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 enc_canon_search(char_u *name); 142 static int dbcs_char2len(int c); 143 static int dbcs_char2bytes(int c, char_u *buf); 144 static int dbcs_ptr2len(char_u *p); 145 static int dbcs_ptr2len_len(char_u *p, int size); 146 static int utf_ptr2cells_len(char_u *p, int size); 147 static int dbcs_char2cells(int c); 148 static int dbcs_ptr2cells_len(char_u *p, int size); 149 static int dbcs_ptr2char(char_u *p); 150 static int utf_safe_read_char_adv(char_u **s, size_t *n); 151 152 /* 153 * Lookup table to quickly get the length in bytes of a UTF-8 character from 154 * the first byte of a UTF-8 string. 155 * Bytes which are illegal when used as the first byte have a 1. 156 * The NUL byte has length 1. 157 */ 158 static char utf8len_tab[256] = 159 { 160 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 161 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 162 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 163 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 164 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 165 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 166 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, 167 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6,1,1, 168 }; 169 170 /* 171 * Like utf8len_tab above, but using a zero for illegal lead bytes. 172 */ 173 static char utf8len_tab_zero[256] = 174 { 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 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 178 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 179 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, 180 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, 181 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, 182 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, 183 }; 184 185 /* 186 * XIM often causes trouble. Define XIM_DEBUG to get a log of XIM callbacks 187 * in the "xim.log" file. 188 */ 189 /* #define XIM_DEBUG */ 190 #ifdef XIM_DEBUG 191 static void 192 xim_log(char *s, ...) 193 { 194 va_list arglist; 195 static FILE *fd = NULL; 196 197 if (fd == (FILE *)-1) 198 return; 199 if (fd == NULL) 200 { 201 fd = mch_fopen("xim.log", "w"); 202 if (fd == NULL) 203 { 204 EMSG("Cannot open xim.log"); 205 fd = (FILE *)-1; 206 return; 207 } 208 } 209 210 va_start(arglist, s); 211 vfprintf(fd, s, arglist); 212 va_end(arglist); 213 } 214 #endif 215 216 #endif 217 218 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 219 /* 220 * Canonical encoding names and their properties. 221 * "iso-8859-n" is handled by enc_canonize() directly. 222 */ 223 static struct 224 { char *name; int prop; int codepage;} 225 enc_canon_table[] = 226 { 227 #define IDX_LATIN_1 0 228 {"latin1", ENC_8BIT + ENC_LATIN1, 1252}, 229 #define IDX_ISO_2 1 230 {"iso-8859-2", ENC_8BIT, 0}, 231 #define IDX_ISO_3 2 232 {"iso-8859-3", ENC_8BIT, 0}, 233 #define IDX_ISO_4 3 234 {"iso-8859-4", ENC_8BIT, 0}, 235 #define IDX_ISO_5 4 236 {"iso-8859-5", ENC_8BIT, 0}, 237 #define IDX_ISO_6 5 238 {"iso-8859-6", ENC_8BIT, 0}, 239 #define IDX_ISO_7 6 240 {"iso-8859-7", ENC_8BIT, 0}, 241 #define IDX_ISO_8 7 242 {"iso-8859-8", ENC_8BIT, 0}, 243 #define IDX_ISO_9 8 244 {"iso-8859-9", ENC_8BIT, 0}, 245 #define IDX_ISO_10 9 246 {"iso-8859-10", ENC_8BIT, 0}, 247 #define IDX_ISO_11 10 248 {"iso-8859-11", ENC_8BIT, 0}, 249 #define IDX_ISO_13 11 250 {"iso-8859-13", ENC_8BIT, 0}, 251 #define IDX_ISO_14 12 252 {"iso-8859-14", ENC_8BIT, 0}, 253 #define IDX_ISO_15 13 254 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0}, 255 #define IDX_KOI8_R 14 256 {"koi8-r", ENC_8BIT, 0}, 257 #define IDX_KOI8_U 15 258 {"koi8-u", ENC_8BIT, 0}, 259 #define IDX_UTF8 16 260 {"utf-8", ENC_UNICODE, 0}, 261 #define IDX_UCS2 17 262 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0}, 263 #define IDX_UCS2LE 18 264 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0}, 265 #define IDX_UTF16 19 266 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0}, 267 #define IDX_UTF16LE 20 268 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0}, 269 #define IDX_UCS4 21 270 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0}, 271 #define IDX_UCS4LE 22 272 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0}, 273 274 /* For debugging DBCS encoding on Unix. */ 275 #define IDX_DEBUG 23 276 {"debug", ENC_DBCS, DBCS_DEBUG}, 277 #define IDX_EUC_JP 24 278 {"euc-jp", ENC_DBCS, DBCS_JPNU}, 279 #define IDX_SJIS 25 280 {"sjis", ENC_DBCS, DBCS_JPN}, 281 #define IDX_EUC_KR 26 282 {"euc-kr", ENC_DBCS, DBCS_KORU}, 283 #define IDX_EUC_CN 27 284 {"euc-cn", ENC_DBCS, DBCS_CHSU}, 285 #define IDX_EUC_TW 28 286 {"euc-tw", ENC_DBCS, DBCS_CHTU}, 287 #define IDX_BIG5 29 288 {"big5", ENC_DBCS, DBCS_CHT}, 289 290 /* MS-DOS and MS-Windows codepages are included here, so that they can be 291 * used on Unix too. Most of them are similar to ISO-8859 encodings, but 292 * not exactly the same. */ 293 #define IDX_CP437 30 294 {"cp437", ENC_8BIT, 437}, /* like iso-8859-1 */ 295 #define IDX_CP737 31 296 {"cp737", ENC_8BIT, 737}, /* like iso-8859-7 */ 297 #define IDX_CP775 32 298 {"cp775", ENC_8BIT, 775}, /* Baltic */ 299 #define IDX_CP850 33 300 {"cp850", ENC_8BIT, 850}, /* like iso-8859-4 */ 301 #define IDX_CP852 34 302 {"cp852", ENC_8BIT, 852}, /* like iso-8859-1 */ 303 #define IDX_CP855 35 304 {"cp855", ENC_8BIT, 855}, /* like iso-8859-2 */ 305 #define IDX_CP857 36 306 {"cp857", ENC_8BIT, 857}, /* like iso-8859-5 */ 307 #define IDX_CP860 37 308 {"cp860", ENC_8BIT, 860}, /* like iso-8859-9 */ 309 #define IDX_CP861 38 310 {"cp861", ENC_8BIT, 861}, /* like iso-8859-1 */ 311 #define IDX_CP862 39 312 {"cp862", ENC_8BIT, 862}, /* like iso-8859-1 */ 313 #define IDX_CP863 40 314 {"cp863", ENC_8BIT, 863}, /* like iso-8859-8 */ 315 #define IDX_CP865 41 316 {"cp865", ENC_8BIT, 865}, /* like iso-8859-1 */ 317 #define IDX_CP866 42 318 {"cp866", ENC_8BIT, 866}, /* like iso-8859-5 */ 319 #define IDX_CP869 43 320 {"cp869", ENC_8BIT, 869}, /* like iso-8859-7 */ 321 #define IDX_CP874 44 322 {"cp874", ENC_8BIT, 874}, /* Thai */ 323 #define IDX_CP932 45 324 {"cp932", ENC_DBCS, DBCS_JPN}, 325 #define IDX_CP936 46 326 {"cp936", ENC_DBCS, DBCS_CHS}, 327 #define IDX_CP949 47 328 {"cp949", ENC_DBCS, DBCS_KOR}, 329 #define IDX_CP950 48 330 {"cp950", ENC_DBCS, DBCS_CHT}, 331 #define IDX_CP1250 49 332 {"cp1250", ENC_8BIT, 1250}, /* Czech, Polish, etc. */ 333 #define IDX_CP1251 50 334 {"cp1251", ENC_8BIT, 1251}, /* Cyrillic */ 335 /* cp1252 is considered to be equal to latin1 */ 336 #define IDX_CP1253 51 337 {"cp1253", ENC_8BIT, 1253}, /* Greek */ 338 #define IDX_CP1254 52 339 {"cp1254", ENC_8BIT, 1254}, /* Turkish */ 340 #define IDX_CP1255 53 341 {"cp1255", ENC_8BIT, 1255}, /* Hebrew */ 342 #define IDX_CP1256 54 343 {"cp1256", ENC_8BIT, 1256}, /* Arabic */ 344 #define IDX_CP1257 55 345 {"cp1257", ENC_8BIT, 1257}, /* Baltic */ 346 #define IDX_CP1258 56 347 {"cp1258", ENC_8BIT, 1258}, /* Vietnamese */ 348 349 #define IDX_MACROMAN 57 350 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, /* Mac OS */ 351 #define IDX_DECMCS 58 352 {"dec-mcs", ENC_8BIT, 0}, /* DEC MCS */ 353 #define IDX_HPROMAN8 59 354 {"hp-roman8", ENC_8BIT, 0}, /* HP Roman8 */ 355 #define IDX_COUNT 60 356 }; 357 358 /* 359 * Aliases for encoding names. 360 */ 361 static struct 362 { char *name; int canon;} 363 enc_alias_table[] = 364 { 365 {"ansi", IDX_LATIN_1}, 366 {"iso-8859-1", IDX_LATIN_1}, 367 {"latin2", IDX_ISO_2}, 368 {"latin3", IDX_ISO_3}, 369 {"latin4", IDX_ISO_4}, 370 {"cyrillic", IDX_ISO_5}, 371 {"arabic", IDX_ISO_6}, 372 {"greek", IDX_ISO_7}, 373 #ifdef WIN3264 374 {"hebrew", IDX_CP1255}, 375 #else 376 {"hebrew", IDX_ISO_8}, 377 #endif 378 {"latin5", IDX_ISO_9}, 379 {"turkish", IDX_ISO_9}, /* ? */ 380 {"latin6", IDX_ISO_10}, 381 {"nordic", IDX_ISO_10}, /* ? */ 382 {"thai", IDX_ISO_11}, /* ? */ 383 {"latin7", IDX_ISO_13}, 384 {"latin8", IDX_ISO_14}, 385 {"latin9", IDX_ISO_15}, 386 {"utf8", IDX_UTF8}, 387 {"unicode", IDX_UCS2}, 388 {"ucs2", IDX_UCS2}, 389 {"ucs2be", IDX_UCS2}, 390 {"ucs-2be", IDX_UCS2}, 391 {"ucs2le", IDX_UCS2LE}, 392 {"utf16", IDX_UTF16}, 393 {"utf16be", IDX_UTF16}, 394 {"utf-16be", IDX_UTF16}, 395 {"utf16le", IDX_UTF16LE}, 396 {"ucs4", IDX_UCS4}, 397 {"ucs4be", IDX_UCS4}, 398 {"ucs-4be", IDX_UCS4}, 399 {"ucs4le", IDX_UCS4LE}, 400 {"utf32", IDX_UCS4}, 401 {"utf-32", IDX_UCS4}, 402 {"utf32be", IDX_UCS4}, 403 {"utf-32be", IDX_UCS4}, 404 {"utf32le", IDX_UCS4LE}, 405 {"utf-32le", IDX_UCS4LE}, 406 {"932", IDX_CP932}, 407 {"949", IDX_CP949}, 408 {"936", IDX_CP936}, 409 {"gbk", IDX_CP936}, 410 {"950", IDX_CP950}, 411 {"eucjp", IDX_EUC_JP}, 412 {"unix-jis", IDX_EUC_JP}, 413 {"ujis", IDX_EUC_JP}, 414 {"shift-jis", IDX_SJIS}, 415 {"pck", IDX_SJIS}, /* Sun: PCK */ 416 {"euckr", IDX_EUC_KR}, 417 {"5601", IDX_EUC_KR}, /* Sun: KS C 5601 */ 418 {"euccn", IDX_EUC_CN}, 419 {"gb2312", IDX_EUC_CN}, 420 {"euctw", IDX_EUC_TW}, 421 #if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS_X) 422 {"japan", IDX_CP932}, 423 {"korea", IDX_CP949}, 424 {"prc", IDX_CP936}, 425 {"chinese", IDX_CP936}, 426 {"taiwan", IDX_CP950}, 427 {"big5", IDX_CP950}, 428 #else 429 {"japan", IDX_EUC_JP}, 430 {"korea", IDX_EUC_KR}, 431 {"prc", IDX_EUC_CN}, 432 {"chinese", IDX_EUC_CN}, 433 {"taiwan", IDX_EUC_TW}, 434 {"cp950", IDX_BIG5}, 435 {"950", IDX_BIG5}, 436 #endif 437 {"mac", IDX_MACROMAN}, 438 {"mac-roman", IDX_MACROMAN}, 439 {NULL, 0} 440 }; 441 442 #ifndef CP_UTF8 443 # define CP_UTF8 65001 /* magic number from winnls.h */ 444 #endif 445 446 /* 447 * Find encoding "name" in the list of canonical encoding names. 448 * Returns -1 if not found. 449 */ 450 static int 451 enc_canon_search(char_u *name) 452 { 453 int i; 454 455 for (i = 0; i < IDX_COUNT; ++i) 456 if (STRCMP(name, enc_canon_table[i].name) == 0) 457 return i; 458 return -1; 459 } 460 461 #endif 462 463 #if defined(FEAT_MBYTE) || defined(PROTO) 464 465 /* 466 * Find canonical encoding "name" in the list and return its properties. 467 * Returns 0 if not found. 468 */ 469 int 470 enc_canon_props(char_u *name) 471 { 472 int i; 473 474 i = enc_canon_search(name); 475 if (i >= 0) 476 return enc_canon_table[i].prop; 477 #ifdef WIN3264 478 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2])) 479 { 480 CPINFO cpinfo; 481 482 /* Get info on this codepage to find out what it is. */ 483 if (GetCPInfo(atoi((char *)name + 2), &cpinfo) != 0) 484 { 485 if (cpinfo.MaxCharSize == 1) /* some single-byte encoding */ 486 return ENC_8BIT; 487 if (cpinfo.MaxCharSize == 2 488 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 489 /* must be a DBCS encoding */ 490 return ENC_DBCS; 491 } 492 return 0; 493 } 494 #endif 495 if (STRNCMP(name, "2byte-", 6) == 0) 496 return ENC_DBCS; 497 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0) 498 return ENC_8BIT; 499 return 0; 500 } 501 502 /* 503 * Set up for using multi-byte characters. 504 * Called in three cases: 505 * - by main() to initialize (p_enc == NULL) 506 * - by set_init_1() after 'encoding' was set to its default. 507 * - by do_set() when 'encoding' has been set. 508 * p_enc must have been passed through enc_canonize() already. 509 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags. 510 * Fills mb_bytelen_tab[] and returns NULL when there are no problems. 511 * When there is something wrong: Returns an error message and doesn't change 512 * anything. 513 */ 514 char_u * 515 mb_init(void) 516 { 517 int i; 518 int idx; 519 int n; 520 int enc_dbcs_new = 0; 521 #if defined(USE_ICONV) && !defined(WIN3264) && !defined(WIN32UNIX) \ 522 && !defined(MACOS_CONVERT) 523 # define LEN_FROM_CONV 524 vimconv_T vimconv; 525 char_u *p; 526 #endif 527 528 if (p_enc == NULL) 529 { 530 /* Just starting up: set the whole table to one's. */ 531 for (i = 0; i < 256; ++i) 532 mb_bytelen_tab[i] = 1; 533 input_conv.vc_type = CONV_NONE; 534 input_conv.vc_factor = 1; 535 output_conv.vc_type = CONV_NONE; 536 return NULL; 537 } 538 539 #ifdef WIN3264 540 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2])) 541 { 542 CPINFO cpinfo; 543 544 /* Get info on this codepage to find out what it is. */ 545 if (GetCPInfo(atoi((char *)p_enc + 2), &cpinfo) != 0) 546 { 547 if (cpinfo.MaxCharSize == 1) 548 { 549 /* some single-byte encoding */ 550 enc_unicode = 0; 551 enc_utf8 = FALSE; 552 } 553 else if (cpinfo.MaxCharSize == 2 554 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 555 { 556 /* must be a DBCS encoding, check below */ 557 enc_dbcs_new = atoi((char *)p_enc + 2); 558 } 559 else 560 goto codepage_invalid; 561 } 562 else if (GetLastError() == ERROR_INVALID_PARAMETER) 563 { 564 codepage_invalid: 565 return (char_u *)N_("E543: Not a valid codepage"); 566 } 567 } 568 #endif 569 else if (STRNCMP(p_enc, "8bit-", 5) == 0 570 || STRNCMP(p_enc, "iso-8859-", 9) == 0) 571 { 572 /* Accept any "8bit-" or "iso-8859-" name. */ 573 enc_unicode = 0; 574 enc_utf8 = FALSE; 575 } 576 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 577 { 578 #ifdef WIN3264 579 /* Windows: accept only valid codepage numbers, check below. */ 580 if (p_enc[6] != 'c' || p_enc[7] != 'p' 581 || (enc_dbcs_new = atoi((char *)p_enc + 8)) == 0) 582 return e_invarg; 583 #else 584 /* Unix: accept any "2byte-" name, assume current locale. */ 585 enc_dbcs_new = DBCS_2BYTE; 586 #endif 587 } 588 else if ((idx = enc_canon_search(p_enc)) >= 0) 589 { 590 i = enc_canon_table[idx].prop; 591 if (i & ENC_UNICODE) 592 { 593 /* Unicode */ 594 enc_utf8 = TRUE; 595 if (i & (ENC_2BYTE | ENC_2WORD)) 596 enc_unicode = 2; 597 else if (i & ENC_4BYTE) 598 enc_unicode = 4; 599 else 600 enc_unicode = 0; 601 } 602 else if (i & ENC_DBCS) 603 { 604 /* 2byte, handle below */ 605 enc_dbcs_new = enc_canon_table[idx].codepage; 606 } 607 else 608 { 609 /* Must be 8-bit. */ 610 enc_unicode = 0; 611 enc_utf8 = FALSE; 612 } 613 } 614 else /* Don't know what encoding this is, reject it. */ 615 return e_invarg; 616 617 if (enc_dbcs_new != 0) 618 { 619 #ifdef WIN3264 620 /* Check if the DBCS code page is OK. */ 621 if (!IsValidCodePage(enc_dbcs_new)) 622 goto codepage_invalid; 623 #endif 624 enc_unicode = 0; 625 enc_utf8 = FALSE; 626 } 627 enc_dbcs = enc_dbcs_new; 628 has_mbyte = (enc_dbcs != 0 || enc_utf8); 629 630 #if defined(WIN3264) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 631 enc_codepage = encname2codepage(p_enc); 632 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0); 633 #endif 634 635 /* Detect an encoding that uses latin1 characters. */ 636 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0 637 || STRCMP(p_enc, "iso-8859-15") == 0); 638 639 /* 640 * Set the function pointers. 641 */ 642 if (enc_utf8) 643 { 644 mb_ptr2len = utfc_ptr2len; 645 mb_ptr2len_len = utfc_ptr2len_len; 646 mb_char2len = utf_char2len; 647 mb_char2bytes = utf_char2bytes; 648 mb_ptr2cells = utf_ptr2cells; 649 mb_ptr2cells_len = utf_ptr2cells_len; 650 mb_char2cells = utf_char2cells; 651 mb_off2cells = utf_off2cells; 652 mb_ptr2char = utf_ptr2char; 653 mb_head_off = utf_head_off; 654 } 655 else if (enc_dbcs != 0) 656 { 657 mb_ptr2len = dbcs_ptr2len; 658 mb_ptr2len_len = dbcs_ptr2len_len; 659 mb_char2len = dbcs_char2len; 660 mb_char2bytes = dbcs_char2bytes; 661 mb_ptr2cells = dbcs_ptr2cells; 662 mb_ptr2cells_len = dbcs_ptr2cells_len; 663 mb_char2cells = dbcs_char2cells; 664 mb_off2cells = dbcs_off2cells; 665 mb_ptr2char = dbcs_ptr2char; 666 mb_head_off = dbcs_head_off; 667 } 668 else 669 { 670 mb_ptr2len = latin_ptr2len; 671 mb_ptr2len_len = latin_ptr2len_len; 672 mb_char2len = latin_char2len; 673 mb_char2bytes = latin_char2bytes; 674 mb_ptr2cells = latin_ptr2cells; 675 mb_ptr2cells_len = latin_ptr2cells_len; 676 mb_char2cells = latin_char2cells; 677 mb_off2cells = latin_off2cells; 678 mb_ptr2char = latin_ptr2char; 679 mb_head_off = latin_head_off; 680 } 681 682 /* 683 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN(). 684 */ 685 #ifdef LEN_FROM_CONV 686 /* When 'encoding' is different from the current locale mblen() won't 687 * work. Use conversion to "utf-8" instead. */ 688 vimconv.vc_type = CONV_NONE; 689 if (enc_dbcs) 690 { 691 p = enc_locale(); 692 if (p == NULL || STRCMP(p, p_enc) != 0) 693 { 694 convert_setup(&vimconv, p_enc, (char_u *)"utf-8"); 695 vimconv.vc_fail = TRUE; 696 } 697 vim_free(p); 698 } 699 #endif 700 701 for (i = 0; i < 256; ++i) 702 { 703 /* Our own function to reliably check the length of UTF-8 characters, 704 * independent of mblen(). */ 705 if (enc_utf8) 706 n = utf8len_tab[i]; 707 else if (enc_dbcs == 0) 708 n = 1; 709 else 710 { 711 #if defined(WIN3264) || defined(WIN32UNIX) 712 /* enc_dbcs is set by setting 'fileencoding'. It becomes a Windows 713 * CodePage identifier, which we can pass directly in to Windows 714 * API */ 715 n = IsDBCSLeadByteEx(enc_dbcs, (WINBYTE)i) ? 2 : 1; 716 #else 717 # if defined(__amigaos4__) || defined(__ANDROID__) || \ 718 !(defined(HAVE_MBLEN) || defined(X_LOCALE)) 719 /* 720 * if mblen() is not available, character which MSB is turned on 721 * are treated as leading byte character. (note : This assumption 722 * is not always true.) 723 */ 724 n = (i & 0x80) ? 2 : 1; 725 # else 726 char buf[MB_MAXBYTES + 1]; 727 728 if (i == NUL) /* just in case mblen() can't handle "" */ 729 n = 1; 730 else 731 { 732 buf[0] = i; 733 buf[1] = 0; 734 # ifdef LEN_FROM_CONV 735 if (vimconv.vc_type != CONV_NONE) 736 { 737 /* 738 * string_convert() should fail when converting the first 739 * byte of a double-byte character. 740 */ 741 p = string_convert(&vimconv, (char_u *)buf, NULL); 742 if (p != NULL) 743 { 744 vim_free(p); 745 n = 1; 746 } 747 else 748 n = 2; 749 } 750 else 751 # endif 752 { 753 /* 754 * mblen() should return -1 for invalid (means the leading 755 * multibyte) character. However there are some platforms 756 * where mblen() returns 0 for invalid character. 757 * Therefore, following condition includes 0. 758 */ 759 ignored = mblen(NULL, 0); /* First reset the state. */ 760 if (mblen(buf, (size_t)1) <= 0) 761 n = 2; 762 else 763 n = 1; 764 } 765 } 766 # endif 767 #endif 768 } 769 770 mb_bytelen_tab[i] = n; 771 } 772 773 #ifdef LEN_FROM_CONV 774 convert_setup(&vimconv, NULL, NULL); 775 #endif 776 777 /* The cell width depends on the type of multi-byte characters. */ 778 (void)init_chartab(); 779 780 /* When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] */ 781 screenalloc(FALSE); 782 783 /* When using Unicode, set default for 'fileencodings'. */ 784 if (enc_utf8 && !option_was_set((char_u *)"fencs")) 785 set_string_option_direct((char_u *)"fencs", -1, 786 (char_u *)"ucs-bom,utf-8,default,latin1", OPT_FREE, 0); 787 788 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT) 789 /* GNU gettext 0.10.37 supports this feature: set the codeset used for 790 * translated messages independently from the current locale. */ 791 (void)bind_textdomain_codeset(VIMPACKAGE, 792 enc_utf8 ? "utf-8" : (char *)p_enc); 793 #endif 794 795 #ifdef WIN32 796 /* When changing 'encoding' while starting up, then convert the command 797 * line arguments from the active codepage to 'encoding'. */ 798 if (starting != 0) 799 fix_arg_enc(); 800 #endif 801 802 /* Fire an autocommand to let people do custom font setup. This must be 803 * after Vim has been setup for the new encoding. */ 804 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); 805 806 #ifdef FEAT_SPELL 807 /* Need to reload spell dictionaries */ 808 spell_reload(); 809 #endif 810 811 return NULL; 812 } 813 814 /* 815 * Return the size of the BOM for the current buffer: 816 * 0 - no BOM 817 * 2 - UCS-2 or UTF-16 BOM 818 * 4 - UCS-4 BOM 819 * 3 - UTF-8 BOM 820 */ 821 int 822 bomb_size(void) 823 { 824 int n = 0; 825 826 if (curbuf->b_p_bomb && !curbuf->b_p_bin) 827 { 828 if (*curbuf->b_p_fenc == NUL) 829 { 830 if (enc_utf8) 831 { 832 if (enc_unicode != 0) 833 n = enc_unicode; 834 else 835 n = 3; 836 } 837 } 838 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) 839 n = 3; 840 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 841 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) 842 n = 2; 843 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) 844 n = 4; 845 } 846 return n; 847 } 848 849 /* 850 * Remove all BOM from "s" by moving remaining text. 851 */ 852 void 853 remove_bom(char_u *s) 854 { 855 if (enc_utf8) 856 { 857 char_u *p = s; 858 859 while ((p = vim_strbyte(p, 0xef)) != NULL) 860 { 861 if (p[1] == 0xbb && p[2] == 0xbf) 862 STRMOVE(p, p + 3); 863 else 864 ++p; 865 } 866 } 867 } 868 869 /* 870 * Get class of pointer: 871 * 0 for blank or NUL 872 * 1 for punctuation 873 * 2 for an (ASCII) word character 874 * >2 for other word characters 875 */ 876 int 877 mb_get_class(char_u *p) 878 { 879 return mb_get_class_buf(p, curbuf); 880 } 881 882 int 883 mb_get_class_buf(char_u *p, buf_T *buf) 884 { 885 if (MB_BYTE2LEN(p[0]) == 1) 886 { 887 if (p[0] == NUL || VIM_ISWHITE(p[0])) 888 return 0; 889 if (vim_iswordc_buf(p[0], buf)) 890 return 2; 891 return 1; 892 } 893 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) 894 return dbcs_class(p[0], p[1]); 895 if (enc_utf8) 896 return utf_class_buf(utf_ptr2char(p), buf); 897 return 0; 898 } 899 900 /* 901 * Get class of a double-byte character. This always returns 3 or bigger. 902 * TODO: Should return 1 for punctuation. 903 */ 904 int 905 dbcs_class(unsigned lead, unsigned trail) 906 { 907 switch (enc_dbcs) 908 { 909 /* please add classify routine for your language in here */ 910 911 case DBCS_JPNU: /* ? */ 912 case DBCS_JPN: 913 { 914 /* JIS code classification */ 915 unsigned char lb = lead; 916 unsigned char tb = trail; 917 918 /* convert process code to JIS */ 919 # if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS_X) 920 /* process code is SJIS */ 921 if (lb <= 0x9f) 922 lb = (lb - 0x81) * 2 + 0x21; 923 else 924 lb = (lb - 0xc1) * 2 + 0x21; 925 if (tb <= 0x7e) 926 tb -= 0x1f; 927 else if (tb <= 0x9e) 928 tb -= 0x20; 929 else 930 { 931 tb -= 0x7e; 932 lb += 1; 933 } 934 # else 935 /* 936 * XXX: Code page identification can not use with all 937 * system! So, some other encoding information 938 * will be needed. 939 * In japanese: SJIS,EUC,UNICODE,(JIS) 940 * Note that JIS-code system don't use as 941 * process code in most system because it uses 942 * escape sequences(JIS is context depend encoding). 943 */ 944 /* assume process code is JAPANESE-EUC */ 945 lb &= 0x7f; 946 tb &= 0x7f; 947 # endif 948 /* exceptions */ 949 switch (lb << 8 | tb) 950 { 951 case 0x2121: /* ZENKAKU space */ 952 return 0; 953 case 0x2122: /* TOU-TEN (Japanese comma) */ 954 case 0x2123: /* KU-TEN (Japanese period) */ 955 case 0x2124: /* ZENKAKU comma */ 956 case 0x2125: /* ZENKAKU period */ 957 return 1; 958 case 0x213c: /* prolongedsound handled as KATAKANA */ 959 return 13; 960 } 961 /* sieved by KU code */ 962 switch (lb) 963 { 964 case 0x21: 965 case 0x22: 966 /* special symbols */ 967 return 10; 968 case 0x23: 969 /* alpha-numeric */ 970 return 11; 971 case 0x24: 972 /* hiragana */ 973 return 12; 974 case 0x25: 975 /* katakana */ 976 return 13; 977 case 0x26: 978 /* greek */ 979 return 14; 980 case 0x27: 981 /* russian */ 982 return 15; 983 case 0x28: 984 /* lines */ 985 return 16; 986 default: 987 /* kanji */ 988 return 17; 989 } 990 } 991 992 case DBCS_KORU: /* ? */ 993 case DBCS_KOR: 994 { 995 /* KS code classification */ 996 unsigned char c1 = lead; 997 unsigned char c2 = trail; 998 999 /* 1000 * 20 : Hangul 1001 * 21 : Hanja 1002 * 22 : Symbols 1003 * 23 : Alpha-numeric/Roman Letter (Full width) 1004 * 24 : Hangul Letter(Alphabet) 1005 * 25 : Roman Numeral/Greek Letter 1006 * 26 : Box Drawings 1007 * 27 : Unit Symbols 1008 * 28 : Circled/Parenthesized Letter 1009 * 29 : Hiragana/Katakana 1010 * 30 : Cyrillic Letter 1011 */ 1012 1013 if (c1 >= 0xB0 && c1 <= 0xC8) 1014 /* Hangul */ 1015 return 20; 1016 #if defined(WIN3264) || defined(WIN32UNIX) 1017 else if (c1 <= 0xA0 || c2 <= 0xA0) 1018 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */ 1019 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE 1020 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 1021 */ 1022 return 20; 1023 #endif 1024 1025 else if (c1 >= 0xCA && c1 <= 0xFD) 1026 /* Hanja */ 1027 return 21; 1028 else switch (c1) 1029 { 1030 case 0xA1: 1031 case 0xA2: 1032 /* Symbols */ 1033 return 22; 1034 case 0xA3: 1035 /* Alpha-numeric */ 1036 return 23; 1037 case 0xA4: 1038 /* Hangul Letter(Alphabet) */ 1039 return 24; 1040 case 0xA5: 1041 /* Roman Numeral/Greek Letter */ 1042 return 25; 1043 case 0xA6: 1044 /* Box Drawings */ 1045 return 26; 1046 case 0xA7: 1047 /* Unit Symbols */ 1048 return 27; 1049 case 0xA8: 1050 case 0xA9: 1051 if (c2 <= 0xAF) 1052 return 25; /* Roman Letter */ 1053 else if (c2 >= 0xF6) 1054 return 22; /* Symbols */ 1055 else 1056 /* Circled/Parenthesized Letter */ 1057 return 28; 1058 case 0xAA: 1059 case 0xAB: 1060 /* Hiragana/Katakana */ 1061 return 29; 1062 case 0xAC: 1063 /* Cyrillic Letter */ 1064 return 30; 1065 } 1066 } 1067 default: 1068 break; 1069 } 1070 return 3; 1071 } 1072 1073 /* 1074 * mb_char2len() function pointer. 1075 * Return length in bytes of character "c". 1076 * Returns 1 for a single-byte character. 1077 */ 1078 int 1079 latin_char2len(int c UNUSED) 1080 { 1081 return 1; 1082 } 1083 1084 static int 1085 dbcs_char2len( 1086 int c) 1087 { 1088 if (c >= 0x100) 1089 return 2; 1090 return 1; 1091 } 1092 1093 /* 1094 * mb_char2bytes() function pointer. 1095 * Convert a character to its bytes. 1096 * Returns the length in bytes. 1097 */ 1098 int 1099 latin_char2bytes(int c, char_u *buf) 1100 { 1101 buf[0] = c; 1102 return 1; 1103 } 1104 1105 static int 1106 dbcs_char2bytes(int c, char_u *buf) 1107 { 1108 if (c >= 0x100) 1109 { 1110 buf[0] = (unsigned)c >> 8; 1111 buf[1] = c; 1112 /* Never use a NUL byte, it causes lots of trouble. It's an invalid 1113 * character anyway. */ 1114 if (buf[1] == NUL) 1115 buf[1] = '\n'; 1116 return 2; 1117 } 1118 buf[0] = c; 1119 return 1; 1120 } 1121 1122 /* 1123 * mb_ptr2len() function pointer. 1124 * Get byte length of character at "*p" but stop at a NUL. 1125 * For UTF-8 this includes following composing characters. 1126 * Returns 0 when *p is NUL. 1127 */ 1128 int 1129 latin_ptr2len(char_u *p) 1130 { 1131 return MB_BYTE2LEN(*p); 1132 } 1133 1134 static int 1135 dbcs_ptr2len( 1136 char_u *p) 1137 { 1138 int len; 1139 1140 /* Check if second byte is not missing. */ 1141 len = MB_BYTE2LEN(*p); 1142 if (len == 2 && p[1] == NUL) 1143 len = 1; 1144 return len; 1145 } 1146 1147 /* 1148 * mb_ptr2len_len() function pointer. 1149 * Like mb_ptr2len(), but limit to read "size" bytes. 1150 * Returns 0 for an empty string. 1151 * Returns 1 for an illegal char or an incomplete byte sequence. 1152 */ 1153 int 1154 latin_ptr2len_len(char_u *p, int size) 1155 { 1156 if (size < 1 || *p == NUL) 1157 return 0; 1158 return 1; 1159 } 1160 1161 static int 1162 dbcs_ptr2len_len(char_u *p, int size) 1163 { 1164 int len; 1165 1166 if (size < 1 || *p == NUL) 1167 return 0; 1168 if (size == 1) 1169 return 1; 1170 /* Check that second byte is not missing. */ 1171 len = MB_BYTE2LEN(*p); 1172 if (len == 2 && p[1] == NUL) 1173 len = 1; 1174 return len; 1175 } 1176 1177 struct interval 1178 { 1179 long first; 1180 long last; 1181 }; 1182 1183 /* 1184 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]". 1185 */ 1186 static int 1187 intable(struct interval *table, size_t size, int c) 1188 { 1189 int mid, bot, top; 1190 1191 /* first quick check for Latin1 etc. characters */ 1192 if (c < table[0].first) 1193 return FALSE; 1194 1195 /* binary search in table */ 1196 bot = 0; 1197 top = (int)(size / sizeof(struct interval) - 1); 1198 while (top >= bot) 1199 { 1200 mid = (bot + top) / 2; 1201 if (table[mid].last < c) 1202 bot = mid + 1; 1203 else if (table[mid].first > c) 1204 top = mid - 1; 1205 else 1206 return TRUE; 1207 } 1208 return FALSE; 1209 } 1210 1211 /* Sorted list of non-overlapping intervals of East Asian Ambiguous 1212 * characters, generated with ../runtime/tools/unicode.vim. */ 1213 static struct interval ambiguous[] = 1214 { 1215 {0x00a1, 0x00a1}, 1216 {0x00a4, 0x00a4}, 1217 {0x00a7, 0x00a8}, 1218 {0x00aa, 0x00aa}, 1219 {0x00ad, 0x00ae}, 1220 {0x00b0, 0x00b4}, 1221 {0x00b6, 0x00ba}, 1222 {0x00bc, 0x00bf}, 1223 {0x00c6, 0x00c6}, 1224 {0x00d0, 0x00d0}, 1225 {0x00d7, 0x00d8}, 1226 {0x00de, 0x00e1}, 1227 {0x00e6, 0x00e6}, 1228 {0x00e8, 0x00ea}, 1229 {0x00ec, 0x00ed}, 1230 {0x00f0, 0x00f0}, 1231 {0x00f2, 0x00f3}, 1232 {0x00f7, 0x00fa}, 1233 {0x00fc, 0x00fc}, 1234 {0x00fe, 0x00fe}, 1235 {0x0101, 0x0101}, 1236 {0x0111, 0x0111}, 1237 {0x0113, 0x0113}, 1238 {0x011b, 0x011b}, 1239 {0x0126, 0x0127}, 1240 {0x012b, 0x012b}, 1241 {0x0131, 0x0133}, 1242 {0x0138, 0x0138}, 1243 {0x013f, 0x0142}, 1244 {0x0144, 0x0144}, 1245 {0x0148, 0x014b}, 1246 {0x014d, 0x014d}, 1247 {0x0152, 0x0153}, 1248 {0x0166, 0x0167}, 1249 {0x016b, 0x016b}, 1250 {0x01ce, 0x01ce}, 1251 {0x01d0, 0x01d0}, 1252 {0x01d2, 0x01d2}, 1253 {0x01d4, 0x01d4}, 1254 {0x01d6, 0x01d6}, 1255 {0x01d8, 0x01d8}, 1256 {0x01da, 0x01da}, 1257 {0x01dc, 0x01dc}, 1258 {0x0251, 0x0251}, 1259 {0x0261, 0x0261}, 1260 {0x02c4, 0x02c4}, 1261 {0x02c7, 0x02c7}, 1262 {0x02c9, 0x02cb}, 1263 {0x02cd, 0x02cd}, 1264 {0x02d0, 0x02d0}, 1265 {0x02d8, 0x02db}, 1266 {0x02dd, 0x02dd}, 1267 {0x02df, 0x02df}, 1268 {0x0300, 0x036f}, 1269 {0x0391, 0x03a1}, 1270 {0x03a3, 0x03a9}, 1271 {0x03b1, 0x03c1}, 1272 {0x03c3, 0x03c9}, 1273 {0x0401, 0x0401}, 1274 {0x0410, 0x044f}, 1275 {0x0451, 0x0451}, 1276 {0x2010, 0x2010}, 1277 {0x2013, 0x2016}, 1278 {0x2018, 0x2019}, 1279 {0x201c, 0x201d}, 1280 {0x2020, 0x2022}, 1281 {0x2024, 0x2027}, 1282 {0x2030, 0x2030}, 1283 {0x2032, 0x2033}, 1284 {0x2035, 0x2035}, 1285 {0x203b, 0x203b}, 1286 {0x203e, 0x203e}, 1287 {0x2074, 0x2074}, 1288 {0x207f, 0x207f}, 1289 {0x2081, 0x2084}, 1290 {0x20ac, 0x20ac}, 1291 {0x2103, 0x2103}, 1292 {0x2105, 0x2105}, 1293 {0x2109, 0x2109}, 1294 {0x2113, 0x2113}, 1295 {0x2116, 0x2116}, 1296 {0x2121, 0x2122}, 1297 {0x2126, 0x2126}, 1298 {0x212b, 0x212b}, 1299 {0x2153, 0x2154}, 1300 {0x215b, 0x215e}, 1301 {0x2160, 0x216b}, 1302 {0x2170, 0x2179}, 1303 {0x2189, 0x2189}, 1304 {0x2190, 0x2199}, 1305 {0x21b8, 0x21b9}, 1306 {0x21d2, 0x21d2}, 1307 {0x21d4, 0x21d4}, 1308 {0x21e7, 0x21e7}, 1309 {0x2200, 0x2200}, 1310 {0x2202, 0x2203}, 1311 {0x2207, 0x2208}, 1312 {0x220b, 0x220b}, 1313 {0x220f, 0x220f}, 1314 {0x2211, 0x2211}, 1315 {0x2215, 0x2215}, 1316 {0x221a, 0x221a}, 1317 {0x221d, 0x2220}, 1318 {0x2223, 0x2223}, 1319 {0x2225, 0x2225}, 1320 {0x2227, 0x222c}, 1321 {0x222e, 0x222e}, 1322 {0x2234, 0x2237}, 1323 {0x223c, 0x223d}, 1324 {0x2248, 0x2248}, 1325 {0x224c, 0x224c}, 1326 {0x2252, 0x2252}, 1327 {0x2260, 0x2261}, 1328 {0x2264, 0x2267}, 1329 {0x226a, 0x226b}, 1330 {0x226e, 0x226f}, 1331 {0x2282, 0x2283}, 1332 {0x2286, 0x2287}, 1333 {0x2295, 0x2295}, 1334 {0x2299, 0x2299}, 1335 {0x22a5, 0x22a5}, 1336 {0x22bf, 0x22bf}, 1337 {0x2312, 0x2312}, 1338 {0x2460, 0x24e9}, 1339 {0x24eb, 0x254b}, 1340 {0x2550, 0x2573}, 1341 {0x2580, 0x258f}, 1342 {0x2592, 0x2595}, 1343 {0x25a0, 0x25a1}, 1344 {0x25a3, 0x25a9}, 1345 {0x25b2, 0x25b3}, 1346 {0x25b6, 0x25b7}, 1347 {0x25bc, 0x25bd}, 1348 {0x25c0, 0x25c1}, 1349 {0x25c6, 0x25c8}, 1350 {0x25cb, 0x25cb}, 1351 {0x25ce, 0x25d1}, 1352 {0x25e2, 0x25e5}, 1353 {0x25ef, 0x25ef}, 1354 {0x2605, 0x2606}, 1355 {0x2609, 0x2609}, 1356 {0x260e, 0x260f}, 1357 {0x261c, 0x261c}, 1358 {0x261e, 0x261e}, 1359 {0x2640, 0x2640}, 1360 {0x2642, 0x2642}, 1361 {0x2660, 0x2661}, 1362 {0x2663, 0x2665}, 1363 {0x2667, 0x266a}, 1364 {0x266c, 0x266d}, 1365 {0x266f, 0x266f}, 1366 {0x269e, 0x269f}, 1367 {0x26bf, 0x26bf}, 1368 {0x26c6, 0x26cd}, 1369 {0x26cf, 0x26d3}, 1370 {0x26d5, 0x26e1}, 1371 {0x26e3, 0x26e3}, 1372 {0x26e8, 0x26e9}, 1373 {0x26eb, 0x26f1}, 1374 {0x26f4, 0x26f4}, 1375 {0x26f6, 0x26f9}, 1376 {0x26fb, 0x26fc}, 1377 {0x26fe, 0x26ff}, 1378 {0x273d, 0x273d}, 1379 {0x2776, 0x277f}, 1380 {0x2b56, 0x2b59}, 1381 {0x3248, 0x324f}, 1382 {0xe000, 0xf8ff}, 1383 {0xfe00, 0xfe0f}, 1384 {0xfffd, 0xfffd}, 1385 {0x1f100, 0x1f10a}, 1386 {0x1f110, 0x1f12d}, 1387 {0x1f130, 0x1f169}, 1388 {0x1f170, 0x1f18d}, 1389 {0x1f18f, 0x1f190}, 1390 {0x1f19b, 0x1f1ac}, 1391 {0xe0100, 0xe01ef}, 1392 {0xf0000, 0xffffd}, 1393 {0x100000, 0x10fffd} 1394 }; 1395 1396 #if defined(FEAT_TERMINAL) || defined(PROTO) 1397 /* 1398 * utf_char2cells() with different argument type for libvterm. 1399 */ 1400 int 1401 utf_uint2cells(UINT32_T c) 1402 { 1403 if (c >= 0x100 && utf_iscomposing((int)c)) 1404 return 0; 1405 return utf_char2cells((int)c); 1406 } 1407 #endif 1408 1409 /* 1410 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1411 * Returns 4 or 6 for an unprintable character. 1412 * Is only correct for characters >= 0x80. 1413 * When p_ambw is "double", return 2 for a character with East Asian Width 1414 * class 'A'(mbiguous). 1415 */ 1416 int 1417 utf_char2cells(int c) 1418 { 1419 /* Sorted list of non-overlapping intervals of East Asian double width 1420 * characters, generated with ../runtime/tools/unicode.vim. */ 1421 static struct interval doublewidth[] = 1422 { 1423 {0x1100, 0x115f}, 1424 {0x231a, 0x231b}, 1425 {0x2329, 0x232a}, 1426 {0x23e9, 0x23ec}, 1427 {0x23f0, 0x23f0}, 1428 {0x23f3, 0x23f3}, 1429 {0x25fd, 0x25fe}, 1430 {0x2614, 0x2615}, 1431 {0x2648, 0x2653}, 1432 {0x267f, 0x267f}, 1433 {0x2693, 0x2693}, 1434 {0x26a1, 0x26a1}, 1435 {0x26aa, 0x26ab}, 1436 {0x26bd, 0x26be}, 1437 {0x26c4, 0x26c5}, 1438 {0x26ce, 0x26ce}, 1439 {0x26d4, 0x26d4}, 1440 {0x26ea, 0x26ea}, 1441 {0x26f2, 0x26f3}, 1442 {0x26f5, 0x26f5}, 1443 {0x26fa, 0x26fa}, 1444 {0x26fd, 0x26fd}, 1445 {0x2705, 0x2705}, 1446 {0x270a, 0x270b}, 1447 {0x2728, 0x2728}, 1448 {0x274c, 0x274c}, 1449 {0x274e, 0x274e}, 1450 {0x2753, 0x2755}, 1451 {0x2757, 0x2757}, 1452 {0x2795, 0x2797}, 1453 {0x27b0, 0x27b0}, 1454 {0x27bf, 0x27bf}, 1455 {0x2b1b, 0x2b1c}, 1456 {0x2b50, 0x2b50}, 1457 {0x2b55, 0x2b55}, 1458 {0x2e80, 0x2e99}, 1459 {0x2e9b, 0x2ef3}, 1460 {0x2f00, 0x2fd5}, 1461 {0x2ff0, 0x2ffb}, 1462 {0x3000, 0x303e}, 1463 {0x3041, 0x3096}, 1464 {0x3099, 0x30ff}, 1465 {0x3105, 0x312e}, 1466 {0x3131, 0x318e}, 1467 {0x3190, 0x31ba}, 1468 {0x31c0, 0x31e3}, 1469 {0x31f0, 0x321e}, 1470 {0x3220, 0x3247}, 1471 {0x3250, 0x32fe}, 1472 {0x3300, 0x4dbf}, 1473 {0x4e00, 0xa48c}, 1474 {0xa490, 0xa4c6}, 1475 {0xa960, 0xa97c}, 1476 {0xac00, 0xd7a3}, 1477 {0xf900, 0xfaff}, 1478 {0xfe10, 0xfe19}, 1479 {0xfe30, 0xfe52}, 1480 {0xfe54, 0xfe66}, 1481 {0xfe68, 0xfe6b}, 1482 {0xff01, 0xff60}, 1483 {0xffe0, 0xffe6}, 1484 {0x16fe0, 0x16fe1}, 1485 {0x17000, 0x187ec}, 1486 {0x18800, 0x18af2}, 1487 {0x1b000, 0x1b11e}, 1488 {0x1b170, 0x1b2fb}, 1489 {0x1f004, 0x1f004}, 1490 {0x1f0cf, 0x1f0cf}, 1491 {0x1f18e, 0x1f18e}, 1492 {0x1f191, 0x1f19a}, 1493 {0x1f200, 0x1f202}, 1494 {0x1f210, 0x1f23b}, 1495 {0x1f240, 0x1f248}, 1496 {0x1f250, 0x1f251}, 1497 {0x1f260, 0x1f265}, 1498 {0x1f300, 0x1f320}, 1499 {0x1f32d, 0x1f335}, 1500 {0x1f337, 0x1f37c}, 1501 {0x1f37e, 0x1f393}, 1502 {0x1f3a0, 0x1f3ca}, 1503 {0x1f3cf, 0x1f3d3}, 1504 {0x1f3e0, 0x1f3f0}, 1505 {0x1f3f4, 0x1f3f4}, 1506 {0x1f3f8, 0x1f43e}, 1507 {0x1f440, 0x1f440}, 1508 {0x1f442, 0x1f4fc}, 1509 {0x1f4ff, 0x1f53d}, 1510 {0x1f54b, 0x1f54e}, 1511 {0x1f550, 0x1f567}, 1512 {0x1f57a, 0x1f57a}, 1513 {0x1f595, 0x1f596}, 1514 {0x1f5a4, 0x1f5a4}, 1515 {0x1f5fb, 0x1f64f}, 1516 {0x1f680, 0x1f6c5}, 1517 {0x1f6cc, 0x1f6cc}, 1518 {0x1f6d0, 0x1f6d2}, 1519 {0x1f6eb, 0x1f6ec}, 1520 {0x1f6f4, 0x1f6f8}, 1521 {0x1f910, 0x1f93e}, 1522 {0x1f940, 0x1f94c}, 1523 {0x1f950, 0x1f96b}, 1524 {0x1f980, 0x1f997}, 1525 {0x1f9c0, 0x1f9c0}, 1526 {0x1f9d0, 0x1f9e6}, 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 {0x0816, 0x0819}, 2352 {0x081b, 0x0823}, 2353 {0x0825, 0x0827}, 2354 {0x0829, 0x082d}, 2355 {0x0859, 0x085b}, 2356 {0x08d4, 0x08e1}, 2357 {0x08e3, 0x0903}, 2358 {0x093a, 0x093c}, 2359 {0x093e, 0x094f}, 2360 {0x0951, 0x0957}, 2361 {0x0962, 0x0963}, 2362 {0x0981, 0x0983}, 2363 {0x09bc, 0x09bc}, 2364 {0x09be, 0x09c4}, 2365 {0x09c7, 0x09c8}, 2366 {0x09cb, 0x09cd}, 2367 {0x09d7, 0x09d7}, 2368 {0x09e2, 0x09e3}, 2369 {0x0a01, 0x0a03}, 2370 {0x0a3c, 0x0a3c}, 2371 {0x0a3e, 0x0a42}, 2372 {0x0a47, 0x0a48}, 2373 {0x0a4b, 0x0a4d}, 2374 {0x0a51, 0x0a51}, 2375 {0x0a70, 0x0a71}, 2376 {0x0a75, 0x0a75}, 2377 {0x0a81, 0x0a83}, 2378 {0x0abc, 0x0abc}, 2379 {0x0abe, 0x0ac5}, 2380 {0x0ac7, 0x0ac9}, 2381 {0x0acb, 0x0acd}, 2382 {0x0ae2, 0x0ae3}, 2383 {0x0afa, 0x0aff}, 2384 {0x0b01, 0x0b03}, 2385 {0x0b3c, 0x0b3c}, 2386 {0x0b3e, 0x0b44}, 2387 {0x0b47, 0x0b48}, 2388 {0x0b4b, 0x0b4d}, 2389 {0x0b56, 0x0b57}, 2390 {0x0b62, 0x0b63}, 2391 {0x0b82, 0x0b82}, 2392 {0x0bbe, 0x0bc2}, 2393 {0x0bc6, 0x0bc8}, 2394 {0x0bca, 0x0bcd}, 2395 {0x0bd7, 0x0bd7}, 2396 {0x0c00, 0x0c03}, 2397 {0x0c3e, 0x0c44}, 2398 {0x0c46, 0x0c48}, 2399 {0x0c4a, 0x0c4d}, 2400 {0x0c55, 0x0c56}, 2401 {0x0c62, 0x0c63}, 2402 {0x0c81, 0x0c83}, 2403 {0x0cbc, 0x0cbc}, 2404 {0x0cbe, 0x0cc4}, 2405 {0x0cc6, 0x0cc8}, 2406 {0x0cca, 0x0ccd}, 2407 {0x0cd5, 0x0cd6}, 2408 {0x0ce2, 0x0ce3}, 2409 {0x0d00, 0x0d03}, 2410 {0x0d3b, 0x0d3c}, 2411 {0x0d3e, 0x0d44}, 2412 {0x0d46, 0x0d48}, 2413 {0x0d4a, 0x0d4d}, 2414 {0x0d57, 0x0d57}, 2415 {0x0d62, 0x0d63}, 2416 {0x0d82, 0x0d83}, 2417 {0x0dca, 0x0dca}, 2418 {0x0dcf, 0x0dd4}, 2419 {0x0dd6, 0x0dd6}, 2420 {0x0dd8, 0x0ddf}, 2421 {0x0df2, 0x0df3}, 2422 {0x0e31, 0x0e31}, 2423 {0x0e34, 0x0e3a}, 2424 {0x0e47, 0x0e4e}, 2425 {0x0eb1, 0x0eb1}, 2426 {0x0eb4, 0x0eb9}, 2427 {0x0ebb, 0x0ebc}, 2428 {0x0ec8, 0x0ecd}, 2429 {0x0f18, 0x0f19}, 2430 {0x0f35, 0x0f35}, 2431 {0x0f37, 0x0f37}, 2432 {0x0f39, 0x0f39}, 2433 {0x0f3e, 0x0f3f}, 2434 {0x0f71, 0x0f84}, 2435 {0x0f86, 0x0f87}, 2436 {0x0f8d, 0x0f97}, 2437 {0x0f99, 0x0fbc}, 2438 {0x0fc6, 0x0fc6}, 2439 {0x102b, 0x103e}, 2440 {0x1056, 0x1059}, 2441 {0x105e, 0x1060}, 2442 {0x1062, 0x1064}, 2443 {0x1067, 0x106d}, 2444 {0x1071, 0x1074}, 2445 {0x1082, 0x108d}, 2446 {0x108f, 0x108f}, 2447 {0x109a, 0x109d}, 2448 {0x135d, 0x135f}, 2449 {0x1712, 0x1714}, 2450 {0x1732, 0x1734}, 2451 {0x1752, 0x1753}, 2452 {0x1772, 0x1773}, 2453 {0x17b4, 0x17d3}, 2454 {0x17dd, 0x17dd}, 2455 {0x180b, 0x180d}, 2456 {0x1885, 0x1886}, 2457 {0x18a9, 0x18a9}, 2458 {0x1920, 0x192b}, 2459 {0x1930, 0x193b}, 2460 {0x1a17, 0x1a1b}, 2461 {0x1a55, 0x1a5e}, 2462 {0x1a60, 0x1a7c}, 2463 {0x1a7f, 0x1a7f}, 2464 {0x1ab0, 0x1abe}, 2465 {0x1b00, 0x1b04}, 2466 {0x1b34, 0x1b44}, 2467 {0x1b6b, 0x1b73}, 2468 {0x1b80, 0x1b82}, 2469 {0x1ba1, 0x1bad}, 2470 {0x1be6, 0x1bf3}, 2471 {0x1c24, 0x1c37}, 2472 {0x1cd0, 0x1cd2}, 2473 {0x1cd4, 0x1ce8}, 2474 {0x1ced, 0x1ced}, 2475 {0x1cf2, 0x1cf4}, 2476 {0x1cf7, 0x1cf9}, 2477 {0x1dc0, 0x1df9}, 2478 {0x1dfb, 0x1dff}, 2479 {0x20d0, 0x20f0}, 2480 {0x2cef, 0x2cf1}, 2481 {0x2d7f, 0x2d7f}, 2482 {0x2de0, 0x2dff}, 2483 {0x302a, 0x302f}, 2484 {0x3099, 0x309a}, 2485 {0xa66f, 0xa672}, 2486 {0xa674, 0xa67d}, 2487 {0xa69e, 0xa69f}, 2488 {0xa6f0, 0xa6f1}, 2489 {0xa802, 0xa802}, 2490 {0xa806, 0xa806}, 2491 {0xa80b, 0xa80b}, 2492 {0xa823, 0xa827}, 2493 {0xa880, 0xa881}, 2494 {0xa8b4, 0xa8c5}, 2495 {0xa8e0, 0xa8f1}, 2496 {0xa926, 0xa92d}, 2497 {0xa947, 0xa953}, 2498 {0xa980, 0xa983}, 2499 {0xa9b3, 0xa9c0}, 2500 {0xa9e5, 0xa9e5}, 2501 {0xaa29, 0xaa36}, 2502 {0xaa43, 0xaa43}, 2503 {0xaa4c, 0xaa4d}, 2504 {0xaa7b, 0xaa7d}, 2505 {0xaab0, 0xaab0}, 2506 {0xaab2, 0xaab4}, 2507 {0xaab7, 0xaab8}, 2508 {0xaabe, 0xaabf}, 2509 {0xaac1, 0xaac1}, 2510 {0xaaeb, 0xaaef}, 2511 {0xaaf5, 0xaaf6}, 2512 {0xabe3, 0xabea}, 2513 {0xabec, 0xabed}, 2514 {0xfb1e, 0xfb1e}, 2515 {0xfe00, 0xfe0f}, 2516 {0xfe20, 0xfe2f}, 2517 {0x101fd, 0x101fd}, 2518 {0x102e0, 0x102e0}, 2519 {0x10376, 0x1037a}, 2520 {0x10a01, 0x10a03}, 2521 {0x10a05, 0x10a06}, 2522 {0x10a0c, 0x10a0f}, 2523 {0x10a38, 0x10a3a}, 2524 {0x10a3f, 0x10a3f}, 2525 {0x10ae5, 0x10ae6}, 2526 {0x11000, 0x11002}, 2527 {0x11038, 0x11046}, 2528 {0x1107f, 0x11082}, 2529 {0x110b0, 0x110ba}, 2530 {0x11100, 0x11102}, 2531 {0x11127, 0x11134}, 2532 {0x11173, 0x11173}, 2533 {0x11180, 0x11182}, 2534 {0x111b3, 0x111c0}, 2535 {0x111ca, 0x111cc}, 2536 {0x1122c, 0x11237}, 2537 {0x1123e, 0x1123e}, 2538 {0x112df, 0x112ea}, 2539 {0x11300, 0x11303}, 2540 {0x1133c, 0x1133c}, 2541 {0x1133e, 0x11344}, 2542 {0x11347, 0x11348}, 2543 {0x1134b, 0x1134d}, 2544 {0x11357, 0x11357}, 2545 {0x11362, 0x11363}, 2546 {0x11366, 0x1136c}, 2547 {0x11370, 0x11374}, 2548 {0x11435, 0x11446}, 2549 {0x114b0, 0x114c3}, 2550 {0x115af, 0x115b5}, 2551 {0x115b8, 0x115c0}, 2552 {0x115dc, 0x115dd}, 2553 {0x11630, 0x11640}, 2554 {0x116ab, 0x116b7}, 2555 {0x1171d, 0x1172b}, 2556 {0x11a01, 0x11a0a}, 2557 {0x11a33, 0x11a39}, 2558 {0x11a3b, 0x11a3e}, 2559 {0x11a47, 0x11a47}, 2560 {0x11a51, 0x11a5b}, 2561 {0x11a8a, 0x11a99}, 2562 {0x11c2f, 0x11c36}, 2563 {0x11c38, 0x11c3f}, 2564 {0x11c92, 0x11ca7}, 2565 {0x11ca9, 0x11cb6}, 2566 {0x11d31, 0x11d36}, 2567 {0x11d3a, 0x11d3a}, 2568 {0x11d3c, 0x11d3d}, 2569 {0x11d3f, 0x11d45}, 2570 {0x11d47, 0x11d47}, 2571 {0x16af0, 0x16af4}, 2572 {0x16b30, 0x16b36}, 2573 {0x16f51, 0x16f7e}, 2574 {0x16f8f, 0x16f92}, 2575 {0x1bc9d, 0x1bc9e}, 2576 {0x1d165, 0x1d169}, 2577 {0x1d16d, 0x1d172}, 2578 {0x1d17b, 0x1d182}, 2579 {0x1d185, 0x1d18b}, 2580 {0x1d1aa, 0x1d1ad}, 2581 {0x1d242, 0x1d244}, 2582 {0x1da00, 0x1da36}, 2583 {0x1da3b, 0x1da6c}, 2584 {0x1da75, 0x1da75}, 2585 {0x1da84, 0x1da84}, 2586 {0x1da9b, 0x1da9f}, 2587 {0x1daa1, 0x1daaf}, 2588 {0x1e000, 0x1e006}, 2589 {0x1e008, 0x1e018}, 2590 {0x1e01b, 0x1e021}, 2591 {0x1e023, 0x1e024}, 2592 {0x1e026, 0x1e02a}, 2593 {0x1e8d0, 0x1e8d6}, 2594 {0x1e944, 0x1e94a}, 2595 {0xe0100, 0xe01ef} 2596 }; 2597 2598 return intable(combining, sizeof(combining), c); 2599 } 2600 2601 /* 2602 * Return TRUE for characters that can be displayed in a normal way. 2603 * Only for characters of 0x100 and above! 2604 */ 2605 int 2606 utf_printable(int c) 2607 { 2608 #ifdef USE_WCHAR_FUNCTIONS 2609 /* 2610 * Assume the iswprint() library function works better than our own stuff. 2611 */ 2612 return iswprint(c); 2613 #else 2614 /* Sorted list of non-overlapping intervals. 2615 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */ 2616 static struct interval nonprint[] = 2617 { 2618 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2619 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2620 {0xfffe, 0xffff} 2621 }; 2622 2623 return !intable(nonprint, sizeof(nonprint), c); 2624 #endif 2625 } 2626 2627 /* Sorted list of non-overlapping intervals of all Emoji characters, 2628 * based on http://unicode.org/emoji/charts/emoji-list.html */ 2629 static struct interval emoji_all[] = 2630 { 2631 {0x203c, 0x203c}, 2632 {0x2049, 0x2049}, 2633 {0x2122, 0x2122}, 2634 {0x2139, 0x2139}, 2635 {0x2194, 0x2199}, 2636 {0x21a9, 0x21aa}, 2637 {0x231a, 0x231b}, 2638 {0x2328, 0x2328}, 2639 {0x23cf, 0x23cf}, 2640 {0x23e9, 0x23f3}, 2641 {0x23f8, 0x23fa}, 2642 {0x24c2, 0x24c2}, 2643 {0x25aa, 0x25ab}, 2644 {0x25b6, 0x25b6}, 2645 {0x25c0, 0x25c0}, 2646 {0x25fb, 0x25fe}, 2647 {0x2600, 0x2604}, 2648 {0x260e, 0x260e}, 2649 {0x2611, 0x2611}, 2650 {0x2614, 0x2615}, 2651 {0x2618, 0x2618}, 2652 {0x261d, 0x261d}, 2653 {0x2620, 0x2620}, 2654 {0x2622, 0x2623}, 2655 {0x2626, 0x2626}, 2656 {0x262a, 0x262a}, 2657 {0x262e, 0x262f}, 2658 {0x2638, 0x263a}, 2659 {0x2640, 0x2640}, 2660 {0x2642, 0x2642}, 2661 {0x2648, 0x2653}, 2662 {0x2660, 0x2660}, 2663 {0x2663, 0x2663}, 2664 {0x2665, 0x2666}, 2665 {0x2668, 0x2668}, 2666 {0x267b, 0x267b}, 2667 {0x267f, 0x267f}, 2668 {0x2692, 0x2697}, 2669 {0x2699, 0x2699}, 2670 {0x269b, 0x269c}, 2671 {0x26a0, 0x26a1}, 2672 {0x26aa, 0x26ab}, 2673 {0x26b0, 0x26b1}, 2674 {0x26bd, 0x26be}, 2675 {0x26c4, 0x26c5}, 2676 {0x26c8, 0x26c8}, 2677 {0x26ce, 0x26cf}, 2678 {0x26d1, 0x26d1}, 2679 {0x26d3, 0x26d4}, 2680 {0x26e9, 0x26ea}, 2681 {0x26f0, 0x26f5}, 2682 {0x26f7, 0x26fa}, 2683 {0x26fd, 0x26fd}, 2684 {0x2702, 0x2702}, 2685 {0x2705, 0x2705}, 2686 {0x2708, 0x270d}, 2687 {0x270f, 0x270f}, 2688 {0x2712, 0x2712}, 2689 {0x2714, 0x2714}, 2690 {0x2716, 0x2716}, 2691 {0x271d, 0x271d}, 2692 {0x2721, 0x2721}, 2693 {0x2728, 0x2728}, 2694 {0x2733, 0x2734}, 2695 {0x2744, 0x2744}, 2696 {0x2747, 0x2747}, 2697 {0x274c, 0x274c}, 2698 {0x274e, 0x274e}, 2699 {0x2753, 0x2755}, 2700 {0x2757, 0x2757}, 2701 {0x2763, 0x2764}, 2702 {0x2795, 0x2797}, 2703 {0x27a1, 0x27a1}, 2704 {0x27b0, 0x27b0}, 2705 {0x27bf, 0x27bf}, 2706 {0x2934, 0x2935}, 2707 {0x2b05, 0x2b07}, 2708 {0x2b1b, 0x2b1c}, 2709 {0x2b50, 0x2b50}, 2710 {0x2b55, 0x2b55}, 2711 {0x3030, 0x3030}, 2712 {0x303d, 0x303d}, 2713 {0x3297, 0x3297}, 2714 {0x3299, 0x3299}, 2715 {0x1f004, 0x1f004}, 2716 {0x1f0cf, 0x1f0cf}, 2717 {0x1f170, 0x1f171}, 2718 {0x1f17e, 0x1f17f}, 2719 {0x1f18e, 0x1f18e}, 2720 {0x1f191, 0x1f19a}, 2721 {0x1f1e6, 0x1f1ff}, 2722 {0x1f201, 0x1f202}, 2723 {0x1f21a, 0x1f21a}, 2724 {0x1f22f, 0x1f22f}, 2725 {0x1f232, 0x1f23a}, 2726 {0x1f250, 0x1f251}, 2727 {0x1f300, 0x1f321}, 2728 {0x1f324, 0x1f393}, 2729 {0x1f396, 0x1f397}, 2730 {0x1f399, 0x1f39b}, 2731 {0x1f39e, 0x1f3f0}, 2732 {0x1f3f3, 0x1f3f5}, 2733 {0x1f3f7, 0x1f4fd}, 2734 {0x1f4ff, 0x1f53d}, 2735 {0x1f549, 0x1f54e}, 2736 {0x1f550, 0x1f567}, 2737 {0x1f56f, 0x1f570}, 2738 {0x1f573, 0x1f57a}, 2739 {0x1f587, 0x1f587}, 2740 {0x1f58a, 0x1f58d}, 2741 {0x1f590, 0x1f590}, 2742 {0x1f595, 0x1f596}, 2743 {0x1f5a4, 0x1f5a5}, 2744 {0x1f5a8, 0x1f5a8}, 2745 {0x1f5b1, 0x1f5b2}, 2746 {0x1f5bc, 0x1f5bc}, 2747 {0x1f5c2, 0x1f5c4}, 2748 {0x1f5d1, 0x1f5d3}, 2749 {0x1f5dc, 0x1f5de}, 2750 {0x1f5e1, 0x1f5e1}, 2751 {0x1f5e3, 0x1f5e3}, 2752 {0x1f5e8, 0x1f5e8}, 2753 {0x1f5ef, 0x1f5ef}, 2754 {0x1f5f3, 0x1f5f3}, 2755 {0x1f5fa, 0x1f64f}, 2756 {0x1f680, 0x1f6c5}, 2757 {0x1f6cb, 0x1f6d2}, 2758 {0x1f6e0, 0x1f6e5}, 2759 {0x1f6e9, 0x1f6e9}, 2760 {0x1f6eb, 0x1f6ec}, 2761 {0x1f6f0, 0x1f6f0}, 2762 {0x1f6f3, 0x1f6f8}, 2763 {0x1f910, 0x1f93a}, 2764 {0x1f93c, 0x1f93e}, 2765 {0x1f940, 0x1f945}, 2766 {0x1f947, 0x1f94c}, 2767 {0x1f950, 0x1f96b}, 2768 {0x1f980, 0x1f997}, 2769 {0x1f9c0, 0x1f9c0}, 2770 {0x1f9d0, 0x1f9e6} 2771 }; 2772 2773 /* 2774 * Get class of a Unicode character. 2775 * 0: white space 2776 * 1: punctuation 2777 * 2 or bigger: some class of word character. 2778 */ 2779 int 2780 utf_class(int c) 2781 { 2782 return utf_class_buf(c, curbuf); 2783 } 2784 2785 int 2786 utf_class_buf(int c, buf_T *buf) 2787 { 2788 /* sorted list of non-overlapping intervals */ 2789 static struct clinterval 2790 { 2791 unsigned int first; 2792 unsigned int last; 2793 unsigned int class; 2794 } classes[] = 2795 { 2796 {0x037e, 0x037e, 1}, /* Greek question mark */ 2797 {0x0387, 0x0387, 1}, /* Greek ano teleia */ 2798 {0x055a, 0x055f, 1}, /* Armenian punctuation */ 2799 {0x0589, 0x0589, 1}, /* Armenian full stop */ 2800 {0x05be, 0x05be, 1}, 2801 {0x05c0, 0x05c0, 1}, 2802 {0x05c3, 0x05c3, 1}, 2803 {0x05f3, 0x05f4, 1}, 2804 {0x060c, 0x060c, 1}, 2805 {0x061b, 0x061b, 1}, 2806 {0x061f, 0x061f, 1}, 2807 {0x066a, 0x066d, 1}, 2808 {0x06d4, 0x06d4, 1}, 2809 {0x0700, 0x070d, 1}, /* Syriac punctuation */ 2810 {0x0964, 0x0965, 1}, 2811 {0x0970, 0x0970, 1}, 2812 {0x0df4, 0x0df4, 1}, 2813 {0x0e4f, 0x0e4f, 1}, 2814 {0x0e5a, 0x0e5b, 1}, 2815 {0x0f04, 0x0f12, 1}, 2816 {0x0f3a, 0x0f3d, 1}, 2817 {0x0f85, 0x0f85, 1}, 2818 {0x104a, 0x104f, 1}, /* Myanmar punctuation */ 2819 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */ 2820 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */ 2821 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */ 2822 {0x1680, 0x1680, 0}, 2823 {0x169b, 0x169c, 1}, 2824 {0x16eb, 0x16ed, 1}, 2825 {0x1735, 0x1736, 1}, 2826 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */ 2827 {0x1800, 0x180a, 1}, /* Mongolian punctuation */ 2828 {0x2000, 0x200b, 0}, /* spaces */ 2829 {0x200c, 0x2027, 1}, /* punctuation and symbols */ 2830 {0x2028, 0x2029, 0}, 2831 {0x202a, 0x202e, 1}, /* punctuation and symbols */ 2832 {0x202f, 0x202f, 0}, 2833 {0x2030, 0x205e, 1}, /* punctuation and symbols */ 2834 {0x205f, 0x205f, 0}, 2835 {0x2060, 0x27ff, 1}, /* punctuation and symbols */ 2836 {0x2070, 0x207f, 0x2070}, /* superscript */ 2837 {0x2080, 0x2094, 0x2080}, /* subscript */ 2838 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */ 2839 {0x2800, 0x28ff, 0x2800}, /* braille */ 2840 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */ 2841 {0x29d8, 0x29db, 1}, 2842 {0x29fc, 0x29fd, 1}, 2843 {0x2e00, 0x2e7f, 1}, /* supplemental punctuation */ 2844 {0x3000, 0x3000, 0}, /* ideographic space */ 2845 {0x3001, 0x3020, 1}, /* ideographic punctuation */ 2846 {0x3030, 0x3030, 1}, 2847 {0x303d, 0x303d, 1}, 2848 {0x3040, 0x309f, 0x3040}, /* Hiragana */ 2849 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */ 2850 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */ 2851 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */ 2852 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */ 2853 {0xfd3e, 0xfd3f, 1}, 2854 {0xfe30, 0xfe6b, 1}, /* punctuation forms */ 2855 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */ 2856 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */ 2857 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */ 2858 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */ 2859 {0x20000, 0x2a6df, 0x4e00}, /* CJK Ideographs */ 2860 {0x2a700, 0x2b73f, 0x4e00}, /* CJK Ideographs */ 2861 {0x2b740, 0x2b81f, 0x4e00}, /* CJK Ideographs */ 2862 {0x2f800, 0x2fa1f, 0x4e00}, /* CJK Ideographs */ 2863 }; 2864 2865 int bot = 0; 2866 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2867 int mid; 2868 2869 /* First quick check for Latin1 characters, use 'iskeyword'. */ 2870 if (c < 0x100) 2871 { 2872 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2873 return 0; /* blank */ 2874 if (vim_iswordc_buf(c, buf)) 2875 return 2; /* word character */ 2876 return 1; /* punctuation */ 2877 } 2878 2879 /* binary search in table */ 2880 while (top >= bot) 2881 { 2882 mid = (bot + top) / 2; 2883 if (classes[mid].last < (unsigned int)c) 2884 bot = mid + 1; 2885 else if (classes[mid].first > (unsigned int)c) 2886 top = mid - 1; 2887 else 2888 return (int)classes[mid].class; 2889 } 2890 2891 /* emoji */ 2892 if (intable(emoji_all, sizeof(emoji_all), c)) 2893 return 3; 2894 2895 /* most other characters are "word" characters */ 2896 return 2; 2897 } 2898 2899 int 2900 utf_ambiguous_width(int c) 2901 { 2902 return c >= 0x80 && (intable(ambiguous, sizeof(ambiguous), c) 2903 || intable(emoji_all, sizeof(emoji_all), c)); 2904 } 2905 2906 /* 2907 * Code for Unicode case-dependent operations. Based on notes in 2908 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2909 * This code uses simple case folding, not full case folding. 2910 * Last updated for Unicode 5.2. 2911 */ 2912 2913 /* 2914 * The following tables are built by ../runtime/tools/unicode.vim. 2915 * They must be in numeric order, because we use binary search. 2916 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2917 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2918 * folded/upper/lower by adding 32. 2919 */ 2920 typedef struct 2921 { 2922 int rangeStart; 2923 int rangeEnd; 2924 int step; 2925 int offset; 2926 } convertStruct; 2927 2928 static convertStruct foldCase[] = 2929 { 2930 {0x41,0x5a,1,32}, 2931 {0xb5,0xb5,-1,775}, 2932 {0xc0,0xd6,1,32}, 2933 {0xd8,0xde,1,32}, 2934 {0x100,0x12e,2,1}, 2935 {0x132,0x136,2,1}, 2936 {0x139,0x147,2,1}, 2937 {0x14a,0x176,2,1}, 2938 {0x178,0x178,-1,-121}, 2939 {0x179,0x17d,2,1}, 2940 {0x17f,0x17f,-1,-268}, 2941 {0x181,0x181,-1,210}, 2942 {0x182,0x184,2,1}, 2943 {0x186,0x186,-1,206}, 2944 {0x187,0x187,-1,1}, 2945 {0x189,0x18a,1,205}, 2946 {0x18b,0x18b,-1,1}, 2947 {0x18e,0x18e,-1,79}, 2948 {0x18f,0x18f,-1,202}, 2949 {0x190,0x190,-1,203}, 2950 {0x191,0x191,-1,1}, 2951 {0x193,0x193,-1,205}, 2952 {0x194,0x194,-1,207}, 2953 {0x196,0x196,-1,211}, 2954 {0x197,0x197,-1,209}, 2955 {0x198,0x198,-1,1}, 2956 {0x19c,0x19c,-1,211}, 2957 {0x19d,0x19d,-1,213}, 2958 {0x19f,0x19f,-1,214}, 2959 {0x1a0,0x1a4,2,1}, 2960 {0x1a6,0x1a6,-1,218}, 2961 {0x1a7,0x1a7,-1,1}, 2962 {0x1a9,0x1a9,-1,218}, 2963 {0x1ac,0x1ac,-1,1}, 2964 {0x1ae,0x1ae,-1,218}, 2965 {0x1af,0x1af,-1,1}, 2966 {0x1b1,0x1b2,1,217}, 2967 {0x1b3,0x1b5,2,1}, 2968 {0x1b7,0x1b7,-1,219}, 2969 {0x1b8,0x1bc,4,1}, 2970 {0x1c4,0x1c4,-1,2}, 2971 {0x1c5,0x1c5,-1,1}, 2972 {0x1c7,0x1c7,-1,2}, 2973 {0x1c8,0x1c8,-1,1}, 2974 {0x1ca,0x1ca,-1,2}, 2975 {0x1cb,0x1db,2,1}, 2976 {0x1de,0x1ee,2,1}, 2977 {0x1f1,0x1f1,-1,2}, 2978 {0x1f2,0x1f4,2,1}, 2979 {0x1f6,0x1f6,-1,-97}, 2980 {0x1f7,0x1f7,-1,-56}, 2981 {0x1f8,0x21e,2,1}, 2982 {0x220,0x220,-1,-130}, 2983 {0x222,0x232,2,1}, 2984 {0x23a,0x23a,-1,10795}, 2985 {0x23b,0x23b,-1,1}, 2986 {0x23d,0x23d,-1,-163}, 2987 {0x23e,0x23e,-1,10792}, 2988 {0x241,0x241,-1,1}, 2989 {0x243,0x243,-1,-195}, 2990 {0x244,0x244,-1,69}, 2991 {0x245,0x245,-1,71}, 2992 {0x246,0x24e,2,1}, 2993 {0x345,0x345,-1,116}, 2994 {0x370,0x372,2,1}, 2995 {0x376,0x376,-1,1}, 2996 {0x37f,0x37f,-1,116}, 2997 {0x386,0x386,-1,38}, 2998 {0x388,0x38a,1,37}, 2999 {0x38c,0x38c,-1,64}, 3000 {0x38e,0x38f,1,63}, 3001 {0x391,0x3a1,1,32}, 3002 {0x3a3,0x3ab,1,32}, 3003 {0x3c2,0x3c2,-1,1}, 3004 {0x3cf,0x3cf,-1,8}, 3005 {0x3d0,0x3d0,-1,-30}, 3006 {0x3d1,0x3d1,-1,-25}, 3007 {0x3d5,0x3d5,-1,-15}, 3008 {0x3d6,0x3d6,-1,-22}, 3009 {0x3d8,0x3ee,2,1}, 3010 {0x3f0,0x3f0,-1,-54}, 3011 {0x3f1,0x3f1,-1,-48}, 3012 {0x3f4,0x3f4,-1,-60}, 3013 {0x3f5,0x3f5,-1,-64}, 3014 {0x3f7,0x3f7,-1,1}, 3015 {0x3f9,0x3f9,-1,-7}, 3016 {0x3fa,0x3fa,-1,1}, 3017 {0x3fd,0x3ff,1,-130}, 3018 {0x400,0x40f,1,80}, 3019 {0x410,0x42f,1,32}, 3020 {0x460,0x480,2,1}, 3021 {0x48a,0x4be,2,1}, 3022 {0x4c0,0x4c0,-1,15}, 3023 {0x4c1,0x4cd,2,1}, 3024 {0x4d0,0x52e,2,1}, 3025 {0x531,0x556,1,48}, 3026 {0x10a0,0x10c5,1,7264}, 3027 {0x10c7,0x10cd,6,7264}, 3028 {0x13f8,0x13fd,1,-8}, 3029 {0x1c80,0x1c80,-1,-6222}, 3030 {0x1c81,0x1c81,-1,-6221}, 3031 {0x1c82,0x1c82,-1,-6212}, 3032 {0x1c83,0x1c84,1,-6210}, 3033 {0x1c85,0x1c85,-1,-6211}, 3034 {0x1c86,0x1c86,-1,-6204}, 3035 {0x1c87,0x1c87,-1,-6180}, 3036 {0x1c88,0x1c88,-1,35267}, 3037 {0x1e00,0x1e94,2,1}, 3038 {0x1e9b,0x1e9b,-1,-58}, 3039 {0x1e9e,0x1e9e,-1,-7615}, 3040 {0x1ea0,0x1efe,2,1}, 3041 {0x1f08,0x1f0f,1,-8}, 3042 {0x1f18,0x1f1d,1,-8}, 3043 {0x1f28,0x1f2f,1,-8}, 3044 {0x1f38,0x1f3f,1,-8}, 3045 {0x1f48,0x1f4d,1,-8}, 3046 {0x1f59,0x1f5f,2,-8}, 3047 {0x1f68,0x1f6f,1,-8}, 3048 {0x1f88,0x1f8f,1,-8}, 3049 {0x1f98,0x1f9f,1,-8}, 3050 {0x1fa8,0x1faf,1,-8}, 3051 {0x1fb8,0x1fb9,1,-8}, 3052 {0x1fba,0x1fbb,1,-74}, 3053 {0x1fbc,0x1fbc,-1,-9}, 3054 {0x1fbe,0x1fbe,-1,-7173}, 3055 {0x1fc8,0x1fcb,1,-86}, 3056 {0x1fcc,0x1fcc,-1,-9}, 3057 {0x1fd8,0x1fd9,1,-8}, 3058 {0x1fda,0x1fdb,1,-100}, 3059 {0x1fe8,0x1fe9,1,-8}, 3060 {0x1fea,0x1feb,1,-112}, 3061 {0x1fec,0x1fec,-1,-7}, 3062 {0x1ff8,0x1ff9,1,-128}, 3063 {0x1ffa,0x1ffb,1,-126}, 3064 {0x1ffc,0x1ffc,-1,-9}, 3065 {0x2126,0x2126,-1,-7517}, 3066 {0x212a,0x212a,-1,-8383}, 3067 {0x212b,0x212b,-1,-8262}, 3068 {0x2132,0x2132,-1,28}, 3069 {0x2160,0x216f,1,16}, 3070 {0x2183,0x2183,-1,1}, 3071 {0x24b6,0x24cf,1,26}, 3072 {0x2c00,0x2c2e,1,48}, 3073 {0x2c60,0x2c60,-1,1}, 3074 {0x2c62,0x2c62,-1,-10743}, 3075 {0x2c63,0x2c63,-1,-3814}, 3076 {0x2c64,0x2c64,-1,-10727}, 3077 {0x2c67,0x2c6b,2,1}, 3078 {0x2c6d,0x2c6d,-1,-10780}, 3079 {0x2c6e,0x2c6e,-1,-10749}, 3080 {0x2c6f,0x2c6f,-1,-10783}, 3081 {0x2c70,0x2c70,-1,-10782}, 3082 {0x2c72,0x2c75,3,1}, 3083 {0x2c7e,0x2c7f,1,-10815}, 3084 {0x2c80,0x2ce2,2,1}, 3085 {0x2ceb,0x2ced,2,1}, 3086 {0x2cf2,0xa640,31054,1}, 3087 {0xa642,0xa66c,2,1}, 3088 {0xa680,0xa69a,2,1}, 3089 {0xa722,0xa72e,2,1}, 3090 {0xa732,0xa76e,2,1}, 3091 {0xa779,0xa77b,2,1}, 3092 {0xa77d,0xa77d,-1,-35332}, 3093 {0xa77e,0xa786,2,1}, 3094 {0xa78b,0xa78b,-1,1}, 3095 {0xa78d,0xa78d,-1,-42280}, 3096 {0xa790,0xa792,2,1}, 3097 {0xa796,0xa7a8,2,1}, 3098 {0xa7aa,0xa7aa,-1,-42308}, 3099 {0xa7ab,0xa7ab,-1,-42319}, 3100 {0xa7ac,0xa7ac,-1,-42315}, 3101 {0xa7ad,0xa7ad,-1,-42305}, 3102 {0xa7ae,0xa7ae,-1,-42308}, 3103 {0xa7b0,0xa7b0,-1,-42258}, 3104 {0xa7b1,0xa7b1,-1,-42282}, 3105 {0xa7b2,0xa7b2,-1,-42261}, 3106 {0xa7b3,0xa7b3,-1,928}, 3107 {0xa7b4,0xa7b6,2,1}, 3108 {0xab70,0xabbf,1,-38864}, 3109 {0xff21,0xff3a,1,32}, 3110 {0x10400,0x10427,1,40}, 3111 {0x104b0,0x104d3,1,40}, 3112 {0x10c80,0x10cb2,1,64}, 3113 {0x118a0,0x118bf,1,32}, 3114 {0x1e900,0x1e921,1,34} 3115 }; 3116 3117 static int utf_convert(int a, convertStruct table[], int tableSize); 3118 static int utf_strnicmp(char_u *s1, char_u *s2, size_t n1, size_t n2); 3119 3120 /* 3121 * Generic conversion function for case operations. 3122 * Return the converted equivalent of "a", which is a UCS-4 character. Use 3123 * the given conversion "table". Uses binary search on "table". 3124 */ 3125 static int 3126 utf_convert( 3127 int a, 3128 convertStruct table[], 3129 int tableSize) 3130 { 3131 int start, mid, end; /* indices into table */ 3132 int entries = tableSize / sizeof(convertStruct); 3133 3134 start = 0; 3135 end = entries; 3136 while (start < end) 3137 { 3138 /* need to search further */ 3139 mid = (end + start) / 2; 3140 if (table[mid].rangeEnd < a) 3141 start = mid + 1; 3142 else 3143 end = mid; 3144 } 3145 if (start < entries 3146 && table[start].rangeStart <= a 3147 && a <= table[start].rangeEnd 3148 && (a - table[start].rangeStart) % table[start].step == 0) 3149 return (a + table[start].offset); 3150 else 3151 return a; 3152 } 3153 3154 /* 3155 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 3156 * simple case folding. 3157 */ 3158 int 3159 utf_fold(int a) 3160 { 3161 if (a < 0x80) 3162 /* be fast for ASCII */ 3163 return a >= 0x41 && a <= 0x5a ? a + 32 : a; 3164 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 3165 } 3166 3167 static convertStruct toLower[] = 3168 { 3169 {0x41,0x5a,1,32}, 3170 {0xc0,0xd6,1,32}, 3171 {0xd8,0xde,1,32}, 3172 {0x100,0x12e,2,1}, 3173 {0x130,0x130,-1,-199}, 3174 {0x132,0x136,2,1}, 3175 {0x139,0x147,2,1}, 3176 {0x14a,0x176,2,1}, 3177 {0x178,0x178,-1,-121}, 3178 {0x179,0x17d,2,1}, 3179 {0x181,0x181,-1,210}, 3180 {0x182,0x184,2,1}, 3181 {0x186,0x186,-1,206}, 3182 {0x187,0x187,-1,1}, 3183 {0x189,0x18a,1,205}, 3184 {0x18b,0x18b,-1,1}, 3185 {0x18e,0x18e,-1,79}, 3186 {0x18f,0x18f,-1,202}, 3187 {0x190,0x190,-1,203}, 3188 {0x191,0x191,-1,1}, 3189 {0x193,0x193,-1,205}, 3190 {0x194,0x194,-1,207}, 3191 {0x196,0x196,-1,211}, 3192 {0x197,0x197,-1,209}, 3193 {0x198,0x198,-1,1}, 3194 {0x19c,0x19c,-1,211}, 3195 {0x19d,0x19d,-1,213}, 3196 {0x19f,0x19f,-1,214}, 3197 {0x1a0,0x1a4,2,1}, 3198 {0x1a6,0x1a6,-1,218}, 3199 {0x1a7,0x1a7,-1,1}, 3200 {0x1a9,0x1a9,-1,218}, 3201 {0x1ac,0x1ac,-1,1}, 3202 {0x1ae,0x1ae,-1,218}, 3203 {0x1af,0x1af,-1,1}, 3204 {0x1b1,0x1b2,1,217}, 3205 {0x1b3,0x1b5,2,1}, 3206 {0x1b7,0x1b7,-1,219}, 3207 {0x1b8,0x1bc,4,1}, 3208 {0x1c4,0x1c4,-1,2}, 3209 {0x1c5,0x1c5,-1,1}, 3210 {0x1c7,0x1c7,-1,2}, 3211 {0x1c8,0x1c8,-1,1}, 3212 {0x1ca,0x1ca,-1,2}, 3213 {0x1cb,0x1db,2,1}, 3214 {0x1de,0x1ee,2,1}, 3215 {0x1f1,0x1f1,-1,2}, 3216 {0x1f2,0x1f4,2,1}, 3217 {0x1f6,0x1f6,-1,-97}, 3218 {0x1f7,0x1f7,-1,-56}, 3219 {0x1f8,0x21e,2,1}, 3220 {0x220,0x220,-1,-130}, 3221 {0x222,0x232,2,1}, 3222 {0x23a,0x23a,-1,10795}, 3223 {0x23b,0x23b,-1,1}, 3224 {0x23d,0x23d,-1,-163}, 3225 {0x23e,0x23e,-1,10792}, 3226 {0x241,0x241,-1,1}, 3227 {0x243,0x243,-1,-195}, 3228 {0x244,0x244,-1,69}, 3229 {0x245,0x245,-1,71}, 3230 {0x246,0x24e,2,1}, 3231 {0x370,0x372,2,1}, 3232 {0x376,0x376,-1,1}, 3233 {0x37f,0x37f,-1,116}, 3234 {0x386,0x386,-1,38}, 3235 {0x388,0x38a,1,37}, 3236 {0x38c,0x38c,-1,64}, 3237 {0x38e,0x38f,1,63}, 3238 {0x391,0x3a1,1,32}, 3239 {0x3a3,0x3ab,1,32}, 3240 {0x3cf,0x3cf,-1,8}, 3241 {0x3d8,0x3ee,2,1}, 3242 {0x3f4,0x3f4,-1,-60}, 3243 {0x3f7,0x3f7,-1,1}, 3244 {0x3f9,0x3f9,-1,-7}, 3245 {0x3fa,0x3fa,-1,1}, 3246 {0x3fd,0x3ff,1,-130}, 3247 {0x400,0x40f,1,80}, 3248 {0x410,0x42f,1,32}, 3249 {0x460,0x480,2,1}, 3250 {0x48a,0x4be,2,1}, 3251 {0x4c0,0x4c0,-1,15}, 3252 {0x4c1,0x4cd,2,1}, 3253 {0x4d0,0x52e,2,1}, 3254 {0x531,0x556,1,48}, 3255 {0x10a0,0x10c5,1,7264}, 3256 {0x10c7,0x10cd,6,7264}, 3257 {0x13a0,0x13ef,1,38864}, 3258 {0x13f0,0x13f5,1,8}, 3259 {0x1e00,0x1e94,2,1}, 3260 {0x1e9e,0x1e9e,-1,-7615}, 3261 {0x1ea0,0x1efe,2,1}, 3262 {0x1f08,0x1f0f,1,-8}, 3263 {0x1f18,0x1f1d,1,-8}, 3264 {0x1f28,0x1f2f,1,-8}, 3265 {0x1f38,0x1f3f,1,-8}, 3266 {0x1f48,0x1f4d,1,-8}, 3267 {0x1f59,0x1f5f,2,-8}, 3268 {0x1f68,0x1f6f,1,-8}, 3269 {0x1f88,0x1f8f,1,-8}, 3270 {0x1f98,0x1f9f,1,-8}, 3271 {0x1fa8,0x1faf,1,-8}, 3272 {0x1fb8,0x1fb9,1,-8}, 3273 {0x1fba,0x1fbb,1,-74}, 3274 {0x1fbc,0x1fbc,-1,-9}, 3275 {0x1fc8,0x1fcb,1,-86}, 3276 {0x1fcc,0x1fcc,-1,-9}, 3277 {0x1fd8,0x1fd9,1,-8}, 3278 {0x1fda,0x1fdb,1,-100}, 3279 {0x1fe8,0x1fe9,1,-8}, 3280 {0x1fea,0x1feb,1,-112}, 3281 {0x1fec,0x1fec,-1,-7}, 3282 {0x1ff8,0x1ff9,1,-128}, 3283 {0x1ffa,0x1ffb,1,-126}, 3284 {0x1ffc,0x1ffc,-1,-9}, 3285 {0x2126,0x2126,-1,-7517}, 3286 {0x212a,0x212a,-1,-8383}, 3287 {0x212b,0x212b,-1,-8262}, 3288 {0x2132,0x2132,-1,28}, 3289 {0x2160,0x216f,1,16}, 3290 {0x2183,0x2183,-1,1}, 3291 {0x24b6,0x24cf,1,26}, 3292 {0x2c00,0x2c2e,1,48}, 3293 {0x2c60,0x2c60,-1,1}, 3294 {0x2c62,0x2c62,-1,-10743}, 3295 {0x2c63,0x2c63,-1,-3814}, 3296 {0x2c64,0x2c64,-1,-10727}, 3297 {0x2c67,0x2c6b,2,1}, 3298 {0x2c6d,0x2c6d,-1,-10780}, 3299 {0x2c6e,0x2c6e,-1,-10749}, 3300 {0x2c6f,0x2c6f,-1,-10783}, 3301 {0x2c70,0x2c70,-1,-10782}, 3302 {0x2c72,0x2c75,3,1}, 3303 {0x2c7e,0x2c7f,1,-10815}, 3304 {0x2c80,0x2ce2,2,1}, 3305 {0x2ceb,0x2ced,2,1}, 3306 {0x2cf2,0xa640,31054,1}, 3307 {0xa642,0xa66c,2,1}, 3308 {0xa680,0xa69a,2,1}, 3309 {0xa722,0xa72e,2,1}, 3310 {0xa732,0xa76e,2,1}, 3311 {0xa779,0xa77b,2,1}, 3312 {0xa77d,0xa77d,-1,-35332}, 3313 {0xa77e,0xa786,2,1}, 3314 {0xa78b,0xa78b,-1,1}, 3315 {0xa78d,0xa78d,-1,-42280}, 3316 {0xa790,0xa792,2,1}, 3317 {0xa796,0xa7a8,2,1}, 3318 {0xa7aa,0xa7aa,-1,-42308}, 3319 {0xa7ab,0xa7ab,-1,-42319}, 3320 {0xa7ac,0xa7ac,-1,-42315}, 3321 {0xa7ad,0xa7ad,-1,-42305}, 3322 {0xa7ae,0xa7ae,-1,-42308}, 3323 {0xa7b0,0xa7b0,-1,-42258}, 3324 {0xa7b1,0xa7b1,-1,-42282}, 3325 {0xa7b2,0xa7b2,-1,-42261}, 3326 {0xa7b3,0xa7b3,-1,928}, 3327 {0xa7b4,0xa7b6,2,1}, 3328 {0xff21,0xff3a,1,32}, 3329 {0x10400,0x10427,1,40}, 3330 {0x104b0,0x104d3,1,40}, 3331 {0x10c80,0x10cb2,1,64}, 3332 {0x118a0,0x118bf,1,32}, 3333 {0x1e900,0x1e921,1,34} 3334 }; 3335 3336 static convertStruct toUpper[] = 3337 { 3338 {0x61,0x7a,1,-32}, 3339 {0xb5,0xb5,-1,743}, 3340 {0xe0,0xf6,1,-32}, 3341 {0xf8,0xfe,1,-32}, 3342 {0xff,0xff,-1,121}, 3343 {0x101,0x12f,2,-1}, 3344 {0x131,0x131,-1,-232}, 3345 {0x133,0x137,2,-1}, 3346 {0x13a,0x148,2,-1}, 3347 {0x14b,0x177,2,-1}, 3348 {0x17a,0x17e,2,-1}, 3349 {0x17f,0x17f,-1,-300}, 3350 {0x180,0x180,-1,195}, 3351 {0x183,0x185,2,-1}, 3352 {0x188,0x18c,4,-1}, 3353 {0x192,0x192,-1,-1}, 3354 {0x195,0x195,-1,97}, 3355 {0x199,0x199,-1,-1}, 3356 {0x19a,0x19a,-1,163}, 3357 {0x19e,0x19e,-1,130}, 3358 {0x1a1,0x1a5,2,-1}, 3359 {0x1a8,0x1ad,5,-1}, 3360 {0x1b0,0x1b4,4,-1}, 3361 {0x1b6,0x1b9,3,-1}, 3362 {0x1bd,0x1bd,-1,-1}, 3363 {0x1bf,0x1bf,-1,56}, 3364 {0x1c5,0x1c5,-1,-1}, 3365 {0x1c6,0x1c6,-1,-2}, 3366 {0x1c8,0x1c8,-1,-1}, 3367 {0x1c9,0x1c9,-1,-2}, 3368 {0x1cb,0x1cb,-1,-1}, 3369 {0x1cc,0x1cc,-1,-2}, 3370 {0x1ce,0x1dc,2,-1}, 3371 {0x1dd,0x1dd,-1,-79}, 3372 {0x1df,0x1ef,2,-1}, 3373 {0x1f2,0x1f2,-1,-1}, 3374 {0x1f3,0x1f3,-1,-2}, 3375 {0x1f5,0x1f9,4,-1}, 3376 {0x1fb,0x21f,2,-1}, 3377 {0x223,0x233,2,-1}, 3378 {0x23c,0x23c,-1,-1}, 3379 {0x23f,0x240,1,10815}, 3380 {0x242,0x247,5,-1}, 3381 {0x249,0x24f,2,-1}, 3382 {0x250,0x250,-1,10783}, 3383 {0x251,0x251,-1,10780}, 3384 {0x252,0x252,-1,10782}, 3385 {0x253,0x253,-1,-210}, 3386 {0x254,0x254,-1,-206}, 3387 {0x256,0x257,1,-205}, 3388 {0x259,0x259,-1,-202}, 3389 {0x25b,0x25b,-1,-203}, 3390 {0x25c,0x25c,-1,42319}, 3391 {0x260,0x260,-1,-205}, 3392 {0x261,0x261,-1,42315}, 3393 {0x263,0x263,-1,-207}, 3394 {0x265,0x265,-1,42280}, 3395 {0x266,0x266,-1,42308}, 3396 {0x268,0x268,-1,-209}, 3397 {0x269,0x269,-1,-211}, 3398 {0x26a,0x26a,-1,42308}, 3399 {0x26b,0x26b,-1,10743}, 3400 {0x26c,0x26c,-1,42305}, 3401 {0x26f,0x26f,-1,-211}, 3402 {0x271,0x271,-1,10749}, 3403 {0x272,0x272,-1,-213}, 3404 {0x275,0x275,-1,-214}, 3405 {0x27d,0x27d,-1,10727}, 3406 {0x280,0x283,3,-218}, 3407 {0x287,0x287,-1,42282}, 3408 {0x288,0x288,-1,-218}, 3409 {0x289,0x289,-1,-69}, 3410 {0x28a,0x28b,1,-217}, 3411 {0x28c,0x28c,-1,-71}, 3412 {0x292,0x292,-1,-219}, 3413 {0x29d,0x29d,-1,42261}, 3414 {0x29e,0x29e,-1,42258}, 3415 {0x345,0x345,-1,84}, 3416 {0x371,0x373,2,-1}, 3417 {0x377,0x377,-1,-1}, 3418 {0x37b,0x37d,1,130}, 3419 {0x3ac,0x3ac,-1,-38}, 3420 {0x3ad,0x3af,1,-37}, 3421 {0x3b1,0x3c1,1,-32}, 3422 {0x3c2,0x3c2,-1,-31}, 3423 {0x3c3,0x3cb,1,-32}, 3424 {0x3cc,0x3cc,-1,-64}, 3425 {0x3cd,0x3ce,1,-63}, 3426 {0x3d0,0x3d0,-1,-62}, 3427 {0x3d1,0x3d1,-1,-57}, 3428 {0x3d5,0x3d5,-1,-47}, 3429 {0x3d6,0x3d6,-1,-54}, 3430 {0x3d7,0x3d7,-1,-8}, 3431 {0x3d9,0x3ef,2,-1}, 3432 {0x3f0,0x3f0,-1,-86}, 3433 {0x3f1,0x3f1,-1,-80}, 3434 {0x3f2,0x3f2,-1,7}, 3435 {0x3f3,0x3f3,-1,-116}, 3436 {0x3f5,0x3f5,-1,-96}, 3437 {0x3f8,0x3fb,3,-1}, 3438 {0x430,0x44f,1,-32}, 3439 {0x450,0x45f,1,-80}, 3440 {0x461,0x481,2,-1}, 3441 {0x48b,0x4bf,2,-1}, 3442 {0x4c2,0x4ce,2,-1}, 3443 {0x4cf,0x4cf,-1,-15}, 3444 {0x4d1,0x52f,2,-1}, 3445 {0x561,0x586,1,-48}, 3446 {0x13f8,0x13fd,1,-8}, 3447 {0x1c80,0x1c80,-1,-6254}, 3448 {0x1c81,0x1c81,-1,-6253}, 3449 {0x1c82,0x1c82,-1,-6244}, 3450 {0x1c83,0x1c84,1,-6242}, 3451 {0x1c85,0x1c85,-1,-6243}, 3452 {0x1c86,0x1c86,-1,-6236}, 3453 {0x1c87,0x1c87,-1,-6181}, 3454 {0x1c88,0x1c88,-1,35266}, 3455 {0x1d79,0x1d79,-1,35332}, 3456 {0x1d7d,0x1d7d,-1,3814}, 3457 {0x1e01,0x1e95,2,-1}, 3458 {0x1e9b,0x1e9b,-1,-59}, 3459 {0x1ea1,0x1eff,2,-1}, 3460 {0x1f00,0x1f07,1,8}, 3461 {0x1f10,0x1f15,1,8}, 3462 {0x1f20,0x1f27,1,8}, 3463 {0x1f30,0x1f37,1,8}, 3464 {0x1f40,0x1f45,1,8}, 3465 {0x1f51,0x1f57,2,8}, 3466 {0x1f60,0x1f67,1,8}, 3467 {0x1f70,0x1f71,1,74}, 3468 {0x1f72,0x1f75,1,86}, 3469 {0x1f76,0x1f77,1,100}, 3470 {0x1f78,0x1f79,1,128}, 3471 {0x1f7a,0x1f7b,1,112}, 3472 {0x1f7c,0x1f7d,1,126}, 3473 {0x1f80,0x1f87,1,8}, 3474 {0x1f90,0x1f97,1,8}, 3475 {0x1fa0,0x1fa7,1,8}, 3476 {0x1fb0,0x1fb1,1,8}, 3477 {0x1fb3,0x1fb3,-1,9}, 3478 {0x1fbe,0x1fbe,-1,-7205}, 3479 {0x1fc3,0x1fc3,-1,9}, 3480 {0x1fd0,0x1fd1,1,8}, 3481 {0x1fe0,0x1fe1,1,8}, 3482 {0x1fe5,0x1fe5,-1,7}, 3483 {0x1ff3,0x1ff3,-1,9}, 3484 {0x214e,0x214e,-1,-28}, 3485 {0x2170,0x217f,1,-16}, 3486 {0x2184,0x2184,-1,-1}, 3487 {0x24d0,0x24e9,1,-26}, 3488 {0x2c30,0x2c5e,1,-48}, 3489 {0x2c61,0x2c61,-1,-1}, 3490 {0x2c65,0x2c65,-1,-10795}, 3491 {0x2c66,0x2c66,-1,-10792}, 3492 {0x2c68,0x2c6c,2,-1}, 3493 {0x2c73,0x2c76,3,-1}, 3494 {0x2c81,0x2ce3,2,-1}, 3495 {0x2cec,0x2cee,2,-1}, 3496 {0x2cf3,0x2cf3,-1,-1}, 3497 {0x2d00,0x2d25,1,-7264}, 3498 {0x2d27,0x2d2d,6,-7264}, 3499 {0xa641,0xa66d,2,-1}, 3500 {0xa681,0xa69b,2,-1}, 3501 {0xa723,0xa72f,2,-1}, 3502 {0xa733,0xa76f,2,-1}, 3503 {0xa77a,0xa77c,2,-1}, 3504 {0xa77f,0xa787,2,-1}, 3505 {0xa78c,0xa791,5,-1}, 3506 {0xa793,0xa797,4,-1}, 3507 {0xa799,0xa7a9,2,-1}, 3508 {0xa7b5,0xa7b7,2,-1}, 3509 {0xab53,0xab53,-1,-928}, 3510 {0xab70,0xabbf,1,-38864}, 3511 {0xff41,0xff5a,1,-32}, 3512 {0x10428,0x1044f,1,-40}, 3513 {0x104d8,0x104fb,1,-40}, 3514 {0x10cc0,0x10cf2,1,-64}, 3515 {0x118c0,0x118df,1,-32}, 3516 {0x1e922,0x1e943,1,-34} 3517 }; 3518 3519 /* 3520 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3521 * simple case folding. 3522 */ 3523 int 3524 utf_toupper(int a) 3525 { 3526 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */ 3527 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3528 return TOUPPER_ASC(a); 3529 3530 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3531 /* If towupper() is available and handles Unicode, use it. */ 3532 if (!(cmp_flags & CMP_INTERNAL)) 3533 return towupper(a); 3534 #endif 3535 3536 /* For characters below 128 use locale sensitive toupper(). */ 3537 if (a < 128) 3538 return TOUPPER_LOC(a); 3539 3540 /* For any other characters use the above mapping table. */ 3541 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3542 } 3543 3544 int 3545 utf_islower(int a) 3546 { 3547 /* German sharp s is lower case but has no upper case equivalent. */ 3548 return (utf_toupper(a) != a) || a == 0xdf; 3549 } 3550 3551 /* 3552 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3553 * simple case folding. 3554 */ 3555 int 3556 utf_tolower(int a) 3557 { 3558 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */ 3559 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3560 return TOLOWER_ASC(a); 3561 3562 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3563 /* If towlower() is available and handles Unicode, use it. */ 3564 if (!(cmp_flags & CMP_INTERNAL)) 3565 return towlower(a); 3566 #endif 3567 3568 /* For characters below 128 use locale sensitive tolower(). */ 3569 if (a < 128) 3570 return TOLOWER_LOC(a); 3571 3572 /* For any other characters use the above mapping table. */ 3573 return utf_convert(a, toLower, (int)sizeof(toLower)); 3574 } 3575 3576 int 3577 utf_isupper(int a) 3578 { 3579 return (utf_tolower(a) != a); 3580 } 3581 3582 static int 3583 utf_strnicmp( 3584 char_u *s1, 3585 char_u *s2, 3586 size_t n1, 3587 size_t n2) 3588 { 3589 int c1, c2, cdiff; 3590 char_u buffer[6]; 3591 3592 for (;;) 3593 { 3594 c1 = utf_safe_read_char_adv(&s1, &n1); 3595 c2 = utf_safe_read_char_adv(&s2, &n2); 3596 3597 if (c1 <= 0 || c2 <= 0) 3598 break; 3599 3600 if (c1 == c2) 3601 continue; 3602 3603 cdiff = utf_fold(c1) - utf_fold(c2); 3604 if (cdiff != 0) 3605 return cdiff; 3606 } 3607 3608 /* some string ended or has an incomplete/illegal character sequence */ 3609 3610 if (c1 == 0 || c2 == 0) 3611 { 3612 /* some string ended. shorter string is smaller */ 3613 if (c1 == 0 && c2 == 0) 3614 return 0; 3615 return c1 == 0 ? -1 : 1; 3616 } 3617 3618 /* Continue with bytewise comparison to produce some result that 3619 * would make comparison operations involving this function transitive. 3620 * 3621 * If only one string had an error, comparison should be made with 3622 * folded version of the other string. In this case it is enough 3623 * to fold just one character to determine the result of comparison. */ 3624 3625 if (c1 != -1 && c2 == -1) 3626 { 3627 n1 = utf_char2bytes(utf_fold(c1), buffer); 3628 s1 = buffer; 3629 } 3630 else if (c2 != -1 && c1 == -1) 3631 { 3632 n2 = utf_char2bytes(utf_fold(c2), buffer); 3633 s2 = buffer; 3634 } 3635 3636 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3637 { 3638 cdiff = (int)(*s1) - (int)(*s2); 3639 if (cdiff != 0) 3640 return cdiff; 3641 3642 s1++; 3643 s2++; 3644 n1--; 3645 n2--; 3646 } 3647 3648 if (n1 > 0 && *s1 == NUL) 3649 n1 = 0; 3650 if (n2 > 0 && *s2 == NUL) 3651 n2 = 0; 3652 3653 if (n1 == 0 && n2 == 0) 3654 return 0; 3655 return n1 == 0 ? -1 : 1; 3656 } 3657 3658 /* 3659 * Version of strnicmp() that handles multi-byte characters. 3660 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3661 * probably use strnicmp(), because there are no ASCII characters in the 3662 * second byte. 3663 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3664 * two characters otherwise. 3665 */ 3666 int 3667 mb_strnicmp(char_u *s1, char_u *s2, size_t nn) 3668 { 3669 int i, l; 3670 int cdiff; 3671 int n = (int)nn; 3672 3673 if (enc_utf8) 3674 { 3675 return utf_strnicmp(s1, s2, nn, nn); 3676 } 3677 else 3678 { 3679 for (i = 0; i < n; i += l) 3680 { 3681 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */ 3682 return 0; 3683 3684 l = (*mb_ptr2len)(s1 + i); 3685 if (l <= 1) 3686 { 3687 /* Single byte: first check normally, then with ignore case. */ 3688 if (s1[i] != s2[i]) 3689 { 3690 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3691 if (cdiff != 0) 3692 return cdiff; 3693 } 3694 } 3695 else 3696 { 3697 /* For non-Unicode multi-byte don't ignore case. */ 3698 if (l > n - i) 3699 l = n - i; 3700 cdiff = STRNCMP(s1 + i, s2 + i, l); 3701 if (cdiff != 0) 3702 return cdiff; 3703 } 3704 } 3705 } 3706 return 0; 3707 } 3708 3709 /* 3710 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3711 * 'encoding' has been set to. 3712 */ 3713 void 3714 show_utf8(void) 3715 { 3716 int len; 3717 int rlen = 0; 3718 char_u *line; 3719 int clen; 3720 int i; 3721 3722 /* Get the byte length of the char under the cursor, including composing 3723 * characters. */ 3724 line = ml_get_cursor(); 3725 len = utfc_ptr2len(line); 3726 if (len == 0) 3727 { 3728 MSG("NUL"); 3729 return; 3730 } 3731 3732 clen = 0; 3733 for (i = 0; i < len; ++i) 3734 { 3735 if (clen == 0) 3736 { 3737 /* start of (composing) character, get its length */ 3738 if (i > 0) 3739 { 3740 STRCPY(IObuff + rlen, "+ "); 3741 rlen += 2; 3742 } 3743 clen = utf_ptr2len(line + i); 3744 } 3745 sprintf((char *)IObuff + rlen, "%02x ", 3746 (line[i] == NL) ? NUL : line[i]); /* NUL is stored as NL */ 3747 --clen; 3748 rlen += (int)STRLEN(IObuff + rlen); 3749 if (rlen > IOSIZE - 20) 3750 break; 3751 } 3752 3753 msg(IObuff); 3754 } 3755 3756 /* 3757 * mb_head_off() function pointer. 3758 * Return offset from "p" to the first byte of the character it points into. 3759 * If "p" points to the NUL at the end of the string return 0. 3760 * Returns 0 when already at the first byte of a character. 3761 */ 3762 int 3763 latin_head_off(char_u *base UNUSED, char_u *p UNUSED) 3764 { 3765 return 0; 3766 } 3767 3768 int 3769 dbcs_head_off(char_u *base, char_u *p) 3770 { 3771 char_u *q; 3772 3773 /* It can't be a trailing byte when not using DBCS, at the start of the 3774 * string or the previous byte can't start a double-byte. */ 3775 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3776 return 0; 3777 3778 /* This is slow: need to start at the base and go forward until the 3779 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */ 3780 q = base; 3781 while (q < p) 3782 q += dbcs_ptr2len(q); 3783 return (q == p) ? 0 : 1; 3784 } 3785 3786 /* 3787 * Special version of dbcs_head_off() that works for ScreenLines[], where 3788 * single-width DBCS_JPNU characters are stored separately. 3789 */ 3790 int 3791 dbcs_screen_head_off(char_u *base, char_u *p) 3792 { 3793 char_u *q; 3794 3795 /* It can't be a trailing byte when not using DBCS, at the start of the 3796 * string or the previous byte can't start a double-byte. 3797 * For euc-jp an 0x8e byte in the previous cell always means we have a 3798 * lead byte in the current cell. */ 3799 if (p <= base 3800 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3801 || MB_BYTE2LEN(p[-1]) == 1 3802 || *p == NUL) 3803 return 0; 3804 3805 /* This is slow: need to start at the base and go forward until the 3806 * byte we are looking for. Return 1 when we went past it, 0 otherwise. 3807 * For DBCS_JPNU look out for 0x8e, which means the second byte is not 3808 * stored as the next byte. */ 3809 q = base; 3810 while (q < p) 3811 { 3812 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3813 ++q; 3814 else 3815 q += dbcs_ptr2len(q); 3816 } 3817 return (q == p) ? 0 : 1; 3818 } 3819 3820 int 3821 utf_head_off(char_u *base, char_u *p) 3822 { 3823 char_u *q; 3824 char_u *s; 3825 int c; 3826 int len; 3827 #ifdef FEAT_ARABIC 3828 char_u *j; 3829 #endif 3830 3831 if (*p < 0x80) /* be quick for ASCII */ 3832 return 0; 3833 3834 /* Skip backwards over trailing bytes: 10xx.xxxx 3835 * Skip backwards again if on a composing char. */ 3836 for (q = p; ; --q) 3837 { 3838 /* Move s to the last byte of this char. */ 3839 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3840 ; 3841 /* Move q to the first byte of this char. */ 3842 while (q > base && (*q & 0xc0) == 0x80) 3843 --q; 3844 /* Check for illegal sequence. Do allow an illegal byte after where we 3845 * started. */ 3846 len = utf8len_tab[*q]; 3847 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3848 return 0; 3849 3850 if (q <= base) 3851 break; 3852 3853 c = utf_ptr2char(q); 3854 if (utf_iscomposing(c)) 3855 continue; 3856 3857 #ifdef FEAT_ARABIC 3858 if (arabic_maycombine(c)) 3859 { 3860 /* Advance to get a sneak-peak at the next char */ 3861 j = q; 3862 --j; 3863 /* Move j to the first byte of this char. */ 3864 while (j > base && (*j & 0xc0) == 0x80) 3865 --j; 3866 if (arabic_combine(utf_ptr2char(j), c)) 3867 continue; 3868 } 3869 #endif 3870 break; 3871 } 3872 3873 return (int)(p - q); 3874 } 3875 3876 /* 3877 * Copy a character from "*fp" to "*tp" and advance the pointers. 3878 */ 3879 void 3880 mb_copy_char(char_u **fp, char_u **tp) 3881 { 3882 int l = (*mb_ptr2len)(*fp); 3883 3884 mch_memmove(*tp, *fp, (size_t)l); 3885 *tp += l; 3886 *fp += l; 3887 } 3888 3889 /* 3890 * Return the offset from "p" to the first byte of a character. When "p" is 3891 * at the start of a character 0 is returned, otherwise the offset to the next 3892 * character. Can start anywhere in a stream of bytes. 3893 */ 3894 int 3895 mb_off_next(char_u *base, char_u *p) 3896 { 3897 int i; 3898 int j; 3899 3900 if (enc_utf8) 3901 { 3902 if (*p < 0x80) /* be quick for ASCII */ 3903 return 0; 3904 3905 /* Find the next character that isn't 10xx.xxxx */ 3906 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 3907 ; 3908 if (i > 0) 3909 { 3910 /* Check for illegal sequence. */ 3911 for (j = 0; p - j > base; ++j) 3912 if ((p[-j] & 0xc0) != 0x80) 3913 break; 3914 if (utf8len_tab[p[-j]] != i + j) 3915 return 0; 3916 } 3917 return i; 3918 } 3919 3920 /* Only need to check if we're on a trail byte, it doesn't matter if we 3921 * want the offset to the next or current character. */ 3922 return (*mb_head_off)(base, p); 3923 } 3924 3925 /* 3926 * Return the offset from "p" to the last byte of the character it points 3927 * into. Can start anywhere in a stream of bytes. 3928 */ 3929 int 3930 mb_tail_off(char_u *base, char_u *p) 3931 { 3932 int i; 3933 int j; 3934 3935 if (*p == NUL) 3936 return 0; 3937 3938 if (enc_utf8) 3939 { 3940 /* Find the last character that is 10xx.xxxx */ 3941 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 3942 ; 3943 /* Check for illegal sequence. */ 3944 for (j = 0; p - j > base; ++j) 3945 if ((p[-j] & 0xc0) != 0x80) 3946 break; 3947 if (utf8len_tab[p[-j]] != i + j + 1) 3948 return 0; 3949 return i; 3950 } 3951 3952 /* It can't be the first byte if a double-byte when not using DBCS, at the 3953 * end of the string or the byte can't start a double-byte. */ 3954 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 3955 return 0; 3956 3957 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3958 return 1 - dbcs_head_off(base, p); 3959 } 3960 3961 /* 3962 * Find the next illegal byte sequence. 3963 */ 3964 void 3965 utf_find_illegal(void) 3966 { 3967 pos_T pos = curwin->w_cursor; 3968 char_u *p; 3969 int len; 3970 vimconv_T vimconv; 3971 char_u *tofree = NULL; 3972 3973 vimconv.vc_type = CONV_NONE; 3974 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 3975 { 3976 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 3977 * possibly a utf-8 file with illegal bytes. Setup for conversion 3978 * from utf-8 to 'fileencoding'. */ 3979 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 3980 } 3981 3982 #ifdef FEAT_VIRTUALEDIT 3983 curwin->w_cursor.coladd = 0; 3984 #endif 3985 for (;;) 3986 { 3987 p = ml_get_cursor(); 3988 if (vimconv.vc_type != CONV_NONE) 3989 { 3990 vim_free(tofree); 3991 tofree = string_convert(&vimconv, p, NULL); 3992 if (tofree == NULL) 3993 break; 3994 p = tofree; 3995 } 3996 3997 while (*p != NUL) 3998 { 3999 /* Illegal means that there are not enough trail bytes (checked by 4000 * utf_ptr2len()) or too many of them (overlong sequence). */ 4001 len = utf_ptr2len(p); 4002 if (*p >= 0x80 && (len == 1 4003 || utf_char2len(utf_ptr2char(p)) != len)) 4004 { 4005 if (vimconv.vc_type == CONV_NONE) 4006 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 4007 else 4008 { 4009 int l; 4010 4011 len = (int)(p - tofree); 4012 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 4013 { 4014 l = utf_ptr2len(p); 4015 curwin->w_cursor.col += l; 4016 } 4017 } 4018 goto theend; 4019 } 4020 p += len; 4021 } 4022 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 4023 break; 4024 ++curwin->w_cursor.lnum; 4025 curwin->w_cursor.col = 0; 4026 } 4027 4028 /* didn't find it: don't move and beep */ 4029 curwin->w_cursor = pos; 4030 beep_flush(); 4031 4032 theend: 4033 vim_free(tofree); 4034 convert_setup(&vimconv, NULL, NULL); 4035 } 4036 4037 #if defined(FEAT_GUI_GTK) || defined(PROTO) 4038 /* 4039 * Return TRUE if string "s" is a valid utf-8 string. 4040 * When "end" is NULL stop at the first NUL. 4041 * When "end" is positive stop there. 4042 */ 4043 int 4044 utf_valid_string(char_u *s, char_u *end) 4045 { 4046 int l; 4047 char_u *p = s; 4048 4049 while (end == NULL ? *p != NUL : p < end) 4050 { 4051 l = utf8len_tab_zero[*p]; 4052 if (l == 0) 4053 return FALSE; /* invalid lead byte */ 4054 if (end != NULL && p + l > end) 4055 return FALSE; /* incomplete byte sequence */ 4056 ++p; 4057 while (--l > 0) 4058 if ((*p++ & 0xc0) != 0x80) 4059 return FALSE; /* invalid trail byte */ 4060 } 4061 return TRUE; 4062 } 4063 #endif 4064 4065 #if defined(FEAT_GUI) || defined(PROTO) 4066 /* 4067 * Special version of mb_tail_off() for use in ScreenLines[]. 4068 */ 4069 int 4070 dbcs_screen_tail_off(char_u *base, char_u *p) 4071 { 4072 /* It can't be the first byte if a double-byte when not using DBCS, at the 4073 * end of the string or the byte can't start a double-byte. 4074 * For euc-jp an 0x8e byte always means we have a lead byte in the current 4075 * cell. */ 4076 if (*p == NUL || p[1] == NUL 4077 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 4078 || MB_BYTE2LEN(*p) == 1) 4079 return 0; 4080 4081 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 4082 return 1 - dbcs_screen_head_off(base, p); 4083 } 4084 #endif 4085 4086 /* 4087 * If the cursor moves on an trail byte, set the cursor on the lead byte. 4088 * Thus it moves left if necessary. 4089 * Return TRUE when the cursor was adjusted. 4090 */ 4091 void 4092 mb_adjust_cursor(void) 4093 { 4094 mb_adjustpos(curbuf, &curwin->w_cursor); 4095 } 4096 4097 /* 4098 * Adjust position "*lp" to point to the first byte of a multi-byte character. 4099 * If it points to a tail byte it's moved backwards to the head byte. 4100 */ 4101 void 4102 mb_adjustpos(buf_T *buf, pos_T *lp) 4103 { 4104 char_u *p; 4105 4106 if (lp->col > 0 4107 #ifdef FEAT_VIRTUALEDIT 4108 || lp->coladd > 1 4109 #endif 4110 ) 4111 { 4112 p = ml_get_buf(buf, lp->lnum, FALSE); 4113 if (*p == NUL || (int)STRLEN(p) < lp->col) 4114 lp->col = 0; 4115 else 4116 lp->col -= (*mb_head_off)(p, p + lp->col); 4117 #ifdef FEAT_VIRTUALEDIT 4118 /* Reset "coladd" when the cursor would be on the right half of a 4119 * double-wide character. */ 4120 if (lp->coladd == 1 4121 && p[lp->col] != TAB 4122 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 4123 && ptr2cells(p + lp->col) > 1) 4124 lp->coladd = 0; 4125 #endif 4126 } 4127 } 4128 4129 /* 4130 * Return a pointer to the character before "*p", if there is one. 4131 */ 4132 char_u * 4133 mb_prevptr( 4134 char_u *line, /* start of the string */ 4135 char_u *p) 4136 { 4137 if (p > line) 4138 MB_PTR_BACK(line, p); 4139 return p; 4140 } 4141 4142 /* 4143 * Return the character length of "str". Each multi-byte character (with 4144 * following composing characters) counts as one. 4145 */ 4146 int 4147 mb_charlen(char_u *str) 4148 { 4149 char_u *p = str; 4150 int count; 4151 4152 if (p == NULL) 4153 return 0; 4154 4155 for (count = 0; *p != NUL; count++) 4156 p += (*mb_ptr2len)(p); 4157 4158 return count; 4159 } 4160 4161 #if defined(FEAT_SPELL) || defined(PROTO) 4162 /* 4163 * Like mb_charlen() but for a string with specified length. 4164 */ 4165 int 4166 mb_charlen_len(char_u *str, int len) 4167 { 4168 char_u *p = str; 4169 int count; 4170 4171 for (count = 0; *p != NUL && p < str + len; count++) 4172 p += (*mb_ptr2len)(p); 4173 4174 return count; 4175 } 4176 #endif 4177 4178 /* 4179 * Try to un-escape a multi-byte character. 4180 * Used for the "to" and "from" part of a mapping. 4181 * Return the un-escaped string if it is a multi-byte character, and advance 4182 * "pp" to just after the bytes that formed it. 4183 * Return NULL if no multi-byte char was found. 4184 */ 4185 char_u * 4186 mb_unescape(char_u **pp) 4187 { 4188 static char_u buf[6]; 4189 int n; 4190 int m = 0; 4191 char_u *str = *pp; 4192 4193 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 4194 * KS_EXTRA KE_CSI to CSI. 4195 * Maximum length of a utf-8 character is 4 bytes. */ 4196 for (n = 0; str[n] != NUL && m < 4; ++n) 4197 { 4198 if (str[n] == K_SPECIAL 4199 && str[n + 1] == KS_SPECIAL 4200 && str[n + 2] == KE_FILLER) 4201 { 4202 buf[m++] = K_SPECIAL; 4203 n += 2; 4204 } 4205 else if ((str[n] == K_SPECIAL 4206 # ifdef FEAT_GUI 4207 || str[n] == CSI 4208 # endif 4209 ) 4210 && str[n + 1] == KS_EXTRA 4211 && str[n + 2] == (int)KE_CSI) 4212 { 4213 buf[m++] = CSI; 4214 n += 2; 4215 } 4216 else if (str[n] == K_SPECIAL 4217 # ifdef FEAT_GUI 4218 || str[n] == CSI 4219 # endif 4220 ) 4221 break; /* a special key can't be a multibyte char */ 4222 else 4223 buf[m++] = str[n]; 4224 buf[m] = NUL; 4225 4226 /* Return a multi-byte character if it's found. An illegal sequence 4227 * will result in a 1 here. */ 4228 if ((*mb_ptr2len)(buf) > 1) 4229 { 4230 *pp = str + n + 1; 4231 return buf; 4232 } 4233 4234 /* Bail out quickly for ASCII. */ 4235 if (buf[0] < 128) 4236 break; 4237 } 4238 return NULL; 4239 } 4240 4241 /* 4242 * Return TRUE if the character at "row"/"col" on the screen is the left side 4243 * of a double-width character. 4244 * Caller must make sure "row" and "col" are not invalid! 4245 */ 4246 int 4247 mb_lefthalve(int row, int col) 4248 { 4249 #ifdef FEAT_HANGULIN 4250 if (composing_hangul) 4251 return TRUE; 4252 #endif 4253 return (*mb_off2cells)(LineOffset[row] + col, 4254 LineOffset[row] + screen_Columns) > 1; 4255 } 4256 4257 /* 4258 * Correct a position on the screen, if it's the right half of a double-wide 4259 * char move it to the left half. Returns the corrected column. 4260 */ 4261 int 4262 mb_fix_col(int col, int row) 4263 { 4264 col = check_col(col); 4265 row = check_row(row); 4266 if (has_mbyte && ScreenLines != NULL && col > 0 4267 && ((enc_dbcs 4268 && ScreenLines[LineOffset[row] + col] != NUL 4269 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 4270 ScreenLines + LineOffset[row] + col)) 4271 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0))) 4272 return col - 1; 4273 return col; 4274 } 4275 #endif 4276 4277 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 4278 static int enc_alias_search(char_u *name); 4279 4280 /* 4281 * Skip the Vim specific head of a 'encoding' name. 4282 */ 4283 char_u * 4284 enc_skip(char_u *p) 4285 { 4286 if (STRNCMP(p, "2byte-", 6) == 0) 4287 return p + 6; 4288 if (STRNCMP(p, "8bit-", 5) == 0) 4289 return p + 5; 4290 return p; 4291 } 4292 4293 /* 4294 * Find the canonical name for encoding "enc". 4295 * When the name isn't recognized, returns "enc" itself, but with all lower 4296 * case characters and '_' replaced with '-'. 4297 * Returns an allocated string. NULL for out-of-memory. 4298 */ 4299 char_u * 4300 enc_canonize(char_u *enc) 4301 { 4302 char_u *r; 4303 char_u *p, *s; 4304 int i; 4305 4306 # ifdef FEAT_MBYTE 4307 if (STRCMP(enc, "default") == 0) 4308 { 4309 /* Use the default encoding as it's found by set_init_1(). */ 4310 r = get_encoding_default(); 4311 if (r == NULL) 4312 r = (char_u *)"latin1"; 4313 return vim_strsave(r); 4314 } 4315 # endif 4316 4317 /* copy "enc" to allocated memory, with room for two '-' */ 4318 r = alloc((unsigned)(STRLEN(enc) + 3)); 4319 if (r != NULL) 4320 { 4321 /* Make it all lower case and replace '_' with '-'. */ 4322 p = r; 4323 for (s = enc; *s != NUL; ++s) 4324 { 4325 if (*s == '_') 4326 *p++ = '-'; 4327 else 4328 *p++ = TOLOWER_ASC(*s); 4329 } 4330 *p = NUL; 4331 4332 /* Skip "2byte-" and "8bit-". */ 4333 p = enc_skip(r); 4334 4335 /* Change "microsoft-cp" to "cp". Used in some spell files. */ 4336 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4337 STRMOVE(p, p + 10); 4338 4339 /* "iso8859" -> "iso-8859" */ 4340 if (STRNCMP(p, "iso8859", 7) == 0) 4341 { 4342 STRMOVE(p + 4, p + 3); 4343 p[3] = '-'; 4344 } 4345 4346 /* "iso-8859n" -> "iso-8859-n" */ 4347 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4348 { 4349 STRMOVE(p + 9, p + 8); 4350 p[8] = '-'; 4351 } 4352 4353 /* "latin-N" -> "latinN" */ 4354 if (STRNCMP(p, "latin-", 6) == 0) 4355 STRMOVE(p + 5, p + 6); 4356 4357 if (enc_canon_search(p) >= 0) 4358 { 4359 /* canonical name can be used unmodified */ 4360 if (p != r) 4361 STRMOVE(r, p); 4362 } 4363 else if ((i = enc_alias_search(p)) >= 0) 4364 { 4365 /* alias recognized, get canonical name */ 4366 vim_free(r); 4367 r = vim_strsave((char_u *)enc_canon_table[i].name); 4368 } 4369 } 4370 return r; 4371 } 4372 4373 /* 4374 * Search for an encoding alias of "name". 4375 * Returns -1 when not found. 4376 */ 4377 static int 4378 enc_alias_search(char_u *name) 4379 { 4380 int i; 4381 4382 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4383 if (STRCMP(name, enc_alias_table[i].name) == 0) 4384 return enc_alias_table[i].canon; 4385 return -1; 4386 } 4387 #endif 4388 4389 #if defined(FEAT_MBYTE) || defined(PROTO) 4390 4391 # ifdef HAVE_LANGINFO_H 4392 # include <langinfo.h> 4393 # endif 4394 4395 # ifndef FEAT_GUI_W32 4396 /* 4397 * Get the canonicalized encoding from the specified locale string "locale" 4398 * or from the environment variables LC_ALL, LC_CTYPE and LANG. 4399 * Returns an allocated string when successful, NULL when not. 4400 */ 4401 char_u * 4402 enc_locale_env(char *locale) 4403 { 4404 char *s = locale; 4405 char *p; 4406 int i; 4407 char buf[50]; 4408 4409 if (s == NULL || *s == NUL) 4410 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4411 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4412 s = getenv("LANG"); 4413 4414 if (s == NULL || *s == NUL) 4415 return NULL; 4416 4417 /* The most generic locale format is: 4418 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4419 * If there is a '.' remove the part before it. 4420 * if there is something after the codeset, remove it. 4421 * Make the name lowercase and replace '_' with '-'. 4422 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4423 * "ko_KR.EUC" == "euc-kr" 4424 */ 4425 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4426 { 4427 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4428 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4429 { 4430 /* copy "XY.EUC" to "euc-XY" to buf[10] */ 4431 STRCPY(buf + 10, "euc-"); 4432 buf[14] = p[-2]; 4433 buf[15] = p[-1]; 4434 buf[16] = 0; 4435 s = buf + 10; 4436 } 4437 else 4438 s = p + 1; 4439 } 4440 for (i = 0; i < (int)sizeof(buf) - 1 && s[i] != NUL; ++i) 4441 { 4442 if (s[i] == '_' || s[i] == '-') 4443 buf[i] = '-'; 4444 else if (isalnum((int)s[i])) 4445 buf[i] = TOLOWER_ASC(s[i]); 4446 else 4447 break; 4448 } 4449 buf[i] = NUL; 4450 4451 return enc_canonize((char_u *)buf); 4452 } 4453 # endif 4454 4455 /* 4456 * Get the canonicalized encoding of the current locale. 4457 * Returns an allocated string when successful, NULL when not. 4458 */ 4459 char_u * 4460 enc_locale(void) 4461 { 4462 # ifdef WIN3264 4463 char buf[50]; 4464 long acp = GetACP(); 4465 4466 if (acp == 1200) 4467 STRCPY(buf, "ucs-2le"); 4468 else if (acp == 1252) /* cp1252 is used as latin1 */ 4469 STRCPY(buf, "latin1"); 4470 else 4471 sprintf(buf, "cp%ld", acp); 4472 4473 return enc_canonize((char_u *)buf); 4474 # else 4475 char *s; 4476 4477 # ifdef HAVE_NL_LANGINFO_CODESET 4478 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4479 # endif 4480 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4481 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4482 # endif 4483 s = NULL; 4484 4485 return enc_locale_env(s); 4486 # endif 4487 } 4488 4489 # if defined(WIN3264) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4490 /* 4491 * Convert an encoding name to an MS-Windows codepage. 4492 * Returns zero if no codepage can be figured out. 4493 */ 4494 int 4495 encname2codepage(char_u *name) 4496 { 4497 int cp; 4498 char_u *p = name; 4499 int idx; 4500 4501 if (STRNCMP(p, "8bit-", 5) == 0) 4502 p += 5; 4503 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4504 p += 6; 4505 4506 if (p[0] == 'c' && p[1] == 'p') 4507 cp = atoi((char *)p + 2); 4508 else if ((idx = enc_canon_search(p)) >= 0) 4509 cp = enc_canon_table[idx].codepage; 4510 else 4511 return 0; 4512 if (IsValidCodePage(cp)) 4513 return cp; 4514 return 0; 4515 } 4516 # endif 4517 4518 # if defined(USE_ICONV) || defined(PROTO) 4519 4520 static char_u *iconv_string(vimconv_T *vcp, char_u *str, int slen, int *unconvlenp, int *resultlenp); 4521 4522 /* 4523 * Call iconv_open() with a check if iconv() works properly (there are broken 4524 * versions). 4525 * Returns (void *)-1 if failed. 4526 * (should return iconv_t, but that causes problems with prototypes). 4527 */ 4528 void * 4529 my_iconv_open(char_u *to, char_u *from) 4530 { 4531 iconv_t fd; 4532 #define ICONV_TESTLEN 400 4533 char_u tobuf[ICONV_TESTLEN]; 4534 char *p; 4535 size_t tolen; 4536 static int iconv_ok = -1; 4537 4538 if (iconv_ok == FALSE) 4539 return (void *)-1; /* detected a broken iconv() previously */ 4540 4541 #ifdef DYNAMIC_ICONV 4542 /* Check if the iconv.dll can be found. */ 4543 if (!iconv_enabled(TRUE)) 4544 return (void *)-1; 4545 #endif 4546 4547 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4548 4549 if (fd != (iconv_t)-1 && iconv_ok == -1) 4550 { 4551 /* 4552 * Do a dummy iconv() call to check if it actually works. There is a 4553 * version of iconv() on Linux that is broken. We can't ignore it, 4554 * because it's wide-spread. The symptoms are that after outputting 4555 * the initial shift state the "to" pointer is NULL and conversion 4556 * stops for no apparent reason after about 8160 characters. 4557 */ 4558 p = (char *)tobuf; 4559 tolen = ICONV_TESTLEN; 4560 (void)iconv(fd, NULL, NULL, &p, &tolen); 4561 if (p == NULL) 4562 { 4563 iconv_ok = FALSE; 4564 iconv_close(fd); 4565 fd = (iconv_t)-1; 4566 } 4567 else 4568 iconv_ok = TRUE; 4569 } 4570 4571 return (void *)fd; 4572 } 4573 4574 /* 4575 * Convert the string "str[slen]" with iconv(). 4576 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4577 * sequence and set "*unconvlenp" to the length of it. 4578 * Returns the converted string in allocated memory. NULL for an error. 4579 * If resultlenp is not NULL, sets it to the result length in bytes. 4580 */ 4581 static char_u * 4582 iconv_string( 4583 vimconv_T *vcp, 4584 char_u *str, 4585 int slen, 4586 int *unconvlenp, 4587 int *resultlenp) 4588 { 4589 const char *from; 4590 size_t fromlen; 4591 char *to; 4592 size_t tolen; 4593 size_t len = 0; 4594 size_t done = 0; 4595 char_u *result = NULL; 4596 char_u *p; 4597 int l; 4598 4599 from = (char *)str; 4600 fromlen = slen; 4601 for (;;) 4602 { 4603 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4604 { 4605 /* Allocate enough room for most conversions. When re-allocating 4606 * increase the buffer size. */ 4607 len = len + fromlen * 2 + 40; 4608 p = alloc((unsigned)len); 4609 if (p != NULL && done > 0) 4610 mch_memmove(p, result, done); 4611 vim_free(result); 4612 result = p; 4613 if (result == NULL) /* out of memory */ 4614 break; 4615 } 4616 4617 to = (char *)result + done; 4618 tolen = len - done - 2; 4619 /* Avoid a warning for systems with a wrong iconv() prototype by 4620 * casting the second argument to void *. */ 4621 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4622 != (size_t)-1) 4623 { 4624 /* Finished, append a NUL. */ 4625 *to = NUL; 4626 break; 4627 } 4628 4629 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4630 * iconv library may use one of them. */ 4631 if (!vcp->vc_fail && unconvlenp != NULL 4632 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4633 { 4634 /* Handle an incomplete sequence at the end. */ 4635 *to = NUL; 4636 *unconvlenp = (int)fromlen; 4637 break; 4638 } 4639 4640 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4641 * iconv library may use one of them. */ 4642 else if (!vcp->vc_fail 4643 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4644 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4645 { 4646 /* Can't convert: insert a '?' and skip a character. This assumes 4647 * conversion from 'encoding' to something else. In other 4648 * situations we don't know what to skip anyway. */ 4649 *to++ = '?'; 4650 if ((*mb_ptr2cells)((char_u *)from) > 1) 4651 *to++ = '?'; 4652 if (enc_utf8) 4653 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4654 else 4655 { 4656 l = (*mb_ptr2len)((char_u *)from); 4657 if (l > (int)fromlen) 4658 l = (int)fromlen; 4659 } 4660 from += l; 4661 fromlen -= l; 4662 } 4663 else if (ICONV_ERRNO != ICONV_E2BIG) 4664 { 4665 /* conversion failed */ 4666 VIM_CLEAR(result); 4667 break; 4668 } 4669 /* Not enough room or skipping illegal sequence. */ 4670 done = to - (char *)result; 4671 } 4672 4673 if (resultlenp != NULL && result != NULL) 4674 *resultlenp = (int)(to - (char *)result); 4675 return result; 4676 } 4677 4678 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4679 /* 4680 * Dynamically load the "iconv.dll" on Win32. 4681 */ 4682 4683 # ifndef DYNAMIC_ICONV /* must be generating prototypes */ 4684 # define HINSTANCE int 4685 # endif 4686 static HINSTANCE hIconvDLL = 0; 4687 static HINSTANCE hMsvcrtDLL = 0; 4688 4689 # ifndef DYNAMIC_ICONV_DLL 4690 # define DYNAMIC_ICONV_DLL "iconv.dll" 4691 # define DYNAMIC_ICONV_DLL_ALT1 "libiconv.dll" 4692 # define DYNAMIC_ICONV_DLL_ALT2 "libiconv2.dll" 4693 # define DYNAMIC_ICONV_DLL_ALT3 "libiconv-2.dll" 4694 # endif 4695 # ifndef DYNAMIC_MSVCRT_DLL 4696 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4697 # endif 4698 4699 /* 4700 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4701 */ 4702 int 4703 iconv_enabled(int verbose) 4704 { 4705 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4706 return TRUE; 4707 4708 /* The iconv DLL file goes under different names, try them all. 4709 * Do the "2" version first, it's newer. */ 4710 #ifdef DYNAMIC_ICONV_DLL_ALT2 4711 if (hIconvDLL == 0) 4712 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT2); 4713 #endif 4714 #ifdef DYNAMIC_ICONV_DLL_ALT3 4715 if (hIconvDLL == 0) 4716 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT3); 4717 #endif 4718 if (hIconvDLL == 0) 4719 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4720 #ifdef DYNAMIC_ICONV_DLL_ALT1 4721 if (hIconvDLL == 0) 4722 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT1); 4723 #endif 4724 4725 if (hIconvDLL != 0) 4726 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4727 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4728 { 4729 /* Only give the message when 'verbose' is set, otherwise it might be 4730 * done whenever a conversion is attempted. */ 4731 if (verbose && p_verbose > 0) 4732 { 4733 verbose_enter(); 4734 EMSG2(_(e_loadlib), 4735 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4736 verbose_leave(); 4737 } 4738 iconv_end(); 4739 return FALSE; 4740 } 4741 4742 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4743 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4744 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4745 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4746 iconv_errno = get_dll_import_func(hIconvDLL, "_errno"); 4747 if (iconv_errno == NULL) 4748 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4749 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4750 || iconvctl == NULL || iconv_errno == NULL) 4751 { 4752 iconv_end(); 4753 if (verbose && p_verbose > 0) 4754 { 4755 verbose_enter(); 4756 EMSG2(_(e_loadfunc), "for libiconv"); 4757 verbose_leave(); 4758 } 4759 return FALSE; 4760 } 4761 return TRUE; 4762 } 4763 4764 void 4765 iconv_end(void) 4766 { 4767 /* Don't use iconv() when inputting or outputting characters. */ 4768 if (input_conv.vc_type == CONV_ICONV) 4769 convert_setup(&input_conv, NULL, NULL); 4770 if (output_conv.vc_type == CONV_ICONV) 4771 convert_setup(&output_conv, NULL, NULL); 4772 4773 if (hIconvDLL != 0) 4774 FreeLibrary(hIconvDLL); 4775 if (hMsvcrtDLL != 0) 4776 FreeLibrary(hMsvcrtDLL); 4777 hIconvDLL = 0; 4778 hMsvcrtDLL = 0; 4779 } 4780 # endif /* DYNAMIC_ICONV */ 4781 # endif /* USE_ICONV */ 4782 4783 #endif /* FEAT_MBYTE */ 4784 4785 #ifdef FEAT_GUI 4786 # define USE_IMACTIVATEFUNC (!gui.in_use && *p_imaf != NUL) 4787 # define USE_IMSTATUSFUNC (!gui.in_use && *p_imsf != NUL) 4788 #else 4789 # define USE_IMACTIVATEFUNC (*p_imaf != NUL) 4790 # define USE_IMSTATUSFUNC (*p_imsf != NUL) 4791 #endif 4792 4793 #if defined(FEAT_EVAL) && defined(FEAT_MBYTE) \ 4794 && (defined(FEAT_XIM) || defined(IME_WITHOUT_XIM)) 4795 static void 4796 call_imactivatefunc(int active) 4797 { 4798 char_u *argv[1]; 4799 4800 if (active) 4801 argv[0] = (char_u *)"1"; 4802 else 4803 argv[0] = (char_u *)"0"; 4804 (void)call_func_retnr(p_imaf, 1, argv, FALSE); 4805 } 4806 4807 static int 4808 call_imstatusfunc(void) 4809 { 4810 int is_active; 4811 4812 /* FIXME: Don't execute user function in unsafe situation. */ 4813 if (exiting || is_autocmd_blocked()) 4814 return FALSE; 4815 /* FIXME: :py print 'xxx' is shown duplicate result. 4816 * Use silent to avoid it. */ 4817 ++msg_silent; 4818 is_active = call_func_retnr(p_imsf, 0, NULL, FALSE); 4819 --msg_silent; 4820 return (is_active > 0); 4821 } 4822 #endif 4823 4824 #if defined(FEAT_XIM) || defined(PROTO) 4825 4826 # if defined(FEAT_GUI_GTK) || defined(PROTO) 4827 static int xim_has_preediting INIT(= FALSE); /* IM current status */ 4828 4829 /* 4830 * Set preedit_start_col to the current cursor position. 4831 */ 4832 static void 4833 init_preedit_start_col(void) 4834 { 4835 if (State & CMDLINE) 4836 preedit_start_col = cmdline_getvcol_cursor(); 4837 else if (curwin != NULL && curwin->w_buffer != NULL) 4838 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); 4839 /* Prevent that preediting marks the buffer as changed. */ 4840 xim_changed_while_preediting = curbuf->b_changed; 4841 } 4842 4843 static int im_is_active = FALSE; /* IM is enabled for current mode */ 4844 static int preedit_is_active = FALSE; 4845 static int im_preedit_cursor = 0; /* cursor offset in characters */ 4846 static int im_preedit_trailing = 0; /* number of characters after cursor */ 4847 4848 static unsigned long im_commit_handler_id = 0; 4849 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; 4850 static unsigned int im_activatekey_state = 0; 4851 4852 static GtkWidget *preedit_window = NULL; 4853 static GtkWidget *preedit_label = NULL; 4854 4855 static void im_preedit_window_set_position(void); 4856 4857 void 4858 im_set_active(int active) 4859 { 4860 int was_active; 4861 4862 was_active = !!im_get_status(); 4863 im_is_active = (active && !p_imdisable); 4864 4865 if (im_is_active != was_active) 4866 xim_reset(); 4867 } 4868 4869 void 4870 xim_set_focus(int focus) 4871 { 4872 if (xic != NULL) 4873 { 4874 if (focus) 4875 gtk_im_context_focus_in(xic); 4876 else 4877 gtk_im_context_focus_out(xic); 4878 } 4879 } 4880 4881 void 4882 im_set_position(int row, int col) 4883 { 4884 if (xic != NULL) 4885 { 4886 GdkRectangle area; 4887 4888 area.x = FILL_X(col); 4889 area.y = FILL_Y(row); 4890 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); 4891 area.height = gui.char_height; 4892 4893 gtk_im_context_set_cursor_location(xic, &area); 4894 4895 if (p_imst == IM_OVER_THE_SPOT) 4896 im_preedit_window_set_position(); 4897 } 4898 } 4899 4900 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */ 4901 void 4902 xim_set_preedit(void) 4903 { 4904 im_set_position(gui.row, gui.col); 4905 } 4906 # endif 4907 4908 static void 4909 im_add_to_input(char_u *str, int len) 4910 { 4911 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */ 4912 if (input_conv.vc_type != CONV_NONE) 4913 { 4914 str = string_convert(&input_conv, str, &len); 4915 g_return_if_fail(str != NULL); 4916 } 4917 4918 add_to_input_buf_csi(str, len); 4919 4920 if (input_conv.vc_type != CONV_NONE) 4921 vim_free(str); 4922 4923 if (p_mh) /* blank out the pointer if necessary */ 4924 gui_mch_mousehide(TRUE); 4925 } 4926 4927 static void 4928 im_preedit_window_set_position(void) 4929 { 4930 int x, y, w, h, sw, sh; 4931 4932 if (preedit_window == NULL) 4933 return; 4934 4935 gui_gtk_get_screen_size_of_win(preedit_window, &sw, &sh); 4936 #if GTK_CHECK_VERSION(3,0,0) 4937 gdk_window_get_origin(gtk_widget_get_window(gui.drawarea), &x, &y); 4938 #else 4939 gdk_window_get_origin(gui.drawarea->window, &x, &y); 4940 #endif 4941 gtk_window_get_size(GTK_WINDOW(preedit_window), &w, &h); 4942 x = x + FILL_X(gui.col); 4943 y = y + FILL_Y(gui.row); 4944 if (x + w > sw) 4945 x = sw - w; 4946 if (y + h > sh) 4947 y = sh - h; 4948 gtk_window_move(GTK_WINDOW(preedit_window), x, y); 4949 } 4950 4951 static void 4952 im_preedit_window_open() 4953 { 4954 char *preedit_string; 4955 #if !GTK_CHECK_VERSION(3,16,0) 4956 char buf[8]; 4957 #endif 4958 PangoAttrList *attr_list; 4959 PangoLayout *layout; 4960 #if GTK_CHECK_VERSION(3,0,0) 4961 # if !GTK_CHECK_VERSION(3,16,0) 4962 GdkRGBA color; 4963 # endif 4964 #else 4965 GdkColor color; 4966 #endif 4967 gint w, h; 4968 4969 if (preedit_window == NULL) 4970 { 4971 preedit_window = gtk_window_new(GTK_WINDOW_POPUP); 4972 gtk_window_set_transient_for(GTK_WINDOW(preedit_window), 4973 GTK_WINDOW(gui.mainwin)); 4974 preedit_label = gtk_label_new(""); 4975 gtk_widget_set_name(preedit_label, "vim-gui-preedit-area"); 4976 gtk_container_add(GTK_CONTAINER(preedit_window), preedit_label); 4977 } 4978 4979 #if GTK_CHECK_VERSION(3,16,0) 4980 { 4981 GtkStyleContext * const context 4982 = gtk_widget_get_style_context(gui.drawarea); 4983 GtkCssProvider * const provider = gtk_css_provider_new(); 4984 gchar *css = NULL; 4985 const char * const fontname 4986 = pango_font_description_get_family(gui.norm_font); 4987 gint fontsize 4988 = pango_font_description_get_size(gui.norm_font) / PANGO_SCALE; 4989 gchar *fontsize_propval = NULL; 4990 4991 if (!pango_font_description_get_size_is_absolute(gui.norm_font)) 4992 { 4993 /* fontsize was given in points. Convert it into that in pixels 4994 * to use with CSS. */ 4995 GdkScreen * const screen 4996 = gdk_window_get_screen(gtk_widget_get_window(gui.mainwin)); 4997 const gdouble dpi = gdk_screen_get_resolution(screen); 4998 fontsize = dpi * fontsize / 72; 4999 } 5000 if (fontsize > 0) 5001 fontsize_propval = g_strdup_printf("%dpx", fontsize); 5002 else 5003 fontsize_propval = g_strdup_printf("inherit"); 5004 5005 css = g_strdup_printf( 5006 "widget#vim-gui-preedit-area {\n" 5007 " font-family: %s,monospace;\n" 5008 " font-size: %s;\n" 5009 " color: #%.2lx%.2lx%.2lx;\n" 5010 " background-color: #%.2lx%.2lx%.2lx;\n" 5011 "}\n", 5012 fontname != NULL ? fontname : "inherit", 5013 fontsize_propval, 5014 (gui.norm_pixel >> 16) & 0xff, 5015 (gui.norm_pixel >> 8) & 0xff, 5016 gui.norm_pixel & 0xff, 5017 (gui.back_pixel >> 16) & 0xff, 5018 (gui.back_pixel >> 8) & 0xff, 5019 gui.back_pixel & 0xff); 5020 5021 gtk_css_provider_load_from_data(provider, css, -1, NULL); 5022 gtk_style_context_add_provider(context, 5023 GTK_STYLE_PROVIDER(provider), G_MAXUINT); 5024 5025 g_free(css); 5026 g_free(fontsize_propval); 5027 g_object_unref(provider); 5028 } 5029 #elif GTK_CHECK_VERSION(3,0,0) 5030 gtk_widget_override_font(preedit_label, gui.norm_font); 5031 5032 vim_snprintf(buf, sizeof(buf), "#%06X", gui.norm_pixel); 5033 gdk_rgba_parse(&color, buf); 5034 gtk_widget_override_color(preedit_label, GTK_STATE_FLAG_NORMAL, &color); 5035 5036 vim_snprintf(buf, sizeof(buf), "#%06X", gui.back_pixel); 5037 gdk_rgba_parse(&color, buf); 5038 gtk_widget_override_background_color(preedit_label, GTK_STATE_FLAG_NORMAL, 5039 &color); 5040 #else 5041 gtk_widget_modify_font(preedit_label, gui.norm_font); 5042 5043 vim_snprintf(buf, sizeof(buf), "#%06X", gui.norm_pixel); 5044 gdk_color_parse(buf, &color); 5045 gtk_widget_modify_fg(preedit_label, GTK_STATE_NORMAL, &color); 5046 5047 vim_snprintf(buf, sizeof(buf), "#%06X", gui.back_pixel); 5048 gdk_color_parse(buf, &color); 5049 gtk_widget_modify_bg(preedit_window, GTK_STATE_NORMAL, &color); 5050 #endif 5051 5052 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5053 5054 if (preedit_string[0] != NUL) 5055 { 5056 gtk_label_set_text(GTK_LABEL(preedit_label), preedit_string); 5057 gtk_label_set_attributes(GTK_LABEL(preedit_label), attr_list); 5058 5059 layout = gtk_label_get_layout(GTK_LABEL(preedit_label)); 5060 pango_layout_get_pixel_size(layout, &w, &h); 5061 h = MAX(h, gui.char_height); 5062 gtk_window_resize(GTK_WINDOW(preedit_window), w, h); 5063 5064 gtk_widget_show_all(preedit_window); 5065 5066 im_preedit_window_set_position(); 5067 } 5068 5069 g_free(preedit_string); 5070 pango_attr_list_unref(attr_list); 5071 } 5072 5073 static void 5074 im_preedit_window_close() 5075 { 5076 if (preedit_window != NULL) 5077 gtk_widget_hide(preedit_window); 5078 } 5079 5080 static void 5081 im_show_preedit() 5082 { 5083 im_preedit_window_open(); 5084 5085 if (p_mh) /* blank out the pointer if necessary */ 5086 gui_mch_mousehide(TRUE); 5087 } 5088 5089 static void 5090 im_delete_preedit(void) 5091 { 5092 char_u bskey[] = {CSI, 'k', 'b'}; 5093 char_u delkey[] = {CSI, 'k', 'D'}; 5094 5095 if (p_imst == IM_OVER_THE_SPOT) 5096 { 5097 im_preedit_window_close(); 5098 return; 5099 } 5100 5101 if (State & NORMAL 5102 #ifdef FEAT_TERMINAL 5103 && !term_use_loop() 5104 #endif 5105 ) 5106 { 5107 im_preedit_cursor = 0; 5108 return; 5109 } 5110 for (; im_preedit_cursor > 0; --im_preedit_cursor) 5111 add_to_input_buf(bskey, (int)sizeof(bskey)); 5112 5113 for (; im_preedit_trailing > 0; --im_preedit_trailing) 5114 add_to_input_buf(delkey, (int)sizeof(delkey)); 5115 } 5116 5117 /* 5118 * Move the cursor left by "num_move_back" characters. 5119 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for 5120 * these K_LEFT keys. 5121 */ 5122 static void 5123 im_correct_cursor(int num_move_back) 5124 { 5125 char_u backkey[] = {CSI, 'k', 'l'}; 5126 5127 if (State & NORMAL) 5128 return; 5129 # ifdef FEAT_RIGHTLEFT 5130 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) 5131 backkey[2] = 'r'; 5132 # endif 5133 for (; num_move_back > 0; --num_move_back) 5134 add_to_input_buf(backkey, (int)sizeof(backkey)); 5135 } 5136 5137 static int xim_expected_char = NUL; 5138 static int xim_ignored_char = FALSE; 5139 5140 /* 5141 * Update the mode and cursor while in an IM callback. 5142 */ 5143 static void 5144 im_show_info(void) 5145 { 5146 int old_vgetc_busy; 5147 5148 old_vgetc_busy = vgetc_busy; 5149 vgetc_busy = TRUE; 5150 showmode(); 5151 vgetc_busy = old_vgetc_busy; 5152 if ((State & NORMAL) || (State & INSERT)) 5153 setcursor(); 5154 out_flush(); 5155 } 5156 5157 /* 5158 * Callback invoked when the user finished preediting. 5159 * Put the final string into the input buffer. 5160 */ 5161 static void 5162 im_commit_cb(GtkIMContext *context UNUSED, 5163 const gchar *str, 5164 gpointer data UNUSED) 5165 { 5166 int slen = (int)STRLEN(str); 5167 int add_to_input = TRUE; 5168 int clen; 5169 int len = slen; 5170 int commit_with_preedit = TRUE; 5171 char_u *im_str; 5172 5173 #ifdef XIM_DEBUG 5174 xim_log("im_commit_cb(): %s\n", str); 5175 #endif 5176 5177 if (p_imst == IM_ON_THE_SPOT) 5178 { 5179 /* The imhangul module doesn't reset the preedit string before 5180 * committing. Call im_delete_preedit() to work around that. */ 5181 im_delete_preedit(); 5182 5183 /* Indicate that preediting has finished. */ 5184 if (preedit_start_col == MAXCOL) 5185 { 5186 init_preedit_start_col(); 5187 commit_with_preedit = FALSE; 5188 } 5189 5190 /* The thing which setting "preedit_start_col" to MAXCOL means that 5191 * "preedit_start_col" will be set forcedly when calling 5192 * preedit_changed_cb() next time. 5193 * "preedit_start_col" should not reset with MAXCOL on this part. Vim 5194 * is simulating the preediting by using add_to_input_str(). when 5195 * preedit begin immediately before committed, the typebuf is not 5196 * flushed to screen, then it can't get correct "preedit_start_col". 5197 * Thus, it should calculate the cells by adding cells of the committed 5198 * string. */ 5199 if (input_conv.vc_type != CONV_NONE) 5200 { 5201 im_str = string_convert(&input_conv, (char_u *)str, &len); 5202 g_return_if_fail(im_str != NULL); 5203 } 5204 else 5205 im_str = (char_u *)str; 5206 5207 clen = mb_string2cells(im_str, len); 5208 5209 if (input_conv.vc_type != CONV_NONE) 5210 vim_free(im_str); 5211 preedit_start_col += clen; 5212 } 5213 5214 /* Is this a single character that matches a keypad key that's just 5215 * been pressed? If so, we don't want it to be entered as such - let 5216 * us carry on processing the raw keycode so that it may be used in 5217 * mappings as <kSomething>. */ 5218 if (xim_expected_char != NUL) 5219 { 5220 /* We're currently processing a keypad or other special key */ 5221 if (slen == 1 && str[0] == xim_expected_char) 5222 { 5223 /* It's a match - don't do it here */ 5224 xim_ignored_char = TRUE; 5225 add_to_input = FALSE; 5226 } 5227 else 5228 { 5229 /* Not a match */ 5230 xim_ignored_char = FALSE; 5231 } 5232 } 5233 5234 if (add_to_input) 5235 im_add_to_input((char_u *)str, slen); 5236 5237 if (p_imst == IM_ON_THE_SPOT) 5238 { 5239 /* Inserting chars while "im_is_active" is set does not cause a 5240 * change of buffer. When the chars are committed the buffer must be 5241 * marked as changed. */ 5242 if (!commit_with_preedit) 5243 preedit_start_col = MAXCOL; 5244 5245 /* This flag is used in changed() at next call. */ 5246 xim_changed_while_preediting = TRUE; 5247 } 5248 5249 if (gtk_main_level() > 0) 5250 gtk_main_quit(); 5251 } 5252 5253 /* 5254 * Callback invoked after start to the preedit. 5255 */ 5256 static void 5257 im_preedit_start_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 5258 { 5259 #ifdef XIM_DEBUG 5260 xim_log("im_preedit_start_cb()\n"); 5261 #endif 5262 5263 im_is_active = TRUE; 5264 preedit_is_active = TRUE; 5265 gui_update_cursor(TRUE, FALSE); 5266 im_show_info(); 5267 } 5268 5269 /* 5270 * Callback invoked after end to the preedit. 5271 */ 5272 static void 5273 im_preedit_end_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 5274 { 5275 #ifdef XIM_DEBUG 5276 xim_log("im_preedit_end_cb()\n"); 5277 #endif 5278 im_delete_preedit(); 5279 5280 /* Indicate that preediting has finished */ 5281 if (p_imst == IM_ON_THE_SPOT) 5282 preedit_start_col = MAXCOL; 5283 xim_has_preediting = FALSE; 5284 5285 #if 0 5286 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was 5287 * switched off unintentionally. We now use preedit_is_active (added by 5288 * SungHyun Nam). */ 5289 im_is_active = FALSE; 5290 #endif 5291 preedit_is_active = FALSE; 5292 gui_update_cursor(TRUE, FALSE); 5293 im_show_info(); 5294 } 5295 5296 /* 5297 * Callback invoked after changes to the preedit string. If the preedit 5298 * string was empty before, remember the preedit start column so we know 5299 * where to apply feedback attributes. Delete the previous preedit string 5300 * if there was one, save the new preedit cursor offset, and put the new 5301 * string into the input buffer. 5302 * 5303 * TODO: The pragmatic "put into input buffer" approach used here has 5304 * several fundamental problems: 5305 * 5306 * - The characters in the preedit string are subject to remapping. 5307 * That's broken, only the finally committed string should be remapped. 5308 * 5309 * - There is a race condition involved: The retrieved value for the 5310 * current cursor position will be wrong if any unprocessed characters 5311 * are still queued in the input buffer. 5312 * 5313 * - Due to the lack of synchronization between the file buffer in memory 5314 * and any typed characters, it's practically impossible to implement the 5315 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM 5316 * modules for languages such as Thai are likely to rely on this feature 5317 * for proper operation. 5318 * 5319 * Conclusions: I think support for preediting needs to be moved to the 5320 * core parts of Vim. Ideally, until it has been committed, the preediting 5321 * string should only be displayed and not affect the buffer content at all. 5322 * The question how to deal with the synchronization issue still remains. 5323 * Circumventing the input buffer is probably not desirable. Anyway, I think 5324 * implementing "retrieve_surrounding" is the only hard problem. 5325 * 5326 * One way to solve all of this in a clean manner would be to queue all key 5327 * press/release events "as is" in the input buffer, and apply the IM filtering 5328 * at the receiving end of the queue. This, however, would have a rather large 5329 * impact on the code base. If there is an easy way to force processing of all 5330 * remaining input from within the "retrieve_surrounding" signal handler, this 5331 * might not be necessary. Gotta ask on vim-dev for opinions. 5332 */ 5333 static void 5334 im_preedit_changed_cb(GtkIMContext *context, gpointer data UNUSED) 5335 { 5336 char *preedit_string = NULL; 5337 int cursor_index = 0; 5338 int num_move_back = 0; 5339 char_u *str; 5340 char_u *p; 5341 int i; 5342 5343 if (p_imst == IM_ON_THE_SPOT) 5344 gtk_im_context_get_preedit_string(context, 5345 &preedit_string, NULL, 5346 &cursor_index); 5347 else 5348 gtk_im_context_get_preedit_string(context, 5349 &preedit_string, NULL, 5350 NULL); 5351 5352 #ifdef XIM_DEBUG 5353 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); 5354 #endif 5355 5356 g_return_if_fail(preedit_string != NULL); /* just in case */ 5357 5358 if (p_imst == IM_OVER_THE_SPOT) 5359 { 5360 if (preedit_string[0] == NUL) 5361 { 5362 xim_has_preediting = FALSE; 5363 im_delete_preedit(); 5364 } 5365 else 5366 { 5367 xim_has_preediting = TRUE; 5368 im_show_preedit(); 5369 } 5370 } 5371 else 5372 { 5373 /* If preedit_start_col is MAXCOL set it to the current cursor position. */ 5374 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') 5375 { 5376 xim_has_preediting = TRUE; 5377 5378 /* Urgh, this breaks if the input buffer isn't empty now */ 5379 init_preedit_start_col(); 5380 } 5381 else if (cursor_index == 0 && preedit_string[0] == '\0') 5382 { 5383 xim_has_preediting = FALSE; 5384 5385 /* If at the start position (after typing backspace) 5386 * preedit_start_col must be reset. */ 5387 preedit_start_col = MAXCOL; 5388 } 5389 5390 im_delete_preedit(); 5391 5392 /* 5393 * Compute the end of the preediting area: "preedit_end_col". 5394 * According to the documentation of gtk_im_context_get_preedit_string(), 5395 * the cursor_pos output argument returns the offset in bytes. This is 5396 * unfortunately not true -- real life shows the offset is in characters, 5397 * and the GTK+ source code agrees with me. Will file a bug later. 5398 */ 5399 if (preedit_start_col != MAXCOL) 5400 preedit_end_col = preedit_start_col; 5401 str = (char_u *)preedit_string; 5402 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) 5403 { 5404 int is_composing; 5405 5406 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); 5407 /* 5408 * These offsets are used as counters when generating <BS> and <Del> 5409 * to delete the preedit string. So don't count composing characters 5410 * unless 'delcombine' is enabled. 5411 */ 5412 if (!is_composing || p_deco) 5413 { 5414 if (i < cursor_index) 5415 ++im_preedit_cursor; 5416 else 5417 ++im_preedit_trailing; 5418 } 5419 if (!is_composing && i >= cursor_index) 5420 { 5421 /* This is essentially the same as im_preedit_trailing, except 5422 * composing characters are not counted even if p_deco is set. */ 5423 ++num_move_back; 5424 } 5425 if (preedit_start_col != MAXCOL) 5426 preedit_end_col += utf_ptr2cells(p); 5427 } 5428 5429 if (p > str) 5430 { 5431 im_add_to_input(str, (int)(p - str)); 5432 im_correct_cursor(num_move_back); 5433 } 5434 } 5435 5436 g_free(preedit_string); 5437 5438 if (gtk_main_level() > 0) 5439 gtk_main_quit(); 5440 } 5441 5442 /* 5443 * Translate the Pango attributes at iter to Vim highlighting attributes. 5444 * Ignore attributes not supported by Vim highlighting. This shouldn't have 5445 * too much impact -- right now we handle even more attributes than necessary 5446 * for the IM modules I tested with. 5447 */ 5448 static int 5449 translate_pango_attributes(PangoAttrIterator *iter) 5450 { 5451 PangoAttribute *attr; 5452 int char_attr = HL_NORMAL; 5453 5454 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); 5455 if (attr != NULL && ((PangoAttrInt *)attr)->value 5456 != (int)PANGO_UNDERLINE_NONE) 5457 char_attr |= HL_UNDERLINE; 5458 5459 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); 5460 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) 5461 char_attr |= HL_BOLD; 5462 5463 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); 5464 if (attr != NULL && ((PangoAttrInt *)attr)->value 5465 != (int)PANGO_STYLE_NORMAL) 5466 char_attr |= HL_ITALIC; 5467 5468 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); 5469 if (attr != NULL) 5470 { 5471 const PangoColor *color = &((PangoAttrColor *)attr)->color; 5472 5473 /* Assume inverse if black background is requested */ 5474 if ((color->red | color->green | color->blue) == 0) 5475 char_attr |= HL_INVERSE; 5476 } 5477 5478 return char_attr; 5479 } 5480 5481 /* 5482 * Retrieve the highlighting attributes at column col in the preedit string. 5483 * Return -1 if not in preediting mode or if col is out of range. 5484 */ 5485 int 5486 im_get_feedback_attr(int col) 5487 { 5488 char *preedit_string = NULL; 5489 PangoAttrList *attr_list = NULL; 5490 int char_attr = -1; 5491 5492 if (xic == NULL) 5493 return char_attr; 5494 5495 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5496 5497 if (preedit_string != NULL && attr_list != NULL) 5498 { 5499 int idx; 5500 5501 /* Get the byte index as used by PangoAttrIterator */ 5502 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col) 5503 idx += utfc_ptr2len((char_u *)preedit_string + idx); 5504 5505 if (preedit_string[idx] != '\0') 5506 { 5507 PangoAttrIterator *iter; 5508 int start, end; 5509 5510 char_attr = HL_NORMAL; 5511 iter = pango_attr_list_get_iterator(attr_list); 5512 5513 /* Extract all relevant attributes from the list. */ 5514 do 5515 { 5516 pango_attr_iterator_range(iter, &start, &end); 5517 5518 if (idx >= start && idx < end) 5519 char_attr |= translate_pango_attributes(iter); 5520 } 5521 while (pango_attr_iterator_next(iter)); 5522 5523 pango_attr_iterator_destroy(iter); 5524 } 5525 } 5526 5527 if (attr_list != NULL) 5528 pango_attr_list_unref(attr_list); 5529 g_free(preedit_string); 5530 5531 return char_attr; 5532 } 5533 5534 void 5535 xim_init(void) 5536 { 5537 #ifdef XIM_DEBUG 5538 xim_log("xim_init()\n"); 5539 #endif 5540 5541 g_return_if_fail(gui.drawarea != NULL); 5542 #if GTK_CHECK_VERSION(3,0,0) 5543 g_return_if_fail(gtk_widget_get_window(gui.drawarea) != NULL); 5544 #else 5545 g_return_if_fail(gui.drawarea->window != NULL); 5546 #endif 5547 5548 xic = gtk_im_multicontext_new(); 5549 g_object_ref(xic); 5550 5551 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", 5552 G_CALLBACK(&im_commit_cb), NULL); 5553 g_signal_connect(G_OBJECT(xic), "preedit_changed", 5554 G_CALLBACK(&im_preedit_changed_cb), NULL); 5555 g_signal_connect(G_OBJECT(xic), "preedit_start", 5556 G_CALLBACK(&im_preedit_start_cb), NULL); 5557 g_signal_connect(G_OBJECT(xic), "preedit_end", 5558 G_CALLBACK(&im_preedit_end_cb), NULL); 5559 5560 #if GTK_CHECK_VERSION(3,0,0) 5561 gtk_im_context_set_client_window(xic, gtk_widget_get_window(gui.drawarea)); 5562 #else 5563 gtk_im_context_set_client_window(xic, gui.drawarea->window); 5564 #endif 5565 } 5566 5567 void 5568 im_shutdown(void) 5569 { 5570 #ifdef XIM_DEBUG 5571 xim_log("im_shutdown()\n"); 5572 #endif 5573 5574 if (xic != NULL) 5575 { 5576 gtk_im_context_focus_out(xic); 5577 g_object_unref(xic); 5578 xic = NULL; 5579 } 5580 im_is_active = FALSE; 5581 im_commit_handler_id = 0; 5582 if (p_imst == IM_ON_THE_SPOT) 5583 preedit_start_col = MAXCOL; 5584 xim_has_preediting = FALSE; 5585 } 5586 5587 /* 5588 * Convert the string argument to keyval and state for GdkEventKey. 5589 * If str is valid return TRUE, otherwise FALSE. 5590 * 5591 * See 'imactivatekey' for documentation of the format. 5592 */ 5593 static int 5594 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) 5595 { 5596 const char *mods_end; 5597 unsigned tmp_keyval; 5598 unsigned tmp_state = 0; 5599 5600 mods_end = strrchr(str, '-'); 5601 mods_end = (mods_end != NULL) ? mods_end + 1 : str; 5602 5603 /* Parse modifier keys */ 5604 while (str < mods_end) 5605 switch (*str++) 5606 { 5607 case '-': break; 5608 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; 5609 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; 5610 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; 5611 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; 5612 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; 5613 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; 5614 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; 5615 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; 5616 default: 5617 return FALSE; 5618 } 5619 5620 tmp_keyval = gdk_keyval_from_name(str); 5621 5622 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) 5623 return FALSE; 5624 5625 if (keyval != NULL) 5626 *keyval = tmp_keyval; 5627 if (state != NULL) 5628 *state = tmp_state; 5629 5630 return TRUE; 5631 } 5632 5633 /* 5634 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an 5635 * empty string is also regarded as valid. 5636 * 5637 * Note: The numerical key value of p_imak is cached if it was valid; thus 5638 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever 5639 * 'imak' changes. This is currently the case but not obvious -- should 5640 * probably rename the function for clarity. 5641 */ 5642 int 5643 im_xim_isvalid_imactivate(void) 5644 { 5645 if (p_imak[0] == NUL) 5646 { 5647 im_activatekey_keyval = GDK_VoidSymbol; 5648 im_activatekey_state = 0; 5649 return TRUE; 5650 } 5651 5652 return im_string_to_keyval((const char *)p_imak, 5653 &im_activatekey_keyval, 5654 &im_activatekey_state); 5655 } 5656 5657 static void 5658 im_synthesize_keypress(unsigned int keyval, unsigned int state) 5659 { 5660 GdkEventKey *event; 5661 5662 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); 5663 # if GTK_CHECK_VERSION(3,0,0) 5664 g_object_ref(gtk_widget_get_window(gui.drawarea)); 5665 /* unreffed by gdk_event_free() */ 5666 # else 5667 g_object_ref(gui.drawarea->window); /* unreffed by gdk_event_free() */ 5668 # endif 5669 # if GTK_CHECK_VERSION(3,0,0) 5670 event->window = gtk_widget_get_window(gui.drawarea); 5671 # else 5672 event->window = gui.drawarea->window; 5673 # endif 5674 event->send_event = TRUE; 5675 event->time = GDK_CURRENT_TIME; 5676 event->state = state; 5677 event->keyval = keyval; 5678 event->hardware_keycode = /* needed for XIM */ 5679 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); 5680 event->length = 0; 5681 event->string = NULL; 5682 5683 gtk_im_context_filter_keypress(xic, event); 5684 5685 /* For consistency, also send the corresponding release event. */ 5686 event->type = GDK_KEY_RELEASE; 5687 event->send_event = FALSE; 5688 gtk_im_context_filter_keypress(xic, event); 5689 5690 gdk_event_free((GdkEvent *)event); 5691 } 5692 5693 void 5694 xim_reset(void) 5695 { 5696 # ifdef FEAT_EVAL 5697 if (USE_IMACTIVATEFUNC) 5698 call_imactivatefunc(im_is_active); 5699 else 5700 # endif 5701 if (xic != NULL) 5702 { 5703 gtk_im_context_reset(xic); 5704 5705 if (p_imdisable) 5706 im_shutdown(); 5707 else 5708 { 5709 xim_set_focus(gui.in_focus); 5710 5711 if (im_activatekey_keyval != GDK_VoidSymbol) 5712 { 5713 if (im_is_active) 5714 { 5715 g_signal_handler_block(xic, im_commit_handler_id); 5716 im_synthesize_keypress(im_activatekey_keyval, 5717 im_activatekey_state); 5718 g_signal_handler_unblock(xic, im_commit_handler_id); 5719 } 5720 } 5721 else 5722 { 5723 im_shutdown(); 5724 xim_init(); 5725 xim_set_focus(gui.in_focus); 5726 } 5727 } 5728 } 5729 5730 if (p_imst == IM_ON_THE_SPOT) 5731 preedit_start_col = MAXCOL; 5732 xim_has_preediting = FALSE; 5733 } 5734 5735 int 5736 xim_queue_key_press_event(GdkEventKey *event, int down) 5737 { 5738 if (down) 5739 { 5740 /* 5741 * Workaround GTK2 XIM 'feature' that always converts keypad keys to 5742 * chars., even when not part of an IM sequence (ref. feature of 5743 * gdk/gdkkeyuni.c). 5744 * Flag any keypad keys that might represent a single char. 5745 * If this (on its own - i.e., not part of an IM sequence) is 5746 * committed while we're processing one of these keys, we can ignore 5747 * that commit and go ahead & process it ourselves. That way we can 5748 * still distinguish keypad keys for use in mappings. 5749 * Also add GDK_space to make <S-Space> work. 5750 */ 5751 switch (event->keyval) 5752 { 5753 case GDK_KP_Add: xim_expected_char = '+'; break; 5754 case GDK_KP_Subtract: xim_expected_char = '-'; break; 5755 case GDK_KP_Divide: xim_expected_char = '/'; break; 5756 case GDK_KP_Multiply: xim_expected_char = '*'; break; 5757 case GDK_KP_Decimal: xim_expected_char = '.'; break; 5758 case GDK_KP_Equal: xim_expected_char = '='; break; 5759 case GDK_KP_0: xim_expected_char = '0'; break; 5760 case GDK_KP_1: xim_expected_char = '1'; break; 5761 case GDK_KP_2: xim_expected_char = '2'; break; 5762 case GDK_KP_3: xim_expected_char = '3'; break; 5763 case GDK_KP_4: xim_expected_char = '4'; break; 5764 case GDK_KP_5: xim_expected_char = '5'; break; 5765 case GDK_KP_6: xim_expected_char = '6'; break; 5766 case GDK_KP_7: xim_expected_char = '7'; break; 5767 case GDK_KP_8: xim_expected_char = '8'; break; 5768 case GDK_KP_9: xim_expected_char = '9'; break; 5769 case GDK_space: xim_expected_char = ' '; break; 5770 default: xim_expected_char = NUL; 5771 } 5772 xim_ignored_char = FALSE; 5773 } 5774 5775 /* 5776 * When typing fFtT, XIM may be activated. Thus it must pass 5777 * gtk_im_context_filter_keypress() in Normal mode. 5778 * And while doing :sh too. 5779 */ 5780 if (xic != NULL && !p_imdisable 5781 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) 5782 { 5783 /* 5784 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is 5785 * always aware of the current status of IM, and can even emulate 5786 * the activation key for modules that don't support one. 5787 */ 5788 if (event->keyval == im_activatekey_keyval 5789 && (event->state & im_activatekey_state) == im_activatekey_state) 5790 { 5791 unsigned int state_mask; 5792 5793 /* Require the state of the 3 most used modifiers to match exactly. 5794 * Otherwise e.g. <S-C-space> would be unusable for other purposes 5795 * if the IM activate key is <S-space>. */ 5796 state_mask = im_activatekey_state; 5797 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK 5798 | (int)GDK_MOD1_MASK); 5799 5800 if ((event->state & state_mask) != im_activatekey_state) 5801 return FALSE; 5802 5803 /* Don't send it a second time on GDK_KEY_RELEASE. */ 5804 if (event->type != GDK_KEY_PRESS) 5805 return TRUE; 5806 5807 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE)) 5808 { 5809 im_set_active(FALSE); 5810 5811 /* ":lmap" mappings exists, toggle use of mappings. */ 5812 State ^= LANGMAP; 5813 if (State & LANGMAP) 5814 { 5815 curbuf->b_p_iminsert = B_IMODE_NONE; 5816 State &= ~LANGMAP; 5817 } 5818 else 5819 { 5820 curbuf->b_p_iminsert = B_IMODE_LMAP; 5821 State |= LANGMAP; 5822 } 5823 return TRUE; 5824 } 5825 5826 return gtk_im_context_filter_keypress(xic, event); 5827 } 5828 5829 /* Don't filter events through the IM context if IM isn't active 5830 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module 5831 * not doing anything before the activation key was sent. */ 5832 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) 5833 { 5834 int imresult = gtk_im_context_filter_keypress(xic, event); 5835 5836 if (p_imst == IM_ON_THE_SPOT) 5837 { 5838 /* Some XIM send following sequence: 5839 * 1. preedited string. 5840 * 2. committed string. 5841 * 3. line changed key. 5842 * 4. preedited string. 5843 * 5. remove preedited string. 5844 * if 3, Vim can't move back the above line for 5. 5845 * thus, this part should not parse the key. */ 5846 if (!imresult && preedit_start_col != MAXCOL 5847 && event->keyval == GDK_Return) 5848 { 5849 im_synthesize_keypress(GDK_Return, 0U); 5850 return FALSE; 5851 } 5852 } 5853 5854 /* If XIM tried to commit a keypad key as a single char., 5855 * ignore it so we can use the keypad key 'raw', for mappings. */ 5856 if (xim_expected_char != NUL && xim_ignored_char) 5857 /* We had a keypad key, and XIM tried to thieve it */ 5858 return FALSE; 5859 5860 /* This is supposed to fix a problem with iBus, that space 5861 * characters don't work in input mode. */ 5862 xim_expected_char = NUL; 5863 5864 /* Normal processing */ 5865 return imresult; 5866 } 5867 } 5868 5869 return FALSE; 5870 } 5871 5872 int 5873 im_get_status(void) 5874 { 5875 # ifdef FEAT_EVAL 5876 if (USE_IMSTATUSFUNC) 5877 return call_imstatusfunc(); 5878 # endif 5879 return im_is_active; 5880 } 5881 5882 int 5883 preedit_get_status(void) 5884 { 5885 return preedit_is_active; 5886 } 5887 5888 int 5889 im_is_preediting(void) 5890 { 5891 return xim_has_preediting; 5892 } 5893 5894 # else /* !FEAT_GUI_GTK */ 5895 5896 static int xim_is_active = FALSE; /* XIM should be active in the current 5897 mode */ 5898 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */ 5899 # ifdef FEAT_GUI_X11 5900 static XIMStyle input_style; 5901 static int status_area_enabled = TRUE; 5902 # endif 5903 5904 /* 5905 * Switch using XIM on/off. This is used by the code that changes "State". 5906 * When 'imactivatefunc' is defined use that function instead. 5907 */ 5908 void 5909 im_set_active(int active_arg) 5910 { 5911 int active = active_arg; 5912 5913 /* If 'imdisable' is set, XIM is never active. */ 5914 if (p_imdisable) 5915 active = FALSE; 5916 else if (input_style & XIMPreeditPosition) 5917 /* There is a problem in switching XIM off when preediting is used, 5918 * and it is not clear how this can be solved. For now, keep XIM on 5919 * all the time, like it was done in Vim 5.8. */ 5920 active = TRUE; 5921 5922 # if defined(FEAT_EVAL) 5923 if (USE_IMACTIVATEFUNC) 5924 { 5925 if (active != im_get_status()) 5926 { 5927 call_imactivatefunc(active); 5928 xim_has_focus = active; 5929 } 5930 return; 5931 } 5932 # endif 5933 5934 if (xic == NULL) 5935 return; 5936 5937 /* Remember the active state, it is needed when Vim gets keyboard focus. */ 5938 xim_is_active = active; 5939 xim_set_preedit(); 5940 } 5941 5942 /* 5943 * Adjust using XIM for gaining or losing keyboard focus. Also called when 5944 * "xim_is_active" changes. 5945 */ 5946 void 5947 xim_set_focus(int focus) 5948 { 5949 if (xic == NULL) 5950 return; 5951 5952 /* 5953 * XIM only gets focus when the Vim window has keyboard focus and XIM has 5954 * been set active for the current mode. 5955 */ 5956 if (focus && xim_is_active) 5957 { 5958 if (!xim_has_focus) 5959 { 5960 xim_has_focus = TRUE; 5961 XSetICFocus(xic); 5962 } 5963 } 5964 else 5965 { 5966 if (xim_has_focus) 5967 { 5968 xim_has_focus = FALSE; 5969 XUnsetICFocus(xic); 5970 } 5971 } 5972 } 5973 5974 void 5975 im_set_position(int row UNUSED, int col UNUSED) 5976 { 5977 xim_set_preedit(); 5978 } 5979 5980 /* 5981 * Set the XIM to the current cursor position. 5982 */ 5983 void 5984 xim_set_preedit(void) 5985 { 5986 XVaNestedList attr_list; 5987 XRectangle spot_area; 5988 XPoint over_spot; 5989 int line_space; 5990 5991 if (xic == NULL) 5992 return; 5993 5994 xim_set_focus(TRUE); 5995 5996 if (!xim_has_focus) 5997 { 5998 /* hide XIM cursor */ 5999 over_spot.x = 0; 6000 over_spot.y = -100; /* arbitrary invisible position */ 6001 attr_list = (XVaNestedList) XVaCreateNestedList(0, 6002 XNSpotLocation, 6003 &over_spot, 6004 NULL); 6005 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); 6006 XFree(attr_list); 6007 return; 6008 } 6009 6010 if (input_style & XIMPreeditPosition) 6011 { 6012 if (xim_fg_color == INVALCOLOR) 6013 { 6014 xim_fg_color = gui.def_norm_pixel; 6015 xim_bg_color = gui.def_back_pixel; 6016 } 6017 over_spot.x = TEXT_X(gui.col); 6018 over_spot.y = TEXT_Y(gui.row); 6019 spot_area.x = 0; 6020 spot_area.y = 0; 6021 spot_area.height = gui.char_height * Rows; 6022 spot_area.width = gui.char_width * Columns; 6023 line_space = gui.char_height; 6024 attr_list = (XVaNestedList) XVaCreateNestedList(0, 6025 XNSpotLocation, &over_spot, 6026 XNForeground, (Pixel) xim_fg_color, 6027 XNBackground, (Pixel) xim_bg_color, 6028 XNArea, &spot_area, 6029 XNLineSpace, line_space, 6030 NULL); 6031 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) 6032 EMSG(_("E284: Cannot set IC values")); 6033 XFree(attr_list); 6034 } 6035 } 6036 6037 # if defined(FEAT_GUI_X11) 6038 static char e_xim[] = N_("E285: Failed to create input context"); 6039 # endif 6040 6041 # if defined(FEAT_GUI_X11) || defined(PROTO) 6042 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(SUN_SYSTEM) 6043 # define USE_X11R6_XIM 6044 # endif 6045 6046 static int xim_real_init(Window x11_window, Display *x11_display); 6047 6048 6049 # ifdef USE_X11R6_XIM 6050 static void xim_destroy_cb(XIM im, XPointer client_data, XPointer call_data); 6051 6052 static void 6053 xim_instantiate_cb( 6054 Display *display, 6055 XPointer client_data UNUSED, 6056 XPointer call_data UNUSED) 6057 { 6058 Window x11_window; 6059 Display *x11_display; 6060 6061 # ifdef XIM_DEBUG 6062 xim_log("xim_instantiate_cb()\n"); 6063 # endif 6064 6065 gui_get_x11_windis(&x11_window, &x11_display); 6066 if (display != x11_display) 6067 return; 6068 6069 xim_real_init(x11_window, x11_display); 6070 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6071 if (xic != NULL) 6072 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6073 xim_instantiate_cb, NULL); 6074 } 6075 6076 static void 6077 xim_destroy_cb( 6078 XIM im UNUSED, 6079 XPointer client_data UNUSED, 6080 XPointer call_data UNUSED) 6081 { 6082 Window x11_window; 6083 Display *x11_display; 6084 6085 # ifdef XIM_DEBUG 6086 xim_log("xim_destroy_cb()\n"); 6087 #endif 6088 gui_get_x11_windis(&x11_window, &x11_display); 6089 6090 xic = NULL; 6091 status_area_enabled = FALSE; 6092 6093 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6094 6095 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6096 xim_instantiate_cb, NULL); 6097 } 6098 # endif 6099 6100 void 6101 xim_init(void) 6102 { 6103 Window x11_window; 6104 Display *x11_display; 6105 6106 # ifdef XIM_DEBUG 6107 xim_log("xim_init()\n"); 6108 # endif 6109 6110 gui_get_x11_windis(&x11_window, &x11_display); 6111 6112 xic = NULL; 6113 6114 if (xim_real_init(x11_window, x11_display)) 6115 return; 6116 6117 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6118 6119 # ifdef USE_X11R6_XIM 6120 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6121 xim_instantiate_cb, NULL); 6122 # endif 6123 } 6124 6125 static int 6126 xim_real_init(Window x11_window, Display *x11_display) 6127 { 6128 int i; 6129 char *p, 6130 *s, 6131 *ns, 6132 *end, 6133 tmp[1024]; 6134 # define IMLEN_MAX 40 6135 char buf[IMLEN_MAX + 7]; 6136 XIM xim = NULL; 6137 XIMStyles *xim_styles; 6138 XIMStyle this_input_style = 0; 6139 Boolean found; 6140 XPoint over_spot; 6141 XVaNestedList preedit_list, status_list; 6142 6143 input_style = 0; 6144 status_area_enabled = FALSE; 6145 6146 if (xic != NULL) 6147 return FALSE; 6148 6149 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) 6150 { 6151 strcpy(tmp, gui.rsrc_input_method); 6152 for (ns = s = tmp; ns != NULL && *s != NUL;) 6153 { 6154 s = (char *)skipwhite((char_u *)s); 6155 if (*s == NUL) 6156 break; 6157 if ((ns = end = strchr(s, ',')) == NULL) 6158 end = s + strlen(s); 6159 while (isspace(((char_u *)end)[-1])) 6160 end--; 6161 *end = NUL; 6162 6163 if (strlen(s) <= IMLEN_MAX) 6164 { 6165 strcpy(buf, "@im="); 6166 strcat(buf, s); 6167 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL 6168 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) 6169 != NULL) 6170 break; 6171 } 6172 6173 s = ns + 1; 6174 } 6175 } 6176 6177 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) 6178 xim = XOpenIM(x11_display, NULL, NULL, NULL); 6179 6180 /* This is supposed to be useful to obtain characters through 6181 * XmbLookupString() without really using a XIM. */ 6182 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL 6183 && *p != NUL) 6184 xim = XOpenIM(x11_display, NULL, NULL, NULL); 6185 6186 if (xim == NULL) 6187 { 6188 /* Only give this message when verbose is set, because too many people 6189 * got this message when they didn't want to use a XIM. */ 6190 if (p_verbose > 0) 6191 { 6192 verbose_enter(); 6193 EMSG(_("E286: Failed to open input method")); 6194 verbose_leave(); 6195 } 6196 return FALSE; 6197 } 6198 6199 # ifdef USE_X11R6_XIM 6200 { 6201 XIMCallback destroy_cb; 6202 6203 destroy_cb.callback = xim_destroy_cb; 6204 destroy_cb.client_data = NULL; 6205 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) 6206 EMSG(_("E287: Warning: Could not set destroy callback to IM")); 6207 } 6208 # endif 6209 6210 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) 6211 { 6212 EMSG(_("E288: input method doesn't support any style")); 6213 XCloseIM(xim); 6214 return FALSE; 6215 } 6216 6217 found = False; 6218 strcpy(tmp, gui.rsrc_preedit_type_name); 6219 for (s = tmp; s && !found; ) 6220 { 6221 while (*s && isspace((unsigned char)*s)) 6222 s++; 6223 if (!*s) 6224 break; 6225 if ((ns = end = strchr(s, ',')) != 0) 6226 ns++; 6227 else 6228 end = s + strlen(s); 6229 while (isspace((unsigned char)*end)) 6230 end--; 6231 *end = '\0'; 6232 6233 if (!strcmp(s, "OverTheSpot")) 6234 this_input_style = (XIMPreeditPosition | XIMStatusArea); 6235 else if (!strcmp(s, "OffTheSpot")) 6236 this_input_style = (XIMPreeditArea | XIMStatusArea); 6237 else if (!strcmp(s, "Root")) 6238 this_input_style = (XIMPreeditNothing | XIMStatusNothing); 6239 6240 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 6241 { 6242 if (this_input_style == xim_styles->supported_styles[i]) 6243 { 6244 found = True; 6245 break; 6246 } 6247 } 6248 if (!found) 6249 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 6250 { 6251 if ((xim_styles->supported_styles[i] & this_input_style) 6252 == (this_input_style & ~XIMStatusArea)) 6253 { 6254 this_input_style &= ~XIMStatusArea; 6255 found = True; 6256 break; 6257 } 6258 } 6259 6260 s = ns; 6261 } 6262 XFree(xim_styles); 6263 6264 if (!found) 6265 { 6266 /* Only give this message when verbose is set, because too many people 6267 * got this message when they didn't want to use a XIM. */ 6268 if (p_verbose > 0) 6269 { 6270 verbose_enter(); 6271 EMSG(_("E289: input method doesn't support my preedit type")); 6272 verbose_leave(); 6273 } 6274 XCloseIM(xim); 6275 return FALSE; 6276 } 6277 6278 over_spot.x = TEXT_X(gui.col); 6279 over_spot.y = TEXT_Y(gui.row); 6280 input_style = this_input_style; 6281 6282 /* A crash was reported when trying to pass gui.norm_font as XNFontSet, 6283 * thus that has been removed. Hopefully the default works... */ 6284 # ifdef FEAT_XFONTSET 6285 if (gui.fontset != NOFONTSET) 6286 { 6287 preedit_list = XVaCreateNestedList(0, 6288 XNSpotLocation, &over_spot, 6289 XNForeground, (Pixel)gui.def_norm_pixel, 6290 XNBackground, (Pixel)gui.def_back_pixel, 6291 XNFontSet, (XFontSet)gui.fontset, 6292 NULL); 6293 status_list = XVaCreateNestedList(0, 6294 XNForeground, (Pixel)gui.def_norm_pixel, 6295 XNBackground, (Pixel)gui.def_back_pixel, 6296 XNFontSet, (XFontSet)gui.fontset, 6297 NULL); 6298 } 6299 else 6300 # endif 6301 { 6302 preedit_list = XVaCreateNestedList(0, 6303 XNSpotLocation, &over_spot, 6304 XNForeground, (Pixel)gui.def_norm_pixel, 6305 XNBackground, (Pixel)gui.def_back_pixel, 6306 NULL); 6307 status_list = XVaCreateNestedList(0, 6308 XNForeground, (Pixel)gui.def_norm_pixel, 6309 XNBackground, (Pixel)gui.def_back_pixel, 6310 NULL); 6311 } 6312 6313 xic = XCreateIC(xim, 6314 XNInputStyle, input_style, 6315 XNClientWindow, x11_window, 6316 XNFocusWindow, gui.wid, 6317 XNPreeditAttributes, preedit_list, 6318 XNStatusAttributes, status_list, 6319 NULL); 6320 XFree(status_list); 6321 XFree(preedit_list); 6322 if (xic != NULL) 6323 { 6324 if (input_style & XIMStatusArea) 6325 { 6326 xim_set_status_area(); 6327 status_area_enabled = TRUE; 6328 } 6329 else 6330 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6331 } 6332 else 6333 { 6334 if (!is_not_a_term()) 6335 EMSG(_(e_xim)); 6336 XCloseIM(xim); 6337 return FALSE; 6338 } 6339 6340 return TRUE; 6341 } 6342 6343 # endif /* FEAT_GUI_X11 */ 6344 6345 /* 6346 * Get IM status. When IM is on, return TRUE. Else return FALSE. 6347 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is 6348 * active, when not having focus XIM may still be active (e.g., when using a 6349 * tear-off menu item). 6350 */ 6351 int 6352 im_get_status(void) 6353 { 6354 # ifdef FEAT_EVAL 6355 if (USE_IMSTATUSFUNC) 6356 return call_imstatusfunc(); 6357 # endif 6358 return xim_has_focus; 6359 } 6360 6361 # endif /* !FEAT_GUI_GTK */ 6362 6363 # if !defined(FEAT_GUI_GTK) || defined(PROTO) 6364 /* 6365 * Set up the status area. 6366 * 6367 * This should use a separate Widget, but that seems not possible, because 6368 * preedit_area and status_area should be set to the same window as for the 6369 * text input. Unfortunately this means the status area pollutes the text 6370 * window... 6371 */ 6372 void 6373 xim_set_status_area(void) 6374 { 6375 XVaNestedList preedit_list = 0, status_list = 0, list = 0; 6376 XRectangle pre_area, status_area; 6377 6378 if (xic == NULL) 6379 return; 6380 6381 if (input_style & XIMStatusArea) 6382 { 6383 if (input_style & XIMPreeditArea) 6384 { 6385 XRectangle *needed_rect; 6386 6387 /* to get status_area width */ 6388 status_list = XVaCreateNestedList(0, XNAreaNeeded, 6389 &needed_rect, NULL); 6390 XGetICValues(xic, XNStatusAttributes, status_list, NULL); 6391 XFree(status_list); 6392 6393 status_area.width = needed_rect->width; 6394 } 6395 else 6396 status_area.width = gui.char_width * Columns; 6397 6398 status_area.x = 0; 6399 status_area.y = gui.char_height * Rows + gui.border_offset; 6400 if (gui.which_scrollbars[SBAR_BOTTOM]) 6401 status_area.y += gui.scrollbar_height; 6402 #ifdef FEAT_MENU 6403 if (gui.menu_is_active) 6404 status_area.y += gui.menu_height; 6405 #endif 6406 status_area.height = gui.char_height; 6407 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); 6408 } 6409 else 6410 { 6411 status_area.x = 0; 6412 status_area.y = gui.char_height * Rows + gui.border_offset; 6413 if (gui.which_scrollbars[SBAR_BOTTOM]) 6414 status_area.y += gui.scrollbar_height; 6415 #ifdef FEAT_MENU 6416 if (gui.menu_is_active) 6417 status_area.y += gui.menu_height; 6418 #endif 6419 status_area.width = 0; 6420 status_area.height = gui.char_height; 6421 } 6422 6423 if (input_style & XIMPreeditArea) /* off-the-spot */ 6424 { 6425 pre_area.x = status_area.x + status_area.width; 6426 pre_area.y = gui.char_height * Rows + gui.border_offset; 6427 pre_area.width = gui.char_width * Columns - pre_area.x; 6428 if (gui.which_scrollbars[SBAR_BOTTOM]) 6429 pre_area.y += gui.scrollbar_height; 6430 #ifdef FEAT_MENU 6431 if (gui.menu_is_active) 6432 pre_area.y += gui.menu_height; 6433 #endif 6434 pre_area.height = gui.char_height; 6435 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6436 } 6437 else if (input_style & XIMPreeditPosition) /* over-the-spot */ 6438 { 6439 pre_area.x = 0; 6440 pre_area.y = 0; 6441 pre_area.height = gui.char_height * Rows; 6442 pre_area.width = gui.char_width * Columns; 6443 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6444 } 6445 6446 if (preedit_list && status_list) 6447 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6448 XNStatusAttributes, status_list, NULL); 6449 else if (preedit_list) 6450 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6451 NULL); 6452 else if (status_list) 6453 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, 6454 NULL); 6455 else 6456 list = NULL; 6457 6458 if (list) 6459 { 6460 XSetICValues(xic, XNVaNestedList, list, NULL); 6461 XFree(list); 6462 } 6463 if (status_list) 6464 XFree(status_list); 6465 if (preedit_list) 6466 XFree(preedit_list); 6467 } 6468 6469 int 6470 xim_get_status_area_height(void) 6471 { 6472 if (status_area_enabled) 6473 return gui.char_height; 6474 return 0; 6475 } 6476 # endif 6477 6478 #else /* !defined(FEAT_XIM) */ 6479 6480 # ifdef IME_WITHOUT_XIM 6481 static int im_was_set_active = FALSE; 6482 6483 int 6484 im_get_status(void) 6485 { 6486 # if defined(FEAT_MBYTE) && defined(FEAT_EVAL) 6487 if (USE_IMSTATUSFUNC) 6488 return call_imstatusfunc(); 6489 # endif 6490 return im_was_set_active; 6491 } 6492 6493 void 6494 im_set_active(int active_arg) 6495 { 6496 # if defined(FEAT_MBYTE) && defined(FEAT_EVAL) 6497 int active = !p_imdisable && active_arg; 6498 6499 if (USE_IMACTIVATEFUNC && active != im_get_status()) 6500 { 6501 call_imactivatefunc(active); 6502 im_was_set_active = active; 6503 } 6504 # endif 6505 } 6506 6507 # ifdef FEAT_GUI 6508 void 6509 im_set_position(int row, int col) 6510 { 6511 } 6512 # endif 6513 # endif 6514 6515 #endif /* FEAT_XIM */ 6516 6517 #if defined(FEAT_MBYTE) || defined(PROTO) 6518 6519 /* 6520 * Setup "vcp" for conversion from "from" to "to". 6521 * The names must have been made canonical with enc_canonize(). 6522 * vcp->vc_type must have been initialized to CONV_NONE. 6523 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 6524 * instead). 6525 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 6526 * Return FAIL when conversion is not supported, OK otherwise. 6527 */ 6528 int 6529 convert_setup(vimconv_T *vcp, char_u *from, char_u *to) 6530 { 6531 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 6532 } 6533 6534 /* 6535 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 6536 * "from" unicode charsets be considered utf-8. Same for "to". 6537 */ 6538 int 6539 convert_setup_ext( 6540 vimconv_T *vcp, 6541 char_u *from, 6542 int from_unicode_is_utf8, 6543 char_u *to, 6544 int to_unicode_is_utf8) 6545 { 6546 int from_prop; 6547 int to_prop; 6548 int from_is_utf8; 6549 int to_is_utf8; 6550 6551 /* Reset to no conversion. */ 6552 # ifdef USE_ICONV 6553 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 6554 iconv_close(vcp->vc_fd); 6555 # endif 6556 vcp->vc_type = CONV_NONE; 6557 vcp->vc_factor = 1; 6558 vcp->vc_fail = FALSE; 6559 6560 /* No conversion when one of the names is empty or they are equal. */ 6561 if (from == NULL || *from == NUL || to == NULL || *to == NUL 6562 || STRCMP(from, to) == 0) 6563 return OK; 6564 6565 from_prop = enc_canon_props(from); 6566 to_prop = enc_canon_props(to); 6567 if (from_unicode_is_utf8) 6568 from_is_utf8 = from_prop & ENC_UNICODE; 6569 else 6570 from_is_utf8 = from_prop == ENC_UNICODE; 6571 if (to_unicode_is_utf8) 6572 to_is_utf8 = to_prop & ENC_UNICODE; 6573 else 6574 to_is_utf8 = to_prop == ENC_UNICODE; 6575 6576 if ((from_prop & ENC_LATIN1) && to_is_utf8) 6577 { 6578 /* Internal latin1 -> utf-8 conversion. */ 6579 vcp->vc_type = CONV_TO_UTF8; 6580 vcp->vc_factor = 2; /* up to twice as long */ 6581 } 6582 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 6583 { 6584 /* Internal latin9 -> utf-8 conversion. */ 6585 vcp->vc_type = CONV_9_TO_UTF8; 6586 vcp->vc_factor = 3; /* up to three as long (euro sign) */ 6587 } 6588 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 6589 { 6590 /* Internal utf-8 -> latin1 conversion. */ 6591 vcp->vc_type = CONV_TO_LATIN1; 6592 } 6593 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 6594 { 6595 /* Internal utf-8 -> latin9 conversion. */ 6596 vcp->vc_type = CONV_TO_LATIN9; 6597 } 6598 #ifdef WIN3264 6599 /* Win32-specific codepage <-> codepage conversion without iconv. */ 6600 else if ((from_is_utf8 || encname2codepage(from) > 0) 6601 && (to_is_utf8 || encname2codepage(to) > 0)) 6602 { 6603 vcp->vc_type = CONV_CODEPAGE; 6604 vcp->vc_factor = 2; /* up to twice as long */ 6605 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 6606 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 6607 } 6608 #endif 6609 #ifdef MACOS_CONVERT 6610 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 6611 { 6612 vcp->vc_type = CONV_MAC_LATIN1; 6613 } 6614 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 6615 { 6616 vcp->vc_type = CONV_MAC_UTF8; 6617 vcp->vc_factor = 2; /* up to twice as long */ 6618 } 6619 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 6620 { 6621 vcp->vc_type = CONV_LATIN1_MAC; 6622 } 6623 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 6624 { 6625 vcp->vc_type = CONV_UTF8_MAC; 6626 } 6627 #endif 6628 # ifdef USE_ICONV 6629 else 6630 { 6631 /* Use iconv() for conversion. */ 6632 vcp->vc_fd = (iconv_t)my_iconv_open( 6633 to_is_utf8 ? (char_u *)"utf-8" : to, 6634 from_is_utf8 ? (char_u *)"utf-8" : from); 6635 if (vcp->vc_fd != (iconv_t)-1) 6636 { 6637 vcp->vc_type = CONV_ICONV; 6638 vcp->vc_factor = 4; /* could be longer too... */ 6639 } 6640 } 6641 # endif 6642 if (vcp->vc_type == CONV_NONE) 6643 return FAIL; 6644 6645 return OK; 6646 } 6647 6648 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \ 6649 || defined(PROTO) 6650 /* 6651 * Do conversion on typed input characters in-place. 6652 * The input and output are not NUL terminated! 6653 * Returns the length after conversion. 6654 */ 6655 int 6656 convert_input(char_u *ptr, int len, int maxlen) 6657 { 6658 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 6659 } 6660 #endif 6661 6662 /* 6663 * Like convert_input(), but when there is an incomplete byte sequence at the 6664 * end return that as an allocated string in "restp" and set "*restlenp" to 6665 * the length. If "restp" is NULL it is not used. 6666 */ 6667 int 6668 convert_input_safe( 6669 char_u *ptr, 6670 int len, 6671 int maxlen, 6672 char_u **restp, 6673 int *restlenp) 6674 { 6675 char_u *d; 6676 int dlen = len; 6677 int unconvertlen = 0; 6678 6679 d = string_convert_ext(&input_conv, ptr, &dlen, 6680 restp == NULL ? NULL : &unconvertlen); 6681 if (d != NULL) 6682 { 6683 if (dlen <= maxlen) 6684 { 6685 if (unconvertlen > 0) 6686 { 6687 /* Move the unconverted characters to allocated memory. */ 6688 *restp = alloc(unconvertlen); 6689 if (*restp != NULL) 6690 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 6691 *restlenp = unconvertlen; 6692 } 6693 mch_memmove(ptr, d, dlen); 6694 } 6695 else 6696 /* result is too long, keep the unconverted text (the caller must 6697 * have done something wrong!) */ 6698 dlen = len; 6699 vim_free(d); 6700 } 6701 return dlen; 6702 } 6703 6704 /* 6705 * Convert text "ptr[*lenp]" according to "vcp". 6706 * Returns the result in allocated memory and sets "*lenp". 6707 * When "lenp" is NULL, use NUL terminated strings. 6708 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 6709 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 6710 */ 6711 char_u * 6712 string_convert( 6713 vimconv_T *vcp, 6714 char_u *ptr, 6715 int *lenp) 6716 { 6717 return string_convert_ext(vcp, ptr, lenp, NULL); 6718 } 6719 6720 /* 6721 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 6722 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 6723 * set to the number of remaining bytes. 6724 */ 6725 char_u * 6726 string_convert_ext( 6727 vimconv_T *vcp, 6728 char_u *ptr, 6729 int *lenp, 6730 int *unconvlenp) 6731 { 6732 char_u *retval = NULL; 6733 char_u *d; 6734 int len; 6735 int i; 6736 int l; 6737 int c; 6738 6739 if (lenp == NULL) 6740 len = (int)STRLEN(ptr); 6741 else 6742 len = *lenp; 6743 if (len == 0) 6744 return vim_strsave((char_u *)""); 6745 6746 switch (vcp->vc_type) 6747 { 6748 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */ 6749 retval = alloc(len * 2 + 1); 6750 if (retval == NULL) 6751 break; 6752 d = retval; 6753 for (i = 0; i < len; ++i) 6754 { 6755 c = ptr[i]; 6756 if (c < 0x80) 6757 *d++ = c; 6758 else 6759 { 6760 *d++ = 0xc0 + ((unsigned)c >> 6); 6761 *d++ = 0x80 + (c & 0x3f); 6762 } 6763 } 6764 *d = NUL; 6765 if (lenp != NULL) 6766 *lenp = (int)(d - retval); 6767 break; 6768 6769 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */ 6770 retval = alloc(len * 3 + 1); 6771 if (retval == NULL) 6772 break; 6773 d = retval; 6774 for (i = 0; i < len; ++i) 6775 { 6776 c = ptr[i]; 6777 switch (c) 6778 { 6779 case 0xa4: c = 0x20ac; break; /* euro */ 6780 case 0xa6: c = 0x0160; break; /* S hat */ 6781 case 0xa8: c = 0x0161; break; /* S -hat */ 6782 case 0xb4: c = 0x017d; break; /* Z hat */ 6783 case 0xb8: c = 0x017e; break; /* Z -hat */ 6784 case 0xbc: c = 0x0152; break; /* OE */ 6785 case 0xbd: c = 0x0153; break; /* oe */ 6786 case 0xbe: c = 0x0178; break; /* Y */ 6787 } 6788 d += utf_char2bytes(c, d); 6789 } 6790 *d = NUL; 6791 if (lenp != NULL) 6792 *lenp = (int)(d - retval); 6793 break; 6794 6795 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */ 6796 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */ 6797 retval = alloc(len + 1); 6798 if (retval == NULL) 6799 break; 6800 d = retval; 6801 for (i = 0; i < len; ++i) 6802 { 6803 l = utf_ptr2len_len(ptr + i, len - i); 6804 if (l == 0) 6805 *d++ = NUL; 6806 else if (l == 1) 6807 { 6808 int l_w = utf8len_tab_zero[ptr[i]]; 6809 6810 if (l_w == 0) 6811 { 6812 /* Illegal utf-8 byte cannot be converted */ 6813 vim_free(retval); 6814 return NULL; 6815 } 6816 if (unconvlenp != NULL && l_w > len - i) 6817 { 6818 /* Incomplete sequence at the end. */ 6819 *unconvlenp = len - i; 6820 break; 6821 } 6822 *d++ = ptr[i]; 6823 } 6824 else 6825 { 6826 c = utf_ptr2char(ptr + i); 6827 if (vcp->vc_type == CONV_TO_LATIN9) 6828 switch (c) 6829 { 6830 case 0x20ac: c = 0xa4; break; /* euro */ 6831 case 0x0160: c = 0xa6; break; /* S hat */ 6832 case 0x0161: c = 0xa8; break; /* S -hat */ 6833 case 0x017d: c = 0xb4; break; /* Z hat */ 6834 case 0x017e: c = 0xb8; break; /* Z -hat */ 6835 case 0x0152: c = 0xbc; break; /* OE */ 6836 case 0x0153: c = 0xbd; break; /* oe */ 6837 case 0x0178: c = 0xbe; break; /* Y */ 6838 case 0xa4: 6839 case 0xa6: 6840 case 0xa8: 6841 case 0xb4: 6842 case 0xb8: 6843 case 0xbc: 6844 case 0xbd: 6845 case 0xbe: c = 0x100; break; /* not in latin9 */ 6846 } 6847 if (!utf_iscomposing(c)) /* skip composing chars */ 6848 { 6849 if (c < 0x100) 6850 *d++ = c; 6851 else if (vcp->vc_fail) 6852 { 6853 vim_free(retval); 6854 return NULL; 6855 } 6856 else 6857 { 6858 *d++ = 0xbf; 6859 if (utf_char2cells(c) > 1) 6860 *d++ = '?'; 6861 } 6862 } 6863 i += l - 1; 6864 } 6865 } 6866 *d = NUL; 6867 if (lenp != NULL) 6868 *lenp = (int)(d - retval); 6869 break; 6870 6871 # ifdef MACOS_CONVERT 6872 case CONV_MAC_LATIN1: 6873 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6874 'm', 'l', unconvlenp); 6875 break; 6876 6877 case CONV_LATIN1_MAC: 6878 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6879 'l', 'm', unconvlenp); 6880 break; 6881 6882 case CONV_MAC_UTF8: 6883 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6884 'm', 'u', unconvlenp); 6885 break; 6886 6887 case CONV_UTF8_MAC: 6888 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6889 'u', 'm', unconvlenp); 6890 break; 6891 # endif 6892 6893 # ifdef USE_ICONV 6894 case CONV_ICONV: /* conversion with output_conv.vc_fd */ 6895 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 6896 break; 6897 # endif 6898 # ifdef WIN3264 6899 case CONV_CODEPAGE: /* codepage -> codepage */ 6900 { 6901 int retlen; 6902 int tmp_len; 6903 short_u *tmp; 6904 6905 /* 1. codepage/UTF-8 -> ucs-2. */ 6906 if (vcp->vc_cpfrom == 0) 6907 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 6908 else 6909 { 6910 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 6911 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 6912 (char *)ptr, len, 0, 0); 6913 if (tmp_len == 0 6914 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 6915 { 6916 if (lenp != NULL) 6917 *lenp = 0; 6918 if (unconvlenp != NULL) 6919 *unconvlenp = len; 6920 retval = alloc(1); 6921 if (retval) 6922 retval[0] = NUL; 6923 return retval; 6924 } 6925 } 6926 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len); 6927 if (tmp == NULL) 6928 break; 6929 if (vcp->vc_cpfrom == 0) 6930 utf8_to_utf16(ptr, len, tmp, unconvlenp); 6931 else 6932 MultiByteToWideChar(vcp->vc_cpfrom, 0, 6933 (char *)ptr, len, tmp, tmp_len); 6934 6935 /* 2. ucs-2 -> codepage/UTF-8. */ 6936 if (vcp->vc_cpto == 0) 6937 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 6938 else 6939 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 6940 tmp, tmp_len, 0, 0, 0, 0); 6941 retval = alloc(retlen + 1); 6942 if (retval != NULL) 6943 { 6944 if (vcp->vc_cpto == 0) 6945 utf16_to_utf8(tmp, tmp_len, retval); 6946 else 6947 WideCharToMultiByte(vcp->vc_cpto, 0, 6948 tmp, tmp_len, 6949 (char *)retval, retlen, 0, 0); 6950 retval[retlen] = NUL; 6951 if (lenp != NULL) 6952 *lenp = retlen; 6953 } 6954 vim_free(tmp); 6955 break; 6956 } 6957 # endif 6958 } 6959 6960 return retval; 6961 } 6962 #endif 6963