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 #ifdef FEAT_AUTOCMD 803 /* Fire an autocommand to let people do custom font setup. This must be 804 * after Vim has been setup for the new encoding. */ 805 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); 806 #endif 807 808 #ifdef FEAT_SPELL 809 /* Need to reload spell dictionaries */ 810 spell_reload(); 811 #endif 812 813 return NULL; 814 } 815 816 /* 817 * Return the size of the BOM for the current buffer: 818 * 0 - no BOM 819 * 2 - UCS-2 or UTF-16 BOM 820 * 4 - UCS-4 BOM 821 * 3 - UTF-8 BOM 822 */ 823 int 824 bomb_size(void) 825 { 826 int n = 0; 827 828 if (curbuf->b_p_bomb && !curbuf->b_p_bin) 829 { 830 if (*curbuf->b_p_fenc == NUL) 831 { 832 if (enc_utf8) 833 { 834 if (enc_unicode != 0) 835 n = enc_unicode; 836 else 837 n = 3; 838 } 839 } 840 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) 841 n = 3; 842 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 843 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) 844 n = 2; 845 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) 846 n = 4; 847 } 848 return n; 849 } 850 851 /* 852 * Remove all BOM from "s" by moving remaining text. 853 */ 854 void 855 remove_bom(char_u *s) 856 { 857 if (enc_utf8) 858 { 859 char_u *p = s; 860 861 while ((p = vim_strbyte(p, 0xef)) != NULL) 862 { 863 if (p[1] == 0xbb && p[2] == 0xbf) 864 STRMOVE(p, p + 3); 865 else 866 ++p; 867 } 868 } 869 } 870 871 /* 872 * Get class of pointer: 873 * 0 for blank or NUL 874 * 1 for punctuation 875 * 2 for an (ASCII) word character 876 * >2 for other word characters 877 */ 878 int 879 mb_get_class(char_u *p) 880 { 881 return mb_get_class_buf(p, curbuf); 882 } 883 884 int 885 mb_get_class_buf(char_u *p, buf_T *buf) 886 { 887 if (MB_BYTE2LEN(p[0]) == 1) 888 { 889 if (p[0] == NUL || VIM_ISWHITE(p[0])) 890 return 0; 891 if (vim_iswordc_buf(p[0], buf)) 892 return 2; 893 return 1; 894 } 895 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) 896 return dbcs_class(p[0], p[1]); 897 if (enc_utf8) 898 return utf_class_buf(utf_ptr2char(p), buf); 899 return 0; 900 } 901 902 /* 903 * Get class of a double-byte character. This always returns 3 or bigger. 904 * TODO: Should return 1 for punctuation. 905 */ 906 int 907 dbcs_class(unsigned lead, unsigned trail) 908 { 909 switch (enc_dbcs) 910 { 911 /* please add classify routine for your language in here */ 912 913 case DBCS_JPNU: /* ? */ 914 case DBCS_JPN: 915 { 916 /* JIS code classification */ 917 unsigned char lb = lead; 918 unsigned char tb = trail; 919 920 /* convert process code to JIS */ 921 # if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS_X) 922 /* process code is SJIS */ 923 if (lb <= 0x9f) 924 lb = (lb - 0x81) * 2 + 0x21; 925 else 926 lb = (lb - 0xc1) * 2 + 0x21; 927 if (tb <= 0x7e) 928 tb -= 0x1f; 929 else if (tb <= 0x9e) 930 tb -= 0x20; 931 else 932 { 933 tb -= 0x7e; 934 lb += 1; 935 } 936 # else 937 /* 938 * XXX: Code page identification can not use with all 939 * system! So, some other encoding information 940 * will be needed. 941 * In japanese: SJIS,EUC,UNICODE,(JIS) 942 * Note that JIS-code system don't use as 943 * process code in most system because it uses 944 * escape sequences(JIS is context depend encoding). 945 */ 946 /* assume process code is JAPANESE-EUC */ 947 lb &= 0x7f; 948 tb &= 0x7f; 949 # endif 950 /* exceptions */ 951 switch (lb << 8 | tb) 952 { 953 case 0x2121: /* ZENKAKU space */ 954 return 0; 955 case 0x2122: /* TOU-TEN (Japanese comma) */ 956 case 0x2123: /* KU-TEN (Japanese period) */ 957 case 0x2124: /* ZENKAKU comma */ 958 case 0x2125: /* ZENKAKU period */ 959 return 1; 960 case 0x213c: /* prolongedsound handled as KATAKANA */ 961 return 13; 962 } 963 /* sieved by KU code */ 964 switch (lb) 965 { 966 case 0x21: 967 case 0x22: 968 /* special symbols */ 969 return 10; 970 case 0x23: 971 /* alpha-numeric */ 972 return 11; 973 case 0x24: 974 /* hiragana */ 975 return 12; 976 case 0x25: 977 /* katakana */ 978 return 13; 979 case 0x26: 980 /* greek */ 981 return 14; 982 case 0x27: 983 /* russian */ 984 return 15; 985 case 0x28: 986 /* lines */ 987 return 16; 988 default: 989 /* kanji */ 990 return 17; 991 } 992 } 993 994 case DBCS_KORU: /* ? */ 995 case DBCS_KOR: 996 { 997 /* KS code classification */ 998 unsigned char c1 = lead; 999 unsigned char c2 = trail; 1000 1001 /* 1002 * 20 : Hangul 1003 * 21 : Hanja 1004 * 22 : Symbols 1005 * 23 : Alpha-numeric/Roman Letter (Full width) 1006 * 24 : Hangul Letter(Alphabet) 1007 * 25 : Roman Numeral/Greek Letter 1008 * 26 : Box Drawings 1009 * 27 : Unit Symbols 1010 * 28 : Circled/Parenthesized Letter 1011 * 29 : Hiragana/Katakana 1012 * 30 : Cyrillic Letter 1013 */ 1014 1015 if (c1 >= 0xB0 && c1 <= 0xC8) 1016 /* Hangul */ 1017 return 20; 1018 #if defined(WIN3264) || defined(WIN32UNIX) 1019 else if (c1 <= 0xA0 || c2 <= 0xA0) 1020 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */ 1021 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE 1022 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 1023 */ 1024 return 20; 1025 #endif 1026 1027 else if (c1 >= 0xCA && c1 <= 0xFD) 1028 /* Hanja */ 1029 return 21; 1030 else switch (c1) 1031 { 1032 case 0xA1: 1033 case 0xA2: 1034 /* Symbols */ 1035 return 22; 1036 case 0xA3: 1037 /* Alpha-numeric */ 1038 return 23; 1039 case 0xA4: 1040 /* Hangul Letter(Alphabet) */ 1041 return 24; 1042 case 0xA5: 1043 /* Roman Numeral/Greek Letter */ 1044 return 25; 1045 case 0xA6: 1046 /* Box Drawings */ 1047 return 26; 1048 case 0xA7: 1049 /* Unit Symbols */ 1050 return 27; 1051 case 0xA8: 1052 case 0xA9: 1053 if (c2 <= 0xAF) 1054 return 25; /* Roman Letter */ 1055 else if (c2 >= 0xF6) 1056 return 22; /* Symbols */ 1057 else 1058 /* Circled/Parenthesized Letter */ 1059 return 28; 1060 case 0xAA: 1061 case 0xAB: 1062 /* Hiragana/Katakana */ 1063 return 29; 1064 case 0xAC: 1065 /* Cyrillic Letter */ 1066 return 30; 1067 } 1068 } 1069 default: 1070 break; 1071 } 1072 return 3; 1073 } 1074 1075 /* 1076 * mb_char2len() function pointer. 1077 * Return length in bytes of character "c". 1078 * Returns 1 for a single-byte character. 1079 */ 1080 int 1081 latin_char2len(int c UNUSED) 1082 { 1083 return 1; 1084 } 1085 1086 static int 1087 dbcs_char2len( 1088 int c) 1089 { 1090 if (c >= 0x100) 1091 return 2; 1092 return 1; 1093 } 1094 1095 /* 1096 * mb_char2bytes() function pointer. 1097 * Convert a character to its bytes. 1098 * Returns the length in bytes. 1099 */ 1100 int 1101 latin_char2bytes(int c, char_u *buf) 1102 { 1103 buf[0] = c; 1104 return 1; 1105 } 1106 1107 static int 1108 dbcs_char2bytes(int c, char_u *buf) 1109 { 1110 if (c >= 0x100) 1111 { 1112 buf[0] = (unsigned)c >> 8; 1113 buf[1] = c; 1114 /* Never use a NUL byte, it causes lots of trouble. It's an invalid 1115 * character anyway. */ 1116 if (buf[1] == NUL) 1117 buf[1] = '\n'; 1118 return 2; 1119 } 1120 buf[0] = c; 1121 return 1; 1122 } 1123 1124 /* 1125 * mb_ptr2len() function pointer. 1126 * Get byte length of character at "*p" but stop at a NUL. 1127 * For UTF-8 this includes following composing characters. 1128 * Returns 0 when *p is NUL. 1129 */ 1130 int 1131 latin_ptr2len(char_u *p) 1132 { 1133 return MB_BYTE2LEN(*p); 1134 } 1135 1136 static int 1137 dbcs_ptr2len( 1138 char_u *p) 1139 { 1140 int len; 1141 1142 /* Check if second byte is not missing. */ 1143 len = MB_BYTE2LEN(*p); 1144 if (len == 2 && p[1] == NUL) 1145 len = 1; 1146 return len; 1147 } 1148 1149 /* 1150 * mb_ptr2len_len() function pointer. 1151 * Like mb_ptr2len(), but limit to read "size" bytes. 1152 * Returns 0 for an empty string. 1153 * Returns 1 for an illegal char or an incomplete byte sequence. 1154 */ 1155 int 1156 latin_ptr2len_len(char_u *p, int size) 1157 { 1158 if (size < 1 || *p == NUL) 1159 return 0; 1160 return 1; 1161 } 1162 1163 static int 1164 dbcs_ptr2len_len(char_u *p, int size) 1165 { 1166 int len; 1167 1168 if (size < 1 || *p == NUL) 1169 return 0; 1170 if (size == 1) 1171 return 1; 1172 /* Check that second byte is not missing. */ 1173 len = MB_BYTE2LEN(*p); 1174 if (len == 2 && p[1] == NUL) 1175 len = 1; 1176 return len; 1177 } 1178 1179 struct interval 1180 { 1181 long first; 1182 long last; 1183 }; 1184 1185 /* 1186 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]". 1187 */ 1188 static int 1189 intable(struct interval *table, size_t size, int c) 1190 { 1191 int mid, bot, top; 1192 1193 /* first quick check for Latin1 etc. characters */ 1194 if (c < table[0].first) 1195 return FALSE; 1196 1197 /* binary search in table */ 1198 bot = 0; 1199 top = (int)(size / sizeof(struct interval) - 1); 1200 while (top >= bot) 1201 { 1202 mid = (bot + top) / 2; 1203 if (table[mid].last < c) 1204 bot = mid + 1; 1205 else if (table[mid].first > c) 1206 top = mid - 1; 1207 else 1208 return TRUE; 1209 } 1210 return FALSE; 1211 } 1212 1213 /* Sorted list of non-overlapping intervals of East Asian Ambiguous 1214 * characters, generated with ../runtime/tools/unicode.vim. */ 1215 static struct interval ambiguous[] = 1216 { 1217 {0x00a1, 0x00a1}, 1218 {0x00a4, 0x00a4}, 1219 {0x00a7, 0x00a8}, 1220 {0x00aa, 0x00aa}, 1221 {0x00ad, 0x00ae}, 1222 {0x00b0, 0x00b4}, 1223 {0x00b6, 0x00ba}, 1224 {0x00bc, 0x00bf}, 1225 {0x00c6, 0x00c6}, 1226 {0x00d0, 0x00d0}, 1227 {0x00d7, 0x00d8}, 1228 {0x00de, 0x00e1}, 1229 {0x00e6, 0x00e6}, 1230 {0x00e8, 0x00ea}, 1231 {0x00ec, 0x00ed}, 1232 {0x00f0, 0x00f0}, 1233 {0x00f2, 0x00f3}, 1234 {0x00f7, 0x00fa}, 1235 {0x00fc, 0x00fc}, 1236 {0x00fe, 0x00fe}, 1237 {0x0101, 0x0101}, 1238 {0x0111, 0x0111}, 1239 {0x0113, 0x0113}, 1240 {0x011b, 0x011b}, 1241 {0x0126, 0x0127}, 1242 {0x012b, 0x012b}, 1243 {0x0131, 0x0133}, 1244 {0x0138, 0x0138}, 1245 {0x013f, 0x0142}, 1246 {0x0144, 0x0144}, 1247 {0x0148, 0x014b}, 1248 {0x014d, 0x014d}, 1249 {0x0152, 0x0153}, 1250 {0x0166, 0x0167}, 1251 {0x016b, 0x016b}, 1252 {0x01ce, 0x01ce}, 1253 {0x01d0, 0x01d0}, 1254 {0x01d2, 0x01d2}, 1255 {0x01d4, 0x01d4}, 1256 {0x01d6, 0x01d6}, 1257 {0x01d8, 0x01d8}, 1258 {0x01da, 0x01da}, 1259 {0x01dc, 0x01dc}, 1260 {0x0251, 0x0251}, 1261 {0x0261, 0x0261}, 1262 {0x02c4, 0x02c4}, 1263 {0x02c7, 0x02c7}, 1264 {0x02c9, 0x02cb}, 1265 {0x02cd, 0x02cd}, 1266 {0x02d0, 0x02d0}, 1267 {0x02d8, 0x02db}, 1268 {0x02dd, 0x02dd}, 1269 {0x02df, 0x02df}, 1270 {0x0300, 0x036f}, 1271 {0x0391, 0x03a1}, 1272 {0x03a3, 0x03a9}, 1273 {0x03b1, 0x03c1}, 1274 {0x03c3, 0x03c9}, 1275 {0x0401, 0x0401}, 1276 {0x0410, 0x044f}, 1277 {0x0451, 0x0451}, 1278 {0x2010, 0x2010}, 1279 {0x2013, 0x2016}, 1280 {0x2018, 0x2019}, 1281 {0x201c, 0x201d}, 1282 {0x2020, 0x2022}, 1283 {0x2024, 0x2027}, 1284 {0x2030, 0x2030}, 1285 {0x2032, 0x2033}, 1286 {0x2035, 0x2035}, 1287 {0x203b, 0x203b}, 1288 {0x203e, 0x203e}, 1289 {0x2074, 0x2074}, 1290 {0x207f, 0x207f}, 1291 {0x2081, 0x2084}, 1292 {0x20ac, 0x20ac}, 1293 {0x2103, 0x2103}, 1294 {0x2105, 0x2105}, 1295 {0x2109, 0x2109}, 1296 {0x2113, 0x2113}, 1297 {0x2116, 0x2116}, 1298 {0x2121, 0x2122}, 1299 {0x2126, 0x2126}, 1300 {0x212b, 0x212b}, 1301 {0x2153, 0x2154}, 1302 {0x215b, 0x215e}, 1303 {0x2160, 0x216b}, 1304 {0x2170, 0x2179}, 1305 {0x2189, 0x2189}, 1306 {0x2190, 0x2199}, 1307 {0x21b8, 0x21b9}, 1308 {0x21d2, 0x21d2}, 1309 {0x21d4, 0x21d4}, 1310 {0x21e7, 0x21e7}, 1311 {0x2200, 0x2200}, 1312 {0x2202, 0x2203}, 1313 {0x2207, 0x2208}, 1314 {0x220b, 0x220b}, 1315 {0x220f, 0x220f}, 1316 {0x2211, 0x2211}, 1317 {0x2215, 0x2215}, 1318 {0x221a, 0x221a}, 1319 {0x221d, 0x2220}, 1320 {0x2223, 0x2223}, 1321 {0x2225, 0x2225}, 1322 {0x2227, 0x222c}, 1323 {0x222e, 0x222e}, 1324 {0x2234, 0x2237}, 1325 {0x223c, 0x223d}, 1326 {0x2248, 0x2248}, 1327 {0x224c, 0x224c}, 1328 {0x2252, 0x2252}, 1329 {0x2260, 0x2261}, 1330 {0x2264, 0x2267}, 1331 {0x226a, 0x226b}, 1332 {0x226e, 0x226f}, 1333 {0x2282, 0x2283}, 1334 {0x2286, 0x2287}, 1335 {0x2295, 0x2295}, 1336 {0x2299, 0x2299}, 1337 {0x22a5, 0x22a5}, 1338 {0x22bf, 0x22bf}, 1339 {0x2312, 0x2312}, 1340 {0x2460, 0x24e9}, 1341 {0x24eb, 0x254b}, 1342 {0x2550, 0x2573}, 1343 {0x2580, 0x258f}, 1344 {0x2592, 0x2595}, 1345 {0x25a0, 0x25a1}, 1346 {0x25a3, 0x25a9}, 1347 {0x25b2, 0x25b3}, 1348 {0x25b6, 0x25b7}, 1349 {0x25bc, 0x25bd}, 1350 {0x25c0, 0x25c1}, 1351 {0x25c6, 0x25c8}, 1352 {0x25cb, 0x25cb}, 1353 {0x25ce, 0x25d1}, 1354 {0x25e2, 0x25e5}, 1355 {0x25ef, 0x25ef}, 1356 {0x2605, 0x2606}, 1357 {0x2609, 0x2609}, 1358 {0x260e, 0x260f}, 1359 {0x261c, 0x261c}, 1360 {0x261e, 0x261e}, 1361 {0x2640, 0x2640}, 1362 {0x2642, 0x2642}, 1363 {0x2660, 0x2661}, 1364 {0x2663, 0x2665}, 1365 {0x2667, 0x266a}, 1366 {0x266c, 0x266d}, 1367 {0x266f, 0x266f}, 1368 {0x269e, 0x269f}, 1369 {0x26bf, 0x26bf}, 1370 {0x26c6, 0x26cd}, 1371 {0x26cf, 0x26d3}, 1372 {0x26d5, 0x26e1}, 1373 {0x26e3, 0x26e3}, 1374 {0x26e8, 0x26e9}, 1375 {0x26eb, 0x26f1}, 1376 {0x26f4, 0x26f4}, 1377 {0x26f6, 0x26f9}, 1378 {0x26fb, 0x26fc}, 1379 {0x26fe, 0x26ff}, 1380 {0x273d, 0x273d}, 1381 {0x2776, 0x277f}, 1382 {0x2b56, 0x2b59}, 1383 {0x3248, 0x324f}, 1384 {0xe000, 0xf8ff}, 1385 {0xfe00, 0xfe0f}, 1386 {0xfffd, 0xfffd}, 1387 {0x1f100, 0x1f10a}, 1388 {0x1f110, 0x1f12d}, 1389 {0x1f130, 0x1f169}, 1390 {0x1f170, 0x1f18d}, 1391 {0x1f18f, 0x1f190}, 1392 {0x1f19b, 0x1f1ac}, 1393 {0xe0100, 0xe01ef}, 1394 {0xf0000, 0xffffd}, 1395 {0x100000, 0x10fffd} 1396 }; 1397 1398 #if defined(FEAT_TERMINAL) || defined(PROTO) 1399 /* 1400 * utf_char2cells() with different argument type for libvterm. 1401 */ 1402 int 1403 utf_uint2cells(UINT32_T c) 1404 { 1405 if (c >= 0x100 && utf_iscomposing((int)c)) 1406 return 0; 1407 return utf_char2cells((int)c); 1408 } 1409 #endif 1410 1411 /* 1412 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1413 * Returns 4 or 6 for an unprintable character. 1414 * Is only correct for characters >= 0x80. 1415 * When p_ambw is "double", return 2 for a character with East Asian Width 1416 * class 'A'(mbiguous). 1417 */ 1418 int 1419 utf_char2cells(int c) 1420 { 1421 /* Sorted list of non-overlapping intervals of East Asian double width 1422 * characters, generated with ../runtime/tools/unicode.vim. */ 1423 static struct interval doublewidth[] = 1424 { 1425 {0x1100, 0x115f}, 1426 {0x231a, 0x231b}, 1427 {0x2329, 0x232a}, 1428 {0x23e9, 0x23ec}, 1429 {0x23f0, 0x23f0}, 1430 {0x23f3, 0x23f3}, 1431 {0x25fd, 0x25fe}, 1432 {0x2614, 0x2615}, 1433 {0x2648, 0x2653}, 1434 {0x267f, 0x267f}, 1435 {0x2693, 0x2693}, 1436 {0x26a1, 0x26a1}, 1437 {0x26aa, 0x26ab}, 1438 {0x26bd, 0x26be}, 1439 {0x26c4, 0x26c5}, 1440 {0x26ce, 0x26ce}, 1441 {0x26d4, 0x26d4}, 1442 {0x26ea, 0x26ea}, 1443 {0x26f2, 0x26f3}, 1444 {0x26f5, 0x26f5}, 1445 {0x26fa, 0x26fa}, 1446 {0x26fd, 0x26fd}, 1447 {0x2705, 0x2705}, 1448 {0x270a, 0x270b}, 1449 {0x2728, 0x2728}, 1450 {0x274c, 0x274c}, 1451 {0x274e, 0x274e}, 1452 {0x2753, 0x2755}, 1453 {0x2757, 0x2757}, 1454 {0x2795, 0x2797}, 1455 {0x27b0, 0x27b0}, 1456 {0x27bf, 0x27bf}, 1457 {0x2b1b, 0x2b1c}, 1458 {0x2b50, 0x2b50}, 1459 {0x2b55, 0x2b55}, 1460 {0x2e80, 0x2e99}, 1461 {0x2e9b, 0x2ef3}, 1462 {0x2f00, 0x2fd5}, 1463 {0x2ff0, 0x2ffb}, 1464 {0x3000, 0x303e}, 1465 {0x3041, 0x3096}, 1466 {0x3099, 0x30ff}, 1467 {0x3105, 0x312e}, 1468 {0x3131, 0x318e}, 1469 {0x3190, 0x31ba}, 1470 {0x31c0, 0x31e3}, 1471 {0x31f0, 0x321e}, 1472 {0x3220, 0x3247}, 1473 {0x3250, 0x32fe}, 1474 {0x3300, 0x4dbf}, 1475 {0x4e00, 0xa48c}, 1476 {0xa490, 0xa4c6}, 1477 {0xa960, 0xa97c}, 1478 {0xac00, 0xd7a3}, 1479 {0xf900, 0xfaff}, 1480 {0xfe10, 0xfe19}, 1481 {0xfe30, 0xfe52}, 1482 {0xfe54, 0xfe66}, 1483 {0xfe68, 0xfe6b}, 1484 {0xff01, 0xff60}, 1485 {0xffe0, 0xffe6}, 1486 {0x16fe0, 0x16fe1}, 1487 {0x17000, 0x187ec}, 1488 {0x18800, 0x18af2}, 1489 {0x1b000, 0x1b11e}, 1490 {0x1b170, 0x1b2fb}, 1491 {0x1f004, 0x1f004}, 1492 {0x1f0cf, 0x1f0cf}, 1493 {0x1f18e, 0x1f18e}, 1494 {0x1f191, 0x1f19a}, 1495 {0x1f200, 0x1f202}, 1496 {0x1f210, 0x1f23b}, 1497 {0x1f240, 0x1f248}, 1498 {0x1f250, 0x1f251}, 1499 {0x1f260, 0x1f265}, 1500 {0x1f300, 0x1f320}, 1501 {0x1f32d, 0x1f335}, 1502 {0x1f337, 0x1f37c}, 1503 {0x1f37e, 0x1f393}, 1504 {0x1f3a0, 0x1f3ca}, 1505 {0x1f3cf, 0x1f3d3}, 1506 {0x1f3e0, 0x1f3f0}, 1507 {0x1f3f4, 0x1f3f4}, 1508 {0x1f3f8, 0x1f43e}, 1509 {0x1f440, 0x1f440}, 1510 {0x1f442, 0x1f4fc}, 1511 {0x1f4ff, 0x1f53d}, 1512 {0x1f54b, 0x1f54e}, 1513 {0x1f550, 0x1f567}, 1514 {0x1f57a, 0x1f57a}, 1515 {0x1f595, 0x1f596}, 1516 {0x1f5a4, 0x1f5a4}, 1517 {0x1f5fb, 0x1f64f}, 1518 {0x1f680, 0x1f6c5}, 1519 {0x1f6cc, 0x1f6cc}, 1520 {0x1f6d0, 0x1f6d2}, 1521 {0x1f6eb, 0x1f6ec}, 1522 {0x1f6f4, 0x1f6f8}, 1523 {0x1f910, 0x1f93e}, 1524 {0x1f940, 0x1f94c}, 1525 {0x1f950, 0x1f96b}, 1526 {0x1f980, 0x1f997}, 1527 {0x1f9c0, 0x1f9c0}, 1528 {0x1f9d0, 0x1f9e6}, 1529 {0x20000, 0x2fffd}, 1530 {0x30000, 0x3fffd} 1531 }; 1532 1533 /* Sorted list of non-overlapping intervals of Emoji characters that don't 1534 * have ambiguous or double width, 1535 * based on http://unicode.org/emoji/charts/emoji-list.html */ 1536 static struct interval emoji_width[] = 1537 { 1538 {0x1f1e6, 0x1f1ff}, 1539 {0x1f321, 0x1f321}, 1540 {0x1f324, 0x1f32c}, 1541 {0x1f336, 0x1f336}, 1542 {0x1f37d, 0x1f37d}, 1543 {0x1f396, 0x1f397}, 1544 {0x1f399, 0x1f39b}, 1545 {0x1f39e, 0x1f39f}, 1546 {0x1f3cb, 0x1f3ce}, 1547 {0x1f3d4, 0x1f3df}, 1548 {0x1f3f3, 0x1f3f5}, 1549 {0x1f3f7, 0x1f3f7}, 1550 {0x1f43f, 0x1f43f}, 1551 {0x1f441, 0x1f441}, 1552 {0x1f4fd, 0x1f4fd}, 1553 {0x1f549, 0x1f54a}, 1554 {0x1f56f, 0x1f570}, 1555 {0x1f573, 0x1f579}, 1556 {0x1f587, 0x1f587}, 1557 {0x1f58a, 0x1f58d}, 1558 {0x1f590, 0x1f590}, 1559 {0x1f5a5, 0x1f5a5}, 1560 {0x1f5a8, 0x1f5a8}, 1561 {0x1f5b1, 0x1f5b2}, 1562 {0x1f5bc, 0x1f5bc}, 1563 {0x1f5c2, 0x1f5c4}, 1564 {0x1f5d1, 0x1f5d3}, 1565 {0x1f5dc, 0x1f5de}, 1566 {0x1f5e1, 0x1f5e1}, 1567 {0x1f5e3, 0x1f5e3}, 1568 {0x1f5e8, 0x1f5e8}, 1569 {0x1f5ef, 0x1f5ef}, 1570 {0x1f5f3, 0x1f5f3}, 1571 {0x1f5fa, 0x1f5fa}, 1572 {0x1f6cb, 0x1f6cf}, 1573 {0x1f6e0, 0x1f6e5}, 1574 {0x1f6e9, 0x1f6e9}, 1575 {0x1f6f0, 0x1f6f0}, 1576 {0x1f6f3, 0x1f6f3} 1577 }; 1578 1579 if (c >= 0x100) 1580 { 1581 #ifdef USE_WCHAR_FUNCTIONS 1582 /* 1583 * Assume the library function wcwidth() works better than our own 1584 * stuff. It should return 1 for ambiguous width chars! 1585 */ 1586 int n = wcwidth(c); 1587 1588 if (n < 0) 1589 return 6; /* unprintable, displays <xxxx> */ 1590 if (n > 1) 1591 return n; 1592 #else 1593 if (!utf_printable(c)) 1594 return 6; /* unprintable, displays <xxxx> */ 1595 if (intable(doublewidth, sizeof(doublewidth), c)) 1596 return 2; 1597 #endif 1598 if (p_emoji && intable(emoji_width, sizeof(emoji_width), c)) 1599 return 2; 1600 } 1601 1602 /* Characters below 0x100 are influenced by 'isprint' option */ 1603 else if (c >= 0x80 && !vim_isprintc(c)) 1604 return 4; /* unprintable, displays <xx> */ 1605 1606 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c)) 1607 return 2; 1608 1609 return 1; 1610 } 1611 1612 /* 1613 * mb_ptr2cells() function pointer. 1614 * Return the number of display cells character at "*p" occupies. 1615 * This doesn't take care of unprintable characters, use ptr2cells() for that. 1616 */ 1617 int 1618 latin_ptr2cells(char_u *p UNUSED) 1619 { 1620 return 1; 1621 } 1622 1623 int 1624 utf_ptr2cells( 1625 char_u *p) 1626 { 1627 int c; 1628 1629 /* Need to convert to a wide character. */ 1630 if (*p >= 0x80) 1631 { 1632 c = utf_ptr2char(p); 1633 /* An illegal byte is displayed as <xx>. */ 1634 if (utf_ptr2len(p) == 1 || c == NUL) 1635 return 4; 1636 /* If the char is ASCII it must be an overlong sequence. */ 1637 if (c < 0x80) 1638 return char2cells(c); 1639 return utf_char2cells(c); 1640 } 1641 return 1; 1642 } 1643 1644 int 1645 dbcs_ptr2cells(char_u *p) 1646 { 1647 /* Number of cells is equal to number of bytes, except for euc-jp when 1648 * the first byte is 0x8e. */ 1649 if (enc_dbcs == DBCS_JPNU && *p == 0x8e) 1650 return 1; 1651 return MB_BYTE2LEN(*p); 1652 } 1653 1654 /* 1655 * mb_ptr2cells_len() function pointer. 1656 * Like mb_ptr2cells(), but limit string length to "size". 1657 * For an empty string or truncated character returns 1. 1658 */ 1659 int 1660 latin_ptr2cells_len(char_u *p UNUSED, int size UNUSED) 1661 { 1662 return 1; 1663 } 1664 1665 static int 1666 utf_ptr2cells_len(char_u *p, int size) 1667 { 1668 int c; 1669 1670 /* Need to convert to a wide character. */ 1671 if (size > 0 && *p >= 0x80) 1672 { 1673 if (utf_ptr2len_len(p, size) < utf8len_tab[*p]) 1674 return 1; /* truncated */ 1675 c = utf_ptr2char(p); 1676 /* An illegal byte is displayed as <xx>. */ 1677 if (utf_ptr2len(p) == 1 || c == NUL) 1678 return 4; 1679 /* If the char is ASCII it must be an overlong sequence. */ 1680 if (c < 0x80) 1681 return char2cells(c); 1682 return utf_char2cells(c); 1683 } 1684 return 1; 1685 } 1686 1687 static int 1688 dbcs_ptr2cells_len(char_u *p, int size) 1689 { 1690 /* Number of cells is equal to number of bytes, except for euc-jp when 1691 * the first byte is 0x8e. */ 1692 if (size <= 1 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)) 1693 return 1; 1694 return MB_BYTE2LEN(*p); 1695 } 1696 1697 /* 1698 * mb_char2cells() function pointer. 1699 * Return the number of display cells character "c" occupies. 1700 * Only takes care of multi-byte chars, not "^C" and such. 1701 */ 1702 int 1703 latin_char2cells(int c UNUSED) 1704 { 1705 return 1; 1706 } 1707 1708 static int 1709 dbcs_char2cells(int c) 1710 { 1711 /* Number of cells is equal to number of bytes, except for euc-jp when 1712 * the first byte is 0x8e. */ 1713 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e) 1714 return 1; 1715 /* use the first byte */ 1716 return MB_BYTE2LEN((unsigned)c >> 8); 1717 } 1718 1719 /* 1720 * Return the number of cells occupied by string "p". 1721 * Stop at a NUL character. When "len" >= 0 stop at character "p[len]". 1722 */ 1723 int 1724 mb_string2cells(char_u *p, int len) 1725 { 1726 int i; 1727 int clen = 0; 1728 1729 for (i = 0; (len < 0 || i < len) && p[i] != NUL; i += (*mb_ptr2len)(p + i)) 1730 clen += (*mb_ptr2cells)(p + i); 1731 return clen; 1732 } 1733 1734 /* 1735 * mb_off2cells() function pointer. 1736 * Return number of display cells for char at ScreenLines[off]. 1737 * We make sure that the offset used is less than "max_off". 1738 */ 1739 int 1740 latin_off2cells(unsigned off UNUSED, unsigned max_off UNUSED) 1741 { 1742 return 1; 1743 } 1744 1745 int 1746 dbcs_off2cells(unsigned off, unsigned max_off) 1747 { 1748 /* never check beyond end of the line */ 1749 if (off >= max_off) 1750 return 1; 1751 1752 /* Number of cells is equal to number of bytes, except for euc-jp when 1753 * the first byte is 0x8e. */ 1754 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e) 1755 return 1; 1756 return MB_BYTE2LEN(ScreenLines[off]); 1757 } 1758 1759 int 1760 utf_off2cells(unsigned off, unsigned max_off) 1761 { 1762 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1; 1763 } 1764 1765 /* 1766 * mb_ptr2char() function pointer. 1767 * Convert a byte sequence into a character. 1768 */ 1769 int 1770 latin_ptr2char(char_u *p) 1771 { 1772 return *p; 1773 } 1774 1775 static int 1776 dbcs_ptr2char(char_u *p) 1777 { 1778 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL) 1779 return (p[0] << 8) + p[1]; 1780 return *p; 1781 } 1782 1783 /* 1784 * Convert a UTF-8 byte sequence to a wide character. 1785 * If the sequence is illegal or truncated by a NUL the first byte is 1786 * returned. 1787 * Does not include composing characters, of course. 1788 */ 1789 int 1790 utf_ptr2char(char_u *p) 1791 { 1792 int len; 1793 1794 if (p[0] < 0x80) /* be quick for ASCII */ 1795 return p[0]; 1796 1797 len = utf8len_tab_zero[p[0]]; 1798 if (len > 1 && (p[1] & 0xc0) == 0x80) 1799 { 1800 if (len == 2) 1801 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); 1802 if ((p[2] & 0xc0) == 0x80) 1803 { 1804 if (len == 3) 1805 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) 1806 + (p[2] & 0x3f); 1807 if ((p[3] & 0xc0) == 0x80) 1808 { 1809 if (len == 4) 1810 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) 1811 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); 1812 if ((p[4] & 0xc0) == 0x80) 1813 { 1814 if (len == 5) 1815 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) 1816 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) 1817 + (p[4] & 0x3f); 1818 if ((p[5] & 0xc0) == 0x80 && len == 6) 1819 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) 1820 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) 1821 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); 1822 } 1823 } 1824 } 1825 } 1826 /* Illegal value, just return the first byte */ 1827 return p[0]; 1828 } 1829 1830 /* 1831 * Convert a UTF-8 byte sequence to a wide character. 1832 * String is assumed to be terminated by NUL or after "n" bytes, whichever 1833 * comes first. 1834 * The function is safe in the sense that it never accesses memory beyond the 1835 * first "n" bytes of "s". 1836 * 1837 * On success, returns decoded codepoint, advances "s" to the beginning of 1838 * next character and decreases "n" accordingly. 1839 * 1840 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past 1841 * NUL byte. 1842 * 1843 * If byte sequence is illegal or incomplete, returns -1 and does not advance 1844 * "s". 1845 */ 1846 static int 1847 utf_safe_read_char_adv(char_u **s, size_t *n) 1848 { 1849 int c, k; 1850 1851 if (*n == 0) /* end of buffer */ 1852 return 0; 1853 1854 k = utf8len_tab_zero[**s]; 1855 1856 if (k == 1) 1857 { 1858 /* ASCII character or NUL */ 1859 (*n)--; 1860 return *(*s)++; 1861 } 1862 1863 if ((size_t)k <= *n) 1864 { 1865 /* We have a multibyte sequence and it isn't truncated by buffer 1866 * limits so utf_ptr2char() is safe to use. Or the first byte is 1867 * illegal (k=0), and it's also safe to use utf_ptr2char(). */ 1868 c = utf_ptr2char(*s); 1869 1870 /* On failure, utf_ptr2char() returns the first byte, so here we 1871 * check equality with the first byte. The only non-ASCII character 1872 * which equals the first byte of its own UTF-8 representation is 1873 * U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too. 1874 * It's safe even if n=1, else we would have k=2 > n. */ 1875 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83)) 1876 { 1877 /* byte sequence was successfully decoded */ 1878 *s += k; 1879 *n -= k; 1880 return c; 1881 } 1882 } 1883 1884 /* byte sequence is incomplete or illegal */ 1885 return -1; 1886 } 1887 1888 /* 1889 * Get character at **pp and advance *pp to the next character. 1890 * Note: composing characters are skipped! 1891 */ 1892 int 1893 mb_ptr2char_adv(char_u **pp) 1894 { 1895 int c; 1896 1897 c = (*mb_ptr2char)(*pp); 1898 *pp += (*mb_ptr2len)(*pp); 1899 return c; 1900 } 1901 1902 /* 1903 * Get character at **pp and advance *pp to the next character. 1904 * Note: composing characters are returned as separate characters. 1905 */ 1906 int 1907 mb_cptr2char_adv(char_u **pp) 1908 { 1909 int c; 1910 1911 c = (*mb_ptr2char)(*pp); 1912 if (enc_utf8) 1913 *pp += utf_ptr2len(*pp); 1914 else 1915 *pp += (*mb_ptr2len)(*pp); 1916 return c; 1917 } 1918 1919 #if defined(FEAT_ARABIC) || defined(PROTO) 1920 /* 1921 * Check whether we are dealing with Arabic combining characters. 1922 * Note: these are NOT really composing characters! 1923 */ 1924 int 1925 arabic_combine( 1926 int one, /* first character */ 1927 int two) /* character just after "one" */ 1928 { 1929 if (one == a_LAM) 1930 return arabic_maycombine(two); 1931 return FALSE; 1932 } 1933 1934 /* 1935 * Check whether we are dealing with a character that could be regarded as an 1936 * Arabic combining character, need to check the character before this. 1937 */ 1938 int 1939 arabic_maycombine(int two) 1940 { 1941 if (p_arshape && !p_tbidi) 1942 return (two == a_ALEF_MADDA 1943 || two == a_ALEF_HAMZA_ABOVE 1944 || two == a_ALEF_HAMZA_BELOW 1945 || two == a_ALEF); 1946 return FALSE; 1947 } 1948 1949 /* 1950 * Check if the character pointed to by "p2" is a composing character when it 1951 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which 1952 * behaves like a composing character. 1953 */ 1954 int 1955 utf_composinglike(char_u *p1, char_u *p2) 1956 { 1957 int c2; 1958 1959 c2 = utf_ptr2char(p2); 1960 if (utf_iscomposing(c2)) 1961 return TRUE; 1962 if (!arabic_maycombine(c2)) 1963 return FALSE; 1964 return arabic_combine(utf_ptr2char(p1), c2); 1965 } 1966 #endif 1967 1968 /* 1969 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1970 * composing characters. 1971 */ 1972 int 1973 utfc_ptr2char( 1974 char_u *p, 1975 int *pcc) /* return: composing chars, last one is 0 */ 1976 { 1977 int len; 1978 int c; 1979 int cc; 1980 int i = 0; 1981 1982 c = utf_ptr2char(p); 1983 len = utf_ptr2len(p); 1984 1985 /* Only accept a composing char when the first char isn't illegal. */ 1986 if ((len > 1 || *p < 0x80) 1987 && p[len] >= 0x80 1988 && UTF_COMPOSINGLIKE(p, p + len)) 1989 { 1990 cc = utf_ptr2char(p + len); 1991 for (;;) 1992 { 1993 pcc[i++] = cc; 1994 if (i == MAX_MCO) 1995 break; 1996 len += utf_ptr2len(p + len); 1997 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1998 break; 1999 } 2000 } 2001 2002 if (i < MAX_MCO) /* last composing char must be 0 */ 2003 pcc[i] = 0; 2004 2005 return c; 2006 } 2007 2008 /* 2009 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 2010 * composing characters. Use no more than p[maxlen]. 2011 */ 2012 int 2013 utfc_ptr2char_len( 2014 char_u *p, 2015 int *pcc, /* return: composing chars, last one is 0 */ 2016 int maxlen) 2017 { 2018 int len; 2019 int c; 2020 int cc; 2021 int i = 0; 2022 2023 c = utf_ptr2char(p); 2024 len = utf_ptr2len_len(p, maxlen); 2025 /* Only accept a composing char when the first char isn't illegal. */ 2026 if ((len > 1 || *p < 0x80) 2027 && len < maxlen 2028 && p[len] >= 0x80 2029 && UTF_COMPOSINGLIKE(p, p + len)) 2030 { 2031 cc = utf_ptr2char(p + len); 2032 for (;;) 2033 { 2034 pcc[i++] = cc; 2035 if (i == MAX_MCO) 2036 break; 2037 len += utf_ptr2len_len(p + len, maxlen - len); 2038 if (len >= maxlen 2039 || p[len] < 0x80 2040 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 2041 break; 2042 } 2043 } 2044 2045 if (i < MAX_MCO) /* last composing char must be 0 */ 2046 pcc[i] = 0; 2047 2048 return c; 2049 } 2050 2051 /* 2052 * Convert the character at screen position "off" to a sequence of bytes. 2053 * Includes the composing characters. 2054 * "buf" must at least have the length MB_MAXBYTES + 1. 2055 * Only to be used when ScreenLinesUC[off] != 0. 2056 * Returns the produced number of bytes. 2057 */ 2058 int 2059 utfc_char2bytes(int off, char_u *buf) 2060 { 2061 int len; 2062 int i; 2063 2064 len = utf_char2bytes(ScreenLinesUC[off], buf); 2065 for (i = 0; i < Screen_mco; ++i) 2066 { 2067 if (ScreenLinesC[i][off] == 0) 2068 break; 2069 len += utf_char2bytes(ScreenLinesC[i][off], buf + len); 2070 } 2071 return len; 2072 } 2073 2074 /* 2075 * Get the length of a UTF-8 byte sequence, not including any following 2076 * composing characters. 2077 * Returns 0 for "". 2078 * Returns 1 for an illegal byte sequence. 2079 */ 2080 int 2081 utf_ptr2len(char_u *p) 2082 { 2083 int len; 2084 int i; 2085 2086 if (*p == NUL) 2087 return 0; 2088 len = utf8len_tab[*p]; 2089 for (i = 1; i < len; ++i) 2090 if ((p[i] & 0xc0) != 0x80) 2091 return 1; 2092 return len; 2093 } 2094 2095 /* 2096 * Return length of UTF-8 character, obtained from the first byte. 2097 * "b" must be between 0 and 255! 2098 * Returns 1 for an invalid first byte value. 2099 */ 2100 int 2101 utf_byte2len(int b) 2102 { 2103 return utf8len_tab[b]; 2104 } 2105 2106 /* 2107 * Get the length of UTF-8 byte sequence "p[size]". Does not include any 2108 * following composing characters. 2109 * Returns 1 for "". 2110 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.). 2111 * Returns number > "size" for an incomplete byte sequence. 2112 * Never returns zero. 2113 */ 2114 int 2115 utf_ptr2len_len(char_u *p, int size) 2116 { 2117 int len; 2118 int i; 2119 int m; 2120 2121 len = utf8len_tab[*p]; 2122 if (len == 1) 2123 return 1; /* NUL, ascii or illegal lead byte */ 2124 if (len > size) 2125 m = size; /* incomplete byte sequence. */ 2126 else 2127 m = len; 2128 for (i = 1; i < m; ++i) 2129 if ((p[i] & 0xc0) != 0x80) 2130 return 1; 2131 return len; 2132 } 2133 2134 /* 2135 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. 2136 * This includes following composing characters. 2137 */ 2138 int 2139 utfc_ptr2len(char_u *p) 2140 { 2141 int len; 2142 int b0 = *p; 2143 #ifdef FEAT_ARABIC 2144 int prevlen; 2145 #endif 2146 2147 if (b0 == NUL) 2148 return 0; 2149 if (b0 < 0x80 && p[1] < 0x80) /* be quick for ASCII */ 2150 return 1; 2151 2152 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2153 len = utf_ptr2len(p); 2154 2155 /* Check for illegal byte. */ 2156 if (len == 1 && b0 >= 0x80) 2157 return 1; 2158 2159 /* 2160 * Check for composing characters. We can handle only the first six, but 2161 * skip all of them (otherwise the cursor would get stuck). 2162 */ 2163 #ifdef FEAT_ARABIC 2164 prevlen = 0; 2165 #endif 2166 for (;;) 2167 { 2168 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2169 return len; 2170 2171 /* Skip over composing char */ 2172 #ifdef FEAT_ARABIC 2173 prevlen = len; 2174 #endif 2175 len += utf_ptr2len(p + len); 2176 } 2177 } 2178 2179 /* 2180 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" 2181 * takes. This includes following composing characters. 2182 * Returns 0 for an empty string. 2183 * Returns 1 for an illegal char or an incomplete byte sequence. 2184 */ 2185 int 2186 utfc_ptr2len_len(char_u *p, int size) 2187 { 2188 int len; 2189 #ifdef FEAT_ARABIC 2190 int prevlen; 2191 #endif 2192 2193 if (size < 1 || *p == NUL) 2194 return 0; 2195 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) /* be quick for ASCII */ 2196 return 1; 2197 2198 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2199 len = utf_ptr2len_len(p, size); 2200 2201 /* Check for illegal byte and incomplete byte sequence. */ 2202 if ((len == 1 && p[0] >= 0x80) || len > size) 2203 return 1; 2204 2205 /* 2206 * Check for composing characters. We can handle only the first six, but 2207 * skip all of them (otherwise the cursor would get stuck). 2208 */ 2209 #ifdef FEAT_ARABIC 2210 prevlen = 0; 2211 #endif 2212 while (len < size) 2213 { 2214 int len_next_char; 2215 2216 if (p[len] < 0x80) 2217 break; 2218 2219 /* 2220 * Next character length should not go beyond size to ensure that 2221 * UTF_COMPOSINGLIKE(...) does not read beyond size. 2222 */ 2223 len_next_char = utf_ptr2len_len(p + len, size - len); 2224 if (len_next_char > size - len) 2225 break; 2226 2227 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2228 break; 2229 2230 /* Skip over composing char */ 2231 #ifdef FEAT_ARABIC 2232 prevlen = len; 2233 #endif 2234 len += len_next_char; 2235 } 2236 return len; 2237 } 2238 2239 /* 2240 * Return the number of bytes the UTF-8 encoding of character "c" takes. 2241 * This does not include composing characters. 2242 */ 2243 int 2244 utf_char2len(int c) 2245 { 2246 if (c < 0x80) 2247 return 1; 2248 if (c < 0x800) 2249 return 2; 2250 if (c < 0x10000) 2251 return 3; 2252 if (c < 0x200000) 2253 return 4; 2254 if (c < 0x4000000) 2255 return 5; 2256 return 6; 2257 } 2258 2259 /* 2260 * Convert Unicode character "c" to UTF-8 string in "buf[]". 2261 * Returns the number of bytes. 2262 * This does not include composing characters. 2263 */ 2264 int 2265 utf_char2bytes(int c, char_u *buf) 2266 { 2267 if (c < 0x80) /* 7 bits */ 2268 { 2269 buf[0] = c; 2270 return 1; 2271 } 2272 if (c < 0x800) /* 11 bits */ 2273 { 2274 buf[0] = 0xc0 + ((unsigned)c >> 6); 2275 buf[1] = 0x80 + (c & 0x3f); 2276 return 2; 2277 } 2278 if (c < 0x10000) /* 16 bits */ 2279 { 2280 buf[0] = 0xe0 + ((unsigned)c >> 12); 2281 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2282 buf[2] = 0x80 + (c & 0x3f); 2283 return 3; 2284 } 2285 if (c < 0x200000) /* 21 bits */ 2286 { 2287 buf[0] = 0xf0 + ((unsigned)c >> 18); 2288 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2289 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2290 buf[3] = 0x80 + (c & 0x3f); 2291 return 4; 2292 } 2293 if (c < 0x4000000) /* 26 bits */ 2294 { 2295 buf[0] = 0xf8 + ((unsigned)c >> 24); 2296 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2297 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2298 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2299 buf[4] = 0x80 + (c & 0x3f); 2300 return 5; 2301 } 2302 /* 31 bits */ 2303 buf[0] = 0xfc + ((unsigned)c >> 30); 2304 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f); 2305 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2306 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2307 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2308 buf[5] = 0x80 + (c & 0x3f); 2309 return 6; 2310 } 2311 2312 #if defined(FEAT_TERMINAL) || defined(PROTO) 2313 /* 2314 * utf_iscomposing() with different argument type for libvterm. 2315 */ 2316 int 2317 utf_iscomposing_uint(UINT32_T c) 2318 { 2319 return utf_iscomposing((int)c); 2320 } 2321 #endif 2322 2323 /* 2324 * Return TRUE if "c" is a composing UTF-8 character. This means it will be 2325 * drawn on top of the preceding character. 2326 * Based on code from Markus Kuhn. 2327 */ 2328 int 2329 utf_iscomposing(int c) 2330 { 2331 /* Sorted list of non-overlapping intervals. 2332 * Generated by ../runtime/tools/unicode.vim. */ 2333 static struct interval combining[] = 2334 { 2335 {0x0300, 0x036f}, 2336 {0x0483, 0x0489}, 2337 {0x0591, 0x05bd}, 2338 {0x05bf, 0x05bf}, 2339 {0x05c1, 0x05c2}, 2340 {0x05c4, 0x05c5}, 2341 {0x05c7, 0x05c7}, 2342 {0x0610, 0x061a}, 2343 {0x064b, 0x065f}, 2344 {0x0670, 0x0670}, 2345 {0x06d6, 0x06dc}, 2346 {0x06df, 0x06e4}, 2347 {0x06e7, 0x06e8}, 2348 {0x06ea, 0x06ed}, 2349 {0x0711, 0x0711}, 2350 {0x0730, 0x074a}, 2351 {0x07a6, 0x07b0}, 2352 {0x07eb, 0x07f3}, 2353 {0x0816, 0x0819}, 2354 {0x081b, 0x0823}, 2355 {0x0825, 0x0827}, 2356 {0x0829, 0x082d}, 2357 {0x0859, 0x085b}, 2358 {0x08d4, 0x08e1}, 2359 {0x08e3, 0x0903}, 2360 {0x093a, 0x093c}, 2361 {0x093e, 0x094f}, 2362 {0x0951, 0x0957}, 2363 {0x0962, 0x0963}, 2364 {0x0981, 0x0983}, 2365 {0x09bc, 0x09bc}, 2366 {0x09be, 0x09c4}, 2367 {0x09c7, 0x09c8}, 2368 {0x09cb, 0x09cd}, 2369 {0x09d7, 0x09d7}, 2370 {0x09e2, 0x09e3}, 2371 {0x0a01, 0x0a03}, 2372 {0x0a3c, 0x0a3c}, 2373 {0x0a3e, 0x0a42}, 2374 {0x0a47, 0x0a48}, 2375 {0x0a4b, 0x0a4d}, 2376 {0x0a51, 0x0a51}, 2377 {0x0a70, 0x0a71}, 2378 {0x0a75, 0x0a75}, 2379 {0x0a81, 0x0a83}, 2380 {0x0abc, 0x0abc}, 2381 {0x0abe, 0x0ac5}, 2382 {0x0ac7, 0x0ac9}, 2383 {0x0acb, 0x0acd}, 2384 {0x0ae2, 0x0ae3}, 2385 {0x0afa, 0x0aff}, 2386 {0x0b01, 0x0b03}, 2387 {0x0b3c, 0x0b3c}, 2388 {0x0b3e, 0x0b44}, 2389 {0x0b47, 0x0b48}, 2390 {0x0b4b, 0x0b4d}, 2391 {0x0b56, 0x0b57}, 2392 {0x0b62, 0x0b63}, 2393 {0x0b82, 0x0b82}, 2394 {0x0bbe, 0x0bc2}, 2395 {0x0bc6, 0x0bc8}, 2396 {0x0bca, 0x0bcd}, 2397 {0x0bd7, 0x0bd7}, 2398 {0x0c00, 0x0c03}, 2399 {0x0c3e, 0x0c44}, 2400 {0x0c46, 0x0c48}, 2401 {0x0c4a, 0x0c4d}, 2402 {0x0c55, 0x0c56}, 2403 {0x0c62, 0x0c63}, 2404 {0x0c81, 0x0c83}, 2405 {0x0cbc, 0x0cbc}, 2406 {0x0cbe, 0x0cc4}, 2407 {0x0cc6, 0x0cc8}, 2408 {0x0cca, 0x0ccd}, 2409 {0x0cd5, 0x0cd6}, 2410 {0x0ce2, 0x0ce3}, 2411 {0x0d00, 0x0d03}, 2412 {0x0d3b, 0x0d3c}, 2413 {0x0d3e, 0x0d44}, 2414 {0x0d46, 0x0d48}, 2415 {0x0d4a, 0x0d4d}, 2416 {0x0d57, 0x0d57}, 2417 {0x0d62, 0x0d63}, 2418 {0x0d82, 0x0d83}, 2419 {0x0dca, 0x0dca}, 2420 {0x0dcf, 0x0dd4}, 2421 {0x0dd6, 0x0dd6}, 2422 {0x0dd8, 0x0ddf}, 2423 {0x0df2, 0x0df3}, 2424 {0x0e31, 0x0e31}, 2425 {0x0e34, 0x0e3a}, 2426 {0x0e47, 0x0e4e}, 2427 {0x0eb1, 0x0eb1}, 2428 {0x0eb4, 0x0eb9}, 2429 {0x0ebb, 0x0ebc}, 2430 {0x0ec8, 0x0ecd}, 2431 {0x0f18, 0x0f19}, 2432 {0x0f35, 0x0f35}, 2433 {0x0f37, 0x0f37}, 2434 {0x0f39, 0x0f39}, 2435 {0x0f3e, 0x0f3f}, 2436 {0x0f71, 0x0f84}, 2437 {0x0f86, 0x0f87}, 2438 {0x0f8d, 0x0f97}, 2439 {0x0f99, 0x0fbc}, 2440 {0x0fc6, 0x0fc6}, 2441 {0x102b, 0x103e}, 2442 {0x1056, 0x1059}, 2443 {0x105e, 0x1060}, 2444 {0x1062, 0x1064}, 2445 {0x1067, 0x106d}, 2446 {0x1071, 0x1074}, 2447 {0x1082, 0x108d}, 2448 {0x108f, 0x108f}, 2449 {0x109a, 0x109d}, 2450 {0x135d, 0x135f}, 2451 {0x1712, 0x1714}, 2452 {0x1732, 0x1734}, 2453 {0x1752, 0x1753}, 2454 {0x1772, 0x1773}, 2455 {0x17b4, 0x17d3}, 2456 {0x17dd, 0x17dd}, 2457 {0x180b, 0x180d}, 2458 {0x1885, 0x1886}, 2459 {0x18a9, 0x18a9}, 2460 {0x1920, 0x192b}, 2461 {0x1930, 0x193b}, 2462 {0x1a17, 0x1a1b}, 2463 {0x1a55, 0x1a5e}, 2464 {0x1a60, 0x1a7c}, 2465 {0x1a7f, 0x1a7f}, 2466 {0x1ab0, 0x1abe}, 2467 {0x1b00, 0x1b04}, 2468 {0x1b34, 0x1b44}, 2469 {0x1b6b, 0x1b73}, 2470 {0x1b80, 0x1b82}, 2471 {0x1ba1, 0x1bad}, 2472 {0x1be6, 0x1bf3}, 2473 {0x1c24, 0x1c37}, 2474 {0x1cd0, 0x1cd2}, 2475 {0x1cd4, 0x1ce8}, 2476 {0x1ced, 0x1ced}, 2477 {0x1cf2, 0x1cf4}, 2478 {0x1cf7, 0x1cf9}, 2479 {0x1dc0, 0x1df9}, 2480 {0x1dfb, 0x1dff}, 2481 {0x20d0, 0x20f0}, 2482 {0x2cef, 0x2cf1}, 2483 {0x2d7f, 0x2d7f}, 2484 {0x2de0, 0x2dff}, 2485 {0x302a, 0x302f}, 2486 {0x3099, 0x309a}, 2487 {0xa66f, 0xa672}, 2488 {0xa674, 0xa67d}, 2489 {0xa69e, 0xa69f}, 2490 {0xa6f0, 0xa6f1}, 2491 {0xa802, 0xa802}, 2492 {0xa806, 0xa806}, 2493 {0xa80b, 0xa80b}, 2494 {0xa823, 0xa827}, 2495 {0xa880, 0xa881}, 2496 {0xa8b4, 0xa8c5}, 2497 {0xa8e0, 0xa8f1}, 2498 {0xa926, 0xa92d}, 2499 {0xa947, 0xa953}, 2500 {0xa980, 0xa983}, 2501 {0xa9b3, 0xa9c0}, 2502 {0xa9e5, 0xa9e5}, 2503 {0xaa29, 0xaa36}, 2504 {0xaa43, 0xaa43}, 2505 {0xaa4c, 0xaa4d}, 2506 {0xaa7b, 0xaa7d}, 2507 {0xaab0, 0xaab0}, 2508 {0xaab2, 0xaab4}, 2509 {0xaab7, 0xaab8}, 2510 {0xaabe, 0xaabf}, 2511 {0xaac1, 0xaac1}, 2512 {0xaaeb, 0xaaef}, 2513 {0xaaf5, 0xaaf6}, 2514 {0xabe3, 0xabea}, 2515 {0xabec, 0xabed}, 2516 {0xfb1e, 0xfb1e}, 2517 {0xfe00, 0xfe0f}, 2518 {0xfe20, 0xfe2f}, 2519 {0x101fd, 0x101fd}, 2520 {0x102e0, 0x102e0}, 2521 {0x10376, 0x1037a}, 2522 {0x10a01, 0x10a03}, 2523 {0x10a05, 0x10a06}, 2524 {0x10a0c, 0x10a0f}, 2525 {0x10a38, 0x10a3a}, 2526 {0x10a3f, 0x10a3f}, 2527 {0x10ae5, 0x10ae6}, 2528 {0x11000, 0x11002}, 2529 {0x11038, 0x11046}, 2530 {0x1107f, 0x11082}, 2531 {0x110b0, 0x110ba}, 2532 {0x11100, 0x11102}, 2533 {0x11127, 0x11134}, 2534 {0x11173, 0x11173}, 2535 {0x11180, 0x11182}, 2536 {0x111b3, 0x111c0}, 2537 {0x111ca, 0x111cc}, 2538 {0x1122c, 0x11237}, 2539 {0x1123e, 0x1123e}, 2540 {0x112df, 0x112ea}, 2541 {0x11300, 0x11303}, 2542 {0x1133c, 0x1133c}, 2543 {0x1133e, 0x11344}, 2544 {0x11347, 0x11348}, 2545 {0x1134b, 0x1134d}, 2546 {0x11357, 0x11357}, 2547 {0x11362, 0x11363}, 2548 {0x11366, 0x1136c}, 2549 {0x11370, 0x11374}, 2550 {0x11435, 0x11446}, 2551 {0x114b0, 0x114c3}, 2552 {0x115af, 0x115b5}, 2553 {0x115b8, 0x115c0}, 2554 {0x115dc, 0x115dd}, 2555 {0x11630, 0x11640}, 2556 {0x116ab, 0x116b7}, 2557 {0x1171d, 0x1172b}, 2558 {0x11a01, 0x11a0a}, 2559 {0x11a33, 0x11a39}, 2560 {0x11a3b, 0x11a3e}, 2561 {0x11a47, 0x11a47}, 2562 {0x11a51, 0x11a5b}, 2563 {0x11a8a, 0x11a99}, 2564 {0x11c2f, 0x11c36}, 2565 {0x11c38, 0x11c3f}, 2566 {0x11c92, 0x11ca7}, 2567 {0x11ca9, 0x11cb6}, 2568 {0x11d31, 0x11d36}, 2569 {0x11d3a, 0x11d3a}, 2570 {0x11d3c, 0x11d3d}, 2571 {0x11d3f, 0x11d45}, 2572 {0x11d47, 0x11d47}, 2573 {0x16af0, 0x16af4}, 2574 {0x16b30, 0x16b36}, 2575 {0x16f51, 0x16f7e}, 2576 {0x16f8f, 0x16f92}, 2577 {0x1bc9d, 0x1bc9e}, 2578 {0x1d165, 0x1d169}, 2579 {0x1d16d, 0x1d172}, 2580 {0x1d17b, 0x1d182}, 2581 {0x1d185, 0x1d18b}, 2582 {0x1d1aa, 0x1d1ad}, 2583 {0x1d242, 0x1d244}, 2584 {0x1da00, 0x1da36}, 2585 {0x1da3b, 0x1da6c}, 2586 {0x1da75, 0x1da75}, 2587 {0x1da84, 0x1da84}, 2588 {0x1da9b, 0x1da9f}, 2589 {0x1daa1, 0x1daaf}, 2590 {0x1e000, 0x1e006}, 2591 {0x1e008, 0x1e018}, 2592 {0x1e01b, 0x1e021}, 2593 {0x1e023, 0x1e024}, 2594 {0x1e026, 0x1e02a}, 2595 {0x1e8d0, 0x1e8d6}, 2596 {0x1e944, 0x1e94a}, 2597 {0xe0100, 0xe01ef} 2598 }; 2599 2600 return intable(combining, sizeof(combining), c); 2601 } 2602 2603 /* 2604 * Return TRUE for characters that can be displayed in a normal way. 2605 * Only for characters of 0x100 and above! 2606 */ 2607 int 2608 utf_printable(int c) 2609 { 2610 #ifdef USE_WCHAR_FUNCTIONS 2611 /* 2612 * Assume the iswprint() library function works better than our own stuff. 2613 */ 2614 return iswprint(c); 2615 #else 2616 /* Sorted list of non-overlapping intervals. 2617 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */ 2618 static struct interval nonprint[] = 2619 { 2620 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2621 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2622 {0xfffe, 0xffff} 2623 }; 2624 2625 return !intable(nonprint, sizeof(nonprint), c); 2626 #endif 2627 } 2628 2629 /* Sorted list of non-overlapping intervals of all Emoji characters, 2630 * based on http://unicode.org/emoji/charts/emoji-list.html */ 2631 static struct interval emoji_all[] = 2632 { 2633 {0x203c, 0x203c}, 2634 {0x2049, 0x2049}, 2635 {0x2122, 0x2122}, 2636 {0x2139, 0x2139}, 2637 {0x2194, 0x2199}, 2638 {0x21a9, 0x21aa}, 2639 {0x231a, 0x231b}, 2640 {0x2328, 0x2328}, 2641 {0x23cf, 0x23cf}, 2642 {0x23e9, 0x23f3}, 2643 {0x23f8, 0x23fa}, 2644 {0x24c2, 0x24c2}, 2645 {0x25aa, 0x25ab}, 2646 {0x25b6, 0x25b6}, 2647 {0x25c0, 0x25c0}, 2648 {0x25fb, 0x25fe}, 2649 {0x2600, 0x2604}, 2650 {0x260e, 0x260e}, 2651 {0x2611, 0x2611}, 2652 {0x2614, 0x2615}, 2653 {0x2618, 0x2618}, 2654 {0x261d, 0x261d}, 2655 {0x2620, 0x2620}, 2656 {0x2622, 0x2623}, 2657 {0x2626, 0x2626}, 2658 {0x262a, 0x262a}, 2659 {0x262e, 0x262f}, 2660 {0x2638, 0x263a}, 2661 {0x2640, 0x2640}, 2662 {0x2642, 0x2642}, 2663 {0x2648, 0x2653}, 2664 {0x2660, 0x2660}, 2665 {0x2663, 0x2663}, 2666 {0x2665, 0x2666}, 2667 {0x2668, 0x2668}, 2668 {0x267b, 0x267b}, 2669 {0x267f, 0x267f}, 2670 {0x2692, 0x2697}, 2671 {0x2699, 0x2699}, 2672 {0x269b, 0x269c}, 2673 {0x26a0, 0x26a1}, 2674 {0x26aa, 0x26ab}, 2675 {0x26b0, 0x26b1}, 2676 {0x26bd, 0x26be}, 2677 {0x26c4, 0x26c5}, 2678 {0x26c8, 0x26c8}, 2679 {0x26ce, 0x26cf}, 2680 {0x26d1, 0x26d1}, 2681 {0x26d3, 0x26d4}, 2682 {0x26e9, 0x26ea}, 2683 {0x26f0, 0x26f5}, 2684 {0x26f7, 0x26fa}, 2685 {0x26fd, 0x26fd}, 2686 {0x2702, 0x2702}, 2687 {0x2705, 0x2705}, 2688 {0x2708, 0x270d}, 2689 {0x270f, 0x270f}, 2690 {0x2712, 0x2712}, 2691 {0x2714, 0x2714}, 2692 {0x2716, 0x2716}, 2693 {0x271d, 0x271d}, 2694 {0x2721, 0x2721}, 2695 {0x2728, 0x2728}, 2696 {0x2733, 0x2734}, 2697 {0x2744, 0x2744}, 2698 {0x2747, 0x2747}, 2699 {0x274c, 0x274c}, 2700 {0x274e, 0x274e}, 2701 {0x2753, 0x2755}, 2702 {0x2757, 0x2757}, 2703 {0x2763, 0x2764}, 2704 {0x2795, 0x2797}, 2705 {0x27a1, 0x27a1}, 2706 {0x27b0, 0x27b0}, 2707 {0x27bf, 0x27bf}, 2708 {0x2934, 0x2935}, 2709 {0x2b05, 0x2b07}, 2710 {0x2b1b, 0x2b1c}, 2711 {0x2b50, 0x2b50}, 2712 {0x2b55, 0x2b55}, 2713 {0x3030, 0x3030}, 2714 {0x303d, 0x303d}, 2715 {0x3297, 0x3297}, 2716 {0x3299, 0x3299}, 2717 {0x1f004, 0x1f004}, 2718 {0x1f0cf, 0x1f0cf}, 2719 {0x1f170, 0x1f171}, 2720 {0x1f17e, 0x1f17f}, 2721 {0x1f18e, 0x1f18e}, 2722 {0x1f191, 0x1f19a}, 2723 {0x1f1e6, 0x1f1ff}, 2724 {0x1f201, 0x1f202}, 2725 {0x1f21a, 0x1f21a}, 2726 {0x1f22f, 0x1f22f}, 2727 {0x1f232, 0x1f23a}, 2728 {0x1f250, 0x1f251}, 2729 {0x1f300, 0x1f321}, 2730 {0x1f324, 0x1f393}, 2731 {0x1f396, 0x1f397}, 2732 {0x1f399, 0x1f39b}, 2733 {0x1f39e, 0x1f3f0}, 2734 {0x1f3f3, 0x1f3f5}, 2735 {0x1f3f7, 0x1f4fd}, 2736 {0x1f4ff, 0x1f53d}, 2737 {0x1f549, 0x1f54e}, 2738 {0x1f550, 0x1f567}, 2739 {0x1f56f, 0x1f570}, 2740 {0x1f573, 0x1f57a}, 2741 {0x1f587, 0x1f587}, 2742 {0x1f58a, 0x1f58d}, 2743 {0x1f590, 0x1f590}, 2744 {0x1f595, 0x1f596}, 2745 {0x1f5a4, 0x1f5a5}, 2746 {0x1f5a8, 0x1f5a8}, 2747 {0x1f5b1, 0x1f5b2}, 2748 {0x1f5bc, 0x1f5bc}, 2749 {0x1f5c2, 0x1f5c4}, 2750 {0x1f5d1, 0x1f5d3}, 2751 {0x1f5dc, 0x1f5de}, 2752 {0x1f5e1, 0x1f5e1}, 2753 {0x1f5e3, 0x1f5e3}, 2754 {0x1f5e8, 0x1f5e8}, 2755 {0x1f5ef, 0x1f5ef}, 2756 {0x1f5f3, 0x1f5f3}, 2757 {0x1f5fa, 0x1f64f}, 2758 {0x1f680, 0x1f6c5}, 2759 {0x1f6cb, 0x1f6d2}, 2760 {0x1f6e0, 0x1f6e5}, 2761 {0x1f6e9, 0x1f6e9}, 2762 {0x1f6eb, 0x1f6ec}, 2763 {0x1f6f0, 0x1f6f0}, 2764 {0x1f6f3, 0x1f6f8}, 2765 {0x1f910, 0x1f93a}, 2766 {0x1f93c, 0x1f93e}, 2767 {0x1f940, 0x1f945}, 2768 {0x1f947, 0x1f94c}, 2769 {0x1f950, 0x1f96b}, 2770 {0x1f980, 0x1f997}, 2771 {0x1f9c0, 0x1f9c0}, 2772 {0x1f9d0, 0x1f9e6} 2773 }; 2774 2775 /* 2776 * Get class of a Unicode character. 2777 * 0: white space 2778 * 1: punctuation 2779 * 2 or bigger: some class of word character. 2780 */ 2781 int 2782 utf_class(int c) 2783 { 2784 return utf_class_buf(c, curbuf); 2785 } 2786 2787 int 2788 utf_class_buf(int c, buf_T *buf) 2789 { 2790 /* sorted list of non-overlapping intervals */ 2791 static struct clinterval 2792 { 2793 unsigned int first; 2794 unsigned int last; 2795 unsigned int class; 2796 } classes[] = 2797 { 2798 {0x037e, 0x037e, 1}, /* Greek question mark */ 2799 {0x0387, 0x0387, 1}, /* Greek ano teleia */ 2800 {0x055a, 0x055f, 1}, /* Armenian punctuation */ 2801 {0x0589, 0x0589, 1}, /* Armenian full stop */ 2802 {0x05be, 0x05be, 1}, 2803 {0x05c0, 0x05c0, 1}, 2804 {0x05c3, 0x05c3, 1}, 2805 {0x05f3, 0x05f4, 1}, 2806 {0x060c, 0x060c, 1}, 2807 {0x061b, 0x061b, 1}, 2808 {0x061f, 0x061f, 1}, 2809 {0x066a, 0x066d, 1}, 2810 {0x06d4, 0x06d4, 1}, 2811 {0x0700, 0x070d, 1}, /* Syriac punctuation */ 2812 {0x0964, 0x0965, 1}, 2813 {0x0970, 0x0970, 1}, 2814 {0x0df4, 0x0df4, 1}, 2815 {0x0e4f, 0x0e4f, 1}, 2816 {0x0e5a, 0x0e5b, 1}, 2817 {0x0f04, 0x0f12, 1}, 2818 {0x0f3a, 0x0f3d, 1}, 2819 {0x0f85, 0x0f85, 1}, 2820 {0x104a, 0x104f, 1}, /* Myanmar punctuation */ 2821 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */ 2822 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */ 2823 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */ 2824 {0x1680, 0x1680, 0}, 2825 {0x169b, 0x169c, 1}, 2826 {0x16eb, 0x16ed, 1}, 2827 {0x1735, 0x1736, 1}, 2828 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */ 2829 {0x1800, 0x180a, 1}, /* Mongolian punctuation */ 2830 {0x2000, 0x200b, 0}, /* spaces */ 2831 {0x200c, 0x2027, 1}, /* punctuation and symbols */ 2832 {0x2028, 0x2029, 0}, 2833 {0x202a, 0x202e, 1}, /* punctuation and symbols */ 2834 {0x202f, 0x202f, 0}, 2835 {0x2030, 0x205e, 1}, /* punctuation and symbols */ 2836 {0x205f, 0x205f, 0}, 2837 {0x2060, 0x27ff, 1}, /* punctuation and symbols */ 2838 {0x2070, 0x207f, 0x2070}, /* superscript */ 2839 {0x2080, 0x2094, 0x2080}, /* subscript */ 2840 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */ 2841 {0x2800, 0x28ff, 0x2800}, /* braille */ 2842 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */ 2843 {0x29d8, 0x29db, 1}, 2844 {0x29fc, 0x29fd, 1}, 2845 {0x2e00, 0x2e7f, 1}, /* supplemental punctuation */ 2846 {0x3000, 0x3000, 0}, /* ideographic space */ 2847 {0x3001, 0x3020, 1}, /* ideographic punctuation */ 2848 {0x3030, 0x3030, 1}, 2849 {0x303d, 0x303d, 1}, 2850 {0x3040, 0x309f, 0x3040}, /* Hiragana */ 2851 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */ 2852 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */ 2853 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */ 2854 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */ 2855 {0xfd3e, 0xfd3f, 1}, 2856 {0xfe30, 0xfe6b, 1}, /* punctuation forms */ 2857 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */ 2858 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */ 2859 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */ 2860 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */ 2861 {0x20000, 0x2a6df, 0x4e00}, /* CJK Ideographs */ 2862 {0x2a700, 0x2b73f, 0x4e00}, /* CJK Ideographs */ 2863 {0x2b740, 0x2b81f, 0x4e00}, /* CJK Ideographs */ 2864 {0x2f800, 0x2fa1f, 0x4e00}, /* CJK Ideographs */ 2865 }; 2866 2867 int bot = 0; 2868 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2869 int mid; 2870 2871 /* First quick check for Latin1 characters, use 'iskeyword'. */ 2872 if (c < 0x100) 2873 { 2874 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2875 return 0; /* blank */ 2876 if (vim_iswordc_buf(c, buf)) 2877 return 2; /* word character */ 2878 return 1; /* punctuation */ 2879 } 2880 2881 /* binary search in table */ 2882 while (top >= bot) 2883 { 2884 mid = (bot + top) / 2; 2885 if (classes[mid].last < (unsigned int)c) 2886 bot = mid + 1; 2887 else if (classes[mid].first > (unsigned int)c) 2888 top = mid - 1; 2889 else 2890 return (int)classes[mid].class; 2891 } 2892 2893 /* emoji */ 2894 if (intable(emoji_all, sizeof(emoji_all), c)) 2895 return 3; 2896 2897 /* most other characters are "word" characters */ 2898 return 2; 2899 } 2900 2901 int 2902 utf_ambiguous_width(int c) 2903 { 2904 return c >= 0x80 && (intable(ambiguous, sizeof(ambiguous), c) 2905 || intable(emoji_all, sizeof(emoji_all), c)); 2906 } 2907 2908 /* 2909 * Code for Unicode case-dependent operations. Based on notes in 2910 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2911 * This code uses simple case folding, not full case folding. 2912 * Last updated for Unicode 5.2. 2913 */ 2914 2915 /* 2916 * The following tables are built by ../runtime/tools/unicode.vim. 2917 * They must be in numeric order, because we use binary search. 2918 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2919 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2920 * folded/upper/lower by adding 32. 2921 */ 2922 typedef struct 2923 { 2924 int rangeStart; 2925 int rangeEnd; 2926 int step; 2927 int offset; 2928 } convertStruct; 2929 2930 static convertStruct foldCase[] = 2931 { 2932 {0x41,0x5a,1,32}, 2933 {0xb5,0xb5,-1,775}, 2934 {0xc0,0xd6,1,32}, 2935 {0xd8,0xde,1,32}, 2936 {0x100,0x12e,2,1}, 2937 {0x132,0x136,2,1}, 2938 {0x139,0x147,2,1}, 2939 {0x14a,0x176,2,1}, 2940 {0x178,0x178,-1,-121}, 2941 {0x179,0x17d,2,1}, 2942 {0x17f,0x17f,-1,-268}, 2943 {0x181,0x181,-1,210}, 2944 {0x182,0x184,2,1}, 2945 {0x186,0x186,-1,206}, 2946 {0x187,0x187,-1,1}, 2947 {0x189,0x18a,1,205}, 2948 {0x18b,0x18b,-1,1}, 2949 {0x18e,0x18e,-1,79}, 2950 {0x18f,0x18f,-1,202}, 2951 {0x190,0x190,-1,203}, 2952 {0x191,0x191,-1,1}, 2953 {0x193,0x193,-1,205}, 2954 {0x194,0x194,-1,207}, 2955 {0x196,0x196,-1,211}, 2956 {0x197,0x197,-1,209}, 2957 {0x198,0x198,-1,1}, 2958 {0x19c,0x19c,-1,211}, 2959 {0x19d,0x19d,-1,213}, 2960 {0x19f,0x19f,-1,214}, 2961 {0x1a0,0x1a4,2,1}, 2962 {0x1a6,0x1a6,-1,218}, 2963 {0x1a7,0x1a7,-1,1}, 2964 {0x1a9,0x1a9,-1,218}, 2965 {0x1ac,0x1ac,-1,1}, 2966 {0x1ae,0x1ae,-1,218}, 2967 {0x1af,0x1af,-1,1}, 2968 {0x1b1,0x1b2,1,217}, 2969 {0x1b3,0x1b5,2,1}, 2970 {0x1b7,0x1b7,-1,219}, 2971 {0x1b8,0x1bc,4,1}, 2972 {0x1c4,0x1c4,-1,2}, 2973 {0x1c5,0x1c5,-1,1}, 2974 {0x1c7,0x1c7,-1,2}, 2975 {0x1c8,0x1c8,-1,1}, 2976 {0x1ca,0x1ca,-1,2}, 2977 {0x1cb,0x1db,2,1}, 2978 {0x1de,0x1ee,2,1}, 2979 {0x1f1,0x1f1,-1,2}, 2980 {0x1f2,0x1f4,2,1}, 2981 {0x1f6,0x1f6,-1,-97}, 2982 {0x1f7,0x1f7,-1,-56}, 2983 {0x1f8,0x21e,2,1}, 2984 {0x220,0x220,-1,-130}, 2985 {0x222,0x232,2,1}, 2986 {0x23a,0x23a,-1,10795}, 2987 {0x23b,0x23b,-1,1}, 2988 {0x23d,0x23d,-1,-163}, 2989 {0x23e,0x23e,-1,10792}, 2990 {0x241,0x241,-1,1}, 2991 {0x243,0x243,-1,-195}, 2992 {0x244,0x244,-1,69}, 2993 {0x245,0x245,-1,71}, 2994 {0x246,0x24e,2,1}, 2995 {0x345,0x345,-1,116}, 2996 {0x370,0x372,2,1}, 2997 {0x376,0x376,-1,1}, 2998 {0x37f,0x37f,-1,116}, 2999 {0x386,0x386,-1,38}, 3000 {0x388,0x38a,1,37}, 3001 {0x38c,0x38c,-1,64}, 3002 {0x38e,0x38f,1,63}, 3003 {0x391,0x3a1,1,32}, 3004 {0x3a3,0x3ab,1,32}, 3005 {0x3c2,0x3c2,-1,1}, 3006 {0x3cf,0x3cf,-1,8}, 3007 {0x3d0,0x3d0,-1,-30}, 3008 {0x3d1,0x3d1,-1,-25}, 3009 {0x3d5,0x3d5,-1,-15}, 3010 {0x3d6,0x3d6,-1,-22}, 3011 {0x3d8,0x3ee,2,1}, 3012 {0x3f0,0x3f0,-1,-54}, 3013 {0x3f1,0x3f1,-1,-48}, 3014 {0x3f4,0x3f4,-1,-60}, 3015 {0x3f5,0x3f5,-1,-64}, 3016 {0x3f7,0x3f7,-1,1}, 3017 {0x3f9,0x3f9,-1,-7}, 3018 {0x3fa,0x3fa,-1,1}, 3019 {0x3fd,0x3ff,1,-130}, 3020 {0x400,0x40f,1,80}, 3021 {0x410,0x42f,1,32}, 3022 {0x460,0x480,2,1}, 3023 {0x48a,0x4be,2,1}, 3024 {0x4c0,0x4c0,-1,15}, 3025 {0x4c1,0x4cd,2,1}, 3026 {0x4d0,0x52e,2,1}, 3027 {0x531,0x556,1,48}, 3028 {0x10a0,0x10c5,1,7264}, 3029 {0x10c7,0x10cd,6,7264}, 3030 {0x13f8,0x13fd,1,-8}, 3031 {0x1c80,0x1c80,-1,-6222}, 3032 {0x1c81,0x1c81,-1,-6221}, 3033 {0x1c82,0x1c82,-1,-6212}, 3034 {0x1c83,0x1c84,1,-6210}, 3035 {0x1c85,0x1c85,-1,-6211}, 3036 {0x1c86,0x1c86,-1,-6204}, 3037 {0x1c87,0x1c87,-1,-6180}, 3038 {0x1c88,0x1c88,-1,35267}, 3039 {0x1e00,0x1e94,2,1}, 3040 {0x1e9b,0x1e9b,-1,-58}, 3041 {0x1e9e,0x1e9e,-1,-7615}, 3042 {0x1ea0,0x1efe,2,1}, 3043 {0x1f08,0x1f0f,1,-8}, 3044 {0x1f18,0x1f1d,1,-8}, 3045 {0x1f28,0x1f2f,1,-8}, 3046 {0x1f38,0x1f3f,1,-8}, 3047 {0x1f48,0x1f4d,1,-8}, 3048 {0x1f59,0x1f5f,2,-8}, 3049 {0x1f68,0x1f6f,1,-8}, 3050 {0x1f88,0x1f8f,1,-8}, 3051 {0x1f98,0x1f9f,1,-8}, 3052 {0x1fa8,0x1faf,1,-8}, 3053 {0x1fb8,0x1fb9,1,-8}, 3054 {0x1fba,0x1fbb,1,-74}, 3055 {0x1fbc,0x1fbc,-1,-9}, 3056 {0x1fbe,0x1fbe,-1,-7173}, 3057 {0x1fc8,0x1fcb,1,-86}, 3058 {0x1fcc,0x1fcc,-1,-9}, 3059 {0x1fd8,0x1fd9,1,-8}, 3060 {0x1fda,0x1fdb,1,-100}, 3061 {0x1fe8,0x1fe9,1,-8}, 3062 {0x1fea,0x1feb,1,-112}, 3063 {0x1fec,0x1fec,-1,-7}, 3064 {0x1ff8,0x1ff9,1,-128}, 3065 {0x1ffa,0x1ffb,1,-126}, 3066 {0x1ffc,0x1ffc,-1,-9}, 3067 {0x2126,0x2126,-1,-7517}, 3068 {0x212a,0x212a,-1,-8383}, 3069 {0x212b,0x212b,-1,-8262}, 3070 {0x2132,0x2132,-1,28}, 3071 {0x2160,0x216f,1,16}, 3072 {0x2183,0x2183,-1,1}, 3073 {0x24b6,0x24cf,1,26}, 3074 {0x2c00,0x2c2e,1,48}, 3075 {0x2c60,0x2c60,-1,1}, 3076 {0x2c62,0x2c62,-1,-10743}, 3077 {0x2c63,0x2c63,-1,-3814}, 3078 {0x2c64,0x2c64,-1,-10727}, 3079 {0x2c67,0x2c6b,2,1}, 3080 {0x2c6d,0x2c6d,-1,-10780}, 3081 {0x2c6e,0x2c6e,-1,-10749}, 3082 {0x2c6f,0x2c6f,-1,-10783}, 3083 {0x2c70,0x2c70,-1,-10782}, 3084 {0x2c72,0x2c75,3,1}, 3085 {0x2c7e,0x2c7f,1,-10815}, 3086 {0x2c80,0x2ce2,2,1}, 3087 {0x2ceb,0x2ced,2,1}, 3088 {0x2cf2,0xa640,31054,1}, 3089 {0xa642,0xa66c,2,1}, 3090 {0xa680,0xa69a,2,1}, 3091 {0xa722,0xa72e,2,1}, 3092 {0xa732,0xa76e,2,1}, 3093 {0xa779,0xa77b,2,1}, 3094 {0xa77d,0xa77d,-1,-35332}, 3095 {0xa77e,0xa786,2,1}, 3096 {0xa78b,0xa78b,-1,1}, 3097 {0xa78d,0xa78d,-1,-42280}, 3098 {0xa790,0xa792,2,1}, 3099 {0xa796,0xa7a8,2,1}, 3100 {0xa7aa,0xa7aa,-1,-42308}, 3101 {0xa7ab,0xa7ab,-1,-42319}, 3102 {0xa7ac,0xa7ac,-1,-42315}, 3103 {0xa7ad,0xa7ad,-1,-42305}, 3104 {0xa7ae,0xa7ae,-1,-42308}, 3105 {0xa7b0,0xa7b0,-1,-42258}, 3106 {0xa7b1,0xa7b1,-1,-42282}, 3107 {0xa7b2,0xa7b2,-1,-42261}, 3108 {0xa7b3,0xa7b3,-1,928}, 3109 {0xa7b4,0xa7b6,2,1}, 3110 {0xab70,0xabbf,1,-38864}, 3111 {0xff21,0xff3a,1,32}, 3112 {0x10400,0x10427,1,40}, 3113 {0x104b0,0x104d3,1,40}, 3114 {0x10c80,0x10cb2,1,64}, 3115 {0x118a0,0x118bf,1,32}, 3116 {0x1e900,0x1e921,1,34} 3117 }; 3118 3119 static int utf_convert(int a, convertStruct table[], int tableSize); 3120 static int utf_strnicmp(char_u *s1, char_u *s2, size_t n1, size_t n2); 3121 3122 /* 3123 * Generic conversion function for case operations. 3124 * Return the converted equivalent of "a", which is a UCS-4 character. Use 3125 * the given conversion "table". Uses binary search on "table". 3126 */ 3127 static int 3128 utf_convert( 3129 int a, 3130 convertStruct table[], 3131 int tableSize) 3132 { 3133 int start, mid, end; /* indices into table */ 3134 int entries = tableSize / sizeof(convertStruct); 3135 3136 start = 0; 3137 end = entries; 3138 while (start < end) 3139 { 3140 /* need to search further */ 3141 mid = (end + start) / 2; 3142 if (table[mid].rangeEnd < a) 3143 start = mid + 1; 3144 else 3145 end = mid; 3146 } 3147 if (start < entries 3148 && table[start].rangeStart <= a 3149 && a <= table[start].rangeEnd 3150 && (a - table[start].rangeStart) % table[start].step == 0) 3151 return (a + table[start].offset); 3152 else 3153 return a; 3154 } 3155 3156 /* 3157 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 3158 * simple case folding. 3159 */ 3160 int 3161 utf_fold(int a) 3162 { 3163 if (a < 0x80) 3164 /* be fast for ASCII */ 3165 return a >= 0x41 && a <= 0x5a ? a + 32 : a; 3166 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 3167 } 3168 3169 static convertStruct toLower[] = 3170 { 3171 {0x41,0x5a,1,32}, 3172 {0xc0,0xd6,1,32}, 3173 {0xd8,0xde,1,32}, 3174 {0x100,0x12e,2,1}, 3175 {0x130,0x130,-1,-199}, 3176 {0x132,0x136,2,1}, 3177 {0x139,0x147,2,1}, 3178 {0x14a,0x176,2,1}, 3179 {0x178,0x178,-1,-121}, 3180 {0x179,0x17d,2,1}, 3181 {0x181,0x181,-1,210}, 3182 {0x182,0x184,2,1}, 3183 {0x186,0x186,-1,206}, 3184 {0x187,0x187,-1,1}, 3185 {0x189,0x18a,1,205}, 3186 {0x18b,0x18b,-1,1}, 3187 {0x18e,0x18e,-1,79}, 3188 {0x18f,0x18f,-1,202}, 3189 {0x190,0x190,-1,203}, 3190 {0x191,0x191,-1,1}, 3191 {0x193,0x193,-1,205}, 3192 {0x194,0x194,-1,207}, 3193 {0x196,0x196,-1,211}, 3194 {0x197,0x197,-1,209}, 3195 {0x198,0x198,-1,1}, 3196 {0x19c,0x19c,-1,211}, 3197 {0x19d,0x19d,-1,213}, 3198 {0x19f,0x19f,-1,214}, 3199 {0x1a0,0x1a4,2,1}, 3200 {0x1a6,0x1a6,-1,218}, 3201 {0x1a7,0x1a7,-1,1}, 3202 {0x1a9,0x1a9,-1,218}, 3203 {0x1ac,0x1ac,-1,1}, 3204 {0x1ae,0x1ae,-1,218}, 3205 {0x1af,0x1af,-1,1}, 3206 {0x1b1,0x1b2,1,217}, 3207 {0x1b3,0x1b5,2,1}, 3208 {0x1b7,0x1b7,-1,219}, 3209 {0x1b8,0x1bc,4,1}, 3210 {0x1c4,0x1c4,-1,2}, 3211 {0x1c5,0x1c5,-1,1}, 3212 {0x1c7,0x1c7,-1,2}, 3213 {0x1c8,0x1c8,-1,1}, 3214 {0x1ca,0x1ca,-1,2}, 3215 {0x1cb,0x1db,2,1}, 3216 {0x1de,0x1ee,2,1}, 3217 {0x1f1,0x1f1,-1,2}, 3218 {0x1f2,0x1f4,2,1}, 3219 {0x1f6,0x1f6,-1,-97}, 3220 {0x1f7,0x1f7,-1,-56}, 3221 {0x1f8,0x21e,2,1}, 3222 {0x220,0x220,-1,-130}, 3223 {0x222,0x232,2,1}, 3224 {0x23a,0x23a,-1,10795}, 3225 {0x23b,0x23b,-1,1}, 3226 {0x23d,0x23d,-1,-163}, 3227 {0x23e,0x23e,-1,10792}, 3228 {0x241,0x241,-1,1}, 3229 {0x243,0x243,-1,-195}, 3230 {0x244,0x244,-1,69}, 3231 {0x245,0x245,-1,71}, 3232 {0x246,0x24e,2,1}, 3233 {0x370,0x372,2,1}, 3234 {0x376,0x376,-1,1}, 3235 {0x37f,0x37f,-1,116}, 3236 {0x386,0x386,-1,38}, 3237 {0x388,0x38a,1,37}, 3238 {0x38c,0x38c,-1,64}, 3239 {0x38e,0x38f,1,63}, 3240 {0x391,0x3a1,1,32}, 3241 {0x3a3,0x3ab,1,32}, 3242 {0x3cf,0x3cf,-1,8}, 3243 {0x3d8,0x3ee,2,1}, 3244 {0x3f4,0x3f4,-1,-60}, 3245 {0x3f7,0x3f7,-1,1}, 3246 {0x3f9,0x3f9,-1,-7}, 3247 {0x3fa,0x3fa,-1,1}, 3248 {0x3fd,0x3ff,1,-130}, 3249 {0x400,0x40f,1,80}, 3250 {0x410,0x42f,1,32}, 3251 {0x460,0x480,2,1}, 3252 {0x48a,0x4be,2,1}, 3253 {0x4c0,0x4c0,-1,15}, 3254 {0x4c1,0x4cd,2,1}, 3255 {0x4d0,0x52e,2,1}, 3256 {0x531,0x556,1,48}, 3257 {0x10a0,0x10c5,1,7264}, 3258 {0x10c7,0x10cd,6,7264}, 3259 {0x13a0,0x13ef,1,38864}, 3260 {0x13f0,0x13f5,1,8}, 3261 {0x1e00,0x1e94,2,1}, 3262 {0x1e9e,0x1e9e,-1,-7615}, 3263 {0x1ea0,0x1efe,2,1}, 3264 {0x1f08,0x1f0f,1,-8}, 3265 {0x1f18,0x1f1d,1,-8}, 3266 {0x1f28,0x1f2f,1,-8}, 3267 {0x1f38,0x1f3f,1,-8}, 3268 {0x1f48,0x1f4d,1,-8}, 3269 {0x1f59,0x1f5f,2,-8}, 3270 {0x1f68,0x1f6f,1,-8}, 3271 {0x1f88,0x1f8f,1,-8}, 3272 {0x1f98,0x1f9f,1,-8}, 3273 {0x1fa8,0x1faf,1,-8}, 3274 {0x1fb8,0x1fb9,1,-8}, 3275 {0x1fba,0x1fbb,1,-74}, 3276 {0x1fbc,0x1fbc,-1,-9}, 3277 {0x1fc8,0x1fcb,1,-86}, 3278 {0x1fcc,0x1fcc,-1,-9}, 3279 {0x1fd8,0x1fd9,1,-8}, 3280 {0x1fda,0x1fdb,1,-100}, 3281 {0x1fe8,0x1fe9,1,-8}, 3282 {0x1fea,0x1feb,1,-112}, 3283 {0x1fec,0x1fec,-1,-7}, 3284 {0x1ff8,0x1ff9,1,-128}, 3285 {0x1ffa,0x1ffb,1,-126}, 3286 {0x1ffc,0x1ffc,-1,-9}, 3287 {0x2126,0x2126,-1,-7517}, 3288 {0x212a,0x212a,-1,-8383}, 3289 {0x212b,0x212b,-1,-8262}, 3290 {0x2132,0x2132,-1,28}, 3291 {0x2160,0x216f,1,16}, 3292 {0x2183,0x2183,-1,1}, 3293 {0x24b6,0x24cf,1,26}, 3294 {0x2c00,0x2c2e,1,48}, 3295 {0x2c60,0x2c60,-1,1}, 3296 {0x2c62,0x2c62,-1,-10743}, 3297 {0x2c63,0x2c63,-1,-3814}, 3298 {0x2c64,0x2c64,-1,-10727}, 3299 {0x2c67,0x2c6b,2,1}, 3300 {0x2c6d,0x2c6d,-1,-10780}, 3301 {0x2c6e,0x2c6e,-1,-10749}, 3302 {0x2c6f,0x2c6f,-1,-10783}, 3303 {0x2c70,0x2c70,-1,-10782}, 3304 {0x2c72,0x2c75,3,1}, 3305 {0x2c7e,0x2c7f,1,-10815}, 3306 {0x2c80,0x2ce2,2,1}, 3307 {0x2ceb,0x2ced,2,1}, 3308 {0x2cf2,0xa640,31054,1}, 3309 {0xa642,0xa66c,2,1}, 3310 {0xa680,0xa69a,2,1}, 3311 {0xa722,0xa72e,2,1}, 3312 {0xa732,0xa76e,2,1}, 3313 {0xa779,0xa77b,2,1}, 3314 {0xa77d,0xa77d,-1,-35332}, 3315 {0xa77e,0xa786,2,1}, 3316 {0xa78b,0xa78b,-1,1}, 3317 {0xa78d,0xa78d,-1,-42280}, 3318 {0xa790,0xa792,2,1}, 3319 {0xa796,0xa7a8,2,1}, 3320 {0xa7aa,0xa7aa,-1,-42308}, 3321 {0xa7ab,0xa7ab,-1,-42319}, 3322 {0xa7ac,0xa7ac,-1,-42315}, 3323 {0xa7ad,0xa7ad,-1,-42305}, 3324 {0xa7ae,0xa7ae,-1,-42308}, 3325 {0xa7b0,0xa7b0,-1,-42258}, 3326 {0xa7b1,0xa7b1,-1,-42282}, 3327 {0xa7b2,0xa7b2,-1,-42261}, 3328 {0xa7b3,0xa7b3,-1,928}, 3329 {0xa7b4,0xa7b6,2,1}, 3330 {0xff21,0xff3a,1,32}, 3331 {0x10400,0x10427,1,40}, 3332 {0x104b0,0x104d3,1,40}, 3333 {0x10c80,0x10cb2,1,64}, 3334 {0x118a0,0x118bf,1,32}, 3335 {0x1e900,0x1e921,1,34} 3336 }; 3337 3338 static convertStruct toUpper[] = 3339 { 3340 {0x61,0x7a,1,-32}, 3341 {0xb5,0xb5,-1,743}, 3342 {0xe0,0xf6,1,-32}, 3343 {0xf8,0xfe,1,-32}, 3344 {0xff,0xff,-1,121}, 3345 {0x101,0x12f,2,-1}, 3346 {0x131,0x131,-1,-232}, 3347 {0x133,0x137,2,-1}, 3348 {0x13a,0x148,2,-1}, 3349 {0x14b,0x177,2,-1}, 3350 {0x17a,0x17e,2,-1}, 3351 {0x17f,0x17f,-1,-300}, 3352 {0x180,0x180,-1,195}, 3353 {0x183,0x185,2,-1}, 3354 {0x188,0x18c,4,-1}, 3355 {0x192,0x192,-1,-1}, 3356 {0x195,0x195,-1,97}, 3357 {0x199,0x199,-1,-1}, 3358 {0x19a,0x19a,-1,163}, 3359 {0x19e,0x19e,-1,130}, 3360 {0x1a1,0x1a5,2,-1}, 3361 {0x1a8,0x1ad,5,-1}, 3362 {0x1b0,0x1b4,4,-1}, 3363 {0x1b6,0x1b9,3,-1}, 3364 {0x1bd,0x1bd,-1,-1}, 3365 {0x1bf,0x1bf,-1,56}, 3366 {0x1c5,0x1c5,-1,-1}, 3367 {0x1c6,0x1c6,-1,-2}, 3368 {0x1c8,0x1c8,-1,-1}, 3369 {0x1c9,0x1c9,-1,-2}, 3370 {0x1cb,0x1cb,-1,-1}, 3371 {0x1cc,0x1cc,-1,-2}, 3372 {0x1ce,0x1dc,2,-1}, 3373 {0x1dd,0x1dd,-1,-79}, 3374 {0x1df,0x1ef,2,-1}, 3375 {0x1f2,0x1f2,-1,-1}, 3376 {0x1f3,0x1f3,-1,-2}, 3377 {0x1f5,0x1f9,4,-1}, 3378 {0x1fb,0x21f,2,-1}, 3379 {0x223,0x233,2,-1}, 3380 {0x23c,0x23c,-1,-1}, 3381 {0x23f,0x240,1,10815}, 3382 {0x242,0x247,5,-1}, 3383 {0x249,0x24f,2,-1}, 3384 {0x250,0x250,-1,10783}, 3385 {0x251,0x251,-1,10780}, 3386 {0x252,0x252,-1,10782}, 3387 {0x253,0x253,-1,-210}, 3388 {0x254,0x254,-1,-206}, 3389 {0x256,0x257,1,-205}, 3390 {0x259,0x259,-1,-202}, 3391 {0x25b,0x25b,-1,-203}, 3392 {0x25c,0x25c,-1,42319}, 3393 {0x260,0x260,-1,-205}, 3394 {0x261,0x261,-1,42315}, 3395 {0x263,0x263,-1,-207}, 3396 {0x265,0x265,-1,42280}, 3397 {0x266,0x266,-1,42308}, 3398 {0x268,0x268,-1,-209}, 3399 {0x269,0x269,-1,-211}, 3400 {0x26a,0x26a,-1,42308}, 3401 {0x26b,0x26b,-1,10743}, 3402 {0x26c,0x26c,-1,42305}, 3403 {0x26f,0x26f,-1,-211}, 3404 {0x271,0x271,-1,10749}, 3405 {0x272,0x272,-1,-213}, 3406 {0x275,0x275,-1,-214}, 3407 {0x27d,0x27d,-1,10727}, 3408 {0x280,0x283,3,-218}, 3409 {0x287,0x287,-1,42282}, 3410 {0x288,0x288,-1,-218}, 3411 {0x289,0x289,-1,-69}, 3412 {0x28a,0x28b,1,-217}, 3413 {0x28c,0x28c,-1,-71}, 3414 {0x292,0x292,-1,-219}, 3415 {0x29d,0x29d,-1,42261}, 3416 {0x29e,0x29e,-1,42258}, 3417 {0x345,0x345,-1,84}, 3418 {0x371,0x373,2,-1}, 3419 {0x377,0x377,-1,-1}, 3420 {0x37b,0x37d,1,130}, 3421 {0x3ac,0x3ac,-1,-38}, 3422 {0x3ad,0x3af,1,-37}, 3423 {0x3b1,0x3c1,1,-32}, 3424 {0x3c2,0x3c2,-1,-31}, 3425 {0x3c3,0x3cb,1,-32}, 3426 {0x3cc,0x3cc,-1,-64}, 3427 {0x3cd,0x3ce,1,-63}, 3428 {0x3d0,0x3d0,-1,-62}, 3429 {0x3d1,0x3d1,-1,-57}, 3430 {0x3d5,0x3d5,-1,-47}, 3431 {0x3d6,0x3d6,-1,-54}, 3432 {0x3d7,0x3d7,-1,-8}, 3433 {0x3d9,0x3ef,2,-1}, 3434 {0x3f0,0x3f0,-1,-86}, 3435 {0x3f1,0x3f1,-1,-80}, 3436 {0x3f2,0x3f2,-1,7}, 3437 {0x3f3,0x3f3,-1,-116}, 3438 {0x3f5,0x3f5,-1,-96}, 3439 {0x3f8,0x3fb,3,-1}, 3440 {0x430,0x44f,1,-32}, 3441 {0x450,0x45f,1,-80}, 3442 {0x461,0x481,2,-1}, 3443 {0x48b,0x4bf,2,-1}, 3444 {0x4c2,0x4ce,2,-1}, 3445 {0x4cf,0x4cf,-1,-15}, 3446 {0x4d1,0x52f,2,-1}, 3447 {0x561,0x586,1,-48}, 3448 {0x13f8,0x13fd,1,-8}, 3449 {0x1c80,0x1c80,-1,-6254}, 3450 {0x1c81,0x1c81,-1,-6253}, 3451 {0x1c82,0x1c82,-1,-6244}, 3452 {0x1c83,0x1c84,1,-6242}, 3453 {0x1c85,0x1c85,-1,-6243}, 3454 {0x1c86,0x1c86,-1,-6236}, 3455 {0x1c87,0x1c87,-1,-6181}, 3456 {0x1c88,0x1c88,-1,35266}, 3457 {0x1d79,0x1d79,-1,35332}, 3458 {0x1d7d,0x1d7d,-1,3814}, 3459 {0x1e01,0x1e95,2,-1}, 3460 {0x1e9b,0x1e9b,-1,-59}, 3461 {0x1ea1,0x1eff,2,-1}, 3462 {0x1f00,0x1f07,1,8}, 3463 {0x1f10,0x1f15,1,8}, 3464 {0x1f20,0x1f27,1,8}, 3465 {0x1f30,0x1f37,1,8}, 3466 {0x1f40,0x1f45,1,8}, 3467 {0x1f51,0x1f57,2,8}, 3468 {0x1f60,0x1f67,1,8}, 3469 {0x1f70,0x1f71,1,74}, 3470 {0x1f72,0x1f75,1,86}, 3471 {0x1f76,0x1f77,1,100}, 3472 {0x1f78,0x1f79,1,128}, 3473 {0x1f7a,0x1f7b,1,112}, 3474 {0x1f7c,0x1f7d,1,126}, 3475 {0x1f80,0x1f87,1,8}, 3476 {0x1f90,0x1f97,1,8}, 3477 {0x1fa0,0x1fa7,1,8}, 3478 {0x1fb0,0x1fb1,1,8}, 3479 {0x1fb3,0x1fb3,-1,9}, 3480 {0x1fbe,0x1fbe,-1,-7205}, 3481 {0x1fc3,0x1fc3,-1,9}, 3482 {0x1fd0,0x1fd1,1,8}, 3483 {0x1fe0,0x1fe1,1,8}, 3484 {0x1fe5,0x1fe5,-1,7}, 3485 {0x1ff3,0x1ff3,-1,9}, 3486 {0x214e,0x214e,-1,-28}, 3487 {0x2170,0x217f,1,-16}, 3488 {0x2184,0x2184,-1,-1}, 3489 {0x24d0,0x24e9,1,-26}, 3490 {0x2c30,0x2c5e,1,-48}, 3491 {0x2c61,0x2c61,-1,-1}, 3492 {0x2c65,0x2c65,-1,-10795}, 3493 {0x2c66,0x2c66,-1,-10792}, 3494 {0x2c68,0x2c6c,2,-1}, 3495 {0x2c73,0x2c76,3,-1}, 3496 {0x2c81,0x2ce3,2,-1}, 3497 {0x2cec,0x2cee,2,-1}, 3498 {0x2cf3,0x2cf3,-1,-1}, 3499 {0x2d00,0x2d25,1,-7264}, 3500 {0x2d27,0x2d2d,6,-7264}, 3501 {0xa641,0xa66d,2,-1}, 3502 {0xa681,0xa69b,2,-1}, 3503 {0xa723,0xa72f,2,-1}, 3504 {0xa733,0xa76f,2,-1}, 3505 {0xa77a,0xa77c,2,-1}, 3506 {0xa77f,0xa787,2,-1}, 3507 {0xa78c,0xa791,5,-1}, 3508 {0xa793,0xa797,4,-1}, 3509 {0xa799,0xa7a9,2,-1}, 3510 {0xa7b5,0xa7b7,2,-1}, 3511 {0xab53,0xab53,-1,-928}, 3512 {0xab70,0xabbf,1,-38864}, 3513 {0xff41,0xff5a,1,-32}, 3514 {0x10428,0x1044f,1,-40}, 3515 {0x104d8,0x104fb,1,-40}, 3516 {0x10cc0,0x10cf2,1,-64}, 3517 {0x118c0,0x118df,1,-32}, 3518 {0x1e922,0x1e943,1,-34} 3519 }; 3520 3521 /* 3522 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3523 * simple case folding. 3524 */ 3525 int 3526 utf_toupper(int a) 3527 { 3528 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */ 3529 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3530 return TOUPPER_ASC(a); 3531 3532 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3533 /* If towupper() is available and handles Unicode, use it. */ 3534 if (!(cmp_flags & CMP_INTERNAL)) 3535 return towupper(a); 3536 #endif 3537 3538 /* For characters below 128 use locale sensitive toupper(). */ 3539 if (a < 128) 3540 return TOUPPER_LOC(a); 3541 3542 /* For any other characters use the above mapping table. */ 3543 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3544 } 3545 3546 int 3547 utf_islower(int a) 3548 { 3549 /* German sharp s is lower case but has no upper case equivalent. */ 3550 return (utf_toupper(a) != a) || a == 0xdf; 3551 } 3552 3553 /* 3554 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3555 * simple case folding. 3556 */ 3557 int 3558 utf_tolower(int a) 3559 { 3560 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */ 3561 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3562 return TOLOWER_ASC(a); 3563 3564 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3565 /* If towlower() is available and handles Unicode, use it. */ 3566 if (!(cmp_flags & CMP_INTERNAL)) 3567 return towlower(a); 3568 #endif 3569 3570 /* For characters below 128 use locale sensitive tolower(). */ 3571 if (a < 128) 3572 return TOLOWER_LOC(a); 3573 3574 /* For any other characters use the above mapping table. */ 3575 return utf_convert(a, toLower, (int)sizeof(toLower)); 3576 } 3577 3578 int 3579 utf_isupper(int a) 3580 { 3581 return (utf_tolower(a) != a); 3582 } 3583 3584 static int 3585 utf_strnicmp( 3586 char_u *s1, 3587 char_u *s2, 3588 size_t n1, 3589 size_t n2) 3590 { 3591 int c1, c2, cdiff; 3592 char_u buffer[6]; 3593 3594 for (;;) 3595 { 3596 c1 = utf_safe_read_char_adv(&s1, &n1); 3597 c2 = utf_safe_read_char_adv(&s2, &n2); 3598 3599 if (c1 <= 0 || c2 <= 0) 3600 break; 3601 3602 if (c1 == c2) 3603 continue; 3604 3605 cdiff = utf_fold(c1) - utf_fold(c2); 3606 if (cdiff != 0) 3607 return cdiff; 3608 } 3609 3610 /* some string ended or has an incomplete/illegal character sequence */ 3611 3612 if (c1 == 0 || c2 == 0) 3613 { 3614 /* some string ended. shorter string is smaller */ 3615 if (c1 == 0 && c2 == 0) 3616 return 0; 3617 return c1 == 0 ? -1 : 1; 3618 } 3619 3620 /* Continue with bytewise comparison to produce some result that 3621 * would make comparison operations involving this function transitive. 3622 * 3623 * If only one string had an error, comparison should be made with 3624 * folded version of the other string. In this case it is enough 3625 * to fold just one character to determine the result of comparison. */ 3626 3627 if (c1 != -1 && c2 == -1) 3628 { 3629 n1 = utf_char2bytes(utf_fold(c1), buffer); 3630 s1 = buffer; 3631 } 3632 else if (c2 != -1 && c1 == -1) 3633 { 3634 n2 = utf_char2bytes(utf_fold(c2), buffer); 3635 s2 = buffer; 3636 } 3637 3638 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3639 { 3640 cdiff = (int)(*s1) - (int)(*s2); 3641 if (cdiff != 0) 3642 return cdiff; 3643 3644 s1++; 3645 s2++; 3646 n1--; 3647 n2--; 3648 } 3649 3650 if (n1 > 0 && *s1 == NUL) 3651 n1 = 0; 3652 if (n2 > 0 && *s2 == NUL) 3653 n2 = 0; 3654 3655 if (n1 == 0 && n2 == 0) 3656 return 0; 3657 return n1 == 0 ? -1 : 1; 3658 } 3659 3660 /* 3661 * Version of strnicmp() that handles multi-byte characters. 3662 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3663 * probably use strnicmp(), because there are no ASCII characters in the 3664 * second byte. 3665 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3666 * two characters otherwise. 3667 */ 3668 int 3669 mb_strnicmp(char_u *s1, char_u *s2, size_t nn) 3670 { 3671 int i, l; 3672 int cdiff; 3673 int n = (int)nn; 3674 3675 if (enc_utf8) 3676 { 3677 return utf_strnicmp(s1, s2, nn, nn); 3678 } 3679 else 3680 { 3681 for (i = 0; i < n; i += l) 3682 { 3683 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */ 3684 return 0; 3685 3686 l = (*mb_ptr2len)(s1 + i); 3687 if (l <= 1) 3688 { 3689 /* Single byte: first check normally, then with ignore case. */ 3690 if (s1[i] != s2[i]) 3691 { 3692 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3693 if (cdiff != 0) 3694 return cdiff; 3695 } 3696 } 3697 else 3698 { 3699 /* For non-Unicode multi-byte don't ignore case. */ 3700 if (l > n - i) 3701 l = n - i; 3702 cdiff = STRNCMP(s1 + i, s2 + i, l); 3703 if (cdiff != 0) 3704 return cdiff; 3705 } 3706 } 3707 } 3708 return 0; 3709 } 3710 3711 /* 3712 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3713 * 'encoding' has been set to. 3714 */ 3715 void 3716 show_utf8(void) 3717 { 3718 int len; 3719 int rlen = 0; 3720 char_u *line; 3721 int clen; 3722 int i; 3723 3724 /* Get the byte length of the char under the cursor, including composing 3725 * characters. */ 3726 line = ml_get_cursor(); 3727 len = utfc_ptr2len(line); 3728 if (len == 0) 3729 { 3730 MSG("NUL"); 3731 return; 3732 } 3733 3734 clen = 0; 3735 for (i = 0; i < len; ++i) 3736 { 3737 if (clen == 0) 3738 { 3739 /* start of (composing) character, get its length */ 3740 if (i > 0) 3741 { 3742 STRCPY(IObuff + rlen, "+ "); 3743 rlen += 2; 3744 } 3745 clen = utf_ptr2len(line + i); 3746 } 3747 sprintf((char *)IObuff + rlen, "%02x ", 3748 (line[i] == NL) ? NUL : line[i]); /* NUL is stored as NL */ 3749 --clen; 3750 rlen += (int)STRLEN(IObuff + rlen); 3751 if (rlen > IOSIZE - 20) 3752 break; 3753 } 3754 3755 msg(IObuff); 3756 } 3757 3758 /* 3759 * mb_head_off() function pointer. 3760 * Return offset from "p" to the first byte of the character it points into. 3761 * If "p" points to the NUL at the end of the string return 0. 3762 * Returns 0 when already at the first byte of a character. 3763 */ 3764 int 3765 latin_head_off(char_u *base UNUSED, char_u *p UNUSED) 3766 { 3767 return 0; 3768 } 3769 3770 int 3771 dbcs_head_off(char_u *base, char_u *p) 3772 { 3773 char_u *q; 3774 3775 /* It can't be a trailing byte when not using DBCS, at the start of the 3776 * string or the previous byte can't start a double-byte. */ 3777 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3778 return 0; 3779 3780 /* This is slow: need to start at the base and go forward until the 3781 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */ 3782 q = base; 3783 while (q < p) 3784 q += dbcs_ptr2len(q); 3785 return (q == p) ? 0 : 1; 3786 } 3787 3788 /* 3789 * Special version of dbcs_head_off() that works for ScreenLines[], where 3790 * single-width DBCS_JPNU characters are stored separately. 3791 */ 3792 int 3793 dbcs_screen_head_off(char_u *base, char_u *p) 3794 { 3795 char_u *q; 3796 3797 /* It can't be a trailing byte when not using DBCS, at the start of the 3798 * string or the previous byte can't start a double-byte. 3799 * For euc-jp an 0x8e byte in the previous cell always means we have a 3800 * lead byte in the current cell. */ 3801 if (p <= base 3802 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3803 || MB_BYTE2LEN(p[-1]) == 1 3804 || *p == NUL) 3805 return 0; 3806 3807 /* This is slow: need to start at the base and go forward until the 3808 * byte we are looking for. Return 1 when we went past it, 0 otherwise. 3809 * For DBCS_JPNU look out for 0x8e, which means the second byte is not 3810 * stored as the next byte. */ 3811 q = base; 3812 while (q < p) 3813 { 3814 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3815 ++q; 3816 else 3817 q += dbcs_ptr2len(q); 3818 } 3819 return (q == p) ? 0 : 1; 3820 } 3821 3822 int 3823 utf_head_off(char_u *base, char_u *p) 3824 { 3825 char_u *q; 3826 char_u *s; 3827 int c; 3828 int len; 3829 #ifdef FEAT_ARABIC 3830 char_u *j; 3831 #endif 3832 3833 if (*p < 0x80) /* be quick for ASCII */ 3834 return 0; 3835 3836 /* Skip backwards over trailing bytes: 10xx.xxxx 3837 * Skip backwards again if on a composing char. */ 3838 for (q = p; ; --q) 3839 { 3840 /* Move s to the last byte of this char. */ 3841 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3842 ; 3843 /* Move q to the first byte of this char. */ 3844 while (q > base && (*q & 0xc0) == 0x80) 3845 --q; 3846 /* Check for illegal sequence. Do allow an illegal byte after where we 3847 * started. */ 3848 len = utf8len_tab[*q]; 3849 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3850 return 0; 3851 3852 if (q <= base) 3853 break; 3854 3855 c = utf_ptr2char(q); 3856 if (utf_iscomposing(c)) 3857 continue; 3858 3859 #ifdef FEAT_ARABIC 3860 if (arabic_maycombine(c)) 3861 { 3862 /* Advance to get a sneak-peak at the next char */ 3863 j = q; 3864 --j; 3865 /* Move j to the first byte of this char. */ 3866 while (j > base && (*j & 0xc0) == 0x80) 3867 --j; 3868 if (arabic_combine(utf_ptr2char(j), c)) 3869 continue; 3870 } 3871 #endif 3872 break; 3873 } 3874 3875 return (int)(p - q); 3876 } 3877 3878 /* 3879 * Copy a character from "*fp" to "*tp" and advance the pointers. 3880 */ 3881 void 3882 mb_copy_char(char_u **fp, char_u **tp) 3883 { 3884 int l = (*mb_ptr2len)(*fp); 3885 3886 mch_memmove(*tp, *fp, (size_t)l); 3887 *tp += l; 3888 *fp += l; 3889 } 3890 3891 /* 3892 * Return the offset from "p" to the first byte of a character. When "p" is 3893 * at the start of a character 0 is returned, otherwise the offset to the next 3894 * character. Can start anywhere in a stream of bytes. 3895 */ 3896 int 3897 mb_off_next(char_u *base, char_u *p) 3898 { 3899 int i; 3900 int j; 3901 3902 if (enc_utf8) 3903 { 3904 if (*p < 0x80) /* be quick for ASCII */ 3905 return 0; 3906 3907 /* Find the next character that isn't 10xx.xxxx */ 3908 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 3909 ; 3910 if (i > 0) 3911 { 3912 /* Check for illegal sequence. */ 3913 for (j = 0; p - j > base; ++j) 3914 if ((p[-j] & 0xc0) != 0x80) 3915 break; 3916 if (utf8len_tab[p[-j]] != i + j) 3917 return 0; 3918 } 3919 return i; 3920 } 3921 3922 /* Only need to check if we're on a trail byte, it doesn't matter if we 3923 * want the offset to the next or current character. */ 3924 return (*mb_head_off)(base, p); 3925 } 3926 3927 /* 3928 * Return the offset from "p" to the last byte of the character it points 3929 * into. Can start anywhere in a stream of bytes. 3930 */ 3931 int 3932 mb_tail_off(char_u *base, char_u *p) 3933 { 3934 int i; 3935 int j; 3936 3937 if (*p == NUL) 3938 return 0; 3939 3940 if (enc_utf8) 3941 { 3942 /* Find the last character that is 10xx.xxxx */ 3943 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 3944 ; 3945 /* Check for illegal sequence. */ 3946 for (j = 0; p - j > base; ++j) 3947 if ((p[-j] & 0xc0) != 0x80) 3948 break; 3949 if (utf8len_tab[p[-j]] != i + j + 1) 3950 return 0; 3951 return i; 3952 } 3953 3954 /* It can't be the first byte if a double-byte when not using DBCS, at the 3955 * end of the string or the byte can't start a double-byte. */ 3956 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 3957 return 0; 3958 3959 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3960 return 1 - dbcs_head_off(base, p); 3961 } 3962 3963 /* 3964 * Find the next illegal byte sequence. 3965 */ 3966 void 3967 utf_find_illegal(void) 3968 { 3969 pos_T pos = curwin->w_cursor; 3970 char_u *p; 3971 int len; 3972 vimconv_T vimconv; 3973 char_u *tofree = NULL; 3974 3975 vimconv.vc_type = CONV_NONE; 3976 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 3977 { 3978 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 3979 * possibly a utf-8 file with illegal bytes. Setup for conversion 3980 * from utf-8 to 'fileencoding'. */ 3981 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 3982 } 3983 3984 #ifdef FEAT_VIRTUALEDIT 3985 curwin->w_cursor.coladd = 0; 3986 #endif 3987 for (;;) 3988 { 3989 p = ml_get_cursor(); 3990 if (vimconv.vc_type != CONV_NONE) 3991 { 3992 vim_free(tofree); 3993 tofree = string_convert(&vimconv, p, NULL); 3994 if (tofree == NULL) 3995 break; 3996 p = tofree; 3997 } 3998 3999 while (*p != NUL) 4000 { 4001 /* Illegal means that there are not enough trail bytes (checked by 4002 * utf_ptr2len()) or too many of them (overlong sequence). */ 4003 len = utf_ptr2len(p); 4004 if (*p >= 0x80 && (len == 1 4005 || utf_char2len(utf_ptr2char(p)) != len)) 4006 { 4007 if (vimconv.vc_type == CONV_NONE) 4008 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 4009 else 4010 { 4011 int l; 4012 4013 len = (int)(p - tofree); 4014 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 4015 { 4016 l = utf_ptr2len(p); 4017 curwin->w_cursor.col += l; 4018 } 4019 } 4020 goto theend; 4021 } 4022 p += len; 4023 } 4024 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 4025 break; 4026 ++curwin->w_cursor.lnum; 4027 curwin->w_cursor.col = 0; 4028 } 4029 4030 /* didn't find it: don't move and beep */ 4031 curwin->w_cursor = pos; 4032 beep_flush(); 4033 4034 theend: 4035 vim_free(tofree); 4036 convert_setup(&vimconv, NULL, NULL); 4037 } 4038 4039 #if defined(FEAT_GUI_GTK) || defined(PROTO) 4040 /* 4041 * Return TRUE if string "s" is a valid utf-8 string. 4042 * When "end" is NULL stop at the first NUL. 4043 * When "end" is positive stop there. 4044 */ 4045 int 4046 utf_valid_string(char_u *s, char_u *end) 4047 { 4048 int l; 4049 char_u *p = s; 4050 4051 while (end == NULL ? *p != NUL : p < end) 4052 { 4053 l = utf8len_tab_zero[*p]; 4054 if (l == 0) 4055 return FALSE; /* invalid lead byte */ 4056 if (end != NULL && p + l > end) 4057 return FALSE; /* incomplete byte sequence */ 4058 ++p; 4059 while (--l > 0) 4060 if ((*p++ & 0xc0) != 0x80) 4061 return FALSE; /* invalid trail byte */ 4062 } 4063 return TRUE; 4064 } 4065 #endif 4066 4067 #if defined(FEAT_GUI) || defined(PROTO) 4068 /* 4069 * Special version of mb_tail_off() for use in ScreenLines[]. 4070 */ 4071 int 4072 dbcs_screen_tail_off(char_u *base, char_u *p) 4073 { 4074 /* It can't be the first byte if a double-byte when not using DBCS, at the 4075 * end of the string or the byte can't start a double-byte. 4076 * For euc-jp an 0x8e byte always means we have a lead byte in the current 4077 * cell. */ 4078 if (*p == NUL || p[1] == NUL 4079 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 4080 || MB_BYTE2LEN(*p) == 1) 4081 return 0; 4082 4083 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 4084 return 1 - dbcs_screen_head_off(base, p); 4085 } 4086 #endif 4087 4088 /* 4089 * If the cursor moves on an trail byte, set the cursor on the lead byte. 4090 * Thus it moves left if necessary. 4091 * Return TRUE when the cursor was adjusted. 4092 */ 4093 void 4094 mb_adjust_cursor(void) 4095 { 4096 mb_adjustpos(curbuf, &curwin->w_cursor); 4097 } 4098 4099 /* 4100 * Adjust position "*lp" to point to the first byte of a multi-byte character. 4101 * If it points to a tail byte it's moved backwards to the head byte. 4102 */ 4103 void 4104 mb_adjustpos(buf_T *buf, pos_T *lp) 4105 { 4106 char_u *p; 4107 4108 if (lp->col > 0 4109 #ifdef FEAT_VIRTUALEDIT 4110 || lp->coladd > 1 4111 #endif 4112 ) 4113 { 4114 p = ml_get_buf(buf, lp->lnum, FALSE); 4115 lp->col -= (*mb_head_off)(p, p + lp->col); 4116 #ifdef FEAT_VIRTUALEDIT 4117 /* Reset "coladd" when the cursor would be on the right half of a 4118 * double-wide character. */ 4119 if (lp->coladd == 1 4120 && p[lp->col] != TAB 4121 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 4122 && ptr2cells(p + lp->col) > 1) 4123 lp->coladd = 0; 4124 #endif 4125 } 4126 } 4127 4128 /* 4129 * Return a pointer to the character before "*p", if there is one. 4130 */ 4131 char_u * 4132 mb_prevptr( 4133 char_u *line, /* start of the string */ 4134 char_u *p) 4135 { 4136 if (p > line) 4137 MB_PTR_BACK(line, p); 4138 return p; 4139 } 4140 4141 /* 4142 * Return the character length of "str". Each multi-byte character (with 4143 * following composing characters) counts as one. 4144 */ 4145 int 4146 mb_charlen(char_u *str) 4147 { 4148 char_u *p = str; 4149 int count; 4150 4151 if (p == NULL) 4152 return 0; 4153 4154 for (count = 0; *p != NUL; count++) 4155 p += (*mb_ptr2len)(p); 4156 4157 return count; 4158 } 4159 4160 #if defined(FEAT_SPELL) || defined(PROTO) 4161 /* 4162 * Like mb_charlen() but for a string with specified length. 4163 */ 4164 int 4165 mb_charlen_len(char_u *str, int len) 4166 { 4167 char_u *p = str; 4168 int count; 4169 4170 for (count = 0; *p != NUL && p < str + len; count++) 4171 p += (*mb_ptr2len)(p); 4172 4173 return count; 4174 } 4175 #endif 4176 4177 /* 4178 * Try to un-escape a multi-byte character. 4179 * Used for the "to" and "from" part of a mapping. 4180 * Return the un-escaped string if it is a multi-byte character, and advance 4181 * "pp" to just after the bytes that formed it. 4182 * Return NULL if no multi-byte char was found. 4183 */ 4184 char_u * 4185 mb_unescape(char_u **pp) 4186 { 4187 static char_u buf[6]; 4188 int n; 4189 int m = 0; 4190 char_u *str = *pp; 4191 4192 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 4193 * KS_EXTRA KE_CSI to CSI. 4194 * Maximum length of a utf-8 character is 4 bytes. */ 4195 for (n = 0; str[n] != NUL && m < 4; ++n) 4196 { 4197 if (str[n] == K_SPECIAL 4198 && str[n + 1] == KS_SPECIAL 4199 && str[n + 2] == KE_FILLER) 4200 { 4201 buf[m++] = K_SPECIAL; 4202 n += 2; 4203 } 4204 else if ((str[n] == K_SPECIAL 4205 # ifdef FEAT_GUI 4206 || str[n] == CSI 4207 # endif 4208 ) 4209 && str[n + 1] == KS_EXTRA 4210 && str[n + 2] == (int)KE_CSI) 4211 { 4212 buf[m++] = CSI; 4213 n += 2; 4214 } 4215 else if (str[n] == K_SPECIAL 4216 # ifdef FEAT_GUI 4217 || str[n] == CSI 4218 # endif 4219 ) 4220 break; /* a special key can't be a multibyte char */ 4221 else 4222 buf[m++] = str[n]; 4223 buf[m] = NUL; 4224 4225 /* Return a multi-byte character if it's found. An illegal sequence 4226 * will result in a 1 here. */ 4227 if ((*mb_ptr2len)(buf) > 1) 4228 { 4229 *pp = str + n + 1; 4230 return buf; 4231 } 4232 4233 /* Bail out quickly for ASCII. */ 4234 if (buf[0] < 128) 4235 break; 4236 } 4237 return NULL; 4238 } 4239 4240 /* 4241 * Return TRUE if the character at "row"/"col" on the screen is the left side 4242 * of a double-width character. 4243 * Caller must make sure "row" and "col" are not invalid! 4244 */ 4245 int 4246 mb_lefthalve(int row, int col) 4247 { 4248 #ifdef FEAT_HANGULIN 4249 if (composing_hangul) 4250 return TRUE; 4251 #endif 4252 return (*mb_off2cells)(LineOffset[row] + col, 4253 LineOffset[row] + screen_Columns) > 1; 4254 } 4255 4256 /* 4257 * Correct a position on the screen, if it's the right half of a double-wide 4258 * char move it to the left half. Returns the corrected column. 4259 */ 4260 int 4261 mb_fix_col(int col, int row) 4262 { 4263 col = check_col(col); 4264 row = check_row(row); 4265 if (has_mbyte && ScreenLines != NULL && col > 0 4266 && ((enc_dbcs 4267 && ScreenLines[LineOffset[row] + col] != NUL 4268 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 4269 ScreenLines + LineOffset[row] + col)) 4270 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0))) 4271 return col - 1; 4272 return col; 4273 } 4274 #endif 4275 4276 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 4277 static int enc_alias_search(char_u *name); 4278 4279 /* 4280 * Skip the Vim specific head of a 'encoding' name. 4281 */ 4282 char_u * 4283 enc_skip(char_u *p) 4284 { 4285 if (STRNCMP(p, "2byte-", 6) == 0) 4286 return p + 6; 4287 if (STRNCMP(p, "8bit-", 5) == 0) 4288 return p + 5; 4289 return p; 4290 } 4291 4292 /* 4293 * Find the canonical name for encoding "enc". 4294 * When the name isn't recognized, returns "enc" itself, but with all lower 4295 * case characters and '_' replaced with '-'. 4296 * Returns an allocated string. NULL for out-of-memory. 4297 */ 4298 char_u * 4299 enc_canonize(char_u *enc) 4300 { 4301 char_u *r; 4302 char_u *p, *s; 4303 int i; 4304 4305 # ifdef FEAT_MBYTE 4306 if (STRCMP(enc, "default") == 0) 4307 { 4308 /* Use the default encoding as it's found by set_init_1(). */ 4309 r = get_encoding_default(); 4310 if (r == NULL) 4311 r = (char_u *)"latin1"; 4312 return vim_strsave(r); 4313 } 4314 # endif 4315 4316 /* copy "enc" to allocated memory, with room for two '-' */ 4317 r = alloc((unsigned)(STRLEN(enc) + 3)); 4318 if (r != NULL) 4319 { 4320 /* Make it all lower case and replace '_' with '-'. */ 4321 p = r; 4322 for (s = enc; *s != NUL; ++s) 4323 { 4324 if (*s == '_') 4325 *p++ = '-'; 4326 else 4327 *p++ = TOLOWER_ASC(*s); 4328 } 4329 *p = NUL; 4330 4331 /* Skip "2byte-" and "8bit-". */ 4332 p = enc_skip(r); 4333 4334 /* Change "microsoft-cp" to "cp". Used in some spell files. */ 4335 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4336 STRMOVE(p, p + 10); 4337 4338 /* "iso8859" -> "iso-8859" */ 4339 if (STRNCMP(p, "iso8859", 7) == 0) 4340 { 4341 STRMOVE(p + 4, p + 3); 4342 p[3] = '-'; 4343 } 4344 4345 /* "iso-8859n" -> "iso-8859-n" */ 4346 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4347 { 4348 STRMOVE(p + 9, p + 8); 4349 p[8] = '-'; 4350 } 4351 4352 /* "latin-N" -> "latinN" */ 4353 if (STRNCMP(p, "latin-", 6) == 0) 4354 STRMOVE(p + 5, p + 6); 4355 4356 if (enc_canon_search(p) >= 0) 4357 { 4358 /* canonical name can be used unmodified */ 4359 if (p != r) 4360 STRMOVE(r, p); 4361 } 4362 else if ((i = enc_alias_search(p)) >= 0) 4363 { 4364 /* alias recognized, get canonical name */ 4365 vim_free(r); 4366 r = vim_strsave((char_u *)enc_canon_table[i].name); 4367 } 4368 } 4369 return r; 4370 } 4371 4372 /* 4373 * Search for an encoding alias of "name". 4374 * Returns -1 when not found. 4375 */ 4376 static int 4377 enc_alias_search(char_u *name) 4378 { 4379 int i; 4380 4381 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4382 if (STRCMP(name, enc_alias_table[i].name) == 0) 4383 return enc_alias_table[i].canon; 4384 return -1; 4385 } 4386 #endif 4387 4388 #if defined(FEAT_MBYTE) || defined(PROTO) 4389 4390 # ifdef HAVE_LANGINFO_H 4391 # include <langinfo.h> 4392 # endif 4393 4394 # ifndef FEAT_GUI_W32 4395 /* 4396 * Get the canonicalized encoding from the specified locale string "locale" 4397 * or from the environment variables LC_ALL, LC_CTYPE and LANG. 4398 * Returns an allocated string when successful, NULL when not. 4399 */ 4400 char_u * 4401 enc_locale_env(char *locale) 4402 { 4403 char *s = locale; 4404 char *p; 4405 int i; 4406 char buf[50]; 4407 4408 if (s == NULL || *s == NUL) 4409 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4410 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4411 s = getenv("LANG"); 4412 4413 if (s == NULL || *s == NUL) 4414 return NULL; 4415 4416 /* The most generic locale format is: 4417 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4418 * If there is a '.' remove the part before it. 4419 * if there is something after the codeset, remove it. 4420 * Make the name lowercase and replace '_' with '-'. 4421 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4422 * "ko_KR.EUC" == "euc-kr" 4423 */ 4424 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4425 { 4426 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4427 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4428 { 4429 /* copy "XY.EUC" to "euc-XY" to buf[10] */ 4430 STRCPY(buf + 10, "euc-"); 4431 buf[14] = p[-2]; 4432 buf[15] = p[-1]; 4433 buf[16] = 0; 4434 s = buf + 10; 4435 } 4436 else 4437 s = p + 1; 4438 } 4439 for (i = 0; i < (int)sizeof(buf) - 1 && s[i] != NUL; ++i) 4440 { 4441 if (s[i] == '_' || s[i] == '-') 4442 buf[i] = '-'; 4443 else if (isalnum((int)s[i])) 4444 buf[i] = TOLOWER_ASC(s[i]); 4445 else 4446 break; 4447 } 4448 buf[i] = NUL; 4449 4450 return enc_canonize((char_u *)buf); 4451 } 4452 # endif 4453 4454 /* 4455 * Get the canonicalized encoding of the current locale. 4456 * Returns an allocated string when successful, NULL when not. 4457 */ 4458 char_u * 4459 enc_locale(void) 4460 { 4461 # ifdef WIN3264 4462 char buf[50]; 4463 long acp = GetACP(); 4464 4465 if (acp == 1200) 4466 STRCPY(buf, "ucs-2le"); 4467 else if (acp == 1252) /* cp1252 is used as latin1 */ 4468 STRCPY(buf, "latin1"); 4469 else 4470 sprintf(buf, "cp%ld", acp); 4471 4472 return enc_canonize((char_u *)buf); 4473 # else 4474 char *s; 4475 4476 # ifdef HAVE_NL_LANGINFO_CODESET 4477 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4478 # endif 4479 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4480 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4481 # endif 4482 s = NULL; 4483 4484 return enc_locale_env(s); 4485 # endif 4486 } 4487 4488 # if defined(WIN3264) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4489 /* 4490 * Convert an encoding name to an MS-Windows codepage. 4491 * Returns zero if no codepage can be figured out. 4492 */ 4493 int 4494 encname2codepage(char_u *name) 4495 { 4496 int cp; 4497 char_u *p = name; 4498 int idx; 4499 4500 if (STRNCMP(p, "8bit-", 5) == 0) 4501 p += 5; 4502 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4503 p += 6; 4504 4505 if (p[0] == 'c' && p[1] == 'p') 4506 cp = atoi((char *)p + 2); 4507 else if ((idx = enc_canon_search(p)) >= 0) 4508 cp = enc_canon_table[idx].codepage; 4509 else 4510 return 0; 4511 if (IsValidCodePage(cp)) 4512 return cp; 4513 return 0; 4514 } 4515 # endif 4516 4517 # if defined(USE_ICONV) || defined(PROTO) 4518 4519 static char_u *iconv_string(vimconv_T *vcp, char_u *str, int slen, int *unconvlenp, int *resultlenp); 4520 4521 /* 4522 * Call iconv_open() with a check if iconv() works properly (there are broken 4523 * versions). 4524 * Returns (void *)-1 if failed. 4525 * (should return iconv_t, but that causes problems with prototypes). 4526 */ 4527 void * 4528 my_iconv_open(char_u *to, char_u *from) 4529 { 4530 iconv_t fd; 4531 #define ICONV_TESTLEN 400 4532 char_u tobuf[ICONV_TESTLEN]; 4533 char *p; 4534 size_t tolen; 4535 static int iconv_ok = -1; 4536 4537 if (iconv_ok == FALSE) 4538 return (void *)-1; /* detected a broken iconv() previously */ 4539 4540 #ifdef DYNAMIC_ICONV 4541 /* Check if the iconv.dll can be found. */ 4542 if (!iconv_enabled(TRUE)) 4543 return (void *)-1; 4544 #endif 4545 4546 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4547 4548 if (fd != (iconv_t)-1 && iconv_ok == -1) 4549 { 4550 /* 4551 * Do a dummy iconv() call to check if it actually works. There is a 4552 * version of iconv() on Linux that is broken. We can't ignore it, 4553 * because it's wide-spread. The symptoms are that after outputting 4554 * the initial shift state the "to" pointer is NULL and conversion 4555 * stops for no apparent reason after about 8160 characters. 4556 */ 4557 p = (char *)tobuf; 4558 tolen = ICONV_TESTLEN; 4559 (void)iconv(fd, NULL, NULL, &p, &tolen); 4560 if (p == NULL) 4561 { 4562 iconv_ok = FALSE; 4563 iconv_close(fd); 4564 fd = (iconv_t)-1; 4565 } 4566 else 4567 iconv_ok = TRUE; 4568 } 4569 4570 return (void *)fd; 4571 } 4572 4573 /* 4574 * Convert the string "str[slen]" with iconv(). 4575 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4576 * sequence and set "*unconvlenp" to the length of it. 4577 * Returns the converted string in allocated memory. NULL for an error. 4578 * If resultlenp is not NULL, sets it to the result length in bytes. 4579 */ 4580 static char_u * 4581 iconv_string( 4582 vimconv_T *vcp, 4583 char_u *str, 4584 int slen, 4585 int *unconvlenp, 4586 int *resultlenp) 4587 { 4588 const char *from; 4589 size_t fromlen; 4590 char *to; 4591 size_t tolen; 4592 size_t len = 0; 4593 size_t done = 0; 4594 char_u *result = NULL; 4595 char_u *p; 4596 int l; 4597 4598 from = (char *)str; 4599 fromlen = slen; 4600 for (;;) 4601 { 4602 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4603 { 4604 /* Allocate enough room for most conversions. When re-allocating 4605 * increase the buffer size. */ 4606 len = len + fromlen * 2 + 40; 4607 p = alloc((unsigned)len); 4608 if (p != NULL && done > 0) 4609 mch_memmove(p, result, done); 4610 vim_free(result); 4611 result = p; 4612 if (result == NULL) /* out of memory */ 4613 break; 4614 } 4615 4616 to = (char *)result + done; 4617 tolen = len - done - 2; 4618 /* Avoid a warning for systems with a wrong iconv() prototype by 4619 * casting the second argument to void *. */ 4620 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4621 != (size_t)-1) 4622 { 4623 /* Finished, append a NUL. */ 4624 *to = NUL; 4625 break; 4626 } 4627 4628 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4629 * iconv library may use one of them. */ 4630 if (!vcp->vc_fail && unconvlenp != NULL 4631 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4632 { 4633 /* Handle an incomplete sequence at the end. */ 4634 *to = NUL; 4635 *unconvlenp = (int)fromlen; 4636 break; 4637 } 4638 4639 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4640 * iconv library may use one of them. */ 4641 else if (!vcp->vc_fail 4642 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4643 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4644 { 4645 /* Can't convert: insert a '?' and skip a character. This assumes 4646 * conversion from 'encoding' to something else. In other 4647 * situations we don't know what to skip anyway. */ 4648 *to++ = '?'; 4649 if ((*mb_ptr2cells)((char_u *)from) > 1) 4650 *to++ = '?'; 4651 if (enc_utf8) 4652 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4653 else 4654 { 4655 l = (*mb_ptr2len)((char_u *)from); 4656 if (l > (int)fromlen) 4657 l = (int)fromlen; 4658 } 4659 from += l; 4660 fromlen -= l; 4661 } 4662 else if (ICONV_ERRNO != ICONV_E2BIG) 4663 { 4664 /* conversion failed */ 4665 vim_free(result); 4666 result = NULL; 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 static void 4795 call_imactivatefunc(int active) 4796 { 4797 char_u *argv[1]; 4798 4799 if (active) 4800 argv[0] = (char_u *)"1"; 4801 else 4802 argv[0] = (char_u *)"0"; 4803 (void)call_func_retnr(p_imaf, 1, argv, FALSE); 4804 } 4805 4806 static int 4807 call_imstatusfunc(void) 4808 { 4809 int is_active; 4810 4811 /* FIXME: Don't execute user function in unsafe situation. */ 4812 if (exiting 4813 # ifdef FEAT_AUTOCMD 4814 || is_autocmd_blocked() 4815 # endif 4816 ) 4817 return FALSE; 4818 /* FIXME: :py print 'xxx' is shown duplicate result. 4819 * Use silent to avoid it. */ 4820 ++msg_silent; 4821 is_active = call_func_retnr(p_imsf, 0, NULL, FALSE); 4822 --msg_silent; 4823 return (is_active > 0); 4824 } 4825 #endif 4826 4827 #if defined(FEAT_XIM) || defined(PROTO) 4828 4829 # if defined(FEAT_GUI_GTK) || defined(PROTO) 4830 static int xim_has_preediting INIT(= FALSE); /* IM current status */ 4831 4832 /* 4833 * Set preedit_start_col to the current cursor position. 4834 */ 4835 static void 4836 init_preedit_start_col(void) 4837 { 4838 if (State & CMDLINE) 4839 preedit_start_col = cmdline_getvcol_cursor(); 4840 else if (curwin != NULL && curwin->w_buffer != NULL) 4841 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); 4842 /* Prevent that preediting marks the buffer as changed. */ 4843 xim_changed_while_preediting = curbuf->b_changed; 4844 } 4845 4846 static int im_is_active = FALSE; /* IM is enabled for current mode */ 4847 static int preedit_is_active = FALSE; 4848 static int im_preedit_cursor = 0; /* cursor offset in characters */ 4849 static int im_preedit_trailing = 0; /* number of characters after cursor */ 4850 4851 static unsigned long im_commit_handler_id = 0; 4852 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; 4853 static unsigned int im_activatekey_state = 0; 4854 4855 static GtkWidget *preedit_window = NULL; 4856 static GtkWidget *preedit_label = NULL; 4857 4858 static void im_preedit_window_set_position(void); 4859 4860 void 4861 im_set_active(int active) 4862 { 4863 int was_active; 4864 4865 was_active = !!im_get_status(); 4866 im_is_active = (active && !p_imdisable); 4867 4868 if (im_is_active != was_active) 4869 xim_reset(); 4870 } 4871 4872 void 4873 xim_set_focus(int focus) 4874 { 4875 if (xic != NULL) 4876 { 4877 if (focus) 4878 gtk_im_context_focus_in(xic); 4879 else 4880 gtk_im_context_focus_out(xic); 4881 } 4882 } 4883 4884 void 4885 im_set_position(int row, int col) 4886 { 4887 if (xic != NULL) 4888 { 4889 GdkRectangle area; 4890 4891 area.x = FILL_X(col); 4892 area.y = FILL_Y(row); 4893 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); 4894 area.height = gui.char_height; 4895 4896 gtk_im_context_set_cursor_location(xic, &area); 4897 4898 if (p_imst == IM_OVER_THE_SPOT) 4899 im_preedit_window_set_position(); 4900 } 4901 } 4902 4903 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */ 4904 void 4905 xim_set_preedit(void) 4906 { 4907 im_set_position(gui.row, gui.col); 4908 } 4909 # endif 4910 4911 static void 4912 im_add_to_input(char_u *str, int len) 4913 { 4914 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */ 4915 if (input_conv.vc_type != CONV_NONE) 4916 { 4917 str = string_convert(&input_conv, str, &len); 4918 g_return_if_fail(str != NULL); 4919 } 4920 4921 add_to_input_buf_csi(str, len); 4922 4923 if (input_conv.vc_type != CONV_NONE) 4924 vim_free(str); 4925 4926 if (p_mh) /* blank out the pointer if necessary */ 4927 gui_mch_mousehide(TRUE); 4928 } 4929 4930 static void 4931 im_preedit_window_set_position(void) 4932 { 4933 int x, y, w, h, sw, sh; 4934 4935 if (preedit_window == NULL) 4936 return; 4937 4938 gui_gtk_get_screen_size_of_win(preedit_window, &sw, &sh); 4939 #if GTK_CHECK_VERSION(3,0,0) 4940 gdk_window_get_origin(gtk_widget_get_window(gui.drawarea), &x, &y); 4941 #else 4942 gdk_window_get_origin(gui.drawarea->window, &x, &y); 4943 #endif 4944 gtk_window_get_size(GTK_WINDOW(preedit_window), &w, &h); 4945 x = x + FILL_X(gui.col); 4946 y = y + FILL_Y(gui.row); 4947 if (x + w > sw) 4948 x = sw - w; 4949 if (y + h > sh) 4950 y = sh - h; 4951 gtk_window_move(GTK_WINDOW(preedit_window), x, y); 4952 } 4953 4954 static void 4955 im_preedit_window_open() 4956 { 4957 char *preedit_string; 4958 #if !GTK_CHECK_VERSION(3,16,0) 4959 char buf[8]; 4960 #endif 4961 PangoAttrList *attr_list; 4962 PangoLayout *layout; 4963 #if GTK_CHECK_VERSION(3,0,0) 4964 # if !GTK_CHECK_VERSION(3,16,0) 4965 GdkRGBA color; 4966 # endif 4967 #else 4968 GdkColor color; 4969 #endif 4970 gint w, h; 4971 4972 if (preedit_window == NULL) 4973 { 4974 preedit_window = gtk_window_new(GTK_WINDOW_POPUP); 4975 gtk_window_set_transient_for(GTK_WINDOW(preedit_window), 4976 GTK_WINDOW(gui.mainwin)); 4977 preedit_label = gtk_label_new(""); 4978 gtk_widget_set_name(preedit_label, "vim-gui-preedit-area"); 4979 gtk_container_add(GTK_CONTAINER(preedit_window), preedit_label); 4980 } 4981 4982 #if GTK_CHECK_VERSION(3,16,0) 4983 { 4984 GtkStyleContext * const context 4985 = gtk_widget_get_style_context(gui.drawarea); 4986 GtkCssProvider * const provider = gtk_css_provider_new(); 4987 gchar *css = NULL; 4988 const char * const fontname 4989 = pango_font_description_get_family(gui.norm_font); 4990 gint fontsize 4991 = pango_font_description_get_size(gui.norm_font) / PANGO_SCALE; 4992 gchar *fontsize_propval = NULL; 4993 4994 if (!pango_font_description_get_size_is_absolute(gui.norm_font)) 4995 { 4996 /* fontsize was given in points. Convert it into that in pixels 4997 * to use with CSS. */ 4998 GdkScreen * const screen 4999 = gdk_window_get_screen(gtk_widget_get_window(gui.mainwin)); 5000 const gdouble dpi = gdk_screen_get_resolution(screen); 5001 fontsize = dpi * fontsize / 72; 5002 } 5003 if (fontsize > 0) 5004 fontsize_propval = g_strdup_printf("%dpx", fontsize); 5005 else 5006 fontsize_propval = g_strdup_printf("inherit"); 5007 5008 css = g_strdup_printf( 5009 "widget#vim-gui-preedit-area {\n" 5010 " font-family: %s,monospace;\n" 5011 " font-size: %s;\n" 5012 " color: #%.2lx%.2lx%.2lx;\n" 5013 " background-color: #%.2lx%.2lx%.2lx;\n" 5014 "}\n", 5015 fontname != NULL ? fontname : "inherit", 5016 fontsize_propval, 5017 (gui.norm_pixel >> 16) & 0xff, 5018 (gui.norm_pixel >> 8) & 0xff, 5019 gui.norm_pixel & 0xff, 5020 (gui.back_pixel >> 16) & 0xff, 5021 (gui.back_pixel >> 8) & 0xff, 5022 gui.back_pixel & 0xff); 5023 5024 gtk_css_provider_load_from_data(provider, css, -1, NULL); 5025 gtk_style_context_add_provider(context, 5026 GTK_STYLE_PROVIDER(provider), G_MAXUINT); 5027 5028 g_free(css); 5029 g_free(fontsize_propval); 5030 g_object_unref(provider); 5031 } 5032 #elif GTK_CHECK_VERSION(3,0,0) 5033 gtk_widget_override_font(preedit_label, gui.norm_font); 5034 5035 vim_snprintf(buf, sizeof(buf), "#%06X", gui.norm_pixel); 5036 gdk_rgba_parse(&color, buf); 5037 gtk_widget_override_color(preedit_label, GTK_STATE_FLAG_NORMAL, &color); 5038 5039 vim_snprintf(buf, sizeof(buf), "#%06X", gui.back_pixel); 5040 gdk_rgba_parse(&color, buf); 5041 gtk_widget_override_background_color(preedit_label, GTK_STATE_FLAG_NORMAL, 5042 &color); 5043 #else 5044 gtk_widget_modify_font(preedit_label, gui.norm_font); 5045 5046 vim_snprintf(buf, sizeof(buf), "#%06X", gui.norm_pixel); 5047 gdk_color_parse(buf, &color); 5048 gtk_widget_modify_fg(preedit_label, GTK_STATE_NORMAL, &color); 5049 5050 vim_snprintf(buf, sizeof(buf), "#%06X", gui.back_pixel); 5051 gdk_color_parse(buf, &color); 5052 gtk_widget_modify_bg(preedit_window, GTK_STATE_NORMAL, &color); 5053 #endif 5054 5055 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5056 5057 if (preedit_string[0] != NUL) 5058 { 5059 gtk_label_set_text(GTK_LABEL(preedit_label), preedit_string); 5060 gtk_label_set_attributes(GTK_LABEL(preedit_label), attr_list); 5061 5062 layout = gtk_label_get_layout(GTK_LABEL(preedit_label)); 5063 pango_layout_get_pixel_size(layout, &w, &h); 5064 h = MAX(h, gui.char_height); 5065 gtk_window_resize(GTK_WINDOW(preedit_window), w, h); 5066 5067 gtk_widget_show_all(preedit_window); 5068 5069 im_preedit_window_set_position(); 5070 } 5071 5072 g_free(preedit_string); 5073 pango_attr_list_unref(attr_list); 5074 } 5075 5076 static void 5077 im_preedit_window_close() 5078 { 5079 if (preedit_window != NULL) 5080 gtk_widget_hide(preedit_window); 5081 } 5082 5083 static void 5084 im_show_preedit() 5085 { 5086 im_preedit_window_open(); 5087 5088 if (p_mh) /* blank out the pointer if necessary */ 5089 gui_mch_mousehide(TRUE); 5090 } 5091 5092 static void 5093 im_delete_preedit(void) 5094 { 5095 char_u bskey[] = {CSI, 'k', 'b'}; 5096 char_u delkey[] = {CSI, 'k', 'D'}; 5097 5098 if (p_imst == IM_OVER_THE_SPOT) 5099 { 5100 im_preedit_window_close(); 5101 return; 5102 } 5103 5104 if (State & NORMAL 5105 #ifdef FEAT_TERMINAL 5106 && !term_use_loop() 5107 #endif 5108 ) 5109 { 5110 im_preedit_cursor = 0; 5111 return; 5112 } 5113 for (; im_preedit_cursor > 0; --im_preedit_cursor) 5114 add_to_input_buf(bskey, (int)sizeof(bskey)); 5115 5116 for (; im_preedit_trailing > 0; --im_preedit_trailing) 5117 add_to_input_buf(delkey, (int)sizeof(delkey)); 5118 } 5119 5120 /* 5121 * Move the cursor left by "num_move_back" characters. 5122 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for 5123 * these K_LEFT keys. 5124 */ 5125 static void 5126 im_correct_cursor(int num_move_back) 5127 { 5128 char_u backkey[] = {CSI, 'k', 'l'}; 5129 5130 if (State & NORMAL) 5131 return; 5132 # ifdef FEAT_RIGHTLEFT 5133 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) 5134 backkey[2] = 'r'; 5135 # endif 5136 for (; num_move_back > 0; --num_move_back) 5137 add_to_input_buf(backkey, (int)sizeof(backkey)); 5138 } 5139 5140 static int xim_expected_char = NUL; 5141 static int xim_ignored_char = FALSE; 5142 5143 /* 5144 * Update the mode and cursor while in an IM callback. 5145 */ 5146 static void 5147 im_show_info(void) 5148 { 5149 int old_vgetc_busy; 5150 5151 old_vgetc_busy = vgetc_busy; 5152 vgetc_busy = TRUE; 5153 showmode(); 5154 vgetc_busy = old_vgetc_busy; 5155 if ((State & NORMAL) || (State & INSERT)) 5156 setcursor(); 5157 out_flush(); 5158 } 5159 5160 /* 5161 * Callback invoked when the user finished preediting. 5162 * Put the final string into the input buffer. 5163 */ 5164 static void 5165 im_commit_cb(GtkIMContext *context UNUSED, 5166 const gchar *str, 5167 gpointer data UNUSED) 5168 { 5169 int slen = (int)STRLEN(str); 5170 int add_to_input = TRUE; 5171 int clen; 5172 int len = slen; 5173 int commit_with_preedit = TRUE; 5174 char_u *im_str; 5175 5176 #ifdef XIM_DEBUG 5177 xim_log("im_commit_cb(): %s\n", str); 5178 #endif 5179 5180 if (p_imst == IM_ON_THE_SPOT) 5181 { 5182 /* The imhangul module doesn't reset the preedit string before 5183 * committing. Call im_delete_preedit() to work around that. */ 5184 im_delete_preedit(); 5185 5186 /* Indicate that preediting has finished. */ 5187 if (preedit_start_col == MAXCOL) 5188 { 5189 init_preedit_start_col(); 5190 commit_with_preedit = FALSE; 5191 } 5192 5193 /* The thing which setting "preedit_start_col" to MAXCOL means that 5194 * "preedit_start_col" will be set forcedly when calling 5195 * preedit_changed_cb() next time. 5196 * "preedit_start_col" should not reset with MAXCOL on this part. Vim 5197 * is simulating the preediting by using add_to_input_str(). when 5198 * preedit begin immediately before committed, the typebuf is not 5199 * flushed to screen, then it can't get correct "preedit_start_col". 5200 * Thus, it should calculate the cells by adding cells of the committed 5201 * string. */ 5202 if (input_conv.vc_type != CONV_NONE) 5203 { 5204 im_str = string_convert(&input_conv, (char_u *)str, &len); 5205 g_return_if_fail(im_str != NULL); 5206 } 5207 else 5208 im_str = (char_u *)str; 5209 5210 clen = mb_string2cells(im_str, len); 5211 5212 if (input_conv.vc_type != CONV_NONE) 5213 vim_free(im_str); 5214 preedit_start_col += clen; 5215 } 5216 5217 /* Is this a single character that matches a keypad key that's just 5218 * been pressed? If so, we don't want it to be entered as such - let 5219 * us carry on processing the raw keycode so that it may be used in 5220 * mappings as <kSomething>. */ 5221 if (xim_expected_char != NUL) 5222 { 5223 /* We're currently processing a keypad or other special key */ 5224 if (slen == 1 && str[0] == xim_expected_char) 5225 { 5226 /* It's a match - don't do it here */ 5227 xim_ignored_char = TRUE; 5228 add_to_input = FALSE; 5229 } 5230 else 5231 { 5232 /* Not a match */ 5233 xim_ignored_char = FALSE; 5234 } 5235 } 5236 5237 if (add_to_input) 5238 im_add_to_input((char_u *)str, slen); 5239 5240 if (p_imst == IM_ON_THE_SPOT) 5241 { 5242 /* Inserting chars while "im_is_active" is set does not cause a 5243 * change of buffer. When the chars are committed the buffer must be 5244 * marked as changed. */ 5245 if (!commit_with_preedit) 5246 preedit_start_col = MAXCOL; 5247 5248 /* This flag is used in changed() at next call. */ 5249 xim_changed_while_preediting = TRUE; 5250 } 5251 5252 if (gtk_main_level() > 0) 5253 gtk_main_quit(); 5254 } 5255 5256 /* 5257 * Callback invoked after start to the preedit. 5258 */ 5259 static void 5260 im_preedit_start_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 5261 { 5262 #ifdef XIM_DEBUG 5263 xim_log("im_preedit_start_cb()\n"); 5264 #endif 5265 5266 im_is_active = TRUE; 5267 preedit_is_active = TRUE; 5268 gui_update_cursor(TRUE, FALSE); 5269 im_show_info(); 5270 } 5271 5272 /* 5273 * Callback invoked after end to the preedit. 5274 */ 5275 static void 5276 im_preedit_end_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 5277 { 5278 #ifdef XIM_DEBUG 5279 xim_log("im_preedit_end_cb()\n"); 5280 #endif 5281 im_delete_preedit(); 5282 5283 /* Indicate that preediting has finished */ 5284 if (p_imst == IM_ON_THE_SPOT) 5285 preedit_start_col = MAXCOL; 5286 xim_has_preediting = FALSE; 5287 5288 #if 0 5289 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was 5290 * switched off unintentionally. We now use preedit_is_active (added by 5291 * SungHyun Nam). */ 5292 im_is_active = FALSE; 5293 #endif 5294 preedit_is_active = FALSE; 5295 gui_update_cursor(TRUE, FALSE); 5296 im_show_info(); 5297 } 5298 5299 /* 5300 * Callback invoked after changes to the preedit string. If the preedit 5301 * string was empty before, remember the preedit start column so we know 5302 * where to apply feedback attributes. Delete the previous preedit string 5303 * if there was one, save the new preedit cursor offset, and put the new 5304 * string into the input buffer. 5305 * 5306 * TODO: The pragmatic "put into input buffer" approach used here has 5307 * several fundamental problems: 5308 * 5309 * - The characters in the preedit string are subject to remapping. 5310 * That's broken, only the finally committed string should be remapped. 5311 * 5312 * - There is a race condition involved: The retrieved value for the 5313 * current cursor position will be wrong if any unprocessed characters 5314 * are still queued in the input buffer. 5315 * 5316 * - Due to the lack of synchronization between the file buffer in memory 5317 * and any typed characters, it's practically impossible to implement the 5318 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM 5319 * modules for languages such as Thai are likely to rely on this feature 5320 * for proper operation. 5321 * 5322 * Conclusions: I think support for preediting needs to be moved to the 5323 * core parts of Vim. Ideally, until it has been committed, the preediting 5324 * string should only be displayed and not affect the buffer content at all. 5325 * The question how to deal with the synchronization issue still remains. 5326 * Circumventing the input buffer is probably not desirable. Anyway, I think 5327 * implementing "retrieve_surrounding" is the only hard problem. 5328 * 5329 * One way to solve all of this in a clean manner would be to queue all key 5330 * press/release events "as is" in the input buffer, and apply the IM filtering 5331 * at the receiving end of the queue. This, however, would have a rather large 5332 * impact on the code base. If there is an easy way to force processing of all 5333 * remaining input from within the "retrieve_surrounding" signal handler, this 5334 * might not be necessary. Gotta ask on vim-dev for opinions. 5335 */ 5336 static void 5337 im_preedit_changed_cb(GtkIMContext *context, gpointer data UNUSED) 5338 { 5339 char *preedit_string = NULL; 5340 int cursor_index = 0; 5341 int num_move_back = 0; 5342 char_u *str; 5343 char_u *p; 5344 int i; 5345 5346 if (p_imst == IM_ON_THE_SPOT) 5347 gtk_im_context_get_preedit_string(context, 5348 &preedit_string, NULL, 5349 &cursor_index); 5350 else 5351 gtk_im_context_get_preedit_string(context, 5352 &preedit_string, NULL, 5353 NULL); 5354 5355 #ifdef XIM_DEBUG 5356 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); 5357 #endif 5358 5359 g_return_if_fail(preedit_string != NULL); /* just in case */ 5360 5361 if (p_imst == IM_OVER_THE_SPOT) 5362 { 5363 if (preedit_string[0] == NUL) 5364 { 5365 xim_has_preediting = FALSE; 5366 im_delete_preedit(); 5367 } 5368 else 5369 { 5370 xim_has_preediting = TRUE; 5371 im_show_preedit(); 5372 } 5373 } 5374 else 5375 { 5376 /* If preedit_start_col is MAXCOL set it to the current cursor position. */ 5377 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') 5378 { 5379 xim_has_preediting = TRUE; 5380 5381 /* Urgh, this breaks if the input buffer isn't empty now */ 5382 init_preedit_start_col(); 5383 } 5384 else if (cursor_index == 0 && preedit_string[0] == '\0') 5385 { 5386 xim_has_preediting = FALSE; 5387 5388 /* If at the start position (after typing backspace) 5389 * preedit_start_col must be reset. */ 5390 preedit_start_col = MAXCOL; 5391 } 5392 5393 im_delete_preedit(); 5394 5395 /* 5396 * Compute the end of the preediting area: "preedit_end_col". 5397 * According to the documentation of gtk_im_context_get_preedit_string(), 5398 * the cursor_pos output argument returns the offset in bytes. This is 5399 * unfortunately not true -- real life shows the offset is in characters, 5400 * and the GTK+ source code agrees with me. Will file a bug later. 5401 */ 5402 if (preedit_start_col != MAXCOL) 5403 preedit_end_col = preedit_start_col; 5404 str = (char_u *)preedit_string; 5405 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) 5406 { 5407 int is_composing; 5408 5409 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); 5410 /* 5411 * These offsets are used as counters when generating <BS> and <Del> 5412 * to delete the preedit string. So don't count composing characters 5413 * unless 'delcombine' is enabled. 5414 */ 5415 if (!is_composing || p_deco) 5416 { 5417 if (i < cursor_index) 5418 ++im_preedit_cursor; 5419 else 5420 ++im_preedit_trailing; 5421 } 5422 if (!is_composing && i >= cursor_index) 5423 { 5424 /* This is essentially the same as im_preedit_trailing, except 5425 * composing characters are not counted even if p_deco is set. */ 5426 ++num_move_back; 5427 } 5428 if (preedit_start_col != MAXCOL) 5429 preedit_end_col += utf_ptr2cells(p); 5430 } 5431 5432 if (p > str) 5433 { 5434 im_add_to_input(str, (int)(p - str)); 5435 im_correct_cursor(num_move_back); 5436 } 5437 } 5438 5439 g_free(preedit_string); 5440 5441 if (gtk_main_level() > 0) 5442 gtk_main_quit(); 5443 } 5444 5445 /* 5446 * Translate the Pango attributes at iter to Vim highlighting attributes. 5447 * Ignore attributes not supported by Vim highlighting. This shouldn't have 5448 * too much impact -- right now we handle even more attributes than necessary 5449 * for the IM modules I tested with. 5450 */ 5451 static int 5452 translate_pango_attributes(PangoAttrIterator *iter) 5453 { 5454 PangoAttribute *attr; 5455 int char_attr = HL_NORMAL; 5456 5457 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); 5458 if (attr != NULL && ((PangoAttrInt *)attr)->value 5459 != (int)PANGO_UNDERLINE_NONE) 5460 char_attr |= HL_UNDERLINE; 5461 5462 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); 5463 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) 5464 char_attr |= HL_BOLD; 5465 5466 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); 5467 if (attr != NULL && ((PangoAttrInt *)attr)->value 5468 != (int)PANGO_STYLE_NORMAL) 5469 char_attr |= HL_ITALIC; 5470 5471 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); 5472 if (attr != NULL) 5473 { 5474 const PangoColor *color = &((PangoAttrColor *)attr)->color; 5475 5476 /* Assume inverse if black background is requested */ 5477 if ((color->red | color->green | color->blue) == 0) 5478 char_attr |= HL_INVERSE; 5479 } 5480 5481 return char_attr; 5482 } 5483 5484 /* 5485 * Retrieve the highlighting attributes at column col in the preedit string. 5486 * Return -1 if not in preediting mode or if col is out of range. 5487 */ 5488 int 5489 im_get_feedback_attr(int col) 5490 { 5491 char *preedit_string = NULL; 5492 PangoAttrList *attr_list = NULL; 5493 int char_attr = -1; 5494 5495 if (xic == NULL) 5496 return char_attr; 5497 5498 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5499 5500 if (preedit_string != NULL && attr_list != NULL) 5501 { 5502 int idx; 5503 5504 /* Get the byte index as used by PangoAttrIterator */ 5505 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col) 5506 idx += utfc_ptr2len((char_u *)preedit_string + idx); 5507 5508 if (preedit_string[idx] != '\0') 5509 { 5510 PangoAttrIterator *iter; 5511 int start, end; 5512 5513 char_attr = HL_NORMAL; 5514 iter = pango_attr_list_get_iterator(attr_list); 5515 5516 /* Extract all relevant attributes from the list. */ 5517 do 5518 { 5519 pango_attr_iterator_range(iter, &start, &end); 5520 5521 if (idx >= start && idx < end) 5522 char_attr |= translate_pango_attributes(iter); 5523 } 5524 while (pango_attr_iterator_next(iter)); 5525 5526 pango_attr_iterator_destroy(iter); 5527 } 5528 } 5529 5530 if (attr_list != NULL) 5531 pango_attr_list_unref(attr_list); 5532 g_free(preedit_string); 5533 5534 return char_attr; 5535 } 5536 5537 void 5538 xim_init(void) 5539 { 5540 #ifdef XIM_DEBUG 5541 xim_log("xim_init()\n"); 5542 #endif 5543 5544 g_return_if_fail(gui.drawarea != NULL); 5545 #if GTK_CHECK_VERSION(3,0,0) 5546 g_return_if_fail(gtk_widget_get_window(gui.drawarea) != NULL); 5547 #else 5548 g_return_if_fail(gui.drawarea->window != NULL); 5549 #endif 5550 5551 xic = gtk_im_multicontext_new(); 5552 g_object_ref(xic); 5553 5554 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", 5555 G_CALLBACK(&im_commit_cb), NULL); 5556 g_signal_connect(G_OBJECT(xic), "preedit_changed", 5557 G_CALLBACK(&im_preedit_changed_cb), NULL); 5558 g_signal_connect(G_OBJECT(xic), "preedit_start", 5559 G_CALLBACK(&im_preedit_start_cb), NULL); 5560 g_signal_connect(G_OBJECT(xic), "preedit_end", 5561 G_CALLBACK(&im_preedit_end_cb), NULL); 5562 5563 #if GTK_CHECK_VERSION(3,0,0) 5564 gtk_im_context_set_client_window(xic, gtk_widget_get_window(gui.drawarea)); 5565 #else 5566 gtk_im_context_set_client_window(xic, gui.drawarea->window); 5567 #endif 5568 } 5569 5570 void 5571 im_shutdown(void) 5572 { 5573 #ifdef XIM_DEBUG 5574 xim_log("im_shutdown()\n"); 5575 #endif 5576 5577 if (xic != NULL) 5578 { 5579 gtk_im_context_focus_out(xic); 5580 g_object_unref(xic); 5581 xic = NULL; 5582 } 5583 im_is_active = FALSE; 5584 im_commit_handler_id = 0; 5585 if (p_imst == IM_ON_THE_SPOT) 5586 preedit_start_col = MAXCOL; 5587 xim_has_preediting = FALSE; 5588 } 5589 5590 /* 5591 * Convert the string argument to keyval and state for GdkEventKey. 5592 * If str is valid return TRUE, otherwise FALSE. 5593 * 5594 * See 'imactivatekey' for documentation of the format. 5595 */ 5596 static int 5597 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) 5598 { 5599 const char *mods_end; 5600 unsigned tmp_keyval; 5601 unsigned tmp_state = 0; 5602 5603 mods_end = strrchr(str, '-'); 5604 mods_end = (mods_end != NULL) ? mods_end + 1 : str; 5605 5606 /* Parse modifier keys */ 5607 while (str < mods_end) 5608 switch (*str++) 5609 { 5610 case '-': break; 5611 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; 5612 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; 5613 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; 5614 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; 5615 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; 5616 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; 5617 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; 5618 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; 5619 default: 5620 return FALSE; 5621 } 5622 5623 tmp_keyval = gdk_keyval_from_name(str); 5624 5625 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) 5626 return FALSE; 5627 5628 if (keyval != NULL) 5629 *keyval = tmp_keyval; 5630 if (state != NULL) 5631 *state = tmp_state; 5632 5633 return TRUE; 5634 } 5635 5636 /* 5637 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an 5638 * empty string is also regarded as valid. 5639 * 5640 * Note: The numerical key value of p_imak is cached if it was valid; thus 5641 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever 5642 * 'imak' changes. This is currently the case but not obvious -- should 5643 * probably rename the function for clarity. 5644 */ 5645 int 5646 im_xim_isvalid_imactivate(void) 5647 { 5648 if (p_imak[0] == NUL) 5649 { 5650 im_activatekey_keyval = GDK_VoidSymbol; 5651 im_activatekey_state = 0; 5652 return TRUE; 5653 } 5654 5655 return im_string_to_keyval((const char *)p_imak, 5656 &im_activatekey_keyval, 5657 &im_activatekey_state); 5658 } 5659 5660 static void 5661 im_synthesize_keypress(unsigned int keyval, unsigned int state) 5662 { 5663 GdkEventKey *event; 5664 5665 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); 5666 # if GTK_CHECK_VERSION(3,0,0) 5667 g_object_ref(gtk_widget_get_window(gui.drawarea)); 5668 /* unreffed by gdk_event_free() */ 5669 # else 5670 g_object_ref(gui.drawarea->window); /* unreffed by gdk_event_free() */ 5671 # endif 5672 # if GTK_CHECK_VERSION(3,0,0) 5673 event->window = gtk_widget_get_window(gui.drawarea); 5674 # else 5675 event->window = gui.drawarea->window; 5676 # endif 5677 event->send_event = TRUE; 5678 event->time = GDK_CURRENT_TIME; 5679 event->state = state; 5680 event->keyval = keyval; 5681 event->hardware_keycode = /* needed for XIM */ 5682 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); 5683 event->length = 0; 5684 event->string = NULL; 5685 5686 gtk_im_context_filter_keypress(xic, event); 5687 5688 /* For consistency, also send the corresponding release event. */ 5689 event->type = GDK_KEY_RELEASE; 5690 event->send_event = FALSE; 5691 gtk_im_context_filter_keypress(xic, event); 5692 5693 gdk_event_free((GdkEvent *)event); 5694 } 5695 5696 void 5697 xim_reset(void) 5698 { 5699 #ifdef FEAT_EVAL 5700 if (USE_IMACTIVATEFUNC) 5701 call_imactivatefunc(im_is_active); 5702 else 5703 #endif 5704 if (xic != NULL) 5705 { 5706 gtk_im_context_reset(xic); 5707 5708 if (p_imdisable) 5709 im_shutdown(); 5710 else 5711 { 5712 xim_set_focus(gui.in_focus); 5713 5714 if (im_activatekey_keyval != GDK_VoidSymbol) 5715 { 5716 if (im_is_active) 5717 { 5718 g_signal_handler_block(xic, im_commit_handler_id); 5719 im_synthesize_keypress(im_activatekey_keyval, 5720 im_activatekey_state); 5721 g_signal_handler_unblock(xic, im_commit_handler_id); 5722 } 5723 } 5724 else 5725 { 5726 im_shutdown(); 5727 xim_init(); 5728 xim_set_focus(gui.in_focus); 5729 } 5730 } 5731 } 5732 5733 if (p_imst == IM_ON_THE_SPOT) 5734 preedit_start_col = MAXCOL; 5735 xim_has_preediting = FALSE; 5736 } 5737 5738 int 5739 xim_queue_key_press_event(GdkEventKey *event, int down) 5740 { 5741 if (down) 5742 { 5743 /* 5744 * Workaround GTK2 XIM 'feature' that always converts keypad keys to 5745 * chars., even when not part of an IM sequence (ref. feature of 5746 * gdk/gdkkeyuni.c). 5747 * Flag any keypad keys that might represent a single char. 5748 * If this (on its own - i.e., not part of an IM sequence) is 5749 * committed while we're processing one of these keys, we can ignore 5750 * that commit and go ahead & process it ourselves. That way we can 5751 * still distinguish keypad keys for use in mappings. 5752 * Also add GDK_space to make <S-Space> work. 5753 */ 5754 switch (event->keyval) 5755 { 5756 case GDK_KP_Add: xim_expected_char = '+'; break; 5757 case GDK_KP_Subtract: xim_expected_char = '-'; break; 5758 case GDK_KP_Divide: xim_expected_char = '/'; break; 5759 case GDK_KP_Multiply: xim_expected_char = '*'; break; 5760 case GDK_KP_Decimal: xim_expected_char = '.'; break; 5761 case GDK_KP_Equal: xim_expected_char = '='; break; 5762 case GDK_KP_0: xim_expected_char = '0'; break; 5763 case GDK_KP_1: xim_expected_char = '1'; break; 5764 case GDK_KP_2: xim_expected_char = '2'; break; 5765 case GDK_KP_3: xim_expected_char = '3'; break; 5766 case GDK_KP_4: xim_expected_char = '4'; break; 5767 case GDK_KP_5: xim_expected_char = '5'; break; 5768 case GDK_KP_6: xim_expected_char = '6'; break; 5769 case GDK_KP_7: xim_expected_char = '7'; break; 5770 case GDK_KP_8: xim_expected_char = '8'; break; 5771 case GDK_KP_9: xim_expected_char = '9'; break; 5772 case GDK_space: xim_expected_char = ' '; break; 5773 default: xim_expected_char = NUL; 5774 } 5775 xim_ignored_char = FALSE; 5776 } 5777 5778 /* 5779 * When typing fFtT, XIM may be activated. Thus it must pass 5780 * gtk_im_context_filter_keypress() in Normal mode. 5781 * And while doing :sh too. 5782 */ 5783 if (xic != NULL && !p_imdisable 5784 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) 5785 { 5786 /* 5787 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is 5788 * always aware of the current status of IM, and can even emulate 5789 * the activation key for modules that don't support one. 5790 */ 5791 if (event->keyval == im_activatekey_keyval 5792 && (event->state & im_activatekey_state) == im_activatekey_state) 5793 { 5794 unsigned int state_mask; 5795 5796 /* Require the state of the 3 most used modifiers to match exactly. 5797 * Otherwise e.g. <S-C-space> would be unusable for other purposes 5798 * if the IM activate key is <S-space>. */ 5799 state_mask = im_activatekey_state; 5800 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK 5801 | (int)GDK_MOD1_MASK); 5802 5803 if ((event->state & state_mask) != im_activatekey_state) 5804 return FALSE; 5805 5806 /* Don't send it a second time on GDK_KEY_RELEASE. */ 5807 if (event->type != GDK_KEY_PRESS) 5808 return TRUE; 5809 5810 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE)) 5811 { 5812 im_set_active(FALSE); 5813 5814 /* ":lmap" mappings exists, toggle use of mappings. */ 5815 State ^= LANGMAP; 5816 if (State & LANGMAP) 5817 { 5818 curbuf->b_p_iminsert = B_IMODE_NONE; 5819 State &= ~LANGMAP; 5820 } 5821 else 5822 { 5823 curbuf->b_p_iminsert = B_IMODE_LMAP; 5824 State |= LANGMAP; 5825 } 5826 return TRUE; 5827 } 5828 5829 return gtk_im_context_filter_keypress(xic, event); 5830 } 5831 5832 /* Don't filter events through the IM context if IM isn't active 5833 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module 5834 * not doing anything before the activation key was sent. */ 5835 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) 5836 { 5837 int imresult = gtk_im_context_filter_keypress(xic, event); 5838 5839 if (p_imst == IM_ON_THE_SPOT) 5840 { 5841 /* Some XIM send following sequence: 5842 * 1. preedited string. 5843 * 2. committed string. 5844 * 3. line changed key. 5845 * 4. preedited string. 5846 * 5. remove preedited string. 5847 * if 3, Vim can't move back the above line for 5. 5848 * thus, this part should not parse the key. */ 5849 if (!imresult && preedit_start_col != MAXCOL 5850 && event->keyval == GDK_Return) 5851 { 5852 im_synthesize_keypress(GDK_Return, 0U); 5853 return FALSE; 5854 } 5855 } 5856 5857 /* If XIM tried to commit a keypad key as a single char., 5858 * ignore it so we can use the keypad key 'raw', for mappings. */ 5859 if (xim_expected_char != NUL && xim_ignored_char) 5860 /* We had a keypad key, and XIM tried to thieve it */ 5861 return FALSE; 5862 5863 /* This is supposed to fix a problem with iBus, that space 5864 * characters don't work in input mode. */ 5865 xim_expected_char = NUL; 5866 5867 /* Normal processing */ 5868 return imresult; 5869 } 5870 } 5871 5872 return FALSE; 5873 } 5874 5875 int 5876 im_get_status(void) 5877 { 5878 # ifdef FEAT_EVAL 5879 if (USE_IMSTATUSFUNC) 5880 return call_imstatusfunc(); 5881 # endif 5882 return im_is_active; 5883 } 5884 5885 int 5886 preedit_get_status(void) 5887 { 5888 return preedit_is_active; 5889 } 5890 5891 int 5892 im_is_preediting(void) 5893 { 5894 return xim_has_preediting; 5895 } 5896 5897 # else /* !FEAT_GUI_GTK */ 5898 5899 static int xim_is_active = FALSE; /* XIM should be active in the current 5900 mode */ 5901 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */ 5902 # ifdef FEAT_GUI_X11 5903 static XIMStyle input_style; 5904 static int status_area_enabled = TRUE; 5905 # endif 5906 5907 /* 5908 * Switch using XIM on/off. This is used by the code that changes "State". 5909 * When 'imactivatefunc' is defined use that function instead. 5910 */ 5911 void 5912 im_set_active(int active_arg) 5913 { 5914 int active = active_arg; 5915 5916 /* If 'imdisable' is set, XIM is never active. */ 5917 if (p_imdisable) 5918 active = FALSE; 5919 else if (input_style & XIMPreeditPosition) 5920 /* There is a problem in switching XIM off when preediting is used, 5921 * and it is not clear how this can be solved. For now, keep XIM on 5922 * all the time, like it was done in Vim 5.8. */ 5923 active = TRUE; 5924 5925 # if defined(FEAT_EVAL) 5926 if (USE_IMACTIVATEFUNC) 5927 { 5928 if (active != im_get_status()) 5929 { 5930 call_imactivatefunc(active); 5931 xim_has_focus = active; 5932 } 5933 return; 5934 } 5935 # endif 5936 5937 if (xic == NULL) 5938 return; 5939 5940 /* Remember the active state, it is needed when Vim gets keyboard focus. */ 5941 xim_is_active = active; 5942 xim_set_preedit(); 5943 } 5944 5945 /* 5946 * Adjust using XIM for gaining or losing keyboard focus. Also called when 5947 * "xim_is_active" changes. 5948 */ 5949 void 5950 xim_set_focus(int focus) 5951 { 5952 if (xic == NULL) 5953 return; 5954 5955 /* 5956 * XIM only gets focus when the Vim window has keyboard focus and XIM has 5957 * been set active for the current mode. 5958 */ 5959 if (focus && xim_is_active) 5960 { 5961 if (!xim_has_focus) 5962 { 5963 xim_has_focus = TRUE; 5964 XSetICFocus(xic); 5965 } 5966 } 5967 else 5968 { 5969 if (xim_has_focus) 5970 { 5971 xim_has_focus = FALSE; 5972 XUnsetICFocus(xic); 5973 } 5974 } 5975 } 5976 5977 void 5978 im_set_position(int row UNUSED, int col UNUSED) 5979 { 5980 xim_set_preedit(); 5981 } 5982 5983 /* 5984 * Set the XIM to the current cursor position. 5985 */ 5986 void 5987 xim_set_preedit(void) 5988 { 5989 XVaNestedList attr_list; 5990 XRectangle spot_area; 5991 XPoint over_spot; 5992 int line_space; 5993 5994 if (xic == NULL) 5995 return; 5996 5997 xim_set_focus(TRUE); 5998 5999 if (!xim_has_focus) 6000 { 6001 /* hide XIM cursor */ 6002 over_spot.x = 0; 6003 over_spot.y = -100; /* arbitrary invisible position */ 6004 attr_list = (XVaNestedList) XVaCreateNestedList(0, 6005 XNSpotLocation, 6006 &over_spot, 6007 NULL); 6008 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); 6009 XFree(attr_list); 6010 return; 6011 } 6012 6013 if (input_style & XIMPreeditPosition) 6014 { 6015 if (xim_fg_color == INVALCOLOR) 6016 { 6017 xim_fg_color = gui.def_norm_pixel; 6018 xim_bg_color = gui.def_back_pixel; 6019 } 6020 over_spot.x = TEXT_X(gui.col); 6021 over_spot.y = TEXT_Y(gui.row); 6022 spot_area.x = 0; 6023 spot_area.y = 0; 6024 spot_area.height = gui.char_height * Rows; 6025 spot_area.width = gui.char_width * Columns; 6026 line_space = gui.char_height; 6027 attr_list = (XVaNestedList) XVaCreateNestedList(0, 6028 XNSpotLocation, &over_spot, 6029 XNForeground, (Pixel) xim_fg_color, 6030 XNBackground, (Pixel) xim_bg_color, 6031 XNArea, &spot_area, 6032 XNLineSpace, line_space, 6033 NULL); 6034 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) 6035 EMSG(_("E284: Cannot set IC values")); 6036 XFree(attr_list); 6037 } 6038 } 6039 6040 # if defined(FEAT_GUI_X11) 6041 static char e_xim[] = N_("E285: Failed to create input context"); 6042 # endif 6043 6044 # if defined(FEAT_GUI_X11) || defined(PROTO) 6045 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(SUN_SYSTEM) 6046 # define USE_X11R6_XIM 6047 # endif 6048 6049 static int xim_real_init(Window x11_window, Display *x11_display); 6050 6051 6052 # ifdef USE_X11R6_XIM 6053 static void xim_destroy_cb(XIM im, XPointer client_data, XPointer call_data); 6054 6055 static void 6056 xim_instantiate_cb( 6057 Display *display, 6058 XPointer client_data UNUSED, 6059 XPointer call_data UNUSED) 6060 { 6061 Window x11_window; 6062 Display *x11_display; 6063 6064 # ifdef XIM_DEBUG 6065 xim_log("xim_instantiate_cb()\n"); 6066 # endif 6067 6068 gui_get_x11_windis(&x11_window, &x11_display); 6069 if (display != x11_display) 6070 return; 6071 6072 xim_real_init(x11_window, x11_display); 6073 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6074 if (xic != NULL) 6075 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6076 xim_instantiate_cb, NULL); 6077 } 6078 6079 static void 6080 xim_destroy_cb( 6081 XIM im UNUSED, 6082 XPointer client_data UNUSED, 6083 XPointer call_data UNUSED) 6084 { 6085 Window x11_window; 6086 Display *x11_display; 6087 6088 # ifdef XIM_DEBUG 6089 xim_log("xim_destroy_cb()\n"); 6090 #endif 6091 gui_get_x11_windis(&x11_window, &x11_display); 6092 6093 xic = NULL; 6094 status_area_enabled = FALSE; 6095 6096 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6097 6098 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6099 xim_instantiate_cb, NULL); 6100 } 6101 # endif 6102 6103 void 6104 xim_init(void) 6105 { 6106 Window x11_window; 6107 Display *x11_display; 6108 6109 # ifdef XIM_DEBUG 6110 xim_log("xim_init()\n"); 6111 # endif 6112 6113 gui_get_x11_windis(&x11_window, &x11_display); 6114 6115 xic = NULL; 6116 6117 if (xim_real_init(x11_window, x11_display)) 6118 return; 6119 6120 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6121 6122 # ifdef USE_X11R6_XIM 6123 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6124 xim_instantiate_cb, NULL); 6125 # endif 6126 } 6127 6128 static int 6129 xim_real_init(Window x11_window, Display *x11_display) 6130 { 6131 int i; 6132 char *p, 6133 *s, 6134 *ns, 6135 *end, 6136 tmp[1024]; 6137 # define IMLEN_MAX 40 6138 char buf[IMLEN_MAX + 7]; 6139 XIM xim = NULL; 6140 XIMStyles *xim_styles; 6141 XIMStyle this_input_style = 0; 6142 Boolean found; 6143 XPoint over_spot; 6144 XVaNestedList preedit_list, status_list; 6145 6146 input_style = 0; 6147 status_area_enabled = FALSE; 6148 6149 if (xic != NULL) 6150 return FALSE; 6151 6152 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) 6153 { 6154 strcpy(tmp, gui.rsrc_input_method); 6155 for (ns = s = tmp; ns != NULL && *s != NUL;) 6156 { 6157 s = (char *)skipwhite((char_u *)s); 6158 if (*s == NUL) 6159 break; 6160 if ((ns = end = strchr(s, ',')) == NULL) 6161 end = s + strlen(s); 6162 while (isspace(((char_u *)end)[-1])) 6163 end--; 6164 *end = NUL; 6165 6166 if (strlen(s) <= IMLEN_MAX) 6167 { 6168 strcpy(buf, "@im="); 6169 strcat(buf, s); 6170 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL 6171 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) 6172 != NULL) 6173 break; 6174 } 6175 6176 s = ns + 1; 6177 } 6178 } 6179 6180 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) 6181 xim = XOpenIM(x11_display, NULL, NULL, NULL); 6182 6183 /* This is supposed to be useful to obtain characters through 6184 * XmbLookupString() without really using a XIM. */ 6185 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL 6186 && *p != NUL) 6187 xim = XOpenIM(x11_display, NULL, NULL, NULL); 6188 6189 if (xim == NULL) 6190 { 6191 /* Only give this message when verbose is set, because too many people 6192 * got this message when they didn't want to use a XIM. */ 6193 if (p_verbose > 0) 6194 { 6195 verbose_enter(); 6196 EMSG(_("E286: Failed to open input method")); 6197 verbose_leave(); 6198 } 6199 return FALSE; 6200 } 6201 6202 # ifdef USE_X11R6_XIM 6203 { 6204 XIMCallback destroy_cb; 6205 6206 destroy_cb.callback = xim_destroy_cb; 6207 destroy_cb.client_data = NULL; 6208 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) 6209 EMSG(_("E287: Warning: Could not set destroy callback to IM")); 6210 } 6211 # endif 6212 6213 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) 6214 { 6215 EMSG(_("E288: input method doesn't support any style")); 6216 XCloseIM(xim); 6217 return FALSE; 6218 } 6219 6220 found = False; 6221 strcpy(tmp, gui.rsrc_preedit_type_name); 6222 for (s = tmp; s && !found; ) 6223 { 6224 while (*s && isspace((unsigned char)*s)) 6225 s++; 6226 if (!*s) 6227 break; 6228 if ((ns = end = strchr(s, ',')) != 0) 6229 ns++; 6230 else 6231 end = s + strlen(s); 6232 while (isspace((unsigned char)*end)) 6233 end--; 6234 *end = '\0'; 6235 6236 if (!strcmp(s, "OverTheSpot")) 6237 this_input_style = (XIMPreeditPosition | XIMStatusArea); 6238 else if (!strcmp(s, "OffTheSpot")) 6239 this_input_style = (XIMPreeditArea | XIMStatusArea); 6240 else if (!strcmp(s, "Root")) 6241 this_input_style = (XIMPreeditNothing | XIMStatusNothing); 6242 6243 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 6244 { 6245 if (this_input_style == xim_styles->supported_styles[i]) 6246 { 6247 found = True; 6248 break; 6249 } 6250 } 6251 if (!found) 6252 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 6253 { 6254 if ((xim_styles->supported_styles[i] & this_input_style) 6255 == (this_input_style & ~XIMStatusArea)) 6256 { 6257 this_input_style &= ~XIMStatusArea; 6258 found = True; 6259 break; 6260 } 6261 } 6262 6263 s = ns; 6264 } 6265 XFree(xim_styles); 6266 6267 if (!found) 6268 { 6269 /* Only give this message when verbose is set, because too many people 6270 * got this message when they didn't want to use a XIM. */ 6271 if (p_verbose > 0) 6272 { 6273 verbose_enter(); 6274 EMSG(_("E289: input method doesn't support my preedit type")); 6275 verbose_leave(); 6276 } 6277 XCloseIM(xim); 6278 return FALSE; 6279 } 6280 6281 over_spot.x = TEXT_X(gui.col); 6282 over_spot.y = TEXT_Y(gui.row); 6283 input_style = this_input_style; 6284 6285 /* A crash was reported when trying to pass gui.norm_font as XNFontSet, 6286 * thus that has been removed. Hopefully the default works... */ 6287 # ifdef FEAT_XFONTSET 6288 if (gui.fontset != NOFONTSET) 6289 { 6290 preedit_list = XVaCreateNestedList(0, 6291 XNSpotLocation, &over_spot, 6292 XNForeground, (Pixel)gui.def_norm_pixel, 6293 XNBackground, (Pixel)gui.def_back_pixel, 6294 XNFontSet, (XFontSet)gui.fontset, 6295 NULL); 6296 status_list = XVaCreateNestedList(0, 6297 XNForeground, (Pixel)gui.def_norm_pixel, 6298 XNBackground, (Pixel)gui.def_back_pixel, 6299 XNFontSet, (XFontSet)gui.fontset, 6300 NULL); 6301 } 6302 else 6303 # endif 6304 { 6305 preedit_list = XVaCreateNestedList(0, 6306 XNSpotLocation, &over_spot, 6307 XNForeground, (Pixel)gui.def_norm_pixel, 6308 XNBackground, (Pixel)gui.def_back_pixel, 6309 NULL); 6310 status_list = XVaCreateNestedList(0, 6311 XNForeground, (Pixel)gui.def_norm_pixel, 6312 XNBackground, (Pixel)gui.def_back_pixel, 6313 NULL); 6314 } 6315 6316 xic = XCreateIC(xim, 6317 XNInputStyle, input_style, 6318 XNClientWindow, x11_window, 6319 XNFocusWindow, gui.wid, 6320 XNPreeditAttributes, preedit_list, 6321 XNStatusAttributes, status_list, 6322 NULL); 6323 XFree(status_list); 6324 XFree(preedit_list); 6325 if (xic != NULL) 6326 { 6327 if (input_style & XIMStatusArea) 6328 { 6329 xim_set_status_area(); 6330 status_area_enabled = TRUE; 6331 } 6332 else 6333 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6334 } 6335 else 6336 { 6337 if (!is_not_a_term()) 6338 EMSG(_(e_xim)); 6339 XCloseIM(xim); 6340 return FALSE; 6341 } 6342 6343 return TRUE; 6344 } 6345 6346 # endif /* FEAT_GUI_X11 */ 6347 6348 /* 6349 * Get IM status. When IM is on, return TRUE. Else return FALSE. 6350 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is 6351 * active, when not having focus XIM may still be active (e.g., when using a 6352 * tear-off menu item). 6353 */ 6354 int 6355 im_get_status(void) 6356 { 6357 # ifdef FEAT_EVAL 6358 if (USE_IMSTATUSFUNC) 6359 return call_imstatusfunc(); 6360 # endif 6361 return xim_has_focus; 6362 } 6363 6364 # endif /* !FEAT_GUI_GTK */ 6365 6366 # if !defined(FEAT_GUI_GTK) || defined(PROTO) 6367 /* 6368 * Set up the status area. 6369 * 6370 * This should use a separate Widget, but that seems not possible, because 6371 * preedit_area and status_area should be set to the same window as for the 6372 * text input. Unfortunately this means the status area pollutes the text 6373 * window... 6374 */ 6375 void 6376 xim_set_status_area(void) 6377 { 6378 XVaNestedList preedit_list = 0, status_list = 0, list = 0; 6379 XRectangle pre_area, status_area; 6380 6381 if (xic == NULL) 6382 return; 6383 6384 if (input_style & XIMStatusArea) 6385 { 6386 if (input_style & XIMPreeditArea) 6387 { 6388 XRectangle *needed_rect; 6389 6390 /* to get status_area width */ 6391 status_list = XVaCreateNestedList(0, XNAreaNeeded, 6392 &needed_rect, NULL); 6393 XGetICValues(xic, XNStatusAttributes, status_list, NULL); 6394 XFree(status_list); 6395 6396 status_area.width = needed_rect->width; 6397 } 6398 else 6399 status_area.width = gui.char_width * Columns; 6400 6401 status_area.x = 0; 6402 status_area.y = gui.char_height * Rows + gui.border_offset; 6403 if (gui.which_scrollbars[SBAR_BOTTOM]) 6404 status_area.y += gui.scrollbar_height; 6405 #ifdef FEAT_MENU 6406 if (gui.menu_is_active) 6407 status_area.y += gui.menu_height; 6408 #endif 6409 status_area.height = gui.char_height; 6410 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); 6411 } 6412 else 6413 { 6414 status_area.x = 0; 6415 status_area.y = gui.char_height * Rows + gui.border_offset; 6416 if (gui.which_scrollbars[SBAR_BOTTOM]) 6417 status_area.y += gui.scrollbar_height; 6418 #ifdef FEAT_MENU 6419 if (gui.menu_is_active) 6420 status_area.y += gui.menu_height; 6421 #endif 6422 status_area.width = 0; 6423 status_area.height = gui.char_height; 6424 } 6425 6426 if (input_style & XIMPreeditArea) /* off-the-spot */ 6427 { 6428 pre_area.x = status_area.x + status_area.width; 6429 pre_area.y = gui.char_height * Rows + gui.border_offset; 6430 pre_area.width = gui.char_width * Columns - pre_area.x; 6431 if (gui.which_scrollbars[SBAR_BOTTOM]) 6432 pre_area.y += gui.scrollbar_height; 6433 #ifdef FEAT_MENU 6434 if (gui.menu_is_active) 6435 pre_area.y += gui.menu_height; 6436 #endif 6437 pre_area.height = gui.char_height; 6438 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6439 } 6440 else if (input_style & XIMPreeditPosition) /* over-the-spot */ 6441 { 6442 pre_area.x = 0; 6443 pre_area.y = 0; 6444 pre_area.height = gui.char_height * Rows; 6445 pre_area.width = gui.char_width * Columns; 6446 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6447 } 6448 6449 if (preedit_list && status_list) 6450 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6451 XNStatusAttributes, status_list, NULL); 6452 else if (preedit_list) 6453 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6454 NULL); 6455 else if (status_list) 6456 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, 6457 NULL); 6458 else 6459 list = NULL; 6460 6461 if (list) 6462 { 6463 XSetICValues(xic, XNVaNestedList, list, NULL); 6464 XFree(list); 6465 } 6466 if (status_list) 6467 XFree(status_list); 6468 if (preedit_list) 6469 XFree(preedit_list); 6470 } 6471 6472 int 6473 xim_get_status_area_height(void) 6474 { 6475 if (status_area_enabled) 6476 return gui.char_height; 6477 return 0; 6478 } 6479 # endif 6480 6481 #else /* !defined(FEAT_XIM) */ 6482 6483 # if !defined(FEAT_GUI_W32) || !(defined(FEAT_MBYTE_IME) || defined(GLOBAL_IME)) 6484 static int im_was_set_active = FALSE; 6485 6486 int 6487 im_get_status() 6488 { 6489 # if defined(FEAT_MBYTE) && defined(FEAT_EVAL) 6490 if (USE_IMSTATUSFUNC) 6491 return call_imstatusfunc(); 6492 # endif 6493 return im_was_set_active; 6494 } 6495 6496 void 6497 im_set_active(int active_arg) 6498 { 6499 # if defined(FEAT_MBYTE) && defined(FEAT_EVAL) 6500 int active = !p_imdisable && active_arg; 6501 6502 if (USE_IMACTIVATEFUNC && active != im_get_status()) 6503 { 6504 call_imactivatefunc(active); 6505 im_was_set_active = active; 6506 } 6507 # endif 6508 } 6509 6510 # ifdef FEAT_GUI 6511 void 6512 im_set_position(int row, int col) 6513 { 6514 } 6515 # endif 6516 # endif 6517 6518 #endif /* FEAT_XIM */ 6519 6520 #if defined(FEAT_MBYTE) || defined(PROTO) 6521 6522 /* 6523 * Setup "vcp" for conversion from "from" to "to". 6524 * The names must have been made canonical with enc_canonize(). 6525 * vcp->vc_type must have been initialized to CONV_NONE. 6526 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 6527 * instead). 6528 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 6529 * Return FAIL when conversion is not supported, OK otherwise. 6530 */ 6531 int 6532 convert_setup(vimconv_T *vcp, char_u *from, char_u *to) 6533 { 6534 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 6535 } 6536 6537 /* 6538 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 6539 * "from" unicode charsets be considered utf-8. Same for "to". 6540 */ 6541 int 6542 convert_setup_ext( 6543 vimconv_T *vcp, 6544 char_u *from, 6545 int from_unicode_is_utf8, 6546 char_u *to, 6547 int to_unicode_is_utf8) 6548 { 6549 int from_prop; 6550 int to_prop; 6551 int from_is_utf8; 6552 int to_is_utf8; 6553 6554 /* Reset to no conversion. */ 6555 # ifdef USE_ICONV 6556 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 6557 iconv_close(vcp->vc_fd); 6558 # endif 6559 vcp->vc_type = CONV_NONE; 6560 vcp->vc_factor = 1; 6561 vcp->vc_fail = FALSE; 6562 6563 /* No conversion when one of the names is empty or they are equal. */ 6564 if (from == NULL || *from == NUL || to == NULL || *to == NUL 6565 || STRCMP(from, to) == 0) 6566 return OK; 6567 6568 from_prop = enc_canon_props(from); 6569 to_prop = enc_canon_props(to); 6570 if (from_unicode_is_utf8) 6571 from_is_utf8 = from_prop & ENC_UNICODE; 6572 else 6573 from_is_utf8 = from_prop == ENC_UNICODE; 6574 if (to_unicode_is_utf8) 6575 to_is_utf8 = to_prop & ENC_UNICODE; 6576 else 6577 to_is_utf8 = to_prop == ENC_UNICODE; 6578 6579 if ((from_prop & ENC_LATIN1) && to_is_utf8) 6580 { 6581 /* Internal latin1 -> utf-8 conversion. */ 6582 vcp->vc_type = CONV_TO_UTF8; 6583 vcp->vc_factor = 2; /* up to twice as long */ 6584 } 6585 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 6586 { 6587 /* Internal latin9 -> utf-8 conversion. */ 6588 vcp->vc_type = CONV_9_TO_UTF8; 6589 vcp->vc_factor = 3; /* up to three as long (euro sign) */ 6590 } 6591 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 6592 { 6593 /* Internal utf-8 -> latin1 conversion. */ 6594 vcp->vc_type = CONV_TO_LATIN1; 6595 } 6596 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 6597 { 6598 /* Internal utf-8 -> latin9 conversion. */ 6599 vcp->vc_type = CONV_TO_LATIN9; 6600 } 6601 #ifdef WIN3264 6602 /* Win32-specific codepage <-> codepage conversion without iconv. */ 6603 else if ((from_is_utf8 || encname2codepage(from) > 0) 6604 && (to_is_utf8 || encname2codepage(to) > 0)) 6605 { 6606 vcp->vc_type = CONV_CODEPAGE; 6607 vcp->vc_factor = 2; /* up to twice as long */ 6608 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 6609 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 6610 } 6611 #endif 6612 #ifdef MACOS_CONVERT 6613 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 6614 { 6615 vcp->vc_type = CONV_MAC_LATIN1; 6616 } 6617 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 6618 { 6619 vcp->vc_type = CONV_MAC_UTF8; 6620 vcp->vc_factor = 2; /* up to twice as long */ 6621 } 6622 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 6623 { 6624 vcp->vc_type = CONV_LATIN1_MAC; 6625 } 6626 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 6627 { 6628 vcp->vc_type = CONV_UTF8_MAC; 6629 } 6630 #endif 6631 # ifdef USE_ICONV 6632 else 6633 { 6634 /* Use iconv() for conversion. */ 6635 vcp->vc_fd = (iconv_t)my_iconv_open( 6636 to_is_utf8 ? (char_u *)"utf-8" : to, 6637 from_is_utf8 ? (char_u *)"utf-8" : from); 6638 if (vcp->vc_fd != (iconv_t)-1) 6639 { 6640 vcp->vc_type = CONV_ICONV; 6641 vcp->vc_factor = 4; /* could be longer too... */ 6642 } 6643 } 6644 # endif 6645 if (vcp->vc_type == CONV_NONE) 6646 return FAIL; 6647 6648 return OK; 6649 } 6650 6651 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \ 6652 || defined(PROTO) 6653 /* 6654 * Do conversion on typed input characters in-place. 6655 * The input and output are not NUL terminated! 6656 * Returns the length after conversion. 6657 */ 6658 int 6659 convert_input(char_u *ptr, int len, int maxlen) 6660 { 6661 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 6662 } 6663 #endif 6664 6665 /* 6666 * Like convert_input(), but when there is an incomplete byte sequence at the 6667 * end return that as an allocated string in "restp" and set "*restlenp" to 6668 * the length. If "restp" is NULL it is not used. 6669 */ 6670 int 6671 convert_input_safe( 6672 char_u *ptr, 6673 int len, 6674 int maxlen, 6675 char_u **restp, 6676 int *restlenp) 6677 { 6678 char_u *d; 6679 int dlen = len; 6680 int unconvertlen = 0; 6681 6682 d = string_convert_ext(&input_conv, ptr, &dlen, 6683 restp == NULL ? NULL : &unconvertlen); 6684 if (d != NULL) 6685 { 6686 if (dlen <= maxlen) 6687 { 6688 if (unconvertlen > 0) 6689 { 6690 /* Move the unconverted characters to allocated memory. */ 6691 *restp = alloc(unconvertlen); 6692 if (*restp != NULL) 6693 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 6694 *restlenp = unconvertlen; 6695 } 6696 mch_memmove(ptr, d, dlen); 6697 } 6698 else 6699 /* result is too long, keep the unconverted text (the caller must 6700 * have done something wrong!) */ 6701 dlen = len; 6702 vim_free(d); 6703 } 6704 return dlen; 6705 } 6706 6707 /* 6708 * Convert text "ptr[*lenp]" according to "vcp". 6709 * Returns the result in allocated memory and sets "*lenp". 6710 * When "lenp" is NULL, use NUL terminated strings. 6711 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 6712 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 6713 */ 6714 char_u * 6715 string_convert( 6716 vimconv_T *vcp, 6717 char_u *ptr, 6718 int *lenp) 6719 { 6720 return string_convert_ext(vcp, ptr, lenp, NULL); 6721 } 6722 6723 /* 6724 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 6725 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 6726 * set to the number of remaining bytes. 6727 */ 6728 char_u * 6729 string_convert_ext( 6730 vimconv_T *vcp, 6731 char_u *ptr, 6732 int *lenp, 6733 int *unconvlenp) 6734 { 6735 char_u *retval = NULL; 6736 char_u *d; 6737 int len; 6738 int i; 6739 int l; 6740 int c; 6741 6742 if (lenp == NULL) 6743 len = (int)STRLEN(ptr); 6744 else 6745 len = *lenp; 6746 if (len == 0) 6747 return vim_strsave((char_u *)""); 6748 6749 switch (vcp->vc_type) 6750 { 6751 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */ 6752 retval = alloc(len * 2 + 1); 6753 if (retval == NULL) 6754 break; 6755 d = retval; 6756 for (i = 0; i < len; ++i) 6757 { 6758 c = ptr[i]; 6759 if (c < 0x80) 6760 *d++ = c; 6761 else 6762 { 6763 *d++ = 0xc0 + ((unsigned)c >> 6); 6764 *d++ = 0x80 + (c & 0x3f); 6765 } 6766 } 6767 *d = NUL; 6768 if (lenp != NULL) 6769 *lenp = (int)(d - retval); 6770 break; 6771 6772 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */ 6773 retval = alloc(len * 3 + 1); 6774 if (retval == NULL) 6775 break; 6776 d = retval; 6777 for (i = 0; i < len; ++i) 6778 { 6779 c = ptr[i]; 6780 switch (c) 6781 { 6782 case 0xa4: c = 0x20ac; break; /* euro */ 6783 case 0xa6: c = 0x0160; break; /* S hat */ 6784 case 0xa8: c = 0x0161; break; /* S -hat */ 6785 case 0xb4: c = 0x017d; break; /* Z hat */ 6786 case 0xb8: c = 0x017e; break; /* Z -hat */ 6787 case 0xbc: c = 0x0152; break; /* OE */ 6788 case 0xbd: c = 0x0153; break; /* oe */ 6789 case 0xbe: c = 0x0178; break; /* Y */ 6790 } 6791 d += utf_char2bytes(c, d); 6792 } 6793 *d = NUL; 6794 if (lenp != NULL) 6795 *lenp = (int)(d - retval); 6796 break; 6797 6798 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */ 6799 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */ 6800 retval = alloc(len + 1); 6801 if (retval == NULL) 6802 break; 6803 d = retval; 6804 for (i = 0; i < len; ++i) 6805 { 6806 l = utf_ptr2len_len(ptr + i, len - i); 6807 if (l == 0) 6808 *d++ = NUL; 6809 else if (l == 1) 6810 { 6811 int l_w = utf8len_tab_zero[ptr[i]]; 6812 6813 if (l_w == 0) 6814 { 6815 /* Illegal utf-8 byte cannot be converted */ 6816 vim_free(retval); 6817 return NULL; 6818 } 6819 if (unconvlenp != NULL && l_w > len - i) 6820 { 6821 /* Incomplete sequence at the end. */ 6822 *unconvlenp = len - i; 6823 break; 6824 } 6825 *d++ = ptr[i]; 6826 } 6827 else 6828 { 6829 c = utf_ptr2char(ptr + i); 6830 if (vcp->vc_type == CONV_TO_LATIN9) 6831 switch (c) 6832 { 6833 case 0x20ac: c = 0xa4; break; /* euro */ 6834 case 0x0160: c = 0xa6; break; /* S hat */ 6835 case 0x0161: c = 0xa8; break; /* S -hat */ 6836 case 0x017d: c = 0xb4; break; /* Z hat */ 6837 case 0x017e: c = 0xb8; break; /* Z -hat */ 6838 case 0x0152: c = 0xbc; break; /* OE */ 6839 case 0x0153: c = 0xbd; break; /* oe */ 6840 case 0x0178: c = 0xbe; break; /* Y */ 6841 case 0xa4: 6842 case 0xa6: 6843 case 0xa8: 6844 case 0xb4: 6845 case 0xb8: 6846 case 0xbc: 6847 case 0xbd: 6848 case 0xbe: c = 0x100; break; /* not in latin9 */ 6849 } 6850 if (!utf_iscomposing(c)) /* skip composing chars */ 6851 { 6852 if (c < 0x100) 6853 *d++ = c; 6854 else if (vcp->vc_fail) 6855 { 6856 vim_free(retval); 6857 return NULL; 6858 } 6859 else 6860 { 6861 *d++ = 0xbf; 6862 if (utf_char2cells(c) > 1) 6863 *d++ = '?'; 6864 } 6865 } 6866 i += l - 1; 6867 } 6868 } 6869 *d = NUL; 6870 if (lenp != NULL) 6871 *lenp = (int)(d - retval); 6872 break; 6873 6874 # ifdef MACOS_CONVERT 6875 case CONV_MAC_LATIN1: 6876 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6877 'm', 'l', unconvlenp); 6878 break; 6879 6880 case CONV_LATIN1_MAC: 6881 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6882 'l', 'm', unconvlenp); 6883 break; 6884 6885 case CONV_MAC_UTF8: 6886 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6887 'm', 'u', unconvlenp); 6888 break; 6889 6890 case CONV_UTF8_MAC: 6891 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6892 'u', 'm', unconvlenp); 6893 break; 6894 # endif 6895 6896 # ifdef USE_ICONV 6897 case CONV_ICONV: /* conversion with output_conv.vc_fd */ 6898 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 6899 break; 6900 # endif 6901 # ifdef WIN3264 6902 case CONV_CODEPAGE: /* codepage -> codepage */ 6903 { 6904 int retlen; 6905 int tmp_len; 6906 short_u *tmp; 6907 6908 /* 1. codepage/UTF-8 -> ucs-2. */ 6909 if (vcp->vc_cpfrom == 0) 6910 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 6911 else 6912 { 6913 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 6914 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 6915 (char *)ptr, len, 0, 0); 6916 if (tmp_len == 0 6917 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 6918 { 6919 if (lenp != NULL) 6920 *lenp = 0; 6921 if (unconvlenp != NULL) 6922 *unconvlenp = len; 6923 retval = alloc(1); 6924 if (retval) 6925 retval[0] = NUL; 6926 return retval; 6927 } 6928 } 6929 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len); 6930 if (tmp == NULL) 6931 break; 6932 if (vcp->vc_cpfrom == 0) 6933 utf8_to_utf16(ptr, len, tmp, unconvlenp); 6934 else 6935 MultiByteToWideChar(vcp->vc_cpfrom, 0, 6936 (char *)ptr, len, tmp, tmp_len); 6937 6938 /* 2. ucs-2 -> codepage/UTF-8. */ 6939 if (vcp->vc_cpto == 0) 6940 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 6941 else 6942 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 6943 tmp, tmp_len, 0, 0, 0, 0); 6944 retval = alloc(retlen + 1); 6945 if (retval != NULL) 6946 { 6947 if (vcp->vc_cpto == 0) 6948 utf16_to_utf8(tmp, tmp_len, retval); 6949 else 6950 WideCharToMultiByte(vcp->vc_cpto, 0, 6951 tmp, tmp_len, 6952 (char *)retval, retlen, 0, 0); 6953 retval[retlen] = NUL; 6954 if (lenp != NULL) 6955 *lenp = retlen; 6956 } 6957 vim_free(tmp); 6958 break; 6959 } 6960 # endif 6961 } 6962 6963 return retval; 6964 } 6965 #endif 6966