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 static int dbcs_head_off(char_u *base, char_u *p); 148 149 /* 150 * Lookup table to quickly get the length in bytes of a UTF-8 character from 151 * the first byte of a UTF-8 string. 152 * Bytes which are illegal when used as the first byte have a 1. 153 * The NUL byte has length 1. 154 */ 155 static char utf8len_tab[256] = 156 { 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 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 163 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, 164 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, 165 }; 166 167 /* 168 * Like utf8len_tab above, but using a zero for illegal lead bytes. 169 */ 170 static char utf8len_tab_zero[256] = 171 { 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 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 176 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 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 178 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, 179 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, 180 }; 181 182 /* 183 * XIM often causes trouble. Define XIM_DEBUG to get a log of XIM callbacks 184 * in the "xim.log" file. 185 */ 186 // #define XIM_DEBUG 187 #ifdef XIM_DEBUG 188 static void 189 xim_log(char *s, ...) 190 { 191 va_list arglist; 192 static FILE *fd = NULL; 193 194 if (fd == (FILE *)-1) 195 return; 196 if (fd == NULL) 197 { 198 fd = mch_fopen("xim.log", "w"); 199 if (fd == NULL) 200 { 201 emsg("Cannot open xim.log"); 202 fd = (FILE *)-1; 203 return; 204 } 205 } 206 207 va_start(arglist, s); 208 vfprintf(fd, s, arglist); 209 va_end(arglist); 210 } 211 #endif 212 213 214 /* 215 * Canonical encoding names and their properties. 216 * "iso-8859-n" is handled by enc_canonize() directly. 217 */ 218 static struct 219 { char *name; int prop; int codepage;} 220 enc_canon_table[] = 221 { 222 #define IDX_LATIN_1 0 223 {"latin1", ENC_8BIT + ENC_LATIN1, 1252}, 224 #define IDX_ISO_2 1 225 {"iso-8859-2", ENC_8BIT, 0}, 226 #define IDX_ISO_3 2 227 {"iso-8859-3", ENC_8BIT, 0}, 228 #define IDX_ISO_4 3 229 {"iso-8859-4", ENC_8BIT, 0}, 230 #define IDX_ISO_5 4 231 {"iso-8859-5", ENC_8BIT, 0}, 232 #define IDX_ISO_6 5 233 {"iso-8859-6", ENC_8BIT, 0}, 234 #define IDX_ISO_7 6 235 {"iso-8859-7", ENC_8BIT, 0}, 236 #define IDX_ISO_8 7 237 {"iso-8859-8", ENC_8BIT, 0}, 238 #define IDX_ISO_9 8 239 {"iso-8859-9", ENC_8BIT, 0}, 240 #define IDX_ISO_10 9 241 {"iso-8859-10", ENC_8BIT, 0}, 242 #define IDX_ISO_11 10 243 {"iso-8859-11", ENC_8BIT, 0}, 244 #define IDX_ISO_13 11 245 {"iso-8859-13", ENC_8BIT, 0}, 246 #define IDX_ISO_14 12 247 {"iso-8859-14", ENC_8BIT, 0}, 248 #define IDX_ISO_15 13 249 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0}, 250 #define IDX_KOI8_R 14 251 {"koi8-r", ENC_8BIT, 0}, 252 #define IDX_KOI8_U 15 253 {"koi8-u", ENC_8BIT, 0}, 254 #define IDX_UTF8 16 255 {"utf-8", ENC_UNICODE, 0}, 256 #define IDX_UCS2 17 257 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0}, 258 #define IDX_UCS2LE 18 259 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0}, 260 #define IDX_UTF16 19 261 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0}, 262 #define IDX_UTF16LE 20 263 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0}, 264 #define IDX_UCS4 21 265 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0}, 266 #define IDX_UCS4LE 22 267 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0}, 268 269 // For debugging DBCS encoding on Unix. 270 #define IDX_DEBUG 23 271 {"debug", ENC_DBCS, DBCS_DEBUG}, 272 #define IDX_EUC_JP 24 273 {"euc-jp", ENC_DBCS, DBCS_JPNU}, 274 #define IDX_SJIS 25 275 {"sjis", ENC_DBCS, DBCS_JPN}, 276 #define IDX_EUC_KR 26 277 {"euc-kr", ENC_DBCS, DBCS_KORU}, 278 #define IDX_EUC_CN 27 279 {"euc-cn", ENC_DBCS, DBCS_CHSU}, 280 #define IDX_EUC_TW 28 281 {"euc-tw", ENC_DBCS, DBCS_CHTU}, 282 #define IDX_BIG5 29 283 {"big5", ENC_DBCS, DBCS_CHT}, 284 285 // MS-DOS and MS-Windows codepages are included here, so that they can be 286 // used on Unix too. Most of them are similar to ISO-8859 encodings, but 287 // not exactly the same. 288 #define IDX_CP437 30 289 {"cp437", ENC_8BIT, 437}, // like iso-8859-1 290 #define IDX_CP737 31 291 {"cp737", ENC_8BIT, 737}, // like iso-8859-7 292 #define IDX_CP775 32 293 {"cp775", ENC_8BIT, 775}, // Baltic 294 #define IDX_CP850 33 295 {"cp850", ENC_8BIT, 850}, // like iso-8859-4 296 #define IDX_CP852 34 297 {"cp852", ENC_8BIT, 852}, // like iso-8859-1 298 #define IDX_CP855 35 299 {"cp855", ENC_8BIT, 855}, // like iso-8859-2 300 #define IDX_CP857 36 301 {"cp857", ENC_8BIT, 857}, // like iso-8859-5 302 #define IDX_CP860 37 303 {"cp860", ENC_8BIT, 860}, // like iso-8859-9 304 #define IDX_CP861 38 305 {"cp861", ENC_8BIT, 861}, // like iso-8859-1 306 #define IDX_CP862 39 307 {"cp862", ENC_8BIT, 862}, // like iso-8859-1 308 #define IDX_CP863 40 309 {"cp863", ENC_8BIT, 863}, // like iso-8859-8 310 #define IDX_CP865 41 311 {"cp865", ENC_8BIT, 865}, // like iso-8859-1 312 #define IDX_CP866 42 313 {"cp866", ENC_8BIT, 866}, // like iso-8859-5 314 #define IDX_CP869 43 315 {"cp869", ENC_8BIT, 869}, // like iso-8859-7 316 #define IDX_CP874 44 317 {"cp874", ENC_8BIT, 874}, // Thai 318 #define IDX_CP932 45 319 {"cp932", ENC_DBCS, DBCS_JPN}, 320 #define IDX_CP936 46 321 {"cp936", ENC_DBCS, DBCS_CHS}, 322 #define IDX_CP949 47 323 {"cp949", ENC_DBCS, DBCS_KOR}, 324 #define IDX_CP950 48 325 {"cp950", ENC_DBCS, DBCS_CHT}, 326 #define IDX_CP1250 49 327 {"cp1250", ENC_8BIT, 1250}, // Czech, Polish, etc. 328 #define IDX_CP1251 50 329 {"cp1251", ENC_8BIT, 1251}, // Cyrillic 330 // cp1252 is considered to be equal to latin1 331 #define IDX_CP1253 51 332 {"cp1253", ENC_8BIT, 1253}, // Greek 333 #define IDX_CP1254 52 334 {"cp1254", ENC_8BIT, 1254}, // Turkish 335 #define IDX_CP1255 53 336 {"cp1255", ENC_8BIT, 1255}, // Hebrew 337 #define IDX_CP1256 54 338 {"cp1256", ENC_8BIT, 1256}, // Arabic 339 #define IDX_CP1257 55 340 {"cp1257", ENC_8BIT, 1257}, // Baltic 341 #define IDX_CP1258 56 342 {"cp1258", ENC_8BIT, 1258}, // Vietnamese 343 344 #define IDX_MACROMAN 57 345 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, // Mac OS 346 #define IDX_DECMCS 58 347 {"dec-mcs", ENC_8BIT, 0}, // DEC MCS 348 #define IDX_HPROMAN8 59 349 {"hp-roman8", ENC_8BIT, 0}, // HP Roman8 350 #define IDX_COUNT 60 351 }; 352 353 /* 354 * Aliases for encoding names. 355 */ 356 static struct 357 { char *name; int canon;} 358 enc_alias_table[] = 359 { 360 {"ansi", IDX_LATIN_1}, 361 {"iso-8859-1", IDX_LATIN_1}, 362 {"latin2", IDX_ISO_2}, 363 {"latin3", IDX_ISO_3}, 364 {"latin4", IDX_ISO_4}, 365 {"cyrillic", IDX_ISO_5}, 366 {"arabic", IDX_ISO_6}, 367 {"greek", IDX_ISO_7}, 368 #ifdef MSWIN 369 {"hebrew", IDX_CP1255}, 370 #else 371 {"hebrew", IDX_ISO_8}, 372 #endif 373 {"latin5", IDX_ISO_9}, 374 {"turkish", IDX_ISO_9}, // ? 375 {"latin6", IDX_ISO_10}, 376 {"nordic", IDX_ISO_10}, // ? 377 {"thai", IDX_ISO_11}, // ? 378 {"latin7", IDX_ISO_13}, 379 {"latin8", IDX_ISO_14}, 380 {"latin9", IDX_ISO_15}, 381 {"utf8", IDX_UTF8}, 382 {"unicode", IDX_UCS2}, 383 {"ucs2", IDX_UCS2}, 384 {"ucs2be", IDX_UCS2}, 385 {"ucs-2be", IDX_UCS2}, 386 {"ucs2le", IDX_UCS2LE}, 387 {"utf16", IDX_UTF16}, 388 {"utf16be", IDX_UTF16}, 389 {"utf-16be", IDX_UTF16}, 390 {"utf16le", IDX_UTF16LE}, 391 {"ucs4", IDX_UCS4}, 392 {"ucs4be", IDX_UCS4}, 393 {"ucs-4be", IDX_UCS4}, 394 {"ucs4le", IDX_UCS4LE}, 395 {"utf32", IDX_UCS4}, 396 {"utf-32", IDX_UCS4}, 397 {"utf32be", IDX_UCS4}, 398 {"utf-32be", IDX_UCS4}, 399 {"utf32le", IDX_UCS4LE}, 400 {"utf-32le", IDX_UCS4LE}, 401 {"932", IDX_CP932}, 402 {"949", IDX_CP949}, 403 {"936", IDX_CP936}, 404 {"gbk", IDX_CP936}, 405 {"950", IDX_CP950}, 406 {"eucjp", IDX_EUC_JP}, 407 {"unix-jis", IDX_EUC_JP}, 408 {"ujis", IDX_EUC_JP}, 409 {"shift-jis", IDX_SJIS}, 410 {"pck", IDX_SJIS}, // Sun: PCK 411 {"euckr", IDX_EUC_KR}, 412 {"5601", IDX_EUC_KR}, // Sun: KS C 5601 413 {"euccn", IDX_EUC_CN}, 414 {"gb2312", IDX_EUC_CN}, 415 {"euctw", IDX_EUC_TW}, 416 #if defined(MSWIN) || defined(WIN32UNIX) || defined(MACOS_X) 417 {"japan", IDX_CP932}, 418 {"korea", IDX_CP949}, 419 {"prc", IDX_CP936}, 420 {"chinese", IDX_CP936}, 421 {"taiwan", IDX_CP950}, 422 {"big5", IDX_CP950}, 423 #else 424 {"japan", IDX_EUC_JP}, 425 {"korea", IDX_EUC_KR}, 426 {"prc", IDX_EUC_CN}, 427 {"chinese", IDX_EUC_CN}, 428 {"taiwan", IDX_EUC_TW}, 429 {"cp950", IDX_BIG5}, 430 {"950", IDX_BIG5}, 431 #endif 432 {"mac", IDX_MACROMAN}, 433 {"mac-roman", IDX_MACROMAN}, 434 {NULL, 0} 435 }; 436 437 #ifndef CP_UTF8 438 # define CP_UTF8 65001 // magic number from winnls.h 439 #endif 440 441 /* 442 * Find encoding "name" in the list of canonical encoding names. 443 * Returns -1 if not found. 444 */ 445 static int 446 enc_canon_search(char_u *name) 447 { 448 int i; 449 450 for (i = 0; i < IDX_COUNT; ++i) 451 if (STRCMP(name, enc_canon_table[i].name) == 0) 452 return i; 453 return -1; 454 } 455 456 457 /* 458 * Find canonical encoding "name" in the list and return its properties. 459 * Returns 0 if not found. 460 */ 461 int 462 enc_canon_props(char_u *name) 463 { 464 int i; 465 466 i = enc_canon_search(name); 467 if (i >= 0) 468 return enc_canon_table[i].prop; 469 #ifdef MSWIN 470 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2])) 471 { 472 CPINFO cpinfo; 473 474 // Get info on this codepage to find out what it is. 475 if (GetCPInfo(atoi((char *)name + 2), &cpinfo) != 0) 476 { 477 if (cpinfo.MaxCharSize == 1) // some single-byte encoding 478 return ENC_8BIT; 479 if (cpinfo.MaxCharSize == 2 480 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 481 // must be a DBCS encoding 482 return ENC_DBCS; 483 } 484 return 0; 485 } 486 #endif 487 if (STRNCMP(name, "2byte-", 6) == 0) 488 return ENC_DBCS; 489 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0) 490 return ENC_8BIT; 491 return 0; 492 } 493 494 /* 495 * Set up for using multi-byte characters. 496 * Called in three cases: 497 * - by main() to initialize (p_enc == NULL) 498 * - by set_init_1() after 'encoding' was set to its default. 499 * - by do_set() when 'encoding' has been set. 500 * p_enc must have been passed through enc_canonize() already. 501 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags. 502 * Fills mb_bytelen_tab[] and returns NULL when there are no problems. 503 * When there is something wrong: Returns an error message and doesn't change 504 * anything. 505 */ 506 char * 507 mb_init(void) 508 { 509 int i; 510 int idx; 511 int n; 512 int enc_dbcs_new = 0; 513 #if defined(USE_ICONV) && !defined(MSWIN) && !defined(WIN32UNIX) \ 514 && !defined(MACOS_CONVERT) 515 # define LEN_FROM_CONV 516 vimconv_T vimconv; 517 char_u *p; 518 #endif 519 520 if (p_enc == NULL) 521 { 522 // Just starting up: set the whole table to one's. 523 for (i = 0; i < 256; ++i) 524 mb_bytelen_tab[i] = 1; 525 input_conv.vc_type = CONV_NONE; 526 input_conv.vc_factor = 1; 527 output_conv.vc_type = CONV_NONE; 528 return NULL; 529 } 530 531 #ifdef MSWIN 532 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2])) 533 { 534 CPINFO cpinfo; 535 536 // Get info on this codepage to find out what it is. 537 if (GetCPInfo(atoi((char *)p_enc + 2), &cpinfo) != 0) 538 { 539 if (cpinfo.MaxCharSize == 1) 540 { 541 // some single-byte encoding 542 enc_unicode = 0; 543 enc_utf8 = FALSE; 544 } 545 else if (cpinfo.MaxCharSize == 2 546 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 547 { 548 // must be a DBCS encoding, check below 549 enc_dbcs_new = atoi((char *)p_enc + 2); 550 } 551 else 552 goto codepage_invalid; 553 } 554 else if (GetLastError() == ERROR_INVALID_PARAMETER) 555 { 556 codepage_invalid: 557 return N_("E543: Not a valid codepage"); 558 } 559 } 560 #endif 561 else if (STRNCMP(p_enc, "8bit-", 5) == 0 562 || STRNCMP(p_enc, "iso-8859-", 9) == 0) 563 { 564 // Accept any "8bit-" or "iso-8859-" name. 565 enc_unicode = 0; 566 enc_utf8 = FALSE; 567 } 568 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 569 { 570 #ifdef MSWIN 571 // Windows: accept only valid codepage numbers, check below. 572 if (p_enc[6] != 'c' || p_enc[7] != 'p' 573 || (enc_dbcs_new = atoi((char *)p_enc + 8)) == 0) 574 return e_invarg; 575 #else 576 // Unix: accept any "2byte-" name, assume current locale. 577 enc_dbcs_new = DBCS_2BYTE; 578 #endif 579 } 580 else if ((idx = enc_canon_search(p_enc)) >= 0) 581 { 582 i = enc_canon_table[idx].prop; 583 if (i & ENC_UNICODE) 584 { 585 // Unicode 586 enc_utf8 = TRUE; 587 if (i & (ENC_2BYTE | ENC_2WORD)) 588 enc_unicode = 2; 589 else if (i & ENC_4BYTE) 590 enc_unicode = 4; 591 else 592 enc_unicode = 0; 593 } 594 else if (i & ENC_DBCS) 595 { 596 // 2byte, handle below 597 enc_dbcs_new = enc_canon_table[idx].codepage; 598 } 599 else 600 { 601 // Must be 8-bit. 602 enc_unicode = 0; 603 enc_utf8 = FALSE; 604 } 605 } 606 else // Don't know what encoding this is, reject it. 607 return e_invarg; 608 609 if (enc_dbcs_new != 0) 610 { 611 #ifdef MSWIN 612 // Check if the DBCS code page is OK. 613 if (!IsValidCodePage(enc_dbcs_new)) 614 goto codepage_invalid; 615 #endif 616 enc_unicode = 0; 617 enc_utf8 = FALSE; 618 } 619 enc_dbcs = enc_dbcs_new; 620 has_mbyte = (enc_dbcs != 0 || enc_utf8); 621 622 #if defined(MSWIN) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 623 enc_codepage = encname2codepage(p_enc); 624 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0); 625 #endif 626 627 // Detect an encoding that uses latin1 characters. 628 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0 629 || STRCMP(p_enc, "iso-8859-15") == 0); 630 631 /* 632 * Set the function pointers. 633 */ 634 if (enc_utf8) 635 { 636 mb_ptr2len = utfc_ptr2len; 637 mb_ptr2len_len = utfc_ptr2len_len; 638 mb_char2len = utf_char2len; 639 mb_char2bytes = utf_char2bytes; 640 mb_ptr2cells = utf_ptr2cells; 641 mb_ptr2cells_len = utf_ptr2cells_len; 642 mb_char2cells = utf_char2cells; 643 mb_off2cells = utf_off2cells; 644 mb_ptr2char = utf_ptr2char; 645 mb_head_off = utf_head_off; 646 } 647 else if (enc_dbcs != 0) 648 { 649 mb_ptr2len = dbcs_ptr2len; 650 mb_ptr2len_len = dbcs_ptr2len_len; 651 mb_char2len = dbcs_char2len; 652 mb_char2bytes = dbcs_char2bytes; 653 mb_ptr2cells = dbcs_ptr2cells; 654 mb_ptr2cells_len = dbcs_ptr2cells_len; 655 mb_char2cells = dbcs_char2cells; 656 mb_off2cells = dbcs_off2cells; 657 mb_ptr2char = dbcs_ptr2char; 658 mb_head_off = dbcs_head_off; 659 } 660 else 661 { 662 mb_ptr2len = latin_ptr2len; 663 mb_ptr2len_len = latin_ptr2len_len; 664 mb_char2len = latin_char2len; 665 mb_char2bytes = latin_char2bytes; 666 mb_ptr2cells = latin_ptr2cells; 667 mb_ptr2cells_len = latin_ptr2cells_len; 668 mb_char2cells = latin_char2cells; 669 mb_off2cells = latin_off2cells; 670 mb_ptr2char = latin_ptr2char; 671 mb_head_off = latin_head_off; 672 } 673 674 /* 675 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN(). 676 */ 677 #ifdef LEN_FROM_CONV 678 // When 'encoding' is different from the current locale mblen() won't 679 // work. Use conversion to "utf-8" instead. 680 vimconv.vc_type = CONV_NONE; 681 if (enc_dbcs) 682 { 683 p = enc_locale(); 684 if (p == NULL || STRCMP(p, p_enc) != 0) 685 { 686 convert_setup(&vimconv, p_enc, (char_u *)"utf-8"); 687 vimconv.vc_fail = TRUE; 688 } 689 vim_free(p); 690 } 691 #endif 692 693 for (i = 0; i < 256; ++i) 694 { 695 // Our own function to reliably check the length of UTF-8 characters, 696 // independent of mblen(). 697 if (enc_utf8) 698 n = utf8len_tab[i]; 699 else if (enc_dbcs == 0) 700 n = 1; 701 else 702 { 703 #if defined(MSWIN) || defined(WIN32UNIX) 704 // enc_dbcs is set by setting 'fileencoding'. It becomes a Windows 705 // CodePage identifier, which we can pass directly in to Windows 706 // API 707 n = IsDBCSLeadByteEx(enc_dbcs, (WINBYTE)i) ? 2 : 1; 708 #else 709 # if defined(__amigaos4__) || defined(__ANDROID__) || \ 710 !(defined(HAVE_MBLEN) || defined(X_LOCALE)) 711 /* 712 * if mblen() is not available, character which MSB is turned on 713 * are treated as leading byte character. (note : This assumption 714 * is not always true.) 715 */ 716 n = (i & 0x80) ? 2 : 1; 717 # else 718 char buf[MB_MAXBYTES + 1]; 719 720 if (i == NUL) // just in case mblen() can't handle "" 721 n = 1; 722 else 723 { 724 buf[0] = i; 725 buf[1] = 0; 726 # ifdef LEN_FROM_CONV 727 if (vimconv.vc_type != CONV_NONE) 728 { 729 /* 730 * string_convert() should fail when converting the first 731 * byte of a double-byte character. 732 */ 733 p = string_convert(&vimconv, (char_u *)buf, NULL); 734 if (p != NULL) 735 { 736 vim_free(p); 737 n = 1; 738 } 739 else 740 n = 2; 741 } 742 else 743 # endif 744 { 745 /* 746 * mblen() should return -1 for invalid (means the leading 747 * multibyte) character. However there are some platforms 748 * where mblen() returns 0 for invalid character. 749 * Therefore, following condition includes 0. 750 */ 751 vim_ignored = mblen(NULL, 0); // First reset the state. 752 if (mblen(buf, (size_t)1) <= 0) 753 n = 2; 754 else 755 n = 1; 756 } 757 } 758 # endif 759 #endif 760 } 761 762 mb_bytelen_tab[i] = n; 763 } 764 765 #ifdef LEN_FROM_CONV 766 convert_setup(&vimconv, NULL, NULL); 767 #endif 768 769 // The cell width depends on the type of multi-byte characters. 770 (void)init_chartab(); 771 772 // When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] 773 screenalloc(FALSE); 774 775 // When using Unicode, set default for 'fileencodings'. 776 if (enc_utf8 && !option_was_set((char_u *)"fencs")) 777 set_string_option_direct((char_u *)"fencs", -1, 778 (char_u *)"ucs-bom,utf-8,default,latin1", OPT_FREE, 0); 779 780 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT) 781 // GNU gettext 0.10.37 supports this feature: set the codeset used for 782 // translated messages independently from the current locale. 783 (void)bind_textdomain_codeset(VIMPACKAGE, 784 enc_utf8 ? "utf-8" : (char *)p_enc); 785 #endif 786 787 #ifdef MSWIN 788 // When changing 'encoding' while starting up, then convert the command 789 // line arguments from the active codepage to 'encoding'. 790 if (starting != 0) 791 fix_arg_enc(); 792 #endif 793 794 // Fire an autocommand to let people do custom font setup. This must be 795 // after Vim has been setup for the new encoding. 796 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); 797 798 #ifdef FEAT_SPELL 799 // Need to reload spell dictionaries 800 spell_reload(); 801 #endif 802 803 return NULL; 804 } 805 806 /* 807 * Return the size of the BOM for the current buffer: 808 * 0 - no BOM 809 * 2 - UCS-2 or UTF-16 BOM 810 * 4 - UCS-4 BOM 811 * 3 - UTF-8 BOM 812 */ 813 int 814 bomb_size(void) 815 { 816 int n = 0; 817 818 if (curbuf->b_p_bomb && !curbuf->b_p_bin) 819 { 820 if (*curbuf->b_p_fenc == NUL) 821 { 822 if (enc_utf8) 823 { 824 if (enc_unicode != 0) 825 n = enc_unicode; 826 else 827 n = 3; 828 } 829 } 830 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) 831 n = 3; 832 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 833 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) 834 n = 2; 835 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) 836 n = 4; 837 } 838 return n; 839 } 840 841 #if defined(FEAT_QUICKFIX) || defined(PROTO) 842 /* 843 * Remove all BOM from "s" by moving remaining text. 844 */ 845 void 846 remove_bom(char_u *s) 847 { 848 if (enc_utf8) 849 { 850 char_u *p = s; 851 852 while ((p = vim_strbyte(p, 0xef)) != NULL) 853 { 854 if (p[1] == 0xbb && p[2] == 0xbf) 855 STRMOVE(p, p + 3); 856 else 857 ++p; 858 } 859 } 860 } 861 #endif 862 863 /* 864 * Get class of pointer: 865 * 0 for blank or NUL 866 * 1 for punctuation 867 * 2 for an (ASCII) word character 868 * >2 for other word characters 869 */ 870 int 871 mb_get_class(char_u *p) 872 { 873 return mb_get_class_buf(p, curbuf); 874 } 875 876 int 877 mb_get_class_buf(char_u *p, buf_T *buf) 878 { 879 if (MB_BYTE2LEN(p[0]) == 1) 880 { 881 if (p[0] == NUL || VIM_ISWHITE(p[0])) 882 return 0; 883 if (vim_iswordc_buf(p[0], buf)) 884 return 2; 885 return 1; 886 } 887 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) 888 return dbcs_class(p[0], p[1]); 889 if (enc_utf8) 890 return utf_class_buf(utf_ptr2char(p), buf); 891 return 0; 892 } 893 894 /* 895 * Get class of a double-byte character. This always returns 3 or bigger. 896 * TODO: Should return 1 for punctuation. 897 */ 898 int 899 dbcs_class(unsigned lead, unsigned trail) 900 { 901 switch (enc_dbcs) 902 { 903 // please add classify routine for your language in here 904 905 case DBCS_JPNU: // ? 906 case DBCS_JPN: 907 { 908 // JIS code classification 909 unsigned char lb = lead; 910 unsigned char tb = trail; 911 912 // convert process code to JIS 913 # if defined(MSWIN) || defined(WIN32UNIX) || defined(MACOS_X) 914 // process code is SJIS 915 if (lb <= 0x9f) 916 lb = (lb - 0x81) * 2 + 0x21; 917 else 918 lb = (lb - 0xc1) * 2 + 0x21; 919 if (tb <= 0x7e) 920 tb -= 0x1f; 921 else if (tb <= 0x9e) 922 tb -= 0x20; 923 else 924 { 925 tb -= 0x7e; 926 lb += 1; 927 } 928 # else 929 /* 930 * XXX: Code page identification can not use with all 931 * system! So, some other encoding information 932 * will be needed. 933 * In japanese: SJIS,EUC,UNICODE,(JIS) 934 * Note that JIS-code system don't use as 935 * process code in most system because it uses 936 * escape sequences(JIS is context depend encoding). 937 */ 938 // assume process code is JAPANESE-EUC 939 lb &= 0x7f; 940 tb &= 0x7f; 941 # endif 942 // exceptions 943 switch (lb << 8 | tb) 944 { 945 case 0x2121: // ZENKAKU space 946 return 0; 947 case 0x2122: // TOU-TEN (Japanese comma) 948 case 0x2123: // KU-TEN (Japanese period) 949 case 0x2124: // ZENKAKU comma 950 case 0x2125: // ZENKAKU period 951 return 1; 952 case 0x213c: // prolongedsound handled as KATAKANA 953 return 13; 954 } 955 // sieved by KU code 956 switch (lb) 957 { 958 case 0x21: 959 case 0x22: 960 // special symbols 961 return 10; 962 case 0x23: 963 // alphanumeric 964 return 11; 965 case 0x24: 966 // hiragana 967 return 12; 968 case 0x25: 969 // katakana 970 return 13; 971 case 0x26: 972 // greek 973 return 14; 974 case 0x27: 975 // russian 976 return 15; 977 case 0x28: 978 // lines 979 return 16; 980 default: 981 // kanji 982 return 17; 983 } 984 } 985 986 case DBCS_KORU: // ? 987 case DBCS_KOR: 988 { 989 // KS code classification 990 unsigned char c1 = lead; 991 unsigned char c2 = trail; 992 993 /* 994 * 20 : Hangul 995 * 21 : Hanja 996 * 22 : Symbols 997 * 23 : Alphanumeric/Roman Letter (Full width) 998 * 24 : Hangul Letter(Alphabet) 999 * 25 : Roman Numeral/Greek Letter 1000 * 26 : Box Drawings 1001 * 27 : Unit Symbols 1002 * 28 : Circled/Parenthesized Letter 1003 * 29 : Hiragana/Katakana 1004 * 30 : Cyrillic Letter 1005 */ 1006 1007 if (c1 >= 0xB0 && c1 <= 0xC8) 1008 // Hangul 1009 return 20; 1010 #if defined(MSWIN) || defined(WIN32UNIX) 1011 else if (c1 <= 0xA0 || c2 <= 0xA0) 1012 // Extended Hangul Region : MS UHC(Unified Hangul Code) 1013 // c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE 1014 // c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 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 // Alphanumeric 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, 0x4dbf}, 1465 {0x4e00, 0xa48c}, 1466 {0xa490, 0xa4c6}, 1467 {0xa960, 0xa97c}, 1468 {0xac00, 0xd7a3}, 1469 {0xf900, 0xfaff}, 1470 {0xfe10, 0xfe19}, 1471 {0xfe30, 0xfe52}, 1472 {0xfe54, 0xfe66}, 1473 {0xfe68, 0xfe6b}, 1474 {0xff01, 0xff60}, 1475 {0xffe0, 0xffe6}, 1476 {0x16fe0, 0x16fe3}, 1477 {0x17000, 0x187f7}, 1478 {0x18800, 0x18af2}, 1479 {0x1b000, 0x1b11e}, 1480 {0x1b150, 0x1b152}, 1481 {0x1b164, 0x1b167}, 1482 {0x1b170, 0x1b2fb}, 1483 {0x1f004, 0x1f004}, 1484 {0x1f0cf, 0x1f0cf}, 1485 {0x1f18e, 0x1f18e}, 1486 {0x1f191, 0x1f19a}, 1487 {0x1f200, 0x1f202}, 1488 {0x1f210, 0x1f23b}, 1489 {0x1f240, 0x1f248}, 1490 {0x1f250, 0x1f251}, 1491 {0x1f260, 0x1f265}, 1492 {0x1f300, 0x1f320}, 1493 {0x1f32d, 0x1f335}, 1494 {0x1f337, 0x1f37c}, 1495 {0x1f37e, 0x1f393}, 1496 {0x1f3a0, 0x1f3ca}, 1497 {0x1f3cf, 0x1f3d3}, 1498 {0x1f3e0, 0x1f3f0}, 1499 {0x1f3f4, 0x1f3f4}, 1500 {0x1f3f8, 0x1f43e}, 1501 {0x1f440, 0x1f440}, 1502 {0x1f442, 0x1f4fc}, 1503 {0x1f4ff, 0x1f53d}, 1504 {0x1f54b, 0x1f54e}, 1505 {0x1f550, 0x1f567}, 1506 {0x1f57a, 0x1f57a}, 1507 {0x1f595, 0x1f596}, 1508 {0x1f5a4, 0x1f5a4}, 1509 {0x1f5fb, 0x1f64f}, 1510 {0x1f680, 0x1f6c5}, 1511 {0x1f6cc, 0x1f6cc}, 1512 {0x1f6d0, 0x1f6d2}, 1513 {0x1f6d5, 0x1f6d5}, 1514 {0x1f6eb, 0x1f6ec}, 1515 {0x1f6f4, 0x1f6fa}, 1516 {0x1f7e0, 0x1f7eb}, 1517 {0x1f90d, 0x1f971}, 1518 {0x1f973, 0x1f976}, 1519 {0x1f97a, 0x1f9a2}, 1520 {0x1f9a5, 0x1f9aa}, 1521 {0x1f9ae, 0x1f9ca}, 1522 {0x1f9cd, 0x1f9ff}, 1523 {0x1fa70, 0x1fa73}, 1524 {0x1fa78, 0x1fa7a}, 1525 {0x1fa80, 0x1fa82}, 1526 {0x1fa90, 0x1fa95}, 1527 {0x20000, 0x2fffd}, 1528 {0x30000, 0x3fffd} 1529 }; 1530 1531 // Sorted list of non-overlapping intervals of Emoji characters that don't 1532 // have ambiguous or double width, 1533 // based on http://unicode.org/emoji/charts/emoji-list.html 1534 static struct interval emoji_width[] = 1535 { 1536 {0x1f1e6, 0x1f1ff}, 1537 {0x1f321, 0x1f321}, 1538 {0x1f324, 0x1f32c}, 1539 {0x1f336, 0x1f336}, 1540 {0x1f37d, 0x1f37d}, 1541 {0x1f396, 0x1f397}, 1542 {0x1f399, 0x1f39b}, 1543 {0x1f39e, 0x1f39f}, 1544 {0x1f3cb, 0x1f3ce}, 1545 {0x1f3d4, 0x1f3df}, 1546 {0x1f3f3, 0x1f3f5}, 1547 {0x1f3f7, 0x1f3f7}, 1548 {0x1f43f, 0x1f43f}, 1549 {0x1f441, 0x1f441}, 1550 {0x1f4fd, 0x1f4fd}, 1551 {0x1f549, 0x1f54a}, 1552 {0x1f56f, 0x1f570}, 1553 {0x1f573, 0x1f579}, 1554 {0x1f587, 0x1f587}, 1555 {0x1f58a, 0x1f58d}, 1556 {0x1f590, 0x1f590}, 1557 {0x1f5a5, 0x1f5a5}, 1558 {0x1f5a8, 0x1f5a8}, 1559 {0x1f5b1, 0x1f5b2}, 1560 {0x1f5bc, 0x1f5bc}, 1561 {0x1f5c2, 0x1f5c4}, 1562 {0x1f5d1, 0x1f5d3}, 1563 {0x1f5dc, 0x1f5de}, 1564 {0x1f5e1, 0x1f5e1}, 1565 {0x1f5e3, 0x1f5e3}, 1566 {0x1f5e8, 0x1f5e8}, 1567 {0x1f5ef, 0x1f5ef}, 1568 {0x1f5f3, 0x1f5f3}, 1569 {0x1f5fa, 0x1f5fa}, 1570 {0x1f6cb, 0x1f6cf}, 1571 {0x1f6e0, 0x1f6e5}, 1572 {0x1f6e9, 0x1f6e9}, 1573 {0x1f6f0, 0x1f6f0}, 1574 {0x1f6f3, 0x1f6f3} 1575 }; 1576 1577 if (c >= 0x100) 1578 { 1579 #ifdef USE_WCHAR_FUNCTIONS 1580 /* 1581 * Assume the library function wcwidth() works better than our own 1582 * stuff. It should return 1 for ambiguous width chars! 1583 */ 1584 int n = wcwidth(c); 1585 1586 if (n < 0) 1587 return 6; // unprintable, displays <xxxx> 1588 if (n > 1) 1589 return n; 1590 #else 1591 if (!utf_printable(c)) 1592 return 6; // unprintable, displays <xxxx> 1593 if (intable(doublewidth, sizeof(doublewidth), c)) 1594 return 2; 1595 #endif 1596 if (p_emoji && intable(emoji_width, sizeof(emoji_width), c)) 1597 return 2; 1598 } 1599 1600 // Characters below 0x100 are influenced by 'isprint' option 1601 else if (c >= 0x80 && !vim_isprintc(c)) 1602 return 4; // unprintable, displays <xx> 1603 1604 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c)) 1605 return 2; 1606 1607 return 1; 1608 } 1609 1610 /* 1611 * mb_ptr2cells() function pointer. 1612 * Return the number of display cells character at "*p" occupies. 1613 * This doesn't take care of unprintable characters, use ptr2cells() for that. 1614 */ 1615 int 1616 latin_ptr2cells(char_u *p UNUSED) 1617 { 1618 return 1; 1619 } 1620 1621 int 1622 utf_ptr2cells( 1623 char_u *p) 1624 { 1625 int c; 1626 1627 // Need to convert to a wide character. 1628 if (*p >= 0x80) 1629 { 1630 c = utf_ptr2char(p); 1631 // An illegal byte is displayed as <xx>. 1632 if (utf_ptr2len(p) == 1 || c == NUL) 1633 return 4; 1634 // If the char is ASCII it must be an overlong sequence. 1635 if (c < 0x80) 1636 return char2cells(c); 1637 return utf_char2cells(c); 1638 } 1639 return 1; 1640 } 1641 1642 int 1643 dbcs_ptr2cells(char_u *p) 1644 { 1645 // Number of cells is equal to number of bytes, except for euc-jp when 1646 // the first byte is 0x8e. 1647 if (enc_dbcs == DBCS_JPNU && *p == 0x8e) 1648 return 1; 1649 return MB_BYTE2LEN(*p); 1650 } 1651 1652 /* 1653 * mb_ptr2cells_len() function pointer. 1654 * Like mb_ptr2cells(), but limit string length to "size". 1655 * For an empty string or truncated character returns 1. 1656 */ 1657 int 1658 latin_ptr2cells_len(char_u *p UNUSED, int size UNUSED) 1659 { 1660 return 1; 1661 } 1662 1663 static int 1664 utf_ptr2cells_len(char_u *p, int size) 1665 { 1666 int c; 1667 1668 // Need to convert to a wide character. 1669 if (size > 0 && *p >= 0x80) 1670 { 1671 if (utf_ptr2len_len(p, size) < utf8len_tab[*p]) 1672 return 1; // truncated 1673 c = utf_ptr2char(p); 1674 // An illegal byte is displayed as <xx>. 1675 if (utf_ptr2len(p) == 1 || c == NUL) 1676 return 4; 1677 // If the char is ASCII it must be an overlong sequence. 1678 if (c < 0x80) 1679 return char2cells(c); 1680 return utf_char2cells(c); 1681 } 1682 return 1; 1683 } 1684 1685 static int 1686 dbcs_ptr2cells_len(char_u *p, int size) 1687 { 1688 // Number of cells is equal to number of bytes, except for euc-jp when 1689 // the first byte is 0x8e. 1690 if (size <= 1 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)) 1691 return 1; 1692 return MB_BYTE2LEN(*p); 1693 } 1694 1695 /* 1696 * mb_char2cells() function pointer. 1697 * Return the number of display cells character "c" occupies. 1698 * Only takes care of multi-byte chars, not "^C" and such. 1699 */ 1700 int 1701 latin_char2cells(int c UNUSED) 1702 { 1703 return 1; 1704 } 1705 1706 static int 1707 dbcs_char2cells(int c) 1708 { 1709 // Number of cells is equal to number of bytes, except for euc-jp when 1710 // the first byte is 0x8e. 1711 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e) 1712 return 1; 1713 // use the first byte 1714 return MB_BYTE2LEN((unsigned)c >> 8); 1715 } 1716 1717 /* 1718 * Return the number of cells occupied by string "p". 1719 * Stop at a NUL character. When "len" >= 0 stop at character "p[len]". 1720 */ 1721 int 1722 mb_string2cells(char_u *p, int len) 1723 { 1724 int i; 1725 int clen = 0; 1726 1727 for (i = 0; (len < 0 || i < len) && p[i] != NUL; i += (*mb_ptr2len)(p + i)) 1728 clen += (*mb_ptr2cells)(p + i); 1729 return clen; 1730 } 1731 1732 /* 1733 * mb_off2cells() function pointer. 1734 * Return number of display cells for char at ScreenLines[off]. 1735 * We make sure that the offset used is less than "max_off". 1736 */ 1737 int 1738 latin_off2cells(unsigned off UNUSED, unsigned max_off UNUSED) 1739 { 1740 return 1; 1741 } 1742 1743 int 1744 dbcs_off2cells(unsigned off, unsigned max_off) 1745 { 1746 // never check beyond end of the line 1747 if (off >= max_off) 1748 return 1; 1749 1750 // Number of cells is equal to number of bytes, except for euc-jp when 1751 // the first byte is 0x8e. 1752 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e) 1753 return 1; 1754 return MB_BYTE2LEN(ScreenLines[off]); 1755 } 1756 1757 int 1758 utf_off2cells(unsigned off, unsigned max_off) 1759 { 1760 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1; 1761 } 1762 1763 /* 1764 * mb_ptr2char() function pointer. 1765 * Convert a byte sequence into a character. 1766 */ 1767 int 1768 latin_ptr2char(char_u *p) 1769 { 1770 return *p; 1771 } 1772 1773 static int 1774 dbcs_ptr2char(char_u *p) 1775 { 1776 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL) 1777 return (p[0] << 8) + p[1]; 1778 return *p; 1779 } 1780 1781 /* 1782 * Convert a UTF-8 byte sequence to a wide character. 1783 * If the sequence is illegal or truncated by a NUL the first byte is 1784 * returned. 1785 * For an overlong sequence this may return zero. 1786 * Does not include composing characters, of course. 1787 */ 1788 int 1789 utf_ptr2char(char_u *p) 1790 { 1791 int len; 1792 1793 if (p[0] < 0x80) // be quick for ASCII 1794 return p[0]; 1795 1796 len = utf8len_tab_zero[p[0]]; 1797 if (len > 1 && (p[1] & 0xc0) == 0x80) 1798 { 1799 if (len == 2) 1800 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); 1801 if ((p[2] & 0xc0) == 0x80) 1802 { 1803 if (len == 3) 1804 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) 1805 + (p[2] & 0x3f); 1806 if ((p[3] & 0xc0) == 0x80) 1807 { 1808 if (len == 4) 1809 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) 1810 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); 1811 if ((p[4] & 0xc0) == 0x80) 1812 { 1813 if (len == 5) 1814 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) 1815 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) 1816 + (p[4] & 0x3f); 1817 if ((p[5] & 0xc0) == 0x80 && len == 6) 1818 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) 1819 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) 1820 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); 1821 } 1822 } 1823 } 1824 } 1825 // Illegal value, just return the first byte 1826 return p[0]; 1827 } 1828 1829 /* 1830 * Convert a UTF-8 byte sequence to a wide character. 1831 * String is assumed to be terminated by NUL or after "n" bytes, whichever 1832 * comes first. 1833 * The function is safe in the sense that it never accesses memory beyond the 1834 * first "n" bytes of "s". 1835 * 1836 * On success, returns decoded codepoint, advances "s" to the beginning of 1837 * next character and decreases "n" accordingly. 1838 * 1839 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past 1840 * NUL byte. 1841 * 1842 * If byte sequence is illegal or incomplete, returns -1 and does not advance 1843 * "s". 1844 */ 1845 static int 1846 utf_safe_read_char_adv(char_u **s, size_t *n) 1847 { 1848 int c, k; 1849 1850 if (*n == 0) // end of buffer 1851 return 0; 1852 1853 k = utf8len_tab_zero[**s]; 1854 1855 if (k == 1) 1856 { 1857 // ASCII character or NUL 1858 (*n)--; 1859 return *(*s)++; 1860 } 1861 1862 if ((size_t)k <= *n) 1863 { 1864 // We have a multibyte sequence and it isn't truncated by buffer 1865 // limits so utf_ptr2char() is safe to use. Or the first byte is 1866 // illegal (k=0), and it's also safe to use utf_ptr2char(). 1867 c = utf_ptr2char(*s); 1868 1869 // On failure, utf_ptr2char() returns the first byte, so here we 1870 // check equality with the first byte. The only non-ASCII character 1871 // which equals the first byte of its own UTF-8 representation is 1872 // U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too. 1873 // It's safe even if n=1, else we would have k=2 > n. 1874 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83)) 1875 { 1876 // byte sequence was successfully decoded 1877 *s += k; 1878 *n -= k; 1879 return c; 1880 } 1881 } 1882 1883 // byte sequence is incomplete or illegal 1884 return -1; 1885 } 1886 1887 /* 1888 * Get character at **pp and advance *pp to the next character. 1889 * Note: composing characters are skipped! 1890 */ 1891 int 1892 mb_ptr2char_adv(char_u **pp) 1893 { 1894 int c; 1895 1896 c = (*mb_ptr2char)(*pp); 1897 *pp += (*mb_ptr2len)(*pp); 1898 return c; 1899 } 1900 1901 /* 1902 * Get character at **pp and advance *pp to the next character. 1903 * Note: composing characters are returned as separate characters. 1904 */ 1905 int 1906 mb_cptr2char_adv(char_u **pp) 1907 { 1908 int c; 1909 1910 c = (*mb_ptr2char)(*pp); 1911 if (enc_utf8) 1912 *pp += utf_ptr2len(*pp); 1913 else 1914 *pp += (*mb_ptr2len)(*pp); 1915 return c; 1916 } 1917 1918 #if defined(FEAT_ARABIC) || defined(PROTO) 1919 /* 1920 * Check if the character pointed to by "p2" is a composing character when it 1921 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which 1922 * behaves like a composing character. 1923 */ 1924 int 1925 utf_composinglike(char_u *p1, char_u *p2) 1926 { 1927 int c2; 1928 1929 c2 = utf_ptr2char(p2); 1930 if (utf_iscomposing(c2)) 1931 return TRUE; 1932 if (!arabic_maycombine(c2)) 1933 return FALSE; 1934 return arabic_combine(utf_ptr2char(p1), c2); 1935 } 1936 #endif 1937 1938 /* 1939 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1940 * composing characters. 1941 */ 1942 int 1943 utfc_ptr2char( 1944 char_u *p, 1945 int *pcc) // return: composing chars, last one is 0 1946 { 1947 int len; 1948 int c; 1949 int cc; 1950 int i = 0; 1951 1952 c = utf_ptr2char(p); 1953 len = utf_ptr2len(p); 1954 1955 // Only accept a composing char when the first char isn't illegal. 1956 if ((len > 1 || *p < 0x80) 1957 && p[len] >= 0x80 1958 && UTF_COMPOSINGLIKE(p, p + len)) 1959 { 1960 cc = utf_ptr2char(p + len); 1961 for (;;) 1962 { 1963 pcc[i++] = cc; 1964 if (i == MAX_MCO) 1965 break; 1966 len += utf_ptr2len(p + len); 1967 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1968 break; 1969 } 1970 } 1971 1972 if (i < MAX_MCO) // last composing char must be 0 1973 pcc[i] = 0; 1974 1975 return c; 1976 } 1977 1978 /* 1979 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1980 * composing characters. Use no more than p[maxlen]. 1981 */ 1982 int 1983 utfc_ptr2char_len( 1984 char_u *p, 1985 int *pcc, // return: composing chars, last one is 0 1986 int maxlen) 1987 { 1988 int len; 1989 int c; 1990 int cc; 1991 int i = 0; 1992 1993 c = utf_ptr2char(p); 1994 len = utf_ptr2len_len(p, maxlen); 1995 // Only accept a composing char when the first char isn't illegal. 1996 if ((len > 1 || *p < 0x80) 1997 && len < maxlen 1998 && p[len] >= 0x80 1999 && UTF_COMPOSINGLIKE(p, p + len)) 2000 { 2001 cc = utf_ptr2char(p + len); 2002 for (;;) 2003 { 2004 pcc[i++] = cc; 2005 if (i == MAX_MCO) 2006 break; 2007 len += utf_ptr2len_len(p + len, maxlen - len); 2008 if (len >= maxlen 2009 || p[len] < 0x80 2010 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 2011 break; 2012 } 2013 } 2014 2015 if (i < MAX_MCO) // last composing char must be 0 2016 pcc[i] = 0; 2017 2018 return c; 2019 } 2020 2021 /* 2022 * Convert the character at screen position "off" to a sequence of bytes. 2023 * Includes the composing characters. 2024 * "buf" must at least have the length MB_MAXBYTES + 1. 2025 * Only to be used when ScreenLinesUC[off] != 0. 2026 * Returns the produced number of bytes. 2027 */ 2028 int 2029 utfc_char2bytes(int off, char_u *buf) 2030 { 2031 int len; 2032 int i; 2033 2034 len = utf_char2bytes(ScreenLinesUC[off], buf); 2035 for (i = 0; i < Screen_mco; ++i) 2036 { 2037 if (ScreenLinesC[i][off] == 0) 2038 break; 2039 len += utf_char2bytes(ScreenLinesC[i][off], buf + len); 2040 } 2041 return len; 2042 } 2043 2044 /* 2045 * Get the length of a UTF-8 byte sequence, not including any following 2046 * composing characters. 2047 * Returns 0 for "". 2048 * Returns 1 for an illegal byte sequence. 2049 */ 2050 int 2051 utf_ptr2len(char_u *p) 2052 { 2053 int len; 2054 int i; 2055 2056 if (*p == NUL) 2057 return 0; 2058 len = utf8len_tab[*p]; 2059 for (i = 1; i < len; ++i) 2060 if ((p[i] & 0xc0) != 0x80) 2061 return 1; 2062 return len; 2063 } 2064 2065 /* 2066 * Return length of UTF-8 character, obtained from the first byte. 2067 * "b" must be between 0 and 255! 2068 * Returns 1 for an invalid first byte value. 2069 */ 2070 int 2071 utf_byte2len(int b) 2072 { 2073 return utf8len_tab[b]; 2074 } 2075 2076 /* 2077 * Get the length of UTF-8 byte sequence "p[size]". Does not include any 2078 * following composing characters. 2079 * Returns 1 for "". 2080 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.). 2081 * Returns number > "size" for an incomplete byte sequence. 2082 * Never returns zero. 2083 */ 2084 int 2085 utf_ptr2len_len(char_u *p, int size) 2086 { 2087 int len; 2088 int i; 2089 int m; 2090 2091 len = utf8len_tab[*p]; 2092 if (len == 1) 2093 return 1; // NUL, ascii or illegal lead byte 2094 if (len > size) 2095 m = size; // incomplete byte sequence. 2096 else 2097 m = len; 2098 for (i = 1; i < m; ++i) 2099 if ((p[i] & 0xc0) != 0x80) 2100 return 1; 2101 return len; 2102 } 2103 2104 /* 2105 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. 2106 * This includes following composing characters. 2107 */ 2108 int 2109 utfc_ptr2len(char_u *p) 2110 { 2111 int len; 2112 int b0 = *p; 2113 #ifdef FEAT_ARABIC 2114 int prevlen; 2115 #endif 2116 2117 if (b0 == NUL) 2118 return 0; 2119 if (b0 < 0x80 && p[1] < 0x80) // be quick for ASCII 2120 return 1; 2121 2122 // Skip over first UTF-8 char, stopping at a NUL byte. 2123 len = utf_ptr2len(p); 2124 2125 // Check for illegal byte. 2126 if (len == 1 && b0 >= 0x80) 2127 return 1; 2128 2129 /* 2130 * Check for composing characters. We can handle only the first six, but 2131 * skip all of them (otherwise the cursor would get stuck). 2132 */ 2133 #ifdef FEAT_ARABIC 2134 prevlen = 0; 2135 #endif 2136 for (;;) 2137 { 2138 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2139 return len; 2140 2141 // Skip over composing char 2142 #ifdef FEAT_ARABIC 2143 prevlen = len; 2144 #endif 2145 len += utf_ptr2len(p + len); 2146 } 2147 } 2148 2149 /* 2150 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" 2151 * takes. This includes following composing characters. 2152 * Returns 0 for an empty string. 2153 * Returns 1 for an illegal char or an incomplete byte sequence. 2154 */ 2155 int 2156 utfc_ptr2len_len(char_u *p, int size) 2157 { 2158 int len; 2159 #ifdef FEAT_ARABIC 2160 int prevlen; 2161 #endif 2162 2163 if (size < 1 || *p == NUL) 2164 return 0; 2165 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) // be quick for ASCII 2166 return 1; 2167 2168 // Skip over first UTF-8 char, stopping at a NUL byte. 2169 len = utf_ptr2len_len(p, size); 2170 2171 // Check for illegal byte and incomplete byte sequence. 2172 if ((len == 1 && p[0] >= 0x80) || len > size) 2173 return 1; 2174 2175 /* 2176 * Check for composing characters. We can handle only the first six, but 2177 * skip all of them (otherwise the cursor would get stuck). 2178 */ 2179 #ifdef FEAT_ARABIC 2180 prevlen = 0; 2181 #endif 2182 while (len < size) 2183 { 2184 int len_next_char; 2185 2186 if (p[len] < 0x80) 2187 break; 2188 2189 /* 2190 * Next character length should not go beyond size to ensure that 2191 * UTF_COMPOSINGLIKE(...) does not read beyond size. 2192 */ 2193 len_next_char = utf_ptr2len_len(p + len, size - len); 2194 if (len_next_char > size - len) 2195 break; 2196 2197 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2198 break; 2199 2200 // Skip over composing char 2201 #ifdef FEAT_ARABIC 2202 prevlen = len; 2203 #endif 2204 len += len_next_char; 2205 } 2206 return len; 2207 } 2208 2209 /* 2210 * Return the number of bytes the UTF-8 encoding of character "c" takes. 2211 * This does not include composing characters. 2212 */ 2213 int 2214 utf_char2len(int c) 2215 { 2216 if (c < 0x80) 2217 return 1; 2218 if (c < 0x800) 2219 return 2; 2220 if (c < 0x10000) 2221 return 3; 2222 if (c < 0x200000) 2223 return 4; 2224 if (c < 0x4000000) 2225 return 5; 2226 return 6; 2227 } 2228 2229 /* 2230 * Convert Unicode character "c" to UTF-8 string in "buf[]". 2231 * Returns the number of bytes. 2232 */ 2233 int 2234 utf_char2bytes(int c, char_u *buf) 2235 { 2236 if (c < 0x80) // 7 bits 2237 { 2238 buf[0] = c; 2239 return 1; 2240 } 2241 if (c < 0x800) // 11 bits 2242 { 2243 buf[0] = 0xc0 + ((unsigned)c >> 6); 2244 buf[1] = 0x80 + (c & 0x3f); 2245 return 2; 2246 } 2247 if (c < 0x10000) // 16 bits 2248 { 2249 buf[0] = 0xe0 + ((unsigned)c >> 12); 2250 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2251 buf[2] = 0x80 + (c & 0x3f); 2252 return 3; 2253 } 2254 if (c < 0x200000) // 21 bits 2255 { 2256 buf[0] = 0xf0 + ((unsigned)c >> 18); 2257 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2258 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2259 buf[3] = 0x80 + (c & 0x3f); 2260 return 4; 2261 } 2262 if (c < 0x4000000) // 26 bits 2263 { 2264 buf[0] = 0xf8 + ((unsigned)c >> 24); 2265 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2266 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2267 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2268 buf[4] = 0x80 + (c & 0x3f); 2269 return 5; 2270 } 2271 // 31 bits 2272 buf[0] = 0xfc + ((unsigned)c >> 30); 2273 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f); 2274 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2275 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2276 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2277 buf[5] = 0x80 + (c & 0x3f); 2278 return 6; 2279 } 2280 2281 #if defined(FEAT_TERMINAL) || defined(PROTO) 2282 /* 2283 * utf_iscomposing() with different argument type for libvterm. 2284 */ 2285 int 2286 utf_iscomposing_uint(UINT32_T c) 2287 { 2288 return utf_iscomposing((int)c); 2289 } 2290 #endif 2291 2292 /* 2293 * Return TRUE if "c" is a composing UTF-8 character. This means it will be 2294 * drawn on top of the preceding character. 2295 * Based on code from Markus Kuhn. 2296 */ 2297 int 2298 utf_iscomposing(int c) 2299 { 2300 // Sorted list of non-overlapping intervals. 2301 // Generated by ../runtime/tools/unicode.vim. 2302 static struct interval combining[] = 2303 { 2304 {0x0300, 0x036f}, 2305 {0x0483, 0x0489}, 2306 {0x0591, 0x05bd}, 2307 {0x05bf, 0x05bf}, 2308 {0x05c1, 0x05c2}, 2309 {0x05c4, 0x05c5}, 2310 {0x05c7, 0x05c7}, 2311 {0x0610, 0x061a}, 2312 {0x064b, 0x065f}, 2313 {0x0670, 0x0670}, 2314 {0x06d6, 0x06dc}, 2315 {0x06df, 0x06e4}, 2316 {0x06e7, 0x06e8}, 2317 {0x06ea, 0x06ed}, 2318 {0x0711, 0x0711}, 2319 {0x0730, 0x074a}, 2320 {0x07a6, 0x07b0}, 2321 {0x07eb, 0x07f3}, 2322 {0x07fd, 0x07fd}, 2323 {0x0816, 0x0819}, 2324 {0x081b, 0x0823}, 2325 {0x0825, 0x0827}, 2326 {0x0829, 0x082d}, 2327 {0x0859, 0x085b}, 2328 {0x08d3, 0x08e1}, 2329 {0x08e3, 0x0903}, 2330 {0x093a, 0x093c}, 2331 {0x093e, 0x094f}, 2332 {0x0951, 0x0957}, 2333 {0x0962, 0x0963}, 2334 {0x0981, 0x0983}, 2335 {0x09bc, 0x09bc}, 2336 {0x09be, 0x09c4}, 2337 {0x09c7, 0x09c8}, 2338 {0x09cb, 0x09cd}, 2339 {0x09d7, 0x09d7}, 2340 {0x09e2, 0x09e3}, 2341 {0x09fe, 0x09fe}, 2342 {0x0a01, 0x0a03}, 2343 {0x0a3c, 0x0a3c}, 2344 {0x0a3e, 0x0a42}, 2345 {0x0a47, 0x0a48}, 2346 {0x0a4b, 0x0a4d}, 2347 {0x0a51, 0x0a51}, 2348 {0x0a70, 0x0a71}, 2349 {0x0a75, 0x0a75}, 2350 {0x0a81, 0x0a83}, 2351 {0x0abc, 0x0abc}, 2352 {0x0abe, 0x0ac5}, 2353 {0x0ac7, 0x0ac9}, 2354 {0x0acb, 0x0acd}, 2355 {0x0ae2, 0x0ae3}, 2356 {0x0afa, 0x0aff}, 2357 {0x0b01, 0x0b03}, 2358 {0x0b3c, 0x0b3c}, 2359 {0x0b3e, 0x0b44}, 2360 {0x0b47, 0x0b48}, 2361 {0x0b4b, 0x0b4d}, 2362 {0x0b56, 0x0b57}, 2363 {0x0b62, 0x0b63}, 2364 {0x0b82, 0x0b82}, 2365 {0x0bbe, 0x0bc2}, 2366 {0x0bc6, 0x0bc8}, 2367 {0x0bca, 0x0bcd}, 2368 {0x0bd7, 0x0bd7}, 2369 {0x0c00, 0x0c04}, 2370 {0x0c3e, 0x0c44}, 2371 {0x0c46, 0x0c48}, 2372 {0x0c4a, 0x0c4d}, 2373 {0x0c55, 0x0c56}, 2374 {0x0c62, 0x0c63}, 2375 {0x0c81, 0x0c83}, 2376 {0x0cbc, 0x0cbc}, 2377 {0x0cbe, 0x0cc4}, 2378 {0x0cc6, 0x0cc8}, 2379 {0x0cca, 0x0ccd}, 2380 {0x0cd5, 0x0cd6}, 2381 {0x0ce2, 0x0ce3}, 2382 {0x0d00, 0x0d03}, 2383 {0x0d3b, 0x0d3c}, 2384 {0x0d3e, 0x0d44}, 2385 {0x0d46, 0x0d48}, 2386 {0x0d4a, 0x0d4d}, 2387 {0x0d57, 0x0d57}, 2388 {0x0d62, 0x0d63}, 2389 {0x0d82, 0x0d83}, 2390 {0x0dca, 0x0dca}, 2391 {0x0dcf, 0x0dd4}, 2392 {0x0dd6, 0x0dd6}, 2393 {0x0dd8, 0x0ddf}, 2394 {0x0df2, 0x0df3}, 2395 {0x0e31, 0x0e31}, 2396 {0x0e34, 0x0e3a}, 2397 {0x0e47, 0x0e4e}, 2398 {0x0eb1, 0x0eb1}, 2399 {0x0eb4, 0x0ebc}, 2400 {0x0ec8, 0x0ecd}, 2401 {0x0f18, 0x0f19}, 2402 {0x0f35, 0x0f35}, 2403 {0x0f37, 0x0f37}, 2404 {0x0f39, 0x0f39}, 2405 {0x0f3e, 0x0f3f}, 2406 {0x0f71, 0x0f84}, 2407 {0x0f86, 0x0f87}, 2408 {0x0f8d, 0x0f97}, 2409 {0x0f99, 0x0fbc}, 2410 {0x0fc6, 0x0fc6}, 2411 {0x102b, 0x103e}, 2412 {0x1056, 0x1059}, 2413 {0x105e, 0x1060}, 2414 {0x1062, 0x1064}, 2415 {0x1067, 0x106d}, 2416 {0x1071, 0x1074}, 2417 {0x1082, 0x108d}, 2418 {0x108f, 0x108f}, 2419 {0x109a, 0x109d}, 2420 {0x135d, 0x135f}, 2421 {0x1712, 0x1714}, 2422 {0x1732, 0x1734}, 2423 {0x1752, 0x1753}, 2424 {0x1772, 0x1773}, 2425 {0x17b4, 0x17d3}, 2426 {0x17dd, 0x17dd}, 2427 {0x180b, 0x180d}, 2428 {0x1885, 0x1886}, 2429 {0x18a9, 0x18a9}, 2430 {0x1920, 0x192b}, 2431 {0x1930, 0x193b}, 2432 {0x1a17, 0x1a1b}, 2433 {0x1a55, 0x1a5e}, 2434 {0x1a60, 0x1a7c}, 2435 {0x1a7f, 0x1a7f}, 2436 {0x1ab0, 0x1abe}, 2437 {0x1b00, 0x1b04}, 2438 {0x1b34, 0x1b44}, 2439 {0x1b6b, 0x1b73}, 2440 {0x1b80, 0x1b82}, 2441 {0x1ba1, 0x1bad}, 2442 {0x1be6, 0x1bf3}, 2443 {0x1c24, 0x1c37}, 2444 {0x1cd0, 0x1cd2}, 2445 {0x1cd4, 0x1ce8}, 2446 {0x1ced, 0x1ced}, 2447 {0x1cf4, 0x1cf4}, 2448 {0x1cf7, 0x1cf9}, 2449 {0x1dc0, 0x1df9}, 2450 {0x1dfb, 0x1dff}, 2451 {0x20d0, 0x20f0}, 2452 {0x2cef, 0x2cf1}, 2453 {0x2d7f, 0x2d7f}, 2454 {0x2de0, 0x2dff}, 2455 {0x302a, 0x302f}, 2456 {0x3099, 0x309a}, 2457 {0xa66f, 0xa672}, 2458 {0xa674, 0xa67d}, 2459 {0xa69e, 0xa69f}, 2460 {0xa6f0, 0xa6f1}, 2461 {0xa802, 0xa802}, 2462 {0xa806, 0xa806}, 2463 {0xa80b, 0xa80b}, 2464 {0xa823, 0xa827}, 2465 {0xa880, 0xa881}, 2466 {0xa8b4, 0xa8c5}, 2467 {0xa8e0, 0xa8f1}, 2468 {0xa8ff, 0xa8ff}, 2469 {0xa926, 0xa92d}, 2470 {0xa947, 0xa953}, 2471 {0xa980, 0xa983}, 2472 {0xa9b3, 0xa9c0}, 2473 {0xa9e5, 0xa9e5}, 2474 {0xaa29, 0xaa36}, 2475 {0xaa43, 0xaa43}, 2476 {0xaa4c, 0xaa4d}, 2477 {0xaa7b, 0xaa7d}, 2478 {0xaab0, 0xaab0}, 2479 {0xaab2, 0xaab4}, 2480 {0xaab7, 0xaab8}, 2481 {0xaabe, 0xaabf}, 2482 {0xaac1, 0xaac1}, 2483 {0xaaeb, 0xaaef}, 2484 {0xaaf5, 0xaaf6}, 2485 {0xabe3, 0xabea}, 2486 {0xabec, 0xabed}, 2487 {0xfb1e, 0xfb1e}, 2488 {0xfe00, 0xfe0f}, 2489 {0xfe20, 0xfe2f}, 2490 {0x101fd, 0x101fd}, 2491 {0x102e0, 0x102e0}, 2492 {0x10376, 0x1037a}, 2493 {0x10a01, 0x10a03}, 2494 {0x10a05, 0x10a06}, 2495 {0x10a0c, 0x10a0f}, 2496 {0x10a38, 0x10a3a}, 2497 {0x10a3f, 0x10a3f}, 2498 {0x10ae5, 0x10ae6}, 2499 {0x10d24, 0x10d27}, 2500 {0x10f46, 0x10f50}, 2501 {0x11000, 0x11002}, 2502 {0x11038, 0x11046}, 2503 {0x1107f, 0x11082}, 2504 {0x110b0, 0x110ba}, 2505 {0x11100, 0x11102}, 2506 {0x11127, 0x11134}, 2507 {0x11145, 0x11146}, 2508 {0x11173, 0x11173}, 2509 {0x11180, 0x11182}, 2510 {0x111b3, 0x111c0}, 2511 {0x111c9, 0x111cc}, 2512 {0x1122c, 0x11237}, 2513 {0x1123e, 0x1123e}, 2514 {0x112df, 0x112ea}, 2515 {0x11300, 0x11303}, 2516 {0x1133b, 0x1133c}, 2517 {0x1133e, 0x11344}, 2518 {0x11347, 0x11348}, 2519 {0x1134b, 0x1134d}, 2520 {0x11357, 0x11357}, 2521 {0x11362, 0x11363}, 2522 {0x11366, 0x1136c}, 2523 {0x11370, 0x11374}, 2524 {0x11435, 0x11446}, 2525 {0x1145e, 0x1145e}, 2526 {0x114b0, 0x114c3}, 2527 {0x115af, 0x115b5}, 2528 {0x115b8, 0x115c0}, 2529 {0x115dc, 0x115dd}, 2530 {0x11630, 0x11640}, 2531 {0x116ab, 0x116b7}, 2532 {0x1171d, 0x1172b}, 2533 {0x1182c, 0x1183a}, 2534 {0x119d1, 0x119d7}, 2535 {0x119da, 0x119e0}, 2536 {0x119e4, 0x119e4}, 2537 {0x11a01, 0x11a0a}, 2538 {0x11a33, 0x11a39}, 2539 {0x11a3b, 0x11a3e}, 2540 {0x11a47, 0x11a47}, 2541 {0x11a51, 0x11a5b}, 2542 {0x11a8a, 0x11a99}, 2543 {0x11c2f, 0x11c36}, 2544 {0x11c38, 0x11c3f}, 2545 {0x11c92, 0x11ca7}, 2546 {0x11ca9, 0x11cb6}, 2547 {0x11d31, 0x11d36}, 2548 {0x11d3a, 0x11d3a}, 2549 {0x11d3c, 0x11d3d}, 2550 {0x11d3f, 0x11d45}, 2551 {0x11d47, 0x11d47}, 2552 {0x11d8a, 0x11d8e}, 2553 {0x11d90, 0x11d91}, 2554 {0x11d93, 0x11d97}, 2555 {0x11ef3, 0x11ef6}, 2556 {0x16af0, 0x16af4}, 2557 {0x16b30, 0x16b36}, 2558 {0x16f4f, 0x16f4f}, 2559 {0x16f51, 0x16f87}, 2560 {0x16f8f, 0x16f92}, 2561 {0x1bc9d, 0x1bc9e}, 2562 {0x1d165, 0x1d169}, 2563 {0x1d16d, 0x1d172}, 2564 {0x1d17b, 0x1d182}, 2565 {0x1d185, 0x1d18b}, 2566 {0x1d1aa, 0x1d1ad}, 2567 {0x1d242, 0x1d244}, 2568 {0x1da00, 0x1da36}, 2569 {0x1da3b, 0x1da6c}, 2570 {0x1da75, 0x1da75}, 2571 {0x1da84, 0x1da84}, 2572 {0x1da9b, 0x1da9f}, 2573 {0x1daa1, 0x1daaf}, 2574 {0x1e000, 0x1e006}, 2575 {0x1e008, 0x1e018}, 2576 {0x1e01b, 0x1e021}, 2577 {0x1e023, 0x1e024}, 2578 {0x1e026, 0x1e02a}, 2579 {0x1e130, 0x1e136}, 2580 {0x1e2ec, 0x1e2ef}, 2581 {0x1e8d0, 0x1e8d6}, 2582 {0x1e944, 0x1e94a}, 2583 {0xe0100, 0xe01ef} 2584 }; 2585 2586 return intable(combining, sizeof(combining), c); 2587 } 2588 2589 /* 2590 * Return TRUE for characters that can be displayed in a normal way. 2591 * Only for characters of 0x100 and above! 2592 */ 2593 int 2594 utf_printable(int c) 2595 { 2596 #ifdef USE_WCHAR_FUNCTIONS 2597 /* 2598 * Assume the iswprint() library function works better than our own stuff. 2599 */ 2600 return iswprint(c); 2601 #else 2602 // Sorted list of non-overlapping intervals. 2603 // 0xd800-0xdfff is reserved for UTF-16, actually illegal. 2604 static struct interval nonprint[] = 2605 { 2606 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2607 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2608 {0xfffe, 0xffff} 2609 }; 2610 2611 return !intable(nonprint, sizeof(nonprint), c); 2612 #endif 2613 } 2614 2615 // Sorted list of non-overlapping intervals of all Emoji characters, 2616 // based on http://unicode.org/emoji/charts/emoji-list.html 2617 static struct interval emoji_all[] = 2618 { 2619 {0x203c, 0x203c}, 2620 {0x2049, 0x2049}, 2621 {0x2122, 0x2122}, 2622 {0x2139, 0x2139}, 2623 {0x2194, 0x2199}, 2624 {0x21a9, 0x21aa}, 2625 {0x231a, 0x231b}, 2626 {0x2328, 0x2328}, 2627 {0x23cf, 0x23cf}, 2628 {0x23e9, 0x23f3}, 2629 {0x23f8, 0x23fa}, 2630 {0x24c2, 0x24c2}, 2631 {0x25aa, 0x25ab}, 2632 {0x25b6, 0x25b6}, 2633 {0x25c0, 0x25c0}, 2634 {0x25fb, 0x25fe}, 2635 {0x2600, 0x2604}, 2636 {0x260e, 0x260e}, 2637 {0x2611, 0x2611}, 2638 {0x2614, 0x2615}, 2639 {0x2618, 0x2618}, 2640 {0x261d, 0x261d}, 2641 {0x2620, 0x2620}, 2642 {0x2622, 0x2623}, 2643 {0x2626, 0x2626}, 2644 {0x262a, 0x262a}, 2645 {0x262e, 0x262f}, 2646 {0x2638, 0x263a}, 2647 {0x2640, 0x2640}, 2648 {0x2642, 0x2642}, 2649 {0x2648, 0x2653}, 2650 {0x265f, 0x2660}, 2651 {0x2663, 0x2663}, 2652 {0x2665, 0x2666}, 2653 {0x2668, 0x2668}, 2654 {0x267b, 0x267b}, 2655 {0x267e, 0x267f}, 2656 {0x2692, 0x2697}, 2657 {0x2699, 0x2699}, 2658 {0x269b, 0x269c}, 2659 {0x26a0, 0x26a1}, 2660 {0x26aa, 0x26ab}, 2661 {0x26b0, 0x26b1}, 2662 {0x26bd, 0x26be}, 2663 {0x26c4, 0x26c5}, 2664 {0x26c8, 0x26c8}, 2665 {0x26ce, 0x26cf}, 2666 {0x26d1, 0x26d1}, 2667 {0x26d3, 0x26d4}, 2668 {0x26e9, 0x26ea}, 2669 {0x26f0, 0x26f5}, 2670 {0x26f7, 0x26fa}, 2671 {0x26fd, 0x26fd}, 2672 {0x2702, 0x2702}, 2673 {0x2705, 0x2705}, 2674 {0x2708, 0x270d}, 2675 {0x270f, 0x270f}, 2676 {0x2712, 0x2712}, 2677 {0x2714, 0x2714}, 2678 {0x2716, 0x2716}, 2679 {0x271d, 0x271d}, 2680 {0x2721, 0x2721}, 2681 {0x2728, 0x2728}, 2682 {0x2733, 0x2734}, 2683 {0x2744, 0x2744}, 2684 {0x2747, 0x2747}, 2685 {0x274c, 0x274c}, 2686 {0x274e, 0x274e}, 2687 {0x2753, 0x2755}, 2688 {0x2757, 0x2757}, 2689 {0x2763, 0x2764}, 2690 {0x2795, 0x2797}, 2691 {0x27a1, 0x27a1}, 2692 {0x27b0, 0x27b0}, 2693 {0x27bf, 0x27bf}, 2694 {0x2934, 0x2935}, 2695 {0x2b05, 0x2b07}, 2696 {0x2b1b, 0x2b1c}, 2697 {0x2b50, 0x2b50}, 2698 {0x2b55, 0x2b55}, 2699 {0x3030, 0x3030}, 2700 {0x303d, 0x303d}, 2701 {0x3297, 0x3297}, 2702 {0x3299, 0x3299}, 2703 {0x1f004, 0x1f004}, 2704 {0x1f0cf, 0x1f0cf}, 2705 {0x1f170, 0x1f171}, 2706 {0x1f17e, 0x1f17f}, 2707 {0x1f18e, 0x1f18e}, 2708 {0x1f191, 0x1f19a}, 2709 {0x1f1e6, 0x1f1ff}, 2710 {0x1f201, 0x1f202}, 2711 {0x1f21a, 0x1f21a}, 2712 {0x1f22f, 0x1f22f}, 2713 {0x1f232, 0x1f23a}, 2714 {0x1f250, 0x1f251}, 2715 {0x1f300, 0x1f321}, 2716 {0x1f324, 0x1f393}, 2717 {0x1f396, 0x1f397}, 2718 {0x1f399, 0x1f39b}, 2719 {0x1f39e, 0x1f3f0}, 2720 {0x1f3f3, 0x1f3f5}, 2721 {0x1f3f7, 0x1f4fd}, 2722 {0x1f4ff, 0x1f53d}, 2723 {0x1f549, 0x1f54e}, 2724 {0x1f550, 0x1f567}, 2725 {0x1f56f, 0x1f570}, 2726 {0x1f573, 0x1f57a}, 2727 {0x1f587, 0x1f587}, 2728 {0x1f58a, 0x1f58d}, 2729 {0x1f590, 0x1f590}, 2730 {0x1f595, 0x1f596}, 2731 {0x1f5a4, 0x1f5a5}, 2732 {0x1f5a8, 0x1f5a8}, 2733 {0x1f5b1, 0x1f5b2}, 2734 {0x1f5bc, 0x1f5bc}, 2735 {0x1f5c2, 0x1f5c4}, 2736 {0x1f5d1, 0x1f5d3}, 2737 {0x1f5dc, 0x1f5de}, 2738 {0x1f5e1, 0x1f5e1}, 2739 {0x1f5e3, 0x1f5e3}, 2740 {0x1f5e8, 0x1f5e8}, 2741 {0x1f5ef, 0x1f5ef}, 2742 {0x1f5f3, 0x1f5f3}, 2743 {0x1f5fa, 0x1f64f}, 2744 {0x1f680, 0x1f6c5}, 2745 {0x1f6cb, 0x1f6d2}, 2746 {0x1f6e0, 0x1f6e5}, 2747 {0x1f6e9, 0x1f6e9}, 2748 {0x1f6eb, 0x1f6ec}, 2749 {0x1f6f0, 0x1f6f0}, 2750 {0x1f6f3, 0x1f6f9}, 2751 {0x1f910, 0x1f93a}, 2752 {0x1f93c, 0x1f93e}, 2753 {0x1f940, 0x1f945}, 2754 {0x1f947, 0x1f970}, 2755 {0x1f973, 0x1f976}, 2756 {0x1f97a, 0x1f97a}, 2757 {0x1f97c, 0x1f9a2}, 2758 {0x1f9b0, 0x1f9b9}, 2759 {0x1f9c0, 0x1f9c2}, 2760 {0x1f9d0, 0x1f9ff} 2761 }; 2762 2763 /* 2764 * Get class of a Unicode character. 2765 * 0: white space 2766 * 1: punctuation 2767 * 2 or bigger: some class of word character. 2768 */ 2769 int 2770 utf_class(int c) 2771 { 2772 return utf_class_buf(c, curbuf); 2773 } 2774 2775 int 2776 utf_class_buf(int c, buf_T *buf) 2777 { 2778 // sorted list of non-overlapping intervals 2779 static struct clinterval 2780 { 2781 unsigned int first; 2782 unsigned int last; 2783 unsigned int class; 2784 } classes[] = 2785 { 2786 {0x037e, 0x037e, 1}, // Greek question mark 2787 {0x0387, 0x0387, 1}, // Greek ano teleia 2788 {0x055a, 0x055f, 1}, // Armenian punctuation 2789 {0x0589, 0x0589, 1}, // Armenian full stop 2790 {0x05be, 0x05be, 1}, 2791 {0x05c0, 0x05c0, 1}, 2792 {0x05c3, 0x05c3, 1}, 2793 {0x05f3, 0x05f4, 1}, 2794 {0x060c, 0x060c, 1}, 2795 {0x061b, 0x061b, 1}, 2796 {0x061f, 0x061f, 1}, 2797 {0x066a, 0x066d, 1}, 2798 {0x06d4, 0x06d4, 1}, 2799 {0x0700, 0x070d, 1}, // Syriac punctuation 2800 {0x0964, 0x0965, 1}, 2801 {0x0970, 0x0970, 1}, 2802 {0x0df4, 0x0df4, 1}, 2803 {0x0e4f, 0x0e4f, 1}, 2804 {0x0e5a, 0x0e5b, 1}, 2805 {0x0f04, 0x0f12, 1}, 2806 {0x0f3a, 0x0f3d, 1}, 2807 {0x0f85, 0x0f85, 1}, 2808 {0x104a, 0x104f, 1}, // Myanmar punctuation 2809 {0x10fb, 0x10fb, 1}, // Georgian punctuation 2810 {0x1361, 0x1368, 1}, // Ethiopic punctuation 2811 {0x166d, 0x166e, 1}, // Canadian Syl. punctuation 2812 {0x1680, 0x1680, 0}, 2813 {0x169b, 0x169c, 1}, 2814 {0x16eb, 0x16ed, 1}, 2815 {0x1735, 0x1736, 1}, 2816 {0x17d4, 0x17dc, 1}, // Khmer punctuation 2817 {0x1800, 0x180a, 1}, // Mongolian punctuation 2818 {0x2000, 0x200b, 0}, // spaces 2819 {0x200c, 0x2027, 1}, // punctuation and symbols 2820 {0x2028, 0x2029, 0}, 2821 {0x202a, 0x202e, 1}, // punctuation and symbols 2822 {0x202f, 0x202f, 0}, 2823 {0x2030, 0x205e, 1}, // punctuation and symbols 2824 {0x205f, 0x205f, 0}, 2825 {0x2060, 0x27ff, 1}, // punctuation and symbols 2826 {0x2070, 0x207f, 0x2070}, // superscript 2827 {0x2080, 0x2094, 0x2080}, // subscript 2828 {0x20a0, 0x27ff, 1}, // all kinds of symbols 2829 {0x2800, 0x28ff, 0x2800}, // braille 2830 {0x2900, 0x2998, 1}, // arrows, brackets, etc. 2831 {0x29d8, 0x29db, 1}, 2832 {0x29fc, 0x29fd, 1}, 2833 {0x2e00, 0x2e7f, 1}, // supplemental punctuation 2834 {0x3000, 0x3000, 0}, // ideographic space 2835 {0x3001, 0x3020, 1}, // ideographic punctuation 2836 {0x3030, 0x3030, 1}, 2837 {0x303d, 0x303d, 1}, 2838 {0x3040, 0x309f, 0x3040}, // Hiragana 2839 {0x30a0, 0x30ff, 0x30a0}, // Katakana 2840 {0x3300, 0x9fff, 0x4e00}, // CJK Ideographs 2841 {0xac00, 0xd7a3, 0xac00}, // Hangul Syllables 2842 {0xf900, 0xfaff, 0x4e00}, // CJK Ideographs 2843 {0xfd3e, 0xfd3f, 1}, 2844 {0xfe30, 0xfe6b, 1}, // punctuation forms 2845 {0xff00, 0xff0f, 1}, // half/fullwidth ASCII 2846 {0xff1a, 0xff20, 1}, // half/fullwidth ASCII 2847 {0xff3b, 0xff40, 1}, // half/fullwidth ASCII 2848 {0xff5b, 0xff65, 1}, // half/fullwidth ASCII 2849 {0x1d000, 0x1d24f, 1}, // Musical notation 2850 {0x1d400, 0x1d7ff, 1}, // Mathematical Alphanumeric Symbols 2851 {0x1f000, 0x1f2ff, 1}, // Game pieces; enclosed characters 2852 {0x1f300, 0x1f9ff, 1}, // Many symbol blocks 2853 {0x20000, 0x2a6df, 0x4e00}, // CJK Ideographs 2854 {0x2a700, 0x2b73f, 0x4e00}, // CJK Ideographs 2855 {0x2b740, 0x2b81f, 0x4e00}, // CJK Ideographs 2856 {0x2f800, 0x2fa1f, 0x4e00}, // CJK Ideographs 2857 }; 2858 2859 int bot = 0; 2860 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2861 int mid; 2862 2863 // First quick check for Latin1 characters, use 'iskeyword'. 2864 if (c < 0x100) 2865 { 2866 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2867 return 0; // blank 2868 if (vim_iswordc_buf(c, buf)) 2869 return 2; // word character 2870 return 1; // punctuation 2871 } 2872 2873 // binary search in table 2874 while (top >= bot) 2875 { 2876 mid = (bot + top) / 2; 2877 if (classes[mid].last < (unsigned int)c) 2878 bot = mid + 1; 2879 else if (classes[mid].first > (unsigned int)c) 2880 top = mid - 1; 2881 else 2882 return (int)classes[mid].class; 2883 } 2884 2885 // emoji 2886 if (intable(emoji_all, sizeof(emoji_all), c)) 2887 return 3; 2888 2889 // most other characters are "word" characters 2890 return 2; 2891 } 2892 2893 int 2894 utf_ambiguous_width(int c) 2895 { 2896 return c >= 0x80 && (intable(ambiguous, sizeof(ambiguous), c) 2897 || intable(emoji_all, sizeof(emoji_all), c)); 2898 } 2899 2900 /* 2901 * Code for Unicode case-dependent operations. Based on notes in 2902 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2903 * This code uses simple case folding, not full case folding. 2904 * Last updated for Unicode 5.2. 2905 */ 2906 2907 /* 2908 * The following tables are built by ../runtime/tools/unicode.vim. 2909 * They must be in numeric order, because we use binary search. 2910 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2911 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2912 * folded/upper/lower by adding 32. 2913 */ 2914 typedef struct 2915 { 2916 int rangeStart; 2917 int rangeEnd; 2918 int step; 2919 int offset; 2920 } convertStruct; 2921 2922 static convertStruct foldCase[] = 2923 { 2924 {0x41,0x5a,1,32}, 2925 {0xb5,0xb5,-1,775}, 2926 {0xc0,0xd6,1,32}, 2927 {0xd8,0xde,1,32}, 2928 {0x100,0x12e,2,1}, 2929 {0x132,0x136,2,1}, 2930 {0x139,0x147,2,1}, 2931 {0x14a,0x176,2,1}, 2932 {0x178,0x178,-1,-121}, 2933 {0x179,0x17d,2,1}, 2934 {0x17f,0x17f,-1,-268}, 2935 {0x181,0x181,-1,210}, 2936 {0x182,0x184,2,1}, 2937 {0x186,0x186,-1,206}, 2938 {0x187,0x187,-1,1}, 2939 {0x189,0x18a,1,205}, 2940 {0x18b,0x18b,-1,1}, 2941 {0x18e,0x18e,-1,79}, 2942 {0x18f,0x18f,-1,202}, 2943 {0x190,0x190,-1,203}, 2944 {0x191,0x191,-1,1}, 2945 {0x193,0x193,-1,205}, 2946 {0x194,0x194,-1,207}, 2947 {0x196,0x196,-1,211}, 2948 {0x197,0x197,-1,209}, 2949 {0x198,0x198,-1,1}, 2950 {0x19c,0x19c,-1,211}, 2951 {0x19d,0x19d,-1,213}, 2952 {0x19f,0x19f,-1,214}, 2953 {0x1a0,0x1a4,2,1}, 2954 {0x1a6,0x1a6,-1,218}, 2955 {0x1a7,0x1a7,-1,1}, 2956 {0x1a9,0x1a9,-1,218}, 2957 {0x1ac,0x1ac,-1,1}, 2958 {0x1ae,0x1ae,-1,218}, 2959 {0x1af,0x1af,-1,1}, 2960 {0x1b1,0x1b2,1,217}, 2961 {0x1b3,0x1b5,2,1}, 2962 {0x1b7,0x1b7,-1,219}, 2963 {0x1b8,0x1bc,4,1}, 2964 {0x1c4,0x1c4,-1,2}, 2965 {0x1c5,0x1c5,-1,1}, 2966 {0x1c7,0x1c7,-1,2}, 2967 {0x1c8,0x1c8,-1,1}, 2968 {0x1ca,0x1ca,-1,2}, 2969 {0x1cb,0x1db,2,1}, 2970 {0x1de,0x1ee,2,1}, 2971 {0x1f1,0x1f1,-1,2}, 2972 {0x1f2,0x1f4,2,1}, 2973 {0x1f6,0x1f6,-1,-97}, 2974 {0x1f7,0x1f7,-1,-56}, 2975 {0x1f8,0x21e,2,1}, 2976 {0x220,0x220,-1,-130}, 2977 {0x222,0x232,2,1}, 2978 {0x23a,0x23a,-1,10795}, 2979 {0x23b,0x23b,-1,1}, 2980 {0x23d,0x23d,-1,-163}, 2981 {0x23e,0x23e,-1,10792}, 2982 {0x241,0x241,-1,1}, 2983 {0x243,0x243,-1,-195}, 2984 {0x244,0x244,-1,69}, 2985 {0x245,0x245,-1,71}, 2986 {0x246,0x24e,2,1}, 2987 {0x345,0x345,-1,116}, 2988 {0x370,0x372,2,1}, 2989 {0x376,0x376,-1,1}, 2990 {0x37f,0x37f,-1,116}, 2991 {0x386,0x386,-1,38}, 2992 {0x388,0x38a,1,37}, 2993 {0x38c,0x38c,-1,64}, 2994 {0x38e,0x38f,1,63}, 2995 {0x391,0x3a1,1,32}, 2996 {0x3a3,0x3ab,1,32}, 2997 {0x3c2,0x3c2,-1,1}, 2998 {0x3cf,0x3cf,-1,8}, 2999 {0x3d0,0x3d0,-1,-30}, 3000 {0x3d1,0x3d1,-1,-25}, 3001 {0x3d5,0x3d5,-1,-15}, 3002 {0x3d6,0x3d6,-1,-22}, 3003 {0x3d8,0x3ee,2,1}, 3004 {0x3f0,0x3f0,-1,-54}, 3005 {0x3f1,0x3f1,-1,-48}, 3006 {0x3f4,0x3f4,-1,-60}, 3007 {0x3f5,0x3f5,-1,-64}, 3008 {0x3f7,0x3f7,-1,1}, 3009 {0x3f9,0x3f9,-1,-7}, 3010 {0x3fa,0x3fa,-1,1}, 3011 {0x3fd,0x3ff,1,-130}, 3012 {0x400,0x40f,1,80}, 3013 {0x410,0x42f,1,32}, 3014 {0x460,0x480,2,1}, 3015 {0x48a,0x4be,2,1}, 3016 {0x4c0,0x4c0,-1,15}, 3017 {0x4c1,0x4cd,2,1}, 3018 {0x4d0,0x52e,2,1}, 3019 {0x531,0x556,1,48}, 3020 {0x10a0,0x10c5,1,7264}, 3021 {0x10c7,0x10cd,6,7264}, 3022 {0x13f8,0x13fd,1,-8}, 3023 {0x1c80,0x1c80,-1,-6222}, 3024 {0x1c81,0x1c81,-1,-6221}, 3025 {0x1c82,0x1c82,-1,-6212}, 3026 {0x1c83,0x1c84,1,-6210}, 3027 {0x1c85,0x1c85,-1,-6211}, 3028 {0x1c86,0x1c86,-1,-6204}, 3029 {0x1c87,0x1c87,-1,-6180}, 3030 {0x1c88,0x1c88,-1,35267}, 3031 {0x1c90,0x1cba,1,-3008}, 3032 {0x1cbd,0x1cbf,1,-3008}, 3033 {0x1e00,0x1e94,2,1}, 3034 {0x1e9b,0x1e9b,-1,-58}, 3035 {0x1e9e,0x1e9e,-1,-7615}, 3036 {0x1ea0,0x1efe,2,1}, 3037 {0x1f08,0x1f0f,1,-8}, 3038 {0x1f18,0x1f1d,1,-8}, 3039 {0x1f28,0x1f2f,1,-8}, 3040 {0x1f38,0x1f3f,1,-8}, 3041 {0x1f48,0x1f4d,1,-8}, 3042 {0x1f59,0x1f5f,2,-8}, 3043 {0x1f68,0x1f6f,1,-8}, 3044 {0x1f88,0x1f8f,1,-8}, 3045 {0x1f98,0x1f9f,1,-8}, 3046 {0x1fa8,0x1faf,1,-8}, 3047 {0x1fb8,0x1fb9,1,-8}, 3048 {0x1fba,0x1fbb,1,-74}, 3049 {0x1fbc,0x1fbc,-1,-9}, 3050 {0x1fbe,0x1fbe,-1,-7173}, 3051 {0x1fc8,0x1fcb,1,-86}, 3052 {0x1fcc,0x1fcc,-1,-9}, 3053 {0x1fd8,0x1fd9,1,-8}, 3054 {0x1fda,0x1fdb,1,-100}, 3055 {0x1fe8,0x1fe9,1,-8}, 3056 {0x1fea,0x1feb,1,-112}, 3057 {0x1fec,0x1fec,-1,-7}, 3058 {0x1ff8,0x1ff9,1,-128}, 3059 {0x1ffa,0x1ffb,1,-126}, 3060 {0x1ffc,0x1ffc,-1,-9}, 3061 {0x2126,0x2126,-1,-7517}, 3062 {0x212a,0x212a,-1,-8383}, 3063 {0x212b,0x212b,-1,-8262}, 3064 {0x2132,0x2132,-1,28}, 3065 {0x2160,0x216f,1,16}, 3066 {0x2183,0x2183,-1,1}, 3067 {0x24b6,0x24cf,1,26}, 3068 {0x2c00,0x2c2e,1,48}, 3069 {0x2c60,0x2c60,-1,1}, 3070 {0x2c62,0x2c62,-1,-10743}, 3071 {0x2c63,0x2c63,-1,-3814}, 3072 {0x2c64,0x2c64,-1,-10727}, 3073 {0x2c67,0x2c6b,2,1}, 3074 {0x2c6d,0x2c6d,-1,-10780}, 3075 {0x2c6e,0x2c6e,-1,-10749}, 3076 {0x2c6f,0x2c6f,-1,-10783}, 3077 {0x2c70,0x2c70,-1,-10782}, 3078 {0x2c72,0x2c75,3,1}, 3079 {0x2c7e,0x2c7f,1,-10815}, 3080 {0x2c80,0x2ce2,2,1}, 3081 {0x2ceb,0x2ced,2,1}, 3082 {0x2cf2,0xa640,31054,1}, 3083 {0xa642,0xa66c,2,1}, 3084 {0xa680,0xa69a,2,1}, 3085 {0xa722,0xa72e,2,1}, 3086 {0xa732,0xa76e,2,1}, 3087 {0xa779,0xa77b,2,1}, 3088 {0xa77d,0xa77d,-1,-35332}, 3089 {0xa77e,0xa786,2,1}, 3090 {0xa78b,0xa78b,-1,1}, 3091 {0xa78d,0xa78d,-1,-42280}, 3092 {0xa790,0xa792,2,1}, 3093 {0xa796,0xa7a8,2,1}, 3094 {0xa7aa,0xa7aa,-1,-42308}, 3095 {0xa7ab,0xa7ab,-1,-42319}, 3096 {0xa7ac,0xa7ac,-1,-42315}, 3097 {0xa7ad,0xa7ad,-1,-42305}, 3098 {0xa7ae,0xa7ae,-1,-42308}, 3099 {0xa7b0,0xa7b0,-1,-42258}, 3100 {0xa7b1,0xa7b1,-1,-42282}, 3101 {0xa7b2,0xa7b2,-1,-42261}, 3102 {0xa7b3,0xa7b3,-1,928}, 3103 {0xa7b4,0xa7be,2,1}, 3104 {0xa7c2,0xa7c2,-1,1}, 3105 {0xa7c4,0xa7c4,-1,-48}, 3106 {0xa7c5,0xa7c5,-1,-42307}, 3107 {0xa7c6,0xa7c6,-1,-35384}, 3108 {0xab70,0xabbf,1,-38864}, 3109 {0xff21,0xff3a,1,32}, 3110 {0x10400,0x10427,1,40}, 3111 {0x104b0,0x104d3,1,40}, 3112 {0x10c80,0x10cb2,1,64}, 3113 {0x118a0,0x118bf,1,32}, 3114 {0x16e40,0x16e5f,1,32}, 3115 {0x1e900,0x1e921,1,34} 3116 }; 3117 3118 /* 3119 * Generic conversion function for case operations. 3120 * Return the converted equivalent of "a", which is a UCS-4 character. Use 3121 * the given conversion "table". Uses binary search on "table". 3122 */ 3123 static int 3124 utf_convert( 3125 int a, 3126 convertStruct table[], 3127 int tableSize) 3128 { 3129 int start, mid, end; // indices into table 3130 int entries = tableSize / sizeof(convertStruct); 3131 3132 start = 0; 3133 end = entries; 3134 while (start < end) 3135 { 3136 // need to search further 3137 mid = (end + start) / 2; 3138 if (table[mid].rangeEnd < a) 3139 start = mid + 1; 3140 else 3141 end = mid; 3142 } 3143 if (start < entries 3144 && table[start].rangeStart <= a 3145 && a <= table[start].rangeEnd 3146 && (a - table[start].rangeStart) % table[start].step == 0) 3147 return (a + table[start].offset); 3148 else 3149 return a; 3150 } 3151 3152 /* 3153 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 3154 * simple case folding. 3155 */ 3156 int 3157 utf_fold(int a) 3158 { 3159 if (a < 0x80) 3160 // be fast for ASCII 3161 return a >= 0x41 && a <= 0x5a ? a + 32 : a; 3162 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 3163 } 3164 3165 static convertStruct toLower[] = 3166 { 3167 {0x41,0x5a,1,32}, 3168 {0xc0,0xd6,1,32}, 3169 {0xd8,0xde,1,32}, 3170 {0x100,0x12e,2,1}, 3171 {0x130,0x130,-1,-199}, 3172 {0x132,0x136,2,1}, 3173 {0x139,0x147,2,1}, 3174 {0x14a,0x176,2,1}, 3175 {0x178,0x178,-1,-121}, 3176 {0x179,0x17d,2,1}, 3177 {0x181,0x181,-1,210}, 3178 {0x182,0x184,2,1}, 3179 {0x186,0x186,-1,206}, 3180 {0x187,0x187,-1,1}, 3181 {0x189,0x18a,1,205}, 3182 {0x18b,0x18b,-1,1}, 3183 {0x18e,0x18e,-1,79}, 3184 {0x18f,0x18f,-1,202}, 3185 {0x190,0x190,-1,203}, 3186 {0x191,0x191,-1,1}, 3187 {0x193,0x193,-1,205}, 3188 {0x194,0x194,-1,207}, 3189 {0x196,0x196,-1,211}, 3190 {0x197,0x197,-1,209}, 3191 {0x198,0x198,-1,1}, 3192 {0x19c,0x19c,-1,211}, 3193 {0x19d,0x19d,-1,213}, 3194 {0x19f,0x19f,-1,214}, 3195 {0x1a0,0x1a4,2,1}, 3196 {0x1a6,0x1a6,-1,218}, 3197 {0x1a7,0x1a7,-1,1}, 3198 {0x1a9,0x1a9,-1,218}, 3199 {0x1ac,0x1ac,-1,1}, 3200 {0x1ae,0x1ae,-1,218}, 3201 {0x1af,0x1af,-1,1}, 3202 {0x1b1,0x1b2,1,217}, 3203 {0x1b3,0x1b5,2,1}, 3204 {0x1b7,0x1b7,-1,219}, 3205 {0x1b8,0x1bc,4,1}, 3206 {0x1c4,0x1c4,-1,2}, 3207 {0x1c5,0x1c5,-1,1}, 3208 {0x1c7,0x1c7,-1,2}, 3209 {0x1c8,0x1c8,-1,1}, 3210 {0x1ca,0x1ca,-1,2}, 3211 {0x1cb,0x1db,2,1}, 3212 {0x1de,0x1ee,2,1}, 3213 {0x1f1,0x1f1,-1,2}, 3214 {0x1f2,0x1f4,2,1}, 3215 {0x1f6,0x1f6,-1,-97}, 3216 {0x1f7,0x1f7,-1,-56}, 3217 {0x1f8,0x21e,2,1}, 3218 {0x220,0x220,-1,-130}, 3219 {0x222,0x232,2,1}, 3220 {0x23a,0x23a,-1,10795}, 3221 {0x23b,0x23b,-1,1}, 3222 {0x23d,0x23d,-1,-163}, 3223 {0x23e,0x23e,-1,10792}, 3224 {0x241,0x241,-1,1}, 3225 {0x243,0x243,-1,-195}, 3226 {0x244,0x244,-1,69}, 3227 {0x245,0x245,-1,71}, 3228 {0x246,0x24e,2,1}, 3229 {0x370,0x372,2,1}, 3230 {0x376,0x376,-1,1}, 3231 {0x37f,0x37f,-1,116}, 3232 {0x386,0x386,-1,38}, 3233 {0x388,0x38a,1,37}, 3234 {0x38c,0x38c,-1,64}, 3235 {0x38e,0x38f,1,63}, 3236 {0x391,0x3a1,1,32}, 3237 {0x3a3,0x3ab,1,32}, 3238 {0x3cf,0x3cf,-1,8}, 3239 {0x3d8,0x3ee,2,1}, 3240 {0x3f4,0x3f4,-1,-60}, 3241 {0x3f7,0x3f7,-1,1}, 3242 {0x3f9,0x3f9,-1,-7}, 3243 {0x3fa,0x3fa,-1,1}, 3244 {0x3fd,0x3ff,1,-130}, 3245 {0x400,0x40f,1,80}, 3246 {0x410,0x42f,1,32}, 3247 {0x460,0x480,2,1}, 3248 {0x48a,0x4be,2,1}, 3249 {0x4c0,0x4c0,-1,15}, 3250 {0x4c1,0x4cd,2,1}, 3251 {0x4d0,0x52e,2,1}, 3252 {0x531,0x556,1,48}, 3253 {0x10a0,0x10c5,1,7264}, 3254 {0x10c7,0x10cd,6,7264}, 3255 {0x13a0,0x13ef,1,38864}, 3256 {0x13f0,0x13f5,1,8}, 3257 {0x1c90,0x1cba,1,-3008}, 3258 {0x1cbd,0x1cbf,1,-3008}, 3259 {0x1e00,0x1e94,2,1}, 3260 {0x1e9e,0x1e9e,-1,-7615}, 3261 {0x1ea0,0x1efe,2,1}, 3262 {0x1f08,0x1f0f,1,-8}, 3263 {0x1f18,0x1f1d,1,-8}, 3264 {0x1f28,0x1f2f,1,-8}, 3265 {0x1f38,0x1f3f,1,-8}, 3266 {0x1f48,0x1f4d,1,-8}, 3267 {0x1f59,0x1f5f,2,-8}, 3268 {0x1f68,0x1f6f,1,-8}, 3269 {0x1f88,0x1f8f,1,-8}, 3270 {0x1f98,0x1f9f,1,-8}, 3271 {0x1fa8,0x1faf,1,-8}, 3272 {0x1fb8,0x1fb9,1,-8}, 3273 {0x1fba,0x1fbb,1,-74}, 3274 {0x1fbc,0x1fbc,-1,-9}, 3275 {0x1fc8,0x1fcb,1,-86}, 3276 {0x1fcc,0x1fcc,-1,-9}, 3277 {0x1fd8,0x1fd9,1,-8}, 3278 {0x1fda,0x1fdb,1,-100}, 3279 {0x1fe8,0x1fe9,1,-8}, 3280 {0x1fea,0x1feb,1,-112}, 3281 {0x1fec,0x1fec,-1,-7}, 3282 {0x1ff8,0x1ff9,1,-128}, 3283 {0x1ffa,0x1ffb,1,-126}, 3284 {0x1ffc,0x1ffc,-1,-9}, 3285 {0x2126,0x2126,-1,-7517}, 3286 {0x212a,0x212a,-1,-8383}, 3287 {0x212b,0x212b,-1,-8262}, 3288 {0x2132,0x2132,-1,28}, 3289 {0x2160,0x216f,1,16}, 3290 {0x2183,0x2183,-1,1}, 3291 {0x24b6,0x24cf,1,26}, 3292 {0x2c00,0x2c2e,1,48}, 3293 {0x2c60,0x2c60,-1,1}, 3294 {0x2c62,0x2c62,-1,-10743}, 3295 {0x2c63,0x2c63,-1,-3814}, 3296 {0x2c64,0x2c64,-1,-10727}, 3297 {0x2c67,0x2c6b,2,1}, 3298 {0x2c6d,0x2c6d,-1,-10780}, 3299 {0x2c6e,0x2c6e,-1,-10749}, 3300 {0x2c6f,0x2c6f,-1,-10783}, 3301 {0x2c70,0x2c70,-1,-10782}, 3302 {0x2c72,0x2c75,3,1}, 3303 {0x2c7e,0x2c7f,1,-10815}, 3304 {0x2c80,0x2ce2,2,1}, 3305 {0x2ceb,0x2ced,2,1}, 3306 {0x2cf2,0xa640,31054,1}, 3307 {0xa642,0xa66c,2,1}, 3308 {0xa680,0xa69a,2,1}, 3309 {0xa722,0xa72e,2,1}, 3310 {0xa732,0xa76e,2,1}, 3311 {0xa779,0xa77b,2,1}, 3312 {0xa77d,0xa77d,-1,-35332}, 3313 {0xa77e,0xa786,2,1}, 3314 {0xa78b,0xa78b,-1,1}, 3315 {0xa78d,0xa78d,-1,-42280}, 3316 {0xa790,0xa792,2,1}, 3317 {0xa796,0xa7a8,2,1}, 3318 {0xa7aa,0xa7aa,-1,-42308}, 3319 {0xa7ab,0xa7ab,-1,-42319}, 3320 {0xa7ac,0xa7ac,-1,-42315}, 3321 {0xa7ad,0xa7ad,-1,-42305}, 3322 {0xa7ae,0xa7ae,-1,-42308}, 3323 {0xa7b0,0xa7b0,-1,-42258}, 3324 {0xa7b1,0xa7b1,-1,-42282}, 3325 {0xa7b2,0xa7b2,-1,-42261}, 3326 {0xa7b3,0xa7b3,-1,928}, 3327 {0xa7b4,0xa7be,2,1}, 3328 {0xa7c2,0xa7c2,-1,1}, 3329 {0xa7c4,0xa7c4,-1,-48}, 3330 {0xa7c5,0xa7c5,-1,-42307}, 3331 {0xa7c6,0xa7c6,-1,-35384}, 3332 {0xff21,0xff3a,1,32}, 3333 {0x10400,0x10427,1,40}, 3334 {0x104b0,0x104d3,1,40}, 3335 {0x10c80,0x10cb2,1,64}, 3336 {0x118a0,0x118bf,1,32}, 3337 {0x16e40,0x16e5f,1,32}, 3338 {0x1e900,0x1e921,1,34} 3339 }; 3340 3341 static convertStruct toUpper[] = 3342 { 3343 {0x61,0x7a,1,-32}, 3344 {0xb5,0xb5,-1,743}, 3345 {0xe0,0xf6,1,-32}, 3346 {0xf8,0xfe,1,-32}, 3347 {0xff,0xff,-1,121}, 3348 {0x101,0x12f,2,-1}, 3349 {0x131,0x131,-1,-232}, 3350 {0x133,0x137,2,-1}, 3351 {0x13a,0x148,2,-1}, 3352 {0x14b,0x177,2,-1}, 3353 {0x17a,0x17e,2,-1}, 3354 {0x17f,0x17f,-1,-300}, 3355 {0x180,0x180,-1,195}, 3356 {0x183,0x185,2,-1}, 3357 {0x188,0x18c,4,-1}, 3358 {0x192,0x192,-1,-1}, 3359 {0x195,0x195,-1,97}, 3360 {0x199,0x199,-1,-1}, 3361 {0x19a,0x19a,-1,163}, 3362 {0x19e,0x19e,-1,130}, 3363 {0x1a1,0x1a5,2,-1}, 3364 {0x1a8,0x1ad,5,-1}, 3365 {0x1b0,0x1b4,4,-1}, 3366 {0x1b6,0x1b9,3,-1}, 3367 {0x1bd,0x1bd,-1,-1}, 3368 {0x1bf,0x1bf,-1,56}, 3369 {0x1c5,0x1c5,-1,-1}, 3370 {0x1c6,0x1c6,-1,-2}, 3371 {0x1c8,0x1c8,-1,-1}, 3372 {0x1c9,0x1c9,-1,-2}, 3373 {0x1cb,0x1cb,-1,-1}, 3374 {0x1cc,0x1cc,-1,-2}, 3375 {0x1ce,0x1dc,2,-1}, 3376 {0x1dd,0x1dd,-1,-79}, 3377 {0x1df,0x1ef,2,-1}, 3378 {0x1f2,0x1f2,-1,-1}, 3379 {0x1f3,0x1f3,-1,-2}, 3380 {0x1f5,0x1f9,4,-1}, 3381 {0x1fb,0x21f,2,-1}, 3382 {0x223,0x233,2,-1}, 3383 {0x23c,0x23c,-1,-1}, 3384 {0x23f,0x240,1,10815}, 3385 {0x242,0x247,5,-1}, 3386 {0x249,0x24f,2,-1}, 3387 {0x250,0x250,-1,10783}, 3388 {0x251,0x251,-1,10780}, 3389 {0x252,0x252,-1,10782}, 3390 {0x253,0x253,-1,-210}, 3391 {0x254,0x254,-1,-206}, 3392 {0x256,0x257,1,-205}, 3393 {0x259,0x259,-1,-202}, 3394 {0x25b,0x25b,-1,-203}, 3395 {0x25c,0x25c,-1,42319}, 3396 {0x260,0x260,-1,-205}, 3397 {0x261,0x261,-1,42315}, 3398 {0x263,0x263,-1,-207}, 3399 {0x265,0x265,-1,42280}, 3400 {0x266,0x266,-1,42308}, 3401 {0x268,0x268,-1,-209}, 3402 {0x269,0x269,-1,-211}, 3403 {0x26a,0x26a,-1,42308}, 3404 {0x26b,0x26b,-1,10743}, 3405 {0x26c,0x26c,-1,42305}, 3406 {0x26f,0x26f,-1,-211}, 3407 {0x271,0x271,-1,10749}, 3408 {0x272,0x272,-1,-213}, 3409 {0x275,0x275,-1,-214}, 3410 {0x27d,0x27d,-1,10727}, 3411 {0x280,0x280,-1,-218}, 3412 {0x282,0x282,-1,42307}, 3413 {0x283,0x283,-1,-218}, 3414 {0x287,0x287,-1,42282}, 3415 {0x288,0x288,-1,-218}, 3416 {0x289,0x289,-1,-69}, 3417 {0x28a,0x28b,1,-217}, 3418 {0x28c,0x28c,-1,-71}, 3419 {0x292,0x292,-1,-219}, 3420 {0x29d,0x29d,-1,42261}, 3421 {0x29e,0x29e,-1,42258}, 3422 {0x345,0x345,-1,84}, 3423 {0x371,0x373,2,-1}, 3424 {0x377,0x377,-1,-1}, 3425 {0x37b,0x37d,1,130}, 3426 {0x3ac,0x3ac,-1,-38}, 3427 {0x3ad,0x3af,1,-37}, 3428 {0x3b1,0x3c1,1,-32}, 3429 {0x3c2,0x3c2,-1,-31}, 3430 {0x3c3,0x3cb,1,-32}, 3431 {0x3cc,0x3cc,-1,-64}, 3432 {0x3cd,0x3ce,1,-63}, 3433 {0x3d0,0x3d0,-1,-62}, 3434 {0x3d1,0x3d1,-1,-57}, 3435 {0x3d5,0x3d5,-1,-47}, 3436 {0x3d6,0x3d6,-1,-54}, 3437 {0x3d7,0x3d7,-1,-8}, 3438 {0x3d9,0x3ef,2,-1}, 3439 {0x3f0,0x3f0,-1,-86}, 3440 {0x3f1,0x3f1,-1,-80}, 3441 {0x3f2,0x3f2,-1,7}, 3442 {0x3f3,0x3f3,-1,-116}, 3443 {0x3f5,0x3f5,-1,-96}, 3444 {0x3f8,0x3fb,3,-1}, 3445 {0x430,0x44f,1,-32}, 3446 {0x450,0x45f,1,-80}, 3447 {0x461,0x481,2,-1}, 3448 {0x48b,0x4bf,2,-1}, 3449 {0x4c2,0x4ce,2,-1}, 3450 {0x4cf,0x4cf,-1,-15}, 3451 {0x4d1,0x52f,2,-1}, 3452 {0x561,0x586,1,-48}, 3453 {0x10d0,0x10fa,1,3008}, 3454 {0x10fd,0x10ff,1,3008}, 3455 {0x13f8,0x13fd,1,-8}, 3456 {0x1c80,0x1c80,-1,-6254}, 3457 {0x1c81,0x1c81,-1,-6253}, 3458 {0x1c82,0x1c82,-1,-6244}, 3459 {0x1c83,0x1c84,1,-6242}, 3460 {0x1c85,0x1c85,-1,-6243}, 3461 {0x1c86,0x1c86,-1,-6236}, 3462 {0x1c87,0x1c87,-1,-6181}, 3463 {0x1c88,0x1c88,-1,35266}, 3464 {0x1d79,0x1d79,-1,35332}, 3465 {0x1d7d,0x1d7d,-1,3814}, 3466 {0x1d8e,0x1d8e,-1,35384}, 3467 {0x1e01,0x1e95,2,-1}, 3468 {0x1e9b,0x1e9b,-1,-59}, 3469 {0x1ea1,0x1eff,2,-1}, 3470 {0x1f00,0x1f07,1,8}, 3471 {0x1f10,0x1f15,1,8}, 3472 {0x1f20,0x1f27,1,8}, 3473 {0x1f30,0x1f37,1,8}, 3474 {0x1f40,0x1f45,1,8}, 3475 {0x1f51,0x1f57,2,8}, 3476 {0x1f60,0x1f67,1,8}, 3477 {0x1f70,0x1f71,1,74}, 3478 {0x1f72,0x1f75,1,86}, 3479 {0x1f76,0x1f77,1,100}, 3480 {0x1f78,0x1f79,1,128}, 3481 {0x1f7a,0x1f7b,1,112}, 3482 {0x1f7c,0x1f7d,1,126}, 3483 {0x1f80,0x1f87,1,8}, 3484 {0x1f90,0x1f97,1,8}, 3485 {0x1fa0,0x1fa7,1,8}, 3486 {0x1fb0,0x1fb1,1,8}, 3487 {0x1fb3,0x1fb3,-1,9}, 3488 {0x1fbe,0x1fbe,-1,-7205}, 3489 {0x1fc3,0x1fc3,-1,9}, 3490 {0x1fd0,0x1fd1,1,8}, 3491 {0x1fe0,0x1fe1,1,8}, 3492 {0x1fe5,0x1fe5,-1,7}, 3493 {0x1ff3,0x1ff3,-1,9}, 3494 {0x214e,0x214e,-1,-28}, 3495 {0x2170,0x217f,1,-16}, 3496 {0x2184,0x2184,-1,-1}, 3497 {0x24d0,0x24e9,1,-26}, 3498 {0x2c30,0x2c5e,1,-48}, 3499 {0x2c61,0x2c61,-1,-1}, 3500 {0x2c65,0x2c65,-1,-10795}, 3501 {0x2c66,0x2c66,-1,-10792}, 3502 {0x2c68,0x2c6c,2,-1}, 3503 {0x2c73,0x2c76,3,-1}, 3504 {0x2c81,0x2ce3,2,-1}, 3505 {0x2cec,0x2cee,2,-1}, 3506 {0x2cf3,0x2cf3,-1,-1}, 3507 {0x2d00,0x2d25,1,-7264}, 3508 {0x2d27,0x2d2d,6,-7264}, 3509 {0xa641,0xa66d,2,-1}, 3510 {0xa681,0xa69b,2,-1}, 3511 {0xa723,0xa72f,2,-1}, 3512 {0xa733,0xa76f,2,-1}, 3513 {0xa77a,0xa77c,2,-1}, 3514 {0xa77f,0xa787,2,-1}, 3515 {0xa78c,0xa791,5,-1}, 3516 {0xa793,0xa793,-1,-1}, 3517 {0xa794,0xa794,-1,48}, 3518 {0xa797,0xa7a9,2,-1}, 3519 {0xa7b5,0xa7bf,2,-1}, 3520 {0xa7c3,0xa7c3,-1,-1}, 3521 {0xab53,0xab53,-1,-928}, 3522 {0xab70,0xabbf,1,-38864}, 3523 {0xff41,0xff5a,1,-32}, 3524 {0x10428,0x1044f,1,-40}, 3525 {0x104d8,0x104fb,1,-40}, 3526 {0x10cc0,0x10cf2,1,-64}, 3527 {0x118c0,0x118df,1,-32}, 3528 {0x16e60,0x16e7f,1,-32}, 3529 {0x1e922,0x1e943,1,-34} 3530 }; 3531 3532 /* 3533 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3534 * simple case folding. 3535 */ 3536 int 3537 utf_toupper(int a) 3538 { 3539 // If 'casemap' contains "keepascii" use ASCII style toupper(). 3540 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3541 return TOUPPER_ASC(a); 3542 3543 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3544 // If towupper() is available and handles Unicode, use it. 3545 if (!(cmp_flags & CMP_INTERNAL)) 3546 return towupper(a); 3547 #endif 3548 3549 // For characters below 128 use locale sensitive toupper(). 3550 if (a < 128) 3551 return TOUPPER_LOC(a); 3552 3553 // For any other characters use the above mapping table. 3554 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3555 } 3556 3557 int 3558 utf_islower(int a) 3559 { 3560 // German sharp s is lower case but has no upper case equivalent. 3561 return (utf_toupper(a) != a) || a == 0xdf; 3562 } 3563 3564 /* 3565 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3566 * simple case folding. 3567 */ 3568 int 3569 utf_tolower(int a) 3570 { 3571 // If 'casemap' contains "keepascii" use ASCII style tolower(). 3572 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3573 return TOLOWER_ASC(a); 3574 3575 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3576 // If towlower() is available and handles Unicode, use it. 3577 if (!(cmp_flags & CMP_INTERNAL)) 3578 return towlower(a); 3579 #endif 3580 3581 // For characters below 128 use locale sensitive tolower(). 3582 if (a < 128) 3583 return TOLOWER_LOC(a); 3584 3585 // For any other characters use the above mapping table. 3586 return utf_convert(a, toLower, (int)sizeof(toLower)); 3587 } 3588 3589 int 3590 utf_isupper(int a) 3591 { 3592 return (utf_tolower(a) != a); 3593 } 3594 3595 static int 3596 utf_strnicmp( 3597 char_u *s1, 3598 char_u *s2, 3599 size_t n1, 3600 size_t n2) 3601 { 3602 int c1, c2, cdiff; 3603 char_u buffer[6]; 3604 3605 for (;;) 3606 { 3607 c1 = utf_safe_read_char_adv(&s1, &n1); 3608 c2 = utf_safe_read_char_adv(&s2, &n2); 3609 3610 if (c1 <= 0 || c2 <= 0) 3611 break; 3612 3613 if (c1 == c2) 3614 continue; 3615 3616 cdiff = utf_fold(c1) - utf_fold(c2); 3617 if (cdiff != 0) 3618 return cdiff; 3619 } 3620 3621 // some string ended or has an incomplete/illegal character sequence 3622 3623 if (c1 == 0 || c2 == 0) 3624 { 3625 // some string ended. shorter string is smaller 3626 if (c1 == 0 && c2 == 0) 3627 return 0; 3628 return c1 == 0 ? -1 : 1; 3629 } 3630 3631 // Continue with bytewise comparison to produce some result that 3632 // would make comparison operations involving this function transitive. 3633 // 3634 // If only one string had an error, comparison should be made with 3635 // folded version of the other string. In this case it is enough 3636 // to fold just one character to determine the result of comparison. 3637 3638 if (c1 != -1 && c2 == -1) 3639 { 3640 n1 = utf_char2bytes(utf_fold(c1), buffer); 3641 s1 = buffer; 3642 } 3643 else if (c2 != -1 && c1 == -1) 3644 { 3645 n2 = utf_char2bytes(utf_fold(c2), buffer); 3646 s2 = buffer; 3647 } 3648 3649 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3650 { 3651 cdiff = (int)(*s1) - (int)(*s2); 3652 if (cdiff != 0) 3653 return cdiff; 3654 3655 s1++; 3656 s2++; 3657 n1--; 3658 n2--; 3659 } 3660 3661 if (n1 > 0 && *s1 == NUL) 3662 n1 = 0; 3663 if (n2 > 0 && *s2 == NUL) 3664 n2 = 0; 3665 3666 if (n1 == 0 && n2 == 0) 3667 return 0; 3668 return n1 == 0 ? -1 : 1; 3669 } 3670 3671 /* 3672 * Version of strnicmp() that handles multi-byte characters. 3673 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3674 * probably use strnicmp(), because there are no ASCII characters in the 3675 * second byte. 3676 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3677 * two characters otherwise. 3678 */ 3679 int 3680 mb_strnicmp(char_u *s1, char_u *s2, size_t nn) 3681 { 3682 int i, l; 3683 int cdiff; 3684 int n = (int)nn; 3685 3686 if (enc_utf8) 3687 { 3688 return utf_strnicmp(s1, s2, nn, nn); 3689 } 3690 else 3691 { 3692 for (i = 0; i < n; i += l) 3693 { 3694 if (s1[i] == NUL && s2[i] == NUL) // both strings end 3695 return 0; 3696 3697 l = (*mb_ptr2len)(s1 + i); 3698 if (l <= 1) 3699 { 3700 // Single byte: first check normally, then with ignore case. 3701 if (s1[i] != s2[i]) 3702 { 3703 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3704 if (cdiff != 0) 3705 return cdiff; 3706 } 3707 } 3708 else 3709 { 3710 // For non-Unicode multi-byte don't ignore case. 3711 if (l > n - i) 3712 l = n - i; 3713 cdiff = STRNCMP(s1 + i, s2 + i, l); 3714 if (cdiff != 0) 3715 return cdiff; 3716 } 3717 } 3718 } 3719 return 0; 3720 } 3721 3722 /* 3723 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3724 * 'encoding' has been set to. 3725 */ 3726 void 3727 show_utf8(void) 3728 { 3729 int len; 3730 int rlen = 0; 3731 char_u *line; 3732 int clen; 3733 int i; 3734 3735 // Get the byte length of the char under the cursor, including composing 3736 // characters. 3737 line = ml_get_cursor(); 3738 len = utfc_ptr2len(line); 3739 if (len == 0) 3740 { 3741 msg("NUL"); 3742 return; 3743 } 3744 3745 clen = 0; 3746 for (i = 0; i < len; ++i) 3747 { 3748 if (clen == 0) 3749 { 3750 // start of (composing) character, get its length 3751 if (i > 0) 3752 { 3753 STRCPY(IObuff + rlen, "+ "); 3754 rlen += 2; 3755 } 3756 clen = utf_ptr2len(line + i); 3757 } 3758 sprintf((char *)IObuff + rlen, "%02x ", 3759 (line[i] == NL) ? NUL : line[i]); // NUL is stored as NL 3760 --clen; 3761 rlen += (int)STRLEN(IObuff + rlen); 3762 if (rlen > IOSIZE - 20) 3763 break; 3764 } 3765 3766 msg((char *)IObuff); 3767 } 3768 3769 /* 3770 * mb_head_off() function pointer. 3771 * Return offset from "p" to the first byte of the character it points into. 3772 * If "p" points to the NUL at the end of the string return 0. 3773 * Returns 0 when already at the first byte of a character. 3774 */ 3775 int 3776 latin_head_off(char_u *base UNUSED, char_u *p UNUSED) 3777 { 3778 return 0; 3779 } 3780 3781 static int 3782 dbcs_head_off(char_u *base, char_u *p) 3783 { 3784 char_u *q; 3785 3786 // It can't be a trailing byte when not using DBCS, at the start of the 3787 // string or the previous byte can't start a double-byte. 3788 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3789 return 0; 3790 3791 // This is slow: need to start at the base and go forward until the 3792 // byte we are looking for. Return 1 when we went past it, 0 otherwise. 3793 q = base; 3794 while (q < p) 3795 q += dbcs_ptr2len(q); 3796 return (q == p) ? 0 : 1; 3797 } 3798 3799 /* 3800 * Special version of dbcs_head_off() that works for ScreenLines[], where 3801 * single-width DBCS_JPNU characters are stored separately. 3802 */ 3803 int 3804 dbcs_screen_head_off(char_u *base, char_u *p) 3805 { 3806 char_u *q; 3807 3808 // It can't be a trailing byte when not using DBCS, at the start of the 3809 // string or the previous byte can't start a double-byte. 3810 // For euc-jp an 0x8e byte in the previous cell always means we have a 3811 // lead byte in the current cell. 3812 if (p <= base 3813 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3814 || MB_BYTE2LEN(p[-1]) == 1 3815 || *p == NUL) 3816 return 0; 3817 3818 // This is slow: need to start at the base and go forward until the 3819 // byte we are looking for. Return 1 when we went past it, 0 otherwise. 3820 // For DBCS_JPNU look out for 0x8e, which means the second byte is not 3821 // stored as the next byte. 3822 q = base; 3823 while (q < p) 3824 { 3825 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3826 ++q; 3827 else 3828 q += dbcs_ptr2len(q); 3829 } 3830 return (q == p) ? 0 : 1; 3831 } 3832 3833 int 3834 utf_head_off(char_u *base, char_u *p) 3835 { 3836 char_u *q; 3837 char_u *s; 3838 int c; 3839 int len; 3840 #ifdef FEAT_ARABIC 3841 char_u *j; 3842 #endif 3843 3844 if (*p < 0x80) // be quick for ASCII 3845 return 0; 3846 3847 // Skip backwards over trailing bytes: 10xx.xxxx 3848 // Skip backwards again if on a composing char. 3849 for (q = p; ; --q) 3850 { 3851 // Move s to the last byte of this char. 3852 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3853 ; 3854 // Move q to the first byte of this char. 3855 while (q > base && (*q & 0xc0) == 0x80) 3856 --q; 3857 // Check for illegal sequence. Do allow an illegal byte after where we 3858 // started. 3859 len = utf8len_tab[*q]; 3860 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3861 return 0; 3862 3863 if (q <= base) 3864 break; 3865 3866 c = utf_ptr2char(q); 3867 if (utf_iscomposing(c)) 3868 continue; 3869 3870 #ifdef FEAT_ARABIC 3871 if (arabic_maycombine(c)) 3872 { 3873 // Advance to get a sneak-peak at the next char 3874 j = q; 3875 --j; 3876 // Move j to the first byte of this char. 3877 while (j > base && (*j & 0xc0) == 0x80) 3878 --j; 3879 if (arabic_combine(utf_ptr2char(j), c)) 3880 continue; 3881 } 3882 #endif 3883 break; 3884 } 3885 3886 return (int)(p - q); 3887 } 3888 3889 /* 3890 * Copy a character from "*fp" to "*tp" and advance the pointers. 3891 */ 3892 void 3893 mb_copy_char(char_u **fp, char_u **tp) 3894 { 3895 int l = (*mb_ptr2len)(*fp); 3896 3897 mch_memmove(*tp, *fp, (size_t)l); 3898 *tp += l; 3899 *fp += l; 3900 } 3901 3902 /* 3903 * Return the offset from "p" to the first byte of a character. When "p" is 3904 * at the start of a character 0 is returned, otherwise the offset to the next 3905 * character. Can start anywhere in a stream of bytes. 3906 */ 3907 int 3908 mb_off_next(char_u *base, char_u *p) 3909 { 3910 int i; 3911 int j; 3912 3913 if (enc_utf8) 3914 { 3915 if (*p < 0x80) // be quick for ASCII 3916 return 0; 3917 3918 // Find the next character that isn't 10xx.xxxx 3919 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 3920 ; 3921 if (i > 0) 3922 { 3923 // Check for illegal sequence. 3924 for (j = 0; p - j > base; ++j) 3925 if ((p[-j] & 0xc0) != 0x80) 3926 break; 3927 if (utf8len_tab[p[-j]] != i + j) 3928 return 0; 3929 } 3930 return i; 3931 } 3932 3933 // Only need to check if we're on a trail byte, it doesn't matter if we 3934 // want the offset to the next or current character. 3935 return (*mb_head_off)(base, p); 3936 } 3937 3938 /* 3939 * Return the offset from "p" to the last byte of the character it points 3940 * into. Can start anywhere in a stream of bytes. 3941 */ 3942 int 3943 mb_tail_off(char_u *base, char_u *p) 3944 { 3945 int i; 3946 int j; 3947 3948 if (*p == NUL) 3949 return 0; 3950 3951 if (enc_utf8) 3952 { 3953 // Find the last character that is 10xx.xxxx 3954 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 3955 ; 3956 // Check for illegal sequence. 3957 for (j = 0; p - j > base; ++j) 3958 if ((p[-j] & 0xc0) != 0x80) 3959 break; 3960 if (utf8len_tab[p[-j]] != i + j + 1) 3961 return 0; 3962 return i; 3963 } 3964 3965 // It can't be the first byte if a double-byte when not using DBCS, at the 3966 // end of the string or the byte can't start a double-byte. 3967 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 3968 return 0; 3969 3970 // Return 1 when on the lead byte, 0 when on the tail byte. 3971 return 1 - dbcs_head_off(base, p); 3972 } 3973 3974 /* 3975 * Find the next illegal byte sequence. 3976 */ 3977 void 3978 utf_find_illegal(void) 3979 { 3980 pos_T pos = curwin->w_cursor; 3981 char_u *p; 3982 int len; 3983 vimconv_T vimconv; 3984 char_u *tofree = NULL; 3985 3986 vimconv.vc_type = CONV_NONE; 3987 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 3988 { 3989 // 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 3990 // possibly a utf-8 file with illegal bytes. Setup for conversion 3991 // from utf-8 to 'fileencoding'. 3992 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 3993 } 3994 3995 curwin->w_cursor.coladd = 0; 3996 for (;;) 3997 { 3998 p = ml_get_cursor(); 3999 if (vimconv.vc_type != CONV_NONE) 4000 { 4001 vim_free(tofree); 4002 tofree = string_convert(&vimconv, p, NULL); 4003 if (tofree == NULL) 4004 break; 4005 p = tofree; 4006 } 4007 4008 while (*p != NUL) 4009 { 4010 // Illegal means that there are not enough trail bytes (checked by 4011 // utf_ptr2len()) or too many of them (overlong sequence). 4012 len = utf_ptr2len(p); 4013 if (*p >= 0x80 && (len == 1 4014 || utf_char2len(utf_ptr2char(p)) != len)) 4015 { 4016 if (vimconv.vc_type == CONV_NONE) 4017 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 4018 else 4019 { 4020 int l; 4021 4022 len = (int)(p - tofree); 4023 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 4024 { 4025 l = utf_ptr2len(p); 4026 curwin->w_cursor.col += l; 4027 } 4028 } 4029 goto theend; 4030 } 4031 p += len; 4032 } 4033 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 4034 break; 4035 ++curwin->w_cursor.lnum; 4036 curwin->w_cursor.col = 0; 4037 } 4038 4039 // didn't find it: don't move and beep 4040 curwin->w_cursor = pos; 4041 beep_flush(); 4042 4043 theend: 4044 vim_free(tofree); 4045 convert_setup(&vimconv, NULL, NULL); 4046 } 4047 4048 #if defined(FEAT_GUI_GTK) || defined(PROTO) 4049 /* 4050 * Return TRUE if string "s" is a valid utf-8 string. 4051 * When "end" is NULL stop at the first NUL. 4052 * When "end" is positive stop there. 4053 */ 4054 int 4055 utf_valid_string(char_u *s, char_u *end) 4056 { 4057 int l; 4058 char_u *p = s; 4059 4060 while (end == NULL ? *p != NUL : p < end) 4061 { 4062 l = utf8len_tab_zero[*p]; 4063 if (l == 0) 4064 return FALSE; // invalid lead byte 4065 if (end != NULL && p + l > end) 4066 return FALSE; // incomplete byte sequence 4067 ++p; 4068 while (--l > 0) 4069 if ((*p++ & 0xc0) != 0x80) 4070 return FALSE; // invalid trail byte 4071 } 4072 return TRUE; 4073 } 4074 #endif 4075 4076 #if defined(FEAT_GUI) || defined(PROTO) 4077 /* 4078 * Special version of mb_tail_off() for use in ScreenLines[]. 4079 */ 4080 int 4081 dbcs_screen_tail_off(char_u *base, char_u *p) 4082 { 4083 // It can't be the first byte if a double-byte when not using DBCS, at the 4084 // end of the string or the byte can't start a double-byte. 4085 // For euc-jp an 0x8e byte always means we have a lead byte in the current 4086 // cell. 4087 if (*p == NUL || p[1] == NUL 4088 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 4089 || MB_BYTE2LEN(*p) == 1) 4090 return 0; 4091 4092 // Return 1 when on the lead byte, 0 when on the tail byte. 4093 return 1 - dbcs_screen_head_off(base, p); 4094 } 4095 #endif 4096 4097 /* 4098 * If the cursor moves on an trail byte, set the cursor on the lead byte. 4099 * Thus it moves left if necessary. 4100 * Return TRUE when the cursor was adjusted. 4101 */ 4102 void 4103 mb_adjust_cursor(void) 4104 { 4105 mb_adjustpos(curbuf, &curwin->w_cursor); 4106 } 4107 4108 /* 4109 * Adjust position "*lp" to point to the first byte of a multi-byte character. 4110 * If it points to a tail byte it's moved backwards to the head byte. 4111 */ 4112 void 4113 mb_adjustpos(buf_T *buf, pos_T *lp) 4114 { 4115 char_u *p; 4116 4117 if (lp->col > 0 || lp->coladd > 1) 4118 { 4119 p = ml_get_buf(buf, lp->lnum, FALSE); 4120 if (*p == NUL || (int)STRLEN(p) < lp->col) 4121 lp->col = 0; 4122 else 4123 lp->col -= (*mb_head_off)(p, p + lp->col); 4124 // Reset "coladd" when the cursor would be on the right half of a 4125 // double-wide character. 4126 if (lp->coladd == 1 4127 && p[lp->col] != TAB 4128 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 4129 && ptr2cells(p + lp->col) > 1) 4130 lp->coladd = 0; 4131 } 4132 } 4133 4134 /* 4135 * Return a pointer to the character before "*p", if there is one. 4136 */ 4137 char_u * 4138 mb_prevptr( 4139 char_u *line, // start of the string 4140 char_u *p) 4141 { 4142 if (p > line) 4143 MB_PTR_BACK(line, p); 4144 return p; 4145 } 4146 4147 /* 4148 * Return the character length of "str". Each multi-byte character (with 4149 * following composing characters) counts as one. 4150 */ 4151 int 4152 mb_charlen(char_u *str) 4153 { 4154 char_u *p = str; 4155 int count; 4156 4157 if (p == NULL) 4158 return 0; 4159 4160 for (count = 0; *p != NUL; count++) 4161 p += (*mb_ptr2len)(p); 4162 4163 return count; 4164 } 4165 4166 #if defined(FEAT_SPELL) || defined(PROTO) 4167 /* 4168 * Like mb_charlen() but for a string with specified length. 4169 */ 4170 int 4171 mb_charlen_len(char_u *str, int len) 4172 { 4173 char_u *p = str; 4174 int count; 4175 4176 for (count = 0; *p != NUL && p < str + len; count++) 4177 p += (*mb_ptr2len)(p); 4178 4179 return count; 4180 } 4181 #endif 4182 4183 /* 4184 * Try to un-escape a multi-byte character. 4185 * Used for the "to" and "from" part of a mapping. 4186 * Return the un-escaped string if it is a multi-byte character, and advance 4187 * "pp" to just after the bytes that formed it. 4188 * Return NULL if no multi-byte char was found. 4189 */ 4190 char_u * 4191 mb_unescape(char_u **pp) 4192 { 4193 static char_u buf[6]; 4194 int n; 4195 int m = 0; 4196 char_u *str = *pp; 4197 4198 // Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 4199 // KS_EXTRA KE_CSI to CSI. 4200 // Maximum length of a utf-8 character is 4 bytes. 4201 for (n = 0; str[n] != NUL && m < 4; ++n) 4202 { 4203 if (str[n] == K_SPECIAL 4204 && str[n + 1] == KS_SPECIAL 4205 && str[n + 2] == KE_FILLER) 4206 { 4207 buf[m++] = K_SPECIAL; 4208 n += 2; 4209 } 4210 else if ((str[n] == K_SPECIAL 4211 # ifdef FEAT_GUI 4212 || str[n] == CSI 4213 # endif 4214 ) 4215 && str[n + 1] == KS_EXTRA 4216 && str[n + 2] == (int)KE_CSI) 4217 { 4218 buf[m++] = CSI; 4219 n += 2; 4220 } 4221 else if (str[n] == K_SPECIAL 4222 # ifdef FEAT_GUI 4223 || str[n] == CSI 4224 # endif 4225 ) 4226 break; // a special key can't be a multibyte char 4227 else 4228 buf[m++] = str[n]; 4229 buf[m] = NUL; 4230 4231 // Return a multi-byte character if it's found. An illegal sequence 4232 // will result in a 1 here. 4233 if ((*mb_ptr2len)(buf) > 1) 4234 { 4235 *pp = str + n + 1; 4236 return buf; 4237 } 4238 4239 // Bail out quickly for ASCII. 4240 if (buf[0] < 128) 4241 break; 4242 } 4243 return NULL; 4244 } 4245 4246 /* 4247 * Return TRUE if the character at "row"/"col" on the screen is the left side 4248 * of a double-width character. 4249 * Caller must make sure "row" and "col" are not invalid! 4250 */ 4251 int 4252 mb_lefthalve(int row, int col) 4253 { 4254 return (*mb_off2cells)(LineOffset[row] + col, 4255 LineOffset[row] + screen_Columns) > 1; 4256 } 4257 4258 /* 4259 * Correct a position on the screen, if it's the right half of a double-wide 4260 * char move it to the left half. Returns the corrected column. 4261 */ 4262 int 4263 mb_fix_col(int col, int row) 4264 { 4265 int off; 4266 4267 col = check_col(col); 4268 row = check_row(row); 4269 off = LineOffset[row] + col; 4270 if (has_mbyte && ScreenLines != NULL && col > 0 4271 && ((enc_dbcs 4272 && ScreenLines[off] != NUL 4273 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 4274 ScreenLines + off)) 4275 || (enc_utf8 && ScreenLines[off] == 0 4276 && ScreenLinesUC[off] == 0))) 4277 return col - 1; 4278 return col; 4279 } 4280 4281 static int enc_alias_search(char_u *name); 4282 4283 /* 4284 * Skip the Vim specific head of a 'encoding' name. 4285 */ 4286 char_u * 4287 enc_skip(char_u *p) 4288 { 4289 if (STRNCMP(p, "2byte-", 6) == 0) 4290 return p + 6; 4291 if (STRNCMP(p, "8bit-", 5) == 0) 4292 return p + 5; 4293 return p; 4294 } 4295 4296 /* 4297 * Find the canonical name for encoding "enc". 4298 * When the name isn't recognized, returns "enc" itself, but with all lower 4299 * case characters and '_' replaced with '-'. 4300 * Returns an allocated string. NULL for out-of-memory. 4301 */ 4302 char_u * 4303 enc_canonize(char_u *enc) 4304 { 4305 char_u *r; 4306 char_u *p, *s; 4307 int i; 4308 4309 if (STRCMP(enc, "default") == 0) 4310 { 4311 // Use the default encoding as it's found by set_init_1(). 4312 r = get_encoding_default(); 4313 if (r == NULL) 4314 r = (char_u *)"latin1"; 4315 return vim_strsave(r); 4316 } 4317 4318 // copy "enc" to allocated memory, with room for two '-' 4319 r = alloc(STRLEN(enc) + 3); 4320 if (r != NULL) 4321 { 4322 // Make it all lower case and replace '_' with '-'. 4323 p = r; 4324 for (s = enc; *s != NUL; ++s) 4325 { 4326 if (*s == '_') 4327 *p++ = '-'; 4328 else 4329 *p++ = TOLOWER_ASC(*s); 4330 } 4331 *p = NUL; 4332 4333 // Skip "2byte-" and "8bit-". 4334 p = enc_skip(r); 4335 4336 // Change "microsoft-cp" to "cp". Used in some spell files. 4337 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4338 STRMOVE(p, p + 10); 4339 4340 // "iso8859" -> "iso-8859" 4341 if (STRNCMP(p, "iso8859", 7) == 0) 4342 { 4343 STRMOVE(p + 4, p + 3); 4344 p[3] = '-'; 4345 } 4346 4347 // "iso-8859n" -> "iso-8859-n" 4348 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4349 { 4350 STRMOVE(p + 9, p + 8); 4351 p[8] = '-'; 4352 } 4353 4354 // "latin-N" -> "latinN" 4355 if (STRNCMP(p, "latin-", 6) == 0) 4356 STRMOVE(p + 5, p + 6); 4357 4358 if (enc_canon_search(p) >= 0) 4359 { 4360 // canonical name can be used unmodified 4361 if (p != r) 4362 STRMOVE(r, p); 4363 } 4364 else if ((i = enc_alias_search(p)) >= 0) 4365 { 4366 // alias recognized, get canonical name 4367 vim_free(r); 4368 r = vim_strsave((char_u *)enc_canon_table[i].name); 4369 } 4370 } 4371 return r; 4372 } 4373 4374 /* 4375 * Search for an encoding alias of "name". 4376 * Returns -1 when not found. 4377 */ 4378 static int 4379 enc_alias_search(char_u *name) 4380 { 4381 int i; 4382 4383 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4384 if (STRCMP(name, enc_alias_table[i].name) == 0) 4385 return enc_alias_table[i].canon; 4386 return -1; 4387 } 4388 4389 4390 #ifdef HAVE_LANGINFO_H 4391 # include <langinfo.h> 4392 #endif 4393 4394 #if !defined(FEAT_GUI_MSWIN) || defined(VIMDLL) 4395 /* 4396 * Get the canonicalized encoding from the specified locale string "locale" 4397 * or from the environment variables LC_ALL, LC_CTYPE and LANG. 4398 * Returns an allocated string when successful, NULL when not. 4399 */ 4400 char_u * 4401 enc_locale_env(char *locale) 4402 { 4403 char *s = locale; 4404 char *p; 4405 int i; 4406 char buf[50]; 4407 4408 if (s == NULL || *s == NUL) 4409 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4410 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4411 s = getenv("LANG"); 4412 4413 if (s == NULL || *s == NUL) 4414 return NULL; 4415 4416 // The most generic locale format is: 4417 // language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4418 // If there is a '.' remove the part before it. 4419 // if there is something after the codeset, remove it. 4420 // Make the name lowercase and replace '_' with '-'. 4421 // Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4422 // "ko_KR.EUC" == "euc-kr" 4423 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4424 { 4425 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4426 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4427 { 4428 // copy "XY.EUC" to "euc-XY" to buf[10] 4429 STRCPY(buf + 10, "euc-"); 4430 buf[14] = p[-2]; 4431 buf[15] = p[-1]; 4432 buf[16] = 0; 4433 s = buf + 10; 4434 } 4435 else 4436 s = p + 1; 4437 } 4438 for (i = 0; i < (int)sizeof(buf) - 1 && s[i] != NUL; ++i) 4439 { 4440 if (s[i] == '_' || s[i] == '-') 4441 buf[i] = '-'; 4442 else if (isalnum((int)s[i])) 4443 buf[i] = TOLOWER_ASC(s[i]); 4444 else 4445 break; 4446 } 4447 buf[i] = NUL; 4448 4449 return enc_canonize((char_u *)buf); 4450 } 4451 #endif 4452 4453 /* 4454 * Get the canonicalized encoding of the current locale. 4455 * Returns an allocated string when successful, NULL when not. 4456 */ 4457 char_u * 4458 enc_locale(void) 4459 { 4460 #ifdef MSWIN 4461 char buf[50]; 4462 long acp = GetACP(); 4463 4464 if (acp == 1200) 4465 STRCPY(buf, "ucs-2le"); 4466 else if (acp == 1252) // cp1252 is used as latin1 4467 STRCPY(buf, "latin1"); 4468 else if (acp == 65001) 4469 STRCPY(buf, "utf-8"); 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(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) && \ 4791 (defined(FEAT_XIM) || defined(IME_WITHOUT_XIM) || defined(VIMDLL)) 4792 static void 4793 call_imactivatefunc(int active) 4794 { 4795 typval_T argv[2]; 4796 4797 argv[0].v_type = VAR_NUMBER; 4798 argv[0].vval.v_number = active ? 1 : 0; 4799 argv[1].v_type = VAR_UNKNOWN; 4800 (void)call_func_retnr(p_imaf, 1, argv); 4801 } 4802 4803 static int 4804 call_imstatusfunc(void) 4805 { 4806 int is_active; 4807 4808 // FIXME: Don't execute user function in unsafe situation. 4809 if (exiting || is_autocmd_blocked()) 4810 return FALSE; 4811 // FIXME: :py print 'xxx' is shown duplicate result. 4812 // Use silent to avoid it. 4813 ++msg_silent; 4814 is_active = call_func_retnr(p_imsf, 0, NULL); 4815 --msg_silent; 4816 return (is_active > 0); 4817 } 4818 #endif 4819 4820 #if defined(FEAT_XIM) || defined(PROTO) 4821 4822 # if defined(FEAT_GUI_GTK) || defined(PROTO) 4823 static int xim_has_preediting INIT(= FALSE); // IM current status 4824 4825 /* 4826 * Set preedit_start_col to the current cursor position. 4827 */ 4828 static void 4829 init_preedit_start_col(void) 4830 { 4831 if (State & CMDLINE) 4832 preedit_start_col = cmdline_getvcol_cursor(); 4833 else if (curwin != NULL && curwin->w_buffer != NULL) 4834 getvcol(curwin, &curwin->w_cursor, &preedit_start_col, NULL, NULL); 4835 // Prevent that preediting marks the buffer as changed. 4836 xim_changed_while_preediting = curbuf->b_changed; 4837 } 4838 4839 static int im_is_active = FALSE; // IM is enabled for current mode 4840 static int preedit_is_active = FALSE; 4841 static int im_preedit_cursor = 0; // cursor offset in characters 4842 static int im_preedit_trailing = 0; // number of characters after cursor 4843 4844 static unsigned long im_commit_handler_id = 0; 4845 static unsigned int im_activatekey_keyval = GDK_VoidSymbol; 4846 static unsigned int im_activatekey_state = 0; 4847 4848 static GtkWidget *preedit_window = NULL; 4849 static GtkWidget *preedit_label = NULL; 4850 4851 static void im_preedit_window_set_position(void); 4852 4853 void 4854 im_set_active(int active) 4855 { 4856 int was_active; 4857 4858 was_active = !!im_get_status(); 4859 im_is_active = (active && !p_imdisable); 4860 4861 if (im_is_active != was_active) 4862 xim_reset(); 4863 } 4864 4865 void 4866 xim_set_focus(int focus) 4867 { 4868 if (xic != NULL) 4869 { 4870 if (focus) 4871 gtk_im_context_focus_in(xic); 4872 else 4873 gtk_im_context_focus_out(xic); 4874 } 4875 } 4876 4877 void 4878 im_set_position(int row, int col) 4879 { 4880 if (xic != NULL) 4881 { 4882 GdkRectangle area; 4883 4884 area.x = FILL_X(col); 4885 area.y = FILL_Y(row); 4886 area.width = gui.char_width * (mb_lefthalve(row, col) ? 2 : 1); 4887 area.height = gui.char_height; 4888 4889 gtk_im_context_set_cursor_location(xic, &area); 4890 4891 if (p_imst == IM_OVER_THE_SPOT) 4892 im_preedit_window_set_position(); 4893 } 4894 } 4895 4896 # if 0 || defined(PROTO) // apparently only used in gui_x11.c 4897 void 4898 xim_set_preedit(void) 4899 { 4900 im_set_position(gui.row, gui.col); 4901 } 4902 # endif 4903 4904 static void 4905 im_add_to_input(char_u *str, int len) 4906 { 4907 // Convert from 'termencoding' (always "utf-8") to 'encoding' 4908 if (input_conv.vc_type != CONV_NONE) 4909 { 4910 str = string_convert(&input_conv, str, &len); 4911 g_return_if_fail(str != NULL); 4912 } 4913 4914 add_to_input_buf_csi(str, len); 4915 4916 if (input_conv.vc_type != CONV_NONE) 4917 vim_free(str); 4918 4919 if (p_mh) // blank out the pointer if necessary 4920 gui_mch_mousehide(TRUE); 4921 } 4922 4923 static void 4924 im_preedit_window_set_position(void) 4925 { 4926 int x, y, width, height; 4927 int screen_x, screen_y, screen_width, screen_height; 4928 4929 if (preedit_window == NULL) 4930 return; 4931 4932 gui_gtk_get_screen_geom_of_win(gui.drawarea, 4933 &screen_x, &screen_y, &screen_width, &screen_height); 4934 gdk_window_get_origin(gtk_widget_get_window(gui.drawarea), &x, &y); 4935 gtk_window_get_size(GTK_WINDOW(preedit_window), &width, &height); 4936 x = x + FILL_X(gui.col); 4937 y = y + FILL_Y(gui.row); 4938 if (x + width > screen_x + screen_width) 4939 x = screen_x + screen_width - width; 4940 if (y + height > screen_y + screen_height) 4941 y = screen_y + screen_height - height; 4942 gtk_window_move(GTK_WINDOW(preedit_window), x, y); 4943 } 4944 4945 static void 4946 im_preedit_window_open() 4947 { 4948 char *preedit_string; 4949 #if !GTK_CHECK_VERSION(3,16,0) 4950 char buf[8]; 4951 #endif 4952 PangoAttrList *attr_list; 4953 PangoLayout *layout; 4954 #if GTK_CHECK_VERSION(3,0,0) 4955 # if !GTK_CHECK_VERSION(3,16,0) 4956 GdkRGBA color; 4957 # endif 4958 #else 4959 GdkColor color; 4960 #endif 4961 gint w, h; 4962 4963 if (preedit_window == NULL) 4964 { 4965 preedit_window = gtk_window_new(GTK_WINDOW_POPUP); 4966 gtk_window_set_transient_for(GTK_WINDOW(preedit_window), 4967 GTK_WINDOW(gui.mainwin)); 4968 preedit_label = gtk_label_new(""); 4969 gtk_widget_set_name(preedit_label, "vim-gui-preedit-area"); 4970 gtk_container_add(GTK_CONTAINER(preedit_window), preedit_label); 4971 } 4972 4973 #if GTK_CHECK_VERSION(3,16,0) 4974 { 4975 GtkStyleContext * const context 4976 = gtk_widget_get_style_context(gui.drawarea); 4977 GtkCssProvider * const provider = gtk_css_provider_new(); 4978 gchar *css = NULL; 4979 const char * const fontname 4980 = pango_font_description_get_family(gui.norm_font); 4981 gint fontsize 4982 = pango_font_description_get_size(gui.norm_font) / PANGO_SCALE; 4983 gchar *fontsize_propval = NULL; 4984 4985 if (!pango_font_description_get_size_is_absolute(gui.norm_font)) 4986 { 4987 // fontsize was given in points. Convert it into that in pixels 4988 // to use with CSS. 4989 GdkScreen * const screen 4990 = gdk_window_get_screen(gtk_widget_get_window(gui.mainwin)); 4991 const gdouble dpi = gdk_screen_get_resolution(screen); 4992 fontsize = dpi * fontsize / 72; 4993 } 4994 if (fontsize > 0) 4995 fontsize_propval = g_strdup_printf("%dpx", fontsize); 4996 else 4997 fontsize_propval = g_strdup_printf("inherit"); 4998 4999 css = g_strdup_printf( 5000 "widget#vim-gui-preedit-area {\n" 5001 " font-family: %s,monospace;\n" 5002 " font-size: %s;\n" 5003 " color: #%.2lx%.2lx%.2lx;\n" 5004 " background-color: #%.2lx%.2lx%.2lx;\n" 5005 "}\n", 5006 fontname != NULL ? fontname : "inherit", 5007 fontsize_propval, 5008 (gui.norm_pixel >> 16) & 0xff, 5009 (gui.norm_pixel >> 8) & 0xff, 5010 gui.norm_pixel & 0xff, 5011 (gui.back_pixel >> 16) & 0xff, 5012 (gui.back_pixel >> 8) & 0xff, 5013 gui.back_pixel & 0xff); 5014 5015 gtk_css_provider_load_from_data(provider, css, -1, NULL); 5016 gtk_style_context_add_provider(context, 5017 GTK_STYLE_PROVIDER(provider), G_MAXUINT); 5018 5019 g_free(css); 5020 g_free(fontsize_propval); 5021 g_object_unref(provider); 5022 } 5023 #elif GTK_CHECK_VERSION(3,0,0) 5024 gtk_widget_override_font(preedit_label, gui.norm_font); 5025 5026 vim_snprintf(buf, sizeof(buf), "#%06X", gui.norm_pixel); 5027 gdk_rgba_parse(&color, buf); 5028 gtk_widget_override_color(preedit_label, GTK_STATE_FLAG_NORMAL, &color); 5029 5030 vim_snprintf(buf, sizeof(buf), "#%06X", gui.back_pixel); 5031 gdk_rgba_parse(&color, buf); 5032 gtk_widget_override_background_color(preedit_label, GTK_STATE_FLAG_NORMAL, 5033 &color); 5034 #else 5035 gtk_widget_modify_font(preedit_label, gui.norm_font); 5036 5037 vim_snprintf(buf, sizeof(buf), "#%06X", (unsigned)gui.norm_pixel); 5038 gdk_color_parse(buf, &color); 5039 gtk_widget_modify_fg(preedit_label, GTK_STATE_NORMAL, &color); 5040 5041 vim_snprintf(buf, sizeof(buf), "#%06X", (unsigned)gui.back_pixel); 5042 gdk_color_parse(buf, &color); 5043 gtk_widget_modify_bg(preedit_window, GTK_STATE_NORMAL, &color); 5044 #endif 5045 5046 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5047 5048 if (preedit_string[0] != NUL) 5049 { 5050 gtk_label_set_text(GTK_LABEL(preedit_label), preedit_string); 5051 gtk_label_set_attributes(GTK_LABEL(preedit_label), attr_list); 5052 5053 layout = gtk_label_get_layout(GTK_LABEL(preedit_label)); 5054 pango_layout_get_pixel_size(layout, &w, &h); 5055 h = MAX(h, gui.char_height); 5056 gtk_window_resize(GTK_WINDOW(preedit_window), w, h); 5057 5058 gtk_widget_show_all(preedit_window); 5059 5060 im_preedit_window_set_position(); 5061 } 5062 5063 g_free(preedit_string); 5064 pango_attr_list_unref(attr_list); 5065 } 5066 5067 static void 5068 im_preedit_window_close() 5069 { 5070 if (preedit_window != NULL) 5071 gtk_widget_hide(preedit_window); 5072 } 5073 5074 static void 5075 im_show_preedit() 5076 { 5077 im_preedit_window_open(); 5078 5079 if (p_mh) // blank out the pointer if necessary 5080 gui_mch_mousehide(TRUE); 5081 } 5082 5083 static void 5084 im_delete_preedit(void) 5085 { 5086 char_u bskey[] = {CSI, 'k', 'b'}; 5087 char_u delkey[] = {CSI, 'k', 'D'}; 5088 5089 if (p_imst == IM_OVER_THE_SPOT) 5090 { 5091 im_preedit_window_close(); 5092 return; 5093 } 5094 5095 if (State & NORMAL 5096 #ifdef FEAT_TERMINAL 5097 && !term_use_loop() 5098 #endif 5099 ) 5100 { 5101 im_preedit_cursor = 0; 5102 return; 5103 } 5104 for (; im_preedit_cursor > 0; --im_preedit_cursor) 5105 add_to_input_buf(bskey, (int)sizeof(bskey)); 5106 5107 for (; im_preedit_trailing > 0; --im_preedit_trailing) 5108 add_to_input_buf(delkey, (int)sizeof(delkey)); 5109 } 5110 5111 /* 5112 * Move the cursor left by "num_move_back" characters. 5113 * Note that ins_left() checks im_is_preediting() to avoid breaking undo for 5114 * these K_LEFT keys. 5115 */ 5116 static void 5117 im_correct_cursor(int num_move_back) 5118 { 5119 char_u backkey[] = {CSI, 'k', 'l'}; 5120 5121 if (State & NORMAL) 5122 return; 5123 # ifdef FEAT_RIGHTLEFT 5124 if ((State & CMDLINE) == 0 && curwin != NULL && curwin->w_p_rl) 5125 backkey[2] = 'r'; 5126 # endif 5127 for (; num_move_back > 0; --num_move_back) 5128 add_to_input_buf(backkey, (int)sizeof(backkey)); 5129 } 5130 5131 static int xim_expected_char = NUL; 5132 static int xim_ignored_char = FALSE; 5133 5134 /* 5135 * Update the mode and cursor while in an IM callback. 5136 */ 5137 static void 5138 im_show_info(void) 5139 { 5140 int old_vgetc_busy; 5141 5142 old_vgetc_busy = vgetc_busy; 5143 vgetc_busy = TRUE; 5144 showmode(); 5145 vgetc_busy = old_vgetc_busy; 5146 if ((State & NORMAL) || (State & INSERT)) 5147 setcursor(); 5148 out_flush(); 5149 } 5150 5151 /* 5152 * Callback invoked when the user finished preediting. 5153 * Put the final string into the input buffer. 5154 */ 5155 static void 5156 im_commit_cb(GtkIMContext *context UNUSED, 5157 const gchar *str, 5158 gpointer data UNUSED) 5159 { 5160 int slen = (int)STRLEN(str); 5161 int add_to_input = TRUE; 5162 int clen; 5163 int len = slen; 5164 int commit_with_preedit = TRUE; 5165 char_u *im_str; 5166 5167 #ifdef XIM_DEBUG 5168 xim_log("im_commit_cb(): %s\n", str); 5169 #endif 5170 5171 if (p_imst == IM_ON_THE_SPOT) 5172 { 5173 // The imhangul module doesn't reset the preedit string before 5174 // committing. Call im_delete_preedit() to work around that. 5175 im_delete_preedit(); 5176 5177 // Indicate that preediting has finished. 5178 if (preedit_start_col == MAXCOL) 5179 { 5180 init_preedit_start_col(); 5181 commit_with_preedit = FALSE; 5182 } 5183 5184 // The thing which setting "preedit_start_col" to MAXCOL means that 5185 // "preedit_start_col" will be set forcedly when calling 5186 // preedit_changed_cb() next time. 5187 // "preedit_start_col" should not reset with MAXCOL on this part. Vim 5188 // is simulating the preediting by using add_to_input_str(). when 5189 // preedit begin immediately before committed, the typebuf is not 5190 // flushed to screen, then it can't get correct "preedit_start_col". 5191 // Thus, it should calculate the cells by adding cells of the committed 5192 // string. 5193 if (input_conv.vc_type != CONV_NONE) 5194 { 5195 im_str = string_convert(&input_conv, (char_u *)str, &len); 5196 g_return_if_fail(im_str != NULL); 5197 } 5198 else 5199 im_str = (char_u *)str; 5200 5201 clen = mb_string2cells(im_str, len); 5202 5203 if (input_conv.vc_type != CONV_NONE) 5204 vim_free(im_str); 5205 preedit_start_col += clen; 5206 } 5207 5208 // Is this a single character that matches a keypad key that's just 5209 // been pressed? If so, we don't want it to be entered as such - let 5210 // us carry on processing the raw keycode so that it may be used in 5211 // mappings as <kSomething>. 5212 if (xim_expected_char != NUL) 5213 { 5214 // We're currently processing a keypad or other special key 5215 if (slen == 1 && str[0] == xim_expected_char) 5216 { 5217 // It's a match - don't do it here 5218 xim_ignored_char = TRUE; 5219 add_to_input = FALSE; 5220 } 5221 else 5222 { 5223 // Not a match 5224 xim_ignored_char = FALSE; 5225 } 5226 } 5227 5228 if (add_to_input) 5229 im_add_to_input((char_u *)str, slen); 5230 5231 if (p_imst == IM_ON_THE_SPOT) 5232 { 5233 // Inserting chars while "im_is_active" is set does not cause a 5234 // change of buffer. When the chars are committed the buffer must be 5235 // marked as changed. 5236 if (!commit_with_preedit) 5237 preedit_start_col = MAXCOL; 5238 5239 // This flag is used in changed() at next call. 5240 xim_changed_while_preediting = TRUE; 5241 } 5242 5243 if (gtk_main_level() > 0) 5244 gtk_main_quit(); 5245 } 5246 5247 /* 5248 * Callback invoked after start to the preedit. 5249 */ 5250 static void 5251 im_preedit_start_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 5252 { 5253 #ifdef XIM_DEBUG 5254 xim_log("im_preedit_start_cb()\n"); 5255 #endif 5256 5257 im_is_active = TRUE; 5258 preedit_is_active = TRUE; 5259 gui_update_cursor(TRUE, FALSE); 5260 im_show_info(); 5261 } 5262 5263 /* 5264 * Callback invoked after end to the preedit. 5265 */ 5266 static void 5267 im_preedit_end_cb(GtkIMContext *context UNUSED, gpointer data UNUSED) 5268 { 5269 #ifdef XIM_DEBUG 5270 xim_log("im_preedit_end_cb()\n"); 5271 #endif 5272 im_delete_preedit(); 5273 5274 // Indicate that preediting has finished 5275 if (p_imst == IM_ON_THE_SPOT) 5276 preedit_start_col = MAXCOL; 5277 xim_has_preediting = FALSE; 5278 5279 #if 0 5280 // Removal of this line suggested by Takuhiro Nishioka. Fixes that IM was 5281 // switched off unintentionally. We now use preedit_is_active (added by 5282 // SungHyun Nam). 5283 im_is_active = FALSE; 5284 #endif 5285 preedit_is_active = FALSE; 5286 gui_update_cursor(TRUE, FALSE); 5287 im_show_info(); 5288 } 5289 5290 /* 5291 * Callback invoked after changes to the preedit string. If the preedit 5292 * string was empty before, remember the preedit start column so we know 5293 * where to apply feedback attributes. Delete the previous preedit string 5294 * if there was one, save the new preedit cursor offset, and put the new 5295 * string into the input buffer. 5296 * 5297 * TODO: The pragmatic "put into input buffer" approach used here has 5298 * several fundamental problems: 5299 * 5300 * - The characters in the preedit string are subject to remapping. 5301 * That's broken, only the finally committed string should be remapped. 5302 * 5303 * - There is a race condition involved: The retrieved value for the 5304 * current cursor position will be wrong if any unprocessed characters 5305 * are still queued in the input buffer. 5306 * 5307 * - Due to the lack of synchronization between the file buffer in memory 5308 * and any typed characters, it's practically impossible to implement the 5309 * "retrieve_surrounding" and "delete_surrounding" signals reliably. IM 5310 * modules for languages such as Thai are likely to rely on this feature 5311 * for proper operation. 5312 * 5313 * Conclusions: I think support for preediting needs to be moved to the 5314 * core parts of Vim. Ideally, until it has been committed, the preediting 5315 * string should only be displayed and not affect the buffer content at all. 5316 * The question how to deal with the synchronization issue still remains. 5317 * Circumventing the input buffer is probably not desirable. Anyway, I think 5318 * implementing "retrieve_surrounding" is the only hard problem. 5319 * 5320 * One way to solve all of this in a clean manner would be to queue all key 5321 * press/release events "as is" in the input buffer, and apply the IM filtering 5322 * at the receiving end of the queue. This, however, would have a rather large 5323 * impact on the code base. If there is an easy way to force processing of all 5324 * remaining input from within the "retrieve_surrounding" signal handler, this 5325 * might not be necessary. Gotta ask on vim-dev for opinions. 5326 */ 5327 static void 5328 im_preedit_changed_cb(GtkIMContext *context, gpointer data UNUSED) 5329 { 5330 char *preedit_string = NULL; 5331 int cursor_index = 0; 5332 int num_move_back = 0; 5333 char_u *str; 5334 char_u *p; 5335 int i; 5336 5337 if (p_imst == IM_ON_THE_SPOT) 5338 gtk_im_context_get_preedit_string(context, 5339 &preedit_string, NULL, 5340 &cursor_index); 5341 else 5342 gtk_im_context_get_preedit_string(context, 5343 &preedit_string, NULL, 5344 NULL); 5345 5346 #ifdef XIM_DEBUG 5347 xim_log("im_preedit_changed_cb(): %s\n", preedit_string); 5348 #endif 5349 5350 g_return_if_fail(preedit_string != NULL); // just in case 5351 5352 if (p_imst == IM_OVER_THE_SPOT) 5353 { 5354 if (preedit_string[0] == NUL) 5355 { 5356 xim_has_preediting = FALSE; 5357 im_delete_preedit(); 5358 } 5359 else 5360 { 5361 xim_has_preediting = TRUE; 5362 im_show_preedit(); 5363 } 5364 } 5365 else 5366 { 5367 // If preedit_start_col is MAXCOL set it to the current cursor position. 5368 if (preedit_start_col == MAXCOL && preedit_string[0] != '\0') 5369 { 5370 xim_has_preediting = TRUE; 5371 5372 // Urgh, this breaks if the input buffer isn't empty now 5373 init_preedit_start_col(); 5374 } 5375 else if (cursor_index == 0 && preedit_string[0] == '\0') 5376 { 5377 xim_has_preediting = FALSE; 5378 5379 // If at the start position (after typing backspace) 5380 // preedit_start_col must be reset. 5381 preedit_start_col = MAXCOL; 5382 } 5383 5384 im_delete_preedit(); 5385 5386 /* 5387 * Compute the end of the preediting area: "preedit_end_col". 5388 * According to the documentation of gtk_im_context_get_preedit_string(), 5389 * the cursor_pos output argument returns the offset in bytes. This is 5390 * unfortunately not true -- real life shows the offset is in characters, 5391 * and the GTK+ source code agrees with me. Will file a bug later. 5392 */ 5393 if (preedit_start_col != MAXCOL) 5394 preedit_end_col = preedit_start_col; 5395 str = (char_u *)preedit_string; 5396 for (p = str, i = 0; *p != NUL; p += utf_byte2len(*p), ++i) 5397 { 5398 int is_composing; 5399 5400 is_composing = ((*p & 0x80) != 0 && utf_iscomposing(utf_ptr2char(p))); 5401 /* 5402 * These offsets are used as counters when generating <BS> and <Del> 5403 * to delete the preedit string. So don't count composing characters 5404 * unless 'delcombine' is enabled. 5405 */ 5406 if (!is_composing || p_deco) 5407 { 5408 if (i < cursor_index) 5409 ++im_preedit_cursor; 5410 else 5411 ++im_preedit_trailing; 5412 } 5413 if (!is_composing && i >= cursor_index) 5414 { 5415 // This is essentially the same as im_preedit_trailing, except 5416 // composing characters are not counted even if p_deco is set. 5417 ++num_move_back; 5418 } 5419 if (preedit_start_col != MAXCOL) 5420 preedit_end_col += utf_ptr2cells(p); 5421 } 5422 5423 if (p > str) 5424 { 5425 im_add_to_input(str, (int)(p - str)); 5426 im_correct_cursor(num_move_back); 5427 } 5428 } 5429 5430 g_free(preedit_string); 5431 5432 if (gtk_main_level() > 0) 5433 gtk_main_quit(); 5434 } 5435 5436 /* 5437 * Translate the Pango attributes at iter to Vim highlighting attributes. 5438 * Ignore attributes not supported by Vim highlighting. This shouldn't have 5439 * too much impact -- right now we handle even more attributes than necessary 5440 * for the IM modules I tested with. 5441 */ 5442 static int 5443 translate_pango_attributes(PangoAttrIterator *iter) 5444 { 5445 PangoAttribute *attr; 5446 int char_attr = HL_NORMAL; 5447 5448 attr = pango_attr_iterator_get(iter, PANGO_ATTR_UNDERLINE); 5449 if (attr != NULL && ((PangoAttrInt *)attr)->value 5450 != (int)PANGO_UNDERLINE_NONE) 5451 char_attr |= HL_UNDERLINE; 5452 5453 attr = pango_attr_iterator_get(iter, PANGO_ATTR_WEIGHT); 5454 if (attr != NULL && ((PangoAttrInt *)attr)->value >= (int)PANGO_WEIGHT_BOLD) 5455 char_attr |= HL_BOLD; 5456 5457 attr = pango_attr_iterator_get(iter, PANGO_ATTR_STYLE); 5458 if (attr != NULL && ((PangoAttrInt *)attr)->value 5459 != (int)PANGO_STYLE_NORMAL) 5460 char_attr |= HL_ITALIC; 5461 5462 attr = pango_attr_iterator_get(iter, PANGO_ATTR_BACKGROUND); 5463 if (attr != NULL) 5464 { 5465 const PangoColor *color = &((PangoAttrColor *)attr)->color; 5466 5467 // Assume inverse if black background is requested 5468 if ((color->red | color->green | color->blue) == 0) 5469 char_attr |= HL_INVERSE; 5470 } 5471 5472 return char_attr; 5473 } 5474 5475 /* 5476 * Retrieve the highlighting attributes at column col in the preedit string. 5477 * Return -1 if not in preediting mode or if col is out of range. 5478 */ 5479 int 5480 im_get_feedback_attr(int col) 5481 { 5482 char *preedit_string = NULL; 5483 PangoAttrList *attr_list = NULL; 5484 int char_attr = -1; 5485 5486 if (xic == NULL) 5487 return char_attr; 5488 5489 gtk_im_context_get_preedit_string(xic, &preedit_string, &attr_list, NULL); 5490 5491 if (preedit_string != NULL && attr_list != NULL) 5492 { 5493 int idx; 5494 5495 // Get the byte index as used by PangoAttrIterator 5496 for (idx = 0; col > 0 && preedit_string[idx] != '\0'; --col) 5497 idx += utfc_ptr2len((char_u *)preedit_string + idx); 5498 5499 if (preedit_string[idx] != '\0') 5500 { 5501 PangoAttrIterator *iter; 5502 int start, end; 5503 5504 char_attr = HL_NORMAL; 5505 iter = pango_attr_list_get_iterator(attr_list); 5506 5507 // Extract all relevant attributes from the list. 5508 do 5509 { 5510 pango_attr_iterator_range(iter, &start, &end); 5511 5512 if (idx >= start && idx < end) 5513 char_attr |= translate_pango_attributes(iter); 5514 } 5515 while (pango_attr_iterator_next(iter)); 5516 5517 pango_attr_iterator_destroy(iter); 5518 } 5519 } 5520 5521 if (attr_list != NULL) 5522 pango_attr_list_unref(attr_list); 5523 g_free(preedit_string); 5524 5525 return char_attr; 5526 } 5527 5528 void 5529 xim_init(void) 5530 { 5531 #ifdef XIM_DEBUG 5532 xim_log("xim_init()\n"); 5533 #endif 5534 5535 g_return_if_fail(gui.drawarea != NULL); 5536 g_return_if_fail(gtk_widget_get_window(gui.drawarea) != NULL); 5537 5538 xic = gtk_im_multicontext_new(); 5539 g_object_ref(xic); 5540 5541 im_commit_handler_id = g_signal_connect(G_OBJECT(xic), "commit", 5542 G_CALLBACK(&im_commit_cb), NULL); 5543 g_signal_connect(G_OBJECT(xic), "preedit_changed", 5544 G_CALLBACK(&im_preedit_changed_cb), NULL); 5545 g_signal_connect(G_OBJECT(xic), "preedit_start", 5546 G_CALLBACK(&im_preedit_start_cb), NULL); 5547 g_signal_connect(G_OBJECT(xic), "preedit_end", 5548 G_CALLBACK(&im_preedit_end_cb), NULL); 5549 5550 gtk_im_context_set_client_window(xic, gtk_widget_get_window(gui.drawarea)); 5551 } 5552 5553 void 5554 im_shutdown(void) 5555 { 5556 #ifdef XIM_DEBUG 5557 xim_log("im_shutdown()\n"); 5558 #endif 5559 5560 if (xic != NULL) 5561 { 5562 gtk_im_context_focus_out(xic); 5563 g_object_unref(xic); 5564 xic = NULL; 5565 } 5566 im_is_active = FALSE; 5567 im_commit_handler_id = 0; 5568 if (p_imst == IM_ON_THE_SPOT) 5569 preedit_start_col = MAXCOL; 5570 xim_has_preediting = FALSE; 5571 } 5572 5573 /* 5574 * Convert the string argument to keyval and state for GdkEventKey. 5575 * If str is valid return TRUE, otherwise FALSE. 5576 * 5577 * See 'imactivatekey' for documentation of the format. 5578 */ 5579 static int 5580 im_string_to_keyval(const char *str, unsigned int *keyval, unsigned int *state) 5581 { 5582 const char *mods_end; 5583 unsigned tmp_keyval; 5584 unsigned tmp_state = 0; 5585 5586 mods_end = strrchr(str, '-'); 5587 mods_end = (mods_end != NULL) ? mods_end + 1 : str; 5588 5589 // Parse modifier keys 5590 while (str < mods_end) 5591 switch (*str++) 5592 { 5593 case '-': break; 5594 case 'S': case 's': tmp_state |= (unsigned)GDK_SHIFT_MASK; break; 5595 case 'L': case 'l': tmp_state |= (unsigned)GDK_LOCK_MASK; break; 5596 case 'C': case 'c': tmp_state |= (unsigned)GDK_CONTROL_MASK;break; 5597 case '1': tmp_state |= (unsigned)GDK_MOD1_MASK; break; 5598 case '2': tmp_state |= (unsigned)GDK_MOD2_MASK; break; 5599 case '3': tmp_state |= (unsigned)GDK_MOD3_MASK; break; 5600 case '4': tmp_state |= (unsigned)GDK_MOD4_MASK; break; 5601 case '5': tmp_state |= (unsigned)GDK_MOD5_MASK; break; 5602 default: 5603 return FALSE; 5604 } 5605 5606 tmp_keyval = gdk_keyval_from_name(str); 5607 5608 if (tmp_keyval == 0 || tmp_keyval == GDK_VoidSymbol) 5609 return FALSE; 5610 5611 if (keyval != NULL) 5612 *keyval = tmp_keyval; 5613 if (state != NULL) 5614 *state = tmp_state; 5615 5616 return TRUE; 5617 } 5618 5619 /* 5620 * Return TRUE if p_imak is valid, otherwise FALSE. As a special case, an 5621 * empty string is also regarded as valid. 5622 * 5623 * Note: The numerical key value of p_imak is cached if it was valid; thus 5624 * boldly assuming im_xim_isvalid_imactivate() will always be called whenever 5625 * 'imak' changes. This is currently the case but not obvious -- should 5626 * probably rename the function for clarity. 5627 */ 5628 int 5629 im_xim_isvalid_imactivate(void) 5630 { 5631 if (p_imak[0] == NUL) 5632 { 5633 im_activatekey_keyval = GDK_VoidSymbol; 5634 im_activatekey_state = 0; 5635 return TRUE; 5636 } 5637 5638 return im_string_to_keyval((const char *)p_imak, 5639 &im_activatekey_keyval, 5640 &im_activatekey_state); 5641 } 5642 5643 static void 5644 im_synthesize_keypress(unsigned int keyval, unsigned int state) 5645 { 5646 GdkEventKey *event; 5647 5648 event = (GdkEventKey *)gdk_event_new(GDK_KEY_PRESS); 5649 g_object_ref(gtk_widget_get_window(gui.drawarea)); 5650 // unreffed by gdk_event_free() 5651 event->window = gtk_widget_get_window(gui.drawarea); 5652 event->send_event = TRUE; 5653 event->time = GDK_CURRENT_TIME; 5654 event->state = state; 5655 event->keyval = keyval; 5656 event->hardware_keycode = // needed for XIM 5657 XKeysymToKeycode(GDK_WINDOW_XDISPLAY(event->window), (KeySym)keyval); 5658 event->length = 0; 5659 event->string = NULL; 5660 5661 gtk_im_context_filter_keypress(xic, event); 5662 5663 // For consistency, also send the corresponding release event. 5664 event->type = GDK_KEY_RELEASE; 5665 event->send_event = FALSE; 5666 gtk_im_context_filter_keypress(xic, event); 5667 5668 gdk_event_free((GdkEvent *)event); 5669 } 5670 5671 void 5672 xim_reset(void) 5673 { 5674 # ifdef FEAT_EVAL 5675 if (USE_IMACTIVATEFUNC) 5676 call_imactivatefunc(im_is_active); 5677 else 5678 # endif 5679 if (xic != NULL) 5680 { 5681 gtk_im_context_reset(xic); 5682 5683 if (p_imdisable) 5684 im_shutdown(); 5685 else 5686 { 5687 xim_set_focus(gui.in_focus); 5688 5689 if (im_activatekey_keyval != GDK_VoidSymbol) 5690 { 5691 if (im_is_active) 5692 { 5693 g_signal_handler_block(xic, im_commit_handler_id); 5694 im_synthesize_keypress(im_activatekey_keyval, 5695 im_activatekey_state); 5696 g_signal_handler_unblock(xic, im_commit_handler_id); 5697 } 5698 } 5699 else 5700 { 5701 im_shutdown(); 5702 xim_init(); 5703 xim_set_focus(gui.in_focus); 5704 } 5705 } 5706 } 5707 5708 if (p_imst == IM_ON_THE_SPOT) 5709 preedit_start_col = MAXCOL; 5710 xim_has_preediting = FALSE; 5711 } 5712 5713 int 5714 xim_queue_key_press_event(GdkEventKey *event, int down) 5715 { 5716 if (down) 5717 { 5718 /* 5719 * Workaround GTK2 XIM 'feature' that always converts keypad keys to 5720 * chars., even when not part of an IM sequence (ref. feature of 5721 * gdk/gdkkeyuni.c). 5722 * Flag any keypad keys that might represent a single char. 5723 * If this (on its own - i.e., not part of an IM sequence) is 5724 * committed while we're processing one of these keys, we can ignore 5725 * that commit and go ahead & process it ourselves. That way we can 5726 * still distinguish keypad keys for use in mappings. 5727 * Also add GDK_space to make <S-Space> work. 5728 */ 5729 switch (event->keyval) 5730 { 5731 case GDK_KP_Add: xim_expected_char = '+'; break; 5732 case GDK_KP_Subtract: xim_expected_char = '-'; break; 5733 case GDK_KP_Divide: xim_expected_char = '/'; break; 5734 case GDK_KP_Multiply: xim_expected_char = '*'; break; 5735 case GDK_KP_Decimal: xim_expected_char = '.'; break; 5736 case GDK_KP_Equal: xim_expected_char = '='; break; 5737 case GDK_KP_0: xim_expected_char = '0'; break; 5738 case GDK_KP_1: xim_expected_char = '1'; break; 5739 case GDK_KP_2: xim_expected_char = '2'; break; 5740 case GDK_KP_3: xim_expected_char = '3'; break; 5741 case GDK_KP_4: xim_expected_char = '4'; break; 5742 case GDK_KP_5: xim_expected_char = '5'; break; 5743 case GDK_KP_6: xim_expected_char = '6'; break; 5744 case GDK_KP_7: xim_expected_char = '7'; break; 5745 case GDK_KP_8: xim_expected_char = '8'; break; 5746 case GDK_KP_9: xim_expected_char = '9'; break; 5747 case GDK_space: xim_expected_char = ' '; break; 5748 default: xim_expected_char = NUL; 5749 } 5750 xim_ignored_char = FALSE; 5751 } 5752 5753 /* 5754 * When typing fFtT, XIM may be activated. Thus it must pass 5755 * gtk_im_context_filter_keypress() in Normal mode. 5756 * And while doing :sh too. 5757 */ 5758 if (xic != NULL && !p_imdisable 5759 && (State & (INSERT | CMDLINE | NORMAL | EXTERNCMD)) != 0) 5760 { 5761 /* 5762 * Filter 'imactivatekey' and map it to CTRL-^. This way, Vim is 5763 * always aware of the current status of IM, and can even emulate 5764 * the activation key for modules that don't support one. 5765 */ 5766 if (event->keyval == im_activatekey_keyval 5767 && (event->state & im_activatekey_state) == im_activatekey_state) 5768 { 5769 unsigned int state_mask; 5770 5771 // Require the state of the 3 most used modifiers to match exactly. 5772 // Otherwise e.g. <S-C-space> would be unusable for other purposes 5773 // if the IM activate key is <S-space>. 5774 state_mask = im_activatekey_state; 5775 state_mask |= ((int)GDK_SHIFT_MASK | (int)GDK_CONTROL_MASK 5776 | (int)GDK_MOD1_MASK); 5777 5778 if ((event->state & state_mask) != im_activatekey_state) 5779 return FALSE; 5780 5781 // Don't send it a second time on GDK_KEY_RELEASE. 5782 if (event->type != GDK_KEY_PRESS) 5783 return TRUE; 5784 5785 if (map_to_exists_mode((char_u *)"", LANGMAP, FALSE)) 5786 { 5787 im_set_active(FALSE); 5788 5789 // ":lmap" mappings exists, toggle use of mappings. 5790 State ^= LANGMAP; 5791 if (State & LANGMAP) 5792 { 5793 curbuf->b_p_iminsert = B_IMODE_NONE; 5794 State &= ~LANGMAP; 5795 } 5796 else 5797 { 5798 curbuf->b_p_iminsert = B_IMODE_LMAP; 5799 State |= LANGMAP; 5800 } 5801 return TRUE; 5802 } 5803 5804 return gtk_im_context_filter_keypress(xic, event); 5805 } 5806 5807 // Don't filter events through the IM context if IM isn't active 5808 // right now. Unlike with GTK+ 1.2 we cannot rely on the IM module 5809 // not doing anything before the activation key was sent. 5810 if (im_activatekey_keyval == GDK_VoidSymbol || im_is_active) 5811 { 5812 int imresult = gtk_im_context_filter_keypress(xic, event); 5813 5814 if (p_imst == IM_ON_THE_SPOT) 5815 { 5816 // Some XIM send following sequence: 5817 // 1. preedited string. 5818 // 2. committed string. 5819 // 3. line changed key. 5820 // 4. preedited string. 5821 // 5. remove preedited string. 5822 // if 3, Vim can't move back the above line for 5. 5823 // thus, this part should not parse the key. 5824 if (!imresult && preedit_start_col != MAXCOL 5825 && event->keyval == GDK_Return) 5826 { 5827 im_synthesize_keypress(GDK_Return, 0U); 5828 return FALSE; 5829 } 5830 } 5831 5832 // If XIM tried to commit a keypad key as a single char., 5833 // ignore it so we can use the keypad key 'raw', for mappings. 5834 if (xim_expected_char != NUL && xim_ignored_char) 5835 // We had a keypad key, and XIM tried to thieve it 5836 return FALSE; 5837 5838 // This is supposed to fix a problem with iBus, that space 5839 // characters don't work in input mode. 5840 xim_expected_char = NUL; 5841 5842 // Normal processing 5843 return imresult; 5844 } 5845 } 5846 5847 return FALSE; 5848 } 5849 5850 int 5851 im_get_status(void) 5852 { 5853 # ifdef FEAT_EVAL 5854 if (USE_IMSTATUSFUNC) 5855 return call_imstatusfunc(); 5856 # endif 5857 return im_is_active; 5858 } 5859 5860 int 5861 preedit_get_status(void) 5862 { 5863 return preedit_is_active; 5864 } 5865 5866 int 5867 im_is_preediting(void) 5868 { 5869 return xim_has_preediting; 5870 } 5871 5872 # else // !FEAT_GUI_GTK 5873 5874 static int xim_is_active = FALSE; // XIM should be active in the current 5875 // mode 5876 static int xim_has_focus = FALSE; // XIM is really being used for Vim 5877 # ifdef FEAT_GUI_X11 5878 static XIMStyle input_style; 5879 static int status_area_enabled = TRUE; 5880 # endif 5881 5882 /* 5883 * Switch using XIM on/off. This is used by the code that changes "State". 5884 * When 'imactivatefunc' is defined use that function instead. 5885 */ 5886 void 5887 im_set_active(int active_arg) 5888 { 5889 int active = active_arg; 5890 5891 // If 'imdisable' is set, XIM is never active. 5892 if (p_imdisable) 5893 active = FALSE; 5894 else if (input_style & XIMPreeditPosition) 5895 // There is a problem in switching XIM off when preediting is used, 5896 // and it is not clear how this can be solved. For now, keep XIM on 5897 // all the time, like it was done in Vim 5.8. 5898 active = TRUE; 5899 5900 # if defined(FEAT_EVAL) 5901 if (USE_IMACTIVATEFUNC) 5902 { 5903 if (active != im_get_status()) 5904 { 5905 call_imactivatefunc(active); 5906 xim_has_focus = active; 5907 } 5908 return; 5909 } 5910 # endif 5911 5912 if (xic == NULL) 5913 return; 5914 5915 // Remember the active state, it is needed when Vim gets keyboard focus. 5916 xim_is_active = active; 5917 xim_set_preedit(); 5918 } 5919 5920 /* 5921 * Adjust using XIM for gaining or losing keyboard focus. Also called when 5922 * "xim_is_active" changes. 5923 */ 5924 void 5925 xim_set_focus(int focus) 5926 { 5927 if (xic == NULL) 5928 return; 5929 5930 /* 5931 * XIM only gets focus when the Vim window has keyboard focus and XIM has 5932 * been set active for the current mode. 5933 */ 5934 if (focus && xim_is_active) 5935 { 5936 if (!xim_has_focus) 5937 { 5938 xim_has_focus = TRUE; 5939 XSetICFocus(xic); 5940 } 5941 } 5942 else 5943 { 5944 if (xim_has_focus) 5945 { 5946 xim_has_focus = FALSE; 5947 XUnsetICFocus(xic); 5948 } 5949 } 5950 } 5951 5952 void 5953 im_set_position(int row UNUSED, int col UNUSED) 5954 { 5955 xim_set_preedit(); 5956 } 5957 5958 /* 5959 * Set the XIM to the current cursor position. 5960 */ 5961 void 5962 xim_set_preedit(void) 5963 { 5964 XVaNestedList attr_list; 5965 XRectangle spot_area; 5966 XPoint over_spot; 5967 int line_space; 5968 5969 if (xic == NULL) 5970 return; 5971 5972 xim_set_focus(TRUE); 5973 5974 if (!xim_has_focus) 5975 { 5976 // hide XIM cursor 5977 over_spot.x = 0; 5978 over_spot.y = -100; // arbitrary invisible position 5979 attr_list = (XVaNestedList) XVaCreateNestedList(0, 5980 XNSpotLocation, 5981 &over_spot, 5982 NULL); 5983 XSetICValues(xic, XNPreeditAttributes, attr_list, NULL); 5984 XFree(attr_list); 5985 return; 5986 } 5987 5988 if (input_style & XIMPreeditPosition) 5989 { 5990 if (xim_fg_color == INVALCOLOR) 5991 { 5992 xim_fg_color = gui.def_norm_pixel; 5993 xim_bg_color = gui.def_back_pixel; 5994 } 5995 over_spot.x = TEXT_X(gui.col); 5996 over_spot.y = TEXT_Y(gui.row); 5997 spot_area.x = 0; 5998 spot_area.y = 0; 5999 spot_area.height = gui.char_height * Rows; 6000 spot_area.width = gui.char_width * Columns; 6001 line_space = gui.char_height; 6002 attr_list = (XVaNestedList) XVaCreateNestedList(0, 6003 XNSpotLocation, &over_spot, 6004 XNForeground, (Pixel) xim_fg_color, 6005 XNBackground, (Pixel) xim_bg_color, 6006 XNArea, &spot_area, 6007 XNLineSpace, line_space, 6008 NULL); 6009 if (XSetICValues(xic, XNPreeditAttributes, attr_list, NULL)) 6010 emsg(_("E284: Cannot set IC values")); 6011 XFree(attr_list); 6012 } 6013 } 6014 6015 # if defined(FEAT_GUI_X11) 6016 static char e_xim[] = N_("E285: Failed to create input context"); 6017 # endif 6018 6019 # if defined(FEAT_GUI_X11) || defined(PROTO) 6020 # if defined(XtSpecificationRelease) && XtSpecificationRelease >= 6 && !defined(SUN_SYSTEM) 6021 # define USE_X11R6_XIM 6022 # endif 6023 6024 static int xim_real_init(Window x11_window, Display *x11_display); 6025 6026 6027 # ifdef USE_X11R6_XIM 6028 static void 6029 xim_instantiate_cb( 6030 Display *display, 6031 XPointer client_data UNUSED, 6032 XPointer call_data UNUSED) 6033 { 6034 Window x11_window; 6035 Display *x11_display; 6036 6037 # ifdef XIM_DEBUG 6038 xim_log("xim_instantiate_cb()\n"); 6039 # endif 6040 6041 gui_get_x11_windis(&x11_window, &x11_display); 6042 if (display != x11_display) 6043 return; 6044 6045 xim_real_init(x11_window, x11_display); 6046 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6047 if (xic != NULL) 6048 XUnregisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6049 xim_instantiate_cb, NULL); 6050 } 6051 6052 static void 6053 xim_destroy_cb( 6054 XIM im UNUSED, 6055 XPointer client_data UNUSED, 6056 XPointer call_data UNUSED) 6057 { 6058 Window x11_window; 6059 Display *x11_display; 6060 6061 # ifdef XIM_DEBUG 6062 xim_log("xim_destroy_cb()\n"); 6063 #endif 6064 gui_get_x11_windis(&x11_window, &x11_display); 6065 6066 xic = NULL; 6067 status_area_enabled = FALSE; 6068 6069 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6070 6071 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6072 xim_instantiate_cb, NULL); 6073 } 6074 # endif 6075 6076 void 6077 xim_init(void) 6078 { 6079 Window x11_window; 6080 Display *x11_display; 6081 6082 # ifdef XIM_DEBUG 6083 xim_log("xim_init()\n"); 6084 # endif 6085 6086 gui_get_x11_windis(&x11_window, &x11_display); 6087 6088 xic = NULL; 6089 6090 if (xim_real_init(x11_window, x11_display)) 6091 return; 6092 6093 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6094 6095 # ifdef USE_X11R6_XIM 6096 XRegisterIMInstantiateCallback(x11_display, NULL, NULL, NULL, 6097 xim_instantiate_cb, NULL); 6098 # endif 6099 } 6100 6101 static int 6102 xim_real_init(Window x11_window, Display *x11_display) 6103 { 6104 int i; 6105 char *p, 6106 *s, 6107 *ns, 6108 *end, 6109 tmp[1024]; 6110 # define IMLEN_MAX 40 6111 char buf[IMLEN_MAX + 7]; 6112 XIM xim = NULL; 6113 XIMStyles *xim_styles; 6114 XIMStyle this_input_style = 0; 6115 Boolean found; 6116 XPoint over_spot; 6117 XVaNestedList preedit_list, status_list; 6118 6119 input_style = 0; 6120 status_area_enabled = FALSE; 6121 6122 if (xic != NULL) 6123 return FALSE; 6124 6125 if (gui.rsrc_input_method != NULL && *gui.rsrc_input_method != NUL) 6126 { 6127 strcpy(tmp, gui.rsrc_input_method); 6128 for (ns = s = tmp; ns != NULL && *s != NUL;) 6129 { 6130 s = (char *)skipwhite((char_u *)s); 6131 if (*s == NUL) 6132 break; 6133 if ((ns = end = strchr(s, ',')) == NULL) 6134 end = s + strlen(s); 6135 while (isspace(((char_u *)end)[-1])) 6136 end--; 6137 *end = NUL; 6138 6139 if (strlen(s) <= IMLEN_MAX) 6140 { 6141 strcpy(buf, "@im="); 6142 strcat(buf, s); 6143 if ((p = XSetLocaleModifiers(buf)) != NULL && *p != NUL 6144 && (xim = XOpenIM(x11_display, NULL, NULL, NULL)) 6145 != NULL) 6146 break; 6147 } 6148 6149 s = ns + 1; 6150 } 6151 } 6152 6153 if (xim == NULL && (p = XSetLocaleModifiers("")) != NULL && *p != NUL) 6154 xim = XOpenIM(x11_display, NULL, NULL, NULL); 6155 6156 // This is supposed to be useful to obtain characters through 6157 // XmbLookupString() without really using a XIM. 6158 if (xim == NULL && (p = XSetLocaleModifiers("@im=none")) != NULL 6159 && *p != NUL) 6160 xim = XOpenIM(x11_display, NULL, NULL, NULL); 6161 6162 if (xim == NULL) 6163 { 6164 // Only give this message when verbose is set, because too many people 6165 // got this message when they didn't want to use a XIM. 6166 if (p_verbose > 0) 6167 { 6168 verbose_enter(); 6169 emsg(_("E286: Failed to open input method")); 6170 verbose_leave(); 6171 } 6172 return FALSE; 6173 } 6174 6175 # ifdef USE_X11R6_XIM 6176 { 6177 XIMCallback destroy_cb; 6178 6179 destroy_cb.callback = xim_destroy_cb; 6180 destroy_cb.client_data = NULL; 6181 if (XSetIMValues(xim, XNDestroyCallback, &destroy_cb, NULL)) 6182 emsg(_("E287: Warning: Could not set destroy callback to IM")); 6183 } 6184 # endif 6185 6186 if (XGetIMValues(xim, XNQueryInputStyle, &xim_styles, NULL) || !xim_styles) 6187 { 6188 emsg(_("E288: input method doesn't support any style")); 6189 XCloseIM(xim); 6190 return FALSE; 6191 } 6192 6193 found = False; 6194 strcpy(tmp, gui.rsrc_preedit_type_name); 6195 for (s = tmp; s && !found; ) 6196 { 6197 while (*s && isspace((unsigned char)*s)) 6198 s++; 6199 if (!*s) 6200 break; 6201 if ((ns = end = strchr(s, ',')) != 0) 6202 ns++; 6203 else 6204 end = s + strlen(s); 6205 while (isspace((unsigned char)*end)) 6206 end--; 6207 *end = '\0'; 6208 6209 if (!strcmp(s, "OverTheSpot")) 6210 this_input_style = (XIMPreeditPosition | XIMStatusArea); 6211 else if (!strcmp(s, "OffTheSpot")) 6212 this_input_style = (XIMPreeditArea | XIMStatusArea); 6213 else if (!strcmp(s, "Root")) 6214 this_input_style = (XIMPreeditNothing | XIMStatusNothing); 6215 6216 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 6217 { 6218 if (this_input_style == xim_styles->supported_styles[i]) 6219 { 6220 found = True; 6221 break; 6222 } 6223 } 6224 if (!found) 6225 for (i = 0; (unsigned short)i < xim_styles->count_styles; i++) 6226 { 6227 if ((xim_styles->supported_styles[i] & this_input_style) 6228 == (this_input_style & ~XIMStatusArea)) 6229 { 6230 this_input_style &= ~XIMStatusArea; 6231 found = True; 6232 break; 6233 } 6234 } 6235 6236 s = ns; 6237 } 6238 XFree(xim_styles); 6239 6240 if (!found) 6241 { 6242 // Only give this message when verbose is set, because too many people 6243 // got this message when they didn't want to use a XIM. 6244 if (p_verbose > 0) 6245 { 6246 verbose_enter(); 6247 emsg(_("E289: input method doesn't support my preedit type")); 6248 verbose_leave(); 6249 } 6250 XCloseIM(xim); 6251 return FALSE; 6252 } 6253 6254 over_spot.x = TEXT_X(gui.col); 6255 over_spot.y = TEXT_Y(gui.row); 6256 input_style = this_input_style; 6257 6258 // A crash was reported when trying to pass gui.norm_font as XNFontSet, 6259 // thus that has been removed. Hopefully the default works... 6260 # ifdef FEAT_XFONTSET 6261 if (gui.fontset != NOFONTSET) 6262 { 6263 preedit_list = XVaCreateNestedList(0, 6264 XNSpotLocation, &over_spot, 6265 XNForeground, (Pixel)gui.def_norm_pixel, 6266 XNBackground, (Pixel)gui.def_back_pixel, 6267 XNFontSet, (XFontSet)gui.fontset, 6268 NULL); 6269 status_list = XVaCreateNestedList(0, 6270 XNForeground, (Pixel)gui.def_norm_pixel, 6271 XNBackground, (Pixel)gui.def_back_pixel, 6272 XNFontSet, (XFontSet)gui.fontset, 6273 NULL); 6274 } 6275 else 6276 # endif 6277 { 6278 preedit_list = XVaCreateNestedList(0, 6279 XNSpotLocation, &over_spot, 6280 XNForeground, (Pixel)gui.def_norm_pixel, 6281 XNBackground, (Pixel)gui.def_back_pixel, 6282 NULL); 6283 status_list = XVaCreateNestedList(0, 6284 XNForeground, (Pixel)gui.def_norm_pixel, 6285 XNBackground, (Pixel)gui.def_back_pixel, 6286 NULL); 6287 } 6288 6289 xic = XCreateIC(xim, 6290 XNInputStyle, input_style, 6291 XNClientWindow, x11_window, 6292 XNFocusWindow, gui.wid, 6293 XNPreeditAttributes, preedit_list, 6294 XNStatusAttributes, status_list, 6295 NULL); 6296 XFree(status_list); 6297 XFree(preedit_list); 6298 if (xic != NULL) 6299 { 6300 if (input_style & XIMStatusArea) 6301 { 6302 xim_set_status_area(); 6303 status_area_enabled = TRUE; 6304 } 6305 else 6306 gui_set_shellsize(FALSE, FALSE, RESIZE_BOTH); 6307 } 6308 else 6309 { 6310 if (!is_not_a_term()) 6311 emsg(_(e_xim)); 6312 XCloseIM(xim); 6313 return FALSE; 6314 } 6315 6316 return TRUE; 6317 } 6318 6319 # endif // FEAT_GUI_X11 6320 6321 /* 6322 * Get IM status. When IM is on, return TRUE. Else return FALSE. 6323 * FIXME: This doesn't work correctly: Having focus doesn't always mean XIM is 6324 * active, when not having focus XIM may still be active (e.g., when using a 6325 * tear-off menu item). 6326 */ 6327 int 6328 im_get_status(void) 6329 { 6330 # ifdef FEAT_EVAL 6331 if (USE_IMSTATUSFUNC) 6332 return call_imstatusfunc(); 6333 # endif 6334 return xim_has_focus; 6335 } 6336 6337 # endif // !FEAT_GUI_GTK 6338 6339 # if !defined(FEAT_GUI_GTK) || defined(PROTO) 6340 /* 6341 * Set up the status area. 6342 * 6343 * This should use a separate Widget, but that seems not possible, because 6344 * preedit_area and status_area should be set to the same window as for the 6345 * text input. Unfortunately this means the status area pollutes the text 6346 * window... 6347 */ 6348 void 6349 xim_set_status_area(void) 6350 { 6351 XVaNestedList preedit_list = 0, status_list = 0, list = 0; 6352 XRectangle pre_area, status_area; 6353 6354 if (xic == NULL) 6355 return; 6356 6357 if (input_style & XIMStatusArea) 6358 { 6359 if (input_style & XIMPreeditArea) 6360 { 6361 XRectangle *needed_rect; 6362 6363 // to get status_area width 6364 status_list = XVaCreateNestedList(0, XNAreaNeeded, 6365 &needed_rect, NULL); 6366 XGetICValues(xic, XNStatusAttributes, status_list, NULL); 6367 XFree(status_list); 6368 6369 status_area.width = needed_rect->width; 6370 } 6371 else 6372 status_area.width = gui.char_width * Columns; 6373 6374 status_area.x = 0; 6375 status_area.y = gui.char_height * Rows + gui.border_offset; 6376 if (gui.which_scrollbars[SBAR_BOTTOM]) 6377 status_area.y += gui.scrollbar_height; 6378 #ifdef FEAT_MENU 6379 if (gui.menu_is_active) 6380 status_area.y += gui.menu_height; 6381 #endif 6382 status_area.height = gui.char_height; 6383 status_list = XVaCreateNestedList(0, XNArea, &status_area, NULL); 6384 } 6385 else 6386 { 6387 status_area.x = 0; 6388 status_area.y = gui.char_height * Rows + gui.border_offset; 6389 if (gui.which_scrollbars[SBAR_BOTTOM]) 6390 status_area.y += gui.scrollbar_height; 6391 #ifdef FEAT_MENU 6392 if (gui.menu_is_active) 6393 status_area.y += gui.menu_height; 6394 #endif 6395 status_area.width = 0; 6396 status_area.height = gui.char_height; 6397 } 6398 6399 if (input_style & XIMPreeditArea) // off-the-spot 6400 { 6401 pre_area.x = status_area.x + status_area.width; 6402 pre_area.y = gui.char_height * Rows + gui.border_offset; 6403 pre_area.width = gui.char_width * Columns - pre_area.x; 6404 if (gui.which_scrollbars[SBAR_BOTTOM]) 6405 pre_area.y += gui.scrollbar_height; 6406 #ifdef FEAT_MENU 6407 if (gui.menu_is_active) 6408 pre_area.y += gui.menu_height; 6409 #endif 6410 pre_area.height = gui.char_height; 6411 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6412 } 6413 else if (input_style & XIMPreeditPosition) // over-the-spot 6414 { 6415 pre_area.x = 0; 6416 pre_area.y = 0; 6417 pre_area.height = gui.char_height * Rows; 6418 pre_area.width = gui.char_width * Columns; 6419 preedit_list = XVaCreateNestedList(0, XNArea, &pre_area, NULL); 6420 } 6421 6422 if (preedit_list && status_list) 6423 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6424 XNStatusAttributes, status_list, NULL); 6425 else if (preedit_list) 6426 list = XVaCreateNestedList(0, XNPreeditAttributes, preedit_list, 6427 NULL); 6428 else if (status_list) 6429 list = XVaCreateNestedList(0, XNStatusAttributes, status_list, 6430 NULL); 6431 else 6432 list = NULL; 6433 6434 if (list) 6435 { 6436 XSetICValues(xic, XNVaNestedList, list, NULL); 6437 XFree(list); 6438 } 6439 if (status_list) 6440 XFree(status_list); 6441 if (preedit_list) 6442 XFree(preedit_list); 6443 } 6444 6445 int 6446 xim_get_status_area_height(void) 6447 { 6448 if (status_area_enabled) 6449 return gui.char_height; 6450 return 0; 6451 } 6452 # endif 6453 6454 #else // !defined(FEAT_XIM) 6455 6456 # if defined(IME_WITHOUT_XIM) || defined(VIMDLL) 6457 static int im_was_set_active = FALSE; 6458 6459 int 6460 # ifdef VIMDLL 6461 mbyte_im_get_status(void) 6462 # else 6463 im_get_status(void) 6464 # endif 6465 { 6466 # if defined(FEAT_EVAL) 6467 if (USE_IMSTATUSFUNC) 6468 return call_imstatusfunc(); 6469 # endif 6470 return im_was_set_active; 6471 } 6472 6473 void 6474 # ifdef VIMDLL 6475 mbyte_im_set_active(int active_arg) 6476 # else 6477 im_set_active(int active_arg) 6478 # endif 6479 { 6480 # if defined(FEAT_EVAL) 6481 int active = !p_imdisable && active_arg; 6482 6483 if (USE_IMACTIVATEFUNC && active != im_get_status()) 6484 { 6485 call_imactivatefunc(active); 6486 im_was_set_active = active; 6487 } 6488 # endif 6489 } 6490 6491 # if defined(FEAT_GUI) && !defined(VIMDLL) 6492 void 6493 im_set_position(int row UNUSED, int col UNUSED) 6494 { 6495 } 6496 # endif 6497 # endif 6498 6499 #endif // FEAT_XIM 6500 6501 #if defined(FEAT_EVAL) || defined(PROTO) 6502 /* 6503 * "getimstatus()" function 6504 */ 6505 void 6506 f_getimstatus(typval_T *argvars UNUSED, typval_T *rettv) 6507 { 6508 # if defined(HAVE_INPUT_METHOD) 6509 rettv->vval.v_number = im_get_status(); 6510 # endif 6511 } 6512 #endif 6513 6514 /* 6515 * Setup "vcp" for conversion from "from" to "to". 6516 * The names must have been made canonical with enc_canonize(). 6517 * vcp->vc_type must have been initialized to CONV_NONE. 6518 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 6519 * instead). 6520 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 6521 * Return FAIL when conversion is not supported, OK otherwise. 6522 */ 6523 int 6524 convert_setup(vimconv_T *vcp, char_u *from, char_u *to) 6525 { 6526 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 6527 } 6528 6529 /* 6530 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 6531 * "from" unicode charsets be considered utf-8. Same for "to". 6532 */ 6533 int 6534 convert_setup_ext( 6535 vimconv_T *vcp, 6536 char_u *from, 6537 int from_unicode_is_utf8, 6538 char_u *to, 6539 int to_unicode_is_utf8) 6540 { 6541 int from_prop; 6542 int to_prop; 6543 int from_is_utf8; 6544 int to_is_utf8; 6545 6546 // Reset to no conversion. 6547 #ifdef USE_ICONV 6548 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 6549 iconv_close(vcp->vc_fd); 6550 #endif 6551 vcp->vc_type = CONV_NONE; 6552 vcp->vc_factor = 1; 6553 vcp->vc_fail = FALSE; 6554 6555 // No conversion when one of the names is empty or they are equal. 6556 if (from == NULL || *from == NUL || to == NULL || *to == NUL 6557 || STRCMP(from, to) == 0) 6558 return OK; 6559 6560 from_prop = enc_canon_props(from); 6561 to_prop = enc_canon_props(to); 6562 if (from_unicode_is_utf8) 6563 from_is_utf8 = from_prop & ENC_UNICODE; 6564 else 6565 from_is_utf8 = from_prop == ENC_UNICODE; 6566 if (to_unicode_is_utf8) 6567 to_is_utf8 = to_prop & ENC_UNICODE; 6568 else 6569 to_is_utf8 = to_prop == ENC_UNICODE; 6570 6571 if ((from_prop & ENC_LATIN1) && to_is_utf8) 6572 { 6573 // Internal latin1 -> utf-8 conversion. 6574 vcp->vc_type = CONV_TO_UTF8; 6575 vcp->vc_factor = 2; // up to twice as long 6576 } 6577 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 6578 { 6579 // Internal latin9 -> utf-8 conversion. 6580 vcp->vc_type = CONV_9_TO_UTF8; 6581 vcp->vc_factor = 3; // up to three as long (euro sign) 6582 } 6583 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 6584 { 6585 // Internal utf-8 -> latin1 conversion. 6586 vcp->vc_type = CONV_TO_LATIN1; 6587 } 6588 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 6589 { 6590 // Internal utf-8 -> latin9 conversion. 6591 vcp->vc_type = CONV_TO_LATIN9; 6592 } 6593 #ifdef MSWIN 6594 // Win32-specific codepage <-> codepage conversion without iconv. 6595 else if ((from_is_utf8 || encname2codepage(from) > 0) 6596 && (to_is_utf8 || encname2codepage(to) > 0)) 6597 { 6598 vcp->vc_type = CONV_CODEPAGE; 6599 vcp->vc_factor = 2; // up to twice as long 6600 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 6601 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 6602 } 6603 #endif 6604 #ifdef MACOS_CONVERT 6605 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 6606 { 6607 vcp->vc_type = CONV_MAC_LATIN1; 6608 } 6609 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 6610 { 6611 vcp->vc_type = CONV_MAC_UTF8; 6612 vcp->vc_factor = 2; // up to twice as long 6613 } 6614 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 6615 { 6616 vcp->vc_type = CONV_LATIN1_MAC; 6617 } 6618 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 6619 { 6620 vcp->vc_type = CONV_UTF8_MAC; 6621 } 6622 #endif 6623 #ifdef USE_ICONV 6624 else 6625 { 6626 // Use iconv() for conversion. 6627 vcp->vc_fd = (iconv_t)my_iconv_open( 6628 to_is_utf8 ? (char_u *)"utf-8" : to, 6629 from_is_utf8 ? (char_u *)"utf-8" : from); 6630 if (vcp->vc_fd != (iconv_t)-1) 6631 { 6632 vcp->vc_type = CONV_ICONV; 6633 vcp->vc_factor = 4; // could be longer too... 6634 } 6635 } 6636 #endif 6637 if (vcp->vc_type == CONV_NONE) 6638 return FAIL; 6639 6640 return OK; 6641 } 6642 6643 #if defined(FEAT_GUI) || defined(AMIGA) || defined(MSWIN) \ 6644 || defined(PROTO) 6645 /* 6646 * Do conversion on typed input characters in-place. 6647 * The input and output are not NUL terminated! 6648 * Returns the length after conversion. 6649 */ 6650 int 6651 convert_input(char_u *ptr, int len, int maxlen) 6652 { 6653 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 6654 } 6655 #endif 6656 6657 /* 6658 * Like convert_input(), but when there is an incomplete byte sequence at the 6659 * end return that as an allocated string in "restp" and set "*restlenp" to 6660 * the length. If "restp" is NULL it is not used. 6661 */ 6662 int 6663 convert_input_safe( 6664 char_u *ptr, 6665 int len, 6666 int maxlen, 6667 char_u **restp, 6668 int *restlenp) 6669 { 6670 char_u *d; 6671 int dlen = len; 6672 int unconvertlen = 0; 6673 6674 d = string_convert_ext(&input_conv, ptr, &dlen, 6675 restp == NULL ? NULL : &unconvertlen); 6676 if (d != NULL) 6677 { 6678 if (dlen <= maxlen) 6679 { 6680 if (unconvertlen > 0) 6681 { 6682 // Move the unconverted characters to allocated memory. 6683 *restp = alloc(unconvertlen); 6684 if (*restp != NULL) 6685 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 6686 *restlenp = unconvertlen; 6687 } 6688 mch_memmove(ptr, d, dlen); 6689 } 6690 else 6691 // result is too long, keep the unconverted text (the caller must 6692 // have done something wrong!) 6693 dlen = len; 6694 vim_free(d); 6695 } 6696 return dlen; 6697 } 6698 6699 /* 6700 * Convert text "ptr[*lenp]" according to "vcp". 6701 * Returns the result in allocated memory and sets "*lenp". 6702 * When "lenp" is NULL, use NUL terminated strings. 6703 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 6704 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 6705 */ 6706 char_u * 6707 string_convert( 6708 vimconv_T *vcp, 6709 char_u *ptr, 6710 int *lenp) 6711 { 6712 return string_convert_ext(vcp, ptr, lenp, NULL); 6713 } 6714 6715 /* 6716 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 6717 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 6718 * set to the number of remaining bytes. 6719 */ 6720 char_u * 6721 string_convert_ext( 6722 vimconv_T *vcp, 6723 char_u *ptr, 6724 int *lenp, 6725 int *unconvlenp) 6726 { 6727 char_u *retval = NULL; 6728 char_u *d; 6729 int len; 6730 int i; 6731 int l; 6732 int c; 6733 6734 if (lenp == NULL) 6735 len = (int)STRLEN(ptr); 6736 else 6737 len = *lenp; 6738 if (len == 0) 6739 return vim_strsave((char_u *)""); 6740 6741 switch (vcp->vc_type) 6742 { 6743 case CONV_TO_UTF8: // latin1 to utf-8 conversion 6744 retval = alloc(len * 2 + 1); 6745 if (retval == NULL) 6746 break; 6747 d = retval; 6748 for (i = 0; i < len; ++i) 6749 { 6750 c = ptr[i]; 6751 if (c < 0x80) 6752 *d++ = c; 6753 else 6754 { 6755 *d++ = 0xc0 + ((unsigned)c >> 6); 6756 *d++ = 0x80 + (c & 0x3f); 6757 } 6758 } 6759 *d = NUL; 6760 if (lenp != NULL) 6761 *lenp = (int)(d - retval); 6762 break; 6763 6764 case CONV_9_TO_UTF8: // latin9 to utf-8 conversion 6765 retval = alloc(len * 3 + 1); 6766 if (retval == NULL) 6767 break; 6768 d = retval; 6769 for (i = 0; i < len; ++i) 6770 { 6771 c = ptr[i]; 6772 switch (c) 6773 { 6774 case 0xa4: c = 0x20ac; break; // euro 6775 case 0xa6: c = 0x0160; break; // S hat 6776 case 0xa8: c = 0x0161; break; // S -hat 6777 case 0xb4: c = 0x017d; break; // Z hat 6778 case 0xb8: c = 0x017e; break; // Z -hat 6779 case 0xbc: c = 0x0152; break; // OE 6780 case 0xbd: c = 0x0153; break; // oe 6781 case 0xbe: c = 0x0178; break; // Y 6782 } 6783 d += utf_char2bytes(c, d); 6784 } 6785 *d = NUL; 6786 if (lenp != NULL) 6787 *lenp = (int)(d - retval); 6788 break; 6789 6790 case CONV_TO_LATIN1: // utf-8 to latin1 conversion 6791 case CONV_TO_LATIN9: // utf-8 to latin9 conversion 6792 retval = alloc(len + 1); 6793 if (retval == NULL) 6794 break; 6795 d = retval; 6796 for (i = 0; i < len; ++i) 6797 { 6798 l = utf_ptr2len_len(ptr + i, len - i); 6799 if (l == 0) 6800 *d++ = NUL; 6801 else if (l == 1) 6802 { 6803 int l_w = utf8len_tab_zero[ptr[i]]; 6804 6805 if (l_w == 0) 6806 { 6807 // Illegal utf-8 byte cannot be converted 6808 vim_free(retval); 6809 return NULL; 6810 } 6811 if (unconvlenp != NULL && l_w > len - i) 6812 { 6813 // Incomplete sequence at the end. 6814 *unconvlenp = len - i; 6815 break; 6816 } 6817 *d++ = ptr[i]; 6818 } 6819 else 6820 { 6821 c = utf_ptr2char(ptr + i); 6822 if (vcp->vc_type == CONV_TO_LATIN9) 6823 switch (c) 6824 { 6825 case 0x20ac: c = 0xa4; break; // euro 6826 case 0x0160: c = 0xa6; break; // S hat 6827 case 0x0161: c = 0xa8; break; // S -hat 6828 case 0x017d: c = 0xb4; break; // Z hat 6829 case 0x017e: c = 0xb8; break; // Z -hat 6830 case 0x0152: c = 0xbc; break; // OE 6831 case 0x0153: c = 0xbd; break; // oe 6832 case 0x0178: c = 0xbe; break; // Y 6833 case 0xa4: 6834 case 0xa6: 6835 case 0xa8: 6836 case 0xb4: 6837 case 0xb8: 6838 case 0xbc: 6839 case 0xbd: 6840 case 0xbe: c = 0x100; break; // not in latin9 6841 } 6842 if (!utf_iscomposing(c)) // skip composing chars 6843 { 6844 if (c < 0x100) 6845 *d++ = c; 6846 else if (vcp->vc_fail) 6847 { 6848 vim_free(retval); 6849 return NULL; 6850 } 6851 else 6852 { 6853 *d++ = 0xbf; 6854 if (utf_char2cells(c) > 1) 6855 *d++ = '?'; 6856 } 6857 } 6858 i += l - 1; 6859 } 6860 } 6861 *d = NUL; 6862 if (lenp != NULL) 6863 *lenp = (int)(d - retval); 6864 break; 6865 6866 # ifdef MACOS_CONVERT 6867 case CONV_MAC_LATIN1: 6868 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6869 'm', 'l', unconvlenp); 6870 break; 6871 6872 case CONV_LATIN1_MAC: 6873 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6874 'l', 'm', unconvlenp); 6875 break; 6876 6877 case CONV_MAC_UTF8: 6878 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6879 'm', 'u', unconvlenp); 6880 break; 6881 6882 case CONV_UTF8_MAC: 6883 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 6884 'u', 'm', unconvlenp); 6885 break; 6886 # endif 6887 6888 # ifdef USE_ICONV 6889 case CONV_ICONV: // conversion with output_conv.vc_fd 6890 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 6891 break; 6892 # endif 6893 # ifdef MSWIN 6894 case CONV_CODEPAGE: // codepage -> codepage 6895 { 6896 int retlen; 6897 int tmp_len; 6898 short_u *tmp; 6899 6900 // 1. codepage/UTF-8 -> ucs-2. 6901 if (vcp->vc_cpfrom == 0) 6902 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 6903 else 6904 { 6905 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 6906 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 6907 (char *)ptr, len, 0, 0); 6908 if (tmp_len == 0 6909 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 6910 { 6911 if (lenp != NULL) 6912 *lenp = 0; 6913 if (unconvlenp != NULL) 6914 *unconvlenp = len; 6915 retval = alloc(1); 6916 if (retval) 6917 retval[0] = NUL; 6918 return retval; 6919 } 6920 } 6921 tmp = ALLOC_MULT(short_u, tmp_len); 6922 if (tmp == NULL) 6923 break; 6924 if (vcp->vc_cpfrom == 0) 6925 utf8_to_utf16(ptr, len, tmp, unconvlenp); 6926 else 6927 MultiByteToWideChar(vcp->vc_cpfrom, 0, 6928 (char *)ptr, len, tmp, tmp_len); 6929 6930 // 2. ucs-2 -> codepage/UTF-8. 6931 if (vcp->vc_cpto == 0) 6932 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 6933 else 6934 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 6935 tmp, tmp_len, 0, 0, 0, 0); 6936 retval = alloc(retlen + 1); 6937 if (retval != NULL) 6938 { 6939 if (vcp->vc_cpto == 0) 6940 utf16_to_utf8(tmp, tmp_len, retval); 6941 else 6942 WideCharToMultiByte(vcp->vc_cpto, 0, 6943 tmp, tmp_len, 6944 (char *)retval, retlen, 0, 0); 6945 retval[retlen] = NUL; 6946 if (lenp != NULL) 6947 *lenp = retlen; 6948 } 6949 vim_free(tmp); 6950 break; 6951 } 6952 # endif 6953 } 6954 6955 return retval; 6956 } 6957