1 /* vi:set ts=8 sts=4 sw=4: 2 * 3 * VIM - Vi IMproved by Bram Moolenaar 4 * Multibyte extensions partly by Sung-Hoon Baek 5 * 6 * Do ":help uganda" in Vim to read copying and usage conditions. 7 * Do ":help credits" in Vim to see a list of people who contributed. 8 * See README.txt for an overview of the Vim source code. 9 */ 10 /* 11 * mbyte.c: Code specifically for handling multi-byte characters. 12 * 13 * The encoding used in the core is set with 'encoding'. When 'encoding' is 14 * changed, the following four variables are set (for speed). 15 * Currently these types of character encodings are supported: 16 * 17 * "enc_dbcs" When non-zero it tells the type of double byte character 18 * encoding (Chinese, Korean, Japanese, etc.). 19 * The cell width on the display is equal to the number of 20 * bytes. (exception: DBCS_JPNU with first byte 0x8e) 21 * Recognizing the first or second byte is difficult, it 22 * requires checking a byte sequence from the start. 23 * "enc_utf8" When TRUE use Unicode characters in UTF-8 encoding. 24 * The cell width on the display needs to be determined from 25 * the character value. 26 * Recognizing bytes is easy: 0xxx.xxxx is a single-byte 27 * char, 10xx.xxxx is a trailing byte, 11xx.xxxx is a leading 28 * byte of a multi-byte character. 29 * To make things complicated, up to six composing characters 30 * are allowed. These are drawn on top of the first char. 31 * For most editing the sequence of bytes with composing 32 * characters included is considered to be one character. 33 * "enc_unicode" When 2 use 16-bit Unicode characters (or UTF-16). 34 * When 4 use 32-but Unicode characters. 35 * Internally characters are stored in UTF-8 encoding to 36 * avoid NUL bytes. Conversion happens when doing I/O. 37 * "enc_utf8" will also be TRUE. 38 * 39 * "has_mbyte" is set when "enc_dbcs" or "enc_utf8" is non-zero. 40 * 41 * If none of these is TRUE, 8-bit bytes are used for a character. The 42 * encoding isn't currently specified (TODO). 43 * 44 * 'encoding' specifies the encoding used in the core. This is in registers, 45 * text manipulation, buffers, etc. Conversion has to be done when characters 46 * in another encoding are received or send: 47 * 48 * clipboard 49 * ^ 50 * | (2) 51 * V 52 * +---------------+ 53 * (1) | | (3) 54 * keyboard ----->| core |-----> display 55 * | | 56 * +---------------+ 57 * ^ 58 * | (4) 59 * V 60 * file 61 * 62 * (1) Typed characters arrive in the current locale. Conversion is to be 63 * done when 'encoding' is different from 'termencoding'. 64 * (2) Text will be made available with the encoding specified with 65 * 'encoding'. If this is not sufficient, system-specific conversion 66 * might be required. 67 * (3) For the GUI the correct font must be selected, no conversion done. 68 * Otherwise, conversion is to be done when 'encoding' differs from 69 * 'termencoding'. (Different in the GTK+ 2 port -- 'termencoding' 70 * is always used for both input and output and must always be set to 71 * "utf-8". gui_mch_init() does this automatically.) 72 * (4) The encoding of the file is specified with 'fileencoding'. Conversion 73 * is to be done when it's different from 'encoding'. 74 * 75 * The viminfo file is a special case: Only text is converted, not file names. 76 * Vim scripts may contain an ":encoding" command. This has an effect for 77 * some commands, like ":menutrans" 78 */ 79 80 #include "vim.h" 81 82 #ifdef WIN32UNIX 83 # ifndef WIN32_LEAN_AND_MEAN 84 # define WIN32_LEAN_AND_MEAN 85 # endif 86 # if defined(FEAT_GUI) || defined(FEAT_XCLIPBOARD) 87 # include <X11/Xwindows.h> 88 # define WINBYTE wBYTE 89 # else 90 # include <windows.h> 91 # define WINBYTE BYTE 92 # endif 93 # ifdef WIN32 94 # undef WIN32 /* Some windows.h define WIN32, we don't want that here. */ 95 # endif 96 #else 97 # define WINBYTE BYTE 98 #endif 99 100 #if (defined(WIN3264) || defined(WIN32UNIX)) && !defined(__MINGW32__) 101 # include <winnls.h> 102 #endif 103 104 #ifdef FEAT_GUI_X11 105 # include <X11/Intrinsic.h> 106 #endif 107 #ifdef X_LOCALE 108 #include <X11/Xlocale.h> 109 #endif 110 111 #if defined(FEAT_GUI_GTK) && defined(FEAT_XIM) 112 # if GTK_CHECK_VERSION(3,0,0) 113 # include <gdk/gdkkeysyms-compat.h> 114 # else 115 # include <gdk/gdkkeysyms.h> 116 # endif 117 # ifdef WIN3264 118 # include <gdk/gdkwin32.h> 119 # else 120 # include <gdk/gdkx.h> 121 # endif 122 #endif 123 124 #ifdef HAVE_WCHAR_H 125 # include <wchar.h> 126 #endif 127 128 #if 0 129 /* This has been disabled, because several people reported problems with the 130 * wcwidth() and iswprint() library functions, esp. for Hebrew. */ 131 # ifdef __STDC_ISO_10646__ 132 # define USE_WCHAR_FUNCTIONS 133 # endif 134 #endif 135 136 #if defined(FEAT_MBYTE) || defined(PROTO) 137 138 static int enc_canon_search(char_u *name); 139 static int dbcs_char2len(int c); 140 static int dbcs_char2bytes(int c, char_u *buf); 141 static int dbcs_ptr2len(char_u *p); 142 static int dbcs_ptr2len_len(char_u *p, int size); 143 static int utf_ptr2cells_len(char_u *p, int size); 144 static int dbcs_char2cells(int c); 145 static int dbcs_ptr2cells_len(char_u *p, int size); 146 static int dbcs_ptr2char(char_u *p); 147 static int utf_safe_read_char_adv(char_u **s, size_t *n); 148 149 /* 150 * Lookup table to quickly get the length in bytes of a UTF-8 character from 151 * the first byte of a UTF-8 string. 152 * Bytes which are illegal when used as the first byte have a 1. 153 * The NUL byte has length 1. 154 */ 155 static char utf8len_tab[256] = 156 { 157 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 158 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 159 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 160 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 161 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 162 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 163 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, 164 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6,1,1, 165 }; 166 167 /* 168 * Like utf8len_tab above, but using a zero for illegal lead bytes. 169 */ 170 static char utf8len_tab_zero[256] = 171 { 172 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 173 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 174 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 175 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 176 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 177 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 178 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, 179 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6,0,0, 180 }; 181 182 /* 183 * XIM often causes trouble. Define XIM_DEBUG to get a log of XIM callbacks 184 * in the "xim.log" file. 185 */ 186 /* #define XIM_DEBUG */ 187 #ifdef XIM_DEBUG 188 static void 189 xim_log(char *s, ...) 190 { 191 va_list arglist; 192 static FILE *fd = NULL; 193 194 if (fd == (FILE *)-1) 195 return; 196 if (fd == NULL) 197 { 198 fd = mch_fopen("xim.log", "w"); 199 if (fd == NULL) 200 { 201 EMSG("Cannot open xim.log"); 202 fd = (FILE *)-1; 203 return; 204 } 205 } 206 207 va_start(arglist, s); 208 vfprintf(fd, s, arglist); 209 va_end(arglist); 210 } 211 #endif 212 213 #endif 214 215 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 216 /* 217 * Canonical encoding names and their properties. 218 * "iso-8859-n" is handled by enc_canonize() directly. 219 */ 220 static struct 221 { char *name; int prop; int codepage;} 222 enc_canon_table[] = 223 { 224 #define IDX_LATIN_1 0 225 {"latin1", ENC_8BIT + ENC_LATIN1, 1252}, 226 #define IDX_ISO_2 1 227 {"iso-8859-2", ENC_8BIT, 0}, 228 #define IDX_ISO_3 2 229 {"iso-8859-3", ENC_8BIT, 0}, 230 #define IDX_ISO_4 3 231 {"iso-8859-4", ENC_8BIT, 0}, 232 #define IDX_ISO_5 4 233 {"iso-8859-5", ENC_8BIT, 0}, 234 #define IDX_ISO_6 5 235 {"iso-8859-6", ENC_8BIT, 0}, 236 #define IDX_ISO_7 6 237 {"iso-8859-7", ENC_8BIT, 0}, 238 #define IDX_ISO_8 7 239 {"iso-8859-8", ENC_8BIT, 0}, 240 #define IDX_ISO_9 8 241 {"iso-8859-9", ENC_8BIT, 0}, 242 #define IDX_ISO_10 9 243 {"iso-8859-10", ENC_8BIT, 0}, 244 #define IDX_ISO_11 10 245 {"iso-8859-11", ENC_8BIT, 0}, 246 #define IDX_ISO_13 11 247 {"iso-8859-13", ENC_8BIT, 0}, 248 #define IDX_ISO_14 12 249 {"iso-8859-14", ENC_8BIT, 0}, 250 #define IDX_ISO_15 13 251 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0}, 252 #define IDX_KOI8_R 14 253 {"koi8-r", ENC_8BIT, 0}, 254 #define IDX_KOI8_U 15 255 {"koi8-u", ENC_8BIT, 0}, 256 #define IDX_UTF8 16 257 {"utf-8", ENC_UNICODE, 0}, 258 #define IDX_UCS2 17 259 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0}, 260 #define IDX_UCS2LE 18 261 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0}, 262 #define IDX_UTF16 19 263 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0}, 264 #define IDX_UTF16LE 20 265 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0}, 266 #define IDX_UCS4 21 267 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0}, 268 #define IDX_UCS4LE 22 269 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0}, 270 271 /* For debugging DBCS encoding on Unix. */ 272 #define IDX_DEBUG 23 273 {"debug", ENC_DBCS, DBCS_DEBUG}, 274 #define IDX_EUC_JP 24 275 {"euc-jp", ENC_DBCS, DBCS_JPNU}, 276 #define IDX_SJIS 25 277 {"sjis", ENC_DBCS, DBCS_JPN}, 278 #define IDX_EUC_KR 26 279 {"euc-kr", ENC_DBCS, DBCS_KORU}, 280 #define IDX_EUC_CN 27 281 {"euc-cn", ENC_DBCS, DBCS_CHSU}, 282 #define IDX_EUC_TW 28 283 {"euc-tw", ENC_DBCS, DBCS_CHTU}, 284 #define IDX_BIG5 29 285 {"big5", ENC_DBCS, DBCS_CHT}, 286 287 /* MS-DOS and MS-Windows codepages are included here, so that they can be 288 * used on Unix too. Most of them are similar to ISO-8859 encodings, but 289 * not exactly the same. */ 290 #define IDX_CP437 30 291 {"cp437", ENC_8BIT, 437}, /* like iso-8859-1 */ 292 #define IDX_CP737 31 293 {"cp737", ENC_8BIT, 737}, /* like iso-8859-7 */ 294 #define IDX_CP775 32 295 {"cp775", ENC_8BIT, 775}, /* Baltic */ 296 #define IDX_CP850 33 297 {"cp850", ENC_8BIT, 850}, /* like iso-8859-4 */ 298 #define IDX_CP852 34 299 {"cp852", ENC_8BIT, 852}, /* like iso-8859-1 */ 300 #define IDX_CP855 35 301 {"cp855", ENC_8BIT, 855}, /* like iso-8859-2 */ 302 #define IDX_CP857 36 303 {"cp857", ENC_8BIT, 857}, /* like iso-8859-5 */ 304 #define IDX_CP860 37 305 {"cp860", ENC_8BIT, 860}, /* like iso-8859-9 */ 306 #define IDX_CP861 38 307 {"cp861", ENC_8BIT, 861}, /* like iso-8859-1 */ 308 #define IDX_CP862 39 309 {"cp862", ENC_8BIT, 862}, /* like iso-8859-1 */ 310 #define IDX_CP863 40 311 {"cp863", ENC_8BIT, 863}, /* like iso-8859-8 */ 312 #define IDX_CP865 41 313 {"cp865", ENC_8BIT, 865}, /* like iso-8859-1 */ 314 #define IDX_CP866 42 315 {"cp866", ENC_8BIT, 866}, /* like iso-8859-5 */ 316 #define IDX_CP869 43 317 {"cp869", ENC_8BIT, 869}, /* like iso-8859-7 */ 318 #define IDX_CP874 44 319 {"cp874", ENC_8BIT, 874}, /* Thai */ 320 #define IDX_CP932 45 321 {"cp932", ENC_DBCS, DBCS_JPN}, 322 #define IDX_CP936 46 323 {"cp936", ENC_DBCS, DBCS_CHS}, 324 #define IDX_CP949 47 325 {"cp949", ENC_DBCS, DBCS_KOR}, 326 #define IDX_CP950 48 327 {"cp950", ENC_DBCS, DBCS_CHT}, 328 #define IDX_CP1250 49 329 {"cp1250", ENC_8BIT, 1250}, /* Czech, Polish, etc. */ 330 #define IDX_CP1251 50 331 {"cp1251", ENC_8BIT, 1251}, /* Cyrillic */ 332 /* cp1252 is considered to be equal to latin1 */ 333 #define IDX_CP1253 51 334 {"cp1253", ENC_8BIT, 1253}, /* Greek */ 335 #define IDX_CP1254 52 336 {"cp1254", ENC_8BIT, 1254}, /* Turkish */ 337 #define IDX_CP1255 53 338 {"cp1255", ENC_8BIT, 1255}, /* Hebrew */ 339 #define IDX_CP1256 54 340 {"cp1256", ENC_8BIT, 1256}, /* Arabic */ 341 #define IDX_CP1257 55 342 {"cp1257", ENC_8BIT, 1257}, /* Baltic */ 343 #define IDX_CP1258 56 344 {"cp1258", ENC_8BIT, 1258}, /* Vietnamese */ 345 346 #define IDX_MACROMAN 57 347 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, /* Mac OS */ 348 #define IDX_DECMCS 58 349 {"dec-mcs", ENC_8BIT, 0}, /* DEC MCS */ 350 #define IDX_HPROMAN8 59 351 {"hp-roman8", ENC_8BIT, 0}, /* HP Roman8 */ 352 #define IDX_COUNT 60 353 }; 354 355 /* 356 * Aliases for encoding names. 357 */ 358 static struct 359 { char *name; int canon;} 360 enc_alias_table[] = 361 { 362 {"ansi", IDX_LATIN_1}, 363 {"iso-8859-1", IDX_LATIN_1}, 364 {"latin2", IDX_ISO_2}, 365 {"latin3", IDX_ISO_3}, 366 {"latin4", IDX_ISO_4}, 367 {"cyrillic", IDX_ISO_5}, 368 {"arabic", IDX_ISO_6}, 369 {"greek", IDX_ISO_7}, 370 #ifdef WIN3264 371 {"hebrew", IDX_CP1255}, 372 #else 373 {"hebrew", IDX_ISO_8}, 374 #endif 375 {"latin5", IDX_ISO_9}, 376 {"turkish", IDX_ISO_9}, /* ? */ 377 {"latin6", IDX_ISO_10}, 378 {"nordic", IDX_ISO_10}, /* ? */ 379 {"thai", IDX_ISO_11}, /* ? */ 380 {"latin7", IDX_ISO_13}, 381 {"latin8", IDX_ISO_14}, 382 {"latin9", IDX_ISO_15}, 383 {"utf8", IDX_UTF8}, 384 {"unicode", IDX_UCS2}, 385 {"ucs2", IDX_UCS2}, 386 {"ucs2be", IDX_UCS2}, 387 {"ucs-2be", IDX_UCS2}, 388 {"ucs2le", IDX_UCS2LE}, 389 {"utf16", IDX_UTF16}, 390 {"utf16be", IDX_UTF16}, 391 {"utf-16be", IDX_UTF16}, 392 {"utf16le", IDX_UTF16LE}, 393 {"ucs4", IDX_UCS4}, 394 {"ucs4be", IDX_UCS4}, 395 {"ucs-4be", IDX_UCS4}, 396 {"ucs4le", IDX_UCS4LE}, 397 {"utf32", IDX_UCS4}, 398 {"utf-32", IDX_UCS4}, 399 {"utf32be", IDX_UCS4}, 400 {"utf-32be", IDX_UCS4}, 401 {"utf32le", IDX_UCS4LE}, 402 {"utf-32le", IDX_UCS4LE}, 403 {"932", IDX_CP932}, 404 {"949", IDX_CP949}, 405 {"936", IDX_CP936}, 406 {"gbk", IDX_CP936}, 407 {"950", IDX_CP950}, 408 {"eucjp", IDX_EUC_JP}, 409 {"unix-jis", IDX_EUC_JP}, 410 {"ujis", IDX_EUC_JP}, 411 {"shift-jis", IDX_SJIS}, 412 {"pck", IDX_SJIS}, /* Sun: PCK */ 413 {"euckr", IDX_EUC_KR}, 414 {"5601", IDX_EUC_KR}, /* Sun: KS C 5601 */ 415 {"euccn", IDX_EUC_CN}, 416 {"gb2312", IDX_EUC_CN}, 417 {"euctw", IDX_EUC_TW}, 418 #if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS) 419 {"japan", IDX_CP932}, 420 {"korea", IDX_CP949}, 421 {"prc", IDX_CP936}, 422 {"chinese", IDX_CP936}, 423 {"taiwan", IDX_CP950}, 424 {"big5", IDX_CP950}, 425 #else 426 {"japan", IDX_EUC_JP}, 427 {"korea", IDX_EUC_KR}, 428 {"prc", IDX_EUC_CN}, 429 {"chinese", IDX_EUC_CN}, 430 {"taiwan", IDX_EUC_TW}, 431 {"cp950", IDX_BIG5}, 432 {"950", IDX_BIG5}, 433 #endif 434 {"mac", IDX_MACROMAN}, 435 {"mac-roman", IDX_MACROMAN}, 436 {NULL, 0} 437 }; 438 439 #ifndef CP_UTF8 440 # define CP_UTF8 65001 /* magic number from winnls.h */ 441 #endif 442 443 /* 444 * Find encoding "name" in the list of canonical encoding names. 445 * Returns -1 if not found. 446 */ 447 static int 448 enc_canon_search(char_u *name) 449 { 450 int i; 451 452 for (i = 0; i < IDX_COUNT; ++i) 453 if (STRCMP(name, enc_canon_table[i].name) == 0) 454 return i; 455 return -1; 456 } 457 458 #endif 459 460 #if defined(FEAT_MBYTE) || defined(PROTO) 461 462 /* 463 * Find canonical encoding "name" in the list and return its properties. 464 * Returns 0 if not found. 465 */ 466 int 467 enc_canon_props(char_u *name) 468 { 469 int i; 470 471 i = enc_canon_search(name); 472 if (i >= 0) 473 return enc_canon_table[i].prop; 474 #ifdef WIN3264 475 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2])) 476 { 477 CPINFO cpinfo; 478 479 /* Get info on this codepage to find out what it is. */ 480 if (GetCPInfo(atoi((char *)name + 2), &cpinfo) != 0) 481 { 482 if (cpinfo.MaxCharSize == 1) /* some single-byte encoding */ 483 return ENC_8BIT; 484 if (cpinfo.MaxCharSize == 2 485 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 486 /* must be a DBCS encoding */ 487 return ENC_DBCS; 488 } 489 return 0; 490 } 491 #endif 492 if (STRNCMP(name, "2byte-", 6) == 0) 493 return ENC_DBCS; 494 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0) 495 return ENC_8BIT; 496 return 0; 497 } 498 499 /* 500 * Set up for using multi-byte characters. 501 * Called in three cases: 502 * - by main() to initialize (p_enc == NULL) 503 * - by set_init_1() after 'encoding' was set to its default. 504 * - by do_set() when 'encoding' has been set. 505 * p_enc must have been passed through enc_canonize() already. 506 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags. 507 * Fills mb_bytelen_tab[] and returns NULL when there are no problems. 508 * When there is something wrong: Returns an error message and doesn't change 509 * anything. 510 */ 511 char_u * 512 mb_init(void) 513 { 514 int i; 515 int idx; 516 int n; 517 int enc_dbcs_new = 0; 518 #if defined(USE_ICONV) && !defined(WIN3264) && !defined(WIN32UNIX) \ 519 && !defined(MACOS) 520 # define LEN_FROM_CONV 521 vimconv_T vimconv; 522 char_u *p; 523 #endif 524 525 if (p_enc == NULL) 526 { 527 /* Just starting up: set the whole table to one's. */ 528 for (i = 0; i < 256; ++i) 529 mb_bytelen_tab[i] = 1; 530 input_conv.vc_type = CONV_NONE; 531 input_conv.vc_factor = 1; 532 output_conv.vc_type = CONV_NONE; 533 return NULL; 534 } 535 536 #ifdef WIN3264 537 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2])) 538 { 539 CPINFO cpinfo; 540 541 /* Get info on this codepage to find out what it is. */ 542 if (GetCPInfo(atoi((char *)p_enc + 2), &cpinfo) != 0) 543 { 544 if (cpinfo.MaxCharSize == 1) 545 { 546 /* some single-byte encoding */ 547 enc_unicode = 0; 548 enc_utf8 = FALSE; 549 } 550 else if (cpinfo.MaxCharSize == 2 551 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 552 { 553 /* must be a DBCS encoding, check below */ 554 enc_dbcs_new = atoi((char *)p_enc + 2); 555 } 556 else 557 goto codepage_invalid; 558 } 559 else if (GetLastError() == ERROR_INVALID_PARAMETER) 560 { 561 codepage_invalid: 562 return (char_u *)N_("E543: Not a valid codepage"); 563 } 564 } 565 #endif 566 else if (STRNCMP(p_enc, "8bit-", 5) == 0 567 || STRNCMP(p_enc, "iso-8859-", 9) == 0) 568 { 569 /* Accept any "8bit-" or "iso-8859-" name. */ 570 enc_unicode = 0; 571 enc_utf8 = FALSE; 572 } 573 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 574 { 575 #ifdef WIN3264 576 /* Windows: accept only valid codepage numbers, check below. */ 577 if (p_enc[6] != 'c' || p_enc[7] != 'p' 578 || (enc_dbcs_new = atoi((char *)p_enc + 8)) == 0) 579 return e_invarg; 580 #else 581 /* Unix: accept any "2byte-" name, assume current locale. */ 582 enc_dbcs_new = DBCS_2BYTE; 583 #endif 584 } 585 else if ((idx = enc_canon_search(p_enc)) >= 0) 586 { 587 i = enc_canon_table[idx].prop; 588 if (i & ENC_UNICODE) 589 { 590 /* Unicode */ 591 enc_utf8 = TRUE; 592 if (i & (ENC_2BYTE | ENC_2WORD)) 593 enc_unicode = 2; 594 else if (i & ENC_4BYTE) 595 enc_unicode = 4; 596 else 597 enc_unicode = 0; 598 } 599 else if (i & ENC_DBCS) 600 { 601 /* 2byte, handle below */ 602 enc_dbcs_new = enc_canon_table[idx].codepage; 603 } 604 else 605 { 606 /* Must be 8-bit. */ 607 enc_unicode = 0; 608 enc_utf8 = FALSE; 609 } 610 } 611 else /* Don't know what encoding this is, reject it. */ 612 return e_invarg; 613 614 if (enc_dbcs_new != 0) 615 { 616 #ifdef WIN3264 617 /* Check if the DBCS code page is OK. */ 618 if (!IsValidCodePage(enc_dbcs_new)) 619 goto codepage_invalid; 620 #endif 621 enc_unicode = 0; 622 enc_utf8 = FALSE; 623 } 624 enc_dbcs = enc_dbcs_new; 625 has_mbyte = (enc_dbcs != 0 || enc_utf8); 626 627 #if defined(WIN3264) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 628 enc_codepage = encname2codepage(p_enc); 629 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0); 630 #endif 631 632 /* Detect an encoding that uses latin1 characters. */ 633 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0 634 || STRCMP(p_enc, "iso-8859-15") == 0); 635 636 /* 637 * Set the function pointers. 638 */ 639 if (enc_utf8) 640 { 641 mb_ptr2len = utfc_ptr2len; 642 mb_ptr2len_len = utfc_ptr2len_len; 643 mb_char2len = utf_char2len; 644 mb_char2bytes = utf_char2bytes; 645 mb_ptr2cells = utf_ptr2cells; 646 mb_ptr2cells_len = utf_ptr2cells_len; 647 mb_char2cells = utf_char2cells; 648 mb_off2cells = utf_off2cells; 649 mb_ptr2char = utf_ptr2char; 650 mb_head_off = utf_head_off; 651 } 652 else if (enc_dbcs != 0) 653 { 654 mb_ptr2len = dbcs_ptr2len; 655 mb_ptr2len_len = dbcs_ptr2len_len; 656 mb_char2len = dbcs_char2len; 657 mb_char2bytes = dbcs_char2bytes; 658 mb_ptr2cells = dbcs_ptr2cells; 659 mb_ptr2cells_len = dbcs_ptr2cells_len; 660 mb_char2cells = dbcs_char2cells; 661 mb_off2cells = dbcs_off2cells; 662 mb_ptr2char = dbcs_ptr2char; 663 mb_head_off = dbcs_head_off; 664 } 665 else 666 { 667 mb_ptr2len = latin_ptr2len; 668 mb_ptr2len_len = latin_ptr2len_len; 669 mb_char2len = latin_char2len; 670 mb_char2bytes = latin_char2bytes; 671 mb_ptr2cells = latin_ptr2cells; 672 mb_ptr2cells_len = latin_ptr2cells_len; 673 mb_char2cells = latin_char2cells; 674 mb_off2cells = latin_off2cells; 675 mb_ptr2char = latin_ptr2char; 676 mb_head_off = latin_head_off; 677 } 678 679 /* 680 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN(). 681 */ 682 #ifdef LEN_FROM_CONV 683 /* When 'encoding' is different from the current locale mblen() won't 684 * work. Use conversion to "utf-8" instead. */ 685 vimconv.vc_type = CONV_NONE; 686 if (enc_dbcs) 687 { 688 p = enc_locale(); 689 if (p == NULL || STRCMP(p, p_enc) != 0) 690 { 691 convert_setup(&vimconv, p_enc, (char_u *)"utf-8"); 692 vimconv.vc_fail = TRUE; 693 } 694 vim_free(p); 695 } 696 #endif 697 698 for (i = 0; i < 256; ++i) 699 { 700 /* Our own function to reliably check the length of UTF-8 characters, 701 * independent of mblen(). */ 702 if (enc_utf8) 703 n = utf8len_tab[i]; 704 else if (enc_dbcs == 0) 705 n = 1; 706 else 707 { 708 #if defined(WIN3264) || defined(WIN32UNIX) 709 /* enc_dbcs is set by setting 'fileencoding'. It becomes a Windows 710 * CodePage identifier, which we can pass directly in to Windows 711 * API */ 712 n = IsDBCSLeadByteEx(enc_dbcs, (WINBYTE)i) ? 2 : 1; 713 #else 714 # if defined(MACOS) || defined(__amigaos4__) || defined(__ANDROID__) 715 /* 716 * if mblen() is not available, character which MSB is turned on 717 * are treated as leading byte character. (note : This assumption 718 * is not always true.) 719 */ 720 n = (i & 0x80) ? 2 : 1; 721 # else 722 char buf[MB_MAXBYTES + 1]; 723 # ifdef X_LOCALE 724 # ifndef mblen 725 # define mblen _Xmblen 726 # endif 727 # endif 728 if (i == NUL) /* just in case mblen() can't handle "" */ 729 n = 1; 730 else 731 { 732 buf[0] = i; 733 buf[1] = 0; 734 #ifdef LEN_FROM_CONV 735 if (vimconv.vc_type != CONV_NONE) 736 { 737 /* 738 * string_convert() should fail when converting the first 739 * byte of a double-byte character. 740 */ 741 p = string_convert(&vimconv, (char_u *)buf, NULL); 742 if (p != NULL) 743 { 744 vim_free(p); 745 n = 1; 746 } 747 else 748 n = 2; 749 } 750 else 751 #endif 752 { 753 /* 754 * mblen() should return -1 for invalid (means the leading 755 * multibyte) character. However there are some platforms 756 * where mblen() returns 0 for invalid character. 757 * Therefore, following condition includes 0. 758 */ 759 ignored = mblen(NULL, 0); /* First reset the state. */ 760 if (mblen(buf, (size_t)1) <= 0) 761 n = 2; 762 else 763 n = 1; 764 } 765 } 766 # endif 767 #endif 768 } 769 770 mb_bytelen_tab[i] = n; 771 } 772 773 #ifdef LEN_FROM_CONV 774 convert_setup(&vimconv, NULL, NULL); 775 #endif 776 777 /* The cell width depends on the type of multi-byte characters. */ 778 (void)init_chartab(); 779 780 /* When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] */ 781 screenalloc(FALSE); 782 783 /* When using Unicode, set default for 'fileencodings'. */ 784 if (enc_utf8 && !option_was_set((char_u *)"fencs")) 785 set_string_option_direct((char_u *)"fencs", -1, 786 (char_u *)"ucs-bom,utf-8,default,latin1", OPT_FREE, 0); 787 788 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT) 789 /* GNU gettext 0.10.37 supports this feature: set the codeset used for 790 * translated messages independently from the current locale. */ 791 (void)bind_textdomain_codeset(VIMPACKAGE, 792 enc_utf8 ? "utf-8" : (char *)p_enc); 793 #endif 794 795 #ifdef WIN32 796 /* When changing 'encoding' while starting up, then convert the command 797 * line arguments from the active codepage to 'encoding'. */ 798 if (starting != 0) 799 fix_arg_enc(); 800 #endif 801 802 #ifdef FEAT_AUTOCMD 803 /* Fire an autocommand to let people do custom font setup. This must be 804 * after Vim has been setup for the new encoding. */ 805 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); 806 #endif 807 808 #ifdef FEAT_SPELL 809 /* Need to reload spell dictionaries */ 810 spell_reload(); 811 #endif 812 813 return NULL; 814 } 815 816 /* 817 * Return the size of the BOM for the current buffer: 818 * 0 - no BOM 819 * 2 - UCS-2 or UTF-16 BOM 820 * 4 - UCS-4 BOM 821 * 3 - UTF-8 BOM 822 */ 823 int 824 bomb_size(void) 825 { 826 int n = 0; 827 828 if (curbuf->b_p_bomb && !curbuf->b_p_bin) 829 { 830 if (*curbuf->b_p_fenc == NUL) 831 { 832 if (enc_utf8) 833 { 834 if (enc_unicode != 0) 835 n = enc_unicode; 836 else 837 n = 3; 838 } 839 } 840 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) 841 n = 3; 842 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 843 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) 844 n = 2; 845 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) 846 n = 4; 847 } 848 return n; 849 } 850 851 /* 852 * Remove all BOM from "s" by moving remaining text. 853 */ 854 void 855 remove_bom(char_u *s) 856 { 857 if (enc_utf8) 858 { 859 char_u *p = s; 860 861 while ((p = vim_strbyte(p, 0xef)) != NULL) 862 { 863 if (p[1] == 0xbb && p[2] == 0xbf) 864 STRMOVE(p, p + 3); 865 else 866 ++p; 867 } 868 } 869 } 870 871 /* 872 * Get class of pointer: 873 * 0 for blank or NUL 874 * 1 for punctuation 875 * 2 for an (ASCII) word character 876 * >2 for other word characters 877 */ 878 int 879 mb_get_class(char_u *p) 880 { 881 return mb_get_class_buf(p, curbuf); 882 } 883 884 int 885 mb_get_class_buf(char_u *p, buf_T *buf) 886 { 887 if (MB_BYTE2LEN(p[0]) == 1) 888 { 889 if (p[0] == NUL || vim_iswhite(p[0])) 890 return 0; 891 if (vim_iswordc_buf(p[0], buf)) 892 return 2; 893 return 1; 894 } 895 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) 896 return dbcs_class(p[0], p[1]); 897 if (enc_utf8) 898 return utf_class(utf_ptr2char(p)); 899 return 0; 900 } 901 902 /* 903 * Get class of a double-byte character. This always returns 3 or bigger. 904 * TODO: Should return 1 for punctuation. 905 */ 906 int 907 dbcs_class(unsigned lead, unsigned trail) 908 { 909 switch (enc_dbcs) 910 { 911 /* please add classify routine for your language in here */ 912 913 case DBCS_JPNU: /* ? */ 914 case DBCS_JPN: 915 { 916 /* JIS code classification */ 917 unsigned char lb = lead; 918 unsigned char tb = trail; 919 920 /* convert process code to JIS */ 921 # if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS) 922 /* process code is SJIS */ 923 if (lb <= 0x9f) 924 lb = (lb - 0x81) * 2 + 0x21; 925 else 926 lb = (lb - 0xc1) * 2 + 0x21; 927 if (tb <= 0x7e) 928 tb -= 0x1f; 929 else if (tb <= 0x9e) 930 tb -= 0x20; 931 else 932 { 933 tb -= 0x7e; 934 lb += 1; 935 } 936 # else 937 /* 938 * XXX: Code page identification can not use with all 939 * system! So, some other encoding information 940 * will be needed. 941 * In japanese: SJIS,EUC,UNICODE,(JIS) 942 * Note that JIS-code system don't use as 943 * process code in most system because it uses 944 * escape sequences(JIS is context depend encoding). 945 */ 946 /* assume process code is JAPANESE-EUC */ 947 lb &= 0x7f; 948 tb &= 0x7f; 949 # endif 950 /* exceptions */ 951 switch (lb << 8 | tb) 952 { 953 case 0x2121: /* ZENKAKU space */ 954 return 0; 955 case 0x2122: /* TOU-TEN (Japanese comma) */ 956 case 0x2123: /* KU-TEN (Japanese period) */ 957 case 0x2124: /* ZENKAKU comma */ 958 case 0x2125: /* ZENKAKU period */ 959 return 1; 960 case 0x213c: /* prolongedsound handled as KATAKANA */ 961 return 13; 962 } 963 /* sieved by KU code */ 964 switch (lb) 965 { 966 case 0x21: 967 case 0x22: 968 /* special symbols */ 969 return 10; 970 case 0x23: 971 /* alpha-numeric */ 972 return 11; 973 case 0x24: 974 /* hiragana */ 975 return 12; 976 case 0x25: 977 /* katakana */ 978 return 13; 979 case 0x26: 980 /* greek */ 981 return 14; 982 case 0x27: 983 /* russian */ 984 return 15; 985 case 0x28: 986 /* lines */ 987 return 16; 988 default: 989 /* kanji */ 990 return 17; 991 } 992 } 993 994 case DBCS_KORU: /* ? */ 995 case DBCS_KOR: 996 { 997 /* KS code classification */ 998 unsigned char c1 = lead; 999 unsigned char c2 = trail; 1000 1001 /* 1002 * 20 : Hangul 1003 * 21 : Hanja 1004 * 22 : Symbols 1005 * 23 : Alpha-numeric/Roman Letter (Full width) 1006 * 24 : Hangul Letter(Alphabet) 1007 * 25 : Roman Numeral/Greek Letter 1008 * 26 : Box Drawings 1009 * 27 : Unit Symbols 1010 * 28 : Circled/Parenthesized Letter 1011 * 29 : Hiragana/Katakana 1012 * 30 : Cyrillic Letter 1013 */ 1014 1015 if (c1 >= 0xB0 && c1 <= 0xC8) 1016 /* Hangul */ 1017 return 20; 1018 #if defined(WIN3264) || defined(WIN32UNIX) 1019 else if (c1 <= 0xA0 || c2 <= 0xA0) 1020 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */ 1021 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE 1022 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 1023 */ 1024 return 20; 1025 #endif 1026 1027 else if (c1 >= 0xCA && c1 <= 0xFD) 1028 /* Hanja */ 1029 return 21; 1030 else switch (c1) 1031 { 1032 case 0xA1: 1033 case 0xA2: 1034 /* Symbols */ 1035 return 22; 1036 case 0xA3: 1037 /* Alpha-numeric */ 1038 return 23; 1039 case 0xA4: 1040 /* Hangul Letter(Alphabet) */ 1041 return 24; 1042 case 0xA5: 1043 /* Roman Numeral/Greek Letter */ 1044 return 25; 1045 case 0xA6: 1046 /* Box Drawings */ 1047 return 26; 1048 case 0xA7: 1049 /* Unit Symbols */ 1050 return 27; 1051 case 0xA8: 1052 case 0xA9: 1053 if (c2 <= 0xAF) 1054 return 25; /* Roman Letter */ 1055 else if (c2 >= 0xF6) 1056 return 22; /* Symbols */ 1057 else 1058 /* Circled/Parenthesized Letter */ 1059 return 28; 1060 case 0xAA: 1061 case 0xAB: 1062 /* Hiragana/Katakana */ 1063 return 29; 1064 case 0xAC: 1065 /* Cyrillic Letter */ 1066 return 30; 1067 } 1068 } 1069 default: 1070 break; 1071 } 1072 return 3; 1073 } 1074 1075 /* 1076 * mb_char2len() function pointer. 1077 * Return length in bytes of character "c". 1078 * Returns 1 for a single-byte character. 1079 */ 1080 int 1081 latin_char2len(int c UNUSED) 1082 { 1083 return 1; 1084 } 1085 1086 static int 1087 dbcs_char2len( 1088 int c) 1089 { 1090 if (c >= 0x100) 1091 return 2; 1092 return 1; 1093 } 1094 1095 /* 1096 * mb_char2bytes() function pointer. 1097 * Convert a character to its bytes. 1098 * Returns the length in bytes. 1099 */ 1100 int 1101 latin_char2bytes(int c, char_u *buf) 1102 { 1103 buf[0] = c; 1104 return 1; 1105 } 1106 1107 static int 1108 dbcs_char2bytes(int c, char_u *buf) 1109 { 1110 if (c >= 0x100) 1111 { 1112 buf[0] = (unsigned)c >> 8; 1113 buf[1] = c; 1114 /* Never use a NUL byte, it causes lots of trouble. It's an invalid 1115 * character anyway. */ 1116 if (buf[1] == NUL) 1117 buf[1] = '\n'; 1118 return 2; 1119 } 1120 buf[0] = c; 1121 return 1; 1122 } 1123 1124 /* 1125 * mb_ptr2len() function pointer. 1126 * Get byte length of character at "*p" but stop at a NUL. 1127 * For UTF-8 this includes following composing characters. 1128 * Returns 0 when *p is NUL. 1129 */ 1130 int 1131 latin_ptr2len(char_u *p) 1132 { 1133 return MB_BYTE2LEN(*p); 1134 } 1135 1136 static int 1137 dbcs_ptr2len( 1138 char_u *p) 1139 { 1140 int len; 1141 1142 /* Check if second byte is not missing. */ 1143 len = MB_BYTE2LEN(*p); 1144 if (len == 2 && p[1] == NUL) 1145 len = 1; 1146 return len; 1147 } 1148 1149 /* 1150 * mb_ptr2len_len() function pointer. 1151 * Like mb_ptr2len(), but limit to read "size" bytes. 1152 * Returns 0 for an empty string. 1153 * Returns 1 for an illegal char or an incomplete byte sequence. 1154 */ 1155 int 1156 latin_ptr2len_len(char_u *p, int size) 1157 { 1158 if (size < 1 || *p == NUL) 1159 return 0; 1160 return 1; 1161 } 1162 1163 static int 1164 dbcs_ptr2len_len(char_u *p, int size) 1165 { 1166 int len; 1167 1168 if (size < 1 || *p == NUL) 1169 return 0; 1170 if (size == 1) 1171 return 1; 1172 /* Check that second byte is not missing. */ 1173 len = MB_BYTE2LEN(*p); 1174 if (len == 2 && p[1] == NUL) 1175 len = 1; 1176 return len; 1177 } 1178 1179 struct interval 1180 { 1181 long first; 1182 long last; 1183 }; 1184 1185 /* 1186 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]". 1187 */ 1188 static int 1189 intable(struct interval *table, size_t size, int c) 1190 { 1191 int mid, bot, top; 1192 1193 /* first quick check for Latin1 etc. characters */ 1194 if (c < table[0].first) 1195 return FALSE; 1196 1197 /* binary search in table */ 1198 bot = 0; 1199 top = (int)(size / sizeof(struct interval) - 1); 1200 while (top >= bot) 1201 { 1202 mid = (bot + top) / 2; 1203 if (table[mid].last < c) 1204 bot = mid + 1; 1205 else if (table[mid].first > c) 1206 top = mid - 1; 1207 else 1208 return TRUE; 1209 } 1210 return FALSE; 1211 } 1212 1213 /* 1214 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1215 * Returns 4 or 6 for an unprintable character. 1216 * Is only correct for characters >= 0x80. 1217 * When p_ambw is "double", return 2 for a character with East Asian Width 1218 * class 'A'(mbiguous). 1219 */ 1220 int 1221 utf_char2cells(int c) 1222 { 1223 /* Sorted list of non-overlapping intervals of East Asian double width 1224 * characters, generated with ../runtime/tools/unicode.vim. */ 1225 static struct interval doublewidth[] = 1226 { 1227 {0x1100, 0x115f}, 1228 {0x2329, 0x232a}, 1229 {0x2e80, 0x2e99}, 1230 {0x2e9b, 0x2ef3}, 1231 {0x2f00, 0x2fd5}, 1232 {0x2ff0, 0x2ffb}, 1233 {0x3000, 0x303e}, 1234 {0x3041, 0x3096}, 1235 {0x3099, 0x30ff}, 1236 {0x3105, 0x312d}, 1237 {0x3131, 0x318e}, 1238 {0x3190, 0x31ba}, 1239 {0x31c0, 0x31e3}, 1240 {0x31f0, 0x321e}, 1241 {0x3220, 0x3247}, 1242 {0x3250, 0x32fe}, 1243 {0x3300, 0x4dbf}, 1244 {0x4e00, 0xa48c}, 1245 {0xa490, 0xa4c6}, 1246 {0xa960, 0xa97c}, 1247 {0xac00, 0xd7a3}, 1248 {0xf900, 0xfaff}, 1249 {0xfe10, 0xfe19}, 1250 {0xfe30, 0xfe52}, 1251 {0xfe54, 0xfe66}, 1252 {0xfe68, 0xfe6b}, 1253 {0xff01, 0xff60}, 1254 {0xffe0, 0xffe6}, 1255 {0x20000, 0x2fffd}, 1256 {0x30000, 0x3fffd} 1257 }; 1258 1259 /* Sorted list of non-overlapping intervals of East Asian Ambiguous 1260 * characters, generated with ../runtime/tools/unicode.vim. */ 1261 static struct interval ambiguous[] = 1262 { 1263 {0x00a1, 0x00a1}, 1264 {0x00a4, 0x00a4}, 1265 {0x00a7, 0x00a8}, 1266 {0x00aa, 0x00aa}, 1267 {0x00ad, 0x00ae}, 1268 {0x00b0, 0x00b4}, 1269 {0x00b6, 0x00ba}, 1270 {0x00bc, 0x00bf}, 1271 {0x00c6, 0x00c6}, 1272 {0x00d0, 0x00d0}, 1273 {0x00d7, 0x00d8}, 1274 {0x00de, 0x00e1}, 1275 {0x00e6, 0x00e6}, 1276 {0x00e8, 0x00ea}, 1277 {0x00ec, 0x00ed}, 1278 {0x00f0, 0x00f0}, 1279 {0x00f2, 0x00f3}, 1280 {0x00f7, 0x00fa}, 1281 {0x00fc, 0x00fc}, 1282 {0x00fe, 0x00fe}, 1283 {0x0101, 0x0101}, 1284 {0x0111, 0x0111}, 1285 {0x0113, 0x0113}, 1286 {0x011b, 0x011b}, 1287 {0x0126, 0x0127}, 1288 {0x012b, 0x012b}, 1289 {0x0131, 0x0133}, 1290 {0x0138, 0x0138}, 1291 {0x013f, 0x0142}, 1292 {0x0144, 0x0144}, 1293 {0x0148, 0x014b}, 1294 {0x014d, 0x014d}, 1295 {0x0152, 0x0153}, 1296 {0x0166, 0x0167}, 1297 {0x016b, 0x016b}, 1298 {0x01ce, 0x01ce}, 1299 {0x01d0, 0x01d0}, 1300 {0x01d2, 0x01d2}, 1301 {0x01d4, 0x01d4}, 1302 {0x01d6, 0x01d6}, 1303 {0x01d8, 0x01d8}, 1304 {0x01da, 0x01da}, 1305 {0x01dc, 0x01dc}, 1306 {0x0251, 0x0251}, 1307 {0x0261, 0x0261}, 1308 {0x02c4, 0x02c4}, 1309 {0x02c7, 0x02c7}, 1310 {0x02c9, 0x02cb}, 1311 {0x02cd, 0x02cd}, 1312 {0x02d0, 0x02d0}, 1313 {0x02d8, 0x02db}, 1314 {0x02dd, 0x02dd}, 1315 {0x02df, 0x02df}, 1316 {0x0300, 0x036f}, 1317 {0x0391, 0x03a1}, 1318 {0x03a3, 0x03a9}, 1319 {0x03b1, 0x03c1}, 1320 {0x03c3, 0x03c9}, 1321 {0x0401, 0x0401}, 1322 {0x0410, 0x044f}, 1323 {0x0451, 0x0451}, 1324 {0x2010, 0x2010}, 1325 {0x2013, 0x2016}, 1326 {0x2018, 0x2019}, 1327 {0x201c, 0x201d}, 1328 {0x2020, 0x2022}, 1329 {0x2024, 0x2027}, 1330 {0x2030, 0x2030}, 1331 {0x2032, 0x2033}, 1332 {0x2035, 0x2035}, 1333 {0x203b, 0x203b}, 1334 {0x203e, 0x203e}, 1335 {0x2074, 0x2074}, 1336 {0x207f, 0x207f}, 1337 {0x2081, 0x2084}, 1338 {0x20ac, 0x20ac}, 1339 {0x2103, 0x2103}, 1340 {0x2105, 0x2105}, 1341 {0x2109, 0x2109}, 1342 {0x2113, 0x2113}, 1343 {0x2116, 0x2116}, 1344 {0x2121, 0x2122}, 1345 {0x2126, 0x2126}, 1346 {0x212b, 0x212b}, 1347 {0x2153, 0x2154}, 1348 {0x215b, 0x215e}, 1349 {0x2160, 0x216b}, 1350 {0x2170, 0x2179}, 1351 {0x2189, 0x2189}, 1352 {0x2190, 0x2199}, 1353 {0x21b8, 0x21b9}, 1354 {0x21d2, 0x21d2}, 1355 {0x21d4, 0x21d4}, 1356 {0x21e7, 0x21e7}, 1357 {0x2200, 0x2200}, 1358 {0x2202, 0x2203}, 1359 {0x2207, 0x2208}, 1360 {0x220b, 0x220b}, 1361 {0x220f, 0x220f}, 1362 {0x2211, 0x2211}, 1363 {0x2215, 0x2215}, 1364 {0x221a, 0x221a}, 1365 {0x221d, 0x2220}, 1366 {0x2223, 0x2223}, 1367 {0x2225, 0x2225}, 1368 {0x2227, 0x222c}, 1369 {0x222e, 0x222e}, 1370 {0x2234, 0x2237}, 1371 {0x223c, 0x223d}, 1372 {0x2248, 0x2248}, 1373 {0x224c, 0x224c}, 1374 {0x2252, 0x2252}, 1375 {0x2260, 0x2261}, 1376 {0x2264, 0x2267}, 1377 {0x226a, 0x226b}, 1378 {0x226e, 0x226f}, 1379 {0x2282, 0x2283}, 1380 {0x2286, 0x2287}, 1381 {0x2295, 0x2295}, 1382 {0x2299, 0x2299}, 1383 {0x22a5, 0x22a5}, 1384 {0x22bf, 0x22bf}, 1385 {0x2312, 0x2312}, 1386 {0x2460, 0x24e9}, 1387 {0x24eb, 0x254b}, 1388 {0x2550, 0x2573}, 1389 {0x2580, 0x258f}, 1390 {0x2592, 0x2595}, 1391 {0x25a0, 0x25a1}, 1392 {0x25a3, 0x25a9}, 1393 {0x25b2, 0x25b3}, 1394 {0x25b6, 0x25b7}, 1395 {0x25bc, 0x25bd}, 1396 {0x25c0, 0x25c1}, 1397 {0x25c6, 0x25c8}, 1398 {0x25cb, 0x25cb}, 1399 {0x25ce, 0x25d1}, 1400 {0x25e2, 0x25e5}, 1401 {0x25ef, 0x25ef}, 1402 {0x2605, 0x2606}, 1403 {0x2609, 0x2609}, 1404 {0x260e, 0x260f}, 1405 {0x2614, 0x2615}, 1406 {0x261c, 0x261c}, 1407 {0x261e, 0x261e}, 1408 {0x2640, 0x2640}, 1409 {0x2642, 0x2642}, 1410 {0x2660, 0x2661}, 1411 {0x2663, 0x2665}, 1412 {0x2667, 0x266a}, 1413 {0x266c, 0x266d}, 1414 {0x266f, 0x266f}, 1415 {0x269e, 0x269f}, 1416 {0x26be, 0x26bf}, 1417 {0x26c4, 0x26cd}, 1418 {0x26cf, 0x26e1}, 1419 {0x26e3, 0x26e3}, 1420 {0x26e8, 0x26ff}, 1421 {0x273d, 0x273d}, 1422 {0x2757, 0x2757}, 1423 {0x2776, 0x277f}, 1424 {0x2b55, 0x2b59}, 1425 {0x3248, 0x324f}, 1426 {0xe000, 0xf8ff}, 1427 {0xfe00, 0xfe0f}, 1428 {0xfffd, 0xfffd}, 1429 {0x1f100, 0x1f10a}, 1430 {0x1f110, 0x1f12d}, 1431 {0x1f130, 0x1f169}, 1432 {0x1f170, 0x1f19a}, 1433 {0xe0100, 0xe01ef}, 1434 {0xf0000, 0xffffd}, 1435 {0x100000, 0x10fffd} 1436 }; 1437 1438 /* Sorted list of non-overlapping intervals of Emoji characters that don't 1439 * have ambiguous or double width, 1440 * based on http://unicode.org/emoji/charts/emoji-list.html */ 1441 static struct interval emoji_width[] = 1442 { 1443 {0x203c, 0x203c}, 1444 {0x2049, 0x2049}, 1445 {0x2139, 0x2139}, 1446 {0x21a9, 0x21aa}, 1447 {0x231a, 0x231b}, 1448 {0x2328, 0x2328}, 1449 {0x23cf, 0x23cf}, 1450 {0x23e9, 0x23f3}, 1451 {0x25aa, 0x25ab}, 1452 {0x25fb, 0x25fe}, 1453 {0x2600, 0x2604}, 1454 {0x2611, 0x2611}, 1455 {0x2618, 0x2618}, 1456 {0x261d, 0x261d}, 1457 {0x2620, 0x2620}, 1458 {0x2622, 0x2623}, 1459 {0x2626, 0x2626}, 1460 {0x262a, 0x262a}, 1461 {0x262e, 0x262f}, 1462 {0x2638, 0x263a}, 1463 {0x2648, 0x2653}, 1464 {0x2666, 0x2666}, 1465 {0x267b, 0x267b}, 1466 {0x267f, 0x267f}, 1467 {0x2692, 0x2694}, 1468 {0x2696, 0x2697}, 1469 {0x2699, 0x2699}, 1470 {0x269b, 0x269c}, 1471 {0x26a0, 0x26a1}, 1472 {0x26aa, 0x26ab}, 1473 {0x26b0, 0x26b1}, 1474 {0x26bd, 0x26bd}, 1475 {0x26ce, 0x26ce}, 1476 {0x2702, 0x2702}, 1477 {0x2705, 0x2705}, 1478 {0x2708, 0x270d}, 1479 {0x270f, 0x270f}, 1480 {0x2712, 0x2712}, 1481 {0x2714, 0x2714}, 1482 {0x2716, 0x2716}, 1483 {0x271d, 0x271d}, 1484 {0x2721, 0x2721}, 1485 {0x2728, 0x2728}, 1486 {0x2733, 0x2734}, 1487 {0x2744, 0x2744}, 1488 {0x2747, 0x2747}, 1489 {0x274c, 0x274c}, 1490 {0x274e, 0x274e}, 1491 {0x2753, 0x2755}, 1492 {0x2763, 0x2764}, 1493 {0x2795, 0x2797}, 1494 {0x27a1, 0x27a1}, 1495 {0x27b0, 0x27b0}, 1496 {0x27bf, 0x27bf}, 1497 {0x2934, 0x2935}, 1498 {0x2b05, 0x2b07}, 1499 {0x2b1b, 0x2b1c}, 1500 {0x2b50, 0x2b50}, 1501 {0x1f004, 0x1f004}, 1502 {0x1f0cf, 0x1f0cf}, 1503 {0x1f1e6, 0x1f1ff}, 1504 {0x1f300, 0x1f320}, 1505 {0x1f330, 0x1f335}, 1506 {0x1f337, 0x1f37c}, 1507 {0x1f380, 0x1f393}, 1508 {0x1f3a0, 0x1f3c4}, 1509 {0x1f3c6, 0x1f3ca}, 1510 {0x1f3e0, 0x1f3f0}, 1511 {0x1f400, 0x1f43e}, 1512 {0x1f440, 0x1f440}, 1513 {0x1f442, 0x1f4f7}, 1514 {0x1f4f9, 0x1f4fc}, 1515 {0x1f500, 0x1f53d}, 1516 {0x1f550, 0x1f567}, 1517 {0x1f5fb, 0x1f640}, 1518 {0x1f645, 0x1f64f}, 1519 {0x1f680, 0x1f6c5} 1520 }; 1521 1522 if (c >= 0x100) 1523 { 1524 #ifdef USE_WCHAR_FUNCTIONS 1525 /* 1526 * Assume the library function wcwidth() works better than our own 1527 * stuff. It should return 1 for ambiguous width chars! 1528 */ 1529 int n = wcwidth(c); 1530 1531 if (n < 0) 1532 return 6; /* unprintable, displays <xxxx> */ 1533 if (n > 1) 1534 return n; 1535 #else 1536 if (!utf_printable(c)) 1537 return 6; /* unprintable, displays <xxxx> */ 1538 if (intable(doublewidth, sizeof(doublewidth), c)) 1539 return 2; 1540 #endif 1541 if (p_emoji && intable(emoji_width, sizeof(emoji_width), c)) 1542 return 2; 1543 } 1544 1545 /* Characters below 0x100 are influenced by 'isprint' option */ 1546 else if (c >= 0x80 && !vim_isprintc(c)) 1547 return 4; /* unprintable, displays <xx> */ 1548 1549 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c)) 1550 return 2; 1551 1552 return 1; 1553 } 1554 1555 /* 1556 * mb_ptr2cells() function pointer. 1557 * Return the number of display cells character at "*p" occupies. 1558 * This doesn't take care of unprintable characters, use ptr2cells() for that. 1559 */ 1560 int 1561 latin_ptr2cells(char_u *p UNUSED) 1562 { 1563 return 1; 1564 } 1565 1566 int 1567 utf_ptr2cells( 1568 char_u *p) 1569 { 1570 int c; 1571 1572 /* Need to convert to a wide character. */ 1573 if (*p >= 0x80) 1574 { 1575 c = utf_ptr2char(p); 1576 /* An illegal byte is displayed as <xx>. */ 1577 if (utf_ptr2len(p) == 1 || c == NUL) 1578 return 4; 1579 /* If the char is ASCII it must be an overlong sequence. */ 1580 if (c < 0x80) 1581 return char2cells(c); 1582 return utf_char2cells(c); 1583 } 1584 return 1; 1585 } 1586 1587 int 1588 dbcs_ptr2cells(char_u *p) 1589 { 1590 /* Number of cells is equal to number of bytes, except for euc-jp when 1591 * the first byte is 0x8e. */ 1592 if (enc_dbcs == DBCS_JPNU && *p == 0x8e) 1593 return 1; 1594 return MB_BYTE2LEN(*p); 1595 } 1596 1597 /* 1598 * mb_ptr2cells_len() function pointer. 1599 * Like mb_ptr2cells(), but limit string length to "size". 1600 * For an empty string or truncated character returns 1. 1601 */ 1602 int 1603 latin_ptr2cells_len(char_u *p UNUSED, int size UNUSED) 1604 { 1605 return 1; 1606 } 1607 1608 static int 1609 utf_ptr2cells_len(char_u *p, int size) 1610 { 1611 int c; 1612 1613 /* Need to convert to a wide character. */ 1614 if (size > 0 && *p >= 0x80) 1615 { 1616 if (utf_ptr2len_len(p, size) < utf8len_tab[*p]) 1617 return 1; /* truncated */ 1618 c = utf_ptr2char(p); 1619 /* An illegal byte is displayed as <xx>. */ 1620 if (utf_ptr2len(p) == 1 || c == NUL) 1621 return 4; 1622 /* If the char is ASCII it must be an overlong sequence. */ 1623 if (c < 0x80) 1624 return char2cells(c); 1625 return utf_char2cells(c); 1626 } 1627 return 1; 1628 } 1629 1630 static int 1631 dbcs_ptr2cells_len(char_u *p, int size) 1632 { 1633 /* Number of cells is equal to number of bytes, except for euc-jp when 1634 * the first byte is 0x8e. */ 1635 if (size <= 1 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)) 1636 return 1; 1637 return MB_BYTE2LEN(*p); 1638 } 1639 1640 /* 1641 * mb_char2cells() function pointer. 1642 * Return the number of display cells character "c" occupies. 1643 * Only takes care of multi-byte chars, not "^C" and such. 1644 */ 1645 int 1646 latin_char2cells(int c UNUSED) 1647 { 1648 return 1; 1649 } 1650 1651 static int 1652 dbcs_char2cells(int c) 1653 { 1654 /* Number of cells is equal to number of bytes, except for euc-jp when 1655 * the first byte is 0x8e. */ 1656 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e) 1657 return 1; 1658 /* use the first byte */ 1659 return MB_BYTE2LEN((unsigned)c >> 8); 1660 } 1661 1662 /* 1663 * Return the number of cells occupied by string "p". 1664 * Stop at a NUL character. When "len" >= 0 stop at character "p[len]". 1665 */ 1666 int 1667 mb_string2cells(char_u *p, int len) 1668 { 1669 int i; 1670 int clen = 0; 1671 1672 for (i = 0; (len < 0 || i < len) && p[i] != NUL; i += (*mb_ptr2len)(p + i)) 1673 clen += (*mb_ptr2cells)(p + i); 1674 return clen; 1675 } 1676 1677 /* 1678 * mb_off2cells() function pointer. 1679 * Return number of display cells for char at ScreenLines[off]. 1680 * We make sure that the offset used is less than "max_off". 1681 */ 1682 int 1683 latin_off2cells(unsigned off UNUSED, unsigned max_off UNUSED) 1684 { 1685 return 1; 1686 } 1687 1688 int 1689 dbcs_off2cells(unsigned off, unsigned max_off) 1690 { 1691 /* never check beyond end of the line */ 1692 if (off >= max_off) 1693 return 1; 1694 1695 /* Number of cells is equal to number of bytes, except for euc-jp when 1696 * the first byte is 0x8e. */ 1697 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e) 1698 return 1; 1699 return MB_BYTE2LEN(ScreenLines[off]); 1700 } 1701 1702 int 1703 utf_off2cells(unsigned off, unsigned max_off) 1704 { 1705 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1; 1706 } 1707 1708 /* 1709 * mb_ptr2char() function pointer. 1710 * Convert a byte sequence into a character. 1711 */ 1712 int 1713 latin_ptr2char(char_u *p) 1714 { 1715 return *p; 1716 } 1717 1718 static int 1719 dbcs_ptr2char(char_u *p) 1720 { 1721 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL) 1722 return (p[0] << 8) + p[1]; 1723 return *p; 1724 } 1725 1726 /* 1727 * Convert a UTF-8 byte sequence to a wide character. 1728 * If the sequence is illegal or truncated by a NUL the first byte is 1729 * returned. 1730 * Does not include composing characters, of course. 1731 */ 1732 int 1733 utf_ptr2char(char_u *p) 1734 { 1735 int len; 1736 1737 if (p[0] < 0x80) /* be quick for ASCII */ 1738 return p[0]; 1739 1740 len = utf8len_tab_zero[p[0]]; 1741 if (len > 1 && (p[1] & 0xc0) == 0x80) 1742 { 1743 if (len == 2) 1744 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); 1745 if ((p[2] & 0xc0) == 0x80) 1746 { 1747 if (len == 3) 1748 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) 1749 + (p[2] & 0x3f); 1750 if ((p[3] & 0xc0) == 0x80) 1751 { 1752 if (len == 4) 1753 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) 1754 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); 1755 if ((p[4] & 0xc0) == 0x80) 1756 { 1757 if (len == 5) 1758 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) 1759 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) 1760 + (p[4] & 0x3f); 1761 if ((p[5] & 0xc0) == 0x80 && len == 6) 1762 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) 1763 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) 1764 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); 1765 } 1766 } 1767 } 1768 } 1769 /* Illegal value, just return the first byte */ 1770 return p[0]; 1771 } 1772 1773 /* 1774 * Convert a UTF-8 byte sequence to a wide character. 1775 * String is assumed to be terminated by NUL or after "n" bytes, whichever 1776 * comes first. 1777 * The function is safe in the sense that it never accesses memory beyond the 1778 * first "n" bytes of "s". 1779 * 1780 * On success, returns decoded codepoint, advances "s" to the beginning of 1781 * next character and decreases "n" accordingly. 1782 * 1783 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past 1784 * NUL byte. 1785 * 1786 * If byte sequence is illegal or incomplete, returns -1 and does not advance 1787 * "s". 1788 */ 1789 static int 1790 utf_safe_read_char_adv(char_u **s, size_t *n) 1791 { 1792 int c, k; 1793 1794 if (*n == 0) /* end of buffer */ 1795 return 0; 1796 1797 k = utf8len_tab_zero[**s]; 1798 1799 if (k == 1) 1800 { 1801 /* ASCII character or NUL */ 1802 (*n)--; 1803 return *(*s)++; 1804 } 1805 1806 if ((size_t)k <= *n) 1807 { 1808 /* We have a multibyte sequence and it isn't truncated by buffer 1809 * limits so utf_ptr2char() is safe to use. Or the first byte is 1810 * illegal (k=0), and it's also safe to use utf_ptr2char(). */ 1811 c = utf_ptr2char(*s); 1812 1813 /* On failure, utf_ptr2char() returns the first byte, so here we 1814 * check equality with the first byte. The only non-ASCII character 1815 * which equals the first byte of its own UTF-8 representation is 1816 * U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too. 1817 * It's safe even if n=1, else we would have k=2 > n. */ 1818 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83)) 1819 { 1820 /* byte sequence was successfully decoded */ 1821 *s += k; 1822 *n -= k; 1823 return c; 1824 } 1825 } 1826 1827 /* byte sequence is incomplete or illegal */ 1828 return -1; 1829 } 1830 1831 /* 1832 * Get character at **pp and advance *pp to the next character. 1833 * Note: composing characters are skipped! 1834 */ 1835 int 1836 mb_ptr2char_adv(char_u **pp) 1837 { 1838 int c; 1839 1840 c = (*mb_ptr2char)(*pp); 1841 *pp += (*mb_ptr2len)(*pp); 1842 return c; 1843 } 1844 1845 /* 1846 * Get character at **pp and advance *pp to the next character. 1847 * Note: composing characters are returned as separate characters. 1848 */ 1849 int 1850 mb_cptr2char_adv(char_u **pp) 1851 { 1852 int c; 1853 1854 c = (*mb_ptr2char)(*pp); 1855 if (enc_utf8) 1856 *pp += utf_ptr2len(*pp); 1857 else 1858 *pp += (*mb_ptr2len)(*pp); 1859 return c; 1860 } 1861 1862 #if defined(FEAT_ARABIC) || defined(PROTO) 1863 /* 1864 * Check whether we are dealing with Arabic combining characters. 1865 * Note: these are NOT really composing characters! 1866 */ 1867 int 1868 arabic_combine( 1869 int one, /* first character */ 1870 int two) /* character just after "one" */ 1871 { 1872 if (one == a_LAM) 1873 return arabic_maycombine(two); 1874 return FALSE; 1875 } 1876 1877 /* 1878 * Check whether we are dealing with a character that could be regarded as an 1879 * Arabic combining character, need to check the character before this. 1880 */ 1881 int 1882 arabic_maycombine(int two) 1883 { 1884 if (p_arshape && !p_tbidi) 1885 return (two == a_ALEF_MADDA 1886 || two == a_ALEF_HAMZA_ABOVE 1887 || two == a_ALEF_HAMZA_BELOW 1888 || two == a_ALEF); 1889 return FALSE; 1890 } 1891 1892 /* 1893 * Check if the character pointed to by "p2" is a composing character when it 1894 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which 1895 * behaves like a composing character. 1896 */ 1897 int 1898 utf_composinglike(char_u *p1, char_u *p2) 1899 { 1900 int c2; 1901 1902 c2 = utf_ptr2char(p2); 1903 if (utf_iscomposing(c2)) 1904 return TRUE; 1905 if (!arabic_maycombine(c2)) 1906 return FALSE; 1907 return arabic_combine(utf_ptr2char(p1), c2); 1908 } 1909 #endif 1910 1911 /* 1912 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1913 * composing characters. 1914 */ 1915 int 1916 utfc_ptr2char( 1917 char_u *p, 1918 int *pcc) /* return: composing chars, last one is 0 */ 1919 { 1920 int len; 1921 int c; 1922 int cc; 1923 int i = 0; 1924 1925 c = utf_ptr2char(p); 1926 len = utf_ptr2len(p); 1927 1928 /* Only accept a composing char when the first char isn't illegal. */ 1929 if ((len > 1 || *p < 0x80) 1930 && p[len] >= 0x80 1931 && UTF_COMPOSINGLIKE(p, p + len)) 1932 { 1933 cc = utf_ptr2char(p + len); 1934 for (;;) 1935 { 1936 pcc[i++] = cc; 1937 if (i == MAX_MCO) 1938 break; 1939 len += utf_ptr2len(p + len); 1940 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1941 break; 1942 } 1943 } 1944 1945 if (i < MAX_MCO) /* last composing char must be 0 */ 1946 pcc[i] = 0; 1947 1948 return c; 1949 } 1950 1951 /* 1952 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1953 * composing characters. Use no more than p[maxlen]. 1954 */ 1955 int 1956 utfc_ptr2char_len( 1957 char_u *p, 1958 int *pcc, /* return: composing chars, last one is 0 */ 1959 int maxlen) 1960 { 1961 int len; 1962 int c; 1963 int cc; 1964 int i = 0; 1965 1966 c = utf_ptr2char(p); 1967 len = utf_ptr2len_len(p, maxlen); 1968 /* Only accept a composing char when the first char isn't illegal. */ 1969 if ((len > 1 || *p < 0x80) 1970 && len < maxlen 1971 && p[len] >= 0x80 1972 && UTF_COMPOSINGLIKE(p, p + len)) 1973 { 1974 cc = utf_ptr2char(p + len); 1975 for (;;) 1976 { 1977 pcc[i++] = cc; 1978 if (i == MAX_MCO) 1979 break; 1980 len += utf_ptr2len_len(p + len, maxlen - len); 1981 if (len >= maxlen 1982 || p[len] < 0x80 1983 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1984 break; 1985 } 1986 } 1987 1988 if (i < MAX_MCO) /* last composing char must be 0 */ 1989 pcc[i] = 0; 1990 1991 return c; 1992 } 1993 1994 /* 1995 * Convert the character at screen position "off" to a sequence of bytes. 1996 * Includes the composing characters. 1997 * "buf" must at least have the length MB_MAXBYTES + 1. 1998 * Only to be used when ScreenLinesUC[off] != 0. 1999 * Returns the produced number of bytes. 2000 */ 2001 int 2002 utfc_char2bytes(int off, char_u *buf) 2003 { 2004 int len; 2005 int i; 2006 2007 len = utf_char2bytes(ScreenLinesUC[off], buf); 2008 for (i = 0; i < Screen_mco; ++i) 2009 { 2010 if (ScreenLinesC[i][off] == 0) 2011 break; 2012 len += utf_char2bytes(ScreenLinesC[i][off], buf + len); 2013 } 2014 return len; 2015 } 2016 2017 /* 2018 * Get the length of a UTF-8 byte sequence, not including any following 2019 * composing characters. 2020 * Returns 0 for "". 2021 * Returns 1 for an illegal byte sequence. 2022 */ 2023 int 2024 utf_ptr2len(char_u *p) 2025 { 2026 int len; 2027 int i; 2028 2029 if (*p == NUL) 2030 return 0; 2031 len = utf8len_tab[*p]; 2032 for (i = 1; i < len; ++i) 2033 if ((p[i] & 0xc0) != 0x80) 2034 return 1; 2035 return len; 2036 } 2037 2038 /* 2039 * Return length of UTF-8 character, obtained from the first byte. 2040 * "b" must be between 0 and 255! 2041 * Returns 1 for an invalid first byte value. 2042 */ 2043 int 2044 utf_byte2len(int b) 2045 { 2046 return utf8len_tab[b]; 2047 } 2048 2049 /* 2050 * Get the length of UTF-8 byte sequence "p[size]". Does not include any 2051 * following composing characters. 2052 * Returns 1 for "". 2053 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.). 2054 * Returns number > "size" for an incomplete byte sequence. 2055 * Never returns zero. 2056 */ 2057 int 2058 utf_ptr2len_len(char_u *p, int size) 2059 { 2060 int len; 2061 int i; 2062 int m; 2063 2064 len = utf8len_tab[*p]; 2065 if (len == 1) 2066 return 1; /* NUL, ascii or illegal lead byte */ 2067 if (len > size) 2068 m = size; /* incomplete byte sequence. */ 2069 else 2070 m = len; 2071 for (i = 1; i < m; ++i) 2072 if ((p[i] & 0xc0) != 0x80) 2073 return 1; 2074 return len; 2075 } 2076 2077 /* 2078 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. 2079 * This includes following composing characters. 2080 */ 2081 int 2082 utfc_ptr2len(char_u *p) 2083 { 2084 int len; 2085 int b0 = *p; 2086 #ifdef FEAT_ARABIC 2087 int prevlen; 2088 #endif 2089 2090 if (b0 == NUL) 2091 return 0; 2092 if (b0 < 0x80 && p[1] < 0x80) /* be quick for ASCII */ 2093 return 1; 2094 2095 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2096 len = utf_ptr2len(p); 2097 2098 /* Check for illegal byte. */ 2099 if (len == 1 && b0 >= 0x80) 2100 return 1; 2101 2102 /* 2103 * Check for composing characters. We can handle only the first six, but 2104 * skip all of them (otherwise the cursor would get stuck). 2105 */ 2106 #ifdef FEAT_ARABIC 2107 prevlen = 0; 2108 #endif 2109 for (;;) 2110 { 2111 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2112 return len; 2113 2114 /* Skip over composing char */ 2115 #ifdef FEAT_ARABIC 2116 prevlen = len; 2117 #endif 2118 len += utf_ptr2len(p + len); 2119 } 2120 } 2121 2122 /* 2123 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" 2124 * takes. This includes following composing characters. 2125 * Returns 0 for an empty string. 2126 * Returns 1 for an illegal char or an incomplete byte sequence. 2127 */ 2128 int 2129 utfc_ptr2len_len(char_u *p, int size) 2130 { 2131 int len; 2132 #ifdef FEAT_ARABIC 2133 int prevlen; 2134 #endif 2135 2136 if (size < 1 || *p == NUL) 2137 return 0; 2138 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) /* be quick for ASCII */ 2139 return 1; 2140 2141 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2142 len = utf_ptr2len_len(p, size); 2143 2144 /* Check for illegal byte and incomplete byte sequence. */ 2145 if ((len == 1 && p[0] >= 0x80) || len > size) 2146 return 1; 2147 2148 /* 2149 * Check for composing characters. We can handle only the first six, but 2150 * skip all of them (otherwise the cursor would get stuck). 2151 */ 2152 #ifdef FEAT_ARABIC 2153 prevlen = 0; 2154 #endif 2155 while (len < size) 2156 { 2157 int len_next_char; 2158 2159 if (p[len] < 0x80) 2160 break; 2161 2162 /* 2163 * Next character length should not go beyond size to ensure that 2164 * UTF_COMPOSINGLIKE(...) does not read beyond size. 2165 */ 2166 len_next_char = utf_ptr2len_len(p + len, size - len); 2167 if (len_next_char > size - len) 2168 break; 2169 2170 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2171 break; 2172 2173 /* Skip over composing char */ 2174 #ifdef FEAT_ARABIC 2175 prevlen = len; 2176 #endif 2177 len += len_next_char; 2178 } 2179 return len; 2180 } 2181 2182 /* 2183 * Return the number of bytes the UTF-8 encoding of character "c" takes. 2184 * This does not include composing characters. 2185 */ 2186 int 2187 utf_char2len(int c) 2188 { 2189 if (c < 0x80) 2190 return 1; 2191 if (c < 0x800) 2192 return 2; 2193 if (c < 0x10000) 2194 return 3; 2195 if (c < 0x200000) 2196 return 4; 2197 if (c < 0x4000000) 2198 return 5; 2199 return 6; 2200 } 2201 2202 /* 2203 * Convert Unicode character "c" to UTF-8 string in "buf[]". 2204 * Returns the number of bytes. 2205 * This does not include composing characters. 2206 */ 2207 int 2208 utf_char2bytes(int c, char_u *buf) 2209 { 2210 if (c < 0x80) /* 7 bits */ 2211 { 2212 buf[0] = c; 2213 return 1; 2214 } 2215 if (c < 0x800) /* 11 bits */ 2216 { 2217 buf[0] = 0xc0 + ((unsigned)c >> 6); 2218 buf[1] = 0x80 + (c & 0x3f); 2219 return 2; 2220 } 2221 if (c < 0x10000) /* 16 bits */ 2222 { 2223 buf[0] = 0xe0 + ((unsigned)c >> 12); 2224 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2225 buf[2] = 0x80 + (c & 0x3f); 2226 return 3; 2227 } 2228 if (c < 0x200000) /* 21 bits */ 2229 { 2230 buf[0] = 0xf0 + ((unsigned)c >> 18); 2231 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2232 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2233 buf[3] = 0x80 + (c & 0x3f); 2234 return 4; 2235 } 2236 if (c < 0x4000000) /* 26 bits */ 2237 { 2238 buf[0] = 0xf8 + ((unsigned)c >> 24); 2239 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2240 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2241 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2242 buf[4] = 0x80 + (c & 0x3f); 2243 return 5; 2244 } 2245 /* 31 bits */ 2246 buf[0] = 0xfc + ((unsigned)c >> 30); 2247 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f); 2248 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2249 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2250 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2251 buf[5] = 0x80 + (c & 0x3f); 2252 return 6; 2253 } 2254 2255 /* 2256 * Return TRUE if "c" is a composing UTF-8 character. This means it will be 2257 * drawn on top of the preceding character. 2258 * Based on code from Markus Kuhn. 2259 */ 2260 int 2261 utf_iscomposing(int c) 2262 { 2263 /* Sorted list of non-overlapping intervals. 2264 * Generated by ../runtime/tools/unicode.vim. */ 2265 static struct interval combining[] = 2266 { 2267 {0x0300, 0x036f}, 2268 {0x0483, 0x0489}, 2269 {0x0591, 0x05bd}, 2270 {0x05bf, 0x05bf}, 2271 {0x05c1, 0x05c2}, 2272 {0x05c4, 0x05c5}, 2273 {0x05c7, 0x05c7}, 2274 {0x0610, 0x061a}, 2275 {0x064b, 0x065f}, 2276 {0x0670, 0x0670}, 2277 {0x06d6, 0x06dc}, 2278 {0x06df, 0x06e4}, 2279 {0x06e7, 0x06e8}, 2280 {0x06ea, 0x06ed}, 2281 {0x0711, 0x0711}, 2282 {0x0730, 0x074a}, 2283 {0x07a6, 0x07b0}, 2284 {0x07eb, 0x07f3}, 2285 {0x0816, 0x0819}, 2286 {0x081b, 0x0823}, 2287 {0x0825, 0x0827}, 2288 {0x0829, 0x082d}, 2289 {0x0859, 0x085b}, 2290 {0x08e3, 0x0903}, 2291 {0x093a, 0x093c}, 2292 {0x093e, 0x094f}, 2293 {0x0951, 0x0957}, 2294 {0x0962, 0x0963}, 2295 {0x0981, 0x0983}, 2296 {0x09bc, 0x09bc}, 2297 {0x09be, 0x09c4}, 2298 {0x09c7, 0x09c8}, 2299 {0x09cb, 0x09cd}, 2300 {0x09d7, 0x09d7}, 2301 {0x09e2, 0x09e3}, 2302 {0x0a01, 0x0a03}, 2303 {0x0a3c, 0x0a3c}, 2304 {0x0a3e, 0x0a42}, 2305 {0x0a47, 0x0a48}, 2306 {0x0a4b, 0x0a4d}, 2307 {0x0a51, 0x0a51}, 2308 {0x0a70, 0x0a71}, 2309 {0x0a75, 0x0a75}, 2310 {0x0a81, 0x0a83}, 2311 {0x0abc, 0x0abc}, 2312 {0x0abe, 0x0ac5}, 2313 {0x0ac7, 0x0ac9}, 2314 {0x0acb, 0x0acd}, 2315 {0x0ae2, 0x0ae3}, 2316 {0x0b01, 0x0b03}, 2317 {0x0b3c, 0x0b3c}, 2318 {0x0b3e, 0x0b44}, 2319 {0x0b47, 0x0b48}, 2320 {0x0b4b, 0x0b4d}, 2321 {0x0b56, 0x0b57}, 2322 {0x0b62, 0x0b63}, 2323 {0x0b82, 0x0b82}, 2324 {0x0bbe, 0x0bc2}, 2325 {0x0bc6, 0x0bc8}, 2326 {0x0bca, 0x0bcd}, 2327 {0x0bd7, 0x0bd7}, 2328 {0x0c00, 0x0c03}, 2329 {0x0c3e, 0x0c44}, 2330 {0x0c46, 0x0c48}, 2331 {0x0c4a, 0x0c4d}, 2332 {0x0c55, 0x0c56}, 2333 {0x0c62, 0x0c63}, 2334 {0x0c81, 0x0c83}, 2335 {0x0cbc, 0x0cbc}, 2336 {0x0cbe, 0x0cc4}, 2337 {0x0cc6, 0x0cc8}, 2338 {0x0cca, 0x0ccd}, 2339 {0x0cd5, 0x0cd6}, 2340 {0x0ce2, 0x0ce3}, 2341 {0x0d01, 0x0d03}, 2342 {0x0d3e, 0x0d44}, 2343 {0x0d46, 0x0d48}, 2344 {0x0d4a, 0x0d4d}, 2345 {0x0d57, 0x0d57}, 2346 {0x0d62, 0x0d63}, 2347 {0x0d82, 0x0d83}, 2348 {0x0dca, 0x0dca}, 2349 {0x0dcf, 0x0dd4}, 2350 {0x0dd6, 0x0dd6}, 2351 {0x0dd8, 0x0ddf}, 2352 {0x0df2, 0x0df3}, 2353 {0x0e31, 0x0e31}, 2354 {0x0e34, 0x0e3a}, 2355 {0x0e47, 0x0e4e}, 2356 {0x0eb1, 0x0eb1}, 2357 {0x0eb4, 0x0eb9}, 2358 {0x0ebb, 0x0ebc}, 2359 {0x0ec8, 0x0ecd}, 2360 {0x0f18, 0x0f19}, 2361 {0x0f35, 0x0f35}, 2362 {0x0f37, 0x0f37}, 2363 {0x0f39, 0x0f39}, 2364 {0x0f3e, 0x0f3f}, 2365 {0x0f71, 0x0f84}, 2366 {0x0f86, 0x0f87}, 2367 {0x0f8d, 0x0f97}, 2368 {0x0f99, 0x0fbc}, 2369 {0x0fc6, 0x0fc6}, 2370 {0x102b, 0x103e}, 2371 {0x1056, 0x1059}, 2372 {0x105e, 0x1060}, 2373 {0x1062, 0x1064}, 2374 {0x1067, 0x106d}, 2375 {0x1071, 0x1074}, 2376 {0x1082, 0x108d}, 2377 {0x108f, 0x108f}, 2378 {0x109a, 0x109d}, 2379 {0x135d, 0x135f}, 2380 {0x1712, 0x1714}, 2381 {0x1732, 0x1734}, 2382 {0x1752, 0x1753}, 2383 {0x1772, 0x1773}, 2384 {0x17b4, 0x17d3}, 2385 {0x17dd, 0x17dd}, 2386 {0x180b, 0x180d}, 2387 {0x18a9, 0x18a9}, 2388 {0x1920, 0x192b}, 2389 {0x1930, 0x193b}, 2390 {0x1a17, 0x1a1b}, 2391 {0x1a55, 0x1a5e}, 2392 {0x1a60, 0x1a7c}, 2393 {0x1a7f, 0x1a7f}, 2394 {0x1ab0, 0x1abe}, 2395 {0x1b00, 0x1b04}, 2396 {0x1b34, 0x1b44}, 2397 {0x1b6b, 0x1b73}, 2398 {0x1b80, 0x1b82}, 2399 {0x1ba1, 0x1bad}, 2400 {0x1be6, 0x1bf3}, 2401 {0x1c24, 0x1c37}, 2402 {0x1cd0, 0x1cd2}, 2403 {0x1cd4, 0x1ce8}, 2404 {0x1ced, 0x1ced}, 2405 {0x1cf2, 0x1cf4}, 2406 {0x1cf8, 0x1cf9}, 2407 {0x1dc0, 0x1df5}, 2408 {0x1dfc, 0x1dff}, 2409 {0x20d0, 0x20f0}, 2410 {0x2cef, 0x2cf1}, 2411 {0x2d7f, 0x2d7f}, 2412 {0x2de0, 0x2dff}, 2413 {0x302a, 0x302f}, 2414 {0x3099, 0x309a}, 2415 {0xa66f, 0xa672}, 2416 {0xa674, 0xa67d}, 2417 {0xa69e, 0xa69f}, 2418 {0xa6f0, 0xa6f1}, 2419 {0xa802, 0xa802}, 2420 {0xa806, 0xa806}, 2421 {0xa80b, 0xa80b}, 2422 {0xa823, 0xa827}, 2423 {0xa880, 0xa881}, 2424 {0xa8b4, 0xa8c4}, 2425 {0xa8e0, 0xa8f1}, 2426 {0xa926, 0xa92d}, 2427 {0xa947, 0xa953}, 2428 {0xa980, 0xa983}, 2429 {0xa9b3, 0xa9c0}, 2430 {0xa9e5, 0xa9e5}, 2431 {0xaa29, 0xaa36}, 2432 {0xaa43, 0xaa43}, 2433 {0xaa4c, 0xaa4d}, 2434 {0xaa7b, 0xaa7d}, 2435 {0xaab0, 0xaab0}, 2436 {0xaab2, 0xaab4}, 2437 {0xaab7, 0xaab8}, 2438 {0xaabe, 0xaabf}, 2439 {0xaac1, 0xaac1}, 2440 {0xaaeb, 0xaaef}, 2441 {0xaaf5, 0xaaf6}, 2442 {0xabe3, 0xabea}, 2443 {0xabec, 0xabed}, 2444 {0xfb1e, 0xfb1e}, 2445 {0xfe00, 0xfe0f}, 2446 {0xfe20, 0xfe2f}, 2447 {0x101fd, 0x101fd}, 2448 {0x102e0, 0x102e0}, 2449 {0x10376, 0x1037a}, 2450 {0x10a01, 0x10a03}, 2451 {0x10a05, 0x10a06}, 2452 {0x10a0c, 0x10a0f}, 2453 {0x10a38, 0x10a3a}, 2454 {0x10a3f, 0x10a3f}, 2455 {0x10ae5, 0x10ae6}, 2456 {0x11000, 0x11002}, 2457 {0x11038, 0x11046}, 2458 {0x1107f, 0x11082}, 2459 {0x110b0, 0x110ba}, 2460 {0x11100, 0x11102}, 2461 {0x11127, 0x11134}, 2462 {0x11173, 0x11173}, 2463 {0x11180, 0x11182}, 2464 {0x111b3, 0x111c0}, 2465 {0x111ca, 0x111cc}, 2466 {0x1122c, 0x11237}, 2467 {0x112df, 0x112ea}, 2468 {0x11300, 0x11303}, 2469 {0x1133c, 0x1133c}, 2470 {0x1133e, 0x11344}, 2471 {0x11347, 0x11348}, 2472 {0x1134b, 0x1134d}, 2473 {0x11357, 0x11357}, 2474 {0x11362, 0x11363}, 2475 {0x11366, 0x1136c}, 2476 {0x11370, 0x11374}, 2477 {0x114b0, 0x114c3}, 2478 {0x115af, 0x115b5}, 2479 {0x115b8, 0x115c0}, 2480 {0x115dc, 0x115dd}, 2481 {0x11630, 0x11640}, 2482 {0x116ab, 0x116b7}, 2483 {0x1171d, 0x1172b}, 2484 {0x16af0, 0x16af4}, 2485 {0x16b30, 0x16b36}, 2486 {0x16f51, 0x16f7e}, 2487 {0x16f8f, 0x16f92}, 2488 {0x1bc9d, 0x1bc9e}, 2489 {0x1d165, 0x1d169}, 2490 {0x1d16d, 0x1d172}, 2491 {0x1d17b, 0x1d182}, 2492 {0x1d185, 0x1d18b}, 2493 {0x1d1aa, 0x1d1ad}, 2494 {0x1d242, 0x1d244}, 2495 {0x1da00, 0x1da36}, 2496 {0x1da3b, 0x1da6c}, 2497 {0x1da75, 0x1da75}, 2498 {0x1da84, 0x1da84}, 2499 {0x1da9b, 0x1da9f}, 2500 {0x1daa1, 0x1daaf}, 2501 {0x1e8d0, 0x1e8d6}, 2502 {0xe0100, 0xe01ef} 2503 }; 2504 2505 return intable(combining, sizeof(combining), c); 2506 } 2507 2508 /* 2509 * Return TRUE for characters that can be displayed in a normal way. 2510 * Only for characters of 0x100 and above! 2511 */ 2512 int 2513 utf_printable(int c) 2514 { 2515 #ifdef USE_WCHAR_FUNCTIONS 2516 /* 2517 * Assume the iswprint() library function works better than our own stuff. 2518 */ 2519 return iswprint(c); 2520 #else 2521 /* Sorted list of non-overlapping intervals. 2522 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */ 2523 static struct interval nonprint[] = 2524 { 2525 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2526 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2527 {0xfffe, 0xffff} 2528 }; 2529 2530 return !intable(nonprint, sizeof(nonprint), c); 2531 #endif 2532 } 2533 2534 /* 2535 * Get class of a Unicode character. 2536 * 0: white space 2537 * 1: punctuation 2538 * 2 or bigger: some class of word character. 2539 */ 2540 int 2541 utf_class(int c) 2542 { 2543 /* sorted list of non-overlapping intervals */ 2544 static struct clinterval 2545 { 2546 unsigned int first; 2547 unsigned int last; 2548 unsigned int class; 2549 } classes[] = 2550 { 2551 {0x037e, 0x037e, 1}, /* Greek question mark */ 2552 {0x0387, 0x0387, 1}, /* Greek ano teleia */ 2553 {0x055a, 0x055f, 1}, /* Armenian punctuation */ 2554 {0x0589, 0x0589, 1}, /* Armenian full stop */ 2555 {0x05be, 0x05be, 1}, 2556 {0x05c0, 0x05c0, 1}, 2557 {0x05c3, 0x05c3, 1}, 2558 {0x05f3, 0x05f4, 1}, 2559 {0x060c, 0x060c, 1}, 2560 {0x061b, 0x061b, 1}, 2561 {0x061f, 0x061f, 1}, 2562 {0x066a, 0x066d, 1}, 2563 {0x06d4, 0x06d4, 1}, 2564 {0x0700, 0x070d, 1}, /* Syriac punctuation */ 2565 {0x0964, 0x0965, 1}, 2566 {0x0970, 0x0970, 1}, 2567 {0x0df4, 0x0df4, 1}, 2568 {0x0e4f, 0x0e4f, 1}, 2569 {0x0e5a, 0x0e5b, 1}, 2570 {0x0f04, 0x0f12, 1}, 2571 {0x0f3a, 0x0f3d, 1}, 2572 {0x0f85, 0x0f85, 1}, 2573 {0x104a, 0x104f, 1}, /* Myanmar punctuation */ 2574 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */ 2575 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */ 2576 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */ 2577 {0x1680, 0x1680, 0}, 2578 {0x169b, 0x169c, 1}, 2579 {0x16eb, 0x16ed, 1}, 2580 {0x1735, 0x1736, 1}, 2581 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */ 2582 {0x1800, 0x180a, 1}, /* Mongolian punctuation */ 2583 {0x2000, 0x200b, 0}, /* spaces */ 2584 {0x200c, 0x2027, 1}, /* punctuation and symbols */ 2585 {0x2028, 0x2029, 0}, 2586 {0x202a, 0x202e, 1}, /* punctuation and symbols */ 2587 {0x202f, 0x202f, 0}, 2588 {0x2030, 0x205e, 1}, /* punctuation and symbols */ 2589 {0x205f, 0x205f, 0}, 2590 {0x2060, 0x27ff, 1}, /* punctuation and symbols */ 2591 {0x2070, 0x207f, 0x2070}, /* superscript */ 2592 {0x2080, 0x2094, 0x2080}, /* subscript */ 2593 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */ 2594 {0x2800, 0x28ff, 0x2800}, /* braille */ 2595 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */ 2596 {0x29d8, 0x29db, 1}, 2597 {0x29fc, 0x29fd, 1}, 2598 {0x2e00, 0x2e7f, 1}, /* supplemental punctuation */ 2599 {0x3000, 0x3000, 0}, /* ideographic space */ 2600 {0x3001, 0x3020, 1}, /* ideographic punctuation */ 2601 {0x3030, 0x3030, 1}, 2602 {0x303d, 0x303d, 1}, 2603 {0x3040, 0x309f, 0x3040}, /* Hiragana */ 2604 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */ 2605 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */ 2606 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */ 2607 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */ 2608 {0xfd3e, 0xfd3f, 1}, 2609 {0xfe30, 0xfe6b, 1}, /* punctuation forms */ 2610 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */ 2611 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */ 2612 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */ 2613 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */ 2614 {0x20000, 0x2a6df, 0x4e00}, /* CJK Ideographs */ 2615 {0x2a700, 0x2b73f, 0x4e00}, /* CJK Ideographs */ 2616 {0x2b740, 0x2b81f, 0x4e00}, /* CJK Ideographs */ 2617 {0x2f800, 0x2fa1f, 0x4e00}, /* CJK Ideographs */ 2618 }; 2619 2620 /* Sorted list of non-overlapping intervals of all Emoji characters, 2621 * based on http://unicode.org/emoji/charts/emoji-list.html */ 2622 static struct interval emoji_all[] = 2623 { 2624 {0x203c, 0x203c}, 2625 {0x2049, 0x2049}, 2626 {0x2122, 0x2122}, 2627 {0x2139, 0x2139}, 2628 {0x2194, 0x2199}, 2629 {0x21a9, 0x21aa}, 2630 {0x231a, 0x231b}, 2631 {0x2328, 0x2328}, 2632 {0x23cf, 0x23cf}, 2633 {0x23e9, 0x23f3}, 2634 {0x24c2, 0x24c2}, 2635 {0x25aa, 0x25ab}, 2636 {0x25b6, 0x25b6}, 2637 {0x25c0, 0x25c0}, 2638 {0x25fb, 0x25fe}, 2639 {0x2600, 0x2604}, 2640 {0x260e, 0x260e}, 2641 {0x2611, 0x2611}, 2642 {0x2614, 0x2615}, 2643 {0x2618, 0x2618}, 2644 {0x261d, 0x261d}, 2645 {0x2620, 0x2620}, 2646 {0x2622, 0x2623}, 2647 {0x2626, 0x2626}, 2648 {0x262a, 0x262a}, 2649 {0x262e, 0x262f}, 2650 {0x2638, 0x263a}, 2651 {0x2648, 0x2653}, 2652 {0x2660, 0x2660}, 2653 {0x2663, 0x2663}, 2654 {0x2665, 0x2666}, 2655 {0x2668, 0x2668}, 2656 {0x267b, 0x267b}, 2657 {0x267f, 0x267f}, 2658 {0x2692, 0x2694}, 2659 {0x2696, 0x2697}, 2660 {0x2699, 0x2699}, 2661 {0x269b, 0x269c}, 2662 {0x26a0, 0x26a1}, 2663 {0x26aa, 0x26ab}, 2664 {0x26b0, 0x26b1}, 2665 {0x26bd, 0x26be}, 2666 {0x26c4, 0x26c5}, 2667 {0x26c8, 0x26c8}, 2668 {0x26ce, 0x26cf}, 2669 {0x26d1, 0x26d1}, 2670 {0x26d3, 0x26d4}, 2671 {0x26e9, 0x26ea}, 2672 {0x26f0, 0x26f5}, 2673 {0x26f7, 0x26fa}, 2674 {0x26fd, 0x26fd}, 2675 {0x2702, 0x2702}, 2676 {0x2705, 0x2705}, 2677 {0x2708, 0x270d}, 2678 {0x270f, 0x270f}, 2679 {0x2712, 0x2712}, 2680 {0x2714, 0x2714}, 2681 {0x2716, 0x2716}, 2682 {0x271d, 0x271d}, 2683 {0x2721, 0x2721}, 2684 {0x2728, 0x2728}, 2685 {0x2733, 0x2734}, 2686 {0x2744, 0x2744}, 2687 {0x2747, 0x2747}, 2688 {0x274c, 0x274c}, 2689 {0x274e, 0x274e}, 2690 {0x2753, 0x2755}, 2691 {0x2757, 0x2757}, 2692 {0x2763, 0x2764}, 2693 {0x2795, 0x2797}, 2694 {0x27a1, 0x27a1}, 2695 {0x27b0, 0x27b0}, 2696 {0x27bf, 0x27bf}, 2697 {0x2934, 0x2935}, 2698 {0x2b05, 0x2b07}, 2699 {0x2b1b, 0x2b1c}, 2700 {0x2b50, 0x2b50}, 2701 {0x2b55, 0x2b55}, 2702 {0x3030, 0x3030}, 2703 {0x303d, 0x303d}, 2704 {0x3297, 0x3297}, 2705 {0x3299, 0x3299}, 2706 {0x1f004, 0x1f004}, 2707 {0x1f0cf, 0x1f0cf}, 2708 {0x1f170, 0x1f171}, 2709 {0x1f17e, 0x1f17f}, 2710 {0x1f18e, 0x1f18e}, 2711 {0x1f191, 0x1f19a}, 2712 {0x1f1e6, 0x1f1ff}, 2713 {0x1f201, 0x1f202}, 2714 {0x1f21a, 0x1f21a}, 2715 {0x1f22f, 0x1f22f}, 2716 {0x1f232, 0x1f23a}, 2717 {0x1f250, 0x1f251}, 2718 {0x1f300, 0x1f320}, 2719 {0x1f330, 0x1f335}, 2720 {0x1f337, 0x1f37c}, 2721 {0x1f380, 0x1f393}, 2722 {0x1f3a0, 0x1f3c4}, 2723 {0x1f3c6, 0x1f3ca}, 2724 {0x1f3e0, 0x1f3f0}, 2725 {0x1f400, 0x1f43e}, 2726 {0x1f440, 0x1f440}, 2727 {0x1f442, 0x1f4f7}, 2728 {0x1f4f9, 0x1f4fc}, 2729 {0x1f500, 0x1f53d}, 2730 {0x1f550, 0x1f567}, 2731 {0x1f5fb, 0x1f640}, 2732 {0x1f645, 0x1f64f}, 2733 {0x1f680, 0x1f6c5} 2734 }; 2735 2736 int bot = 0; 2737 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2738 int mid; 2739 2740 /* First quick check for Latin1 characters, use 'iskeyword'. */ 2741 if (c < 0x100) 2742 { 2743 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2744 return 0; /* blank */ 2745 if (vim_iswordc(c)) 2746 return 2; /* word character */ 2747 return 1; /* punctuation */ 2748 } 2749 2750 /* binary search in table */ 2751 while (top >= bot) 2752 { 2753 mid = (bot + top) / 2; 2754 if (classes[mid].last < (unsigned int)c) 2755 bot = mid + 1; 2756 else if (classes[mid].first > (unsigned int)c) 2757 top = mid - 1; 2758 else 2759 return (int)classes[mid].class; 2760 } 2761 2762 /* emoji */ 2763 if (intable(emoji_all, sizeof(emoji_all), c)) 2764 return 3; 2765 2766 /* most other characters are "word" characters */ 2767 return 2; 2768 } 2769 2770 /* 2771 * Code for Unicode case-dependent operations. Based on notes in 2772 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2773 * This code uses simple case folding, not full case folding. 2774 * Last updated for Unicode 5.2. 2775 */ 2776 2777 /* 2778 * The following tables are built by ../runtime/tools/unicode.vim. 2779 * They must be in numeric order, because we use binary search. 2780 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2781 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2782 * folded/upper/lower by adding 32. 2783 */ 2784 typedef struct 2785 { 2786 int rangeStart; 2787 int rangeEnd; 2788 int step; 2789 int offset; 2790 } convertStruct; 2791 2792 static convertStruct foldCase[] = 2793 { 2794 {0x41,0x5a,1,32}, 2795 {0xb5,0xb5,-1,775}, 2796 {0xc0,0xd6,1,32}, 2797 {0xd8,0xde,1,32}, 2798 {0x100,0x12e,2,1}, 2799 {0x132,0x136,2,1}, 2800 {0x139,0x147,2,1}, 2801 {0x14a,0x176,2,1}, 2802 {0x178,0x178,-1,-121}, 2803 {0x179,0x17d,2,1}, 2804 {0x17f,0x17f,-1,-268}, 2805 {0x181,0x181,-1,210}, 2806 {0x182,0x184,2,1}, 2807 {0x186,0x186,-1,206}, 2808 {0x187,0x187,-1,1}, 2809 {0x189,0x18a,1,205}, 2810 {0x18b,0x18b,-1,1}, 2811 {0x18e,0x18e,-1,79}, 2812 {0x18f,0x18f,-1,202}, 2813 {0x190,0x190,-1,203}, 2814 {0x191,0x191,-1,1}, 2815 {0x193,0x193,-1,205}, 2816 {0x194,0x194,-1,207}, 2817 {0x196,0x196,-1,211}, 2818 {0x197,0x197,-1,209}, 2819 {0x198,0x198,-1,1}, 2820 {0x19c,0x19c,-1,211}, 2821 {0x19d,0x19d,-1,213}, 2822 {0x19f,0x19f,-1,214}, 2823 {0x1a0,0x1a4,2,1}, 2824 {0x1a6,0x1a6,-1,218}, 2825 {0x1a7,0x1a7,-1,1}, 2826 {0x1a9,0x1a9,-1,218}, 2827 {0x1ac,0x1ac,-1,1}, 2828 {0x1ae,0x1ae,-1,218}, 2829 {0x1af,0x1af,-1,1}, 2830 {0x1b1,0x1b2,1,217}, 2831 {0x1b3,0x1b5,2,1}, 2832 {0x1b7,0x1b7,-1,219}, 2833 {0x1b8,0x1bc,4,1}, 2834 {0x1c4,0x1c4,-1,2}, 2835 {0x1c5,0x1c5,-1,1}, 2836 {0x1c7,0x1c7,-1,2}, 2837 {0x1c8,0x1c8,-1,1}, 2838 {0x1ca,0x1ca,-1,2}, 2839 {0x1cb,0x1db,2,1}, 2840 {0x1de,0x1ee,2,1}, 2841 {0x1f1,0x1f1,-1,2}, 2842 {0x1f2,0x1f4,2,1}, 2843 {0x1f6,0x1f6,-1,-97}, 2844 {0x1f7,0x1f7,-1,-56}, 2845 {0x1f8,0x21e,2,1}, 2846 {0x220,0x220,-1,-130}, 2847 {0x222,0x232,2,1}, 2848 {0x23a,0x23a,-1,10795}, 2849 {0x23b,0x23b,-1,1}, 2850 {0x23d,0x23d,-1,-163}, 2851 {0x23e,0x23e,-1,10792}, 2852 {0x241,0x241,-1,1}, 2853 {0x243,0x243,-1,-195}, 2854 {0x244,0x244,-1,69}, 2855 {0x245,0x245,-1,71}, 2856 {0x246,0x24e,2,1}, 2857 {0x345,0x345,-1,116}, 2858 {0x370,0x372,2,1}, 2859 {0x376,0x376,-1,1}, 2860 {0x37f,0x37f,-1,116}, 2861 {0x386,0x386,-1,38}, 2862 {0x388,0x38a,1,37}, 2863 {0x38c,0x38c,-1,64}, 2864 {0x38e,0x38f,1,63}, 2865 {0x391,0x3a1,1,32}, 2866 {0x3a3,0x3ab,1,32}, 2867 {0x3c2,0x3c2,-1,1}, 2868 {0x3cf,0x3cf,-1,8}, 2869 {0x3d0,0x3d0,-1,-30}, 2870 {0x3d1,0x3d1,-1,-25}, 2871 {0x3d5,0x3d5,-1,-15}, 2872 {0x3d6,0x3d6,-1,-22}, 2873 {0x3d8,0x3ee,2,1}, 2874 {0x3f0,0x3f0,-1,-54}, 2875 {0x3f1,0x3f1,-1,-48}, 2876 {0x3f4,0x3f4,-1,-60}, 2877 {0x3f5,0x3f5,-1,-64}, 2878 {0x3f7,0x3f7,-1,1}, 2879 {0x3f9,0x3f9,-1,-7}, 2880 {0x3fa,0x3fa,-1,1}, 2881 {0x3fd,0x3ff,1,-130}, 2882 {0x400,0x40f,1,80}, 2883 {0x410,0x42f,1,32}, 2884 {0x460,0x480,2,1}, 2885 {0x48a,0x4be,2,1}, 2886 {0x4c0,0x4c0,-1,15}, 2887 {0x4c1,0x4cd,2,1}, 2888 {0x4d0,0x52e,2,1}, 2889 {0x531,0x556,1,48}, 2890 {0x10a0,0x10c5,1,7264}, 2891 {0x10c7,0x10cd,6,7264}, 2892 {0x13f8,0x13fd,1,-8}, 2893 {0x1e00,0x1e94,2,1}, 2894 {0x1e9b,0x1e9b,-1,-58}, 2895 {0x1e9e,0x1e9e,-1,-7615}, 2896 {0x1ea0,0x1efe,2,1}, 2897 {0x1f08,0x1f0f,1,-8}, 2898 {0x1f18,0x1f1d,1,-8}, 2899 {0x1f28,0x1f2f,1,-8}, 2900 {0x1f38,0x1f3f,1,-8}, 2901 {0x1f48,0x1f4d,1,-8}, 2902 {0x1f59,0x1f5f,2,-8}, 2903 {0x1f68,0x1f6f,1,-8}, 2904 {0x1f88,0x1f8f,1,-8}, 2905 {0x1f98,0x1f9f,1,-8}, 2906 {0x1fa8,0x1faf,1,-8}, 2907 {0x1fb8,0x1fb9,1,-8}, 2908 {0x1fba,0x1fbb,1,-74}, 2909 {0x1fbc,0x1fbc,-1,-9}, 2910 {0x1fbe,0x1fbe,-1,-7173}, 2911 {0x1fc8,0x1fcb,1,-86}, 2912 {0x1fcc,0x1fcc,-1,-9}, 2913 {0x1fd8,0x1fd9,1,-8}, 2914 {0x1fda,0x1fdb,1,-100}, 2915 {0x1fe8,0x1fe9,1,-8}, 2916 {0x1fea,0x1feb,1,-112}, 2917 {0x1fec,0x1fec,-1,-7}, 2918 {0x1ff8,0x1ff9,1,-128}, 2919 {0x1ffa,0x1ffb,1,-126}, 2920 {0x1ffc,0x1ffc,-1,-9}, 2921 {0x2126,0x2126,-1,-7517}, 2922 {0x212a,0x212a,-1,-8383}, 2923 {0x212b,0x212b,-1,-8262}, 2924 {0x2132,0x2132,-1,28}, 2925 {0x2160,0x216f,1,16}, 2926 {0x2183,0x2183,-1,1}, 2927 {0x24b6,0x24cf,1,26}, 2928 {0x2c00,0x2c2e,1,48}, 2929 {0x2c60,0x2c60,-1,1}, 2930 {0x2c62,0x2c62,-1,-10743}, 2931 {0x2c63,0x2c63,-1,-3814}, 2932 {0x2c64,0x2c64,-1,-10727}, 2933 {0x2c67,0x2c6b,2,1}, 2934 {0x2c6d,0x2c6d,-1,-10780}, 2935 {0x2c6e,0x2c6e,-1,-10749}, 2936 {0x2c6f,0x2c6f,-1,-10783}, 2937 {0x2c70,0x2c70,-1,-10782}, 2938 {0x2c72,0x2c75,3,1}, 2939 {0x2c7e,0x2c7f,1,-10815}, 2940 {0x2c80,0x2ce2,2,1}, 2941 {0x2ceb,0x2ced,2,1}, 2942 {0x2cf2,0xa640,31054,1}, 2943 {0xa642,0xa66c,2,1}, 2944 {0xa680,0xa69a,2,1}, 2945 {0xa722,0xa72e,2,1}, 2946 {0xa732,0xa76e,2,1}, 2947 {0xa779,0xa77b,2,1}, 2948 {0xa77d,0xa77d,-1,-35332}, 2949 {0xa77e,0xa786,2,1}, 2950 {0xa78b,0xa78b,-1,1}, 2951 {0xa78d,0xa78d,-1,-42280}, 2952 {0xa790,0xa792,2,1}, 2953 {0xa796,0xa7a8,2,1}, 2954 {0xa7aa,0xa7aa,-1,-42308}, 2955 {0xa7ab,0xa7ab,-1,-42319}, 2956 {0xa7ac,0xa7ac,-1,-42315}, 2957 {0xa7ad,0xa7ad,-1,-42305}, 2958 {0xa7b0,0xa7b0,-1,-42258}, 2959 {0xa7b1,0xa7b1,-1,-42282}, 2960 {0xa7b2,0xa7b2,-1,-42261}, 2961 {0xa7b3,0xa7b3,-1,928}, 2962 {0xa7b4,0xa7b6,2,1}, 2963 {0xab70,0xabbf,1,-38864}, 2964 {0xff21,0xff3a,1,32}, 2965 {0x10400,0x10427,1,40}, 2966 {0x10c80,0x10cb2,1,64}, 2967 {0x118a0,0x118bf,1,32} 2968 }; 2969 2970 static int utf_convert(int a, convertStruct table[], int tableSize); 2971 static int utf_strnicmp(char_u *s1, char_u *s2, size_t n1, size_t n2); 2972 2973 /* 2974 * Generic conversion function for case operations. 2975 * Return the converted equivalent of "a", which is a UCS-4 character. Use 2976 * the given conversion "table". Uses binary search on "table". 2977 */ 2978 static int 2979 utf_convert( 2980 int a, 2981 convertStruct table[], 2982 int tableSize) 2983 { 2984 int start, mid, end; /* indices into table */ 2985 int entries = tableSize / sizeof(convertStruct); 2986 2987 start = 0; 2988 end = entries; 2989 while (start < end) 2990 { 2991 /* need to search further */ 2992 mid = (end + start) / 2; 2993 if (table[mid].rangeEnd < a) 2994 start = mid + 1; 2995 else 2996 end = mid; 2997 } 2998 if (start < entries 2999 && table[start].rangeStart <= a 3000 && a <= table[start].rangeEnd 3001 && (a - table[start].rangeStart) % table[start].step == 0) 3002 return (a + table[start].offset); 3003 else 3004 return a; 3005 } 3006 3007 /* 3008 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 3009 * simple case folding. 3010 */ 3011 int 3012 utf_fold(int a) 3013 { 3014 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 3015 } 3016 3017 static convertStruct toLower[] = 3018 { 3019 {0x41,0x5a,1,32}, 3020 {0xc0,0xd6,1,32}, 3021 {0xd8,0xde,1,32}, 3022 {0x100,0x12e,2,1}, 3023 {0x130,0x130,-1,-199}, 3024 {0x132,0x136,2,1}, 3025 {0x139,0x147,2,1}, 3026 {0x14a,0x176,2,1}, 3027 {0x178,0x178,-1,-121}, 3028 {0x179,0x17d,2,1}, 3029 {0x181,0x181,-1,210}, 3030 {0x182,0x184,2,1}, 3031 {0x186,0x186,-1,206}, 3032 {0x187,0x187,-1,1}, 3033 {0x189,0x18a,1,205}, 3034 {0x18b,0x18b,-1,1}, 3035 {0x18e,0x18e,-1,79}, 3036 {0x18f,0x18f,-1,202}, 3037 {0x190,0x190,-1,203}, 3038 {0x191,0x191,-1,1}, 3039 {0x193,0x193,-1,205}, 3040 {0x194,0x194,-1,207}, 3041 {0x196,0x196,-1,211}, 3042 {0x197,0x197,-1,209}, 3043 {0x198,0x198,-1,1}, 3044 {0x19c,0x19c,-1,211}, 3045 {0x19d,0x19d,-1,213}, 3046 {0x19f,0x19f,-1,214}, 3047 {0x1a0,0x1a4,2,1}, 3048 {0x1a6,0x1a6,-1,218}, 3049 {0x1a7,0x1a7,-1,1}, 3050 {0x1a9,0x1a9,-1,218}, 3051 {0x1ac,0x1ac,-1,1}, 3052 {0x1ae,0x1ae,-1,218}, 3053 {0x1af,0x1af,-1,1}, 3054 {0x1b1,0x1b2,1,217}, 3055 {0x1b3,0x1b5,2,1}, 3056 {0x1b7,0x1b7,-1,219}, 3057 {0x1b8,0x1bc,4,1}, 3058 {0x1c4,0x1c4,-1,2}, 3059 {0x1c5,0x1c5,-1,1}, 3060 {0x1c7,0x1c7,-1,2}, 3061 {0x1c8,0x1c8,-1,1}, 3062 {0x1ca,0x1ca,-1,2}, 3063 {0x1cb,0x1db,2,1}, 3064 {0x1de,0x1ee,2,1}, 3065 {0x1f1,0x1f1,-1,2}, 3066 {0x1f2,0x1f4,2,1}, 3067 {0x1f6,0x1f6,-1,-97}, 3068 {0x1f7,0x1f7,-1,-56}, 3069 {0x1f8,0x21e,2,1}, 3070 {0x220,0x220,-1,-130}, 3071 {0x222,0x232,2,1}, 3072 {0x23a,0x23a,-1,10795}, 3073 {0x23b,0x23b,-1,1}, 3074 {0x23d,0x23d,-1,-163}, 3075 {0x23e,0x23e,-1,10792}, 3076 {0x241,0x241,-1,1}, 3077 {0x243,0x243,-1,-195}, 3078 {0x244,0x244,-1,69}, 3079 {0x245,0x245,-1,71}, 3080 {0x246,0x24e,2,1}, 3081 {0x370,0x372,2,1}, 3082 {0x376,0x376,-1,1}, 3083 {0x37f,0x37f,-1,116}, 3084 {0x386,0x386,-1,38}, 3085 {0x388,0x38a,1,37}, 3086 {0x38c,0x38c,-1,64}, 3087 {0x38e,0x38f,1,63}, 3088 {0x391,0x3a1,1,32}, 3089 {0x3a3,0x3ab,1,32}, 3090 {0x3cf,0x3cf,-1,8}, 3091 {0x3d8,0x3ee,2,1}, 3092 {0x3f4,0x3f4,-1,-60}, 3093 {0x3f7,0x3f7,-1,1}, 3094 {0x3f9,0x3f9,-1,-7}, 3095 {0x3fa,0x3fa,-1,1}, 3096 {0x3fd,0x3ff,1,-130}, 3097 {0x400,0x40f,1,80}, 3098 {0x410,0x42f,1,32}, 3099 {0x460,0x480,2,1}, 3100 {0x48a,0x4be,2,1}, 3101 {0x4c0,0x4c0,-1,15}, 3102 {0x4c1,0x4cd,2,1}, 3103 {0x4d0,0x52e,2,1}, 3104 {0x531,0x556,1,48}, 3105 {0x10a0,0x10c5,1,7264}, 3106 {0x10c7,0x10cd,6,7264}, 3107 {0x13a0,0x13ef,1,38864}, 3108 {0x13f0,0x13f5,1,8}, 3109 {0x1e00,0x1e94,2,1}, 3110 {0x1e9e,0x1e9e,-1,-7615}, 3111 {0x1ea0,0x1efe,2,1}, 3112 {0x1f08,0x1f0f,1,-8}, 3113 {0x1f18,0x1f1d,1,-8}, 3114 {0x1f28,0x1f2f,1,-8}, 3115 {0x1f38,0x1f3f,1,-8}, 3116 {0x1f48,0x1f4d,1,-8}, 3117 {0x1f59,0x1f5f,2,-8}, 3118 {0x1f68,0x1f6f,1,-8}, 3119 {0x1f88,0x1f8f,1,-8}, 3120 {0x1f98,0x1f9f,1,-8}, 3121 {0x1fa8,0x1faf,1,-8}, 3122 {0x1fb8,0x1fb9,1,-8}, 3123 {0x1fba,0x1fbb,1,-74}, 3124 {0x1fbc,0x1fbc,-1,-9}, 3125 {0x1fc8,0x1fcb,1,-86}, 3126 {0x1fcc,0x1fcc,-1,-9}, 3127 {0x1fd8,0x1fd9,1,-8}, 3128 {0x1fda,0x1fdb,1,-100}, 3129 {0x1fe8,0x1fe9,1,-8}, 3130 {0x1fea,0x1feb,1,-112}, 3131 {0x1fec,0x1fec,-1,-7}, 3132 {0x1ff8,0x1ff9,1,-128}, 3133 {0x1ffa,0x1ffb,1,-126}, 3134 {0x1ffc,0x1ffc,-1,-9}, 3135 {0x2126,0x2126,-1,-7517}, 3136 {0x212a,0x212a,-1,-8383}, 3137 {0x212b,0x212b,-1,-8262}, 3138 {0x2132,0x2132,-1,28}, 3139 {0x2160,0x216f,1,16}, 3140 {0x2183,0x2183,-1,1}, 3141 {0x24b6,0x24cf,1,26}, 3142 {0x2c00,0x2c2e,1,48}, 3143 {0x2c60,0x2c60,-1,1}, 3144 {0x2c62,0x2c62,-1,-10743}, 3145 {0x2c63,0x2c63,-1,-3814}, 3146 {0x2c64,0x2c64,-1,-10727}, 3147 {0x2c67,0x2c6b,2,1}, 3148 {0x2c6d,0x2c6d,-1,-10780}, 3149 {0x2c6e,0x2c6e,-1,-10749}, 3150 {0x2c6f,0x2c6f,-1,-10783}, 3151 {0x2c70,0x2c70,-1,-10782}, 3152 {0x2c72,0x2c75,3,1}, 3153 {0x2c7e,0x2c7f,1,-10815}, 3154 {0x2c80,0x2ce2,2,1}, 3155 {0x2ceb,0x2ced,2,1}, 3156 {0x2cf2,0xa640,31054,1}, 3157 {0xa642,0xa66c,2,1}, 3158 {0xa680,0xa69a,2,1}, 3159 {0xa722,0xa72e,2,1}, 3160 {0xa732,0xa76e,2,1}, 3161 {0xa779,0xa77b,2,1}, 3162 {0xa77d,0xa77d,-1,-35332}, 3163 {0xa77e,0xa786,2,1}, 3164 {0xa78b,0xa78b,-1,1}, 3165 {0xa78d,0xa78d,-1,-42280}, 3166 {0xa790,0xa792,2,1}, 3167 {0xa796,0xa7a8,2,1}, 3168 {0xa7aa,0xa7aa,-1,-42308}, 3169 {0xa7ab,0xa7ab,-1,-42319}, 3170 {0xa7ac,0xa7ac,-1,-42315}, 3171 {0xa7ad,0xa7ad,-1,-42305}, 3172 {0xa7b0,0xa7b0,-1,-42258}, 3173 {0xa7b1,0xa7b1,-1,-42282}, 3174 {0xa7b2,0xa7b2,-1,-42261}, 3175 {0xa7b3,0xa7b3,-1,928}, 3176 {0xa7b4,0xa7b6,2,1}, 3177 {0xff21,0xff3a,1,32}, 3178 {0x10400,0x10427,1,40}, 3179 {0x10c80,0x10cb2,1,64}, 3180 {0x118a0,0x118bf,1,32} 3181 }; 3182 3183 static convertStruct toUpper[] = 3184 { 3185 {0x61,0x7a,1,-32}, 3186 {0xb5,0xb5,-1,743}, 3187 {0xe0,0xf6,1,-32}, 3188 {0xf8,0xfe,1,-32}, 3189 {0xff,0xff,-1,121}, 3190 {0x101,0x12f,2,-1}, 3191 {0x131,0x131,-1,-232}, 3192 {0x133,0x137,2,-1}, 3193 {0x13a,0x148,2,-1}, 3194 {0x14b,0x177,2,-1}, 3195 {0x17a,0x17e,2,-1}, 3196 {0x17f,0x17f,-1,-300}, 3197 {0x180,0x180,-1,195}, 3198 {0x183,0x185,2,-1}, 3199 {0x188,0x18c,4,-1}, 3200 {0x192,0x192,-1,-1}, 3201 {0x195,0x195,-1,97}, 3202 {0x199,0x199,-1,-1}, 3203 {0x19a,0x19a,-1,163}, 3204 {0x19e,0x19e,-1,130}, 3205 {0x1a1,0x1a5,2,-1}, 3206 {0x1a8,0x1ad,5,-1}, 3207 {0x1b0,0x1b4,4,-1}, 3208 {0x1b6,0x1b9,3,-1}, 3209 {0x1bd,0x1bd,-1,-1}, 3210 {0x1bf,0x1bf,-1,56}, 3211 {0x1c5,0x1c5,-1,-1}, 3212 {0x1c6,0x1c6,-1,-2}, 3213 {0x1c8,0x1c8,-1,-1}, 3214 {0x1c9,0x1c9,-1,-2}, 3215 {0x1cb,0x1cb,-1,-1}, 3216 {0x1cc,0x1cc,-1,-2}, 3217 {0x1ce,0x1dc,2,-1}, 3218 {0x1dd,0x1dd,-1,-79}, 3219 {0x1df,0x1ef,2,-1}, 3220 {0x1f2,0x1f2,-1,-1}, 3221 {0x1f3,0x1f3,-1,-2}, 3222 {0x1f5,0x1f9,4,-1}, 3223 {0x1fb,0x21f,2,-1}, 3224 {0x223,0x233,2,-1}, 3225 {0x23c,0x23c,-1,-1}, 3226 {0x23f,0x240,1,10815}, 3227 {0x242,0x247,5,-1}, 3228 {0x249,0x24f,2,-1}, 3229 {0x250,0x250,-1,10783}, 3230 {0x251,0x251,-1,10780}, 3231 {0x252,0x252,-1,10782}, 3232 {0x253,0x253,-1,-210}, 3233 {0x254,0x254,-1,-206}, 3234 {0x256,0x257,1,-205}, 3235 {0x259,0x259,-1,-202}, 3236 {0x25b,0x25b,-1,-203}, 3237 {0x25c,0x25c,-1,42319}, 3238 {0x260,0x260,-1,-205}, 3239 {0x261,0x261,-1,42315}, 3240 {0x263,0x263,-1,-207}, 3241 {0x265,0x265,-1,42280}, 3242 {0x266,0x266,-1,42308}, 3243 {0x268,0x268,-1,-209}, 3244 {0x269,0x269,-1,-211}, 3245 {0x26b,0x26b,-1,10743}, 3246 {0x26c,0x26c,-1,42305}, 3247 {0x26f,0x26f,-1,-211}, 3248 {0x271,0x271,-1,10749}, 3249 {0x272,0x272,-1,-213}, 3250 {0x275,0x275,-1,-214}, 3251 {0x27d,0x27d,-1,10727}, 3252 {0x280,0x283,3,-218}, 3253 {0x287,0x287,-1,42282}, 3254 {0x288,0x288,-1,-218}, 3255 {0x289,0x289,-1,-69}, 3256 {0x28a,0x28b,1,-217}, 3257 {0x28c,0x28c,-1,-71}, 3258 {0x292,0x292,-1,-219}, 3259 {0x29d,0x29d,-1,42261}, 3260 {0x29e,0x29e,-1,42258}, 3261 {0x345,0x345,-1,84}, 3262 {0x371,0x373,2,-1}, 3263 {0x377,0x377,-1,-1}, 3264 {0x37b,0x37d,1,130}, 3265 {0x3ac,0x3ac,-1,-38}, 3266 {0x3ad,0x3af,1,-37}, 3267 {0x3b1,0x3c1,1,-32}, 3268 {0x3c2,0x3c2,-1,-31}, 3269 {0x3c3,0x3cb,1,-32}, 3270 {0x3cc,0x3cc,-1,-64}, 3271 {0x3cd,0x3ce,1,-63}, 3272 {0x3d0,0x3d0,-1,-62}, 3273 {0x3d1,0x3d1,-1,-57}, 3274 {0x3d5,0x3d5,-1,-47}, 3275 {0x3d6,0x3d6,-1,-54}, 3276 {0x3d7,0x3d7,-1,-8}, 3277 {0x3d9,0x3ef,2,-1}, 3278 {0x3f0,0x3f0,-1,-86}, 3279 {0x3f1,0x3f1,-1,-80}, 3280 {0x3f2,0x3f2,-1,7}, 3281 {0x3f3,0x3f3,-1,-116}, 3282 {0x3f5,0x3f5,-1,-96}, 3283 {0x3f8,0x3fb,3,-1}, 3284 {0x430,0x44f,1,-32}, 3285 {0x450,0x45f,1,-80}, 3286 {0x461,0x481,2,-1}, 3287 {0x48b,0x4bf,2,-1}, 3288 {0x4c2,0x4ce,2,-1}, 3289 {0x4cf,0x4cf,-1,-15}, 3290 {0x4d1,0x52f,2,-1}, 3291 {0x561,0x586,1,-48}, 3292 {0x13f8,0x13fd,1,-8}, 3293 {0x1d79,0x1d79,-1,35332}, 3294 {0x1d7d,0x1d7d,-1,3814}, 3295 {0x1e01,0x1e95,2,-1}, 3296 {0x1e9b,0x1e9b,-1,-59}, 3297 {0x1ea1,0x1eff,2,-1}, 3298 {0x1f00,0x1f07,1,8}, 3299 {0x1f10,0x1f15,1,8}, 3300 {0x1f20,0x1f27,1,8}, 3301 {0x1f30,0x1f37,1,8}, 3302 {0x1f40,0x1f45,1,8}, 3303 {0x1f51,0x1f57,2,8}, 3304 {0x1f60,0x1f67,1,8}, 3305 {0x1f70,0x1f71,1,74}, 3306 {0x1f72,0x1f75,1,86}, 3307 {0x1f76,0x1f77,1,100}, 3308 {0x1f78,0x1f79,1,128}, 3309 {0x1f7a,0x1f7b,1,112}, 3310 {0x1f7c,0x1f7d,1,126}, 3311 {0x1f80,0x1f87,1,8}, 3312 {0x1f90,0x1f97,1,8}, 3313 {0x1fa0,0x1fa7,1,8}, 3314 {0x1fb0,0x1fb1,1,8}, 3315 {0x1fb3,0x1fb3,-1,9}, 3316 {0x1fbe,0x1fbe,-1,-7205}, 3317 {0x1fc3,0x1fc3,-1,9}, 3318 {0x1fd0,0x1fd1,1,8}, 3319 {0x1fe0,0x1fe1,1,8}, 3320 {0x1fe5,0x1fe5,-1,7}, 3321 {0x1ff3,0x1ff3,-1,9}, 3322 {0x214e,0x214e,-1,-28}, 3323 {0x2170,0x217f,1,-16}, 3324 {0x2184,0x2184,-1,-1}, 3325 {0x24d0,0x24e9,1,-26}, 3326 {0x2c30,0x2c5e,1,-48}, 3327 {0x2c61,0x2c61,-1,-1}, 3328 {0x2c65,0x2c65,-1,-10795}, 3329 {0x2c66,0x2c66,-1,-10792}, 3330 {0x2c68,0x2c6c,2,-1}, 3331 {0x2c73,0x2c76,3,-1}, 3332 {0x2c81,0x2ce3,2,-1}, 3333 {0x2cec,0x2cee,2,-1}, 3334 {0x2cf3,0x2cf3,-1,-1}, 3335 {0x2d00,0x2d25,1,-7264}, 3336 {0x2d27,0x2d2d,6,-7264}, 3337 {0xa641,0xa66d,2,-1}, 3338 {0xa681,0xa69b,2,-1}, 3339 {0xa723,0xa72f,2,-1}, 3340 {0xa733,0xa76f,2,-1}, 3341 {0xa77a,0xa77c,2,-1}, 3342 {0xa77f,0xa787,2,-1}, 3343 {0xa78c,0xa791,5,-1}, 3344 {0xa793,0xa797,4,-1}, 3345 {0xa799,0xa7a9,2,-1}, 3346 {0xa7b5,0xa7b7,2,-1}, 3347 {0xab53,0xab53,-1,-928}, 3348 {0xab70,0xabbf,1,-38864}, 3349 {0xff41,0xff5a,1,-32}, 3350 {0x10428,0x1044f,1,-40}, 3351 {0x10cc0,0x10cf2,1,-64}, 3352 {0x118c0,0x118df,1,-32} 3353 }; 3354 /* 3355 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3356 * simple case folding. 3357 */ 3358 int 3359 utf_toupper(int a) 3360 { 3361 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */ 3362 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3363 return TOUPPER_ASC(a); 3364 3365 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3366 /* If towupper() is available and handles Unicode, use it. */ 3367 if (!(cmp_flags & CMP_INTERNAL)) 3368 return towupper(a); 3369 #endif 3370 3371 /* For characters below 128 use locale sensitive toupper(). */ 3372 if (a < 128) 3373 return TOUPPER_LOC(a); 3374 3375 /* For any other characters use the above mapping table. */ 3376 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3377 } 3378 3379 int 3380 utf_islower(int a) 3381 { 3382 /* German sharp s is lower case but has no upper case equivalent. */ 3383 return (utf_toupper(a) != a) || a == 0xdf; 3384 } 3385 3386 /* 3387 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3388 * simple case folding. 3389 */ 3390 int 3391 utf_tolower(int a) 3392 { 3393 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */ 3394 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3395 return TOLOWER_ASC(a); 3396 3397 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3398 /* If towlower() is available and handles Unicode, use it. */ 3399 if (!(cmp_flags & CMP_INTERNAL)) 3400 return towlower(a); 3401 #endif 3402 3403 /* For characters below 128 use locale sensitive tolower(). */ 3404 if (a < 128) 3405 return TOLOWER_LOC(a); 3406 3407 /* For any other characters use the above mapping table. */ 3408 return utf_convert(a, toLower, (int)sizeof(toLower)); 3409 } 3410 3411 int 3412 utf_isupper(int a) 3413 { 3414 return (utf_tolower(a) != a); 3415 } 3416 3417 static int 3418 utf_strnicmp( 3419 char_u *s1, 3420 char_u *s2, 3421 size_t n1, 3422 size_t n2) 3423 { 3424 int c1, c2, cdiff; 3425 char_u buffer[6]; 3426 3427 for (;;) 3428 { 3429 c1 = utf_safe_read_char_adv(&s1, &n1); 3430 c2 = utf_safe_read_char_adv(&s2, &n2); 3431 3432 if (c1 <= 0 || c2 <= 0) 3433 break; 3434 3435 if (c1 == c2) 3436 continue; 3437 3438 cdiff = utf_fold(c1) - utf_fold(c2); 3439 if (cdiff != 0) 3440 return cdiff; 3441 } 3442 3443 /* some string ended or has an incomplete/illegal character sequence */ 3444 3445 if (c1 == 0 || c2 == 0) 3446 { 3447 /* some string ended. shorter string is smaller */ 3448 if (c1 == 0 && c2 == 0) 3449 return 0; 3450 return c1 == 0 ? -1 : 1; 3451 } 3452 3453 /* Continue with bytewise comparison to produce some result that 3454 * would make comparison operations involving this function transitive. 3455 * 3456 * If only one string had an error, comparison should be made with 3457 * folded version of the other string. In this case it is enough 3458 * to fold just one character to determine the result of comparison. */ 3459 3460 if (c1 != -1 && c2 == -1) 3461 { 3462 n1 = utf_char2bytes(utf_fold(c1), buffer); 3463 s1 = buffer; 3464 } 3465 else if (c2 != -1 && c1 == -1) 3466 { 3467 n2 = utf_char2bytes(utf_fold(c2), buffer); 3468 s2 = buffer; 3469 } 3470 3471 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3472 { 3473 cdiff = (int)(*s1) - (int)(*s2); 3474 if (cdiff != 0) 3475 return cdiff; 3476 3477 s1++; 3478 s2++; 3479 n1--; 3480 n2--; 3481 } 3482 3483 if (n1 > 0 && *s1 == NUL) 3484 n1 = 0; 3485 if (n2 > 0 && *s2 == NUL) 3486 n2 = 0; 3487 3488 if (n1 == 0 && n2 == 0) 3489 return 0; 3490 return n1 == 0 ? -1 : 1; 3491 } 3492 3493 /* 3494 * Version of strnicmp() that handles multi-byte characters. 3495 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3496 * probably use strnicmp(), because there are no ASCII characters in the 3497 * second byte. 3498 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3499 * two characters otherwise. 3500 */ 3501 int 3502 mb_strnicmp(char_u *s1, char_u *s2, size_t nn) 3503 { 3504 int i, l; 3505 int cdiff; 3506 int n = (int)nn; 3507 3508 if (enc_utf8) 3509 { 3510 return utf_strnicmp(s1, s2, nn, nn); 3511 } 3512 else 3513 { 3514 for (i = 0; i < n; i += l) 3515 { 3516 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */ 3517 return 0; 3518 3519 l = (*mb_ptr2len)(s1 + i); 3520 if (l <= 1) 3521 { 3522 /* Single byte: first check normally, then with ignore case. */ 3523 if (s1[i] != s2[i]) 3524 { 3525 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3526 if (cdiff != 0) 3527 return cdiff; 3528 } 3529 } 3530 else 3531 { 3532 /* For non-Unicode multi-byte don't ignore case. */ 3533 if (l > n - i) 3534 l = n - i; 3535 cdiff = STRNCMP(s1 + i, s2 + i, l); 3536 if (cdiff != 0) 3537 return cdiff; 3538 } 3539 } 3540 } 3541 return 0; 3542 } 3543 3544 /* 3545 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3546 * 'encoding' has been set to. 3547 */ 3548 void 3549 show_utf8(void) 3550 { 3551 int len; 3552 int rlen = 0; 3553 char_u *line; 3554 int clen; 3555 int i; 3556 3557 /* Get the byte length of the char under the cursor, including composing 3558 * characters. */ 3559 line = ml_get_cursor(); 3560 len = utfc_ptr2len(line); 3561 if (len == 0) 3562 { 3563 MSG("NUL"); 3564 return; 3565 } 3566 3567 clen = 0; 3568 for (i = 0; i < len; ++i) 3569 { 3570 if (clen == 0) 3571 { 3572 /* start of (composing) character, get its length */ 3573 if (i > 0) 3574 { 3575 STRCPY(IObuff + rlen, "+ "); 3576 rlen += 2; 3577 } 3578 clen = utf_ptr2len(line + i); 3579 } 3580 sprintf((char *)IObuff + rlen, "%02x ", 3581 (line[i] == NL) ? NUL : line[i]); /* NUL is stored as NL */ 3582 --clen; 3583 rlen += (int)STRLEN(IObuff + rlen); 3584 if (rlen > IOSIZE - 20) 3585 break; 3586 } 3587 3588 msg(IObuff); 3589 } 3590 3591 /* 3592 * mb_head_off() function pointer. 3593 * Return offset from "p" to the first byte of the character it points into. 3594 * If "p" points to the NUL at the end of the string return 0. 3595 * Returns 0 when already at the first byte of a character. 3596 */ 3597 int 3598 latin_head_off(char_u *base UNUSED, char_u *p UNUSED) 3599 { 3600 return 0; 3601 } 3602 3603 int 3604 dbcs_head_off(char_u *base, char_u *p) 3605 { 3606 char_u *q; 3607 3608 /* It can't be a trailing byte when not using DBCS, at the start of the 3609 * string or the previous byte can't start a double-byte. */ 3610 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3611 return 0; 3612 3613 /* This is slow: need to start at the base and go forward until the 3614 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */ 3615 q = base; 3616 while (q < p) 3617 q += dbcs_ptr2len(q); 3618 return (q == p) ? 0 : 1; 3619 } 3620 3621 /* 3622 * Special version of dbcs_head_off() that works for ScreenLines[], where 3623 * single-width DBCS_JPNU characters are stored separately. 3624 */ 3625 int 3626 dbcs_screen_head_off(char_u *base, char_u *p) 3627 { 3628 char_u *q; 3629 3630 /* It can't be a trailing byte when not using DBCS, at the start of the 3631 * string or the previous byte can't start a double-byte. 3632 * For euc-jp an 0x8e byte in the previous cell always means we have a 3633 * lead byte in the current cell. */ 3634 if (p <= base 3635 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3636 || MB_BYTE2LEN(p[-1]) == 1 3637 || *p == NUL) 3638 return 0; 3639 3640 /* This is slow: need to start at the base and go forward until the 3641 * byte we are looking for. Return 1 when we went past it, 0 otherwise. 3642 * For DBCS_JPNU look out for 0x8e, which means the second byte is not 3643 * stored as the next byte. */ 3644 q = base; 3645 while (q < p) 3646 { 3647 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3648 ++q; 3649 else 3650 q += dbcs_ptr2len(q); 3651 } 3652 return (q == p) ? 0 : 1; 3653 } 3654 3655 int 3656 utf_head_off(char_u *base, char_u *p) 3657 { 3658 char_u *q; 3659 char_u *s; 3660 int c; 3661 int len; 3662 #ifdef FEAT_ARABIC 3663 char_u *j; 3664 #endif 3665 3666 if (*p < 0x80) /* be quick for ASCII */ 3667 return 0; 3668 3669 /* Skip backwards over trailing bytes: 10xx.xxxx 3670 * Skip backwards again if on a composing char. */ 3671 for (q = p; ; --q) 3672 { 3673 /* Move s to the last byte of this char. */ 3674 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3675 ; 3676 /* Move q to the first byte of this char. */ 3677 while (q > base && (*q & 0xc0) == 0x80) 3678 --q; 3679 /* Check for illegal sequence. Do allow an illegal byte after where we 3680 * started. */ 3681 len = utf8len_tab[*q]; 3682 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3683 return 0; 3684 3685 if (q <= base) 3686 break; 3687 3688 c = utf_ptr2char(q); 3689 if (utf_iscomposing(c)) 3690 continue; 3691 3692 #ifdef FEAT_ARABIC 3693 if (arabic_maycombine(c)) 3694 { 3695 /* Advance to get a sneak-peak at the next char */ 3696 j = q; 3697 --j; 3698 /* Move j to the first byte of this char. */ 3699 while (j > base && (*j & 0xc0) == 0x80) 3700 --j; 3701 if (arabic_combine(utf_ptr2char(j), c)) 3702 continue; 3703 } 3704 #endif 3705 break; 3706 } 3707 3708 return (int)(p - q); 3709 } 3710 3711 /* 3712 * Copy a character from "*fp" to "*tp" and advance the pointers. 3713 */ 3714 void 3715 mb_copy_char(char_u **fp, char_u **tp) 3716 { 3717 int l = (*mb_ptr2len)(*fp); 3718 3719 mch_memmove(*tp, *fp, (size_t)l); 3720 *tp += l; 3721 *fp += l; 3722 } 3723 3724 /* 3725 * Return the offset from "p" to the first byte of a character. When "p" is 3726 * at the start of a character 0 is returned, otherwise the offset to the next 3727 * character. Can start anywhere in a stream of bytes. 3728 */ 3729 int 3730 mb_off_next(char_u *base, char_u *p) 3731 { 3732 int i; 3733 int j; 3734 3735 if (enc_utf8) 3736 { 3737 if (*p < 0x80) /* be quick for ASCII */ 3738 return 0; 3739 3740 /* Find the next character that isn't 10xx.xxxx */ 3741 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 3742 ; 3743 if (i > 0) 3744 { 3745 /* Check for illegal sequence. */ 3746 for (j = 0; p - j > base; ++j) 3747 if ((p[-j] & 0xc0) != 0x80) 3748 break; 3749 if (utf8len_tab[p[-j]] != i + j) 3750 return 0; 3751 } 3752 return i; 3753 } 3754 3755 /* Only need to check if we're on a trail byte, it doesn't matter if we 3756 * want the offset to the next or current character. */ 3757 return (*mb_head_off)(base, p); 3758 } 3759 3760 /* 3761 * Return the offset from "p" to the last byte of the character it points 3762 * into. Can start anywhere in a stream of bytes. 3763 */ 3764 int 3765 mb_tail_off(char_u *base, char_u *p) 3766 { 3767 int i; 3768 int j; 3769 3770 if (*p == NUL) 3771 return 0; 3772 3773 if (enc_utf8) 3774 { 3775 /* Find the last character that is 10xx.xxxx */ 3776 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 3777 ; 3778 /* Check for illegal sequence. */ 3779 for (j = 0; p - j > base; ++j) 3780 if ((p[-j] & 0xc0) != 0x80) 3781 break; 3782 if (utf8len_tab[p[-j]] != i + j + 1) 3783 return 0; 3784 return i; 3785 } 3786 3787 /* It can't be the first byte if a double-byte when not using DBCS, at the 3788 * end of the string or the byte can't start a double-byte. */ 3789 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 3790 return 0; 3791 3792 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3793 return 1 - dbcs_head_off(base, p); 3794 } 3795 3796 /* 3797 * Find the next illegal byte sequence. 3798 */ 3799 void 3800 utf_find_illegal(void) 3801 { 3802 pos_T pos = curwin->w_cursor; 3803 char_u *p; 3804 int len; 3805 vimconv_T vimconv; 3806 char_u *tofree = NULL; 3807 3808 vimconv.vc_type = CONV_NONE; 3809 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 3810 { 3811 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 3812 * possibly a utf-8 file with illegal bytes. Setup for conversion 3813 * from utf-8 to 'fileencoding'. */ 3814 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 3815 } 3816 3817 #ifdef FEAT_VIRTUALEDIT 3818 curwin->w_cursor.coladd = 0; 3819 #endif 3820 for (;;) 3821 { 3822 p = ml_get_cursor(); 3823 if (vimconv.vc_type != CONV_NONE) 3824 { 3825 vim_free(tofree); 3826 tofree = string_convert(&vimconv, p, NULL); 3827 if (tofree == NULL) 3828 break; 3829 p = tofree; 3830 } 3831 3832 while (*p != NUL) 3833 { 3834 /* Illegal means that there are not enough trail bytes (checked by 3835 * utf_ptr2len()) or too many of them (overlong sequence). */ 3836 len = utf_ptr2len(p); 3837 if (*p >= 0x80 && (len == 1 3838 || utf_char2len(utf_ptr2char(p)) != len)) 3839 { 3840 if (vimconv.vc_type == CONV_NONE) 3841 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 3842 else 3843 { 3844 int l; 3845 3846 len = (int)(p - tofree); 3847 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 3848 { 3849 l = utf_ptr2len(p); 3850 curwin->w_cursor.col += l; 3851 } 3852 } 3853 goto theend; 3854 } 3855 p += len; 3856 } 3857 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 3858 break; 3859 ++curwin->w_cursor.lnum; 3860 curwin->w_cursor.col = 0; 3861 } 3862 3863 /* didn't find it: don't move and beep */ 3864 curwin->w_cursor = pos; 3865 beep_flush(); 3866 3867 theend: 3868 vim_free(tofree); 3869 convert_setup(&vimconv, NULL, NULL); 3870 } 3871 3872 #if defined(FEAT_GUI_GTK) || defined(PROTO) 3873 /* 3874 * Return TRUE if string "s" is a valid utf-8 string. 3875 * When "end" is NULL stop at the first NUL. 3876 * When "end" is positive stop there. 3877 */ 3878 int 3879 utf_valid_string(char_u *s, char_u *end) 3880 { 3881 int l; 3882 char_u *p = s; 3883 3884 while (end == NULL ? *p != NUL : p < end) 3885 { 3886 l = utf8len_tab_zero[*p]; 3887 if (l == 0) 3888 return FALSE; /* invalid lead byte */ 3889 if (end != NULL && p + l > end) 3890 return FALSE; /* incomplete byte sequence */ 3891 ++p; 3892 while (--l > 0) 3893 if ((*p++ & 0xc0) != 0x80) 3894 return FALSE; /* invalid trail byte */ 3895 } 3896 return TRUE; 3897 } 3898 #endif 3899 3900 #if defined(FEAT_GUI) || defined(PROTO) 3901 /* 3902 * Special version of mb_tail_off() for use in ScreenLines[]. 3903 */ 3904 int 3905 dbcs_screen_tail_off(char_u *base, char_u *p) 3906 { 3907 /* It can't be the first byte if a double-byte when not using DBCS, at the 3908 * end of the string or the byte can't start a double-byte. 3909 * For euc-jp an 0x8e byte always means we have a lead byte in the current 3910 * cell. */ 3911 if (*p == NUL || p[1] == NUL 3912 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 3913 || MB_BYTE2LEN(*p) == 1) 3914 return 0; 3915 3916 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3917 return 1 - dbcs_screen_head_off(base, p); 3918 } 3919 #endif 3920 3921 /* 3922 * If the cursor moves on an trail byte, set the cursor on the lead byte. 3923 * Thus it moves left if necessary. 3924 * Return TRUE when the cursor was adjusted. 3925 */ 3926 void 3927 mb_adjust_cursor(void) 3928 { 3929 mb_adjustpos(curbuf, &curwin->w_cursor); 3930 } 3931 3932 /* 3933 * Adjust position "*lp" to point to the first byte of a multi-byte character. 3934 * If it points to a tail byte it's moved backwards to the head byte. 3935 */ 3936 void 3937 mb_adjustpos(buf_T *buf, pos_T *lp) 3938 { 3939 char_u *p; 3940 3941 if (lp->col > 0 3942 #ifdef FEAT_VIRTUALEDIT 3943 || lp->coladd > 1 3944 #endif 3945 ) 3946 { 3947 p = ml_get_buf(buf, lp->lnum, FALSE); 3948 lp->col -= (*mb_head_off)(p, p + lp->col); 3949 #ifdef FEAT_VIRTUALEDIT 3950 /* Reset "coladd" when the cursor would be on the right half of a 3951 * double-wide character. */ 3952 if (lp->coladd == 1 3953 && p[lp->col] != TAB 3954 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 3955 && ptr2cells(p + lp->col) > 1) 3956 lp->coladd = 0; 3957 #endif 3958 } 3959 } 3960 3961 /* 3962 * Return a pointer to the character before "*p", if there is one. 3963 */ 3964 char_u * 3965 mb_prevptr( 3966 char_u *line, /* start of the string */ 3967 char_u *p) 3968 { 3969 if (p > line) 3970 mb_ptr_back(line, p); 3971 return p; 3972 } 3973 3974 /* 3975 * Return the character length of "str". Each multi-byte character (with 3976 * following composing characters) counts as one. 3977 */ 3978 int 3979 mb_charlen(char_u *str) 3980 { 3981 char_u *p = str; 3982 int count; 3983 3984 if (p == NULL) 3985 return 0; 3986 3987 for (count = 0; *p != NUL; count++) 3988 p += (*mb_ptr2len)(p); 3989 3990 return count; 3991 } 3992 3993 #if defined(FEAT_SPELL) || defined(PROTO) 3994 /* 3995 * Like mb_charlen() but for a string with specified length. 3996 */ 3997 int 3998 mb_charlen_len(char_u *str, int len) 3999 { 4000 char_u *p = str; 4001 int count; 4002 4003 for (count = 0; *p != NUL && p < str + len; count++) 4004 p += (*mb_ptr2len)(p); 4005 4006 return count; 4007 } 4008 #endif 4009 4010 /* 4011 * Try to un-escape a multi-byte character. 4012 * Used for the "to" and "from" part of a mapping. 4013 * Return the un-escaped string if it is a multi-byte character, and advance 4014 * "pp" to just after the bytes that formed it. 4015 * Return NULL if no multi-byte char was found. 4016 */ 4017 char_u * 4018 mb_unescape(char_u **pp) 4019 { 4020 static char_u buf[6]; 4021 int n; 4022 int m = 0; 4023 char_u *str = *pp; 4024 4025 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 4026 * KS_EXTRA KE_CSI to CSI. 4027 * Maximum length of a utf-8 character is 4 bytes. */ 4028 for (n = 0; str[n] != NUL && m < 4; ++n) 4029 { 4030 if (str[n] == K_SPECIAL 4031 && str[n + 1] == KS_SPECIAL 4032 && str[n + 2] == KE_FILLER) 4033 { 4034 buf[m++] = K_SPECIAL; 4035 n += 2; 4036 } 4037 else if ((str[n] == K_SPECIAL 4038 # ifdef FEAT_GUI 4039 || str[n] == CSI 4040 # endif 4041 ) 4042 && str[n + 1] == KS_EXTRA 4043 && str[n + 2] == (int)KE_CSI) 4044 { 4045 buf[m++] = CSI; 4046 n += 2; 4047 } 4048 else if (str[n] == K_SPECIAL 4049 # ifdef FEAT_GUI 4050 || str[n] == CSI 4051 # endif 4052 ) 4053 break; /* a special key can't be a multibyte char */ 4054 else 4055 buf[m++] = str[n]; 4056 buf[m] = NUL; 4057 4058 /* Return a multi-byte character if it's found. An illegal sequence 4059 * will result in a 1 here. */ 4060 if ((*mb_ptr2len)(buf) > 1) 4061 { 4062 *pp = str + n + 1; 4063 return buf; 4064 } 4065 4066 /* Bail out quickly for ASCII. */ 4067 if (buf[0] < 128) 4068 break; 4069 } 4070 return NULL; 4071 } 4072 4073 /* 4074 * Return TRUE if the character at "row"/"col" on the screen is the left side 4075 * of a double-width character. 4076 * Caller must make sure "row" and "col" are not invalid! 4077 */ 4078 int 4079 mb_lefthalve(int row, int col) 4080 { 4081 #ifdef FEAT_HANGULIN 4082 if (composing_hangul) 4083 return TRUE; 4084 #endif 4085 return (*mb_off2cells)(LineOffset[row] + col, 4086 LineOffset[row] + screen_Columns) > 1; 4087 } 4088 4089 /* 4090 * Correct a position on the screen, if it's the right half of a double-wide 4091 * char move it to the left half. Returns the corrected column. 4092 */ 4093 int 4094 mb_fix_col(int col, int row) 4095 { 4096 col = check_col(col); 4097 row = check_row(row); 4098 if (has_mbyte && ScreenLines != NULL && col > 0 4099 && ((enc_dbcs 4100 && ScreenLines[LineOffset[row] + col] != NUL 4101 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 4102 ScreenLines + LineOffset[row] + col)) 4103 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0))) 4104 return col - 1; 4105 return col; 4106 } 4107 #endif 4108 4109 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 4110 static int enc_alias_search(char_u *name); 4111 4112 /* 4113 * Skip the Vim specific head of a 'encoding' name. 4114 */ 4115 char_u * 4116 enc_skip(char_u *p) 4117 { 4118 if (STRNCMP(p, "2byte-", 6) == 0) 4119 return p + 6; 4120 if (STRNCMP(p, "8bit-", 5) == 0) 4121 return p + 5; 4122 return p; 4123 } 4124 4125 /* 4126 * Find the canonical name for encoding "enc". 4127 * When the name isn't recognized, returns "enc" itself, but with all lower 4128 * case characters and '_' replaced with '-'. 4129 * Returns an allocated string. NULL for out-of-memory. 4130 */ 4131 char_u * 4132 enc_canonize(char_u *enc) 4133 { 4134 char_u *r; 4135 char_u *p, *s; 4136 int i; 4137 4138 # ifdef FEAT_MBYTE 4139 if (STRCMP(enc, "default") == 0) 4140 { 4141 /* Use the default encoding as it's found by set_init_1(). */ 4142 r = get_encoding_default(); 4143 if (r == NULL) 4144 r = (char_u *)"latin1"; 4145 return vim_strsave(r); 4146 } 4147 # endif 4148 4149 /* copy "enc" to allocated memory, with room for two '-' */ 4150 r = alloc((unsigned)(STRLEN(enc) + 3)); 4151 if (r != NULL) 4152 { 4153 /* Make it all lower case and replace '_' with '-'. */ 4154 p = r; 4155 for (s = enc; *s != NUL; ++s) 4156 { 4157 if (*s == '_') 4158 *p++ = '-'; 4159 else 4160 *p++ = TOLOWER_ASC(*s); 4161 } 4162 *p = NUL; 4163 4164 /* Skip "2byte-" and "8bit-". */ 4165 p = enc_skip(r); 4166 4167 /* Change "microsoft-cp" to "cp". Used in some spell files. */ 4168 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4169 STRMOVE(p, p + 10); 4170 4171 /* "iso8859" -> "iso-8859" */ 4172 if (STRNCMP(p, "iso8859", 7) == 0) 4173 { 4174 STRMOVE(p + 4, p + 3); 4175 p[3] = '-'; 4176 } 4177 4178 /* "iso-8859n" -> "iso-8859-n" */ 4179 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4180 { 4181 STRMOVE(p + 9, p + 8); 4182 p[8] = '-'; 4183 } 4184 4185 /* "latin-N" -> "latinN" */ 4186 if (STRNCMP(p, "latin-", 6) == 0) 4187 STRMOVE(p + 5, p + 6); 4188 4189 if (enc_canon_search(p) >= 0) 4190 { 4191 /* canonical name can be used unmodified */ 4192 if (p != r) 4193 STRMOVE(r, p); 4194 } 4195 else if ((i = enc_alias_search(p)) >= 0) 4196 { 4197 /* alias recognized, get canonical name */ 4198 vim_free(r); 4199 r = vim_strsave((char_u *)enc_canon_table[i].name); 4200 } 4201 } 4202 return r; 4203 } 4204 4205 /* 4206 * Search for an encoding alias of "name". 4207 * Returns -1 when not found. 4208 */ 4209 static int 4210 enc_alias_search(char_u *name) 4211 { 4212 int i; 4213 4214 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4215 if (STRCMP(name, enc_alias_table[i].name) == 0) 4216 return enc_alias_table[i].canon; 4217 return -1; 4218 } 4219 #endif 4220 4221 #if defined(FEAT_MBYTE) || defined(PROTO) 4222 4223 #ifdef HAVE_LANGINFO_H 4224 # include <langinfo.h> 4225 #endif 4226 4227 /* 4228 * Get the canonicalized encoding of the current locale. 4229 * Returns an allocated string when successful, NULL when not. 4230 */ 4231 char_u * 4232 enc_locale(void) 4233 { 4234 #ifndef WIN3264 4235 char *s; 4236 char *p; 4237 int i; 4238 #endif 4239 char buf[50]; 4240 #ifdef WIN3264 4241 long acp = GetACP(); 4242 4243 if (acp == 1200) 4244 STRCPY(buf, "ucs-2le"); 4245 else if (acp == 1252) /* cp1252 is used as latin1 */ 4246 STRCPY(buf, "latin1"); 4247 else 4248 sprintf(buf, "cp%ld", acp); 4249 #else 4250 # ifdef HAVE_NL_LANGINFO_CODESET 4251 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4252 # endif 4253 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4254 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4255 # endif 4256 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4257 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4258 s = getenv("LANG"); 4259 4260 if (s == NULL || *s == NUL) 4261 return FAIL; 4262 4263 /* The most generic locale format is: 4264 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4265 * If there is a '.' remove the part before it. 4266 * if there is something after the codeset, remove it. 4267 * Make the name lowercase and replace '_' with '-'. 4268 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4269 * "ko_KR.EUC" == "euc-kr" 4270 */ 4271 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4272 { 4273 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4274 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4275 { 4276 /* copy "XY.EUC" to "euc-XY" to buf[10] */ 4277 STRCPY(buf + 10, "euc-"); 4278 buf[14] = p[-2]; 4279 buf[15] = p[-1]; 4280 buf[16] = 0; 4281 s = buf + 10; 4282 } 4283 else 4284 s = p + 1; 4285 } 4286 for (i = 0; s[i] != NUL && i < (int)sizeof(buf) - 1; ++i) 4287 { 4288 if (s[i] == '_' || s[i] == '-') 4289 buf[i] = '-'; 4290 else if (isalnum((int)s[i])) 4291 buf[i] = TOLOWER_ASC(s[i]); 4292 else 4293 break; 4294 } 4295 buf[i] = NUL; 4296 #endif 4297 4298 return enc_canonize((char_u *)buf); 4299 } 4300 4301 #if defined(WIN3264) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4302 /* 4303 * Convert an encoding name to an MS-Windows codepage. 4304 * Returns zero if no codepage can be figured out. 4305 */ 4306 int 4307 encname2codepage(char_u *name) 4308 { 4309 int cp; 4310 char_u *p = name; 4311 int idx; 4312 4313 if (STRNCMP(p, "8bit-", 5) == 0) 4314 p += 5; 4315 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4316 p += 6; 4317 4318 if (p[0] == 'c' && p[1] == 'p') 4319 cp = atoi((char *)p + 2); 4320 else if ((idx = enc_canon_search(p)) >= 0) 4321 cp = enc_canon_table[idx].codepage; 4322 else 4323 return 0; 4324 if (IsValidCodePage(cp)) 4325 return cp; 4326 return 0; 4327 } 4328 #endif 4329 4330 # if defined(USE_ICONV) || defined(PROTO) 4331 4332 static char_u *iconv_string(vimconv_T *vcp, char_u *str, int slen, int *unconvlenp, int *resultlenp); 4333 4334 /* 4335 * Call iconv_open() with a check if iconv() works properly (there are broken 4336 * versions). 4337 * Returns (void *)-1 if failed. 4338 * (should return iconv_t, but that causes problems with prototypes). 4339 */ 4340 void * 4341 my_iconv_open(char_u *to, char_u *from) 4342 { 4343 iconv_t fd; 4344 #define ICONV_TESTLEN 400 4345 char_u tobuf[ICONV_TESTLEN]; 4346 char *p; 4347 size_t tolen; 4348 static int iconv_ok = -1; 4349 4350 if (iconv_ok == FALSE) 4351 return (void *)-1; /* detected a broken iconv() previously */ 4352 4353 #ifdef DYNAMIC_ICONV 4354 /* Check if the iconv.dll can be found. */ 4355 if (!iconv_enabled(TRUE)) 4356 return (void *)-1; 4357 #endif 4358 4359 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4360 4361 if (fd != (iconv_t)-1 && iconv_ok == -1) 4362 { 4363 /* 4364 * Do a dummy iconv() call to check if it actually works. There is a 4365 * version of iconv() on Linux that is broken. We can't ignore it, 4366 * because it's wide-spread. The symptoms are that after outputting 4367 * the initial shift state the "to" pointer is NULL and conversion 4368 * stops for no apparent reason after about 8160 characters. 4369 */ 4370 p = (char *)tobuf; 4371 tolen = ICONV_TESTLEN; 4372 (void)iconv(fd, NULL, NULL, &p, &tolen); 4373 if (p == NULL) 4374 { 4375 iconv_ok = FALSE; 4376 iconv_close(fd); 4377 fd = (iconv_t)-1; 4378 } 4379 else 4380 iconv_ok = TRUE; 4381 } 4382 4383 return (void *)fd; 4384 } 4385 4386 /* 4387 * Convert the string "str[slen]" with iconv(). 4388 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4389 * sequence and set "*unconvlenp" to the length of it. 4390 * Returns the converted string in allocated memory. NULL for an error. 4391 * If resultlenp is not NULL, sets it to the result length in bytes. 4392 */ 4393 static char_u * 4394 iconv_string( 4395 vimconv_T *vcp, 4396 char_u *str, 4397 int slen, 4398 int *unconvlenp, 4399 int *resultlenp) 4400 { 4401 const char *from; 4402 size_t fromlen; 4403 char *to; 4404 size_t tolen; 4405 size_t len = 0; 4406 size_t done = 0; 4407 char_u *result = NULL; 4408 char_u *p; 4409 int l; 4410 4411 from = (char *)str; 4412 fromlen = slen; 4413 for (;;) 4414 { 4415 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4416 { 4417 /* Allocate enough room for most conversions. When re-allocating 4418 * increase the buffer size. */ 4419 len = len + fromlen * 2 + 40; 4420 p = alloc((unsigned)len); 4421 if (p != NULL && done > 0) 4422 mch_memmove(p, result, done); 4423 vim_free(result); 4424 result = p; 4425 if (result == NULL) /* out of memory */ 4426 break; 4427 } 4428 4429 to = (char *)result + done; 4430 tolen = len - done - 2; 4431 /* Avoid a warning for systems with a wrong iconv() prototype by 4432 * casting the second argument to void *. */ 4433 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4434 != (size_t)-1) 4435 { 4436 /* Finished, append a NUL. */ 4437 *to = NUL; 4438 break; 4439 } 4440 4441 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4442 * iconv library may use one of them. */ 4443 if (!vcp->vc_fail && unconvlenp != NULL 4444 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4445 { 4446 /* Handle an incomplete sequence at the end. */ 4447 *to = NUL; 4448 *unconvlenp = (int)fromlen; 4449 break; 4450 } 4451 4452 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4453 * iconv library may use one of them. */ 4454 else if (!vcp->vc_fail 4455 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4456 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4457 { 4458 /* Can't convert: insert a '?' and skip a character. This assumes 4459 * conversion from 'encoding' to something else. In other 4460 * situations we don't know what to skip anyway. */ 4461 *to++ = '?'; 4462 if ((*mb_ptr2cells)((char_u *)from) > 1) 4463 *to++ = '?'; 4464 if (enc_utf8) 4465 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4466 else 4467 { 4468 l = (*mb_ptr2len)((char_u *)from); 4469 if (l > (int)fromlen) 4470 l = (int)fromlen; 4471 } 4472 from += l; 4473 fromlen -= l; 4474 } 4475 else if (ICONV_ERRNO != ICONV_E2BIG) 4476 { 4477 /* conversion failed */ 4478 vim_free(result); 4479 result = NULL; 4480 break; 4481 } 4482 /* Not enough room or skipping illegal sequence. */ 4483 done = to - (char *)result; 4484 } 4485 4486 if (resultlenp != NULL && result != NULL) 4487 *resultlenp = (int)(to - (char *)result); 4488 return result; 4489 } 4490 4491 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4492 /* 4493 * Dynamically load the "iconv.dll" on Win32. 4494 */ 4495 4496 # ifndef DYNAMIC_ICONV /* must be generating prototypes */ 4497 # define HINSTANCE int 4498 # endif 4499 static HINSTANCE hIconvDLL = 0; 4500 static HINSTANCE hMsvcrtDLL = 0; 4501 4502 # ifndef DYNAMIC_ICONV_DLL 4503 # define DYNAMIC_ICONV_DLL "iconv.dll" 4504 # define DYNAMIC_ICONV_DLL_ALT1 "libiconv.dll" 4505 # define DYNAMIC_ICONV_DLL_ALT2 "libiconv2.dll" 4506 # define DYNAMIC_ICONV_DLL_ALT3 "libiconv-2.dll" 4507 # endif 4508 # ifndef DYNAMIC_MSVCRT_DLL 4509 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4510 # endif 4511 4512 /* 4513 * Get the address of 'funcname' which is imported by 'hInst' DLL. 4514 */ 4515 static void * 4516 get_iconv_import_func(HINSTANCE hInst, const char *funcname) 4517 { 4518 PBYTE pImage = (PBYTE)hInst; 4519 PIMAGE_DOS_HEADER pDOS = (PIMAGE_DOS_HEADER)hInst; 4520 PIMAGE_NT_HEADERS pPE; 4521 PIMAGE_IMPORT_DESCRIPTOR pImpDesc; 4522 PIMAGE_THUNK_DATA pIAT; /* Import Address Table */ 4523 PIMAGE_THUNK_DATA pINT; /* Import Name Table */ 4524 PIMAGE_IMPORT_BY_NAME pImpName; 4525 4526 if (pDOS->e_magic != IMAGE_DOS_SIGNATURE) 4527 return NULL; 4528 pPE = (PIMAGE_NT_HEADERS)(pImage + pDOS->e_lfanew); 4529 if (pPE->Signature != IMAGE_NT_SIGNATURE) 4530 return NULL; 4531 pImpDesc = (PIMAGE_IMPORT_DESCRIPTOR)(pImage 4532 + pPE->OptionalHeader.DataDirectory[IMAGE_DIRECTORY_ENTRY_IMPORT] 4533 .VirtualAddress); 4534 for (; pImpDesc->FirstThunk; ++pImpDesc) 4535 { 4536 if (!pImpDesc->OriginalFirstThunk) 4537 continue; 4538 pIAT = (PIMAGE_THUNK_DATA)(pImage + pImpDesc->FirstThunk); 4539 pINT = (PIMAGE_THUNK_DATA)(pImage + pImpDesc->OriginalFirstThunk); 4540 for (; pIAT->u1.Function; ++pIAT, ++pINT) 4541 { 4542 if (IMAGE_SNAP_BY_ORDINAL(pINT->u1.Ordinal)) 4543 continue; 4544 pImpName = (PIMAGE_IMPORT_BY_NAME)(pImage 4545 + (UINT_PTR)(pINT->u1.AddressOfData)); 4546 if (strcmp((char *)pImpName->Name, funcname) == 0) 4547 return (void *)pIAT->u1.Function; 4548 } 4549 } 4550 return NULL; 4551 } 4552 4553 /* 4554 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4555 */ 4556 int 4557 iconv_enabled(int verbose) 4558 { 4559 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4560 return TRUE; 4561 4562 /* The iconv DLL file goes under different names, try them all. 4563 * Do the "2" version first, it's newer. */ 4564 #ifdef DYNAMIC_ICONV_DLL_ALT2 4565 if (hIconvDLL == 0) 4566 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT2); 4567 #endif 4568 #ifdef DYNAMIC_ICONV_DLL_ALT3 4569 if (hIconvDLL == 0) 4570 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT3); 4571 #endif 4572 if (hIconvDLL == 0) 4573 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4574 #ifdef DYNAMIC_ICONV_DLL_ALT1 4575 if (hIconvDLL == 0) 4576 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT1); 4577 #endif 4578 4579 if (hIconvDLL != 0) 4580 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4581 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4582 { 4583 /* Only give the message when 'verbose' is set, otherwise it might be 4584 * done whenever a conversion is attempted. */ 4585 if (verbose && p_verbose > 0) 4586 { 4587 verbose_enter(); 4588 EMSG2(_(e_loadlib), 4589 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4590 verbose_leave(); 4591 } 4592 iconv_end(); 4593 return FALSE; 4594 } 4595 4596 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4597 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4598 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4599 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4600 iconv_errno = get_iconv_import_func(hIconvDLL, "_errno"); 4601 if (iconv_errno == NULL) 4602 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4603 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4604 || iconvctl == NULL || iconv_errno == NULL) 4605 { 4606 iconv_end(); 4607 if (verbose && p_verbose > 0) 4608 { 4609 verbose_enter(); 4610 EMSG2(_(e_loadfunc), "for libiconv"); 4611 verbose_leave(); 4612 } 4613 return FALSE; 4614 } 4615 return TRUE; 4616 } 4617 4618 void 4619 iconv_end(void) 4620 { 4621 /* Don't use iconv() when inputting or outputting characters. */ 4622 if (input_conv.vc_type == CONV_ICONV) 4623 convert_setup(&input_conv, NULL, NULL); 4624 if (output_conv.vc_type == CONV_ICONV) 4625 convert_setup(&output_conv, NULL, NULL); 4626 4627 if (hIconvDLL != 0) 4628 FreeLibrary(hIconvDLL); 4629 if (hMsvcrtDLL != 0) 4630 FreeLibrary(hMsvcrtDLL); 4631 hIconvDLL = 0; 4632 hMsvcrtDLL = 0; 4633 } 4634 # endif /* DYNAMIC_ICONV */ 4635 # endif /* USE_ICONV */ 4636 4637 #endif /* FEAT_MBYTE */ 4638 4639 #if defined(FEAT_XIM) || defined(PROTO) 4640 4641 # if defined(FEAT_GUI_GTK) || defined(PROTO) 4642 static int xim_has_preediting INIT(= FALSE); /* IM current status */ 4643 4644 /* 4645 * Set preedit_start_col to the current cursor position. 4646 */ 4647 static void 4648 init_preedit_start_col(void) 4649 { 4650 if (State & CMDLINE) 4651 preedit_start_col = cmdline_getvcol_cursor(); 4652 else if (curwin != NULL && curwin->w_buffer != NULL) 4653 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); 4654 /* Prevent that preediting marks the buffer as changed. */ 4655 xim_changed_while_preediting = curbuf->b_changed; 4656 } 4657 4658 static int im_is_active = FALSE; /* IM is enabled for current mode */ 4659 static int preedit_is_active = FALSE; 4660 static int im_preedit_cursor = 0; /* cursor offset in characters */ 4661 static int im_preedit_trailing = 0; /* number of characters after cursor */ 4662 4663 static unsigned long im_commit_handler_id = 0; 4664 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; 4665 static unsigned int im_activatekey_state = 0; 4666 4667 void 4668 im_set_active(int active) 4669 { 4670 int was_active; 4671 4672 was_active = !!im_get_status(); 4673 im_is_active = (active && !p_imdisable); 4674 4675 if (im_is_active != was_active) 4676 xim_reset(); 4677 } 4678 4679 void 4680 xim_set_focus(int focus) 4681 { 4682 if (xic != NULL) 4683 { 4684 if (focus) 4685 gtk_im_context_focus_in(xic); 4686 else 4687 gtk_im_context_focus_out(xic); 4688 } 4689 } 4690 4691 void 4692 im_set_position(int row, int col) 4693 { 4694 if (xic != NULL) 4695 { 4696 GdkRectangle area; 4697 4698 area.x = FILL_X(col); 4699 area.y = FILL_Y(row); 4700 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); 4701 area.height = gui.char_height; 4702 4703 gtk_im_context_set_cursor_location(xic, &area); 4704 } 4705 } 4706 4707 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */ 4708 void 4709 xim_set_preedit(void) 4710 { 4711 im_set_position(gui.row, gui.col); 4712 } 4713 # endif 4714 4715 static void 4716 im_add_to_input(char_u *str, int len) 4717 { 4718 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */ 4719 if (input_conv.vc_type != CONV_NONE) 4720 { 4721 str = string_convert(&input_conv, str, &len); 4722 g_return_if_fail(str != NULL); 4723 } 4724 4725 add_to_input_buf_csi(str, len); 4726 4727 if (input_conv.vc_type != CONV_NONE) 4728 vim_free(str); 4729 4730 if (p_mh) /* blank out the pointer if necessary */ 4731 gui_mch_mousehide(TRUE); 4732 } 4733 4734 static void 4735 im_delete_preedit(void) 4736 { 4737 char_u bskey[] = {CSI, 'k', 'b'}; 4738 char_u delkey[] = {CSI, 'k', 'D'}; 4739 4740 if (State & NORMAL) 4741 { 4742 im_preedit_cursor = 0; 4743 return; 4744 } 4745 for (; im_preedit_cursor > 0; --im_preedit_cursor) 4746 add_to_input_buf(bskey, (int)sizeof(bskey)); 4747 4748 for (; im_preedit_trailing > 0; --im_preedit_trailing) 4749 add_to_input_buf(delkey, (int)sizeof(delkey)); 4750 } 4751 4752 /* 4753 * Move the cursor left by "num_move_back" characters. 4754 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for 4755 * these K_LEFT keys. 4756 */ 4757 static void 4758 im_correct_cursor(int num_move_back) 4759 { 4760 char_u backkey[] = {CSI, 'k', 'l'}; 4761 4762 if (State & NORMAL) 4763 return; 4764 # ifdef FEAT_RIGHTLEFT 4765 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) 4766 backkey[2] = 'r'; 4767 # endif 4768 for (; num_move_back > 0; --num_move_back) 4769 add_to_input_buf(backkey, (int)sizeof(backkey)); 4770 } 4771 4772 static int xim_expected_char = NUL; 4773 static int xim_ignored_char = FALSE; 4774 4775 /* 4776 * Update the mode and cursor while in an IM callback. 4777 */ 4778 static void 4779 im_show_info(void) 4780 { 4781 int old_vgetc_busy; 4782 4783 old_vgetc_busy = vgetc_busy; 4784 vgetc_busy = TRUE; 4785 showmode(); 4786 vgetc_busy = old_vgetc_busy; 4787 if ((State & NORMAL) || (State & INSERT)) 4788 setcursor(); 4789 out_flush(); 4790 } 4791 4792 /* 4793 * Callback invoked when the user finished preediting. 4794 * Put the final string into the input buffer. 4795 */ 4796 static void 4797 im_commit_cb(GtkIMContext *context UNUSED, 4798 const gchar *str, 4799 gpointer data UNUSED) 4800 { 4801 int slen = (int)STRLEN(str); 4802 int add_to_input = TRUE; 4803 int clen; 4804 int len = slen; 4805 int commit_with_preedit = TRUE; 4806 char_u *im_str; 4807 4808 #ifdef XIM_DEBUG 4809 xim_log("im_commit_cb(): %s\n", str); 4810 #endif 4811 4812 /* The imhangul module doesn't reset the preedit string before 4813 * committing. Call im_delete_preedit() to work around that. */ 4814 im_delete_preedit(); 4815 4816 /* Indicate that preediting has finished. */ 4817 if (preedit_start_col == MAXCOL) 4818 { 4819 init_preedit_start_col(); 4820 commit_with_preedit = FALSE; 4821 } 4822 4823 /* The thing which setting "preedit_start_col" to MAXCOL means that 4824 * "preedit_start_col" will be set forcedly when calling 4825 * preedit_changed_cb() next time. 4826 * "preedit_start_col" should not reset with MAXCOL on this part. Vim 4827 * is simulating the preediting by using add_to_input_str(). when 4828 * preedit begin immediately before committed, the typebuf is not 4829 * flushed to screen, then it can't get correct "preedit_start_col". 4830 * Thus, it should calculate the cells by adding cells of the committed 4831 * string. */ 4832 if (input_conv.vc_type != CONV_NONE) 4833 { 4834 im_str = string_convert(&input_conv, (char_u *)str, &len); 4835 g_return_if_fail(im_str != NULL); 4836 } 4837 else 4838 im_str = (char_u *)str; 4839 4840 clen = mb_string2cells(im_str, len); 4841 4842 if (input_conv.vc_type != CONV_NONE) 4843 vim_free(im_str); 4844 preedit_start_col += clen; 4845 4846 /* Is this a single character that matches a keypad key that's just 4847 * been pressed? If so, we don't want it to be entered as such - let 4848 * us carry on processing the raw keycode so that it may be used in 4849 * mappings as <kSomething>. */ 4850 if (xim_expected_char != NUL) 4851 { 4852 /* We're currently processing a keypad or other special key */ 4853 if (slen == 1 && str[0] == xim_expected_char) 4854 { 4855 /* It's a match - don't do it here */ 4856 xim_ignored_char = TRUE; 4857 add_to_input = FALSE; 4858 } 4859 else 4860 { 4861 /* Not a match */ 4862 xim_ignored_char = FALSE; 4863 } 4864 } 4865 4866 if (add_to_input) 4867 im_add_to_input((char_u *)str, slen); 4868 4869 /* Inserting chars while "im_is_active" is set does not cause a change of 4870 * buffer. When the chars are committed the buffer must be marked as 4871 * changed. */ 4872 if (!commit_with_preedit) 4873 preedit_start_col = MAXCOL; 4874 4875 /* This flag is used in changed() at next call. */ 4876 xim_changed_while_preediting = TRUE; 4877 4878 if (gtk_main_level() > 0) 4879 gtk_main_quit(); 4880 } 4881 4882 /* 4883 * Callback invoked after start to the preedit. 4884 */ 4885 static void 4886 im_preedit_start_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4887 { 4888 #ifdef XIM_DEBUG 4889 xim_log("im_preedit_start_cb()\n"); 4890 #endif 4891 4892 im_is_active = TRUE; 4893 preedit_is_active = TRUE; 4894 gui_update_cursor(TRUE, FALSE); 4895 im_show_info(); 4896 } 4897 4898 /* 4899 * Callback invoked after end to the preedit. 4900 */ 4901 static void 4902 im_preedit_end_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 4903 { 4904 #ifdef XIM_DEBUG 4905 xim_log("im_preedit_end_cb()\n"); 4906 #endif 4907 im_delete_preedit(); 4908 4909 /* Indicate that preediting has finished */ 4910 preedit_start_col = MAXCOL; 4911 xim_has_preediting = FALSE; 4912 4913 #if 0 4914 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was 4915 * switched off unintentionally. We now use preedit_is_active (added by 4916 * SungHyun Nam). */ 4917 im_is_active = FALSE; 4918 #endif 4919 preedit_is_active = FALSE; 4920 gui_update_cursor(TRUE, FALSE); 4921 im_show_info(); 4922 } 4923 4924 /* 4925 * Callback invoked after changes to the preedit string. If the preedit 4926 * string was empty before, remember the preedit start column so we know 4927 * where to apply feedback attributes. Delete the previous preedit string 4928 * if there was one, save the new preedit cursor offset, and put the new 4929 * string into the input buffer. 4930 * 4931 * TODO: The pragmatic "put into input buffer" approach used here has 4932 * several fundamental problems: 4933 * 4934 * - The characters in the preedit string are subject to remapping. 4935 * That's broken, only the finally committed string should be remapped. 4936 * 4937 * - There is a race condition involved: The retrieved value for the 4938 * current cursor position will be wrong if any unprocessed characters 4939 * are still queued in the input buffer. 4940 * 4941 * - Due to the lack of synchronization between the file buffer in memory 4942 * and any typed characters, it's practically impossible to implement the 4943 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM 4944 * modules for languages such as Thai are likely to rely on this feature 4945 * for proper operation. 4946 * 4947 * Conclusions: I think support for preediting needs to be moved to the 4948 * core parts of Vim. Ideally, until it has been committed, the preediting 4949 * string should only be displayed and not affect the buffer content at all. 4950 * The question how to deal with the synchronization issue still remains. 4951 * Circumventing the input buffer is probably not desirable. Anyway, I think 4952 * implementing "retrieve_surrounding" is the only hard problem. 4953 * 4954 * One way to solve all of this in a clean manner would be to queue all key 4955 * press/release events "as is" in the input buffer, and apply the IM filtering 4956 * at the receiving end of the queue. This, however, would have a rather large 4957 * impact on the code base. If there is an easy way to force processing of all 4958 * remaining input from within the "retrieve_surrounding" signal handler, this 4959 * might not be necessary. Gotta ask on vim-dev for opinions. 4960 */ 4961 static void 4962 im_preedit_changed_cb(GtkIMContext *context, gpointer data UNUSED) 4963 { 4964 char *preedit_string = NULL; 4965 int cursor_index = 0; 4966 int num_move_back = 0; 4967 char_u *str; 4968 char_u *p; 4969 int i; 4970 4971 gtk_im_context_get_preedit_string(context, 4972 &preedit_string, NULL, 4973 &cursor_index); 4974 4975 #ifdef XIM_DEBUG 4976 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); 4977 #endif 4978 4979 g_return_if_fail(preedit_string != NULL); /* just in case */ 4980 4981 /* If preedit_start_col is MAXCOL set it to the current cursor position. */ 4982 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') 4983 { 4984 xim_has_preediting = TRUE; 4985 4986 /* Urgh, this breaks if the input buffer isn't empty now */ 4987 init_preedit_start_col(); 4988 } 4989 else if (cursor_index == 0 && preedit_string[0] == '\0') 4990 { 4991 xim_has_preediting = FALSE; 4992 4993 /* If at the start position (after typing backspace) 4994 * preedit_start_col must be reset. */ 4995 preedit_start_col = MAXCOL; 4996 } 4997 4998 im_delete_preedit(); 4999 5000 /* 5001 * Compute the end of the preediting area: "preedit_end_col". 5002 * According to the documentation of gtk_im_context_get_preedit_string(), 5003 * the cursor_pos output argument returns the offset in bytes. This is 5004 * unfortunately not true -- real life shows the offset is in characters, 5005 * and the GTK+ source code agrees with me. Will file a bug later. 5006 */ 5007 if (preedit_start_col != MAXCOL) 5008 preedit_end_col = preedit_start_col; 5009 str = (char_u *)preedit_string; 5010 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) 5011 { 5012 int is_composing; 5013 5014 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); 5015 /* 5016 * These offsets are used as counters when generating <BS> and <Del> 5017 * to delete the preedit string. So don't count composing characters 5018 * unless 'delcombine' is enabled. 5019 */ 5020 if (!is_composing || p_deco) 5021 { 5022 if (i < cursor_index) 5023 ++im_preedit_cursor; 5024 else 5025 ++im_preedit_trailing; 5026 } 5027 if (!is_composing && i >= cursor_index) 5028 { 5029 /* This is essentially the same as im_preedit_trailing, except 5030 * composing characters are not counted even if p_deco is set. */ 5031 ++num_move_back; 5032 } 5033 if (preedit_start_col != MAXCOL) 5034 preedit_end_col += utf_ptr2cells(p); 5035 } 5036 5037 if (p > str) 5038 { 5039 im_add_to_input(str, (int)(p - str)); 5040 im_correct_cursor(num_move_back); 5041 } 5042 5043 g_free(preedit_string); 5044 5045 if (gtk_main_level() > 0) 5046 gtk_main_quit(); 5047 } 5048 5049 /* 5050 * Translate the Pango attributes at iter to Vim highlighting attributes. 5051 * Ignore attributes not supported by Vim highlighting. This shouldn't have 5052 * too much impact -- right now we handle even more attributes than necessary 5053 * for the IM modules I tested with. 5054 */ 5055 static int 5056 translate_pango_attributes(PangoAttrIterator *iter) 5057 { 5058 PangoAttribute *attr; 5059 int char_attr = HL_NORMAL; 5060 5061 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); 5062 if (attr != NULL && ((PangoAttrInt *)attr)->value 5063 != (int)PANGO_UNDERLINE_NONE) 5064 char_attr |= HL_UNDERLINE; 5065 5066 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); 5067 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) 5068 char_attr |= HL_BOLD; 5069 5070 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); 5071 if (attr != NULL && ((PangoAttrInt *)attr)->value 5072 != (int)PANGO_STYLE_NORMAL) 5073 char_attr |= HL_ITALIC; 5074 5075 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); 5076 if (attr != NULL) 5077 { 5078 const PangoColor *color = &((PangoAttrColor *)attr)->color; 5079 5080 /* Assume inverse if black background is requested */ 5081 if ((color->red | color->green | color->blue) == 0) 5082 char_attr |= HL_INVERSE; 5083 } 5084 5085 return char_attr; 5086 } 5087 5088 /* 5089 * Retrieve the highlighting attributes at column col in the preedit string. 5090 * Return -1 if not in preediting mode or if col is out of range. 5091 */ 5092 int 5093 im_get_feedback_attr(int col) 5094 { 5095 char *preedit_string = NULL; 5096 PangoAttrList *attr_list = NULL; 5097 int char_attr = -1; 5098 5099 if (xic == NULL) 5100 return char_attr; 5101 5102 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5103 5104 if (preedit_string != NULL && attr_list != NULL) 5105 { 5106 int idx; 5107 5108 /* Get the byte index as used by PangoAttrIterator */ 5109 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col) 5110 idx += utfc_ptr2len((char_u *)preedit_string + idx); 5111 5112 if (preedit_string[idx] != '\0') 5113 { 5114 PangoAttrIterator *iter; 5115 int start, end; 5116 5117 char_attr = HL_NORMAL; 5118 iter = pango_attr_list_get_iterator(attr_list); 5119 5120 /* Extract all relevant attributes from the list. */ 5121 do 5122 { 5123 pango_attr_iterator_range(iter, &start, &end); 5124 5125 if (idx >= start && idx < end) 5126 char_attr |= translate_pango_attributes(iter); 5127 } 5128 while (pango_attr_iterator_next(iter)); 5129 5130 pango_attr_iterator_destroy(iter); 5131 } 5132 } 5133 5134 if (attr_list != NULL) 5135 pango_attr_list_unref(attr_list); 5136 g_free(preedit_string); 5137 5138 return char_attr; 5139 } 5140 5141 void 5142 xim_init(void) 5143 { 5144 #ifdef XIM_DEBUG 5145 xim_log("xim_init()\n"); 5146 #endif 5147 5148 g_return_if_fail(gui.drawarea != NULL); 5149 #if GTK_CHECK_VERSION(3,0,0) 5150 g_return_if_fail(gtk_widget_get_window(gui.drawarea) != NULL); 5151 #else 5152 g_return_if_fail(gui.drawarea->window != NULL); 5153 #endif 5154 5155 xic = gtk_im_multicontext_new(); 5156 g_object_ref(xic); 5157 5158 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", 5159 G_CALLBACK(&im_commit_cb), NULL); 5160 g_signal_connect(G_OBJECT(xic), "preedit_changed", 5161 G_CALLBACK(&im_preedit_changed_cb), NULL); 5162 g_signal_connect(G_OBJECT(xic), "preedit_start", 5163 G_CALLBACK(&im_preedit_start_cb), NULL); 5164 g_signal_connect(G_OBJECT(xic), "preedit_end", 5165 G_CALLBACK(&im_preedit_end_cb), NULL); 5166 5167 #if GTK_CHECK_VERSION(3,0,0) 5168 gtk_im_context_set_client_window(xic, gtk_widget_get_window(gui.drawarea)); 5169 #else 5170 gtk_im_context_set_client_window(xic, gui.drawarea->window); 5171 #endif 5172 } 5173 5174 void 5175 im_shutdown(void) 5176 { 5177 #ifdef XIM_DEBUG 5178 xim_log("im_shutdown()\n"); 5179 #endif 5180 5181 if (xic != NULL) 5182 { 5183 gtk_im_context_focus_out(xic); 5184 g_object_unref(xic); 5185 xic = NULL; 5186 } 5187 im_is_active = FALSE; 5188 im_commit_handler_id = 0; 5189 preedit_start_col = MAXCOL; 5190 xim_has_preediting = FALSE; 5191 } 5192 5193 /* 5194 * Convert the string argument to keyval and state for GdkEventKey. 5195 * If str is valid return TRUE, otherwise FALSE. 5196 * 5197 * See 'imactivatekey' for documentation of the format. 5198 */ 5199 static int 5200 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) 5201 { 5202 const char *mods_end; 5203 unsigned tmp_keyval; 5204 unsigned tmp_state = 0; 5205 5206 mods_end = strrchr(str, '-'); 5207 mods_end = (mods_end != NULL) ? mods_end + 1 : str; 5208 5209 /* Parse modifier keys */ 5210 while (str < mods_end) 5211 switch (*str++) 5212 { 5213 case '-': break; 5214 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; 5215 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; 5216 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; 5217 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; 5218 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; 5219 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; 5220 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; 5221 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; 5222 default: 5223 return FALSE; 5224 } 5225 5226 tmp_keyval = gdk_keyval_from_name(str); 5227 5228 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) 5229 return FALSE; 5230 5231 if (keyval != NULL) 5232 *keyval = tmp_keyval; 5233 if (state != NULL) 5234 *state = tmp_state; 5235 5236 return TRUE; 5237 } 5238 5239 /* 5240 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an 5241 * empty string is also regarded as valid. 5242 * 5243 * Note: The numerical key value of p_imak is cached if it was valid; thus 5244 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever 5245 * 'imak' changes. This is currently the case but not obvious -- should 5246 * probably rename the function for clarity. 5247 */ 5248 int 5249 im_xim_isvalid_imactivate(void) 5250 { 5251 if (p_imak[0] == NUL) 5252 { 5253 im_activatekey_keyval = GDK_VoidSymbol; 5254 im_activatekey_state = 0; 5255 return TRUE; 5256 } 5257 5258 return im_string_to_keyval((const char *)p_imak, 5259 &im_activatekey_keyval, 5260 &im_activatekey_state); 5261 } 5262 5263 static void 5264 im_synthesize_keypress(unsigned int keyval, unsigned int state) 5265 { 5266 GdkEventKey *event; 5267 5268 # ifdef HAVE_GTK_MULTIHEAD 5269 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); 5270 # if GTK_CHECK_VERSION(3,0,0) 5271 g_object_ref(gtk_widget_get_window(gui.drawarea)); 5272 /* unreffed by gdk_event_free() */ 5273 # else 5274 g_object_ref(gui.drawarea->window); /* unreffed by gdk_event_free() */ 5275 # endif 5276 # else 5277 event = (GdkEventKey *)g_malloc0((gulong)sizeof(GdkEvent)); 5278 event->type = GDK_KEY_PRESS; 5279 # endif 5280 # if GTK_CHECK_VERSION(3,0,0) 5281 event->window = gtk_widget_get_window(gui.drawarea); 5282 # else 5283 event->window = gui.drawarea->window; 5284 # endif 5285 event->send_event = TRUE; 5286 event->time = GDK_CURRENT_TIME; 5287 event->state = state; 5288 event->keyval = keyval; 5289 event->hardware_keycode = /* needed for XIM */ 5290 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); 5291 event->length = 0; 5292 event->string = NULL; 5293 5294 gtk_im_context_filter_keypress(xic, event); 5295 5296 /* For consistency, also send the corresponding release event. */ 5297 event->type = GDK_KEY_RELEASE; 5298 event->send_event = FALSE; 5299 gtk_im_context_filter_keypress(xic, event); 5300 5301 # ifdef HAVE_GTK_MULTIHEAD 5302 gdk_event_free((GdkEvent *)event); 5303 # else 5304 g_free(event); 5305 # endif 5306 } 5307 5308 void 5309 xim_reset(void) 5310 { 5311 if (xic != NULL) 5312 { 5313 gtk_im_context_reset(xic); 5314 5315 if (p_imdisable) 5316 im_shutdown(); 5317 else 5318 { 5319 xim_set_focus(gui.in_focus); 5320 5321 # ifdef FEAT_EVAL 5322 if (p_imaf[0] != NUL) 5323 { 5324 char_u *argv[1]; 5325 5326 if (im_is_active) 5327 argv[0] = (char_u *)"1"; 5328 else 5329 argv[0] = (char_u *)"0"; 5330 (void)call_func_retnr(p_imaf, 1, argv, FALSE); 5331 } 5332 else 5333 # endif 5334 if (im_activatekey_keyval != GDK_VoidSymbol) 5335 { 5336 if (im_is_active) 5337 { 5338 g_signal_handler_block(xic, im_commit_handler_id); 5339 im_synthesize_keypress(im_activatekey_keyval, 5340 im_activatekey_state); 5341 g_signal_handler_unblock(xic, im_commit_handler_id); 5342 } 5343 } 5344 else 5345 { 5346 im_shutdown(); 5347 xim_init(); 5348 xim_set_focus(gui.in_focus); 5349 } 5350 } 5351 } 5352 5353 preedit_start_col = MAXCOL; 5354 xim_has_preediting = FALSE; 5355 } 5356 5357 int 5358 xim_queue_key_press_event(GdkEventKey *event, int down) 5359 { 5360 if (down) 5361 { 5362 /* 5363 * Workaround GTK2 XIM 'feature' that always converts keypad keys to 5364 * chars., even when not part of an IM sequence (ref. feature of 5365 * gdk/gdkkeyuni.c). 5366 * Flag any keypad keys that might represent a single char. 5367 * If this (on its own - i.e., not part of an IM sequence) is 5368 * committed while we're processing one of these keys, we can ignore 5369 * that commit and go ahead & process it ourselves. That way we can 5370 * still distinguish keypad keys for use in mappings. 5371 * Also add GDK_space to make <S-Space> work. 5372 */ 5373 switch (event->keyval) 5374 { 5375 case GDK_KP_Add: xim_expected_char = '+'; break; 5376 case GDK_KP_Subtract: xim_expected_char = '-'; break; 5377 case GDK_KP_Divide: xim_expected_char = '/'; break; 5378 case GDK_KP_Multiply: xim_expected_char = '*'; break; 5379 case GDK_KP_Decimal: xim_expected_char = '.'; break; 5380 case GDK_KP_Equal: xim_expected_char = '='; break; 5381 case GDK_KP_0: xim_expected_char = '0'; break; 5382 case GDK_KP_1: xim_expected_char = '1'; break; 5383 case GDK_KP_2: xim_expected_char = '2'; break; 5384 case GDK_KP_3: xim_expected_char = '3'; break; 5385 case GDK_KP_4: xim_expected_char = '4'; break; 5386 case GDK_KP_5: xim_expected_char = '5'; break; 5387 case GDK_KP_6: xim_expected_char = '6'; break; 5388 case GDK_KP_7: xim_expected_char = '7'; break; 5389 case GDK_KP_8: xim_expected_char = '8'; break; 5390 case GDK_KP_9: xim_expected_char = '9'; break; 5391 case GDK_space: xim_expected_char = ' '; break; 5392 default: xim_expected_char = NUL; 5393 } 5394 xim_ignored_char = FALSE; 5395 } 5396 5397 /* 5398 * When typing fFtT, XIM may be activated. Thus it must pass 5399 * gtk_im_context_filter_keypress() in Normal mode. 5400 * And while doing :sh too. 5401 */ 5402 if (xic != NULL && !p_imdisable 5403 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) 5404 { 5405 /* 5406 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is 5407 * always aware of the current status of IM, and can even emulate 5408 * the activation key for modules that don't support one. 5409 */ 5410 if (event->keyval == im_activatekey_keyval 5411 && (event->state & im_activatekey_state) == im_activatekey_state) 5412 { 5413 unsigned int state_mask; 5414 5415 /* Require the state of the 3 most used modifiers to match exactly. 5416 * Otherwise e.g. <S-C-space> would be unusable for other purposes 5417 * if the IM activate key is <S-space>. */ 5418 state_mask = im_activatekey_state; 5419 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK 5420 | (int)GDK_MOD1_MASK); 5421 5422 if ((event->state & state_mask) != im_activatekey_state) 5423 return FALSE; 5424 5425 /* Don't send it a second time on GDK_KEY_RELEASE. */ 5426 if (event->type != GDK_KEY_PRESS) 5427 return TRUE; 5428 5429 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE)) 5430 { 5431 im_set_active(FALSE); 5432 5433 /* ":lmap" mappings exists, toggle use of mappings. */ 5434 State ^= LANGMAP; 5435 if (State & LANGMAP) 5436 { 5437 curbuf->b_p_iminsert = B_IMODE_NONE; 5438 State &= ~LANGMAP; 5439 } 5440 else 5441 { 5442 curbuf->b_p_iminsert = B_IMODE_LMAP; 5443 State |= LANGMAP; 5444 } 5445 return TRUE; 5446 } 5447 5448 return gtk_im_context_filter_keypress(xic, event); 5449 } 5450 5451 /* Don't filter events through the IM context if IM isn't active 5452 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module 5453 * not doing anything before the activation key was sent. */ 5454 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) 5455 { 5456 int imresult = gtk_im_context_filter_keypress(xic, event); 5457 5458 /* Some XIM send following sequence: 5459 * 1. preedited string. 5460 * 2. committed string. 5461 * 3. line changed key. 5462 * 4. preedited string. 5463 * 5. remove preedited string. 5464 * if 3, Vim can't move back the above line for 5. 5465 * thus, this part should not parse the key. */ 5466 if (!imresult && preedit_start_col != MAXCOL 5467 && event->keyval == GDK_Return) 5468 { 5469 im_synthesize_keypress(GDK_Return, 0U); 5470 return FALSE; 5471 } 5472 5473 /* If XIM tried to commit a keypad key as a single char., 5474 * ignore it so we can use the keypad key 'raw', for mappings. */ 5475 if (xim_expected_char != NUL && xim_ignored_char) 5476 /* We had a keypad key, and XIM tried to thieve it */ 5477 return FALSE; 5478 5479 /* This is supposed to fix a problem with iBus, that space 5480 * characters don't work in input mode. */ 5481 xim_expected_char = NUL; 5482 5483 /* Normal processing */ 5484 return imresult; 5485 } 5486 } 5487 5488 return FALSE; 5489 } 5490 5491 int 5492 im_get_status(void) 5493 { 5494 # ifdef FEAT_EVAL 5495 if (p_imsf[0] != NUL) 5496 { 5497 int is_active; 5498 5499 /* FIXME: Don't execute user function in unsafe situation. */ 5500 if (exiting 5501 # ifdef FEAT_AUTOCMD 5502 || is_autocmd_blocked() 5503 # endif 5504 ) 5505 return FALSE; 5506 /* FIXME: :py print 'xxx' is shown duplicate result. 5507 * Use silent to avoid it. */ 5508 ++msg_silent; 5509 is_active = call_func_retnr(p_imsf, 0, NULL, FALSE); 5510 --msg_silent; 5511 return (is_active > 0); 5512 } 5513 # endif 5514 return im_is_active; 5515 } 5516 5517 int 5518 preedit_get_status(void) 5519 { 5520 return preedit_is_active; 5521 } 5522 5523 int 5524 im_is_preediting(void) 5525 { 5526 return xim_has_preediting; 5527 } 5528 5529 # else /* !FEAT_GUI_GTK */ 5530 5531 static int xim_is_active = FALSE; /* XIM should be active in the current 5532 mode */ 5533 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */ 5534 #ifdef FEAT_GUI_X11 5535 static XIMStyle input_style; 5536 static int status_area_enabled = TRUE; 5537 #endif 5538 5539 /* 5540 * Switch using XIM on/off. This is used by the code that changes "State". 5541 */ 5542 void 5543 im_set_active(int active) 5544 { 5545 if (xic == NULL) 5546 return; 5547 5548 /* If 'imdisable' is set, XIM is never active. */ 5549 if (p_imdisable) 5550 active = FALSE; 5551 #if !defined(FEAT_GUI_GTK) 5552 else if (input_style & XIMPreeditPosition) 5553 /* There is a problem in switching XIM off when preediting is used, 5554 * and it is not clear how this can be solved. For now, keep XIM on 5555 * all the time, like it was done in Vim 5.8. */ 5556 active = TRUE; 5557 #endif 5558 5559 /* Remember the active state, it is needed when Vim gets keyboard focus. */ 5560 xim_is_active = active; 5561 xim_set_preedit(); 5562 } 5563 5564 /* 5565 * Adjust using XIM for gaining or losing keyboard focus. Also called when 5566 * "xim_is_active" changes. 5567 */ 5568 void 5569 xim_set_focus(int focus) 5570 { 5571 if (xic == NULL) 5572 return; 5573 5574 /* 5575 * XIM only gets focus when the Vim window has keyboard focus and XIM has 5576 * been set active for the current mode. 5577 */ 5578 if (focus && xim_is_active) 5579 { 5580 if (!xim_has_focus) 5581 { 5582 xim_has_focus = TRUE; 5583 XSetICFocus(xic); 5584 } 5585 } 5586 else 5587 { 5588 if (xim_has_focus) 5589 { 5590 xim_has_focus = FALSE; 5591 XUnsetICFocus(xic); 5592 } 5593 } 5594 } 5595 5596 void 5597 im_set_position(int row UNUSED, int col UNUSED) 5598 { 5599 xim_set_preedit(); 5600 } 5601 5602 /* 5603 * Set the XIM to the current cursor position. 5604 */ 5605 void 5606 xim_set_preedit(void) 5607 { 5608 XVaNestedList attr_list; 5609 XRectangle spot_area; 5610 XPoint over_spot; 5611 int line_space; 5612 5613 if (xic == NULL) 5614 return; 5615 5616 xim_set_focus(TRUE); 5617 5618 if (!xim_has_focus) 5619 { 5620 /* hide XIM cursor */ 5621 over_spot.x = 0; 5622 over_spot.y = -100; /* arbitrary invisible position */ 5623 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5624 XNSpotLocation, 5625 &over_spot, 5626 NULL); 5627 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); 5628 XFree(attr_list); 5629 return; 5630 } 5631 5632 if (input_style & XIMPreeditPosition) 5633 { 5634 if (xim_fg_color == INVALCOLOR) 5635 { 5636 xim_fg_color = gui.def_norm_pixel; 5637 xim_bg_color = gui.def_back_pixel; 5638 } 5639 over_spot.x = TEXT_X(gui.col); 5640 over_spot.y = TEXT_Y(gui.row); 5641 spot_area.x = 0; 5642 spot_area.y = 0; 5643 spot_area.height = gui.char_height * Rows; 5644 spot_area.width = gui.char_width * Columns; 5645 line_space = gui.char_height; 5646 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5647 XNSpotLocation, &over_spot, 5648 XNForeground, (Pixel) xim_fg_color, 5649 XNBackground, (Pixel) xim_bg_color, 5650 XNArea, &spot_area, 5651 XNLineSpace, line_space, 5652 NULL); 5653 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) 5654 EMSG(_("E284: Cannot set IC values")); 5655 XFree(attr_list); 5656 } 5657 } 5658 5659 #if defined(FEAT_GUI_X11) 5660 static char e_xim[] = N_("E285: Failed to create input context"); 5661 #endif 5662 5663 #if defined(FEAT_GUI_X11) || defined(PROTO) 5664 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(sun) 5665 # define USE_X11R6_XIM 5666 # endif 5667 5668 static int xim_real_init(Window x11_window, Display *x11_display); 5669 5670 5671 #ifdef USE_X11R6_XIM 5672 static void xim_instantiate_cb(Display *display, XPointer client_data, XPointer call_data); 5673 static void xim_destroy_cb(XIM im, XPointer client_data, XPointer call_data); 5674 5675 static void 5676 xim_instantiate_cb( 5677 Display *display, 5678 XPointer client_data UNUSED, 5679 XPointer call_data UNUSED) 5680 { 5681 Window x11_window; 5682 Display *x11_display; 5683 5684 #ifdef XIM_DEBUG 5685 xim_log("xim_instantiate_cb()\n"); 5686 #endif 5687 5688 gui_get_x11_windis(&x11_window, &x11_display); 5689 if (display != x11_display) 5690 return; 5691 5692 xim_real_init(x11_window, x11_display); 5693 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5694 if (xic != NULL) 5695 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5696 xim_instantiate_cb, NULL); 5697 } 5698 5699 static void 5700 xim_destroy_cb( 5701 XIM im UNUSED, 5702 XPointer client_data UNUSED, 5703 XPointer call_data UNUSED) 5704 { 5705 Window x11_window; 5706 Display *x11_display; 5707 5708 #ifdef XIM_DEBUG 5709 xim_log("xim_destroy_cb()\n"); 5710 #endif 5711 gui_get_x11_windis(&x11_window, &x11_display); 5712 5713 xic = NULL; 5714 status_area_enabled = FALSE; 5715 5716 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5717 5718 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5719 xim_instantiate_cb, NULL); 5720 } 5721 #endif 5722 5723 void 5724 xim_init(void) 5725 { 5726 Window x11_window; 5727 Display *x11_display; 5728 5729 #ifdef XIM_DEBUG 5730 xim_log("xim_init()\n"); 5731 #endif 5732 5733 gui_get_x11_windis(&x11_window, &x11_display); 5734 5735 xic = NULL; 5736 5737 if (xim_real_init(x11_window, x11_display)) 5738 return; 5739 5740 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5741 5742 #ifdef USE_X11R6_XIM 5743 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 5744 xim_instantiate_cb, NULL); 5745 #endif 5746 } 5747 5748 static int 5749 xim_real_init(Window x11_window, Display *x11_display) 5750 { 5751 int i; 5752 char *p, 5753 *s, 5754 *ns, 5755 *end, 5756 tmp[1024]; 5757 #define IMLEN_MAX 40 5758 char buf[IMLEN_MAX + 7]; 5759 XIM xim = NULL; 5760 XIMStyles *xim_styles; 5761 XIMStyle this_input_style = 0; 5762 Boolean found; 5763 XPoint over_spot; 5764 XVaNestedList preedit_list, status_list; 5765 5766 input_style = 0; 5767 status_area_enabled = FALSE; 5768 5769 if (xic != NULL) 5770 return FALSE; 5771 5772 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) 5773 { 5774 strcpy(tmp, gui.rsrc_input_method); 5775 for (ns = s = tmp; ns != NULL && *s != NUL;) 5776 { 5777 s = (char *)skipwhite((char_u *)s); 5778 if (*s == NUL) 5779 break; 5780 if ((ns = end = strchr(s, ',')) == NULL) 5781 end = s + strlen(s); 5782 while (isspace(((char_u *)end)[-1])) 5783 end--; 5784 *end = NUL; 5785 5786 if (strlen(s) <= IMLEN_MAX) 5787 { 5788 strcpy(buf, "@im="); 5789 strcat(buf, s); 5790 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL 5791 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) 5792 != NULL) 5793 break; 5794 } 5795 5796 s = ns + 1; 5797 } 5798 } 5799 5800 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) 5801 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5802 5803 /* This is supposed to be useful to obtain characters through 5804 * XmbLookupString() without really using a XIM. */ 5805 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL 5806 && *p != NUL) 5807 xim = XOpenIM(x11_display, NULL, NULL, NULL); 5808 5809 if (xim == NULL) 5810 { 5811 /* Only give this message when verbose is set, because too many people 5812 * got this message when they didn't want to use a XIM. */ 5813 if (p_verbose > 0) 5814 { 5815 verbose_enter(); 5816 EMSG(_("E286: Failed to open input method")); 5817 verbose_leave(); 5818 } 5819 return FALSE; 5820 } 5821 5822 #ifdef USE_X11R6_XIM 5823 { 5824 XIMCallback destroy_cb; 5825 5826 destroy_cb.callback = xim_destroy_cb; 5827 destroy_cb.client_data = NULL; 5828 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) 5829 EMSG(_("E287: Warning: Could not set destroy callback to IM")); 5830 } 5831 #endif 5832 5833 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) 5834 { 5835 EMSG(_("E288: input method doesn't support any style")); 5836 XCloseIM(xim); 5837 return FALSE; 5838 } 5839 5840 found = False; 5841 strcpy(tmp, gui.rsrc_preedit_type_name); 5842 for (s = tmp; s && !found; ) 5843 { 5844 while (*s && isspace((unsigned char)*s)) 5845 s++; 5846 if (!*s) 5847 break; 5848 if ((ns = end = strchr(s, ',')) != 0) 5849 ns++; 5850 else 5851 end = s + strlen(s); 5852 while (isspace((unsigned char)*end)) 5853 end--; 5854 *end = '\0'; 5855 5856 if (!strcmp(s, "OverTheSpot")) 5857 this_input_style = (XIMPreeditPosition | XIMStatusArea); 5858 else if (!strcmp(s, "OffTheSpot")) 5859 this_input_style = (XIMPreeditArea | XIMStatusArea); 5860 else if (!strcmp(s, "Root")) 5861 this_input_style = (XIMPreeditNothing | XIMStatusNothing); 5862 5863 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5864 { 5865 if (this_input_style == xim_styles->supported_styles[i]) 5866 { 5867 found = True; 5868 break; 5869 } 5870 } 5871 if (!found) 5872 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 5873 { 5874 if ((xim_styles->supported_styles[i] & this_input_style) 5875 == (this_input_style & ~XIMStatusArea)) 5876 { 5877 this_input_style &= ~XIMStatusArea; 5878 found = True; 5879 break; 5880 } 5881 } 5882 5883 s = ns; 5884 } 5885 XFree(xim_styles); 5886 5887 if (!found) 5888 { 5889 /* Only give this message when verbose is set, because too many people 5890 * got this message when they didn't want to use a XIM. */ 5891 if (p_verbose > 0) 5892 { 5893 verbose_enter(); 5894 EMSG(_("E289: input method doesn't support my preedit type")); 5895 verbose_leave(); 5896 } 5897 XCloseIM(xim); 5898 return FALSE; 5899 } 5900 5901 over_spot.x = TEXT_X(gui.col); 5902 over_spot.y = TEXT_Y(gui.row); 5903 input_style = this_input_style; 5904 5905 /* A crash was reported when trying to pass gui.norm_font as XNFontSet, 5906 * thus that has been removed. Hopefully the default works... */ 5907 #ifdef FEAT_XFONTSET 5908 if (gui.fontset != NOFONTSET) 5909 { 5910 preedit_list = XVaCreateNestedList(0, 5911 XNSpotLocation, &over_spot, 5912 XNForeground, (Pixel)gui.def_norm_pixel, 5913 XNBackground, (Pixel)gui.def_back_pixel, 5914 XNFontSet, (XFontSet)gui.fontset, 5915 NULL); 5916 status_list = XVaCreateNestedList(0, 5917 XNForeground, (Pixel)gui.def_norm_pixel, 5918 XNBackground, (Pixel)gui.def_back_pixel, 5919 XNFontSet, (XFontSet)gui.fontset, 5920 NULL); 5921 } 5922 else 5923 #endif 5924 { 5925 preedit_list = XVaCreateNestedList(0, 5926 XNSpotLocation, &over_spot, 5927 XNForeground, (Pixel)gui.def_norm_pixel, 5928 XNBackground, (Pixel)gui.def_back_pixel, 5929 NULL); 5930 status_list = XVaCreateNestedList(0, 5931 XNForeground, (Pixel)gui.def_norm_pixel, 5932 XNBackground, (Pixel)gui.def_back_pixel, 5933 NULL); 5934 } 5935 5936 xic = XCreateIC(xim, 5937 XNInputStyle, input_style, 5938 XNClientWindow, x11_window, 5939 XNFocusWindow, gui.wid, 5940 XNPreeditAttributes, preedit_list, 5941 XNStatusAttributes, status_list, 5942 NULL); 5943 XFree(status_list); 5944 XFree(preedit_list); 5945 if (xic != NULL) 5946 { 5947 if (input_style & XIMStatusArea) 5948 { 5949 xim_set_status_area(); 5950 status_area_enabled = TRUE; 5951 } 5952 else 5953 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 5954 } 5955 else 5956 { 5957 EMSG(_(e_xim)); 5958 XCloseIM(xim); 5959 return FALSE; 5960 } 5961 5962 return TRUE; 5963 } 5964 5965 #endif /* FEAT_GUI_X11 */ 5966 5967 /* 5968 * Get IM status. When IM is on, return TRUE. Else return FALSE. 5969 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is 5970 * active, when not having focus XIM may still be active (e.g., when using a 5971 * tear-off menu item). 5972 */ 5973 int 5974 im_get_status(void) 5975 { 5976 return xim_has_focus; 5977 } 5978 5979 # endif /* !FEAT_GUI_GTK */ 5980 5981 # if !defined(FEAT_GUI_GTK) || defined(PROTO) 5982 /* 5983 * Set up the status area. 5984 * 5985 * This should use a separate Widget, but that seems not possible, because 5986 * preedit_area and status_area should be set to the same window as for the 5987 * text input. Unfortunately this means the status area pollutes the text 5988 * window... 5989 */ 5990 void 5991 xim_set_status_area(void) 5992 { 5993 XVaNestedList preedit_list = 0, status_list = 0, list = 0; 5994 XRectangle pre_area, status_area; 5995 5996 if (xic == NULL) 5997 return; 5998 5999 if (input_style & XIMStatusArea) 6000 { 6001 if (input_style & XIMPreeditArea) 6002 { 6003 XRectangle *needed_rect; 6004 6005 /* to get status_area width */ 6006 status_list = XVaCreateNestedList(0, XNAreaNeeded, 6007 &needed_rect, NULL); 6008 XGetICValues(xic, XNStatusAttributes, status_list, NULL); 6009 XFree(status_list); 6010 6011 status_area.width = needed_rect->width; 6012 } 6013 else 6014 status_area.width = gui.char_width * Columns; 6015 6016 status_area.x = 0; 6017 status_area.y = gui.char_height * Rows + gui.border_offset; 6018 if (gui.which_scrollbars[SBAR_BOTTOM]) 6019 status_area.y += gui.scrollbar_height; 6020 #ifdef FEAT_MENU 6021 if (gui.menu_is_active) 6022 status_area.y += gui.menu_height; 6023 #endif 6024 status_area.height = gui.char_height; 6025 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); 6026 } 6027 else 6028 { 6029 status_area.x = 0; 6030 status_area.y = gui.char_height * Rows + gui.border_offset; 6031 if (gui.which_scrollbars[SBAR_BOTTOM]) 6032 status_area.y += gui.scrollbar_height; 6033 #ifdef FEAT_MENU 6034 if (gui.menu_is_active) 6035 status_area.y += gui.menu_height; 6036 #endif 6037 status_area.width = 0; 6038 status_area.height = gui.char_height; 6039 } 6040 6041 if (input_style & XIMPreeditArea) /* off-the-spot */ 6042 { 6043 pre_area.x = status_area.x + status_area.width; 6044 pre_area.y = gui.char_height * Rows + gui.border_offset; 6045 pre_area.width = gui.char_width * Columns - pre_area.x; 6046 if (gui.which_scrollbars[SBAR_BOTTOM]) 6047 pre_area.y += gui.scrollbar_height; 6048 #ifdef FEAT_MENU 6049 if (gui.menu_is_active) 6050 pre_area.y += gui.menu_height; 6051 #endif 6052 pre_area.height = gui.char_height; 6053 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6054 } 6055 else if (input_style & XIMPreeditPosition) /* over-the-spot */ 6056 { 6057 pre_area.x = 0; 6058 pre_area.y = 0; 6059 pre_area.height = gui.char_height * Rows; 6060 pre_area.width = gui.char_width * Columns; 6061 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6062 } 6063 6064 if (preedit_list && status_list) 6065 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6066 XNStatusAttributes, status_list, NULL); 6067 else if (preedit_list) 6068 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6069 NULL); 6070 else if (status_list) 6071 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, 6072 NULL); 6073 else 6074 list = NULL; 6075 6076 if (list) 6077 { 6078 XSetICValues(xic, XNVaNestedList, list, NULL); 6079 XFree(list); 6080 } 6081 if (status_list) 6082 XFree(status_list); 6083 if (preedit_list) 6084 XFree(preedit_list); 6085 } 6086 6087 int 6088 xim_get_status_area_height(void) 6089 { 6090 if (status_area_enabled) 6091 return gui.char_height; 6092 return 0; 6093 } 6094 # endif 6095 6096 #endif /* FEAT_XIM */ 6097 6098 #if defined(FEAT_MBYTE) || defined(PROTO) 6099 6100 /* 6101 * Setup "vcp" for conversion from "from" to "to". 6102 * The names must have been made canonical with enc_canonize(). 6103 * vcp->vc_type must have been initialized to CONV_NONE. 6104 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 6105 * instead). 6106 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 6107 * Return FAIL when conversion is not supported, OK otherwise. 6108 */ 6109 int 6110 convert_setup(vimconv_T *vcp, char_u *from, char_u *to) 6111 { 6112 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 6113 } 6114 6115 /* 6116 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 6117 * "from" unicode charsets be considered utf-8. Same for "to". 6118 */ 6119 int 6120 convert_setup_ext( 6121 vimconv_T *vcp, 6122 char_u *from, 6123 int from_unicode_is_utf8, 6124 char_u *to, 6125 int to_unicode_is_utf8) 6126 { 6127 int from_prop; 6128 int to_prop; 6129 int from_is_utf8; 6130 int to_is_utf8; 6131 6132 /* Reset to no conversion. */ 6133 # ifdef USE_ICONV 6134 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 6135 iconv_close(vcp->vc_fd); 6136 # endif 6137 vcp->vc_type = CONV_NONE; 6138 vcp->vc_factor = 1; 6139 vcp->vc_fail = FALSE; 6140 6141 /* No conversion when one of the names is empty or they are equal. */ 6142 if (from == NULL || *from == NUL || to == NULL || *to == NUL 6143 || STRCMP(from, to) == 0) 6144 return OK; 6145 6146 from_prop = enc_canon_props(from); 6147 to_prop = enc_canon_props(to); 6148 if (from_unicode_is_utf8) 6149 from_is_utf8 = from_prop & ENC_UNICODE; 6150 else 6151 from_is_utf8 = from_prop == ENC_UNICODE; 6152 if (to_unicode_is_utf8) 6153 to_is_utf8 = to_prop & ENC_UNICODE; 6154 else 6155 to_is_utf8 = to_prop == ENC_UNICODE; 6156 6157 if ((from_prop & ENC_LATIN1) && to_is_utf8) 6158 { 6159 /* Internal latin1 -> utf-8 conversion. */ 6160 vcp->vc_type = CONV_TO_UTF8; 6161 vcp->vc_factor = 2; /* up to twice as long */ 6162 } 6163 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 6164 { 6165 /* Internal latin9 -> utf-8 conversion. */ 6166 vcp->vc_type = CONV_9_TO_UTF8; 6167 vcp->vc_factor = 3; /* up to three as long (euro sign) */ 6168 } 6169 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 6170 { 6171 /* Internal utf-8 -> latin1 conversion. */ 6172 vcp->vc_type = CONV_TO_LATIN1; 6173 } 6174 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 6175 { 6176 /* Internal utf-8 -> latin9 conversion. */ 6177 vcp->vc_type = CONV_TO_LATIN9; 6178 } 6179 #ifdef WIN3264 6180 /* Win32-specific codepage <-> codepage conversion without iconv. */ 6181 else if ((from_is_utf8 || encname2codepage(from) > 0) 6182 && (to_is_utf8 || encname2codepage(to) > 0)) 6183 { 6184 vcp->vc_type = CONV_CODEPAGE; 6185 vcp->vc_factor = 2; /* up to twice as long */ 6186 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 6187 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 6188 } 6189 #endif 6190 #ifdef MACOS_X 6191 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 6192 { 6193 vcp->vc_type = CONV_MAC_LATIN1; 6194 } 6195 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 6196 { 6197 vcp->vc_type = CONV_MAC_UTF8; 6198 vcp->vc_factor = 2; /* up to twice as long */ 6199 } 6200 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 6201 { 6202 vcp->vc_type = CONV_LATIN1_MAC; 6203 } 6204 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 6205 { 6206 vcp->vc_type = CONV_UTF8_MAC; 6207 } 6208 #endif 6209 # ifdef USE_ICONV 6210 else 6211 { 6212 /* Use iconv() for conversion. */ 6213 vcp->vc_fd = (iconv_t)my_iconv_open( 6214 to_is_utf8 ? (char_u *)"utf-8" : to, 6215 from_is_utf8 ? (char_u *)"utf-8" : from); 6216 if (vcp->vc_fd != (iconv_t)-1) 6217 { 6218 vcp->vc_type = CONV_ICONV; 6219 vcp->vc_factor = 4; /* could be longer too... */ 6220 } 6221 } 6222 # endif 6223 if (vcp->vc_type == CONV_NONE) 6224 return FAIL; 6225 6226 return OK; 6227 } 6228 6229 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \ 6230 || defined(PROTO) 6231 /* 6232 * Do conversion on typed input characters in-place. 6233 * The input and output are not NUL terminated! 6234 * Returns the length after conversion. 6235 */ 6236 int 6237 convert_input(char_u *ptr, int len, int maxlen) 6238 { 6239 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 6240 } 6241 #endif 6242 6243 /* 6244 * Like convert_input(), but when there is an incomplete byte sequence at the 6245 * end return that as an allocated string in "restp" and set "*restlenp" to 6246 * the length. If "restp" is NULL it is not used. 6247 */ 6248 int 6249 convert_input_safe( 6250 char_u *ptr, 6251 int len, 6252 int maxlen, 6253 char_u **restp, 6254 int *restlenp) 6255 { 6256 char_u *d; 6257 int dlen = len; 6258 int unconvertlen = 0; 6259 6260 d = string_convert_ext(&input_conv, ptr, &dlen, 6261 restp == NULL ? NULL : &unconvertlen); 6262 if (d != NULL) 6263 { 6264 if (dlen <= maxlen) 6265 { 6266 if (unconvertlen > 0) 6267 { 6268 /* Move the unconverted characters to allocated memory. */ 6269 *restp = alloc(unconvertlen); 6270 if (*restp != NULL) 6271 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 6272 *restlenp = unconvertlen; 6273 } 6274 mch_memmove(ptr, d, dlen); 6275 } 6276 else 6277 /* result is too long, keep the unconverted text (the caller must 6278 * have done something wrong!) */ 6279 dlen = len; 6280 vim_free(d); 6281 } 6282 return dlen; 6283 } 6284 6285 /* 6286 * Convert text "ptr[*lenp]" according to "vcp". 6287 * Returns the result in allocated memory and sets "*lenp". 6288 * When "lenp" is NULL, use NUL terminated strings. 6289 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 6290 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 6291 */ 6292 char_u * 6293 string_convert( 6294 vimconv_T *vcp, 6295 char_u *ptr, 6296 int *lenp) 6297 { 6298 return string_convert_ext(vcp, ptr, lenp, NULL); 6299 } 6300 6301 /* 6302 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 6303 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 6304 * set to the number of remaining bytes. 6305 */ 6306 char_u * 6307 string_convert_ext( 6308 vimconv_T *vcp, 6309 char_u *ptr, 6310 int *lenp, 6311 int *unconvlenp) 6312 { 6313 char_u *retval = NULL; 6314 char_u *d; 6315 int len; 6316 int i; 6317 int l; 6318 int c; 6319 6320 if (lenp == NULL) 6321 len = (int)STRLEN(ptr); 6322 else 6323 len = *lenp; 6324 if (len == 0) 6325 return vim_strsave((char_u *)""); 6326 6327 switch (vcp->vc_type) 6328 { 6329 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */ 6330 retval = alloc(len * 2 + 1); 6331 if (retval == NULL) 6332 break; 6333 d = retval; 6334 for (i = 0; i < len; ++i) 6335 { 6336 c = ptr[i]; 6337 if (c < 0x80) 6338 *d++ = c; 6339 else 6340 { 6341 *d++ = 0xc0 + ((unsigned)c >> 6); 6342 *d++ = 0x80 + (c & 0x3f); 6343 } 6344 } 6345 *d = NUL; 6346 if (lenp != NULL) 6347 *lenp = (int)(d - retval); 6348 break; 6349 6350 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */ 6351 retval = alloc(len * 3 + 1); 6352 if (retval == NULL) 6353 break; 6354 d = retval; 6355 for (i = 0; i < len; ++i) 6356 { 6357 c = ptr[i]; 6358 switch (c) 6359 { 6360 case 0xa4: c = 0x20ac; break; /* euro */ 6361 case 0xa6: c = 0x0160; break; /* S hat */ 6362 case 0xa8: c = 0x0161; break; /* S -hat */ 6363 case 0xb4: c = 0x017d; break; /* Z hat */ 6364 case 0xb8: c = 0x017e; break; /* Z -hat */ 6365 case 0xbc: c = 0x0152; break; /* OE */ 6366 case 0xbd: c = 0x0153; break; /* oe */ 6367 case 0xbe: c = 0x0178; break; /* Y */ 6368 } 6369 d += utf_char2bytes(c, d); 6370 } 6371 *d = NUL; 6372 if (lenp != NULL) 6373 *lenp = (int)(d - retval); 6374 break; 6375 6376 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */ 6377 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */ 6378 retval = alloc(len + 1); 6379 if (retval == NULL) 6380 break; 6381 d = retval; 6382 for (i = 0; i < len; ++i) 6383 { 6384 l = utf_ptr2len_len(ptr + i, len - i); 6385 if (l == 0) 6386 *d++ = NUL; 6387 else if (l == 1) 6388 { 6389 int l_w = utf8len_tab_zero[ptr[i]]; 6390 6391 if (l_w == 0) 6392 { 6393 /* Illegal utf-8 byte cannot be converted */ 6394 vim_free(retval); 6395 return NULL; 6396 } 6397 if (unconvlenp != NULL && l_w > len - i) 6398 { 6399 /* Incomplete sequence at the end. */ 6400 *unconvlenp = len - i; 6401 break; 6402 } 6403 *d++ = ptr[i]; 6404 } 6405 else 6406 { 6407 c = utf_ptr2char(ptr + i); 6408 if (vcp->vc_type == CONV_TO_LATIN9) 6409 switch (c) 6410 { 6411 case 0x20ac: c = 0xa4; break; /* euro */ 6412 case 0x0160: c = 0xa6; break; /* S hat */ 6413 case 0x0161: c = 0xa8; break; /* S -hat */ 6414 case 0x017d: c = 0xb4; break; /* Z hat */ 6415 case 0x017e: c = 0xb8; break; /* Z -hat */ 6416 case 0x0152: c = 0xbc; break; /* OE */ 6417 case 0x0153: c = 0xbd; break; /* oe */ 6418 case 0x0178: c = 0xbe; break; /* Y */ 6419 case 0xa4: 6420 case 0xa6: 6421 case 0xa8: 6422 case 0xb4: 6423 case 0xb8: 6424 case 0xbc: 6425 case 0xbd: 6426 case 0xbe: c = 0x100; break; /* not in latin9 */ 6427 } 6428 if (!utf_iscomposing(c)) /* skip composing chars */ 6429 { 6430 if (c < 0x100) 6431 *d++ = c; 6432 else if (vcp->vc_fail) 6433 { 6434 vim_free(retval); 6435 return NULL; 6436 } 6437 else 6438 { 6439 *d++ = 0xbf; 6440 if (utf_char2cells(c) > 1) 6441 *d++ = '?'; 6442 } 6443 } 6444 i += l - 1; 6445 } 6446 } 6447 *d = NUL; 6448 if (lenp != NULL) 6449 *lenp = (int)(d - retval); 6450 break; 6451 6452 # ifdef MACOS_CONVERT 6453 case CONV_MAC_LATIN1: 6454 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6455 'm', 'l', unconvlenp); 6456 break; 6457 6458 case CONV_LATIN1_MAC: 6459 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6460 'l', 'm', unconvlenp); 6461 break; 6462 6463 case CONV_MAC_UTF8: 6464 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6465 'm', 'u', unconvlenp); 6466 break; 6467 6468 case CONV_UTF8_MAC: 6469 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6470 'u', 'm', unconvlenp); 6471 break; 6472 # endif 6473 6474 # ifdef USE_ICONV 6475 case CONV_ICONV: /* conversion with output_conv.vc_fd */ 6476 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 6477 break; 6478 # endif 6479 # ifdef WIN3264 6480 case CONV_CODEPAGE: /* codepage -> codepage */ 6481 { 6482 int retlen; 6483 int tmp_len; 6484 short_u *tmp; 6485 6486 /* 1. codepage/UTF-8 -> ucs-2. */ 6487 if (vcp->vc_cpfrom == 0) 6488 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 6489 else 6490 { 6491 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 6492 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 6493 (char *)ptr, len, 0, 0); 6494 if (tmp_len == 0 6495 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 6496 { 6497 if (lenp != NULL) 6498 *lenp = 0; 6499 if (unconvlenp != NULL) 6500 *unconvlenp = len; 6501 retval = alloc(1); 6502 if (retval) 6503 retval[0] = NUL; 6504 return retval; 6505 } 6506 } 6507 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len); 6508 if (tmp == NULL) 6509 break; 6510 if (vcp->vc_cpfrom == 0) 6511 utf8_to_utf16(ptr, len, tmp, unconvlenp); 6512 else 6513 MultiByteToWideChar(vcp->vc_cpfrom, 0, 6514 (char *)ptr, len, tmp, tmp_len); 6515 6516 /* 2. ucs-2 -> codepage/UTF-8. */ 6517 if (vcp->vc_cpto == 0) 6518 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 6519 else 6520 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 6521 tmp, tmp_len, 0, 0, 0, 0); 6522 retval = alloc(retlen + 1); 6523 if (retval != NULL) 6524 { 6525 if (vcp->vc_cpto == 0) 6526 utf16_to_utf8(tmp, tmp_len, retval); 6527 else 6528 WideCharToMultiByte(vcp->vc_cpto, 0, 6529 tmp, tmp_len, 6530 (char *)retval, retlen, 0, 0); 6531 retval[retlen] = NUL; 6532 if (lenp != NULL) 6533 *lenp = retlen; 6534 } 6535 vim_free(tmp); 6536 break; 6537 } 6538 # endif 6539 } 6540 6541 return retval; 6542 } 6543 #endif 6544