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(MSWIN) || 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 MSWIN 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 static int dbcs_char2len(int c); 140 static int dbcs_char2bytes(int c, char_u *buf); 141 static int dbcs_ptr2len(char_u *p); 142 static int dbcs_ptr2len_len(char_u *p, int size); 143 static int utf_ptr2cells_len(char_u *p, int size); 144 static int dbcs_char2cells(int c); 145 static int dbcs_ptr2cells_len(char_u *p, int size); 146 static int dbcs_ptr2char(char_u *p); 147 148 /* 149 * Lookup table to quickly get the length in bytes of a UTF-8 character from 150 * the first byte of a UTF-8 string. 151 * Bytes which are illegal when used as the first byte have a 1. 152 * The NUL byte has length 1. 153 */ 154 static char utf8len_tab[256] = 155 { 156 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 157 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 158 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 159 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 160 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 161 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 162 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, 163 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, 164 }; 165 166 /* 167 * Like utf8len_tab above, but using a zero for illegal lead bytes. 168 */ 169 static char utf8len_tab_zero[256] = 170 { 171 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 172 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 173 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 174 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 175 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, 176 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 177 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, 178 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, 179 }; 180 181 /* 182 * XIM often causes trouble. Define XIM_DEBUG to get a log of XIM callbacks 183 * in the "xim.log" file. 184 */ 185 /* #define XIM_DEBUG */ 186 #ifdef XIM_DEBUG 187 static void 188 xim_log(char *s, ...) 189 { 190 va_list arglist; 191 static FILE *fd = NULL; 192 193 if (fd == (FILE *)-1) 194 return; 195 if (fd == NULL) 196 { 197 fd = mch_fopen("xim.log", "w"); 198 if (fd == NULL) 199 { 200 emsg("Cannot open xim.log"); 201 fd = (FILE *)-1; 202 return; 203 } 204 } 205 206 va_start(arglist, s); 207 vfprintf(fd, s, arglist); 208 va_end(arglist); 209 } 210 #endif 211 212 213 /* 214 * Canonical encoding names and their properties. 215 * "iso-8859-n" is handled by enc_canonize() directly. 216 */ 217 static struct 218 { char *name; int prop; int codepage;} 219 enc_canon_table[] = 220 { 221 #define IDX_LATIN_1 0 222 {"latin1", ENC_8BIT + ENC_LATIN1, 1252}, 223 #define IDX_ISO_2 1 224 {"iso-8859-2", ENC_8BIT, 0}, 225 #define IDX_ISO_3 2 226 {"iso-8859-3", ENC_8BIT, 0}, 227 #define IDX_ISO_4 3 228 {"iso-8859-4", ENC_8BIT, 0}, 229 #define IDX_ISO_5 4 230 {"iso-8859-5", ENC_8BIT, 0}, 231 #define IDX_ISO_6 5 232 {"iso-8859-6", ENC_8BIT, 0}, 233 #define IDX_ISO_7 6 234 {"iso-8859-7", ENC_8BIT, 0}, 235 #define IDX_ISO_8 7 236 {"iso-8859-8", ENC_8BIT, 0}, 237 #define IDX_ISO_9 8 238 {"iso-8859-9", ENC_8BIT, 0}, 239 #define IDX_ISO_10 9 240 {"iso-8859-10", ENC_8BIT, 0}, 241 #define IDX_ISO_11 10 242 {"iso-8859-11", ENC_8BIT, 0}, 243 #define IDX_ISO_13 11 244 {"iso-8859-13", ENC_8BIT, 0}, 245 #define IDX_ISO_14 12 246 {"iso-8859-14", ENC_8BIT, 0}, 247 #define IDX_ISO_15 13 248 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0}, 249 #define IDX_KOI8_R 14 250 {"koi8-r", ENC_8BIT, 0}, 251 #define IDX_KOI8_U 15 252 {"koi8-u", ENC_8BIT, 0}, 253 #define IDX_UTF8 16 254 {"utf-8", ENC_UNICODE, 0}, 255 #define IDX_UCS2 17 256 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0}, 257 #define IDX_UCS2LE 18 258 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0}, 259 #define IDX_UTF16 19 260 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0}, 261 #define IDX_UTF16LE 20 262 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0}, 263 #define IDX_UCS4 21 264 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0}, 265 #define IDX_UCS4LE 22 266 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0}, 267 268 /* For debugging DBCS encoding on Unix. */ 269 #define IDX_DEBUG 23 270 {"debug", ENC_DBCS, DBCS_DEBUG}, 271 #define IDX_EUC_JP 24 272 {"euc-jp", ENC_DBCS, DBCS_JPNU}, 273 #define IDX_SJIS 25 274 {"sjis", ENC_DBCS, DBCS_JPN}, 275 #define IDX_EUC_KR 26 276 {"euc-kr", ENC_DBCS, DBCS_KORU}, 277 #define IDX_EUC_CN 27 278 {"euc-cn", ENC_DBCS, DBCS_CHSU}, 279 #define IDX_EUC_TW 28 280 {"euc-tw", ENC_DBCS, DBCS_CHTU}, 281 #define IDX_BIG5 29 282 {"big5", ENC_DBCS, DBCS_CHT}, 283 284 /* MS-DOS and MS-Windows codepages are included here, so that they can be 285 * used on Unix too. Most of them are similar to ISO-8859 encodings, but 286 * not exactly the same. */ 287 #define IDX_CP437 30 288 {"cp437", ENC_8BIT, 437}, /* like iso-8859-1 */ 289 #define IDX_CP737 31 290 {"cp737", ENC_8BIT, 737}, /* like iso-8859-7 */ 291 #define IDX_CP775 32 292 {"cp775", ENC_8BIT, 775}, /* Baltic */ 293 #define IDX_CP850 33 294 {"cp850", ENC_8BIT, 850}, /* like iso-8859-4 */ 295 #define IDX_CP852 34 296 {"cp852", ENC_8BIT, 852}, /* like iso-8859-1 */ 297 #define IDX_CP855 35 298 {"cp855", ENC_8BIT, 855}, /* like iso-8859-2 */ 299 #define IDX_CP857 36 300 {"cp857", ENC_8BIT, 857}, /* like iso-8859-5 */ 301 #define IDX_CP860 37 302 {"cp860", ENC_8BIT, 860}, /* like iso-8859-9 */ 303 #define IDX_CP861 38 304 {"cp861", ENC_8BIT, 861}, /* like iso-8859-1 */ 305 #define IDX_CP862 39 306 {"cp862", ENC_8BIT, 862}, /* like iso-8859-1 */ 307 #define IDX_CP863 40 308 {"cp863", ENC_8BIT, 863}, /* like iso-8859-8 */ 309 #define IDX_CP865 41 310 {"cp865", ENC_8BIT, 865}, /* like iso-8859-1 */ 311 #define IDX_CP866 42 312 {"cp866", ENC_8BIT, 866}, /* like iso-8859-5 */ 313 #define IDX_CP869 43 314 {"cp869", ENC_8BIT, 869}, /* like iso-8859-7 */ 315 #define IDX_CP874 44 316 {"cp874", ENC_8BIT, 874}, /* Thai */ 317 #define IDX_CP932 45 318 {"cp932", ENC_DBCS, DBCS_JPN}, 319 #define IDX_CP936 46 320 {"cp936", ENC_DBCS, DBCS_CHS}, 321 #define IDX_CP949 47 322 {"cp949", ENC_DBCS, DBCS_KOR}, 323 #define IDX_CP950 48 324 {"cp950", ENC_DBCS, DBCS_CHT}, 325 #define IDX_CP1250 49 326 {"cp1250", ENC_8BIT, 1250}, /* Czech, Polish, etc. */ 327 #define IDX_CP1251 50 328 {"cp1251", ENC_8BIT, 1251}, /* Cyrillic */ 329 /* cp1252 is considered to be equal to latin1 */ 330 #define IDX_CP1253 51 331 {"cp1253", ENC_8BIT, 1253}, /* Greek */ 332 #define IDX_CP1254 52 333 {"cp1254", ENC_8BIT, 1254}, /* Turkish */ 334 #define IDX_CP1255 53 335 {"cp1255", ENC_8BIT, 1255}, /* Hebrew */ 336 #define IDX_CP1256 54 337 {"cp1256", ENC_8BIT, 1256}, /* Arabic */ 338 #define IDX_CP1257 55 339 {"cp1257", ENC_8BIT, 1257}, /* Baltic */ 340 #define IDX_CP1258 56 341 {"cp1258", ENC_8BIT, 1258}, /* Vietnamese */ 342 343 #define IDX_MACROMAN 57 344 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, /* Mac OS */ 345 #define IDX_DECMCS 58 346 {"dec-mcs", ENC_8BIT, 0}, /* DEC MCS */ 347 #define IDX_HPROMAN8 59 348 {"hp-roman8", ENC_8BIT, 0}, /* HP Roman8 */ 349 #define IDX_COUNT 60 350 }; 351 352 /* 353 * Aliases for encoding names. 354 */ 355 static struct 356 { char *name; int canon;} 357 enc_alias_table[] = 358 { 359 {"ansi", IDX_LATIN_1}, 360 {"iso-8859-1", IDX_LATIN_1}, 361 {"latin2", IDX_ISO_2}, 362 {"latin3", IDX_ISO_3}, 363 {"latin4", IDX_ISO_4}, 364 {"cyrillic", IDX_ISO_5}, 365 {"arabic", IDX_ISO_6}, 366 {"greek", IDX_ISO_7}, 367 #ifdef MSWIN 368 {"hebrew", IDX_CP1255}, 369 #else 370 {"hebrew", IDX_ISO_8}, 371 #endif 372 {"latin5", IDX_ISO_9}, 373 {"turkish", IDX_ISO_9}, /* ? */ 374 {"latin6", IDX_ISO_10}, 375 {"nordic", IDX_ISO_10}, /* ? */ 376 {"thai", IDX_ISO_11}, /* ? */ 377 {"latin7", IDX_ISO_13}, 378 {"latin8", IDX_ISO_14}, 379 {"latin9", IDX_ISO_15}, 380 {"utf8", IDX_UTF8}, 381 {"unicode", IDX_UCS2}, 382 {"ucs2", IDX_UCS2}, 383 {"ucs2be", IDX_UCS2}, 384 {"ucs-2be", IDX_UCS2}, 385 {"ucs2le", IDX_UCS2LE}, 386 {"utf16", IDX_UTF16}, 387 {"utf16be", IDX_UTF16}, 388 {"utf-16be", IDX_UTF16}, 389 {"utf16le", IDX_UTF16LE}, 390 {"ucs4", IDX_UCS4}, 391 {"ucs4be", IDX_UCS4}, 392 {"ucs-4be", IDX_UCS4}, 393 {"ucs4le", IDX_UCS4LE}, 394 {"utf32", IDX_UCS4}, 395 {"utf-32", IDX_UCS4}, 396 {"utf32be", IDX_UCS4}, 397 {"utf-32be", IDX_UCS4}, 398 {"utf32le", IDX_UCS4LE}, 399 {"utf-32le", IDX_UCS4LE}, 400 {"932", IDX_CP932}, 401 {"949", IDX_CP949}, 402 {"936", IDX_CP936}, 403 {"gbk", IDX_CP936}, 404 {"950", IDX_CP950}, 405 {"eucjp", IDX_EUC_JP}, 406 {"unix-jis", IDX_EUC_JP}, 407 {"ujis", IDX_EUC_JP}, 408 {"shift-jis", IDX_SJIS}, 409 {"pck", IDX_SJIS}, /* Sun: PCK */ 410 {"euckr", IDX_EUC_KR}, 411 {"5601", IDX_EUC_KR}, /* Sun: KS C 5601 */ 412 {"euccn", IDX_EUC_CN}, 413 {"gb2312", IDX_EUC_CN}, 414 {"euctw", IDX_EUC_TW}, 415 #if defined(MSWIN) || defined(WIN32UNIX) || defined(MACOS_X) 416 {"japan", IDX_CP932}, 417 {"korea", IDX_CP949}, 418 {"prc", IDX_CP936}, 419 {"chinese", IDX_CP936}, 420 {"taiwan", IDX_CP950}, 421 {"big5", IDX_CP950}, 422 #else 423 {"japan", IDX_EUC_JP}, 424 {"korea", IDX_EUC_KR}, 425 {"prc", IDX_EUC_CN}, 426 {"chinese", IDX_EUC_CN}, 427 {"taiwan", IDX_EUC_TW}, 428 {"cp950", IDX_BIG5}, 429 {"950", IDX_BIG5}, 430 #endif 431 {"mac", IDX_MACROMAN}, 432 {"mac-roman", IDX_MACROMAN}, 433 {NULL, 0} 434 }; 435 436 #ifndef CP_UTF8 437 # define CP_UTF8 65001 /* magic number from winnls.h */ 438 #endif 439 440 /* 441 * Find encoding "name" in the list of canonical encoding names. 442 * Returns -1 if not found. 443 */ 444 static int 445 enc_canon_search(char_u *name) 446 { 447 int i; 448 449 for (i = 0; i < IDX_COUNT; ++i) 450 if (STRCMP(name, enc_canon_table[i].name) == 0) 451 return i; 452 return -1; 453 } 454 455 456 /* 457 * Find canonical encoding "name" in the list and return its properties. 458 * Returns 0 if not found. 459 */ 460 int 461 enc_canon_props(char_u *name) 462 { 463 int i; 464 465 i = enc_canon_search(name); 466 if (i >= 0) 467 return enc_canon_table[i].prop; 468 #ifdef MSWIN 469 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2])) 470 { 471 CPINFO cpinfo; 472 473 /* Get info on this codepage to find out what it is. */ 474 if (GetCPInfo(atoi((char *)name + 2), &cpinfo) != 0) 475 { 476 if (cpinfo.MaxCharSize == 1) /* some single-byte encoding */ 477 return ENC_8BIT; 478 if (cpinfo.MaxCharSize == 2 479 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 480 /* must be a DBCS encoding */ 481 return ENC_DBCS; 482 } 483 return 0; 484 } 485 #endif 486 if (STRNCMP(name, "2byte-", 6) == 0) 487 return ENC_DBCS; 488 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0) 489 return ENC_8BIT; 490 return 0; 491 } 492 493 /* 494 * Set up for using multi-byte characters. 495 * Called in three cases: 496 * - by main() to initialize (p_enc == NULL) 497 * - by set_init_1() after 'encoding' was set to its default. 498 * - by do_set() when 'encoding' has been set. 499 * p_enc must have been passed through enc_canonize() already. 500 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags. 501 * Fills mb_bytelen_tab[] and returns NULL when there are no problems. 502 * When there is something wrong: Returns an error message and doesn't change 503 * anything. 504 */ 505 char * 506 mb_init(void) 507 { 508 int i; 509 int idx; 510 int n; 511 int enc_dbcs_new = 0; 512 #if defined(USE_ICONV) && !defined(MSWIN) && !defined(WIN32UNIX) \ 513 && !defined(MACOS_CONVERT) 514 # define LEN_FROM_CONV 515 vimconv_T vimconv; 516 char_u *p; 517 #endif 518 519 if (p_enc == NULL) 520 { 521 /* Just starting up: set the whole table to one's. */ 522 for (i = 0; i < 256; ++i) 523 mb_bytelen_tab[i] = 1; 524 input_conv.vc_type = CONV_NONE; 525 input_conv.vc_factor = 1; 526 output_conv.vc_type = CONV_NONE; 527 return NULL; 528 } 529 530 #ifdef MSWIN 531 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2])) 532 { 533 CPINFO cpinfo; 534 535 /* Get info on this codepage to find out what it is. */ 536 if (GetCPInfo(atoi((char *)p_enc + 2), &cpinfo) != 0) 537 { 538 if (cpinfo.MaxCharSize == 1) 539 { 540 /* some single-byte encoding */ 541 enc_unicode = 0; 542 enc_utf8 = FALSE; 543 } 544 else if (cpinfo.MaxCharSize == 2 545 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 546 { 547 /* must be a DBCS encoding, check below */ 548 enc_dbcs_new = atoi((char *)p_enc + 2); 549 } 550 else 551 goto codepage_invalid; 552 } 553 else if (GetLastError() == ERROR_INVALID_PARAMETER) 554 { 555 codepage_invalid: 556 return N_("E543: Not a valid codepage"); 557 } 558 } 559 #endif 560 else if (STRNCMP(p_enc, "8bit-", 5) == 0 561 || STRNCMP(p_enc, "iso-8859-", 9) == 0) 562 { 563 /* Accept any "8bit-" or "iso-8859-" name. */ 564 enc_unicode = 0; 565 enc_utf8 = FALSE; 566 } 567 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 568 { 569 #ifdef MSWIN 570 /* Windows: accept only valid codepage numbers, check below. */ 571 if (p_enc[6] != 'c' || p_enc[7] != 'p' 572 || (enc_dbcs_new = atoi((char *)p_enc + 8)) == 0) 573 return e_invarg; 574 #else 575 /* Unix: accept any "2byte-" name, assume current locale. */ 576 enc_dbcs_new = DBCS_2BYTE; 577 #endif 578 } 579 else if ((idx = enc_canon_search(p_enc)) >= 0) 580 { 581 i = enc_canon_table[idx].prop; 582 if (i & ENC_UNICODE) 583 { 584 /* Unicode */ 585 enc_utf8 = TRUE; 586 if (i & (ENC_2BYTE | ENC_2WORD)) 587 enc_unicode = 2; 588 else if (i & ENC_4BYTE) 589 enc_unicode = 4; 590 else 591 enc_unicode = 0; 592 } 593 else if (i & ENC_DBCS) 594 { 595 /* 2byte, handle below */ 596 enc_dbcs_new = enc_canon_table[idx].codepage; 597 } 598 else 599 { 600 /* Must be 8-bit. */ 601 enc_unicode = 0; 602 enc_utf8 = FALSE; 603 } 604 } 605 else /* Don't know what encoding this is, reject it. */ 606 return e_invarg; 607 608 if (enc_dbcs_new != 0) 609 { 610 #ifdef MSWIN 611 /* Check if the DBCS code page is OK. */ 612 if (!IsValidCodePage(enc_dbcs_new)) 613 goto codepage_invalid; 614 #endif 615 enc_unicode = 0; 616 enc_utf8 = FALSE; 617 } 618 enc_dbcs = enc_dbcs_new; 619 has_mbyte = (enc_dbcs != 0 || enc_utf8); 620 621 #if defined(MSWIN) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 622 enc_codepage = encname2codepage(p_enc); 623 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0); 624 #endif 625 626 /* Detect an encoding that uses latin1 characters. */ 627 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0 628 || STRCMP(p_enc, "iso-8859-15") == 0); 629 630 /* 631 * Set the function pointers. 632 */ 633 if (enc_utf8) 634 { 635 mb_ptr2len = utfc_ptr2len; 636 mb_ptr2len_len = utfc_ptr2len_len; 637 mb_char2len = utf_char2len; 638 mb_char2bytes = utf_char2bytes; 639 mb_ptr2cells = utf_ptr2cells; 640 mb_ptr2cells_len = utf_ptr2cells_len; 641 mb_char2cells = utf_char2cells; 642 mb_off2cells = utf_off2cells; 643 mb_ptr2char = utf_ptr2char; 644 mb_head_off = utf_head_off; 645 } 646 else if (enc_dbcs != 0) 647 { 648 mb_ptr2len = dbcs_ptr2len; 649 mb_ptr2len_len = dbcs_ptr2len_len; 650 mb_char2len = dbcs_char2len; 651 mb_char2bytes = dbcs_char2bytes; 652 mb_ptr2cells = dbcs_ptr2cells; 653 mb_ptr2cells_len = dbcs_ptr2cells_len; 654 mb_char2cells = dbcs_char2cells; 655 mb_off2cells = dbcs_off2cells; 656 mb_ptr2char = dbcs_ptr2char; 657 mb_head_off = dbcs_head_off; 658 } 659 else 660 { 661 mb_ptr2len = latin_ptr2len; 662 mb_ptr2len_len = latin_ptr2len_len; 663 mb_char2len = latin_char2len; 664 mb_char2bytes = latin_char2bytes; 665 mb_ptr2cells = latin_ptr2cells; 666 mb_ptr2cells_len = latin_ptr2cells_len; 667 mb_char2cells = latin_char2cells; 668 mb_off2cells = latin_off2cells; 669 mb_ptr2char = latin_ptr2char; 670 mb_head_off = latin_head_off; 671 } 672 673 /* 674 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN(). 675 */ 676 #ifdef LEN_FROM_CONV 677 /* When 'encoding' is different from the current locale mblen() won't 678 * work. Use conversion to "utf-8" instead. */ 679 vimconv.vc_type = CONV_NONE; 680 if (enc_dbcs) 681 { 682 p = enc_locale(); 683 if (p == NULL || STRCMP(p, p_enc) != 0) 684 { 685 convert_setup(&vimconv, p_enc, (char_u *)"utf-8"); 686 vimconv.vc_fail = TRUE; 687 } 688 vim_free(p); 689 } 690 #endif 691 692 for (i = 0; i < 256; ++i) 693 { 694 /* Our own function to reliably check the length of UTF-8 characters, 695 * independent of mblen(). */ 696 if (enc_utf8) 697 n = utf8len_tab[i]; 698 else if (enc_dbcs == 0) 699 n = 1; 700 else 701 { 702 #if defined(MSWIN) || defined(WIN32UNIX) 703 /* enc_dbcs is set by setting 'fileencoding'. It becomes a Windows 704 * CodePage identifier, which we can pass directly in to Windows 705 * API */ 706 n = IsDBCSLeadByteEx(enc_dbcs, (WINBYTE)i) ? 2 : 1; 707 #else 708 # if defined(__amigaos4__) || defined(__ANDROID__) || \ 709 !(defined(HAVE_MBLEN) || defined(X_LOCALE)) 710 /* 711 * if mblen() is not available, character which MSB is turned on 712 * are treated as leading byte character. (note : This assumption 713 * is not always true.) 714 */ 715 n = (i & 0x80) ? 2 : 1; 716 # else 717 char buf[MB_MAXBYTES + 1]; 718 719 if (i == NUL) /* just in case mblen() can't handle "" */ 720 n = 1; 721 else 722 { 723 buf[0] = i; 724 buf[1] = 0; 725 # ifdef LEN_FROM_CONV 726 if (vimconv.vc_type != CONV_NONE) 727 { 728 /* 729 * string_convert() should fail when converting the first 730 * byte of a double-byte character. 731 */ 732 p = string_convert(&vimconv, (char_u *)buf, NULL); 733 if (p != NULL) 734 { 735 vim_free(p); 736 n = 1; 737 } 738 else 739 n = 2; 740 } 741 else 742 # endif 743 { 744 /* 745 * mblen() should return -1 for invalid (means the leading 746 * multibyte) character. However there are some platforms 747 * where mblen() returns 0 for invalid character. 748 * Therefore, following condition includes 0. 749 */ 750 vim_ignored = mblen(NULL, 0); // First reset the state. 751 if (mblen(buf, (size_t)1) <= 0) 752 n = 2; 753 else 754 n = 1; 755 } 756 } 757 # endif 758 #endif 759 } 760 761 mb_bytelen_tab[i] = n; 762 } 763 764 #ifdef LEN_FROM_CONV 765 convert_setup(&vimconv, NULL, NULL); 766 #endif 767 768 /* The cell width depends on the type of multi-byte characters. */ 769 (void)init_chartab(); 770 771 /* When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] */ 772 screenalloc(FALSE); 773 774 /* When using Unicode, set default for 'fileencodings'. */ 775 if (enc_utf8 && !option_was_set((char_u *)"fencs")) 776 set_string_option_direct((char_u *)"fencs", -1, 777 (char_u *)"ucs-bom,utf-8,default,latin1", OPT_FREE, 0); 778 779 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT) 780 /* GNU gettext 0.10.37 supports this feature: set the codeset used for 781 * translated messages independently from the current locale. */ 782 (void)bind_textdomain_codeset(VIMPACKAGE, 783 enc_utf8 ? "utf-8" : (char *)p_enc); 784 #endif 785 786 #ifdef MSWIN 787 /* When changing 'encoding' while starting up, then convert the command 788 * line arguments from the active codepage to 'encoding'. */ 789 if (starting != 0) 790 fix_arg_enc(); 791 #endif 792 793 /* Fire an autocommand to let people do custom font setup. This must be 794 * after Vim has been setup for the new encoding. */ 795 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); 796 797 #ifdef FEAT_SPELL 798 /* Need to reload spell dictionaries */ 799 spell_reload(); 800 #endif 801 802 return NULL; 803 } 804 805 /* 806 * Return the size of the BOM for the current buffer: 807 * 0 - no BOM 808 * 2 - UCS-2 or UTF-16 BOM 809 * 4 - UCS-4 BOM 810 * 3 - UTF-8 BOM 811 */ 812 int 813 bomb_size(void) 814 { 815 int n = 0; 816 817 if (curbuf->b_p_bomb && !curbuf->b_p_bin) 818 { 819 if (*curbuf->b_p_fenc == NUL) 820 { 821 if (enc_utf8) 822 { 823 if (enc_unicode != 0) 824 n = enc_unicode; 825 else 826 n = 3; 827 } 828 } 829 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) 830 n = 3; 831 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 832 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) 833 n = 2; 834 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) 835 n = 4; 836 } 837 return n; 838 } 839 840 #if defined(FEAT_QUICKFIX) || defined(PROTO) 841 /* 842 * Remove all BOM from "s" by moving remaining text. 843 */ 844 void 845 remove_bom(char_u *s) 846 { 847 if (enc_utf8) 848 { 849 char_u *p = s; 850 851 while ((p = vim_strbyte(p, 0xef)) != NULL) 852 { 853 if (p[1] == 0xbb && p[2] == 0xbf) 854 STRMOVE(p, p + 3); 855 else 856 ++p; 857 } 858 } 859 } 860 #endif 861 862 /* 863 * Get class of pointer: 864 * 0 for blank or NUL 865 * 1 for punctuation 866 * 2 for an (ASCII) word character 867 * >2 for other word characters 868 */ 869 int 870 mb_get_class(char_u *p) 871 { 872 return mb_get_class_buf(p, curbuf); 873 } 874 875 int 876 mb_get_class_buf(char_u *p, buf_T *buf) 877 { 878 if (MB_BYTE2LEN(p[0]) == 1) 879 { 880 if (p[0] == NUL || VIM_ISWHITE(p[0])) 881 return 0; 882 if (vim_iswordc_buf(p[0], buf)) 883 return 2; 884 return 1; 885 } 886 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) 887 return dbcs_class(p[0], p[1]); 888 if (enc_utf8) 889 return utf_class_buf(utf_ptr2char(p), buf); 890 return 0; 891 } 892 893 /* 894 * Get class of a double-byte character. This always returns 3 or bigger. 895 * TODO: Should return 1 for punctuation. 896 */ 897 int 898 dbcs_class(unsigned lead, unsigned trail) 899 { 900 switch (enc_dbcs) 901 { 902 /* please add classify routine for your language in here */ 903 904 case DBCS_JPNU: /* ? */ 905 case DBCS_JPN: 906 { 907 /* JIS code classification */ 908 unsigned char lb = lead; 909 unsigned char tb = trail; 910 911 /* convert process code to JIS */ 912 # if defined(MSWIN) || defined(WIN32UNIX) || defined(MACOS_X) 913 /* process code is SJIS */ 914 if (lb <= 0x9f) 915 lb = (lb - 0x81) * 2 + 0x21; 916 else 917 lb = (lb - 0xc1) * 2 + 0x21; 918 if (tb <= 0x7e) 919 tb -= 0x1f; 920 else if (tb <= 0x9e) 921 tb -= 0x20; 922 else 923 { 924 tb -= 0x7e; 925 lb += 1; 926 } 927 # else 928 /* 929 * XXX: Code page identification can not use with all 930 * system! So, some other encoding information 931 * will be needed. 932 * In japanese: SJIS,EUC,UNICODE,(JIS) 933 * Note that JIS-code system don't use as 934 * process code in most system because it uses 935 * escape sequences(JIS is context depend encoding). 936 */ 937 /* assume process code is JAPANESE-EUC */ 938 lb &= 0x7f; 939 tb &= 0x7f; 940 # endif 941 /* exceptions */ 942 switch (lb << 8 | tb) 943 { 944 case 0x2121: /* ZENKAKU space */ 945 return 0; 946 case 0x2122: /* TOU-TEN (Japanese comma) */ 947 case 0x2123: /* KU-TEN (Japanese period) */ 948 case 0x2124: /* ZENKAKU comma */ 949 case 0x2125: /* ZENKAKU period */ 950 return 1; 951 case 0x213c: /* prolongedsound handled as KATAKANA */ 952 return 13; 953 } 954 /* sieved by KU code */ 955 switch (lb) 956 { 957 case 0x21: 958 case 0x22: 959 /* special symbols */ 960 return 10; 961 case 0x23: 962 /* alpha-numeric */ 963 return 11; 964 case 0x24: 965 /* hiragana */ 966 return 12; 967 case 0x25: 968 /* katakana */ 969 return 13; 970 case 0x26: 971 /* greek */ 972 return 14; 973 case 0x27: 974 /* russian */ 975 return 15; 976 case 0x28: 977 /* lines */ 978 return 16; 979 default: 980 /* kanji */ 981 return 17; 982 } 983 } 984 985 case DBCS_KORU: /* ? */ 986 case DBCS_KOR: 987 { 988 /* KS code classification */ 989 unsigned char c1 = lead; 990 unsigned char c2 = trail; 991 992 /* 993 * 20 : Hangul 994 * 21 : Hanja 995 * 22 : Symbols 996 * 23 : Alpha-numeric/Roman Letter (Full width) 997 * 24 : Hangul Letter(Alphabet) 998 * 25 : Roman Numeral/Greek Letter 999 * 26 : Box Drawings 1000 * 27 : Unit Symbols 1001 * 28 : Circled/Parenthesized Letter 1002 * 29 : Hiragana/Katakana 1003 * 30 : Cyrillic Letter 1004 */ 1005 1006 if (c1 >= 0xB0 && c1 <= 0xC8) 1007 /* Hangul */ 1008 return 20; 1009 #if defined(MSWIN) || defined(WIN32UNIX) 1010 else if (c1 <= 0xA0 || c2 <= 0xA0) 1011 /* Extended Hangul Region : MS UHC(Unified Hangul Code) */ 1012 /* c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE 1013 * c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 1014 */ 1015 return 20; 1016 #endif 1017 1018 else if (c1 >= 0xCA && c1 <= 0xFD) 1019 /* Hanja */ 1020 return 21; 1021 else switch (c1) 1022 { 1023 case 0xA1: 1024 case 0xA2: 1025 /* Symbols */ 1026 return 22; 1027 case 0xA3: 1028 /* Alpha-numeric */ 1029 return 23; 1030 case 0xA4: 1031 /* Hangul Letter(Alphabet) */ 1032 return 24; 1033 case 0xA5: 1034 /* Roman Numeral/Greek Letter */ 1035 return 25; 1036 case 0xA6: 1037 /* Box Drawings */ 1038 return 26; 1039 case 0xA7: 1040 /* Unit Symbols */ 1041 return 27; 1042 case 0xA8: 1043 case 0xA9: 1044 if (c2 <= 0xAF) 1045 return 25; /* Roman Letter */ 1046 else if (c2 >= 0xF6) 1047 return 22; /* Symbols */ 1048 else 1049 /* Circled/Parenthesized Letter */ 1050 return 28; 1051 case 0xAA: 1052 case 0xAB: 1053 /* Hiragana/Katakana */ 1054 return 29; 1055 case 0xAC: 1056 /* Cyrillic Letter */ 1057 return 30; 1058 } 1059 } 1060 default: 1061 break; 1062 } 1063 return 3; 1064 } 1065 1066 /* 1067 * mb_char2len() function pointer. 1068 * Return length in bytes of character "c". 1069 * Returns 1 for a single-byte character. 1070 */ 1071 int 1072 latin_char2len(int c UNUSED) 1073 { 1074 return 1; 1075 } 1076 1077 static int 1078 dbcs_char2len( 1079 int c) 1080 { 1081 if (c >= 0x100) 1082 return 2; 1083 return 1; 1084 } 1085 1086 /* 1087 * mb_char2bytes() function pointer. 1088 * Convert a character to its bytes. 1089 * Returns the length in bytes. 1090 */ 1091 int 1092 latin_char2bytes(int c, char_u *buf) 1093 { 1094 buf[0] = c; 1095 return 1; 1096 } 1097 1098 static int 1099 dbcs_char2bytes(int c, char_u *buf) 1100 { 1101 if (c >= 0x100) 1102 { 1103 buf[0] = (unsigned)c >> 8; 1104 buf[1] = c; 1105 /* Never use a NUL byte, it causes lots of trouble. It's an invalid 1106 * character anyway. */ 1107 if (buf[1] == NUL) 1108 buf[1] = '\n'; 1109 return 2; 1110 } 1111 buf[0] = c; 1112 return 1; 1113 } 1114 1115 /* 1116 * mb_ptr2len() function pointer. 1117 * Get byte length of character at "*p" but stop at a NUL. 1118 * For UTF-8 this includes following composing characters. 1119 * Returns 0 when *p is NUL. 1120 */ 1121 int 1122 latin_ptr2len(char_u *p) 1123 { 1124 return MB_BYTE2LEN(*p); 1125 } 1126 1127 static int 1128 dbcs_ptr2len( 1129 char_u *p) 1130 { 1131 int len; 1132 1133 /* Check if second byte is not missing. */ 1134 len = MB_BYTE2LEN(*p); 1135 if (len == 2 && p[1] == NUL) 1136 len = 1; 1137 return len; 1138 } 1139 1140 /* 1141 * mb_ptr2len_len() function pointer. 1142 * Like mb_ptr2len(), but limit to read "size" bytes. 1143 * Returns 0 for an empty string. 1144 * Returns 1 for an illegal char or an incomplete byte sequence. 1145 */ 1146 int 1147 latin_ptr2len_len(char_u *p, int size) 1148 { 1149 if (size < 1 || *p == NUL) 1150 return 0; 1151 return 1; 1152 } 1153 1154 static int 1155 dbcs_ptr2len_len(char_u *p, int size) 1156 { 1157 int len; 1158 1159 if (size < 1 || *p == NUL) 1160 return 0; 1161 if (size == 1) 1162 return 1; 1163 /* Check that second byte is not missing. */ 1164 len = MB_BYTE2LEN(*p); 1165 if (len == 2 && p[1] == NUL) 1166 len = 1; 1167 return len; 1168 } 1169 1170 struct interval 1171 { 1172 long first; 1173 long last; 1174 }; 1175 1176 /* 1177 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]". 1178 */ 1179 static int 1180 intable(struct interval *table, size_t size, int c) 1181 { 1182 int mid, bot, top; 1183 1184 /* first quick check for Latin1 etc. characters */ 1185 if (c < table[0].first) 1186 return FALSE; 1187 1188 /* binary search in table */ 1189 bot = 0; 1190 top = (int)(size / sizeof(struct interval) - 1); 1191 while (top >= bot) 1192 { 1193 mid = (bot + top) / 2; 1194 if (table[mid].last < c) 1195 bot = mid + 1; 1196 else if (table[mid].first > c) 1197 top = mid - 1; 1198 else 1199 return TRUE; 1200 } 1201 return FALSE; 1202 } 1203 1204 /* Sorted list of non-overlapping intervals of East Asian Ambiguous 1205 * characters, generated with ../runtime/tools/unicode.vim. */ 1206 static struct interval ambiguous[] = 1207 { 1208 {0x00a1, 0x00a1}, 1209 {0x00a4, 0x00a4}, 1210 {0x00a7, 0x00a8}, 1211 {0x00aa, 0x00aa}, 1212 {0x00ad, 0x00ae}, 1213 {0x00b0, 0x00b4}, 1214 {0x00b6, 0x00ba}, 1215 {0x00bc, 0x00bf}, 1216 {0x00c6, 0x00c6}, 1217 {0x00d0, 0x00d0}, 1218 {0x00d7, 0x00d8}, 1219 {0x00de, 0x00e1}, 1220 {0x00e6, 0x00e6}, 1221 {0x00e8, 0x00ea}, 1222 {0x00ec, 0x00ed}, 1223 {0x00f0, 0x00f0}, 1224 {0x00f2, 0x00f3}, 1225 {0x00f7, 0x00fa}, 1226 {0x00fc, 0x00fc}, 1227 {0x00fe, 0x00fe}, 1228 {0x0101, 0x0101}, 1229 {0x0111, 0x0111}, 1230 {0x0113, 0x0113}, 1231 {0x011b, 0x011b}, 1232 {0x0126, 0x0127}, 1233 {0x012b, 0x012b}, 1234 {0x0131, 0x0133}, 1235 {0x0138, 0x0138}, 1236 {0x013f, 0x0142}, 1237 {0x0144, 0x0144}, 1238 {0x0148, 0x014b}, 1239 {0x014d, 0x014d}, 1240 {0x0152, 0x0153}, 1241 {0x0166, 0x0167}, 1242 {0x016b, 0x016b}, 1243 {0x01ce, 0x01ce}, 1244 {0x01d0, 0x01d0}, 1245 {0x01d2, 0x01d2}, 1246 {0x01d4, 0x01d4}, 1247 {0x01d6, 0x01d6}, 1248 {0x01d8, 0x01d8}, 1249 {0x01da, 0x01da}, 1250 {0x01dc, 0x01dc}, 1251 {0x0251, 0x0251}, 1252 {0x0261, 0x0261}, 1253 {0x02c4, 0x02c4}, 1254 {0x02c7, 0x02c7}, 1255 {0x02c9, 0x02cb}, 1256 {0x02cd, 0x02cd}, 1257 {0x02d0, 0x02d0}, 1258 {0x02d8, 0x02db}, 1259 {0x02dd, 0x02dd}, 1260 {0x02df, 0x02df}, 1261 {0x0300, 0x036f}, 1262 {0x0391, 0x03a1}, 1263 {0x03a3, 0x03a9}, 1264 {0x03b1, 0x03c1}, 1265 {0x03c3, 0x03c9}, 1266 {0x0401, 0x0401}, 1267 {0x0410, 0x044f}, 1268 {0x0451, 0x0451}, 1269 {0x2010, 0x2010}, 1270 {0x2013, 0x2016}, 1271 {0x2018, 0x2019}, 1272 {0x201c, 0x201d}, 1273 {0x2020, 0x2022}, 1274 {0x2024, 0x2027}, 1275 {0x2030, 0x2030}, 1276 {0x2032, 0x2033}, 1277 {0x2035, 0x2035}, 1278 {0x203b, 0x203b}, 1279 {0x203e, 0x203e}, 1280 {0x2074, 0x2074}, 1281 {0x207f, 0x207f}, 1282 {0x2081, 0x2084}, 1283 {0x20ac, 0x20ac}, 1284 {0x2103, 0x2103}, 1285 {0x2105, 0x2105}, 1286 {0x2109, 0x2109}, 1287 {0x2113, 0x2113}, 1288 {0x2116, 0x2116}, 1289 {0x2121, 0x2122}, 1290 {0x2126, 0x2126}, 1291 {0x212b, 0x212b}, 1292 {0x2153, 0x2154}, 1293 {0x215b, 0x215e}, 1294 {0x2160, 0x216b}, 1295 {0x2170, 0x2179}, 1296 {0x2189, 0x2189}, 1297 {0x2190, 0x2199}, 1298 {0x21b8, 0x21b9}, 1299 {0x21d2, 0x21d2}, 1300 {0x21d4, 0x21d4}, 1301 {0x21e7, 0x21e7}, 1302 {0x2200, 0x2200}, 1303 {0x2202, 0x2203}, 1304 {0x2207, 0x2208}, 1305 {0x220b, 0x220b}, 1306 {0x220f, 0x220f}, 1307 {0x2211, 0x2211}, 1308 {0x2215, 0x2215}, 1309 {0x221a, 0x221a}, 1310 {0x221d, 0x2220}, 1311 {0x2223, 0x2223}, 1312 {0x2225, 0x2225}, 1313 {0x2227, 0x222c}, 1314 {0x222e, 0x222e}, 1315 {0x2234, 0x2237}, 1316 {0x223c, 0x223d}, 1317 {0x2248, 0x2248}, 1318 {0x224c, 0x224c}, 1319 {0x2252, 0x2252}, 1320 {0x2260, 0x2261}, 1321 {0x2264, 0x2267}, 1322 {0x226a, 0x226b}, 1323 {0x226e, 0x226f}, 1324 {0x2282, 0x2283}, 1325 {0x2286, 0x2287}, 1326 {0x2295, 0x2295}, 1327 {0x2299, 0x2299}, 1328 {0x22a5, 0x22a5}, 1329 {0x22bf, 0x22bf}, 1330 {0x2312, 0x2312}, 1331 {0x2460, 0x24e9}, 1332 {0x24eb, 0x254b}, 1333 {0x2550, 0x2573}, 1334 {0x2580, 0x258f}, 1335 {0x2592, 0x2595}, 1336 {0x25a0, 0x25a1}, 1337 {0x25a3, 0x25a9}, 1338 {0x25b2, 0x25b3}, 1339 {0x25b6, 0x25b7}, 1340 {0x25bc, 0x25bd}, 1341 {0x25c0, 0x25c1}, 1342 {0x25c6, 0x25c8}, 1343 {0x25cb, 0x25cb}, 1344 {0x25ce, 0x25d1}, 1345 {0x25e2, 0x25e5}, 1346 {0x25ef, 0x25ef}, 1347 {0x2605, 0x2606}, 1348 {0x2609, 0x2609}, 1349 {0x260e, 0x260f}, 1350 {0x261c, 0x261c}, 1351 {0x261e, 0x261e}, 1352 {0x2640, 0x2640}, 1353 {0x2642, 0x2642}, 1354 {0x2660, 0x2661}, 1355 {0x2663, 0x2665}, 1356 {0x2667, 0x266a}, 1357 {0x266c, 0x266d}, 1358 {0x266f, 0x266f}, 1359 {0x269e, 0x269f}, 1360 {0x26bf, 0x26bf}, 1361 {0x26c6, 0x26cd}, 1362 {0x26cf, 0x26d3}, 1363 {0x26d5, 0x26e1}, 1364 {0x26e3, 0x26e3}, 1365 {0x26e8, 0x26e9}, 1366 {0x26eb, 0x26f1}, 1367 {0x26f4, 0x26f4}, 1368 {0x26f6, 0x26f9}, 1369 {0x26fb, 0x26fc}, 1370 {0x26fe, 0x26ff}, 1371 {0x273d, 0x273d}, 1372 {0x2776, 0x277f}, 1373 {0x2b56, 0x2b59}, 1374 {0x3248, 0x324f}, 1375 {0xe000, 0xf8ff}, 1376 {0xfe00, 0xfe0f}, 1377 {0xfffd, 0xfffd}, 1378 {0x1f100, 0x1f10a}, 1379 {0x1f110, 0x1f12d}, 1380 {0x1f130, 0x1f169}, 1381 {0x1f170, 0x1f18d}, 1382 {0x1f18f, 0x1f190}, 1383 {0x1f19b, 0x1f1ac}, 1384 {0xe0100, 0xe01ef}, 1385 {0xf0000, 0xffffd}, 1386 {0x100000, 0x10fffd} 1387 }; 1388 1389 #if defined(FEAT_TERMINAL) || defined(PROTO) 1390 /* 1391 * utf_char2cells() with different argument type for libvterm. 1392 */ 1393 int 1394 utf_uint2cells(UINT32_T c) 1395 { 1396 if (c >= 0x100 && utf_iscomposing((int)c)) 1397 return 0; 1398 return utf_char2cells((int)c); 1399 } 1400 #endif 1401 1402 /* 1403 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1404 * Returns 4 or 6 for an unprintable character. 1405 * Is only correct for characters >= 0x80. 1406 * When p_ambw is "double", return 2 for a character with East Asian Width 1407 * class 'A'(mbiguous). 1408 */ 1409 int 1410 utf_char2cells(int c) 1411 { 1412 /* Sorted list of non-overlapping intervals of East Asian double width 1413 * characters, generated with ../runtime/tools/unicode.vim. */ 1414 static struct interval doublewidth[] = 1415 { 1416 {0x1100, 0x115f}, 1417 {0x231a, 0x231b}, 1418 {0x2329, 0x232a}, 1419 {0x23e9, 0x23ec}, 1420 {0x23f0, 0x23f0}, 1421 {0x23f3, 0x23f3}, 1422 {0x25fd, 0x25fe}, 1423 {0x2614, 0x2615}, 1424 {0x2648, 0x2653}, 1425 {0x267f, 0x267f}, 1426 {0x2693, 0x2693}, 1427 {0x26a1, 0x26a1}, 1428 {0x26aa, 0x26ab}, 1429 {0x26bd, 0x26be}, 1430 {0x26c4, 0x26c5}, 1431 {0x26ce, 0x26ce}, 1432 {0x26d4, 0x26d4}, 1433 {0x26ea, 0x26ea}, 1434 {0x26f2, 0x26f3}, 1435 {0x26f5, 0x26f5}, 1436 {0x26fa, 0x26fa}, 1437 {0x26fd, 0x26fd}, 1438 {0x2705, 0x2705}, 1439 {0x270a, 0x270b}, 1440 {0x2728, 0x2728}, 1441 {0x274c, 0x274c}, 1442 {0x274e, 0x274e}, 1443 {0x2753, 0x2755}, 1444 {0x2757, 0x2757}, 1445 {0x2795, 0x2797}, 1446 {0x27b0, 0x27b0}, 1447 {0x27bf, 0x27bf}, 1448 {0x2b1b, 0x2b1c}, 1449 {0x2b50, 0x2b50}, 1450 {0x2b55, 0x2b55}, 1451 {0x2e80, 0x2e99}, 1452 {0x2e9b, 0x2ef3}, 1453 {0x2f00, 0x2fd5}, 1454 {0x2ff0, 0x2ffb}, 1455 {0x3000, 0x303e}, 1456 {0x3041, 0x3096}, 1457 {0x3099, 0x30ff}, 1458 {0x3105, 0x312f}, 1459 {0x3131, 0x318e}, 1460 {0x3190, 0x31ba}, 1461 {0x31c0, 0x31e3}, 1462 {0x31f0, 0x321e}, 1463 {0x3220, 0x3247}, 1464 {0x3250, 0x32fe}, 1465 {0x3300, 0x4dbf}, 1466 {0x4e00, 0xa48c}, 1467 {0xa490, 0xa4c6}, 1468 {0xa960, 0xa97c}, 1469 {0xac00, 0xd7a3}, 1470 {0xf900, 0xfaff}, 1471 {0xfe10, 0xfe19}, 1472 {0xfe30, 0xfe52}, 1473 {0xfe54, 0xfe66}, 1474 {0xfe68, 0xfe6b}, 1475 {0xff01, 0xff60}, 1476 {0xffe0, 0xffe6}, 1477 {0x16fe0, 0x16fe3}, 1478 {0x17000, 0x187f7}, 1479 {0x18800, 0x18af2}, 1480 {0x1b000, 0x1b11e}, 1481 {0x1b150, 0x1b152}, 1482 {0x1b164, 0x1b167}, 1483 {0x1b170, 0x1b2fb}, 1484 {0x1f004, 0x1f004}, 1485 {0x1f0cf, 0x1f0cf}, 1486 {0x1f18e, 0x1f18e}, 1487 {0x1f191, 0x1f19a}, 1488 {0x1f200, 0x1f202}, 1489 {0x1f210, 0x1f23b}, 1490 {0x1f240, 0x1f248}, 1491 {0x1f250, 0x1f251}, 1492 {0x1f260, 0x1f265}, 1493 {0x1f300, 0x1f320}, 1494 {0x1f32d, 0x1f335}, 1495 {0x1f337, 0x1f37c}, 1496 {0x1f37e, 0x1f393}, 1497 {0x1f3a0, 0x1f3ca}, 1498 {0x1f3cf, 0x1f3d3}, 1499 {0x1f3e0, 0x1f3f0}, 1500 {0x1f3f4, 0x1f3f4}, 1501 {0x1f3f8, 0x1f43e}, 1502 {0x1f440, 0x1f440}, 1503 {0x1f442, 0x1f4fc}, 1504 {0x1f4ff, 0x1f53d}, 1505 {0x1f54b, 0x1f54e}, 1506 {0x1f550, 0x1f567}, 1507 {0x1f57a, 0x1f57a}, 1508 {0x1f595, 0x1f596}, 1509 {0x1f5a4, 0x1f5a4}, 1510 {0x1f5fb, 0x1f64f}, 1511 {0x1f680, 0x1f6c5}, 1512 {0x1f6cc, 0x1f6cc}, 1513 {0x1f6d0, 0x1f6d2}, 1514 {0x1f6d5, 0x1f6d5}, 1515 {0x1f6eb, 0x1f6ec}, 1516 {0x1f6f4, 0x1f6fa}, 1517 {0x1f7e0, 0x1f7eb}, 1518 {0x1f90d, 0x1f971}, 1519 {0x1f973, 0x1f976}, 1520 {0x1f97a, 0x1f9a2}, 1521 {0x1f9a5, 0x1f9aa}, 1522 {0x1f9ae, 0x1f9ca}, 1523 {0x1f9cd, 0x1f9ff}, 1524 {0x1fa70, 0x1fa73}, 1525 {0x1fa78, 0x1fa7a}, 1526 {0x1fa80, 0x1fa82}, 1527 {0x1fa90, 0x1fa95}, 1528 {0x20000, 0x2fffd}, 1529 {0x30000, 0x3fffd} 1530 }; 1531 1532 /* Sorted list of non-overlapping intervals of Emoji characters that don't 1533 * have ambiguous or double width, 1534 * based on http://unicode.org/emoji/charts/emoji-list.html */ 1535 static struct interval emoji_width[] = 1536 { 1537 {0x1f1e6, 0x1f1ff}, 1538 {0x1f321, 0x1f321}, 1539 {0x1f324, 0x1f32c}, 1540 {0x1f336, 0x1f336}, 1541 {0x1f37d, 0x1f37d}, 1542 {0x1f396, 0x1f397}, 1543 {0x1f399, 0x1f39b}, 1544 {0x1f39e, 0x1f39f}, 1545 {0x1f3cb, 0x1f3ce}, 1546 {0x1f3d4, 0x1f3df}, 1547 {0x1f3f3, 0x1f3f5}, 1548 {0x1f3f7, 0x1f3f7}, 1549 {0x1f43f, 0x1f43f}, 1550 {0x1f441, 0x1f441}, 1551 {0x1f4fd, 0x1f4fd}, 1552 {0x1f549, 0x1f54a}, 1553 {0x1f56f, 0x1f570}, 1554 {0x1f573, 0x1f579}, 1555 {0x1f587, 0x1f587}, 1556 {0x1f58a, 0x1f58d}, 1557 {0x1f590, 0x1f590}, 1558 {0x1f5a5, 0x1f5a5}, 1559 {0x1f5a8, 0x1f5a8}, 1560 {0x1f5b1, 0x1f5b2}, 1561 {0x1f5bc, 0x1f5bc}, 1562 {0x1f5c2, 0x1f5c4}, 1563 {0x1f5d1, 0x1f5d3}, 1564 {0x1f5dc, 0x1f5de}, 1565 {0x1f5e1, 0x1f5e1}, 1566 {0x1f5e3, 0x1f5e3}, 1567 {0x1f5e8, 0x1f5e8}, 1568 {0x1f5ef, 0x1f5ef}, 1569 {0x1f5f3, 0x1f5f3}, 1570 {0x1f5fa, 0x1f5fa}, 1571 {0x1f6cb, 0x1f6cf}, 1572 {0x1f6e0, 0x1f6e5}, 1573 {0x1f6e9, 0x1f6e9}, 1574 {0x1f6f0, 0x1f6f0}, 1575 {0x1f6f3, 0x1f6f3} 1576 }; 1577 1578 if (c >= 0x100) 1579 { 1580 #ifdef USE_WCHAR_FUNCTIONS 1581 /* 1582 * Assume the library function wcwidth() works better than our own 1583 * stuff. It should return 1 for ambiguous width chars! 1584 */ 1585 int n = wcwidth(c); 1586 1587 if (n < 0) 1588 return 6; /* unprintable, displays <xxxx> */ 1589 if (n > 1) 1590 return n; 1591 #else 1592 if (!utf_printable(c)) 1593 return 6; /* unprintable, displays <xxxx> */ 1594 if (intable(doublewidth, sizeof(doublewidth), c)) 1595 return 2; 1596 #endif 1597 if (p_emoji && intable(emoji_width, sizeof(emoji_width), c)) 1598 return 2; 1599 } 1600 1601 /* Characters below 0x100 are influenced by 'isprint' option */ 1602 else if (c >= 0x80 && !vim_isprintc(c)) 1603 return 4; /* unprintable, displays <xx> */ 1604 1605 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c)) 1606 return 2; 1607 1608 return 1; 1609 } 1610 1611 /* 1612 * mb_ptr2cells() function pointer. 1613 * Return the number of display cells character at "*p" occupies. 1614 * This doesn't take care of unprintable characters, use ptr2cells() for that. 1615 */ 1616 int 1617 latin_ptr2cells(char_u *p UNUSED) 1618 { 1619 return 1; 1620 } 1621 1622 int 1623 utf_ptr2cells( 1624 char_u *p) 1625 { 1626 int c; 1627 1628 /* Need to convert to a wide character. */ 1629 if (*p >= 0x80) 1630 { 1631 c = utf_ptr2char(p); 1632 /* An illegal byte is displayed as <xx>. */ 1633 if (utf_ptr2len(p) == 1 || c == NUL) 1634 return 4; 1635 /* If the char is ASCII it must be an overlong sequence. */ 1636 if (c < 0x80) 1637 return char2cells(c); 1638 return utf_char2cells(c); 1639 } 1640 return 1; 1641 } 1642 1643 int 1644 dbcs_ptr2cells(char_u *p) 1645 { 1646 /* Number of cells is equal to number of bytes, except for euc-jp when 1647 * the first byte is 0x8e. */ 1648 if (enc_dbcs == DBCS_JPNU && *p == 0x8e) 1649 return 1; 1650 return MB_BYTE2LEN(*p); 1651 } 1652 1653 /* 1654 * mb_ptr2cells_len() function pointer. 1655 * Like mb_ptr2cells(), but limit string length to "size". 1656 * For an empty string or truncated character returns 1. 1657 */ 1658 int 1659 latin_ptr2cells_len(char_u *p UNUSED, int size UNUSED) 1660 { 1661 return 1; 1662 } 1663 1664 static int 1665 utf_ptr2cells_len(char_u *p, int size) 1666 { 1667 int c; 1668 1669 /* Need to convert to a wide character. */ 1670 if (size > 0 && *p >= 0x80) 1671 { 1672 if (utf_ptr2len_len(p, size) < utf8len_tab[*p]) 1673 return 1; /* truncated */ 1674 c = utf_ptr2char(p); 1675 /* An illegal byte is displayed as <xx>. */ 1676 if (utf_ptr2len(p) == 1 || c == NUL) 1677 return 4; 1678 /* If the char is ASCII it must be an overlong sequence. */ 1679 if (c < 0x80) 1680 return char2cells(c); 1681 return utf_char2cells(c); 1682 } 1683 return 1; 1684 } 1685 1686 static int 1687 dbcs_ptr2cells_len(char_u *p, int size) 1688 { 1689 /* Number of cells is equal to number of bytes, except for euc-jp when 1690 * the first byte is 0x8e. */ 1691 if (size <= 1 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)) 1692 return 1; 1693 return MB_BYTE2LEN(*p); 1694 } 1695 1696 /* 1697 * mb_char2cells() function pointer. 1698 * Return the number of display cells character "c" occupies. 1699 * Only takes care of multi-byte chars, not "^C" and such. 1700 */ 1701 int 1702 latin_char2cells(int c UNUSED) 1703 { 1704 return 1; 1705 } 1706 1707 static int 1708 dbcs_char2cells(int c) 1709 { 1710 /* Number of cells is equal to number of bytes, except for euc-jp when 1711 * the first byte is 0x8e. */ 1712 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e) 1713 return 1; 1714 /* use the first byte */ 1715 return MB_BYTE2LEN((unsigned)c >> 8); 1716 } 1717 1718 /* 1719 * Return the number of cells occupied by string "p". 1720 * Stop at a NUL character. When "len" >= 0 stop at character "p[len]". 1721 */ 1722 int 1723 mb_string2cells(char_u *p, int len) 1724 { 1725 int i; 1726 int clen = 0; 1727 1728 for (i = 0; (len < 0 || i < len) && p[i] != NUL; i += (*mb_ptr2len)(p + i)) 1729 clen += (*mb_ptr2cells)(p + i); 1730 return clen; 1731 } 1732 1733 /* 1734 * mb_off2cells() function pointer. 1735 * Return number of display cells for char at ScreenLines[off]. 1736 * We make sure that the offset used is less than "max_off". 1737 */ 1738 int 1739 latin_off2cells(unsigned off UNUSED, unsigned max_off UNUSED) 1740 { 1741 return 1; 1742 } 1743 1744 int 1745 dbcs_off2cells(unsigned off, unsigned max_off) 1746 { 1747 /* never check beyond end of the line */ 1748 if (off >= max_off) 1749 return 1; 1750 1751 /* Number of cells is equal to number of bytes, except for euc-jp when 1752 * the first byte is 0x8e. */ 1753 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e) 1754 return 1; 1755 return MB_BYTE2LEN(ScreenLines[off]); 1756 } 1757 1758 int 1759 utf_off2cells(unsigned off, unsigned max_off) 1760 { 1761 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1; 1762 } 1763 1764 /* 1765 * mb_ptr2char() function pointer. 1766 * Convert a byte sequence into a character. 1767 */ 1768 int 1769 latin_ptr2char(char_u *p) 1770 { 1771 return *p; 1772 } 1773 1774 static int 1775 dbcs_ptr2char(char_u *p) 1776 { 1777 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL) 1778 return (p[0] << 8) + p[1]; 1779 return *p; 1780 } 1781 1782 /* 1783 * Convert a UTF-8 byte sequence to a wide character. 1784 * If the sequence is illegal or truncated by a NUL the first byte is 1785 * returned. 1786 * For an overlong sequence this may return zero. 1787 * Does not include composing characters, of course. 1788 */ 1789 int 1790 utf_ptr2char(char_u *p) 1791 { 1792 int len; 1793 1794 if (p[0] < 0x80) /* be quick for ASCII */ 1795 return p[0]; 1796 1797 len = utf8len_tab_zero[p[0]]; 1798 if (len > 1 && (p[1] & 0xc0) == 0x80) 1799 { 1800 if (len == 2) 1801 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); 1802 if ((p[2] & 0xc0) == 0x80) 1803 { 1804 if (len == 3) 1805 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) 1806 + (p[2] & 0x3f); 1807 if ((p[3] & 0xc0) == 0x80) 1808 { 1809 if (len == 4) 1810 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) 1811 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); 1812 if ((p[4] & 0xc0) == 0x80) 1813 { 1814 if (len == 5) 1815 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) 1816 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) 1817 + (p[4] & 0x3f); 1818 if ((p[5] & 0xc0) == 0x80 && len == 6) 1819 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) 1820 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) 1821 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); 1822 } 1823 } 1824 } 1825 } 1826 /* Illegal value, just return the first byte */ 1827 return p[0]; 1828 } 1829 1830 /* 1831 * Convert a UTF-8 byte sequence to a wide character. 1832 * String is assumed to be terminated by NUL or after "n" bytes, whichever 1833 * comes first. 1834 * The function is safe in the sense that it never accesses memory beyond the 1835 * first "n" bytes of "s". 1836 * 1837 * On success, returns decoded codepoint, advances "s" to the beginning of 1838 * next character and decreases "n" accordingly. 1839 * 1840 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past 1841 * NUL byte. 1842 * 1843 * If byte sequence is illegal or incomplete, returns -1 and does not advance 1844 * "s". 1845 */ 1846 static int 1847 utf_safe_read_char_adv(char_u **s, size_t *n) 1848 { 1849 int c, k; 1850 1851 if (*n == 0) /* end of buffer */ 1852 return 0; 1853 1854 k = utf8len_tab_zero[**s]; 1855 1856 if (k == 1) 1857 { 1858 /* ASCII character or NUL */ 1859 (*n)--; 1860 return *(*s)++; 1861 } 1862 1863 if ((size_t)k <= *n) 1864 { 1865 /* We have a multibyte sequence and it isn't truncated by buffer 1866 * limits so utf_ptr2char() is safe to use. Or the first byte is 1867 * illegal (k=0), and it's also safe to use utf_ptr2char(). */ 1868 c = utf_ptr2char(*s); 1869 1870 /* On failure, utf_ptr2char() returns the first byte, so here we 1871 * check equality with the first byte. The only non-ASCII character 1872 * which equals the first byte of its own UTF-8 representation is 1873 * U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too. 1874 * It's safe even if n=1, else we would have k=2 > n. */ 1875 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83)) 1876 { 1877 /* byte sequence was successfully decoded */ 1878 *s += k; 1879 *n -= k; 1880 return c; 1881 } 1882 } 1883 1884 /* byte sequence is incomplete or illegal */ 1885 return -1; 1886 } 1887 1888 /* 1889 * Get character at **pp and advance *pp to the next character. 1890 * Note: composing characters are skipped! 1891 */ 1892 int 1893 mb_ptr2char_adv(char_u **pp) 1894 { 1895 int c; 1896 1897 c = (*mb_ptr2char)(*pp); 1898 *pp += (*mb_ptr2len)(*pp); 1899 return c; 1900 } 1901 1902 /* 1903 * Get character at **pp and advance *pp to the next character. 1904 * Note: composing characters are returned as separate characters. 1905 */ 1906 int 1907 mb_cptr2char_adv(char_u **pp) 1908 { 1909 int c; 1910 1911 c = (*mb_ptr2char)(*pp); 1912 if (enc_utf8) 1913 *pp += utf_ptr2len(*pp); 1914 else 1915 *pp += (*mb_ptr2len)(*pp); 1916 return c; 1917 } 1918 1919 #if defined(FEAT_ARABIC) || defined(PROTO) 1920 /* 1921 * Check if the character pointed to by "p2" is a composing character when it 1922 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which 1923 * behaves like a composing character. 1924 */ 1925 int 1926 utf_composinglike(char_u *p1, char_u *p2) 1927 { 1928 int c2; 1929 1930 c2 = utf_ptr2char(p2); 1931 if (utf_iscomposing(c2)) 1932 return TRUE; 1933 if (!arabic_maycombine(c2)) 1934 return FALSE; 1935 return arabic_combine(utf_ptr2char(p1), c2); 1936 } 1937 #endif 1938 1939 /* 1940 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1941 * composing characters. 1942 */ 1943 int 1944 utfc_ptr2char( 1945 char_u *p, 1946 int *pcc) /* return: composing chars, last one is 0 */ 1947 { 1948 int len; 1949 int c; 1950 int cc; 1951 int i = 0; 1952 1953 c = utf_ptr2char(p); 1954 len = utf_ptr2len(p); 1955 1956 /* Only accept a composing char when the first char isn't illegal. */ 1957 if ((len > 1 || *p < 0x80) 1958 && p[len] >= 0x80 1959 && UTF_COMPOSINGLIKE(p, p + len)) 1960 { 1961 cc = utf_ptr2char(p + len); 1962 for (;;) 1963 { 1964 pcc[i++] = cc; 1965 if (i == MAX_MCO) 1966 break; 1967 len += utf_ptr2len(p + len); 1968 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1969 break; 1970 } 1971 } 1972 1973 if (i < MAX_MCO) /* last composing char must be 0 */ 1974 pcc[i] = 0; 1975 1976 return c; 1977 } 1978 1979 /* 1980 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1981 * composing characters. Use no more than p[maxlen]. 1982 */ 1983 int 1984 utfc_ptr2char_len( 1985 char_u *p, 1986 int *pcc, /* return: composing chars, last one is 0 */ 1987 int maxlen) 1988 { 1989 int len; 1990 int c; 1991 int cc; 1992 int i = 0; 1993 1994 c = utf_ptr2char(p); 1995 len = utf_ptr2len_len(p, maxlen); 1996 /* Only accept a composing char when the first char isn't illegal. */ 1997 if ((len > 1 || *p < 0x80) 1998 && len < maxlen 1999 && p[len] >= 0x80 2000 && UTF_COMPOSINGLIKE(p, p + len)) 2001 { 2002 cc = utf_ptr2char(p + len); 2003 for (;;) 2004 { 2005 pcc[i++] = cc; 2006 if (i == MAX_MCO) 2007 break; 2008 len += utf_ptr2len_len(p + len, maxlen - len); 2009 if (len >= maxlen 2010 || p[len] < 0x80 2011 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 2012 break; 2013 } 2014 } 2015 2016 if (i < MAX_MCO) /* last composing char must be 0 */ 2017 pcc[i] = 0; 2018 2019 return c; 2020 } 2021 2022 /* 2023 * Convert the character at screen position "off" to a sequence of bytes. 2024 * Includes the composing characters. 2025 * "buf" must at least have the length MB_MAXBYTES + 1. 2026 * Only to be used when ScreenLinesUC[off] != 0. 2027 * Returns the produced number of bytes. 2028 */ 2029 int 2030 utfc_char2bytes(int off, char_u *buf) 2031 { 2032 int len; 2033 int i; 2034 2035 len = utf_char2bytes(ScreenLinesUC[off], buf); 2036 for (i = 0; i < Screen_mco; ++i) 2037 { 2038 if (ScreenLinesC[i][off] == 0) 2039 break; 2040 len += utf_char2bytes(ScreenLinesC[i][off], buf + len); 2041 } 2042 return len; 2043 } 2044 2045 /* 2046 * Get the length of a UTF-8 byte sequence, not including any following 2047 * composing characters. 2048 * Returns 0 for "". 2049 * Returns 1 for an illegal byte sequence. 2050 */ 2051 int 2052 utf_ptr2len(char_u *p) 2053 { 2054 int len; 2055 int i; 2056 2057 if (*p == NUL) 2058 return 0; 2059 len = utf8len_tab[*p]; 2060 for (i = 1; i < len; ++i) 2061 if ((p[i] & 0xc0) != 0x80) 2062 return 1; 2063 return len; 2064 } 2065 2066 /* 2067 * Return length of UTF-8 character, obtained from the first byte. 2068 * "b" must be between 0 and 255! 2069 * Returns 1 for an invalid first byte value. 2070 */ 2071 int 2072 utf_byte2len(int b) 2073 { 2074 return utf8len_tab[b]; 2075 } 2076 2077 /* 2078 * Get the length of UTF-8 byte sequence "p[size]". Does not include any 2079 * following composing characters. 2080 * Returns 1 for "". 2081 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.). 2082 * Returns number > "size" for an incomplete byte sequence. 2083 * Never returns zero. 2084 */ 2085 int 2086 utf_ptr2len_len(char_u *p, int size) 2087 { 2088 int len; 2089 int i; 2090 int m; 2091 2092 len = utf8len_tab[*p]; 2093 if (len == 1) 2094 return 1; /* NUL, ascii or illegal lead byte */ 2095 if (len > size) 2096 m = size; /* incomplete byte sequence. */ 2097 else 2098 m = len; 2099 for (i = 1; i < m; ++i) 2100 if ((p[i] & 0xc0) != 0x80) 2101 return 1; 2102 return len; 2103 } 2104 2105 /* 2106 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. 2107 * This includes following composing characters. 2108 */ 2109 int 2110 utfc_ptr2len(char_u *p) 2111 { 2112 int len; 2113 int b0 = *p; 2114 #ifdef FEAT_ARABIC 2115 int prevlen; 2116 #endif 2117 2118 if (b0 == NUL) 2119 return 0; 2120 if (b0 < 0x80 && p[1] < 0x80) /* be quick for ASCII */ 2121 return 1; 2122 2123 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2124 len = utf_ptr2len(p); 2125 2126 /* Check for illegal byte. */ 2127 if (len == 1 && b0 >= 0x80) 2128 return 1; 2129 2130 /* 2131 * Check for composing characters. We can handle only the first six, but 2132 * skip all of them (otherwise the cursor would get stuck). 2133 */ 2134 #ifdef FEAT_ARABIC 2135 prevlen = 0; 2136 #endif 2137 for (;;) 2138 { 2139 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2140 return len; 2141 2142 /* Skip over composing char */ 2143 #ifdef FEAT_ARABIC 2144 prevlen = len; 2145 #endif 2146 len += utf_ptr2len(p + len); 2147 } 2148 } 2149 2150 /* 2151 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" 2152 * takes. This includes following composing characters. 2153 * Returns 0 for an empty string. 2154 * Returns 1 for an illegal char or an incomplete byte sequence. 2155 */ 2156 int 2157 utfc_ptr2len_len(char_u *p, int size) 2158 { 2159 int len; 2160 #ifdef FEAT_ARABIC 2161 int prevlen; 2162 #endif 2163 2164 if (size < 1 || *p == NUL) 2165 return 0; 2166 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) /* be quick for ASCII */ 2167 return 1; 2168 2169 /* Skip over first UTF-8 char, stopping at a NUL byte. */ 2170 len = utf_ptr2len_len(p, size); 2171 2172 /* Check for illegal byte and incomplete byte sequence. */ 2173 if ((len == 1 && p[0] >= 0x80) || len > size) 2174 return 1; 2175 2176 /* 2177 * Check for composing characters. We can handle only the first six, but 2178 * skip all of them (otherwise the cursor would get stuck). 2179 */ 2180 #ifdef FEAT_ARABIC 2181 prevlen = 0; 2182 #endif 2183 while (len < size) 2184 { 2185 int len_next_char; 2186 2187 if (p[len] < 0x80) 2188 break; 2189 2190 /* 2191 * Next character length should not go beyond size to ensure that 2192 * UTF_COMPOSINGLIKE(...) does not read beyond size. 2193 */ 2194 len_next_char = utf_ptr2len_len(p + len, size - len); 2195 if (len_next_char > size - len) 2196 break; 2197 2198 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2199 break; 2200 2201 /* Skip over composing char */ 2202 #ifdef FEAT_ARABIC 2203 prevlen = len; 2204 #endif 2205 len += len_next_char; 2206 } 2207 return len; 2208 } 2209 2210 /* 2211 * Return the number of bytes the UTF-8 encoding of character "c" takes. 2212 * This does not include composing characters. 2213 */ 2214 int 2215 utf_char2len(int c) 2216 { 2217 if (c < 0x80) 2218 return 1; 2219 if (c < 0x800) 2220 return 2; 2221 if (c < 0x10000) 2222 return 3; 2223 if (c < 0x200000) 2224 return 4; 2225 if (c < 0x4000000) 2226 return 5; 2227 return 6; 2228 } 2229 2230 /* 2231 * Convert Unicode character "c" to UTF-8 string in "buf[]". 2232 * Returns the number of bytes. 2233 */ 2234 int 2235 utf_char2bytes(int c, char_u *buf) 2236 { 2237 if (c < 0x80) /* 7 bits */ 2238 { 2239 buf[0] = c; 2240 return 1; 2241 } 2242 if (c < 0x800) /* 11 bits */ 2243 { 2244 buf[0] = 0xc0 + ((unsigned)c >> 6); 2245 buf[1] = 0x80 + (c & 0x3f); 2246 return 2; 2247 } 2248 if (c < 0x10000) /* 16 bits */ 2249 { 2250 buf[0] = 0xe0 + ((unsigned)c >> 12); 2251 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2252 buf[2] = 0x80 + (c & 0x3f); 2253 return 3; 2254 } 2255 if (c < 0x200000) /* 21 bits */ 2256 { 2257 buf[0] = 0xf0 + ((unsigned)c >> 18); 2258 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2259 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2260 buf[3] = 0x80 + (c & 0x3f); 2261 return 4; 2262 } 2263 if (c < 0x4000000) /* 26 bits */ 2264 { 2265 buf[0] = 0xf8 + ((unsigned)c >> 24); 2266 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2267 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2268 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2269 buf[4] = 0x80 + (c & 0x3f); 2270 return 5; 2271 } 2272 /* 31 bits */ 2273 buf[0] = 0xfc + ((unsigned)c >> 30); 2274 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f); 2275 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2276 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2277 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2278 buf[5] = 0x80 + (c & 0x3f); 2279 return 6; 2280 } 2281 2282 #if defined(FEAT_TERMINAL) || defined(PROTO) 2283 /* 2284 * utf_iscomposing() with different argument type for libvterm. 2285 */ 2286 int 2287 utf_iscomposing_uint(UINT32_T c) 2288 { 2289 return utf_iscomposing((int)c); 2290 } 2291 #endif 2292 2293 /* 2294 * Return TRUE if "c" is a composing UTF-8 character. This means it will be 2295 * drawn on top of the preceding character. 2296 * Based on code from Markus Kuhn. 2297 */ 2298 int 2299 utf_iscomposing(int c) 2300 { 2301 /* Sorted list of non-overlapping intervals. 2302 * Generated by ../runtime/tools/unicode.vim. */ 2303 static struct interval combining[] = 2304 { 2305 {0x0300, 0x036f}, 2306 {0x0483, 0x0489}, 2307 {0x0591, 0x05bd}, 2308 {0x05bf, 0x05bf}, 2309 {0x05c1, 0x05c2}, 2310 {0x05c4, 0x05c5}, 2311 {0x05c7, 0x05c7}, 2312 {0x0610, 0x061a}, 2313 {0x064b, 0x065f}, 2314 {0x0670, 0x0670}, 2315 {0x06d6, 0x06dc}, 2316 {0x06df, 0x06e4}, 2317 {0x06e7, 0x06e8}, 2318 {0x06ea, 0x06ed}, 2319 {0x0711, 0x0711}, 2320 {0x0730, 0x074a}, 2321 {0x07a6, 0x07b0}, 2322 {0x07eb, 0x07f3}, 2323 {0x07fd, 0x07fd}, 2324 {0x0816, 0x0819}, 2325 {0x081b, 0x0823}, 2326 {0x0825, 0x0827}, 2327 {0x0829, 0x082d}, 2328 {0x0859, 0x085b}, 2329 {0x08d3, 0x08e1}, 2330 {0x08e3, 0x0903}, 2331 {0x093a, 0x093c}, 2332 {0x093e, 0x094f}, 2333 {0x0951, 0x0957}, 2334 {0x0962, 0x0963}, 2335 {0x0981, 0x0983}, 2336 {0x09bc, 0x09bc}, 2337 {0x09be, 0x09c4}, 2338 {0x09c7, 0x09c8}, 2339 {0x09cb, 0x09cd}, 2340 {0x09d7, 0x09d7}, 2341 {0x09e2, 0x09e3}, 2342 {0x09fe, 0x09fe}, 2343 {0x0a01, 0x0a03}, 2344 {0x0a3c, 0x0a3c}, 2345 {0x0a3e, 0x0a42}, 2346 {0x0a47, 0x0a48}, 2347 {0x0a4b, 0x0a4d}, 2348 {0x0a51, 0x0a51}, 2349 {0x0a70, 0x0a71}, 2350 {0x0a75, 0x0a75}, 2351 {0x0a81, 0x0a83}, 2352 {0x0abc, 0x0abc}, 2353 {0x0abe, 0x0ac5}, 2354 {0x0ac7, 0x0ac9}, 2355 {0x0acb, 0x0acd}, 2356 {0x0ae2, 0x0ae3}, 2357 {0x0afa, 0x0aff}, 2358 {0x0b01, 0x0b03}, 2359 {0x0b3c, 0x0b3c}, 2360 {0x0b3e, 0x0b44}, 2361 {0x0b47, 0x0b48}, 2362 {0x0b4b, 0x0b4d}, 2363 {0x0b56, 0x0b57}, 2364 {0x0b62, 0x0b63}, 2365 {0x0b82, 0x0b82}, 2366 {0x0bbe, 0x0bc2}, 2367 {0x0bc6, 0x0bc8}, 2368 {0x0bca, 0x0bcd}, 2369 {0x0bd7, 0x0bd7}, 2370 {0x0c00, 0x0c04}, 2371 {0x0c3e, 0x0c44}, 2372 {0x0c46, 0x0c48}, 2373 {0x0c4a, 0x0c4d}, 2374 {0x0c55, 0x0c56}, 2375 {0x0c62, 0x0c63}, 2376 {0x0c81, 0x0c83}, 2377 {0x0cbc, 0x0cbc}, 2378 {0x0cbe, 0x0cc4}, 2379 {0x0cc6, 0x0cc8}, 2380 {0x0cca, 0x0ccd}, 2381 {0x0cd5, 0x0cd6}, 2382 {0x0ce2, 0x0ce3}, 2383 {0x0d00, 0x0d03}, 2384 {0x0d3b, 0x0d3c}, 2385 {0x0d3e, 0x0d44}, 2386 {0x0d46, 0x0d48}, 2387 {0x0d4a, 0x0d4d}, 2388 {0x0d57, 0x0d57}, 2389 {0x0d62, 0x0d63}, 2390 {0x0d82, 0x0d83}, 2391 {0x0dca, 0x0dca}, 2392 {0x0dcf, 0x0dd4}, 2393 {0x0dd6, 0x0dd6}, 2394 {0x0dd8, 0x0ddf}, 2395 {0x0df2, 0x0df3}, 2396 {0x0e31, 0x0e31}, 2397 {0x0e34, 0x0e3a}, 2398 {0x0e47, 0x0e4e}, 2399 {0x0eb1, 0x0eb1}, 2400 {0x0eb4, 0x0ebc}, 2401 {0x0ec8, 0x0ecd}, 2402 {0x0f18, 0x0f19}, 2403 {0x0f35, 0x0f35}, 2404 {0x0f37, 0x0f37}, 2405 {0x0f39, 0x0f39}, 2406 {0x0f3e, 0x0f3f}, 2407 {0x0f71, 0x0f84}, 2408 {0x0f86, 0x0f87}, 2409 {0x0f8d, 0x0f97}, 2410 {0x0f99, 0x0fbc}, 2411 {0x0fc6, 0x0fc6}, 2412 {0x102b, 0x103e}, 2413 {0x1056, 0x1059}, 2414 {0x105e, 0x1060}, 2415 {0x1062, 0x1064}, 2416 {0x1067, 0x106d}, 2417 {0x1071, 0x1074}, 2418 {0x1082, 0x108d}, 2419 {0x108f, 0x108f}, 2420 {0x109a, 0x109d}, 2421 {0x135d, 0x135f}, 2422 {0x1712, 0x1714}, 2423 {0x1732, 0x1734}, 2424 {0x1752, 0x1753}, 2425 {0x1772, 0x1773}, 2426 {0x17b4, 0x17d3}, 2427 {0x17dd, 0x17dd}, 2428 {0x180b, 0x180d}, 2429 {0x1885, 0x1886}, 2430 {0x18a9, 0x18a9}, 2431 {0x1920, 0x192b}, 2432 {0x1930, 0x193b}, 2433 {0x1a17, 0x1a1b}, 2434 {0x1a55, 0x1a5e}, 2435 {0x1a60, 0x1a7c}, 2436 {0x1a7f, 0x1a7f}, 2437 {0x1ab0, 0x1abe}, 2438 {0x1b00, 0x1b04}, 2439 {0x1b34, 0x1b44}, 2440 {0x1b6b, 0x1b73}, 2441 {0x1b80, 0x1b82}, 2442 {0x1ba1, 0x1bad}, 2443 {0x1be6, 0x1bf3}, 2444 {0x1c24, 0x1c37}, 2445 {0x1cd0, 0x1cd2}, 2446 {0x1cd4, 0x1ce8}, 2447 {0x1ced, 0x1ced}, 2448 {0x1cf4, 0x1cf4}, 2449 {0x1cf7, 0x1cf9}, 2450 {0x1dc0, 0x1df9}, 2451 {0x1dfb, 0x1dff}, 2452 {0x20d0, 0x20f0}, 2453 {0x2cef, 0x2cf1}, 2454 {0x2d7f, 0x2d7f}, 2455 {0x2de0, 0x2dff}, 2456 {0x302a, 0x302f}, 2457 {0x3099, 0x309a}, 2458 {0xa66f, 0xa672}, 2459 {0xa674, 0xa67d}, 2460 {0xa69e, 0xa69f}, 2461 {0xa6f0, 0xa6f1}, 2462 {0xa802, 0xa802}, 2463 {0xa806, 0xa806}, 2464 {0xa80b, 0xa80b}, 2465 {0xa823, 0xa827}, 2466 {0xa880, 0xa881}, 2467 {0xa8b4, 0xa8c5}, 2468 {0xa8e0, 0xa8f1}, 2469 {0xa8ff, 0xa8ff}, 2470 {0xa926, 0xa92d}, 2471 {0xa947, 0xa953}, 2472 {0xa980, 0xa983}, 2473 {0xa9b3, 0xa9c0}, 2474 {0xa9e5, 0xa9e5}, 2475 {0xaa29, 0xaa36}, 2476 {0xaa43, 0xaa43}, 2477 {0xaa4c, 0xaa4d}, 2478 {0xaa7b, 0xaa7d}, 2479 {0xaab0, 0xaab0}, 2480 {0xaab2, 0xaab4}, 2481 {0xaab7, 0xaab8}, 2482 {0xaabe, 0xaabf}, 2483 {0xaac1, 0xaac1}, 2484 {0xaaeb, 0xaaef}, 2485 {0xaaf5, 0xaaf6}, 2486 {0xabe3, 0xabea}, 2487 {0xabec, 0xabed}, 2488 {0xfb1e, 0xfb1e}, 2489 {0xfe00, 0xfe0f}, 2490 {0xfe20, 0xfe2f}, 2491 {0x101fd, 0x101fd}, 2492 {0x102e0, 0x102e0}, 2493 {0x10376, 0x1037a}, 2494 {0x10a01, 0x10a03}, 2495 {0x10a05, 0x10a06}, 2496 {0x10a0c, 0x10a0f}, 2497 {0x10a38, 0x10a3a}, 2498 {0x10a3f, 0x10a3f}, 2499 {0x10ae5, 0x10ae6}, 2500 {0x10d24, 0x10d27}, 2501 {0x10f46, 0x10f50}, 2502 {0x11000, 0x11002}, 2503 {0x11038, 0x11046}, 2504 {0x1107f, 0x11082}, 2505 {0x110b0, 0x110ba}, 2506 {0x11100, 0x11102}, 2507 {0x11127, 0x11134}, 2508 {0x11145, 0x11146}, 2509 {0x11173, 0x11173}, 2510 {0x11180, 0x11182}, 2511 {0x111b3, 0x111c0}, 2512 {0x111c9, 0x111cc}, 2513 {0x1122c, 0x11237}, 2514 {0x1123e, 0x1123e}, 2515 {0x112df, 0x112ea}, 2516 {0x11300, 0x11303}, 2517 {0x1133b, 0x1133c}, 2518 {0x1133e, 0x11344}, 2519 {0x11347, 0x11348}, 2520 {0x1134b, 0x1134d}, 2521 {0x11357, 0x11357}, 2522 {0x11362, 0x11363}, 2523 {0x11366, 0x1136c}, 2524 {0x11370, 0x11374}, 2525 {0x11435, 0x11446}, 2526 {0x1145e, 0x1145e}, 2527 {0x114b0, 0x114c3}, 2528 {0x115af, 0x115b5}, 2529 {0x115b8, 0x115c0}, 2530 {0x115dc, 0x115dd}, 2531 {0x11630, 0x11640}, 2532 {0x116ab, 0x116b7}, 2533 {0x1171d, 0x1172b}, 2534 {0x1182c, 0x1183a}, 2535 {0x119d1, 0x119d7}, 2536 {0x119da, 0x119e0}, 2537 {0x119e4, 0x119e4}, 2538 {0x11a01, 0x11a0a}, 2539 {0x11a33, 0x11a39}, 2540 {0x11a3b, 0x11a3e}, 2541 {0x11a47, 0x11a47}, 2542 {0x11a51, 0x11a5b}, 2543 {0x11a8a, 0x11a99}, 2544 {0x11c2f, 0x11c36}, 2545 {0x11c38, 0x11c3f}, 2546 {0x11c92, 0x11ca7}, 2547 {0x11ca9, 0x11cb6}, 2548 {0x11d31, 0x11d36}, 2549 {0x11d3a, 0x11d3a}, 2550 {0x11d3c, 0x11d3d}, 2551 {0x11d3f, 0x11d45}, 2552 {0x11d47, 0x11d47}, 2553 {0x11d8a, 0x11d8e}, 2554 {0x11d90, 0x11d91}, 2555 {0x11d93, 0x11d97}, 2556 {0x11ef3, 0x11ef6}, 2557 {0x16af0, 0x16af4}, 2558 {0x16b30, 0x16b36}, 2559 {0x16f4f, 0x16f4f}, 2560 {0x16f51, 0x16f87}, 2561 {0x16f8f, 0x16f92}, 2562 {0x1bc9d, 0x1bc9e}, 2563 {0x1d165, 0x1d169}, 2564 {0x1d16d, 0x1d172}, 2565 {0x1d17b, 0x1d182}, 2566 {0x1d185, 0x1d18b}, 2567 {0x1d1aa, 0x1d1ad}, 2568 {0x1d242, 0x1d244}, 2569 {0x1da00, 0x1da36}, 2570 {0x1da3b, 0x1da6c}, 2571 {0x1da75, 0x1da75}, 2572 {0x1da84, 0x1da84}, 2573 {0x1da9b, 0x1da9f}, 2574 {0x1daa1, 0x1daaf}, 2575 {0x1e000, 0x1e006}, 2576 {0x1e008, 0x1e018}, 2577 {0x1e01b, 0x1e021}, 2578 {0x1e023, 0x1e024}, 2579 {0x1e026, 0x1e02a}, 2580 {0x1e130, 0x1e136}, 2581 {0x1e2ec, 0x1e2ef}, 2582 {0x1e8d0, 0x1e8d6}, 2583 {0x1e944, 0x1e94a}, 2584 {0xe0100, 0xe01ef} 2585 }; 2586 2587 return intable(combining, sizeof(combining), c); 2588 } 2589 2590 /* 2591 * Return TRUE for characters that can be displayed in a normal way. 2592 * Only for characters of 0x100 and above! 2593 */ 2594 int 2595 utf_printable(int c) 2596 { 2597 #ifdef USE_WCHAR_FUNCTIONS 2598 /* 2599 * Assume the iswprint() library function works better than our own stuff. 2600 */ 2601 return iswprint(c); 2602 #else 2603 /* Sorted list of non-overlapping intervals. 2604 * 0xd800-0xdfff is reserved for UTF-16, actually illegal. */ 2605 static struct interval nonprint[] = 2606 { 2607 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2608 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2609 {0xfffe, 0xffff} 2610 }; 2611 2612 return !intable(nonprint, sizeof(nonprint), c); 2613 #endif 2614 } 2615 2616 /* Sorted list of non-overlapping intervals of all Emoji characters, 2617 * based on http://unicode.org/emoji/charts/emoji-list.html */ 2618 static struct interval emoji_all[] = 2619 { 2620 {0x203c, 0x203c}, 2621 {0x2049, 0x2049}, 2622 {0x2122, 0x2122}, 2623 {0x2139, 0x2139}, 2624 {0x2194, 0x2199}, 2625 {0x21a9, 0x21aa}, 2626 {0x231a, 0x231b}, 2627 {0x2328, 0x2328}, 2628 {0x23cf, 0x23cf}, 2629 {0x23e9, 0x23f3}, 2630 {0x23f8, 0x23fa}, 2631 {0x24c2, 0x24c2}, 2632 {0x25aa, 0x25ab}, 2633 {0x25b6, 0x25b6}, 2634 {0x25c0, 0x25c0}, 2635 {0x25fb, 0x25fe}, 2636 {0x2600, 0x2604}, 2637 {0x260e, 0x260e}, 2638 {0x2611, 0x2611}, 2639 {0x2614, 0x2615}, 2640 {0x2618, 0x2618}, 2641 {0x261d, 0x261d}, 2642 {0x2620, 0x2620}, 2643 {0x2622, 0x2623}, 2644 {0x2626, 0x2626}, 2645 {0x262a, 0x262a}, 2646 {0x262e, 0x262f}, 2647 {0x2638, 0x263a}, 2648 {0x2640, 0x2640}, 2649 {0x2642, 0x2642}, 2650 {0x2648, 0x2653}, 2651 {0x265f, 0x2660}, 2652 {0x2663, 0x2663}, 2653 {0x2665, 0x2666}, 2654 {0x2668, 0x2668}, 2655 {0x267b, 0x267b}, 2656 {0x267e, 0x267f}, 2657 {0x2692, 0x2697}, 2658 {0x2699, 0x2699}, 2659 {0x269b, 0x269c}, 2660 {0x26a0, 0x26a1}, 2661 {0x26aa, 0x26ab}, 2662 {0x26b0, 0x26b1}, 2663 {0x26bd, 0x26be}, 2664 {0x26c4, 0x26c5}, 2665 {0x26c8, 0x26c8}, 2666 {0x26ce, 0x26cf}, 2667 {0x26d1, 0x26d1}, 2668 {0x26d3, 0x26d4}, 2669 {0x26e9, 0x26ea}, 2670 {0x26f0, 0x26f5}, 2671 {0x26f7, 0x26fa}, 2672 {0x26fd, 0x26fd}, 2673 {0x2702, 0x2702}, 2674 {0x2705, 0x2705}, 2675 {0x2708, 0x270d}, 2676 {0x270f, 0x270f}, 2677 {0x2712, 0x2712}, 2678 {0x2714, 0x2714}, 2679 {0x2716, 0x2716}, 2680 {0x271d, 0x271d}, 2681 {0x2721, 0x2721}, 2682 {0x2728, 0x2728}, 2683 {0x2733, 0x2734}, 2684 {0x2744, 0x2744}, 2685 {0x2747, 0x2747}, 2686 {0x274c, 0x274c}, 2687 {0x274e, 0x274e}, 2688 {0x2753, 0x2755}, 2689 {0x2757, 0x2757}, 2690 {0x2763, 0x2764}, 2691 {0x2795, 0x2797}, 2692 {0x27a1, 0x27a1}, 2693 {0x27b0, 0x27b0}, 2694 {0x27bf, 0x27bf}, 2695 {0x2934, 0x2935}, 2696 {0x2b05, 0x2b07}, 2697 {0x2b1b, 0x2b1c}, 2698 {0x2b50, 0x2b50}, 2699 {0x2b55, 0x2b55}, 2700 {0x3030, 0x3030}, 2701 {0x303d, 0x303d}, 2702 {0x3297, 0x3297}, 2703 {0x3299, 0x3299}, 2704 {0x1f004, 0x1f004}, 2705 {0x1f0cf, 0x1f0cf}, 2706 {0x1f170, 0x1f171}, 2707 {0x1f17e, 0x1f17f}, 2708 {0x1f18e, 0x1f18e}, 2709 {0x1f191, 0x1f19a}, 2710 {0x1f1e6, 0x1f1ff}, 2711 {0x1f201, 0x1f202}, 2712 {0x1f21a, 0x1f21a}, 2713 {0x1f22f, 0x1f22f}, 2714 {0x1f232, 0x1f23a}, 2715 {0x1f250, 0x1f251}, 2716 {0x1f300, 0x1f321}, 2717 {0x1f324, 0x1f393}, 2718 {0x1f396, 0x1f397}, 2719 {0x1f399, 0x1f39b}, 2720 {0x1f39e, 0x1f3f0}, 2721 {0x1f3f3, 0x1f3f5}, 2722 {0x1f3f7, 0x1f4fd}, 2723 {0x1f4ff, 0x1f53d}, 2724 {0x1f549, 0x1f54e}, 2725 {0x1f550, 0x1f567}, 2726 {0x1f56f, 0x1f570}, 2727 {0x1f573, 0x1f57a}, 2728 {0x1f587, 0x1f587}, 2729 {0x1f58a, 0x1f58d}, 2730 {0x1f590, 0x1f590}, 2731 {0x1f595, 0x1f596}, 2732 {0x1f5a4, 0x1f5a5}, 2733 {0x1f5a8, 0x1f5a8}, 2734 {0x1f5b1, 0x1f5b2}, 2735 {0x1f5bc, 0x1f5bc}, 2736 {0x1f5c2, 0x1f5c4}, 2737 {0x1f5d1, 0x1f5d3}, 2738 {0x1f5dc, 0x1f5de}, 2739 {0x1f5e1, 0x1f5e1}, 2740 {0x1f5e3, 0x1f5e3}, 2741 {0x1f5e8, 0x1f5e8}, 2742 {0x1f5ef, 0x1f5ef}, 2743 {0x1f5f3, 0x1f5f3}, 2744 {0x1f5fa, 0x1f64f}, 2745 {0x1f680, 0x1f6c5}, 2746 {0x1f6cb, 0x1f6d2}, 2747 {0x1f6e0, 0x1f6e5}, 2748 {0x1f6e9, 0x1f6e9}, 2749 {0x1f6eb, 0x1f6ec}, 2750 {0x1f6f0, 0x1f6f0}, 2751 {0x1f6f3, 0x1f6f9}, 2752 {0x1f910, 0x1f93a}, 2753 {0x1f93c, 0x1f93e}, 2754 {0x1f940, 0x1f945}, 2755 {0x1f947, 0x1f970}, 2756 {0x1f973, 0x1f976}, 2757 {0x1f97a, 0x1f97a}, 2758 {0x1f97c, 0x1f9a2}, 2759 {0x1f9b0, 0x1f9b9}, 2760 {0x1f9c0, 0x1f9c2}, 2761 {0x1f9d0, 0x1f9ff} 2762 }; 2763 2764 /* 2765 * Get class of a Unicode character. 2766 * 0: white space 2767 * 1: punctuation 2768 * 2 or bigger: some class of word character. 2769 */ 2770 int 2771 utf_class(int c) 2772 { 2773 return utf_class_buf(c, curbuf); 2774 } 2775 2776 int 2777 utf_class_buf(int c, buf_T *buf) 2778 { 2779 /* sorted list of non-overlapping intervals */ 2780 static struct clinterval 2781 { 2782 unsigned int first; 2783 unsigned int last; 2784 unsigned int class; 2785 } classes[] = 2786 { 2787 {0x037e, 0x037e, 1}, /* Greek question mark */ 2788 {0x0387, 0x0387, 1}, /* Greek ano teleia */ 2789 {0x055a, 0x055f, 1}, /* Armenian punctuation */ 2790 {0x0589, 0x0589, 1}, /* Armenian full stop */ 2791 {0x05be, 0x05be, 1}, 2792 {0x05c0, 0x05c0, 1}, 2793 {0x05c3, 0x05c3, 1}, 2794 {0x05f3, 0x05f4, 1}, 2795 {0x060c, 0x060c, 1}, 2796 {0x061b, 0x061b, 1}, 2797 {0x061f, 0x061f, 1}, 2798 {0x066a, 0x066d, 1}, 2799 {0x06d4, 0x06d4, 1}, 2800 {0x0700, 0x070d, 1}, /* Syriac punctuation */ 2801 {0x0964, 0x0965, 1}, 2802 {0x0970, 0x0970, 1}, 2803 {0x0df4, 0x0df4, 1}, 2804 {0x0e4f, 0x0e4f, 1}, 2805 {0x0e5a, 0x0e5b, 1}, 2806 {0x0f04, 0x0f12, 1}, 2807 {0x0f3a, 0x0f3d, 1}, 2808 {0x0f85, 0x0f85, 1}, 2809 {0x104a, 0x104f, 1}, /* Myanmar punctuation */ 2810 {0x10fb, 0x10fb, 1}, /* Georgian punctuation */ 2811 {0x1361, 0x1368, 1}, /* Ethiopic punctuation */ 2812 {0x166d, 0x166e, 1}, /* Canadian Syl. punctuation */ 2813 {0x1680, 0x1680, 0}, 2814 {0x169b, 0x169c, 1}, 2815 {0x16eb, 0x16ed, 1}, 2816 {0x1735, 0x1736, 1}, 2817 {0x17d4, 0x17dc, 1}, /* Khmer punctuation */ 2818 {0x1800, 0x180a, 1}, /* Mongolian punctuation */ 2819 {0x2000, 0x200b, 0}, /* spaces */ 2820 {0x200c, 0x2027, 1}, /* punctuation and symbols */ 2821 {0x2028, 0x2029, 0}, 2822 {0x202a, 0x202e, 1}, /* punctuation and symbols */ 2823 {0x202f, 0x202f, 0}, 2824 {0x2030, 0x205e, 1}, /* punctuation and symbols */ 2825 {0x205f, 0x205f, 0}, 2826 {0x2060, 0x27ff, 1}, /* punctuation and symbols */ 2827 {0x2070, 0x207f, 0x2070}, /* superscript */ 2828 {0x2080, 0x2094, 0x2080}, /* subscript */ 2829 {0x20a0, 0x27ff, 1}, /* all kinds of symbols */ 2830 {0x2800, 0x28ff, 0x2800}, /* braille */ 2831 {0x2900, 0x2998, 1}, /* arrows, brackets, etc. */ 2832 {0x29d8, 0x29db, 1}, 2833 {0x29fc, 0x29fd, 1}, 2834 {0x2e00, 0x2e7f, 1}, /* supplemental punctuation */ 2835 {0x3000, 0x3000, 0}, /* ideographic space */ 2836 {0x3001, 0x3020, 1}, /* ideographic punctuation */ 2837 {0x3030, 0x3030, 1}, 2838 {0x303d, 0x303d, 1}, 2839 {0x3040, 0x309f, 0x3040}, /* Hiragana */ 2840 {0x30a0, 0x30ff, 0x30a0}, /* Katakana */ 2841 {0x3300, 0x9fff, 0x4e00}, /* CJK Ideographs */ 2842 {0xac00, 0xd7a3, 0xac00}, /* Hangul Syllables */ 2843 {0xf900, 0xfaff, 0x4e00}, /* CJK Ideographs */ 2844 {0xfd3e, 0xfd3f, 1}, 2845 {0xfe30, 0xfe6b, 1}, /* punctuation forms */ 2846 {0xff00, 0xff0f, 1}, /* half/fullwidth ASCII */ 2847 {0xff1a, 0xff20, 1}, /* half/fullwidth ASCII */ 2848 {0xff3b, 0xff40, 1}, /* half/fullwidth ASCII */ 2849 {0xff5b, 0xff65, 1}, /* half/fullwidth ASCII */ 2850 {0x1d000, 0x1d24f, 1}, /* Musical notation */ 2851 {0x1d400, 0x1d7ff, 1}, /* Mathematical Alphanumeric Symbols */ 2852 {0x1f000, 0x1f2ff, 1}, /* Game pieces; enclosed characters */ 2853 {0x1f300, 0x1f9ff, 1}, /* Many symbol blocks */ 2854 {0x20000, 0x2a6df, 0x4e00}, /* CJK Ideographs */ 2855 {0x2a700, 0x2b73f, 0x4e00}, /* CJK Ideographs */ 2856 {0x2b740, 0x2b81f, 0x4e00}, /* CJK Ideographs */ 2857 {0x2f800, 0x2fa1f, 0x4e00}, /* CJK Ideographs */ 2858 }; 2859 2860 int bot = 0; 2861 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2862 int mid; 2863 2864 /* First quick check for Latin1 characters, use 'iskeyword'. */ 2865 if (c < 0x100) 2866 { 2867 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2868 return 0; /* blank */ 2869 if (vim_iswordc_buf(c, buf)) 2870 return 2; /* word character */ 2871 return 1; /* punctuation */ 2872 } 2873 2874 /* binary search in table */ 2875 while (top >= bot) 2876 { 2877 mid = (bot + top) / 2; 2878 if (classes[mid].last < (unsigned int)c) 2879 bot = mid + 1; 2880 else if (classes[mid].first > (unsigned int)c) 2881 top = mid - 1; 2882 else 2883 return (int)classes[mid].class; 2884 } 2885 2886 /* emoji */ 2887 if (intable(emoji_all, sizeof(emoji_all), c)) 2888 return 3; 2889 2890 /* most other characters are "word" characters */ 2891 return 2; 2892 } 2893 2894 int 2895 utf_ambiguous_width(int c) 2896 { 2897 return c >= 0x80 && (intable(ambiguous, sizeof(ambiguous), c) 2898 || intable(emoji_all, sizeof(emoji_all), c)); 2899 } 2900 2901 /* 2902 * Code for Unicode case-dependent operations. Based on notes in 2903 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2904 * This code uses simple case folding, not full case folding. 2905 * Last updated for Unicode 5.2. 2906 */ 2907 2908 /* 2909 * The following tables are built by ../runtime/tools/unicode.vim. 2910 * They must be in numeric order, because we use binary search. 2911 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2912 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2913 * folded/upper/lower by adding 32. 2914 */ 2915 typedef struct 2916 { 2917 int rangeStart; 2918 int rangeEnd; 2919 int step; 2920 int offset; 2921 } convertStruct; 2922 2923 static convertStruct foldCase[] = 2924 { 2925 {0x41,0x5a,1,32}, 2926 {0xb5,0xb5,-1,775}, 2927 {0xc0,0xd6,1,32}, 2928 {0xd8,0xde,1,32}, 2929 {0x100,0x12e,2,1}, 2930 {0x132,0x136,2,1}, 2931 {0x139,0x147,2,1}, 2932 {0x14a,0x176,2,1}, 2933 {0x178,0x178,-1,-121}, 2934 {0x179,0x17d,2,1}, 2935 {0x17f,0x17f,-1,-268}, 2936 {0x181,0x181,-1,210}, 2937 {0x182,0x184,2,1}, 2938 {0x186,0x186,-1,206}, 2939 {0x187,0x187,-1,1}, 2940 {0x189,0x18a,1,205}, 2941 {0x18b,0x18b,-1,1}, 2942 {0x18e,0x18e,-1,79}, 2943 {0x18f,0x18f,-1,202}, 2944 {0x190,0x190,-1,203}, 2945 {0x191,0x191,-1,1}, 2946 {0x193,0x193,-1,205}, 2947 {0x194,0x194,-1,207}, 2948 {0x196,0x196,-1,211}, 2949 {0x197,0x197,-1,209}, 2950 {0x198,0x198,-1,1}, 2951 {0x19c,0x19c,-1,211}, 2952 {0x19d,0x19d,-1,213}, 2953 {0x19f,0x19f,-1,214}, 2954 {0x1a0,0x1a4,2,1}, 2955 {0x1a6,0x1a6,-1,218}, 2956 {0x1a7,0x1a7,-1,1}, 2957 {0x1a9,0x1a9,-1,218}, 2958 {0x1ac,0x1ac,-1,1}, 2959 {0x1ae,0x1ae,-1,218}, 2960 {0x1af,0x1af,-1,1}, 2961 {0x1b1,0x1b2,1,217}, 2962 {0x1b3,0x1b5,2,1}, 2963 {0x1b7,0x1b7,-1,219}, 2964 {0x1b8,0x1bc,4,1}, 2965 {0x1c4,0x1c4,-1,2}, 2966 {0x1c5,0x1c5,-1,1}, 2967 {0x1c7,0x1c7,-1,2}, 2968 {0x1c8,0x1c8,-1,1}, 2969 {0x1ca,0x1ca,-1,2}, 2970 {0x1cb,0x1db,2,1}, 2971 {0x1de,0x1ee,2,1}, 2972 {0x1f1,0x1f1,-1,2}, 2973 {0x1f2,0x1f4,2,1}, 2974 {0x1f6,0x1f6,-1,-97}, 2975 {0x1f7,0x1f7,-1,-56}, 2976 {0x1f8,0x21e,2,1}, 2977 {0x220,0x220,-1,-130}, 2978 {0x222,0x232,2,1}, 2979 {0x23a,0x23a,-1,10795}, 2980 {0x23b,0x23b,-1,1}, 2981 {0x23d,0x23d,-1,-163}, 2982 {0x23e,0x23e,-1,10792}, 2983 {0x241,0x241,-1,1}, 2984 {0x243,0x243,-1,-195}, 2985 {0x244,0x244,-1,69}, 2986 {0x245,0x245,-1,71}, 2987 {0x246,0x24e,2,1}, 2988 {0x345,0x345,-1,116}, 2989 {0x370,0x372,2,1}, 2990 {0x376,0x376,-1,1}, 2991 {0x37f,0x37f,-1,116}, 2992 {0x386,0x386,-1,38}, 2993 {0x388,0x38a,1,37}, 2994 {0x38c,0x38c,-1,64}, 2995 {0x38e,0x38f,1,63}, 2996 {0x391,0x3a1,1,32}, 2997 {0x3a3,0x3ab,1,32}, 2998 {0x3c2,0x3c2,-1,1}, 2999 {0x3cf,0x3cf,-1,8}, 3000 {0x3d0,0x3d0,-1,-30}, 3001 {0x3d1,0x3d1,-1,-25}, 3002 {0x3d5,0x3d5,-1,-15}, 3003 {0x3d6,0x3d6,-1,-22}, 3004 {0x3d8,0x3ee,2,1}, 3005 {0x3f0,0x3f0,-1,-54}, 3006 {0x3f1,0x3f1,-1,-48}, 3007 {0x3f4,0x3f4,-1,-60}, 3008 {0x3f5,0x3f5,-1,-64}, 3009 {0x3f7,0x3f7,-1,1}, 3010 {0x3f9,0x3f9,-1,-7}, 3011 {0x3fa,0x3fa,-1,1}, 3012 {0x3fd,0x3ff,1,-130}, 3013 {0x400,0x40f,1,80}, 3014 {0x410,0x42f,1,32}, 3015 {0x460,0x480,2,1}, 3016 {0x48a,0x4be,2,1}, 3017 {0x4c0,0x4c0,-1,15}, 3018 {0x4c1,0x4cd,2,1}, 3019 {0x4d0,0x52e,2,1}, 3020 {0x531,0x556,1,48}, 3021 {0x10a0,0x10c5,1,7264}, 3022 {0x10c7,0x10cd,6,7264}, 3023 {0x13f8,0x13fd,1,-8}, 3024 {0x1c80,0x1c80,-1,-6222}, 3025 {0x1c81,0x1c81,-1,-6221}, 3026 {0x1c82,0x1c82,-1,-6212}, 3027 {0x1c83,0x1c84,1,-6210}, 3028 {0x1c85,0x1c85,-1,-6211}, 3029 {0x1c86,0x1c86,-1,-6204}, 3030 {0x1c87,0x1c87,-1,-6180}, 3031 {0x1c88,0x1c88,-1,35267}, 3032 {0x1c90,0x1cba,1,-3008}, 3033 {0x1cbd,0x1cbf,1,-3008}, 3034 {0x1e00,0x1e94,2,1}, 3035 {0x1e9b,0x1e9b,-1,-58}, 3036 {0x1e9e,0x1e9e,-1,-7615}, 3037 {0x1ea0,0x1efe,2,1}, 3038 {0x1f08,0x1f0f,1,-8}, 3039 {0x1f18,0x1f1d,1,-8}, 3040 {0x1f28,0x1f2f,1,-8}, 3041 {0x1f38,0x1f3f,1,-8}, 3042 {0x1f48,0x1f4d,1,-8}, 3043 {0x1f59,0x1f5f,2,-8}, 3044 {0x1f68,0x1f6f,1,-8}, 3045 {0x1f88,0x1f8f,1,-8}, 3046 {0x1f98,0x1f9f,1,-8}, 3047 {0x1fa8,0x1faf,1,-8}, 3048 {0x1fb8,0x1fb9,1,-8}, 3049 {0x1fba,0x1fbb,1,-74}, 3050 {0x1fbc,0x1fbc,-1,-9}, 3051 {0x1fbe,0x1fbe,-1,-7173}, 3052 {0x1fc8,0x1fcb,1,-86}, 3053 {0x1fcc,0x1fcc,-1,-9}, 3054 {0x1fd8,0x1fd9,1,-8}, 3055 {0x1fda,0x1fdb,1,-100}, 3056 {0x1fe8,0x1fe9,1,-8}, 3057 {0x1fea,0x1feb,1,-112}, 3058 {0x1fec,0x1fec,-1,-7}, 3059 {0x1ff8,0x1ff9,1,-128}, 3060 {0x1ffa,0x1ffb,1,-126}, 3061 {0x1ffc,0x1ffc,-1,-9}, 3062 {0x2126,0x2126,-1,-7517}, 3063 {0x212a,0x212a,-1,-8383}, 3064 {0x212b,0x212b,-1,-8262}, 3065 {0x2132,0x2132,-1,28}, 3066 {0x2160,0x216f,1,16}, 3067 {0x2183,0x2183,-1,1}, 3068 {0x24b6,0x24cf,1,26}, 3069 {0x2c00,0x2c2e,1,48}, 3070 {0x2c60,0x2c60,-1,1}, 3071 {0x2c62,0x2c62,-1,-10743}, 3072 {0x2c63,0x2c63,-1,-3814}, 3073 {0x2c64,0x2c64,-1,-10727}, 3074 {0x2c67,0x2c6b,2,1}, 3075 {0x2c6d,0x2c6d,-1,-10780}, 3076 {0x2c6e,0x2c6e,-1,-10749}, 3077 {0x2c6f,0x2c6f,-1,-10783}, 3078 {0x2c70,0x2c70,-1,-10782}, 3079 {0x2c72,0x2c75,3,1}, 3080 {0x2c7e,0x2c7f,1,-10815}, 3081 {0x2c80,0x2ce2,2,1}, 3082 {0x2ceb,0x2ced,2,1}, 3083 {0x2cf2,0xa640,31054,1}, 3084 {0xa642,0xa66c,2,1}, 3085 {0xa680,0xa69a,2,1}, 3086 {0xa722,0xa72e,2,1}, 3087 {0xa732,0xa76e,2,1}, 3088 {0xa779,0xa77b,2,1}, 3089 {0xa77d,0xa77d,-1,-35332}, 3090 {0xa77e,0xa786,2,1}, 3091 {0xa78b,0xa78b,-1,1}, 3092 {0xa78d,0xa78d,-1,-42280}, 3093 {0xa790,0xa792,2,1}, 3094 {0xa796,0xa7a8,2,1}, 3095 {0xa7aa,0xa7aa,-1,-42308}, 3096 {0xa7ab,0xa7ab,-1,-42319}, 3097 {0xa7ac,0xa7ac,-1,-42315}, 3098 {0xa7ad,0xa7ad,-1,-42305}, 3099 {0xa7ae,0xa7ae,-1,-42308}, 3100 {0xa7b0,0xa7b0,-1,-42258}, 3101 {0xa7b1,0xa7b1,-1,-42282}, 3102 {0xa7b2,0xa7b2,-1,-42261}, 3103 {0xa7b3,0xa7b3,-1,928}, 3104 {0xa7b4,0xa7be,2,1}, 3105 {0xa7c2,0xa7c2,-1,1}, 3106 {0xa7c4,0xa7c4,-1,-48}, 3107 {0xa7c5,0xa7c5,-1,-42307}, 3108 {0xa7c6,0xa7c6,-1,-35384}, 3109 {0xab70,0xabbf,1,-38864}, 3110 {0xff21,0xff3a,1,32}, 3111 {0x10400,0x10427,1,40}, 3112 {0x104b0,0x104d3,1,40}, 3113 {0x10c80,0x10cb2,1,64}, 3114 {0x118a0,0x118bf,1,32}, 3115 {0x16e40,0x16e5f,1,32}, 3116 {0x1e900,0x1e921,1,34} 3117 }; 3118 3119 /* 3120 * Generic conversion function for case operations. 3121 * Return the converted equivalent of "a", which is a UCS-4 character. Use 3122 * the given conversion "table". Uses binary search on "table". 3123 */ 3124 static int 3125 utf_convert( 3126 int a, 3127 convertStruct table[], 3128 int tableSize) 3129 { 3130 int start, mid, end; /* indices into table */ 3131 int entries = tableSize / sizeof(convertStruct); 3132 3133 start = 0; 3134 end = entries; 3135 while (start < end) 3136 { 3137 /* need to search further */ 3138 mid = (end + start) / 2; 3139 if (table[mid].rangeEnd < a) 3140 start = mid + 1; 3141 else 3142 end = mid; 3143 } 3144 if (start < entries 3145 && table[start].rangeStart <= a 3146 && a <= table[start].rangeEnd 3147 && (a - table[start].rangeStart) % table[start].step == 0) 3148 return (a + table[start].offset); 3149 else 3150 return a; 3151 } 3152 3153 /* 3154 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 3155 * simple case folding. 3156 */ 3157 int 3158 utf_fold(int a) 3159 { 3160 if (a < 0x80) 3161 /* be fast for ASCII */ 3162 return a >= 0x41 && a <= 0x5a ? a + 32 : a; 3163 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 3164 } 3165 3166 static convertStruct toLower[] = 3167 { 3168 {0x41,0x5a,1,32}, 3169 {0xc0,0xd6,1,32}, 3170 {0xd8,0xde,1,32}, 3171 {0x100,0x12e,2,1}, 3172 {0x130,0x130,-1,-199}, 3173 {0x132,0x136,2,1}, 3174 {0x139,0x147,2,1}, 3175 {0x14a,0x176,2,1}, 3176 {0x178,0x178,-1,-121}, 3177 {0x179,0x17d,2,1}, 3178 {0x181,0x181,-1,210}, 3179 {0x182,0x184,2,1}, 3180 {0x186,0x186,-1,206}, 3181 {0x187,0x187,-1,1}, 3182 {0x189,0x18a,1,205}, 3183 {0x18b,0x18b,-1,1}, 3184 {0x18e,0x18e,-1,79}, 3185 {0x18f,0x18f,-1,202}, 3186 {0x190,0x190,-1,203}, 3187 {0x191,0x191,-1,1}, 3188 {0x193,0x193,-1,205}, 3189 {0x194,0x194,-1,207}, 3190 {0x196,0x196,-1,211}, 3191 {0x197,0x197,-1,209}, 3192 {0x198,0x198,-1,1}, 3193 {0x19c,0x19c,-1,211}, 3194 {0x19d,0x19d,-1,213}, 3195 {0x19f,0x19f,-1,214}, 3196 {0x1a0,0x1a4,2,1}, 3197 {0x1a6,0x1a6,-1,218}, 3198 {0x1a7,0x1a7,-1,1}, 3199 {0x1a9,0x1a9,-1,218}, 3200 {0x1ac,0x1ac,-1,1}, 3201 {0x1ae,0x1ae,-1,218}, 3202 {0x1af,0x1af,-1,1}, 3203 {0x1b1,0x1b2,1,217}, 3204 {0x1b3,0x1b5,2,1}, 3205 {0x1b7,0x1b7,-1,219}, 3206 {0x1b8,0x1bc,4,1}, 3207 {0x1c4,0x1c4,-1,2}, 3208 {0x1c5,0x1c5,-1,1}, 3209 {0x1c7,0x1c7,-1,2}, 3210 {0x1c8,0x1c8,-1,1}, 3211 {0x1ca,0x1ca,-1,2}, 3212 {0x1cb,0x1db,2,1}, 3213 {0x1de,0x1ee,2,1}, 3214 {0x1f1,0x1f1,-1,2}, 3215 {0x1f2,0x1f4,2,1}, 3216 {0x1f6,0x1f6,-1,-97}, 3217 {0x1f7,0x1f7,-1,-56}, 3218 {0x1f8,0x21e,2,1}, 3219 {0x220,0x220,-1,-130}, 3220 {0x222,0x232,2,1}, 3221 {0x23a,0x23a,-1,10795}, 3222 {0x23b,0x23b,-1,1}, 3223 {0x23d,0x23d,-1,-163}, 3224 {0x23e,0x23e,-1,10792}, 3225 {0x241,0x241,-1,1}, 3226 {0x243,0x243,-1,-195}, 3227 {0x244,0x244,-1,69}, 3228 {0x245,0x245,-1,71}, 3229 {0x246,0x24e,2,1}, 3230 {0x370,0x372,2,1}, 3231 {0x376,0x376,-1,1}, 3232 {0x37f,0x37f,-1,116}, 3233 {0x386,0x386,-1,38}, 3234 {0x388,0x38a,1,37}, 3235 {0x38c,0x38c,-1,64}, 3236 {0x38e,0x38f,1,63}, 3237 {0x391,0x3a1,1,32}, 3238 {0x3a3,0x3ab,1,32}, 3239 {0x3cf,0x3cf,-1,8}, 3240 {0x3d8,0x3ee,2,1}, 3241 {0x3f4,0x3f4,-1,-60}, 3242 {0x3f7,0x3f7,-1,1}, 3243 {0x3f9,0x3f9,-1,-7}, 3244 {0x3fa,0x3fa,-1,1}, 3245 {0x3fd,0x3ff,1,-130}, 3246 {0x400,0x40f,1,80}, 3247 {0x410,0x42f,1,32}, 3248 {0x460,0x480,2,1}, 3249 {0x48a,0x4be,2,1}, 3250 {0x4c0,0x4c0,-1,15}, 3251 {0x4c1,0x4cd,2,1}, 3252 {0x4d0,0x52e,2,1}, 3253 {0x531,0x556,1,48}, 3254 {0x10a0,0x10c5,1,7264}, 3255 {0x10c7,0x10cd,6,7264}, 3256 {0x13a0,0x13ef,1,38864}, 3257 {0x13f0,0x13f5,1,8}, 3258 {0x1c90,0x1cba,1,-3008}, 3259 {0x1cbd,0x1cbf,1,-3008}, 3260 {0x1e00,0x1e94,2,1}, 3261 {0x1e9e,0x1e9e,-1,-7615}, 3262 {0x1ea0,0x1efe,2,1}, 3263 {0x1f08,0x1f0f,1,-8}, 3264 {0x1f18,0x1f1d,1,-8}, 3265 {0x1f28,0x1f2f,1,-8}, 3266 {0x1f38,0x1f3f,1,-8}, 3267 {0x1f48,0x1f4d,1,-8}, 3268 {0x1f59,0x1f5f,2,-8}, 3269 {0x1f68,0x1f6f,1,-8}, 3270 {0x1f88,0x1f8f,1,-8}, 3271 {0x1f98,0x1f9f,1,-8}, 3272 {0x1fa8,0x1faf,1,-8}, 3273 {0x1fb8,0x1fb9,1,-8}, 3274 {0x1fba,0x1fbb,1,-74}, 3275 {0x1fbc,0x1fbc,-1,-9}, 3276 {0x1fc8,0x1fcb,1,-86}, 3277 {0x1fcc,0x1fcc,-1,-9}, 3278 {0x1fd8,0x1fd9,1,-8}, 3279 {0x1fda,0x1fdb,1,-100}, 3280 {0x1fe8,0x1fe9,1,-8}, 3281 {0x1fea,0x1feb,1,-112}, 3282 {0x1fec,0x1fec,-1,-7}, 3283 {0x1ff8,0x1ff9,1,-128}, 3284 {0x1ffa,0x1ffb,1,-126}, 3285 {0x1ffc,0x1ffc,-1,-9}, 3286 {0x2126,0x2126,-1,-7517}, 3287 {0x212a,0x212a,-1,-8383}, 3288 {0x212b,0x212b,-1,-8262}, 3289 {0x2132,0x2132,-1,28}, 3290 {0x2160,0x216f,1,16}, 3291 {0x2183,0x2183,-1,1}, 3292 {0x24b6,0x24cf,1,26}, 3293 {0x2c00,0x2c2e,1,48}, 3294 {0x2c60,0x2c60,-1,1}, 3295 {0x2c62,0x2c62,-1,-10743}, 3296 {0x2c63,0x2c63,-1,-3814}, 3297 {0x2c64,0x2c64,-1,-10727}, 3298 {0x2c67,0x2c6b,2,1}, 3299 {0x2c6d,0x2c6d,-1,-10780}, 3300 {0x2c6e,0x2c6e,-1,-10749}, 3301 {0x2c6f,0x2c6f,-1,-10783}, 3302 {0x2c70,0x2c70,-1,-10782}, 3303 {0x2c72,0x2c75,3,1}, 3304 {0x2c7e,0x2c7f,1,-10815}, 3305 {0x2c80,0x2ce2,2,1}, 3306 {0x2ceb,0x2ced,2,1}, 3307 {0x2cf2,0xa640,31054,1}, 3308 {0xa642,0xa66c,2,1}, 3309 {0xa680,0xa69a,2,1}, 3310 {0xa722,0xa72e,2,1}, 3311 {0xa732,0xa76e,2,1}, 3312 {0xa779,0xa77b,2,1}, 3313 {0xa77d,0xa77d,-1,-35332}, 3314 {0xa77e,0xa786,2,1}, 3315 {0xa78b,0xa78b,-1,1}, 3316 {0xa78d,0xa78d,-1,-42280}, 3317 {0xa790,0xa792,2,1}, 3318 {0xa796,0xa7a8,2,1}, 3319 {0xa7aa,0xa7aa,-1,-42308}, 3320 {0xa7ab,0xa7ab,-1,-42319}, 3321 {0xa7ac,0xa7ac,-1,-42315}, 3322 {0xa7ad,0xa7ad,-1,-42305}, 3323 {0xa7ae,0xa7ae,-1,-42308}, 3324 {0xa7b0,0xa7b0,-1,-42258}, 3325 {0xa7b1,0xa7b1,-1,-42282}, 3326 {0xa7b2,0xa7b2,-1,-42261}, 3327 {0xa7b3,0xa7b3,-1,928}, 3328 {0xa7b4,0xa7be,2,1}, 3329 {0xa7c2,0xa7c2,-1,1}, 3330 {0xa7c4,0xa7c4,-1,-48}, 3331 {0xa7c5,0xa7c5,-1,-42307}, 3332 {0xa7c6,0xa7c6,-1,-35384}, 3333 {0xff21,0xff3a,1,32}, 3334 {0x10400,0x10427,1,40}, 3335 {0x104b0,0x104d3,1,40}, 3336 {0x10c80,0x10cb2,1,64}, 3337 {0x118a0,0x118bf,1,32}, 3338 {0x16e40,0x16e5f,1,32}, 3339 {0x1e900,0x1e921,1,34} 3340 }; 3341 3342 static convertStruct toUpper[] = 3343 { 3344 {0x61,0x7a,1,-32}, 3345 {0xb5,0xb5,-1,743}, 3346 {0xe0,0xf6,1,-32}, 3347 {0xf8,0xfe,1,-32}, 3348 {0xff,0xff,-1,121}, 3349 {0x101,0x12f,2,-1}, 3350 {0x131,0x131,-1,-232}, 3351 {0x133,0x137,2,-1}, 3352 {0x13a,0x148,2,-1}, 3353 {0x14b,0x177,2,-1}, 3354 {0x17a,0x17e,2,-1}, 3355 {0x17f,0x17f,-1,-300}, 3356 {0x180,0x180,-1,195}, 3357 {0x183,0x185,2,-1}, 3358 {0x188,0x18c,4,-1}, 3359 {0x192,0x192,-1,-1}, 3360 {0x195,0x195,-1,97}, 3361 {0x199,0x199,-1,-1}, 3362 {0x19a,0x19a,-1,163}, 3363 {0x19e,0x19e,-1,130}, 3364 {0x1a1,0x1a5,2,-1}, 3365 {0x1a8,0x1ad,5,-1}, 3366 {0x1b0,0x1b4,4,-1}, 3367 {0x1b6,0x1b9,3,-1}, 3368 {0x1bd,0x1bd,-1,-1}, 3369 {0x1bf,0x1bf,-1,56}, 3370 {0x1c5,0x1c5,-1,-1}, 3371 {0x1c6,0x1c6,-1,-2}, 3372 {0x1c8,0x1c8,-1,-1}, 3373 {0x1c9,0x1c9,-1,-2}, 3374 {0x1cb,0x1cb,-1,-1}, 3375 {0x1cc,0x1cc,-1,-2}, 3376 {0x1ce,0x1dc,2,-1}, 3377 {0x1dd,0x1dd,-1,-79}, 3378 {0x1df,0x1ef,2,-1}, 3379 {0x1f2,0x1f2,-1,-1}, 3380 {0x1f3,0x1f3,-1,-2}, 3381 {0x1f5,0x1f9,4,-1}, 3382 {0x1fb,0x21f,2,-1}, 3383 {0x223,0x233,2,-1}, 3384 {0x23c,0x23c,-1,-1}, 3385 {0x23f,0x240,1,10815}, 3386 {0x242,0x247,5,-1}, 3387 {0x249,0x24f,2,-1}, 3388 {0x250,0x250,-1,10783}, 3389 {0x251,0x251,-1,10780}, 3390 {0x252,0x252,-1,10782}, 3391 {0x253,0x253,-1,-210}, 3392 {0x254,0x254,-1,-206}, 3393 {0x256,0x257,1,-205}, 3394 {0x259,0x259,-1,-202}, 3395 {0x25b,0x25b,-1,-203}, 3396 {0x25c,0x25c,-1,42319}, 3397 {0x260,0x260,-1,-205}, 3398 {0x261,0x261,-1,42315}, 3399 {0x263,0x263,-1,-207}, 3400 {0x265,0x265,-1,42280}, 3401 {0x266,0x266,-1,42308}, 3402 {0x268,0x268,-1,-209}, 3403 {0x269,0x269,-1,-211}, 3404 {0x26a,0x26a,-1,42308}, 3405 {0x26b,0x26b,-1,10743}, 3406 {0x26c,0x26c,-1,42305}, 3407 {0x26f,0x26f,-1,-211}, 3408 {0x271,0x271,-1,10749}, 3409 {0x272,0x272,-1,-213}, 3410 {0x275,0x275,-1,-214}, 3411 {0x27d,0x27d,-1,10727}, 3412 {0x280,0x280,-1,-218}, 3413 {0x282,0x282,-1,42307}, 3414 {0x283,0x283,-1,-218}, 3415 {0x287,0x287,-1,42282}, 3416 {0x288,0x288,-1,-218}, 3417 {0x289,0x289,-1,-69}, 3418 {0x28a,0x28b,1,-217}, 3419 {0x28c,0x28c,-1,-71}, 3420 {0x292,0x292,-1,-219}, 3421 {0x29d,0x29d,-1,42261}, 3422 {0x29e,0x29e,-1,42258}, 3423 {0x345,0x345,-1,84}, 3424 {0x371,0x373,2,-1}, 3425 {0x377,0x377,-1,-1}, 3426 {0x37b,0x37d,1,130}, 3427 {0x3ac,0x3ac,-1,-38}, 3428 {0x3ad,0x3af,1,-37}, 3429 {0x3b1,0x3c1,1,-32}, 3430 {0x3c2,0x3c2,-1,-31}, 3431 {0x3c3,0x3cb,1,-32}, 3432 {0x3cc,0x3cc,-1,-64}, 3433 {0x3cd,0x3ce,1,-63}, 3434 {0x3d0,0x3d0,-1,-62}, 3435 {0x3d1,0x3d1,-1,-57}, 3436 {0x3d5,0x3d5,-1,-47}, 3437 {0x3d6,0x3d6,-1,-54}, 3438 {0x3d7,0x3d7,-1,-8}, 3439 {0x3d9,0x3ef,2,-1}, 3440 {0x3f0,0x3f0,-1,-86}, 3441 {0x3f1,0x3f1,-1,-80}, 3442 {0x3f2,0x3f2,-1,7}, 3443 {0x3f3,0x3f3,-1,-116}, 3444 {0x3f5,0x3f5,-1,-96}, 3445 {0x3f8,0x3fb,3,-1}, 3446 {0x430,0x44f,1,-32}, 3447 {0x450,0x45f,1,-80}, 3448 {0x461,0x481,2,-1}, 3449 {0x48b,0x4bf,2,-1}, 3450 {0x4c2,0x4ce,2,-1}, 3451 {0x4cf,0x4cf,-1,-15}, 3452 {0x4d1,0x52f,2,-1}, 3453 {0x561,0x586,1,-48}, 3454 {0x10d0,0x10fa,1,3008}, 3455 {0x10fd,0x10ff,1,3008}, 3456 {0x13f8,0x13fd,1,-8}, 3457 {0x1c80,0x1c80,-1,-6254}, 3458 {0x1c81,0x1c81,-1,-6253}, 3459 {0x1c82,0x1c82,-1,-6244}, 3460 {0x1c83,0x1c84,1,-6242}, 3461 {0x1c85,0x1c85,-1,-6243}, 3462 {0x1c86,0x1c86,-1,-6236}, 3463 {0x1c87,0x1c87,-1,-6181}, 3464 {0x1c88,0x1c88,-1,35266}, 3465 {0x1d79,0x1d79,-1,35332}, 3466 {0x1d7d,0x1d7d,-1,3814}, 3467 {0x1d8e,0x1d8e,-1,35384}, 3468 {0x1e01,0x1e95,2,-1}, 3469 {0x1e9b,0x1e9b,-1,-59}, 3470 {0x1ea1,0x1eff,2,-1}, 3471 {0x1f00,0x1f07,1,8}, 3472 {0x1f10,0x1f15,1,8}, 3473 {0x1f20,0x1f27,1,8}, 3474 {0x1f30,0x1f37,1,8}, 3475 {0x1f40,0x1f45,1,8}, 3476 {0x1f51,0x1f57,2,8}, 3477 {0x1f60,0x1f67,1,8}, 3478 {0x1f70,0x1f71,1,74}, 3479 {0x1f72,0x1f75,1,86}, 3480 {0x1f76,0x1f77,1,100}, 3481 {0x1f78,0x1f79,1,128}, 3482 {0x1f7a,0x1f7b,1,112}, 3483 {0x1f7c,0x1f7d,1,126}, 3484 {0x1f80,0x1f87,1,8}, 3485 {0x1f90,0x1f97,1,8}, 3486 {0x1fa0,0x1fa7,1,8}, 3487 {0x1fb0,0x1fb1,1,8}, 3488 {0x1fb3,0x1fb3,-1,9}, 3489 {0x1fbe,0x1fbe,-1,-7205}, 3490 {0x1fc3,0x1fc3,-1,9}, 3491 {0x1fd0,0x1fd1,1,8}, 3492 {0x1fe0,0x1fe1,1,8}, 3493 {0x1fe5,0x1fe5,-1,7}, 3494 {0x1ff3,0x1ff3,-1,9}, 3495 {0x214e,0x214e,-1,-28}, 3496 {0x2170,0x217f,1,-16}, 3497 {0x2184,0x2184,-1,-1}, 3498 {0x24d0,0x24e9,1,-26}, 3499 {0x2c30,0x2c5e,1,-48}, 3500 {0x2c61,0x2c61,-1,-1}, 3501 {0x2c65,0x2c65,-1,-10795}, 3502 {0x2c66,0x2c66,-1,-10792}, 3503 {0x2c68,0x2c6c,2,-1}, 3504 {0x2c73,0x2c76,3,-1}, 3505 {0x2c81,0x2ce3,2,-1}, 3506 {0x2cec,0x2cee,2,-1}, 3507 {0x2cf3,0x2cf3,-1,-1}, 3508 {0x2d00,0x2d25,1,-7264}, 3509 {0x2d27,0x2d2d,6,-7264}, 3510 {0xa641,0xa66d,2,-1}, 3511 {0xa681,0xa69b,2,-1}, 3512 {0xa723,0xa72f,2,-1}, 3513 {0xa733,0xa76f,2,-1}, 3514 {0xa77a,0xa77c,2,-1}, 3515 {0xa77f,0xa787,2,-1}, 3516 {0xa78c,0xa791,5,-1}, 3517 {0xa793,0xa793,-1,-1}, 3518 {0xa794,0xa794,-1,48}, 3519 {0xa797,0xa7a9,2,-1}, 3520 {0xa7b5,0xa7bf,2,-1}, 3521 {0xa7c3,0xa7c3,-1,-1}, 3522 {0xab53,0xab53,-1,-928}, 3523 {0xab70,0xabbf,1,-38864}, 3524 {0xff41,0xff5a,1,-32}, 3525 {0x10428,0x1044f,1,-40}, 3526 {0x104d8,0x104fb,1,-40}, 3527 {0x10cc0,0x10cf2,1,-64}, 3528 {0x118c0,0x118df,1,-32}, 3529 {0x16e60,0x16e7f,1,-32}, 3530 {0x1e922,0x1e943,1,-34} 3531 }; 3532 3533 /* 3534 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3535 * simple case folding. 3536 */ 3537 int 3538 utf_toupper(int a) 3539 { 3540 /* If 'casemap' contains "keepascii" use ASCII style toupper(). */ 3541 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3542 return TOUPPER_ASC(a); 3543 3544 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3545 /* If towupper() is available and handles Unicode, use it. */ 3546 if (!(cmp_flags & CMP_INTERNAL)) 3547 return towupper(a); 3548 #endif 3549 3550 /* For characters below 128 use locale sensitive toupper(). */ 3551 if (a < 128) 3552 return TOUPPER_LOC(a); 3553 3554 /* For any other characters use the above mapping table. */ 3555 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3556 } 3557 3558 int 3559 utf_islower(int a) 3560 { 3561 /* German sharp s is lower case but has no upper case equivalent. */ 3562 return (utf_toupper(a) != a) || a == 0xdf; 3563 } 3564 3565 /* 3566 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3567 * simple case folding. 3568 */ 3569 int 3570 utf_tolower(int a) 3571 { 3572 /* If 'casemap' contains "keepascii" use ASCII style tolower(). */ 3573 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3574 return TOLOWER_ASC(a); 3575 3576 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3577 /* If towlower() is available and handles Unicode, use it. */ 3578 if (!(cmp_flags & CMP_INTERNAL)) 3579 return towlower(a); 3580 #endif 3581 3582 /* For characters below 128 use locale sensitive tolower(). */ 3583 if (a < 128) 3584 return TOLOWER_LOC(a); 3585 3586 /* For any other characters use the above mapping table. */ 3587 return utf_convert(a, toLower, (int)sizeof(toLower)); 3588 } 3589 3590 int 3591 utf_isupper(int a) 3592 { 3593 return (utf_tolower(a) != a); 3594 } 3595 3596 static int 3597 utf_strnicmp( 3598 char_u *s1, 3599 char_u *s2, 3600 size_t n1, 3601 size_t n2) 3602 { 3603 int c1, c2, cdiff; 3604 char_u buffer[6]; 3605 3606 for (;;) 3607 { 3608 c1 = utf_safe_read_char_adv(&s1, &n1); 3609 c2 = utf_safe_read_char_adv(&s2, &n2); 3610 3611 if (c1 <= 0 || c2 <= 0) 3612 break; 3613 3614 if (c1 == c2) 3615 continue; 3616 3617 cdiff = utf_fold(c1) - utf_fold(c2); 3618 if (cdiff != 0) 3619 return cdiff; 3620 } 3621 3622 /* some string ended or has an incomplete/illegal character sequence */ 3623 3624 if (c1 == 0 || c2 == 0) 3625 { 3626 /* some string ended. shorter string is smaller */ 3627 if (c1 == 0 && c2 == 0) 3628 return 0; 3629 return c1 == 0 ? -1 : 1; 3630 } 3631 3632 /* Continue with bytewise comparison to produce some result that 3633 * would make comparison operations involving this function transitive. 3634 * 3635 * If only one string had an error, comparison should be made with 3636 * folded version of the other string. In this case it is enough 3637 * to fold just one character to determine the result of comparison. */ 3638 3639 if (c1 != -1 && c2 == -1) 3640 { 3641 n1 = utf_char2bytes(utf_fold(c1), buffer); 3642 s1 = buffer; 3643 } 3644 else if (c2 != -1 && c1 == -1) 3645 { 3646 n2 = utf_char2bytes(utf_fold(c2), buffer); 3647 s2 = buffer; 3648 } 3649 3650 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3651 { 3652 cdiff = (int)(*s1) - (int)(*s2); 3653 if (cdiff != 0) 3654 return cdiff; 3655 3656 s1++; 3657 s2++; 3658 n1--; 3659 n2--; 3660 } 3661 3662 if (n1 > 0 && *s1 == NUL) 3663 n1 = 0; 3664 if (n2 > 0 && *s2 == NUL) 3665 n2 = 0; 3666 3667 if (n1 == 0 && n2 == 0) 3668 return 0; 3669 return n1 == 0 ? -1 : 1; 3670 } 3671 3672 /* 3673 * Version of strnicmp() that handles multi-byte characters. 3674 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3675 * probably use strnicmp(), because there are no ASCII characters in the 3676 * second byte. 3677 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3678 * two characters otherwise. 3679 */ 3680 int 3681 mb_strnicmp(char_u *s1, char_u *s2, size_t nn) 3682 { 3683 int i, l; 3684 int cdiff; 3685 int n = (int)nn; 3686 3687 if (enc_utf8) 3688 { 3689 return utf_strnicmp(s1, s2, nn, nn); 3690 } 3691 else 3692 { 3693 for (i = 0; i < n; i += l) 3694 { 3695 if (s1[i] == NUL && s2[i] == NUL) /* both strings end */ 3696 return 0; 3697 3698 l = (*mb_ptr2len)(s1 + i); 3699 if (l <= 1) 3700 { 3701 /* Single byte: first check normally, then with ignore case. */ 3702 if (s1[i] != s2[i]) 3703 { 3704 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3705 if (cdiff != 0) 3706 return cdiff; 3707 } 3708 } 3709 else 3710 { 3711 /* For non-Unicode multi-byte don't ignore case. */ 3712 if (l > n - i) 3713 l = n - i; 3714 cdiff = STRNCMP(s1 + i, s2 + i, l); 3715 if (cdiff != 0) 3716 return cdiff; 3717 } 3718 } 3719 } 3720 return 0; 3721 } 3722 3723 /* 3724 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3725 * 'encoding' has been set to. 3726 */ 3727 void 3728 show_utf8(void) 3729 { 3730 int len; 3731 int rlen = 0; 3732 char_u *line; 3733 int clen; 3734 int i; 3735 3736 /* Get the byte length of the char under the cursor, including composing 3737 * characters. */ 3738 line = ml_get_cursor(); 3739 len = utfc_ptr2len(line); 3740 if (len == 0) 3741 { 3742 msg("NUL"); 3743 return; 3744 } 3745 3746 clen = 0; 3747 for (i = 0; i < len; ++i) 3748 { 3749 if (clen == 0) 3750 { 3751 /* start of (composing) character, get its length */ 3752 if (i > 0) 3753 { 3754 STRCPY(IObuff + rlen, "+ "); 3755 rlen += 2; 3756 } 3757 clen = utf_ptr2len(line + i); 3758 } 3759 sprintf((char *)IObuff + rlen, "%02x ", 3760 (line[i] == NL) ? NUL : line[i]); /* NUL is stored as NL */ 3761 --clen; 3762 rlen += (int)STRLEN(IObuff + rlen); 3763 if (rlen > IOSIZE - 20) 3764 break; 3765 } 3766 3767 msg((char *)IObuff); 3768 } 3769 3770 /* 3771 * mb_head_off() function pointer. 3772 * Return offset from "p" to the first byte of the character it points into. 3773 * If "p" points to the NUL at the end of the string return 0. 3774 * Returns 0 when already at the first byte of a character. 3775 */ 3776 int 3777 latin_head_off(char_u *base UNUSED, char_u *p UNUSED) 3778 { 3779 return 0; 3780 } 3781 3782 int 3783 dbcs_head_off(char_u *base, char_u *p) 3784 { 3785 char_u *q; 3786 3787 /* It can't be a trailing byte when not using DBCS, at the start of the 3788 * string or the previous byte can't start a double-byte. */ 3789 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3790 return 0; 3791 3792 /* This is slow: need to start at the base and go forward until the 3793 * byte we are looking for. Return 1 when we went past it, 0 otherwise. */ 3794 q = base; 3795 while (q < p) 3796 q += dbcs_ptr2len(q); 3797 return (q == p) ? 0 : 1; 3798 } 3799 3800 /* 3801 * Special version of dbcs_head_off() that works for ScreenLines[], where 3802 * single-width DBCS_JPNU characters are stored separately. 3803 */ 3804 int 3805 dbcs_screen_head_off(char_u *base, char_u *p) 3806 { 3807 char_u *q; 3808 3809 /* It can't be a trailing byte when not using DBCS, at the start of the 3810 * string or the previous byte can't start a double-byte. 3811 * For euc-jp an 0x8e byte in the previous cell always means we have a 3812 * lead byte in the current cell. */ 3813 if (p <= base 3814 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3815 || MB_BYTE2LEN(p[-1]) == 1 3816 || *p == NUL) 3817 return 0; 3818 3819 /* This is slow: need to start at the base and go forward until the 3820 * byte we are looking for. Return 1 when we went past it, 0 otherwise. 3821 * For DBCS_JPNU look out for 0x8e, which means the second byte is not 3822 * stored as the next byte. */ 3823 q = base; 3824 while (q < p) 3825 { 3826 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3827 ++q; 3828 else 3829 q += dbcs_ptr2len(q); 3830 } 3831 return (q == p) ? 0 : 1; 3832 } 3833 3834 int 3835 utf_head_off(char_u *base, char_u *p) 3836 { 3837 char_u *q; 3838 char_u *s; 3839 int c; 3840 int len; 3841 #ifdef FEAT_ARABIC 3842 char_u *j; 3843 #endif 3844 3845 if (*p < 0x80) /* be quick for ASCII */ 3846 return 0; 3847 3848 /* Skip backwards over trailing bytes: 10xx.xxxx 3849 * Skip backwards again if on a composing char. */ 3850 for (q = p; ; --q) 3851 { 3852 /* Move s to the last byte of this char. */ 3853 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3854 ; 3855 /* Move q to the first byte of this char. */ 3856 while (q > base && (*q & 0xc0) == 0x80) 3857 --q; 3858 /* Check for illegal sequence. Do allow an illegal byte after where we 3859 * started. */ 3860 len = utf8len_tab[*q]; 3861 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3862 return 0; 3863 3864 if (q <= base) 3865 break; 3866 3867 c = utf_ptr2char(q); 3868 if (utf_iscomposing(c)) 3869 continue; 3870 3871 #ifdef FEAT_ARABIC 3872 if (arabic_maycombine(c)) 3873 { 3874 /* Advance to get a sneak-peak at the next char */ 3875 j = q; 3876 --j; 3877 /* Move j to the first byte of this char. */ 3878 while (j > base && (*j & 0xc0) == 0x80) 3879 --j; 3880 if (arabic_combine(utf_ptr2char(j), c)) 3881 continue; 3882 } 3883 #endif 3884 break; 3885 } 3886 3887 return (int)(p - q); 3888 } 3889 3890 /* 3891 * Copy a character from "*fp" to "*tp" and advance the pointers. 3892 */ 3893 void 3894 mb_copy_char(char_u **fp, char_u **tp) 3895 { 3896 int l = (*mb_ptr2len)(*fp); 3897 3898 mch_memmove(*tp, *fp, (size_t)l); 3899 *tp += l; 3900 *fp += l; 3901 } 3902 3903 /* 3904 * Return the offset from "p" to the first byte of a character. When "p" is 3905 * at the start of a character 0 is returned, otherwise the offset to the next 3906 * character. Can start anywhere in a stream of bytes. 3907 */ 3908 int 3909 mb_off_next(char_u *base, char_u *p) 3910 { 3911 int i; 3912 int j; 3913 3914 if (enc_utf8) 3915 { 3916 if (*p < 0x80) /* be quick for ASCII */ 3917 return 0; 3918 3919 /* Find the next character that isn't 10xx.xxxx */ 3920 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 3921 ; 3922 if (i > 0) 3923 { 3924 /* Check for illegal sequence. */ 3925 for (j = 0; p - j > base; ++j) 3926 if ((p[-j] & 0xc0) != 0x80) 3927 break; 3928 if (utf8len_tab[p[-j]] != i + j) 3929 return 0; 3930 } 3931 return i; 3932 } 3933 3934 /* Only need to check if we're on a trail byte, it doesn't matter if we 3935 * want the offset to the next or current character. */ 3936 return (*mb_head_off)(base, p); 3937 } 3938 3939 /* 3940 * Return the offset from "p" to the last byte of the character it points 3941 * into. Can start anywhere in a stream of bytes. 3942 */ 3943 int 3944 mb_tail_off(char_u *base, char_u *p) 3945 { 3946 int i; 3947 int j; 3948 3949 if (*p == NUL) 3950 return 0; 3951 3952 if (enc_utf8) 3953 { 3954 /* Find the last character that is 10xx.xxxx */ 3955 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 3956 ; 3957 /* Check for illegal sequence. */ 3958 for (j = 0; p - j > base; ++j) 3959 if ((p[-j] & 0xc0) != 0x80) 3960 break; 3961 if (utf8len_tab[p[-j]] != i + j + 1) 3962 return 0; 3963 return i; 3964 } 3965 3966 /* It can't be the first byte if a double-byte when not using DBCS, at the 3967 * end of the string or the byte can't start a double-byte. */ 3968 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 3969 return 0; 3970 3971 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 3972 return 1 - dbcs_head_off(base, p); 3973 } 3974 3975 /* 3976 * Find the next illegal byte sequence. 3977 */ 3978 void 3979 utf_find_illegal(void) 3980 { 3981 pos_T pos = curwin->w_cursor; 3982 char_u *p; 3983 int len; 3984 vimconv_T vimconv; 3985 char_u *tofree = NULL; 3986 3987 vimconv.vc_type = CONV_NONE; 3988 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 3989 { 3990 /* 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 3991 * possibly a utf-8 file with illegal bytes. Setup for conversion 3992 * from utf-8 to 'fileencoding'. */ 3993 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 3994 } 3995 3996 curwin->w_cursor.coladd = 0; 3997 for (;;) 3998 { 3999 p = ml_get_cursor(); 4000 if (vimconv.vc_type != CONV_NONE) 4001 { 4002 vim_free(tofree); 4003 tofree = string_convert(&vimconv, p, NULL); 4004 if (tofree == NULL) 4005 break; 4006 p = tofree; 4007 } 4008 4009 while (*p != NUL) 4010 { 4011 /* Illegal means that there are not enough trail bytes (checked by 4012 * utf_ptr2len()) or too many of them (overlong sequence). */ 4013 len = utf_ptr2len(p); 4014 if (*p >= 0x80 && (len == 1 4015 || utf_char2len(utf_ptr2char(p)) != len)) 4016 { 4017 if (vimconv.vc_type == CONV_NONE) 4018 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 4019 else 4020 { 4021 int l; 4022 4023 len = (int)(p - tofree); 4024 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 4025 { 4026 l = utf_ptr2len(p); 4027 curwin->w_cursor.col += l; 4028 } 4029 } 4030 goto theend; 4031 } 4032 p += len; 4033 } 4034 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 4035 break; 4036 ++curwin->w_cursor.lnum; 4037 curwin->w_cursor.col = 0; 4038 } 4039 4040 /* didn't find it: don't move and beep */ 4041 curwin->w_cursor = pos; 4042 beep_flush(); 4043 4044 theend: 4045 vim_free(tofree); 4046 convert_setup(&vimconv, NULL, NULL); 4047 } 4048 4049 #if defined(FEAT_GUI_GTK) || defined(PROTO) 4050 /* 4051 * Return TRUE if string "s" is a valid utf-8 string. 4052 * When "end" is NULL stop at the first NUL. 4053 * When "end" is positive stop there. 4054 */ 4055 int 4056 utf_valid_string(char_u *s, char_u *end) 4057 { 4058 int l; 4059 char_u *p = s; 4060 4061 while (end == NULL ? *p != NUL : p < end) 4062 { 4063 l = utf8len_tab_zero[*p]; 4064 if (l == 0) 4065 return FALSE; /* invalid lead byte */ 4066 if (end != NULL && p + l > end) 4067 return FALSE; /* incomplete byte sequence */ 4068 ++p; 4069 while (--l > 0) 4070 if ((*p++ & 0xc0) != 0x80) 4071 return FALSE; /* invalid trail byte */ 4072 } 4073 return TRUE; 4074 } 4075 #endif 4076 4077 #if defined(FEAT_GUI) || defined(PROTO) 4078 /* 4079 * Special version of mb_tail_off() for use in ScreenLines[]. 4080 */ 4081 int 4082 dbcs_screen_tail_off(char_u *base, char_u *p) 4083 { 4084 /* It can't be the first byte if a double-byte when not using DBCS, at the 4085 * end of the string or the byte can't start a double-byte. 4086 * For euc-jp an 0x8e byte always means we have a lead byte in the current 4087 * cell. */ 4088 if (*p == NUL || p[1] == NUL 4089 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 4090 || MB_BYTE2LEN(*p) == 1) 4091 return 0; 4092 4093 /* Return 1 when on the lead byte, 0 when on the tail byte. */ 4094 return 1 - dbcs_screen_head_off(base, p); 4095 } 4096 #endif 4097 4098 /* 4099 * If the cursor moves on an trail byte, set the cursor on the lead byte. 4100 * Thus it moves left if necessary. 4101 * Return TRUE when the cursor was adjusted. 4102 */ 4103 void 4104 mb_adjust_cursor(void) 4105 { 4106 mb_adjustpos(curbuf, &curwin->w_cursor); 4107 } 4108 4109 /* 4110 * Adjust position "*lp" to point to the first byte of a multi-byte character. 4111 * If it points to a tail byte it's moved backwards to the head byte. 4112 */ 4113 void 4114 mb_adjustpos(buf_T *buf, pos_T *lp) 4115 { 4116 char_u *p; 4117 4118 if (lp->col > 0 || lp->coladd > 1) 4119 { 4120 p = ml_get_buf(buf, lp->lnum, FALSE); 4121 if (*p == NUL || (int)STRLEN(p) < lp->col) 4122 lp->col = 0; 4123 else 4124 lp->col -= (*mb_head_off)(p, p + lp->col); 4125 /* Reset "coladd" when the cursor would be on the right half of a 4126 * double-wide character. */ 4127 if (lp->coladd == 1 4128 && p[lp->col] != TAB 4129 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 4130 && ptr2cells(p + lp->col) > 1) 4131 lp->coladd = 0; 4132 } 4133 } 4134 4135 /* 4136 * Return a pointer to the character before "*p", if there is one. 4137 */ 4138 char_u * 4139 mb_prevptr( 4140 char_u *line, /* start of the string */ 4141 char_u *p) 4142 { 4143 if (p > line) 4144 MB_PTR_BACK(line, p); 4145 return p; 4146 } 4147 4148 /* 4149 * Return the character length of "str". Each multi-byte character (with 4150 * following composing characters) counts as one. 4151 */ 4152 int 4153 mb_charlen(char_u *str) 4154 { 4155 char_u *p = str; 4156 int count; 4157 4158 if (p == NULL) 4159 return 0; 4160 4161 for (count = 0; *p != NUL; count++) 4162 p += (*mb_ptr2len)(p); 4163 4164 return count; 4165 } 4166 4167 #if defined(FEAT_SPELL) || defined(PROTO) 4168 /* 4169 * Like mb_charlen() but for a string with specified length. 4170 */ 4171 int 4172 mb_charlen_len(char_u *str, int len) 4173 { 4174 char_u *p = str; 4175 int count; 4176 4177 for (count = 0; *p != NUL && p < str + len; count++) 4178 p += (*mb_ptr2len)(p); 4179 4180 return count; 4181 } 4182 #endif 4183 4184 /* 4185 * Try to un-escape a multi-byte character. 4186 * Used for the "to" and "from" part of a mapping. 4187 * Return the un-escaped string if it is a multi-byte character, and advance 4188 * "pp" to just after the bytes that formed it. 4189 * Return NULL if no multi-byte char was found. 4190 */ 4191 char_u * 4192 mb_unescape(char_u **pp) 4193 { 4194 static char_u buf[6]; 4195 int n; 4196 int m = 0; 4197 char_u *str = *pp; 4198 4199 /* Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 4200 * KS_EXTRA KE_CSI to CSI. 4201 * Maximum length of a utf-8 character is 4 bytes. */ 4202 for (n = 0; str[n] != NUL && m < 4; ++n) 4203 { 4204 if (str[n] == K_SPECIAL 4205 && str[n + 1] == KS_SPECIAL 4206 && str[n + 2] == KE_FILLER) 4207 { 4208 buf[m++] = K_SPECIAL; 4209 n += 2; 4210 } 4211 else if ((str[n] == K_SPECIAL 4212 # ifdef FEAT_GUI 4213 || str[n] == CSI 4214 # endif 4215 ) 4216 && str[n + 1] == KS_EXTRA 4217 && str[n + 2] == (int)KE_CSI) 4218 { 4219 buf[m++] = CSI; 4220 n += 2; 4221 } 4222 else if (str[n] == K_SPECIAL 4223 # ifdef FEAT_GUI 4224 || str[n] == CSI 4225 # endif 4226 ) 4227 break; /* a special key can't be a multibyte char */ 4228 else 4229 buf[m++] = str[n]; 4230 buf[m] = NUL; 4231 4232 /* Return a multi-byte character if it's found. An illegal sequence 4233 * will result in a 1 here. */ 4234 if ((*mb_ptr2len)(buf) > 1) 4235 { 4236 *pp = str + n + 1; 4237 return buf; 4238 } 4239 4240 /* Bail out quickly for ASCII. */ 4241 if (buf[0] < 128) 4242 break; 4243 } 4244 return NULL; 4245 } 4246 4247 /* 4248 * Return TRUE if the character at "row"/"col" on the screen is the left side 4249 * of a double-width character. 4250 * Caller must make sure "row" and "col" are not invalid! 4251 */ 4252 int 4253 mb_lefthalve(int row, int col) 4254 { 4255 #ifdef FEAT_HANGULIN 4256 if (composing_hangul) 4257 return TRUE; 4258 #endif 4259 return (*mb_off2cells)(LineOffset[row] + col, 4260 LineOffset[row] + screen_Columns) > 1; 4261 } 4262 4263 /* 4264 * Correct a position on the screen, if it's the right half of a double-wide 4265 * char move it to the left half. Returns the corrected column. 4266 */ 4267 int 4268 mb_fix_col(int col, int row) 4269 { 4270 col = check_col(col); 4271 row = check_row(row); 4272 if (has_mbyte && ScreenLines != NULL && col > 0 4273 && ((enc_dbcs 4274 && ScreenLines[LineOffset[row] + col] != NUL 4275 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 4276 ScreenLines + LineOffset[row] + col)) 4277 || (enc_utf8 && ScreenLines[LineOffset[row] + col] == 0))) 4278 return col - 1; 4279 return col; 4280 } 4281 4282 static int enc_alias_search(char_u *name); 4283 4284 /* 4285 * Skip the Vim specific head of a 'encoding' name. 4286 */ 4287 char_u * 4288 enc_skip(char_u *p) 4289 { 4290 if (STRNCMP(p, "2byte-", 6) == 0) 4291 return p + 6; 4292 if (STRNCMP(p, "8bit-", 5) == 0) 4293 return p + 5; 4294 return p; 4295 } 4296 4297 /* 4298 * Find the canonical name for encoding "enc". 4299 * When the name isn't recognized, returns "enc" itself, but with all lower 4300 * case characters and '_' replaced with '-'. 4301 * Returns an allocated string. NULL for out-of-memory. 4302 */ 4303 char_u * 4304 enc_canonize(char_u *enc) 4305 { 4306 char_u *r; 4307 char_u *p, *s; 4308 int i; 4309 4310 if (STRCMP(enc, "default") == 0) 4311 { 4312 /* Use the default encoding as it's found by set_init_1(). */ 4313 r = get_encoding_default(); 4314 if (r == NULL) 4315 r = (char_u *)"latin1"; 4316 return vim_strsave(r); 4317 } 4318 4319 /* copy "enc" to allocated memory, with room for two '-' */ 4320 r = alloc((unsigned)(STRLEN(enc) + 3)); 4321 if (r != NULL) 4322 { 4323 /* Make it all lower case and replace '_' with '-'. */ 4324 p = r; 4325 for (s = enc; *s != NUL; ++s) 4326 { 4327 if (*s == '_') 4328 *p++ = '-'; 4329 else 4330 *p++ = TOLOWER_ASC(*s); 4331 } 4332 *p = NUL; 4333 4334 /* Skip "2byte-" and "8bit-". */ 4335 p = enc_skip(r); 4336 4337 /* Change "microsoft-cp" to "cp". Used in some spell files. */ 4338 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4339 STRMOVE(p, p + 10); 4340 4341 /* "iso8859" -> "iso-8859" */ 4342 if (STRNCMP(p, "iso8859", 7) == 0) 4343 { 4344 STRMOVE(p + 4, p + 3); 4345 p[3] = '-'; 4346 } 4347 4348 /* "iso-8859n" -> "iso-8859-n" */ 4349 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4350 { 4351 STRMOVE(p + 9, p + 8); 4352 p[8] = '-'; 4353 } 4354 4355 /* "latin-N" -> "latinN" */ 4356 if (STRNCMP(p, "latin-", 6) == 0) 4357 STRMOVE(p + 5, p + 6); 4358 4359 if (enc_canon_search(p) >= 0) 4360 { 4361 /* canonical name can be used unmodified */ 4362 if (p != r) 4363 STRMOVE(r, p); 4364 } 4365 else if ((i = enc_alias_search(p)) >= 0) 4366 { 4367 /* alias recognized, get canonical name */ 4368 vim_free(r); 4369 r = vim_strsave((char_u *)enc_canon_table[i].name); 4370 } 4371 } 4372 return r; 4373 } 4374 4375 /* 4376 * Search for an encoding alias of "name". 4377 * Returns -1 when not found. 4378 */ 4379 static int 4380 enc_alias_search(char_u *name) 4381 { 4382 int i; 4383 4384 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4385 if (STRCMP(name, enc_alias_table[i].name) == 0) 4386 return enc_alias_table[i].canon; 4387 return -1; 4388 } 4389 4390 4391 #ifdef HAVE_LANGINFO_H 4392 # include <langinfo.h> 4393 #endif 4394 4395 #if !defined(FEAT_GUI_MSWIN) || defined(VIMDLL) 4396 /* 4397 * Get the canonicalized encoding from the specified locale string "locale" 4398 * or from the environment variables LC_ALL, LC_CTYPE and LANG. 4399 * Returns an allocated string when successful, NULL when not. 4400 */ 4401 char_u * 4402 enc_locale_env(char *locale) 4403 { 4404 char *s = locale; 4405 char *p; 4406 int i; 4407 char buf[50]; 4408 4409 if (s == NULL || *s == NUL) 4410 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4411 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4412 s = getenv("LANG"); 4413 4414 if (s == NULL || *s == NUL) 4415 return NULL; 4416 4417 /* The most generic locale format is: 4418 * language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4419 * If there is a '.' remove the part before it. 4420 * if there is something after the codeset, remove it. 4421 * Make the name lowercase and replace '_' with '-'. 4422 * Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4423 * "ko_KR.EUC" == "euc-kr" 4424 */ 4425 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4426 { 4427 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4428 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4429 { 4430 /* copy "XY.EUC" to "euc-XY" to buf[10] */ 4431 STRCPY(buf + 10, "euc-"); 4432 buf[14] = p[-2]; 4433 buf[15] = p[-1]; 4434 buf[16] = 0; 4435 s = buf + 10; 4436 } 4437 else 4438 s = p + 1; 4439 } 4440 for (i = 0; i < (int)sizeof(buf) - 1 && s[i] != NUL; ++i) 4441 { 4442 if (s[i] == '_' || s[i] == '-') 4443 buf[i] = '-'; 4444 else if (isalnum((int)s[i])) 4445 buf[i] = TOLOWER_ASC(s[i]); 4446 else 4447 break; 4448 } 4449 buf[i] = NUL; 4450 4451 return enc_canonize((char_u *)buf); 4452 } 4453 #endif 4454 4455 /* 4456 * Get the canonicalized encoding of the current locale. 4457 * Returns an allocated string when successful, NULL when not. 4458 */ 4459 char_u * 4460 enc_locale(void) 4461 { 4462 #ifdef MSWIN 4463 char buf[50]; 4464 long acp = GetACP(); 4465 4466 if (acp == 1200) 4467 STRCPY(buf, "ucs-2le"); 4468 else if (acp == 1252) /* cp1252 is used as latin1 */ 4469 STRCPY(buf, "latin1"); 4470 else 4471 sprintf(buf, "cp%ld", acp); 4472 4473 return enc_canonize((char_u *)buf); 4474 #else 4475 char *s; 4476 4477 # ifdef HAVE_NL_LANGINFO_CODESET 4478 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4479 # endif 4480 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4481 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4482 # endif 4483 s = NULL; 4484 4485 return enc_locale_env(s); 4486 #endif 4487 } 4488 4489 # if defined(MSWIN) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4490 /* 4491 * Convert an encoding name to an MS-Windows codepage. 4492 * Returns zero if no codepage can be figured out. 4493 */ 4494 int 4495 encname2codepage(char_u *name) 4496 { 4497 int cp; 4498 char_u *p = name; 4499 int idx; 4500 4501 if (STRNCMP(p, "8bit-", 5) == 0) 4502 p += 5; 4503 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4504 p += 6; 4505 4506 if (p[0] == 'c' && p[1] == 'p') 4507 cp = atoi((char *)p + 2); 4508 else if ((idx = enc_canon_search(p)) >= 0) 4509 cp = enc_canon_table[idx].codepage; 4510 else 4511 return 0; 4512 if (IsValidCodePage(cp)) 4513 return cp; 4514 return 0; 4515 } 4516 # endif 4517 4518 # if defined(USE_ICONV) || defined(PROTO) 4519 4520 /* 4521 * Call iconv_open() with a check if iconv() works properly (there are broken 4522 * versions). 4523 * Returns (void *)-1 if failed. 4524 * (should return iconv_t, but that causes problems with prototypes). 4525 */ 4526 void * 4527 my_iconv_open(char_u *to, char_u *from) 4528 { 4529 iconv_t fd; 4530 #define ICONV_TESTLEN 400 4531 char_u tobuf[ICONV_TESTLEN]; 4532 char *p; 4533 size_t tolen; 4534 static int iconv_ok = -1; 4535 4536 if (iconv_ok == FALSE) 4537 return (void *)-1; /* detected a broken iconv() previously */ 4538 4539 #ifdef DYNAMIC_ICONV 4540 /* Check if the iconv.dll can be found. */ 4541 if (!iconv_enabled(TRUE)) 4542 return (void *)-1; 4543 #endif 4544 4545 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4546 4547 if (fd != (iconv_t)-1 && iconv_ok == -1) 4548 { 4549 /* 4550 * Do a dummy iconv() call to check if it actually works. There is a 4551 * version of iconv() on Linux that is broken. We can't ignore it, 4552 * because it's wide-spread. The symptoms are that after outputting 4553 * the initial shift state the "to" pointer is NULL and conversion 4554 * stops for no apparent reason after about 8160 characters. 4555 */ 4556 p = (char *)tobuf; 4557 tolen = ICONV_TESTLEN; 4558 (void)iconv(fd, NULL, NULL, &p, &tolen); 4559 if (p == NULL) 4560 { 4561 iconv_ok = FALSE; 4562 iconv_close(fd); 4563 fd = (iconv_t)-1; 4564 } 4565 else 4566 iconv_ok = TRUE; 4567 } 4568 4569 return (void *)fd; 4570 } 4571 4572 /* 4573 * Convert the string "str[slen]" with iconv(). 4574 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4575 * sequence and set "*unconvlenp" to the length of it. 4576 * Returns the converted string in allocated memory. NULL for an error. 4577 * If resultlenp is not NULL, sets it to the result length in bytes. 4578 */ 4579 static char_u * 4580 iconv_string( 4581 vimconv_T *vcp, 4582 char_u *str, 4583 int slen, 4584 int *unconvlenp, 4585 int *resultlenp) 4586 { 4587 const char *from; 4588 size_t fromlen; 4589 char *to; 4590 size_t tolen; 4591 size_t len = 0; 4592 size_t done = 0; 4593 char_u *result = NULL; 4594 char_u *p; 4595 int l; 4596 4597 from = (char *)str; 4598 fromlen = slen; 4599 for (;;) 4600 { 4601 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4602 { 4603 /* Allocate enough room for most conversions. When re-allocating 4604 * increase the buffer size. */ 4605 len = len + fromlen * 2 + 40; 4606 p = alloc((unsigned)len); 4607 if (p != NULL && done > 0) 4608 mch_memmove(p, result, done); 4609 vim_free(result); 4610 result = p; 4611 if (result == NULL) /* out of memory */ 4612 break; 4613 } 4614 4615 to = (char *)result + done; 4616 tolen = len - done - 2; 4617 /* Avoid a warning for systems with a wrong iconv() prototype by 4618 * casting the second argument to void *. */ 4619 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4620 != (size_t)-1) 4621 { 4622 /* Finished, append a NUL. */ 4623 *to = NUL; 4624 break; 4625 } 4626 4627 /* Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4628 * iconv library may use one of them. */ 4629 if (!vcp->vc_fail && unconvlenp != NULL 4630 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4631 { 4632 /* Handle an incomplete sequence at the end. */ 4633 *to = NUL; 4634 *unconvlenp = (int)fromlen; 4635 break; 4636 } 4637 4638 /* Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4639 * iconv library may use one of them. */ 4640 else if (!vcp->vc_fail 4641 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4642 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4643 { 4644 /* Can't convert: insert a '?' and skip a character. This assumes 4645 * conversion from 'encoding' to something else. In other 4646 * situations we don't know what to skip anyway. */ 4647 *to++ = '?'; 4648 if ((*mb_ptr2cells)((char_u *)from) > 1) 4649 *to++ = '?'; 4650 if (enc_utf8) 4651 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4652 else 4653 { 4654 l = (*mb_ptr2len)((char_u *)from); 4655 if (l > (int)fromlen) 4656 l = (int)fromlen; 4657 } 4658 from += l; 4659 fromlen -= l; 4660 } 4661 else if (ICONV_ERRNO != ICONV_E2BIG) 4662 { 4663 /* conversion failed */ 4664 VIM_CLEAR(result); 4665 break; 4666 } 4667 /* Not enough room or skipping illegal sequence. */ 4668 done = to - (char *)result; 4669 } 4670 4671 if (resultlenp != NULL && result != NULL) 4672 *resultlenp = (int)(to - (char *)result); 4673 return result; 4674 } 4675 4676 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4677 /* 4678 * Dynamically load the "iconv.dll" on Win32. 4679 */ 4680 4681 # ifndef DYNAMIC_ICONV /* must be generating prototypes */ 4682 # define HINSTANCE int 4683 # endif 4684 static HINSTANCE hIconvDLL = 0; 4685 static HINSTANCE hMsvcrtDLL = 0; 4686 4687 # ifndef DYNAMIC_ICONV_DLL 4688 # define DYNAMIC_ICONV_DLL "iconv.dll" 4689 # define DYNAMIC_ICONV_DLL_ALT1 "libiconv.dll" 4690 # define DYNAMIC_ICONV_DLL_ALT2 "libiconv2.dll" 4691 # define DYNAMIC_ICONV_DLL_ALT3 "libiconv-2.dll" 4692 # endif 4693 # ifndef DYNAMIC_MSVCRT_DLL 4694 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4695 # endif 4696 4697 /* 4698 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4699 */ 4700 int 4701 iconv_enabled(int verbose) 4702 { 4703 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4704 return TRUE; 4705 4706 /* The iconv DLL file goes under different names, try them all. 4707 * Do the "2" version first, it's newer. */ 4708 #ifdef DYNAMIC_ICONV_DLL_ALT2 4709 if (hIconvDLL == 0) 4710 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT2); 4711 #endif 4712 #ifdef DYNAMIC_ICONV_DLL_ALT3 4713 if (hIconvDLL == 0) 4714 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT3); 4715 #endif 4716 if (hIconvDLL == 0) 4717 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4718 #ifdef DYNAMIC_ICONV_DLL_ALT1 4719 if (hIconvDLL == 0) 4720 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT1); 4721 #endif 4722 4723 if (hIconvDLL != 0) 4724 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4725 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4726 { 4727 /* Only give the message when 'verbose' is set, otherwise it might be 4728 * done whenever a conversion is attempted. */ 4729 if (verbose && p_verbose > 0) 4730 { 4731 verbose_enter(); 4732 semsg(_(e_loadlib), 4733 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4734 verbose_leave(); 4735 } 4736 iconv_end(); 4737 return FALSE; 4738 } 4739 4740 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4741 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4742 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4743 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4744 iconv_errno = get_dll_import_func(hIconvDLL, "_errno"); 4745 if (iconv_errno == NULL) 4746 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4747 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4748 || iconvctl == NULL || iconv_errno == NULL) 4749 { 4750 iconv_end(); 4751 if (verbose && p_verbose > 0) 4752 { 4753 verbose_enter(); 4754 semsg(_(e_loadfunc), "for libiconv"); 4755 verbose_leave(); 4756 } 4757 return FALSE; 4758 } 4759 return TRUE; 4760 } 4761 4762 void 4763 iconv_end(void) 4764 { 4765 /* Don't use iconv() when inputting or outputting characters. */ 4766 if (input_conv.vc_type == CONV_ICONV) 4767 convert_setup(&input_conv, NULL, NULL); 4768 if (output_conv.vc_type == CONV_ICONV) 4769 convert_setup(&output_conv, NULL, NULL); 4770 4771 if (hIconvDLL != 0) 4772 FreeLibrary(hIconvDLL); 4773 if (hMsvcrtDLL != 0) 4774 FreeLibrary(hMsvcrtDLL); 4775 hIconvDLL = 0; 4776 hMsvcrtDLL = 0; 4777 } 4778 # endif /* DYNAMIC_ICONV */ 4779 # endif /* USE_ICONV */ 4780 4781 4782 #ifdef FEAT_GUI 4783 # define USE_IMACTIVATEFUNC (!gui.in_use && *p_imaf != NUL) 4784 # define USE_IMSTATUSFUNC (!gui.in_use && *p_imsf != NUL) 4785 #else 4786 # define USE_IMACTIVATEFUNC (*p_imaf != NUL) 4787 # define USE_IMSTATUSFUNC (*p_imsf != NUL) 4788 #endif 4789 4790 #if defined(FEAT_EVAL) && (defined(FEAT_XIM) || defined(IME_WITHOUT_XIM)) 4791 static void 4792 call_imactivatefunc(int active) 4793 { 4794 typval_T argv[2]; 4795 4796 argv[0].v_type = VAR_NUMBER; 4797 argv[0].vval.v_number = active ? 1 : 0; 4798 argv[1].v_type = VAR_UNKNOWN; 4799 (void)call_func_retnr(p_imaf, 1, argv); 4800 } 4801 4802 static int 4803 call_imstatusfunc(void) 4804 { 4805 int is_active; 4806 4807 /* FIXME: Don't execute user function in unsafe situation. */ 4808 if (exiting || is_autocmd_blocked()) 4809 return FALSE; 4810 /* FIXME: :py print 'xxx' is shown duplicate result. 4811 * Use silent to avoid it. */ 4812 ++msg_silent; 4813 is_active = call_func_retnr(p_imsf, 0, NULL); 4814 --msg_silent; 4815 return (is_active > 0); 4816 } 4817 #endif 4818 4819 #if defined(FEAT_XIM) || defined(PROTO) 4820 4821 # if defined(FEAT_GUI_GTK) || defined(PROTO) 4822 static int xim_has_preediting INIT(= FALSE); /* IM current status */ 4823 4824 /* 4825 * Set preedit_start_col to the current cursor position. 4826 */ 4827 static void 4828 init_preedit_start_col(void) 4829 { 4830 if (State & CMDLINE) 4831 preedit_start_col = cmdline_getvcol_cursor(); 4832 else if (curwin != NULL && curwin->w_buffer != NULL) 4833 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); 4834 /* Prevent that preediting marks the buffer as changed. */ 4835 xim_changed_while_preediting = curbuf->b_changed; 4836 } 4837 4838 static int im_is_active = FALSE; /* IM is enabled for current mode */ 4839 static int preedit_is_active = FALSE; 4840 static int im_preedit_cursor = 0; /* cursor offset in characters */ 4841 static int im_preedit_trailing = 0; /* number of characters after cursor */ 4842 4843 static unsigned long im_commit_handler_id = 0; 4844 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; 4845 static unsigned int im_activatekey_state = 0; 4846 4847 static GtkWidget *preedit_window = NULL; 4848 static GtkWidget *preedit_label = NULL; 4849 4850 static void im_preedit_window_set_position(void); 4851 4852 void 4853 im_set_active(int active) 4854 { 4855 int was_active; 4856 4857 was_active = !!im_get_status(); 4858 im_is_active = (active && !p_imdisable); 4859 4860 if (im_is_active != was_active) 4861 xim_reset(); 4862 } 4863 4864 void 4865 xim_set_focus(int focus) 4866 { 4867 if (xic != NULL) 4868 { 4869 if (focus) 4870 gtk_im_context_focus_in(xic); 4871 else 4872 gtk_im_context_focus_out(xic); 4873 } 4874 } 4875 4876 void 4877 im_set_position(int row, int col) 4878 { 4879 if (xic != NULL) 4880 { 4881 GdkRectangle area; 4882 4883 area.x = FILL_X(col); 4884 area.y = FILL_Y(row); 4885 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); 4886 area.height = gui.char_height; 4887 4888 gtk_im_context_set_cursor_location(xic, &area); 4889 4890 if (p_imst == IM_OVER_THE_SPOT) 4891 im_preedit_window_set_position(); 4892 } 4893 } 4894 4895 # if 0 || defined(PROTO) /* apparently only used in gui_x11.c */ 4896 void 4897 xim_set_preedit(void) 4898 { 4899 im_set_position(gui.row, gui.col); 4900 } 4901 # endif 4902 4903 static void 4904 im_add_to_input(char_u *str, int len) 4905 { 4906 /* Convert from 'termencoding' (always "utf-8") to 'encoding' */ 4907 if (input_conv.vc_type != CONV_NONE) 4908 { 4909 str = string_convert(&input_conv, str, &len); 4910 g_return_if_fail(str != NULL); 4911 } 4912 4913 add_to_input_buf_csi(str, len); 4914 4915 if (input_conv.vc_type != CONV_NONE) 4916 vim_free(str); 4917 4918 if (p_mh) /* blank out the pointer if necessary */ 4919 gui_mch_mousehide(TRUE); 4920 } 4921 4922 static void 4923 im_preedit_window_set_position(void) 4924 { 4925 int x, y, width, height; 4926 int screen_x, screen_y, screen_width, screen_height; 4927 4928 if (preedit_window == NULL) 4929 return; 4930 4931 gui_gtk_get_screen_geom_of_win(gui.drawarea, 4932 &screen_x, &screen_y, &screen_width, &screen_height); 4933 gdk_window_get_origin(gtk_widget_get_window(gui.drawarea), &x, &y); 4934 gtk_window_get_size(GTK_WINDOW(preedit_window), &width, &height); 4935 x = x + FILL_X(gui.col); 4936 y = y + FILL_Y(gui.row); 4937 if (x + width > screen_x + screen_width) 4938 x = screen_x + screen_width - width; 4939 if (y + height > screen_y + screen_height) 4940 y = screen_y + screen_height - height; 4941 gtk_window_move(GTK_WINDOW(preedit_window), x, y); 4942 } 4943 4944 static void 4945 im_preedit_window_open() 4946 { 4947 char *preedit_string; 4948 #if !GTK_CHECK_VERSION(3,16,0) 4949 char buf[8]; 4950 #endif 4951 PangoAttrList *attr_list; 4952 PangoLayout *layout; 4953 #if GTK_CHECK_VERSION(3,0,0) 4954 # if !GTK_CHECK_VERSION(3,16,0) 4955 GdkRGBA color; 4956 # endif 4957 #else 4958 GdkColor color; 4959 #endif 4960 gint w, h; 4961 4962 if (preedit_window == NULL) 4963 { 4964 preedit_window = gtk_window_new(GTK_WINDOW_POPUP); 4965 gtk_window_set_transient_for(GTK_WINDOW(preedit_window), 4966 GTK_WINDOW(gui.mainwin)); 4967 preedit_label = gtk_label_new(""); 4968 gtk_widget_set_name(preedit_label, "vim-gui-preedit-area"); 4969 gtk_container_add(GTK_CONTAINER(preedit_window), preedit_label); 4970 } 4971 4972 #if GTK_CHECK_VERSION(3,16,0) 4973 { 4974 GtkStyleContext * const context 4975 = gtk_widget_get_style_context(gui.drawarea); 4976 GtkCssProvider * const provider = gtk_css_provider_new(); 4977 gchar *css = NULL; 4978 const char * const fontname 4979 = pango_font_description_get_family(gui.norm_font); 4980 gint fontsize 4981 = pango_font_description_get_size(gui.norm_font) / PANGO_SCALE; 4982 gchar *fontsize_propval = NULL; 4983 4984 if (!pango_font_description_get_size_is_absolute(gui.norm_font)) 4985 { 4986 /* fontsize was given in points. Convert it into that in pixels 4987 * to use with CSS. */ 4988 GdkScreen * const screen 4989 = gdk_window_get_screen(gtk_widget_get_window(gui.mainwin)); 4990 const gdouble dpi = gdk_screen_get_resolution(screen); 4991 fontsize = dpi * fontsize / 72; 4992 } 4993 if (fontsize > 0) 4994 fontsize_propval = g_strdup_printf("%dpx", fontsize); 4995 else 4996 fontsize_propval = g_strdup_printf("inherit"); 4997 4998 css = g_strdup_printf( 4999 "widget#vim-gui-preedit-area {\n" 5000 " font-family: %s,monospace;\n" 5001 " font-size: %s;\n" 5002 " color: #%.2lx%.2lx%.2lx;\n" 5003 " background-color: #%.2lx%.2lx%.2lx;\n" 5004 "}\n", 5005 fontname != NULL ? fontname : "inherit", 5006 fontsize_propval, 5007 (gui.norm_pixel >> 16) & 0xff, 5008 (gui.norm_pixel >> 8) & 0xff, 5009 gui.norm_pixel & 0xff, 5010 (gui.back_pixel >> 16) & 0xff, 5011 (gui.back_pixel >> 8) & 0xff, 5012 gui.back_pixel & 0xff); 5013 5014 gtk_css_provider_load_from_data(provider, css, -1, NULL); 5015 gtk_style_context_add_provider(context, 5016 GTK_STYLE_PROVIDER(provider), G_MAXUINT); 5017 5018 g_free(css); 5019 g_free(fontsize_propval); 5020 g_object_unref(provider); 5021 } 5022 #elif GTK_CHECK_VERSION(3,0,0) 5023 gtk_widget_override_font(preedit_label, gui.norm_font); 5024 5025 vim_snprintf(buf, sizeof(buf), "#%06X", gui.norm_pixel); 5026 gdk_rgba_parse(&color, buf); 5027 gtk_widget_override_color(preedit_label, GTK_STATE_FLAG_NORMAL, &color); 5028 5029 vim_snprintf(buf, sizeof(buf), "#%06X", gui.back_pixel); 5030 gdk_rgba_parse(&color, buf); 5031 gtk_widget_override_background_color(preedit_label, GTK_STATE_FLAG_NORMAL, 5032 &color); 5033 #else 5034 gtk_widget_modify_font(preedit_label, gui.norm_font); 5035 5036 vim_snprintf(buf, sizeof(buf), "#%06X", (unsigned)gui.norm_pixel); 5037 gdk_color_parse(buf, &color); 5038 gtk_widget_modify_fg(preedit_label, GTK_STATE_NORMAL, &color); 5039 5040 vim_snprintf(buf, sizeof(buf), "#%06X", (unsigned)gui.back_pixel); 5041 gdk_color_parse(buf, &color); 5042 gtk_widget_modify_bg(preedit_window, GTK_STATE_NORMAL, &color); 5043 #endif 5044 5045 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5046 5047 if (preedit_string[0] != NUL) 5048 { 5049 gtk_label_set_text(GTK_LABEL(preedit_label), preedit_string); 5050 gtk_label_set_attributes(GTK_LABEL(preedit_label), attr_list); 5051 5052 layout = gtk_label_get_layout(GTK_LABEL(preedit_label)); 5053 pango_layout_get_pixel_size(layout, &w, &h); 5054 h = MAX(h, gui.char_height); 5055 gtk_window_resize(GTK_WINDOW(preedit_window), w, h); 5056 5057 gtk_widget_show_all(preedit_window); 5058 5059 im_preedit_window_set_position(); 5060 } 5061 5062 g_free(preedit_string); 5063 pango_attr_list_unref(attr_list); 5064 } 5065 5066 static void 5067 im_preedit_window_close() 5068 { 5069 if (preedit_window != NULL) 5070 gtk_widget_hide(preedit_window); 5071 } 5072 5073 static void 5074 im_show_preedit() 5075 { 5076 im_preedit_window_open(); 5077 5078 if (p_mh) /* blank out the pointer if necessary */ 5079 gui_mch_mousehide(TRUE); 5080 } 5081 5082 static void 5083 im_delete_preedit(void) 5084 { 5085 char_u bskey[] = {CSI, 'k', 'b'}; 5086 char_u delkey[] = {CSI, 'k', 'D'}; 5087 5088 if (p_imst == IM_OVER_THE_SPOT) 5089 { 5090 im_preedit_window_close(); 5091 return; 5092 } 5093 5094 if (State & NORMAL 5095 #ifdef FEAT_TERMINAL 5096 && !term_use_loop() 5097 #endif 5098 ) 5099 { 5100 im_preedit_cursor = 0; 5101 return; 5102 } 5103 for (; im_preedit_cursor > 0; --im_preedit_cursor) 5104 add_to_input_buf(bskey, (int)sizeof(bskey)); 5105 5106 for (; im_preedit_trailing > 0; --im_preedit_trailing) 5107 add_to_input_buf(delkey, (int)sizeof(delkey)); 5108 } 5109 5110 /* 5111 * Move the cursor left by "num_move_back" characters. 5112 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for 5113 * these K_LEFT keys. 5114 */ 5115 static void 5116 im_correct_cursor(int num_move_back) 5117 { 5118 char_u backkey[] = {CSI, 'k', 'l'}; 5119 5120 if (State & NORMAL) 5121 return; 5122 # ifdef FEAT_RIGHTLEFT 5123 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) 5124 backkey[2] = 'r'; 5125 # endif 5126 for (; num_move_back > 0; --num_move_back) 5127 add_to_input_buf(backkey, (int)sizeof(backkey)); 5128 } 5129 5130 static int xim_expected_char = NUL; 5131 static int xim_ignored_char = FALSE; 5132 5133 /* 5134 * Update the mode and cursor while in an IM callback. 5135 */ 5136 static void 5137 im_show_info(void) 5138 { 5139 int old_vgetc_busy; 5140 5141 old_vgetc_busy = vgetc_busy; 5142 vgetc_busy = TRUE; 5143 showmode(); 5144 vgetc_busy = old_vgetc_busy; 5145 if ((State & NORMAL) || (State & INSERT)) 5146 setcursor(); 5147 out_flush(); 5148 } 5149 5150 /* 5151 * Callback invoked when the user finished preediting. 5152 * Put the final string into the input buffer. 5153 */ 5154 static void 5155 im_commit_cb(GtkIMContext *context UNUSED, 5156 const gchar *str, 5157 gpointer data UNUSED) 5158 { 5159 int slen = (int)STRLEN(str); 5160 int add_to_input = TRUE; 5161 int clen; 5162 int len = slen; 5163 int commit_with_preedit = TRUE; 5164 char_u *im_str; 5165 5166 #ifdef XIM_DEBUG 5167 xim_log("im_commit_cb(): %s\n", str); 5168 #endif 5169 5170 if (p_imst == IM_ON_THE_SPOT) 5171 { 5172 /* The imhangul module doesn't reset the preedit string before 5173 * committing. Call im_delete_preedit() to work around that. */ 5174 im_delete_preedit(); 5175 5176 /* Indicate that preediting has finished. */ 5177 if (preedit_start_col == MAXCOL) 5178 { 5179 init_preedit_start_col(); 5180 commit_with_preedit = FALSE; 5181 } 5182 5183 /* The thing which setting "preedit_start_col" to MAXCOL means that 5184 * "preedit_start_col" will be set forcedly when calling 5185 * preedit_changed_cb() next time. 5186 * "preedit_start_col" should not reset with MAXCOL on this part. Vim 5187 * is simulating the preediting by using add_to_input_str(). when 5188 * preedit begin immediately before committed, the typebuf is not 5189 * flushed to screen, then it can't get correct "preedit_start_col". 5190 * Thus, it should calculate the cells by adding cells of the committed 5191 * string. */ 5192 if (input_conv.vc_type != CONV_NONE) 5193 { 5194 im_str = string_convert(&input_conv, (char_u *)str, &len); 5195 g_return_if_fail(im_str != NULL); 5196 } 5197 else 5198 im_str = (char_u *)str; 5199 5200 clen = mb_string2cells(im_str, len); 5201 5202 if (input_conv.vc_type != CONV_NONE) 5203 vim_free(im_str); 5204 preedit_start_col += clen; 5205 } 5206 5207 /* Is this a single character that matches a keypad key that's just 5208 * been pressed? If so, we don't want it to be entered as such - let 5209 * us carry on processing the raw keycode so that it may be used in 5210 * mappings as <kSomething>. */ 5211 if (xim_expected_char != NUL) 5212 { 5213 /* We're currently processing a keypad or other special key */ 5214 if (slen == 1 && str[0] == xim_expected_char) 5215 { 5216 /* It's a match - don't do it here */ 5217 xim_ignored_char = TRUE; 5218 add_to_input = FALSE; 5219 } 5220 else 5221 { 5222 /* Not a match */ 5223 xim_ignored_char = FALSE; 5224 } 5225 } 5226 5227 if (add_to_input) 5228 im_add_to_input((char_u *)str, slen); 5229 5230 if (p_imst == IM_ON_THE_SPOT) 5231 { 5232 /* Inserting chars while "im_is_active" is set does not cause a 5233 * change of buffer. When the chars are committed the buffer must be 5234 * marked as changed. */ 5235 if (!commit_with_preedit) 5236 preedit_start_col = MAXCOL; 5237 5238 /* This flag is used in changed() at next call. */ 5239 xim_changed_while_preediting = TRUE; 5240 } 5241 5242 if (gtk_main_level() > 0) 5243 gtk_main_quit(); 5244 } 5245 5246 /* 5247 * Callback invoked after start to the preedit. 5248 */ 5249 static void 5250 im_preedit_start_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 5251 { 5252 #ifdef XIM_DEBUG 5253 xim_log("im_preedit_start_cb()\n"); 5254 #endif 5255 5256 im_is_active = TRUE; 5257 preedit_is_active = TRUE; 5258 gui_update_cursor(TRUE, FALSE); 5259 im_show_info(); 5260 } 5261 5262 /* 5263 * Callback invoked after end to the preedit. 5264 */ 5265 static void 5266 im_preedit_end_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 5267 { 5268 #ifdef XIM_DEBUG 5269 xim_log("im_preedit_end_cb()\n"); 5270 #endif 5271 im_delete_preedit(); 5272 5273 /* Indicate that preediting has finished */ 5274 if (p_imst == IM_ON_THE_SPOT) 5275 preedit_start_col = MAXCOL; 5276 xim_has_preediting = FALSE; 5277 5278 #if 0 5279 /* Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was 5280 * switched off unintentionally. We now use preedit_is_active (added by 5281 * SungHyun Nam). */ 5282 im_is_active = FALSE; 5283 #endif 5284 preedit_is_active = FALSE; 5285 gui_update_cursor(TRUE, FALSE); 5286 im_show_info(); 5287 } 5288 5289 /* 5290 * Callback invoked after changes to the preedit string. If the preedit 5291 * string was empty before, remember the preedit start column so we know 5292 * where to apply feedback attributes. Delete the previous preedit string 5293 * if there was one, save the new preedit cursor offset, and put the new 5294 * string into the input buffer. 5295 * 5296 * TODO: The pragmatic "put into input buffer" approach used here has 5297 * several fundamental problems: 5298 * 5299 * - The characters in the preedit string are subject to remapping. 5300 * That's broken, only the finally committed string should be remapped. 5301 * 5302 * - There is a race condition involved: The retrieved value for the 5303 * current cursor position will be wrong if any unprocessed characters 5304 * are still queued in the input buffer. 5305 * 5306 * - Due to the lack of synchronization between the file buffer in memory 5307 * and any typed characters, it's practically impossible to implement the 5308 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM 5309 * modules for languages such as Thai are likely to rely on this feature 5310 * for proper operation. 5311 * 5312 * Conclusions: I think support for preediting needs to be moved to the 5313 * core parts of Vim. Ideally, until it has been committed, the preediting 5314 * string should only be displayed and not affect the buffer content at all. 5315 * The question how to deal with the synchronization issue still remains. 5316 * Circumventing the input buffer is probably not desirable. Anyway, I think 5317 * implementing "retrieve_surrounding" is the only hard problem. 5318 * 5319 * One way to solve all of this in a clean manner would be to queue all key 5320 * press/release events "as is" in the input buffer, and apply the IM filtering 5321 * at the receiving end of the queue. This, however, would have a rather large 5322 * impact on the code base. If there is an easy way to force processing of all 5323 * remaining input from within the "retrieve_surrounding" signal handler, this 5324 * might not be necessary. Gotta ask on vim-dev for opinions. 5325 */ 5326 static void 5327 im_preedit_changed_cb(GtkIMContext *context, gpointer data UNUSED) 5328 { 5329 char *preedit_string = NULL; 5330 int cursor_index = 0; 5331 int num_move_back = 0; 5332 char_u *str; 5333 char_u *p; 5334 int i; 5335 5336 if (p_imst == IM_ON_THE_SPOT) 5337 gtk_im_context_get_preedit_string(context, 5338 &preedit_string, NULL, 5339 &cursor_index); 5340 else 5341 gtk_im_context_get_preedit_string(context, 5342 &preedit_string, NULL, 5343 NULL); 5344 5345 #ifdef XIM_DEBUG 5346 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); 5347 #endif 5348 5349 g_return_if_fail(preedit_string != NULL); /* just in case */ 5350 5351 if (p_imst == IM_OVER_THE_SPOT) 5352 { 5353 if (preedit_string[0] == NUL) 5354 { 5355 xim_has_preediting = FALSE; 5356 im_delete_preedit(); 5357 } 5358 else 5359 { 5360 xim_has_preediting = TRUE; 5361 im_show_preedit(); 5362 } 5363 } 5364 else 5365 { 5366 /* If preedit_start_col is MAXCOL set it to the current cursor position. */ 5367 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') 5368 { 5369 xim_has_preediting = TRUE; 5370 5371 /* Urgh, this breaks if the input buffer isn't empty now */ 5372 init_preedit_start_col(); 5373 } 5374 else if (cursor_index == 0 && preedit_string[0] == '\0') 5375 { 5376 xim_has_preediting = FALSE; 5377 5378 /* If at the start position (after typing backspace) 5379 * preedit_start_col must be reset. */ 5380 preedit_start_col = MAXCOL; 5381 } 5382 5383 im_delete_preedit(); 5384 5385 /* 5386 * Compute the end of the preediting area: "preedit_end_col". 5387 * According to the documentation of gtk_im_context_get_preedit_string(), 5388 * the cursor_pos output argument returns the offset in bytes. This is 5389 * unfortunately not true -- real life shows the offset is in characters, 5390 * and the GTK+ source code agrees with me. Will file a bug later. 5391 */ 5392 if (preedit_start_col != MAXCOL) 5393 preedit_end_col = preedit_start_col; 5394 str = (char_u *)preedit_string; 5395 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) 5396 { 5397 int is_composing; 5398 5399 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); 5400 /* 5401 * These offsets are used as counters when generating <BS> and <Del> 5402 * to delete the preedit string. So don't count composing characters 5403 * unless 'delcombine' is enabled. 5404 */ 5405 if (!is_composing || p_deco) 5406 { 5407 if (i < cursor_index) 5408 ++im_preedit_cursor; 5409 else 5410 ++im_preedit_trailing; 5411 } 5412 if (!is_composing && i >= cursor_index) 5413 { 5414 /* This is essentially the same as im_preedit_trailing, except 5415 * composing characters are not counted even if p_deco is set. */ 5416 ++num_move_back; 5417 } 5418 if (preedit_start_col != MAXCOL) 5419 preedit_end_col += utf_ptr2cells(p); 5420 } 5421 5422 if (p > str) 5423 { 5424 im_add_to_input(str, (int)(p - str)); 5425 im_correct_cursor(num_move_back); 5426 } 5427 } 5428 5429 g_free(preedit_string); 5430 5431 if (gtk_main_level() > 0) 5432 gtk_main_quit(); 5433 } 5434 5435 /* 5436 * Translate the Pango attributes at iter to Vim highlighting attributes. 5437 * Ignore attributes not supported by Vim highlighting. This shouldn't have 5438 * too much impact -- right now we handle even more attributes than necessary 5439 * for the IM modules I tested with. 5440 */ 5441 static int 5442 translate_pango_attributes(PangoAttrIterator *iter) 5443 { 5444 PangoAttribute *attr; 5445 int char_attr = HL_NORMAL; 5446 5447 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); 5448 if (attr != NULL && ((PangoAttrInt *)attr)->value 5449 != (int)PANGO_UNDERLINE_NONE) 5450 char_attr |= HL_UNDERLINE; 5451 5452 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); 5453 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) 5454 char_attr |= HL_BOLD; 5455 5456 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); 5457 if (attr != NULL && ((PangoAttrInt *)attr)->value 5458 != (int)PANGO_STYLE_NORMAL) 5459 char_attr |= HL_ITALIC; 5460 5461 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); 5462 if (attr != NULL) 5463 { 5464 const PangoColor *color = &((PangoAttrColor *)attr)->color; 5465 5466 /* Assume inverse if black background is requested */ 5467 if ((color->red | color->green | color->blue) == 0) 5468 char_attr |= HL_INVERSE; 5469 } 5470 5471 return char_attr; 5472 } 5473 5474 /* 5475 * Retrieve the highlighting attributes at column col in the preedit string. 5476 * Return -1 if not in preediting mode or if col is out of range. 5477 */ 5478 int 5479 im_get_feedback_attr(int col) 5480 { 5481 char *preedit_string = NULL; 5482 PangoAttrList *attr_list = NULL; 5483 int char_attr = -1; 5484 5485 if (xic == NULL) 5486 return char_attr; 5487 5488 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5489 5490 if (preedit_string != NULL && attr_list != NULL) 5491 { 5492 int idx; 5493 5494 /* Get the byte index as used by PangoAttrIterator */ 5495 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col) 5496 idx += utfc_ptr2len((char_u *)preedit_string + idx); 5497 5498 if (preedit_string[idx] != '\0') 5499 { 5500 PangoAttrIterator *iter; 5501 int start, end; 5502 5503 char_attr = HL_NORMAL; 5504 iter = pango_attr_list_get_iterator(attr_list); 5505 5506 /* Extract all relevant attributes from the list. */ 5507 do 5508 { 5509 pango_attr_iterator_range(iter, &start, &end); 5510 5511 if (idx >= start && idx < end) 5512 char_attr |= translate_pango_attributes(iter); 5513 } 5514 while (pango_attr_iterator_next(iter)); 5515 5516 pango_attr_iterator_destroy(iter); 5517 } 5518 } 5519 5520 if (attr_list != NULL) 5521 pango_attr_list_unref(attr_list); 5522 g_free(preedit_string); 5523 5524 return char_attr; 5525 } 5526 5527 void 5528 xim_init(void) 5529 { 5530 #ifdef XIM_DEBUG 5531 xim_log("xim_init()\n"); 5532 #endif 5533 5534 g_return_if_fail(gui.drawarea != NULL); 5535 g_return_if_fail(gtk_widget_get_window(gui.drawarea) != NULL); 5536 5537 xic = gtk_im_multicontext_new(); 5538 g_object_ref(xic); 5539 5540 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", 5541 G_CALLBACK(&im_commit_cb), NULL); 5542 g_signal_connect(G_OBJECT(xic), "preedit_changed", 5543 G_CALLBACK(&im_preedit_changed_cb), NULL); 5544 g_signal_connect(G_OBJECT(xic), "preedit_start", 5545 G_CALLBACK(&im_preedit_start_cb), NULL); 5546 g_signal_connect(G_OBJECT(xic), "preedit_end", 5547 G_CALLBACK(&im_preedit_end_cb), NULL); 5548 5549 gtk_im_context_set_client_window(xic, gtk_widget_get_window(gui.drawarea)); 5550 } 5551 5552 void 5553 im_shutdown(void) 5554 { 5555 #ifdef XIM_DEBUG 5556 xim_log("im_shutdown()\n"); 5557 #endif 5558 5559 if (xic != NULL) 5560 { 5561 gtk_im_context_focus_out(xic); 5562 g_object_unref(xic); 5563 xic = NULL; 5564 } 5565 im_is_active = FALSE; 5566 im_commit_handler_id = 0; 5567 if (p_imst == IM_ON_THE_SPOT) 5568 preedit_start_col = MAXCOL; 5569 xim_has_preediting = FALSE; 5570 } 5571 5572 /* 5573 * Convert the string argument to keyval and state for GdkEventKey. 5574 * If str is valid return TRUE, otherwise FALSE. 5575 * 5576 * See 'imactivatekey' for documentation of the format. 5577 */ 5578 static int 5579 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) 5580 { 5581 const char *mods_end; 5582 unsigned tmp_keyval; 5583 unsigned tmp_state = 0; 5584 5585 mods_end = strrchr(str, '-'); 5586 mods_end = (mods_end != NULL) ? mods_end + 1 : str; 5587 5588 /* Parse modifier keys */ 5589 while (str < mods_end) 5590 switch (*str++) 5591 { 5592 case '-': break; 5593 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; 5594 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; 5595 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; 5596 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; 5597 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; 5598 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; 5599 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; 5600 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; 5601 default: 5602 return FALSE; 5603 } 5604 5605 tmp_keyval = gdk_keyval_from_name(str); 5606 5607 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) 5608 return FALSE; 5609 5610 if (keyval != NULL) 5611 *keyval = tmp_keyval; 5612 if (state != NULL) 5613 *state = tmp_state; 5614 5615 return TRUE; 5616 } 5617 5618 /* 5619 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an 5620 * empty string is also regarded as valid. 5621 * 5622 * Note: The numerical key value of p_imak is cached if it was valid; thus 5623 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever 5624 * 'imak' changes. This is currently the case but not obvious -- should 5625 * probably rename the function for clarity. 5626 */ 5627 int 5628 im_xim_isvalid_imactivate(void) 5629 { 5630 if (p_imak[0] == NUL) 5631 { 5632 im_activatekey_keyval = GDK_VoidSymbol; 5633 im_activatekey_state = 0; 5634 return TRUE; 5635 } 5636 5637 return im_string_to_keyval((const char *)p_imak, 5638 &im_activatekey_keyval, 5639 &im_activatekey_state); 5640 } 5641 5642 static void 5643 im_synthesize_keypress(unsigned int keyval, unsigned int state) 5644 { 5645 GdkEventKey *event; 5646 5647 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); 5648 g_object_ref(gtk_widget_get_window(gui.drawarea)); 5649 /* unreffed by gdk_event_free() */ 5650 event->window = gtk_widget_get_window(gui.drawarea); 5651 event->send_event = TRUE; 5652 event->time = GDK_CURRENT_TIME; 5653 event->state = state; 5654 event->keyval = keyval; 5655 event->hardware_keycode = /* needed for XIM */ 5656 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); 5657 event->length = 0; 5658 event->string = NULL; 5659 5660 gtk_im_context_filter_keypress(xic, event); 5661 5662 /* For consistency, also send the corresponding release event. */ 5663 event->type = GDK_KEY_RELEASE; 5664 event->send_event = FALSE; 5665 gtk_im_context_filter_keypress(xic, event); 5666 5667 gdk_event_free((GdkEvent *)event); 5668 } 5669 5670 void 5671 xim_reset(void) 5672 { 5673 # ifdef FEAT_EVAL 5674 if (USE_IMACTIVATEFUNC) 5675 call_imactivatefunc(im_is_active); 5676 else 5677 # endif 5678 if (xic != NULL) 5679 { 5680 gtk_im_context_reset(xic); 5681 5682 if (p_imdisable) 5683 im_shutdown(); 5684 else 5685 { 5686 xim_set_focus(gui.in_focus); 5687 5688 if (im_activatekey_keyval != GDK_VoidSymbol) 5689 { 5690 if (im_is_active) 5691 { 5692 g_signal_handler_block(xic, im_commit_handler_id); 5693 im_synthesize_keypress(im_activatekey_keyval, 5694 im_activatekey_state); 5695 g_signal_handler_unblock(xic, im_commit_handler_id); 5696 } 5697 } 5698 else 5699 { 5700 im_shutdown(); 5701 xim_init(); 5702 xim_set_focus(gui.in_focus); 5703 } 5704 } 5705 } 5706 5707 if (p_imst == IM_ON_THE_SPOT) 5708 preedit_start_col = MAXCOL; 5709 xim_has_preediting = FALSE; 5710 } 5711 5712 int 5713 xim_queue_key_press_event(GdkEventKey *event, int down) 5714 { 5715 if (down) 5716 { 5717 /* 5718 * Workaround GTK2 XIM 'feature' that always converts keypad keys to 5719 * chars., even when not part of an IM sequence (ref. feature of 5720 * gdk/gdkkeyuni.c). 5721 * Flag any keypad keys that might represent a single char. 5722 * If this (on its own - i.e., not part of an IM sequence) is 5723 * committed while we're processing one of these keys, we can ignore 5724 * that commit and go ahead & process it ourselves. That way we can 5725 * still distinguish keypad keys for use in mappings. 5726 * Also add GDK_space to make <S-Space> work. 5727 */ 5728 switch (event->keyval) 5729 { 5730 case GDK_KP_Add: xim_expected_char = '+'; break; 5731 case GDK_KP_Subtract: xim_expected_char = '-'; break; 5732 case GDK_KP_Divide: xim_expected_char = '/'; break; 5733 case GDK_KP_Multiply: xim_expected_char = '*'; break; 5734 case GDK_KP_Decimal: xim_expected_char = '.'; break; 5735 case GDK_KP_Equal: xim_expected_char = '='; break; 5736 case GDK_KP_0: xim_expected_char = '0'; break; 5737 case GDK_KP_1: xim_expected_char = '1'; break; 5738 case GDK_KP_2: xim_expected_char = '2'; break; 5739 case GDK_KP_3: xim_expected_char = '3'; break; 5740 case GDK_KP_4: xim_expected_char = '4'; break; 5741 case GDK_KP_5: xim_expected_char = '5'; break; 5742 case GDK_KP_6: xim_expected_char = '6'; break; 5743 case GDK_KP_7: xim_expected_char = '7'; break; 5744 case GDK_KP_8: xim_expected_char = '8'; break; 5745 case GDK_KP_9: xim_expected_char = '9'; break; 5746 case GDK_space: xim_expected_char = ' '; break; 5747 default: xim_expected_char = NUL; 5748 } 5749 xim_ignored_char = FALSE; 5750 } 5751 5752 /* 5753 * When typing fFtT, XIM may be activated. Thus it must pass 5754 * gtk_im_context_filter_keypress() in Normal mode. 5755 * And while doing :sh too. 5756 */ 5757 if (xic != NULL && !p_imdisable 5758 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) 5759 { 5760 /* 5761 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is 5762 * always aware of the current status of IM, and can even emulate 5763 * the activation key for modules that don't support one. 5764 */ 5765 if (event->keyval == im_activatekey_keyval 5766 && (event->state & im_activatekey_state) == im_activatekey_state) 5767 { 5768 unsigned int state_mask; 5769 5770 /* Require the state of the 3 most used modifiers to match exactly. 5771 * Otherwise e.g. <S-C-space> would be unusable for other purposes 5772 * if the IM activate key is <S-space>. */ 5773 state_mask = im_activatekey_state; 5774 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK 5775 | (int)GDK_MOD1_MASK); 5776 5777 if ((event->state & state_mask) != im_activatekey_state) 5778 return FALSE; 5779 5780 /* Don't send it a second time on GDK_KEY_RELEASE. */ 5781 if (event->type != GDK_KEY_PRESS) 5782 return TRUE; 5783 5784 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE)) 5785 { 5786 im_set_active(FALSE); 5787 5788 /* ":lmap" mappings exists, toggle use of mappings. */ 5789 State ^= LANGMAP; 5790 if (State & LANGMAP) 5791 { 5792 curbuf->b_p_iminsert = B_IMODE_NONE; 5793 State &= ~LANGMAP; 5794 } 5795 else 5796 { 5797 curbuf->b_p_iminsert = B_IMODE_LMAP; 5798 State |= LANGMAP; 5799 } 5800 return TRUE; 5801 } 5802 5803 return gtk_im_context_filter_keypress(xic, event); 5804 } 5805 5806 /* Don't filter events through the IM context if IM isn't active 5807 * right now. Unlike with GTK+ 1.2 we cannot rely on the IM module 5808 * not doing anything before the activation key was sent. */ 5809 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) 5810 { 5811 int imresult = gtk_im_context_filter_keypress(xic, event); 5812 5813 if (p_imst == IM_ON_THE_SPOT) 5814 { 5815 /* Some XIM send following sequence: 5816 * 1. preedited string. 5817 * 2. committed string. 5818 * 3. line changed key. 5819 * 4. preedited string. 5820 * 5. remove preedited string. 5821 * if 3, Vim can't move back the above line for 5. 5822 * thus, this part should not parse the key. */ 5823 if (!imresult && preedit_start_col != MAXCOL 5824 && event->keyval == GDK_Return) 5825 { 5826 im_synthesize_keypress(GDK_Return, 0U); 5827 return FALSE; 5828 } 5829 } 5830 5831 /* If XIM tried to commit a keypad key as a single char., 5832 * ignore it so we can use the keypad key 'raw', for mappings. */ 5833 if (xim_expected_char != NUL && xim_ignored_char) 5834 /* We had a keypad key, and XIM tried to thieve it */ 5835 return FALSE; 5836 5837 /* This is supposed to fix a problem with iBus, that space 5838 * characters don't work in input mode. */ 5839 xim_expected_char = NUL; 5840 5841 /* Normal processing */ 5842 return imresult; 5843 } 5844 } 5845 5846 return FALSE; 5847 } 5848 5849 int 5850 im_get_status(void) 5851 { 5852 # ifdef FEAT_EVAL 5853 if (USE_IMSTATUSFUNC) 5854 return call_imstatusfunc(); 5855 # endif 5856 return im_is_active; 5857 } 5858 5859 int 5860 preedit_get_status(void) 5861 { 5862 return preedit_is_active; 5863 } 5864 5865 int 5866 im_is_preediting(void) 5867 { 5868 return xim_has_preediting; 5869 } 5870 5871 # else /* !FEAT_GUI_GTK */ 5872 5873 static int xim_is_active = FALSE; /* XIM should be active in the current 5874 mode */ 5875 static int xim_has_focus = FALSE; /* XIM is really being used for Vim */ 5876 # ifdef FEAT_GUI_X11 5877 static XIMStyle input_style; 5878 static int status_area_enabled = TRUE; 5879 # endif 5880 5881 /* 5882 * Switch using XIM on/off. This is used by the code that changes "State". 5883 * When 'imactivatefunc' is defined use that function instead. 5884 */ 5885 void 5886 im_set_active(int active_arg) 5887 { 5888 int active = active_arg; 5889 5890 /* If 'imdisable' is set, XIM is never active. */ 5891 if (p_imdisable) 5892 active = FALSE; 5893 else if (input_style & XIMPreeditPosition) 5894 /* There is a problem in switching XIM off when preediting is used, 5895 * and it is not clear how this can be solved. For now, keep XIM on 5896 * all the time, like it was done in Vim 5.8. */ 5897 active = TRUE; 5898 5899 # if defined(FEAT_EVAL) 5900 if (USE_IMACTIVATEFUNC) 5901 { 5902 if (active != im_get_status()) 5903 { 5904 call_imactivatefunc(active); 5905 xim_has_focus = active; 5906 } 5907 return; 5908 } 5909 # endif 5910 5911 if (xic == NULL) 5912 return; 5913 5914 /* Remember the active state, it is needed when Vim gets keyboard focus. */ 5915 xim_is_active = active; 5916 xim_set_preedit(); 5917 } 5918 5919 /* 5920 * Adjust using XIM for gaining or losing keyboard focus. Also called when 5921 * "xim_is_active" changes. 5922 */ 5923 void 5924 xim_set_focus(int focus) 5925 { 5926 if (xic == NULL) 5927 return; 5928 5929 /* 5930 * XIM only gets focus when the Vim window has keyboard focus and XIM has 5931 * been set active for the current mode. 5932 */ 5933 if (focus && xim_is_active) 5934 { 5935 if (!xim_has_focus) 5936 { 5937 xim_has_focus = TRUE; 5938 XSetICFocus(xic); 5939 } 5940 } 5941 else 5942 { 5943 if (xim_has_focus) 5944 { 5945 xim_has_focus = FALSE; 5946 XUnsetICFocus(xic); 5947 } 5948 } 5949 } 5950 5951 void 5952 im_set_position(int row UNUSED, int col UNUSED) 5953 { 5954 xim_set_preedit(); 5955 } 5956 5957 /* 5958 * Set the XIM to the current cursor position. 5959 */ 5960 void 5961 xim_set_preedit(void) 5962 { 5963 XVaNestedList attr_list; 5964 XRectangle spot_area; 5965 XPoint over_spot; 5966 int line_space; 5967 5968 if (xic == NULL) 5969 return; 5970 5971 xim_set_focus(TRUE); 5972 5973 if (!xim_has_focus) 5974 { 5975 /* hide XIM cursor */ 5976 over_spot.x = 0; 5977 over_spot.y = -100; /* arbitrary invisible position */ 5978 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5979 XNSpotLocation, 5980 &over_spot, 5981 NULL); 5982 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); 5983 XFree(attr_list); 5984 return; 5985 } 5986 5987 if (input_style & XIMPreeditPosition) 5988 { 5989 if (xim_fg_color == INVALCOLOR) 5990 { 5991 xim_fg_color = gui.def_norm_pixel; 5992 xim_bg_color = gui.def_back_pixel; 5993 } 5994 over_spot.x = TEXT_X(gui.col); 5995 over_spot.y = TEXT_Y(gui.row); 5996 spot_area.x = 0; 5997 spot_area.y = 0; 5998 spot_area.height = gui.char_height * Rows; 5999 spot_area.width = gui.char_width * Columns; 6000 line_space = gui.char_height; 6001 attr_list = (XVaNestedList) XVaCreateNestedList(0, 6002 XNSpotLocation, &over_spot, 6003 XNForeground, (Pixel) xim_fg_color, 6004 XNBackground, (Pixel) xim_bg_color, 6005 XNArea, &spot_area, 6006 XNLineSpace, line_space, 6007 NULL); 6008 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) 6009 emsg(_("E284: Cannot set IC values")); 6010 XFree(attr_list); 6011 } 6012 } 6013 6014 # if defined(FEAT_GUI_X11) 6015 static char e_xim[] = N_("E285: Failed to create input context"); 6016 # endif 6017 6018 # if defined(FEAT_GUI_X11) || defined(PROTO) 6019 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(SUN_SYSTEM) 6020 # define USE_X11R6_XIM 6021 # endif 6022 6023 static int xim_real_init(Window x11_window, Display *x11_display); 6024 6025 6026 # ifdef USE_X11R6_XIM 6027 static void 6028 xim_instantiate_cb( 6029 Display *display, 6030 XPointer client_data UNUSED, 6031 XPointer call_data UNUSED) 6032 { 6033 Window x11_window; 6034 Display *x11_display; 6035 6036 # ifdef XIM_DEBUG 6037 xim_log("xim_instantiate_cb()\n"); 6038 # endif 6039 6040 gui_get_x11_windis(&x11_window, &x11_display); 6041 if (display != x11_display) 6042 return; 6043 6044 xim_real_init(x11_window, x11_display); 6045 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6046 if (xic != NULL) 6047 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6048 xim_instantiate_cb, NULL); 6049 } 6050 6051 static void 6052 xim_destroy_cb( 6053 XIM im UNUSED, 6054 XPointer client_data UNUSED, 6055 XPointer call_data UNUSED) 6056 { 6057 Window x11_window; 6058 Display *x11_display; 6059 6060 # ifdef XIM_DEBUG 6061 xim_log("xim_destroy_cb()\n"); 6062 #endif 6063 gui_get_x11_windis(&x11_window, &x11_display); 6064 6065 xic = NULL; 6066 status_area_enabled = FALSE; 6067 6068 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6069 6070 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6071 xim_instantiate_cb, NULL); 6072 } 6073 # endif 6074 6075 void 6076 xim_init(void) 6077 { 6078 Window x11_window; 6079 Display *x11_display; 6080 6081 # ifdef XIM_DEBUG 6082 xim_log("xim_init()\n"); 6083 # endif 6084 6085 gui_get_x11_windis(&x11_window, &x11_display); 6086 6087 xic = NULL; 6088 6089 if (xim_real_init(x11_window, x11_display)) 6090 return; 6091 6092 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6093 6094 # ifdef USE_X11R6_XIM 6095 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6096 xim_instantiate_cb, NULL); 6097 # endif 6098 } 6099 6100 static int 6101 xim_real_init(Window x11_window, Display *x11_display) 6102 { 6103 int i; 6104 char *p, 6105 *s, 6106 *ns, 6107 *end, 6108 tmp[1024]; 6109 # define IMLEN_MAX 40 6110 char buf[IMLEN_MAX + 7]; 6111 XIM xim = NULL; 6112 XIMStyles *xim_styles; 6113 XIMStyle this_input_style = 0; 6114 Boolean found; 6115 XPoint over_spot; 6116 XVaNestedList preedit_list, status_list; 6117 6118 input_style = 0; 6119 status_area_enabled = FALSE; 6120 6121 if (xic != NULL) 6122 return FALSE; 6123 6124 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) 6125 { 6126 strcpy(tmp, gui.rsrc_input_method); 6127 for (ns = s = tmp; ns != NULL && *s != NUL;) 6128 { 6129 s = (char *)skipwhite((char_u *)s); 6130 if (*s == NUL) 6131 break; 6132 if ((ns = end = strchr(s, ',')) == NULL) 6133 end = s + strlen(s); 6134 while (isspace(((char_u *)end)[-1])) 6135 end--; 6136 *end = NUL; 6137 6138 if (strlen(s) <= IMLEN_MAX) 6139 { 6140 strcpy(buf, "@im="); 6141 strcat(buf, s); 6142 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL 6143 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) 6144 != NULL) 6145 break; 6146 } 6147 6148 s = ns + 1; 6149 } 6150 } 6151 6152 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) 6153 xim = XOpenIM(x11_display, NULL, NULL, NULL); 6154 6155 /* This is supposed to be useful to obtain characters through 6156 * XmbLookupString() without really using a XIM. */ 6157 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL 6158 && *p != NUL) 6159 xim = XOpenIM(x11_display, NULL, NULL, NULL); 6160 6161 if (xim == NULL) 6162 { 6163 /* Only give this message when verbose is set, because too many people 6164 * got this message when they didn't want to use a XIM. */ 6165 if (p_verbose > 0) 6166 { 6167 verbose_enter(); 6168 emsg(_("E286: Failed to open input method")); 6169 verbose_leave(); 6170 } 6171 return FALSE; 6172 } 6173 6174 # ifdef USE_X11R6_XIM 6175 { 6176 XIMCallback destroy_cb; 6177 6178 destroy_cb.callback = xim_destroy_cb; 6179 destroy_cb.client_data = NULL; 6180 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) 6181 emsg(_("E287: Warning: Could not set destroy callback to IM")); 6182 } 6183 # endif 6184 6185 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) 6186 { 6187 emsg(_("E288: input method doesn't support any style")); 6188 XCloseIM(xim); 6189 return FALSE; 6190 } 6191 6192 found = False; 6193 strcpy(tmp, gui.rsrc_preedit_type_name); 6194 for (s = tmp; s && !found; ) 6195 { 6196 while (*s && isspace((unsigned char)*s)) 6197 s++; 6198 if (!*s) 6199 break; 6200 if ((ns = end = strchr(s, ',')) != 0) 6201 ns++; 6202 else 6203 end = s + strlen(s); 6204 while (isspace((unsigned char)*end)) 6205 end--; 6206 *end = '\0'; 6207 6208 if (!strcmp(s, "OverTheSpot")) 6209 this_input_style = (XIMPreeditPosition | XIMStatusArea); 6210 else if (!strcmp(s, "OffTheSpot")) 6211 this_input_style = (XIMPreeditArea | XIMStatusArea); 6212 else if (!strcmp(s, "Root")) 6213 this_input_style = (XIMPreeditNothing | XIMStatusNothing); 6214 6215 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 6216 { 6217 if (this_input_style == xim_styles->supported_styles[i]) 6218 { 6219 found = True; 6220 break; 6221 } 6222 } 6223 if (!found) 6224 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 6225 { 6226 if ((xim_styles->supported_styles[i] & this_input_style) 6227 == (this_input_style & ~XIMStatusArea)) 6228 { 6229 this_input_style &= ~XIMStatusArea; 6230 found = True; 6231 break; 6232 } 6233 } 6234 6235 s = ns; 6236 } 6237 XFree(xim_styles); 6238 6239 if (!found) 6240 { 6241 /* Only give this message when verbose is set, because too many people 6242 * got this message when they didn't want to use a XIM. */ 6243 if (p_verbose > 0) 6244 { 6245 verbose_enter(); 6246 emsg(_("E289: input method doesn't support my preedit type")); 6247 verbose_leave(); 6248 } 6249 XCloseIM(xim); 6250 return FALSE; 6251 } 6252 6253 over_spot.x = TEXT_X(gui.col); 6254 over_spot.y = TEXT_Y(gui.row); 6255 input_style = this_input_style; 6256 6257 /* A crash was reported when trying to pass gui.norm_font as XNFontSet, 6258 * thus that has been removed. Hopefully the default works... */ 6259 # ifdef FEAT_XFONTSET 6260 if (gui.fontset != NOFONTSET) 6261 { 6262 preedit_list = XVaCreateNestedList(0, 6263 XNSpotLocation, &over_spot, 6264 XNForeground, (Pixel)gui.def_norm_pixel, 6265 XNBackground, (Pixel)gui.def_back_pixel, 6266 XNFontSet, (XFontSet)gui.fontset, 6267 NULL); 6268 status_list = XVaCreateNestedList(0, 6269 XNForeground, (Pixel)gui.def_norm_pixel, 6270 XNBackground, (Pixel)gui.def_back_pixel, 6271 XNFontSet, (XFontSet)gui.fontset, 6272 NULL); 6273 } 6274 else 6275 # endif 6276 { 6277 preedit_list = XVaCreateNestedList(0, 6278 XNSpotLocation, &over_spot, 6279 XNForeground, (Pixel)gui.def_norm_pixel, 6280 XNBackground, (Pixel)gui.def_back_pixel, 6281 NULL); 6282 status_list = XVaCreateNestedList(0, 6283 XNForeground, (Pixel)gui.def_norm_pixel, 6284 XNBackground, (Pixel)gui.def_back_pixel, 6285 NULL); 6286 } 6287 6288 xic = XCreateIC(xim, 6289 XNInputStyle, input_style, 6290 XNClientWindow, x11_window, 6291 XNFocusWindow, gui.wid, 6292 XNPreeditAttributes, preedit_list, 6293 XNStatusAttributes, status_list, 6294 NULL); 6295 XFree(status_list); 6296 XFree(preedit_list); 6297 if (xic != NULL) 6298 { 6299 if (input_style & XIMStatusArea) 6300 { 6301 xim_set_status_area(); 6302 status_area_enabled = TRUE; 6303 } 6304 else 6305 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6306 } 6307 else 6308 { 6309 if (!is_not_a_term()) 6310 emsg(_(e_xim)); 6311 XCloseIM(xim); 6312 return FALSE; 6313 } 6314 6315 return TRUE; 6316 } 6317 6318 # endif /* FEAT_GUI_X11 */ 6319 6320 /* 6321 * Get IM status. When IM is on, return TRUE. Else return FALSE. 6322 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is 6323 * active, when not having focus XIM may still be active (e.g., when using a 6324 * tear-off menu item). 6325 */ 6326 int 6327 im_get_status(void) 6328 { 6329 # ifdef FEAT_EVAL 6330 if (USE_IMSTATUSFUNC) 6331 return call_imstatusfunc(); 6332 # endif 6333 return xim_has_focus; 6334 } 6335 6336 # endif /* !FEAT_GUI_GTK */ 6337 6338 # if !defined(FEAT_GUI_GTK) || defined(PROTO) 6339 /* 6340 * Set up the status area. 6341 * 6342 * This should use a separate Widget, but that seems not possible, because 6343 * preedit_area and status_area should be set to the same window as for the 6344 * text input. Unfortunately this means the status area pollutes the text 6345 * window... 6346 */ 6347 void 6348 xim_set_status_area(void) 6349 { 6350 XVaNestedList preedit_list = 0, status_list = 0, list = 0; 6351 XRectangle pre_area, status_area; 6352 6353 if (xic == NULL) 6354 return; 6355 6356 if (input_style & XIMStatusArea) 6357 { 6358 if (input_style & XIMPreeditArea) 6359 { 6360 XRectangle *needed_rect; 6361 6362 /* to get status_area width */ 6363 status_list = XVaCreateNestedList(0, XNAreaNeeded, 6364 &needed_rect, NULL); 6365 XGetICValues(xic, XNStatusAttributes, status_list, NULL); 6366 XFree(status_list); 6367 6368 status_area.width = needed_rect->width; 6369 } 6370 else 6371 status_area.width = gui.char_width * Columns; 6372 6373 status_area.x = 0; 6374 status_area.y = gui.char_height * Rows + gui.border_offset; 6375 if (gui.which_scrollbars[SBAR_BOTTOM]) 6376 status_area.y += gui.scrollbar_height; 6377 #ifdef FEAT_MENU 6378 if (gui.menu_is_active) 6379 status_area.y += gui.menu_height; 6380 #endif 6381 status_area.height = gui.char_height; 6382 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); 6383 } 6384 else 6385 { 6386 status_area.x = 0; 6387 status_area.y = gui.char_height * Rows + gui.border_offset; 6388 if (gui.which_scrollbars[SBAR_BOTTOM]) 6389 status_area.y += gui.scrollbar_height; 6390 #ifdef FEAT_MENU 6391 if (gui.menu_is_active) 6392 status_area.y += gui.menu_height; 6393 #endif 6394 status_area.width = 0; 6395 status_area.height = gui.char_height; 6396 } 6397 6398 if (input_style & XIMPreeditArea) /* off-the-spot */ 6399 { 6400 pre_area.x = status_area.x + status_area.width; 6401 pre_area.y = gui.char_height * Rows + gui.border_offset; 6402 pre_area.width = gui.char_width * Columns - pre_area.x; 6403 if (gui.which_scrollbars[SBAR_BOTTOM]) 6404 pre_area.y += gui.scrollbar_height; 6405 #ifdef FEAT_MENU 6406 if (gui.menu_is_active) 6407 pre_area.y += gui.menu_height; 6408 #endif 6409 pre_area.height = gui.char_height; 6410 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6411 } 6412 else if (input_style & XIMPreeditPosition) /* over-the-spot */ 6413 { 6414 pre_area.x = 0; 6415 pre_area.y = 0; 6416 pre_area.height = gui.char_height * Rows; 6417 pre_area.width = gui.char_width * Columns; 6418 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6419 } 6420 6421 if (preedit_list && status_list) 6422 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6423 XNStatusAttributes, status_list, NULL); 6424 else if (preedit_list) 6425 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6426 NULL); 6427 else if (status_list) 6428 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, 6429 NULL); 6430 else 6431 list = NULL; 6432 6433 if (list) 6434 { 6435 XSetICValues(xic, XNVaNestedList, list, NULL); 6436 XFree(list); 6437 } 6438 if (status_list) 6439 XFree(status_list); 6440 if (preedit_list) 6441 XFree(preedit_list); 6442 } 6443 6444 int 6445 xim_get_status_area_height(void) 6446 { 6447 if (status_area_enabled) 6448 return gui.char_height; 6449 return 0; 6450 } 6451 # endif 6452 6453 #else /* !defined(FEAT_XIM) */ 6454 6455 # ifdef IME_WITHOUT_XIM 6456 static int im_was_set_active = FALSE; 6457 6458 int 6459 im_get_status(void) 6460 { 6461 # if defined(FEAT_EVAL) 6462 if (USE_IMSTATUSFUNC) 6463 return call_imstatusfunc(); 6464 # endif 6465 return im_was_set_active; 6466 } 6467 6468 void 6469 im_set_active(int active_arg) 6470 { 6471 # if defined(FEAT_EVAL) 6472 int active = !p_imdisable && active_arg; 6473 6474 if (USE_IMACTIVATEFUNC && active != im_get_status()) 6475 { 6476 call_imactivatefunc(active); 6477 im_was_set_active = active; 6478 } 6479 # endif 6480 } 6481 6482 # ifdef FEAT_GUI 6483 void 6484 im_set_position(int row UNUSED, int col UNUSED) 6485 { 6486 } 6487 # endif 6488 # endif 6489 6490 #endif /* FEAT_XIM */ 6491 6492 6493 /* 6494 * Setup "vcp" for conversion from "from" to "to". 6495 * The names must have been made canonical with enc_canonize(). 6496 * vcp->vc_type must have been initialized to CONV_NONE. 6497 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 6498 * instead). 6499 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 6500 * Return FAIL when conversion is not supported, OK otherwise. 6501 */ 6502 int 6503 convert_setup(vimconv_T *vcp, char_u *from, char_u *to) 6504 { 6505 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 6506 } 6507 6508 /* 6509 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 6510 * "from" unicode charsets be considered utf-8. Same for "to". 6511 */ 6512 int 6513 convert_setup_ext( 6514 vimconv_T *vcp, 6515 char_u *from, 6516 int from_unicode_is_utf8, 6517 char_u *to, 6518 int to_unicode_is_utf8) 6519 { 6520 int from_prop; 6521 int to_prop; 6522 int from_is_utf8; 6523 int to_is_utf8; 6524 6525 /* Reset to no conversion. */ 6526 #ifdef USE_ICONV 6527 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 6528 iconv_close(vcp->vc_fd); 6529 #endif 6530 vcp->vc_type = CONV_NONE; 6531 vcp->vc_factor = 1; 6532 vcp->vc_fail = FALSE; 6533 6534 /* No conversion when one of the names is empty or they are equal. */ 6535 if (from == NULL || *from == NUL || to == NULL || *to == NUL 6536 || STRCMP(from, to) == 0) 6537 return OK; 6538 6539 from_prop = enc_canon_props(from); 6540 to_prop = enc_canon_props(to); 6541 if (from_unicode_is_utf8) 6542 from_is_utf8 = from_prop & ENC_UNICODE; 6543 else 6544 from_is_utf8 = from_prop == ENC_UNICODE; 6545 if (to_unicode_is_utf8) 6546 to_is_utf8 = to_prop & ENC_UNICODE; 6547 else 6548 to_is_utf8 = to_prop == ENC_UNICODE; 6549 6550 if ((from_prop & ENC_LATIN1) && to_is_utf8) 6551 { 6552 /* Internal latin1 -> utf-8 conversion. */ 6553 vcp->vc_type = CONV_TO_UTF8; 6554 vcp->vc_factor = 2; /* up to twice as long */ 6555 } 6556 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 6557 { 6558 /* Internal latin9 -> utf-8 conversion. */ 6559 vcp->vc_type = CONV_9_TO_UTF8; 6560 vcp->vc_factor = 3; /* up to three as long (euro sign) */ 6561 } 6562 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 6563 { 6564 /* Internal utf-8 -> latin1 conversion. */ 6565 vcp->vc_type = CONV_TO_LATIN1; 6566 } 6567 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 6568 { 6569 /* Internal utf-8 -> latin9 conversion. */ 6570 vcp->vc_type = CONV_TO_LATIN9; 6571 } 6572 #ifdef MSWIN 6573 /* Win32-specific codepage <-> codepage conversion without iconv. */ 6574 else if ((from_is_utf8 || encname2codepage(from) > 0) 6575 && (to_is_utf8 || encname2codepage(to) > 0)) 6576 { 6577 vcp->vc_type = CONV_CODEPAGE; 6578 vcp->vc_factor = 2; /* up to twice as long */ 6579 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 6580 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 6581 } 6582 #endif 6583 #ifdef MACOS_CONVERT 6584 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 6585 { 6586 vcp->vc_type = CONV_MAC_LATIN1; 6587 } 6588 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 6589 { 6590 vcp->vc_type = CONV_MAC_UTF8; 6591 vcp->vc_factor = 2; /* up to twice as long */ 6592 } 6593 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 6594 { 6595 vcp->vc_type = CONV_LATIN1_MAC; 6596 } 6597 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 6598 { 6599 vcp->vc_type = CONV_UTF8_MAC; 6600 } 6601 #endif 6602 #ifdef USE_ICONV 6603 else 6604 { 6605 /* Use iconv() for conversion. */ 6606 vcp->vc_fd = (iconv_t)my_iconv_open( 6607 to_is_utf8 ? (char_u *)"utf-8" : to, 6608 from_is_utf8 ? (char_u *)"utf-8" : from); 6609 if (vcp->vc_fd != (iconv_t)-1) 6610 { 6611 vcp->vc_type = CONV_ICONV; 6612 vcp->vc_factor = 4; /* could be longer too... */ 6613 } 6614 } 6615 #endif 6616 if (vcp->vc_type == CONV_NONE) 6617 return FAIL; 6618 6619 return OK; 6620 } 6621 6622 #if defined(FEAT_GUI) || defined(AMIGA) || defined(MSWIN) \ 6623 || defined(PROTO) 6624 /* 6625 * Do conversion on typed input characters in-place. 6626 * The input and output are not NUL terminated! 6627 * Returns the length after conversion. 6628 */ 6629 int 6630 convert_input(char_u *ptr, int len, int maxlen) 6631 { 6632 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 6633 } 6634 #endif 6635 6636 /* 6637 * Like convert_input(), but when there is an incomplete byte sequence at the 6638 * end return that as an allocated string in "restp" and set "*restlenp" to 6639 * the length. If "restp" is NULL it is not used. 6640 */ 6641 int 6642 convert_input_safe( 6643 char_u *ptr, 6644 int len, 6645 int maxlen, 6646 char_u **restp, 6647 int *restlenp) 6648 { 6649 char_u *d; 6650 int dlen = len; 6651 int unconvertlen = 0; 6652 6653 d = string_convert_ext(&input_conv, ptr, &dlen, 6654 restp == NULL ? NULL : &unconvertlen); 6655 if (d != NULL) 6656 { 6657 if (dlen <= maxlen) 6658 { 6659 if (unconvertlen > 0) 6660 { 6661 /* Move the unconverted characters to allocated memory. */ 6662 *restp = alloc(unconvertlen); 6663 if (*restp != NULL) 6664 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 6665 *restlenp = unconvertlen; 6666 } 6667 mch_memmove(ptr, d, dlen); 6668 } 6669 else 6670 /* result is too long, keep the unconverted text (the caller must 6671 * have done something wrong!) */ 6672 dlen = len; 6673 vim_free(d); 6674 } 6675 return dlen; 6676 } 6677 6678 /* 6679 * Convert text "ptr[*lenp]" according to "vcp". 6680 * Returns the result in allocated memory and sets "*lenp". 6681 * When "lenp" is NULL, use NUL terminated strings. 6682 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 6683 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 6684 */ 6685 char_u * 6686 string_convert( 6687 vimconv_T *vcp, 6688 char_u *ptr, 6689 int *lenp) 6690 { 6691 return string_convert_ext(vcp, ptr, lenp, NULL); 6692 } 6693 6694 /* 6695 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 6696 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 6697 * set to the number of remaining bytes. 6698 */ 6699 char_u * 6700 string_convert_ext( 6701 vimconv_T *vcp, 6702 char_u *ptr, 6703 int *lenp, 6704 int *unconvlenp) 6705 { 6706 char_u *retval = NULL; 6707 char_u *d; 6708 int len; 6709 int i; 6710 int l; 6711 int c; 6712 6713 if (lenp == NULL) 6714 len = (int)STRLEN(ptr); 6715 else 6716 len = *lenp; 6717 if (len == 0) 6718 return vim_strsave((char_u *)""); 6719 6720 switch (vcp->vc_type) 6721 { 6722 case CONV_TO_UTF8: /* latin1 to utf-8 conversion */ 6723 retval = alloc(len * 2 + 1); 6724 if (retval == NULL) 6725 break; 6726 d = retval; 6727 for (i = 0; i < len; ++i) 6728 { 6729 c = ptr[i]; 6730 if (c < 0x80) 6731 *d++ = c; 6732 else 6733 { 6734 *d++ = 0xc0 + ((unsigned)c >> 6); 6735 *d++ = 0x80 + (c & 0x3f); 6736 } 6737 } 6738 *d = NUL; 6739 if (lenp != NULL) 6740 *lenp = (int)(d - retval); 6741 break; 6742 6743 case CONV_9_TO_UTF8: /* latin9 to utf-8 conversion */ 6744 retval = alloc(len * 3 + 1); 6745 if (retval == NULL) 6746 break; 6747 d = retval; 6748 for (i = 0; i < len; ++i) 6749 { 6750 c = ptr[i]; 6751 switch (c) 6752 { 6753 case 0xa4: c = 0x20ac; break; /* euro */ 6754 case 0xa6: c = 0x0160; break; /* S hat */ 6755 case 0xa8: c = 0x0161; break; /* S -hat */ 6756 case 0xb4: c = 0x017d; break; /* Z hat */ 6757 case 0xb8: c = 0x017e; break; /* Z -hat */ 6758 case 0xbc: c = 0x0152; break; /* OE */ 6759 case 0xbd: c = 0x0153; break; /* oe */ 6760 case 0xbe: c = 0x0178; break; /* Y */ 6761 } 6762 d += utf_char2bytes(c, d); 6763 } 6764 *d = NUL; 6765 if (lenp != NULL) 6766 *lenp = (int)(d - retval); 6767 break; 6768 6769 case CONV_TO_LATIN1: /* utf-8 to latin1 conversion */ 6770 case CONV_TO_LATIN9: /* utf-8 to latin9 conversion */ 6771 retval = alloc(len + 1); 6772 if (retval == NULL) 6773 break; 6774 d = retval; 6775 for (i = 0; i < len; ++i) 6776 { 6777 l = utf_ptr2len_len(ptr + i, len - i); 6778 if (l == 0) 6779 *d++ = NUL; 6780 else if (l == 1) 6781 { 6782 int l_w = utf8len_tab_zero[ptr[i]]; 6783 6784 if (l_w == 0) 6785 { 6786 /* Illegal utf-8 byte cannot be converted */ 6787 vim_free(retval); 6788 return NULL; 6789 } 6790 if (unconvlenp != NULL && l_w > len - i) 6791 { 6792 /* Incomplete sequence at the end. */ 6793 *unconvlenp = len - i; 6794 break; 6795 } 6796 *d++ = ptr[i]; 6797 } 6798 else 6799 { 6800 c = utf_ptr2char(ptr + i); 6801 if (vcp->vc_type == CONV_TO_LATIN9) 6802 switch (c) 6803 { 6804 case 0x20ac: c = 0xa4; break; /* euro */ 6805 case 0x0160: c = 0xa6; break; /* S hat */ 6806 case 0x0161: c = 0xa8; break; /* S -hat */ 6807 case 0x017d: c = 0xb4; break; /* Z hat */ 6808 case 0x017e: c = 0xb8; break; /* Z -hat */ 6809 case 0x0152: c = 0xbc; break; /* OE */ 6810 case 0x0153: c = 0xbd; break; /* oe */ 6811 case 0x0178: c = 0xbe; break; /* Y */ 6812 case 0xa4: 6813 case 0xa6: 6814 case 0xa8: 6815 case 0xb4: 6816 case 0xb8: 6817 case 0xbc: 6818 case 0xbd: 6819 case 0xbe: c = 0x100; break; /* not in latin9 */ 6820 } 6821 if (!utf_iscomposing(c)) /* skip composing chars */ 6822 { 6823 if (c < 0x100) 6824 *d++ = c; 6825 else if (vcp->vc_fail) 6826 { 6827 vim_free(retval); 6828 return NULL; 6829 } 6830 else 6831 { 6832 *d++ = 0xbf; 6833 if (utf_char2cells(c) > 1) 6834 *d++ = '?'; 6835 } 6836 } 6837 i += l - 1; 6838 } 6839 } 6840 *d = NUL; 6841 if (lenp != NULL) 6842 *lenp = (int)(d - retval); 6843 break; 6844 6845 # ifdef MACOS_CONVERT 6846 case CONV_MAC_LATIN1: 6847 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6848 'm', 'l', unconvlenp); 6849 break; 6850 6851 case CONV_LATIN1_MAC: 6852 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6853 'l', 'm', unconvlenp); 6854 break; 6855 6856 case CONV_MAC_UTF8: 6857 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6858 'm', 'u', unconvlenp); 6859 break; 6860 6861 case CONV_UTF8_MAC: 6862 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6863 'u', 'm', unconvlenp); 6864 break; 6865 # endif 6866 6867 # ifdef USE_ICONV 6868 case CONV_ICONV: /* conversion with output_conv.vc_fd */ 6869 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 6870 break; 6871 # endif 6872 # ifdef MSWIN 6873 case CONV_CODEPAGE: /* codepage -> codepage */ 6874 { 6875 int retlen; 6876 int tmp_len; 6877 short_u *tmp; 6878 6879 /* 1. codepage/UTF-8 -> ucs-2. */ 6880 if (vcp->vc_cpfrom == 0) 6881 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 6882 else 6883 { 6884 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 6885 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 6886 (char *)ptr, len, 0, 0); 6887 if (tmp_len == 0 6888 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 6889 { 6890 if (lenp != NULL) 6891 *lenp = 0; 6892 if (unconvlenp != NULL) 6893 *unconvlenp = len; 6894 retval = alloc(1); 6895 if (retval) 6896 retval[0] = NUL; 6897 return retval; 6898 } 6899 } 6900 tmp = (short_u *)alloc(sizeof(short_u) * tmp_len); 6901 if (tmp == NULL) 6902 break; 6903 if (vcp->vc_cpfrom == 0) 6904 utf8_to_utf16(ptr, len, tmp, unconvlenp); 6905 else 6906 MultiByteToWideChar(vcp->vc_cpfrom, 0, 6907 (char *)ptr, len, tmp, tmp_len); 6908 6909 /* 2. ucs-2 -> codepage/UTF-8. */ 6910 if (vcp->vc_cpto == 0) 6911 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 6912 else 6913 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 6914 tmp, tmp_len, 0, 0, 0, 0); 6915 retval = alloc(retlen + 1); 6916 if (retval != NULL) 6917 { 6918 if (vcp->vc_cpto == 0) 6919 utf16_to_utf8(tmp, tmp_len, retval); 6920 else 6921 WideCharToMultiByte(vcp->vc_cpto, 0, 6922 tmp, tmp_len, 6923 (char *)retval, retlen, 0, 0); 6924 retval[retlen] = NUL; 6925 if (lenp != NULL) 6926 *lenp = retlen; 6927 } 6928 vim_free(tmp); 6929 break; 6930 } 6931 # endif 6932 } 6933 6934 return retval; 6935 } 6936