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