 2274 {0x0a3c, 0x0a3c}, 2275 {0x0a3e, 0x0a42}, 2276 {0x0a47, 0x0a48}, 2277 {0x0a4b, 0x0a4d}, 2278 {0x0a51, 0x0a51}, 2279 {0x0a70, 0x0a71}, 2280 {0x0a75, 0x0a75}, 2281 {0x0a81, 0x0a83}, 2282 {0x0abc, 0x0abc}, 2283 {0x0abe, 0x0ac5}, 2284 {0x0ac7, 0x0ac9}, 2285 {0x0acb, 0x0acd}, 2286 {0x0ae2, 0x0ae3}, 2287 {0x0b01, 0x0b03}, 2288 {0x0b3c, 0x0b3c}, 2289 {0x0b3e, 0x0b44}, 2290 {0x0b47, 0x0b48}, 2291 {0x0b4b, 0x0b4d}, 2292 {0x0b56, 0x0b57}, 2293 {0x0b62, 0x0b63}, 2294 {0x0b82, 0x0b82}, 2295 {0x0bbe, 0x0bc2}, 2296 {0x0bc6, 0x0bc8}, 2297 {0x0bca, 0x0bcd}, 2298 {0x0bd7, 0x0bd7}, 2299 {0x0c01, 0x0c03}, 2300 {0x0c3e, 0x0c44}, 2301 {0x0c46, 0x0c48}, 2302 {0x0c4a, 0x0c4d}, 2303 {0x0c55, 0x0c56}, 2304 {0x0c62, 0x0c63}, 2305 {0x0c82, 0x0c83}, 2306 {0x0cbc, 0x0cbc}, 2307 {0x0cbe, 0x0cc4}, 2308 {0x0cc6, 0x0cc8}, 2309 {0x0cca, 0x0ccd}, 2310 {0x0cd5, 0x0cd6}, 2311 {0x0ce2, 0x0ce3}, 2312 {0x0d02, 0x0d03}, 2313 {0x0d3e, 0x0d44}, 2314 {0x0d46, 0x0d48}, 2315 {0x0d4a, 0x0d4d}, 2316 {0x0d57, 0x0d57}, 2317 {0x0d62, 0x0d63}, 2318 {0x0d82, 0x0d83}, 2319 {0x0dca, 0x0dca}, 2320 {0x0dcf, 0x0dd4}, 2321 {0x0dd6, 0x0dd6}, 2322 {0x0dd8, 0x0ddf}, 2323 {0x0df2, 0x0df3}, 2324 {0x0e31, 0x0e31}, 2325 {0x0e34, 0x0e3a}, 2326 {0x0e47, 0x0e4e}, 2327 {0x0eb1, 0x0eb1}, 2328 {0x0eb4, 0x0eb9}, 2329 {0x0ebb, 0x0ebc}, 2330 {0x0ec8, 0x0ecd}, 2331 {0x0f18, 0x0f19}, 2332 {0x0f35, 0x0f35}, 2333 {0x0f37, 0x0f37}, 2334 {0x0f39, 0x0f39}, 2335 {0x0f3e, 0x0f3f}, 2336 {0x0f71, 0x0f84}, 2337 {0x0f86, 0x0f87}, 2338 {0x0f90, 0x0f97}, 2339 {0x0f99, 0x0fbc}, 2340 {0x0fc6, 0x0fc6}, 2341 {0x102b, 0x103e}, 2342 {0x1056, 0x1059}, 2343 {0x105e, 0x1060}, 2344 {0x1062, 0x1064}, 2345 {0x1067, 0x106d}, 2346 {0x1071, 0x1074}, 2347 {0x1082, 0x108d}, 2348 {0x108f, 0x108f}, 2349 {0x109a, 0x109d}, 2350 {0x135f, 0x135f}, 2351 {0x1712, 0x1714}, 2352 {0x1732, 0x1734}, 2353 {0x1752, 0x1753}, 2354 {0x1772, 0x1773}, 2355 {0x17b6, 0x17d3}, 2356 {0x17dd, 0x17dd}, 2357 {0x180b, 0x180d}, 2358 {0x18a9, 0x18a9}, 2359 {0x1920, 0x192b}, 2360 {0x1930, 0x193b}, 2361 {0x19b0, 0x19c0}, 2362 {0x19c8, 0x19c9}, 2363 {0x1a17, 0x1a1b}, 2364 {0x1a55, 0x1a5e}, 2365 {0x1a60, 0x1a7c}, 2366 {0x1a7f, 0x1a7f}, 2367 {0x1b00, 0x1b04}, 2368 {0x1b34, 0x1b44}, 2369 {0x1b6b, 0x1b73}, 2370 {0x1b80, 0x1b82}, 2371 {0x1ba1, 0x1baa}, 2372 {0x1c24, 0x1c37}, 2373 {0x1cd0, 0x1cd2}, 2374 {0x1cd4, 0x1ce8}, 2375 {0x1ced, 0x1ced}, 2376 {0x1cf2, 0x1cf2}, 2377 {0x1dc0, 0x1de6}, 2378 {0x1dfd, 0x1dff}, 2379 {0x20d0, 0x20f0}, 2380 {0x2cef, 0x2cf1}, 2381 {0x2de0, 0x2dff}, 2382 {0x302a, 0x302f}, 2383 {0x3099, 0x309a}, 2384 {0xa66f, 0xa672}, 2385 {0xa67c, 0xa67d}, 2386 {0xa6f0, 0xa6f1}, 2387 {0xa802, 0xa802}, 2388 {0xa806, 0xa806}, 2389 {0xa80b, 0xa80b}, 2390 {0xa823, 0xa827}, 2391 {0xa880, 0xa881}, 2392 {0xa8b4, 0xa8c4}, 2393 {0xa8e0, 0xa8f1}, 2394 {0xa926, 0xa92d}, 2395 {0xa947, 0xa953}, 2396 {0xa980, 0xa983}, 2397 {0xa9b3, 0xa9c0}, 2398 {0xaa29, 0xaa36}, 2399 {0xaa43, 0xaa43}, 2400 {0xaa4c, 0xaa4d}, 2401 {0xaa7b, 0xaa7b}, 2402 {0xaab0, 0xaab0}, 2403 {0xaab2, 0xaab4}, 2404 {0xaab7, 0xaab8}, 2405 {0xaabe, 0xaabf}, 2406 {0xaac1, 0xaac1}, 2407 {0xabe3, 0xabea}, 2408 {0xabec, 0xabed}, 2409 {0xfb1e, 0xfb1e}, 2410 {0xfe00, 0xfe0f}, 2411 {0xfe20, 0xfe26}, 2412 {0x101fd, 0x101fd}, 2413 {0x10a01, 0x10a03}, 2414 {0x10a05, 0x10a06}, 2415 {0x10a0c, 0x10a0f}, 2416 {0x10a38, 0x10a3a}, 2417 {0x10a3f, 0x10a3f}, 2418 {0x11080, 0x11082}, 2419 {0x110b0, 0x110ba}, 2420 {0x1d165, 0x1d169}, 2421 {0x1d16d, 0x1d172}, 2422 {0x1d17b, 0x1d182}, 2423 {0x1d185, 0x1d18b}, 2424 {0x1d1aa, 0x1d1ad}, 2425 {0x1d242, 0x1d244}, 2426 {0xe0100, 0xe01ef} 2427 }; 2428 2429 return intable(combining, sizeof(combining), c); 2430 } 2431 2432 /* 2433 * Return TRUE for characters that can be displayed in a normal way. 2434 * Only for characters of 0x100 and above! 2435 */ 2436 int 2437 utf_printable(c) 2438 int c; 2439 { 2440 #ifdef USE_WCHAR_FUNCTIONS 2441 /* 2442 * Assume the iswprint() library function works better than our own stuff. 2443 */ 2444 return iswprint(c); 2445 #else 2446 /* Sorted list of non-overlapping intervals. 2447 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */ 2448 static struct interval nonprint[] = 2449 { 2450 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2451 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2452 {0xfffe, 0xffff} 2453 }; 2454 2455 return !intable(nonprint, sizeof(nonprint), c); 2456 #endif 2457 } 2458 2459 /* 2460 * Get class of a Unicode character. 2461 * 0: white space 2462 * 1: punctuation 2463 * 2 or bigger: some class of word character. 2464 */ 2465 int 2466 utf_class(c) 2467 int c; 2468 { 2469 /* sorted list of non-overlapping intervals */ 2470 static struct clinterval 2471 { 2472 unsigned short first; 2473 unsigned short last; 2474 unsigned short class; 2475 } classes[] = 2476 { 2477 {0x037e, 0x037e, 1}, /* Greek question mark */ 2478 {0x0387, 0x0387, 1}, /* Greek ano teleia */ 2479 {0x055a, 0x055f, 1}, /* Armenian punctuation */ 2480 {0x0589, 0x0589, 1}, /* Armenian full stop */ 2481 {0x05be, 0x05be, 1}, 2482 {0x05c0, 0x05c0, 1}, 2483 {0x05c3, 0x05c3, 1}, 2484 {0x05f3, 0x05f4, 1}, 2485 {0x060c, 0x060c, 1}, 2486 {0x061b, 0x061b, 1}, 2487 {0x061f, 0x061f, 1}, 2488 {0x066a, 0x066d, 1}, 2489 {0x06d4, 0x06d4, 1}, 2490 {0x0700, 0x070d, 1}, /* Syriac punctuation */ 2491 {0x0964, 0x0965, 1}, 2492 {0x0970, 0x0970, 1}, 2493 {0x0df4, 0x0df4, 1}, 2494 {0x0e4f, 0x0e4f, 1}, 2495 {0x0e5a, 0x0e5b, 1}, 2496 {0x0f04, 0x0f12, 1}, 2497 {0x0f3a, 0x0f3d, 1}, 2498 {0x0f85, 0x0f85, 1}, 2499 {0x104a, 0x104f, 1}, /* Myanmar punctuation */ 2500 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */ 2501 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */ 2502 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */ 2503 {0x1680, 0x1680, 0}, 2504 {0x169b, 0x169c, 1}, 2505 {0x16eb, 0x16ed, 1}, 2506 {0x1735, 0x1736, 1}, 2507 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */ 2508 {0x1800, 0x180a, 1}, /* Mongolian punctuation */ 2509 {0x2000, 0x200b, 0}, /* spaces */ 2510 {0x200c, 0x2027, 1}, /* punctuation and symbols */ 2511 {0x2028, 0x2029, 0}, 2512 {0x202a, 0x202e, 1}, /* punctuation and symbols */ 2513 {0x202f, 0x202f, 0}, 2514 {0x2030, 0x205e, 1}, /* punctuation and symbols */ 2515 {0x205f, 0x205f, 0}, 2516 {0x2060, 0x27ff, 1}, /* punctuation and symbols */ 2517 {0x2070, 0x207f, 0x2070}, /* superscript */ 2518 {0x2080, 0x2094, 0x2080}, /* subscript */ 2519 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */ 2520 {0x2800, 0x28ff, 0x2800}, /* braille */ 2521 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */ 2522 {0x29d8, 0x29db, 1}, 2523 {0x29fc, 0x29fd, 1}, 2524 {0x3000, 0x3000, 0}, /* ideographic space */ 2525 {0x3001, 0x3020, 1}, /* ideographic punctuation */ 2526 {0x3030, 0x3030, 1}, 2527 {0x303d, 0x303d, 1}, 2528 {0x3040, 0x309f, 0x3040}, /* Hiragana */ 2529 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */ 2530 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */ 2531 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */ 2532 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */ 2533 {0xfd3e, 0xfd3f, 1}, 2534 {0xfe30, 0xfe6b, 1}, /* punctuation forms */ 2535 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */ 2536 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */ 2537 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */ 2538 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */ 2539 }; 2540 int bot = 0; 2541 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2542 int mid; 2543 2544 /* First quick check for Latin1 characters, use 'iskeyword'. */ 2545 if (c < 0x100) 2546 { 2547 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2548 return 0; /* blank */ 2549 if (vim_iswordc(c)) 2550 return 2; /* word character */ 2551 return 1; /* punctuation */ 2552 } 2553 2554 /* binary search in table */ 2555 while (top >= bot) 2556 { 2557 mid = (bot + top) / 2; 2558 if (classes[mid].last < c) 2559 bot = mid + 1; 2560 else if (classes[mid].first > c) 2561 top = mid - 1; 2562 else 2563 return (int)classes[mid].class; 2564 } 2565 2566 /* most other characters are "word" characters */ 2567 return 2; 2568 } 2569 2570 /* 2571 * Code for Unicode case-dependent operations. Based on notes in 2572 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2573 * This code uses simple case folding, not full case folding. 2574 * Last updated for Unicode 5.2. 2575 */ 2576 2577 /* 2578 * The following tables are built by ../runtime/tools/unicode.vim. 2579 * They must be in numeric order, because we use binary search. 2580 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2581 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2582 * folded/upper/lower by adding 32. 2583 */ 2584 typedef struct 2585 { 2586 int rangeStart; 2587 int rangeEnd; 2588 int step; 2589 int offset; 2590 } convertStruct; 2591 2592 static convertStruct foldCase[] = 2593 { 2594 {0x41,0x5a,1,32}, 2595 {0xb5,0xb5,-1,775}, 2596 {0xc0,0xd6,1,32}, 2597 {0xd8,0xde,1,32}, 2598 {0x100,0x12e,2,1}, 2599 {0x132,0x136,2,1}, 2600 {0x139,0x147,2,1}, 2601 {0x14a,0x176,2,1}, 2602 {0x178,0x178,-1,-121}, 2603 {0x179,0x17d,2,1}, 2604 {0x17f,0x17f,-1,-268}, 2605 {0x181,0x181,-1,210}, 2606 {0x182,0x184,2,1}, 2607 {0x186,0x186,-1,206}, 2608 {0x187,0x187,-1,1}, 2609 {0x189,0x18a,1,205}, 2610 {0x18b,0x18b,-1,1}, 2611 {0x18e,0x18e,-1,79}, 2612 {0x18f,0x18f,-1,202}, 2613 {0x190,0x190,-1,203}, 2614 {0x191,0x191,-1,1}, 2615 {0x193,0x193,-1,205}, 2616 {0x194,0x194,-1,207}, 2617 {0x196,0x196,-1,211}, 2618 {0x197,0x197,-1,209}, 2619 {0x198,0x198,-1,1}, 2620 {0x19c,0x19c,-1,211}, 2621 {0x19d,0x19d,-1,213}, 2622 {0x19f,0x19f,-1,214}, 2623 {0x1a0,0x1a4,2,1}, 2624 {0x1a6,0x1a6,-1,218}, 2625 {0x1a7,0x1a7,-1,1}, 2626 {0x1a9,0x1a9,-1,218}, 2627 {0x1ac,0x1ac,-1,1}, 2628 {0x1ae,0x1ae,-1,218}, 2629 {0x1af,0x1af,-1,1}, 2630 {0x1b1,0x1b2,1,217}, 2631 {0x1b3,0x1b5,2,1}, 2632 {0x1b7,0x1b7,-1,219}, 2633 {0x1b8,0x1bc,4,1}, 2634 {0x1c4,0x1c4,-1,2}, 2635 {0x1c5,0x1c5,-1,1}, 2636 {0x1c7,0x1c7,-1,2}, 2637 {0x1c8,0x1c8,-1,1}, 2638 {0x1ca,0x1ca,-1,2}, 2639 {0x1cb,0x1db,2,1}, 2640 {0x1de,0x1ee,2,1}, 2641 {0x1f1,0x1f1,-1,2}, 2642 {0x1f2,0x1f4,2,1}, 2643 {0x1f6,0x1f6,-1,-97}, 2644 {0x1f7,0x1f7,-1,-56}, 2645 {0x1f8,0x21e,2,1}, 2646 {0x220,0x220,-1,-130}, 2647 {0x222,0x232,2,1}, 2648 {0x23a,0x23a,-1,10795}, 2649 {0x23b,0x23b,-1,1}, 2650 {0x23d,0x23d,-1,-163}, 2651 {0x23e,0x23e,-1,10792}, 2652 {0x241,0x241,-1,1}, 2653 {0x243,0x243,-1,-195}, 2654 {0x244,0x244,-1,69}, 2655 {0x245,0x245,-1,71}, 2656 {0x246,0x24e,2,1}, 2657 {0x345,0x345,-1,116}, 2658 {0x370,0x372,2,1}, 2659 {0x376,0x376,-1,1}, 2660 {0x386,0x386,-1,38}, 2661 {0x388,0x38a,1,37}, 2662 {0x38c,0x38c,-1,64}, 2663 {0x38e,0x38f,1,63}, 2664 {0x391,0x3a1,1,32}, 2665 {0x3a3,0x3ab,1,32}, 2666 {0x3c2,0x3c2,-1,1}, 2667 {0x3cf,0x3cf,-1,8}, 2668 {0x3d0,0x3d0,-1,-30}, 2669 {0x3d1,0x3d1,-1,-25}, 2670 {0x3d5,0x3d5,-1,-15}, 2671 {0x3d6,0x3d6,-1,-22}, 2672 {0x3d8,0x3ee,2,1}, 2673 {0x3f0,0x3f0,-1,-54}, 2674 {0x3f1,0x3f1,-1,-48}, 2675 {0x3f4,0x3f4,-1,-60}, 2676 {0x3f5,0x3f5,-1,-64}, 2677 {0x3f7,0x3f7,-1,1}, 2678 {0x3f9,0x3f9,-1,-7}, 2679 {0x3fa,0x3fa,-1,1}, 2680 {0x3fd,0x3ff,1,-130}, 2681 {0x400,0x40f,1,80}, 2682 {0x410,0x42f,1,32}, 2683 {0x460,0x480,2,1}, 2684 {0x48a,0x4be,2,1}, 2685 {0x4c0,0x4c0,-1,15}, 2686 {0x4c1,0x4cd,2,1}, 2687 {0x4d0,0x524,2,1}, 2688 {0x531,0x556,1,48}, 2689 {0x10a0,0x10c5,1,7264}, 2690 {0x1e00,0x1e94,2,1}, 2691 {0x1e9b,0x1e9b,-1,-58}, 2692 {0x1e9e,0x1e9e,-1,-7615}, 2693 {0x1ea0,0x1efe,2,1}, 2694 {0x1f08,0x1f0f,1,-8}, 2695 {0x1f18,0x1f1d,1,-8}, 2696 {0x1f28,0x1f2f,1,-8}, 2697 {0x1f38,0x1f3f,1,-8}, 2698 {0x1f48,0x1f4d,1,-8}, 2699 {0x1f59,0x1f5f,2,-8}, 2700 {0x1f68,0x1f6f,1,-8}, 2701 {0x1f88,0x1f8f,1,-8}, 2702 {0x1f98,0x1f9f,1,-8}, 2703 {0x1fa8,0x1faf,1,-8}, 2704 {0x1fb8,0x1fb9,1,-8}, 2705 {0x1fba,0x1fbb,1,-74}, 2706 {0x1fbc,0x1fbc,-1,-9}, 2707 {0x1fbe,0x1fbe,-1,-7173}, 2708 {0x1fc8,0x1fcb,1,-86}, 2709 {0x1fcc,0x1fcc,-1,-9}, 2710 {0x1fd8,0x1fd9,1,-8}, 2711 {0x1fda,0x1fdb,1,-100}, 2712 {0x1fe8,0x1fe9,1,-8}, 2713 {0x1fea,0x1feb,1,-112}, 2714 {0x1fec,0x1fec,-1,-7}, 2715 {0x1ff8,0x1ff9,1,-128}, 2716 {0x1ffa,0x1ffb,1,-126}, 2717 {0x1ffc,0x1ffc,-1,-9}, 2718 {0x2126,0x2126,-1,-7517}, 2719 {0x212a,0x212a,-1,-8383}, 2720 {0x212b,0x212b,-1,-8262}, 2721 {0x2132,0x2132,-1,28}, 2722 {0x2160,0x216f,1,16}, 2723 {0x2183,0x2183,-1,1}, 2724 {0x24b6,0x24cf,1,26}, 2725 {0x2c00,0x2c2e,1,48}, 2726 {0x2c60,0x2c60,-1,1}, 2727 {0x2c62,0x2c62,-1,-10743}, 2728 {0x2c63,0x2c63,-1,-3814}, 2729 {0x2c64,0x2c64,-1,-10727}, 2730 {0x2c67,0x2c6b,2,1}, 2731 {0x2c6d,0x2c6d,-1,-10780}, 2732 {0x2c6e,0x2c6e,-1,-10749}, 2733 {0x2c6f,0x2c6f,-1,-10783}, 2734 {0x2c70,0x2c70,-1,-10782}, 2735 {0x2c72,0x2c75,3,1}, 2736 {0x2c7e,0x2c7f,1,-10815}, 2737 {0x2c80,0x2ce2,2,1}, 2738 {0x2ceb,0x2ced,2,1}, 2739 {0xa640,0xa65e,2,1}, 2740 {0xa662,0xa66c,2,1}, 2741 {0xa680,0xa696,2,1}, 2742 {0xa722,0xa72e,2,1}, 2743 {0xa732,0xa76e,2,1}, 2744 {0xa779,0xa77b,2,1}, 2745 {0xa77d,0xa77d,-1,-35332}, 2746 {0xa77e,0xa786,2,1}, 2747 {0xa78b,0xa78b,-1,1}, 2748 {0xff21,0xff3a,1,32}, 2749 {0x10400,0x10427,1,40} 2750 }; 2751 2752 static int utf_convert __ARGS((int a, convertStruct table[], int tableSize)); 2753 static int utf_strnicmp __ARGS((char_u *s1, char_u *s2, size_t n1, size_t n2)); 2754 2755 /* 2756 * Generic conversion function for case operations. 2757 * Return the converted equivalent of "a", which is a UCS-4 character. Use 2758 * the given conversion "table". Uses binary search on "table". 2759 */ 2760 static int 2761 utf_convert(a, table, tableSize) 2762 int a; 2763 convertStruct table[]; 2764 int tableSize; 2765 { 2766 int start, mid, end; /* indices into table */ 2767 int entries = tableSize / sizeof(convertStruct); 2768 2769 start = 0; 2770 end = entries; 2771 while (start < end) 2772 { 2773 /* need to search further */ 2774 mid = (end + start) / 2; 2775 if (table[mid].rangeEnd < a) 2776 start = mid + 1; 2777 else 2778 end = mid; 2779 } 2780 if (start < entries 2781 && table[start].rangeStart <= a 2782 && a <= table[start].rangeEnd 2783 && (a - table[start].rangeStart) % table[start].step == 0) 2784 return (a + table[start].offset); 2785 else 2786 return a; 2787 } 2788 2789 /* 2790 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 2791 * simple case folding. 2792 */ 2793 int 2794 utf_fold(a) 2795 int a; 2796 { 2797 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 2798 } 2799 2800 static convertStruct toLower[] = 2801 { 2802 {0x41,0x5a,1,32}, 2803 {0xc0,0xd6,1,32}, 2804 {0xd8,0xde,1,32}, 2805 {0x100,0x12e,2,1}, 2806 {0x130,0x130,-1,-199}, 2807 {0x132,0x136,2,1}, 2808 {0x139,0x147,2,1}, 2809 {0x14a,0x176,2,1}, 2810 {0x178,0x178,-1,-121}, 2811 {0x179,0x17d,2,1}, 2812 {0x181,0x181,-1,210}, 2813 {0x182,0x184,2,1}, 2814 {0x186,0x186,-1,206}, 2815 {0x187,0x187,-1,1}, 2816 {0x189,0x18a,1,205}, 2817 {0x18b,0x18b,-1,1}, 2818 {0x18e,0x18e,-1,79}, 2819 {0x18f,0x18f,-1,202}, 2820 {0x190,0x190,-1,203}, 2821 {0x191,0x191,-1,1}, 2822 {0x193,0x193,-1,205}, 2823 {0x194,0x194,-1,207}, 2824 {0x196,0x196,-1,211}, 2825 {0x197,0x197,-1,209}, 2826 {0x198,0x198,-1,1}, 2827 {0x19c,0x19c,-1,211}, 2828 {0x19d,0x19d,-1,213}, 2829 {0x19f,0x19f,-1,214}, 2830 {0x1a0,0x1a4,2,1}, 2831 {0x1a6,0x1a6,-1,218}, 2832 {0x1a7,0x1a7,-1,1}, 2833 {0x1a9,0x1a9,-1,218}, 2834 {0x1ac,0x1ac,-1,1}, 2835 {0x1ae,0x1ae,-1,218}, 2836 {0x1af,0x1af,-1,1}, 2837 {0x1b1,0x1b2,1,217}, 2838 {0x1b3,0x1b5,2,1}, 2839 {0x1b7,0x1b7,-1,219}, 2840 {0x1b8,0x1bc,4,1}, 2841 {0x1c4,0x1c4,-1,2}, 2842 {0x1c5,0x1c5,-1,1}, 2843 {0x1c7,0x1c7,-1,2}, 2844 {0x1c8,0x1c8,-1,1}, 2845 {0x1ca,0x1ca,-1,2}, 2846 {0x1cb,0x1db,2,1}, 2847 {0x1de,0x1ee,2,1}, 2848 {0x1f1,0x1f1,-1,2}, 2849 {0x1f2,0x1f4,2,1}, 2850 {0x1f6,0x1f6,-1,-97}, 2851 {0x1f7,0x1f7,-1,-56}, 2852 {0x1f8,0x21e,2,1}, 2853 {0x220,0x220,-1,-130}, 2854 {0x222,0x232,2,1}, 2855 {0x23a,0x23a,-1,10795}, 2856 {0x23b,0x23b,-1,1}, 2857 {0x23d,0x23d,-1,-163}, 2858 {0x23e,0x23e,-1,10792}, 2859 {0x241,0x241,-1,1}, 2860 {0x243,0x243,-1,-195}, 2861 {0x244,0x244,-1,69}, 2862 {0x245,0x245,-1,71}, 2863 {0x246,0x24e,2,1}, 2864 {0x370,0x372,2,1}, 2865 {0x376,0x376,-1,1}, 2866 {0x386,0x386,-1,38}, 2867 {0x388,0x38a,1,37}, 2868 {0x38c,0x38c,-1,64}, 2869 {0x38e,0x38f,1,63}, 2870 {0x391,0x3a1,1,32}, 2871 {0x3a3,0x3ab,1,32}, 2872 {0x3cf,0x3cf,-1,8}, 2873 {0x3d8,0x3ee,2,1}, 2874 {0x3f4,0x3f4,-1,-60}, 2875 {0x3f7,0x3f7,-1,1}, 2876 {0x3f9,0x3f9,-1,-7}, 2877 {0x3fa,0x3fa,-1,1}, 2878 {0x3fd,0x3ff,1,-130}, 2879 {0x400,0x40f,1,80}, 2880 {0x410,0x42f,1,32}, 2881 {0x460,0x480,2,1}, 2882 {0x48a,0x4be,2,1}, 2883 {0x4c0,0x4c0,-1,15}, 2884 {0x4c1,0x4cd,2,1}, 2885 {0x4d0,0x524,2,1}, 2886 {0x531,0x556,1,48}, 2887 {0x10a0,0x10c5,1,7264}, 2888 {0x1e00,0x1e94,2,1}, 2889 {0x1e9e,0x1e9e,-1,-7615}, 2890 {0x1ea0,0x1efe,2,1}, 2891 {0x1f08,0x1f0f,1,-8}, 2892 {0x1f18,0x1f1d,1,-8}, 2893 {0x1f28,0x1f2f,1,-8}, 2894 {0x1f38,0x1f3f,1,-8}, 2895 {0x1f48,0x1f4d,1,-8}, 2896 {0x1f59,0x1f5f,2,-8}, 2897 {0x1f68,0x1f6f,1,-8}, 2898 {0x1f88,0x1f8f,1,-8}, 2899 {0x1f98,0x1f9f,1,-8}, 2900 {0x1fa8,0x1faf,1,-8}, 2901 {0x1fb8,0x1fb9,1,-8}, 2902 {0x1fba,0x1fbb,1,-74}, 2903 {0x1fbc,0x1fbc,-1,-9}, 2904 {0x1fc8,0x1fcb,1,-86}, 2905 {0x1fcc,0x1fcc,-1,-9}, 2906 {0x1fd8,0x1fd9,1,-8}, 2907 {0x1fda,0x1fdb,1,-100}, 2908 {0x1fe8,0x1fe9,1,-8}, 2909 {0x1fea,0x1feb,1,-112}, 2910 {0x1fec,0x1fec,-1,-7}, 2911 {0x1ff8,0x1ff9,1,-128}, 2912 {0x1ffa,0x1ffb,1,-126}, 2913 {0x1ffc,0x1ffc,-1,-9}, 2914 {0x2126,0x2126,-1,-7517}, 2915 {0x212a,0x212a,-1,-8383}, 2916 {0x212b,0x212b,-1,-8262}, 2917 {0x2132,0x2132,-1,28}, 2918 {0x2160,0x216f,1,16}, 2919 {0x2183,0x2183,-1,1}, 2920 {0x24b6,0x24cf,1,26}, 2921 {0x2c00,0x2c2e,1,48}, 2922 {0x2c60,0x2c60,-1,1}, 2923 {0x2c62,0x2c62,-1,-10743}, 2924 {0x2c63,0x2c63,-1,-3814}, 2925 {0x2c64,0x2c64,-1,-10727}, 2926 {0x2c67,0x2c6b,2,1}, 2927 {0x2c6d,0x2c6d,-1,-10780}, 2928 {0x2c6e,0x2c6e,-1,-10749}, 2929 {0x2c6f,0x2c6f,-1,-10783}, 2930 {0x2c70,0x2c70,-1,-10782}, 2931 {0x2c72,0x2c75,3,1}, 2932 {0x2c7e,0x2c7f,1,-10815}, 2933 {0x2c80,0x2ce2,2,1}, 2934 {0x2ceb,0x2ced,2,1}, 2935 {0xa640,0xa65e,2,1}, 2936 {0xa662,0xa66c,2,1}, 2937 {0xa680,0xa696,2,1}, 2938 {0xa722,0xa72e,2,1}, 2939 {0xa732,0xa76e,2,1}, 2940 {0xa779,0xa77b,2,1}, 2941 {0xa77d,0xa77d,-1,-35332}, 2942 {0xa77e,0xa786,2,1}, 2943 {0xa78b,0xa78b,-1,1}, 2944 {0xff21,0xff3a,1,32}, 2945 {0x10400,0x10427,1,40} 2946 }; 2947 2948 static convertStruct toUpper[] = 2949 { 2950 {0x61,0x7a,1,-32}, 2951 {0xb5,0xb5,-1,743}, 2952 {0xe0,0xf6,1,-32}, /* 0xdf (German sharp s) is not upper-cased */ 2953 {0xf8,0xfe,1,-32}, 2954 {0xff,0xff,-1,121}, 2955 {0x101,0x12f,2,-1}, 2956 {0x131,0x131,-1,-232}, 2957 {0x133,0x137,2,-1}, 2958 {0x13a,0x148,2,-1}, 2959 {0x14b,0x177,2,-1}, 2960 {0x17a,0x17e,2,-1}, 2961 {0x17f,0x17f,-1,-300}, 2962 {0x180,0x180,-1,195}, 2963 {0x183,0x185,2,-1}, 2964 {0x188,0x18c,4,-1}, 2965 {0x192,0x192,-1,-1}, 2966 {0x195,0x195,-1,97}, 2967 {0x199,0x199,-1,-1}, 2968 {0x19a,0x19a,-1,163}, 2969 {0x19e,0x19e,-1,130}, 2970 {0x1a1,0x1a5,2,-1}, 2971 {0x1a8,0x1ad,5,-1}, 2972 {0x1b0,0x1b4,4,-1}, 2973 {0x1b6,0x1b9,3,-1}, 2974 {0x1bd,0x1bd,-1,-1}, 2975 {0x1bf,0x1bf,-1,56}, 2976 {0x1c5,0x1c5,-1,-1}, 2977 {0x1c6,0x1c6,-1,-2}, 2978 {0x1c8,0x1c8,-1,-1}, 2979 {0x1c9,0x1c9,-1,-2}, 2980 {0x1cb,0x1cb,-1,-1}, 2981 {0x1cc,0x1cc,-1,-2}, 2982 {0x1ce,0x1dc,2,-1}, 2983 {0x1dd,0x1dd,-1,-79}, 2984 {0x1df,0x1ef,2,-1}, 2985 {0x1f2,0x1f2,-1,-1}, 2986 {0x1f3,0x1f3,-1,-2}, 2987 {0x1f5,0x1f9,4,-1}, 2988 {0x1fb,0x21f,2,-1}, 2989 {0x223,0x233,2,-1}, 2990 {0x23c,0x23c,-1,-1}, 2991 {0x23f,0x240,1,10815}, 2992 {0x242,0x247,5,-1}, 2993 {0x249,0x24f,2,-1}, 2994 {0x250,0x250,-1,10783}, 2995 {0x251,0x251,-1,10780}, 2996 {0x252,0x252,-1,10782}, 2997 {0x253,0x253,-1,-210}, 2998 {0x254,0x254,-1,-206}, 2999 {0x256,0x257,1,-205}, 3000 {0x259,0x259,-1,-202}, 3001 {0x25b,0x25b,-1,-203}, 3002 {0x260,0x260,-1,-205}, 3003 {0x263,0x263,-1,-207}, 3004 {0x268,0x268,-1,-209}, 3005 {0x269,0x269,-1,-211}, 3006 {0x26b,0x26b,-1,10743}, 3007 {0x26f,0x26f,-1,-211}, 3008 {0x271,0x271,-1,10749}, 3009 {0x272,0x272,-1,-213}, 3010 {0x275,0x275,-1,-214}, 3011 {0x27d,0x27d,-1,10727}, 3012 {0x280,0x283,3,-218}, 3013 {0x288,0x288,-1,-218}, 3014 {0x289,0x289,-1,-69}, 3015 {0x28a,0x28b,1,-217}, 3016 {0x28c,0x28c,-1,-71}, 3017 {0x292,0x292,-1,-219}, 3018 {0x345,0x345,-1,84}, 3019 {0x371,0x373,2,-1}, 3020 {0x377,0x377,-1,-1}, 3021 {0x37b,0x37d,1,130}, 3022 {0x3ac,0x3ac,-1,-38}, 3023 {0x3ad,0x3af,1,-37}, 3024 {0x3b1,0x3c1,1,-32}, 3025 {0x3c2,0x3c2,-1,-31}, 3026 {0x3c3,0x3cb,1,-32}, 3027 {0x3cc,0x3cc,-1,-64}, 3028 {0x3cd,0x3ce,1,-63}, 3029 {0x3d0,0x3d0,-1,-62}, 3030 {0x3d1,0x3d1,-1,-57}, 3031 {0x3d5,0x3d5,-1,-47}, 3032 {0x3d6,0x3d6,-1,-54}, 3033 {0x3d7,0x3d7,-1,-8}, 3034 {0x3d9,0x3ef,2,-1}, 3035 {0x3f0,0x3f0,-1,-86}, 3036 {0x3f1,0x3f1,-1,-80}, 3037 {0x3f2,0x3f2,-1,7}, 3038 {0x3f5,0x3f5,-1,-96}, 3039 {0x3f8,0x3fb,3,-1}, 3040 {0x430,0x44f,1,-32}, 3041 {0x450,0x45f,1,-80}, 3042 {0x461,0x481,2,-1}, 3043 {0x48b,0x4bf,2,-1}, 3044 {0x4c2,0x4ce,2,-1}, 3045 {0x4cf,0x4cf,-1,-15}, 3046 {0x4d1,0x525,2,-1}, 3047 {0x561,0x586,1,-48}, 3048 {0x1d79,0x1d79,-1,35332}, 3049 {0x1d7d,0x1d7d,-1,3814}, 3050 {0x1e01,0x1e95,2,-1}, 3051 {0x1e9b,0x1e9b,-1,-59}, 3052 {0x1ea1,0x1eff,2,-1}, 3053 {0x1f00,0x1f07,1,8}, 3054 {0x1f10,0x1f15,1,8}, 3055 {0x1f20,0x1f27,1,8}, 3056 {0x1f30,0x1f37,1,8}, 3057 {0x1f40,0x1f45,1,8}, 3058 {0x1f51,0x1f57,2,8}, 3059 {0x1f60,0x1f67,1,8}, 3060 {0x1f70,0x1f71,1,74}, 3061 {0x1f72,0x1f75,1,86}, 3062 {0x1f76,0x1f77,1,100}, 3063 {0x1f78,0x1f79,1,128}, 3064 {0x1f7a,0x1f7b,1,112}, 3065 {0x1f7c,0x1f7d,1,126}, 3066 {0x1f80,0x1f87,1,8}, 3067 {0x1f90,0x1f97,1,8}, 3068 {0x1fa0,0x1fa7,1,8}, 3069 {0x1fb0,0x1fb1,1,8}, 3070 {0x1fb3,0x1fb3,-1,9}, 3071 {0x1fbe,0x1fbe,-1,-7205}, 3072 {0x1fc3,0x1fc3,-1,9}, 3073 {0x1fd0,0x1fd1,1,8}, 3074 {0x1fe0,0x1fe1,1,8}, 3075 {0x1fe5,0x1fe5,-1,7}, 3076 {0x1ff3,0x1ff3,-1,9}, 3077 {0x214e,0x214e,-1,-28}, 3078 {0x2170,0x217f,1,-16}, 3079 {0x2184,0x2184,-1,-1}, 3080 {0x24d0,0x24e9,1,-26}, 3081 {0x2c30,0x2c5e,1,-48}, 3082 {0x2c61,0x2c61,-1,-1}, 3083 {0x2c65,0x2c65,-1,-10795}, 3084 {0x2c66,0x2c66,-1,-10792}, 3085 {0x2c68,0x2c6c,2,-1}, 3086 {0x2c73,0x2c76,3,-1}, 3087 {0x2c81,0x2ce3,2,-1}, 3088 {0x2cec,0x2cee,2,-1}, 3089 {0x2d00,0x2d25,1,-7264}, 3090 {0xa641,0xa65f,2,-1}, 3091 {0xa663,0xa66d,2,-1}, 3092 {0xa681,0xa697,2,-1}, 3093 {0xa723,0xa72f,2,-1}, 3094 {0xa733,0xa76f,2,-1}, 3095 {0xa77a,0xa77c,2,-1}, 3096 {0xa77f,0xa787,2,-1}, 3097 {0xa78c,0xa78c,-1,-1}, 3098 {0xff41,0xff5a,1,-32}, 3099 {0x10428,0x1044f,1,-40} 3100 }; 3101 3102 /* 3103 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3104 * simple case folding. 3105 */ 3106 int 3107 utf_toupper(a) 3108 int a; 3109 { 3110 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */ 3111 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3112 return TOUPPER_ASC(a); 3113 3114 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3115 /* If towupper() is available and handles Unicode, use it. */ 3116 if (!(cmp_flags & CMP_INTERNAL)) 3117 return towupper(a); 3118 #endif 3119 3120 /* For characters below 128 use locale sensitive toupper(). */ 3121 if (a < 128) 3122 return TOUPPER_LOC(a); 3123 3124 /* For any other characters use the above mapping table. */ 3125 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3126 } 3127 3128 int 3129 utf_islower(a) 3130 int a; 3131 { 3132 /* German sharp s is lower case but has no upper case equivalent. */ 3133 return (utf_toupper(a) != a) || a == 0xdf; 3134 } 3135 3136 /* 3137 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3138 * simple case folding. 3139 */ 3140 int 3141 utf_tolower(a) 3142 int a; 3143 { 3144 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */ 3145 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3146 return TOLOWER_ASC(a); 3147 3148 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3149 /* If towlower() is available and handles Unicode, use it. */ 3150 if (!(cmp_flags & CMP_INTERNAL)) 3151 return towlower(a); 3152 #endif 3153 3154 /* For characters below 128 use locale sensitive tolower(). */ 3155 if (a < 128) 3156 return TOLOWER_LOC(a); 3157 3158 /* For any other characters use the above mapping table. */ 3159 return utf_convert(a, toLower, (int)sizeof(toLower)); 3160 } 3161 3162 int 3163 utf_isupper(a) 3164 int a; 3165 { 3166 return (utf_tolower(a) != a); 3167 } 3168 3169 static int 3170 utf_strnicmp(s1, s2, n1, n2) 3171 char_u *s1, *s2; 3172 size_t n1, n2; 3173 { 3174 int c1, c2, cdiff; 3175 char_u buffer[6]; 3176 3177 for (;;) 3178 { 3179 c1 = utf_safe_read_char_adv(&s1, &n1); 3180 c2 = utf_safe_read_char_adv(&s2, &n2); 3181 3182 if (c1 <= 0 || c2 <= 0) 3183 break; 3184 3185 if (c1 == c2) 3186 continue; 3187 3188 cdiff = utf_fold(c1) - utf_fold(c2); 3189 if (cdiff != 0) 3190 return cdiff; 3191 } 3192 3193 /* some string ended or has an incomplete/illegal character sequence */ 3194 3195 if (c1 == 0 || c2 == 0) 3196 { 3197 /* some string ended. shorter string is smaller */ 3198 if (c1 == 0 && c2 == 0) 3199 return 0; 3200 return c1 == 0 ? -1 : 1; 3201 } 3202 3203 /* Continue with bytewise comparison to produce some result that 3204 * would make comparison operations involving this function transitive. 3205 * 3206 * If only one string had an error, comparison should be made with 3207 * folded version of the other string. In this case it is enough 3208 * to fold just one character to determine the result of comparison. */ 3209 3210 if (c1 != -1 && c2 == -1) 3211 { 3212 n1 = utf_char2bytes(utf_fold(c1), buffer); 3213 s1 = buffer; 3214 } 3215 else if (c2 != -1 && c1 == -1) 3216 { 3217 n2 = utf_char2bytes(utf_fold(c2), buffer); 3218 s2 = buffer; 3219 } 3220 3221 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3222 { 3223 cdiff = (int)(*s1) - (int)(*s2); 3224 if (cdiff != 0) 3225 return cdiff; 3226 3227 s1++; 3228 s2++; 3229 n1--; 3230 n2--; 3231 } 3232 3233 if (n1 > 0 && *s1 == NUL) 3234 n1 = 0; 3235 if (n2 > 0 && *s2 == NUL) 3236 n2 = 0; 3237 3238 if (n1 == 0 && n2 == 0) 3239 return 0; 3240 return n1 == 0 ? -1 : 1; 3241 } 3242 3243 /* 3244 * Version of strnicmp() that handles multi-byte characters. 3245 * Needed for Big5, Sjift-JIS and UTF-8 encoding. Other DBCS encodings can 3246 * probably use strnicmp(), because there are no ASCII characters in the 3247 * second byte. 3248 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3249 * two characters otherwise. 3250 */ 3251 int 3252 mb_strnicmp(s1, s2, nn) 3253 char_u *s1, *s2; 3254 size_t nn; 3255 { 3256 int i, l; 3257 int cdiff; 3258 int n = (int)nn; 3259 3260 if (enc_utf8) 3261 { 3262 return utf_strnicmp(s1, s2, nn, nn); 3263 } 3264 else 3265 { 3266 for (i = 0; i < n; i += l) 3267 { 3268 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */ 3269 return 0; 3270 3271 l = (*mb_ptr2len)(s1 + i); 3272 if (l <= 1) 3273 { 3274 /* Single byte: first check normally, then with ignore case. */ 3275 if (s1[i] != s2[i]) 3276 { 3277 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3278 if (cdiff != 0) 3279 return cdiff; 3280 } 3281 } 3282 else 3283 { 3284 /* For non-Unicode multi-byte don't ignore case. */ 3285 if (l > n - i) 3286 l = n - i; 3287 cdiff = STRNCMP(s1 + i, s2 + i, l); 3288 if (cdiff != 0) 3289 return cdiff; 3290 } 3291 } 3292 } 3293 return 0; 3294 } 3295 3296 /* 3297 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3298 * 'encoding' has been set to. 3299 */ 3300 void 3301 show_utf8() 3302 { 3303 int len; 3304 int rlen = 0; 3305 char_u *line; 3306 int clen; 3307 int i; 3308 3309 /* Get the byte length of the char under the cursor, including composing 3310 * characters. */ 3311 line = ml_get_cursor(); 3312 len = utfc_ptr2len(line); 3313 if (len == 0) 3314 { 3315 MSG("NUL"); 3316 return; 3317 } 3318 3319 clen = 0; 3320 for (i = 0; i < len; ++i) 3321 { 3322 if (clen == 0) 3323 { 3324 /* start of (composing) character, get its length */ 3325 if (i > 0) 3326 { 3327 STRCPY(IObuff + rlen, "+ "); 3328 rlen += 2; 3329 } 3330 clen = utf_ptr2len(line + i); 3331 } 3332 sprintf((char *)IObuff + rlen, "%02x ", 3333 (line[i] == NL) ? NUL : line[i]); /* NUL is stored as NL */ 3334 --clen; 3335 rlen += (int)STRLEN(IObuff + rlen); 3336 if (rlen > IOSIZE - 20) 3337 break; 3338 } 3339 3340 msg(IObuff); 3341 } 3342 3343 /* 3344 * mb_head_off() function pointer. 3345 * Return offset from "p" to the first byte of the character it points into. 3346 * If "p" points to the NUL at the end of the string return 0. 3347 * Returns 0 when already at the first byte of a character. 3348 */ 3349 int 3350 latin_head_off(base, p) 3351 char_u *base UNUSED; 3352 char_u *p UNUSED; 3353 { 3354 return 0; 3355 } 3356 3357 int 3358 dbcs_head_off(base, p) 3359 char_u *base; 3360 char_u *p; 3361 { 3362 char_u *q; 3363 3364 /* It can't be a trailing byte when not using DBCS, at the start of the 3365 * string or the previous byte can't start a double-byte. */ 3366 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3367 return 0; 3368 3369 /* This is slow: need to start at the base and go forward until the 3370 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */ 3371 q = base; 3372 while (q < p) 3373 q += dbcs_ptr2len(q); 3374 return (q == p) ? 0 : 1; 3375 } 3376 3377 /* 3378 * Special version of dbcs_head_off() that works for ScreenLines[], where 3379 * single-width DBCS_JPNU characters are stored separately. 3380 */ 3381 int 3382 dbcs_screen_head_off(base, p) 3383 char_u *base; 3384 char_u *p; 3385 { 3386 char_u *q; 3387 3388 /* It can't be a trailing byte when not using DBCS, at the start of the 3389 * string or the previous byte can't start a double-byte. 3390 * For euc-jp an 0x8e byte in the previous cell always means we have a 3391 * lead byte in the current cell. */ 3392 if (p <= base 3393 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3394 || MB_BYTE2LEN(p[-1]) == 1 3395 || *p == NUL) 3396 return 0; 3397 3398 /* This is slow: need to start at the base and go forward until the 3399 * byte we are looking for. Return 1 when we went past it, 0 otherwise. 3400 * For DBCS_JPNU look out for 0x8e, which means the second byte is not 3401 * stored as the next byte. */ 3402 q = base; 3403 while (q < p) 3404 { 3405 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3406 ++q; 3407 else 3408 q += dbcs_ptr2len(q); 3409 } 3410 return (q == p) ? 0 : 1; 3411 } 3412 3413 int 3414 utf_head_off(base, p) 3415 char_u *base; 3416 char_u *p; 3417 { 3418 char_u *q; 3419 char_u *s; 3420 int c; 3421 int len; 3422 #ifdef FEAT_ARABIC 3423 char_u *j; 3424 #endif 3425 3426 if (*p < 0x80) /* be quick for ASCII */ 3427 return 0; 3428 3429 /* Skip backwards over trailing bytes: 10xx.xxxx 3430 * Skip backwards again if on a composing char. */ 3431 for (q = p; ; --q) 3432 { 3433 /* Move s to the last byte of this char. */ 3434 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3435 ; 3436 /* Move q to the first byte of this char. */ 3437 while (q > base && (*q & 0xc0) == 0x80) 3438 --q; 3439 /* Check for illegal sequence. Do allow an illegal byte after where we 3440 * started. */ 3441 len = utf8len_tab[*q]; 3442 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3443 return 0; 3444 3445 if (q <= base) 3446 break; 3447 3448 c = utf_ptr2char(q); 3449 if (utf_iscomposing(c)) 3450 continue; 3451 3452 #ifdef FEAT_ARABIC 3453 if (arabic_maycombine(c)) 3454 { 3455 /* Advance to get a sneak-peak at the next char */ 3456 j = q; 3457 --j; 3458 /* Move j to the first byte of this char. */ 3459 while (j > base && (*j & 0xc0) == 0x80) 3460 --j; 3461 if (arabic_combine(utf_ptr2char(j), c)) 3462 continue; 3463 } 3464 #endif 3465 break; 3466 } 3467 3468 return (int)(p - q); 3469 } 3470 3471 /* 3472 * Copy a character from "*fp" to "*tp" and advance the pointers. 3473 */ 3474 void 3475 mb_copy_char(fp, tp) 3476 char_u **fp; 3477 char_u **tp; 3478 { 3479 int l = (*mb_ptr2len)(*fp); 3480 3481 mch_memmove(*tp, *fp, (size_t)l); 3482 *tp += l; 3483 *fp += l; 3484 } 3485 3486 /* 3487 * Return the offset from "p" to the first byte of a character. When "p" is 3488 * at the start of a character 0 is returned, otherwise the offset to the next 3489 * character. Can start anywhere in a stream of bytes. 3490 */ 3491 int 3492 mb_off_next(base, p) 3493 char_u *base; 3494 char_u *p; 3495 { 3496 int i; 3497 int j; 3498 3499 if (enc_utf8) 3500 { 3501 if (*p < 0x80) /* be quick for ASCII */ 3502 return 0; 3503 3504 /* Find the next character that isn't 10xx.xxxx */ 3505 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 3506 ; 3507 if (i > 0) 3508 { 3509 /* Check for illegal sequence. */ 3510 for (j = 0; p - j > base; ++j) 3511 if ((p[-j] & 0xc0) != 0x80) 3512 break; 3513 if (utf8len_tab[p[-j]] != i + j) 3514 return 0; 3515 } 3516 return i; 3517 } 3518 3519 /* Only need to check if we're on a trail byte, it doesn't matter if we 3520 * want the offset to the next or current character. */ 3521 return (*mb_head_off)(base, p); 3522 } 3523 3524 /* 3525 * Return the offset from "p" to the last byte of the character it points 3526 * into. Can start anywhere in a stream of bytes. 3527 */ 3528 int 3529 mb_tail_off(base, p) 3530 char_u *base; 3531 char_u *p; 3532 { 3533 int i; 3534 int j; 3535 3536 if (*p == NUL) 3537 return 0; 3538 3539 if (enc_utf8) 3540 { 3541 /* Find the last character that is 10xx.xxxx */ 3542 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 3543 ; 3544 /* Check for illegal sequence. */ 3545 for (j = 0; p - j > base; ++j) 3546 if ((p[-j] & 0xc0) != 0x80) 3547 break; 3548 if (utf8len_tab[p[-j]] != i + j + 1) 3549 return 0; 3550 return i; 3551 } 3552 3553 /* It can't be the first byte if a double-byte when not using DBCS, at the 3554 * end of the string or the byte can't start a double-byte. */ 3555 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 3556 return 0; 3557 3558 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3559 return 1 - dbcs_head_off(base, p); 3560 } 3561 3562 /* 3563 * Find the next illegal byte sequence. 3564 */ 3565 void 3566 utf_find_illegal() 3567 { 3568 pos_T pos = curwin->w_cursor; 3569 char_u *p; 3570 int len; 3571 vimconv_T vimconv; 3572 char_u *tofree = NULL; 3573 3574 vimconv.vc_type = CONV_NONE; 3575 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 3576 { 3577 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 3578 * possibly a utf-8 file with illegal bytes. Setup for conversion 3579 * from utf-8 to 'fileencoding'. */ 3580 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 3581 } 3582 3583 #ifdef FEAT_VIRTUALEDIT 3584 curwin->w_cursor.coladd = 0; 3585 #endif 3586 for (;;) 3587 { 3588 p = ml_get_cursor(); 3589 if (vimconv.vc_type != CONV_NONE) 3590 { 3591 vim_free(tofree); 3592 tofree = string_convert(&vimconv, p, NULL); 3593 if (tofree == NULL) 3594 break; 3595 p = tofree; 3596 } 3597 3598 while (*p != NUL) 3599 { 3600 /* Illegal means that there are not enough trail bytes (checked by 3601 * utf_ptr2len()) or too many of them (overlong sequence). */ 3602 len = utf_ptr2len(p); 3603 if (*p >= 0x80 && (len == 1 3604 || utf_char2len(utf_ptr2char(p)) != len)) 3605 { 3606 if (vimconv.vc_type == CONV_NONE) 3607 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 3608 else 3609 { 3610 int l; 3611 3612 len = (int)(p - tofree); 3613 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 3614 { 3615 l = utf_ptr2len(p); 3616 curwin->w_cursor.col += l; 3617 } 3618 } 3619 goto theend; 3620 } 3621 p += len; 3622 } 3623 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 3624 break; 3625 ++curwin->w_cursor.lnum; 3626 curwin->w_cursor.col = 0; 3627 } 3628 3629 /* didn't find it: don't move and beep */ 3630 curwin->w_cursor = pos; 3631 beep_flush(); 3632 3633 theend: 3634 vim_free(tofree); 3635 convert_setup(&vimconv, NULL, NULL); 3636 } 3637 3638 #if defined(FEAT_GUI_GTK) || defined(PROTO) 3639 /* 3640 * Return TRUE if string "s" is a valid utf-8 string. 3641 * When "end" is NULL stop at the first NUL. 3642 * When "end" is positive stop there. 3643 */ 3644 int 3645 utf_valid_string(s, end) 3646 char_u *s; 3647 char_u *end; 3648 { 3649 int l; 3650 char_u *p = s; 3651 3652 while (end == NULL ? *p != NUL : p < end) 3653 { 3654 l = utf8len_tab_zero[*p]; 3655 if (l == 0) 3656 return FALSE; /* invalid lead byte */ 3657 if (end != NULL && p + l > end) 3658 return FALSE; /* incomplete byte sequence */ 3659 ++p; 3660 while (--l > 0) 3661 if ((*p++ & 0xc0) != 0x80) 3662 return FALSE; /* invalid trail byte */ 3663 } 3664 return TRUE; 3665 } 3666 #endif 3667 3668 #if defined(FEAT_GUI) || defined(PROTO) 3669 /* 3670 * Special version of mb_tail_off() for use in ScreenLines[]. 3671 */ 3672 int 3673 dbcs_screen_tail_off(base, p) 3674 char_u *base; 3675 char_u *p; 3676 { 3677 /* It can't be the first byte if a double-byte when not using DBCS, at the 3678 * end of the string or the byte can't start a double-byte. 3679 * For euc-jp an 0x8e byte always means we have a lead byte in the current 3680 * cell. */ 3681 if (*p == NUL || p[1] == NUL 3682 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 3683 || MB_BYTE2LEN(*p) == 1) 3684 return 0; 3685 3686 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3687 return 1 - dbcs_screen_head_off(base, p); 3688 } 3689 #endif 3690 3691 /* 3692 * If the cursor moves on an trail byte, set the cursor on the lead byte. 3693 * Thus it moves left if necessary. 3694 * Return TRUE when the cursor was adjusted. 3695 */ 3696 void 3697 mb_adjust_cursor() 3698 { 3699 mb_adjustpos(curbuf, &curwin->w_cursor); 3700 } 3701 3702 /* 3703 * Adjust position "*lp" to point to the first byte of a multi-byte character. 3704 * If it points to a tail byte it's moved backwards to the head byte. 3705 */ 3706 void 3707 mb_adjustpos(buf, lp) 3708 buf_T *buf; 3709 pos_T *lp; 3710 { 3711 char_u *p; 3712 3713 if (lp->col > 0 3714 #ifdef FEAT_VIRTUALEDIT 3715 || lp->coladd > 1 3716 #endif 3717 ) 3718 { 3719 p = ml_get_buf(buf, lp->lnum, FALSE); 3720 lp->col -= (*mb_head_off)(p, p + lp->col); 3721 #ifdef FEAT_VIRTUALEDIT 3722 /* Reset "coladd" when the cursor would be on the right half of a 3723 * double-wide character. */ 3724 if (lp->coladd == 1 3725 && p[lp->col] != TAB 3726 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 3727 && ptr2cells(p + lp->col) > 1) 3728 lp->coladd = 0; 3729 #endif 3730 } 3731 } 3732 3733 /* 3734 * Return a pointer to the character before "*p", if there is one. 3735 */ 3736 char_u * 3737 mb_prevptr(line, p) 3738 char_u *line; /* start of the string */ 3739 char_u *p; 3740 { 3741 if (p > line) 3742 mb_ptr_back(line, p); 3743 return p; 3744 } 3745 3746 /* 3747 * Return the character length of "str". Each multi-byte character (with 3748 * following composing characters) counts as one. 3749 */ 3750 int 3751 mb_charlen(str) 3752 char_u *str; 3753 { 3754 char_u *p = str; 3755 int count; 3756 3757 if (p == NULL) 3758 return 0; 3759 3760 for (count = 0; *p != NUL; count++) 3761 p += (*mb_ptr2len)(p); 3762 3763 return count; 3764 } 3765 3766 #if defined(FEAT_SPELL) || defined(PROTO) 3767 /* 3768 * Like mb_charlen() but for a string with specified length. 3769 */ 3770 int 3771 mb_charlen_len(str, len) 3772 char_u *str; 3773 int len; 3774 { 3775 char_u *p = str; 3776 int count; 3777 3778 for (count = 0; *p != NUL && p < str + len; count++) 3779 p += (*mb_ptr2len)(p); 3780 3781 return count; 3782 } 3783 #endif 3784 3785 /* 3786 * Try to un-escape a multi-byte character. 3787 * Used for the "to" and "from" part of a mapping. 3788 * Return the un-escaped string if it is a multi-byte character, and advance 3789 * "pp" to just after the bytes that formed it. 3790 * Return NULL if no multi-byte char was found. 3791 */ 3792 char_u * 3793 mb_unescape(pp) 3794 char_u **pp; 3795 { 3796 static char_u buf[6]; 3797 int n; 3798 int m = 0; 3799 char_u *str = *pp; 3800 3801 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 3802 * KS_EXTRA KE_CSI to CSI. 3803 * Maximum length of a utf-8 character is 4 bytes. */ 3804 for (n = 0; str[n] != NUL && m < 4; ++n) 3805 { 3806 if (str[n] == K_SPECIAL 3807 && str[n + 1] == KS_SPECIAL 3808 && str[n + 2] == KE_FILLER) 3809 { 3810 buf[m++] = K_SPECIAL; 3811 n += 2; 3812 } 3813 else if ((str[n] == K_SPECIAL 3814 # ifdef FEAT_GUI 3815 || str[n] == CSI 3816 # endif 3817 ) 3818 && str[n + 1] == KS_EXTRA 3819 && str[n + 2] == (int)KE_CSI) 3820 { 3821 buf[m++] = CSI; 3822 n += 2; 3823 } 3824 else if (str[n] == K_SPECIAL 3825 # ifdef FEAT_GUI 3826 || str[n] == CSI 3827 # endif 3828 ) 3829 break; /* a special key can't be a multibyte char */ 3830 else 3831 buf[m++] = str[n]; 3832 buf[m] = NUL; 3833 3834 /* Return a multi-byte character if it's found. An illegal sequence 3835 * will result in a 1 here. */ 3836 if ((*mb_ptr2len)(buf) > 1) 3837 { 3838 *pp = str + n + 1; 3839 return buf; 3840 } 3841 3842 /* Bail out quickly for ASCII. */ 3843 if (buf[0] < 128) 3844 break; 3845 } 3846 return NULL; 3847 } 3848 3849 /* 3850 * Return TRUE if the character at "row"/"col" on the screen is the left side 3851 * of a double-width character. 3852 * Caller must make sure "row" and "col" are not invalid! 3853 */ 3854 int 3855 mb_lefthalve(row, col) 3856 int row; 3857 int col; 3858 { 3859 #ifdef FEAT_HANGULIN 3860 if (composing_hangul) 3861 return TRUE; 3862 #endif 3863 return (*mb_off2cells)(LineOffset[row] + col, 3864 LineOffset[row] + screen_Columns) > 1; 3865 } 3866 3867 /* 3868 * Correct a position on the screen, if it's the right half of a double-wide 3869 * char move it to the left half. Returns the corrected column. 3870 */ 3871 int 3872 mb_fix_col(col, row) 3873 int col; 3874 int row; 3875 { 3876 col = check_col(col); 3877 row = check_row(row); 3878 if (has_mbyte && ScreenLines != NULL && col > 0 3879 && ((enc_dbcs 3880 && ScreenLines[LineOffset[row] + col] != NUL 3881 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 3882 ScreenLines + LineOffset[row] + col)) 3883 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0))) 3884 return col - 1; 3885 return col; 3886 } 3887 #endif 3888 3889 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 3890 static int enc_alias_search __ARGS((char_u *name)); 3891 3892 /* 3893 * Skip the Vim specific head of a 'encoding' name. 3894 */ 3895 char_u * 3896 enc_skip(p) 3897 char_u *p; 3898 { 3899 if (STRNCMP(p, "2byte-", 6) == 0) 3900 return p + 6; 3901 if (STRNCMP(p, "8bit-", 5) == 0) 3902 return p + 5; 3903 return p; 3904 } 3905 3906 /* 3907 * Find the canonical name for encoding "enc". 3908 * When the name isn't recognized, returns "enc" itself, but with all lower 3909 * case characters and '_' replaced with '-'. 3910 * Returns an allocated string. NULL for out-of-memory. 3911 */ 3912 char_u * 3913 enc_canonize(enc) 3914 char_u *enc; 3915 { 3916 char_u *r; 3917 char_u *p, *s; 3918 int i; 3919 3920 # ifdef FEAT_MBYTE 3921 if (STRCMP(enc, "default") == 0) 3922 { 3923 /* Use the default encoding as it's found by set_init_1(). */ 3924 r = get_encoding_default(); 3925 if (r == NULL) 3926 r = (char_u *)"latin1"; 3927 return vim_strsave(r); 3928 } 3929 # endif 3930 3931 /* copy "enc" to allocated memory, with room for two '-' */ 3932 r = alloc((unsigned)(STRLEN(enc) + 3)); 3933 if (r != NULL) 3934 { 3935 /* Make it all lower case and replace '_' with '-'. */ 3936 p = r; 3937 for (s = enc; *s != NUL; ++s) 3938 { 3939 if (*s == '_') 3940 *p++ = '-'; 3941 else 3942 *p++ = TOLOWER_ASC(*s); 3943 } 3944 *p = NUL; 3945 3946 /* Skip "2byte-" and "8bit-". */ 3947 p = enc_skip(r); 3948 3949 /* Change "microsoft-cp" to "cp". Used in some spell files. */ 3950 if (STRNCMP(p, "microsoft-cp", 12) == 0) 3951 STRMOVE(p, p + 10); 3952 3953 /* "iso8859" -> "iso-8859" */ 3954 if (STRNCMP(p, "iso8859", 7) == 0) 3955 { 3956 STRMOVE(p + 4, p + 3); 3957 p[3] = '-'; 3958 } 3959 3960 /* "iso-8859n" -> "iso-8859-n" */ 3961 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 3962 { 3963 STRMOVE(p + 9, p + 8); 3964 p[8] = '-'; 3965 } 3966 3967 /* "latin-N" -> "latinN" */ 3968 if (STRNCMP(p, "latin-", 6) == 0) 3969 STRMOVE(p + 5, p + 6); 3970 3971 if (enc_canon_search(p) >= 0) 3972 { 3973 /* canonical name can be used unmodified */ 3974 if (p != r) 3975 STRMOVE(r, p); 3976 } 3977 else if ((i = enc_alias_search(p)) >= 0) 3978 { 3979 /* alias recognized, get canonical name */ 3980 vim_free(r); 3981 r = vim_strsave((char_u *)enc_canon_table[i].name); 3982 } 3983 } 3984 return r; 3985 } 3986 3987 /* 3988 * Search for an encoding alias of "name". 3989 * Returns -1 when not found. 3990 */ 3991 static int 3992 enc_alias_search(name) 3993 char_u *name; 3994 { 3995 int i; 3996 3997 for (i = 0; enc_alias_table[i].name != NULL; ++i) 3998 if (STRCMP(name, enc_alias_table[i].name) == 0) 3999 return enc_alias_table[i].canon; 4000 return -1; 4001 } 4002 #endif 4003 4004 #if defined(FEAT_MBYTE) || defined(PROTO) 4005 4006 #ifdef HAVE_LANGINFO_H 4007 # include <langinfo.h> 4008 #endif 4009 4010 /* 4011 * Get the canonicalized encoding of the current locale. 4012 * Returns an allocated string when successful, NULL when not. 4013 */ 4014 char_u * 4015 enc_locale() 4016 { 4017 #ifndef WIN3264 4018 char *s; 4019 char *p; 4020 int i; 4021 #endif 4022 char buf[50]; 4023 #ifdef WIN3264 4024 long acp = GetACP(); 4025 4026 if (acp == 1200) 4027 STRCPY(buf, "ucs-2le"); 4028 else if (acp == 1252) /* cp1252 is used as latin1 */ 4029 STRCPY(buf, "latin1"); 4030 else 4031 sprintf(buf, "cp%ld", acp); 4032 #else 4033 # ifdef HAVE_NL_LANGINFO_CODESET 4034 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4035 # endif 4036 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4037 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4038 # endif 4039 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4040 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4041 s = getenv("LANG"); 4042 4043 if (s == NULL || *s == NUL) 4044 return FAIL; 4045 4046 /* The most generic locale format is: 4047 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4048 * If there is a '.' remove the part before it. 4049 * if there is something after the codeset, remove it. 4050 * Make the name lowercase and replace '_' with '-'. 4051 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4052 * "ko_KR.EUC" == "euc-kr" 4053 */ 4054 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4055 { 4056 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4057 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4058 { 4059 /* copy "XY.EUC" to "euc-XY" to buf[10] */ 4060 STRCPY(buf + 10, "euc-"); 4061 buf[14] = p[-2]; 4062 buf[15] = p[-1]; 4063 buf[16] = 0; 4064 s = buf + 10; 4065 } 4066 else 4067 s = p + 1; 4068 } 4069 for (i = 0; s[i] != NUL && i < (int)sizeof(buf) - 1; ++i) 4070 { 4071 if (s[i] == '_' || s[i] == '-') 4072 buf[i] = '-'; 4073 else if (isalnum((int)s[i])) 4074 buf[i] = TOLOWER_ASC(s[i]); 4075 else 4076 break; 4077 } 4078 buf[i] = NUL; 4079 #endif 4080 4081 return enc_canonize((char_u *)buf); 4082 } 4083 4084 #if defined(WIN3264) || defined(PROTO) 4085 /* 4086 * Convert an encoding name to an MS-Windows codepage. 4087 * Returns zero if no codepage can be figured out. 4088 */ 4089 int 4090 encname2codepage(name) 4091 char_u *name; 4092 { 4093 int cp; 4094 char_u *p = name; 4095 int idx; 4096 4097 if (STRNCMP(p, "8bit-", 5) == 0) 4098 p += 5; 4099 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4100 p += 6; 4101 4102 if (p[0] == 'c' && p[1] == 'p') 4103 cp = atoi(p + 2); 4104 else if ((idx = enc_canon_search(p)) >= 0) 4105 cp = enc_canon_table[idx].codepage; 4106 else 4107 return 0; 4108 if (IsValidCodePage(cp)) 4109 return cp; 4110 return 0; 4111 } 4112 #endif 4113 4114 # if defined(USE_ICONV) || defined(PROTO) 4115 4116 static char_u *iconv_string __ARGS((vimconv_T *vcp, char_u *str, int slen, int *unconvlenp, int *resultlenp)); 4117 4118 /* 4119 * Call iconv_open() with a check if iconv() works properly (there are broken 4120 * versions). 4121 * Returns (void *)-1 if failed. 4122 * (should return iconv_t, but that causes problems with prototypes). 4123 */ 4124 void * 4125 my_iconv_open(to, from) 4126 char_u *to; 4127 char_u *from; 4128 { 4129 iconv_t fd; 4130 #define ICONV_TESTLEN 400 4131 char_u tobuf[ICONV_TESTLEN]; 4132 char *p; 4133 size_t tolen; 4134 static int iconv_ok = -1; 4135 4136 if (iconv_ok == FALSE) 4137 return (void *)-1; /* detected a broken iconv() previously */ 4138 4139 #ifdef DYNAMIC_ICONV 4140 /* Check if the iconv.dll can be found. */ 4141 if (!iconv_enabled(TRUE)) 4142 return (void *)-1; 4143 #endif 4144 4145 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4146 4147 if (fd != (iconv_t)-1 && iconv_ok == -1) 4148 { 4149 /* 4150 * Do a dummy iconv() call to check if it actually works. There is a 4151 * version of iconv() on Linux that is broken. We can't ignore it, 4152 * because it's wide-spread. The symptoms are that after outputting 4153 * the initial shift state the "to" pointer is NULL and conversion 4154 * stops for no apparent reason after about 8160 characters. 4155 */ 4156 p = (char *)tobuf; 4157 tolen = ICONV_TESTLEN; 4158 (void)iconv(fd, NULL, NULL, &p, &tolen); 4159 if (p == NULL) 4160 { 4161 iconv_ok = FALSE; 4162 iconv_close(fd); 4163 fd = (iconv_t)-1; 4164 } 4165 else 4166 iconv_ok = TRUE; 4167 } 4168 4169 return (void *)fd; 4170 } 4171 4172 /* 4173 * Convert the string "str[slen]" with iconv(). 4174 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4175 * sequence and set "*unconvlenp" to the length of it. 4176 * Returns the converted string in allocated memory. NULL for an error. 4177 * If resultlenp is not NULL, sets it to the result length in bytes. 4178 */ 4179 static char_u * 4180 iconv_string(vcp, str, slen, unconvlenp, resultlenp) 4181 vimconv_T *vcp; 4182 char_u *str; 4183 int slen; 4184 int *unconvlenp; 4185 int *resultlenp; 4186 { 4187 const char *from; 4188 size_t fromlen; 4189 char *to; 4190 size_t tolen; 4191 size_t len = 0; 4192 size_t done = 0; 4193 char_u *result = NULL; 4194 char_u *p; 4195 int l; 4196 4197 from = (char *)str; 4198 fromlen = slen; 4199 for (;;) 4200 { 4201 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4202 { 4203 /* Allocate enough room for most conversions. When re-allocating 4204 * increase the buffer size. */ 4205 len = len + fromlen * 2 + 40; 4206 p = alloc((unsigned)len); 4207 if (p != NULL && done > 0) 4208 mch_memmove(p, result, done); 4209 vim_free(result); 4210 result = p; 4211 if (result == NULL) /* out of memory */ 4212 break; 4213 } 4214 4215 to = (char *)result + done; 4216 tolen = len - done - 2; 4217 /* Avoid a warning for systems with a wrong iconv() prototype by 4218 * casting the second argument to void *. */ 4219 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4220 != (size_t)-1) 4221 { 4222 /* Finished, append a NUL. */ 4223 *to = NUL; 4224 break; 4225 } 4226 4227 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4228 * iconv library may use one of them. */ 4229 if (!vcp->vc_fail && unconvlenp != NULL 4230 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4231 { 4232 /* Handle an incomplete sequence at the end. */ 4233 *to = NUL; 4234 *unconvlenp = (int)fromlen; 4235 break; 4236 } 4237 4238 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4239 * iconv library may use one of them. */ 4240 else if (!vcp->vc_fail 4241 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4242 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4243 { 4244 /* Can't convert: insert a '?' and skip a character. This assumes 4245 * conversion from 'encoding' to something else. In other 4246 * situations we don't know what to skip anyway. */ 4247 *to++ = '?'; 4248 if ((*mb_ptr2cells)((char_u *)from) > 1) 4249 *to++ = '?'; 4250 if (enc_utf8) 4251 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4252 else 4253 { 4254 l = (*mb_ptr2len)((char_u *)from); 4255 if (l > (int)fromlen) 4256 l = (int)fromlen; 4257 } 4258 from += l; 4259 fromlen -= l; 4260 } 4261 else if (ICONV_ERRNO != ICONV_E2BIG) 4262 { 4263 /* conversion failed */ 4264 vim_free(result); 4265 result = NULL; 4266 break; 4267 } 4268 /* Not enough room or skipping illegal sequence. */ 4269 done = to - (char *)result; 4270 } 4271 4272 if (resultlenp != NULL && result != NULL) 4273 *resultlenp = (int)(to - (char *)result); 4274 return result; 4275 } 4276 4277 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4278 /* 4279 * Dynamically load the "iconv.dll" on Win32. 4280 */ 4281 4282 #ifndef DYNAMIC_ICONV /* just generating prototypes */ 4283 # define HINSTANCE int 4284 #endif 4285 static HINSTANCE hIconvDLL = 0; 4286 static HINSTANCE hMsvcrtDLL = 0; 4287 4288 # ifndef DYNAMIC_ICONV_DLL 4289 # define DYNAMIC_ICONV_DLL "iconv.dll" 4290 # define DYNAMIC_ICONV_DLL_ALT "libiconv.dll" 4291 # endif 4292 # ifndef DYNAMIC_MSVCRT_DLL 4293 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4294 # endif 4295 4296 /* 4297 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4298 */ 4299 int 4300 iconv_enabled(verbose) 4301 int verbose; 4302 { 4303 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4304 return TRUE; 4305 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4306 if (hIconvDLL == 0) /* sometimes it's called libiconv.dll */ 4307 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT); 4308 if (hIconvDLL != 0) 4309 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4310 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4311 { 4312 /* Only give the message when 'verbose' is set, otherwise it might be 4313 * done whenever a conversion is attempted. */ 4314 if (verbose && p_verbose > 0) 4315 { 4316 verbose_enter(); 4317 EMSG2(_(e_loadlib), 4318 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4319 verbose_leave(); 4320 } 4321 iconv_end(); 4322 return FALSE; 4323 } 4324 4325 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4326 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4327 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4328 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4329 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4330 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4331 || iconvctl == NULL || iconv_errno == NULL) 4332 { 4333 iconv_end(); 4334 if (verbose && p_verbose > 0) 4335 { 4336 verbose_enter(); 4337 EMSG2(_(e_loadfunc), "for libiconv"); 4338 verbose_leave(); 4339 } 4340 return FALSE; 4341 } 4342 return TRUE; 4343 } 4344 4345 void 4346 iconv_end() 4347 { 4348 /* Don't use iconv() when inputting or outputting characters. */ 4349 if (input_conv.vc_type == CONV_ICONV) 4350 convert_setup(&input_conv, NULL, NULL); 4351 if (output_conv.vc_type == CONV_ICONV) 4352 convert_setup(&output_conv, NULL, NULL); 4353 4354 if (hIconvDLL != 0) 4355 FreeLibrary(hIconvDLL); 4356 if (hMsvcrtDLL != 0) 4357 FreeLibrary(hMsvcrtDLL); 4358 hIconvDLL = 0; 4359 hMsvcrtDLL = 0; 4360 } 4361 # endif /* DYNAMIC_ICONV */ 4362 # endif /* USE_ICONV */ 4363 4364 #endif /* FEAT_MBYTE */ 4365 4366 #if defined(FEAT_XIM) || defined(PROTO) 4367 4368 # if defined(FEAT_GUI_GTK) || defined(PROTO) 4369 static int xim_has_preediting INIT(= FALSE); /* IM current status */ 4370 4371 /* 4372 * Set preedit_start_col to the current cursor position. 4373 */ 4374 static void 4375 init_preedit_start_col(void) 4376 { 4377 if (State & CMDLINE) 4378 preedit_start_col = cmdline_getvcol_cursor(); 4379 else if (curwin != NULL) 4380 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); 4381 /* Prevent that preediting marks the buffer as changed. */ 4382 xim_changed_while_preediting = curbuf->b_changed; 4383 } 4384 4385 static int im_is_active = FALSE; /* IM is enabled for current mode */ 4386 static int preedit_is_active = FALSE; 4387 static int im_preedit_cursor = 0; /* cursor offset in characters */ 4388 static int im_preedit_trailing = 0; /* number of characters after cursor */ 4389 4390 static unsigned long im_commit_handler_id = 0; 4391 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; 4392 static unsigned int im_activatekey_state = 0; 4393 4394 void 4395 im_set_active(int active) 4396 { 4397 int was_active; 4398 4399 was_active = !!im_is_active; 4400 im_is_active = (active && !p_imdisable); 4401 4402 if (im_is_active != was_active) 4403 xim_reset(); 4404 } 4405 4406 void 4407 xim_set_focus(int focus) 4408 { 4409 if (xic != NULL) 4410 { 4411 if (focus) 4412 gtk_im_context_focus_in(xic); 4413 else 4414 gtk_im_context_focus_out(xic); 4415 } 4416 } 4417 4418 void 4419 im_set_position(int row, int col) 4420 { 4421 if (xic != NULL) 4422 { 4423 GdkRectangle area; 4424 4425 area.x = FILL_X(col); 4426 area.y = FILL_Y(row); 4427 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); 4428 area.height = gui.char_height; 4429 4430 gtk_im_context_set_cursor_location(xic, &area); 4431 } 4432 } 4433 4434 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */ 4435 void 4436 xim_set_preedit(void) 4437 { 4438 im_set_position(gui.row, gui.col); 4439 } 4440 # endif 4441 4442 static void 4443 im_add_to_input(char_u *str, int len) 4444 { 4445 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */ 4446 if (input_conv.vc_type != CONV_NONE) 4447 { 4448 str = string_convert(&input_conv, str, &len); 4449 g_return_if_fail(str != NULL); 4450 } 4451 4452 add_to_input_buf_csi(str, len); 4453 4454 if (input_conv.vc_type != CONV_NONE) 4455 vim_free(str); 4456 4457 if (p_mh) /* blank out the pointer if necessary */ 4458 gui_mch_mousehide(TRUE); 4459 } 4460 4461 static void 4462 im_delete_preedit(void) 4463 { 4464 char_u bskey[] = {CSI, 'k', 'b'}; 4465 char_u delkey[] = {CSI, 'k', 'D'}; 4466 4467 if (State & NORMAL) 4468 { 4469 im_preedit_cursor = 0; 4470 return; 4471 } 4472 for (; im_preedit_cursor > 0; --im_preedit_cursor) 4473 add_to_input_buf(bskey, (int)sizeof(bskey)); 4474 4475 for (; im_preedit_trailing > 0; --im_preedit_trailing) 4476 add_to_input_buf(delkey, (int)sizeof(delkey)); 4477 } 4478 4479 /* 4480 * Move the cursor left by "num_move_back" characters. 4481 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for 4482 * these K_LEFT keys. 4483 */ 4484 static void 4485 im_correct_cursor(int num_move_back) 4486 { 4487 char_u backkey[] = {CSI, 'k', 'l'}; 4488 4489 if (State & NORMAL) 4490 return; 4491 # ifdef FEAT_RIGHTLEFT 4492 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) 4493 backkey[2] = 'r'; 4494 # endif 4495 for (; num_move_back > 0; --num_move_back) 4496 add_to_input_buf(backkey, (int)sizeof(backkey)); 4497 } 4498 4499 static int xim_expected_char = NUL; 4500 static int xim_ignored_char = FALSE; 4501 4502 /* 4503 * Update the mode and cursor while in an IM callback. 4504 */ 4505 static void 4506 im_show_info(void) 4507 { 4508 int old_vgetc_busy; 4509 4510 old_vgetc_busy = vgetc_busy; 4511 vgetc_busy = TRUE; 4512 showmode(); 4513 vgetc_busy = old_vgetc_busy; 4514 if ((State & NORMAL) || (State & INSERT)) 4515 setcursor(); 4516 out_flush(); 4517 } 4518 4519 /* 4520 * Callback invoked when the user finished preediting. 4521 * Put the final string into the input buffer. 4522 */ 4523 static void 4524 im_commit_cb(GtkIMContext *context UNUSED, 4525 const gchar *str, 4526 gpointer data UNUSED) 4527 { 4528 int slen = (int)STRLEN(str); 4529 int add_to_input = TRUE; 4530 int clen; 4531 int len = slen; 4532 int commit_with_preedit = TRUE; 4533 char_u *im_str; 4534 4535 #ifdef XIM_DEBUG 4536 xim_log("im_commit_cb(): %s\n", str); 4537 #endif 4538 4539 /* The imhangul module doesn't reset the preedit string before 4540 * committing. Call im_delete_preedit() to work around that. */ 4541 im_delete_preedit(); 4542 4543 /* Indicate that preediting has finished. */ 4544 if (preedit_start_col == MAXCOL) 4545 { 4546 init_preedit_start_col(); 4547 commit_with_preedit = FALSE; 4548 } 4549 4550 /* The thing which setting "preedit_start_col" to MAXCOL means that 4551 * "preedit_start_col" will be set forcely when calling 4552 * preedit_changed_cb() next time. 4553 * "preedit_start_col" should not reset with MAXCOL on this part. Vim 4554 * is simulating the preediting by using add_to_input_str(). when 4555 * preedit begin immediately before committed, the typebuf is not 4556 * flushed to screen, then it can't get correct "preedit_start_col". 4557 * Thus, it should calculate the cells by adding cells of the committed 4558 * string. */ 4559 if (input_conv.vc_type != CONV_NONE) 4560 { 4561 im_str = string_convert(&input_conv, (char_u *)str, &len); 4562 g_return_if_fail(im_str != NULL); 4563 } 4564 else 4565 im_str = (char_u *)str; 4566 4567 clen = mb_string2cells(im_str, len); 4568 4569 if (input_conv.vc_type != CONV_NONE) 4570 vim_free(im_str); 4571 preedit_start_col += clen; 4572 4573 /* Is this a single character that matches a keypad key that's just 4574 * been pressed? If so, we don't want it to be entered as such - let 4575 * us carry on processing the raw keycode so that it may be used in 4576 * mappings as <kSomething>. */ 4577 if (xim_expected_char != NUL) 4578 { 4579 /* We're currently processing a keypad or other special key */ 4580 if (slen == 1 && str[0] == xim_expected_char) 4581 { 4582 /* It's a match - don't do it here */ 4583 xim_ignored_char = TRUE; 4584 add_to_input = FALSE; 4585 } 4586 else 4587 { 4588 /* Not a match */ 4589 xim_ignored_char = FALSE; 4590 } 4591 } 4592 4593 if (add_to_input) 4594 im_add_to_input((char_u *)str, slen); 4595 4596 /* Inserting chars while "im_is_active" is set does not cause a change of 4597 * buffer. When the chars are committed the buffer must be marked as 4598 * changed. */ 4599 if (!commit_with_preedit) 4600 preedit_start_col = MAXCOL; 4601 4602 /* This flag is used in changed() at next call. */ 4603 xim_changed_while_preediting = TRUE; 4604 4605 if (gtk_main_level() > 0) 4606 gtk_main_quit(); 4607 } 4608 4609 /* 4610 * Callback invoked after start to the preedit. 4611 */ 4612 static void 4613 im_preedit_start_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4614 { 4615 #ifdef XIM_DEBUG 4616 xim_log("im_preedit_start_cb()\n"); 4617 #endif 4618 4619 im_is_active = TRUE; 4620 preedit_is_active = TRUE; 4621 gui_update_cursor(TRUE, FALSE); 4622 im_show_info(); 4623 } 4624 4625 /* 4626 * Callback invoked after end to the preedit. 4627 */ 4628 static void 4629 im_preedit_end_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4630 { 4631 #ifdef XIM_DEBUG 4632 xim_log("im_preedit_end_cb()\n"); 4633 #endif 4634 im_delete_preedit(); 4635 4636 /* Indicate that preediting has finished */ 4637 preedit_start_col = MAXCOL; 4638 xim_has_preediting = FALSE; 4639 4640 #if 0 4641 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was 4642 * switched off unintentionally. We now use preedit_is_active (added by 4643 * SungHyun Nam). */ 4644 im_is_active = FALSE; 4645 #endif 4646 preedit_is_active = FALSE; 4647 gui_update_cursor(TRUE, FALSE); 4648 im_show_info(); 4649 } 4650 4651 /* 4652 * Callback invoked after changes to the preedit string. If the preedit 4653 * string was empty before, remember the preedit start column so we know 4654 * where to apply feedback attributes. Delete the previous preedit string 4655 * if there was one, save the new preedit cursor offset, and put the new 4656 * string into the input buffer. 4657 * 4658 * TODO: The pragmatic "put into input buffer" approach used here has 4659 * several fundamental problems: 4660 * 4661 * - The characters in the preedit string are subject to remapping. 4662 * That's broken, only the finally committed string should be remapped. 4663 * 4664 * - There is a race condition involved: The retrieved value for the 4665 * current cursor position will be wrong if any unprocessed characters 4666 * are still queued in the input buffer. 4667 * 4668 * - Due to the lack of synchronization between the file buffer in memory 4669 * and any typed characters, it's practically impossible to implement the 4670 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM 4671 * modules for languages such as Thai are likely to rely on this feature 4672 * for proper operation. 4673 * 4674 * Conclusions: I think support for preediting needs to be moved to the 4675 * core parts of Vim. Ideally, until it has been committed, the preediting 4676 * string should only be displayed and not affect the buffer content at all. 4677 * The question how to deal with the synchronization issue still remains. 4678 * Circumventing the input buffer is probably not desirable. Anyway, I think 4679 * implementing "retrieve_surrounding" is the only hard problem. 4680 * 4681 * One way to solve all of this in a clean manner would be to queue all key 4682 * press/release events "as is" in the input buffer, and apply the IM filtering 4683 * at the receiving end of the queue. This, however, would have a rather large 4684 * impact on the code base. If there is an easy way to force processing of all 4685 * remaining input from within the "retrieve_surrounding" signal handler, this 4686 * might not be necessary. Gotta ask on vim-dev for opinions. 4687 */ 4688 static void 4689 im_preedit_changed_cb(GtkIMContext *context, gpointer data UNUSED) 4690 { 4691 char *preedit_string = NULL; 4692 int cursor_index = 0; 4693 int num_move_back = 0; 4694 char_u *str; 4695 char_u *p; 4696 int i; 4697 4698 gtk_im_context_get_preedit_string(context, 4699 &preedit_string, NULL, 4700 &cursor_index); 4701 4702 #ifdef XIM_DEBUG 4703 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); 4704 #endif 4705 4706 g_return_if_fail(preedit_string != NULL); /* just in case */ 4707 4708 /* If preedit_start_col is MAXCOL set it to the current cursor position. */ 4709 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') 4710 { 4711 xim_has_preediting = TRUE; 4712 4713 /* Urgh, this breaks if the input buffer isn't empty now */ 4714 init_preedit_start_col(); 4715 } 4716 else if (cursor_index == 0 && preedit_string[0] == '\0') 4717 { 4718 xim_has_preediting = FALSE; 4719 4720 /* If at the start position (after typing backspace) 4721 * preedit_start_col must be reset. */ 4722 preedit_start_col = MAXCOL; 4723 } 4724 4725 im_delete_preedit(); 4726 4727 /* 4728 * Compute the end of the preediting area: "preedit_end_col". 4729 * According to the documentation of gtk_im_context_get_preedit_string(), 4730 * the cursor_pos output argument returns the offset in bytes. This is 4731 * unfortunately not true -- real life shows the offset is in characters, 4732 * and the GTK+ source code agrees with me. Will file a bug later. 4733 */ 4734 if (preedit_start_col != MAXCOL) 4735 preedit_end_col = preedit_start_col; 4736 str = (char_u *)preedit_string; 4737 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) 4738 { 4739 int is_composing; 4740 4741 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); 4742 /* 4743 * These offsets are used as counters when generating <BS> and <Del> 4744 * to delete the preedit string. So don't count composing characters 4745 * unless 'delcombine' is enabled. 4746 */ 4747 if (!is_composing || p_deco) 4748 { 4749 if (i < cursor_index) 4750 ++im_preedit_cursor; 4751 else 4752 ++im_preedit_trailing; 4753 } 4754 if (!is_composing && i >= cursor_index) 4755 { 4756 /* This is essentially the same as im_preedit_trailing, except 4757 * composing characters are not counted even if p_deco is set. */ 4758 ++num_move_back; 4759 } 4760 if (preedit_start_col != MAXCOL) 4761 preedit_end_col += utf_ptr2cells(p); 4762 } 4763 4764 if (p > str) 4765 { 4766 im_add_to_input(str, (int)(p - str)); 4767 im_correct_cursor(num_move_back); 4768 } 4769 4770 g_free(preedit_string); 4771 4772 if (gtk_main_level() > 0) 4773 gtk_main_quit(); 4774 } 4775 4776 /* 4777 * Translate the Pango attributes at iter to Vim highlighting attributes. 4778 * Ignore attributes not supported by Vim highlighting. This shouldn't have 4779 * too much impact -- right now we handle even more attributes than necessary 4780 * for the IM modules I tested with. 4781 */ 4782 static int 4783 translate_pango_attributes(PangoAttrIterator *iter) 4784 { 4785 PangoAttribute *attr; 4786 int char_attr = HL_NORMAL; 4787 4788 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); 4789 if (attr != NULL && ((PangoAttrInt *)attr)->value 4790 != (int)PANGO_UNDERLINE_NONE) 4791 char_attr |= HL_UNDERLINE; 4792 4793 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); 4794 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) 4795 char_attr |= HL_BOLD; 4796 4797 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); 4798 if (attr != NULL && ((PangoAttrInt *)attr)->value 4799 != (int)PANGO_STYLE_NORMAL) 4800 char_attr |= HL_ITALIC; 4801 4802 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); 4803 if (attr != NULL) 4804 { 4805 const PangoColor *color = &((PangoAttrColor *)attr)->color; 4806 4807 /* Assume inverse if black background is requested */ 4808 if ((color->red | color->green | color->blue) == 0) 4809 char_attr |= HL_INVERSE; 4810 } 4811 4812 return char_attr; 4813 } 4814 4815 /* 4816 * Retrieve the highlighting attributes at column col in the preedit string. 4817 * Return -1 if not in preediting mode or if col is out of range. 4818 */ 4819 int 4820 im_get_feedback_attr(int col) 4821 { 4822 char *preedit_string = NULL; 4823 PangoAttrList *attr_list = NULL; 4824 int char_attr = -1; 4825 4826 if (xic == NULL) 4827 return char_attr; 4828 4829 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 4830 4831 if (preedit_string != NULL && attr_list != NULL) 4832 { 4833 int idx; 4834 4835 /* Get the byte index as used by PangoAttrIterator */ 4836 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col) 4837 idx += utfc_ptr2len((char_u *)preedit_string + idx); 4838 4839 if (preedit_string[idx] != '\0') 4840 { 4841 PangoAttrIterator *iter; 4842 int start, end; 4843 4844 char_attr = HL_NORMAL; 4845 iter = pango_attr_list_get_iterator(attr_list); 4846 4847 /* Extract all relevant attributes from the list. */ 4848 do 4849 { 4850 pango_attr_iterator_range(iter, &start, &end); 4851 4852 if (idx >= start && idx < end) 4853 char_attr |= translate_pango_attributes(iter); 4854 } 4855 while (pango_attr_iterator_next(iter)); 4856 4857 pango_attr_iterator_destroy(iter); 4858 } 4859 } 4860 4861 if (attr_list != NULL) 4862 pango_attr_list_unref(attr_list); 4863 g_free(preedit_string); 4864 4865 return char_attr; 4866 } 4867 4868 void 4869 xim_init(void) 4870 { 4871 #ifdef XIM_DEBUG 4872 xim_log("xim_init()\n"); 4873 #endif 4874 4875 g_return_if_fail(gui.drawarea != NULL); 4876 g_return_if_fail(gui.drawarea->window != NULL); 4877 4878 xic = gtk_im_multicontext_new(); 4879 g_object_ref(xic); 4880 4881 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", 4882 G_CALLBACK(&im_commit_cb), NULL); 4883 g_signal_connect(G_OBJECT(xic), "preedit_changed", 4884 G_CALLBACK(&im_preedit_changed_cb), NULL); 4885 g_signal_connect(G_OBJECT(xic), "preedit_start", 4886 G_CALLBACK(&im_preedit_start_cb), NULL); 4887 g_signal_connect(G_OBJECT(xic), "preedit_end", 4888 G_CALLBACK(&im_preedit_end_cb), NULL); 4889 4890 gtk_im_context_set_client_window(xic, gui.drawarea->window); 4891 } 4892 4893 void 4894 im_shutdown(void) 4895 { 4896 #ifdef XIM_DEBUG 4897 xim_log("im_shutdown()\n"); 4898 #endif 4899 4900 if (xic != NULL) 4901 { 4902 gtk_im_context_focus_out(xic); 4903 g_object_unref(xic); 4904 xic = NULL; 4905 } 4906 im_is_active = FALSE; 4907 im_commit_handler_id = 0; 4908 preedit_start_col = MAXCOL; 4909 xim_has_preediting = FALSE; 4910 } 4911 4912 /* 4913 * Convert the string argument to keyval and state for GdkEventKey. 4914 * If str is valid return TRUE, otherwise FALSE. 4915 * 4916 * See 'imactivatekey' for documentation of the format. 4917 */ 4918 static int 4919 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) 4920 { 4921 const char *mods_end; 4922 unsigned tmp_keyval; 4923 unsigned tmp_state = 0; 4924 4925 mods_end = strrchr(str, '-'); 4926 mods_end = (mods_end != NULL) ? mods_end + 1 : str; 4927 4928 /* Parse modifier keys */ 4929 while (str < mods_end) 4930 switch (*str++) 4931 { 4932 case '-': break; 4933 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; 4934 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; 4935 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; 4936 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; 4937 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; 4938 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; 4939 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; 4940 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; 4941 default: 4942 return FALSE; 4943 } 4944 4945 tmp_keyval = gdk_keyval_from_name(str); 4946 4947 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) 4948 return FALSE; 4949 4950 if (keyval != NULL) 4951 *keyval = tmp_keyval; 4952 if (state != NULL) 4953 *state = tmp_state; 4954 4955 return TRUE; 4956 } 4957 4958 /* 4959 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an 4960 * empty string is also regarded as valid. 4961 * 4962 * Note: The numerical key value of p_imak is cached if it was valid; thus 4963 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever 4964 * 'imak' changes. This is currently the case but not obvious -- should 4965 * probably rename the function for clarity. 4966 */ 4967 int 4968 im_xim_isvalid_imactivate(void) 4969 { 4970 if (p_imak[0] == NUL) 4971 { 4972 im_activatekey_keyval = GDK_VoidSymbol; 4973 im_activatekey_state = 0; 4974 return TRUE; 4975 } 4976 4977 return im_string_to_keyval((const char *)p_imak, 4978 &im_activatekey_keyval, 4979 &im_activatekey_state); 4980 } 4981 4982 static void 4983 im_synthesize_keypress(unsigned int keyval, unsigned int state) 4984 { 4985 GdkEventKey *event; 4986 4987 # ifdef HAVE_GTK_MULTIHEAD 4988 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); 4989 g_object_ref(gui.drawarea->window); /* unreffed by gdk_event_free() */ 4990 # else 4991 event = (GdkEventKey *)g_malloc0((gulong)sizeof(GdkEvent)); 4992 event->type = GDK_KEY_PRESS; 4993 # endif 4994 event->window = gui.drawarea->window; 4995 event->send_event = TRUE; 4996 event->time = GDK_CURRENT_TIME; 4997 event->state = state; 4998 event->keyval = keyval; 4999 event->hardware_keycode = /* needed for XIM */ 5000 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); 5001 event->length = 0; 5002 event->string = NULL; 5003 5004 gtk_im_context_filter_keypress(xic, event); 5005 5006 /* For consistency, also send the corresponding release event. */ 5007 event->type = GDK_KEY_RELEASE; 5008 event->send_event = FALSE; 5009 gtk_im_context_filter_keypress(xic, event); 5010 5011 # ifdef HAVE_GTK_MULTIHEAD 5012 gdk_event_free((GdkEvent *)event); 5013 # else 5014 g_free(event); 5015 # endif 5016 } 5017 5018 void 5019 xim_reset(void) 5020 { 5021 if (xic != NULL) 5022 { 5023 /* 5024 * The third-party imhangul module (and maybe others too) ignores 5025 * gtk_im_context_reset() or at least doesn't reset the active state. 5026 * Thus sending imactivatekey would turn it off if it was on before, 5027 * which is clearly not what we want. Fortunately we can work around 5028 * that for imhangul by sending GDK_Escape, but I don't know if it 5029 * works with all IM modules that support an activation key :/ 5030 * 5031 * An alternative approach would be to destroy the IM context and 5032 * recreate it. But that means loading/unloading the IM module on 5033 * every mode switch, which causes a quite noticeable delay even on 5034 * my rather fast box... 5035 * * 5036 * Moreover, there are some XIM which cannot respond to 5037 * im_synthesize_keypress(). we hope that they reset by 5038 * xim_shutdown(). 5039 */ 5040 if (im_activatekey_keyval != GDK_VoidSymbol && im_is_active) 5041 im_synthesize_keypress(GDK_Escape, 0U); 5042 5043 gtk_im_context_reset(xic); 5044 5045 /* 5046 * HACK for Ami: This sequence of function calls makes Ami handle 5047 * the IM reset graciously, without breaking loads of other stuff. 5048 * It seems to force English mode as well, which is exactly what we 5049 * want because it makes the Ami status display work reliably. 5050 */ 5051 gtk_im_context_set_use_preedit(xic, FALSE); 5052 5053 if (p_imdisable) 5054 im_shutdown(); 5055 else 5056 { 5057 gtk_im_context_set_use_preedit(xic, TRUE); 5058 xim_set_focus(gui.in_focus); 5059 5060 if (im_activatekey_keyval != GDK_VoidSymbol) 5061 { 5062 if (im_is_active) 5063 { 5064 g_signal_handler_block(xic, im_commit_handler_id); 5065 im_synthesize_keypress(im_activatekey_keyval, 5066 im_activatekey_state); 5067 g_signal_handler_unblock(xic, im_commit_handler_id); 5068 } 5069 } 5070 else 5071 { 5072 im_shutdown(); 5073 xim_init(); 5074 xim_set_focus(gui.in_focus); 5075 } 5076 } 5077 } 5078 5079 preedit_start_col = MAXCOL; 5080 xim_has_preediting = FALSE; 5081 } 5082 5083 int 5084 xim_queue_key_press_event(GdkEventKey *event, int down) 5085 { 5086 if (down) 5087 { 5088 /* 5089 * Workaround GTK2 XIM 'feature' that always converts keypad keys to 5090 * chars., even when not part of an IM sequence (ref. feature of 5091 * gdk/gdkkeyuni.c). 5092 * Flag any keypad keys that might represent a single char. 5093 * If this (on its own - i.e., not part of an IM sequence) is 5094 * committed while we're processing one of these keys, we can ignore 5095 * that commit and go ahead & process it ourselves. That way we can 5096 * still distinguish keypad keys for use in mappings. 5097 * Also add GDK_space to make <S-Space> work. 5098 */ 5099 switch (event->keyval) 5100 { 5101 case GDK_KP_Add: xim_expected_char = '+'; break; 5102 case GDK_KP_Subtract: xim_expected_char = '-'; break; 5103 case GDK_KP_Divide: xim_expected_char = '/'; break; 5104 case GDK_KP_Multiply: xim_expected_char = '*'; break; 5105 case GDK_KP_Decimal: xim_expected_char = '.'; break; 5106 case GDK_KP_Equal: xim_expected_char = '='; break; 5107 case GDK_KP_0: xim_expected_char = '0'; break; 5108 case GDK_KP_1: xim_expected_char = '1'; break; 5109 case GDK_KP_2: xim_expected_char = '2'; break; 5110 case GDK_KP_3: xim_expected_char = '3'; break; 5111 case GDK_KP_4: xim_expected_char = '4'; break; 5112 case GDK_KP_5: xim_expected_char = '5'; break; 5113 case GDK_KP_6: xim_expected_char = '6'; break; 5114 case GDK_KP_7: xim_expected_char = '7'; break; 5115 case GDK_KP_8: xim_expected_char = '8'; break; 5116 case GDK_KP_9: xim_expected_char = '9'; break; 5117 case GDK_space: xim_expected_char = ' '; break; 5118 default: xim_expected_char = NUL; 5119 } 5120 xim_ignored_char = FALSE; 5121 } 5122 5123 /* 5124 * When typing fFtT, XIM may be activated. Thus it must pass 5125 * gtk_im_context_filter_keypress() in Normal mode. 5126 * And while doing :sh too. 5127 */ 5128 if (xic != NULL && !p_imdisable 5129 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) 5130 { 5131 /* 5132 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is 5133 * always aware of the current status of IM, and can even emulate 5134 * the activation key for modules that don't support one. 5135 */ 5136 if (event->keyval == im_activatekey_keyval 5137 && (event->state & im_activatekey_state) == im_activatekey_state) 5138 { 5139 unsigned int state_mask; 5140 5141 /* Require the state of the 3 most used modifiers to match exactly. 5142 * Otherwise e.g. <S-C-space> would be unusable for other purposes 5143 * if the IM activate key is <S-space>. */ 5144 state_mask = im_activatekey_state; 5145 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK 5146 | (int)GDK_MOD1_MASK); 5147 5148 if ((event->state & state_mask) != im_activatekey_state) 5149 return FALSE; 5150 5151 /* Don't send it a second time on GDK_KEY_RELEASE. */ 5152 if (event->type != GDK_KEY_PRESS) 5153 return TRUE; 5154 5155 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE)) 5156 { 5157 im_set_active(FALSE); 5158 5159 /* ":lmap" mappings exists, toggle use of mappings. */ 5160 State ^= LANGMAP; 5161 if (State & LANGMAP) 5162 { 5163 curbuf->b_p_iminsert = B_IMODE_NONE; 5164 State &= ~LANGMAP; 5165 } 5166 else 5167 { 5168 curbuf->b_p_iminsert = B_IMODE_LMAP; 5169 State |= LANGMAP; 5170 } 5171 return TRUE; 5172 } 5173 5174 return gtk_im_context_filter_keypress(xic, event); 5175 } 5176 5177 /* Don't filter events through the IM context if IM isn't active 5178 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module 5179 * not doing anything before the activation key was sent. */ 5180 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) 5181 { 5182 int imresult = gtk_im_context_filter_keypress(xic, event); 5183 5184 /* Some XIM send following sequence: 5185 * 1. preedited string. 5186 * 2. committed string. 5187 * 3. line changed key. 5188 * 4. preedited string. 5189 * 5. remove preedited string. 5190 * if 3, Vim can't move back the above line for 5. 5191 * thus, this part should not parse the key. */ 5192 if (!imresult && preedit_start_col != MAXCOL 5193 && event->keyval == GDK_Return) 5194 { 5195 im_synthesize_keypress(GDK_Return, 0U); 5196 return FALSE; 5197 } 5198 5199 /* If XIM tried to commit a keypad key as a single char., 5200 * ignore it so we can use the keypad key 'raw', for mappings. */ 5201 if (xim_expected_char != NUL && xim_ignored_char) 5202 /* We had a keypad key, and XIM tried to thieve it */ 5203 return FALSE; 5204 5205 /* This is supposed to fix a problem with iBus, that space 5206 * characters don't work in input mode. */ 5207 xim_expected_char = NUL; 5208 5209 /* Normal processing */ 5210 return imresult; 5211 } 5212 } 5213 5214 return FALSE; 5215 } 5216 5217 int 5218 im_get_status(void) 5219 { 5220 return im_is_active; 5221 } 5222 5223 int 5224 preedit_get_status(void) 5225 { 5226 return preedit_is_active; 5227 } 5228 5229 int 5230 im_is_preediting() 5231 { 5232 return xim_has_preediting; 5233 } 5234 5235 # else /* !FEAT_GUI_GTK */ 5236 5237 static int xim_is_active = FALSE; /* XIM should be active in the current 5238 mode */ 5239 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */ 5240 #ifdef FEAT_GUI_X11 5241 static XIMStyle input_style; 5242 static int status_area_enabled = TRUE; 5243 #endif 5244 5245 /* 5246 * Switch using XIM on/off. This is used by the code that changes "State". 5247 */ 5248 void 5249 im_set_active(active) 5250 int active; 5251 { 5252 if (xic == NULL) 5253 return; 5254 5255 /* If 'imdisable' is set, XIM is never active. */ 5256 if (p_imdisable) 5257 active = FALSE; 5258 #if !defined(FEAT_GUI_GTK) 5259 else if (input_style & XIMPreeditPosition) 5260 /* There is a problem in switching XIM off when preediting is used, 5261 * and it is not clear how this can be solved. For now, keep XIM on 5262 * all the time, like it was done in Vim 5.8. */ 5263 active = TRUE; 5264 #endif 5265 5266 /* Remember the active state, it is needed when Vim gets keyboard focus. */ 5267 xim_is_active = active; 5268 xim_set_preedit(); 5269 } 5270 5271 /* 5272 * Adjust using XIM for gaining or losing keyboard focus. Also called when 5273 * "xim_is_active" changes. 5274 */ 5275 void 5276 xim_set_focus(focus) 5277 int focus; 5278 { 5279 if (xic == NULL) 5280 return; 5281 5282 /* 5283 * XIM only gets focus when the Vim window has keyboard focus and XIM has 5284 * been set active for the current mode. 5285 */ 5286 if (focus && xim_is_active) 5287 { 5288 if (!xim_has_focus) 5289 { 5290 xim_has_focus = TRUE; 5291 XSetICFocus(xic); 5292 } 5293 } 5294 else 5295 { 5296 if (xim_has_focus) 5297 { 5298 xim_has_focus = FALSE; 5299 XUnsetICFocus(xic); 5300 } 5301 } 5302 } 5303 5304 void 5305 im_set_position(row, col) 5306 int row UNUSED; 5307 int col UNUSED; 5308 { 5309 xim_set_preedit(); 5310 } 5311 5312 /* 5313 * Set the XIM to the current cursor position. 5314 */ 5315 void 5316 xim_set_preedit() 5317 { 5318 XVaNestedList attr_list; 5319 XRectangle spot_area; 5320 XPoint over_spot; 5321 int line_space; 5322 5323 if (xic == NULL) 5324 return; 5325 5326 xim_set_focus(TRUE); 5327 5328 if (!xim_has_focus) 5329 { 5330 /* hide XIM cursor */ 5331 over_spot.x = 0; 5332 over_spot.y = -100; /* arbitrary invisible position */ 5333 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5334 XNSpotLocation, 5335 &over_spot, 5336 NULL); 5337 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); 5338 XFree(attr_list); 5339 return; 5340 } 5341 5342 if (input_style & XIMPreeditPosition) 5343 { 5344 if (xim_fg_color == INVALCOLOR) 5345 { 5346 xim_fg_color = gui.def_norm_pixel; 5347 xim_bg_color = gui.def_back_pixel; 5348 } 5349 over_spot.x = TEXT_X(gui.col); 5350 over_spot.y = TEXT_Y(gui.row); 5351 spot_area.x = 0; 5352 spot_area.y = 0; 5353 spot_area.height = gui.char_height * Rows; 5354 spot_area.width = gui.char_width * Columns; 5355 line_space = gui.char_height; 5356 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5357 XNSpotLocation, &over_spot, 5358 XNForeground, (Pixel) xim_fg_color, 5359 XNBackground, (Pixel) xim_bg_color, 5360 XNArea, &spot_area, 5361 XNLineSpace, line_space, 5362 NULL); 5363 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) 5364 EMSG(_("E284: Cannot set IC values")); 5365 XFree(attr_list); 5366 } 5367 } 5368 5369 #if defined(FEAT_GUI_X11) 5370 static char e_xim[] = N_("E285: Failed to create input context"); 5371 #endif 5372 5373 #if defined(FEAT_GUI_X11) || defined(PROTO) 5374 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(sun) 5375 # define USE_X11R6_XIM 5376 # endif 5377 5378 static int xim_real_init __ARGS((Window x11_window, Display *x11_display)); 5379 5380 5381 #ifdef USE_X11R6_XIM 5382 static void xim_instantiate_cb __ARGS((Display *display, XPointer client_data, XPointer call_data)); 5383 static void xim_destroy_cb __ARGS((XIM im, XPointer client_data, XPointer call_data)); 5384 5385 static void 5386 xim_instantiate_cb(display, client_data, call_data) 5387 Display *display; 5388 XPointer client_data UNUSED; 5389 XPointer call_data UNUSED; 5390 { 5391 Window x11_window; 5392 Display *x11_display; 5393 5394 #ifdef XIM_DEBUG 5395 xim_log("xim_instantiate_cb()\n"); 5396 #endif 5397 5398 gui_get_x11_windis(&x11_window, &x11_display); 5399 if (display != x11_display) 5400 return; 5401 5402 xim_real_init(x11_window, x11_display); 5403 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5404 if (xic != NULL) 5405 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5406 xim_instantiate_cb, NULL); 5407 } 5408 5409 static void 5410 xim_destroy_cb(im, client_data, call_data) 5411 XIM im UNUSED; 5412 XPointer client_data UNUSED; 5413 XPointer call_data UNUSED; 5414 { 5415 Window x11_window; 5416 Display *x11_display; 5417 5418 #ifdef XIM_DEBUG 5419 xim_log("xim_destroy_cb()\n"); 5420 #endif 5421 gui_get_x11_windis(&x11_window, &x11_display); 5422 5423 xic = NULL; 5424 status_area_enabled = FALSE; 5425 5426 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5427 5428 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5429 xim_instantiate_cb, NULL); 5430 } 5431 #endif 5432 5433 void 5434 xim_init() 5435 { 5436 Window x11_window; 5437 Display *x11_display; 5438 5439 #ifdef XIM_DEBUG 5440 xim_log("xim_init()\n"); 5441 #endif 5442 5443 gui_get_x11_windis(&x11_window, &x11_display); 5444 5445 xic = NULL; 5446 5447 if (xim_real_init(x11_window, x11_display)) 5448 return; 5449 5450 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5451 5452 #ifdef USE_X11R6_XIM 5453 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5454 xim_instantiate_cb, NULL); 5455 #endif 5456 } 5457 5458 static int 5459 xim_real_init(x11_window, x11_display) 5460 Window x11_window; 5461 Display *x11_display; 5462 { 5463 int i; 5464 char *p, 5465 *s, 5466 *ns, 5467 *end, 5468 tmp[1024]; 5469 #define IMLEN_MAX 40 5470 char buf[IMLEN_MAX + 7]; 5471 XIM xim = NULL; 5472 XIMStyles *xim_styles; 5473 XIMStyle this_input_style = 0; 5474 Boolean found; 5475 XPoint over_spot; 5476 XVaNestedList preedit_list, status_list; 5477 5478 input_style = 0; 5479 status_area_enabled = FALSE; 5480 5481 if (xic != NULL) 5482 return FALSE; 5483 5484 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) 5485 { 5486 strcpy(tmp, gui.rsrc_input_method); 5487 for (ns = s = tmp; ns != NULL && *s != NUL;) 5488 { 5489 s = (char *)skipwhite((char_u *)s); 5490 if (*s == NUL) 5491 break; 5492 if ((ns = end = strchr(s, ',')) == NULL) 5493 end = s + strlen(s); 5494 while (isspace(((char_u *)end)[-1])) 5495 end--; 5496 *end = NUL; 5497 5498 if (strlen(s) <= IMLEN_MAX) 5499 { 5500 strcpy(buf, "@im="); 5501 strcat(buf, s); 5502 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL 5503 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) 5504 != NULL) 5505 break; 5506 } 5507 5508 s = ns + 1; 5509 } 5510 } 5511 5512 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) 5513 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5514 5515 /* This is supposed to be useful to obtain characters through 5516 * XmbLookupString() without really using a XIM. */ 5517 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL 5518 && *p != NUL) 5519 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5520 5521 if (xim == NULL) 5522 { 5523 /* Only give this message when verbose is set, because too many people 5524 * got this message when they didn't want to use a XIM. */ 5525 if (p_verbose > 0) 5526 { 5527 verbose_enter(); 5528 EMSG(_("E286: Failed to open input method")); 5529 verbose_leave(); 5530 } 5531 return FALSE; 5532 } 5533 5534 #ifdef USE_X11R6_XIM 5535 { 5536 XIMCallback destroy_cb; 5537 5538 destroy_cb.callback = xim_destroy_cb; 5539 destroy_cb.client_data = NULL; 5540 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) 5541 EMSG(_("E287: Warning: Could not set destroy callback to IM")); 5542 } 5543 #endif 5544 5545 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) 5546 { 5547 EMSG(_("E288: input method doesn't support any style")); 5548 XCloseIM(xim); 5549 return FALSE; 5550 } 5551 5552 found = False; 5553 strcpy(tmp, gui.rsrc_preedit_type_name); 5554 for (s = tmp; s && !found; ) 5555 { 5556 while (*s && isspace((unsigned char)*s)) 5557 s++; 5558 if (!*s) 5559 break; 5560 if ((ns = end = strchr(s, ',')) != 0) 5561 ns++; 5562 else 5563 end = s + strlen(s); 5564 while (isspace((unsigned char)*end)) 5565 end--; 5566 *end = '\0'; 5567 5568 if (!strcmp(s, "OverTheSpot")) 5569 this_input_style = (XIMPreeditPosition | XIMStatusArea); 5570 else if (!strcmp(s, "OffTheSpot")) 5571 this_input_style = (XIMPreeditArea | XIMStatusArea); 5572 else if (!strcmp(s, "Root")) 5573 this_input_style = (XIMPreeditNothing | XIMStatusNothing); 5574 5575 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5576 { 5577 if (this_input_style == xim_styles->supported_styles[i]) 5578 { 5579 found = True; 5580 break; 5581 } 5582 } 5583 if (!found) 5584 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5585 { 5586 if ((xim_styles->supported_styles[i] & this_input_style) 5587 == (this_input_style & ~XIMStatusArea)) 5588 { 5589 this_input_style &= ~XIMStatusArea; 5590 found = True; 5591 break; 5592 } 5593 } 5594 5595 s = ns; 5596 } 5597 XFree(xim_styles); 5598 5599 if (!found) 5600 { 5601 /* Only give this message when verbose is set, because too many people 5602 * got this message when they didn't want to use a XIM. */ 5603 if (p_verbose > 0) 5604 { 5605 verbose_enter(); 5606 EMSG(_("E289: input method doesn't support my preedit type")); 5607 verbose_leave(); 5608 } 5609 XCloseIM(xim); 5610 return FALSE; 5611 } 5612 5613 over_spot.x = TEXT_X(gui.col); 5614 over_spot.y = TEXT_Y(gui.row); 5615 input_style = this_input_style; 5616 5617 /* A crash was reported when trying to pass gui.norm_font as XNFontSet, 5618 * thus that has been removed. Hopefully the default works... */ 5619 #ifdef FEAT_XFONTSET 5620 if (gui.fontset != NOFONTSET) 5621 { 5622 preedit_list = XVaCreateNestedList(0, 5623 XNSpotLocation, &over_spot, 5624 XNForeground, (Pixel)gui.def_norm_pixel, 5625 XNBackground, (Pixel)gui.def_back_pixel, 5626 XNFontSet, (XFontSet)gui.fontset, 5627 NULL); 5628 status_list = XVaCreateNestedList(0, 5629 XNForeground, (Pixel)gui.def_norm_pixel, 5630 XNBackground, (Pixel)gui.def_back_pixel, 5631 XNFontSet, (XFontSet)gui.fontset, 5632 NULL); 5633 } 5634 else 5635 #endif 5636 { 5637 preedit_list = XVaCreateNestedList(0, 5638 XNSpotLocation, &over_spot, 5639 XNForeground, (Pixel)gui.def_norm_pixel, 5640 XNBackground, (Pixel)gui.def_back_pixel, 5641 NULL); 5642 status_list = XVaCreateNestedList(0, 5643 XNForeground, (Pixel)gui.def_norm_pixel, 5644 XNBackground, (Pixel)gui.def_back_pixel, 5645 NULL); 5646 } 5647 5648 xic = XCreateIC(xim, 5649 XNInputStyle, input_style, 5650 XNClientWindow, x11_window, 5651 XNFocusWindow, gui.wid, 5652 XNPreeditAttributes, preedit_list, 5653 XNStatusAttributes, status_list, 5654 NULL); 5655 XFree(status_list); 5656 XFree(preedit_list); 5657 if (xic != NULL) 5658 { 5659 if (input_style & XIMStatusArea) 5660 { 5661 xim_set_status_area(); 5662 status_area_enabled = TRUE; 5663 } 5664 else 5665 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5666 } 5667 else 5668 { 5669 EMSG(_(e_xim)); 5670 XCloseIM(xim); 5671 return FALSE; 5672 } 5673 5674 return TRUE; 5675 } 5676 5677 #endif /* FEAT_GUI_X11 */ 5678 5679 /* 5680 * Get IM status. When IM is on, return TRUE. Else return FALSE. 5681 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is 5682 * active, when not having focus XIM may still be active (e.g., when using a 5683 * tear-off menu item). 5684 */ 5685 int 5686 im_get_status() 5687 { 5688 return xim_has_focus; 5689 } 5690 5691 # endif /* !FEAT_GUI_GTK */ 5692 5693 # if !defined(FEAT_GUI_GTK) || defined(PROTO) 5694 /* 5695 * Set up the status area. 5696 * 5697 * This should use a separate Widget, but that seems not possible, because 5698 * preedit_area and status_area should be set to the same window as for the 5699 * text input. Unfortunately this means the status area pollutes the text 5700 * window... 5701 */ 5702 void 5703 xim_set_status_area() 5704 { 5705 XVaNestedList preedit_list = 0, status_list = 0, list = 0; 5706 XRectangle pre_area, status_area; 5707 5708 if (xic == NULL) 5709 return; 5710 5711 if (input_style & XIMStatusArea) 5712 { 5713 if (input_style & XIMPreeditArea) 5714 { 5715 XRectangle *needed_rect; 5716 5717 /* to get status_area width */ 5718 status_list = XVaCreateNestedList(0, XNAreaNeeded, 5719 &needed_rect, NULL); 5720 XGetICValues(xic, XNStatusAttributes, status_list, NULL); 5721 XFree(status_list); 5722 5723 status_area.width = needed_rect->width; 5724 } 5725 else 5726 status_area.width = gui.char_width * Columns; 5727 5728 status_area.x = 0; 5729 status_area.y = gui.char_height * Rows + gui.border_offset; 5730 if (gui.which_scrollbars[SBAR_BOTTOM]) 5731 status_area.y += gui.scrollbar_height; 5732 #ifdef FEAT_MENU 5733 if (gui.menu_is_active) 5734 status_area.y += gui.menu_height; 5735 #endif 5736 status_area.height = gui.char_height; 5737 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); 5738 } 5739 else 5740 { 5741 status_area.x = 0; 5742 status_area.y = gui.char_height * Rows + gui.border_offset; 5743 if (gui.which_scrollbars[SBAR_BOTTOM]) 5744 status_area.y += gui.scrollbar_height; 5745 #ifdef FEAT_MENU 5746 if (gui.menu_is_active) 5747 status_area.y += gui.menu_height; 5748 #endif 5749 status_area.width = 0; 5750 status_area.height = gui.char_height; 5751 } 5752 5753 if (input_style & XIMPreeditArea) /* off-the-spot */ 5754 { 5755 pre_area.x = status_area.x + status_area.width; 5756 pre_area.y = gui.char_height * Rows + gui.border_offset; 5757 pre_area.width = gui.char_width * Columns - pre_area.x; 5758 if (gui.which_scrollbars[SBAR_BOTTOM]) 5759 pre_area.y += gui.scrollbar_height; 5760 #ifdef FEAT_MENU 5761 if (gui.menu_is_active) 5762 pre_area.y += gui.menu_height; 5763 #endif 5764 pre_area.height = gui.char_height; 5765 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 5766 } 5767 else if (input_style & XIMPreeditPosition) /* over-the-spot */ 5768 { 5769 pre_area.x = 0; 5770 pre_area.y = 0; 5771 pre_area.height = gui.char_height * Rows; 5772 pre_area.width = gui.char_width * Columns; 5773 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 5774 } 5775 5776 if (preedit_list && status_list) 5777 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 5778 XNStatusAttributes, status_list, NULL); 5779 else if (preedit_list) 5780 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 5781 NULL); 5782 else if (status_list) 5783 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, 5784 NULL); 5785 else 5786 list = NULL; 5787 5788 if (list) 5789 { 5790 XSetICValues(xic, XNVaNestedList, list, NULL); 5791 XFree(list); 5792 } 5793 if (status_list) 5794 XFree(status_list); 5795 if (preedit_list) 5796 XFree(preedit_list); 5797 } 5798 5799 int 5800 xim_get_status_area_height() 5801 { 5802 if (status_area_enabled) 5803 return gui.char_height; 5804 return 0; 5805 } 5806 # endif 5807 5808 #endif /* FEAT_XIM */ 5809 5810 #if defined(FEAT_MBYTE) || defined(PROTO) 5811 5812 /* 5813 * Setup "vcp" for conversion from "from" to "to". 5814 * The names must have been made canonical with enc_canonize(). 5815 * vcp->vc_type must have been initialized to CONV_NONE. 5816 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 5817 * instead). 5818 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 5819 * Return FAIL when conversion is not supported, OK otherwise. 5820 */ 5821 int 5822 convert_setup(vcp, from, to) 5823 vimconv_T *vcp; 5824 char_u *from; 5825 char_u *to; 5826 { 5827 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 5828 } 5829 5830 /* 5831 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 5832 * "from" unicode charsets be considered utf-8. Same for "to". 5833 */ 5834 int 5835 convert_setup_ext(vcp, from, from_unicode_is_utf8, to, to_unicode_is_utf8) 5836 vimconv_T *vcp; 5837 char_u *from; 5838 int from_unicode_is_utf8; 5839 char_u *to; 5840 int to_unicode_is_utf8; 5841 { 5842 int from_prop; 5843 int to_prop; 5844 int from_is_utf8; 5845 int to_is_utf8; 5846 5847 /* Reset to no conversion. */ 5848 # ifdef USE_ICONV 5849 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 5850 iconv_close(vcp->vc_fd); 5851 # endif 5852 vcp->vc_type = CONV_NONE; 5853 vcp->vc_factor = 1; 5854 vcp->vc_fail = FALSE; 5855 5856 /* No conversion when one of the names is empty or they are equal. */ 5857 if (from == NULL || *from == NUL || to == NULL || *to == NUL 5858 || STRCMP(from, to) == 0) 5859 return OK; 5860 5861 from_prop = enc_canon_props(from); 5862 to_prop = enc_canon_props(to); 5863 if (from_unicode_is_utf8) 5864 from_is_utf8 = from_prop & ENC_UNICODE; 5865 else 5866 from_is_utf8 = from_prop == ENC_UNICODE; 5867 if (to_unicode_is_utf8) 5868 to_is_utf8 = to_prop & ENC_UNICODE; 5869 else 5870 to_is_utf8 = to_prop == ENC_UNICODE; 5871 5872 if ((from_prop & ENC_LATIN1) && to_is_utf8) 5873 { 5874 /* Internal latin1 -> utf-8 conversion. */ 5875 vcp->vc_type = CONV_TO_UTF8; 5876 vcp->vc_factor = 2; /* up to twice as long */ 5877 } 5878 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 5879 { 5880 /* Internal latin9 -> utf-8 conversion. */ 5881 vcp->vc_type = CONV_9_TO_UTF8; 5882 vcp->vc_factor = 3; /* up to three as long (euro sign) */ 5883 } 5884 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 5885 { 5886 /* Internal utf-8 -> latin1 conversion. */ 5887 vcp->vc_type = CONV_TO_LATIN1; 5888 } 5889 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 5890 { 5891 /* Internal utf-8 -> latin9 conversion. */ 5892 vcp->vc_type = CONV_TO_LATIN9; 5893 } 5894 #ifdef WIN3264 5895 /* Win32-specific codepage <-> codepage conversion without iconv. */ 5896 else if ((from_is_utf8 || encname2codepage(from) > 0) 5897 && (to_is_utf8 || encname2codepage(to) > 0)) 5898 { 5899 vcp->vc_type = CONV_CODEPAGE; 5900 vcp->vc_factor = 2; /* up to twice as long */ 5901 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 5902 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 5903 } 5904 #endif 5905 #ifdef MACOS_X 5906 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 5907 { 5908 vcp->vc_type = CONV_MAC_LATIN1; 5909 } 5910 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 5911 { 5912 vcp->vc_type = CONV_MAC_UTF8; 5913 vcp->vc_factor = 2; /* up to twice as long */ 5914 } 5915 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 5916 { 5917 vcp->vc_type = CONV_LATIN1_MAC; 5918 } 5919 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 5920 { 5921 vcp->vc_type = CONV_UTF8_MAC; 5922 } 5923 #endif 5924 # ifdef USE_ICONV 5925 else 5926 { 5927 /* Use iconv() for conversion. */ 5928 vcp->vc_fd = (iconv_t)my_iconv_open( 5929 to_is_utf8 ? (char_u *)"utf-8" : to, 5930 from_is_utf8 ? (char_u *)"utf-8" : from); 5931 if (vcp->vc_fd != (iconv_t)-1) 5932 { 5933 vcp->vc_type = CONV_ICONV; 5934 vcp->vc_factor = 4; /* could be longer too... */ 5935 } 5936 } 5937 # endif 5938 if (vcp->vc_type == CONV_NONE) 5939 return FAIL; 5940 5941 return OK; 5942 } 5943 5944 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \ 5945 || defined(MSDOS) || defined(PROTO) 5946 /* 5947 * Do conversion on typed input characters in-place. 5948 * The input and output are not NUL terminated! 5949 * Returns the length after conversion. 5950 */ 5951 int 5952 convert_input(ptr, len, maxlen) 5953 char_u *ptr; 5954 int len; 5955 int maxlen; 5956 { 5957 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 5958 } 5959 #endif 5960 5961 /* 5962 * Like convert_input(), but when there is an incomplete byte sequence at the 5963 * end return that as an allocated string in "restp" and set "*restlenp" to 5964 * the length. If "restp" is NULL it is not used. 5965 */ 5966 int 5967 convert_input_safe(ptr, len, maxlen, restp, restlenp) 5968 char_u *ptr; 5969 int len; 5970 int maxlen; 5971 char_u **restp; 5972 int *restlenp; 5973 { 5974 char_u *d; 5975 int dlen = len; 5976 int unconvertlen = 0; 5977 5978 d = string_convert_ext(&input_conv, ptr, &dlen, 5979 restp == NULL ? NULL : &unconvertlen); 5980 if (d != NULL) 5981 { 5982 if (dlen <= maxlen) 5983 { 5984 if (unconvertlen > 0) 5985 { 5986 /* Move the unconverted characters to allocated memory. */ 5987 *restp = alloc(unconvertlen); 5988 if (*restp != NULL) 5989 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 5990 *restlenp = unconvertlen; 5991 } 5992 mch_memmove(ptr, d, dlen); 5993 } 5994 else 5995 /* result is too long, keep the unconverted text (the caller must 5996 * have done something wrong!) */ 5997 dlen = len; 5998 vim_free(d); 5999 } 6000 return dlen; 6001 } 6002 6003 /* 6004 * Convert text "ptr[*lenp]" according to "vcp". 6005 * Returns the result in allocated memory and sets "*lenp". 6006 * When "lenp" is NULL, use NUL terminated strings. 6007 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 6008 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 6009 */ 6010 char_u * 6011 string_convert(vcp, ptr, lenp) 6012 vimconv_T *vcp; 6013 char_u *ptr; 6014 int *lenp; 6015 { 6016 return string_convert_ext(vcp, ptr, lenp, NULL); 6017 } 6018 6019 /* 6020 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 6021 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 6022 * set to the number of remaining bytes. 6023 */ 6024 char_u * 6025 string_convert_ext(vcp, ptr, lenp, unconvlenp) 6026 vimconv_T *vcp; 6027 char_u *ptr; 6028 int *lenp; 6029 int *unconvlenp; 6030 { 6031 char_u *retval = NULL; 6032 char_u *d; 6033 int len; 6034 int i; 6035 int l; 6036 int c; 6037 6038 if (lenp == NULL) 6039 len = (int)STRLEN(ptr); 6040 else 6041 len = *lenp; 6042 if (len == 0) 6043 return vim_strsave((char_u *)""); 6044 6045 switch (vcp->vc_type) 6046 { 6047 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */ 6048 retval = alloc(len * 2 + 1); 6049 if (retval == NULL) 6050 break; 6051 d = retval; 6052 for (i = 0; i < len; ++i) 6053 { 6054 c = ptr[i]; 6055 if (c < 0x80) 6056 *d++ = c; 6057 else 6058 { 6059 *d++ = 0xc0 + ((unsigned)c >> 6); 6060 *d++ = 0x80 + (c & 0x3f); 6061 } 6062 } 6063 *d = NUL; 6064 if (lenp != NULL) 6065 *lenp = (int)(d - retval); 6066 break; 6067 6068 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */ 6069 retval = alloc(len * 3 + 1); 6070 if (retval == NULL) 6071 break; 6072 d = retval; 6073 for (i = 0; i < len; ++i) 6074 { 6075 c = ptr[i]; 6076 switch (c) 6077 { 6078 case 0xa4: c = 0x20ac; break; /* euro */ 6079 case 0xa6: c = 0x0160; break; /* S hat */ 6080 case 0xa8: c = 0x0161; break; /* S -hat */ 6081 case 0xb4: c = 0x017d; break; /* Z hat */ 6082 case 0xb8: c = 0x017e; break; /* Z -hat */ 6083 case 0xbc: c = 0x0152; break; /* OE */ 6084 case 0xbd: c = 0x0153; break; /* oe */ 6085 case 0xbe: c = 0x0178; break; /* Y */ 6086 } 6087 d += utf_char2bytes(c, d); 6088 } 6089 *d = NUL; 6090 if (lenp != NULL) 6091 *lenp = (int)(d - retval); 6092 break; 6093 6094 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */ 6095 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */ 6096 retval = alloc(len + 1); 6097 if (retval == NULL) 6098 break; 6099 d = retval; 6100 for (i = 0; i < len; ++i) 6101 { 6102 l = utf_ptr2len_len(ptr + i, len - i); 6103 if (l == 0) 6104 *d++ = NUL; 6105 else if (l == 1) 6106 { 6107 int l_w = utf8len_tab_zero[ptr[i]]; 6108 6109 if (l_w == 0) 6110 { 6111 /* Illegal utf-8 byte cannot be converted */ 6112 vim_free(retval); 6113 return NULL; 6114 } 6115 if (unconvlenp != NULL && l_w > len - i) 6116 { 6117 /* Incomplete sequence at the end. */ 6118 *unconvlenp = len - i; 6119 break; 6120 } 6121 *d++ = ptr[i]; 6122 } 6123 else 6124 { 6125 c = utf_ptr2char(ptr + i); 6126 if (vcp->vc_type == CONV_TO_LATIN9) 6127 switch (c) 6128 { 6129 case 0x20ac: c = 0xa4; break; /* euro */ 6130 case 0x0160: c = 0xa6; break; /* S hat */ 6131 case 0x0161: c = 0xa8; break; /* S -hat */ 6132 case 0x017d: c = 0xb4; break; /* Z hat */ 6133 case 0x017e: c = 0xb8; break; /* Z -hat */ 6134 case 0x0152: c = 0xbc; break; /* OE */ 6135 case 0x0153: c = 0xbd; break; /* oe */ 6136 case 0x0178: c = 0xbe; break; /* Y */ 6137 case 0xa4: 6138 case 0xa6: 6139 case 0xa8: 6140 case 0xb4: 6141 case 0xb8: 6142 case 0xbc: 6143 case 0xbd: 6144 case 0xbe: c = 0x100; break; /* not in latin9 */ 6145 } 6146 if (!utf_iscomposing(c)) /* skip composing chars */ 6147 { 6148 if (c < 0x100) 6149 *d++ = c; 6150 else if (vcp->vc_fail) 6151 { 6152 vim_free(retval); 6153 return NULL; 6154 } 6155 else 6156 { 6157 *d++ = 0xbf; 6158 if (utf_char2cells(c) > 1) 6159 *d++ = '?'; 6160 } 6161 } 6162 i += l - 1; 6163 } 6164 } 6165 *d = NUL; 6166 if (lenp != NULL) 6167 *lenp = (int)(d - retval); 6168 break; 6169 6170 # ifdef MACOS_CONVERT 6171 case CONV_MAC_LATIN1: 6172 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6173 'm', 'l', unconvlenp); 6174 break; 6175 6176 case CONV_LATIN1_MAC: 6177 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6178 'l', 'm', unconvlenp); 6179 break; 6180 6181 case CONV_MAC_UTF8: 6182 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6183 'm', 'u', unconvlenp); 6184 break; 6185 6186 case CONV_UTF8_MAC: 6187 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6188 'u', 'm', unconvlenp); 6189 break; 6190 # endif 6191 6192 # ifdef USE_ICONV 6193 case CONV_ICONV: /* conversion with output_conv.vc_fd */ 6194 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 6195 break; 6196 # endif 6197 # ifdef WIN3264 6198 case CONV_CODEPAGE: /* codepage -> codepage */ 6199 { 6200 int retlen; 6201 int tmp_len; 6202 short_u *tmp; 6203 6204 /* 1. codepage/UTF-8 -> ucs-2. */ 6205 if (vcp->vc_cpfrom == 0) 6206 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 6207 else 6208 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 0, 6209 ptr, len, 0, 0); 6210 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len); 6211 if (tmp == NULL) 6212 break; 6213 if (vcp->vc_cpfrom == 0) 6214 utf8_to_utf16(ptr, len, tmp, unconvlenp); 6215 else 6216 MultiByteToWideChar(vcp->vc_cpfrom, 0, ptr, len, tmp, tmp_len); 6217 6218 /* 2. ucs-2 -> codepage/UTF-8. */ 6219 if (vcp->vc_cpto == 0) 6220 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 6221 else 6222 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 6223 tmp, tmp_len, 0, 0, 0, 0); 6224 retval = alloc(retlen + 1); 6225 if (retval != NULL) 6226 { 6227 if (vcp->vc_cpto == 0) 6228 utf16_to_utf8(tmp, tmp_len, retval); 6229 else 6230 WideCharToMultiByte(vcp->vc_cpto, 0, 6231 tmp, tmp_len, retval, retlen, 0, 0); 6232 retval[retlen] = NUL; 6233 if (lenp != NULL) 6234 *lenp = retlen; 6235 } 6236 vim_free(tmp); 6237 break; 6238 } 6239 # endif 6240 } 6241 6242 return retval; 6243 } 6244 #endif 6245