1 /* vi:set ts=8 sts=4 sw=4 noet: 2 * 3 * VIM - Vi IMproved by Bram Moolenaar 4 * Multibyte extensions partly by Sung-Hoon Baek 5 * 6 * Do ":help uganda" in Vim to read copying and usage conditions. 7 * Do ":help credits" in Vim to see a list of people who contributed. 8 * See README.txt for an overview of the Vim source code. 9 */ 10 /* 11 * mbyte.c: Code specifically for handling multi-byte characters. 12 * 13 * The encoding used in the core is set with 'encoding'. When 'encoding' is 14 * changed, the following four variables are set (for speed). 15 * Currently these types of character encodings are supported: 16 * 17 * "enc_dbcs" When non-zero it tells the type of double byte character 18 * encoding (Chinese, Korean, Japanese, etc.). 19 * The cell width on the display is equal to the number of 20 * bytes. (exception: DBCS_JPNU with first byte 0x8e) 21 * Recognizing the first or second byte is difficult, it 22 * requires checking a byte sequence from the start. 23 * "enc_utf8" When TRUE use Unicode characters in UTF-8 encoding. 24 * The cell width on the display needs to be determined from 25 * the character value. 26 * Recognizing bytes is easy: 0xxx.xxxx is a single-byte 27 * char, 10xx.xxxx is a trailing byte, 11xx.xxxx is a leading 28 * byte of a multi-byte character. 29 * To make things complicated, up to six composing characters 30 * are allowed. These are drawn on top of the first char. 31 * For most editing the sequence of bytes with composing 32 * characters included is considered to be one character. 33 * "enc_unicode" When 2 use 16-bit Unicode characters (or UTF-16). 34 * When 4 use 32-but Unicode characters. 35 * Internally characters are stored in UTF-8 encoding to 36 * avoid NUL bytes. Conversion happens when doing I/O. 37 * "enc_utf8" will also be TRUE. 38 * 39 * "has_mbyte" is set when "enc_dbcs" or "enc_utf8" is non-zero. 40 * 41 * If none of these is TRUE, 8-bit bytes are used for a character. The 42 * encoding isn't currently specified (TODO). 43 * 44 * 'encoding' specifies the encoding used in the core. This is in registers, 45 * text manipulation, buffers, etc. Conversion has to be done when characters 46 * in another encoding are received or send: 47 * 48 * clipboard 49 * ^ 50 * | (2) 51 * V 52 * +---------------+ 53 * (1) | | (3) 54 * keyboard ----->| core |-----> display 55 * | | 56 * +---------------+ 57 * ^ 58 * | (4) 59 * V 60 * file 61 * 62 * (1) Typed characters arrive in the current locale. Conversion is to be 63 * done when 'encoding' is different from 'termencoding'. 64 * (2) Text will be made available with the encoding specified with 65 * 'encoding'. If this is not sufficient, system-specific conversion 66 * might be required. 67 * (3) For the GUI the correct font must be selected, no conversion done. 68 * Otherwise, conversion is to be done when 'encoding' differs from 69 * 'termencoding'. (Different in the GTK+ 2 port -- 'termencoding' 70 * is always used for both input and output and must always be set to 71 * "utf-8". gui_mch_init() does this automatically.) 72 * (4) The encoding of the file is specified with 'fileencoding'. Conversion 73 * is to be done when it's different from 'encoding'. 74 * 75 * The viminfo file is a special case: Only text is converted, not file names. 76 * Vim scripts may contain an ":encoding" command. This has an effect for 77 * some commands, like ":menutrans" 78 */ 79 80 #include "vim.h" 81 82 #ifdef WIN32UNIX 83 # ifndef WIN32_LEAN_AND_MEAN 84 # define WIN32_LEAN_AND_MEAN 85 # endif 86 # if defined(FEAT_GUI) || defined(FEAT_XCLIPBOARD) 87 # include <X11/Xwindows.h> 88 # define WINBYTE wBYTE 89 # else 90 # include <windows.h> 91 # define WINBYTE BYTE 92 # endif 93 # ifdef WIN32 94 # undef WIN32 /* Some windows.h define WIN32, we don't want that here. */ 95 # endif 96 #else 97 # define WINBYTE BYTE 98 #endif 99 100 #if (defined(WIN3264) || defined(WIN32UNIX)) && !defined(__MINGW32__) 101 # include <winnls.h> 102 #endif 103 104 #ifdef FEAT_GUI_X11 105 # include <X11/Intrinsic.h> 106 #endif 107 #ifdef X_LOCALE 108 # include <X11/Xlocale.h> 109 # if !defined(HAVE_MBLEN) && !defined(mblen) 110 # define mblen _Xmblen 111 # endif 112 #endif 113 114 #if defined(FEAT_GUI_GTK) && defined(FEAT_XIM) 115 # if GTK_CHECK_VERSION(3,0,0) 116 # include <gdk/gdkkeysyms-compat.h> 117 # else 118 # include <gdk/gdkkeysyms.h> 119 # endif 120 # ifdef WIN3264 121 # include <gdk/gdkwin32.h> 122 # else 123 # include <gdk/gdkx.h> 124 # endif 125 #endif 126 127 #ifdef HAVE_WCHAR_H 128 # include <wchar.h> 129 #endif 130 131 #if 0 132 /* This has been disabled, because several people reported problems with the 133 * wcwidth() and iswprint() library functions, esp. for Hebrew. */ 134 # ifdef __STDC_ISO_10646__ 135 # define USE_WCHAR_FUNCTIONS 136 # endif 137 #endif 138 139 #if defined(FEAT_MBYTE) || defined(PROTO) 140 141 static int enc_canon_search(char_u *name); 142 static int dbcs_char2len(int c); 143 static int dbcs_char2bytes(int c, char_u *buf); 144 static int dbcs_ptr2len(char_u *p); 145 static int dbcs_ptr2len_len(char_u *p, int size); 146 static int utf_ptr2cells_len(char_u *p, int size); 147 static int dbcs_char2cells(int c); 148 static int dbcs_ptr2cells_len(char_u *p, int size); 149 static int dbcs_ptr2char(char_u *p); 150 static int utf_safe_read_char_adv(char_u **s, size_t *n); 151 152 /* 153 * Lookup table to quickly get the length in bytes of a UTF-8 character from 154 * the first byte of a UTF-8 string. 155 * Bytes which are illegal when used as the first byte have a 1. 156 * The NUL byte has length 1. 157 */ 158 static char utf8len_tab[256] = 159 { 160 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 161 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 162 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 163 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 164 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 165 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 166 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, 167 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6,1,1, 168 }; 169 170 /* 171 * Like utf8len_tab above, but using a zero for illegal lead bytes. 172 */ 173 static char utf8len_tab_zero[256] = 174 { 175 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 176 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 177 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 178 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 179 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 180 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 181 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, 182 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6,0,0, 183 }; 184 185 /* 186 * XIM often causes trouble. Define XIM_DEBUG to get a log of XIM callbacks 187 * in the "xim.log" file. 188 */ 189 /* #define XIM_DEBUG */ 190 #ifdef XIM_DEBUG 191 static void 192 xim_log(char *s, ...) 193 { 194 va_list arglist; 195 static FILE *fd = NULL; 196 197 if (fd == (FILE *)-1) 198 return; 199 if (fd == NULL) 200 { 201 fd = mch_fopen("xim.log", "w"); 202 if (fd == NULL) 203 { 204 EMSG("Cannot open xim.log"); 205 fd = (FILE *)-1; 206 return; 207 } 208 } 209 210 va_start(arglist, s); 211 vfprintf(fd, s, arglist); 212 va_end(arglist); 213 } 214 #endif 215 216 #endif 217 218 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 219 /* 220 * Canonical encoding names and their properties. 221 * "iso-8859-n" is handled by enc_canonize() directly. 222 */ 223 static struct 224 { char *name; int prop; int codepage;} 225 enc_canon_table[] = 226 { 227 #define IDX_LATIN_1 0 228 {"latin1", ENC_8BIT + ENC_LATIN1, 1252}, 229 #define IDX_ISO_2 1 230 {"iso-8859-2", ENC_8BIT, 0}, 231 #define IDX_ISO_3 2 232 {"iso-8859-3", ENC_8BIT, 0}, 233 #define IDX_ISO_4 3 234 {"iso-8859-4", ENC_8BIT, 0}, 235 #define IDX_ISO_5 4 236 {"iso-8859-5", ENC_8BIT, 0}, 237 #define IDX_ISO_6 5 238 {"iso-8859-6", ENC_8BIT, 0}, 239 #define IDX_ISO_7 6 240 {"iso-8859-7", ENC_8BIT, 0}, 241 #define IDX_ISO_8 7 242 {"iso-8859-8", ENC_8BIT, 0}, 243 #define IDX_ISO_9 8 244 {"iso-8859-9", ENC_8BIT, 0}, 245 #define IDX_ISO_10 9 246 {"iso-8859-10", ENC_8BIT, 0}, 247 #define IDX_ISO_11 10 248 {"iso-8859-11", ENC_8BIT, 0}, 249 #define IDX_ISO_13 11 250 {"iso-8859-13", ENC_8BIT, 0}, 251 #define IDX_ISO_14 12 252 {"iso-8859-14", ENC_8BIT, 0}, 253 #define IDX_ISO_15 13 254 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0}, 255 #define IDX_KOI8_R 14 256 {"koi8-r", ENC_8BIT, 0}, 257 #define IDX_KOI8_U 15 258 {"koi8-u", ENC_8BIT, 0}, 259 #define IDX_UTF8 16 260 {"utf-8", ENC_UNICODE, 0}, 261 #define IDX_UCS2 17 262 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0}, 263 #define IDX_UCS2LE 18 264 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0}, 265 #define IDX_UTF16 19 266 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0}, 267 #define IDX_UTF16LE 20 268 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0}, 269 #define IDX_UCS4 21 270 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0}, 271 #define IDX_UCS4LE 22 272 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0}, 273 274 /* For debugging DBCS encoding on Unix. */ 275 #define IDX_DEBUG 23 276 {"debug", ENC_DBCS, DBCS_DEBUG}, 277 #define IDX_EUC_JP 24 278 {"euc-jp", ENC_DBCS, DBCS_JPNU}, 279 #define IDX_SJIS 25 280 {"sjis", ENC_DBCS, DBCS_JPN}, 281 #define IDX_EUC_KR 26 282 {"euc-kr", ENC_DBCS, DBCS_KORU}, 283 #define IDX_EUC_CN 27 284 {"euc-cn", ENC_DBCS, DBCS_CHSU}, 285 #define IDX_EUC_TW 28 286 {"euc-tw", ENC_DBCS, DBCS_CHTU}, 287 #define IDX_BIG5 29 288 {"big5", ENC_DBCS, DBCS_CHT}, 289 290 /* MS-DOS and MS-Windows codepages are included here, so that they can be 291 * used on Unix too. Most of them are similar to ISO-8859 encodings, but 292 * not exactly the same. */ 293 #define IDX_CP437 30 294 {"cp437", ENC_8BIT, 437}, /* like iso-8859-1 */ 295 #define IDX_CP737 31 296 {"cp737", ENC_8BIT, 737}, /* like iso-8859-7 */ 297 #define IDX_CP775 32 298 {"cp775", ENC_8BIT, 775}, /* Baltic */ 299 #define IDX_CP850 33 300 {"cp850", ENC_8BIT, 850}, /* like iso-8859-4 */ 301 #define IDX_CP852 34 302 {"cp852", ENC_8BIT, 852}, /* like iso-8859-1 */ 303 #define IDX_CP855 35 304 {"cp855", ENC_8BIT, 855}, /* like iso-8859-2 */ 305 #define IDX_CP857 36 306 {"cp857", ENC_8BIT, 857}, /* like iso-8859-5 */ 307 #define IDX_CP860 37 308 {"cp860", ENC_8BIT, 860}, /* like iso-8859-9 */ 309 #define IDX_CP861 38 310 {"cp861", ENC_8BIT, 861}, /* like iso-8859-1 */ 311 #define IDX_CP862 39 312 {"cp862", ENC_8BIT, 862}, /* like iso-8859-1 */ 313 #define IDX_CP863 40 314 {"cp863", ENC_8BIT, 863}, /* like iso-8859-8 */ 315 #define IDX_CP865 41 316 {"cp865", ENC_8BIT, 865}, /* like iso-8859-1 */ 317 #define IDX_CP866 42 318 {"cp866", ENC_8BIT, 866}, /* like iso-8859-5 */ 319 #define IDX_CP869 43 320 {"cp869", ENC_8BIT, 869}, /* like iso-8859-7 */ 321 #define IDX_CP874 44 322 {"cp874", ENC_8BIT, 874}, /* Thai */ 323 #define IDX_CP932 45 324 {"cp932", ENC_DBCS, DBCS_JPN}, 325 #define IDX_CP936 46 326 {"cp936", ENC_DBCS, DBCS_CHS}, 327 #define IDX_CP949 47 328 {"cp949", ENC_DBCS, DBCS_KOR}, 329 #define IDX_CP950 48 330 {"cp950", ENC_DBCS, DBCS_CHT}, 331 #define IDX_CP1250 49 332 {"cp1250", ENC_8BIT, 1250}, /* Czech, Polish, etc. */ 333 #define IDX_CP1251 50 334 {"cp1251", ENC_8BIT, 1251}, /* Cyrillic */ 335 /* cp1252 is considered to be equal to latin1 */ 336 #define IDX_CP1253 51 337 {"cp1253", ENC_8BIT, 1253}, /* Greek */ 338 #define IDX_CP1254 52 339 {"cp1254", ENC_8BIT, 1254}, /* Turkish */ 340 #define IDX_CP1255 53 341 {"cp1255", ENC_8BIT, 1255}, /* Hebrew */ 342 #define IDX_CP1256 54 343 {"cp1256", ENC_8BIT, 1256}, /* Arabic */ 344 #define IDX_CP1257 55 345 {"cp1257", ENC_8BIT, 1257}, /* Baltic */ 346 #define IDX_CP1258 56 347 {"cp1258", ENC_8BIT, 1258}, /* Vietnamese */ 348 349 #define IDX_MACROMAN 57 350 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, /* Mac OS */ 351 #define IDX_DECMCS 58 352 {"dec-mcs", ENC_8BIT, 0}, /* DEC MCS */ 353 #define IDX_HPROMAN8 59 354 {"hp-roman8", ENC_8BIT, 0}, /* HP Roman8 */ 355 #define IDX_COUNT 60 356 }; 357 358 /* 359 * Aliases for encoding names. 360 */ 361 static struct 362 { char *name; int canon;} 363 enc_alias_table[] = 364 { 365 {"ansi", IDX_LATIN_1}, 366 {"iso-8859-1", IDX_LATIN_1}, 367 {"latin2", IDX_ISO_2}, 368 {"latin3", IDX_ISO_3}, 369 {"latin4", IDX_ISO_4}, 370 {"cyrillic", IDX_ISO_5}, 371 {"arabic", IDX_ISO_6}, 372 {"greek", IDX_ISO_7}, 373 #ifdef WIN3264 374 {"hebrew", IDX_CP1255}, 375 #else 376 {"hebrew", IDX_ISO_8}, 377 #endif 378 {"latin5", IDX_ISO_9}, 379 {"turkish", IDX_ISO_9}, /* ? */ 380 {"latin6", IDX_ISO_10}, 381 {"nordic", IDX_ISO_10}, /* ? */ 382 {"thai", IDX_ISO_11}, /* ? */ 383 {"latin7", IDX_ISO_13}, 384 {"latin8", IDX_ISO_14}, 385 {"latin9", IDX_ISO_15}, 386 {"utf8", IDX_UTF8}, 387 {"unicode", IDX_UCS2}, 388 {"ucs2", IDX_UCS2}, 389 {"ucs2be", IDX_UCS2}, 390 {"ucs-2be", IDX_UCS2}, 391 {"ucs2le", IDX_UCS2LE}, 392 {"utf16", IDX_UTF16}, 393 {"utf16be", IDX_UTF16}, 394 {"utf-16be", IDX_UTF16}, 395 {"utf16le", IDX_UTF16LE}, 396 {"ucs4", IDX_UCS4}, 397 {"ucs4be", IDX_UCS4}, 398 {"ucs-4be", IDX_UCS4}, 399 {"ucs4le", IDX_UCS4LE}, 400 {"utf32", IDX_UCS4}, 401 {"utf-32", IDX_UCS4}, 402 {"utf32be", IDX_UCS4}, 403 {"utf-32be", IDX_UCS4}, 404 {"utf32le", IDX_UCS4LE}, 405 {"utf-32le", IDX_UCS4LE}, 406 {"932", IDX_CP932}, 407 {"949", IDX_CP949}, 408 {"936", IDX_CP936}, 409 {"gbk", IDX_CP936}, 410 {"950", IDX_CP950}, 411 {"eucjp", IDX_EUC_JP}, 412 {"unix-jis", IDX_EUC_JP}, 413 {"ujis", IDX_EUC_JP}, 414 {"shift-jis", IDX_SJIS}, 415 {"pck", IDX_SJIS}, /* Sun: PCK */ 416 {"euckr", IDX_EUC_KR}, 417 {"5601", IDX_EUC_KR}, /* Sun: KS C 5601 */ 418 {"euccn", IDX_EUC_CN}, 419 {"gb2312", IDX_EUC_CN}, 420 {"euctw", IDX_EUC_TW}, 421 #if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS_X) 422 {"japan", IDX_CP932}, 423 {"korea", IDX_CP949}, 424 {"prc", IDX_CP936}, 425 {"chinese", IDX_CP936}, 426 {"taiwan", IDX_CP950}, 427 {"big5", IDX_CP950}, 428 #else 429 {"japan", IDX_EUC_JP}, 430 {"korea", IDX_EUC_KR}, 431 {"prc", IDX_EUC_CN}, 432 {"chinese", IDX_EUC_CN}, 433 {"taiwan", IDX_EUC_TW}, 434 {"cp950", IDX_BIG5}, 435 {"950", IDX_BIG5}, 436 #endif 437 {"mac", IDX_MACROMAN}, 438 {"mac-roman", IDX_MACROMAN}, 439 {NULL, 0} 440 }; 441 442 #ifndef CP_UTF8 443 # define CP_UTF8 65001 /* magic number from winnls.h */ 444 #endif 445 446 /* 447 * Find encoding "name" in the list of canonical encoding names. 448 * Returns -1 if not found. 449 */ 450 static int 451 enc_canon_search(char_u *name) 452 { 453 int i; 454 455 for (i = 0; i < IDX_COUNT; ++i) 456 if (STRCMP(name, enc_canon_table[i].name) == 0) 457 return i; 458 return -1; 459 } 460 461 #endif 462 463 #if defined(FEAT_MBYTE) || defined(PROTO) 464 465 /* 466 * Find canonical encoding "name" in the list and return its properties. 467 * Returns 0 if not found. 468 */ 469 int 470 enc_canon_props(char_u *name) 471 { 472 int i; 473 474 i = enc_canon_search(name); 475 if (i >= 0) 476 return enc_canon_table[i].prop; 477 #ifdef WIN3264 478 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2])) 479 { 480 CPINFO cpinfo; 481 482 /* Get info on this codepage to find out what it is. */ 483 if (GetCPInfo(atoi((char *)name + 2), &cpinfo) != 0) 484 { 485 if (cpinfo.MaxCharSize == 1) /* some single-byte encoding */ 486 return ENC_8BIT; 487 if (cpinfo.MaxCharSize == 2 488 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 489 /* must be a DBCS encoding */ 490 return ENC_DBCS; 491 } 492 return 0; 493 } 494 #endif 495 if (STRNCMP(name, "2byte-", 6) == 0) 496 return ENC_DBCS; 497 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0) 498 return ENC_8BIT; 499 return 0; 500 } 501 502 /* 503 * Set up for using multi-byte characters. 504 * Called in three cases: 505 * - by main() to initialize (p_enc == NULL) 506 * - by set_init_1() after 'encoding' was set to its default. 507 * - by do_set() when 'encoding' has been set. 508 * p_enc must have been passed through enc_canonize() already. 509 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags. 510 * Fills mb_bytelen_tab[] and returns NULL when there are no problems. 511 * When there is something wrong: Returns an error message and doesn't change 512 * anything. 513 */ 514 char_u * 515 mb_init(void) 516 { 517 int i; 518 int idx; 519 int n; 520 int enc_dbcs_new = 0; 521 #if defined(USE_ICONV) && !defined(WIN3264) && !defined(WIN32UNIX) \ 522 && !defined(MACOS_CONVERT) 523 # define LEN_FROM_CONV 524 vimconv_T vimconv; 525 char_u *p; 526 #endif 527 528 if (p_enc == NULL) 529 { 530 /* Just starting up: set the whole table to one's. */ 531 for (i = 0; i < 256; ++i) 532 mb_bytelen_tab[i] = 1; 533 input_conv.vc_type = CONV_NONE; 534 input_conv.vc_factor = 1; 535 output_conv.vc_type = CONV_NONE; 536 return NULL; 537 } 538 539 #ifdef WIN3264 540 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2])) 541 { 542 CPINFO cpinfo; 543 544 /* Get info on this codepage to find out what it is. */ 545 if (GetCPInfo(atoi((char *)p_enc + 2), &cpinfo) != 0) 546 { 547 if (cpinfo.MaxCharSize == 1) 548 { 549 /* some single-byte encoding */ 550 enc_unicode = 0; 551 enc_utf8 = FALSE; 552 } 553 else if (cpinfo.MaxCharSize == 2 554 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 555 { 556 /* must be a DBCS encoding, check below */ 557 enc_dbcs_new = atoi((char *)p_enc + 2); 558 } 559 else 560 goto codepage_invalid; 561 } 562 else if (GetLastError() == ERROR_INVALID_PARAMETER) 563 { 564 codepage_invalid: 565 return (char_u *)N_("E543: Not a valid codepage"); 566 } 567 } 568 #endif 569 else if (STRNCMP(p_enc, "8bit-", 5) == 0 570 || STRNCMP(p_enc, "iso-8859-", 9) == 0) 571 { 572 /* Accept any "8bit-" or "iso-8859-" name. */ 573 enc_unicode = 0; 574 enc_utf8 = FALSE; 575 } 576 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 577 { 578 #ifdef WIN3264 579 /* Windows: accept only valid codepage numbers, check below. */ 580 if (p_enc[6] != 'c' || p_enc[7] != 'p' 581 || (enc_dbcs_new = atoi((char *)p_enc + 8)) == 0) 582 return e_invarg; 583 #else 584 /* Unix: accept any "2byte-" name, assume current locale. */ 585 enc_dbcs_new = DBCS_2BYTE; 586 #endif 587 } 588 else if ((idx = enc_canon_search(p_enc)) >= 0) 589 { 590 i = enc_canon_table[idx].prop; 591 if (i & ENC_UNICODE) 592 { 593 /* Unicode */ 594 enc_utf8 = TRUE; 595 if (i & (ENC_2BYTE | ENC_2WORD)) 596 enc_unicode = 2; 597 else if (i & ENC_4BYTE) 598 enc_unicode = 4; 599 else 600 enc_unicode = 0; 601 } 602 else if (i & ENC_DBCS) 603 { 604 /* 2byte, handle below */ 605 enc_dbcs_new = enc_canon_table[idx].codepage; 606 } 607 else 608 { 609 /* Must be 8-bit. */ 610 enc_unicode = 0; 611 enc_utf8 = FALSE; 612 } 613 } 614 else /* Don't know what encoding this is, reject it. */ 615 return e_invarg; 616 617 if (enc_dbcs_new != 0) 618 { 619 #ifdef WIN3264 620 /* Check if the DBCS code page is OK. */ 621 if (!IsValidCodePage(enc_dbcs_new)) 622 goto codepage_invalid; 623 #endif 624 enc_unicode = 0; 625 enc_utf8 = FALSE; 626 } 627 enc_dbcs = enc_dbcs_new; 628 has_mbyte = (enc_dbcs != 0 || enc_utf8); 629 630 #if defined(WIN3264) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 631 enc_codepage = encname2codepage(p_enc); 632 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0); 633 #endif 634 635 /* Detect an encoding that uses latin1 characters. */ 636 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0 637 || STRCMP(p_enc, "iso-8859-15") == 0); 638 639 /* 640 * Set the function pointers. 641 */ 642 if (enc_utf8) 643 { 644 mb_ptr2len = utfc_ptr2len; 645 mb_ptr2len_len = utfc_ptr2len_len; 646 mb_char2len = utf_char2len; 647 mb_char2bytes = utf_char2bytes; 648 mb_ptr2cells = utf_ptr2cells; 649 mb_ptr2cells_len = utf_ptr2cells_len; 650 mb_char2cells = utf_char2cells; 651 mb_off2cells = utf_off2cells; 652 mb_ptr2char = utf_ptr2char; 653 mb_head_off = utf_head_off; 654 } 655 else if (enc_dbcs != 0) 656 { 657 mb_ptr2len = dbcs_ptr2len; 658 mb_ptr2len_len = dbcs_ptr2len_len; 659 mb_char2len = dbcs_char2len; 660 mb_char2bytes = dbcs_char2bytes; 661 mb_ptr2cells = dbcs_ptr2cells; 662 mb_ptr2cells_len = dbcs_ptr2cells_len; 663 mb_char2cells = dbcs_char2cells; 664 mb_off2cells = dbcs_off2cells; 665 mb_ptr2char = dbcs_ptr2char; 666 mb_head_off = dbcs_head_off; 667 } 668 else 669 { 670 mb_ptr2len = latin_ptr2len; 671 mb_ptr2len_len = latin_ptr2len_len; 672 mb_char2len = latin_char2len; 673 mb_char2bytes = latin_char2bytes; 674 mb_ptr2cells = latin_ptr2cells; 675 mb_ptr2cells_len = latin_ptr2cells_len; 676 mb_char2cells = latin_char2cells; 677 mb_off2cells = latin_off2cells; 678 mb_ptr2char = latin_ptr2char; 679 mb_head_off = latin_head_off; 680 } 681 682 /* 683 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN(). 684 */ 685 #ifdef LEN_FROM_CONV 686 /* When 'encoding' is different from the current locale mblen() won't 687 * work. Use conversion to "utf-8" instead. */ 688 vimconv.vc_type = CONV_NONE; 689 if (enc_dbcs) 690 { 691 p = enc_locale(); 692 if (p == NULL || STRCMP(p, p_enc) != 0) 693 { 694 convert_setup(&vimconv, p_enc, (char_u *)"utf-8"); 695 vimconv.vc_fail = TRUE; 696 } 697 vim_free(p); 698 } 699 #endif 700 701 for (i = 0; i < 256; ++i) 702 { 703 /* Our own function to reliably check the length of UTF-8 characters, 704 * independent of mblen(). */ 705 if (enc_utf8) 706 n = utf8len_tab[i]; 707 else if (enc_dbcs == 0) 708 n = 1; 709 else 710 { 711 #if defined(WIN3264) || defined(WIN32UNIX) 712 /* enc_dbcs is set by setting 'fileencoding'. It becomes a Windows 713 * CodePage identifier, which we can pass directly in to Windows 714 * API */ 715 n = IsDBCSLeadByteEx(enc_dbcs, (WINBYTE)i) ? 2 : 1; 716 #else 717 # if defined(__amigaos4__) || defined(__ANDROID__) || \ 718 !(defined(HAVE_MBLEN) || defined(X_LOCALE)) 719 /* 720 * if mblen() is not available, character which MSB is turned on 721 * are treated as leading byte character. (note : This assumption 722 * is not always true.) 723 */ 724 n = (i & 0x80) ? 2 : 1; 725 # else 726 char buf[MB_MAXBYTES + 1]; 727 728 if (i == NUL) /* just in case mblen() can't handle "" */ 729 n = 1; 730 else 731 { 732 buf[0] = i; 733 buf[1] = 0; 734 # ifdef LEN_FROM_CONV 735 if (vimconv.vc_type != CONV_NONE) 736 { 737 /* 738 * string_convert() should fail when converting the first 739 * byte of a double-byte character. 740 */ 741 p = string_convert(&vimconv, (char_u *)buf, NULL); 742 if (p != NULL) 743 { 744 vim_free(p); 745 n = 1; 746 } 747 else 748 n = 2; 749 } 750 else 751 # endif 752 { 753 /* 754 * mblen() should return -1 for invalid (means the leading 755 * multibyte) character. However there are some platforms 756 * where mblen() returns 0 for invalid character. 757 * Therefore, following condition includes 0. 758 */ 759 ignored = mblen(NULL, 0); /* First reset the state. */ 760 if (mblen(buf, (size_t)1) <= 0) 761 n = 2; 762 else 763 n = 1; 764 } 765 } 766 # endif 767 #endif 768 } 769 770 mb_bytelen_tab[i] = n; 771 } 772 773 #ifdef LEN_FROM_CONV 774 convert_setup(&vimconv, NULL, NULL); 775 #endif 776 777 /* The cell width depends on the type of multi-byte characters. */ 778 (void)init_chartab(); 779 780 /* When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] */ 781 screenalloc(FALSE); 782 783 /* When using Unicode, set default for 'fileencodings'. */ 784 if (enc_utf8 && !option_was_set((char_u *)"fencs")) 785 set_string_option_direct((char_u *)"fencs", -1, 786 (char_u *)"ucs-bom,utf-8,default,latin1", OPT_FREE, 0); 787 788 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT) 789 /* GNU gettext 0.10.37 supports this feature: set the codeset used for 790 * translated messages independently from the current locale. */ 791 (void)bind_textdomain_codeset(VIMPACKAGE, 792 enc_utf8 ? "utf-8" : (char *)p_enc); 793 #endif 794 795 #ifdef WIN32 796 /* When changing 'encoding' while starting up, then convert the command 797 * line arguments from the active codepage to 'encoding'. */ 798 if (starting != 0) 799 fix_arg_enc(); 800 #endif 801 802 /* Fire an autocommand to let people do custom font setup. This must be 803 * after Vim has been setup for the new encoding. */ 804 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); 805 806 #ifdef FEAT_SPELL 807 /* Need to reload spell dictionaries */ 808 spell_reload(); 809 #endif 810 811 return NULL; 812 } 813 814 /* 815 * Return the size of the BOM for the current buffer: 816 * 0 - no BOM 817 * 2 - UCS-2 or UTF-16 BOM 818 * 4 - UCS-4 BOM 819 * 3 - UTF-8 BOM 820 */ 821 int 822 bomb_size(void) 823 { 824 int n = 0; 825 826 if (curbuf->b_p_bomb && !curbuf->b_p_bin) 827 { 828 if (*curbuf->b_p_fenc == NUL) 829 { 830 if (enc_utf8) 831 { 832 if (enc_unicode != 0) 833 n = enc_unicode; 834 else 835 n = 3; 836 } 837 } 838 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) 839 n = 3; 840 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 841 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) 842 n = 2; 843 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) 844 n = 4; 845 } 846 return n; 847 } 848 849 /* 850 * Remove all BOM from "s" by moving remaining text. 851 */ 852 void 853 remove_bom(char_u *s) 854 { 855 if (enc_utf8) 856 { 857 char_u *p = s; 858 859 while ((p = vim_strbyte(p, 0xef)) != NULL) 860 { 861 if (p[1] == 0xbb && p[2] == 0xbf) 862 STRMOVE(p, p + 3); 863 else 864 ++p; 865 } 866 } 867 } 868 869 /* 870 * Get class of pointer: 871 * 0 for blank or NUL 872 * 1 for punctuation 873 * 2 for an (ASCII) word character 874 * >2 for other word characters 875 */ 876 int 877 mb_get_class(char_u *p) 878 { 879 return mb_get_class_buf(p, curbuf); 880 } 881 882 int 883 mb_get_class_buf(char_u *p, buf_T *buf) 884 { 885 if (MB_BYTE2LEN(p[0]) == 1) 886 { 887 if (p[0] == NUL || VIM_ISWHITE(p[0])) 888 return 0; 889 if (vim_iswordc_buf(p[0], buf)) 890 return 2; 891 return 1; 892 } 893 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) 894 return dbcs_class(p[0], p[1]); 895 if (enc_utf8) 896 return utf_class_buf(utf_ptr2char(p), buf); 897 return 0; 898 } 899 900 /* 901 * Get class of a double-byte character. This always returns 3 or bigger. 902 * TODO: Should return 1 for punctuation. 903 */ 904 int 905 dbcs_class(unsigned lead, unsigned trail) 906 { 907 switch (enc_dbcs) 908 { 909 /* please add classify routine for your language in here */ 910 911 case DBCS_JPNU: /* ? */ 912 case DBCS_JPN: 913 { 914 /* JIS code classification */ 915 unsigned char lb = lead; 916 unsigned char tb = trail; 917 918 /* convert process code to JIS */ 919 # if defined(WIN3264) || defined(WIN32UNIX) || defined(MACOS_X) 920 /* process code is SJIS */ 921 if (lb <= 0x9f) 922 lb = (lb - 0x81) * 2 + 0x21; 923 else 924 lb = (lb - 0xc1) * 2 + 0x21; 925 if (tb <= 0x7e) 926 tb -= 0x1f; 927 else if (tb <= 0x9e) 928 tb -= 0x20; 929 else 930 { 931 tb -= 0x7e; 932 lb += 1; 933 } 934 # else 935 /* 936 * XXX: Code page identification can not use with all 937 * system! So, some other encoding information 938 * will be needed. 939 * In japanese: SJIS,EUC,UNICODE,(JIS) 940 * Note that JIS-code system don't use as 941 * process code in most system because it uses 942 * escape sequences(JIS is context depend encoding). 943 */ 944 /* assume process code is JAPANESE-EUC */ 945 lb &= 0x7f; 946 tb &= 0x7f; 947 # endif 948 /* exceptions */ 949 switch (lb << 8 | tb) 950 { 951 case 0x2121: /* ZENKAKU space */ 952 return 0; 953 case 0x2122: /* TOU-TEN (Japanese comma) */ 954 case 0x2123: /* KU-TEN (Japanese period) */ 955 case 0x2124: /* ZENKAKU comma */ 956 case 0x2125: /* ZENKAKU period */ 957 return 1; 958 case 0x213c: /* prolongedsound handled as KATAKANA */ 959 return 13; 960 } 961 /* sieved by KU code */ 962 switch (lb) 963 { 964 case 0x21: 965 case 0x22: 966 /* special symbols */ 967 return 10; 968 case 0x23: 969 /* alpha-numeric */ 970 return 11; 971 case 0x24: 972 /* hiragana */ 973 return 12; 974 case 0x25: 975 /* katakana */ 976 return 13; 977 case 0x26: 978 /* greek */ 979 return 14; 980 case 0x27: 981 /* russian */ 982 return 15; 983 case 0x28: 984 /* lines */ 985 return 16; 986 default: 987 /* kanji */ 988 return 17; 989 } 990 } 991 992 case DBCS_KORU: /* ? */ 993 case DBCS_KOR: 994 { 995 /* KS code classification */ 996 unsigned char c1 = lead; 997 unsigned char c2 = trail; 998 999 /* 1000 * 20 : Hangul 1001 * 21 : Hanja 1002 * 22 : Symbols 1003 * 23 : Alpha-numeric/Roman Letter (Full width) 1004 * 24 : Hangul Letter(Alphabet) 1005 * 25 : Roman Numeral/Greek Letter 1006 * 26 : Box Drawings 1007 * 27 : Unit Symbols 1008 * 28 : Circled/Parenthesized Letter 1009 * 29 : Hiragana/Katakana 1010 * 30 : Cyrillic Letter 1011 */ 1012 1013 if (c1 >= 0xB0 && c1 <= 0xC8) 1014 /* Hangul */ 1015 return 20; 1016 #if defined(WIN3264) || defined(WIN32UNIX) 1017 else if (c1 <= 0xA0 || c2 <= 0xA0) 1018 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */ 1019 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE 1020 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 1021 */ 1022 return 20; 1023 #endif 1024 1025 else if (c1 >= 0xCA && c1 <= 0xFD) 1026 /* Hanja */ 1027 return 21; 1028 else switch (c1) 1029 { 1030 case 0xA1: 1031 case 0xA2: 1032 /* Symbols */ 1033 return 22; 1034 case 0xA3: 1035 /* Alpha-numeric */ 1036 return 23; 1037 case 0xA4: 1038 /* Hangul Letter(Alphabet) */ 1039 return 24; 1040 case 0xA5: 1041 /* Roman Numeral/Greek Letter */ 1042 return 25; 1043 case 0xA6: 1044 /* Box Drawings */ 1045 return 26; 1046 case 0xA7: 1047 /* Unit Symbols */ 1048 return 27; 1049 case 0xA8: 1050 case 0xA9: 1051 if (c2 <= 0xAF) 1052 return 25; /* Roman Letter */ 1053 else if (c2 >= 0xF6) 1054 return 22; /* Symbols */ 1055 else 1056 /* Circled/Parenthesized Letter */ 1057 return 28; 1058 case 0xAA: 1059 case 0xAB: 1060 /* Hiragana/Katakana */ 1061 return 29; 1062 case 0xAC: 1063 /* Cyrillic Letter */ 1064 return 30; 1065 } 1066 } 1067 default: 1068 break; 1069 } 1070 return 3; 1071 } 1072 1073 /* 1074 * mb_char2len() function pointer. 1075 * Return length in bytes of character "c". 1076 * Returns 1 for a single-byte character. 1077 */ 1078 int 1079 latin_char2len(int c UNUSED) 1080 { 1081 return 1; 1082 } 1083 1084 static int 1085 dbcs_char2len( 1086 int c) 1087 { 1088 if (c >= 0x100) 1089 return 2; 1090 return 1; 1091 } 1092 1093 /* 1094 * mb_char2bytes() function pointer. 1095 * Convert a character to its bytes. 1096 * Returns the length in bytes. 1097 */ 1098 int 1099 latin_char2bytes(int c, char_u *buf) 1100 { 1101 buf[0] = c; 1102 return 1; 1103 } 1104 1105 static int 1106 dbcs_char2bytes(int c, char_u *buf) 1107 { 1108 if (c >= 0x100) 1109 { 1110 buf[0] = (unsigned)c >> 8; 1111 buf[1] = c; 1112 /* Never use a NUL byte, it causes lots of trouble. It's an invalid 1113 * character anyway. */ 1114 if (buf[1] == NUL) 1115 buf[1] = '\n'; 1116 return 2; 1117 } 1118 buf[0] = c; 1119 return 1; 1120 } 1121 1122 /* 1123 * mb_ptr2len() function pointer. 1124 * Get byte length of character at "*p" but stop at a NUL. 1125 * For UTF-8 this includes following composing characters. 1126 * Returns 0 when *p is NUL. 1127 */ 1128 int 1129 latin_ptr2len(char_u *p) 1130 { 1131 return MB_BYTE2LEN(*p); 1132 } 1133 1134 static int 1135 dbcs_ptr2len( 1136 char_u *p) 1137 { 1138 int len; 1139 1140 /* Check if second byte is not missing. */ 1141 len = MB_BYTE2LEN(*p); 1142 if (len == 2 && p[1] == NUL) 1143 len = 1; 1144 return len; 1145 } 1146 1147 /* 1148 * mb_ptr2len_len() function pointer. 1149 * Like mb_ptr2len(), but limit to read "size" bytes. 1150 * Returns 0 for an empty string. 1151 * Returns 1 for an illegal char or an incomplete byte sequence. 1152 */ 1153 int 1154 latin_ptr2len_len(char_u *p, int size) 1155 { 1156 if (size < 1 || *p == NUL) 1157 return 0; 1158 return 1; 1159 } 1160 1161 static int 1162 dbcs_ptr2len_len(char_u *p, int size) 1163 { 1164 int len; 1165 1166 if (size < 1 || *p == NUL) 1167 return 0; 1168 if (size == 1) 1169 return 1; 1170 /* Check that second byte is not missing. */ 1171 len = MB_BYTE2LEN(*p); 1172 if (len == 2 && p[1] == NUL) 1173 len = 1; 1174 return len; 1175 } 1176 1177 struct interval 1178 { 1179 long first; 1180 long last; 1181 }; 1182 1183 /* 1184 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]". 1185 */ 1186 static int 1187 intable(struct interval *table, size_t size, int c) 1188 { 1189 int mid, bot, top; 1190 1191 /* first quick check for Latin1 etc. characters */ 1192 if (c < table[0].first) 1193 return FALSE; 1194 1195 /* binary search in table */ 1196 bot = 0; 1197 top = (int)(size / sizeof(struct interval) - 1); 1198 while (top >= bot) 1199 { 1200 mid = (bot + top) / 2; 1201 if (table[mid].last < c) 1202 bot = mid + 1; 1203 else if (table[mid].first > c) 1204 top = mid - 1; 1205 else 1206 return TRUE; 1207 } 1208 return FALSE; 1209 } 1210 1211 /* Sorted list of non-overlapping intervals of East Asian Ambiguous 1212 * characters, generated with ../runtime/tools/unicode.vim. */ 1213 static struct interval ambiguous[] = 1214 { 1215 {0x00a1, 0x00a1}, 1216 {0x00a4, 0x00a4}, 1217 {0x00a7, 0x00a8}, 1218 {0x00aa, 0x00aa}, 1219 {0x00ad, 0x00ae}, 1220 {0x00b0, 0x00b4}, 1221 {0x00b6, 0x00ba}, 1222 {0x00bc, 0x00bf}, 1223 {0x00c6, 0x00c6}, 1224 {0x00d0, 0x00d0}, 1225 {0x00d7, 0x00d8}, 1226 {0x00de, 0x00e1}, 1227 {0x00e6, 0x00e6}, 1228 {0x00e8, 0x00ea}, 1229 {0x00ec, 0x00ed}, 1230 {0x00f0, 0x00f0}, 1231 {0x00f2, 0x00f3}, 1232 {0x00f7, 0x00fa}, 1233 {0x00fc, 0x00fc}, 1234 {0x00fe, 0x00fe}, 1235 {0x0101, 0x0101}, 1236 {0x0111, 0x0111}, 1237 {0x0113, 0x0113}, 1238 {0x011b, 0x011b}, 1239 {0x0126, 0x0127}, 1240 {0x012b, 0x012b}, 1241 {0x0131, 0x0133}, 1242 {0x0138, 0x0138}, 1243 {0x013f, 0x0142}, 1244 {0x0144, 0x0144}, 1245 {0x0148, 0x014b}, 1246 {0x014d, 0x014d}, 1247 {0x0152, 0x0153}, 1248 {0x0166, 0x0167}, 1249 {0x016b, 0x016b}, 1250 {0x01ce, 0x01ce}, 1251 {0x01d0, 0x01d0}, 1252 {0x01d2, 0x01d2}, 1253 {0x01d4, 0x01d4}, 1254 {0x01d6, 0x01d6}, 1255 {0x01d8, 0x01d8}, 1256 {0x01da, 0x01da}, 1257 {0x01dc, 0x01dc}, 1258 {0x0251, 0x0251}, 1259 {0x0261, 0x0261}, 1260 {0x02c4, 0x02c4}, 1261 {0x02c7, 0x02c7}, 1262 {0x02c9, 0x02cb}, 1263 {0x02cd, 0x02cd}, 1264 {0x02d0, 0x02d0}, 1265 {0x02d8, 0x02db}, 1266 {0x02dd, 0x02dd}, 1267 {0x02df, 0x02df}, 1268 {0x0300, 0x036f}, 1269 {0x0391, 0x03a1}, 1270 {0x03a3, 0x03a9}, 1271 {0x03b1, 0x03c1}, 1272 {0x03c3, 0x03c9}, 1273 {0x0401, 0x0401}, 1274 {0x0410, 0x044f}, 1275 {0x0451, 0x0451}, 1276 {0x2010, 0x2010}, 1277 {0x2013, 0x2016}, 1278 {0x2018, 0x2019}, 1279 {0x201c, 0x201d}, 1280 {0x2020, 0x2022}, 1281 {0x2024, 0x2027}, 1282 {0x2030, 0x2030}, 1283 {0x2032, 0x2033}, 1284 {0x2035, 0x2035}, 1285 {0x203b, 0x203b}, 1286 {0x203e, 0x203e}, 1287 {0x2074, 0x2074}, 1288 {0x207f, 0x207f}, 1289 {0x2081, 0x2084}, 1290 {0x20ac, 0x20ac}, 1291 {0x2103, 0x2103}, 1292 {0x2105, 0x2105}, 1293 {0x2109, 0x2109}, 1294 {0x2113, 0x2113}, 1295 {0x2116, 0x2116}, 1296 {0x2121, 0x2122}, 1297 {0x2126, 0x2126}, 1298 {0x212b, 0x212b}, 1299 {0x2153, 0x2154}, 1300 {0x215b, 0x215e}, 1301 {0x2160, 0x216b}, 1302 {0x2170, 0x2179}, 1303 {0x2189, 0x2189}, 1304 {0x2190, 0x2199}, 1305 {0x21b8, 0x21b9}, 1306 {0x21d2, 0x21d2}, 1307 {0x21d4, 0x21d4}, 1308 {0x21e7, 0x21e7}, 1309 {0x2200, 0x2200}, 1310 {0x2202, 0x2203}, 1311 {0x2207, 0x2208}, 1312 {0x220b, 0x220b}, 1313 {0x220f, 0x220f}, 1314 {0x2211, 0x2211}, 1315 {0x2215, 0x2215}, 1316 {0x221a, 0x221a}, 1317 {0x221d, 0x2220}, 1318 {0x2223, 0x2223}, 1319 {0x2225, 0x2225}, 1320 {0x2227, 0x222c}, 1321 {0x222e, 0x222e}, 1322 {0x2234, 0x2237}, 1323 {0x223c, 0x223d}, 1324 {0x2248, 0x2248}, 1325 {0x224c, 0x224c}, 1326 {0x2252, 0x2252}, 1327 {0x2260, 0x2261}, 1328 {0x2264, 0x2267}, 1329 {0x226a, 0x226b}, 1330 {0x226e, 0x226f}, 1331 {0x2282, 0x2283}, 1332 {0x2286, 0x2287}, 1333 {0x2295, 0x2295}, 1334 {0x2299, 0x2299}, 1335 {0x22a5, 0x22a5}, 1336 {0x22bf, 0x22bf}, 1337 {0x2312, 0x2312}, 1338 {0x2460, 0x24e9}, 1339 {0x24eb, 0x254b}, 1340 {0x2550, 0x2573}, 1341 {0x2580, 0x258f}, 1342 {0x2592, 0x2595}, 1343 {0x25a0, 0x25a1}, 1344 {0x25a3, 0x25a9}, 1345 {0x25b2, 0x25b3}, 1346 {0x25b6, 0x25b7}, 1347 {0x25bc, 0x25bd}, 1348 {0x25c0, 0x25c1}, 1349 {0x25c6, 0x25c8}, 1350 {0x25cb, 0x25cb}, 1351 {0x25ce, 0x25d1}, 1352 {0x25e2, 0x25e5}, 1353 {0x25ef, 0x25ef}, 1354 {0x2605, 0x2606}, 1355 {0x2609, 0x2609}, 1356 {0x260e, 0x260f}, 1357 {0x261c, 0x261c}, 1358 {0x261e, 0x261e}, 1359 {0x2640, 0x2640}, 1360 {0x2642, 0x2642}, 1361 {0x2660, 0x2661}, 1362 {0x2663, 0x2665}, 1363 {0x2667, 0x266a}, 1364 {0x266c, 0x266d}, 1365 {0x266f, 0x266f}, 1366 {0x269e, 0x269f}, 1367 {0x26bf, 0x26bf}, 1368 {0x26c6, 0x26cd}, 1369 {0x26cf, 0x26d3}, 1370 {0x26d5, 0x26e1}, 1371 {0x26e3, 0x26e3}, 1372 {0x26e8, 0x26e9}, 1373 {0x26eb, 0x26f1}, 1374 {0x26f4, 0x26f4}, 1375 {0x26f6, 0x26f9}, 1376 {0x26fb, 0x26fc}, 1377 {0x26fe, 0x26ff}, 1378 {0x273d, 0x273d}, 1379 {0x2776, 0x277f}, 1380 {0x2b56, 0x2b59}, 1381 {0x3248, 0x324f}, 1382 {0xe000, 0xf8ff}, 1383 {0xfe00, 0xfe0f}, 1384 {0xfffd, 0xfffd}, 1385 {0x1f100, 0x1f10a}, 1386 {0x1f110, 0x1f12d}, 1387 {0x1f130, 0x1f169}, 1388 {0x1f170, 0x1f18d}, 1389 {0x1f18f, 0x1f190}, 1390 {0x1f19b, 0x1f1ac}, 1391 {0xe0100, 0xe01ef}, 1392 {0xf0000, 0xffffd}, 1393 {0x100000, 0x10fffd} 1394 }; 1395 1396 #if defined(FEAT_TERMINAL) || defined(PROTO) 1397 /* 1398 * utf_char2cells() with different argument type for libvterm. 1399 */ 1400 int 1401 utf_uint2cells(UINT32_T c) 1402 { 1403 if (c >= 0x100 && utf_iscomposing((int)c)) 1404 return 0; 1405 return utf_char2cells((int)c); 1406 } 1407 #endif 1408 1409 /* 1410 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1411 * Returns 4 or 6 for an unprintable character. 1412 * Is only correct for characters >= 0x80. 1413 * When p_ambw is "double", return 2 for a character with East Asian Width 1414 * class 'A'(mbiguous). 1415 */ 1416 int 1417 utf_char2cells(int c) 1418 { 1419 /* Sorted list of non-overlapping intervals of East Asian double width 1420 * characters, generated with ../runtime/tools/unicode.vim. */ 1421 static struct interval doublewidth[] = 1422 { 1423 {0x1100, 0x115f}, 1424 {0x231a, 0x231b}, 1425 {0x2329, 0x232a}, 1426 {0x23e9, 0x23ec}, 1427 {0x23f0, 0x23f0}, 1428 {0x23f3, 0x23f3}, 1429 {0x25fd, 0x25fe}, 1430 {0x2614, 0x2615}, 1431 {0x2648, 0x2653}, 1432 {0x267f, 0x267f}, 1433 {0x2693, 0x2693}, 1434 {0x26a1, 0x26a1}, 1435 {0x26aa, 0x26ab}, 1436 {0x26bd, 0x26be}, 1437 {0x26c4, 0x26c5}, 1438 {0x26ce, 0x26ce}, 1439 {0x26d4, 0x26d4}, 1440 {0x26ea, 0x26ea}, 1441 {0x26f2, 0x26f3}, 1442 {0x26f5, 0x26f5}, 1443 {0x26fa, 0x26fa}, 1444 {0x26fd, 0x26fd}, 1445 {0x2705, 0x2705}, 1446 {0x270a, 0x270b}, 1447 {0x2728, 0x2728}, 1448 {0x274c, 0x274c}, 1449 {0x274e, 0x274e}, 1450 {0x2753, 0x2755}, 1451 {0x2757, 0x2757}, 1452 {0x2795, 0x2797}, 1453 {0x27b0, 0x27b0}, 1454 {0x27bf, 0x27bf}, 1455 {0x2b1b, 0x2b1c}, 1456 {0x2b50, 0x2b50}, 1457 {0x2b55, 0x2b55}, 1458 {0x2e80, 0x2e99}, 1459 {0x2e9b, 0x2ef3}, 1460 {0x2f00, 0x2fd5}, 1461 {0x2ff0, 0x2ffb}, 1462 {0x3000, 0x303e}, 1463 {0x3041, 0x3096}, 1464 {0x3099, 0x30ff}, 1465 {0x3105, 0x312f}, 1466 {0x3131, 0x318e}, 1467 {0x3190, 0x31ba}, 1468 {0x31c0, 0x31e3}, 1469 {0x31f0, 0x321e}, 1470 {0x3220, 0x3247}, 1471 {0x3250, 0x32fe}, 1472 {0x3300, 0x4dbf}, 1473 {0x4e00, 0xa48c}, 1474 {0xa490, 0xa4c6}, 1475 {0xa960, 0xa97c}, 1476 {0xac00, 0xd7a3}, 1477 {0xf900, 0xfaff}, 1478 {0xfe10, 0xfe19}, 1479 {0xfe30, 0xfe52}, 1480 {0xfe54, 0xfe66}, 1481 {0xfe68, 0xfe6b}, 1482 {0xff01, 0xff60}, 1483 {0xffe0, 0xffe6}, 1484 {0x16fe0, 0x16fe1}, 1485 {0x17000, 0x187f1}, 1486 {0x18800, 0x18af2}, 1487 {0x1b000, 0x1b11e}, 1488 {0x1b170, 0x1b2fb}, 1489 {0x1f004, 0x1f004}, 1490 {0x1f0cf, 0x1f0cf}, 1491 {0x1f18e, 0x1f18e}, 1492 {0x1f191, 0x1f19a}, 1493 {0x1f200, 0x1f202}, 1494 {0x1f210, 0x1f23b}, 1495 {0x1f240, 0x1f248}, 1496 {0x1f250, 0x1f251}, 1497 {0x1f260, 0x1f265}, 1498 {0x1f300, 0x1f320}, 1499 {0x1f32d, 0x1f335}, 1500 {0x1f337, 0x1f37c}, 1501 {0x1f37e, 0x1f393}, 1502 {0x1f3a0, 0x1f3ca}, 1503 {0x1f3cf, 0x1f3d3}, 1504 {0x1f3e0, 0x1f3f0}, 1505 {0x1f3f4, 0x1f3f4}, 1506 {0x1f3f8, 0x1f43e}, 1507 {0x1f440, 0x1f440}, 1508 {0x1f442, 0x1f4fc}, 1509 {0x1f4ff, 0x1f53d}, 1510 {0x1f54b, 0x1f54e}, 1511 {0x1f550, 0x1f567}, 1512 {0x1f57a, 0x1f57a}, 1513 {0x1f595, 0x1f596}, 1514 {0x1f5a4, 0x1f5a4}, 1515 {0x1f5fb, 0x1f64f}, 1516 {0x1f680, 0x1f6c5}, 1517 {0x1f6cc, 0x1f6cc}, 1518 {0x1f6d0, 0x1f6d2}, 1519 {0x1f6eb, 0x1f6ec}, 1520 {0x1f6f4, 0x1f6f9}, 1521 {0x1f910, 0x1f93e}, 1522 {0x1f940, 0x1f970}, 1523 {0x1f973, 0x1f976}, 1524 {0x1f97a, 0x1f97a}, 1525 {0x1f97c, 0x1f9a2}, 1526 {0x1f9b0, 0x1f9b9}, 1527 {0x1f9c0, 0x1f9c2}, 1528 {0x1f9d0, 0x1f9ff}, 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 * For an overlong sequence this may return zero. 1788 * Does not include composing characters, of course. 1789 */ 1790 int 1791 utf_ptr2char(char_u *p) 1792 { 1793 int len; 1794 1795 if (p[0] < 0x80) /* be quick for ASCII */ 1796 return p[0]; 1797 1798 len = utf8len_tab_zero[p[0]]; 1799 if (len > 1 && (p[1] & 0xc0) == 0x80) 1800 { 1801 if (len == 2) 1802 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); 1803 if ((p[2] & 0xc0) == 0x80) 1804 { 1805 if (len == 3) 1806 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) 1807 + (p[2] & 0x3f); 1808 if ((p[3] & 0xc0) == 0x80) 1809 { 1810 if (len == 4) 1811 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) 1812 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); 1813 if ((p[4] & 0xc0) == 0x80) 1814 { 1815 if (len == 5) 1816 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) 1817 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) 1818 + (p[4] & 0x3f); 1819 if ((p[5] & 0xc0) == 0x80 && len == 6) 1820 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) 1821 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) 1822 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); 1823 } 1824 } 1825 } 1826 } 1827 /* Illegal value, just return the first byte */ 1828 return p[0]; 1829 } 1830 1831 /* 1832 * Convert a UTF-8 byte sequence to a wide character. 1833 * String is assumed to be terminated by NUL or after "n" bytes, whichever 1834 * comes first. 1835 * The function is safe in the sense that it never accesses memory beyond the 1836 * first "n" bytes of "s". 1837 * 1838 * On success, returns decoded codepoint, advances "s" to the beginning of 1839 * next character and decreases "n" accordingly. 1840 * 1841 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past 1842 * NUL byte. 1843 * 1844 * If byte sequence is illegal or incomplete, returns -1 and does not advance 1845 * "s". 1846 */ 1847 static int 1848 utf_safe_read_char_adv(char_u **s, size_t *n) 1849 { 1850 int c, k; 1851 1852 if (*n == 0) /* end of buffer */ 1853 return 0; 1854 1855 k = utf8len_tab_zero[**s]; 1856 1857 if (k == 1) 1858 { 1859 /* ASCII character or NUL */ 1860 (*n)--; 1861 return *(*s)++; 1862 } 1863 1864 if ((size_t)k <= *n) 1865 { 1866 /* We have a multibyte sequence and it isn't truncated by buffer 1867 * limits so utf_ptr2char() is safe to use. Or the first byte is 1868 * illegal (k=0), and it's also safe to use utf_ptr2char(). */ 1869 c = utf_ptr2char(*s); 1870 1871 /* On failure, utf_ptr2char() returns the first byte, so here we 1872 * check equality with the first byte. The only non-ASCII character 1873 * which equals the first byte of its own UTF-8 representation is 1874 * U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too. 1875 * It's safe even if n=1, else we would have k=2 > n. */ 1876 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83)) 1877 { 1878 /* byte sequence was successfully decoded */ 1879 *s += k; 1880 *n -= k; 1881 return c; 1882 } 1883 } 1884 1885 /* byte sequence is incomplete or illegal */ 1886 return -1; 1887 } 1888 1889 /* 1890 * Get character at **pp and advance *pp to the next character. 1891 * Note: composing characters are skipped! 1892 */ 1893 int 1894 mb_ptr2char_adv(char_u **pp) 1895 { 1896 int c; 1897 1898 c = (*mb_ptr2char)(*pp); 1899 *pp += (*mb_ptr2len)(*pp); 1900 return c; 1901 } 1902 1903 /* 1904 * Get character at **pp and advance *pp to the next character. 1905 * Note: composing characters are returned as separate characters. 1906 */ 1907 int 1908 mb_cptr2char_adv(char_u **pp) 1909 { 1910 int c; 1911 1912 c = (*mb_ptr2char)(*pp); 1913 if (enc_utf8) 1914 *pp += utf_ptr2len(*pp); 1915 else 1916 *pp += (*mb_ptr2len)(*pp); 1917 return c; 1918 } 1919 1920 #if defined(FEAT_ARABIC) || defined(PROTO) 1921 /* 1922 * Check whether we are dealing with Arabic combining characters. 1923 * Note: these are NOT really composing characters! 1924 */ 1925 int 1926 arabic_combine( 1927 int one, /* first character */ 1928 int two) /* character just after "one" */ 1929 { 1930 if (one == a_LAM) 1931 return arabic_maycombine(two); 1932 return FALSE; 1933 } 1934 1935 /* 1936 * Check whether we are dealing with a character that could be regarded as an 1937 * Arabic combining character, need to check the character before this. 1938 */ 1939 int 1940 arabic_maycombine(int two) 1941 { 1942 if (p_arshape && !p_tbidi) 1943 return (two == a_ALEF_MADDA 1944 || two == a_ALEF_HAMZA_ABOVE 1945 || two == a_ALEF_HAMZA_BELOW 1946 || two == a_ALEF); 1947 return FALSE; 1948 } 1949 1950 /* 1951 * Check if the character pointed to by "p2" is a composing character when it 1952 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which 1953 * behaves like a composing character. 1954 */ 1955 int 1956 utf_composinglike(char_u *p1, char_u *p2) 1957 { 1958 int c2; 1959 1960 c2 = utf_ptr2char(p2); 1961 if (utf_iscomposing(c2)) 1962 return TRUE; 1963 if (!arabic_maycombine(c2)) 1964 return FALSE; 1965 return arabic_combine(utf_ptr2char(p1), c2); 1966 } 1967 #endif 1968 1969 /* 1970 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1971 * composing characters. 1972 */ 1973 int 1974 utfc_ptr2char( 1975 char_u *p, 1976 int *pcc) /* return: composing chars, last one is 0 */ 1977 { 1978 int len; 1979 int c; 1980 int cc; 1981 int i = 0; 1982 1983 c = utf_ptr2char(p); 1984 len = utf_ptr2len(p); 1985 1986 /* Only accept a composing char when the first char isn't illegal. */ 1987 if ((len > 1 || *p < 0x80) 1988 && p[len] >= 0x80 1989 && UTF_COMPOSINGLIKE(p, p + len)) 1990 { 1991 cc = utf_ptr2char(p + len); 1992 for (;;) 1993 { 1994 pcc[i++] = cc; 1995 if (i == MAX_MCO) 1996 break; 1997 len += utf_ptr2len(p + len); 1998 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1999 break; 2000 } 2001 } 2002 2003 if (i < MAX_MCO) /* last composing char must be 0 */ 2004 pcc[i] = 0; 2005 2006 return c; 2007 } 2008 2009 /* 2010 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 2011 * composing characters. Use no more than p[maxlen]. 2012 */ 2013 int 2014 utfc_ptr2char_len( 2015 char_u *p, 2016 int *pcc, /* return: composing chars, last one is 0 */ 2017 int maxlen) 2018 { 2019 int len; 2020 int c; 2021 int cc; 2022 int i = 0; 2023 2024 c = utf_ptr2char(p); 2025 len = utf_ptr2len_len(p, maxlen); 2026 /* Only accept a composing char when the first char isn't illegal. */ 2027 if ((len > 1 || *p < 0x80) 2028 && len < maxlen 2029 && p[len] >= 0x80 2030 && UTF_COMPOSINGLIKE(p, p + len)) 2031 { 2032 cc = utf_ptr2char(p + len); 2033 for (;;) 2034 { 2035 pcc[i++] = cc; 2036 if (i == MAX_MCO) 2037 break; 2038 len += utf_ptr2len_len(p + len, maxlen - len); 2039 if (len >= maxlen 2040 || p[len] < 0x80 2041 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 2042 break; 2043 } 2044 } 2045 2046 if (i < MAX_MCO) /* last composing char must be 0 */ 2047 pcc[i] = 0; 2048 2049 return c; 2050 } 2051 2052 /* 2053 * Convert the character at screen position "off" to a sequence of bytes. 2054 * Includes the composing characters. 2055 * "buf" must at least have the length MB_MAXBYTES + 1. 2056 * Only to be used when ScreenLinesUC[off] != 0. 2057 * Returns the produced number of bytes. 2058 */ 2059 int 2060 utfc_char2bytes(int off, char_u *buf) 2061 { 2062 int len; 2063 int i; 2064 2065 len = utf_char2bytes(ScreenLinesUC[off], buf); 2066 for (i = 0; i < Screen_mco; ++i) 2067 { 2068 if (ScreenLinesC[i][off] == 0) 2069 break; 2070 len += utf_char2bytes(ScreenLinesC[i][off], buf + len); 2071 } 2072 return len; 2073 } 2074 2075 /* 2076 * Get the length of a UTF-8 byte sequence, not including any following 2077 * composing characters. 2078 * Returns 0 for "". 2079 * Returns 1 for an illegal byte sequence. 2080 */ 2081 int 2082 utf_ptr2len(char_u *p) 2083 { 2084 int len; 2085 int i; 2086 2087 if (*p == NUL) 2088 return 0; 2089 len = utf8len_tab[*p]; 2090 for (i = 1; i < len; ++i) 2091 if ((p[i] & 0xc0) != 0x80) 2092 return 1; 2093 return len; 2094 } 2095 2096 /* 2097 * Return length of UTF-8 character, obtained from the first byte. 2098 * "b" must be between 0 and 255! 2099 * Returns 1 for an invalid first byte value. 2100 */ 2101 int 2102 utf_byte2len(int b) 2103 { 2104 return utf8len_tab[b]; 2105 } 2106 2107 /* 2108 * Get the length of UTF-8 byte sequence "p[size]". Does not include any 2109 * following composing characters. 2110 * Returns 1 for "". 2111 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.). 2112 * Returns number > "size" for an incomplete byte sequence. 2113 * Never returns zero. 2114 */ 2115 int 2116 utf_ptr2len_len(char_u *p, int size) 2117 { 2118 int len; 2119 int i; 2120 int m; 2121 2122 len = utf8len_tab[*p]; 2123 if (len == 1) 2124 return 1; /* NUL, ascii or illegal lead byte */ 2125 if (len > size) 2126 m = size; /* incomplete byte sequence. */ 2127 else 2128 m = len; 2129 for (i = 1; i < m; ++i) 2130 if ((p[i] & 0xc0) != 0x80) 2131 return 1; 2132 return len; 2133 } 2134 2135 /* 2136 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. 2137 * This includes following composing characters. 2138 */ 2139 int 2140 utfc_ptr2len(char_u *p) 2141 { 2142 int len; 2143 int b0 = *p; 2144 #ifdef FEAT_ARABIC 2145 int prevlen; 2146 #endif 2147 2148 if (b0 == NUL) 2149 return 0; 2150 if (b0 < 0x80 && p[1] < 0x80) /* be quick for ASCII */ 2151 return 1; 2152 2153 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2154 len = utf_ptr2len(p); 2155 2156 /* Check for illegal byte. */ 2157 if (len == 1 && b0 >= 0x80) 2158 return 1; 2159 2160 /* 2161 * Check for composing characters. We can handle only the first six, but 2162 * skip all of them (otherwise the cursor would get stuck). 2163 */ 2164 #ifdef FEAT_ARABIC 2165 prevlen = 0; 2166 #endif 2167 for (;;) 2168 { 2169 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2170 return len; 2171 2172 /* Skip over composing char */ 2173 #ifdef FEAT_ARABIC 2174 prevlen = len; 2175 #endif 2176 len += utf_ptr2len(p + len); 2177 } 2178 } 2179 2180 /* 2181 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" 2182 * takes. This includes following composing characters. 2183 * Returns 0 for an empty string. 2184 * Returns 1 for an illegal char or an incomplete byte sequence. 2185 */ 2186 int 2187 utfc_ptr2len_len(char_u *p, int size) 2188 { 2189 int len; 2190 #ifdef FEAT_ARABIC 2191 int prevlen; 2192 #endif 2193 2194 if (size < 1 || *p == NUL) 2195 return 0; 2196 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) /* be quick for ASCII */ 2197 return 1; 2198 2199 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2200 len = utf_ptr2len_len(p, size); 2201 2202 /* Check for illegal byte and incomplete byte sequence. */ 2203 if ((len == 1 && p[0] >= 0x80) || len > size) 2204 return 1; 2205 2206 /* 2207 * Check for composing characters. We can handle only the first six, but 2208 * skip all of them (otherwise the cursor would get stuck). 2209 */ 2210 #ifdef FEAT_ARABIC 2211 prevlen = 0; 2212 #endif 2213 while (len < size) 2214 { 2215 int len_next_char; 2216 2217 if (p[len] < 0x80) 2218 break; 2219 2220 /* 2221 * Next character length should not go beyond size to ensure that 2222 * UTF_COMPOSINGLIKE(...) does not read beyond size. 2223 */ 2224 len_next_char = utf_ptr2len_len(p + len, size - len); 2225 if (len_next_char > size - len) 2226 break; 2227 2228 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2229 break; 2230 2231 /* Skip over composing char */ 2232 #ifdef FEAT_ARABIC 2233 prevlen = len; 2234 #endif 2235 len += len_next_char; 2236 } 2237 return len; 2238 } 2239 2240 /* 2241 * Return the number of bytes the UTF-8 encoding of character "c" takes. 2242 * This does not include composing characters. 2243 */ 2244 int 2245 utf_char2len(int c) 2246 { 2247 if (c < 0x80) 2248 return 1; 2249 if (c < 0x800) 2250 return 2; 2251 if (c < 0x10000) 2252 return 3; 2253 if (c < 0x200000) 2254 return 4; 2255 if (c < 0x4000000) 2256 return 5; 2257 return 6; 2258 } 2259 2260 /* 2261 * Convert Unicode character "c" to UTF-8 string in "buf[]". 2262 * Returns the number of bytes. 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 {0x07fd, 0x07fd}, 2354 {0x0816, 0x0819}, 2355 {0x081b, 0x0823}, 2356 {0x0825, 0x0827}, 2357 {0x0829, 0x082d}, 2358 {0x0859, 0x085b}, 2359 {0x08d3, 0x08e1}, 2360 {0x08e3, 0x0903}, 2361 {0x093a, 0x093c}, 2362 {0x093e, 0x094f}, 2363 {0x0951, 0x0957}, 2364 {0x0962, 0x0963}, 2365 {0x0981, 0x0983}, 2366 {0x09bc, 0x09bc}, 2367 {0x09be, 0x09c4}, 2368 {0x09c7, 0x09c8}, 2369 {0x09cb, 0x09cd}, 2370 {0x09d7, 0x09d7}, 2371 {0x09e2, 0x09e3}, 2372 {0x09fe, 0x09fe}, 2373 {0x0a01, 0x0a03}, 2374 {0x0a3c, 0x0a3c}, 2375 {0x0a3e, 0x0a42}, 2376 {0x0a47, 0x0a48}, 2377 {0x0a4b, 0x0a4d}, 2378 {0x0a51, 0x0a51}, 2379 {0x0a70, 0x0a71}, 2380 {0x0a75, 0x0a75}, 2381 {0x0a81, 0x0a83}, 2382 {0x0abc, 0x0abc}, 2383 {0x0abe, 0x0ac5}, 2384 {0x0ac7, 0x0ac9}, 2385 {0x0acb, 0x0acd}, 2386 {0x0ae2, 0x0ae3}, 2387 {0x0afa, 0x0aff}, 2388 {0x0b01, 0x0b03}, 2389 {0x0b3c, 0x0b3c}, 2390 {0x0b3e, 0x0b44}, 2391 {0x0b47, 0x0b48}, 2392 {0x0b4b, 0x0b4d}, 2393 {0x0b56, 0x0b57}, 2394 {0x0b62, 0x0b63}, 2395 {0x0b82, 0x0b82}, 2396 {0x0bbe, 0x0bc2}, 2397 {0x0bc6, 0x0bc8}, 2398 {0x0bca, 0x0bcd}, 2399 {0x0bd7, 0x0bd7}, 2400 {0x0c00, 0x0c04}, 2401 {0x0c3e, 0x0c44}, 2402 {0x0c46, 0x0c48}, 2403 {0x0c4a, 0x0c4d}, 2404 {0x0c55, 0x0c56}, 2405 {0x0c62, 0x0c63}, 2406 {0x0c81, 0x0c83}, 2407 {0x0cbc, 0x0cbc}, 2408 {0x0cbe, 0x0cc4}, 2409 {0x0cc6, 0x0cc8}, 2410 {0x0cca, 0x0ccd}, 2411 {0x0cd5, 0x0cd6}, 2412 {0x0ce2, 0x0ce3}, 2413 {0x0d00, 0x0d03}, 2414 {0x0d3b, 0x0d3c}, 2415 {0x0d3e, 0x0d44}, 2416 {0x0d46, 0x0d48}, 2417 {0x0d4a, 0x0d4d}, 2418 {0x0d57, 0x0d57}, 2419 {0x0d62, 0x0d63}, 2420 {0x0d82, 0x0d83}, 2421 {0x0dca, 0x0dca}, 2422 {0x0dcf, 0x0dd4}, 2423 {0x0dd6, 0x0dd6}, 2424 {0x0dd8, 0x0ddf}, 2425 {0x0df2, 0x0df3}, 2426 {0x0e31, 0x0e31}, 2427 {0x0e34, 0x0e3a}, 2428 {0x0e47, 0x0e4e}, 2429 {0x0eb1, 0x0eb1}, 2430 {0x0eb4, 0x0eb9}, 2431 {0x0ebb, 0x0ebc}, 2432 {0x0ec8, 0x0ecd}, 2433 {0x0f18, 0x0f19}, 2434 {0x0f35, 0x0f35}, 2435 {0x0f37, 0x0f37}, 2436 {0x0f39, 0x0f39}, 2437 {0x0f3e, 0x0f3f}, 2438 {0x0f71, 0x0f84}, 2439 {0x0f86, 0x0f87}, 2440 {0x0f8d, 0x0f97}, 2441 {0x0f99, 0x0fbc}, 2442 {0x0fc6, 0x0fc6}, 2443 {0x102b, 0x103e}, 2444 {0x1056, 0x1059}, 2445 {0x105e, 0x1060}, 2446 {0x1062, 0x1064}, 2447 {0x1067, 0x106d}, 2448 {0x1071, 0x1074}, 2449 {0x1082, 0x108d}, 2450 {0x108f, 0x108f}, 2451 {0x109a, 0x109d}, 2452 {0x135d, 0x135f}, 2453 {0x1712, 0x1714}, 2454 {0x1732, 0x1734}, 2455 {0x1752, 0x1753}, 2456 {0x1772, 0x1773}, 2457 {0x17b4, 0x17d3}, 2458 {0x17dd, 0x17dd}, 2459 {0x180b, 0x180d}, 2460 {0x1885, 0x1886}, 2461 {0x18a9, 0x18a9}, 2462 {0x1920, 0x192b}, 2463 {0x1930, 0x193b}, 2464 {0x1a17, 0x1a1b}, 2465 {0x1a55, 0x1a5e}, 2466 {0x1a60, 0x1a7c}, 2467 {0x1a7f, 0x1a7f}, 2468 {0x1ab0, 0x1abe}, 2469 {0x1b00, 0x1b04}, 2470 {0x1b34, 0x1b44}, 2471 {0x1b6b, 0x1b73}, 2472 {0x1b80, 0x1b82}, 2473 {0x1ba1, 0x1bad}, 2474 {0x1be6, 0x1bf3}, 2475 {0x1c24, 0x1c37}, 2476 {0x1cd0, 0x1cd2}, 2477 {0x1cd4, 0x1ce8}, 2478 {0x1ced, 0x1ced}, 2479 {0x1cf2, 0x1cf4}, 2480 {0x1cf7, 0x1cf9}, 2481 {0x1dc0, 0x1df9}, 2482 {0x1dfb, 0x1dff}, 2483 {0x20d0, 0x20f0}, 2484 {0x2cef, 0x2cf1}, 2485 {0x2d7f, 0x2d7f}, 2486 {0x2de0, 0x2dff}, 2487 {0x302a, 0x302f}, 2488 {0x3099, 0x309a}, 2489 {0xa66f, 0xa672}, 2490 {0xa674, 0xa67d}, 2491 {0xa69e, 0xa69f}, 2492 {0xa6f0, 0xa6f1}, 2493 {0xa802, 0xa802}, 2494 {0xa806, 0xa806}, 2495 {0xa80b, 0xa80b}, 2496 {0xa823, 0xa827}, 2497 {0xa880, 0xa881}, 2498 {0xa8b4, 0xa8c5}, 2499 {0xa8e0, 0xa8f1}, 2500 {0xa8ff, 0xa8ff}, 2501 {0xa926, 0xa92d}, 2502 {0xa947, 0xa953}, 2503 {0xa980, 0xa983}, 2504 {0xa9b3, 0xa9c0}, 2505 {0xa9e5, 0xa9e5}, 2506 {0xaa29, 0xaa36}, 2507 {0xaa43, 0xaa43}, 2508 {0xaa4c, 0xaa4d}, 2509 {0xaa7b, 0xaa7d}, 2510 {0xaab0, 0xaab0}, 2511 {0xaab2, 0xaab4}, 2512 {0xaab7, 0xaab8}, 2513 {0xaabe, 0xaabf}, 2514 {0xaac1, 0xaac1}, 2515 {0xaaeb, 0xaaef}, 2516 {0xaaf5, 0xaaf6}, 2517 {0xabe3, 0xabea}, 2518 {0xabec, 0xabed}, 2519 {0xfb1e, 0xfb1e}, 2520 {0xfe00, 0xfe0f}, 2521 {0xfe20, 0xfe2f}, 2522 {0x101fd, 0x101fd}, 2523 {0x102e0, 0x102e0}, 2524 {0x10376, 0x1037a}, 2525 {0x10a01, 0x10a03}, 2526 {0x10a05, 0x10a06}, 2527 {0x10a0c, 0x10a0f}, 2528 {0x10a38, 0x10a3a}, 2529 {0x10a3f, 0x10a3f}, 2530 {0x10ae5, 0x10ae6}, 2531 {0x10d24, 0x10d27}, 2532 {0x10f46, 0x10f50}, 2533 {0x11000, 0x11002}, 2534 {0x11038, 0x11046}, 2535 {0x1107f, 0x11082}, 2536 {0x110b0, 0x110ba}, 2537 {0x11100, 0x11102}, 2538 {0x11127, 0x11134}, 2539 {0x11145, 0x11146}, 2540 {0x11173, 0x11173}, 2541 {0x11180, 0x11182}, 2542 {0x111b3, 0x111c0}, 2543 {0x111c9, 0x111cc}, 2544 {0x1122c, 0x11237}, 2545 {0x1123e, 0x1123e}, 2546 {0x112df, 0x112ea}, 2547 {0x11300, 0x11303}, 2548 {0x1133b, 0x1133c}, 2549 {0x1133e, 0x11344}, 2550 {0x11347, 0x11348}, 2551 {0x1134b, 0x1134d}, 2552 {0x11357, 0x11357}, 2553 {0x11362, 0x11363}, 2554 {0x11366, 0x1136c}, 2555 {0x11370, 0x11374}, 2556 {0x11435, 0x11446}, 2557 {0x1145e, 0x1145e}, 2558 {0x114b0, 0x114c3}, 2559 {0x115af, 0x115b5}, 2560 {0x115b8, 0x115c0}, 2561 {0x115dc, 0x115dd}, 2562 {0x11630, 0x11640}, 2563 {0x116ab, 0x116b7}, 2564 {0x1171d, 0x1172b}, 2565 {0x1182c, 0x1183a}, 2566 {0x11a01, 0x11a0a}, 2567 {0x11a33, 0x11a39}, 2568 {0x11a3b, 0x11a3e}, 2569 {0x11a47, 0x11a47}, 2570 {0x11a51, 0x11a5b}, 2571 {0x11a8a, 0x11a99}, 2572 {0x11c2f, 0x11c36}, 2573 {0x11c38, 0x11c3f}, 2574 {0x11c92, 0x11ca7}, 2575 {0x11ca9, 0x11cb6}, 2576 {0x11d31, 0x11d36}, 2577 {0x11d3a, 0x11d3a}, 2578 {0x11d3c, 0x11d3d}, 2579 {0x11d3f, 0x11d45}, 2580 {0x11d47, 0x11d47}, 2581 {0x11d8a, 0x11d8e}, 2582 {0x11d90, 0x11d91}, 2583 {0x11d93, 0x11d97}, 2584 {0x11ef3, 0x11ef6}, 2585 {0x16af0, 0x16af4}, 2586 {0x16b30, 0x16b36}, 2587 {0x16f51, 0x16f7e}, 2588 {0x16f8f, 0x16f92}, 2589 {0x1bc9d, 0x1bc9e}, 2590 {0x1d165, 0x1d169}, 2591 {0x1d16d, 0x1d172}, 2592 {0x1d17b, 0x1d182}, 2593 {0x1d185, 0x1d18b}, 2594 {0x1d1aa, 0x1d1ad}, 2595 {0x1d242, 0x1d244}, 2596 {0x1da00, 0x1da36}, 2597 {0x1da3b, 0x1da6c}, 2598 {0x1da75, 0x1da75}, 2599 {0x1da84, 0x1da84}, 2600 {0x1da9b, 0x1da9f}, 2601 {0x1daa1, 0x1daaf}, 2602 {0x1e000, 0x1e006}, 2603 {0x1e008, 0x1e018}, 2604 {0x1e01b, 0x1e021}, 2605 {0x1e023, 0x1e024}, 2606 {0x1e026, 0x1e02a}, 2607 {0x1e8d0, 0x1e8d6}, 2608 {0x1e944, 0x1e94a}, 2609 {0xe0100, 0xe01ef} 2610 }; 2611 2612 return intable(combining, sizeof(combining), c); 2613 } 2614 2615 /* 2616 * Return TRUE for characters that can be displayed in a normal way. 2617 * Only for characters of 0x100 and above! 2618 */ 2619 int 2620 utf_printable(int c) 2621 { 2622 #ifdef USE_WCHAR_FUNCTIONS 2623 /* 2624 * Assume the iswprint() library function works better than our own stuff. 2625 */ 2626 return iswprint(c); 2627 #else 2628 /* Sorted list of non-overlapping intervals. 2629 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */ 2630 static struct interval nonprint[] = 2631 { 2632 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2633 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2634 {0xfffe, 0xffff} 2635 }; 2636 2637 return !intable(nonprint, sizeof(nonprint), c); 2638 #endif 2639 } 2640 2641 /* Sorted list of non-overlapping intervals of all Emoji characters, 2642 * based on http://unicode.org/emoji/charts/emoji-list.html */ 2643 static struct interval emoji_all[] = 2644 { 2645 {0x203c, 0x203c}, 2646 {0x2049, 0x2049}, 2647 {0x2122, 0x2122}, 2648 {0x2139, 0x2139}, 2649 {0x2194, 0x2199}, 2650 {0x21a9, 0x21aa}, 2651 {0x231a, 0x231b}, 2652 {0x2328, 0x2328}, 2653 {0x23cf, 0x23cf}, 2654 {0x23e9, 0x23f3}, 2655 {0x23f8, 0x23fa}, 2656 {0x24c2, 0x24c2}, 2657 {0x25aa, 0x25ab}, 2658 {0x25b6, 0x25b6}, 2659 {0x25c0, 0x25c0}, 2660 {0x25fb, 0x25fe}, 2661 {0x2600, 0x2604}, 2662 {0x260e, 0x260e}, 2663 {0x2611, 0x2611}, 2664 {0x2614, 0x2615}, 2665 {0x2618, 0x2618}, 2666 {0x261d, 0x261d}, 2667 {0x2620, 0x2620}, 2668 {0x2622, 0x2623}, 2669 {0x2626, 0x2626}, 2670 {0x262a, 0x262a}, 2671 {0x262e, 0x262f}, 2672 {0x2638, 0x263a}, 2673 {0x2640, 0x2640}, 2674 {0x2642, 0x2642}, 2675 {0x2648, 0x2653}, 2676 {0x265f, 0x2660}, 2677 {0x2663, 0x2663}, 2678 {0x2665, 0x2666}, 2679 {0x2668, 0x2668}, 2680 {0x267b, 0x267b}, 2681 {0x267e, 0x267f}, 2682 {0x2692, 0x2697}, 2683 {0x2699, 0x2699}, 2684 {0x269b, 0x269c}, 2685 {0x26a0, 0x26a1}, 2686 {0x26aa, 0x26ab}, 2687 {0x26b0, 0x26b1}, 2688 {0x26bd, 0x26be}, 2689 {0x26c4, 0x26c5}, 2690 {0x26c8, 0x26c8}, 2691 {0x26ce, 0x26cf}, 2692 {0x26d1, 0x26d1}, 2693 {0x26d3, 0x26d4}, 2694 {0x26e9, 0x26ea}, 2695 {0x26f0, 0x26f5}, 2696 {0x26f7, 0x26fa}, 2697 {0x26fd, 0x26fd}, 2698 {0x2702, 0x2702}, 2699 {0x2705, 0x2705}, 2700 {0x2708, 0x270d}, 2701 {0x270f, 0x270f}, 2702 {0x2712, 0x2712}, 2703 {0x2714, 0x2714}, 2704 {0x2716, 0x2716}, 2705 {0x271d, 0x271d}, 2706 {0x2721, 0x2721}, 2707 {0x2728, 0x2728}, 2708 {0x2733, 0x2734}, 2709 {0x2744, 0x2744}, 2710 {0x2747, 0x2747}, 2711 {0x274c, 0x274c}, 2712 {0x274e, 0x274e}, 2713 {0x2753, 0x2755}, 2714 {0x2757, 0x2757}, 2715 {0x2763, 0x2764}, 2716 {0x2795, 0x2797}, 2717 {0x27a1, 0x27a1}, 2718 {0x27b0, 0x27b0}, 2719 {0x27bf, 0x27bf}, 2720 {0x2934, 0x2935}, 2721 {0x2b05, 0x2b07}, 2722 {0x2b1b, 0x2b1c}, 2723 {0x2b50, 0x2b50}, 2724 {0x2b55, 0x2b55}, 2725 {0x3030, 0x3030}, 2726 {0x303d, 0x303d}, 2727 {0x3297, 0x3297}, 2728 {0x3299, 0x3299}, 2729 {0x1f004, 0x1f004}, 2730 {0x1f0cf, 0x1f0cf}, 2731 {0x1f170, 0x1f171}, 2732 {0x1f17e, 0x1f17f}, 2733 {0x1f18e, 0x1f18e}, 2734 {0x1f191, 0x1f19a}, 2735 {0x1f1e6, 0x1f1ff}, 2736 {0x1f201, 0x1f202}, 2737 {0x1f21a, 0x1f21a}, 2738 {0x1f22f, 0x1f22f}, 2739 {0x1f232, 0x1f23a}, 2740 {0x1f250, 0x1f251}, 2741 {0x1f300, 0x1f321}, 2742 {0x1f324, 0x1f393}, 2743 {0x1f396, 0x1f397}, 2744 {0x1f399, 0x1f39b}, 2745 {0x1f39e, 0x1f3f0}, 2746 {0x1f3f3, 0x1f3f5}, 2747 {0x1f3f7, 0x1f4fd}, 2748 {0x1f4ff, 0x1f53d}, 2749 {0x1f549, 0x1f54e}, 2750 {0x1f550, 0x1f567}, 2751 {0x1f56f, 0x1f570}, 2752 {0x1f573, 0x1f57a}, 2753 {0x1f587, 0x1f587}, 2754 {0x1f58a, 0x1f58d}, 2755 {0x1f590, 0x1f590}, 2756 {0x1f595, 0x1f596}, 2757 {0x1f5a4, 0x1f5a5}, 2758 {0x1f5a8, 0x1f5a8}, 2759 {0x1f5b1, 0x1f5b2}, 2760 {0x1f5bc, 0x1f5bc}, 2761 {0x1f5c2, 0x1f5c4}, 2762 {0x1f5d1, 0x1f5d3}, 2763 {0x1f5dc, 0x1f5de}, 2764 {0x1f5e1, 0x1f5e1}, 2765 {0x1f5e3, 0x1f5e3}, 2766 {0x1f5e8, 0x1f5e8}, 2767 {0x1f5ef, 0x1f5ef}, 2768 {0x1f5f3, 0x1f5f3}, 2769 {0x1f5fa, 0x1f64f}, 2770 {0x1f680, 0x1f6c5}, 2771 {0x1f6cb, 0x1f6d2}, 2772 {0x1f6e0, 0x1f6e5}, 2773 {0x1f6e9, 0x1f6e9}, 2774 {0x1f6eb, 0x1f6ec}, 2775 {0x1f6f0, 0x1f6f0}, 2776 {0x1f6f3, 0x1f6f9}, 2777 {0x1f910, 0x1f93a}, 2778 {0x1f93c, 0x1f93e}, 2779 {0x1f940, 0x1f945}, 2780 {0x1f947, 0x1f970}, 2781 {0x1f973, 0x1f976}, 2782 {0x1f97a, 0x1f97a}, 2783 {0x1f97c, 0x1f9a2}, 2784 {0x1f9b0, 0x1f9b9}, 2785 {0x1f9c0, 0x1f9c2}, 2786 {0x1f9d0, 0x1f9ff} 2787 }; 2788 2789 /* 2790 * Get class of a Unicode character. 2791 * 0: white space 2792 * 1: punctuation 2793 * 2 or bigger: some class of word character. 2794 */ 2795 int 2796 utf_class(int c) 2797 { 2798 return utf_class_buf(c, curbuf); 2799 } 2800 2801 int 2802 utf_class_buf(int c, buf_T *buf) 2803 { 2804 /* sorted list of non-overlapping intervals */ 2805 static struct clinterval 2806 { 2807 unsigned int first; 2808 unsigned int last; 2809 unsigned int class; 2810 } classes[] = 2811 { 2812 {0x037e, 0x037e, 1}, /* Greek question mark */ 2813 {0x0387, 0x0387, 1}, /* Greek ano teleia */ 2814 {0x055a, 0x055f, 1}, /* Armenian punctuation */ 2815 {0x0589, 0x0589, 1}, /* Armenian full stop */ 2816 {0x05be, 0x05be, 1}, 2817 {0x05c0, 0x05c0, 1}, 2818 {0x05c3, 0x05c3, 1}, 2819 {0x05f3, 0x05f4, 1}, 2820 {0x060c, 0x060c, 1}, 2821 {0x061b, 0x061b, 1}, 2822 {0x061f, 0x061f, 1}, 2823 {0x066a, 0x066d, 1}, 2824 {0x06d4, 0x06d4, 1}, 2825 {0x0700, 0x070d, 1}, /* Syriac punctuation */ 2826 {0x0964, 0x0965, 1}, 2827 {0x0970, 0x0970, 1}, 2828 {0x0df4, 0x0df4, 1}, 2829 {0x0e4f, 0x0e4f, 1}, 2830 {0x0e5a, 0x0e5b, 1}, 2831 {0x0f04, 0x0f12, 1}, 2832 {0x0f3a, 0x0f3d, 1}, 2833 {0x0f85, 0x0f85, 1}, 2834 {0x104a, 0x104f, 1}, /* Myanmar punctuation */ 2835 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */ 2836 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */ 2837 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */ 2838 {0x1680, 0x1680, 0}, 2839 {0x169b, 0x169c, 1}, 2840 {0x16eb, 0x16ed, 1}, 2841 {0x1735, 0x1736, 1}, 2842 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */ 2843 {0x1800, 0x180a, 1}, /* Mongolian punctuation */ 2844 {0x2000, 0x200b, 0}, /* spaces */ 2845 {0x200c, 0x2027, 1}, /* punctuation and symbols */ 2846 {0x2028, 0x2029, 0}, 2847 {0x202a, 0x202e, 1}, /* punctuation and symbols */ 2848 {0x202f, 0x202f, 0}, 2849 {0x2030, 0x205e, 1}, /* punctuation and symbols */ 2850 {0x205f, 0x205f, 0}, 2851 {0x2060, 0x27ff, 1}, /* punctuation and symbols */ 2852 {0x2070, 0x207f, 0x2070}, /* superscript */ 2853 {0x2080, 0x2094, 0x2080}, /* subscript */ 2854 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */ 2855 {0x2800, 0x28ff, 0x2800}, /* braille */ 2856 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */ 2857 {0x29d8, 0x29db, 1}, 2858 {0x29fc, 0x29fd, 1}, 2859 {0x2e00, 0x2e7f, 1}, /* supplemental punctuation */ 2860 {0x3000, 0x3000, 0}, /* ideographic space */ 2861 {0x3001, 0x3020, 1}, /* ideographic punctuation */ 2862 {0x3030, 0x3030, 1}, 2863 {0x303d, 0x303d, 1}, 2864 {0x3040, 0x309f, 0x3040}, /* Hiragana */ 2865 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */ 2866 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */ 2867 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */ 2868 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */ 2869 {0xfd3e, 0xfd3f, 1}, 2870 {0xfe30, 0xfe6b, 1}, /* punctuation forms */ 2871 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */ 2872 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */ 2873 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */ 2874 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */ 2875 {0x20000, 0x2a6df, 0x4e00}, /* CJK Ideographs */ 2876 {0x2a700, 0x2b73f, 0x4e00}, /* CJK Ideographs */ 2877 {0x2b740, 0x2b81f, 0x4e00}, /* CJK Ideographs */ 2878 {0x2f800, 0x2fa1f, 0x4e00}, /* CJK Ideographs */ 2879 }; 2880 2881 int bot = 0; 2882 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2883 int mid; 2884 2885 /* First quick check for Latin1 characters, use 'iskeyword'. */ 2886 if (c < 0x100) 2887 { 2888 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2889 return 0; /* blank */ 2890 if (vim_iswordc_buf(c, buf)) 2891 return 2; /* word character */ 2892 return 1; /* punctuation */ 2893 } 2894 2895 /* binary search in table */ 2896 while (top >= bot) 2897 { 2898 mid = (bot + top) / 2; 2899 if (classes[mid].last < (unsigned int)c) 2900 bot = mid + 1; 2901 else if (classes[mid].first > (unsigned int)c) 2902 top = mid - 1; 2903 else 2904 return (int)classes[mid].class; 2905 } 2906 2907 /* emoji */ 2908 if (intable(emoji_all, sizeof(emoji_all), c)) 2909 return 3; 2910 2911 /* most other characters are "word" characters */ 2912 return 2; 2913 } 2914 2915 int 2916 utf_ambiguous_width(int c) 2917 { 2918 return c >= 0x80 && (intable(ambiguous, sizeof(ambiguous), c) 2919 || intable(emoji_all, sizeof(emoji_all), c)); 2920 } 2921 2922 /* 2923 * Code for Unicode case-dependent operations. Based on notes in 2924 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2925 * This code uses simple case folding, not full case folding. 2926 * Last updated for Unicode 5.2. 2927 */ 2928 2929 /* 2930 * The following tables are built by ../runtime/tools/unicode.vim. 2931 * They must be in numeric order, because we use binary search. 2932 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2933 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2934 * folded/upper/lower by adding 32. 2935 */ 2936 typedef struct 2937 { 2938 int rangeStart; 2939 int rangeEnd; 2940 int step; 2941 int offset; 2942 } convertStruct; 2943 2944 static convertStruct foldCase[] = 2945 { 2946 {0x41,0x5a,1,32}, 2947 {0xb5,0xb5,-1,775}, 2948 {0xc0,0xd6,1,32}, 2949 {0xd8,0xde,1,32}, 2950 {0x100,0x12e,2,1}, 2951 {0x132,0x136,2,1}, 2952 {0x139,0x147,2,1}, 2953 {0x14a,0x176,2,1}, 2954 {0x178,0x178,-1,-121}, 2955 {0x179,0x17d,2,1}, 2956 {0x17f,0x17f,-1,-268}, 2957 {0x181,0x181,-1,210}, 2958 {0x182,0x184,2,1}, 2959 {0x186,0x186,-1,206}, 2960 {0x187,0x187,-1,1}, 2961 {0x189,0x18a,1,205}, 2962 {0x18b,0x18b,-1,1}, 2963 {0x18e,0x18e,-1,79}, 2964 {0x18f,0x18f,-1,202}, 2965 {0x190,0x190,-1,203}, 2966 {0x191,0x191,-1,1}, 2967 {0x193,0x193,-1,205}, 2968 {0x194,0x194,-1,207}, 2969 {0x196,0x196,-1,211}, 2970 {0x197,0x197,-1,209}, 2971 {0x198,0x198,-1,1}, 2972 {0x19c,0x19c,-1,211}, 2973 {0x19d,0x19d,-1,213}, 2974 {0x19f,0x19f,-1,214}, 2975 {0x1a0,0x1a4,2,1}, 2976 {0x1a6,0x1a6,-1,218}, 2977 {0x1a7,0x1a7,-1,1}, 2978 {0x1a9,0x1a9,-1,218}, 2979 {0x1ac,0x1ac,-1,1}, 2980 {0x1ae,0x1ae,-1,218}, 2981 {0x1af,0x1af,-1,1}, 2982 {0x1b1,0x1b2,1,217}, 2983 {0x1b3,0x1b5,2,1}, 2984 {0x1b7,0x1b7,-1,219}, 2985 {0x1b8,0x1bc,4,1}, 2986 {0x1c4,0x1c4,-1,2}, 2987 {0x1c5,0x1c5,-1,1}, 2988 {0x1c7,0x1c7,-1,2}, 2989 {0x1c8,0x1c8,-1,1}, 2990 {0x1ca,0x1ca,-1,2}, 2991 {0x1cb,0x1db,2,1}, 2992 {0x1de,0x1ee,2,1}, 2993 {0x1f1,0x1f1,-1,2}, 2994 {0x1f2,0x1f4,2,1}, 2995 {0x1f6,0x1f6,-1,-97}, 2996 {0x1f7,0x1f7,-1,-56}, 2997 {0x1f8,0x21e,2,1}, 2998 {0x220,0x220,-1,-130}, 2999 {0x222,0x232,2,1}, 3000 {0x23a,0x23a,-1,10795}, 3001 {0x23b,0x23b,-1,1}, 3002 {0x23d,0x23d,-1,-163}, 3003 {0x23e,0x23e,-1,10792}, 3004 {0x241,0x241,-1,1}, 3005 {0x243,0x243,-1,-195}, 3006 {0x244,0x244,-1,69}, 3007 {0x245,0x245,-1,71}, 3008 {0x246,0x24e,2,1}, 3009 {0x345,0x345,-1,116}, 3010 {0x370,0x372,2,1}, 3011 {0x376,0x376,-1,1}, 3012 {0x37f,0x37f,-1,116}, 3013 {0x386,0x386,-1,38}, 3014 {0x388,0x38a,1,37}, 3015 {0x38c,0x38c,-1,64}, 3016 {0x38e,0x38f,1,63}, 3017 {0x391,0x3a1,1,32}, 3018 {0x3a3,0x3ab,1,32}, 3019 {0x3c2,0x3c2,-1,1}, 3020 {0x3cf,0x3cf,-1,8}, 3021 {0x3d0,0x3d0,-1,-30}, 3022 {0x3d1,0x3d1,-1,-25}, 3023 {0x3d5,0x3d5,-1,-15}, 3024 {0x3d6,0x3d6,-1,-22}, 3025 {0x3d8,0x3ee,2,1}, 3026 {0x3f0,0x3f0,-1,-54}, 3027 {0x3f1,0x3f1,-1,-48}, 3028 {0x3f4,0x3f4,-1,-60}, 3029 {0x3f5,0x3f5,-1,-64}, 3030 {0x3f7,0x3f7,-1,1}, 3031 {0x3f9,0x3f9,-1,-7}, 3032 {0x3fa,0x3fa,-1,1}, 3033 {0x3fd,0x3ff,1,-130}, 3034 {0x400,0x40f,1,80}, 3035 {0x410,0x42f,1,32}, 3036 {0x460,0x480,2,1}, 3037 {0x48a,0x4be,2,1}, 3038 {0x4c0,0x4c0,-1,15}, 3039 {0x4c1,0x4cd,2,1}, 3040 {0x4d0,0x52e,2,1}, 3041 {0x531,0x556,1,48}, 3042 {0x10a0,0x10c5,1,7264}, 3043 {0x10c7,0x10cd,6,7264}, 3044 {0x13f8,0x13fd,1,-8}, 3045 {0x1c80,0x1c80,-1,-6222}, 3046 {0x1c81,0x1c81,-1,-6221}, 3047 {0x1c82,0x1c82,-1,-6212}, 3048 {0x1c83,0x1c84,1,-6210}, 3049 {0x1c85,0x1c85,-1,-6211}, 3050 {0x1c86,0x1c86,-1,-6204}, 3051 {0x1c87,0x1c87,-1,-6180}, 3052 {0x1c88,0x1c88,-1,35267}, 3053 {0x1c90,0x1cba,1,-3008}, 3054 {0x1cbd,0x1cbf,1,-3008}, 3055 {0x1e00,0x1e94,2,1}, 3056 {0x1e9b,0x1e9b,-1,-58}, 3057 {0x1e9e,0x1e9e,-1,-7615}, 3058 {0x1ea0,0x1efe,2,1}, 3059 {0x1f08,0x1f0f,1,-8}, 3060 {0x1f18,0x1f1d,1,-8}, 3061 {0x1f28,0x1f2f,1,-8}, 3062 {0x1f38,0x1f3f,1,-8}, 3063 {0x1f48,0x1f4d,1,-8}, 3064 {0x1f59,0x1f5f,2,-8}, 3065 {0x1f68,0x1f6f,1,-8}, 3066 {0x1f88,0x1f8f,1,-8}, 3067 {0x1f98,0x1f9f,1,-8}, 3068 {0x1fa8,0x1faf,1,-8}, 3069 {0x1fb8,0x1fb9,1,-8}, 3070 {0x1fba,0x1fbb,1,-74}, 3071 {0x1fbc,0x1fbc,-1,-9}, 3072 {0x1fbe,0x1fbe,-1,-7173}, 3073 {0x1fc8,0x1fcb,1,-86}, 3074 {0x1fcc,0x1fcc,-1,-9}, 3075 {0x1fd8,0x1fd9,1,-8}, 3076 {0x1fda,0x1fdb,1,-100}, 3077 {0x1fe8,0x1fe9,1,-8}, 3078 {0x1fea,0x1feb,1,-112}, 3079 {0x1fec,0x1fec,-1,-7}, 3080 {0x1ff8,0x1ff9,1,-128}, 3081 {0x1ffa,0x1ffb,1,-126}, 3082 {0x1ffc,0x1ffc,-1,-9}, 3083 {0x2126,0x2126,-1,-7517}, 3084 {0x212a,0x212a,-1,-8383}, 3085 {0x212b,0x212b,-1,-8262}, 3086 {0x2132,0x2132,-1,28}, 3087 {0x2160,0x216f,1,16}, 3088 {0x2183,0x2183,-1,1}, 3089 {0x24b6,0x24cf,1,26}, 3090 {0x2c00,0x2c2e,1,48}, 3091 {0x2c60,0x2c60,-1,1}, 3092 {0x2c62,0x2c62,-1,-10743}, 3093 {0x2c63,0x2c63,-1,-3814}, 3094 {0x2c64,0x2c64,-1,-10727}, 3095 {0x2c67,0x2c6b,2,1}, 3096 {0x2c6d,0x2c6d,-1,-10780}, 3097 {0x2c6e,0x2c6e,-1,-10749}, 3098 {0x2c6f,0x2c6f,-1,-10783}, 3099 {0x2c70,0x2c70,-1,-10782}, 3100 {0x2c72,0x2c75,3,1}, 3101 {0x2c7e,0x2c7f,1,-10815}, 3102 {0x2c80,0x2ce2,2,1}, 3103 {0x2ceb,0x2ced,2,1}, 3104 {0x2cf2,0xa640,31054,1}, 3105 {0xa642,0xa66c,2,1}, 3106 {0xa680,0xa69a,2,1}, 3107 {0xa722,0xa72e,2,1}, 3108 {0xa732,0xa76e,2,1}, 3109 {0xa779,0xa77b,2,1}, 3110 {0xa77d,0xa77d,-1,-35332}, 3111 {0xa77e,0xa786,2,1}, 3112 {0xa78b,0xa78b,-1,1}, 3113 {0xa78d,0xa78d,-1,-42280}, 3114 {0xa790,0xa792,2,1}, 3115 {0xa796,0xa7a8,2,1}, 3116 {0xa7aa,0xa7aa,-1,-42308}, 3117 {0xa7ab,0xa7ab,-1,-42319}, 3118 {0xa7ac,0xa7ac,-1,-42315}, 3119 {0xa7ad,0xa7ad,-1,-42305}, 3120 {0xa7ae,0xa7ae,-1,-42308}, 3121 {0xa7b0,0xa7b0,-1,-42258}, 3122 {0xa7b1,0xa7b1,-1,-42282}, 3123 {0xa7b2,0xa7b2,-1,-42261}, 3124 {0xa7b3,0xa7b3,-1,928}, 3125 {0xa7b4,0xa7b8,2,1}, 3126 {0xab70,0xabbf,1,-38864}, 3127 {0xff21,0xff3a,1,32}, 3128 {0x10400,0x10427,1,40}, 3129 {0x104b0,0x104d3,1,40}, 3130 {0x10c80,0x10cb2,1,64}, 3131 {0x118a0,0x118bf,1,32}, 3132 {0x16e40,0x16e5f,1,32}, 3133 {0x1e900,0x1e921,1,34} 3134 }; 3135 3136 static int utf_convert(int a, convertStruct table[], int tableSize); 3137 static int utf_strnicmp(char_u *s1, char_u *s2, size_t n1, size_t n2); 3138 3139 /* 3140 * Generic conversion function for case operations. 3141 * Return the converted equivalent of "a", which is a UCS-4 character. Use 3142 * the given conversion "table". Uses binary search on "table". 3143 */ 3144 static int 3145 utf_convert( 3146 int a, 3147 convertStruct table[], 3148 int tableSize) 3149 { 3150 int start, mid, end; /* indices into table */ 3151 int entries = tableSize / sizeof(convertStruct); 3152 3153 start = 0; 3154 end = entries; 3155 while (start < end) 3156 { 3157 /* need to search further */ 3158 mid = (end + start) / 2; 3159 if (table[mid].rangeEnd < a) 3160 start = mid + 1; 3161 else 3162 end = mid; 3163 } 3164 if (start < entries 3165 && table[start].rangeStart <= a 3166 && a <= table[start].rangeEnd 3167 && (a - table[start].rangeStart) % table[start].step == 0) 3168 return (a + table[start].offset); 3169 else 3170 return a; 3171 } 3172 3173 /* 3174 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 3175 * simple case folding. 3176 */ 3177 int 3178 utf_fold(int a) 3179 { 3180 if (a < 0x80) 3181 /* be fast for ASCII */ 3182 return a >= 0x41 && a <= 0x5a ? a + 32 : a; 3183 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 3184 } 3185 3186 static convertStruct toLower[] = 3187 { 3188 {0x41,0x5a,1,32}, 3189 {0xc0,0xd6,1,32}, 3190 {0xd8,0xde,1,32}, 3191 {0x100,0x12e,2,1}, 3192 {0x130,0x130,-1,-199}, 3193 {0x132,0x136,2,1}, 3194 {0x139,0x147,2,1}, 3195 {0x14a,0x176,2,1}, 3196 {0x178,0x178,-1,-121}, 3197 {0x179,0x17d,2,1}, 3198 {0x181,0x181,-1,210}, 3199 {0x182,0x184,2,1}, 3200 {0x186,0x186,-1,206}, 3201 {0x187,0x187,-1,1}, 3202 {0x189,0x18a,1,205}, 3203 {0x18b,0x18b,-1,1}, 3204 {0x18e,0x18e,-1,79}, 3205 {0x18f,0x18f,-1,202}, 3206 {0x190,0x190,-1,203}, 3207 {0x191,0x191,-1,1}, 3208 {0x193,0x193,-1,205}, 3209 {0x194,0x194,-1,207}, 3210 {0x196,0x196,-1,211}, 3211 {0x197,0x197,-1,209}, 3212 {0x198,0x198,-1,1}, 3213 {0x19c,0x19c,-1,211}, 3214 {0x19d,0x19d,-1,213}, 3215 {0x19f,0x19f,-1,214}, 3216 {0x1a0,0x1a4,2,1}, 3217 {0x1a6,0x1a6,-1,218}, 3218 {0x1a7,0x1a7,-1,1}, 3219 {0x1a9,0x1a9,-1,218}, 3220 {0x1ac,0x1ac,-1,1}, 3221 {0x1ae,0x1ae,-1,218}, 3222 {0x1af,0x1af,-1,1}, 3223 {0x1b1,0x1b2,1,217}, 3224 {0x1b3,0x1b5,2,1}, 3225 {0x1b7,0x1b7,-1,219}, 3226 {0x1b8,0x1bc,4,1}, 3227 {0x1c4,0x1c4,-1,2}, 3228 {0x1c5,0x1c5,-1,1}, 3229 {0x1c7,0x1c7,-1,2}, 3230 {0x1c8,0x1c8,-1,1}, 3231 {0x1ca,0x1ca,-1,2}, 3232 {0x1cb,0x1db,2,1}, 3233 {0x1de,0x1ee,2,1}, 3234 {0x1f1,0x1f1,-1,2}, 3235 {0x1f2,0x1f4,2,1}, 3236 {0x1f6,0x1f6,-1,-97}, 3237 {0x1f7,0x1f7,-1,-56}, 3238 {0x1f8,0x21e,2,1}, 3239 {0x220,0x220,-1,-130}, 3240 {0x222,0x232,2,1}, 3241 {0x23a,0x23a,-1,10795}, 3242 {0x23b,0x23b,-1,1}, 3243 {0x23d,0x23d,-1,-163}, 3244 {0x23e,0x23e,-1,10792}, 3245 {0x241,0x241,-1,1}, 3246 {0x243,0x243,-1,-195}, 3247 {0x244,0x244,-1,69}, 3248 {0x245,0x245,-1,71}, 3249 {0x246,0x24e,2,1}, 3250 {0x370,0x372,2,1}, 3251 {0x376,0x376,-1,1}, 3252 {0x37f,0x37f,-1,116}, 3253 {0x386,0x386,-1,38}, 3254 {0x388,0x38a,1,37}, 3255 {0x38c,0x38c,-1,64}, 3256 {0x38e,0x38f,1,63}, 3257 {0x391,0x3a1,1,32}, 3258 {0x3a3,0x3ab,1,32}, 3259 {0x3cf,0x3cf,-1,8}, 3260 {0x3d8,0x3ee,2,1}, 3261 {0x3f4,0x3f4,-1,-60}, 3262 {0x3f7,0x3f7,-1,1}, 3263 {0x3f9,0x3f9,-1,-7}, 3264 {0x3fa,0x3fa,-1,1}, 3265 {0x3fd,0x3ff,1,-130}, 3266 {0x400,0x40f,1,80}, 3267 {0x410,0x42f,1,32}, 3268 {0x460,0x480,2,1}, 3269 {0x48a,0x4be,2,1}, 3270 {0x4c0,0x4c0,-1,15}, 3271 {0x4c1,0x4cd,2,1}, 3272 {0x4d0,0x52e,2,1}, 3273 {0x531,0x556,1,48}, 3274 {0x10a0,0x10c5,1,7264}, 3275 {0x10c7,0x10cd,6,7264}, 3276 {0x13a0,0x13ef,1,38864}, 3277 {0x13f0,0x13f5,1,8}, 3278 {0x1c90,0x1cba,1,-3008}, 3279 {0x1cbd,0x1cbf,1,-3008}, 3280 {0x1e00,0x1e94,2,1}, 3281 {0x1e9e,0x1e9e,-1,-7615}, 3282 {0x1ea0,0x1efe,2,1}, 3283 {0x1f08,0x1f0f,1,-8}, 3284 {0x1f18,0x1f1d,1,-8}, 3285 {0x1f28,0x1f2f,1,-8}, 3286 {0x1f38,0x1f3f,1,-8}, 3287 {0x1f48,0x1f4d,1,-8}, 3288 {0x1f59,0x1f5f,2,-8}, 3289 {0x1f68,0x1f6f,1,-8}, 3290 {0x1f88,0x1f8f,1,-8}, 3291 {0x1f98,0x1f9f,1,-8}, 3292 {0x1fa8,0x1faf,1,-8}, 3293 {0x1fb8,0x1fb9,1,-8}, 3294 {0x1fba,0x1fbb,1,-74}, 3295 {0x1fbc,0x1fbc,-1,-9}, 3296 {0x1fc8,0x1fcb,1,-86}, 3297 {0x1fcc,0x1fcc,-1,-9}, 3298 {0x1fd8,0x1fd9,1,-8}, 3299 {0x1fda,0x1fdb,1,-100}, 3300 {0x1fe8,0x1fe9,1,-8}, 3301 {0x1fea,0x1feb,1,-112}, 3302 {0x1fec,0x1fec,-1,-7}, 3303 {0x1ff8,0x1ff9,1,-128}, 3304 {0x1ffa,0x1ffb,1,-126}, 3305 {0x1ffc,0x1ffc,-1,-9}, 3306 {0x2126,0x2126,-1,-7517}, 3307 {0x212a,0x212a,-1,-8383}, 3308 {0x212b,0x212b,-1,-8262}, 3309 {0x2132,0x2132,-1,28}, 3310 {0x2160,0x216f,1,16}, 3311 {0x2183,0x2183,-1,1}, 3312 {0x24b6,0x24cf,1,26}, 3313 {0x2c00,0x2c2e,1,48}, 3314 {0x2c60,0x2c60,-1,1}, 3315 {0x2c62,0x2c62,-1,-10743}, 3316 {0x2c63,0x2c63,-1,-3814}, 3317 {0x2c64,0x2c64,-1,-10727}, 3318 {0x2c67,0x2c6b,2,1}, 3319 {0x2c6d,0x2c6d,-1,-10780}, 3320 {0x2c6e,0x2c6e,-1,-10749}, 3321 {0x2c6f,0x2c6f,-1,-10783}, 3322 {0x2c70,0x2c70,-1,-10782}, 3323 {0x2c72,0x2c75,3,1}, 3324 {0x2c7e,0x2c7f,1,-10815}, 3325 {0x2c80,0x2ce2,2,1}, 3326 {0x2ceb,0x2ced,2,1}, 3327 {0x2cf2,0xa640,31054,1}, 3328 {0xa642,0xa66c,2,1}, 3329 {0xa680,0xa69a,2,1}, 3330 {0xa722,0xa72e,2,1}, 3331 {0xa732,0xa76e,2,1}, 3332 {0xa779,0xa77b,2,1}, 3333 {0xa77d,0xa77d,-1,-35332}, 3334 {0xa77e,0xa786,2,1}, 3335 {0xa78b,0xa78b,-1,1}, 3336 {0xa78d,0xa78d,-1,-42280}, 3337 {0xa790,0xa792,2,1}, 3338 {0xa796,0xa7a8,2,1}, 3339 {0xa7aa,0xa7aa,-1,-42308}, 3340 {0xa7ab,0xa7ab,-1,-42319}, 3341 {0xa7ac,0xa7ac,-1,-42315}, 3342 {0xa7ad,0xa7ad,-1,-42305}, 3343 {0xa7ae,0xa7ae,-1,-42308}, 3344 {0xa7b0,0xa7b0,-1,-42258}, 3345 {0xa7b1,0xa7b1,-1,-42282}, 3346 {0xa7b2,0xa7b2,-1,-42261}, 3347 {0xa7b3,0xa7b3,-1,928}, 3348 {0xa7b4,0xa7b8,2,1}, 3349 {0xff21,0xff3a,1,32}, 3350 {0x10400,0x10427,1,40}, 3351 {0x104b0,0x104d3,1,40}, 3352 {0x10c80,0x10cb2,1,64}, 3353 {0x118a0,0x118bf,1,32}, 3354 {0x16e40,0x16e5f,1,32}, 3355 {0x1e900,0x1e921,1,34} 3356 }; 3357 3358 static convertStruct toUpper[] = 3359 { 3360 {0x61,0x7a,1,-32}, 3361 {0xb5,0xb5,-1,743}, 3362 {0xe0,0xf6,1,-32}, 3363 {0xf8,0xfe,1,-32}, 3364 {0xff,0xff,-1,121}, 3365 {0x101,0x12f,2,-1}, 3366 {0x131,0x131,-1,-232}, 3367 {0x133,0x137,2,-1}, 3368 {0x13a,0x148,2,-1}, 3369 {0x14b,0x177,2,-1}, 3370 {0x17a,0x17e,2,-1}, 3371 {0x17f,0x17f,-1,-300}, 3372 {0x180,0x180,-1,195}, 3373 {0x183,0x185,2,-1}, 3374 {0x188,0x18c,4,-1}, 3375 {0x192,0x192,-1,-1}, 3376 {0x195,0x195,-1,97}, 3377 {0x199,0x199,-1,-1}, 3378 {0x19a,0x19a,-1,163}, 3379 {0x19e,0x19e,-1,130}, 3380 {0x1a1,0x1a5,2,-1}, 3381 {0x1a8,0x1ad,5,-1}, 3382 {0x1b0,0x1b4,4,-1}, 3383 {0x1b6,0x1b9,3,-1}, 3384 {0x1bd,0x1bd,-1,-1}, 3385 {0x1bf,0x1bf,-1,56}, 3386 {0x1c5,0x1c5,-1,-1}, 3387 {0x1c6,0x1c6,-1,-2}, 3388 {0x1c8,0x1c8,-1,-1}, 3389 {0x1c9,0x1c9,-1,-2}, 3390 {0x1cb,0x1cb,-1,-1}, 3391 {0x1cc,0x1cc,-1,-2}, 3392 {0x1ce,0x1dc,2,-1}, 3393 {0x1dd,0x1dd,-1,-79}, 3394 {0x1df,0x1ef,2,-1}, 3395 {0x1f2,0x1f2,-1,-1}, 3396 {0x1f3,0x1f3,-1,-2}, 3397 {0x1f5,0x1f9,4,-1}, 3398 {0x1fb,0x21f,2,-1}, 3399 {0x223,0x233,2,-1}, 3400 {0x23c,0x23c,-1,-1}, 3401 {0x23f,0x240,1,10815}, 3402 {0x242,0x247,5,-1}, 3403 {0x249,0x24f,2,-1}, 3404 {0x250,0x250,-1,10783}, 3405 {0x251,0x251,-1,10780}, 3406 {0x252,0x252,-1,10782}, 3407 {0x253,0x253,-1,-210}, 3408 {0x254,0x254,-1,-206}, 3409 {0x256,0x257,1,-205}, 3410 {0x259,0x259,-1,-202}, 3411 {0x25b,0x25b,-1,-203}, 3412 {0x25c,0x25c,-1,42319}, 3413 {0x260,0x260,-1,-205}, 3414 {0x261,0x261,-1,42315}, 3415 {0x263,0x263,-1,-207}, 3416 {0x265,0x265,-1,42280}, 3417 {0x266,0x266,-1,42308}, 3418 {0x268,0x268,-1,-209}, 3419 {0x269,0x269,-1,-211}, 3420 {0x26a,0x26a,-1,42308}, 3421 {0x26b,0x26b,-1,10743}, 3422 {0x26c,0x26c,-1,42305}, 3423 {0x26f,0x26f,-1,-211}, 3424 {0x271,0x271,-1,10749}, 3425 {0x272,0x272,-1,-213}, 3426 {0x275,0x275,-1,-214}, 3427 {0x27d,0x27d,-1,10727}, 3428 {0x280,0x283,3,-218}, 3429 {0x287,0x287,-1,42282}, 3430 {0x288,0x288,-1,-218}, 3431 {0x289,0x289,-1,-69}, 3432 {0x28a,0x28b,1,-217}, 3433 {0x28c,0x28c,-1,-71}, 3434 {0x292,0x292,-1,-219}, 3435 {0x29d,0x29d,-1,42261}, 3436 {0x29e,0x29e,-1,42258}, 3437 {0x345,0x345,-1,84}, 3438 {0x371,0x373,2,-1}, 3439 {0x377,0x377,-1,-1}, 3440 {0x37b,0x37d,1,130}, 3441 {0x3ac,0x3ac,-1,-38}, 3442 {0x3ad,0x3af,1,-37}, 3443 {0x3b1,0x3c1,1,-32}, 3444 {0x3c2,0x3c2,-1,-31}, 3445 {0x3c3,0x3cb,1,-32}, 3446 {0x3cc,0x3cc,-1,-64}, 3447 {0x3cd,0x3ce,1,-63}, 3448 {0x3d0,0x3d0,-1,-62}, 3449 {0x3d1,0x3d1,-1,-57}, 3450 {0x3d5,0x3d5,-1,-47}, 3451 {0x3d6,0x3d6,-1,-54}, 3452 {0x3d7,0x3d7,-1,-8}, 3453 {0x3d9,0x3ef,2,-1}, 3454 {0x3f0,0x3f0,-1,-86}, 3455 {0x3f1,0x3f1,-1,-80}, 3456 {0x3f2,0x3f2,-1,7}, 3457 {0x3f3,0x3f3,-1,-116}, 3458 {0x3f5,0x3f5,-1,-96}, 3459 {0x3f8,0x3fb,3,-1}, 3460 {0x430,0x44f,1,-32}, 3461 {0x450,0x45f,1,-80}, 3462 {0x461,0x481,2,-1}, 3463 {0x48b,0x4bf,2,-1}, 3464 {0x4c2,0x4ce,2,-1}, 3465 {0x4cf,0x4cf,-1,-15}, 3466 {0x4d1,0x52f,2,-1}, 3467 {0x561,0x586,1,-48}, 3468 {0x10d0,0x10fa,1,3008}, 3469 {0x10fd,0x10ff,1,3008}, 3470 {0x13f8,0x13fd,1,-8}, 3471 {0x1c80,0x1c80,-1,-6254}, 3472 {0x1c81,0x1c81,-1,-6253}, 3473 {0x1c82,0x1c82,-1,-6244}, 3474 {0x1c83,0x1c84,1,-6242}, 3475 {0x1c85,0x1c85,-1,-6243}, 3476 {0x1c86,0x1c86,-1,-6236}, 3477 {0x1c87,0x1c87,-1,-6181}, 3478 {0x1c88,0x1c88,-1,35266}, 3479 {0x1d79,0x1d79,-1,35332}, 3480 {0x1d7d,0x1d7d,-1,3814}, 3481 {0x1e01,0x1e95,2,-1}, 3482 {0x1e9b,0x1e9b,-1,-59}, 3483 {0x1ea1,0x1eff,2,-1}, 3484 {0x1f00,0x1f07,1,8}, 3485 {0x1f10,0x1f15,1,8}, 3486 {0x1f20,0x1f27,1,8}, 3487 {0x1f30,0x1f37,1,8}, 3488 {0x1f40,0x1f45,1,8}, 3489 {0x1f51,0x1f57,2,8}, 3490 {0x1f60,0x1f67,1,8}, 3491 {0x1f70,0x1f71,1,74}, 3492 {0x1f72,0x1f75,1,86}, 3493 {0x1f76,0x1f77,1,100}, 3494 {0x1f78,0x1f79,1,128}, 3495 {0x1f7a,0x1f7b,1,112}, 3496 {0x1f7c,0x1f7d,1,126}, 3497 {0x1f80,0x1f87,1,8}, 3498 {0x1f90,0x1f97,1,8}, 3499 {0x1fa0,0x1fa7,1,8}, 3500 {0x1fb0,0x1fb1,1,8}, 3501 {0x1fb3,0x1fb3,-1,9}, 3502 {0x1fbe,0x1fbe,-1,-7205}, 3503 {0x1fc3,0x1fc3,-1,9}, 3504 {0x1fd0,0x1fd1,1,8}, 3505 {0x1fe0,0x1fe1,1,8}, 3506 {0x1fe5,0x1fe5,-1,7}, 3507 {0x1ff3,0x1ff3,-1,9}, 3508 {0x214e,0x214e,-1,-28}, 3509 {0x2170,0x217f,1,-16}, 3510 {0x2184,0x2184,-1,-1}, 3511 {0x24d0,0x24e9,1,-26}, 3512 {0x2c30,0x2c5e,1,-48}, 3513 {0x2c61,0x2c61,-1,-1}, 3514 {0x2c65,0x2c65,-1,-10795}, 3515 {0x2c66,0x2c66,-1,-10792}, 3516 {0x2c68,0x2c6c,2,-1}, 3517 {0x2c73,0x2c76,3,-1}, 3518 {0x2c81,0x2ce3,2,-1}, 3519 {0x2cec,0x2cee,2,-1}, 3520 {0x2cf3,0x2cf3,-1,-1}, 3521 {0x2d00,0x2d25,1,-7264}, 3522 {0x2d27,0x2d2d,6,-7264}, 3523 {0xa641,0xa66d,2,-1}, 3524 {0xa681,0xa69b,2,-1}, 3525 {0xa723,0xa72f,2,-1}, 3526 {0xa733,0xa76f,2,-1}, 3527 {0xa77a,0xa77c,2,-1}, 3528 {0xa77f,0xa787,2,-1}, 3529 {0xa78c,0xa791,5,-1}, 3530 {0xa793,0xa797,4,-1}, 3531 {0xa799,0xa7a9,2,-1}, 3532 {0xa7b5,0xa7b9,2,-1}, 3533 {0xab53,0xab53,-1,-928}, 3534 {0xab70,0xabbf,1,-38864}, 3535 {0xff41,0xff5a,1,-32}, 3536 {0x10428,0x1044f,1,-40}, 3537 {0x104d8,0x104fb,1,-40}, 3538 {0x10cc0,0x10cf2,1,-64}, 3539 {0x118c0,0x118df,1,-32}, 3540 {0x16e60,0x16e7f,1,-32}, 3541 {0x1e922,0x1e943,1,-34} 3542 }; 3543 3544 /* 3545 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3546 * simple case folding. 3547 */ 3548 int 3549 utf_toupper(int a) 3550 { 3551 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */ 3552 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3553 return TOUPPER_ASC(a); 3554 3555 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3556 /* If towupper() is available and handles Unicode, use it. */ 3557 if (!(cmp_flags & CMP_INTERNAL)) 3558 return towupper(a); 3559 #endif 3560 3561 /* For characters below 128 use locale sensitive toupper(). */ 3562 if (a < 128) 3563 return TOUPPER_LOC(a); 3564 3565 /* For any other characters use the above mapping table. */ 3566 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3567 } 3568 3569 int 3570 utf_islower(int a) 3571 { 3572 /* German sharp s is lower case but has no upper case equivalent. */ 3573 return (utf_toupper(a) != a) || a == 0xdf; 3574 } 3575 3576 /* 3577 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3578 * simple case folding. 3579 */ 3580 int 3581 utf_tolower(int a) 3582 { 3583 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */ 3584 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3585 return TOLOWER_ASC(a); 3586 3587 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3588 /* If towlower() is available and handles Unicode, use it. */ 3589 if (!(cmp_flags & CMP_INTERNAL)) 3590 return towlower(a); 3591 #endif 3592 3593 /* For characters below 128 use locale sensitive tolower(). */ 3594 if (a < 128) 3595 return TOLOWER_LOC(a); 3596 3597 /* For any other characters use the above mapping table. */ 3598 return utf_convert(a, toLower, (int)sizeof(toLower)); 3599 } 3600 3601 int 3602 utf_isupper(int a) 3603 { 3604 return (utf_tolower(a) != a); 3605 } 3606 3607 static int 3608 utf_strnicmp( 3609 char_u *s1, 3610 char_u *s2, 3611 size_t n1, 3612 size_t n2) 3613 { 3614 int c1, c2, cdiff; 3615 char_u buffer[6]; 3616 3617 for (;;) 3618 { 3619 c1 = utf_safe_read_char_adv(&s1, &n1); 3620 c2 = utf_safe_read_char_adv(&s2, &n2); 3621 3622 if (c1 <= 0 || c2 <= 0) 3623 break; 3624 3625 if (c1 == c2) 3626 continue; 3627 3628 cdiff = utf_fold(c1) - utf_fold(c2); 3629 if (cdiff != 0) 3630 return cdiff; 3631 } 3632 3633 /* some string ended or has an incomplete/illegal character sequence */ 3634 3635 if (c1 == 0 || c2 == 0) 3636 { 3637 /* some string ended. shorter string is smaller */ 3638 if (c1 == 0 && c2 == 0) 3639 return 0; 3640 return c1 == 0 ? -1 : 1; 3641 } 3642 3643 /* Continue with bytewise comparison to produce some result that 3644 * would make comparison operations involving this function transitive. 3645 * 3646 * If only one string had an error, comparison should be made with 3647 * folded version of the other string. In this case it is enough 3648 * to fold just one character to determine the result of comparison. */ 3649 3650 if (c1 != -1 && c2 == -1) 3651 { 3652 n1 = utf_char2bytes(utf_fold(c1), buffer); 3653 s1 = buffer; 3654 } 3655 else if (c2 != -1 && c1 == -1) 3656 { 3657 n2 = utf_char2bytes(utf_fold(c2), buffer); 3658 s2 = buffer; 3659 } 3660 3661 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3662 { 3663 cdiff = (int)(*s1) - (int)(*s2); 3664 if (cdiff != 0) 3665 return cdiff; 3666 3667 s1++; 3668 s2++; 3669 n1--; 3670 n2--; 3671 } 3672 3673 if (n1 > 0 && *s1 == NUL) 3674 n1 = 0; 3675 if (n2 > 0 && *s2 == NUL) 3676 n2 = 0; 3677 3678 if (n1 == 0 && n2 == 0) 3679 return 0; 3680 return n1 == 0 ? -1 : 1; 3681 } 3682 3683 /* 3684 * Version of strnicmp() that handles multi-byte characters. 3685 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3686 * probably use strnicmp(), because there are no ASCII characters in the 3687 * second byte. 3688 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3689 * two characters otherwise. 3690 */ 3691 int 3692 mb_strnicmp(char_u *s1, char_u *s2, size_t nn) 3693 { 3694 int i, l; 3695 int cdiff; 3696 int n = (int)nn; 3697 3698 if (enc_utf8) 3699 { 3700 return utf_strnicmp(s1, s2, nn, nn); 3701 } 3702 else 3703 { 3704 for (i = 0; i < n; i += l) 3705 { 3706 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */ 3707 return 0; 3708 3709 l = (*mb_ptr2len)(s1 + i); 3710 if (l <= 1) 3711 { 3712 /* Single byte: first check normally, then with ignore case. */ 3713 if (s1[i] != s2[i]) 3714 { 3715 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3716 if (cdiff != 0) 3717 return cdiff; 3718 } 3719 } 3720 else 3721 { 3722 /* For non-Unicode multi-byte don't ignore case. */ 3723 if (l > n - i) 3724 l = n - i; 3725 cdiff = STRNCMP(s1 + i, s2 + i, l); 3726 if (cdiff != 0) 3727 return cdiff; 3728 } 3729 } 3730 } 3731 return 0; 3732 } 3733 3734 /* 3735 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3736 * 'encoding' has been set to. 3737 */ 3738 void 3739 show_utf8(void) 3740 { 3741 int len; 3742 int rlen = 0; 3743 char_u *line; 3744 int clen; 3745 int i; 3746 3747 /* Get the byte length of the char under the cursor, including composing 3748 * characters. */ 3749 line = ml_get_cursor(); 3750 len = utfc_ptr2len(line); 3751 if (len == 0) 3752 { 3753 MSG("NUL"); 3754 return; 3755 } 3756 3757 clen = 0; 3758 for (i = 0; i < len; ++i) 3759 { 3760 if (clen == 0) 3761 { 3762 /* start of (composing) character, get its length */ 3763 if (i > 0) 3764 { 3765 STRCPY(IObuff + rlen, "+ "); 3766 rlen += 2; 3767 } 3768 clen = utf_ptr2len(line + i); 3769 } 3770 sprintf((char *)IObuff + rlen, "%02x ", 3771 (line[i] == NL) ? NUL : line[i]); /* NUL is stored as NL */ 3772 --clen; 3773 rlen += (int)STRLEN(IObuff + rlen); 3774 if (rlen > IOSIZE - 20) 3775 break; 3776 } 3777 3778 msg(IObuff); 3779 } 3780 3781 /* 3782 * mb_head_off() function pointer. 3783 * Return offset from "p" to the first byte of the character it points into. 3784 * If "p" points to the NUL at the end of the string return 0. 3785 * Returns 0 when already at the first byte of a character. 3786 */ 3787 int 3788 latin_head_off(char_u *base UNUSED, char_u *p UNUSED) 3789 { 3790 return 0; 3791 } 3792 3793 int 3794 dbcs_head_off(char_u *base, char_u *p) 3795 { 3796 char_u *q; 3797 3798 /* It can't be a trailing byte when not using DBCS, at the start of the 3799 * string or the previous byte can't start a double-byte. */ 3800 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3801 return 0; 3802 3803 /* This is slow: need to start at the base and go forward until the 3804 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */ 3805 q = base; 3806 while (q < p) 3807 q += dbcs_ptr2len(q); 3808 return (q == p) ? 0 : 1; 3809 } 3810 3811 /* 3812 * Special version of dbcs_head_off() that works for ScreenLines[], where 3813 * single-width DBCS_JPNU characters are stored separately. 3814 */ 3815 int 3816 dbcs_screen_head_off(char_u *base, char_u *p) 3817 { 3818 char_u *q; 3819 3820 /* It can't be a trailing byte when not using DBCS, at the start of the 3821 * string or the previous byte can't start a double-byte. 3822 * For euc-jp an 0x8e byte in the previous cell always means we have a 3823 * lead byte in the current cell. */ 3824 if (p <= base 3825 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3826 || MB_BYTE2LEN(p[-1]) == 1 3827 || *p == NUL) 3828 return 0; 3829 3830 /* This is slow: need to start at the base and go forward until the 3831 * byte we are looking for. Return 1 when we went past it, 0 otherwise. 3832 * For DBCS_JPNU look out for 0x8e, which means the second byte is not 3833 * stored as the next byte. */ 3834 q = base; 3835 while (q < p) 3836 { 3837 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3838 ++q; 3839 else 3840 q += dbcs_ptr2len(q); 3841 } 3842 return (q == p) ? 0 : 1; 3843 } 3844 3845 int 3846 utf_head_off(char_u *base, char_u *p) 3847 { 3848 char_u *q; 3849 char_u *s; 3850 int c; 3851 int len; 3852 #ifdef FEAT_ARABIC 3853 char_u *j; 3854 #endif 3855 3856 if (*p < 0x80) /* be quick for ASCII */ 3857 return 0; 3858 3859 /* Skip backwards over trailing bytes: 10xx.xxxx 3860 * Skip backwards again if on a composing char. */ 3861 for (q = p; ; --q) 3862 { 3863 /* Move s to the last byte of this char. */ 3864 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3865 ; 3866 /* Move q to the first byte of this char. */ 3867 while (q > base && (*q & 0xc0) == 0x80) 3868 --q; 3869 /* Check for illegal sequence. Do allow an illegal byte after where we 3870 * started. */ 3871 len = utf8len_tab[*q]; 3872 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3873 return 0; 3874 3875 if (q <= base) 3876 break; 3877 3878 c = utf_ptr2char(q); 3879 if (utf_iscomposing(c)) 3880 continue; 3881 3882 #ifdef FEAT_ARABIC 3883 if (arabic_maycombine(c)) 3884 { 3885 /* Advance to get a sneak-peak at the next char */ 3886 j = q; 3887 --j; 3888 /* Move j to the first byte of this char. */ 3889 while (j > base && (*j & 0xc0) == 0x80) 3890 --j; 3891 if (arabic_combine(utf_ptr2char(j), c)) 3892 continue; 3893 } 3894 #endif 3895 break; 3896 } 3897 3898 return (int)(p - q); 3899 } 3900 3901 /* 3902 * Copy a character from "*fp" to "*tp" and advance the pointers. 3903 */ 3904 void 3905 mb_copy_char(char_u **fp, char_u **tp) 3906 { 3907 int l = (*mb_ptr2len)(*fp); 3908 3909 mch_memmove(*tp, *fp, (size_t)l); 3910 *tp += l; 3911 *fp += l; 3912 } 3913 3914 /* 3915 * Return the offset from "p" to the first byte of a character. When "p" is 3916 * at the start of a character 0 is returned, otherwise the offset to the next 3917 * character. Can start anywhere in a stream of bytes. 3918 */ 3919 int 3920 mb_off_next(char_u *base, char_u *p) 3921 { 3922 int i; 3923 int j; 3924 3925 if (enc_utf8) 3926 { 3927 if (*p < 0x80) /* be quick for ASCII */ 3928 return 0; 3929 3930 /* Find the next character that isn't 10xx.xxxx */ 3931 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 3932 ; 3933 if (i > 0) 3934 { 3935 /* Check for illegal sequence. */ 3936 for (j = 0; p - j > base; ++j) 3937 if ((p[-j] & 0xc0) != 0x80) 3938 break; 3939 if (utf8len_tab[p[-j]] != i + j) 3940 return 0; 3941 } 3942 return i; 3943 } 3944 3945 /* Only need to check if we're on a trail byte, it doesn't matter if we 3946 * want the offset to the next or current character. */ 3947 return (*mb_head_off)(base, p); 3948 } 3949 3950 /* 3951 * Return the offset from "p" to the last byte of the character it points 3952 * into. Can start anywhere in a stream of bytes. 3953 */ 3954 int 3955 mb_tail_off(char_u *base, char_u *p) 3956 { 3957 int i; 3958 int j; 3959 3960 if (*p == NUL) 3961 return 0; 3962 3963 if (enc_utf8) 3964 { 3965 /* Find the last character that is 10xx.xxxx */ 3966 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 3967 ; 3968 /* Check for illegal sequence. */ 3969 for (j = 0; p - j > base; ++j) 3970 if ((p[-j] & 0xc0) != 0x80) 3971 break; 3972 if (utf8len_tab[p[-j]] != i + j + 1) 3973 return 0; 3974 return i; 3975 } 3976 3977 /* It can't be the first byte if a double-byte when not using DBCS, at the 3978 * end of the string or the byte can't start a double-byte. */ 3979 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 3980 return 0; 3981 3982 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3983 return 1 - dbcs_head_off(base, p); 3984 } 3985 3986 /* 3987 * Find the next illegal byte sequence. 3988 */ 3989 void 3990 utf_find_illegal(void) 3991 { 3992 pos_T pos = curwin->w_cursor; 3993 char_u *p; 3994 int len; 3995 vimconv_T vimconv; 3996 char_u *tofree = NULL; 3997 3998 vimconv.vc_type = CONV_NONE; 3999 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 4000 { 4001 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 4002 * possibly a utf-8 file with illegal bytes. Setup for conversion 4003 * from utf-8 to 'fileencoding'. */ 4004 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 4005 } 4006 4007 #ifdef FEAT_VIRTUALEDIT 4008 curwin->w_cursor.coladd = 0; 4009 #endif 4010 for (;;) 4011 { 4012 p = ml_get_cursor(); 4013 if (vimconv.vc_type != CONV_NONE) 4014 { 4015 vim_free(tofree); 4016 tofree = string_convert(&vimconv, p, NULL); 4017 if (tofree == NULL) 4018 break; 4019 p = tofree; 4020 } 4021 4022 while (*p != NUL) 4023 { 4024 /* Illegal means that there are not enough trail bytes (checked by 4025 * utf_ptr2len()) or too many of them (overlong sequence). */ 4026 len = utf_ptr2len(p); 4027 if (*p >= 0x80 && (len == 1 4028 || utf_char2len(utf_ptr2char(p)) != len)) 4029 { 4030 if (vimconv.vc_type == CONV_NONE) 4031 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 4032 else 4033 { 4034 int l; 4035 4036 len = (int)(p - tofree); 4037 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 4038 { 4039 l = utf_ptr2len(p); 4040 curwin->w_cursor.col += l; 4041 } 4042 } 4043 goto theend; 4044 } 4045 p += len; 4046 } 4047 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 4048 break; 4049 ++curwin->w_cursor.lnum; 4050 curwin->w_cursor.col = 0; 4051 } 4052 4053 /* didn't find it: don't move and beep */ 4054 curwin->w_cursor = pos; 4055 beep_flush(); 4056 4057 theend: 4058 vim_free(tofree); 4059 convert_setup(&vimconv, NULL, NULL); 4060 } 4061 4062 #if defined(FEAT_GUI_GTK) || defined(PROTO) 4063 /* 4064 * Return TRUE if string "s" is a valid utf-8 string. 4065 * When "end" is NULL stop at the first NUL. 4066 * When "end" is positive stop there. 4067 */ 4068 int 4069 utf_valid_string(char_u *s, char_u *end) 4070 { 4071 int l; 4072 char_u *p = s; 4073 4074 while (end == NULL ? *p != NUL : p < end) 4075 { 4076 l = utf8len_tab_zero[*p]; 4077 if (l == 0) 4078 return FALSE; /* invalid lead byte */ 4079 if (end != NULL && p + l > end) 4080 return FALSE; /* incomplete byte sequence */ 4081 ++p; 4082 while (--l > 0) 4083 if ((*p++ & 0xc0) != 0x80) 4084 return FALSE; /* invalid trail byte */ 4085 } 4086 return TRUE; 4087 } 4088 #endif 4089 4090 #if defined(FEAT_GUI) || defined(PROTO) 4091 /* 4092 * Special version of mb_tail_off() for use in ScreenLines[]. 4093 */ 4094 int 4095 dbcs_screen_tail_off(char_u *base, char_u *p) 4096 { 4097 /* It can't be the first byte if a double-byte when not using DBCS, at the 4098 * end of the string or the byte can't start a double-byte. 4099 * For euc-jp an 0x8e byte always means we have a lead byte in the current 4100 * cell. */ 4101 if (*p == NUL || p[1] == NUL 4102 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 4103 || MB_BYTE2LEN(*p) == 1) 4104 return 0; 4105 4106 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 4107 return 1 - dbcs_screen_head_off(base, p); 4108 } 4109 #endif 4110 4111 /* 4112 * If the cursor moves on an trail byte, set the cursor on the lead byte. 4113 * Thus it moves left if necessary. 4114 * Return TRUE when the cursor was adjusted. 4115 */ 4116 void 4117 mb_adjust_cursor(void) 4118 { 4119 mb_adjustpos(curbuf, &curwin->w_cursor); 4120 } 4121 4122 /* 4123 * Adjust position "*lp" to point to the first byte of a multi-byte character. 4124 * If it points to a tail byte it's moved backwards to the head byte. 4125 */ 4126 void 4127 mb_adjustpos(buf_T *buf, pos_T *lp) 4128 { 4129 char_u *p; 4130 4131 if (lp->col > 0 4132 #ifdef FEAT_VIRTUALEDIT 4133 || lp->coladd > 1 4134 #endif 4135 ) 4136 { 4137 p = ml_get_buf(buf, lp->lnum, FALSE); 4138 if (*p == NUL || (int)STRLEN(p) < lp->col) 4139 lp->col = 0; 4140 else 4141 lp->col -= (*mb_head_off)(p, p + lp->col); 4142 #ifdef FEAT_VIRTUALEDIT 4143 /* Reset "coladd" when the cursor would be on the right half of a 4144 * double-wide character. */ 4145 if (lp->coladd == 1 4146 && p[lp->col] != TAB 4147 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 4148 && ptr2cells(p + lp->col) > 1) 4149 lp->coladd = 0; 4150 #endif 4151 } 4152 } 4153 4154 /* 4155 * Return a pointer to the character before "*p", if there is one. 4156 */ 4157 char_u * 4158 mb_prevptr( 4159 char_u *line, /* start of the string */ 4160 char_u *p) 4161 { 4162 if (p > line) 4163 MB_PTR_BACK(line, p); 4164 return p; 4165 } 4166 4167 /* 4168 * Return the character length of "str". Each multi-byte character (with 4169 * following composing characters) counts as one. 4170 */ 4171 int 4172 mb_charlen(char_u *str) 4173 { 4174 char_u *p = str; 4175 int count; 4176 4177 if (p == NULL) 4178 return 0; 4179 4180 for (count = 0; *p != NUL; count++) 4181 p += (*mb_ptr2len)(p); 4182 4183 return count; 4184 } 4185 4186 #if defined(FEAT_SPELL) || defined(PROTO) 4187 /* 4188 * Like mb_charlen() but for a string with specified length. 4189 */ 4190 int 4191 mb_charlen_len(char_u *str, int len) 4192 { 4193 char_u *p = str; 4194 int count; 4195 4196 for (count = 0; *p != NUL && p < str + len; count++) 4197 p += (*mb_ptr2len)(p); 4198 4199 return count; 4200 } 4201 #endif 4202 4203 /* 4204 * Try to un-escape a multi-byte character. 4205 * Used for the "to" and "from" part of a mapping. 4206 * Return the un-escaped string if it is a multi-byte character, and advance 4207 * "pp" to just after the bytes that formed it. 4208 * Return NULL if no multi-byte char was found. 4209 */ 4210 char_u * 4211 mb_unescape(char_u **pp) 4212 { 4213 static char_u buf[6]; 4214 int n; 4215 int m = 0; 4216 char_u *str = *pp; 4217 4218 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 4219 * KS_EXTRA KE_CSI to CSI. 4220 * Maximum length of a utf-8 character is 4 bytes. */ 4221 for (n = 0; str[n] != NUL && m < 4; ++n) 4222 { 4223 if (str[n] == K_SPECIAL 4224 && str[n + 1] == KS_SPECIAL 4225 && str[n + 2] == KE_FILLER) 4226 { 4227 buf[m++] = K_SPECIAL; 4228 n += 2; 4229 } 4230 else if ((str[n] == K_SPECIAL 4231 # ifdef FEAT_GUI 4232 || str[n] == CSI 4233 # endif 4234 ) 4235 && str[n + 1] == KS_EXTRA 4236 && str[n + 2] == (int)KE_CSI) 4237 { 4238 buf[m++] = CSI; 4239 n += 2; 4240 } 4241 else if (str[n] == K_SPECIAL 4242 # ifdef FEAT_GUI 4243 || str[n] == CSI 4244 # endif 4245 ) 4246 break; /* a special key can't be a multibyte char */ 4247 else 4248 buf[m++] = str[n]; 4249 buf[m] = NUL; 4250 4251 /* Return a multi-byte character if it's found. An illegal sequence 4252 * will result in a 1 here. */ 4253 if ((*mb_ptr2len)(buf) > 1) 4254 { 4255 *pp = str + n + 1; 4256 return buf; 4257 } 4258 4259 /* Bail out quickly for ASCII. */ 4260 if (buf[0] < 128) 4261 break; 4262 } 4263 return NULL; 4264 } 4265 4266 /* 4267 * Return TRUE if the character at "row"/"col" on the screen is the left side 4268 * of a double-width character. 4269 * Caller must make sure "row" and "col" are not invalid! 4270 */ 4271 int 4272 mb_lefthalve(int row, int col) 4273 { 4274 #ifdef FEAT_HANGULIN 4275 if (composing_hangul) 4276 return TRUE; 4277 #endif 4278 return (*mb_off2cells)(LineOffset[row] + col, 4279 LineOffset[row] + screen_Columns) > 1; 4280 } 4281 4282 /* 4283 * Correct a position on the screen, if it's the right half of a double-wide 4284 * char move it to the left half. Returns the corrected column. 4285 */ 4286 int 4287 mb_fix_col(int col, int row) 4288 { 4289 col = check_col(col); 4290 row = check_row(row); 4291 if (has_mbyte && ScreenLines != NULL && col > 0 4292 && ((enc_dbcs 4293 && ScreenLines[LineOffset[row] + col] != NUL 4294 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 4295 ScreenLines + LineOffset[row] + col)) 4296 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0))) 4297 return col - 1; 4298 return col; 4299 } 4300 #endif 4301 4302 #if defined(FEAT_MBYTE) || defined(FEAT_POSTSCRIPT) || defined(PROTO) 4303 static int enc_alias_search(char_u *name); 4304 4305 /* 4306 * Skip the Vim specific head of a 'encoding' name. 4307 */ 4308 char_u * 4309 enc_skip(char_u *p) 4310 { 4311 if (STRNCMP(p, "2byte-", 6) == 0) 4312 return p + 6; 4313 if (STRNCMP(p, "8bit-", 5) == 0) 4314 return p + 5; 4315 return p; 4316 } 4317 4318 /* 4319 * Find the canonical name for encoding "enc". 4320 * When the name isn't recognized, returns "enc" itself, but with all lower 4321 * case characters and '_' replaced with '-'. 4322 * Returns an allocated string. NULL for out-of-memory. 4323 */ 4324 char_u * 4325 enc_canonize(char_u *enc) 4326 { 4327 char_u *r; 4328 char_u *p, *s; 4329 int i; 4330 4331 # ifdef FEAT_MBYTE 4332 if (STRCMP(enc, "default") == 0) 4333 { 4334 /* Use the default encoding as it's found by set_init_1(). */ 4335 r = get_encoding_default(); 4336 if (r == NULL) 4337 r = (char_u *)"latin1"; 4338 return vim_strsave(r); 4339 } 4340 # endif 4341 4342 /* copy "enc" to allocated memory, with room for two '-' */ 4343 r = alloc((unsigned)(STRLEN(enc) + 3)); 4344 if (r != NULL) 4345 { 4346 /* Make it all lower case and replace '_' with '-'. */ 4347 p = r; 4348 for (s = enc; *s != NUL; ++s) 4349 { 4350 if (*s == '_') 4351 *p++ = '-'; 4352 else 4353 *p++ = TOLOWER_ASC(*s); 4354 } 4355 *p = NUL; 4356 4357 /* Skip "2byte-" and "8bit-". */ 4358 p = enc_skip(r); 4359 4360 /* Change "microsoft-cp" to "cp". Used in some spell files. */ 4361 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4362 STRMOVE(p, p + 10); 4363 4364 /* "iso8859" -> "iso-8859" */ 4365 if (STRNCMP(p, "iso8859", 7) == 0) 4366 { 4367 STRMOVE(p + 4, p + 3); 4368 p[3] = '-'; 4369 } 4370 4371 /* "iso-8859n" -> "iso-8859-n" */ 4372 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4373 { 4374 STRMOVE(p + 9, p + 8); 4375 p[8] = '-'; 4376 } 4377 4378 /* "latin-N" -> "latinN" */ 4379 if (STRNCMP(p, "latin-", 6) == 0) 4380 STRMOVE(p + 5, p + 6); 4381 4382 if (enc_canon_search(p) >= 0) 4383 { 4384 /* canonical name can be used unmodified */ 4385 if (p != r) 4386 STRMOVE(r, p); 4387 } 4388 else if ((i = enc_alias_search(p)) >= 0) 4389 { 4390 /* alias recognized, get canonical name */ 4391 vim_free(r); 4392 r = vim_strsave((char_u *)enc_canon_table[i].name); 4393 } 4394 } 4395 return r; 4396 } 4397 4398 /* 4399 * Search for an encoding alias of "name". 4400 * Returns -1 when not found. 4401 */ 4402 static int 4403 enc_alias_search(char_u *name) 4404 { 4405 int i; 4406 4407 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4408 if (STRCMP(name, enc_alias_table[i].name) == 0) 4409 return enc_alias_table[i].canon; 4410 return -1; 4411 } 4412 #endif 4413 4414 #if defined(FEAT_MBYTE) || defined(PROTO) 4415 4416 # ifdef HAVE_LANGINFO_H 4417 # include <langinfo.h> 4418 # endif 4419 4420 # ifndef FEAT_GUI_W32 4421 /* 4422 * Get the canonicalized encoding from the specified locale string "locale" 4423 * or from the environment variables LC_ALL, LC_CTYPE and LANG. 4424 * Returns an allocated string when successful, NULL when not. 4425 */ 4426 char_u * 4427 enc_locale_env(char *locale) 4428 { 4429 char *s = locale; 4430 char *p; 4431 int i; 4432 char buf[50]; 4433 4434 if (s == NULL || *s == NUL) 4435 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4436 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4437 s = getenv("LANG"); 4438 4439 if (s == NULL || *s == NUL) 4440 return NULL; 4441 4442 /* The most generic locale format is: 4443 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4444 * If there is a '.' remove the part before it. 4445 * if there is something after the codeset, remove it. 4446 * Make the name lowercase and replace '_' with '-'. 4447 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4448 * "ko_KR.EUC" == "euc-kr" 4449 */ 4450 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4451 { 4452 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4453 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4454 { 4455 /* copy "XY.EUC" to "euc-XY" to buf[10] */ 4456 STRCPY(buf + 10, "euc-"); 4457 buf[14] = p[-2]; 4458 buf[15] = p[-1]; 4459 buf[16] = 0; 4460 s = buf + 10; 4461 } 4462 else 4463 s = p + 1; 4464 } 4465 for (i = 0; i < (int)sizeof(buf) - 1 && s[i] != NUL; ++i) 4466 { 4467 if (s[i] == '_' || s[i] == '-') 4468 buf[i] = '-'; 4469 else if (isalnum((int)s[i])) 4470 buf[i] = TOLOWER_ASC(s[i]); 4471 else 4472 break; 4473 } 4474 buf[i] = NUL; 4475 4476 return enc_canonize((char_u *)buf); 4477 } 4478 # endif 4479 4480 /* 4481 * Get the canonicalized encoding of the current locale. 4482 * Returns an allocated string when successful, NULL when not. 4483 */ 4484 char_u * 4485 enc_locale(void) 4486 { 4487 # ifdef WIN3264 4488 char buf[50]; 4489 long acp = GetACP(); 4490 4491 if (acp == 1200) 4492 STRCPY(buf, "ucs-2le"); 4493 else if (acp == 1252) /* cp1252 is used as latin1 */ 4494 STRCPY(buf, "latin1"); 4495 else 4496 sprintf(buf, "cp%ld", acp); 4497 4498 return enc_canonize((char_u *)buf); 4499 # else 4500 char *s; 4501 4502 # ifdef HAVE_NL_LANGINFO_CODESET 4503 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4504 # endif 4505 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4506 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4507 # endif 4508 s = NULL; 4509 4510 return enc_locale_env(s); 4511 # endif 4512 } 4513 4514 # if defined(WIN3264) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4515 /* 4516 * Convert an encoding name to an MS-Windows codepage. 4517 * Returns zero if no codepage can be figured out. 4518 */ 4519 int 4520 encname2codepage(char_u *name) 4521 { 4522 int cp; 4523 char_u *p = name; 4524 int idx; 4525 4526 if (STRNCMP(p, "8bit-", 5) == 0) 4527 p += 5; 4528 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4529 p += 6; 4530 4531 if (p[0] == 'c' && p[1] == 'p') 4532 cp = atoi((char *)p + 2); 4533 else if ((idx = enc_canon_search(p)) >= 0) 4534 cp = enc_canon_table[idx].codepage; 4535 else 4536 return 0; 4537 if (IsValidCodePage(cp)) 4538 return cp; 4539 return 0; 4540 } 4541 # endif 4542 4543 # if defined(USE_ICONV) || defined(PROTO) 4544 4545 static char_u *iconv_string(vimconv_T *vcp, char_u *str, int slen, int *unconvlenp, int *resultlenp); 4546 4547 /* 4548 * Call iconv_open() with a check if iconv() works properly (there are broken 4549 * versions). 4550 * Returns (void *)-1 if failed. 4551 * (should return iconv_t, but that causes problems with prototypes). 4552 */ 4553 void * 4554 my_iconv_open(char_u *to, char_u *from) 4555 { 4556 iconv_t fd; 4557 #define ICONV_TESTLEN 400 4558 char_u tobuf[ICONV_TESTLEN]; 4559 char *p; 4560 size_t tolen; 4561 static int iconv_ok = -1; 4562 4563 if (iconv_ok == FALSE) 4564 return (void *)-1; /* detected a broken iconv() previously */ 4565 4566 #ifdef DYNAMIC_ICONV 4567 /* Check if the iconv.dll can be found. */ 4568 if (!iconv_enabled(TRUE)) 4569 return (void *)-1; 4570 #endif 4571 4572 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4573 4574 if (fd != (iconv_t)-1 && iconv_ok == -1) 4575 { 4576 /* 4577 * Do a dummy iconv() call to check if it actually works. There is a 4578 * version of iconv() on Linux that is broken. We can't ignore it, 4579 * because it's wide-spread. The symptoms are that after outputting 4580 * the initial shift state the "to" pointer is NULL and conversion 4581 * stops for no apparent reason after about 8160 characters. 4582 */ 4583 p = (char *)tobuf; 4584 tolen = ICONV_TESTLEN; 4585 (void)iconv(fd, NULL, NULL, &p, &tolen); 4586 if (p == NULL) 4587 { 4588 iconv_ok = FALSE; 4589 iconv_close(fd); 4590 fd = (iconv_t)-1; 4591 } 4592 else 4593 iconv_ok = TRUE; 4594 } 4595 4596 return (void *)fd; 4597 } 4598 4599 /* 4600 * Convert the string "str[slen]" with iconv(). 4601 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4602 * sequence and set "*unconvlenp" to the length of it. 4603 * Returns the converted string in allocated memory. NULL for an error. 4604 * If resultlenp is not NULL, sets it to the result length in bytes. 4605 */ 4606 static char_u * 4607 iconv_string( 4608 vimconv_T *vcp, 4609 char_u *str, 4610 int slen, 4611 int *unconvlenp, 4612 int *resultlenp) 4613 { 4614 const char *from; 4615 size_t fromlen; 4616 char *to; 4617 size_t tolen; 4618 size_t len = 0; 4619 size_t done = 0; 4620 char_u *result = NULL; 4621 char_u *p; 4622 int l; 4623 4624 from = (char *)str; 4625 fromlen = slen; 4626 for (;;) 4627 { 4628 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4629 { 4630 /* Allocate enough room for most conversions. When re-allocating 4631 * increase the buffer size. */ 4632 len = len + fromlen * 2 + 40; 4633 p = alloc((unsigned)len); 4634 if (p != NULL && done > 0) 4635 mch_memmove(p, result, done); 4636 vim_free(result); 4637 result = p; 4638 if (result == NULL) /* out of memory */ 4639 break; 4640 } 4641 4642 to = (char *)result + done; 4643 tolen = len - done - 2; 4644 /* Avoid a warning for systems with a wrong iconv() prototype by 4645 * casting the second argument to void *. */ 4646 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4647 != (size_t)-1) 4648 { 4649 /* Finished, append a NUL. */ 4650 *to = NUL; 4651 break; 4652 } 4653 4654 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4655 * iconv library may use one of them. */ 4656 if (!vcp->vc_fail && unconvlenp != NULL 4657 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4658 { 4659 /* Handle an incomplete sequence at the end. */ 4660 *to = NUL; 4661 *unconvlenp = (int)fromlen; 4662 break; 4663 } 4664 4665 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4666 * iconv library may use one of them. */ 4667 else if (!vcp->vc_fail 4668 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4669 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4670 { 4671 /* Can't convert: insert a '?' and skip a character. This assumes 4672 * conversion from 'encoding' to something else. In other 4673 * situations we don't know what to skip anyway. */ 4674 *to++ = '?'; 4675 if ((*mb_ptr2cells)((char_u *)from) > 1) 4676 *to++ = '?'; 4677 if (enc_utf8) 4678 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4679 else 4680 { 4681 l = (*mb_ptr2len)((char_u *)from); 4682 if (l > (int)fromlen) 4683 l = (int)fromlen; 4684 } 4685 from += l; 4686 fromlen -= l; 4687 } 4688 else if (ICONV_ERRNO != ICONV_E2BIG) 4689 { 4690 /* conversion failed */ 4691 VIM_CLEAR(result); 4692 break; 4693 } 4694 /* Not enough room or skipping illegal sequence. */ 4695 done = to - (char *)result; 4696 } 4697 4698 if (resultlenp != NULL && result != NULL) 4699 *resultlenp = (int)(to - (char *)result); 4700 return result; 4701 } 4702 4703 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4704 /* 4705 * Dynamically load the "iconv.dll" on Win32. 4706 */ 4707 4708 # ifndef DYNAMIC_ICONV /* must be generating prototypes */ 4709 # define HINSTANCE int 4710 # endif 4711 static HINSTANCE hIconvDLL = 0; 4712 static HINSTANCE hMsvcrtDLL = 0; 4713 4714 # ifndef DYNAMIC_ICONV_DLL 4715 # define DYNAMIC_ICONV_DLL "iconv.dll" 4716 # define DYNAMIC_ICONV_DLL_ALT1 "libiconv.dll" 4717 # define DYNAMIC_ICONV_DLL_ALT2 "libiconv2.dll" 4718 # define DYNAMIC_ICONV_DLL_ALT3 "libiconv-2.dll" 4719 # endif 4720 # ifndef DYNAMIC_MSVCRT_DLL 4721 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4722 # endif 4723 4724 /* 4725 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4726 */ 4727 int 4728 iconv_enabled(int verbose) 4729 { 4730 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4731 return TRUE; 4732 4733 /* The iconv DLL file goes under different names, try them all. 4734 * Do the "2" version first, it's newer. */ 4735 #ifdef DYNAMIC_ICONV_DLL_ALT2 4736 if (hIconvDLL == 0) 4737 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT2); 4738 #endif 4739 #ifdef DYNAMIC_ICONV_DLL_ALT3 4740 if (hIconvDLL == 0) 4741 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT3); 4742 #endif 4743 if (hIconvDLL == 0) 4744 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4745 #ifdef DYNAMIC_ICONV_DLL_ALT1 4746 if (hIconvDLL == 0) 4747 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT1); 4748 #endif 4749 4750 if (hIconvDLL != 0) 4751 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4752 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4753 { 4754 /* Only give the message when 'verbose' is set, otherwise it might be 4755 * done whenever a conversion is attempted. */ 4756 if (verbose && p_verbose > 0) 4757 { 4758 verbose_enter(); 4759 EMSG2(_(e_loadlib), 4760 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4761 verbose_leave(); 4762 } 4763 iconv_end(); 4764 return FALSE; 4765 } 4766 4767 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4768 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4769 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4770 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4771 iconv_errno = get_dll_import_func(hIconvDLL, "_errno"); 4772 if (iconv_errno == NULL) 4773 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4774 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4775 || iconvctl == NULL || iconv_errno == NULL) 4776 { 4777 iconv_end(); 4778 if (verbose && p_verbose > 0) 4779 { 4780 verbose_enter(); 4781 EMSG2(_(e_loadfunc), "for libiconv"); 4782 verbose_leave(); 4783 } 4784 return FALSE; 4785 } 4786 return TRUE; 4787 } 4788 4789 void 4790 iconv_end(void) 4791 { 4792 /* Don't use iconv() when inputting or outputting characters. */ 4793 if (input_conv.vc_type == CONV_ICONV) 4794 convert_setup(&input_conv, NULL, NULL); 4795 if (output_conv.vc_type == CONV_ICONV) 4796 convert_setup(&output_conv, NULL, NULL); 4797 4798 if (hIconvDLL != 0) 4799 FreeLibrary(hIconvDLL); 4800 if (hMsvcrtDLL != 0) 4801 FreeLibrary(hMsvcrtDLL); 4802 hIconvDLL = 0; 4803 hMsvcrtDLL = 0; 4804 } 4805 # endif /* DYNAMIC_ICONV */ 4806 # endif /* USE_ICONV */ 4807 4808 #endif /* FEAT_MBYTE */ 4809 4810 #ifdef FEAT_GUI 4811 # define USE_IMACTIVATEFUNC (!gui.in_use && *p_imaf != NUL) 4812 # define USE_IMSTATUSFUNC (!gui.in_use && *p_imsf != NUL) 4813 #else 4814 # define USE_IMACTIVATEFUNC (*p_imaf != NUL) 4815 # define USE_IMSTATUSFUNC (*p_imsf != NUL) 4816 #endif 4817 4818 #if defined(FEAT_EVAL) && defined(FEAT_MBYTE) \ 4819 && (defined(FEAT_XIM) || defined(IME_WITHOUT_XIM)) 4820 static void 4821 call_imactivatefunc(int active) 4822 { 4823 typval_T argv[2]; 4824 4825 argv[0].v_type = VAR_NUMBER; 4826 argv[0].vval.v_number = active ? 1 : 0; 4827 argv[1].v_type = VAR_UNKNOWN; 4828 (void)call_func_retnr(p_imaf, 1, argv); 4829 } 4830 4831 static int 4832 call_imstatusfunc(void) 4833 { 4834 int is_active; 4835 4836 /* FIXME: Don't execute user function in unsafe situation. */ 4837 if (exiting || is_autocmd_blocked()) 4838 return FALSE; 4839 /* FIXME: :py print 'xxx' is shown duplicate result. 4840 * Use silent to avoid it. */ 4841 ++msg_silent; 4842 is_active = call_func_retnr(p_imsf, 0, NULL); 4843 --msg_silent; 4844 return (is_active > 0); 4845 } 4846 #endif 4847 4848 #if defined(FEAT_XIM) || defined(PROTO) 4849 4850 # if defined(FEAT_GUI_GTK) || defined(PROTO) 4851 static int xim_has_preediting INIT(= FALSE); /* IM current status */ 4852 4853 /* 4854 * Set preedit_start_col to the current cursor position. 4855 */ 4856 static void 4857 init_preedit_start_col(void) 4858 { 4859 if (State & CMDLINE) 4860 preedit_start_col = cmdline_getvcol_cursor(); 4861 else if (curwin != NULL && curwin->w_buffer != NULL) 4862 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); 4863 /* Prevent that preediting marks the buffer as changed. */ 4864 xim_changed_while_preediting = curbuf->b_changed; 4865 } 4866 4867 static int im_is_active = FALSE; /* IM is enabled for current mode */ 4868 static int preedit_is_active = FALSE; 4869 static int im_preedit_cursor = 0; /* cursor offset in characters */ 4870 static int im_preedit_trailing = 0; /* number of characters after cursor */ 4871 4872 static unsigned long im_commit_handler_id = 0; 4873 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; 4874 static unsigned int im_activatekey_state = 0; 4875 4876 static GtkWidget *preedit_window = NULL; 4877 static GtkWidget *preedit_label = NULL; 4878 4879 static void im_preedit_window_set_position(void); 4880 4881 void 4882 im_set_active(int active) 4883 { 4884 int was_active; 4885 4886 was_active = !!im_get_status(); 4887 im_is_active = (active && !p_imdisable); 4888 4889 if (im_is_active != was_active) 4890 xim_reset(); 4891 } 4892 4893 void 4894 xim_set_focus(int focus) 4895 { 4896 if (xic != NULL) 4897 { 4898 if (focus) 4899 gtk_im_context_focus_in(xic); 4900 else 4901 gtk_im_context_focus_out(xic); 4902 } 4903 } 4904 4905 void 4906 im_set_position(int row, int col) 4907 { 4908 if (xic != NULL) 4909 { 4910 GdkRectangle area; 4911 4912 area.x = FILL_X(col); 4913 area.y = FILL_Y(row); 4914 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); 4915 area.height = gui.char_height; 4916 4917 gtk_im_context_set_cursor_location(xic, &area); 4918 4919 if (p_imst == IM_OVER_THE_SPOT) 4920 im_preedit_window_set_position(); 4921 } 4922 } 4923 4924 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */ 4925 void 4926 xim_set_preedit(void) 4927 { 4928 im_set_position(gui.row, gui.col); 4929 } 4930 # endif 4931 4932 static void 4933 im_add_to_input(char_u *str, int len) 4934 { 4935 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */ 4936 if (input_conv.vc_type != CONV_NONE) 4937 { 4938 str = string_convert(&input_conv, str, &len); 4939 g_return_if_fail(str != NULL); 4940 } 4941 4942 add_to_input_buf_csi(str, len); 4943 4944 if (input_conv.vc_type != CONV_NONE) 4945 vim_free(str); 4946 4947 if (p_mh) /* blank out the pointer if necessary */ 4948 gui_mch_mousehide(TRUE); 4949 } 4950 4951 static void 4952 im_preedit_window_set_position(void) 4953 { 4954 int x, y, width, height; 4955 int screen_x, screen_y, screen_width, screen_height; 4956 4957 if (preedit_window == NULL) 4958 return; 4959 4960 gui_gtk_get_screen_geom_of_win(gui.drawarea, 4961 &screen_x, &screen_y, &screen_width, &screen_height); 4962 #if GTK_CHECK_VERSION(3,0,0) 4963 gdk_window_get_origin(gtk_widget_get_window(gui.drawarea), &x, &y); 4964 #else 4965 gdk_window_get_origin(gui.drawarea->window, &x, &y); 4966 #endif 4967 gtk_window_get_size(GTK_WINDOW(preedit_window), &width, &height); 4968 x = x + FILL_X(gui.col); 4969 y = y + FILL_Y(gui.row); 4970 if (x + width > screen_x + screen_width) 4971 x = screen_x + screen_width - width; 4972 if (y + height > screen_y + screen_height) 4973 y = screen_y + screen_height - height; 4974 gtk_window_move(GTK_WINDOW(preedit_window), x, y); 4975 } 4976 4977 static void 4978 im_preedit_window_open() 4979 { 4980 char *preedit_string; 4981 #if !GTK_CHECK_VERSION(3,16,0) 4982 char buf[8]; 4983 #endif 4984 PangoAttrList *attr_list; 4985 PangoLayout *layout; 4986 #if GTK_CHECK_VERSION(3,0,0) 4987 # if !GTK_CHECK_VERSION(3,16,0) 4988 GdkRGBA color; 4989 # endif 4990 #else 4991 GdkColor color; 4992 #endif 4993 gint w, h; 4994 4995 if (preedit_window == NULL) 4996 { 4997 preedit_window = gtk_window_new(GTK_WINDOW_POPUP); 4998 gtk_window_set_transient_for(GTK_WINDOW(preedit_window), 4999 GTK_WINDOW(gui.mainwin)); 5000 preedit_label = gtk_label_new(""); 5001 gtk_widget_set_name(preedit_label, "vim-gui-preedit-area"); 5002 gtk_container_add(GTK_CONTAINER(preedit_window), preedit_label); 5003 } 5004 5005 #if GTK_CHECK_VERSION(3,16,0) 5006 { 5007 GtkStyleContext * const context 5008 = gtk_widget_get_style_context(gui.drawarea); 5009 GtkCssProvider * const provider = gtk_css_provider_new(); 5010 gchar *css = NULL; 5011 const char * const fontname 5012 = pango_font_description_get_family(gui.norm_font); 5013 gint fontsize 5014 = pango_font_description_get_size(gui.norm_font) / PANGO_SCALE; 5015 gchar *fontsize_propval = NULL; 5016 5017 if (!pango_font_description_get_size_is_absolute(gui.norm_font)) 5018 { 5019 /* fontsize was given in points. Convert it into that in pixels 5020 * to use with CSS. */ 5021 GdkScreen * const screen 5022 = gdk_window_get_screen(gtk_widget_get_window(gui.mainwin)); 5023 const gdouble dpi = gdk_screen_get_resolution(screen); 5024 fontsize = dpi * fontsize / 72; 5025 } 5026 if (fontsize > 0) 5027 fontsize_propval = g_strdup_printf("%dpx", fontsize); 5028 else 5029 fontsize_propval = g_strdup_printf("inherit"); 5030 5031 css = g_strdup_printf( 5032 "widget#vim-gui-preedit-area {\n" 5033 " font-family: %s,monospace;\n" 5034 " font-size: %s;\n" 5035 " color: #%.2lx%.2lx%.2lx;\n" 5036 " background-color: #%.2lx%.2lx%.2lx;\n" 5037 "}\n", 5038 fontname != NULL ? fontname : "inherit", 5039 fontsize_propval, 5040 (gui.norm_pixel >> 16) & 0xff, 5041 (gui.norm_pixel >> 8) & 0xff, 5042 gui.norm_pixel & 0xff, 5043 (gui.back_pixel >> 16) & 0xff, 5044 (gui.back_pixel >> 8) & 0xff, 5045 gui.back_pixel & 0xff); 5046 5047 gtk_css_provider_load_from_data(provider, css, -1, NULL); 5048 gtk_style_context_add_provider(context, 5049 GTK_STYLE_PROVIDER(provider), G_MAXUINT); 5050 5051 g_free(css); 5052 g_free(fontsize_propval); 5053 g_object_unref(provider); 5054 } 5055 #elif GTK_CHECK_VERSION(3,0,0) 5056 gtk_widget_override_font(preedit_label, gui.norm_font); 5057 5058 vim_snprintf(buf, sizeof(buf), "#%06X", gui.norm_pixel); 5059 gdk_rgba_parse(&color, buf); 5060 gtk_widget_override_color(preedit_label, GTK_STATE_FLAG_NORMAL, &color); 5061 5062 vim_snprintf(buf, sizeof(buf), "#%06X", gui.back_pixel); 5063 gdk_rgba_parse(&color, buf); 5064 gtk_widget_override_background_color(preedit_label, GTK_STATE_FLAG_NORMAL, 5065 &color); 5066 #else 5067 gtk_widget_modify_font(preedit_label, gui.norm_font); 5068 5069 vim_snprintf(buf, sizeof(buf), "#%06X", (unsigned)gui.norm_pixel); 5070 gdk_color_parse(buf, &color); 5071 gtk_widget_modify_fg(preedit_label, GTK_STATE_NORMAL, &color); 5072 5073 vim_snprintf(buf, sizeof(buf), "#%06X", (unsigned)gui.back_pixel); 5074 gdk_color_parse(buf, &color); 5075 gtk_widget_modify_bg(preedit_window, GTK_STATE_NORMAL, &color); 5076 #endif 5077 5078 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5079 5080 if (preedit_string[0] != NUL) 5081 { 5082 gtk_label_set_text(GTK_LABEL(preedit_label), preedit_string); 5083 gtk_label_set_attributes(GTK_LABEL(preedit_label), attr_list); 5084 5085 layout = gtk_label_get_layout(GTK_LABEL(preedit_label)); 5086 pango_layout_get_pixel_size(layout, &w, &h); 5087 h = MAX(h, gui.char_height); 5088 gtk_window_resize(GTK_WINDOW(preedit_window), w, h); 5089 5090 gtk_widget_show_all(preedit_window); 5091 5092 im_preedit_window_set_position(); 5093 } 5094 5095 g_free(preedit_string); 5096 pango_attr_list_unref(attr_list); 5097 } 5098 5099 static void 5100 im_preedit_window_close() 5101 { 5102 if (preedit_window != NULL) 5103 gtk_widget_hide(preedit_window); 5104 } 5105 5106 static void 5107 im_show_preedit() 5108 { 5109 im_preedit_window_open(); 5110 5111 if (p_mh) /* blank out the pointer if necessary */ 5112 gui_mch_mousehide(TRUE); 5113 } 5114 5115 static void 5116 im_delete_preedit(void) 5117 { 5118 char_u bskey[] = {CSI, 'k', 'b'}; 5119 char_u delkey[] = {CSI, 'k', 'D'}; 5120 5121 if (p_imst == IM_OVER_THE_SPOT) 5122 { 5123 im_preedit_window_close(); 5124 return; 5125 } 5126 5127 if (State & NORMAL 5128 #ifdef FEAT_TERMINAL 5129 && !term_use_loop() 5130 #endif 5131 ) 5132 { 5133 im_preedit_cursor = 0; 5134 return; 5135 } 5136 for (; im_preedit_cursor > 0; --im_preedit_cursor) 5137 add_to_input_buf(bskey, (int)sizeof(bskey)); 5138 5139 for (; im_preedit_trailing > 0; --im_preedit_trailing) 5140 add_to_input_buf(delkey, (int)sizeof(delkey)); 5141 } 5142 5143 /* 5144 * Move the cursor left by "num_move_back" characters. 5145 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for 5146 * these K_LEFT keys. 5147 */ 5148 static void 5149 im_correct_cursor(int num_move_back) 5150 { 5151 char_u backkey[] = {CSI, 'k', 'l'}; 5152 5153 if (State & NORMAL) 5154 return; 5155 # ifdef FEAT_RIGHTLEFT 5156 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) 5157 backkey[2] = 'r'; 5158 # endif 5159 for (; num_move_back > 0; --num_move_back) 5160 add_to_input_buf(backkey, (int)sizeof(backkey)); 5161 } 5162 5163 static int xim_expected_char = NUL; 5164 static int xim_ignored_char = FALSE; 5165 5166 /* 5167 * Update the mode and cursor while in an IM callback. 5168 */ 5169 static void 5170 im_show_info(void) 5171 { 5172 int old_vgetc_busy; 5173 5174 old_vgetc_busy = vgetc_busy; 5175 vgetc_busy = TRUE; 5176 showmode(); 5177 vgetc_busy = old_vgetc_busy; 5178 if ((State & NORMAL) || (State & INSERT)) 5179 setcursor(); 5180 out_flush(); 5181 } 5182 5183 /* 5184 * Callback invoked when the user finished preediting. 5185 * Put the final string into the input buffer. 5186 */ 5187 static void 5188 im_commit_cb(GtkIMContext *context UNUSED, 5189 const gchar *str, 5190 gpointer data UNUSED) 5191 { 5192 int slen = (int)STRLEN(str); 5193 int add_to_input = TRUE; 5194 int clen; 5195 int len = slen; 5196 int commit_with_preedit = TRUE; 5197 char_u *im_str; 5198 5199 #ifdef XIM_DEBUG 5200 xim_log("im_commit_cb(): %s\n", str); 5201 #endif 5202 5203 if (p_imst == IM_ON_THE_SPOT) 5204 { 5205 /* The imhangul module doesn't reset the preedit string before 5206 * committing. Call im_delete_preedit() to work around that. */ 5207 im_delete_preedit(); 5208 5209 /* Indicate that preediting has finished. */ 5210 if (preedit_start_col == MAXCOL) 5211 { 5212 init_preedit_start_col(); 5213 commit_with_preedit = FALSE; 5214 } 5215 5216 /* The thing which setting "preedit_start_col" to MAXCOL means that 5217 * "preedit_start_col" will be set forcedly when calling 5218 * preedit_changed_cb() next time. 5219 * "preedit_start_col" should not reset with MAXCOL on this part. Vim 5220 * is simulating the preediting by using add_to_input_str(). when 5221 * preedit begin immediately before committed, the typebuf is not 5222 * flushed to screen, then it can't get correct "preedit_start_col". 5223 * Thus, it should calculate the cells by adding cells of the committed 5224 * string. */ 5225 if (input_conv.vc_type != CONV_NONE) 5226 { 5227 im_str = string_convert(&input_conv, (char_u *)str, &len); 5228 g_return_if_fail(im_str != NULL); 5229 } 5230 else 5231 im_str = (char_u *)str; 5232 5233 clen = mb_string2cells(im_str, len); 5234 5235 if (input_conv.vc_type != CONV_NONE) 5236 vim_free(im_str); 5237 preedit_start_col += clen; 5238 } 5239 5240 /* Is this a single character that matches a keypad key that's just 5241 * been pressed? If so, we don't want it to be entered as such - let 5242 * us carry on processing the raw keycode so that it may be used in 5243 * mappings as <kSomething>. */ 5244 if (xim_expected_char != NUL) 5245 { 5246 /* We're currently processing a keypad or other special key */ 5247 if (slen == 1 && str[0] == xim_expected_char) 5248 { 5249 /* It's a match - don't do it here */ 5250 xim_ignored_char = TRUE; 5251 add_to_input = FALSE; 5252 } 5253 else 5254 { 5255 /* Not a match */ 5256 xim_ignored_char = FALSE; 5257 } 5258 } 5259 5260 if (add_to_input) 5261 im_add_to_input((char_u *)str, slen); 5262 5263 if (p_imst == IM_ON_THE_SPOT) 5264 { 5265 /* Inserting chars while "im_is_active" is set does not cause a 5266 * change of buffer. When the chars are committed the buffer must be 5267 * marked as changed. */ 5268 if (!commit_with_preedit) 5269 preedit_start_col = MAXCOL; 5270 5271 /* This flag is used in changed() at next call. */ 5272 xim_changed_while_preediting = TRUE; 5273 } 5274 5275 if (gtk_main_level() > 0) 5276 gtk_main_quit(); 5277 } 5278 5279 /* 5280 * Callback invoked after start to the preedit. 5281 */ 5282 static void 5283 im_preedit_start_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 5284 { 5285 #ifdef XIM_DEBUG 5286 xim_log("im_preedit_start_cb()\n"); 5287 #endif 5288 5289 im_is_active = TRUE; 5290 preedit_is_active = TRUE; 5291 gui_update_cursor(TRUE, FALSE); 5292 im_show_info(); 5293 } 5294 5295 /* 5296 * Callback invoked after end to the preedit. 5297 */ 5298 static void 5299 im_preedit_end_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 5300 { 5301 #ifdef XIM_DEBUG 5302 xim_log("im_preedit_end_cb()\n"); 5303 #endif 5304 im_delete_preedit(); 5305 5306 /* Indicate that preediting has finished */ 5307 if (p_imst == IM_ON_THE_SPOT) 5308 preedit_start_col = MAXCOL; 5309 xim_has_preediting = FALSE; 5310 5311 #if 0 5312 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was 5313 * switched off unintentionally. We now use preedit_is_active (added by 5314 * SungHyun Nam). */ 5315 im_is_active = FALSE; 5316 #endif 5317 preedit_is_active = FALSE; 5318 gui_update_cursor(TRUE, FALSE); 5319 im_show_info(); 5320 } 5321 5322 /* 5323 * Callback invoked after changes to the preedit string. If the preedit 5324 * string was empty before, remember the preedit start column so we know 5325 * where to apply feedback attributes. Delete the previous preedit string 5326 * if there was one, save the new preedit cursor offset, and put the new 5327 * string into the input buffer. 5328 * 5329 * TODO: The pragmatic "put into input buffer" approach used here has 5330 * several fundamental problems: 5331 * 5332 * - The characters in the preedit string are subject to remapping. 5333 * That's broken, only the finally committed string should be remapped. 5334 * 5335 * - There is a race condition involved: The retrieved value for the 5336 * current cursor position will be wrong if any unprocessed characters 5337 * are still queued in the input buffer. 5338 * 5339 * - Due to the lack of synchronization between the file buffer in memory 5340 * and any typed characters, it's practically impossible to implement the 5341 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM 5342 * modules for languages such as Thai are likely to rely on this feature 5343 * for proper operation. 5344 * 5345 * Conclusions: I think support for preediting needs to be moved to the 5346 * core parts of Vim. Ideally, until it has been committed, the preediting 5347 * string should only be displayed and not affect the buffer content at all. 5348 * The question how to deal with the synchronization issue still remains. 5349 * Circumventing the input buffer is probably not desirable. Anyway, I think 5350 * implementing "retrieve_surrounding" is the only hard problem. 5351 * 5352 * One way to solve all of this in a clean manner would be to queue all key 5353 * press/release events "as is" in the input buffer, and apply the IM filtering 5354 * at the receiving end of the queue. This, however, would have a rather large 5355 * impact on the code base. If there is an easy way to force processing of all 5356 * remaining input from within the "retrieve_surrounding" signal handler, this 5357 * might not be necessary. Gotta ask on vim-dev for opinions. 5358 */ 5359 static void 5360 im_preedit_changed_cb(GtkIMContext *context, gpointer data UNUSED) 5361 { 5362 char *preedit_string = NULL; 5363 int cursor_index = 0; 5364 int num_move_back = 0; 5365 char_u *str; 5366 char_u *p; 5367 int i; 5368 5369 if (p_imst == IM_ON_THE_SPOT) 5370 gtk_im_context_get_preedit_string(context, 5371 &preedit_string, NULL, 5372 &cursor_index); 5373 else 5374 gtk_im_context_get_preedit_string(context, 5375 &preedit_string, NULL, 5376 NULL); 5377 5378 #ifdef XIM_DEBUG 5379 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); 5380 #endif 5381 5382 g_return_if_fail(preedit_string != NULL); /* just in case */ 5383 5384 if (p_imst == IM_OVER_THE_SPOT) 5385 { 5386 if (preedit_string[0] == NUL) 5387 { 5388 xim_has_preediting = FALSE; 5389 im_delete_preedit(); 5390 } 5391 else 5392 { 5393 xim_has_preediting = TRUE; 5394 im_show_preedit(); 5395 } 5396 } 5397 else 5398 { 5399 /* If preedit_start_col is MAXCOL set it to the current cursor position. */ 5400 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') 5401 { 5402 xim_has_preediting = TRUE; 5403 5404 /* Urgh, this breaks if the input buffer isn't empty now */ 5405 init_preedit_start_col(); 5406 } 5407 else if (cursor_index == 0 && preedit_string[0] == '\0') 5408 { 5409 xim_has_preediting = FALSE; 5410 5411 /* If at the start position (after typing backspace) 5412 * preedit_start_col must be reset. */ 5413 preedit_start_col = MAXCOL; 5414 } 5415 5416 im_delete_preedit(); 5417 5418 /* 5419 * Compute the end of the preediting area: "preedit_end_col". 5420 * According to the documentation of gtk_im_context_get_preedit_string(), 5421 * the cursor_pos output argument returns the offset in bytes. This is 5422 * unfortunately not true -- real life shows the offset is in characters, 5423 * and the GTK+ source code agrees with me. Will file a bug later. 5424 */ 5425 if (preedit_start_col != MAXCOL) 5426 preedit_end_col = preedit_start_col; 5427 str = (char_u *)preedit_string; 5428 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) 5429 { 5430 int is_composing; 5431 5432 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); 5433 /* 5434 * These offsets are used as counters when generating <BS> and <Del> 5435 * to delete the preedit string. So don't count composing characters 5436 * unless 'delcombine' is enabled. 5437 */ 5438 if (!is_composing || p_deco) 5439 { 5440 if (i < cursor_index) 5441 ++im_preedit_cursor; 5442 else 5443 ++im_preedit_trailing; 5444 } 5445 if (!is_composing && i >= cursor_index) 5446 { 5447 /* This is essentially the same as im_preedit_trailing, except 5448 * composing characters are not counted even if p_deco is set. */ 5449 ++num_move_back; 5450 } 5451 if (preedit_start_col != MAXCOL) 5452 preedit_end_col += utf_ptr2cells(p); 5453 } 5454 5455 if (p > str) 5456 { 5457 im_add_to_input(str, (int)(p - str)); 5458 im_correct_cursor(num_move_back); 5459 } 5460 } 5461 5462 g_free(preedit_string); 5463 5464 if (gtk_main_level() > 0) 5465 gtk_main_quit(); 5466 } 5467 5468 /* 5469 * Translate the Pango attributes at iter to Vim highlighting attributes. 5470 * Ignore attributes not supported by Vim highlighting. This shouldn't have 5471 * too much impact -- right now we handle even more attributes than necessary 5472 * for the IM modules I tested with. 5473 */ 5474 static int 5475 translate_pango_attributes(PangoAttrIterator *iter) 5476 { 5477 PangoAttribute *attr; 5478 int char_attr = HL_NORMAL; 5479 5480 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); 5481 if (attr != NULL && ((PangoAttrInt *)attr)->value 5482 != (int)PANGO_UNDERLINE_NONE) 5483 char_attr |= HL_UNDERLINE; 5484 5485 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); 5486 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) 5487 char_attr |= HL_BOLD; 5488 5489 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); 5490 if (attr != NULL && ((PangoAttrInt *)attr)->value 5491 != (int)PANGO_STYLE_NORMAL) 5492 char_attr |= HL_ITALIC; 5493 5494 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); 5495 if (attr != NULL) 5496 { 5497 const PangoColor *color = &((PangoAttrColor *)attr)->color; 5498 5499 /* Assume inverse if black background is requested */ 5500 if ((color->red | color->green | color->blue) == 0) 5501 char_attr |= HL_INVERSE; 5502 } 5503 5504 return char_attr; 5505 } 5506 5507 /* 5508 * Retrieve the highlighting attributes at column col in the preedit string. 5509 * Return -1 if not in preediting mode or if col is out of range. 5510 */ 5511 int 5512 im_get_feedback_attr(int col) 5513 { 5514 char *preedit_string = NULL; 5515 PangoAttrList *attr_list = NULL; 5516 int char_attr = -1; 5517 5518 if (xic == NULL) 5519 return char_attr; 5520 5521 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5522 5523 if (preedit_string != NULL && attr_list != NULL) 5524 { 5525 int idx; 5526 5527 /* Get the byte index as used by PangoAttrIterator */ 5528 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col) 5529 idx += utfc_ptr2len((char_u *)preedit_string + idx); 5530 5531 if (preedit_string[idx] != '\0') 5532 { 5533 PangoAttrIterator *iter; 5534 int start, end; 5535 5536 char_attr = HL_NORMAL; 5537 iter = pango_attr_list_get_iterator(attr_list); 5538 5539 /* Extract all relevant attributes from the list. */ 5540 do 5541 { 5542 pango_attr_iterator_range(iter, &start, &end); 5543 5544 if (idx >= start && idx < end) 5545 char_attr |= translate_pango_attributes(iter); 5546 } 5547 while (pango_attr_iterator_next(iter)); 5548 5549 pango_attr_iterator_destroy(iter); 5550 } 5551 } 5552 5553 if (attr_list != NULL) 5554 pango_attr_list_unref(attr_list); 5555 g_free(preedit_string); 5556 5557 return char_attr; 5558 } 5559 5560 void 5561 xim_init(void) 5562 { 5563 #ifdef XIM_DEBUG 5564 xim_log("xim_init()\n"); 5565 #endif 5566 5567 g_return_if_fail(gui.drawarea != NULL); 5568 #if GTK_CHECK_VERSION(3,0,0) 5569 g_return_if_fail(gtk_widget_get_window(gui.drawarea) != NULL); 5570 #else 5571 g_return_if_fail(gui.drawarea->window != NULL); 5572 #endif 5573 5574 xic = gtk_im_multicontext_new(); 5575 g_object_ref(xic); 5576 5577 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", 5578 G_CALLBACK(&im_commit_cb), NULL); 5579 g_signal_connect(G_OBJECT(xic), "preedit_changed", 5580 G_CALLBACK(&im_preedit_changed_cb), NULL); 5581 g_signal_connect(G_OBJECT(xic), "preedit_start", 5582 G_CALLBACK(&im_preedit_start_cb), NULL); 5583 g_signal_connect(G_OBJECT(xic), "preedit_end", 5584 G_CALLBACK(&im_preedit_end_cb), NULL); 5585 5586 #if GTK_CHECK_VERSION(3,0,0) 5587 gtk_im_context_set_client_window(xic, gtk_widget_get_window(gui.drawarea)); 5588 #else 5589 gtk_im_context_set_client_window(xic, gui.drawarea->window); 5590 #endif 5591 } 5592 5593 void 5594 im_shutdown(void) 5595 { 5596 #ifdef XIM_DEBUG 5597 xim_log("im_shutdown()\n"); 5598 #endif 5599 5600 if (xic != NULL) 5601 { 5602 gtk_im_context_focus_out(xic); 5603 g_object_unref(xic); 5604 xic = NULL; 5605 } 5606 im_is_active = FALSE; 5607 im_commit_handler_id = 0; 5608 if (p_imst == IM_ON_THE_SPOT) 5609 preedit_start_col = MAXCOL; 5610 xim_has_preediting = FALSE; 5611 } 5612 5613 /* 5614 * Convert the string argument to keyval and state for GdkEventKey. 5615 * If str is valid return TRUE, otherwise FALSE. 5616 * 5617 * See 'imactivatekey' for documentation of the format. 5618 */ 5619 static int 5620 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) 5621 { 5622 const char *mods_end; 5623 unsigned tmp_keyval; 5624 unsigned tmp_state = 0; 5625 5626 mods_end = strrchr(str, '-'); 5627 mods_end = (mods_end != NULL) ? mods_end + 1 : str; 5628 5629 /* Parse modifier keys */ 5630 while (str < mods_end) 5631 switch (*str++) 5632 { 5633 case '-': break; 5634 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; 5635 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; 5636 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; 5637 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; 5638 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; 5639 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; 5640 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; 5641 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; 5642 default: 5643 return FALSE; 5644 } 5645 5646 tmp_keyval = gdk_keyval_from_name(str); 5647 5648 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) 5649 return FALSE; 5650 5651 if (keyval != NULL) 5652 *keyval = tmp_keyval; 5653 if (state != NULL) 5654 *state = tmp_state; 5655 5656 return TRUE; 5657 } 5658 5659 /* 5660 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an 5661 * empty string is also regarded as valid. 5662 * 5663 * Note: The numerical key value of p_imak is cached if it was valid; thus 5664 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever 5665 * 'imak' changes. This is currently the case but not obvious -- should 5666 * probably rename the function for clarity. 5667 */ 5668 int 5669 im_xim_isvalid_imactivate(void) 5670 { 5671 if (p_imak[0] == NUL) 5672 { 5673 im_activatekey_keyval = GDK_VoidSymbol; 5674 im_activatekey_state = 0; 5675 return TRUE; 5676 } 5677 5678 return im_string_to_keyval((const char *)p_imak, 5679 &im_activatekey_keyval, 5680 &im_activatekey_state); 5681 } 5682 5683 static void 5684 im_synthesize_keypress(unsigned int keyval, unsigned int state) 5685 { 5686 GdkEventKey *event; 5687 5688 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); 5689 # if GTK_CHECK_VERSION(3,0,0) 5690 g_object_ref(gtk_widget_get_window(gui.drawarea)); 5691 /* unreffed by gdk_event_free() */ 5692 # else 5693 g_object_ref(gui.drawarea->window); /* unreffed by gdk_event_free() */ 5694 # endif 5695 # if GTK_CHECK_VERSION(3,0,0) 5696 event->window = gtk_widget_get_window(gui.drawarea); 5697 # else 5698 event->window = gui.drawarea->window; 5699 # endif 5700 event->send_event = TRUE; 5701 event->time = GDK_CURRENT_TIME; 5702 event->state = state; 5703 event->keyval = keyval; 5704 event->hardware_keycode = /* needed for XIM */ 5705 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); 5706 event->length = 0; 5707 event->string = NULL; 5708 5709 gtk_im_context_filter_keypress(xic, event); 5710 5711 /* For consistency, also send the corresponding release event. */ 5712 event->type = GDK_KEY_RELEASE; 5713 event->send_event = FALSE; 5714 gtk_im_context_filter_keypress(xic, event); 5715 5716 gdk_event_free((GdkEvent *)event); 5717 } 5718 5719 void 5720 xim_reset(void) 5721 { 5722 # ifdef FEAT_EVAL 5723 if (USE_IMACTIVATEFUNC) 5724 call_imactivatefunc(im_is_active); 5725 else 5726 # endif 5727 if (xic != NULL) 5728 { 5729 gtk_im_context_reset(xic); 5730 5731 if (p_imdisable) 5732 im_shutdown(); 5733 else 5734 { 5735 xim_set_focus(gui.in_focus); 5736 5737 if (im_activatekey_keyval != GDK_VoidSymbol) 5738 { 5739 if (im_is_active) 5740 { 5741 g_signal_handler_block(xic, im_commit_handler_id); 5742 im_synthesize_keypress(im_activatekey_keyval, 5743 im_activatekey_state); 5744 g_signal_handler_unblock(xic, im_commit_handler_id); 5745 } 5746 } 5747 else 5748 { 5749 im_shutdown(); 5750 xim_init(); 5751 xim_set_focus(gui.in_focus); 5752 } 5753 } 5754 } 5755 5756 if (p_imst == IM_ON_THE_SPOT) 5757 preedit_start_col = MAXCOL; 5758 xim_has_preediting = FALSE; 5759 } 5760 5761 int 5762 xim_queue_key_press_event(GdkEventKey *event, int down) 5763 { 5764 if (down) 5765 { 5766 /* 5767 * Workaround GTK2 XIM 'feature' that always converts keypad keys to 5768 * chars., even when not part of an IM sequence (ref. feature of 5769 * gdk/gdkkeyuni.c). 5770 * Flag any keypad keys that might represent a single char. 5771 * If this (on its own - i.e., not part of an IM sequence) is 5772 * committed while we're processing one of these keys, we can ignore 5773 * that commit and go ahead & process it ourselves. That way we can 5774 * still distinguish keypad keys for use in mappings. 5775 * Also add GDK_space to make <S-Space> work. 5776 */ 5777 switch (event->keyval) 5778 { 5779 case GDK_KP_Add: xim_expected_char = '+'; break; 5780 case GDK_KP_Subtract: xim_expected_char = '-'; break; 5781 case GDK_KP_Divide: xim_expected_char = '/'; break; 5782 case GDK_KP_Multiply: xim_expected_char = '*'; break; 5783 case GDK_KP_Decimal: xim_expected_char = '.'; break; 5784 case GDK_KP_Equal: xim_expected_char = '='; break; 5785 case GDK_KP_0: xim_expected_char = '0'; break; 5786 case GDK_KP_1: xim_expected_char = '1'; break; 5787 case GDK_KP_2: xim_expected_char = '2'; break; 5788 case GDK_KP_3: xim_expected_char = '3'; break; 5789 case GDK_KP_4: xim_expected_char = '4'; break; 5790 case GDK_KP_5: xim_expected_char = '5'; break; 5791 case GDK_KP_6: xim_expected_char = '6'; break; 5792 case GDK_KP_7: xim_expected_char = '7'; break; 5793 case GDK_KP_8: xim_expected_char = '8'; break; 5794 case GDK_KP_9: xim_expected_char = '9'; break; 5795 case GDK_space: xim_expected_char = ' '; break; 5796 default: xim_expected_char = NUL; 5797 } 5798 xim_ignored_char = FALSE; 5799 } 5800 5801 /* 5802 * When typing fFtT, XIM may be activated. Thus it must pass 5803 * gtk_im_context_filter_keypress() in Normal mode. 5804 * And while doing :sh too. 5805 */ 5806 if (xic != NULL && !p_imdisable 5807 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) 5808 { 5809 /* 5810 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is 5811 * always aware of the current status of IM, and can even emulate 5812 * the activation key for modules that don't support one. 5813 */ 5814 if (event->keyval == im_activatekey_keyval 5815 && (event->state & im_activatekey_state) == im_activatekey_state) 5816 { 5817 unsigned int state_mask; 5818 5819 /* Require the state of the 3 most used modifiers to match exactly. 5820 * Otherwise e.g. <S-C-space> would be unusable for other purposes 5821 * if the IM activate key is <S-space>. */ 5822 state_mask = im_activatekey_state; 5823 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK 5824 | (int)GDK_MOD1_MASK); 5825 5826 if ((event->state & state_mask) != im_activatekey_state) 5827 return FALSE; 5828 5829 /* Don't send it a second time on GDK_KEY_RELEASE. */ 5830 if (event->type != GDK_KEY_PRESS) 5831 return TRUE; 5832 5833 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE)) 5834 { 5835 im_set_active(FALSE); 5836 5837 /* ":lmap" mappings exists, toggle use of mappings. */ 5838 State ^= LANGMAP; 5839 if (State & LANGMAP) 5840 { 5841 curbuf->b_p_iminsert = B_IMODE_NONE; 5842 State &= ~LANGMAP; 5843 } 5844 else 5845 { 5846 curbuf->b_p_iminsert = B_IMODE_LMAP; 5847 State |= LANGMAP; 5848 } 5849 return TRUE; 5850 } 5851 5852 return gtk_im_context_filter_keypress(xic, event); 5853 } 5854 5855 /* Don't filter events through the IM context if IM isn't active 5856 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module 5857 * not doing anything before the activation key was sent. */ 5858 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) 5859 { 5860 int imresult = gtk_im_context_filter_keypress(xic, event); 5861 5862 if (p_imst == IM_ON_THE_SPOT) 5863 { 5864 /* Some XIM send following sequence: 5865 * 1. preedited string. 5866 * 2. committed string. 5867 * 3. line changed key. 5868 * 4. preedited string. 5869 * 5. remove preedited string. 5870 * if 3, Vim can't move back the above line for 5. 5871 * thus, this part should not parse the key. */ 5872 if (!imresult && preedit_start_col != MAXCOL 5873 && event->keyval == GDK_Return) 5874 { 5875 im_synthesize_keypress(GDK_Return, 0U); 5876 return FALSE; 5877 } 5878 } 5879 5880 /* If XIM tried to commit a keypad key as a single char., 5881 * ignore it so we can use the keypad key 'raw', for mappings. */ 5882 if (xim_expected_char != NUL && xim_ignored_char) 5883 /* We had a keypad key, and XIM tried to thieve it */ 5884 return FALSE; 5885 5886 /* This is supposed to fix a problem with iBus, that space 5887 * characters don't work in input mode. */ 5888 xim_expected_char = NUL; 5889 5890 /* Normal processing */ 5891 return imresult; 5892 } 5893 } 5894 5895 return FALSE; 5896 } 5897 5898 int 5899 im_get_status(void) 5900 { 5901 # ifdef FEAT_EVAL 5902 if (USE_IMSTATUSFUNC) 5903 return call_imstatusfunc(); 5904 # endif 5905 return im_is_active; 5906 } 5907 5908 int 5909 preedit_get_status(void) 5910 { 5911 return preedit_is_active; 5912 } 5913 5914 int 5915 im_is_preediting(void) 5916 { 5917 return xim_has_preediting; 5918 } 5919 5920 # else /* !FEAT_GUI_GTK */ 5921 5922 static int xim_is_active = FALSE; /* XIM should be active in the current 5923 mode */ 5924 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */ 5925 # ifdef FEAT_GUI_X11 5926 static XIMStyle input_style; 5927 static int status_area_enabled = TRUE; 5928 # endif 5929 5930 /* 5931 * Switch using XIM on/off. This is used by the code that changes "State". 5932 * When 'imactivatefunc' is defined use that function instead. 5933 */ 5934 void 5935 im_set_active(int active_arg) 5936 { 5937 int active = active_arg; 5938 5939 /* If 'imdisable' is set, XIM is never active. */ 5940 if (p_imdisable) 5941 active = FALSE; 5942 else if (input_style & XIMPreeditPosition) 5943 /* There is a problem in switching XIM off when preediting is used, 5944 * and it is not clear how this can be solved. For now, keep XIM on 5945 * all the time, like it was done in Vim 5.8. */ 5946 active = TRUE; 5947 5948 # if defined(FEAT_EVAL) 5949 if (USE_IMACTIVATEFUNC) 5950 { 5951 if (active != im_get_status()) 5952 { 5953 call_imactivatefunc(active); 5954 xim_has_focus = active; 5955 } 5956 return; 5957 } 5958 # endif 5959 5960 if (xic == NULL) 5961 return; 5962 5963 /* Remember the active state, it is needed when Vim gets keyboard focus. */ 5964 xim_is_active = active; 5965 xim_set_preedit(); 5966 } 5967 5968 /* 5969 * Adjust using XIM for gaining or losing keyboard focus. Also called when 5970 * "xim_is_active" changes. 5971 */ 5972 void 5973 xim_set_focus(int focus) 5974 { 5975 if (xic == NULL) 5976 return; 5977 5978 /* 5979 * XIM only gets focus when the Vim window has keyboard focus and XIM has 5980 * been set active for the current mode. 5981 */ 5982 if (focus && xim_is_active) 5983 { 5984 if (!xim_has_focus) 5985 { 5986 xim_has_focus = TRUE; 5987 XSetICFocus(xic); 5988 } 5989 } 5990 else 5991 { 5992 if (xim_has_focus) 5993 { 5994 xim_has_focus = FALSE; 5995 XUnsetICFocus(xic); 5996 } 5997 } 5998 } 5999 6000 void 6001 im_set_position(int row UNUSED, int col UNUSED) 6002 { 6003 xim_set_preedit(); 6004 } 6005 6006 /* 6007 * Set the XIM to the current cursor position. 6008 */ 6009 void 6010 xim_set_preedit(void) 6011 { 6012 XVaNestedList attr_list; 6013 XRectangle spot_area; 6014 XPoint over_spot; 6015 int line_space; 6016 6017 if (xic == NULL) 6018 return; 6019 6020 xim_set_focus(TRUE); 6021 6022 if (!xim_has_focus) 6023 { 6024 /* hide XIM cursor */ 6025 over_spot.x = 0; 6026 over_spot.y = -100; /* arbitrary invisible position */ 6027 attr_list = (XVaNestedList) XVaCreateNestedList(0, 6028 XNSpotLocation, 6029 &over_spot, 6030 NULL); 6031 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); 6032 XFree(attr_list); 6033 return; 6034 } 6035 6036 if (input_style & XIMPreeditPosition) 6037 { 6038 if (xim_fg_color == INVALCOLOR) 6039 { 6040 xim_fg_color = gui.def_norm_pixel; 6041 xim_bg_color = gui.def_back_pixel; 6042 } 6043 over_spot.x = TEXT_X(gui.col); 6044 over_spot.y = TEXT_Y(gui.row); 6045 spot_area.x = 0; 6046 spot_area.y = 0; 6047 spot_area.height = gui.char_height * Rows; 6048 spot_area.width = gui.char_width * Columns; 6049 line_space = gui.char_height; 6050 attr_list = (XVaNestedList) XVaCreateNestedList(0, 6051 XNSpotLocation, &over_spot, 6052 XNForeground, (Pixel) xim_fg_color, 6053 XNBackground, (Pixel) xim_bg_color, 6054 XNArea, &spot_area, 6055 XNLineSpace, line_space, 6056 NULL); 6057 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) 6058 EMSG(_("E284: Cannot set IC values")); 6059 XFree(attr_list); 6060 } 6061 } 6062 6063 # if defined(FEAT_GUI_X11) 6064 static char e_xim[] = N_("E285: Failed to create input context"); 6065 # endif 6066 6067 # if defined(FEAT_GUI_X11) || defined(PROTO) 6068 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(SUN_SYSTEM) 6069 # define USE_X11R6_XIM 6070 # endif 6071 6072 static int xim_real_init(Window x11_window, Display *x11_display); 6073 6074 6075 # ifdef USE_X11R6_XIM 6076 static void xim_destroy_cb(XIM im, XPointer client_data, XPointer call_data); 6077 6078 static void 6079 xim_instantiate_cb( 6080 Display *display, 6081 XPointer client_data UNUSED, 6082 XPointer call_data UNUSED) 6083 { 6084 Window x11_window; 6085 Display *x11_display; 6086 6087 # ifdef XIM_DEBUG 6088 xim_log("xim_instantiate_cb()\n"); 6089 # endif 6090 6091 gui_get_x11_windis(&x11_window, &x11_display); 6092 if (display != x11_display) 6093 return; 6094 6095 xim_real_init(x11_window, x11_display); 6096 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6097 if (xic != NULL) 6098 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6099 xim_instantiate_cb, NULL); 6100 } 6101 6102 static void 6103 xim_destroy_cb( 6104 XIM im UNUSED, 6105 XPointer client_data UNUSED, 6106 XPointer call_data UNUSED) 6107 { 6108 Window x11_window; 6109 Display *x11_display; 6110 6111 # ifdef XIM_DEBUG 6112 xim_log("xim_destroy_cb()\n"); 6113 #endif 6114 gui_get_x11_windis(&x11_window, &x11_display); 6115 6116 xic = NULL; 6117 status_area_enabled = FALSE; 6118 6119 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6120 6121 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6122 xim_instantiate_cb, NULL); 6123 } 6124 # endif 6125 6126 void 6127 xim_init(void) 6128 { 6129 Window x11_window; 6130 Display *x11_display; 6131 6132 # ifdef XIM_DEBUG 6133 xim_log("xim_init()\n"); 6134 # endif 6135 6136 gui_get_x11_windis(&x11_window, &x11_display); 6137 6138 xic = NULL; 6139 6140 if (xim_real_init(x11_window, x11_display)) 6141 return; 6142 6143 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6144 6145 # ifdef USE_X11R6_XIM 6146 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6147 xim_instantiate_cb, NULL); 6148 # endif 6149 } 6150 6151 static int 6152 xim_real_init(Window x11_window, Display *x11_display) 6153 { 6154 int i; 6155 char *p, 6156 *s, 6157 *ns, 6158 *end, 6159 tmp[1024]; 6160 # define IMLEN_MAX 40 6161 char buf[IMLEN_MAX + 7]; 6162 XIM xim = NULL; 6163 XIMStyles *xim_styles; 6164 XIMStyle this_input_style = 0; 6165 Boolean found; 6166 XPoint over_spot; 6167 XVaNestedList preedit_list, status_list; 6168 6169 input_style = 0; 6170 status_area_enabled = FALSE; 6171 6172 if (xic != NULL) 6173 return FALSE; 6174 6175 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) 6176 { 6177 strcpy(tmp, gui.rsrc_input_method); 6178 for (ns = s = tmp; ns != NULL && *s != NUL;) 6179 { 6180 s = (char *)skipwhite((char_u *)s); 6181 if (*s == NUL) 6182 break; 6183 if ((ns = end = strchr(s, ',')) == NULL) 6184 end = s + strlen(s); 6185 while (isspace(((char_u *)end)[-1])) 6186 end--; 6187 *end = NUL; 6188 6189 if (strlen(s) <= IMLEN_MAX) 6190 { 6191 strcpy(buf, "@im="); 6192 strcat(buf, s); 6193 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL 6194 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) 6195 != NULL) 6196 break; 6197 } 6198 6199 s = ns + 1; 6200 } 6201 } 6202 6203 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) 6204 xim = XOpenIM(x11_display, NULL, NULL, NULL); 6205 6206 /* This is supposed to be useful to obtain characters through 6207 * XmbLookupString() without really using a XIM. */ 6208 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL 6209 && *p != NUL) 6210 xim = XOpenIM(x11_display, NULL, NULL, NULL); 6211 6212 if (xim == NULL) 6213 { 6214 /* Only give this message when verbose is set, because too many people 6215 * got this message when they didn't want to use a XIM. */ 6216 if (p_verbose > 0) 6217 { 6218 verbose_enter(); 6219 EMSG(_("E286: Failed to open input method")); 6220 verbose_leave(); 6221 } 6222 return FALSE; 6223 } 6224 6225 # ifdef USE_X11R6_XIM 6226 { 6227 XIMCallback destroy_cb; 6228 6229 destroy_cb.callback = xim_destroy_cb; 6230 destroy_cb.client_data = NULL; 6231 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) 6232 EMSG(_("E287: Warning: Could not set destroy callback to IM")); 6233 } 6234 # endif 6235 6236 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) 6237 { 6238 EMSG(_("E288: input method doesn't support any style")); 6239 XCloseIM(xim); 6240 return FALSE; 6241 } 6242 6243 found = False; 6244 strcpy(tmp, gui.rsrc_preedit_type_name); 6245 for (s = tmp; s && !found; ) 6246 { 6247 while (*s && isspace((unsigned char)*s)) 6248 s++; 6249 if (!*s) 6250 break; 6251 if ((ns = end = strchr(s, ',')) != 0) 6252 ns++; 6253 else 6254 end = s + strlen(s); 6255 while (isspace((unsigned char)*end)) 6256 end--; 6257 *end = '\0'; 6258 6259 if (!strcmp(s, "OverTheSpot")) 6260 this_input_style = (XIMPreeditPosition | XIMStatusArea); 6261 else if (!strcmp(s, "OffTheSpot")) 6262 this_input_style = (XIMPreeditArea | XIMStatusArea); 6263 else if (!strcmp(s, "Root")) 6264 this_input_style = (XIMPreeditNothing | XIMStatusNothing); 6265 6266 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 6267 { 6268 if (this_input_style == xim_styles->supported_styles[i]) 6269 { 6270 found = True; 6271 break; 6272 } 6273 } 6274 if (!found) 6275 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 6276 { 6277 if ((xim_styles->supported_styles[i] & this_input_style) 6278 == (this_input_style & ~XIMStatusArea)) 6279 { 6280 this_input_style &= ~XIMStatusArea; 6281 found = True; 6282 break; 6283 } 6284 } 6285 6286 s = ns; 6287 } 6288 XFree(xim_styles); 6289 6290 if (!found) 6291 { 6292 /* Only give this message when verbose is set, because too many people 6293 * got this message when they didn't want to use a XIM. */ 6294 if (p_verbose > 0) 6295 { 6296 verbose_enter(); 6297 EMSG(_("E289: input method doesn't support my preedit type")); 6298 verbose_leave(); 6299 } 6300 XCloseIM(xim); 6301 return FALSE; 6302 } 6303 6304 over_spot.x = TEXT_X(gui.col); 6305 over_spot.y = TEXT_Y(gui.row); 6306 input_style = this_input_style; 6307 6308 /* A crash was reported when trying to pass gui.norm_font as XNFontSet, 6309 * thus that has been removed. Hopefully the default works... */ 6310 # ifdef FEAT_XFONTSET 6311 if (gui.fontset != NOFONTSET) 6312 { 6313 preedit_list = XVaCreateNestedList(0, 6314 XNSpotLocation, &over_spot, 6315 XNForeground, (Pixel)gui.def_norm_pixel, 6316 XNBackground, (Pixel)gui.def_back_pixel, 6317 XNFontSet, (XFontSet)gui.fontset, 6318 NULL); 6319 status_list = XVaCreateNestedList(0, 6320 XNForeground, (Pixel)gui.def_norm_pixel, 6321 XNBackground, (Pixel)gui.def_back_pixel, 6322 XNFontSet, (XFontSet)gui.fontset, 6323 NULL); 6324 } 6325 else 6326 # endif 6327 { 6328 preedit_list = XVaCreateNestedList(0, 6329 XNSpotLocation, &over_spot, 6330 XNForeground, (Pixel)gui.def_norm_pixel, 6331 XNBackground, (Pixel)gui.def_back_pixel, 6332 NULL); 6333 status_list = XVaCreateNestedList(0, 6334 XNForeground, (Pixel)gui.def_norm_pixel, 6335 XNBackground, (Pixel)gui.def_back_pixel, 6336 NULL); 6337 } 6338 6339 xic = XCreateIC(xim, 6340 XNInputStyle, input_style, 6341 XNClientWindow, x11_window, 6342 XNFocusWindow, gui.wid, 6343 XNPreeditAttributes, preedit_list, 6344 XNStatusAttributes, status_list, 6345 NULL); 6346 XFree(status_list); 6347 XFree(preedit_list); 6348 if (xic != NULL) 6349 { 6350 if (input_style & XIMStatusArea) 6351 { 6352 xim_set_status_area(); 6353 status_area_enabled = TRUE; 6354 } 6355 else 6356 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6357 } 6358 else 6359 { 6360 if (!is_not_a_term()) 6361 EMSG(_(e_xim)); 6362 XCloseIM(xim); 6363 return FALSE; 6364 } 6365 6366 return TRUE; 6367 } 6368 6369 # endif /* FEAT_GUI_X11 */ 6370 6371 /* 6372 * Get IM status. When IM is on, return TRUE. Else return FALSE. 6373 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is 6374 * active, when not having focus XIM may still be active (e.g., when using a 6375 * tear-off menu item). 6376 */ 6377 int 6378 im_get_status(void) 6379 { 6380 # ifdef FEAT_EVAL 6381 if (USE_IMSTATUSFUNC) 6382 return call_imstatusfunc(); 6383 # endif 6384 return xim_has_focus; 6385 } 6386 6387 # endif /* !FEAT_GUI_GTK */ 6388 6389 # if !defined(FEAT_GUI_GTK) || defined(PROTO) 6390 /* 6391 * Set up the status area. 6392 * 6393 * This should use a separate Widget, but that seems not possible, because 6394 * preedit_area and status_area should be set to the same window as for the 6395 * text input. Unfortunately this means the status area pollutes the text 6396 * window... 6397 */ 6398 void 6399 xim_set_status_area(void) 6400 { 6401 XVaNestedList preedit_list = 0, status_list = 0, list = 0; 6402 XRectangle pre_area, status_area; 6403 6404 if (xic == NULL) 6405 return; 6406 6407 if (input_style & XIMStatusArea) 6408 { 6409 if (input_style & XIMPreeditArea) 6410 { 6411 XRectangle *needed_rect; 6412 6413 /* to get status_area width */ 6414 status_list = XVaCreateNestedList(0, XNAreaNeeded, 6415 &needed_rect, NULL); 6416 XGetICValues(xic, XNStatusAttributes, status_list, NULL); 6417 XFree(status_list); 6418 6419 status_area.width = needed_rect->width; 6420 } 6421 else 6422 status_area.width = gui.char_width * Columns; 6423 6424 status_area.x = 0; 6425 status_area.y = gui.char_height * Rows + gui.border_offset; 6426 if (gui.which_scrollbars[SBAR_BOTTOM]) 6427 status_area.y += gui.scrollbar_height; 6428 #ifdef FEAT_MENU 6429 if (gui.menu_is_active) 6430 status_area.y += gui.menu_height; 6431 #endif 6432 status_area.height = gui.char_height; 6433 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); 6434 } 6435 else 6436 { 6437 status_area.x = 0; 6438 status_area.y = gui.char_height * Rows + gui.border_offset; 6439 if (gui.which_scrollbars[SBAR_BOTTOM]) 6440 status_area.y += gui.scrollbar_height; 6441 #ifdef FEAT_MENU 6442 if (gui.menu_is_active) 6443 status_area.y += gui.menu_height; 6444 #endif 6445 status_area.width = 0; 6446 status_area.height = gui.char_height; 6447 } 6448 6449 if (input_style & XIMPreeditArea) /* off-the-spot */ 6450 { 6451 pre_area.x = status_area.x + status_area.width; 6452 pre_area.y = gui.char_height * Rows + gui.border_offset; 6453 pre_area.width = gui.char_width * Columns - pre_area.x; 6454 if (gui.which_scrollbars[SBAR_BOTTOM]) 6455 pre_area.y += gui.scrollbar_height; 6456 #ifdef FEAT_MENU 6457 if (gui.menu_is_active) 6458 pre_area.y += gui.menu_height; 6459 #endif 6460 pre_area.height = gui.char_height; 6461 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6462 } 6463 else if (input_style & XIMPreeditPosition) /* over-the-spot */ 6464 { 6465 pre_area.x = 0; 6466 pre_area.y = 0; 6467 pre_area.height = gui.char_height * Rows; 6468 pre_area.width = gui.char_width * Columns; 6469 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6470 } 6471 6472 if (preedit_list && status_list) 6473 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6474 XNStatusAttributes, status_list, NULL); 6475 else if (preedit_list) 6476 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6477 NULL); 6478 else if (status_list) 6479 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, 6480 NULL); 6481 else 6482 list = NULL; 6483 6484 if (list) 6485 { 6486 XSetICValues(xic, XNVaNestedList, list, NULL); 6487 XFree(list); 6488 } 6489 if (status_list) 6490 XFree(status_list); 6491 if (preedit_list) 6492 XFree(preedit_list); 6493 } 6494 6495 int 6496 xim_get_status_area_height(void) 6497 { 6498 if (status_area_enabled) 6499 return gui.char_height; 6500 return 0; 6501 } 6502 # endif 6503 6504 #else /* !defined(FEAT_XIM) */ 6505 6506 # ifdef IME_WITHOUT_XIM 6507 static int im_was_set_active = FALSE; 6508 6509 int 6510 im_get_status(void) 6511 { 6512 # if defined(FEAT_MBYTE) && defined(FEAT_EVAL) 6513 if (USE_IMSTATUSFUNC) 6514 return call_imstatusfunc(); 6515 # endif 6516 return im_was_set_active; 6517 } 6518 6519 void 6520 im_set_active(int active_arg) 6521 { 6522 # if defined(FEAT_MBYTE) && defined(FEAT_EVAL) 6523 int active = !p_imdisable && active_arg; 6524 6525 if (USE_IMACTIVATEFUNC && active != im_get_status()) 6526 { 6527 call_imactivatefunc(active); 6528 im_was_set_active = active; 6529 } 6530 # endif 6531 } 6532 6533 # ifdef FEAT_GUI 6534 void 6535 im_set_position(int row UNUSED, int col UNUSED) 6536 { 6537 } 6538 # endif 6539 # endif 6540 6541 #endif /* FEAT_XIM */ 6542 6543 #if defined(FEAT_MBYTE) || defined(PROTO) 6544 6545 /* 6546 * Setup "vcp" for conversion from "from" to "to". 6547 * The names must have been made canonical with enc_canonize(). 6548 * vcp->vc_type must have been initialized to CONV_NONE. 6549 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 6550 * instead). 6551 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 6552 * Return FAIL when conversion is not supported, OK otherwise. 6553 */ 6554 int 6555 convert_setup(vimconv_T *vcp, char_u *from, char_u *to) 6556 { 6557 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 6558 } 6559 6560 /* 6561 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 6562 * "from" unicode charsets be considered utf-8. Same for "to". 6563 */ 6564 int 6565 convert_setup_ext( 6566 vimconv_T *vcp, 6567 char_u *from, 6568 int from_unicode_is_utf8, 6569 char_u *to, 6570 int to_unicode_is_utf8) 6571 { 6572 int from_prop; 6573 int to_prop; 6574 int from_is_utf8; 6575 int to_is_utf8; 6576 6577 /* Reset to no conversion. */ 6578 # ifdef USE_ICONV 6579 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 6580 iconv_close(vcp->vc_fd); 6581 # endif 6582 vcp->vc_type = CONV_NONE; 6583 vcp->vc_factor = 1; 6584 vcp->vc_fail = FALSE; 6585 6586 /* No conversion when one of the names is empty or they are equal. */ 6587 if (from == NULL || *from == NUL || to == NULL || *to == NUL 6588 || STRCMP(from, to) == 0) 6589 return OK; 6590 6591 from_prop = enc_canon_props(from); 6592 to_prop = enc_canon_props(to); 6593 if (from_unicode_is_utf8) 6594 from_is_utf8 = from_prop & ENC_UNICODE; 6595 else 6596 from_is_utf8 = from_prop == ENC_UNICODE; 6597 if (to_unicode_is_utf8) 6598 to_is_utf8 = to_prop & ENC_UNICODE; 6599 else 6600 to_is_utf8 = to_prop == ENC_UNICODE; 6601 6602 if ((from_prop & ENC_LATIN1) && to_is_utf8) 6603 { 6604 /* Internal latin1 -> utf-8 conversion. */ 6605 vcp->vc_type = CONV_TO_UTF8; 6606 vcp->vc_factor = 2; /* up to twice as long */ 6607 } 6608 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 6609 { 6610 /* Internal latin9 -> utf-8 conversion. */ 6611 vcp->vc_type = CONV_9_TO_UTF8; 6612 vcp->vc_factor = 3; /* up to three as long (euro sign) */ 6613 } 6614 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 6615 { 6616 /* Internal utf-8 -> latin1 conversion. */ 6617 vcp->vc_type = CONV_TO_LATIN1; 6618 } 6619 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 6620 { 6621 /* Internal utf-8 -> latin9 conversion. */ 6622 vcp->vc_type = CONV_TO_LATIN9; 6623 } 6624 #ifdef WIN3264 6625 /* Win32-specific codepage <-> codepage conversion without iconv. */ 6626 else if ((from_is_utf8 || encname2codepage(from) > 0) 6627 && (to_is_utf8 || encname2codepage(to) > 0)) 6628 { 6629 vcp->vc_type = CONV_CODEPAGE; 6630 vcp->vc_factor = 2; /* up to twice as long */ 6631 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 6632 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 6633 } 6634 #endif 6635 #ifdef MACOS_CONVERT 6636 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 6637 { 6638 vcp->vc_type = CONV_MAC_LATIN1; 6639 } 6640 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 6641 { 6642 vcp->vc_type = CONV_MAC_UTF8; 6643 vcp->vc_factor = 2; /* up to twice as long */ 6644 } 6645 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 6646 { 6647 vcp->vc_type = CONV_LATIN1_MAC; 6648 } 6649 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 6650 { 6651 vcp->vc_type = CONV_UTF8_MAC; 6652 } 6653 #endif 6654 # ifdef USE_ICONV 6655 else 6656 { 6657 /* Use iconv() for conversion. */ 6658 vcp->vc_fd = (iconv_t)my_iconv_open( 6659 to_is_utf8 ? (char_u *)"utf-8" : to, 6660 from_is_utf8 ? (char_u *)"utf-8" : from); 6661 if (vcp->vc_fd != (iconv_t)-1) 6662 { 6663 vcp->vc_type = CONV_ICONV; 6664 vcp->vc_factor = 4; /* could be longer too... */ 6665 } 6666 } 6667 # endif 6668 if (vcp->vc_type == CONV_NONE) 6669 return FAIL; 6670 6671 return OK; 6672 } 6673 6674 #if defined(FEAT_GUI) || defined(AMIGA) || defined(WIN3264) \ 6675 || defined(PROTO) 6676 /* 6677 * Do conversion on typed input characters in-place. 6678 * The input and output are not NUL terminated! 6679 * Returns the length after conversion. 6680 */ 6681 int 6682 convert_input(char_u *ptr, int len, int maxlen) 6683 { 6684 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 6685 } 6686 #endif 6687 6688 /* 6689 * Like convert_input(), but when there is an incomplete byte sequence at the 6690 * end return that as an allocated string in "restp" and set "*restlenp" to 6691 * the length. If "restp" is NULL it is not used. 6692 */ 6693 int 6694 convert_input_safe( 6695 char_u *ptr, 6696 int len, 6697 int maxlen, 6698 char_u **restp, 6699 int *restlenp) 6700 { 6701 char_u *d; 6702 int dlen = len; 6703 int unconvertlen = 0; 6704 6705 d = string_convert_ext(&input_conv, ptr, &dlen, 6706 restp == NULL ? NULL : &unconvertlen); 6707 if (d != NULL) 6708 { 6709 if (dlen <= maxlen) 6710 { 6711 if (unconvertlen > 0) 6712 { 6713 /* Move the unconverted characters to allocated memory. */ 6714 *restp = alloc(unconvertlen); 6715 if (*restp != NULL) 6716 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 6717 *restlenp = unconvertlen; 6718 } 6719 mch_memmove(ptr, d, dlen); 6720 } 6721 else 6722 /* result is too long, keep the unconverted text (the caller must 6723 * have done something wrong!) */ 6724 dlen = len; 6725 vim_free(d); 6726 } 6727 return dlen; 6728 } 6729 6730 /* 6731 * Convert text "ptr[*lenp]" according to "vcp". 6732 * Returns the result in allocated memory and sets "*lenp". 6733 * When "lenp" is NULL, use NUL terminated strings. 6734 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 6735 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 6736 */ 6737 char_u * 6738 string_convert( 6739 vimconv_T *vcp, 6740 char_u *ptr, 6741 int *lenp) 6742 { 6743 return string_convert_ext(vcp, ptr, lenp, NULL); 6744 } 6745 6746 /* 6747 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 6748 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 6749 * set to the number of remaining bytes. 6750 */ 6751 char_u * 6752 string_convert_ext( 6753 vimconv_T *vcp, 6754 char_u *ptr, 6755 int *lenp, 6756 int *unconvlenp) 6757 { 6758 char_u *retval = NULL; 6759 char_u *d; 6760 int len; 6761 int i; 6762 int l; 6763 int c; 6764 6765 if (lenp == NULL) 6766 len = (int)STRLEN(ptr); 6767 else 6768 len = *lenp; 6769 if (len == 0) 6770 return vim_strsave((char_u *)""); 6771 6772 switch (vcp->vc_type) 6773 { 6774 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */ 6775 retval = alloc(len * 2 + 1); 6776 if (retval == NULL) 6777 break; 6778 d = retval; 6779 for (i = 0; i < len; ++i) 6780 { 6781 c = ptr[i]; 6782 if (c < 0x80) 6783 *d++ = c; 6784 else 6785 { 6786 *d++ = 0xc0 + ((unsigned)c >> 6); 6787 *d++ = 0x80 + (c & 0x3f); 6788 } 6789 } 6790 *d = NUL; 6791 if (lenp != NULL) 6792 *lenp = (int)(d - retval); 6793 break; 6794 6795 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */ 6796 retval = alloc(len * 3 + 1); 6797 if (retval == NULL) 6798 break; 6799 d = retval; 6800 for (i = 0; i < len; ++i) 6801 { 6802 c = ptr[i]; 6803 switch (c) 6804 { 6805 case 0xa4: c = 0x20ac; break; /* euro */ 6806 case 0xa6: c = 0x0160; break; /* S hat */ 6807 case 0xa8: c = 0x0161; break; /* S -hat */ 6808 case 0xb4: c = 0x017d; break; /* Z hat */ 6809 case 0xb8: c = 0x017e; break; /* Z -hat */ 6810 case 0xbc: c = 0x0152; break; /* OE */ 6811 case 0xbd: c = 0x0153; break; /* oe */ 6812 case 0xbe: c = 0x0178; break; /* Y */ 6813 } 6814 d += utf_char2bytes(c, d); 6815 } 6816 *d = NUL; 6817 if (lenp != NULL) 6818 *lenp = (int)(d - retval); 6819 break; 6820 6821 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */ 6822 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */ 6823 retval = alloc(len + 1); 6824 if (retval == NULL) 6825 break; 6826 d = retval; 6827 for (i = 0; i < len; ++i) 6828 { 6829 l = utf_ptr2len_len(ptr + i, len - i); 6830 if (l == 0) 6831 *d++ = NUL; 6832 else if (l == 1) 6833 { 6834 int l_w = utf8len_tab_zero[ptr[i]]; 6835 6836 if (l_w == 0) 6837 { 6838 /* Illegal utf-8 byte cannot be converted */ 6839 vim_free(retval); 6840 return NULL; 6841 } 6842 if (unconvlenp != NULL && l_w > len - i) 6843 { 6844 /* Incomplete sequence at the end. */ 6845 *unconvlenp = len - i; 6846 break; 6847 } 6848 *d++ = ptr[i]; 6849 } 6850 else 6851 { 6852 c = utf_ptr2char(ptr + i); 6853 if (vcp->vc_type == CONV_TO_LATIN9) 6854 switch (c) 6855 { 6856 case 0x20ac: c = 0xa4; break; /* euro */ 6857 case 0x0160: c = 0xa6; break; /* S hat */ 6858 case 0x0161: c = 0xa8; break; /* S -hat */ 6859 case 0x017d: c = 0xb4; break; /* Z hat */ 6860 case 0x017e: c = 0xb8; break; /* Z -hat */ 6861 case 0x0152: c = 0xbc; break; /* OE */ 6862 case 0x0153: c = 0xbd; break; /* oe */ 6863 case 0x0178: c = 0xbe; break; /* Y */ 6864 case 0xa4: 6865 case 0xa6: 6866 case 0xa8: 6867 case 0xb4: 6868 case 0xb8: 6869 case 0xbc: 6870 case 0xbd: 6871 case 0xbe: c = 0x100; break; /* not in latin9 */ 6872 } 6873 if (!utf_iscomposing(c)) /* skip composing chars */ 6874 { 6875 if (c < 0x100) 6876 *d++ = c; 6877 else if (vcp->vc_fail) 6878 { 6879 vim_free(retval); 6880 return NULL; 6881 } 6882 else 6883 { 6884 *d++ = 0xbf; 6885 if (utf_char2cells(c) > 1) 6886 *d++ = '?'; 6887 } 6888 } 6889 i += l - 1; 6890 } 6891 } 6892 *d = NUL; 6893 if (lenp != NULL) 6894 *lenp = (int)(d - retval); 6895 break; 6896 6897 # ifdef MACOS_CONVERT 6898 case CONV_MAC_LATIN1: 6899 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6900 'm', 'l', unconvlenp); 6901 break; 6902 6903 case CONV_LATIN1_MAC: 6904 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6905 'l', 'm', unconvlenp); 6906 break; 6907 6908 case CONV_MAC_UTF8: 6909 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6910 'm', 'u', unconvlenp); 6911 break; 6912 6913 case CONV_UTF8_MAC: 6914 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6915 'u', 'm', unconvlenp); 6916 break; 6917 # endif 6918 6919 # ifdef USE_ICONV 6920 case CONV_ICONV: /* conversion with output_conv.vc_fd */ 6921 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 6922 break; 6923 # endif 6924 # ifdef WIN3264 6925 case CONV_CODEPAGE: /* codepage -> codepage */ 6926 { 6927 int retlen; 6928 int tmp_len; 6929 short_u *tmp; 6930 6931 /* 1. codepage/UTF-8 -> ucs-2. */ 6932 if (vcp->vc_cpfrom == 0) 6933 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 6934 else 6935 { 6936 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 6937 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 6938 (char *)ptr, len, 0, 0); 6939 if (tmp_len == 0 6940 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 6941 { 6942 if (lenp != NULL) 6943 *lenp = 0; 6944 if (unconvlenp != NULL) 6945 *unconvlenp = len; 6946 retval = alloc(1); 6947 if (retval) 6948 retval[0] = NUL; 6949 return retval; 6950 } 6951 } 6952 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len); 6953 if (tmp == NULL) 6954 break; 6955 if (vcp->vc_cpfrom == 0) 6956 utf8_to_utf16(ptr, len, tmp, unconvlenp); 6957 else 6958 MultiByteToWideChar(vcp->vc_cpfrom, 0, 6959 (char *)ptr, len, tmp, tmp_len); 6960 6961 /* 2. ucs-2 -> codepage/UTF-8. */ 6962 if (vcp->vc_cpto == 0) 6963 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 6964 else 6965 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 6966 tmp, tmp_len, 0, 0, 0, 0); 6967 retval = alloc(retlen + 1); 6968 if (retval != NULL) 6969 { 6970 if (vcp->vc_cpto == 0) 6971 utf16_to_utf8(tmp, tmp_len, retval); 6972 else 6973 WideCharToMultiByte(vcp->vc_cpto, 0, 6974 tmp, tmp_len, 6975 (char *)retval, retlen, 0, 0); 6976 retval[retlen] = NUL; 6977 if (lenp != NULL) 6978 *lenp = retlen; 6979 } 6980 vim_free(tmp); 6981 break; 6982 } 6983 # endif 6984 } 6985 6986 return retval; 6987 } 6988 #endif 6989