1 /* vi:set ts=8 sts=4 sw=4 noet: 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 # if !defined(HAVE_MBLEN) && !defined(mblen) 110 # define mblen _Xmblen 111 # endif 112 #endif 113 114 #if defined(FEAT_GUI_GTK) && defined(FEAT_XIM) 115 # if GTK_CHECK_VERSION(3,0,0) 116 # include <gdk/gdkkeysyms-compat.h> 117 # else 118 # include <gdk/gdkkeysyms.h> 119 # endif 120 # ifdef WIN3264 121 # include <gdk/gdkwin32.h> 122 # else 123 # include <gdk/gdkx.h> 124 # endif 125 #endif 126 127 #ifdef HAVE_WCHAR_H 128 # include <wchar.h> 129 #endif 130 131 #if 0 132 /* This has been disabled, because several people reported problems with the 133 * wcwidth() and iswprint() library functions, esp. for Hebrew. */ 134 # ifdef __STDC_ISO_10646__ 135 # define USE_WCHAR_FUNCTIONS 136 # endif 137 #endif 138 139 #if defined(FEAT_MBYTE) || defined(PROTO) 140 141 static int dbcs_char2len(int c); 142 static int dbcs_char2bytes(int c, char_u *buf); 143 static int dbcs_ptr2len(char_u *p); 144 static int dbcs_ptr2len_len(char_u *p, int size); 145 static int utf_ptr2cells_len(char_u *p, int size); 146 static int dbcs_char2cells(int c); 147 static int dbcs_ptr2cells_len(char_u *p, int size); 148 static int dbcs_ptr2char(char_u *p); 149 150 /* 151 * Lookup table to quickly get the length in bytes of a UTF-8 character from 152 * the first byte of a UTF-8 string. 153 * Bytes which are illegal when used as the first byte have a 1. 154 * The NUL byte has length 1. 155 */ 156 static char utf8len_tab[256] = 157 { 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 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 164 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, 165 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, 166 }; 167 168 /* 169 * Like utf8len_tab above, but using a zero for illegal lead bytes. 170 */ 171 static char utf8len_tab_zero[256] = 172 { 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 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 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 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, 179 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, 180 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, 181 }; 182 183 /* 184 * XIM often causes trouble. Define XIM_DEBUG to get a log of XIM callbacks 185 * in the "xim.log" file. 186 */ 187 /* #define XIM_DEBUG */ 188 #ifdef XIM_DEBUG 189 static void 190 xim_log(char *s, ...) 191 { 192 va_list arglist; 193 static FILE *fd = NULL; 194 195 if (fd == (FILE *)-1) 196 return; 197 if (fd == NULL) 198 { 199 fd = mch_fopen("xim.log", "w"); 200 if (fd == NULL) 201 { 202 emsg("Cannot open xim.log"); 203 fd = (FILE *)-1; 204 return; 205 } 206 } 207 208 va_start(arglist, s); 209 vfprintf(fd, s, arglist); 210 va_end(arglist); 211 } 212 #endif 213 214 #endif 215 216 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 217 /* 218 * Canonical encoding names and their properties. 219 * "iso-8859-n" is handled by enc_canonize() directly. 220 */ 221 static struct 222 { char *name; int prop; int codepage;} 223 enc_canon_table[] = 224 { 225 #define IDX_LATIN_1 0 226 {"latin1", ENC_8BIT + ENC_LATIN1, 1252}, 227 #define IDX_ISO_2 1 228 {"iso-8859-2", ENC_8BIT, 0}, 229 #define IDX_ISO_3 2 230 {"iso-8859-3", ENC_8BIT, 0}, 231 #define IDX_ISO_4 3 232 {"iso-8859-4", ENC_8BIT, 0}, 233 #define IDX_ISO_5 4 234 {"iso-8859-5", ENC_8BIT, 0}, 235 #define IDX_ISO_6 5 236 {"iso-8859-6", ENC_8BIT, 0}, 237 #define IDX_ISO_7 6 238 {"iso-8859-7", ENC_8BIT, 0}, 239 #define IDX_ISO_8 7 240 {"iso-8859-8", ENC_8BIT, 0}, 241 #define IDX_ISO_9 8 242 {"iso-8859-9", ENC_8BIT, 0}, 243 #define IDX_ISO_10 9 244 {"iso-8859-10", ENC_8BIT, 0}, 245 #define IDX_ISO_11 10 246 {"iso-8859-11", ENC_8BIT, 0}, 247 #define IDX_ISO_13 11 248 {"iso-8859-13", ENC_8BIT, 0}, 249 #define IDX_ISO_14 12 250 {"iso-8859-14", ENC_8BIT, 0}, 251 #define IDX_ISO_15 13 252 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0}, 253 #define IDX_KOI8_R 14 254 {"koi8-r", ENC_8BIT, 0}, 255 #define IDX_KOI8_U 15 256 {"koi8-u", ENC_8BIT, 0}, 257 #define IDX_UTF8 16 258 {"utf-8", ENC_UNICODE, 0}, 259 #define IDX_UCS2 17 260 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0}, 261 #define IDX_UCS2LE 18 262 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0}, 263 #define IDX_UTF16 19 264 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0}, 265 #define IDX_UTF16LE 20 266 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0}, 267 #define IDX_UCS4 21 268 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0}, 269 #define IDX_UCS4LE 22 270 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0}, 271 272 /* For debugging DBCS encoding on Unix. */ 273 #define IDX_DEBUG 23 274 {"debug", ENC_DBCS, DBCS_DEBUG}, 275 #define IDX_EUC_JP 24 276 {"euc-jp", ENC_DBCS, DBCS_JPNU}, 277 #define IDX_SJIS 25 278 {"sjis", ENC_DBCS, DBCS_JPN}, 279 #define IDX_EUC_KR 26 280 {"euc-kr", ENC_DBCS, DBCS_KORU}, 281 #define IDX_EUC_CN 27 282 {"euc-cn", ENC_DBCS, DBCS_CHSU}, 283 #define IDX_EUC_TW 28 284 {"euc-tw", ENC_DBCS, DBCS_CHTU}, 285 #define IDX_BIG5 29 286 {"big5", ENC_DBCS, DBCS_CHT}, 287 288 /* MS-DOS and MS-Windows codepages are included here, so that they can be 289 * used on Unix too. Most of them are similar to ISO-8859 encodings, but 290 * not exactly the same. */ 291 #define IDX_CP437 30 292 {"cp437", ENC_8BIT, 437}, /* like iso-8859-1 */ 293 #define IDX_CP737 31 294 {"cp737", ENC_8BIT, 737}, /* like iso-8859-7 */ 295 #define IDX_CP775 32 296 {"cp775", ENC_8BIT, 775}, /* Baltic */ 297 #define IDX_CP850 33 298 {"cp850", ENC_8BIT, 850}, /* like iso-8859-4 */ 299 #define IDX_CP852 34 300 {"cp852", ENC_8BIT, 852}, /* like iso-8859-1 */ 301 #define IDX_CP855 35 302 {"cp855", ENC_8BIT, 855}, /* like iso-8859-2 */ 303 #define IDX_CP857 36 304 {"cp857", ENC_8BIT, 857}, /* like iso-8859-5 */ 305 #define IDX_CP860 37 306 {"cp860", ENC_8BIT, 860}, /* like iso-8859-9 */ 307 #define IDX_CP861 38 308 {"cp861", ENC_8BIT, 861}, /* like iso-8859-1 */ 309 #define IDX_CP862 39 310 {"cp862", ENC_8BIT, 862}, /* like iso-8859-1 */ 311 #define IDX_CP863 40 312 {"cp863", ENC_8BIT, 863}, /* like iso-8859-8 */ 313 #define IDX_CP865 41 314 {"cp865", ENC_8BIT, 865}, /* like iso-8859-1 */ 315 #define IDX_CP866 42 316 {"cp866", ENC_8BIT, 866}, /* like iso-8859-5 */ 317 #define IDX_CP869 43 318 {"cp869", ENC_8BIT, 869}, /* like iso-8859-7 */ 319 #define IDX_CP874 44 320 {"cp874", ENC_8BIT, 874}, /* Thai */ 321 #define IDX_CP932 45 322 {"cp932", ENC_DBCS, DBCS_JPN}, 323 #define IDX_CP936 46 324 {"cp936", ENC_DBCS, DBCS_CHS}, 325 #define IDX_CP949 47 326 {"cp949", ENC_DBCS, DBCS_KOR}, 327 #define IDX_CP950 48 328 {"cp950", ENC_DBCS, DBCS_CHT}, 329 #define IDX_CP1250 49 330 {"cp1250", ENC_8BIT, 1250}, /* Czech, Polish, etc. */ 331 #define IDX_CP1251 50 332 {"cp1251", ENC_8BIT, 1251}, /* Cyrillic */ 333 /* cp1252 is considered to be equal to latin1 */ 334 #define IDX_CP1253 51 335 {"cp1253", ENC_8BIT, 1253}, /* Greek */ 336 #define IDX_CP1254 52 337 {"cp1254", ENC_8BIT, 1254}, /* Turkish */ 338 #define IDX_CP1255 53 339 {"cp1255", ENC_8BIT, 1255}, /* Hebrew */ 340 #define IDX_CP1256 54 341 {"cp1256", ENC_8BIT, 1256}, /* Arabic */ 342 #define IDX_CP1257 55 343 {"cp1257", ENC_8BIT, 1257}, /* Baltic */ 344 #define IDX_CP1258 56 345 {"cp1258", ENC_8BIT, 1258}, /* Vietnamese */ 346 347 #define IDX_MACROMAN 57 348 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, /* Mac OS */ 349 #define IDX_DECMCS 58 350 {"dec-mcs", ENC_8BIT, 0}, /* DEC MCS */ 351 #define IDX_HPROMAN8 59 352 {"hp-roman8", ENC_8BIT, 0}, /* HP Roman8 */ 353 #define IDX_COUNT 60 354 }; 355 356 /* 357 * Aliases for encoding names. 358 */ 359 static struct 360 { char *name; int canon;} 361 enc_alias_table[] = 362 { 363 {"ansi", IDX_LATIN_1}, 364 {"iso-8859-1", IDX_LATIN_1}, 365 {"latin2", IDX_ISO_2}, 366 {"latin3", IDX_ISO_3}, 367 {"latin4", IDX_ISO_4}, 368 {"cyrillic", IDX_ISO_5}, 369 {"arabic", IDX_ISO_6}, 370 {"greek", IDX_ISO_7}, 371 #ifdef WIN3264 372 {"hebrew", IDX_CP1255}, 373 #else 374 {"hebrew", IDX_ISO_8}, 375 #endif 376 {"latin5", IDX_ISO_9}, 377 {"turkish", IDX_ISO_9}, /* ? */ 378 {"latin6", IDX_ISO_10}, 379 {"nordic", IDX_ISO_10}, /* ? */ 380 {"thai", IDX_ISO_11}, /* ? */ 381 {"latin7", IDX_ISO_13}, 382 {"latin8", IDX_ISO_14}, 383 {"latin9", IDX_ISO_15}, 384 {"utf8", IDX_UTF8}, 385 {"unicode", IDX_UCS2}, 386 {"ucs2", IDX_UCS2}, 387 {"ucs2be", IDX_UCS2}, 388 {"ucs-2be", IDX_UCS2}, 389 {"ucs2le", IDX_UCS2LE}, 390 {"utf16", IDX_UTF16}, 391 {"utf16be", IDX_UTF16}, 392 {"utf-16be", IDX_UTF16}, 393 {"utf16le", IDX_UTF16LE}, 394 {"ucs4", IDX_UCS4}, 395 {"ucs4be", IDX_UCS4}, 396 {"ucs-4be", IDX_UCS4}, 397 {"ucs4le", IDX_UCS4LE}, 398 {"utf32", IDX_UCS4}, 399 {"utf-32", IDX_UCS4}, 400 {"utf32be", IDX_UCS4}, 401 {"utf-32be", IDX_UCS4}, 402 {"utf32le", IDX_UCS4LE}, 403 {"utf-32le", IDX_UCS4LE}, 404 {"932", IDX_CP932}, 405 {"949", IDX_CP949}, 406 {"936", IDX_CP936}, 407 {"gbk", IDX_CP936}, 408 {"950", IDX_CP950}, 409 {"eucjp", IDX_EUC_JP}, 410 {"unix-jis", IDX_EUC_JP}, 411 {"ujis", IDX_EUC_JP}, 412 {"shift-jis", IDX_SJIS}, 413 {"pck", IDX_SJIS}, /* Sun: PCK */ 414 {"euckr", IDX_EUC_KR}, 415 {"5601", IDX_EUC_KR}, /* Sun: KS C 5601 */ 416 {"euccn", IDX_EUC_CN}, 417 {"gb2312", IDX_EUC_CN}, 418 {"euctw", IDX_EUC_TW}, 419 #if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS_X) 420 {"japan", IDX_CP932}, 421 {"korea", IDX_CP949}, 422 {"prc", IDX_CP936}, 423 {"chinese", IDX_CP936}, 424 {"taiwan", IDX_CP950}, 425 {"big5", IDX_CP950}, 426 #else 427 {"japan", IDX_EUC_JP}, 428 {"korea", IDX_EUC_KR}, 429 {"prc", IDX_EUC_CN}, 430 {"chinese", IDX_EUC_CN}, 431 {"taiwan", IDX_EUC_TW}, 432 {"cp950", IDX_BIG5}, 433 {"950", IDX_BIG5}, 434 #endif 435 {"mac", IDX_MACROMAN}, 436 {"mac-roman", IDX_MACROMAN}, 437 {NULL, 0} 438 }; 439 440 #ifndef CP_UTF8 441 # define CP_UTF8 65001 /* magic number from winnls.h */ 442 #endif 443 444 /* 445 * Find encoding "name" in the list of canonical encoding names. 446 * Returns -1 if not found. 447 */ 448 static int 449 enc_canon_search(char_u *name) 450 { 451 int i; 452 453 for (i = 0; i < IDX_COUNT; ++i) 454 if (STRCMP(name, enc_canon_table[i].name) == 0) 455 return i; 456 return -1; 457 } 458 459 #endif 460 461 #if defined(FEAT_MBYTE) || defined(PROTO) 462 463 /* 464 * Find canonical encoding "name" in the list and return its properties. 465 * Returns 0 if not found. 466 */ 467 int 468 enc_canon_props(char_u *name) 469 { 470 int i; 471 472 i = enc_canon_search(name); 473 if (i >= 0) 474 return enc_canon_table[i].prop; 475 #ifdef WIN3264 476 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2])) 477 { 478 CPINFO cpinfo; 479 480 /* Get info on this codepage to find out what it is. */ 481 if (GetCPInfo(atoi((char *)name + 2), &cpinfo) != 0) 482 { 483 if (cpinfo.MaxCharSize == 1) /* some single-byte encoding */ 484 return ENC_8BIT; 485 if (cpinfo.MaxCharSize == 2 486 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 487 /* must be a DBCS encoding */ 488 return ENC_DBCS; 489 } 490 return 0; 491 } 492 #endif 493 if (STRNCMP(name, "2byte-", 6) == 0) 494 return ENC_DBCS; 495 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0) 496 return ENC_8BIT; 497 return 0; 498 } 499 500 /* 501 * Set up for using multi-byte characters. 502 * Called in three cases: 503 * - by main() to initialize (p_enc == NULL) 504 * - by set_init_1() after 'encoding' was set to its default. 505 * - by do_set() when 'encoding' has been set. 506 * p_enc must have been passed through enc_canonize() already. 507 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags. 508 * Fills mb_bytelen_tab[] and returns NULL when there are no problems. 509 * When there is something wrong: Returns an error message and doesn't change 510 * anything. 511 */ 512 char * 513 mb_init(void) 514 { 515 int i; 516 int idx; 517 int n; 518 int enc_dbcs_new = 0; 519 #if defined(USE_ICONV) && !defined(WIN3264) && !defined(WIN32UNIX) \ 520 && !defined(MACOS_CONVERT) 521 # define LEN_FROM_CONV 522 vimconv_T vimconv; 523 char_u *p; 524 #endif 525 526 if (p_enc == NULL) 527 { 528 /* Just starting up: set the whole table to one's. */ 529 for (i = 0; i < 256; ++i) 530 mb_bytelen_tab[i] = 1; 531 input_conv.vc_type = CONV_NONE; 532 input_conv.vc_factor = 1; 533 output_conv.vc_type = CONV_NONE; 534 return NULL; 535 } 536 537 #ifdef WIN3264 538 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2])) 539 { 540 CPINFO cpinfo; 541 542 /* Get info on this codepage to find out what it is. */ 543 if (GetCPInfo(atoi((char *)p_enc + 2), &cpinfo) != 0) 544 { 545 if (cpinfo.MaxCharSize == 1) 546 { 547 /* some single-byte encoding */ 548 enc_unicode = 0; 549 enc_utf8 = FALSE; 550 } 551 else if (cpinfo.MaxCharSize == 2 552 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 553 { 554 /* must be a DBCS encoding, check below */ 555 enc_dbcs_new = atoi((char *)p_enc + 2); 556 } 557 else 558 goto codepage_invalid; 559 } 560 else if (GetLastError() == ERROR_INVALID_PARAMETER) 561 { 562 codepage_invalid: 563 return N_("E543: Not a valid codepage"); 564 } 565 } 566 #endif 567 else if (STRNCMP(p_enc, "8bit-", 5) == 0 568 || STRNCMP(p_enc, "iso-8859-", 9) == 0) 569 { 570 /* Accept any "8bit-" or "iso-8859-" name. */ 571 enc_unicode = 0; 572 enc_utf8 = FALSE; 573 } 574 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 575 { 576 #ifdef WIN3264 577 /* Windows: accept only valid codepage numbers, check below. */ 578 if (p_enc[6] != 'c' || p_enc[7] != 'p' 579 || (enc_dbcs_new = atoi((char *)p_enc + 8)) == 0) 580 return e_invarg; 581 #else 582 /* Unix: accept any "2byte-" name, assume current locale. */ 583 enc_dbcs_new = DBCS_2BYTE; 584 #endif 585 } 586 else if ((idx = enc_canon_search(p_enc)) >= 0) 587 { 588 i = enc_canon_table[idx].prop; 589 if (i & ENC_UNICODE) 590 { 591 /* Unicode */ 592 enc_utf8 = TRUE; 593 if (i & (ENC_2BYTE | ENC_2WORD)) 594 enc_unicode = 2; 595 else if (i & ENC_4BYTE) 596 enc_unicode = 4; 597 else 598 enc_unicode = 0; 599 } 600 else if (i & ENC_DBCS) 601 { 602 /* 2byte, handle below */ 603 enc_dbcs_new = enc_canon_table[idx].codepage; 604 } 605 else 606 { 607 /* Must be 8-bit. */ 608 enc_unicode = 0; 609 enc_utf8 = FALSE; 610 } 611 } 612 else /* Don't know what encoding this is, reject it. */ 613 return e_invarg; 614 615 if (enc_dbcs_new != 0) 616 { 617 #ifdef WIN3264 618 /* Check if the DBCS code page is OK. */ 619 if (!IsValidCodePage(enc_dbcs_new)) 620 goto codepage_invalid; 621 #endif 622 enc_unicode = 0; 623 enc_utf8 = FALSE; 624 } 625 enc_dbcs = enc_dbcs_new; 626 has_mbyte = (enc_dbcs != 0 || enc_utf8); 627 628 #if defined(WIN3264) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 629 enc_codepage = encname2codepage(p_enc); 630 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0); 631 #endif 632 633 /* Detect an encoding that uses latin1 characters. */ 634 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0 635 || STRCMP(p_enc, "iso-8859-15") == 0); 636 637 /* 638 * Set the function pointers. 639 */ 640 if (enc_utf8) 641 { 642 mb_ptr2len = utfc_ptr2len; 643 mb_ptr2len_len = utfc_ptr2len_len; 644 mb_char2len = utf_char2len; 645 mb_char2bytes = utf_char2bytes; 646 mb_ptr2cells = utf_ptr2cells; 647 mb_ptr2cells_len = utf_ptr2cells_len; 648 mb_char2cells = utf_char2cells; 649 mb_off2cells = utf_off2cells; 650 mb_ptr2char = utf_ptr2char; 651 mb_head_off = utf_head_off; 652 } 653 else if (enc_dbcs != 0) 654 { 655 mb_ptr2len = dbcs_ptr2len; 656 mb_ptr2len_len = dbcs_ptr2len_len; 657 mb_char2len = dbcs_char2len; 658 mb_char2bytes = dbcs_char2bytes; 659 mb_ptr2cells = dbcs_ptr2cells; 660 mb_ptr2cells_len = dbcs_ptr2cells_len; 661 mb_char2cells = dbcs_char2cells; 662 mb_off2cells = dbcs_off2cells; 663 mb_ptr2char = dbcs_ptr2char; 664 mb_head_off = dbcs_head_off; 665 } 666 else 667 { 668 mb_ptr2len = latin_ptr2len; 669 mb_ptr2len_len = latin_ptr2len_len; 670 mb_char2len = latin_char2len; 671 mb_char2bytes = latin_char2bytes; 672 mb_ptr2cells = latin_ptr2cells; 673 mb_ptr2cells_len = latin_ptr2cells_len; 674 mb_char2cells = latin_char2cells; 675 mb_off2cells = latin_off2cells; 676 mb_ptr2char = latin_ptr2char; 677 mb_head_off = latin_head_off; 678 } 679 680 /* 681 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN(). 682 */ 683 #ifdef LEN_FROM_CONV 684 /* When 'encoding' is different from the current locale mblen() won't 685 * work. Use conversion to "utf-8" instead. */ 686 vimconv.vc_type = CONV_NONE; 687 if (enc_dbcs) 688 { 689 p = enc_locale(); 690 if (p == NULL || STRCMP(p, p_enc) != 0) 691 { 692 convert_setup(&vimconv, p_enc, (char_u *)"utf-8"); 693 vimconv.vc_fail = TRUE; 694 } 695 vim_free(p); 696 } 697 #endif 698 699 for (i = 0; i < 256; ++i) 700 { 701 /* Our own function to reliably check the length of UTF-8 characters, 702 * independent of mblen(). */ 703 if (enc_utf8) 704 n = utf8len_tab[i]; 705 else if (enc_dbcs == 0) 706 n = 1; 707 else 708 { 709 #if defined(WIN3264) || defined(WIN32UNIX) 710 /* enc_dbcs is set by setting 'fileencoding'. It becomes a Windows 711 * CodePage identifier, which we can pass directly in to Windows 712 * API */ 713 n = IsDBCSLeadByteEx(enc_dbcs, (WINBYTE)i) ? 2 : 1; 714 #else 715 # if defined(__amigaos4__) || defined(__ANDROID__) || \ 716 !(defined(HAVE_MBLEN) || defined(X_LOCALE)) 717 /* 718 * if mblen() is not available, character which MSB is turned on 719 * are treated as leading byte character. (note : This assumption 720 * is not always true.) 721 */ 722 n = (i & 0x80) ? 2 : 1; 723 # else 724 char buf[MB_MAXBYTES + 1]; 725 726 if (i == NUL) /* just in case mblen() can't handle "" */ 727 n = 1; 728 else 729 { 730 buf[0] = i; 731 buf[1] = 0; 732 # ifdef LEN_FROM_CONV 733 if (vimconv.vc_type != CONV_NONE) 734 { 735 /* 736 * string_convert() should fail when converting the first 737 * byte of a double-byte character. 738 */ 739 p = string_convert(&vimconv, (char_u *)buf, NULL); 740 if (p != NULL) 741 { 742 vim_free(p); 743 n = 1; 744 } 745 else 746 n = 2; 747 } 748 else 749 # endif 750 { 751 /* 752 * mblen() should return -1 for invalid (means the leading 753 * multibyte) character. However there are some platforms 754 * where mblen() returns 0 for invalid character. 755 * Therefore, following condition includes 0. 756 */ 757 vim_ignored = mblen(NULL, 0); // First reset the state. 758 if (mblen(buf, (size_t)1) <= 0) 759 n = 2; 760 else 761 n = 1; 762 } 763 } 764 # endif 765 #endif 766 } 767 768 mb_bytelen_tab[i] = n; 769 } 770 771 #ifdef LEN_FROM_CONV 772 convert_setup(&vimconv, NULL, NULL); 773 #endif 774 775 /* The cell width depends on the type of multi-byte characters. */ 776 (void)init_chartab(); 777 778 /* When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] */ 779 screenalloc(FALSE); 780 781 /* When using Unicode, set default for 'fileencodings'. */ 782 if (enc_utf8 && !option_was_set((char_u *)"fencs")) 783 set_string_option_direct((char_u *)"fencs", -1, 784 (char_u *)"ucs-bom,utf-8,default,latin1", OPT_FREE, 0); 785 786 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT) 787 /* GNU gettext 0.10.37 supports this feature: set the codeset used for 788 * translated messages independently from the current locale. */ 789 (void)bind_textdomain_codeset(VIMPACKAGE, 790 enc_utf8 ? "utf-8" : (char *)p_enc); 791 #endif 792 793 #ifdef WIN32 794 /* When changing 'encoding' while starting up, then convert the command 795 * line arguments from the active codepage to 'encoding'. */ 796 if (starting != 0) 797 fix_arg_enc(); 798 #endif 799 800 /* Fire an autocommand to let people do custom font setup. This must be 801 * after Vim has been setup for the new encoding. */ 802 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); 803 804 #ifdef FEAT_SPELL 805 /* Need to reload spell dictionaries */ 806 spell_reload(); 807 #endif 808 809 return NULL; 810 } 811 812 /* 813 * Return the size of the BOM for the current buffer: 814 * 0 - no BOM 815 * 2 - UCS-2 or UTF-16 BOM 816 * 4 - UCS-4 BOM 817 * 3 - UTF-8 BOM 818 */ 819 int 820 bomb_size(void) 821 { 822 int n = 0; 823 824 if (curbuf->b_p_bomb && !curbuf->b_p_bin) 825 { 826 if (*curbuf->b_p_fenc == NUL) 827 { 828 if (enc_utf8) 829 { 830 if (enc_unicode != 0) 831 n = enc_unicode; 832 else 833 n = 3; 834 } 835 } 836 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) 837 n = 3; 838 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 839 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) 840 n = 2; 841 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) 842 n = 4; 843 } 844 return n; 845 } 846 847 /* 848 * Remove all BOM from "s" by moving remaining text. 849 */ 850 void 851 remove_bom(char_u *s) 852 { 853 if (enc_utf8) 854 { 855 char_u *p = s; 856 857 while ((p = vim_strbyte(p, 0xef)) != NULL) 858 { 859 if (p[1] == 0xbb && p[2] == 0xbf) 860 STRMOVE(p, p + 3); 861 else 862 ++p; 863 } 864 } 865 } 866 867 /* 868 * Get class of pointer: 869 * 0 for blank or NUL 870 * 1 for punctuation 871 * 2 for an (ASCII) word character 872 * >2 for other word characters 873 */ 874 int 875 mb_get_class(char_u *p) 876 { 877 return mb_get_class_buf(p, curbuf); 878 } 879 880 int 881 mb_get_class_buf(char_u *p, buf_T *buf) 882 { 883 if (MB_BYTE2LEN(p[0]) == 1) 884 { 885 if (p[0] == NUL || VIM_ISWHITE(p[0])) 886 return 0; 887 if (vim_iswordc_buf(p[0], buf)) 888 return 2; 889 return 1; 890 } 891 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) 892 return dbcs_class(p[0], p[1]); 893 if (enc_utf8) 894 return utf_class_buf(utf_ptr2char(p), buf); 895 return 0; 896 } 897 898 /* 899 * Get class of a double-byte character. This always returns 3 or bigger. 900 * TODO: Should return 1 for punctuation. 901 */ 902 int 903 dbcs_class(unsigned lead, unsigned trail) 904 { 905 switch (enc_dbcs) 906 { 907 /* please add classify routine for your language in here */ 908 909 case DBCS_JPNU: /* ? */ 910 case DBCS_JPN: 911 { 912 /* JIS code classification */ 913 unsigned char lb = lead; 914 unsigned char tb = trail; 915 916 /* convert process code to JIS */ 917 # if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS_X) 918 /* process code is SJIS */ 919 if (lb <= 0x9f) 920 lb = (lb - 0x81) * 2 + 0x21; 921 else 922 lb = (lb - 0xc1) * 2 + 0x21; 923 if (tb <= 0x7e) 924 tb -= 0x1f; 925 else if (tb <= 0x9e) 926 tb -= 0x20; 927 else 928 { 929 tb -= 0x7e; 930 lb += 1; 931 } 932 # else 933 /* 934 * XXX: Code page identification can not use with all 935 * system! So, some other encoding information 936 * will be needed. 937 * In japanese: SJIS,EUC,UNICODE,(JIS) 938 * Note that JIS-code system don't use as 939 * process code in most system because it uses 940 * escape sequences(JIS is context depend encoding). 941 */ 942 /* assume process code is JAPANESE-EUC */ 943 lb &= 0x7f; 944 tb &= 0x7f; 945 # endif 946 /* exceptions */ 947 switch (lb << 8 | tb) 948 { 949 case 0x2121: /* ZENKAKU space */ 950 return 0; 951 case 0x2122: /* TOU-TEN (Japanese comma) */ 952 case 0x2123: /* KU-TEN (Japanese period) */ 953 case 0x2124: /* ZENKAKU comma */ 954 case 0x2125: /* ZENKAKU period */ 955 return 1; 956 case 0x213c: /* prolongedsound handled as KATAKANA */ 957 return 13; 958 } 959 /* sieved by KU code */ 960 switch (lb) 961 { 962 case 0x21: 963 case 0x22: 964 /* special symbols */ 965 return 10; 966 case 0x23: 967 /* alpha-numeric */ 968 return 11; 969 case 0x24: 970 /* hiragana */ 971 return 12; 972 case 0x25: 973 /* katakana */ 974 return 13; 975 case 0x26: 976 /* greek */ 977 return 14; 978 case 0x27: 979 /* russian */ 980 return 15; 981 case 0x28: 982 /* lines */ 983 return 16; 984 default: 985 /* kanji */ 986 return 17; 987 } 988 } 989 990 case DBCS_KORU: /* ? */ 991 case DBCS_KOR: 992 { 993 /* KS code classification */ 994 unsigned char c1 = lead; 995 unsigned char c2 = trail; 996 997 /* 998 * 20 : Hangul 999 * 21 : Hanja 1000 * 22 : Symbols 1001 * 23 : Alpha-numeric/Roman Letter (Full width) 1002 * 24 : Hangul Letter(Alphabet) 1003 * 25 : Roman Numeral/Greek Letter 1004 * 26 : Box Drawings 1005 * 27 : Unit Symbols 1006 * 28 : Circled/Parenthesized Letter 1007 * 29 : Hiragana/Katakana 1008 * 30 : Cyrillic Letter 1009 */ 1010 1011 if (c1 >= 0xB0 && c1 <= 0xC8) 1012 /* Hangul */ 1013 return 20; 1014 #if defined(WIN3264) || defined(WIN32UNIX) 1015 else if (c1 <= 0xA0 || c2 <= 0xA0) 1016 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */ 1017 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE 1018 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 1019 */ 1020 return 20; 1021 #endif 1022 1023 else if (c1 >= 0xCA && c1 <= 0xFD) 1024 /* Hanja */ 1025 return 21; 1026 else switch (c1) 1027 { 1028 case 0xA1: 1029 case 0xA2: 1030 /* Symbols */ 1031 return 22; 1032 case 0xA3: 1033 /* Alpha-numeric */ 1034 return 23; 1035 case 0xA4: 1036 /* Hangul Letter(Alphabet) */ 1037 return 24; 1038 case 0xA5: 1039 /* Roman Numeral/Greek Letter */ 1040 return 25; 1041 case 0xA6: 1042 /* Box Drawings */ 1043 return 26; 1044 case 0xA7: 1045 /* Unit Symbols */ 1046 return 27; 1047 case 0xA8: 1048 case 0xA9: 1049 if (c2 <= 0xAF) 1050 return 25; /* Roman Letter */ 1051 else if (c2 >= 0xF6) 1052 return 22; /* Symbols */ 1053 else 1054 /* Circled/Parenthesized Letter */ 1055 return 28; 1056 case 0xAA: 1057 case 0xAB: 1058 /* Hiragana/Katakana */ 1059 return 29; 1060 case 0xAC: 1061 /* Cyrillic Letter */ 1062 return 30; 1063 } 1064 } 1065 default: 1066 break; 1067 } 1068 return 3; 1069 } 1070 1071 /* 1072 * mb_char2len() function pointer. 1073 * Return length in bytes of character "c". 1074 * Returns 1 for a single-byte character. 1075 */ 1076 int 1077 latin_char2len(int c UNUSED) 1078 { 1079 return 1; 1080 } 1081 1082 static int 1083 dbcs_char2len( 1084 int c) 1085 { 1086 if (c >= 0x100) 1087 return 2; 1088 return 1; 1089 } 1090 1091 /* 1092 * mb_char2bytes() function pointer. 1093 * Convert a character to its bytes. 1094 * Returns the length in bytes. 1095 */ 1096 int 1097 latin_char2bytes(int c, char_u *buf) 1098 { 1099 buf[0] = c; 1100 return 1; 1101 } 1102 1103 static int 1104 dbcs_char2bytes(int c, char_u *buf) 1105 { 1106 if (c >= 0x100) 1107 { 1108 buf[0] = (unsigned)c >> 8; 1109 buf[1] = c; 1110 /* Never use a NUL byte, it causes lots of trouble. It's an invalid 1111 * character anyway. */ 1112 if (buf[1] == NUL) 1113 buf[1] = '\n'; 1114 return 2; 1115 } 1116 buf[0] = c; 1117 return 1; 1118 } 1119 1120 /* 1121 * mb_ptr2len() function pointer. 1122 * Get byte length of character at "*p" but stop at a NUL. 1123 * For UTF-8 this includes following composing characters. 1124 * Returns 0 when *p is NUL. 1125 */ 1126 int 1127 latin_ptr2len(char_u *p) 1128 { 1129 return MB_BYTE2LEN(*p); 1130 } 1131 1132 static int 1133 dbcs_ptr2len( 1134 char_u *p) 1135 { 1136 int len; 1137 1138 /* Check if second byte is not missing. */ 1139 len = MB_BYTE2LEN(*p); 1140 if (len == 2 && p[1] == NUL) 1141 len = 1; 1142 return len; 1143 } 1144 1145 /* 1146 * mb_ptr2len_len() function pointer. 1147 * Like mb_ptr2len(), but limit to read "size" bytes. 1148 * Returns 0 for an empty string. 1149 * Returns 1 for an illegal char or an incomplete byte sequence. 1150 */ 1151 int 1152 latin_ptr2len_len(char_u *p, int size) 1153 { 1154 if (size < 1 || *p == NUL) 1155 return 0; 1156 return 1; 1157 } 1158 1159 static int 1160 dbcs_ptr2len_len(char_u *p, int size) 1161 { 1162 int len; 1163 1164 if (size < 1 || *p == NUL) 1165 return 0; 1166 if (size == 1) 1167 return 1; 1168 /* Check that second byte is not missing. */ 1169 len = MB_BYTE2LEN(*p); 1170 if (len == 2 && p[1] == NUL) 1171 len = 1; 1172 return len; 1173 } 1174 1175 struct interval 1176 { 1177 long first; 1178 long last; 1179 }; 1180 1181 /* 1182 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]". 1183 */ 1184 static int 1185 intable(struct interval *table, size_t size, int c) 1186 { 1187 int mid, bot, top; 1188 1189 /* first quick check for Latin1 etc. characters */ 1190 if (c < table[0].first) 1191 return FALSE; 1192 1193 /* binary search in table */ 1194 bot = 0; 1195 top = (int)(size / sizeof(struct interval) - 1); 1196 while (top >= bot) 1197 { 1198 mid = (bot + top) / 2; 1199 if (table[mid].last < c) 1200 bot = mid + 1; 1201 else if (table[mid].first > c) 1202 top = mid - 1; 1203 else 1204 return TRUE; 1205 } 1206 return FALSE; 1207 } 1208 1209 /* Sorted list of non-overlapping intervals of East Asian Ambiguous 1210 * characters, generated with ../runtime/tools/unicode.vim. */ 1211 static struct interval ambiguous[] = 1212 { 1213 {0x00a1, 0x00a1}, 1214 {0x00a4, 0x00a4}, 1215 {0x00a7, 0x00a8}, 1216 {0x00aa, 0x00aa}, 1217 {0x00ad, 0x00ae}, 1218 {0x00b0, 0x00b4}, 1219 {0x00b6, 0x00ba}, 1220 {0x00bc, 0x00bf}, 1221 {0x00c6, 0x00c6}, 1222 {0x00d0, 0x00d0}, 1223 {0x00d7, 0x00d8}, 1224 {0x00de, 0x00e1}, 1225 {0x00e6, 0x00e6}, 1226 {0x00e8, 0x00ea}, 1227 {0x00ec, 0x00ed}, 1228 {0x00f0, 0x00f0}, 1229 {0x00f2, 0x00f3}, 1230 {0x00f7, 0x00fa}, 1231 {0x00fc, 0x00fc}, 1232 {0x00fe, 0x00fe}, 1233 {0x0101, 0x0101}, 1234 {0x0111, 0x0111}, 1235 {0x0113, 0x0113}, 1236 {0x011b, 0x011b}, 1237 {0x0126, 0x0127}, 1238 {0x012b, 0x012b}, 1239 {0x0131, 0x0133}, 1240 {0x0138, 0x0138}, 1241 {0x013f, 0x0142}, 1242 {0x0144, 0x0144}, 1243 {0x0148, 0x014b}, 1244 {0x014d, 0x014d}, 1245 {0x0152, 0x0153}, 1246 {0x0166, 0x0167}, 1247 {0x016b, 0x016b}, 1248 {0x01ce, 0x01ce}, 1249 {0x01d0, 0x01d0}, 1250 {0x01d2, 0x01d2}, 1251 {0x01d4, 0x01d4}, 1252 {0x01d6, 0x01d6}, 1253 {0x01d8, 0x01d8}, 1254 {0x01da, 0x01da}, 1255 {0x01dc, 0x01dc}, 1256 {0x0251, 0x0251}, 1257 {0x0261, 0x0261}, 1258 {0x02c4, 0x02c4}, 1259 {0x02c7, 0x02c7}, 1260 {0x02c9, 0x02cb}, 1261 {0x02cd, 0x02cd}, 1262 {0x02d0, 0x02d0}, 1263 {0x02d8, 0x02db}, 1264 {0x02dd, 0x02dd}, 1265 {0x02df, 0x02df}, 1266 {0x0300, 0x036f}, 1267 {0x0391, 0x03a1}, 1268 {0x03a3, 0x03a9}, 1269 {0x03b1, 0x03c1}, 1270 {0x03c3, 0x03c9}, 1271 {0x0401, 0x0401}, 1272 {0x0410, 0x044f}, 1273 {0x0451, 0x0451}, 1274 {0x2010, 0x2010}, 1275 {0x2013, 0x2016}, 1276 {0x2018, 0x2019}, 1277 {0x201c, 0x201d}, 1278 {0x2020, 0x2022}, 1279 {0x2024, 0x2027}, 1280 {0x2030, 0x2030}, 1281 {0x2032, 0x2033}, 1282 {0x2035, 0x2035}, 1283 {0x203b, 0x203b}, 1284 {0x203e, 0x203e}, 1285 {0x2074, 0x2074}, 1286 {0x207f, 0x207f}, 1287 {0x2081, 0x2084}, 1288 {0x20ac, 0x20ac}, 1289 {0x2103, 0x2103}, 1290 {0x2105, 0x2105}, 1291 {0x2109, 0x2109}, 1292 {0x2113, 0x2113}, 1293 {0x2116, 0x2116}, 1294 {0x2121, 0x2122}, 1295 {0x2126, 0x2126}, 1296 {0x212b, 0x212b}, 1297 {0x2153, 0x2154}, 1298 {0x215b, 0x215e}, 1299 {0x2160, 0x216b}, 1300 {0x2170, 0x2179}, 1301 {0x2189, 0x2189}, 1302 {0x2190, 0x2199}, 1303 {0x21b8, 0x21b9}, 1304 {0x21d2, 0x21d2}, 1305 {0x21d4, 0x21d4}, 1306 {0x21e7, 0x21e7}, 1307 {0x2200, 0x2200}, 1308 {0x2202, 0x2203}, 1309 {0x2207, 0x2208}, 1310 {0x220b, 0x220b}, 1311 {0x220f, 0x220f}, 1312 {0x2211, 0x2211}, 1313 {0x2215, 0x2215}, 1314 {0x221a, 0x221a}, 1315 {0x221d, 0x2220}, 1316 {0x2223, 0x2223}, 1317 {0x2225, 0x2225}, 1318 {0x2227, 0x222c}, 1319 {0x222e, 0x222e}, 1320 {0x2234, 0x2237}, 1321 {0x223c, 0x223d}, 1322 {0x2248, 0x2248}, 1323 {0x224c, 0x224c}, 1324 {0x2252, 0x2252}, 1325 {0x2260, 0x2261}, 1326 {0x2264, 0x2267}, 1327 {0x226a, 0x226b}, 1328 {0x226e, 0x226f}, 1329 {0x2282, 0x2283}, 1330 {0x2286, 0x2287}, 1331 {0x2295, 0x2295}, 1332 {0x2299, 0x2299}, 1333 {0x22a5, 0x22a5}, 1334 {0x22bf, 0x22bf}, 1335 {0x2312, 0x2312}, 1336 {0x2460, 0x24e9}, 1337 {0x24eb, 0x254b}, 1338 {0x2550, 0x2573}, 1339 {0x2580, 0x258f}, 1340 {0x2592, 0x2595}, 1341 {0x25a0, 0x25a1}, 1342 {0x25a3, 0x25a9}, 1343 {0x25b2, 0x25b3}, 1344 {0x25b6, 0x25b7}, 1345 {0x25bc, 0x25bd}, 1346 {0x25c0, 0x25c1}, 1347 {0x25c6, 0x25c8}, 1348 {0x25cb, 0x25cb}, 1349 {0x25ce, 0x25d1}, 1350 {0x25e2, 0x25e5}, 1351 {0x25ef, 0x25ef}, 1352 {0x2605, 0x2606}, 1353 {0x2609, 0x2609}, 1354 {0x260e, 0x260f}, 1355 {0x261c, 0x261c}, 1356 {0x261e, 0x261e}, 1357 {0x2640, 0x2640}, 1358 {0x2642, 0x2642}, 1359 {0x2660, 0x2661}, 1360 {0x2663, 0x2665}, 1361 {0x2667, 0x266a}, 1362 {0x266c, 0x266d}, 1363 {0x266f, 0x266f}, 1364 {0x269e, 0x269f}, 1365 {0x26bf, 0x26bf}, 1366 {0x26c6, 0x26cd}, 1367 {0x26cf, 0x26d3}, 1368 {0x26d5, 0x26e1}, 1369 {0x26e3, 0x26e3}, 1370 {0x26e8, 0x26e9}, 1371 {0x26eb, 0x26f1}, 1372 {0x26f4, 0x26f4}, 1373 {0x26f6, 0x26f9}, 1374 {0x26fb, 0x26fc}, 1375 {0x26fe, 0x26ff}, 1376 {0x273d, 0x273d}, 1377 {0x2776, 0x277f}, 1378 {0x2b56, 0x2b59}, 1379 {0x3248, 0x324f}, 1380 {0xe000, 0xf8ff}, 1381 {0xfe00, 0xfe0f}, 1382 {0xfffd, 0xfffd}, 1383 {0x1f100, 0x1f10a}, 1384 {0x1f110, 0x1f12d}, 1385 {0x1f130, 0x1f169}, 1386 {0x1f170, 0x1f18d}, 1387 {0x1f18f, 0x1f190}, 1388 {0x1f19b, 0x1f1ac}, 1389 {0xe0100, 0xe01ef}, 1390 {0xf0000, 0xffffd}, 1391 {0x100000, 0x10fffd} 1392 }; 1393 1394 #if defined(FEAT_TERMINAL) || defined(PROTO) 1395 /* 1396 * utf_char2cells() with different argument type for libvterm. 1397 */ 1398 int 1399 utf_uint2cells(UINT32_T c) 1400 { 1401 if (c >= 0x100 && utf_iscomposing((int)c)) 1402 return 0; 1403 return utf_char2cells((int)c); 1404 } 1405 #endif 1406 1407 /* 1408 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1409 * Returns 4 or 6 for an unprintable character. 1410 * Is only correct for characters >= 0x80. 1411 * When p_ambw is "double", return 2 for a character with East Asian Width 1412 * class 'A'(mbiguous). 1413 */ 1414 int 1415 utf_char2cells(int c) 1416 { 1417 /* Sorted list of non-overlapping intervals of East Asian double width 1418 * characters, generated with ../runtime/tools/unicode.vim. */ 1419 static struct interval doublewidth[] = 1420 { 1421 {0x1100, 0x115f}, 1422 {0x231a, 0x231b}, 1423 {0x2329, 0x232a}, 1424 {0x23e9, 0x23ec}, 1425 {0x23f0, 0x23f0}, 1426 {0x23f3, 0x23f3}, 1427 {0x25fd, 0x25fe}, 1428 {0x2614, 0x2615}, 1429 {0x2648, 0x2653}, 1430 {0x267f, 0x267f}, 1431 {0x2693, 0x2693}, 1432 {0x26a1, 0x26a1}, 1433 {0x26aa, 0x26ab}, 1434 {0x26bd, 0x26be}, 1435 {0x26c4, 0x26c5}, 1436 {0x26ce, 0x26ce}, 1437 {0x26d4, 0x26d4}, 1438 {0x26ea, 0x26ea}, 1439 {0x26f2, 0x26f3}, 1440 {0x26f5, 0x26f5}, 1441 {0x26fa, 0x26fa}, 1442 {0x26fd, 0x26fd}, 1443 {0x2705, 0x2705}, 1444 {0x270a, 0x270b}, 1445 {0x2728, 0x2728}, 1446 {0x274c, 0x274c}, 1447 {0x274e, 0x274e}, 1448 {0x2753, 0x2755}, 1449 {0x2757, 0x2757}, 1450 {0x2795, 0x2797}, 1451 {0x27b0, 0x27b0}, 1452 {0x27bf, 0x27bf}, 1453 {0x2b1b, 0x2b1c}, 1454 {0x2b50, 0x2b50}, 1455 {0x2b55, 0x2b55}, 1456 {0x2e80, 0x2e99}, 1457 {0x2e9b, 0x2ef3}, 1458 {0x2f00, 0x2fd5}, 1459 {0x2ff0, 0x2ffb}, 1460 {0x3000, 0x303e}, 1461 {0x3041, 0x3096}, 1462 {0x3099, 0x30ff}, 1463 {0x3105, 0x312f}, 1464 {0x3131, 0x318e}, 1465 {0x3190, 0x31ba}, 1466 {0x31c0, 0x31e3}, 1467 {0x31f0, 0x321e}, 1468 {0x3220, 0x3247}, 1469 {0x3250, 0x32fe}, 1470 {0x3300, 0x4dbf}, 1471 {0x4e00, 0xa48c}, 1472 {0xa490, 0xa4c6}, 1473 {0xa960, 0xa97c}, 1474 {0xac00, 0xd7a3}, 1475 {0xf900, 0xfaff}, 1476 {0xfe10, 0xfe19}, 1477 {0xfe30, 0xfe52}, 1478 {0xfe54, 0xfe66}, 1479 {0xfe68, 0xfe6b}, 1480 {0xff01, 0xff60}, 1481 {0xffe0, 0xffe6}, 1482 {0x16fe0, 0x16fe1}, 1483 {0x17000, 0x187f1}, 1484 {0x18800, 0x18af2}, 1485 {0x1b000, 0x1b11e}, 1486 {0x1b170, 0x1b2fb}, 1487 {0x1f004, 0x1f004}, 1488 {0x1f0cf, 0x1f0cf}, 1489 {0x1f18e, 0x1f18e}, 1490 {0x1f191, 0x1f19a}, 1491 {0x1f200, 0x1f202}, 1492 {0x1f210, 0x1f23b}, 1493 {0x1f240, 0x1f248}, 1494 {0x1f250, 0x1f251}, 1495 {0x1f260, 0x1f265}, 1496 {0x1f300, 0x1f320}, 1497 {0x1f32d, 0x1f335}, 1498 {0x1f337, 0x1f37c}, 1499 {0x1f37e, 0x1f393}, 1500 {0x1f3a0, 0x1f3ca}, 1501 {0x1f3cf, 0x1f3d3}, 1502 {0x1f3e0, 0x1f3f0}, 1503 {0x1f3f4, 0x1f3f4}, 1504 {0x1f3f8, 0x1f43e}, 1505 {0x1f440, 0x1f440}, 1506 {0x1f442, 0x1f4fc}, 1507 {0x1f4ff, 0x1f53d}, 1508 {0x1f54b, 0x1f54e}, 1509 {0x1f550, 0x1f567}, 1510 {0x1f57a, 0x1f57a}, 1511 {0x1f595, 0x1f596}, 1512 {0x1f5a4, 0x1f5a4}, 1513 {0x1f5fb, 0x1f64f}, 1514 {0x1f680, 0x1f6c5}, 1515 {0x1f6cc, 0x1f6cc}, 1516 {0x1f6d0, 0x1f6d2}, 1517 {0x1f6eb, 0x1f6ec}, 1518 {0x1f6f4, 0x1f6f9}, 1519 {0x1f910, 0x1f93e}, 1520 {0x1f940, 0x1f970}, 1521 {0x1f973, 0x1f976}, 1522 {0x1f97a, 0x1f97a}, 1523 {0x1f97c, 0x1f9a2}, 1524 {0x1f9b0, 0x1f9b9}, 1525 {0x1f9c0, 0x1f9c2}, 1526 {0x1f9d0, 0x1f9ff}, 1527 {0x20000, 0x2fffd}, 1528 {0x30000, 0x3fffd} 1529 }; 1530 1531 /* Sorted list of non-overlapping intervals of Emoji characters that don't 1532 * have ambiguous or double width, 1533 * based on http://unicode.org/emoji/charts/emoji-list.html */ 1534 static struct interval emoji_width[] = 1535 { 1536 {0x1f1e6, 0x1f1ff}, 1537 {0x1f321, 0x1f321}, 1538 {0x1f324, 0x1f32c}, 1539 {0x1f336, 0x1f336}, 1540 {0x1f37d, 0x1f37d}, 1541 {0x1f396, 0x1f397}, 1542 {0x1f399, 0x1f39b}, 1543 {0x1f39e, 0x1f39f}, 1544 {0x1f3cb, 0x1f3ce}, 1545 {0x1f3d4, 0x1f3df}, 1546 {0x1f3f3, 0x1f3f5}, 1547 {0x1f3f7, 0x1f3f7}, 1548 {0x1f43f, 0x1f43f}, 1549 {0x1f441, 0x1f441}, 1550 {0x1f4fd, 0x1f4fd}, 1551 {0x1f549, 0x1f54a}, 1552 {0x1f56f, 0x1f570}, 1553 {0x1f573, 0x1f579}, 1554 {0x1f587, 0x1f587}, 1555 {0x1f58a, 0x1f58d}, 1556 {0x1f590, 0x1f590}, 1557 {0x1f5a5, 0x1f5a5}, 1558 {0x1f5a8, 0x1f5a8}, 1559 {0x1f5b1, 0x1f5b2}, 1560 {0x1f5bc, 0x1f5bc}, 1561 {0x1f5c2, 0x1f5c4}, 1562 {0x1f5d1, 0x1f5d3}, 1563 {0x1f5dc, 0x1f5de}, 1564 {0x1f5e1, 0x1f5e1}, 1565 {0x1f5e3, 0x1f5e3}, 1566 {0x1f5e8, 0x1f5e8}, 1567 {0x1f5ef, 0x1f5ef}, 1568 {0x1f5f3, 0x1f5f3}, 1569 {0x1f5fa, 0x1f5fa}, 1570 {0x1f6cb, 0x1f6cf}, 1571 {0x1f6e0, 0x1f6e5}, 1572 {0x1f6e9, 0x1f6e9}, 1573 {0x1f6f0, 0x1f6f0}, 1574 {0x1f6f3, 0x1f6f3} 1575 }; 1576 1577 if (c >= 0x100) 1578 { 1579 #ifdef USE_WCHAR_FUNCTIONS 1580 /* 1581 * Assume the library function wcwidth() works better than our own 1582 * stuff. It should return 1 for ambiguous width chars! 1583 */ 1584 int n = wcwidth(c); 1585 1586 if (n < 0) 1587 return 6; /* unprintable, displays <xxxx> */ 1588 if (n > 1) 1589 return n; 1590 #else 1591 if (!utf_printable(c)) 1592 return 6; /* unprintable, displays <xxxx> */ 1593 if (intable(doublewidth, sizeof(doublewidth), c)) 1594 return 2; 1595 #endif 1596 if (p_emoji && intable(emoji_width, sizeof(emoji_width), c)) 1597 return 2; 1598 } 1599 1600 /* Characters below 0x100 are influenced by 'isprint' option */ 1601 else if (c >= 0x80 && !vim_isprintc(c)) 1602 return 4; /* unprintable, displays <xx> */ 1603 1604 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c)) 1605 return 2; 1606 1607 return 1; 1608 } 1609 1610 /* 1611 * mb_ptr2cells() function pointer. 1612 * Return the number of display cells character at "*p" occupies. 1613 * This doesn't take care of unprintable characters, use ptr2cells() for that. 1614 */ 1615 int 1616 latin_ptr2cells(char_u *p UNUSED) 1617 { 1618 return 1; 1619 } 1620 1621 int 1622 utf_ptr2cells( 1623 char_u *p) 1624 { 1625 int c; 1626 1627 /* Need to convert to a wide character. */ 1628 if (*p >= 0x80) 1629 { 1630 c = utf_ptr2char(p); 1631 /* An illegal byte is displayed as <xx>. */ 1632 if (utf_ptr2len(p) == 1 || c == NUL) 1633 return 4; 1634 /* If the char is ASCII it must be an overlong sequence. */ 1635 if (c < 0x80) 1636 return char2cells(c); 1637 return utf_char2cells(c); 1638 } 1639 return 1; 1640 } 1641 1642 int 1643 dbcs_ptr2cells(char_u *p) 1644 { 1645 /* Number of cells is equal to number of bytes, except for euc-jp when 1646 * the first byte is 0x8e. */ 1647 if (enc_dbcs == DBCS_JPNU && *p == 0x8e) 1648 return 1; 1649 return MB_BYTE2LEN(*p); 1650 } 1651 1652 /* 1653 * mb_ptr2cells_len() function pointer. 1654 * Like mb_ptr2cells(), but limit string length to "size". 1655 * For an empty string or truncated character returns 1. 1656 */ 1657 int 1658 latin_ptr2cells_len(char_u *p UNUSED, int size UNUSED) 1659 { 1660 return 1; 1661 } 1662 1663 static int 1664 utf_ptr2cells_len(char_u *p, int size) 1665 { 1666 int c; 1667 1668 /* Need to convert to a wide character. */ 1669 if (size > 0 && *p >= 0x80) 1670 { 1671 if (utf_ptr2len_len(p, size) < utf8len_tab[*p]) 1672 return 1; /* truncated */ 1673 c = utf_ptr2char(p); 1674 /* An illegal byte is displayed as <xx>. */ 1675 if (utf_ptr2len(p) == 1 || c == NUL) 1676 return 4; 1677 /* If the char is ASCII it must be an overlong sequence. */ 1678 if (c < 0x80) 1679 return char2cells(c); 1680 return utf_char2cells(c); 1681 } 1682 return 1; 1683 } 1684 1685 static int 1686 dbcs_ptr2cells_len(char_u *p, int size) 1687 { 1688 /* Number of cells is equal to number of bytes, except for euc-jp when 1689 * the first byte is 0x8e. */ 1690 if (size <= 1 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)) 1691 return 1; 1692 return MB_BYTE2LEN(*p); 1693 } 1694 1695 /* 1696 * mb_char2cells() function pointer. 1697 * Return the number of display cells character "c" occupies. 1698 * Only takes care of multi-byte chars, not "^C" and such. 1699 */ 1700 int 1701 latin_char2cells(int c UNUSED) 1702 { 1703 return 1; 1704 } 1705 1706 static int 1707 dbcs_char2cells(int c) 1708 { 1709 /* Number of cells is equal to number of bytes, except for euc-jp when 1710 * the first byte is 0x8e. */ 1711 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e) 1712 return 1; 1713 /* use the first byte */ 1714 return MB_BYTE2LEN((unsigned)c >> 8); 1715 } 1716 1717 /* 1718 * Return the number of cells occupied by string "p". 1719 * Stop at a NUL character. When "len" >= 0 stop at character "p[len]". 1720 */ 1721 int 1722 mb_string2cells(char_u *p, int len) 1723 { 1724 int i; 1725 int clen = 0; 1726 1727 for (i = 0; (len < 0 || i < len) && p[i] != NUL; i += (*mb_ptr2len)(p + i)) 1728 clen += (*mb_ptr2cells)(p + i); 1729 return clen; 1730 } 1731 1732 /* 1733 * mb_off2cells() function pointer. 1734 * Return number of display cells for char at ScreenLines[off]. 1735 * We make sure that the offset used is less than "max_off". 1736 */ 1737 int 1738 latin_off2cells(unsigned off UNUSED, unsigned max_off UNUSED) 1739 { 1740 return 1; 1741 } 1742 1743 int 1744 dbcs_off2cells(unsigned off, unsigned max_off) 1745 { 1746 /* never check beyond end of the line */ 1747 if (off >= max_off) 1748 return 1; 1749 1750 /* Number of cells is equal to number of bytes, except for euc-jp when 1751 * the first byte is 0x8e. */ 1752 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e) 1753 return 1; 1754 return MB_BYTE2LEN(ScreenLines[off]); 1755 } 1756 1757 int 1758 utf_off2cells(unsigned off, unsigned max_off) 1759 { 1760 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1; 1761 } 1762 1763 /* 1764 * mb_ptr2char() function pointer. 1765 * Convert a byte sequence into a character. 1766 */ 1767 int 1768 latin_ptr2char(char_u *p) 1769 { 1770 return *p; 1771 } 1772 1773 static int 1774 dbcs_ptr2char(char_u *p) 1775 { 1776 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL) 1777 return (p[0] << 8) + p[1]; 1778 return *p; 1779 } 1780 1781 /* 1782 * Convert a UTF-8 byte sequence to a wide character. 1783 * If the sequence is illegal or truncated by a NUL the first byte is 1784 * returned. 1785 * For an overlong sequence this may return zero. 1786 * Does not include composing characters, of course. 1787 */ 1788 int 1789 utf_ptr2char(char_u *p) 1790 { 1791 int len; 1792 1793 if (p[0] < 0x80) /* be quick for ASCII */ 1794 return p[0]; 1795 1796 len = utf8len_tab_zero[p[0]]; 1797 if (len > 1 && (p[1] & 0xc0) == 0x80) 1798 { 1799 if (len == 2) 1800 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); 1801 if ((p[2] & 0xc0) == 0x80) 1802 { 1803 if (len == 3) 1804 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) 1805 + (p[2] & 0x3f); 1806 if ((p[3] & 0xc0) == 0x80) 1807 { 1808 if (len == 4) 1809 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) 1810 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); 1811 if ((p[4] & 0xc0) == 0x80) 1812 { 1813 if (len == 5) 1814 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) 1815 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) 1816 + (p[4] & 0x3f); 1817 if ((p[5] & 0xc0) == 0x80 && len == 6) 1818 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) 1819 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) 1820 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); 1821 } 1822 } 1823 } 1824 } 1825 /* Illegal value, just return the first byte */ 1826 return p[0]; 1827 } 1828 1829 /* 1830 * Convert a UTF-8 byte sequence to a wide character. 1831 * String is assumed to be terminated by NUL or after "n" bytes, whichever 1832 * comes first. 1833 * The function is safe in the sense that it never accesses memory beyond the 1834 * first "n" bytes of "s". 1835 * 1836 * On success, returns decoded codepoint, advances "s" to the beginning of 1837 * next character and decreases "n" accordingly. 1838 * 1839 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past 1840 * NUL byte. 1841 * 1842 * If byte sequence is illegal or incomplete, returns -1 and does not advance 1843 * "s". 1844 */ 1845 static int 1846 utf_safe_read_char_adv(char_u **s, size_t *n) 1847 { 1848 int c, k; 1849 1850 if (*n == 0) /* end of buffer */ 1851 return 0; 1852 1853 k = utf8len_tab_zero[**s]; 1854 1855 if (k == 1) 1856 { 1857 /* ASCII character or NUL */ 1858 (*n)--; 1859 return *(*s)++; 1860 } 1861 1862 if ((size_t)k <= *n) 1863 { 1864 /* We have a multibyte sequence and it isn't truncated by buffer 1865 * limits so utf_ptr2char() is safe to use. Or the first byte is 1866 * illegal (k=0), and it's also safe to use utf_ptr2char(). */ 1867 c = utf_ptr2char(*s); 1868 1869 /* On failure, utf_ptr2char() returns the first byte, so here we 1870 * check equality with the first byte. The only non-ASCII character 1871 * which equals the first byte of its own UTF-8 representation is 1872 * U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too. 1873 * It's safe even if n=1, else we would have k=2 > n. */ 1874 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83)) 1875 { 1876 /* byte sequence was successfully decoded */ 1877 *s += k; 1878 *n -= k; 1879 return c; 1880 } 1881 } 1882 1883 /* byte sequence is incomplete or illegal */ 1884 return -1; 1885 } 1886 1887 /* 1888 * Get character at **pp and advance *pp to the next character. 1889 * Note: composing characters are skipped! 1890 */ 1891 int 1892 mb_ptr2char_adv(char_u **pp) 1893 { 1894 int c; 1895 1896 c = (*mb_ptr2char)(*pp); 1897 *pp += (*mb_ptr2len)(*pp); 1898 return c; 1899 } 1900 1901 /* 1902 * Get character at **pp and advance *pp to the next character. 1903 * Note: composing characters are returned as separate characters. 1904 */ 1905 int 1906 mb_cptr2char_adv(char_u **pp) 1907 { 1908 int c; 1909 1910 c = (*mb_ptr2char)(*pp); 1911 if (enc_utf8) 1912 *pp += utf_ptr2len(*pp); 1913 else 1914 *pp += (*mb_ptr2len)(*pp); 1915 return c; 1916 } 1917 1918 #if defined(FEAT_ARABIC) || defined(PROTO) 1919 /* 1920 * Check whether we are dealing with Arabic combining characters. 1921 * Note: these are NOT really composing characters! 1922 */ 1923 int 1924 arabic_combine( 1925 int one, /* first character */ 1926 int two) /* character just after "one" */ 1927 { 1928 if (one == a_LAM) 1929 return arabic_maycombine(two); 1930 return FALSE; 1931 } 1932 1933 /* 1934 * Check whether we are dealing with a character that could be regarded as an 1935 * Arabic combining character, need to check the character before this. 1936 */ 1937 int 1938 arabic_maycombine(int two) 1939 { 1940 if (p_arshape && !p_tbidi) 1941 return (two == a_ALEF_MADDA 1942 || two == a_ALEF_HAMZA_ABOVE 1943 || two == a_ALEF_HAMZA_BELOW 1944 || two == a_ALEF); 1945 return FALSE; 1946 } 1947 1948 /* 1949 * Check if the character pointed to by "p2" is a composing character when it 1950 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which 1951 * behaves like a composing character. 1952 */ 1953 int 1954 utf_composinglike(char_u *p1, char_u *p2) 1955 { 1956 int c2; 1957 1958 c2 = utf_ptr2char(p2); 1959 if (utf_iscomposing(c2)) 1960 return TRUE; 1961 if (!arabic_maycombine(c2)) 1962 return FALSE; 1963 return arabic_combine(utf_ptr2char(p1), c2); 1964 } 1965 #endif 1966 1967 /* 1968 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1969 * composing characters. 1970 */ 1971 int 1972 utfc_ptr2char( 1973 char_u *p, 1974 int *pcc) /* return: composing chars, last one is 0 */ 1975 { 1976 int len; 1977 int c; 1978 int cc; 1979 int i = 0; 1980 1981 c = utf_ptr2char(p); 1982 len = utf_ptr2len(p); 1983 1984 /* Only accept a composing char when the first char isn't illegal. */ 1985 if ((len > 1 || *p < 0x80) 1986 && p[len] >= 0x80 1987 && UTF_COMPOSINGLIKE(p, p + len)) 1988 { 1989 cc = utf_ptr2char(p + len); 1990 for (;;) 1991 { 1992 pcc[i++] = cc; 1993 if (i == MAX_MCO) 1994 break; 1995 len += utf_ptr2len(p + len); 1996 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1997 break; 1998 } 1999 } 2000 2001 if (i < MAX_MCO) /* last composing char must be 0 */ 2002 pcc[i] = 0; 2003 2004 return c; 2005 } 2006 2007 /* 2008 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 2009 * composing characters. Use no more than p[maxlen]. 2010 */ 2011 int 2012 utfc_ptr2char_len( 2013 char_u *p, 2014 int *pcc, /* return: composing chars, last one is 0 */ 2015 int maxlen) 2016 { 2017 int len; 2018 int c; 2019 int cc; 2020 int i = 0; 2021 2022 c = utf_ptr2char(p); 2023 len = utf_ptr2len_len(p, maxlen); 2024 /* Only accept a composing char when the first char isn't illegal. */ 2025 if ((len > 1 || *p < 0x80) 2026 && len < maxlen 2027 && p[len] >= 0x80 2028 && UTF_COMPOSINGLIKE(p, p + len)) 2029 { 2030 cc = utf_ptr2char(p + len); 2031 for (;;) 2032 { 2033 pcc[i++] = cc; 2034 if (i == MAX_MCO) 2035 break; 2036 len += utf_ptr2len_len(p + len, maxlen - len); 2037 if (len >= maxlen 2038 || p[len] < 0x80 2039 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 2040 break; 2041 } 2042 } 2043 2044 if (i < MAX_MCO) /* last composing char must be 0 */ 2045 pcc[i] = 0; 2046 2047 return c; 2048 } 2049 2050 /* 2051 * Convert the character at screen position "off" to a sequence of bytes. 2052 * Includes the composing characters. 2053 * "buf" must at least have the length MB_MAXBYTES + 1. 2054 * Only to be used when ScreenLinesUC[off] != 0. 2055 * Returns the produced number of bytes. 2056 */ 2057 int 2058 utfc_char2bytes(int off, char_u *buf) 2059 { 2060 int len; 2061 int i; 2062 2063 len = utf_char2bytes(ScreenLinesUC[off], buf); 2064 for (i = 0; i < Screen_mco; ++i) 2065 { 2066 if (ScreenLinesC[i][off] == 0) 2067 break; 2068 len += utf_char2bytes(ScreenLinesC[i][off], buf + len); 2069 } 2070 return len; 2071 } 2072 2073 /* 2074 * Get the length of a UTF-8 byte sequence, not including any following 2075 * composing characters. 2076 * Returns 0 for "". 2077 * Returns 1 for an illegal byte sequence. 2078 */ 2079 int 2080 utf_ptr2len(char_u *p) 2081 { 2082 int len; 2083 int i; 2084 2085 if (*p == NUL) 2086 return 0; 2087 len = utf8len_tab[*p]; 2088 for (i = 1; i < len; ++i) 2089 if ((p[i] & 0xc0) != 0x80) 2090 return 1; 2091 return len; 2092 } 2093 2094 /* 2095 * Return length of UTF-8 character, obtained from the first byte. 2096 * "b" must be between 0 and 255! 2097 * Returns 1 for an invalid first byte value. 2098 */ 2099 int 2100 utf_byte2len(int b) 2101 { 2102 return utf8len_tab[b]; 2103 } 2104 2105 /* 2106 * Get the length of UTF-8 byte sequence "p[size]". Does not include any 2107 * following composing characters. 2108 * Returns 1 for "". 2109 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.). 2110 * Returns number > "size" for an incomplete byte sequence. 2111 * Never returns zero. 2112 */ 2113 int 2114 utf_ptr2len_len(char_u *p, int size) 2115 { 2116 int len; 2117 int i; 2118 int m; 2119 2120 len = utf8len_tab[*p]; 2121 if (len == 1) 2122 return 1; /* NUL, ascii or illegal lead byte */ 2123 if (len > size) 2124 m = size; /* incomplete byte sequence. */ 2125 else 2126 m = len; 2127 for (i = 1; i < m; ++i) 2128 if ((p[i] & 0xc0) != 0x80) 2129 return 1; 2130 return len; 2131 } 2132 2133 /* 2134 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. 2135 * This includes following composing characters. 2136 */ 2137 int 2138 utfc_ptr2len(char_u *p) 2139 { 2140 int len; 2141 int b0 = *p; 2142 #ifdef FEAT_ARABIC 2143 int prevlen; 2144 #endif 2145 2146 if (b0 == NUL) 2147 return 0; 2148 if (b0 < 0x80 && p[1] < 0x80) /* be quick for ASCII */ 2149 return 1; 2150 2151 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2152 len = utf_ptr2len(p); 2153 2154 /* Check for illegal byte. */ 2155 if (len == 1 && b0 >= 0x80) 2156 return 1; 2157 2158 /* 2159 * Check for composing characters. We can handle only the first six, but 2160 * skip all of them (otherwise the cursor would get stuck). 2161 */ 2162 #ifdef FEAT_ARABIC 2163 prevlen = 0; 2164 #endif 2165 for (;;) 2166 { 2167 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2168 return len; 2169 2170 /* Skip over composing char */ 2171 #ifdef FEAT_ARABIC 2172 prevlen = len; 2173 #endif 2174 len += utf_ptr2len(p + len); 2175 } 2176 } 2177 2178 /* 2179 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" 2180 * takes. This includes following composing characters. 2181 * Returns 0 for an empty string. 2182 * Returns 1 for an illegal char or an incomplete byte sequence. 2183 */ 2184 int 2185 utfc_ptr2len_len(char_u *p, int size) 2186 { 2187 int len; 2188 #ifdef FEAT_ARABIC 2189 int prevlen; 2190 #endif 2191 2192 if (size < 1 || *p == NUL) 2193 return 0; 2194 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) /* be quick for ASCII */ 2195 return 1; 2196 2197 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2198 len = utf_ptr2len_len(p, size); 2199 2200 /* Check for illegal byte and incomplete byte sequence. */ 2201 if ((len == 1 && p[0] >= 0x80) || len > size) 2202 return 1; 2203 2204 /* 2205 * Check for composing characters. We can handle only the first six, but 2206 * skip all of them (otherwise the cursor would get stuck). 2207 */ 2208 #ifdef FEAT_ARABIC 2209 prevlen = 0; 2210 #endif 2211 while (len < size) 2212 { 2213 int len_next_char; 2214 2215 if (p[len] < 0x80) 2216 break; 2217 2218 /* 2219 * Next character length should not go beyond size to ensure that 2220 * UTF_COMPOSINGLIKE(...) does not read beyond size. 2221 */ 2222 len_next_char = utf_ptr2len_len(p + len, size - len); 2223 if (len_next_char > size - len) 2224 break; 2225 2226 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2227 break; 2228 2229 /* Skip over composing char */ 2230 #ifdef FEAT_ARABIC 2231 prevlen = len; 2232 #endif 2233 len += len_next_char; 2234 } 2235 return len; 2236 } 2237 2238 /* 2239 * Return the number of bytes the UTF-8 encoding of character "c" takes. 2240 * This does not include composing characters. 2241 */ 2242 int 2243 utf_char2len(int c) 2244 { 2245 if (c < 0x80) 2246 return 1; 2247 if (c < 0x800) 2248 return 2; 2249 if (c < 0x10000) 2250 return 3; 2251 if (c < 0x200000) 2252 return 4; 2253 if (c < 0x4000000) 2254 return 5; 2255 return 6; 2256 } 2257 2258 /* 2259 * Convert Unicode character "c" to UTF-8 string in "buf[]". 2260 * Returns the number of bytes. 2261 */ 2262 int 2263 utf_char2bytes(int c, char_u *buf) 2264 { 2265 if (c < 0x80) /* 7 bits */ 2266 { 2267 buf[0] = c; 2268 return 1; 2269 } 2270 if (c < 0x800) /* 11 bits */ 2271 { 2272 buf[0] = 0xc0 + ((unsigned)c >> 6); 2273 buf[1] = 0x80 + (c & 0x3f); 2274 return 2; 2275 } 2276 if (c < 0x10000) /* 16 bits */ 2277 { 2278 buf[0] = 0xe0 + ((unsigned)c >> 12); 2279 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2280 buf[2] = 0x80 + (c & 0x3f); 2281 return 3; 2282 } 2283 if (c < 0x200000) /* 21 bits */ 2284 { 2285 buf[0] = 0xf0 + ((unsigned)c >> 18); 2286 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2287 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2288 buf[3] = 0x80 + (c & 0x3f); 2289 return 4; 2290 } 2291 if (c < 0x4000000) /* 26 bits */ 2292 { 2293 buf[0] = 0xf8 + ((unsigned)c >> 24); 2294 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2295 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2296 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2297 buf[4] = 0x80 + (c & 0x3f); 2298 return 5; 2299 } 2300 /* 31 bits */ 2301 buf[0] = 0xfc + ((unsigned)c >> 30); 2302 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f); 2303 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2304 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2305 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2306 buf[5] = 0x80 + (c & 0x3f); 2307 return 6; 2308 } 2309 2310 #if defined(FEAT_TERMINAL) || defined(PROTO) 2311 /* 2312 * utf_iscomposing() with different argument type for libvterm. 2313 */ 2314 int 2315 utf_iscomposing_uint(UINT32_T c) 2316 { 2317 return utf_iscomposing((int)c); 2318 } 2319 #endif 2320 2321 /* 2322 * Return TRUE if "c" is a composing UTF-8 character. This means it will be 2323 * drawn on top of the preceding character. 2324 * Based on code from Markus Kuhn. 2325 */ 2326 int 2327 utf_iscomposing(int c) 2328 { 2329 /* Sorted list of non-overlapping intervals. 2330 * Generated by ../runtime/tools/unicode.vim. */ 2331 static struct interval combining[] = 2332 { 2333 {0x0300, 0x036f}, 2334 {0x0483, 0x0489}, 2335 {0x0591, 0x05bd}, 2336 {0x05bf, 0x05bf}, 2337 {0x05c1, 0x05c2}, 2338 {0x05c4, 0x05c5}, 2339 {0x05c7, 0x05c7}, 2340 {0x0610, 0x061a}, 2341 {0x064b, 0x065f}, 2342 {0x0670, 0x0670}, 2343 {0x06d6, 0x06dc}, 2344 {0x06df, 0x06e4}, 2345 {0x06e7, 0x06e8}, 2346 {0x06ea, 0x06ed}, 2347 {0x0711, 0x0711}, 2348 {0x0730, 0x074a}, 2349 {0x07a6, 0x07b0}, 2350 {0x07eb, 0x07f3}, 2351 {0x07fd, 0x07fd}, 2352 {0x0816, 0x0819}, 2353 {0x081b, 0x0823}, 2354 {0x0825, 0x0827}, 2355 {0x0829, 0x082d}, 2356 {0x0859, 0x085b}, 2357 {0x08d3, 0x08e1}, 2358 {0x08e3, 0x0903}, 2359 {0x093a, 0x093c}, 2360 {0x093e, 0x094f}, 2361 {0x0951, 0x0957}, 2362 {0x0962, 0x0963}, 2363 {0x0981, 0x0983}, 2364 {0x09bc, 0x09bc}, 2365 {0x09be, 0x09c4}, 2366 {0x09c7, 0x09c8}, 2367 {0x09cb, 0x09cd}, 2368 {0x09d7, 0x09d7}, 2369 {0x09e2, 0x09e3}, 2370 {0x09fe, 0x09fe}, 2371 {0x0a01, 0x0a03}, 2372 {0x0a3c, 0x0a3c}, 2373 {0x0a3e, 0x0a42}, 2374 {0x0a47, 0x0a48}, 2375 {0x0a4b, 0x0a4d}, 2376 {0x0a51, 0x0a51}, 2377 {0x0a70, 0x0a71}, 2378 {0x0a75, 0x0a75}, 2379 {0x0a81, 0x0a83}, 2380 {0x0abc, 0x0abc}, 2381 {0x0abe, 0x0ac5}, 2382 {0x0ac7, 0x0ac9}, 2383 {0x0acb, 0x0acd}, 2384 {0x0ae2, 0x0ae3}, 2385 {0x0afa, 0x0aff}, 2386 {0x0b01, 0x0b03}, 2387 {0x0b3c, 0x0b3c}, 2388 {0x0b3e, 0x0b44}, 2389 {0x0b47, 0x0b48}, 2390 {0x0b4b, 0x0b4d}, 2391 {0x0b56, 0x0b57}, 2392 {0x0b62, 0x0b63}, 2393 {0x0b82, 0x0b82}, 2394 {0x0bbe, 0x0bc2}, 2395 {0x0bc6, 0x0bc8}, 2396 {0x0bca, 0x0bcd}, 2397 {0x0bd7, 0x0bd7}, 2398 {0x0c00, 0x0c04}, 2399 {0x0c3e, 0x0c44}, 2400 {0x0c46, 0x0c48}, 2401 {0x0c4a, 0x0c4d}, 2402 {0x0c55, 0x0c56}, 2403 {0x0c62, 0x0c63}, 2404 {0x0c81, 0x0c83}, 2405 {0x0cbc, 0x0cbc}, 2406 {0x0cbe, 0x0cc4}, 2407 {0x0cc6, 0x0cc8}, 2408 {0x0cca, 0x0ccd}, 2409 {0x0cd5, 0x0cd6}, 2410 {0x0ce2, 0x0ce3}, 2411 {0x0d00, 0x0d03}, 2412 {0x0d3b, 0x0d3c}, 2413 {0x0d3e, 0x0d44}, 2414 {0x0d46, 0x0d48}, 2415 {0x0d4a, 0x0d4d}, 2416 {0x0d57, 0x0d57}, 2417 {0x0d62, 0x0d63}, 2418 {0x0d82, 0x0d83}, 2419 {0x0dca, 0x0dca}, 2420 {0x0dcf, 0x0dd4}, 2421 {0x0dd6, 0x0dd6}, 2422 {0x0dd8, 0x0ddf}, 2423 {0x0df2, 0x0df3}, 2424 {0x0e31, 0x0e31}, 2425 {0x0e34, 0x0e3a}, 2426 {0x0e47, 0x0e4e}, 2427 {0x0eb1, 0x0eb1}, 2428 {0x0eb4, 0x0eb9}, 2429 {0x0ebb, 0x0ebc}, 2430 {0x0ec8, 0x0ecd}, 2431 {0x0f18, 0x0f19}, 2432 {0x0f35, 0x0f35}, 2433 {0x0f37, 0x0f37}, 2434 {0x0f39, 0x0f39}, 2435 {0x0f3e, 0x0f3f}, 2436 {0x0f71, 0x0f84}, 2437 {0x0f86, 0x0f87}, 2438 {0x0f8d, 0x0f97}, 2439 {0x0f99, 0x0fbc}, 2440 {0x0fc6, 0x0fc6}, 2441 {0x102b, 0x103e}, 2442 {0x1056, 0x1059}, 2443 {0x105e, 0x1060}, 2444 {0x1062, 0x1064}, 2445 {0x1067, 0x106d}, 2446 {0x1071, 0x1074}, 2447 {0x1082, 0x108d}, 2448 {0x108f, 0x108f}, 2449 {0x109a, 0x109d}, 2450 {0x135d, 0x135f}, 2451 {0x1712, 0x1714}, 2452 {0x1732, 0x1734}, 2453 {0x1752, 0x1753}, 2454 {0x1772, 0x1773}, 2455 {0x17b4, 0x17d3}, 2456 {0x17dd, 0x17dd}, 2457 {0x180b, 0x180d}, 2458 {0x1885, 0x1886}, 2459 {0x18a9, 0x18a9}, 2460 {0x1920, 0x192b}, 2461 {0x1930, 0x193b}, 2462 {0x1a17, 0x1a1b}, 2463 {0x1a55, 0x1a5e}, 2464 {0x1a60, 0x1a7c}, 2465 {0x1a7f, 0x1a7f}, 2466 {0x1ab0, 0x1abe}, 2467 {0x1b00, 0x1b04}, 2468 {0x1b34, 0x1b44}, 2469 {0x1b6b, 0x1b73}, 2470 {0x1b80, 0x1b82}, 2471 {0x1ba1, 0x1bad}, 2472 {0x1be6, 0x1bf3}, 2473 {0x1c24, 0x1c37}, 2474 {0x1cd0, 0x1cd2}, 2475 {0x1cd4, 0x1ce8}, 2476 {0x1ced, 0x1ced}, 2477 {0x1cf2, 0x1cf4}, 2478 {0x1cf7, 0x1cf9}, 2479 {0x1dc0, 0x1df9}, 2480 {0x1dfb, 0x1dff}, 2481 {0x20d0, 0x20f0}, 2482 {0x2cef, 0x2cf1}, 2483 {0x2d7f, 0x2d7f}, 2484 {0x2de0, 0x2dff}, 2485 {0x302a, 0x302f}, 2486 {0x3099, 0x309a}, 2487 {0xa66f, 0xa672}, 2488 {0xa674, 0xa67d}, 2489 {0xa69e, 0xa69f}, 2490 {0xa6f0, 0xa6f1}, 2491 {0xa802, 0xa802}, 2492 {0xa806, 0xa806}, 2493 {0xa80b, 0xa80b}, 2494 {0xa823, 0xa827}, 2495 {0xa880, 0xa881}, 2496 {0xa8b4, 0xa8c5}, 2497 {0xa8e0, 0xa8f1}, 2498 {0xa8ff, 0xa8ff}, 2499 {0xa926, 0xa92d}, 2500 {0xa947, 0xa953}, 2501 {0xa980, 0xa983}, 2502 {0xa9b3, 0xa9c0}, 2503 {0xa9e5, 0xa9e5}, 2504 {0xaa29, 0xaa36}, 2505 {0xaa43, 0xaa43}, 2506 {0xaa4c, 0xaa4d}, 2507 {0xaa7b, 0xaa7d}, 2508 {0xaab0, 0xaab0}, 2509 {0xaab2, 0xaab4}, 2510 {0xaab7, 0xaab8}, 2511 {0xaabe, 0xaabf}, 2512 {0xaac1, 0xaac1}, 2513 {0xaaeb, 0xaaef}, 2514 {0xaaf5, 0xaaf6}, 2515 {0xabe3, 0xabea}, 2516 {0xabec, 0xabed}, 2517 {0xfb1e, 0xfb1e}, 2518 {0xfe00, 0xfe0f}, 2519 {0xfe20, 0xfe2f}, 2520 {0x101fd, 0x101fd}, 2521 {0x102e0, 0x102e0}, 2522 {0x10376, 0x1037a}, 2523 {0x10a01, 0x10a03}, 2524 {0x10a05, 0x10a06}, 2525 {0x10a0c, 0x10a0f}, 2526 {0x10a38, 0x10a3a}, 2527 {0x10a3f, 0x10a3f}, 2528 {0x10ae5, 0x10ae6}, 2529 {0x10d24, 0x10d27}, 2530 {0x10f46, 0x10f50}, 2531 {0x11000, 0x11002}, 2532 {0x11038, 0x11046}, 2533 {0x1107f, 0x11082}, 2534 {0x110b0, 0x110ba}, 2535 {0x11100, 0x11102}, 2536 {0x11127, 0x11134}, 2537 {0x11145, 0x11146}, 2538 {0x11173, 0x11173}, 2539 {0x11180, 0x11182}, 2540 {0x111b3, 0x111c0}, 2541 {0x111c9, 0x111cc}, 2542 {0x1122c, 0x11237}, 2543 {0x1123e, 0x1123e}, 2544 {0x112df, 0x112ea}, 2545 {0x11300, 0x11303}, 2546 {0x1133b, 0x1133c}, 2547 {0x1133e, 0x11344}, 2548 {0x11347, 0x11348}, 2549 {0x1134b, 0x1134d}, 2550 {0x11357, 0x11357}, 2551 {0x11362, 0x11363}, 2552 {0x11366, 0x1136c}, 2553 {0x11370, 0x11374}, 2554 {0x11435, 0x11446}, 2555 {0x1145e, 0x1145e}, 2556 {0x114b0, 0x114c3}, 2557 {0x115af, 0x115b5}, 2558 {0x115b8, 0x115c0}, 2559 {0x115dc, 0x115dd}, 2560 {0x11630, 0x11640}, 2561 {0x116ab, 0x116b7}, 2562 {0x1171d, 0x1172b}, 2563 {0x1182c, 0x1183a}, 2564 {0x11a01, 0x11a0a}, 2565 {0x11a33, 0x11a39}, 2566 {0x11a3b, 0x11a3e}, 2567 {0x11a47, 0x11a47}, 2568 {0x11a51, 0x11a5b}, 2569 {0x11a8a, 0x11a99}, 2570 {0x11c2f, 0x11c36}, 2571 {0x11c38, 0x11c3f}, 2572 {0x11c92, 0x11ca7}, 2573 {0x11ca9, 0x11cb6}, 2574 {0x11d31, 0x11d36}, 2575 {0x11d3a, 0x11d3a}, 2576 {0x11d3c, 0x11d3d}, 2577 {0x11d3f, 0x11d45}, 2578 {0x11d47, 0x11d47}, 2579 {0x11d8a, 0x11d8e}, 2580 {0x11d90, 0x11d91}, 2581 {0x11d93, 0x11d97}, 2582 {0x11ef3, 0x11ef6}, 2583 {0x16af0, 0x16af4}, 2584 {0x16b30, 0x16b36}, 2585 {0x16f51, 0x16f7e}, 2586 {0x16f8f, 0x16f92}, 2587 {0x1bc9d, 0x1bc9e}, 2588 {0x1d165, 0x1d169}, 2589 {0x1d16d, 0x1d172}, 2590 {0x1d17b, 0x1d182}, 2591 {0x1d185, 0x1d18b}, 2592 {0x1d1aa, 0x1d1ad}, 2593 {0x1d242, 0x1d244}, 2594 {0x1da00, 0x1da36}, 2595 {0x1da3b, 0x1da6c}, 2596 {0x1da75, 0x1da75}, 2597 {0x1da84, 0x1da84}, 2598 {0x1da9b, 0x1da9f}, 2599 {0x1daa1, 0x1daaf}, 2600 {0x1e000, 0x1e006}, 2601 {0x1e008, 0x1e018}, 2602 {0x1e01b, 0x1e021}, 2603 {0x1e023, 0x1e024}, 2604 {0x1e026, 0x1e02a}, 2605 {0x1e8d0, 0x1e8d6}, 2606 {0x1e944, 0x1e94a}, 2607 {0xe0100, 0xe01ef} 2608 }; 2609 2610 return intable(combining, sizeof(combining), c); 2611 } 2612 2613 /* 2614 * Return TRUE for characters that can be displayed in a normal way. 2615 * Only for characters of 0x100 and above! 2616 */ 2617 int 2618 utf_printable(int c) 2619 { 2620 #ifdef USE_WCHAR_FUNCTIONS 2621 /* 2622 * Assume the iswprint() library function works better than our own stuff. 2623 */ 2624 return iswprint(c); 2625 #else 2626 /* Sorted list of non-overlapping intervals. 2627 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */ 2628 static struct interval nonprint[] = 2629 { 2630 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2631 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2632 {0xfffe, 0xffff} 2633 }; 2634 2635 return !intable(nonprint, sizeof(nonprint), c); 2636 #endif 2637 } 2638 2639 /* Sorted list of non-overlapping intervals of all Emoji characters, 2640 * based on http://unicode.org/emoji/charts/emoji-list.html */ 2641 static struct interval emoji_all[] = 2642 { 2643 {0x203c, 0x203c}, 2644 {0x2049, 0x2049}, 2645 {0x2122, 0x2122}, 2646 {0x2139, 0x2139}, 2647 {0x2194, 0x2199}, 2648 {0x21a9, 0x21aa}, 2649 {0x231a, 0x231b}, 2650 {0x2328, 0x2328}, 2651 {0x23cf, 0x23cf}, 2652 {0x23e9, 0x23f3}, 2653 {0x23f8, 0x23fa}, 2654 {0x24c2, 0x24c2}, 2655 {0x25aa, 0x25ab}, 2656 {0x25b6, 0x25b6}, 2657 {0x25c0, 0x25c0}, 2658 {0x25fb, 0x25fe}, 2659 {0x2600, 0x2604}, 2660 {0x260e, 0x260e}, 2661 {0x2611, 0x2611}, 2662 {0x2614, 0x2615}, 2663 {0x2618, 0x2618}, 2664 {0x261d, 0x261d}, 2665 {0x2620, 0x2620}, 2666 {0x2622, 0x2623}, 2667 {0x2626, 0x2626}, 2668 {0x262a, 0x262a}, 2669 {0x262e, 0x262f}, 2670 {0x2638, 0x263a}, 2671 {0x2640, 0x2640}, 2672 {0x2642, 0x2642}, 2673 {0x2648, 0x2653}, 2674 {0x265f, 0x2660}, 2675 {0x2663, 0x2663}, 2676 {0x2665, 0x2666}, 2677 {0x2668, 0x2668}, 2678 {0x267b, 0x267b}, 2679 {0x267e, 0x267f}, 2680 {0x2692, 0x2697}, 2681 {0x2699, 0x2699}, 2682 {0x269b, 0x269c}, 2683 {0x26a0, 0x26a1}, 2684 {0x26aa, 0x26ab}, 2685 {0x26b0, 0x26b1}, 2686 {0x26bd, 0x26be}, 2687 {0x26c4, 0x26c5}, 2688 {0x26c8, 0x26c8}, 2689 {0x26ce, 0x26cf}, 2690 {0x26d1, 0x26d1}, 2691 {0x26d3, 0x26d4}, 2692 {0x26e9, 0x26ea}, 2693 {0x26f0, 0x26f5}, 2694 {0x26f7, 0x26fa}, 2695 {0x26fd, 0x26fd}, 2696 {0x2702, 0x2702}, 2697 {0x2705, 0x2705}, 2698 {0x2708, 0x270d}, 2699 {0x270f, 0x270f}, 2700 {0x2712, 0x2712}, 2701 {0x2714, 0x2714}, 2702 {0x2716, 0x2716}, 2703 {0x271d, 0x271d}, 2704 {0x2721, 0x2721}, 2705 {0x2728, 0x2728}, 2706 {0x2733, 0x2734}, 2707 {0x2744, 0x2744}, 2708 {0x2747, 0x2747}, 2709 {0x274c, 0x274c}, 2710 {0x274e, 0x274e}, 2711 {0x2753, 0x2755}, 2712 {0x2757, 0x2757}, 2713 {0x2763, 0x2764}, 2714 {0x2795, 0x2797}, 2715 {0x27a1, 0x27a1}, 2716 {0x27b0, 0x27b0}, 2717 {0x27bf, 0x27bf}, 2718 {0x2934, 0x2935}, 2719 {0x2b05, 0x2b07}, 2720 {0x2b1b, 0x2b1c}, 2721 {0x2b50, 0x2b50}, 2722 {0x2b55, 0x2b55}, 2723 {0x3030, 0x3030}, 2724 {0x303d, 0x303d}, 2725 {0x3297, 0x3297}, 2726 {0x3299, 0x3299}, 2727 {0x1f004, 0x1f004}, 2728 {0x1f0cf, 0x1f0cf}, 2729 {0x1f170, 0x1f171}, 2730 {0x1f17e, 0x1f17f}, 2731 {0x1f18e, 0x1f18e}, 2732 {0x1f191, 0x1f19a}, 2733 {0x1f1e6, 0x1f1ff}, 2734 {0x1f201, 0x1f202}, 2735 {0x1f21a, 0x1f21a}, 2736 {0x1f22f, 0x1f22f}, 2737 {0x1f232, 0x1f23a}, 2738 {0x1f250, 0x1f251}, 2739 {0x1f300, 0x1f321}, 2740 {0x1f324, 0x1f393}, 2741 {0x1f396, 0x1f397}, 2742 {0x1f399, 0x1f39b}, 2743 {0x1f39e, 0x1f3f0}, 2744 {0x1f3f3, 0x1f3f5}, 2745 {0x1f3f7, 0x1f4fd}, 2746 {0x1f4ff, 0x1f53d}, 2747 {0x1f549, 0x1f54e}, 2748 {0x1f550, 0x1f567}, 2749 {0x1f56f, 0x1f570}, 2750 {0x1f573, 0x1f57a}, 2751 {0x1f587, 0x1f587}, 2752 {0x1f58a, 0x1f58d}, 2753 {0x1f590, 0x1f590}, 2754 {0x1f595, 0x1f596}, 2755 {0x1f5a4, 0x1f5a5}, 2756 {0x1f5a8, 0x1f5a8}, 2757 {0x1f5b1, 0x1f5b2}, 2758 {0x1f5bc, 0x1f5bc}, 2759 {0x1f5c2, 0x1f5c4}, 2760 {0x1f5d1, 0x1f5d3}, 2761 {0x1f5dc, 0x1f5de}, 2762 {0x1f5e1, 0x1f5e1}, 2763 {0x1f5e3, 0x1f5e3}, 2764 {0x1f5e8, 0x1f5e8}, 2765 {0x1f5ef, 0x1f5ef}, 2766 {0x1f5f3, 0x1f5f3}, 2767 {0x1f5fa, 0x1f64f}, 2768 {0x1f680, 0x1f6c5}, 2769 {0x1f6cb, 0x1f6d2}, 2770 {0x1f6e0, 0x1f6e5}, 2771 {0x1f6e9, 0x1f6e9}, 2772 {0x1f6eb, 0x1f6ec}, 2773 {0x1f6f0, 0x1f6f0}, 2774 {0x1f6f3, 0x1f6f9}, 2775 {0x1f910, 0x1f93a}, 2776 {0x1f93c, 0x1f93e}, 2777 {0x1f940, 0x1f945}, 2778 {0x1f947, 0x1f970}, 2779 {0x1f973, 0x1f976}, 2780 {0x1f97a, 0x1f97a}, 2781 {0x1f97c, 0x1f9a2}, 2782 {0x1f9b0, 0x1f9b9}, 2783 {0x1f9c0, 0x1f9c2}, 2784 {0x1f9d0, 0x1f9ff} 2785 }; 2786 2787 /* 2788 * Get class of a Unicode character. 2789 * 0: white space 2790 * 1: punctuation 2791 * 2 or bigger: some class of word character. 2792 */ 2793 int 2794 utf_class(int c) 2795 { 2796 return utf_class_buf(c, curbuf); 2797 } 2798 2799 int 2800 utf_class_buf(int c, buf_T *buf) 2801 { 2802 /* sorted list of non-overlapping intervals */ 2803 static struct clinterval 2804 { 2805 unsigned int first; 2806 unsigned int last; 2807 unsigned int class; 2808 } classes[] = 2809 { 2810 {0x037e, 0x037e, 1}, /* Greek question mark */ 2811 {0x0387, 0x0387, 1}, /* Greek ano teleia */ 2812 {0x055a, 0x055f, 1}, /* Armenian punctuation */ 2813 {0x0589, 0x0589, 1}, /* Armenian full stop */ 2814 {0x05be, 0x05be, 1}, 2815 {0x05c0, 0x05c0, 1}, 2816 {0x05c3, 0x05c3, 1}, 2817 {0x05f3, 0x05f4, 1}, 2818 {0x060c, 0x060c, 1}, 2819 {0x061b, 0x061b, 1}, 2820 {0x061f, 0x061f, 1}, 2821 {0x066a, 0x066d, 1}, 2822 {0x06d4, 0x06d4, 1}, 2823 {0x0700, 0x070d, 1}, /* Syriac punctuation */ 2824 {0x0964, 0x0965, 1}, 2825 {0x0970, 0x0970, 1}, 2826 {0x0df4, 0x0df4, 1}, 2827 {0x0e4f, 0x0e4f, 1}, 2828 {0x0e5a, 0x0e5b, 1}, 2829 {0x0f04, 0x0f12, 1}, 2830 {0x0f3a, 0x0f3d, 1}, 2831 {0x0f85, 0x0f85, 1}, 2832 {0x104a, 0x104f, 1}, /* Myanmar punctuation */ 2833 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */ 2834 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */ 2835 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */ 2836 {0x1680, 0x1680, 0}, 2837 {0x169b, 0x169c, 1}, 2838 {0x16eb, 0x16ed, 1}, 2839 {0x1735, 0x1736, 1}, 2840 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */ 2841 {0x1800, 0x180a, 1}, /* Mongolian punctuation */ 2842 {0x2000, 0x200b, 0}, /* spaces */ 2843 {0x200c, 0x2027, 1}, /* punctuation and symbols */ 2844 {0x2028, 0x2029, 0}, 2845 {0x202a, 0x202e, 1}, /* punctuation and symbols */ 2846 {0x202f, 0x202f, 0}, 2847 {0x2030, 0x205e, 1}, /* punctuation and symbols */ 2848 {0x205f, 0x205f, 0}, 2849 {0x2060, 0x27ff, 1}, /* punctuation and symbols */ 2850 {0x2070, 0x207f, 0x2070}, /* superscript */ 2851 {0x2080, 0x2094, 0x2080}, /* subscript */ 2852 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */ 2853 {0x2800, 0x28ff, 0x2800}, /* braille */ 2854 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */ 2855 {0x29d8, 0x29db, 1}, 2856 {0x29fc, 0x29fd, 1}, 2857 {0x2e00, 0x2e7f, 1}, /* supplemental punctuation */ 2858 {0x3000, 0x3000, 0}, /* ideographic space */ 2859 {0x3001, 0x3020, 1}, /* ideographic punctuation */ 2860 {0x3030, 0x3030, 1}, 2861 {0x303d, 0x303d, 1}, 2862 {0x3040, 0x309f, 0x3040}, /* Hiragana */ 2863 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */ 2864 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */ 2865 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */ 2866 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */ 2867 {0xfd3e, 0xfd3f, 1}, 2868 {0xfe30, 0xfe6b, 1}, /* punctuation forms */ 2869 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */ 2870 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */ 2871 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */ 2872 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */ 2873 {0x20000, 0x2a6df, 0x4e00}, /* CJK Ideographs */ 2874 {0x2a700, 0x2b73f, 0x4e00}, /* CJK Ideographs */ 2875 {0x2b740, 0x2b81f, 0x4e00}, /* CJK Ideographs */ 2876 {0x2f800, 0x2fa1f, 0x4e00}, /* CJK Ideographs */ 2877 }; 2878 2879 int bot = 0; 2880 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2881 int mid; 2882 2883 /* First quick check for Latin1 characters, use 'iskeyword'. */ 2884 if (c < 0x100) 2885 { 2886 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2887 return 0; /* blank */ 2888 if (vim_iswordc_buf(c, buf)) 2889 return 2; /* word character */ 2890 return 1; /* punctuation */ 2891 } 2892 2893 /* binary search in table */ 2894 while (top >= bot) 2895 { 2896 mid = (bot + top) / 2; 2897 if (classes[mid].last < (unsigned int)c) 2898 bot = mid + 1; 2899 else if (classes[mid].first > (unsigned int)c) 2900 top = mid - 1; 2901 else 2902 return (int)classes[mid].class; 2903 } 2904 2905 /* emoji */ 2906 if (intable(emoji_all, sizeof(emoji_all), c)) 2907 return 3; 2908 2909 /* most other characters are "word" characters */ 2910 return 2; 2911 } 2912 2913 int 2914 utf_ambiguous_width(int c) 2915 { 2916 return c >= 0x80 && (intable(ambiguous, sizeof(ambiguous), c) 2917 || intable(emoji_all, sizeof(emoji_all), c)); 2918 } 2919 2920 /* 2921 * Code for Unicode case-dependent operations. Based on notes in 2922 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2923 * This code uses simple case folding, not full case folding. 2924 * Last updated for Unicode 5.2. 2925 */ 2926 2927 /* 2928 * The following tables are built by ../runtime/tools/unicode.vim. 2929 * They must be in numeric order, because we use binary search. 2930 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2931 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2932 * folded/upper/lower by adding 32. 2933 */ 2934 typedef struct 2935 { 2936 int rangeStart; 2937 int rangeEnd; 2938 int step; 2939 int offset; 2940 } convertStruct; 2941 2942 static convertStruct foldCase[] = 2943 { 2944 {0x41,0x5a,1,32}, 2945 {0xb5,0xb5,-1,775}, 2946 {0xc0,0xd6,1,32}, 2947 {0xd8,0xde,1,32}, 2948 {0x100,0x12e,2,1}, 2949 {0x132,0x136,2,1}, 2950 {0x139,0x147,2,1}, 2951 {0x14a,0x176,2,1}, 2952 {0x178,0x178,-1,-121}, 2953 {0x179,0x17d,2,1}, 2954 {0x17f,0x17f,-1,-268}, 2955 {0x181,0x181,-1,210}, 2956 {0x182,0x184,2,1}, 2957 {0x186,0x186,-1,206}, 2958 {0x187,0x187,-1,1}, 2959 {0x189,0x18a,1,205}, 2960 {0x18b,0x18b,-1,1}, 2961 {0x18e,0x18e,-1,79}, 2962 {0x18f,0x18f,-1,202}, 2963 {0x190,0x190,-1,203}, 2964 {0x191,0x191,-1,1}, 2965 {0x193,0x193,-1,205}, 2966 {0x194,0x194,-1,207}, 2967 {0x196,0x196,-1,211}, 2968 {0x197,0x197,-1,209}, 2969 {0x198,0x198,-1,1}, 2970 {0x19c,0x19c,-1,211}, 2971 {0x19d,0x19d,-1,213}, 2972 {0x19f,0x19f,-1,214}, 2973 {0x1a0,0x1a4,2,1}, 2974 {0x1a6,0x1a6,-1,218}, 2975 {0x1a7,0x1a7,-1,1}, 2976 {0x1a9,0x1a9,-1,218}, 2977 {0x1ac,0x1ac,-1,1}, 2978 {0x1ae,0x1ae,-1,218}, 2979 {0x1af,0x1af,-1,1}, 2980 {0x1b1,0x1b2,1,217}, 2981 {0x1b3,0x1b5,2,1}, 2982 {0x1b7,0x1b7,-1,219}, 2983 {0x1b8,0x1bc,4,1}, 2984 {0x1c4,0x1c4,-1,2}, 2985 {0x1c5,0x1c5,-1,1}, 2986 {0x1c7,0x1c7,-1,2}, 2987 {0x1c8,0x1c8,-1,1}, 2988 {0x1ca,0x1ca,-1,2}, 2989 {0x1cb,0x1db,2,1}, 2990 {0x1de,0x1ee,2,1}, 2991 {0x1f1,0x1f1,-1,2}, 2992 {0x1f2,0x1f4,2,1}, 2993 {0x1f6,0x1f6,-1,-97}, 2994 {0x1f7,0x1f7,-1,-56}, 2995 {0x1f8,0x21e,2,1}, 2996 {0x220,0x220,-1,-130}, 2997 {0x222,0x232,2,1}, 2998 {0x23a,0x23a,-1,10795}, 2999 {0x23b,0x23b,-1,1}, 3000 {0x23d,0x23d,-1,-163}, 3001 {0x23e,0x23e,-1,10792}, 3002 {0x241,0x241,-1,1}, 3003 {0x243,0x243,-1,-195}, 3004 {0x244,0x244,-1,69}, 3005 {0x245,0x245,-1,71}, 3006 {0x246,0x24e,2,1}, 3007 {0x345,0x345,-1,116}, 3008 {0x370,0x372,2,1}, 3009 {0x376,0x376,-1,1}, 3010 {0x37f,0x37f,-1,116}, 3011 {0x386,0x386,-1,38}, 3012 {0x388,0x38a,1,37}, 3013 {0x38c,0x38c,-1,64}, 3014 {0x38e,0x38f,1,63}, 3015 {0x391,0x3a1,1,32}, 3016 {0x3a3,0x3ab,1,32}, 3017 {0x3c2,0x3c2,-1,1}, 3018 {0x3cf,0x3cf,-1,8}, 3019 {0x3d0,0x3d0,-1,-30}, 3020 {0x3d1,0x3d1,-1,-25}, 3021 {0x3d5,0x3d5,-1,-15}, 3022 {0x3d6,0x3d6,-1,-22}, 3023 {0x3d8,0x3ee,2,1}, 3024 {0x3f0,0x3f0,-1,-54}, 3025 {0x3f1,0x3f1,-1,-48}, 3026 {0x3f4,0x3f4,-1,-60}, 3027 {0x3f5,0x3f5,-1,-64}, 3028 {0x3f7,0x3f7,-1,1}, 3029 {0x3f9,0x3f9,-1,-7}, 3030 {0x3fa,0x3fa,-1,1}, 3031 {0x3fd,0x3ff,1,-130}, 3032 {0x400,0x40f,1,80}, 3033 {0x410,0x42f,1,32}, 3034 {0x460,0x480,2,1}, 3035 {0x48a,0x4be,2,1}, 3036 {0x4c0,0x4c0,-1,15}, 3037 {0x4c1,0x4cd,2,1}, 3038 {0x4d0,0x52e,2,1}, 3039 {0x531,0x556,1,48}, 3040 {0x10a0,0x10c5,1,7264}, 3041 {0x10c7,0x10cd,6,7264}, 3042 {0x13f8,0x13fd,1,-8}, 3043 {0x1c80,0x1c80,-1,-6222}, 3044 {0x1c81,0x1c81,-1,-6221}, 3045 {0x1c82,0x1c82,-1,-6212}, 3046 {0x1c83,0x1c84,1,-6210}, 3047 {0x1c85,0x1c85,-1,-6211}, 3048 {0x1c86,0x1c86,-1,-6204}, 3049 {0x1c87,0x1c87,-1,-6180}, 3050 {0x1c88,0x1c88,-1,35267}, 3051 {0x1c90,0x1cba,1,-3008}, 3052 {0x1cbd,0x1cbf,1,-3008}, 3053 {0x1e00,0x1e94,2,1}, 3054 {0x1e9b,0x1e9b,-1,-58}, 3055 {0x1e9e,0x1e9e,-1,-7615}, 3056 {0x1ea0,0x1efe,2,1}, 3057 {0x1f08,0x1f0f,1,-8}, 3058 {0x1f18,0x1f1d,1,-8}, 3059 {0x1f28,0x1f2f,1,-8}, 3060 {0x1f38,0x1f3f,1,-8}, 3061 {0x1f48,0x1f4d,1,-8}, 3062 {0x1f59,0x1f5f,2,-8}, 3063 {0x1f68,0x1f6f,1,-8}, 3064 {0x1f88,0x1f8f,1,-8}, 3065 {0x1f98,0x1f9f,1,-8}, 3066 {0x1fa8,0x1faf,1,-8}, 3067 {0x1fb8,0x1fb9,1,-8}, 3068 {0x1fba,0x1fbb,1,-74}, 3069 {0x1fbc,0x1fbc,-1,-9}, 3070 {0x1fbe,0x1fbe,-1,-7173}, 3071 {0x1fc8,0x1fcb,1,-86}, 3072 {0x1fcc,0x1fcc,-1,-9}, 3073 {0x1fd8,0x1fd9,1,-8}, 3074 {0x1fda,0x1fdb,1,-100}, 3075 {0x1fe8,0x1fe9,1,-8}, 3076 {0x1fea,0x1feb,1,-112}, 3077 {0x1fec,0x1fec,-1,-7}, 3078 {0x1ff8,0x1ff9,1,-128}, 3079 {0x1ffa,0x1ffb,1,-126}, 3080 {0x1ffc,0x1ffc,-1,-9}, 3081 {0x2126,0x2126,-1,-7517}, 3082 {0x212a,0x212a,-1,-8383}, 3083 {0x212b,0x212b,-1,-8262}, 3084 {0x2132,0x2132,-1,28}, 3085 {0x2160,0x216f,1,16}, 3086 {0x2183,0x2183,-1,1}, 3087 {0x24b6,0x24cf,1,26}, 3088 {0x2c00,0x2c2e,1,48}, 3089 {0x2c60,0x2c60,-1,1}, 3090 {0x2c62,0x2c62,-1,-10743}, 3091 {0x2c63,0x2c63,-1,-3814}, 3092 {0x2c64,0x2c64,-1,-10727}, 3093 {0x2c67,0x2c6b,2,1}, 3094 {0x2c6d,0x2c6d,-1,-10780}, 3095 {0x2c6e,0x2c6e,-1,-10749}, 3096 {0x2c6f,0x2c6f,-1,-10783}, 3097 {0x2c70,0x2c70,-1,-10782}, 3098 {0x2c72,0x2c75,3,1}, 3099 {0x2c7e,0x2c7f,1,-10815}, 3100 {0x2c80,0x2ce2,2,1}, 3101 {0x2ceb,0x2ced,2,1}, 3102 {0x2cf2,0xa640,31054,1}, 3103 {0xa642,0xa66c,2,1}, 3104 {0xa680,0xa69a,2,1}, 3105 {0xa722,0xa72e,2,1}, 3106 {0xa732,0xa76e,2,1}, 3107 {0xa779,0xa77b,2,1}, 3108 {0xa77d,0xa77d,-1,-35332}, 3109 {0xa77e,0xa786,2,1}, 3110 {0xa78b,0xa78b,-1,1}, 3111 {0xa78d,0xa78d,-1,-42280}, 3112 {0xa790,0xa792,2,1}, 3113 {0xa796,0xa7a8,2,1}, 3114 {0xa7aa,0xa7aa,-1,-42308}, 3115 {0xa7ab,0xa7ab,-1,-42319}, 3116 {0xa7ac,0xa7ac,-1,-42315}, 3117 {0xa7ad,0xa7ad,-1,-42305}, 3118 {0xa7ae,0xa7ae,-1,-42308}, 3119 {0xa7b0,0xa7b0,-1,-42258}, 3120 {0xa7b1,0xa7b1,-1,-42282}, 3121 {0xa7b2,0xa7b2,-1,-42261}, 3122 {0xa7b3,0xa7b3,-1,928}, 3123 {0xa7b4,0xa7b8,2,1}, 3124 {0xab70,0xabbf,1,-38864}, 3125 {0xff21,0xff3a,1,32}, 3126 {0x10400,0x10427,1,40}, 3127 {0x104b0,0x104d3,1,40}, 3128 {0x10c80,0x10cb2,1,64}, 3129 {0x118a0,0x118bf,1,32}, 3130 {0x16e40,0x16e5f,1,32}, 3131 {0x1e900,0x1e921,1,34} 3132 }; 3133 3134 /* 3135 * Generic conversion function for case operations. 3136 * Return the converted equivalent of "a", which is a UCS-4 character. Use 3137 * the given conversion "table". Uses binary search on "table". 3138 */ 3139 static int 3140 utf_convert( 3141 int a, 3142 convertStruct table[], 3143 int tableSize) 3144 { 3145 int start, mid, end; /* indices into table */ 3146 int entries = tableSize / sizeof(convertStruct); 3147 3148 start = 0; 3149 end = entries; 3150 while (start < end) 3151 { 3152 /* need to search further */ 3153 mid = (end + start) / 2; 3154 if (table[mid].rangeEnd < a) 3155 start = mid + 1; 3156 else 3157 end = mid; 3158 } 3159 if (start < entries 3160 && table[start].rangeStart <= a 3161 && a <= table[start].rangeEnd 3162 && (a - table[start].rangeStart) % table[start].step == 0) 3163 return (a + table[start].offset); 3164 else 3165 return a; 3166 } 3167 3168 /* 3169 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 3170 * simple case folding. 3171 */ 3172 int 3173 utf_fold(int a) 3174 { 3175 if (a < 0x80) 3176 /* be fast for ASCII */ 3177 return a >= 0x41 && a <= 0x5a ? a + 32 : a; 3178 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 3179 } 3180 3181 static convertStruct toLower[] = 3182 { 3183 {0x41,0x5a,1,32}, 3184 {0xc0,0xd6,1,32}, 3185 {0xd8,0xde,1,32}, 3186 {0x100,0x12e,2,1}, 3187 {0x130,0x130,-1,-199}, 3188 {0x132,0x136,2,1}, 3189 {0x139,0x147,2,1}, 3190 {0x14a,0x176,2,1}, 3191 {0x178,0x178,-1,-121}, 3192 {0x179,0x17d,2,1}, 3193 {0x181,0x181,-1,210}, 3194 {0x182,0x184,2,1}, 3195 {0x186,0x186,-1,206}, 3196 {0x187,0x187,-1,1}, 3197 {0x189,0x18a,1,205}, 3198 {0x18b,0x18b,-1,1}, 3199 {0x18e,0x18e,-1,79}, 3200 {0x18f,0x18f,-1,202}, 3201 {0x190,0x190,-1,203}, 3202 {0x191,0x191,-1,1}, 3203 {0x193,0x193,-1,205}, 3204 {0x194,0x194,-1,207}, 3205 {0x196,0x196,-1,211}, 3206 {0x197,0x197,-1,209}, 3207 {0x198,0x198,-1,1}, 3208 {0x19c,0x19c,-1,211}, 3209 {0x19d,0x19d,-1,213}, 3210 {0x19f,0x19f,-1,214}, 3211 {0x1a0,0x1a4,2,1}, 3212 {0x1a6,0x1a6,-1,218}, 3213 {0x1a7,0x1a7,-1,1}, 3214 {0x1a9,0x1a9,-1,218}, 3215 {0x1ac,0x1ac,-1,1}, 3216 {0x1ae,0x1ae,-1,218}, 3217 {0x1af,0x1af,-1,1}, 3218 {0x1b1,0x1b2,1,217}, 3219 {0x1b3,0x1b5,2,1}, 3220 {0x1b7,0x1b7,-1,219}, 3221 {0x1b8,0x1bc,4,1}, 3222 {0x1c4,0x1c4,-1,2}, 3223 {0x1c5,0x1c5,-1,1}, 3224 {0x1c7,0x1c7,-1,2}, 3225 {0x1c8,0x1c8,-1,1}, 3226 {0x1ca,0x1ca,-1,2}, 3227 {0x1cb,0x1db,2,1}, 3228 {0x1de,0x1ee,2,1}, 3229 {0x1f1,0x1f1,-1,2}, 3230 {0x1f2,0x1f4,2,1}, 3231 {0x1f6,0x1f6,-1,-97}, 3232 {0x1f7,0x1f7,-1,-56}, 3233 {0x1f8,0x21e,2,1}, 3234 {0x220,0x220,-1,-130}, 3235 {0x222,0x232,2,1}, 3236 {0x23a,0x23a,-1,10795}, 3237 {0x23b,0x23b,-1,1}, 3238 {0x23d,0x23d,-1,-163}, 3239 {0x23e,0x23e,-1,10792}, 3240 {0x241,0x241,-1,1}, 3241 {0x243,0x243,-1,-195}, 3242 {0x244,0x244,-1,69}, 3243 {0x245,0x245,-1,71}, 3244 {0x246,0x24e,2,1}, 3245 {0x370,0x372,2,1}, 3246 {0x376,0x376,-1,1}, 3247 {0x37f,0x37f,-1,116}, 3248 {0x386,0x386,-1,38}, 3249 {0x388,0x38a,1,37}, 3250 {0x38c,0x38c,-1,64}, 3251 {0x38e,0x38f,1,63}, 3252 {0x391,0x3a1,1,32}, 3253 {0x3a3,0x3ab,1,32}, 3254 {0x3cf,0x3cf,-1,8}, 3255 {0x3d8,0x3ee,2,1}, 3256 {0x3f4,0x3f4,-1,-60}, 3257 {0x3f7,0x3f7,-1,1}, 3258 {0x3f9,0x3f9,-1,-7}, 3259 {0x3fa,0x3fa,-1,1}, 3260 {0x3fd,0x3ff,1,-130}, 3261 {0x400,0x40f,1,80}, 3262 {0x410,0x42f,1,32}, 3263 {0x460,0x480,2,1}, 3264 {0x48a,0x4be,2,1}, 3265 {0x4c0,0x4c0,-1,15}, 3266 {0x4c1,0x4cd,2,1}, 3267 {0x4d0,0x52e,2,1}, 3268 {0x531,0x556,1,48}, 3269 {0x10a0,0x10c5,1,7264}, 3270 {0x10c7,0x10cd,6,7264}, 3271 {0x13a0,0x13ef,1,38864}, 3272 {0x13f0,0x13f5,1,8}, 3273 {0x1c90,0x1cba,1,-3008}, 3274 {0x1cbd,0x1cbf,1,-3008}, 3275 {0x1e00,0x1e94,2,1}, 3276 {0x1e9e,0x1e9e,-1,-7615}, 3277 {0x1ea0,0x1efe,2,1}, 3278 {0x1f08,0x1f0f,1,-8}, 3279 {0x1f18,0x1f1d,1,-8}, 3280 {0x1f28,0x1f2f,1,-8}, 3281 {0x1f38,0x1f3f,1,-8}, 3282 {0x1f48,0x1f4d,1,-8}, 3283 {0x1f59,0x1f5f,2,-8}, 3284 {0x1f68,0x1f6f,1,-8}, 3285 {0x1f88,0x1f8f,1,-8}, 3286 {0x1f98,0x1f9f,1,-8}, 3287 {0x1fa8,0x1faf,1,-8}, 3288 {0x1fb8,0x1fb9,1,-8}, 3289 {0x1fba,0x1fbb,1,-74}, 3290 {0x1fbc,0x1fbc,-1,-9}, 3291 {0x1fc8,0x1fcb,1,-86}, 3292 {0x1fcc,0x1fcc,-1,-9}, 3293 {0x1fd8,0x1fd9,1,-8}, 3294 {0x1fda,0x1fdb,1,-100}, 3295 {0x1fe8,0x1fe9,1,-8}, 3296 {0x1fea,0x1feb,1,-112}, 3297 {0x1fec,0x1fec,-1,-7}, 3298 {0x1ff8,0x1ff9,1,-128}, 3299 {0x1ffa,0x1ffb,1,-126}, 3300 {0x1ffc,0x1ffc,-1,-9}, 3301 {0x2126,0x2126,-1,-7517}, 3302 {0x212a,0x212a,-1,-8383}, 3303 {0x212b,0x212b,-1,-8262}, 3304 {0x2132,0x2132,-1,28}, 3305 {0x2160,0x216f,1,16}, 3306 {0x2183,0x2183,-1,1}, 3307 {0x24b6,0x24cf,1,26}, 3308 {0x2c00,0x2c2e,1,48}, 3309 {0x2c60,0x2c60,-1,1}, 3310 {0x2c62,0x2c62,-1,-10743}, 3311 {0x2c63,0x2c63,-1,-3814}, 3312 {0x2c64,0x2c64,-1,-10727}, 3313 {0x2c67,0x2c6b,2,1}, 3314 {0x2c6d,0x2c6d,-1,-10780}, 3315 {0x2c6e,0x2c6e,-1,-10749}, 3316 {0x2c6f,0x2c6f,-1,-10783}, 3317 {0x2c70,0x2c70,-1,-10782}, 3318 {0x2c72,0x2c75,3,1}, 3319 {0x2c7e,0x2c7f,1,-10815}, 3320 {0x2c80,0x2ce2,2,1}, 3321 {0x2ceb,0x2ced,2,1}, 3322 {0x2cf2,0xa640,31054,1}, 3323 {0xa642,0xa66c,2,1}, 3324 {0xa680,0xa69a,2,1}, 3325 {0xa722,0xa72e,2,1}, 3326 {0xa732,0xa76e,2,1}, 3327 {0xa779,0xa77b,2,1}, 3328 {0xa77d,0xa77d,-1,-35332}, 3329 {0xa77e,0xa786,2,1}, 3330 {0xa78b,0xa78b,-1,1}, 3331 {0xa78d,0xa78d,-1,-42280}, 3332 {0xa790,0xa792,2,1}, 3333 {0xa796,0xa7a8,2,1}, 3334 {0xa7aa,0xa7aa,-1,-42308}, 3335 {0xa7ab,0xa7ab,-1,-42319}, 3336 {0xa7ac,0xa7ac,-1,-42315}, 3337 {0xa7ad,0xa7ad,-1,-42305}, 3338 {0xa7ae,0xa7ae,-1,-42308}, 3339 {0xa7b0,0xa7b0,-1,-42258}, 3340 {0xa7b1,0xa7b1,-1,-42282}, 3341 {0xa7b2,0xa7b2,-1,-42261}, 3342 {0xa7b3,0xa7b3,-1,928}, 3343 {0xa7b4,0xa7b8,2,1}, 3344 {0xff21,0xff3a,1,32}, 3345 {0x10400,0x10427,1,40}, 3346 {0x104b0,0x104d3,1,40}, 3347 {0x10c80,0x10cb2,1,64}, 3348 {0x118a0,0x118bf,1,32}, 3349 {0x16e40,0x16e5f,1,32}, 3350 {0x1e900,0x1e921,1,34} 3351 }; 3352 3353 static convertStruct toUpper[] = 3354 { 3355 {0x61,0x7a,1,-32}, 3356 {0xb5,0xb5,-1,743}, 3357 {0xe0,0xf6,1,-32}, 3358 {0xf8,0xfe,1,-32}, 3359 {0xff,0xff,-1,121}, 3360 {0x101,0x12f,2,-1}, 3361 {0x131,0x131,-1,-232}, 3362 {0x133,0x137,2,-1}, 3363 {0x13a,0x148,2,-1}, 3364 {0x14b,0x177,2,-1}, 3365 {0x17a,0x17e,2,-1}, 3366 {0x17f,0x17f,-1,-300}, 3367 {0x180,0x180,-1,195}, 3368 {0x183,0x185,2,-1}, 3369 {0x188,0x18c,4,-1}, 3370 {0x192,0x192,-1,-1}, 3371 {0x195,0x195,-1,97}, 3372 {0x199,0x199,-1,-1}, 3373 {0x19a,0x19a,-1,163}, 3374 {0x19e,0x19e,-1,130}, 3375 {0x1a1,0x1a5,2,-1}, 3376 {0x1a8,0x1ad,5,-1}, 3377 {0x1b0,0x1b4,4,-1}, 3378 {0x1b6,0x1b9,3,-1}, 3379 {0x1bd,0x1bd,-1,-1}, 3380 {0x1bf,0x1bf,-1,56}, 3381 {0x1c5,0x1c5,-1,-1}, 3382 {0x1c6,0x1c6,-1,-2}, 3383 {0x1c8,0x1c8,-1,-1}, 3384 {0x1c9,0x1c9,-1,-2}, 3385 {0x1cb,0x1cb,-1,-1}, 3386 {0x1cc,0x1cc,-1,-2}, 3387 {0x1ce,0x1dc,2,-1}, 3388 {0x1dd,0x1dd,-1,-79}, 3389 {0x1df,0x1ef,2,-1}, 3390 {0x1f2,0x1f2,-1,-1}, 3391 {0x1f3,0x1f3,-1,-2}, 3392 {0x1f5,0x1f9,4,-1}, 3393 {0x1fb,0x21f,2,-1}, 3394 {0x223,0x233,2,-1}, 3395 {0x23c,0x23c,-1,-1}, 3396 {0x23f,0x240,1,10815}, 3397 {0x242,0x247,5,-1}, 3398 {0x249,0x24f,2,-1}, 3399 {0x250,0x250,-1,10783}, 3400 {0x251,0x251,-1,10780}, 3401 {0x252,0x252,-1,10782}, 3402 {0x253,0x253,-1,-210}, 3403 {0x254,0x254,-1,-206}, 3404 {0x256,0x257,1,-205}, 3405 {0x259,0x259,-1,-202}, 3406 {0x25b,0x25b,-1,-203}, 3407 {0x25c,0x25c,-1,42319}, 3408 {0x260,0x260,-1,-205}, 3409 {0x261,0x261,-1,42315}, 3410 {0x263,0x263,-1,-207}, 3411 {0x265,0x265,-1,42280}, 3412 {0x266,0x266,-1,42308}, 3413 {0x268,0x268,-1,-209}, 3414 {0x269,0x269,-1,-211}, 3415 {0x26a,0x26a,-1,42308}, 3416 {0x26b,0x26b,-1,10743}, 3417 {0x26c,0x26c,-1,42305}, 3418 {0x26f,0x26f,-1,-211}, 3419 {0x271,0x271,-1,10749}, 3420 {0x272,0x272,-1,-213}, 3421 {0x275,0x275,-1,-214}, 3422 {0x27d,0x27d,-1,10727}, 3423 {0x280,0x283,3,-218}, 3424 {0x287,0x287,-1,42282}, 3425 {0x288,0x288,-1,-218}, 3426 {0x289,0x289,-1,-69}, 3427 {0x28a,0x28b,1,-217}, 3428 {0x28c,0x28c,-1,-71}, 3429 {0x292,0x292,-1,-219}, 3430 {0x29d,0x29d,-1,42261}, 3431 {0x29e,0x29e,-1,42258}, 3432 {0x345,0x345,-1,84}, 3433 {0x371,0x373,2,-1}, 3434 {0x377,0x377,-1,-1}, 3435 {0x37b,0x37d,1,130}, 3436 {0x3ac,0x3ac,-1,-38}, 3437 {0x3ad,0x3af,1,-37}, 3438 {0x3b1,0x3c1,1,-32}, 3439 {0x3c2,0x3c2,-1,-31}, 3440 {0x3c3,0x3cb,1,-32}, 3441 {0x3cc,0x3cc,-1,-64}, 3442 {0x3cd,0x3ce,1,-63}, 3443 {0x3d0,0x3d0,-1,-62}, 3444 {0x3d1,0x3d1,-1,-57}, 3445 {0x3d5,0x3d5,-1,-47}, 3446 {0x3d6,0x3d6,-1,-54}, 3447 {0x3d7,0x3d7,-1,-8}, 3448 {0x3d9,0x3ef,2,-1}, 3449 {0x3f0,0x3f0,-1,-86}, 3450 {0x3f1,0x3f1,-1,-80}, 3451 {0x3f2,0x3f2,-1,7}, 3452 {0x3f3,0x3f3,-1,-116}, 3453 {0x3f5,0x3f5,-1,-96}, 3454 {0x3f8,0x3fb,3,-1}, 3455 {0x430,0x44f,1,-32}, 3456 {0x450,0x45f,1,-80}, 3457 {0x461,0x481,2,-1}, 3458 {0x48b,0x4bf,2,-1}, 3459 {0x4c2,0x4ce,2,-1}, 3460 {0x4cf,0x4cf,-1,-15}, 3461 {0x4d1,0x52f,2,-1}, 3462 {0x561,0x586,1,-48}, 3463 {0x10d0,0x10fa,1,3008}, 3464 {0x10fd,0x10ff,1,3008}, 3465 {0x13f8,0x13fd,1,-8}, 3466 {0x1c80,0x1c80,-1,-6254}, 3467 {0x1c81,0x1c81,-1,-6253}, 3468 {0x1c82,0x1c82,-1,-6244}, 3469 {0x1c83,0x1c84,1,-6242}, 3470 {0x1c85,0x1c85,-1,-6243}, 3471 {0x1c86,0x1c86,-1,-6236}, 3472 {0x1c87,0x1c87,-1,-6181}, 3473 {0x1c88,0x1c88,-1,35266}, 3474 {0x1d79,0x1d79,-1,35332}, 3475 {0x1d7d,0x1d7d,-1,3814}, 3476 {0x1e01,0x1e95,2,-1}, 3477 {0x1e9b,0x1e9b,-1,-59}, 3478 {0x1ea1,0x1eff,2,-1}, 3479 {0x1f00,0x1f07,1,8}, 3480 {0x1f10,0x1f15,1,8}, 3481 {0x1f20,0x1f27,1,8}, 3482 {0x1f30,0x1f37,1,8}, 3483 {0x1f40,0x1f45,1,8}, 3484 {0x1f51,0x1f57,2,8}, 3485 {0x1f60,0x1f67,1,8}, 3486 {0x1f70,0x1f71,1,74}, 3487 {0x1f72,0x1f75,1,86}, 3488 {0x1f76,0x1f77,1,100}, 3489 {0x1f78,0x1f79,1,128}, 3490 {0x1f7a,0x1f7b,1,112}, 3491 {0x1f7c,0x1f7d,1,126}, 3492 {0x1f80,0x1f87,1,8}, 3493 {0x1f90,0x1f97,1,8}, 3494 {0x1fa0,0x1fa7,1,8}, 3495 {0x1fb0,0x1fb1,1,8}, 3496 {0x1fb3,0x1fb3,-1,9}, 3497 {0x1fbe,0x1fbe,-1,-7205}, 3498 {0x1fc3,0x1fc3,-1,9}, 3499 {0x1fd0,0x1fd1,1,8}, 3500 {0x1fe0,0x1fe1,1,8}, 3501 {0x1fe5,0x1fe5,-1,7}, 3502 {0x1ff3,0x1ff3,-1,9}, 3503 {0x214e,0x214e,-1,-28}, 3504 {0x2170,0x217f,1,-16}, 3505 {0x2184,0x2184,-1,-1}, 3506 {0x24d0,0x24e9,1,-26}, 3507 {0x2c30,0x2c5e,1,-48}, 3508 {0x2c61,0x2c61,-1,-1}, 3509 {0x2c65,0x2c65,-1,-10795}, 3510 {0x2c66,0x2c66,-1,-10792}, 3511 {0x2c68,0x2c6c,2,-1}, 3512 {0x2c73,0x2c76,3,-1}, 3513 {0x2c81,0x2ce3,2,-1}, 3514 {0x2cec,0x2cee,2,-1}, 3515 {0x2cf3,0x2cf3,-1,-1}, 3516 {0x2d00,0x2d25,1,-7264}, 3517 {0x2d27,0x2d2d,6,-7264}, 3518 {0xa641,0xa66d,2,-1}, 3519 {0xa681,0xa69b,2,-1}, 3520 {0xa723,0xa72f,2,-1}, 3521 {0xa733,0xa76f,2,-1}, 3522 {0xa77a,0xa77c,2,-1}, 3523 {0xa77f,0xa787,2,-1}, 3524 {0xa78c,0xa791,5,-1}, 3525 {0xa793,0xa797,4,-1}, 3526 {0xa799,0xa7a9,2,-1}, 3527 {0xa7b5,0xa7b9,2,-1}, 3528 {0xab53,0xab53,-1,-928}, 3529 {0xab70,0xabbf,1,-38864}, 3530 {0xff41,0xff5a,1,-32}, 3531 {0x10428,0x1044f,1,-40}, 3532 {0x104d8,0x104fb,1,-40}, 3533 {0x10cc0,0x10cf2,1,-64}, 3534 {0x118c0,0x118df,1,-32}, 3535 {0x16e60,0x16e7f,1,-32}, 3536 {0x1e922,0x1e943,1,-34} 3537 }; 3538 3539 /* 3540 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3541 * simple case folding. 3542 */ 3543 int 3544 utf_toupper(int a) 3545 { 3546 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */ 3547 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3548 return TOUPPER_ASC(a); 3549 3550 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3551 /* If towupper() is available and handles Unicode, use it. */ 3552 if (!(cmp_flags & CMP_INTERNAL)) 3553 return towupper(a); 3554 #endif 3555 3556 /* For characters below 128 use locale sensitive toupper(). */ 3557 if (a < 128) 3558 return TOUPPER_LOC(a); 3559 3560 /* For any other characters use the above mapping table. */ 3561 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3562 } 3563 3564 int 3565 utf_islower(int a) 3566 { 3567 /* German sharp s is lower case but has no upper case equivalent. */ 3568 return (utf_toupper(a) != a) || a == 0xdf; 3569 } 3570 3571 /* 3572 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3573 * simple case folding. 3574 */ 3575 int 3576 utf_tolower(int a) 3577 { 3578 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */ 3579 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3580 return TOLOWER_ASC(a); 3581 3582 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3583 /* If towlower() is available and handles Unicode, use it. */ 3584 if (!(cmp_flags & CMP_INTERNAL)) 3585 return towlower(a); 3586 #endif 3587 3588 /* For characters below 128 use locale sensitive tolower(). */ 3589 if (a < 128) 3590 return TOLOWER_LOC(a); 3591 3592 /* For any other characters use the above mapping table. */ 3593 return utf_convert(a, toLower, (int)sizeof(toLower)); 3594 } 3595 3596 int 3597 utf_isupper(int a) 3598 { 3599 return (utf_tolower(a) != a); 3600 } 3601 3602 static int 3603 utf_strnicmp( 3604 char_u *s1, 3605 char_u *s2, 3606 size_t n1, 3607 size_t n2) 3608 { 3609 int c1, c2, cdiff; 3610 char_u buffer[6]; 3611 3612 for (;;) 3613 { 3614 c1 = utf_safe_read_char_adv(&s1, &n1); 3615 c2 = utf_safe_read_char_adv(&s2, &n2); 3616 3617 if (c1 <= 0 || c2 <= 0) 3618 break; 3619 3620 if (c1 == c2) 3621 continue; 3622 3623 cdiff = utf_fold(c1) - utf_fold(c2); 3624 if (cdiff != 0) 3625 return cdiff; 3626 } 3627 3628 /* some string ended or has an incomplete/illegal character sequence */ 3629 3630 if (c1 == 0 || c2 == 0) 3631 { 3632 /* some string ended. shorter string is smaller */ 3633 if (c1 == 0 && c2 == 0) 3634 return 0; 3635 return c1 == 0 ? -1 : 1; 3636 } 3637 3638 /* Continue with bytewise comparison to produce some result that 3639 * would make comparison operations involving this function transitive. 3640 * 3641 * If only one string had an error, comparison should be made with 3642 * folded version of the other string. In this case it is enough 3643 * to fold just one character to determine the result of comparison. */ 3644 3645 if (c1 != -1 && c2 == -1) 3646 { 3647 n1 = utf_char2bytes(utf_fold(c1), buffer); 3648 s1 = buffer; 3649 } 3650 else if (c2 != -1 && c1 == -1) 3651 { 3652 n2 = utf_char2bytes(utf_fold(c2), buffer); 3653 s2 = buffer; 3654 } 3655 3656 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3657 { 3658 cdiff = (int)(*s1) - (int)(*s2); 3659 if (cdiff != 0) 3660 return cdiff; 3661 3662 s1++; 3663 s2++; 3664 n1--; 3665 n2--; 3666 } 3667 3668 if (n1 > 0 && *s1 == NUL) 3669 n1 = 0; 3670 if (n2 > 0 && *s2 == NUL) 3671 n2 = 0; 3672 3673 if (n1 == 0 && n2 == 0) 3674 return 0; 3675 return n1 == 0 ? -1 : 1; 3676 } 3677 3678 /* 3679 * Version of strnicmp() that handles multi-byte characters. 3680 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3681 * probably use strnicmp(), because there are no ASCII characters in the 3682 * second byte. 3683 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3684 * two characters otherwise. 3685 */ 3686 int 3687 mb_strnicmp(char_u *s1, char_u *s2, size_t nn) 3688 { 3689 int i, l; 3690 int cdiff; 3691 int n = (int)nn; 3692 3693 if (enc_utf8) 3694 { 3695 return utf_strnicmp(s1, s2, nn, nn); 3696 } 3697 else 3698 { 3699 for (i = 0; i < n; i += l) 3700 { 3701 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */ 3702 return 0; 3703 3704 l = (*mb_ptr2len)(s1 + i); 3705 if (l <= 1) 3706 { 3707 /* Single byte: first check normally, then with ignore case. */ 3708 if (s1[i] != s2[i]) 3709 { 3710 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3711 if (cdiff != 0) 3712 return cdiff; 3713 } 3714 } 3715 else 3716 { 3717 /* For non-Unicode multi-byte don't ignore case. */ 3718 if (l > n - i) 3719 l = n - i; 3720 cdiff = STRNCMP(s1 + i, s2 + i, l); 3721 if (cdiff != 0) 3722 return cdiff; 3723 } 3724 } 3725 } 3726 return 0; 3727 } 3728 3729 /* 3730 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3731 * 'encoding' has been set to. 3732 */ 3733 void 3734 show_utf8(void) 3735 { 3736 int len; 3737 int rlen = 0; 3738 char_u *line; 3739 int clen; 3740 int i; 3741 3742 /* Get the byte length of the char under the cursor, including composing 3743 * characters. */ 3744 line = ml_get_cursor(); 3745 len = utfc_ptr2len(line); 3746 if (len == 0) 3747 { 3748 MSG("NUL"); 3749 return; 3750 } 3751 3752 clen = 0; 3753 for (i = 0; i < len; ++i) 3754 { 3755 if (clen == 0) 3756 { 3757 /* start of (composing) character, get its length */ 3758 if (i > 0) 3759 { 3760 STRCPY(IObuff + rlen, "+ "); 3761 rlen += 2; 3762 } 3763 clen = utf_ptr2len(line + i); 3764 } 3765 sprintf((char *)IObuff + rlen, "%02x ", 3766 (line[i] == NL) ? NUL : line[i]); /* NUL is stored as NL */ 3767 --clen; 3768 rlen += (int)STRLEN(IObuff + rlen); 3769 if (rlen > IOSIZE - 20) 3770 break; 3771 } 3772 3773 msg(IObuff); 3774 } 3775 3776 /* 3777 * mb_head_off() function pointer. 3778 * Return offset from "p" to the first byte of the character it points into. 3779 * If "p" points to the NUL at the end of the string return 0. 3780 * Returns 0 when already at the first byte of a character. 3781 */ 3782 int 3783 latin_head_off(char_u *base UNUSED, char_u *p UNUSED) 3784 { 3785 return 0; 3786 } 3787 3788 int 3789 dbcs_head_off(char_u *base, char_u *p) 3790 { 3791 char_u *q; 3792 3793 /* It can't be a trailing byte when not using DBCS, at the start of the 3794 * string or the previous byte can't start a double-byte. */ 3795 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3796 return 0; 3797 3798 /* This is slow: need to start at the base and go forward until the 3799 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */ 3800 q = base; 3801 while (q < p) 3802 q += dbcs_ptr2len(q); 3803 return (q == p) ? 0 : 1; 3804 } 3805 3806 /* 3807 * Special version of dbcs_head_off() that works for ScreenLines[], where 3808 * single-width DBCS_JPNU characters are stored separately. 3809 */ 3810 int 3811 dbcs_screen_head_off(char_u *base, char_u *p) 3812 { 3813 char_u *q; 3814 3815 /* It can't be a trailing byte when not using DBCS, at the start of the 3816 * string or the previous byte can't start a double-byte. 3817 * For euc-jp an 0x8e byte in the previous cell always means we have a 3818 * lead byte in the current cell. */ 3819 if (p <= base 3820 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3821 || MB_BYTE2LEN(p[-1]) == 1 3822 || *p == NUL) 3823 return 0; 3824 3825 /* This is slow: need to start at the base and go forward until the 3826 * byte we are looking for. Return 1 when we went past it, 0 otherwise. 3827 * For DBCS_JPNU look out for 0x8e, which means the second byte is not 3828 * stored as the next byte. */ 3829 q = base; 3830 while (q < p) 3831 { 3832 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3833 ++q; 3834 else 3835 q += dbcs_ptr2len(q); 3836 } 3837 return (q == p) ? 0 : 1; 3838 } 3839 3840 int 3841 utf_head_off(char_u *base, char_u *p) 3842 { 3843 char_u *q; 3844 char_u *s; 3845 int c; 3846 int len; 3847 #ifdef FEAT_ARABIC 3848 char_u *j; 3849 #endif 3850 3851 if (*p < 0x80) /* be quick for ASCII */ 3852 return 0; 3853 3854 /* Skip backwards over trailing bytes: 10xx.xxxx 3855 * Skip backwards again if on a composing char. */ 3856 for (q = p; ; --q) 3857 { 3858 /* Move s to the last byte of this char. */ 3859 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3860 ; 3861 /* Move q to the first byte of this char. */ 3862 while (q > base && (*q & 0xc0) == 0x80) 3863 --q; 3864 /* Check for illegal sequence. Do allow an illegal byte after where we 3865 * started. */ 3866 len = utf8len_tab[*q]; 3867 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3868 return 0; 3869 3870 if (q <= base) 3871 break; 3872 3873 c = utf_ptr2char(q); 3874 if (utf_iscomposing(c)) 3875 continue; 3876 3877 #ifdef FEAT_ARABIC 3878 if (arabic_maycombine(c)) 3879 { 3880 /* Advance to get a sneak-peak at the next char */ 3881 j = q; 3882 --j; 3883 /* Move j to the first byte of this char. */ 3884 while (j > base && (*j & 0xc0) == 0x80) 3885 --j; 3886 if (arabic_combine(utf_ptr2char(j), c)) 3887 continue; 3888 } 3889 #endif 3890 break; 3891 } 3892 3893 return (int)(p - q); 3894 } 3895 3896 /* 3897 * Copy a character from "*fp" to "*tp" and advance the pointers. 3898 */ 3899 void 3900 mb_copy_char(char_u **fp, char_u **tp) 3901 { 3902 int l = (*mb_ptr2len)(*fp); 3903 3904 mch_memmove(*tp, *fp, (size_t)l); 3905 *tp += l; 3906 *fp += l; 3907 } 3908 3909 /* 3910 * Return the offset from "p" to the first byte of a character. When "p" is 3911 * at the start of a character 0 is returned, otherwise the offset to the next 3912 * character. Can start anywhere in a stream of bytes. 3913 */ 3914 int 3915 mb_off_next(char_u *base, char_u *p) 3916 { 3917 int i; 3918 int j; 3919 3920 if (enc_utf8) 3921 { 3922 if (*p < 0x80) /* be quick for ASCII */ 3923 return 0; 3924 3925 /* Find the next character that isn't 10xx.xxxx */ 3926 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 3927 ; 3928 if (i > 0) 3929 { 3930 /* Check for illegal sequence. */ 3931 for (j = 0; p - j > base; ++j) 3932 if ((p[-j] & 0xc0) != 0x80) 3933 break; 3934 if (utf8len_tab[p[-j]] != i + j) 3935 return 0; 3936 } 3937 return i; 3938 } 3939 3940 /* Only need to check if we're on a trail byte, it doesn't matter if we 3941 * want the offset to the next or current character. */ 3942 return (*mb_head_off)(base, p); 3943 } 3944 3945 /* 3946 * Return the offset from "p" to the last byte of the character it points 3947 * into. Can start anywhere in a stream of bytes. 3948 */ 3949 int 3950 mb_tail_off(char_u *base, char_u *p) 3951 { 3952 int i; 3953 int j; 3954 3955 if (*p == NUL) 3956 return 0; 3957 3958 if (enc_utf8) 3959 { 3960 /* Find the last character that is 10xx.xxxx */ 3961 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 3962 ; 3963 /* Check for illegal sequence. */ 3964 for (j = 0; p - j > base; ++j) 3965 if ((p[-j] & 0xc0) != 0x80) 3966 break; 3967 if (utf8len_tab[p[-j]] != i + j + 1) 3968 return 0; 3969 return i; 3970 } 3971 3972 /* It can't be the first byte if a double-byte when not using DBCS, at the 3973 * end of the string or the byte can't start a double-byte. */ 3974 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 3975 return 0; 3976 3977 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3978 return 1 - dbcs_head_off(base, p); 3979 } 3980 3981 /* 3982 * Find the next illegal byte sequence. 3983 */ 3984 void 3985 utf_find_illegal(void) 3986 { 3987 pos_T pos = curwin->w_cursor; 3988 char_u *p; 3989 int len; 3990 vimconv_T vimconv; 3991 char_u *tofree = NULL; 3992 3993 vimconv.vc_type = CONV_NONE; 3994 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 3995 { 3996 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 3997 * possibly a utf-8 file with illegal bytes. Setup for conversion 3998 * from utf-8 to 'fileencoding'. */ 3999 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 4000 } 4001 4002 #ifdef FEAT_VIRTUALEDIT 4003 curwin->w_cursor.coladd = 0; 4004 #endif 4005 for (;;) 4006 { 4007 p = ml_get_cursor(); 4008 if (vimconv.vc_type != CONV_NONE) 4009 { 4010 vim_free(tofree); 4011 tofree = string_convert(&vimconv, p, NULL); 4012 if (tofree == NULL) 4013 break; 4014 p = tofree; 4015 } 4016 4017 while (*p != NUL) 4018 { 4019 /* Illegal means that there are not enough trail bytes (checked by 4020 * utf_ptr2len()) or too many of them (overlong sequence). */ 4021 len = utf_ptr2len(p); 4022 if (*p >= 0x80 && (len == 1 4023 || utf_char2len(utf_ptr2char(p)) != len)) 4024 { 4025 if (vimconv.vc_type == CONV_NONE) 4026 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 4027 else 4028 { 4029 int l; 4030 4031 len = (int)(p - tofree); 4032 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 4033 { 4034 l = utf_ptr2len(p); 4035 curwin->w_cursor.col += l; 4036 } 4037 } 4038 goto theend; 4039 } 4040 p += len; 4041 } 4042 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 4043 break; 4044 ++curwin->w_cursor.lnum; 4045 curwin->w_cursor.col = 0; 4046 } 4047 4048 /* didn't find it: don't move and beep */ 4049 curwin->w_cursor = pos; 4050 beep_flush(); 4051 4052 theend: 4053 vim_free(tofree); 4054 convert_setup(&vimconv, NULL, NULL); 4055 } 4056 4057 #if defined(FEAT_GUI_GTK) || defined(PROTO) 4058 /* 4059 * Return TRUE if string "s" is a valid utf-8 string. 4060 * When "end" is NULL stop at the first NUL. 4061 * When "end" is positive stop there. 4062 */ 4063 int 4064 utf_valid_string(char_u *s, char_u *end) 4065 { 4066 int l; 4067 char_u *p = s; 4068 4069 while (end == NULL ? *p != NUL : p < end) 4070 { 4071 l = utf8len_tab_zero[*p]; 4072 if (l == 0) 4073 return FALSE; /* invalid lead byte */ 4074 if (end != NULL && p + l > end) 4075 return FALSE; /* incomplete byte sequence */ 4076 ++p; 4077 while (--l > 0) 4078 if ((*p++ & 0xc0) != 0x80) 4079 return FALSE; /* invalid trail byte */ 4080 } 4081 return TRUE; 4082 } 4083 #endif 4084 4085 #if defined(FEAT_GUI) || defined(PROTO) 4086 /* 4087 * Special version of mb_tail_off() for use in ScreenLines[]. 4088 */ 4089 int 4090 dbcs_screen_tail_off(char_u *base, char_u *p) 4091 { 4092 /* It can't be the first byte if a double-byte when not using DBCS, at the 4093 * end of the string or the byte can't start a double-byte. 4094 * For euc-jp an 0x8e byte always means we have a lead byte in the current 4095 * cell. */ 4096 if (*p == NUL || p[1] == NUL 4097 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 4098 || MB_BYTE2LEN(*p) == 1) 4099 return 0; 4100 4101 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 4102 return 1 - dbcs_screen_head_off(base, p); 4103 } 4104 #endif 4105 4106 /* 4107 * If the cursor moves on an trail byte, set the cursor on the lead byte. 4108 * Thus it moves left if necessary. 4109 * Return TRUE when the cursor was adjusted. 4110 */ 4111 void 4112 mb_adjust_cursor(void) 4113 { 4114 mb_adjustpos(curbuf, &curwin->w_cursor); 4115 } 4116 4117 /* 4118 * Adjust position "*lp" to point to the first byte of a multi-byte character. 4119 * If it points to a tail byte it's moved backwards to the head byte. 4120 */ 4121 void 4122 mb_adjustpos(buf_T *buf, pos_T *lp) 4123 { 4124 char_u *p; 4125 4126 if (lp->col > 0 4127 #ifdef FEAT_VIRTUALEDIT 4128 || lp->coladd > 1 4129 #endif 4130 ) 4131 { 4132 p = ml_get_buf(buf, lp->lnum, FALSE); 4133 if (*p == NUL || (int)STRLEN(p) < lp->col) 4134 lp->col = 0; 4135 else 4136 lp->col -= (*mb_head_off)(p, p + lp->col); 4137 #ifdef FEAT_VIRTUALEDIT 4138 /* Reset "coladd" when the cursor would be on the right half of a 4139 * double-wide character. */ 4140 if (lp->coladd == 1 4141 && p[lp->col] != TAB 4142 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 4143 && ptr2cells(p + lp->col) > 1) 4144 lp->coladd = 0; 4145 #endif 4146 } 4147 } 4148 4149 /* 4150 * Return a pointer to the character before "*p", if there is one. 4151 */ 4152 char_u * 4153 mb_prevptr( 4154 char_u *line, /* start of the string */ 4155 char_u *p) 4156 { 4157 if (p > line) 4158 MB_PTR_BACK(line, p); 4159 return p; 4160 } 4161 4162 /* 4163 * Return the character length of "str". Each multi-byte character (with 4164 * following composing characters) counts as one. 4165 */ 4166 int 4167 mb_charlen(char_u *str) 4168 { 4169 char_u *p = str; 4170 int count; 4171 4172 if (p == NULL) 4173 return 0; 4174 4175 for (count = 0; *p != NUL; count++) 4176 p += (*mb_ptr2len)(p); 4177 4178 return count; 4179 } 4180 4181 #if defined(FEAT_SPELL) || defined(PROTO) 4182 /* 4183 * Like mb_charlen() but for a string with specified length. 4184 */ 4185 int 4186 mb_charlen_len(char_u *str, int len) 4187 { 4188 char_u *p = str; 4189 int count; 4190 4191 for (count = 0; *p != NUL && p < str + len; count++) 4192 p += (*mb_ptr2len)(p); 4193 4194 return count; 4195 } 4196 #endif 4197 4198 /* 4199 * Try to un-escape a multi-byte character. 4200 * Used for the "to" and "from" part of a mapping. 4201 * Return the un-escaped string if it is a multi-byte character, and advance 4202 * "pp" to just after the bytes that formed it. 4203 * Return NULL if no multi-byte char was found. 4204 */ 4205 char_u * 4206 mb_unescape(char_u **pp) 4207 { 4208 static char_u buf[6]; 4209 int n; 4210 int m = 0; 4211 char_u *str = *pp; 4212 4213 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 4214 * KS_EXTRA KE_CSI to CSI. 4215 * Maximum length of a utf-8 character is 4 bytes. */ 4216 for (n = 0; str[n] != NUL && m < 4; ++n) 4217 { 4218 if (str[n] == K_SPECIAL 4219 && str[n + 1] == KS_SPECIAL 4220 && str[n + 2] == KE_FILLER) 4221 { 4222 buf[m++] = K_SPECIAL; 4223 n += 2; 4224 } 4225 else if ((str[n] == K_SPECIAL 4226 # ifdef FEAT_GUI 4227 || str[n] == CSI 4228 # endif 4229 ) 4230 && str[n + 1] == KS_EXTRA 4231 && str[n + 2] == (int)KE_CSI) 4232 { 4233 buf[m++] = CSI; 4234 n += 2; 4235 } 4236 else if (str[n] == K_SPECIAL 4237 # ifdef FEAT_GUI 4238 || str[n] == CSI 4239 # endif 4240 ) 4241 break; /* a special key can't be a multibyte char */ 4242 else 4243 buf[m++] = str[n]; 4244 buf[m] = NUL; 4245 4246 /* Return a multi-byte character if it's found. An illegal sequence 4247 * will result in a 1 here. */ 4248 if ((*mb_ptr2len)(buf) > 1) 4249 { 4250 *pp = str + n + 1; 4251 return buf; 4252 } 4253 4254 /* Bail out quickly for ASCII. */ 4255 if (buf[0] < 128) 4256 break; 4257 } 4258 return NULL; 4259 } 4260 4261 /* 4262 * Return TRUE if the character at "row"/"col" on the screen is the left side 4263 * of a double-width character. 4264 * Caller must make sure "row" and "col" are not invalid! 4265 */ 4266 int 4267 mb_lefthalve(int row, int col) 4268 { 4269 #ifdef FEAT_HANGULIN 4270 if (composing_hangul) 4271 return TRUE; 4272 #endif 4273 return (*mb_off2cells)(LineOffset[row] + col, 4274 LineOffset[row] + screen_Columns) > 1; 4275 } 4276 4277 /* 4278 * Correct a position on the screen, if it's the right half of a double-wide 4279 * char move it to the left half. Returns the corrected column. 4280 */ 4281 int 4282 mb_fix_col(int col, int row) 4283 { 4284 col = check_col(col); 4285 row = check_row(row); 4286 if (has_mbyte && ScreenLines != NULL && col > 0 4287 && ((enc_dbcs 4288 && ScreenLines[LineOffset[row] + col] != NUL 4289 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 4290 ScreenLines + LineOffset[row] + col)) 4291 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0))) 4292 return col - 1; 4293 return col; 4294 } 4295 #endif 4296 4297 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 4298 static int enc_alias_search(char_u *name); 4299 4300 /* 4301 * Skip the Vim specific head of a 'encoding' name. 4302 */ 4303 char_u * 4304 enc_skip(char_u *p) 4305 { 4306 if (STRNCMP(p, "2byte-", 6) == 0) 4307 return p + 6; 4308 if (STRNCMP(p, "8bit-", 5) == 0) 4309 return p + 5; 4310 return p; 4311 } 4312 4313 /* 4314 * Find the canonical name for encoding "enc". 4315 * When the name isn't recognized, returns "enc" itself, but with all lower 4316 * case characters and '_' replaced with '-'. 4317 * Returns an allocated string. NULL for out-of-memory. 4318 */ 4319 char_u * 4320 enc_canonize(char_u *enc) 4321 { 4322 char_u *r; 4323 char_u *p, *s; 4324 int i; 4325 4326 # ifdef FEAT_MBYTE 4327 if (STRCMP(enc, "default") == 0) 4328 { 4329 /* Use the default encoding as it's found by set_init_1(). */ 4330 r = get_encoding_default(); 4331 if (r == NULL) 4332 r = (char_u *)"latin1"; 4333 return vim_strsave(r); 4334 } 4335 # endif 4336 4337 /* copy "enc" to allocated memory, with room for two '-' */ 4338 r = alloc((unsigned)(STRLEN(enc) + 3)); 4339 if (r != NULL) 4340 { 4341 /* Make it all lower case and replace '_' with '-'. */ 4342 p = r; 4343 for (s = enc; *s != NUL; ++s) 4344 { 4345 if (*s == '_') 4346 *p++ = '-'; 4347 else 4348 *p++ = TOLOWER_ASC(*s); 4349 } 4350 *p = NUL; 4351 4352 /* Skip "2byte-" and "8bit-". */ 4353 p = enc_skip(r); 4354 4355 /* Change "microsoft-cp" to "cp". Used in some spell files. */ 4356 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4357 STRMOVE(p, p + 10); 4358 4359 /* "iso8859" -> "iso-8859" */ 4360 if (STRNCMP(p, "iso8859", 7) == 0) 4361 { 4362 STRMOVE(p + 4, p + 3); 4363 p[3] = '-'; 4364 } 4365 4366 /* "iso-8859n" -> "iso-8859-n" */ 4367 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4368 { 4369 STRMOVE(p + 9, p + 8); 4370 p[8] = '-'; 4371 } 4372 4373 /* "latin-N" -> "latinN" */ 4374 if (STRNCMP(p, "latin-", 6) == 0) 4375 STRMOVE(p + 5, p + 6); 4376 4377 if (enc_canon_search(p) >= 0) 4378 { 4379 /* canonical name can be used unmodified */ 4380 if (p != r) 4381 STRMOVE(r, p); 4382 } 4383 else if ((i = enc_alias_search(p)) >= 0) 4384 { 4385 /* alias recognized, get canonical name */ 4386 vim_free(r); 4387 r = vim_strsave((char_u *)enc_canon_table[i].name); 4388 } 4389 } 4390 return r; 4391 } 4392 4393 /* 4394 * Search for an encoding alias of "name". 4395 * Returns -1 when not found. 4396 */ 4397 static int 4398 enc_alias_search(char_u *name) 4399 { 4400 int i; 4401 4402 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4403 if (STRCMP(name, enc_alias_table[i].name) == 0) 4404 return enc_alias_table[i].canon; 4405 return -1; 4406 } 4407 #endif 4408 4409 #if defined(FEAT_MBYTE) || defined(PROTO) 4410 4411 # ifdef HAVE_LANGINFO_H 4412 # include <langinfo.h> 4413 # endif 4414 4415 # ifndef FEAT_GUI_W32 4416 /* 4417 * Get the canonicalized encoding from the specified locale string "locale" 4418 * or from the environment variables LC_ALL, LC_CTYPE and LANG. 4419 * Returns an allocated string when successful, NULL when not. 4420 */ 4421 char_u * 4422 enc_locale_env(char *locale) 4423 { 4424 char *s = locale; 4425 char *p; 4426 int i; 4427 char buf[50]; 4428 4429 if (s == NULL || *s == NUL) 4430 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4431 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4432 s = getenv("LANG"); 4433 4434 if (s == NULL || *s == NUL) 4435 return NULL; 4436 4437 /* The most generic locale format is: 4438 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4439 * If there is a '.' remove the part before it. 4440 * if there is something after the codeset, remove it. 4441 * Make the name lowercase and replace '_' with '-'. 4442 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4443 * "ko_KR.EUC" == "euc-kr" 4444 */ 4445 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4446 { 4447 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4448 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4449 { 4450 /* copy "XY.EUC" to "euc-XY" to buf[10] */ 4451 STRCPY(buf + 10, "euc-"); 4452 buf[14] = p[-2]; 4453 buf[15] = p[-1]; 4454 buf[16] = 0; 4455 s = buf + 10; 4456 } 4457 else 4458 s = p + 1; 4459 } 4460 for (i = 0; i < (int)sizeof(buf) - 1 && s[i] != NUL; ++i) 4461 { 4462 if (s[i] == '_' || s[i] == '-') 4463 buf[i] = '-'; 4464 else if (isalnum((int)s[i])) 4465 buf[i] = TOLOWER_ASC(s[i]); 4466 else 4467 break; 4468 } 4469 buf[i] = NUL; 4470 4471 return enc_canonize((char_u *)buf); 4472 } 4473 # endif 4474 4475 /* 4476 * Get the canonicalized encoding of the current locale. 4477 * Returns an allocated string when successful, NULL when not. 4478 */ 4479 char_u * 4480 enc_locale(void) 4481 { 4482 # ifdef WIN3264 4483 char buf[50]; 4484 long acp = GetACP(); 4485 4486 if (acp == 1200) 4487 STRCPY(buf, "ucs-2le"); 4488 else if (acp == 1252) /* cp1252 is used as latin1 */ 4489 STRCPY(buf, "latin1"); 4490 else 4491 sprintf(buf, "cp%ld", acp); 4492 4493 return enc_canonize((char_u *)buf); 4494 # else 4495 char *s; 4496 4497 # ifdef HAVE_NL_LANGINFO_CODESET 4498 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4499 # endif 4500 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4501 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4502 # endif 4503 s = NULL; 4504 4505 return enc_locale_env(s); 4506 # endif 4507 } 4508 4509 # if defined(WIN3264) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4510 /* 4511 * Convert an encoding name to an MS-Windows codepage. 4512 * Returns zero if no codepage can be figured out. 4513 */ 4514 int 4515 encname2codepage(char_u *name) 4516 { 4517 int cp; 4518 char_u *p = name; 4519 int idx; 4520 4521 if (STRNCMP(p, "8bit-", 5) == 0) 4522 p += 5; 4523 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4524 p += 6; 4525 4526 if (p[0] == 'c' && p[1] == 'p') 4527 cp = atoi((char *)p + 2); 4528 else if ((idx = enc_canon_search(p)) >= 0) 4529 cp = enc_canon_table[idx].codepage; 4530 else 4531 return 0; 4532 if (IsValidCodePage(cp)) 4533 return cp; 4534 return 0; 4535 } 4536 # endif 4537 4538 # if defined(USE_ICONV) || defined(PROTO) 4539 4540 /* 4541 * Call iconv_open() with a check if iconv() works properly (there are broken 4542 * versions). 4543 * Returns (void *)-1 if failed. 4544 * (should return iconv_t, but that causes problems with prototypes). 4545 */ 4546 void * 4547 my_iconv_open(char_u *to, char_u *from) 4548 { 4549 iconv_t fd; 4550 #define ICONV_TESTLEN 400 4551 char_u tobuf[ICONV_TESTLEN]; 4552 char *p; 4553 size_t tolen; 4554 static int iconv_ok = -1; 4555 4556 if (iconv_ok == FALSE) 4557 return (void *)-1; /* detected a broken iconv() previously */ 4558 4559 #ifdef DYNAMIC_ICONV 4560 /* Check if the iconv.dll can be found. */ 4561 if (!iconv_enabled(TRUE)) 4562 return (void *)-1; 4563 #endif 4564 4565 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4566 4567 if (fd != (iconv_t)-1 && iconv_ok == -1) 4568 { 4569 /* 4570 * Do a dummy iconv() call to check if it actually works. There is a 4571 * version of iconv() on Linux that is broken. We can't ignore it, 4572 * because it's wide-spread. The symptoms are that after outputting 4573 * the initial shift state the "to" pointer is NULL and conversion 4574 * stops for no apparent reason after about 8160 characters. 4575 */ 4576 p = (char *)tobuf; 4577 tolen = ICONV_TESTLEN; 4578 (void)iconv(fd, NULL, NULL, &p, &tolen); 4579 if (p == NULL) 4580 { 4581 iconv_ok = FALSE; 4582 iconv_close(fd); 4583 fd = (iconv_t)-1; 4584 } 4585 else 4586 iconv_ok = TRUE; 4587 } 4588 4589 return (void *)fd; 4590 } 4591 4592 /* 4593 * Convert the string "str[slen]" with iconv(). 4594 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4595 * sequence and set "*unconvlenp" to the length of it. 4596 * Returns the converted string in allocated memory. NULL for an error. 4597 * If resultlenp is not NULL, sets it to the result length in bytes. 4598 */ 4599 static char_u * 4600 iconv_string( 4601 vimconv_T *vcp, 4602 char_u *str, 4603 int slen, 4604 int *unconvlenp, 4605 int *resultlenp) 4606 { 4607 const char *from; 4608 size_t fromlen; 4609 char *to; 4610 size_t tolen; 4611 size_t len = 0; 4612 size_t done = 0; 4613 char_u *result = NULL; 4614 char_u *p; 4615 int l; 4616 4617 from = (char *)str; 4618 fromlen = slen; 4619 for (;;) 4620 { 4621 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4622 { 4623 /* Allocate enough room for most conversions. When re-allocating 4624 * increase the buffer size. */ 4625 len = len + fromlen * 2 + 40; 4626 p = alloc((unsigned)len); 4627 if (p != NULL && done > 0) 4628 mch_memmove(p, result, done); 4629 vim_free(result); 4630 result = p; 4631 if (result == NULL) /* out of memory */ 4632 break; 4633 } 4634 4635 to = (char *)result + done; 4636 tolen = len - done - 2; 4637 /* Avoid a warning for systems with a wrong iconv() prototype by 4638 * casting the second argument to void *. */ 4639 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4640 != (size_t)-1) 4641 { 4642 /* Finished, append a NUL. */ 4643 *to = NUL; 4644 break; 4645 } 4646 4647 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4648 * iconv library may use one of them. */ 4649 if (!vcp->vc_fail && unconvlenp != NULL 4650 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4651 { 4652 /* Handle an incomplete sequence at the end. */ 4653 *to = NUL; 4654 *unconvlenp = (int)fromlen; 4655 break; 4656 } 4657 4658 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4659 * iconv library may use one of them. */ 4660 else if (!vcp->vc_fail 4661 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4662 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4663 { 4664 /* Can't convert: insert a '?' and skip a character. This assumes 4665 * conversion from 'encoding' to something else. In other 4666 * situations we don't know what to skip anyway. */ 4667 *to++ = '?'; 4668 if ((*mb_ptr2cells)((char_u *)from) > 1) 4669 *to++ = '?'; 4670 if (enc_utf8) 4671 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4672 else 4673 { 4674 l = (*mb_ptr2len)((char_u *)from); 4675 if (l > (int)fromlen) 4676 l = (int)fromlen; 4677 } 4678 from += l; 4679 fromlen -= l; 4680 } 4681 else if (ICONV_ERRNO != ICONV_E2BIG) 4682 { 4683 /* conversion failed */ 4684 VIM_CLEAR(result); 4685 break; 4686 } 4687 /* Not enough room or skipping illegal sequence. */ 4688 done = to - (char *)result; 4689 } 4690 4691 if (resultlenp != NULL && result != NULL) 4692 *resultlenp = (int)(to - (char *)result); 4693 return result; 4694 } 4695 4696 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4697 /* 4698 * Dynamically load the "iconv.dll" on Win32. 4699 */ 4700 4701 # ifndef DYNAMIC_ICONV /* must be generating prototypes */ 4702 # define HINSTANCE int 4703 # endif 4704 static HINSTANCE hIconvDLL = 0; 4705 static HINSTANCE hMsvcrtDLL = 0; 4706 4707 # ifndef DYNAMIC_ICONV_DLL 4708 # define DYNAMIC_ICONV_DLL "iconv.dll" 4709 # define DYNAMIC_ICONV_DLL_ALT1 "libiconv.dll" 4710 # define DYNAMIC_ICONV_DLL_ALT2 "libiconv2.dll" 4711 # define DYNAMIC_ICONV_DLL_ALT3 "libiconv-2.dll" 4712 # endif 4713 # ifndef DYNAMIC_MSVCRT_DLL 4714 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4715 # endif 4716 4717 /* 4718 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4719 */ 4720 int 4721 iconv_enabled(int verbose) 4722 { 4723 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4724 return TRUE; 4725 4726 /* The iconv DLL file goes under different names, try them all. 4727 * Do the "2" version first, it's newer. */ 4728 #ifdef DYNAMIC_ICONV_DLL_ALT2 4729 if (hIconvDLL == 0) 4730 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT2); 4731 #endif 4732 #ifdef DYNAMIC_ICONV_DLL_ALT3 4733 if (hIconvDLL == 0) 4734 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT3); 4735 #endif 4736 if (hIconvDLL == 0) 4737 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4738 #ifdef DYNAMIC_ICONV_DLL_ALT1 4739 if (hIconvDLL == 0) 4740 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT1); 4741 #endif 4742 4743 if (hIconvDLL != 0) 4744 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4745 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4746 { 4747 /* Only give the message when 'verbose' is set, otherwise it might be 4748 * done whenever a conversion is attempted. */ 4749 if (verbose && p_verbose > 0) 4750 { 4751 verbose_enter(); 4752 semsg(_(e_loadlib), 4753 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4754 verbose_leave(); 4755 } 4756 iconv_end(); 4757 return FALSE; 4758 } 4759 4760 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4761 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4762 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4763 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4764 iconv_errno = get_dll_import_func(hIconvDLL, "_errno"); 4765 if (iconv_errno == NULL) 4766 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4767 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4768 || iconvctl == NULL || iconv_errno == NULL) 4769 { 4770 iconv_end(); 4771 if (verbose && p_verbose > 0) 4772 { 4773 verbose_enter(); 4774 semsg(_(e_loadfunc), "for libiconv"); 4775 verbose_leave(); 4776 } 4777 return FALSE; 4778 } 4779 return TRUE; 4780 } 4781 4782 void 4783 iconv_end(void) 4784 { 4785 /* Don't use iconv() when inputting or outputting characters. */ 4786 if (input_conv.vc_type == CONV_ICONV) 4787 convert_setup(&input_conv, NULL, NULL); 4788 if (output_conv.vc_type == CONV_ICONV) 4789 convert_setup(&output_conv, NULL, NULL); 4790 4791 if (hIconvDLL != 0) 4792 FreeLibrary(hIconvDLL); 4793 if (hMsvcrtDLL != 0) 4794 FreeLibrary(hMsvcrtDLL); 4795 hIconvDLL = 0; 4796 hMsvcrtDLL = 0; 4797 } 4798 # endif /* DYNAMIC_ICONV */ 4799 # endif /* USE_ICONV */ 4800 4801 #endif /* FEAT_MBYTE */ 4802 4803 #ifdef FEAT_GUI 4804 # define USE_IMACTIVATEFUNC (!gui.in_use && *p_imaf != NUL) 4805 # define USE_IMSTATUSFUNC (!gui.in_use && *p_imsf != NUL) 4806 #else 4807 # define USE_IMACTIVATEFUNC (*p_imaf != NUL) 4808 # define USE_IMSTATUSFUNC (*p_imsf != NUL) 4809 #endif 4810 4811 #if defined(FEAT_EVAL) && defined(FEAT_MBYTE) \ 4812 && (defined(FEAT_XIM) || defined(IME_WITHOUT_XIM)) 4813 static void 4814 call_imactivatefunc(int active) 4815 { 4816 typval_T argv[2]; 4817 4818 argv[0].v_type = VAR_NUMBER; 4819 argv[0].vval.v_number = active ? 1 : 0; 4820 argv[1].v_type = VAR_UNKNOWN; 4821 (void)call_func_retnr(p_imaf, 1, argv); 4822 } 4823 4824 static int 4825 call_imstatusfunc(void) 4826 { 4827 int is_active; 4828 4829 /* FIXME: Don't execute user function in unsafe situation. */ 4830 if (exiting || is_autocmd_blocked()) 4831 return FALSE; 4832 /* FIXME: :py print 'xxx' is shown duplicate result. 4833 * Use silent to avoid it. */ 4834 ++msg_silent; 4835 is_active = call_func_retnr(p_imsf, 0, NULL); 4836 --msg_silent; 4837 return (is_active > 0); 4838 } 4839 #endif 4840 4841 #if defined(FEAT_XIM) || defined(PROTO) 4842 4843 # if defined(FEAT_GUI_GTK) || defined(PROTO) 4844 static int xim_has_preediting INIT(= FALSE); /* IM current status */ 4845 4846 /* 4847 * Set preedit_start_col to the current cursor position. 4848 */ 4849 static void 4850 init_preedit_start_col(void) 4851 { 4852 if (State & CMDLINE) 4853 preedit_start_col = cmdline_getvcol_cursor(); 4854 else if (curwin != NULL && curwin->w_buffer != NULL) 4855 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); 4856 /* Prevent that preediting marks the buffer as changed. */ 4857 xim_changed_while_preediting = curbuf->b_changed; 4858 } 4859 4860 static int im_is_active = FALSE; /* IM is enabled for current mode */ 4861 static int preedit_is_active = FALSE; 4862 static int im_preedit_cursor = 0; /* cursor offset in characters */ 4863 static int im_preedit_trailing = 0; /* number of characters after cursor */ 4864 4865 static unsigned long im_commit_handler_id = 0; 4866 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; 4867 static unsigned int im_activatekey_state = 0; 4868 4869 static GtkWidget *preedit_window = NULL; 4870 static GtkWidget *preedit_label = NULL; 4871 4872 static void im_preedit_window_set_position(void); 4873 4874 void 4875 im_set_active(int active) 4876 { 4877 int was_active; 4878 4879 was_active = !!im_get_status(); 4880 im_is_active = (active && !p_imdisable); 4881 4882 if (im_is_active != was_active) 4883 xim_reset(); 4884 } 4885 4886 void 4887 xim_set_focus(int focus) 4888 { 4889 if (xic != NULL) 4890 { 4891 if (focus) 4892 gtk_im_context_focus_in(xic); 4893 else 4894 gtk_im_context_focus_out(xic); 4895 } 4896 } 4897 4898 void 4899 im_set_position(int row, int col) 4900 { 4901 if (xic != NULL) 4902 { 4903 GdkRectangle area; 4904 4905 area.x = FILL_X(col); 4906 area.y = FILL_Y(row); 4907 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); 4908 area.height = gui.char_height; 4909 4910 gtk_im_context_set_cursor_location(xic, &area); 4911 4912 if (p_imst == IM_OVER_THE_SPOT) 4913 im_preedit_window_set_position(); 4914 } 4915 } 4916 4917 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */ 4918 void 4919 xim_set_preedit(void) 4920 { 4921 im_set_position(gui.row, gui.col); 4922 } 4923 # endif 4924 4925 static void 4926 im_add_to_input(char_u *str, int len) 4927 { 4928 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */ 4929 if (input_conv.vc_type != CONV_NONE) 4930 { 4931 str = string_convert(&input_conv, str, &len); 4932 g_return_if_fail(str != NULL); 4933 } 4934 4935 add_to_input_buf_csi(str, len); 4936 4937 if (input_conv.vc_type != CONV_NONE) 4938 vim_free(str); 4939 4940 if (p_mh) /* blank out the pointer if necessary */ 4941 gui_mch_mousehide(TRUE); 4942 } 4943 4944 static void 4945 im_preedit_window_set_position(void) 4946 { 4947 int x, y, width, height; 4948 int screen_x, screen_y, screen_width, screen_height; 4949 4950 if (preedit_window == NULL) 4951 return; 4952 4953 gui_gtk_get_screen_geom_of_win(gui.drawarea, 4954 &screen_x, &screen_y, &screen_width, &screen_height); 4955 gdk_window_get_origin(gtk_widget_get_window(gui.drawarea), &x, &y); 4956 gtk_window_get_size(GTK_WINDOW(preedit_window), &width, &height); 4957 x = x + FILL_X(gui.col); 4958 y = y + FILL_Y(gui.row); 4959 if (x + width > screen_x + screen_width) 4960 x = screen_x + screen_width - width; 4961 if (y + height > screen_y + screen_height) 4962 y = screen_y + screen_height - height; 4963 gtk_window_move(GTK_WINDOW(preedit_window), x, y); 4964 } 4965 4966 static void 4967 im_preedit_window_open() 4968 { 4969 char *preedit_string; 4970 #if !GTK_CHECK_VERSION(3,16,0) 4971 char buf[8]; 4972 #endif 4973 PangoAttrList *attr_list; 4974 PangoLayout *layout; 4975 #if GTK_CHECK_VERSION(3,0,0) 4976 # if !GTK_CHECK_VERSION(3,16,0) 4977 GdkRGBA color; 4978 # endif 4979 #else 4980 GdkColor color; 4981 #endif 4982 gint w, h; 4983 4984 if (preedit_window == NULL) 4985 { 4986 preedit_window = gtk_window_new(GTK_WINDOW_POPUP); 4987 gtk_window_set_transient_for(GTK_WINDOW(preedit_window), 4988 GTK_WINDOW(gui.mainwin)); 4989 preedit_label = gtk_label_new(""); 4990 gtk_widget_set_name(preedit_label, "vim-gui-preedit-area"); 4991 gtk_container_add(GTK_CONTAINER(preedit_window), preedit_label); 4992 } 4993 4994 #if GTK_CHECK_VERSION(3,16,0) 4995 { 4996 GtkStyleContext * const context 4997 = gtk_widget_get_style_context(gui.drawarea); 4998 GtkCssProvider * const provider = gtk_css_provider_new(); 4999 gchar *css = NULL; 5000 const char * const fontname 5001 = pango_font_description_get_family(gui.norm_font); 5002 gint fontsize 5003 = pango_font_description_get_size(gui.norm_font) / PANGO_SCALE; 5004 gchar *fontsize_propval = NULL; 5005 5006 if (!pango_font_description_get_size_is_absolute(gui.norm_font)) 5007 { 5008 /* fontsize was given in points. Convert it into that in pixels 5009 * to use with CSS. */ 5010 GdkScreen * const screen 5011 = gdk_window_get_screen(gtk_widget_get_window(gui.mainwin)); 5012 const gdouble dpi = gdk_screen_get_resolution(screen); 5013 fontsize = dpi * fontsize / 72; 5014 } 5015 if (fontsize > 0) 5016 fontsize_propval = g_strdup_printf("%dpx", fontsize); 5017 else 5018 fontsize_propval = g_strdup_printf("inherit"); 5019 5020 css = g_strdup_printf( 5021 "widget#vim-gui-preedit-area {\n" 5022 " font-family: %s,monospace;\n" 5023 " font-size: %s;\n" 5024 " color: #%.2lx%.2lx%.2lx;\n" 5025 " background-color: #%.2lx%.2lx%.2lx;\n" 5026 "}\n", 5027 fontname != NULL ? fontname : "inherit", 5028 fontsize_propval, 5029 (gui.norm_pixel >> 16) & 0xff, 5030 (gui.norm_pixel >> 8) & 0xff, 5031 gui.norm_pixel & 0xff, 5032 (gui.back_pixel >> 16) & 0xff, 5033 (gui.back_pixel >> 8) & 0xff, 5034 gui.back_pixel & 0xff); 5035 5036 gtk_css_provider_load_from_data(provider, css, -1, NULL); 5037 gtk_style_context_add_provider(context, 5038 GTK_STYLE_PROVIDER(provider), G_MAXUINT); 5039 5040 g_free(css); 5041 g_free(fontsize_propval); 5042 g_object_unref(provider); 5043 } 5044 #elif GTK_CHECK_VERSION(3,0,0) 5045 gtk_widget_override_font(preedit_label, gui.norm_font); 5046 5047 vim_snprintf(buf, sizeof(buf), "#%06X", gui.norm_pixel); 5048 gdk_rgba_parse(&color, buf); 5049 gtk_widget_override_color(preedit_label, GTK_STATE_FLAG_NORMAL, &color); 5050 5051 vim_snprintf(buf, sizeof(buf), "#%06X", gui.back_pixel); 5052 gdk_rgba_parse(&color, buf); 5053 gtk_widget_override_background_color(preedit_label, GTK_STATE_FLAG_NORMAL, 5054 &color); 5055 #else 5056 gtk_widget_modify_font(preedit_label, gui.norm_font); 5057 5058 vim_snprintf(buf, sizeof(buf), "#%06X", (unsigned)gui.norm_pixel); 5059 gdk_color_parse(buf, &color); 5060 gtk_widget_modify_fg(preedit_label, GTK_STATE_NORMAL, &color); 5061 5062 vim_snprintf(buf, sizeof(buf), "#%06X", (unsigned)gui.back_pixel); 5063 gdk_color_parse(buf, &color); 5064 gtk_widget_modify_bg(preedit_window, GTK_STATE_NORMAL, &color); 5065 #endif 5066 5067 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5068 5069 if (preedit_string[0] != NUL) 5070 { 5071 gtk_label_set_text(GTK_LABEL(preedit_label), preedit_string); 5072 gtk_label_set_attributes(GTK_LABEL(preedit_label), attr_list); 5073 5074 layout = gtk_label_get_layout(GTK_LABEL(preedit_label)); 5075 pango_layout_get_pixel_size(layout, &w, &h); 5076 h = MAX(h, gui.char_height); 5077 gtk_window_resize(GTK_WINDOW(preedit_window), w, h); 5078 5079 gtk_widget_show_all(preedit_window); 5080 5081 im_preedit_window_set_position(); 5082 } 5083 5084 g_free(preedit_string); 5085 pango_attr_list_unref(attr_list); 5086 } 5087 5088 static void 5089 im_preedit_window_close() 5090 { 5091 if (preedit_window != NULL) 5092 gtk_widget_hide(preedit_window); 5093 } 5094 5095 static void 5096 im_show_preedit() 5097 { 5098 im_preedit_window_open(); 5099 5100 if (p_mh) /* blank out the pointer if necessary */ 5101 gui_mch_mousehide(TRUE); 5102 } 5103 5104 static void 5105 im_delete_preedit(void) 5106 { 5107 char_u bskey[] = {CSI, 'k', 'b'}; 5108 char_u delkey[] = {CSI, 'k', 'D'}; 5109 5110 if (p_imst == IM_OVER_THE_SPOT) 5111 { 5112 im_preedit_window_close(); 5113 return; 5114 } 5115 5116 if (State & NORMAL 5117 #ifdef FEAT_TERMINAL 5118 && !term_use_loop() 5119 #endif 5120 ) 5121 { 5122 im_preedit_cursor = 0; 5123 return; 5124 } 5125 for (; im_preedit_cursor > 0; --im_preedit_cursor) 5126 add_to_input_buf(bskey, (int)sizeof(bskey)); 5127 5128 for (; im_preedit_trailing > 0; --im_preedit_trailing) 5129 add_to_input_buf(delkey, (int)sizeof(delkey)); 5130 } 5131 5132 /* 5133 * Move the cursor left by "num_move_back" characters. 5134 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for 5135 * these K_LEFT keys. 5136 */ 5137 static void 5138 im_correct_cursor(int num_move_back) 5139 { 5140 char_u backkey[] = {CSI, 'k', 'l'}; 5141 5142 if (State & NORMAL) 5143 return; 5144 # ifdef FEAT_RIGHTLEFT 5145 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) 5146 backkey[2] = 'r'; 5147 # endif 5148 for (; num_move_back > 0; --num_move_back) 5149 add_to_input_buf(backkey, (int)sizeof(backkey)); 5150 } 5151 5152 static int xim_expected_char = NUL; 5153 static int xim_ignored_char = FALSE; 5154 5155 /* 5156 * Update the mode and cursor while in an IM callback. 5157 */ 5158 static void 5159 im_show_info(void) 5160 { 5161 int old_vgetc_busy; 5162 5163 old_vgetc_busy = vgetc_busy; 5164 vgetc_busy = TRUE; 5165 showmode(); 5166 vgetc_busy = old_vgetc_busy; 5167 if ((State & NORMAL) || (State & INSERT)) 5168 setcursor(); 5169 out_flush(); 5170 } 5171 5172 /* 5173 * Callback invoked when the user finished preediting. 5174 * Put the final string into the input buffer. 5175 */ 5176 static void 5177 im_commit_cb(GtkIMContext *context UNUSED, 5178 const gchar *str, 5179 gpointer data UNUSED) 5180 { 5181 int slen = (int)STRLEN(str); 5182 int add_to_input = TRUE; 5183 int clen; 5184 int len = slen; 5185 int commit_with_preedit = TRUE; 5186 char_u *im_str; 5187 5188 #ifdef XIM_DEBUG 5189 xim_log("im_commit_cb(): %s\n", str); 5190 #endif 5191 5192 if (p_imst == IM_ON_THE_SPOT) 5193 { 5194 /* The imhangul module doesn't reset the preedit string before 5195 * committing. Call im_delete_preedit() to work around that. */ 5196 im_delete_preedit(); 5197 5198 /* Indicate that preediting has finished. */ 5199 if (preedit_start_col == MAXCOL) 5200 { 5201 init_preedit_start_col(); 5202 commit_with_preedit = FALSE; 5203 } 5204 5205 /* The thing which setting "preedit_start_col" to MAXCOL means that 5206 * "preedit_start_col" will be set forcedly when calling 5207 * preedit_changed_cb() next time. 5208 * "preedit_start_col" should not reset with MAXCOL on this part. Vim 5209 * is simulating the preediting by using add_to_input_str(). when 5210 * preedit begin immediately before committed, the typebuf is not 5211 * flushed to screen, then it can't get correct "preedit_start_col". 5212 * Thus, it should calculate the cells by adding cells of the committed 5213 * string. */ 5214 if (input_conv.vc_type != CONV_NONE) 5215 { 5216 im_str = string_convert(&input_conv, (char_u *)str, &len); 5217 g_return_if_fail(im_str != NULL); 5218 } 5219 else 5220 im_str = (char_u *)str; 5221 5222 clen = mb_string2cells(im_str, len); 5223 5224 if (input_conv.vc_type != CONV_NONE) 5225 vim_free(im_str); 5226 preedit_start_col += clen; 5227 } 5228 5229 /* Is this a single character that matches a keypad key that's just 5230 * been pressed? If so, we don't want it to be entered as such - let 5231 * us carry on processing the raw keycode so that it may be used in 5232 * mappings as <kSomething>. */ 5233 if (xim_expected_char != NUL) 5234 { 5235 /* We're currently processing a keypad or other special key */ 5236 if (slen == 1 && str[0] == xim_expected_char) 5237 { 5238 /* It's a match - don't do it here */ 5239 xim_ignored_char = TRUE; 5240 add_to_input = FALSE; 5241 } 5242 else 5243 { 5244 /* Not a match */ 5245 xim_ignored_char = FALSE; 5246 } 5247 } 5248 5249 if (add_to_input) 5250 im_add_to_input((char_u *)str, slen); 5251 5252 if (p_imst == IM_ON_THE_SPOT) 5253 { 5254 /* Inserting chars while "im_is_active" is set does not cause a 5255 * change of buffer. When the chars are committed the buffer must be 5256 * marked as changed. */ 5257 if (!commit_with_preedit) 5258 preedit_start_col = MAXCOL; 5259 5260 /* This flag is used in changed() at next call. */ 5261 xim_changed_while_preediting = TRUE; 5262 } 5263 5264 if (gtk_main_level() > 0) 5265 gtk_main_quit(); 5266 } 5267 5268 /* 5269 * Callback invoked after start to the preedit. 5270 */ 5271 static void 5272 im_preedit_start_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 5273 { 5274 #ifdef XIM_DEBUG 5275 xim_log("im_preedit_start_cb()\n"); 5276 #endif 5277 5278 im_is_active = TRUE; 5279 preedit_is_active = TRUE; 5280 gui_update_cursor(TRUE, FALSE); 5281 im_show_info(); 5282 } 5283 5284 /* 5285 * Callback invoked after end to the preedit. 5286 */ 5287 static void 5288 im_preedit_end_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 5289 { 5290 #ifdef XIM_DEBUG 5291 xim_log("im_preedit_end_cb()\n"); 5292 #endif 5293 im_delete_preedit(); 5294 5295 /* Indicate that preediting has finished */ 5296 if (p_imst == IM_ON_THE_SPOT) 5297 preedit_start_col = MAXCOL; 5298 xim_has_preediting = FALSE; 5299 5300 #if 0 5301 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was 5302 * switched off unintentionally. We now use preedit_is_active (added by 5303 * SungHyun Nam). */ 5304 im_is_active = FALSE; 5305 #endif 5306 preedit_is_active = FALSE; 5307 gui_update_cursor(TRUE, FALSE); 5308 im_show_info(); 5309 } 5310 5311 /* 5312 * Callback invoked after changes to the preedit string. If the preedit 5313 * string was empty before, remember the preedit start column so we know 5314 * where to apply feedback attributes. Delete the previous preedit string 5315 * if there was one, save the new preedit cursor offset, and put the new 5316 * string into the input buffer. 5317 * 5318 * TODO: The pragmatic "put into input buffer" approach used here has 5319 * several fundamental problems: 5320 * 5321 * - The characters in the preedit string are subject to remapping. 5322 * That's broken, only the finally committed string should be remapped. 5323 * 5324 * - There is a race condition involved: The retrieved value for the 5325 * current cursor position will be wrong if any unprocessed characters 5326 * are still queued in the input buffer. 5327 * 5328 * - Due to the lack of synchronization between the file buffer in memory 5329 * and any typed characters, it's practically impossible to implement the 5330 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM 5331 * modules for languages such as Thai are likely to rely on this feature 5332 * for proper operation. 5333 * 5334 * Conclusions: I think support for preediting needs to be moved to the 5335 * core parts of Vim. Ideally, until it has been committed, the preediting 5336 * string should only be displayed and not affect the buffer content at all. 5337 * The question how to deal with the synchronization issue still remains. 5338 * Circumventing the input buffer is probably not desirable. Anyway, I think 5339 * implementing "retrieve_surrounding" is the only hard problem. 5340 * 5341 * One way to solve all of this in a clean manner would be to queue all key 5342 * press/release events "as is" in the input buffer, and apply the IM filtering 5343 * at the receiving end of the queue. This, however, would have a rather large 5344 * impact on the code base. If there is an easy way to force processing of all 5345 * remaining input from within the "retrieve_surrounding" signal handler, this 5346 * might not be necessary. Gotta ask on vim-dev for opinions. 5347 */ 5348 static void 5349 im_preedit_changed_cb(GtkIMContext *context, gpointer data UNUSED) 5350 { 5351 char *preedit_string = NULL; 5352 int cursor_index = 0; 5353 int num_move_back = 0; 5354 char_u *str; 5355 char_u *p; 5356 int i; 5357 5358 if (p_imst == IM_ON_THE_SPOT) 5359 gtk_im_context_get_preedit_string(context, 5360 &preedit_string, NULL, 5361 &cursor_index); 5362 else 5363 gtk_im_context_get_preedit_string(context, 5364 &preedit_string, NULL, 5365 NULL); 5366 5367 #ifdef XIM_DEBUG 5368 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); 5369 #endif 5370 5371 g_return_if_fail(preedit_string != NULL); /* just in case */ 5372 5373 if (p_imst == IM_OVER_THE_SPOT) 5374 { 5375 if (preedit_string[0] == NUL) 5376 { 5377 xim_has_preediting = FALSE; 5378 im_delete_preedit(); 5379 } 5380 else 5381 { 5382 xim_has_preediting = TRUE; 5383 im_show_preedit(); 5384 } 5385 } 5386 else 5387 { 5388 /* If preedit_start_col is MAXCOL set it to the current cursor position. */ 5389 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') 5390 { 5391 xim_has_preediting = TRUE; 5392 5393 /* Urgh, this breaks if the input buffer isn't empty now */ 5394 init_preedit_start_col(); 5395 } 5396 else if (cursor_index == 0 && preedit_string[0] == '\0') 5397 { 5398 xim_has_preediting = FALSE; 5399 5400 /* If at the start position (after typing backspace) 5401 * preedit_start_col must be reset. */ 5402 preedit_start_col = MAXCOL; 5403 } 5404 5405 im_delete_preedit(); 5406 5407 /* 5408 * Compute the end of the preediting area: "preedit_end_col". 5409 * According to the documentation of gtk_im_context_get_preedit_string(), 5410 * the cursor_pos output argument returns the offset in bytes. This is 5411 * unfortunately not true -- real life shows the offset is in characters, 5412 * and the GTK+ source code agrees with me. Will file a bug later. 5413 */ 5414 if (preedit_start_col != MAXCOL) 5415 preedit_end_col = preedit_start_col; 5416 str = (char_u *)preedit_string; 5417 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) 5418 { 5419 int is_composing; 5420 5421 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); 5422 /* 5423 * These offsets are used as counters when generating <BS> and <Del> 5424 * to delete the preedit string. So don't count composing characters 5425 * unless 'delcombine' is enabled. 5426 */ 5427 if (!is_composing || p_deco) 5428 { 5429 if (i < cursor_index) 5430 ++im_preedit_cursor; 5431 else 5432 ++im_preedit_trailing; 5433 } 5434 if (!is_composing && i >= cursor_index) 5435 { 5436 /* This is essentially the same as im_preedit_trailing, except 5437 * composing characters are not counted even if p_deco is set. */ 5438 ++num_move_back; 5439 } 5440 if (preedit_start_col != MAXCOL) 5441 preedit_end_col += utf_ptr2cells(p); 5442 } 5443 5444 if (p > str) 5445 { 5446 im_add_to_input(str, (int)(p - str)); 5447 im_correct_cursor(num_move_back); 5448 } 5449 } 5450 5451 g_free(preedit_string); 5452 5453 if (gtk_main_level() > 0) 5454 gtk_main_quit(); 5455 } 5456 5457 /* 5458 * Translate the Pango attributes at iter to Vim highlighting attributes. 5459 * Ignore attributes not supported by Vim highlighting. This shouldn't have 5460 * too much impact -- right now we handle even more attributes than necessary 5461 * for the IM modules I tested with. 5462 */ 5463 static int 5464 translate_pango_attributes(PangoAttrIterator *iter) 5465 { 5466 PangoAttribute *attr; 5467 int char_attr = HL_NORMAL; 5468 5469 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); 5470 if (attr != NULL && ((PangoAttrInt *)attr)->value 5471 != (int)PANGO_UNDERLINE_NONE) 5472 char_attr |= HL_UNDERLINE; 5473 5474 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); 5475 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) 5476 char_attr |= HL_BOLD; 5477 5478 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); 5479 if (attr != NULL && ((PangoAttrInt *)attr)->value 5480 != (int)PANGO_STYLE_NORMAL) 5481 char_attr |= HL_ITALIC; 5482 5483 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); 5484 if (attr != NULL) 5485 { 5486 const PangoColor *color = &((PangoAttrColor *)attr)->color; 5487 5488 /* Assume inverse if black background is requested */ 5489 if ((color->red | color->green | color->blue) == 0) 5490 char_attr |= HL_INVERSE; 5491 } 5492 5493 return char_attr; 5494 } 5495 5496 /* 5497 * Retrieve the highlighting attributes at column col in the preedit string. 5498 * Return -1 if not in preediting mode or if col is out of range. 5499 */ 5500 int 5501 im_get_feedback_attr(int col) 5502 { 5503 char *preedit_string = NULL; 5504 PangoAttrList *attr_list = NULL; 5505 int char_attr = -1; 5506 5507 if (xic == NULL) 5508 return char_attr; 5509 5510 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5511 5512 if (preedit_string != NULL && attr_list != NULL) 5513 { 5514 int idx; 5515 5516 /* Get the byte index as used by PangoAttrIterator */ 5517 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col) 5518 idx += utfc_ptr2len((char_u *)preedit_string + idx); 5519 5520 if (preedit_string[idx] != '\0') 5521 { 5522 PangoAttrIterator *iter; 5523 int start, end; 5524 5525 char_attr = HL_NORMAL; 5526 iter = pango_attr_list_get_iterator(attr_list); 5527 5528 /* Extract all relevant attributes from the list. */ 5529 do 5530 { 5531 pango_attr_iterator_range(iter, &start, &end); 5532 5533 if (idx >= start && idx < end) 5534 char_attr |= translate_pango_attributes(iter); 5535 } 5536 while (pango_attr_iterator_next(iter)); 5537 5538 pango_attr_iterator_destroy(iter); 5539 } 5540 } 5541 5542 if (attr_list != NULL) 5543 pango_attr_list_unref(attr_list); 5544 g_free(preedit_string); 5545 5546 return char_attr; 5547 } 5548 5549 void 5550 xim_init(void) 5551 { 5552 #ifdef XIM_DEBUG 5553 xim_log("xim_init()\n"); 5554 #endif 5555 5556 g_return_if_fail(gui.drawarea != NULL); 5557 g_return_if_fail(gtk_widget_get_window(gui.drawarea) != NULL); 5558 5559 xic = gtk_im_multicontext_new(); 5560 g_object_ref(xic); 5561 5562 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", 5563 G_CALLBACK(&im_commit_cb), NULL); 5564 g_signal_connect(G_OBJECT(xic), "preedit_changed", 5565 G_CALLBACK(&im_preedit_changed_cb), NULL); 5566 g_signal_connect(G_OBJECT(xic), "preedit_start", 5567 G_CALLBACK(&im_preedit_start_cb), NULL); 5568 g_signal_connect(G_OBJECT(xic), "preedit_end", 5569 G_CALLBACK(&im_preedit_end_cb), NULL); 5570 5571 gtk_im_context_set_client_window(xic, gtk_widget_get_window(gui.drawarea)); 5572 } 5573 5574 void 5575 im_shutdown(void) 5576 { 5577 #ifdef XIM_DEBUG 5578 xim_log("im_shutdown()\n"); 5579 #endif 5580 5581 if (xic != NULL) 5582 { 5583 gtk_im_context_focus_out(xic); 5584 g_object_unref(xic); 5585 xic = NULL; 5586 } 5587 im_is_active = FALSE; 5588 im_commit_handler_id = 0; 5589 if (p_imst == IM_ON_THE_SPOT) 5590 preedit_start_col = MAXCOL; 5591 xim_has_preediting = FALSE; 5592 } 5593 5594 /* 5595 * Convert the string argument to keyval and state for GdkEventKey. 5596 * If str is valid return TRUE, otherwise FALSE. 5597 * 5598 * See 'imactivatekey' for documentation of the format. 5599 */ 5600 static int 5601 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) 5602 { 5603 const char *mods_end; 5604 unsigned tmp_keyval; 5605 unsigned tmp_state = 0; 5606 5607 mods_end = strrchr(str, '-'); 5608 mods_end = (mods_end != NULL) ? mods_end + 1 : str; 5609 5610 /* Parse modifier keys */ 5611 while (str < mods_end) 5612 switch (*str++) 5613 { 5614 case '-': break; 5615 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; 5616 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; 5617 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; 5618 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; 5619 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; 5620 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; 5621 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; 5622 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; 5623 default: 5624 return FALSE; 5625 } 5626 5627 tmp_keyval = gdk_keyval_from_name(str); 5628 5629 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) 5630 return FALSE; 5631 5632 if (keyval != NULL) 5633 *keyval = tmp_keyval; 5634 if (state != NULL) 5635 *state = tmp_state; 5636 5637 return TRUE; 5638 } 5639 5640 /* 5641 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an 5642 * empty string is also regarded as valid. 5643 * 5644 * Note: The numerical key value of p_imak is cached if it was valid; thus 5645 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever 5646 * 'imak' changes. This is currently the case but not obvious -- should 5647 * probably rename the function for clarity. 5648 */ 5649 int 5650 im_xim_isvalid_imactivate(void) 5651 { 5652 if (p_imak[0] == NUL) 5653 { 5654 im_activatekey_keyval = GDK_VoidSymbol; 5655 im_activatekey_state = 0; 5656 return TRUE; 5657 } 5658 5659 return im_string_to_keyval((const char *)p_imak, 5660 &im_activatekey_keyval, 5661 &im_activatekey_state); 5662 } 5663 5664 static void 5665 im_synthesize_keypress(unsigned int keyval, unsigned int state) 5666 { 5667 GdkEventKey *event; 5668 5669 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); 5670 g_object_ref(gtk_widget_get_window(gui.drawarea)); 5671 /* unreffed by gdk_event_free() */ 5672 event->window = gtk_widget_get_window(gui.drawarea); 5673 event->send_event = TRUE; 5674 event->time = GDK_CURRENT_TIME; 5675 event->state = state; 5676 event->keyval = keyval; 5677 event->hardware_keycode = /* needed for XIM */ 5678 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); 5679 event->length = 0; 5680 event->string = NULL; 5681 5682 gtk_im_context_filter_keypress(xic, event); 5683 5684 /* For consistency, also send the corresponding release event. */ 5685 event->type = GDK_KEY_RELEASE; 5686 event->send_event = FALSE; 5687 gtk_im_context_filter_keypress(xic, event); 5688 5689 gdk_event_free((GdkEvent *)event); 5690 } 5691 5692 void 5693 xim_reset(void) 5694 { 5695 # ifdef FEAT_EVAL 5696 if (USE_IMACTIVATEFUNC) 5697 call_imactivatefunc(im_is_active); 5698 else 5699 # endif 5700 if (xic != NULL) 5701 { 5702 gtk_im_context_reset(xic); 5703 5704 if (p_imdisable) 5705 im_shutdown(); 5706 else 5707 { 5708 xim_set_focus(gui.in_focus); 5709 5710 if (im_activatekey_keyval != GDK_VoidSymbol) 5711 { 5712 if (im_is_active) 5713 { 5714 g_signal_handler_block(xic, im_commit_handler_id); 5715 im_synthesize_keypress(im_activatekey_keyval, 5716 im_activatekey_state); 5717 g_signal_handler_unblock(xic, im_commit_handler_id); 5718 } 5719 } 5720 else 5721 { 5722 im_shutdown(); 5723 xim_init(); 5724 xim_set_focus(gui.in_focus); 5725 } 5726 } 5727 } 5728 5729 if (p_imst == IM_ON_THE_SPOT) 5730 preedit_start_col = MAXCOL; 5731 xim_has_preediting = FALSE; 5732 } 5733 5734 int 5735 xim_queue_key_press_event(GdkEventKey *event, int down) 5736 { 5737 if (down) 5738 { 5739 /* 5740 * Workaround GTK2 XIM 'feature' that always converts keypad keys to 5741 * chars., even when not part of an IM sequence (ref. feature of 5742 * gdk/gdkkeyuni.c). 5743 * Flag any keypad keys that might represent a single char. 5744 * If this (on its own - i.e., not part of an IM sequence) is 5745 * committed while we're processing one of these keys, we can ignore 5746 * that commit and go ahead & process it ourselves. That way we can 5747 * still distinguish keypad keys for use in mappings. 5748 * Also add GDK_space to make <S-Space> work. 5749 */ 5750 switch (event->keyval) 5751 { 5752 case GDK_KP_Add: xim_expected_char = '+'; break; 5753 case GDK_KP_Subtract: xim_expected_char = '-'; break; 5754 case GDK_KP_Divide: xim_expected_char = '/'; break; 5755 case GDK_KP_Multiply: xim_expected_char = '*'; break; 5756 case GDK_KP_Decimal: xim_expected_char = '.'; break; 5757 case GDK_KP_Equal: xim_expected_char = '='; break; 5758 case GDK_KP_0: xim_expected_char = '0'; break; 5759 case GDK_KP_1: xim_expected_char = '1'; break; 5760 case GDK_KP_2: xim_expected_char = '2'; break; 5761 case GDK_KP_3: xim_expected_char = '3'; break; 5762 case GDK_KP_4: xim_expected_char = '4'; break; 5763 case GDK_KP_5: xim_expected_char = '5'; break; 5764 case GDK_KP_6: xim_expected_char = '6'; break; 5765 case GDK_KP_7: xim_expected_char = '7'; break; 5766 case GDK_KP_8: xim_expected_char = '8'; break; 5767 case GDK_KP_9: xim_expected_char = '9'; break; 5768 case GDK_space: xim_expected_char = ' '; break; 5769 default: xim_expected_char = NUL; 5770 } 5771 xim_ignored_char = FALSE; 5772 } 5773 5774 /* 5775 * When typing fFtT, XIM may be activated. Thus it must pass 5776 * gtk_im_context_filter_keypress() in Normal mode. 5777 * And while doing :sh too. 5778 */ 5779 if (xic != NULL && !p_imdisable 5780 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) 5781 { 5782 /* 5783 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is 5784 * always aware of the current status of IM, and can even emulate 5785 * the activation key for modules that don't support one. 5786 */ 5787 if (event->keyval == im_activatekey_keyval 5788 && (event->state & im_activatekey_state) == im_activatekey_state) 5789 { 5790 unsigned int state_mask; 5791 5792 /* Require the state of the 3 most used modifiers to match exactly. 5793 * Otherwise e.g. <S-C-space> would be unusable for other purposes 5794 * if the IM activate key is <S-space>. */ 5795 state_mask = im_activatekey_state; 5796 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK 5797 | (int)GDK_MOD1_MASK); 5798 5799 if ((event->state & state_mask) != im_activatekey_state) 5800 return FALSE; 5801 5802 /* Don't send it a second time on GDK_KEY_RELEASE. */ 5803 if (event->type != GDK_KEY_PRESS) 5804 return TRUE; 5805 5806 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE)) 5807 { 5808 im_set_active(FALSE); 5809 5810 /* ":lmap" mappings exists, toggle use of mappings. */ 5811 State ^= LANGMAP; 5812 if (State & LANGMAP) 5813 { 5814 curbuf->b_p_iminsert = B_IMODE_NONE; 5815 State &= ~LANGMAP; 5816 } 5817 else 5818 { 5819 curbuf->b_p_iminsert = B_IMODE_LMAP; 5820 State |= LANGMAP; 5821 } 5822 return TRUE; 5823 } 5824 5825 return gtk_im_context_filter_keypress(xic, event); 5826 } 5827 5828 /* Don't filter events through the IM context if IM isn't active 5829 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module 5830 * not doing anything before the activation key was sent. */ 5831 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) 5832 { 5833 int imresult = gtk_im_context_filter_keypress(xic, event); 5834 5835 if (p_imst == IM_ON_THE_SPOT) 5836 { 5837 /* Some XIM send following sequence: 5838 * 1. preedited string. 5839 * 2. committed string. 5840 * 3. line changed key. 5841 * 4. preedited string. 5842 * 5. remove preedited string. 5843 * if 3, Vim can't move back the above line for 5. 5844 * thus, this part should not parse the key. */ 5845 if (!imresult && preedit_start_col != MAXCOL 5846 && event->keyval == GDK_Return) 5847 { 5848 im_synthesize_keypress(GDK_Return, 0U); 5849 return FALSE; 5850 } 5851 } 5852 5853 /* If XIM tried to commit a keypad key as a single char., 5854 * ignore it so we can use the keypad key 'raw', for mappings. */ 5855 if (xim_expected_char != NUL && xim_ignored_char) 5856 /* We had a keypad key, and XIM tried to thieve it */ 5857 return FALSE; 5858 5859 /* This is supposed to fix a problem with iBus, that space 5860 * characters don't work in input mode. */ 5861 xim_expected_char = NUL; 5862 5863 /* Normal processing */ 5864 return imresult; 5865 } 5866 } 5867 5868 return FALSE; 5869 } 5870 5871 int 5872 im_get_status(void) 5873 { 5874 # ifdef FEAT_EVAL 5875 if (USE_IMSTATUSFUNC) 5876 return call_imstatusfunc(); 5877 # endif 5878 return im_is_active; 5879 } 5880 5881 int 5882 preedit_get_status(void) 5883 { 5884 return preedit_is_active; 5885 } 5886 5887 int 5888 im_is_preediting(void) 5889 { 5890 return xim_has_preediting; 5891 } 5892 5893 # else /* !FEAT_GUI_GTK */ 5894 5895 static int xim_is_active = FALSE; /* XIM should be active in the current 5896 mode */ 5897 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */ 5898 # ifdef FEAT_GUI_X11 5899 static XIMStyle input_style; 5900 static int status_area_enabled = TRUE; 5901 # endif 5902 5903 /* 5904 * Switch using XIM on/off. This is used by the code that changes "State". 5905 * When 'imactivatefunc' is defined use that function instead. 5906 */ 5907 void 5908 im_set_active(int active_arg) 5909 { 5910 int active = active_arg; 5911 5912 /* If 'imdisable' is set, XIM is never active. */ 5913 if (p_imdisable) 5914 active = FALSE; 5915 else if (input_style & XIMPreeditPosition) 5916 /* There is a problem in switching XIM off when preediting is used, 5917 * and it is not clear how this can be solved. For now, keep XIM on 5918 * all the time, like it was done in Vim 5.8. */ 5919 active = TRUE; 5920 5921 # if defined(FEAT_EVAL) 5922 if (USE_IMACTIVATEFUNC) 5923 { 5924 if (active != im_get_status()) 5925 { 5926 call_imactivatefunc(active); 5927 xim_has_focus = active; 5928 } 5929 return; 5930 } 5931 # endif 5932 5933 if (xic == NULL) 5934 return; 5935 5936 /* Remember the active state, it is needed when Vim gets keyboard focus. */ 5937 xim_is_active = active; 5938 xim_set_preedit(); 5939 } 5940 5941 /* 5942 * Adjust using XIM for gaining or losing keyboard focus. Also called when 5943 * "xim_is_active" changes. 5944 */ 5945 void 5946 xim_set_focus(int focus) 5947 { 5948 if (xic == NULL) 5949 return; 5950 5951 /* 5952 * XIM only gets focus when the Vim window has keyboard focus and XIM has 5953 * been set active for the current mode. 5954 */ 5955 if (focus && xim_is_active) 5956 { 5957 if (!xim_has_focus) 5958 { 5959 xim_has_focus = TRUE; 5960 XSetICFocus(xic); 5961 } 5962 } 5963 else 5964 { 5965 if (xim_has_focus) 5966 { 5967 xim_has_focus = FALSE; 5968 XUnsetICFocus(xic); 5969 } 5970 } 5971 } 5972 5973 void 5974 im_set_position(int row UNUSED, int col UNUSED) 5975 { 5976 xim_set_preedit(); 5977 } 5978 5979 /* 5980 * Set the XIM to the current cursor position. 5981 */ 5982 void 5983 xim_set_preedit(void) 5984 { 5985 XVaNestedList attr_list; 5986 XRectangle spot_area; 5987 XPoint over_spot; 5988 int line_space; 5989 5990 if (xic == NULL) 5991 return; 5992 5993 xim_set_focus(TRUE); 5994 5995 if (!xim_has_focus) 5996 { 5997 /* hide XIM cursor */ 5998 over_spot.x = 0; 5999 over_spot.y = -100; /* arbitrary invisible position */ 6000 attr_list = (XVaNestedList) XVaCreateNestedList(0, 6001 XNSpotLocation, 6002 &over_spot, 6003 NULL); 6004 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); 6005 XFree(attr_list); 6006 return; 6007 } 6008 6009 if (input_style & XIMPreeditPosition) 6010 { 6011 if (xim_fg_color == INVALCOLOR) 6012 { 6013 xim_fg_color = gui.def_norm_pixel; 6014 xim_bg_color = gui.def_back_pixel; 6015 } 6016 over_spot.x = TEXT_X(gui.col); 6017 over_spot.y = TEXT_Y(gui.row); 6018 spot_area.x = 0; 6019 spot_area.y = 0; 6020 spot_area.height = gui.char_height * Rows; 6021 spot_area.width = gui.char_width * Columns; 6022 line_space = gui.char_height; 6023 attr_list = (XVaNestedList) XVaCreateNestedList(0, 6024 XNSpotLocation, &over_spot, 6025 XNForeground, (Pixel) xim_fg_color, 6026 XNBackground, (Pixel) xim_bg_color, 6027 XNArea, &spot_area, 6028 XNLineSpace, line_space, 6029 NULL); 6030 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) 6031 emsg(_("E284: Cannot set IC values")); 6032 XFree(attr_list); 6033 } 6034 } 6035 6036 # if defined(FEAT_GUI_X11) 6037 static char e_xim[] = N_("E285: Failed to create input context"); 6038 # endif 6039 6040 # if defined(FEAT_GUI_X11) || defined(PROTO) 6041 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(SUN_SYSTEM) 6042 # define USE_X11R6_XIM 6043 # endif 6044 6045 static int xim_real_init(Window x11_window, Display *x11_display); 6046 6047 6048 # ifdef USE_X11R6_XIM 6049 static void 6050 xim_instantiate_cb( 6051 Display *display, 6052 XPointer client_data UNUSED, 6053 XPointer call_data UNUSED) 6054 { 6055 Window x11_window; 6056 Display *x11_display; 6057 6058 # ifdef XIM_DEBUG 6059 xim_log("xim_instantiate_cb()\n"); 6060 # endif 6061 6062 gui_get_x11_windis(&x11_window, &x11_display); 6063 if (display != x11_display) 6064 return; 6065 6066 xim_real_init(x11_window, x11_display); 6067 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6068 if (xic != NULL) 6069 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6070 xim_instantiate_cb, NULL); 6071 } 6072 6073 static void 6074 xim_destroy_cb( 6075 XIM im UNUSED, 6076 XPointer client_data UNUSED, 6077 XPointer call_data UNUSED) 6078 { 6079 Window x11_window; 6080 Display *x11_display; 6081 6082 # ifdef XIM_DEBUG 6083 xim_log("xim_destroy_cb()\n"); 6084 #endif 6085 gui_get_x11_windis(&x11_window, &x11_display); 6086 6087 xic = NULL; 6088 status_area_enabled = FALSE; 6089 6090 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6091 6092 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6093 xim_instantiate_cb, NULL); 6094 } 6095 # endif 6096 6097 void 6098 xim_init(void) 6099 { 6100 Window x11_window; 6101 Display *x11_display; 6102 6103 # ifdef XIM_DEBUG 6104 xim_log("xim_init()\n"); 6105 # endif 6106 6107 gui_get_x11_windis(&x11_window, &x11_display); 6108 6109 xic = NULL; 6110 6111 if (xim_real_init(x11_window, x11_display)) 6112 return; 6113 6114 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6115 6116 # ifdef USE_X11R6_XIM 6117 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6118 xim_instantiate_cb, NULL); 6119 # endif 6120 } 6121 6122 static int 6123 xim_real_init(Window x11_window, Display *x11_display) 6124 { 6125 int i; 6126 char *p, 6127 *s, 6128 *ns, 6129 *end, 6130 tmp[1024]; 6131 # define IMLEN_MAX 40 6132 char buf[IMLEN_MAX + 7]; 6133 XIM xim = NULL; 6134 XIMStyles *xim_styles; 6135 XIMStyle this_input_style = 0; 6136 Boolean found; 6137 XPoint over_spot; 6138 XVaNestedList preedit_list, status_list; 6139 6140 input_style = 0; 6141 status_area_enabled = FALSE; 6142 6143 if (xic != NULL) 6144 return FALSE; 6145 6146 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) 6147 { 6148 strcpy(tmp, gui.rsrc_input_method); 6149 for (ns = s = tmp; ns != NULL && *s != NUL;) 6150 { 6151 s = (char *)skipwhite((char_u *)s); 6152 if (*s == NUL) 6153 break; 6154 if ((ns = end = strchr(s, ',')) == NULL) 6155 end = s + strlen(s); 6156 while (isspace(((char_u *)end)[-1])) 6157 end--; 6158 *end = NUL; 6159 6160 if (strlen(s) <= IMLEN_MAX) 6161 { 6162 strcpy(buf, "@im="); 6163 strcat(buf, s); 6164 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL 6165 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) 6166 != NULL) 6167 break; 6168 } 6169 6170 s = ns + 1; 6171 } 6172 } 6173 6174 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) 6175 xim = XOpenIM(x11_display, NULL, NULL, NULL); 6176 6177 /* This is supposed to be useful to obtain characters through 6178 * XmbLookupString() without really using a XIM. */ 6179 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL 6180 && *p != NUL) 6181 xim = XOpenIM(x11_display, NULL, NULL, NULL); 6182 6183 if (xim == NULL) 6184 { 6185 /* Only give this message when verbose is set, because too many people 6186 * got this message when they didn't want to use a XIM. */ 6187 if (p_verbose > 0) 6188 { 6189 verbose_enter(); 6190 emsg(_("E286: Failed to open input method")); 6191 verbose_leave(); 6192 } 6193 return FALSE; 6194 } 6195 6196 # ifdef USE_X11R6_XIM 6197 { 6198 XIMCallback destroy_cb; 6199 6200 destroy_cb.callback = xim_destroy_cb; 6201 destroy_cb.client_data = NULL; 6202 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) 6203 emsg(_("E287: Warning: Could not set destroy callback to IM")); 6204 } 6205 # endif 6206 6207 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) 6208 { 6209 emsg(_("E288: input method doesn't support any style")); 6210 XCloseIM(xim); 6211 return FALSE; 6212 } 6213 6214 found = False; 6215 strcpy(tmp, gui.rsrc_preedit_type_name); 6216 for (s = tmp; s && !found; ) 6217 { 6218 while (*s && isspace((unsigned char)*s)) 6219 s++; 6220 if (!*s) 6221 break; 6222 if ((ns = end = strchr(s, ',')) != 0) 6223 ns++; 6224 else 6225 end = s + strlen(s); 6226 while (isspace((unsigned char)*end)) 6227 end--; 6228 *end = '\0'; 6229 6230 if (!strcmp(s, "OverTheSpot")) 6231 this_input_style = (XIMPreeditPosition | XIMStatusArea); 6232 else if (!strcmp(s, "OffTheSpot")) 6233 this_input_style = (XIMPreeditArea | XIMStatusArea); 6234 else if (!strcmp(s, "Root")) 6235 this_input_style = (XIMPreeditNothing | XIMStatusNothing); 6236 6237 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 6238 { 6239 if (this_input_style == xim_styles->supported_styles[i]) 6240 { 6241 found = True; 6242 break; 6243 } 6244 } 6245 if (!found) 6246 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 6247 { 6248 if ((xim_styles->supported_styles[i] & this_input_style) 6249 == (this_input_style & ~XIMStatusArea)) 6250 { 6251 this_input_style &= ~XIMStatusArea; 6252 found = True; 6253 break; 6254 } 6255 } 6256 6257 s = ns; 6258 } 6259 XFree(xim_styles); 6260 6261 if (!found) 6262 { 6263 /* Only give this message when verbose is set, because too many people 6264 * got this message when they didn't want to use a XIM. */ 6265 if (p_verbose > 0) 6266 { 6267 verbose_enter(); 6268 emsg(_("E289: input method doesn't support my preedit type")); 6269 verbose_leave(); 6270 } 6271 XCloseIM(xim); 6272 return FALSE; 6273 } 6274 6275 over_spot.x = TEXT_X(gui.col); 6276 over_spot.y = TEXT_Y(gui.row); 6277 input_style = this_input_style; 6278 6279 /* A crash was reported when trying to pass gui.norm_font as XNFontSet, 6280 * thus that has been removed. Hopefully the default works... */ 6281 # ifdef FEAT_XFONTSET 6282 if (gui.fontset != NOFONTSET) 6283 { 6284 preedit_list = XVaCreateNestedList(0, 6285 XNSpotLocation, &over_spot, 6286 XNForeground, (Pixel)gui.def_norm_pixel, 6287 XNBackground, (Pixel)gui.def_back_pixel, 6288 XNFontSet, (XFontSet)gui.fontset, 6289 NULL); 6290 status_list = XVaCreateNestedList(0, 6291 XNForeground, (Pixel)gui.def_norm_pixel, 6292 XNBackground, (Pixel)gui.def_back_pixel, 6293 XNFontSet, (XFontSet)gui.fontset, 6294 NULL); 6295 } 6296 else 6297 # endif 6298 { 6299 preedit_list = XVaCreateNestedList(0, 6300 XNSpotLocation, &over_spot, 6301 XNForeground, (Pixel)gui.def_norm_pixel, 6302 XNBackground, (Pixel)gui.def_back_pixel, 6303 NULL); 6304 status_list = XVaCreateNestedList(0, 6305 XNForeground, (Pixel)gui.def_norm_pixel, 6306 XNBackground, (Pixel)gui.def_back_pixel, 6307 NULL); 6308 } 6309 6310 xic = XCreateIC(xim, 6311 XNInputStyle, input_style, 6312 XNClientWindow, x11_window, 6313 XNFocusWindow, gui.wid, 6314 XNPreeditAttributes, preedit_list, 6315 XNStatusAttributes, status_list, 6316 NULL); 6317 XFree(status_list); 6318 XFree(preedit_list); 6319 if (xic != NULL) 6320 { 6321 if (input_style & XIMStatusArea) 6322 { 6323 xim_set_status_area(); 6324 status_area_enabled = TRUE; 6325 } 6326 else 6327 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6328 } 6329 else 6330 { 6331 if (!is_not_a_term()) 6332 emsg(_(e_xim)); 6333 XCloseIM(xim); 6334 return FALSE; 6335 } 6336 6337 return TRUE; 6338 } 6339 6340 # endif /* FEAT_GUI_X11 */ 6341 6342 /* 6343 * Get IM status. When IM is on, return TRUE. Else return FALSE. 6344 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is 6345 * active, when not having focus XIM may still be active (e.g., when using a 6346 * tear-off menu item). 6347 */ 6348 int 6349 im_get_status(void) 6350 { 6351 # ifdef FEAT_EVAL 6352 if (USE_IMSTATUSFUNC) 6353 return call_imstatusfunc(); 6354 # endif 6355 return xim_has_focus; 6356 } 6357 6358 # endif /* !FEAT_GUI_GTK */ 6359 6360 # if !defined(FEAT_GUI_GTK) || defined(PROTO) 6361 /* 6362 * Set up the status area. 6363 * 6364 * This should use a separate Widget, but that seems not possible, because 6365 * preedit_area and status_area should be set to the same window as for the 6366 * text input. Unfortunately this means the status area pollutes the text 6367 * window... 6368 */ 6369 void 6370 xim_set_status_area(void) 6371 { 6372 XVaNestedList preedit_list = 0, status_list = 0, list = 0; 6373 XRectangle pre_area, status_area; 6374 6375 if (xic == NULL) 6376 return; 6377 6378 if (input_style & XIMStatusArea) 6379 { 6380 if (input_style & XIMPreeditArea) 6381 { 6382 XRectangle *needed_rect; 6383 6384 /* to get status_area width */ 6385 status_list = XVaCreateNestedList(0, XNAreaNeeded, 6386 &needed_rect, NULL); 6387 XGetICValues(xic, XNStatusAttributes, status_list, NULL); 6388 XFree(status_list); 6389 6390 status_area.width = needed_rect->width; 6391 } 6392 else 6393 status_area.width = gui.char_width * Columns; 6394 6395 status_area.x = 0; 6396 status_area.y = gui.char_height * Rows + gui.border_offset; 6397 if (gui.which_scrollbars[SBAR_BOTTOM]) 6398 status_area.y += gui.scrollbar_height; 6399 #ifdef FEAT_MENU 6400 if (gui.menu_is_active) 6401 status_area.y += gui.menu_height; 6402 #endif 6403 status_area.height = gui.char_height; 6404 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); 6405 } 6406 else 6407 { 6408 status_area.x = 0; 6409 status_area.y = gui.char_height * Rows + gui.border_offset; 6410 if (gui.which_scrollbars[SBAR_BOTTOM]) 6411 status_area.y += gui.scrollbar_height; 6412 #ifdef FEAT_MENU 6413 if (gui.menu_is_active) 6414 status_area.y += gui.menu_height; 6415 #endif 6416 status_area.width = 0; 6417 status_area.height = gui.char_height; 6418 } 6419 6420 if (input_style & XIMPreeditArea) /* off-the-spot */ 6421 { 6422 pre_area.x = status_area.x + status_area.width; 6423 pre_area.y = gui.char_height * Rows + gui.border_offset; 6424 pre_area.width = gui.char_width * Columns - pre_area.x; 6425 if (gui.which_scrollbars[SBAR_BOTTOM]) 6426 pre_area.y += gui.scrollbar_height; 6427 #ifdef FEAT_MENU 6428 if (gui.menu_is_active) 6429 pre_area.y += gui.menu_height; 6430 #endif 6431 pre_area.height = gui.char_height; 6432 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6433 } 6434 else if (input_style & XIMPreeditPosition) /* over-the-spot */ 6435 { 6436 pre_area.x = 0; 6437 pre_area.y = 0; 6438 pre_area.height = gui.char_height * Rows; 6439 pre_area.width = gui.char_width * Columns; 6440 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6441 } 6442 6443 if (preedit_list && status_list) 6444 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6445 XNStatusAttributes, status_list, NULL); 6446 else if (preedit_list) 6447 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6448 NULL); 6449 else if (status_list) 6450 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, 6451 NULL); 6452 else 6453 list = NULL; 6454 6455 if (list) 6456 { 6457 XSetICValues(xic, XNVaNestedList, list, NULL); 6458 XFree(list); 6459 } 6460 if (status_list) 6461 XFree(status_list); 6462 if (preedit_list) 6463 XFree(preedit_list); 6464 } 6465 6466 int 6467 xim_get_status_area_height(void) 6468 { 6469 if (status_area_enabled) 6470 return gui.char_height; 6471 return 0; 6472 } 6473 # endif 6474 6475 #else /* !defined(FEAT_XIM) */ 6476 6477 # ifdef IME_WITHOUT_XIM 6478 static int im_was_set_active = FALSE; 6479 6480 int 6481 im_get_status(void) 6482 { 6483 # if defined(FEAT_MBYTE) && defined(FEAT_EVAL) 6484 if (USE_IMSTATUSFUNC) 6485 return call_imstatusfunc(); 6486 # endif 6487 return im_was_set_active; 6488 } 6489 6490 void 6491 im_set_active(int active_arg) 6492 { 6493 # if defined(FEAT_MBYTE) && defined(FEAT_EVAL) 6494 int active = !p_imdisable && active_arg; 6495 6496 if (USE_IMACTIVATEFUNC && active != im_get_status()) 6497 { 6498 call_imactivatefunc(active); 6499 im_was_set_active = active; 6500 } 6501 # endif 6502 } 6503 6504 # ifdef FEAT_GUI 6505 void 6506 im_set_position(int row UNUSED, int col UNUSED) 6507 { 6508 } 6509 # endif 6510 # endif 6511 6512 #endif /* FEAT_XIM */ 6513 6514 #if defined(FEAT_MBYTE) || defined(PROTO) 6515 6516 /* 6517 * Setup "vcp" for conversion from "from" to "to". 6518 * The names must have been made canonical with enc_canonize(). 6519 * vcp->vc_type must have been initialized to CONV_NONE. 6520 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 6521 * instead). 6522 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 6523 * Return FAIL when conversion is not supported, OK otherwise. 6524 */ 6525 int 6526 convert_setup(vimconv_T *vcp, char_u *from, char_u *to) 6527 { 6528 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 6529 } 6530 6531 /* 6532 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 6533 * "from" unicode charsets be considered utf-8. Same for "to". 6534 */ 6535 int 6536 convert_setup_ext( 6537 vimconv_T *vcp, 6538 char_u *from, 6539 int from_unicode_is_utf8, 6540 char_u *to, 6541 int to_unicode_is_utf8) 6542 { 6543 int from_prop; 6544 int to_prop; 6545 int from_is_utf8; 6546 int to_is_utf8; 6547 6548 /* Reset to no conversion. */ 6549 # ifdef USE_ICONV 6550 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 6551 iconv_close(vcp->vc_fd); 6552 # endif 6553 vcp->vc_type = CONV_NONE; 6554 vcp->vc_factor = 1; 6555 vcp->vc_fail = FALSE; 6556 6557 /* No conversion when one of the names is empty or they are equal. */ 6558 if (from == NULL || *from == NUL || to == NULL || *to == NUL 6559 || STRCMP(from, to) == 0) 6560 return OK; 6561 6562 from_prop = enc_canon_props(from); 6563 to_prop = enc_canon_props(to); 6564 if (from_unicode_is_utf8) 6565 from_is_utf8 = from_prop & ENC_UNICODE; 6566 else 6567 from_is_utf8 = from_prop == ENC_UNICODE; 6568 if (to_unicode_is_utf8) 6569 to_is_utf8 = to_prop & ENC_UNICODE; 6570 else 6571 to_is_utf8 = to_prop == ENC_UNICODE; 6572 6573 if ((from_prop & ENC_LATIN1) && to_is_utf8) 6574 { 6575 /* Internal latin1 -> utf-8 conversion. */ 6576 vcp->vc_type = CONV_TO_UTF8; 6577 vcp->vc_factor = 2; /* up to twice as long */ 6578 } 6579 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 6580 { 6581 /* Internal latin9 -> utf-8 conversion. */ 6582 vcp->vc_type = CONV_9_TO_UTF8; 6583 vcp->vc_factor = 3; /* up to three as long (euro sign) */ 6584 } 6585 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 6586 { 6587 /* Internal utf-8 -> latin1 conversion. */ 6588 vcp->vc_type = CONV_TO_LATIN1; 6589 } 6590 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 6591 { 6592 /* Internal utf-8 -> latin9 conversion. */ 6593 vcp->vc_type = CONV_TO_LATIN9; 6594 } 6595 #ifdef WIN3264 6596 /* Win32-specific codepage <-> codepage conversion without iconv. */ 6597 else if ((from_is_utf8 || encname2codepage(from) > 0) 6598 && (to_is_utf8 || encname2codepage(to) > 0)) 6599 { 6600 vcp->vc_type = CONV_CODEPAGE; 6601 vcp->vc_factor = 2; /* up to twice as long */ 6602 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 6603 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 6604 } 6605 #endif 6606 #ifdef MACOS_CONVERT 6607 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 6608 { 6609 vcp->vc_type = CONV_MAC_LATIN1; 6610 } 6611 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 6612 { 6613 vcp->vc_type = CONV_MAC_UTF8; 6614 vcp->vc_factor = 2; /* up to twice as long */ 6615 } 6616 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 6617 { 6618 vcp->vc_type = CONV_LATIN1_MAC; 6619 } 6620 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 6621 { 6622 vcp->vc_type = CONV_UTF8_MAC; 6623 } 6624 #endif 6625 # ifdef USE_ICONV 6626 else 6627 { 6628 /* Use iconv() for conversion. */ 6629 vcp->vc_fd = (iconv_t)my_iconv_open( 6630 to_is_utf8 ? (char_u *)"utf-8" : to, 6631 from_is_utf8 ? (char_u *)"utf-8" : from); 6632 if (vcp->vc_fd != (iconv_t)-1) 6633 { 6634 vcp->vc_type = CONV_ICONV; 6635 vcp->vc_factor = 4; /* could be longer too... */ 6636 } 6637 } 6638 # endif 6639 if (vcp->vc_type == CONV_NONE) 6640 return FAIL; 6641 6642 return OK; 6643 } 6644 6645 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \ 6646 || defined(PROTO) 6647 /* 6648 * Do conversion on typed input characters in-place. 6649 * The input and output are not NUL terminated! 6650 * Returns the length after conversion. 6651 */ 6652 int 6653 convert_input(char_u *ptr, int len, int maxlen) 6654 { 6655 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 6656 } 6657 #endif 6658 6659 /* 6660 * Like convert_input(), but when there is an incomplete byte sequence at the 6661 * end return that as an allocated string in "restp" and set "*restlenp" to 6662 * the length. If "restp" is NULL it is not used. 6663 */ 6664 int 6665 convert_input_safe( 6666 char_u *ptr, 6667 int len, 6668 int maxlen, 6669 char_u **restp, 6670 int *restlenp) 6671 { 6672 char_u *d; 6673 int dlen = len; 6674 int unconvertlen = 0; 6675 6676 d = string_convert_ext(&input_conv, ptr, &dlen, 6677 restp == NULL ? NULL : &unconvertlen); 6678 if (d != NULL) 6679 { 6680 if (dlen <= maxlen) 6681 { 6682 if (unconvertlen > 0) 6683 { 6684 /* Move the unconverted characters to allocated memory. */ 6685 *restp = alloc(unconvertlen); 6686 if (*restp != NULL) 6687 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 6688 *restlenp = unconvertlen; 6689 } 6690 mch_memmove(ptr, d, dlen); 6691 } 6692 else 6693 /* result is too long, keep the unconverted text (the caller must 6694 * have done something wrong!) */ 6695 dlen = len; 6696 vim_free(d); 6697 } 6698 return dlen; 6699 } 6700 6701 /* 6702 * Convert text "ptr[*lenp]" according to "vcp". 6703 * Returns the result in allocated memory and sets "*lenp". 6704 * When "lenp" is NULL, use NUL terminated strings. 6705 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 6706 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 6707 */ 6708 char_u * 6709 string_convert( 6710 vimconv_T *vcp, 6711 char_u *ptr, 6712 int *lenp) 6713 { 6714 return string_convert_ext(vcp, ptr, lenp, NULL); 6715 } 6716 6717 /* 6718 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 6719 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 6720 * set to the number of remaining bytes. 6721 */ 6722 char_u * 6723 string_convert_ext( 6724 vimconv_T *vcp, 6725 char_u *ptr, 6726 int *lenp, 6727 int *unconvlenp) 6728 { 6729 char_u *retval = NULL; 6730 char_u *d; 6731 int len; 6732 int i; 6733 int l; 6734 int c; 6735 6736 if (lenp == NULL) 6737 len = (int)STRLEN(ptr); 6738 else 6739 len = *lenp; 6740 if (len == 0) 6741 return vim_strsave((char_u *)""); 6742 6743 switch (vcp->vc_type) 6744 { 6745 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */ 6746 retval = alloc(len * 2 + 1); 6747 if (retval == NULL) 6748 break; 6749 d = retval; 6750 for (i = 0; i < len; ++i) 6751 { 6752 c = ptr[i]; 6753 if (c < 0x80) 6754 *d++ = c; 6755 else 6756 { 6757 *d++ = 0xc0 + ((unsigned)c >> 6); 6758 *d++ = 0x80 + (c & 0x3f); 6759 } 6760 } 6761 *d = NUL; 6762 if (lenp != NULL) 6763 *lenp = (int)(d - retval); 6764 break; 6765 6766 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */ 6767 retval = alloc(len * 3 + 1); 6768 if (retval == NULL) 6769 break; 6770 d = retval; 6771 for (i = 0; i < len; ++i) 6772 { 6773 c = ptr[i]; 6774 switch (c) 6775 { 6776 case 0xa4: c = 0x20ac; break; /* euro */ 6777 case 0xa6: c = 0x0160; break; /* S hat */ 6778 case 0xa8: c = 0x0161; break; /* S -hat */ 6779 case 0xb4: c = 0x017d; break; /* Z hat */ 6780 case 0xb8: c = 0x017e; break; /* Z -hat */ 6781 case 0xbc: c = 0x0152; break; /* OE */ 6782 case 0xbd: c = 0x0153; break; /* oe */ 6783 case 0xbe: c = 0x0178; break; /* Y */ 6784 } 6785 d += utf_char2bytes(c, d); 6786 } 6787 *d = NUL; 6788 if (lenp != NULL) 6789 *lenp = (int)(d - retval); 6790 break; 6791 6792 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */ 6793 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */ 6794 retval = alloc(len + 1); 6795 if (retval == NULL) 6796 break; 6797 d = retval; 6798 for (i = 0; i < len; ++i) 6799 { 6800 l = utf_ptr2len_len(ptr + i, len - i); 6801 if (l == 0) 6802 *d++ = NUL; 6803 else if (l == 1) 6804 { 6805 int l_w = utf8len_tab_zero[ptr[i]]; 6806 6807 if (l_w == 0) 6808 { 6809 /* Illegal utf-8 byte cannot be converted */ 6810 vim_free(retval); 6811 return NULL; 6812 } 6813 if (unconvlenp != NULL && l_w > len - i) 6814 { 6815 /* Incomplete sequence at the end. */ 6816 *unconvlenp = len - i; 6817 break; 6818 } 6819 *d++ = ptr[i]; 6820 } 6821 else 6822 { 6823 c = utf_ptr2char(ptr + i); 6824 if (vcp->vc_type == CONV_TO_LATIN9) 6825 switch (c) 6826 { 6827 case 0x20ac: c = 0xa4; break; /* euro */ 6828 case 0x0160: c = 0xa6; break; /* S hat */ 6829 case 0x0161: c = 0xa8; break; /* S -hat */ 6830 case 0x017d: c = 0xb4; break; /* Z hat */ 6831 case 0x017e: c = 0xb8; break; /* Z -hat */ 6832 case 0x0152: c = 0xbc; break; /* OE */ 6833 case 0x0153: c = 0xbd; break; /* oe */ 6834 case 0x0178: c = 0xbe; break; /* Y */ 6835 case 0xa4: 6836 case 0xa6: 6837 case 0xa8: 6838 case 0xb4: 6839 case 0xb8: 6840 case 0xbc: 6841 case 0xbd: 6842 case 0xbe: c = 0x100; break; /* not in latin9 */ 6843 } 6844 if (!utf_iscomposing(c)) /* skip composing chars */ 6845 { 6846 if (c < 0x100) 6847 *d++ = c; 6848 else if (vcp->vc_fail) 6849 { 6850 vim_free(retval); 6851 return NULL; 6852 } 6853 else 6854 { 6855 *d++ = 0xbf; 6856 if (utf_char2cells(c) > 1) 6857 *d++ = '?'; 6858 } 6859 } 6860 i += l - 1; 6861 } 6862 } 6863 *d = NUL; 6864 if (lenp != NULL) 6865 *lenp = (int)(d - retval); 6866 break; 6867 6868 # ifdef MACOS_CONVERT 6869 case CONV_MAC_LATIN1: 6870 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6871 'm', 'l', unconvlenp); 6872 break; 6873 6874 case CONV_LATIN1_MAC: 6875 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6876 'l', 'm', unconvlenp); 6877 break; 6878 6879 case CONV_MAC_UTF8: 6880 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6881 'm', 'u', unconvlenp); 6882 break; 6883 6884 case CONV_UTF8_MAC: 6885 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6886 'u', 'm', unconvlenp); 6887 break; 6888 # endif 6889 6890 # ifdef USE_ICONV 6891 case CONV_ICONV: /* conversion with output_conv.vc_fd */ 6892 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 6893 break; 6894 # endif 6895 # ifdef WIN3264 6896 case CONV_CODEPAGE: /* codepage -> codepage */ 6897 { 6898 int retlen; 6899 int tmp_len; 6900 short_u *tmp; 6901 6902 /* 1. codepage/UTF-8 -> ucs-2. */ 6903 if (vcp->vc_cpfrom == 0) 6904 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 6905 else 6906 { 6907 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 6908 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 6909 (char *)ptr, len, 0, 0); 6910 if (tmp_len == 0 6911 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 6912 { 6913 if (lenp != NULL) 6914 *lenp = 0; 6915 if (unconvlenp != NULL) 6916 *unconvlenp = len; 6917 retval = alloc(1); 6918 if (retval) 6919 retval[0] = NUL; 6920 return retval; 6921 } 6922 } 6923 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len); 6924 if (tmp == NULL) 6925 break; 6926 if (vcp->vc_cpfrom == 0) 6927 utf8_to_utf16(ptr, len, tmp, unconvlenp); 6928 else 6929 MultiByteToWideChar(vcp->vc_cpfrom, 0, 6930 (char *)ptr, len, tmp, tmp_len); 6931 6932 /* 2. ucs-2 -> codepage/UTF-8. */ 6933 if (vcp->vc_cpto == 0) 6934 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 6935 else 6936 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 6937 tmp, tmp_len, 0, 0, 0, 0); 6938 retval = alloc(retlen + 1); 6939 if (retval != NULL) 6940 { 6941 if (vcp->vc_cpto == 0) 6942 utf16_to_utf8(tmp, tmp_len, retval); 6943 else 6944 WideCharToMultiByte(vcp->vc_cpto, 0, 6945 tmp, tmp_len, 6946 (char *)retval, retlen, 0, 0); 6947 retval[retlen] = NUL; 6948 if (lenp != NULL) 6949 *lenp = retlen; 6950 } 6951 vim_free(tmp); 6952 break; 6953 } 6954 # endif 6955 } 6956 6957 return retval; 6958 } 6959 #endif 6960