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 #ifdef HAVE_WCHAR_H 115 # include <wchar.h> 116 #endif 117 118 #if 0 119 // This has been disabled, because several people reported problems with the 120 // wcwidth() and iswprint() library functions, esp. for Hebrew. 121 # ifdef __STDC_ISO_10646__ 122 # define USE_WCHAR_FUNCTIONS 123 # endif 124 #endif 125 126 static int dbcs_char2len(int c); 127 static int dbcs_char2bytes(int c, char_u *buf); 128 static int dbcs_ptr2len(char_u *p); 129 static int dbcs_ptr2len_len(char_u *p, int size); 130 static int utf_ptr2cells_len(char_u *p, int size); 131 static int dbcs_char2cells(int c); 132 static int dbcs_ptr2cells_len(char_u *p, int size); 133 static int dbcs_ptr2char(char_u *p); 134 static int dbcs_head_off(char_u *base, char_u *p); 135 #ifdef FEAT_EVAL 136 static int cw_value(int c); 137 #endif 138 139 /* 140 * Lookup table to quickly get the length in bytes of a UTF-8 character from 141 * the first byte of a UTF-8 string. 142 * Bytes which are illegal when used as the first byte have a 1. 143 * The NUL byte has length 1. 144 */ 145 static char utf8len_tab[256] = 146 { 147 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 148 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 149 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 150 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 151 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 152 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 153 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, 154 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, 155 }; 156 157 /* 158 * Like utf8len_tab above, but using a zero for illegal lead bytes. 159 */ 160 static char utf8len_tab_zero[256] = 161 { 162 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 163 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 164 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 165 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 166 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, 167 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, 168 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, 169 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, 170 }; 171 172 173 /* 174 * Canonical encoding names and their properties. 175 * "iso-8859-n" is handled by enc_canonize() directly. 176 */ 177 static struct 178 { char *name; int prop; int codepage;} 179 enc_canon_table[] = 180 { 181 #define IDX_LATIN_1 0 182 {"latin1", ENC_8BIT + ENC_LATIN1, 1252}, 183 #define IDX_ISO_2 1 184 {"iso-8859-2", ENC_8BIT, 0}, 185 #define IDX_ISO_3 2 186 {"iso-8859-3", ENC_8BIT, 0}, 187 #define IDX_ISO_4 3 188 {"iso-8859-4", ENC_8BIT, 0}, 189 #define IDX_ISO_5 4 190 {"iso-8859-5", ENC_8BIT, 0}, 191 #define IDX_ISO_6 5 192 {"iso-8859-6", ENC_8BIT, 0}, 193 #define IDX_ISO_7 6 194 {"iso-8859-7", ENC_8BIT, 0}, 195 #define IDX_ISO_8 7 196 {"iso-8859-8", ENC_8BIT, 0}, 197 #define IDX_ISO_9 8 198 {"iso-8859-9", ENC_8BIT, 0}, 199 #define IDX_ISO_10 9 200 {"iso-8859-10", ENC_8BIT, 0}, 201 #define IDX_ISO_11 10 202 {"iso-8859-11", ENC_8BIT, 0}, 203 #define IDX_ISO_13 11 204 {"iso-8859-13", ENC_8BIT, 0}, 205 #define IDX_ISO_14 12 206 {"iso-8859-14", ENC_8BIT, 0}, 207 #define IDX_ISO_15 13 208 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0}, 209 #define IDX_KOI8_R 14 210 {"koi8-r", ENC_8BIT, 0}, 211 #define IDX_KOI8_U 15 212 {"koi8-u", ENC_8BIT, 0}, 213 #define IDX_UTF8 16 214 {"utf-8", ENC_UNICODE, 0}, 215 #define IDX_UCS2 17 216 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0}, 217 #define IDX_UCS2LE 18 218 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0}, 219 #define IDX_UTF16 19 220 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0}, 221 #define IDX_UTF16LE 20 222 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0}, 223 #define IDX_UCS4 21 224 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0}, 225 #define IDX_UCS4LE 22 226 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0}, 227 228 // For debugging DBCS encoding on Unix. 229 #define IDX_DEBUG 23 230 {"debug", ENC_DBCS, DBCS_DEBUG}, 231 #define IDX_EUC_JP 24 232 {"euc-jp", ENC_DBCS, DBCS_JPNU}, 233 #define IDX_SJIS 25 234 {"sjis", ENC_DBCS, DBCS_JPN}, 235 #define IDX_EUC_KR 26 236 {"euc-kr", ENC_DBCS, DBCS_KORU}, 237 #define IDX_EUC_CN 27 238 {"euc-cn", ENC_DBCS, DBCS_CHSU}, 239 #define IDX_EUC_TW 28 240 {"euc-tw", ENC_DBCS, DBCS_CHTU}, 241 #define IDX_BIG5 29 242 {"big5", ENC_DBCS, DBCS_CHT}, 243 244 // MS-DOS and MS-Windows codepages are included here, so that they can be 245 // used on Unix too. Most of them are similar to ISO-8859 encodings, but 246 // not exactly the same. 247 #define IDX_CP437 30 248 {"cp437", ENC_8BIT, 437}, // like iso-8859-1 249 #define IDX_CP737 31 250 {"cp737", ENC_8BIT, 737}, // like iso-8859-7 251 #define IDX_CP775 32 252 {"cp775", ENC_8BIT, 775}, // Baltic 253 #define IDX_CP850 33 254 {"cp850", ENC_8BIT, 850}, // like iso-8859-4 255 #define IDX_CP852 34 256 {"cp852", ENC_8BIT, 852}, // like iso-8859-1 257 #define IDX_CP855 35 258 {"cp855", ENC_8BIT, 855}, // like iso-8859-2 259 #define IDX_CP857 36 260 {"cp857", ENC_8BIT, 857}, // like iso-8859-5 261 #define IDX_CP860 37 262 {"cp860", ENC_8BIT, 860}, // like iso-8859-9 263 #define IDX_CP861 38 264 {"cp861", ENC_8BIT, 861}, // like iso-8859-1 265 #define IDX_CP862 39 266 {"cp862", ENC_8BIT, 862}, // like iso-8859-1 267 #define IDX_CP863 40 268 {"cp863", ENC_8BIT, 863}, // like iso-8859-8 269 #define IDX_CP865 41 270 {"cp865", ENC_8BIT, 865}, // like iso-8859-1 271 #define IDX_CP866 42 272 {"cp866", ENC_8BIT, 866}, // like iso-8859-5 273 #define IDX_CP869 43 274 {"cp869", ENC_8BIT, 869}, // like iso-8859-7 275 #define IDX_CP874 44 276 {"cp874", ENC_8BIT, 874}, // Thai 277 #define IDX_CP932 45 278 {"cp932", ENC_DBCS, DBCS_JPN}, 279 #define IDX_CP936 46 280 {"cp936", ENC_DBCS, DBCS_CHS}, 281 #define IDX_CP949 47 282 {"cp949", ENC_DBCS, DBCS_KOR}, 283 #define IDX_CP950 48 284 {"cp950", ENC_DBCS, DBCS_CHT}, 285 #define IDX_CP1250 49 286 {"cp1250", ENC_8BIT, 1250}, // Czech, Polish, etc. 287 #define IDX_CP1251 50 288 {"cp1251", ENC_8BIT, 1251}, // Cyrillic 289 // cp1252 is considered to be equal to latin1 290 #define IDX_CP1253 51 291 {"cp1253", ENC_8BIT, 1253}, // Greek 292 #define IDX_CP1254 52 293 {"cp1254", ENC_8BIT, 1254}, // Turkish 294 #define IDX_CP1255 53 295 {"cp1255", ENC_8BIT, 1255}, // Hebrew 296 #define IDX_CP1256 54 297 {"cp1256", ENC_8BIT, 1256}, // Arabic 298 #define IDX_CP1257 55 299 {"cp1257", ENC_8BIT, 1257}, // Baltic 300 #define IDX_CP1258 56 301 {"cp1258", ENC_8BIT, 1258}, // Vietnamese 302 303 #define IDX_MACROMAN 57 304 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, // Mac OS 305 #define IDX_DECMCS 58 306 {"dec-mcs", ENC_8BIT, 0}, // DEC MCS 307 #define IDX_HPROMAN8 59 308 {"hp-roman8", ENC_8BIT, 0}, // HP Roman8 309 #define IDX_COUNT 60 310 }; 311 312 /* 313 * Aliases for encoding names. 314 */ 315 static struct 316 { char *name; int canon;} 317 enc_alias_table[] = 318 { 319 {"ansi", IDX_LATIN_1}, 320 {"iso-8859-1", IDX_LATIN_1}, 321 {"latin2", IDX_ISO_2}, 322 {"latin3", IDX_ISO_3}, 323 {"latin4", IDX_ISO_4}, 324 {"cyrillic", IDX_ISO_5}, 325 {"arabic", IDX_ISO_6}, 326 {"greek", IDX_ISO_7}, 327 #ifdef MSWIN 328 {"hebrew", IDX_CP1255}, 329 #else 330 {"hebrew", IDX_ISO_8}, 331 #endif 332 {"latin5", IDX_ISO_9}, 333 {"turkish", IDX_ISO_9}, // ? 334 {"latin6", IDX_ISO_10}, 335 {"nordic", IDX_ISO_10}, // ? 336 {"thai", IDX_ISO_11}, // ? 337 {"latin7", IDX_ISO_13}, 338 {"latin8", IDX_ISO_14}, 339 {"latin9", IDX_ISO_15}, 340 {"utf8", IDX_UTF8}, 341 {"unicode", IDX_UCS2}, 342 {"ucs2", IDX_UCS2}, 343 {"ucs2be", IDX_UCS2}, 344 {"ucs-2be", IDX_UCS2}, 345 {"ucs2le", IDX_UCS2LE}, 346 {"utf16", IDX_UTF16}, 347 {"utf16be", IDX_UTF16}, 348 {"utf-16be", IDX_UTF16}, 349 {"utf16le", IDX_UTF16LE}, 350 {"ucs4", IDX_UCS4}, 351 {"ucs4be", IDX_UCS4}, 352 {"ucs-4be", IDX_UCS4}, 353 {"ucs4le", IDX_UCS4LE}, 354 {"utf32", IDX_UCS4}, 355 {"utf-32", IDX_UCS4}, 356 {"utf32be", IDX_UCS4}, 357 {"utf-32be", IDX_UCS4}, 358 {"utf32le", IDX_UCS4LE}, 359 {"utf-32le", IDX_UCS4LE}, 360 {"932", IDX_CP932}, 361 {"949", IDX_CP949}, 362 {"936", IDX_CP936}, 363 {"gbk", IDX_CP936}, 364 {"950", IDX_CP950}, 365 {"eucjp", IDX_EUC_JP}, 366 {"unix-jis", IDX_EUC_JP}, 367 {"ujis", IDX_EUC_JP}, 368 {"shift-jis", IDX_SJIS}, 369 {"pck", IDX_SJIS}, // Sun: PCK 370 {"euckr", IDX_EUC_KR}, 371 {"5601", IDX_EUC_KR}, // Sun: KS C 5601 372 {"euccn", IDX_EUC_CN}, 373 {"gb2312", IDX_EUC_CN}, 374 {"euctw", IDX_EUC_TW}, 375 #if defined(MSWIN) || defined(WIN32UNIX) || defined(MACOS_X) 376 {"japan", IDX_CP932}, 377 {"korea", IDX_CP949}, 378 {"prc", IDX_CP936}, 379 {"chinese", IDX_CP936}, 380 {"taiwan", IDX_CP950}, 381 {"big5", IDX_CP950}, 382 #else 383 {"japan", IDX_EUC_JP}, 384 {"korea", IDX_EUC_KR}, 385 {"prc", IDX_EUC_CN}, 386 {"chinese", IDX_EUC_CN}, 387 {"taiwan", IDX_EUC_TW}, 388 {"cp950", IDX_BIG5}, 389 {"950", IDX_BIG5}, 390 #endif 391 {"mac", IDX_MACROMAN}, 392 {"mac-roman", IDX_MACROMAN}, 393 {NULL, 0} 394 }; 395 396 #ifndef CP_UTF8 397 # define CP_UTF8 65001 // magic number from winnls.h 398 #endif 399 400 /* 401 * Find encoding "name" in the list of canonical encoding names. 402 * Returns -1 if not found. 403 */ 404 static int 405 enc_canon_search(char_u *name) 406 { 407 int i; 408 409 for (i = 0; i < IDX_COUNT; ++i) 410 if (STRCMP(name, enc_canon_table[i].name) == 0) 411 return i; 412 return -1; 413 } 414 415 416 /* 417 * Find canonical encoding "name" in the list and return its properties. 418 * Returns 0 if not found. 419 */ 420 int 421 enc_canon_props(char_u *name) 422 { 423 int i; 424 425 i = enc_canon_search(name); 426 if (i >= 0) 427 return enc_canon_table[i].prop; 428 #ifdef MSWIN 429 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2])) 430 { 431 CPINFO cpinfo; 432 433 // Get info on this codepage to find out what it is. 434 if (GetCPInfo(atoi((char *)name + 2), &cpinfo) != 0) 435 { 436 if (cpinfo.MaxCharSize == 1) // some single-byte encoding 437 return ENC_8BIT; 438 if (cpinfo.MaxCharSize == 2 439 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 440 // must be a DBCS encoding 441 return ENC_DBCS; 442 } 443 return 0; 444 } 445 #endif 446 if (STRNCMP(name, "2byte-", 6) == 0) 447 return ENC_DBCS; 448 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0) 449 return ENC_8BIT; 450 return 0; 451 } 452 453 /* 454 * Set up for using multi-byte characters. 455 * Called in three cases: 456 * - by main() to initialize (p_enc == NULL) 457 * - by set_init_1() after 'encoding' was set to its default. 458 * - by do_set() when 'encoding' has been set. 459 * p_enc must have been passed through enc_canonize() already. 460 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags. 461 * Fills mb_bytelen_tab[] and returns NULL when there are no problems. 462 * When there is something wrong: Returns an error message and doesn't change 463 * anything. 464 */ 465 char * 466 mb_init(void) 467 { 468 int i; 469 int idx; 470 int n; 471 int enc_dbcs_new = 0; 472 #if defined(USE_ICONV) && !defined(MSWIN) && !defined(WIN32UNIX) \ 473 && !defined(MACOS_CONVERT) 474 # define LEN_FROM_CONV 475 vimconv_T vimconv; 476 char_u *p; 477 #endif 478 479 if (p_enc == NULL) 480 { 481 // Just starting up: set the whole table to one's. 482 for (i = 0; i < 256; ++i) 483 mb_bytelen_tab[i] = 1; 484 input_conv.vc_type = CONV_NONE; 485 input_conv.vc_factor = 1; 486 output_conv.vc_type = CONV_NONE; 487 return NULL; 488 } 489 490 #ifdef MSWIN 491 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2])) 492 { 493 CPINFO cpinfo; 494 495 // Get info on this codepage to find out what it is. 496 if (GetCPInfo(atoi((char *)p_enc + 2), &cpinfo) != 0) 497 { 498 if (cpinfo.MaxCharSize == 1) 499 { 500 // some single-byte encoding 501 enc_unicode = 0; 502 enc_utf8 = FALSE; 503 } 504 else if (cpinfo.MaxCharSize == 2 505 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 506 { 507 // must be a DBCS encoding, check below 508 enc_dbcs_new = atoi((char *)p_enc + 2); 509 } 510 else 511 goto codepage_invalid; 512 } 513 else if (GetLastError() == ERROR_INVALID_PARAMETER) 514 { 515 codepage_invalid: 516 return N_("E543: Not a valid codepage"); 517 } 518 } 519 #endif 520 else if (STRNCMP(p_enc, "8bit-", 5) == 0 521 || STRNCMP(p_enc, "iso-8859-", 9) == 0) 522 { 523 // Accept any "8bit-" or "iso-8859-" name. 524 enc_unicode = 0; 525 enc_utf8 = FALSE; 526 } 527 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 528 { 529 #ifdef MSWIN 530 // Windows: accept only valid codepage numbers, check below. 531 if (p_enc[6] != 'c' || p_enc[7] != 'p' 532 || (enc_dbcs_new = atoi((char *)p_enc + 8)) == 0) 533 return e_invarg; 534 #else 535 // Unix: accept any "2byte-" name, assume current locale. 536 enc_dbcs_new = DBCS_2BYTE; 537 #endif 538 } 539 else if ((idx = enc_canon_search(p_enc)) >= 0) 540 { 541 i = enc_canon_table[idx].prop; 542 if (i & ENC_UNICODE) 543 { 544 // Unicode 545 enc_utf8 = TRUE; 546 if (i & (ENC_2BYTE | ENC_2WORD)) 547 enc_unicode = 2; 548 else if (i & ENC_4BYTE) 549 enc_unicode = 4; 550 else 551 enc_unicode = 0; 552 } 553 else if (i & ENC_DBCS) 554 { 555 // 2byte, handle below 556 enc_dbcs_new = enc_canon_table[idx].codepage; 557 } 558 else 559 { 560 // Must be 8-bit. 561 enc_unicode = 0; 562 enc_utf8 = FALSE; 563 } 564 } 565 else // Don't know what encoding this is, reject it. 566 return e_invarg; 567 568 if (enc_dbcs_new != 0) 569 { 570 #ifdef MSWIN 571 // Check if the DBCS code page is OK. 572 if (!IsValidCodePage(enc_dbcs_new)) 573 goto codepage_invalid; 574 #endif 575 enc_unicode = 0; 576 enc_utf8 = FALSE; 577 } 578 enc_dbcs = enc_dbcs_new; 579 has_mbyte = (enc_dbcs != 0 || enc_utf8); 580 581 #if defined(MSWIN) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 582 enc_codepage = encname2codepage(p_enc); 583 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0); 584 #endif 585 586 // Detect an encoding that uses latin1 characters. 587 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0 588 || STRCMP(p_enc, "iso-8859-15") == 0); 589 590 /* 591 * Set the function pointers. 592 */ 593 if (enc_utf8) 594 { 595 mb_ptr2len = utfc_ptr2len; 596 mb_ptr2len_len = utfc_ptr2len_len; 597 mb_char2len = utf_char2len; 598 mb_char2bytes = utf_char2bytes; 599 mb_ptr2cells = utf_ptr2cells; 600 mb_ptr2cells_len = utf_ptr2cells_len; 601 mb_char2cells = utf_char2cells; 602 mb_off2cells = utf_off2cells; 603 mb_ptr2char = utf_ptr2char; 604 mb_head_off = utf_head_off; 605 } 606 else if (enc_dbcs != 0) 607 { 608 mb_ptr2len = dbcs_ptr2len; 609 mb_ptr2len_len = dbcs_ptr2len_len; 610 mb_char2len = dbcs_char2len; 611 mb_char2bytes = dbcs_char2bytes; 612 mb_ptr2cells = dbcs_ptr2cells; 613 mb_ptr2cells_len = dbcs_ptr2cells_len; 614 mb_char2cells = dbcs_char2cells; 615 mb_off2cells = dbcs_off2cells; 616 mb_ptr2char = dbcs_ptr2char; 617 mb_head_off = dbcs_head_off; 618 } 619 else 620 { 621 mb_ptr2len = latin_ptr2len; 622 mb_ptr2len_len = latin_ptr2len_len; 623 mb_char2len = latin_char2len; 624 mb_char2bytes = latin_char2bytes; 625 mb_ptr2cells = latin_ptr2cells; 626 mb_ptr2cells_len = latin_ptr2cells_len; 627 mb_char2cells = latin_char2cells; 628 mb_off2cells = latin_off2cells; 629 mb_ptr2char = latin_ptr2char; 630 mb_head_off = latin_head_off; 631 } 632 633 /* 634 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN(). 635 */ 636 #ifdef LEN_FROM_CONV 637 // When 'encoding' is different from the current locale mblen() won't 638 // work. Use conversion to "utf-8" instead. 639 vimconv.vc_type = CONV_NONE; 640 if (enc_dbcs) 641 { 642 p = enc_locale(); 643 if (p == NULL || STRCMP(p, p_enc) != 0) 644 { 645 convert_setup(&vimconv, p_enc, (char_u *)"utf-8"); 646 vimconv.vc_fail = TRUE; 647 } 648 vim_free(p); 649 } 650 #endif 651 652 for (i = 0; i < 256; ++i) 653 { 654 // Our own function to reliably check the length of UTF-8 characters, 655 // independent of mblen(). 656 if (enc_utf8) 657 n = utf8len_tab[i]; 658 else if (enc_dbcs == 0) 659 n = 1; 660 else 661 { 662 #if defined(MSWIN) || defined(WIN32UNIX) 663 // enc_dbcs is set by setting 'fileencoding'. It becomes a Windows 664 // CodePage identifier, which we can pass directly in to Windows 665 // API 666 n = IsDBCSLeadByteEx(enc_dbcs, (WINBYTE)i) ? 2 : 1; 667 #else 668 # if defined(__amigaos4__) || defined(__ANDROID__) || \ 669 !(defined(HAVE_MBLEN) || defined(X_LOCALE)) 670 /* 671 * if mblen() is not available, character which MSB is turned on 672 * are treated as leading byte character. (note : This assumption 673 * is not always true.) 674 */ 675 n = (i & 0x80) ? 2 : 1; 676 # else 677 char buf[MB_MAXBYTES + 1]; 678 679 if (i == NUL) // just in case mblen() can't handle "" 680 n = 1; 681 else 682 { 683 buf[0] = i; 684 buf[1] = 0; 685 # ifdef LEN_FROM_CONV 686 if (vimconv.vc_type != CONV_NONE) 687 { 688 /* 689 * string_convert() should fail when converting the first 690 * byte of a double-byte character. 691 */ 692 p = string_convert(&vimconv, (char_u *)buf, NULL); 693 if (p != NULL) 694 { 695 vim_free(p); 696 n = 1; 697 } 698 else 699 n = 2; 700 } 701 else 702 # endif 703 { 704 /* 705 * mblen() should return -1 for invalid (means the leading 706 * multibyte) character. However there are some platforms 707 * where mblen() returns 0 for invalid character. 708 * Therefore, following condition includes 0. 709 */ 710 vim_ignored = mblen(NULL, 0); // First reset the state. 711 if (mblen(buf, (size_t)1) <= 0) 712 n = 2; 713 else 714 n = 1; 715 } 716 } 717 # endif 718 #endif 719 } 720 721 mb_bytelen_tab[i] = n; 722 } 723 724 #ifdef LEN_FROM_CONV 725 convert_setup(&vimconv, NULL, NULL); 726 #endif 727 728 // The cell width depends on the type of multi-byte characters. 729 (void)init_chartab(); 730 731 // When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] 732 screenalloc(FALSE); 733 734 // When using Unicode, set default for 'fileencodings'. 735 if (enc_utf8 && !option_was_set((char_u *)"fencs")) 736 set_string_option_direct((char_u *)"fencs", -1, 737 (char_u *)"ucs-bom,utf-8,default,latin1", OPT_FREE, 0); 738 739 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT) 740 // GNU gettext 0.10.37 supports this feature: set the codeset used for 741 // translated messages independently from the current locale. 742 (void)bind_textdomain_codeset(VIMPACKAGE, 743 enc_utf8 ? "utf-8" : (char *)p_enc); 744 #endif 745 746 #ifdef MSWIN 747 // When changing 'encoding' while starting up, then convert the command 748 // line arguments from the active codepage to 'encoding'. 749 if (starting != 0) 750 fix_arg_enc(); 751 #endif 752 753 // Fire an autocommand to let people do custom font setup. This must be 754 // after Vim has been setup for the new encoding. 755 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); 756 757 #ifdef FEAT_SPELL 758 // Need to reload spell dictionaries 759 spell_reload(); 760 #endif 761 762 return NULL; 763 } 764 765 /* 766 * Return the size of the BOM for the current buffer: 767 * 0 - no BOM 768 * 2 - UCS-2 or UTF-16 BOM 769 * 4 - UCS-4 BOM 770 * 3 - UTF-8 BOM 771 */ 772 int 773 bomb_size(void) 774 { 775 int n = 0; 776 777 if (curbuf->b_p_bomb && !curbuf->b_p_bin) 778 { 779 if (*curbuf->b_p_fenc == NUL) 780 { 781 if (enc_utf8) 782 { 783 if (enc_unicode != 0) 784 n = enc_unicode; 785 else 786 n = 3; 787 } 788 } 789 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) 790 n = 3; 791 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 792 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) 793 n = 2; 794 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) 795 n = 4; 796 } 797 return n; 798 } 799 800 #if defined(FEAT_QUICKFIX) || defined(PROTO) 801 /* 802 * Remove all BOM from "s" by moving remaining text. 803 */ 804 void 805 remove_bom(char_u *s) 806 { 807 if (enc_utf8) 808 { 809 char_u *p = s; 810 811 while ((p = vim_strbyte(p, 0xef)) != NULL) 812 { 813 if (p[1] == 0xbb && p[2] == 0xbf) 814 STRMOVE(p, p + 3); 815 else 816 ++p; 817 } 818 } 819 } 820 #endif 821 822 /* 823 * Get class of pointer: 824 * 0 for blank or NUL 825 * 1 for punctuation 826 * 2 for an (ASCII) word character 827 * >2 for other word characters 828 */ 829 int 830 mb_get_class(char_u *p) 831 { 832 return mb_get_class_buf(p, curbuf); 833 } 834 835 int 836 mb_get_class_buf(char_u *p, buf_T *buf) 837 { 838 if (MB_BYTE2LEN(p[0]) == 1) 839 { 840 if (p[0] == NUL || VIM_ISWHITE(p[0])) 841 return 0; 842 if (vim_iswordc_buf(p[0], buf)) 843 return 2; 844 return 1; 845 } 846 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) 847 return dbcs_class(p[0], p[1]); 848 if (enc_utf8) 849 return utf_class_buf(utf_ptr2char(p), buf); 850 return 0; 851 } 852 853 /* 854 * Get class of a double-byte character. This always returns 3 or bigger. 855 * TODO: Should return 1 for punctuation. 856 */ 857 int 858 dbcs_class(unsigned lead, unsigned trail) 859 { 860 switch (enc_dbcs) 861 { 862 // please add classify routine for your language in here 863 864 case DBCS_JPNU: // ? 865 case DBCS_JPN: 866 { 867 // JIS code classification 868 unsigned char lb = lead; 869 unsigned char tb = trail; 870 871 // convert process code to JIS 872 # if defined(MSWIN) || defined(WIN32UNIX) || defined(MACOS_X) 873 // process code is SJIS 874 if (lb <= 0x9f) 875 lb = (lb - 0x81) * 2 + 0x21; 876 else 877 lb = (lb - 0xc1) * 2 + 0x21; 878 if (tb <= 0x7e) 879 tb -= 0x1f; 880 else if (tb <= 0x9e) 881 tb -= 0x20; 882 else 883 { 884 tb -= 0x7e; 885 lb += 1; 886 } 887 # else 888 /* 889 * XXX: Code page identification can not use with all 890 * system! So, some other encoding information 891 * will be needed. 892 * In japanese: SJIS,EUC,UNICODE,(JIS) 893 * Note that JIS-code system don't use as 894 * process code in most system because it uses 895 * escape sequences(JIS is context depend encoding). 896 */ 897 // assume process code is JAPANESE-EUC 898 lb &= 0x7f; 899 tb &= 0x7f; 900 # endif 901 // exceptions 902 switch (lb << 8 | tb) 903 { 904 case 0x2121: // ZENKAKU space 905 return 0; 906 case 0x2122: // TOU-TEN (Japanese comma) 907 case 0x2123: // KU-TEN (Japanese period) 908 case 0x2124: // ZENKAKU comma 909 case 0x2125: // ZENKAKU period 910 return 1; 911 case 0x213c: // prolongedsound handled as KATAKANA 912 return 13; 913 } 914 // sieved by KU code 915 switch (lb) 916 { 917 case 0x21: 918 case 0x22: 919 // special symbols 920 return 10; 921 case 0x23: 922 // alphanumeric 923 return 11; 924 case 0x24: 925 // hiragana 926 return 12; 927 case 0x25: 928 // katakana 929 return 13; 930 case 0x26: 931 // greek 932 return 14; 933 case 0x27: 934 // russian 935 return 15; 936 case 0x28: 937 // lines 938 return 16; 939 default: 940 // kanji 941 return 17; 942 } 943 } 944 945 case DBCS_KORU: // ? 946 case DBCS_KOR: 947 { 948 // KS code classification 949 unsigned char c1 = lead; 950 unsigned char c2 = trail; 951 952 /* 953 * 20 : Hangul 954 * 21 : Hanja 955 * 22 : Symbols 956 * 23 : Alphanumeric/Roman Letter (Full width) 957 * 24 : Hangul Letter(Alphabet) 958 * 25 : Roman Numeral/Greek Letter 959 * 26 : Box Drawings 960 * 27 : Unit Symbols 961 * 28 : Circled/Parenthesized Letter 962 * 29 : Hiragana/Katakana 963 * 30 : Cyrillic Letter 964 */ 965 966 if (c1 >= 0xB0 && c1 <= 0xC8) 967 // Hangul 968 return 20; 969 #if defined(MSWIN) || defined(WIN32UNIX) 970 else if (c1 <= 0xA0 || c2 <= 0xA0) 971 // Extended Hangul Region : MS UHC(Unified Hangul Code) 972 // c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE 973 // c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 974 return 20; 975 #endif 976 977 else if (c1 >= 0xCA && c1 <= 0xFD) 978 // Hanja 979 return 21; 980 else switch (c1) 981 { 982 case 0xA1: 983 case 0xA2: 984 // Symbols 985 return 22; 986 case 0xA3: 987 // Alphanumeric 988 return 23; 989 case 0xA4: 990 // Hangul Letter(Alphabet) 991 return 24; 992 case 0xA5: 993 // Roman Numeral/Greek Letter 994 return 25; 995 case 0xA6: 996 // Box Drawings 997 return 26; 998 case 0xA7: 999 // Unit Symbols 1000 return 27; 1001 case 0xA8: 1002 case 0xA9: 1003 if (c2 <= 0xAF) 1004 return 25; // Roman Letter 1005 else if (c2 >= 0xF6) 1006 return 22; // Symbols 1007 else 1008 // Circled/Parenthesized Letter 1009 return 28; 1010 case 0xAA: 1011 case 0xAB: 1012 // Hiragana/Katakana 1013 return 29; 1014 case 0xAC: 1015 // Cyrillic Letter 1016 return 30; 1017 } 1018 } 1019 default: 1020 break; 1021 } 1022 return 3; 1023 } 1024 1025 /* 1026 * mb_char2len() function pointer. 1027 * Return length in bytes of character "c". 1028 * Returns 1 for a single-byte character. 1029 */ 1030 int 1031 latin_char2len(int c UNUSED) 1032 { 1033 return 1; 1034 } 1035 1036 static int 1037 dbcs_char2len( 1038 int c) 1039 { 1040 if (c >= 0x100) 1041 return 2; 1042 return 1; 1043 } 1044 1045 /* 1046 * mb_char2bytes() function pointer. 1047 * Convert a character to its bytes. 1048 * Returns the length in bytes. 1049 */ 1050 int 1051 latin_char2bytes(int c, char_u *buf) 1052 { 1053 buf[0] = c; 1054 return 1; 1055 } 1056 1057 static int 1058 dbcs_char2bytes(int c, char_u *buf) 1059 { 1060 if (c >= 0x100) 1061 { 1062 buf[0] = (unsigned)c >> 8; 1063 buf[1] = c; 1064 // Never use a NUL byte, it causes lots of trouble. It's an invalid 1065 // character anyway. 1066 if (buf[1] == NUL) 1067 buf[1] = '\n'; 1068 return 2; 1069 } 1070 buf[0] = c; 1071 return 1; 1072 } 1073 1074 /* 1075 * mb_ptr2len() function pointer. 1076 * Get byte length of character at "*p" but stop at a NUL. 1077 * For UTF-8 this includes following composing characters. 1078 * Returns 0 when *p is NUL. 1079 */ 1080 int 1081 latin_ptr2len(char_u *p) 1082 { 1083 return MB_BYTE2LEN(*p); 1084 } 1085 1086 static int 1087 dbcs_ptr2len( 1088 char_u *p) 1089 { 1090 int len; 1091 1092 // Check if second byte is not missing. 1093 len = MB_BYTE2LEN(*p); 1094 if (len == 2 && p[1] == NUL) 1095 len = 1; 1096 return len; 1097 } 1098 1099 /* 1100 * mb_ptr2len_len() function pointer. 1101 * Like mb_ptr2len(), but limit to read "size" bytes. 1102 * Returns 0 for an empty string. 1103 * Returns 1 for an illegal char or an incomplete byte sequence. 1104 */ 1105 int 1106 latin_ptr2len_len(char_u *p, int size) 1107 { 1108 if (size < 1 || *p == NUL) 1109 return 0; 1110 return 1; 1111 } 1112 1113 static int 1114 dbcs_ptr2len_len(char_u *p, int size) 1115 { 1116 int len; 1117 1118 if (size < 1 || *p == NUL) 1119 return 0; 1120 if (size == 1) 1121 return 1; 1122 // Check that second byte is not missing. 1123 len = MB_BYTE2LEN(*p); 1124 if (len == 2 && p[1] == NUL) 1125 len = 1; 1126 return len; 1127 } 1128 1129 struct interval 1130 { 1131 long first; 1132 long last; 1133 }; 1134 1135 /* 1136 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]". 1137 */ 1138 static int 1139 intable(struct interval *table, size_t size, int c) 1140 { 1141 int mid, bot, top; 1142 1143 // first quick check for Latin1 etc. characters 1144 if (c < table[0].first) 1145 return FALSE; 1146 1147 // binary search in table 1148 bot = 0; 1149 top = (int)(size / sizeof(struct interval) - 1); 1150 while (top >= bot) 1151 { 1152 mid = (bot + top) / 2; 1153 if (table[mid].last < c) 1154 bot = mid + 1; 1155 else if (table[mid].first > c) 1156 top = mid - 1; 1157 else 1158 return TRUE; 1159 } 1160 return FALSE; 1161 } 1162 1163 // Sorted list of non-overlapping intervals of East Asian Ambiguous 1164 // characters, generated with ../runtime/tools/unicode.vim. 1165 static struct interval ambiguous[] = 1166 { 1167 {0x00a1, 0x00a1}, 1168 {0x00a4, 0x00a4}, 1169 {0x00a7, 0x00a8}, 1170 {0x00aa, 0x00aa}, 1171 {0x00ad, 0x00ae}, 1172 {0x00b0, 0x00b4}, 1173 {0x00b6, 0x00ba}, 1174 {0x00bc, 0x00bf}, 1175 {0x00c6, 0x00c6}, 1176 {0x00d0, 0x00d0}, 1177 {0x00d7, 0x00d8}, 1178 {0x00de, 0x00e1}, 1179 {0x00e6, 0x00e6}, 1180 {0x00e8, 0x00ea}, 1181 {0x00ec, 0x00ed}, 1182 {0x00f0, 0x00f0}, 1183 {0x00f2, 0x00f3}, 1184 {0x00f7, 0x00fa}, 1185 {0x00fc, 0x00fc}, 1186 {0x00fe, 0x00fe}, 1187 {0x0101, 0x0101}, 1188 {0x0111, 0x0111}, 1189 {0x0113, 0x0113}, 1190 {0x011b, 0x011b}, 1191 {0x0126, 0x0127}, 1192 {0x012b, 0x012b}, 1193 {0x0131, 0x0133}, 1194 {0x0138, 0x0138}, 1195 {0x013f, 0x0142}, 1196 {0x0144, 0x0144}, 1197 {0x0148, 0x014b}, 1198 {0x014d, 0x014d}, 1199 {0x0152, 0x0153}, 1200 {0x0166, 0x0167}, 1201 {0x016b, 0x016b}, 1202 {0x01ce, 0x01ce}, 1203 {0x01d0, 0x01d0}, 1204 {0x01d2, 0x01d2}, 1205 {0x01d4, 0x01d4}, 1206 {0x01d6, 0x01d6}, 1207 {0x01d8, 0x01d8}, 1208 {0x01da, 0x01da}, 1209 {0x01dc, 0x01dc}, 1210 {0x0251, 0x0251}, 1211 {0x0261, 0x0261}, 1212 {0x02c4, 0x02c4}, 1213 {0x02c7, 0x02c7}, 1214 {0x02c9, 0x02cb}, 1215 {0x02cd, 0x02cd}, 1216 {0x02d0, 0x02d0}, 1217 {0x02d8, 0x02db}, 1218 {0x02dd, 0x02dd}, 1219 {0x02df, 0x02df}, 1220 {0x0300, 0x036f}, 1221 {0x0391, 0x03a1}, 1222 {0x03a3, 0x03a9}, 1223 {0x03b1, 0x03c1}, 1224 {0x03c3, 0x03c9}, 1225 {0x0401, 0x0401}, 1226 {0x0410, 0x044f}, 1227 {0x0451, 0x0451}, 1228 {0x2010, 0x2010}, 1229 {0x2013, 0x2016}, 1230 {0x2018, 0x2019}, 1231 {0x201c, 0x201d}, 1232 {0x2020, 0x2022}, 1233 {0x2024, 0x2027}, 1234 {0x2030, 0x2030}, 1235 {0x2032, 0x2033}, 1236 {0x2035, 0x2035}, 1237 {0x203b, 0x203b}, 1238 {0x203e, 0x203e}, 1239 {0x2074, 0x2074}, 1240 {0x207f, 0x207f}, 1241 {0x2081, 0x2084}, 1242 {0x20ac, 0x20ac}, 1243 {0x2103, 0x2103}, 1244 {0x2105, 0x2105}, 1245 {0x2109, 0x2109}, 1246 {0x2113, 0x2113}, 1247 {0x2116, 0x2116}, 1248 {0x2121, 0x2122}, 1249 {0x2126, 0x2126}, 1250 {0x212b, 0x212b}, 1251 {0x2153, 0x2154}, 1252 {0x215b, 0x215e}, 1253 {0x2160, 0x216b}, 1254 {0x2170, 0x2179}, 1255 {0x2189, 0x2189}, 1256 {0x2190, 0x2199}, 1257 {0x21b8, 0x21b9}, 1258 {0x21d2, 0x21d2}, 1259 {0x21d4, 0x21d4}, 1260 {0x21e7, 0x21e7}, 1261 {0x2200, 0x2200}, 1262 {0x2202, 0x2203}, 1263 {0x2207, 0x2208}, 1264 {0x220b, 0x220b}, 1265 {0x220f, 0x220f}, 1266 {0x2211, 0x2211}, 1267 {0x2215, 0x2215}, 1268 {0x221a, 0x221a}, 1269 {0x221d, 0x2220}, 1270 {0x2223, 0x2223}, 1271 {0x2225, 0x2225}, 1272 {0x2227, 0x222c}, 1273 {0x222e, 0x222e}, 1274 {0x2234, 0x2237}, 1275 {0x223c, 0x223d}, 1276 {0x2248, 0x2248}, 1277 {0x224c, 0x224c}, 1278 {0x2252, 0x2252}, 1279 {0x2260, 0x2261}, 1280 {0x2264, 0x2267}, 1281 {0x226a, 0x226b}, 1282 {0x226e, 0x226f}, 1283 {0x2282, 0x2283}, 1284 {0x2286, 0x2287}, 1285 {0x2295, 0x2295}, 1286 {0x2299, 0x2299}, 1287 {0x22a5, 0x22a5}, 1288 {0x22bf, 0x22bf}, 1289 {0x2312, 0x2312}, 1290 {0x2460, 0x24e9}, 1291 {0x24eb, 0x254b}, 1292 {0x2550, 0x2573}, 1293 {0x2580, 0x258f}, 1294 {0x2592, 0x2595}, 1295 {0x25a0, 0x25a1}, 1296 {0x25a3, 0x25a9}, 1297 {0x25b2, 0x25b3}, 1298 {0x25b6, 0x25b7}, 1299 {0x25bc, 0x25bd}, 1300 {0x25c0, 0x25c1}, 1301 {0x25c6, 0x25c8}, 1302 {0x25cb, 0x25cb}, 1303 {0x25ce, 0x25d1}, 1304 {0x25e2, 0x25e5}, 1305 {0x25ef, 0x25ef}, 1306 {0x2605, 0x2606}, 1307 {0x2609, 0x2609}, 1308 {0x260e, 0x260f}, 1309 {0x261c, 0x261c}, 1310 {0x261e, 0x261e}, 1311 {0x2640, 0x2640}, 1312 {0x2642, 0x2642}, 1313 {0x2660, 0x2661}, 1314 {0x2663, 0x2665}, 1315 {0x2667, 0x266a}, 1316 {0x266c, 0x266d}, 1317 {0x266f, 0x266f}, 1318 {0x269e, 0x269f}, 1319 {0x26bf, 0x26bf}, 1320 {0x26c6, 0x26cd}, 1321 {0x26cf, 0x26d3}, 1322 {0x26d5, 0x26e1}, 1323 {0x26e3, 0x26e3}, 1324 {0x26e8, 0x26e9}, 1325 {0x26eb, 0x26f1}, 1326 {0x26f4, 0x26f4}, 1327 {0x26f6, 0x26f9}, 1328 {0x26fb, 0x26fc}, 1329 {0x26fe, 0x26ff}, 1330 {0x273d, 0x273d}, 1331 {0x2776, 0x277f}, 1332 {0x2b56, 0x2b59}, 1333 {0x3248, 0x324f}, 1334 {0xe000, 0xf8ff}, 1335 {0xfe00, 0xfe0f}, 1336 {0xfffd, 0xfffd}, 1337 {0x1f100, 0x1f10a}, 1338 {0x1f110, 0x1f12d}, 1339 {0x1f130, 0x1f169}, 1340 {0x1f170, 0x1f18d}, 1341 {0x1f18f, 0x1f190}, 1342 {0x1f19b, 0x1f1ac}, 1343 {0xe0100, 0xe01ef}, 1344 {0xf0000, 0xffffd}, 1345 {0x100000, 0x10fffd} 1346 }; 1347 1348 #if defined(FEAT_TERMINAL) || defined(PROTO) 1349 /* 1350 * utf_char2cells() with different argument type for libvterm. 1351 */ 1352 int 1353 utf_uint2cells(UINT32_T c) 1354 { 1355 if (c >= 0x100 && utf_iscomposing((int)c)) 1356 return 0; 1357 return utf_char2cells((int)c); 1358 } 1359 #endif 1360 1361 /* 1362 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1363 * Returns 4 or 6 for an unprintable character. 1364 * Is only correct for characters >= 0x80. 1365 * When p_ambw is "double", return 2 for a character with East Asian Width 1366 * class 'A'(mbiguous). 1367 */ 1368 int 1369 utf_char2cells(int c) 1370 { 1371 // Sorted list of non-overlapping intervals of East Asian double width 1372 // characters, generated with ../runtime/tools/unicode.vim. 1373 static struct interval doublewidth[] = 1374 { 1375 {0x1100, 0x115f}, 1376 {0x231a, 0x231b}, 1377 {0x2329, 0x232a}, 1378 {0x23e9, 0x23ec}, 1379 {0x23f0, 0x23f0}, 1380 {0x23f3, 0x23f3}, 1381 {0x25fd, 0x25fe}, 1382 {0x2614, 0x2615}, 1383 {0x2648, 0x2653}, 1384 {0x267f, 0x267f}, 1385 {0x2693, 0x2693}, 1386 {0x26a1, 0x26a1}, 1387 {0x26aa, 0x26ab}, 1388 {0x26bd, 0x26be}, 1389 {0x26c4, 0x26c5}, 1390 {0x26ce, 0x26ce}, 1391 {0x26d4, 0x26d4}, 1392 {0x26ea, 0x26ea}, 1393 {0x26f2, 0x26f5}, 1394 {0x26fa, 0x26fa}, 1395 {0x26fd, 0x26fd}, 1396 {0x2705, 0x2705}, 1397 {0x270a, 0x270b}, 1398 {0x2728, 0x2728}, 1399 {0x274c, 0x274c}, 1400 {0x274e, 0x274e}, 1401 {0x2753, 0x2755}, 1402 {0x2757, 0x2757}, 1403 {0x2795, 0x2797}, 1404 {0x27b0, 0x27b0}, 1405 {0x27bf, 0x27bf}, 1406 {0x2b1b, 0x2b1c}, 1407 {0x2b50, 0x2b50}, 1408 {0x2b55, 0x2b55}, 1409 {0x2e80, 0x2e99}, 1410 {0x2e9b, 0x2ef3}, 1411 {0x2f00, 0x2fd5}, 1412 {0x2ff0, 0x2ffb}, 1413 {0x3000, 0x303e}, 1414 {0x3041, 0x3096}, 1415 {0x3099, 0x30ff}, 1416 {0x3105, 0x312f}, 1417 {0x3131, 0x318e}, 1418 {0x3190, 0x31ba}, 1419 {0x31c0, 0x31e3}, 1420 {0x31f0, 0x321e}, 1421 {0x3220, 0x3247}, 1422 {0x3250, 0x4dbf}, 1423 {0x4e00, 0xa48c}, 1424 {0xa490, 0xa4c6}, 1425 {0xa960, 0xa97c}, 1426 {0xac00, 0xd7a3}, 1427 {0xf900, 0xfaff}, 1428 {0xfe10, 0xfe19}, 1429 {0xfe30, 0xfe52}, 1430 {0xfe54, 0xfe66}, 1431 {0xfe68, 0xfe6b}, 1432 {0xff01, 0xff60}, 1433 {0xffe0, 0xffe6}, 1434 {0x16fe0, 0x16fe3}, 1435 {0x17000, 0x187f7}, 1436 {0x18800, 0x18af2}, 1437 {0x1b000, 0x1b11e}, 1438 {0x1b150, 0x1b152}, 1439 {0x1b164, 0x1b167}, 1440 {0x1b170, 0x1b2fb}, 1441 {0x1f004, 0x1f004}, 1442 {0x1f0cf, 0x1f0cf}, 1443 {0x1f18e, 0x1f18e}, 1444 {0x1f191, 0x1f19a}, 1445 {0x1f200, 0x1f202}, 1446 {0x1f210, 0x1f23b}, 1447 {0x1f240, 0x1f248}, 1448 {0x1f250, 0x1f251}, 1449 {0x1f260, 0x1f265}, 1450 {0x1f300, 0x1f320}, 1451 {0x1f32d, 0x1f335}, 1452 {0x1f337, 0x1f37c}, 1453 {0x1f37e, 0x1f393}, 1454 {0x1f3a0, 0x1f3ca}, 1455 {0x1f3cf, 0x1f3d3}, 1456 {0x1f3e0, 0x1f3f0}, 1457 {0x1f3f4, 0x1f3f4}, 1458 {0x1f3f8, 0x1f43e}, 1459 {0x1f440, 0x1f440}, 1460 {0x1f442, 0x1f4fc}, 1461 {0x1f4ff, 0x1f53d}, 1462 {0x1f54b, 0x1f54e}, 1463 {0x1f550, 0x1f567}, 1464 {0x1f57a, 0x1f57a}, 1465 {0x1f595, 0x1f596}, 1466 {0x1f5a4, 0x1f5a4}, 1467 {0x1f5fb, 0x1f64f}, 1468 {0x1f680, 0x1f6c5}, 1469 {0x1f6cc, 0x1f6cc}, 1470 {0x1f6d0, 0x1f6d2}, 1471 {0x1f6d5, 0x1f6d5}, 1472 {0x1f6eb, 0x1f6ec}, 1473 {0x1f6f4, 0x1f6fa}, 1474 {0x1f7e0, 0x1f7eb}, 1475 {0x1f90d, 0x1f971}, 1476 {0x1f973, 0x1f976}, 1477 {0x1f97a, 0x1f9a2}, 1478 {0x1f9a5, 0x1f9aa}, 1479 {0x1f9ae, 0x1f9ca}, 1480 {0x1f9cd, 0x1f9ff}, 1481 {0x1fa70, 0x1fa73}, 1482 {0x1fa78, 0x1fa7a}, 1483 {0x1fa80, 0x1fa82}, 1484 {0x1fa90, 0x1fa95}, 1485 {0x20000, 0x2fffd}, 1486 {0x30000, 0x3fffd} 1487 }; 1488 1489 // Sorted list of non-overlapping intervals of Emoji characters that don't 1490 // have ambiguous or double width, 1491 // based on http://unicode.org/emoji/charts/emoji-list.html 1492 static struct interval emoji_wide[] = 1493 { 1494 {0x23ed, 0x23ef}, 1495 {0x23f1, 0x23f2}, 1496 {0x23f8, 0x23fa}, 1497 {0x24c2, 0x24c2}, 1498 {0x261d, 0x261d}, 1499 {0x26c8, 0x26c8}, 1500 {0x26cf, 0x26cf}, 1501 {0x26d1, 0x26d1}, 1502 {0x26d3, 0x26d3}, 1503 {0x26e9, 0x26e9}, 1504 {0x26f0, 0x26f1}, 1505 {0x26f7, 0x26f9}, 1506 {0x270c, 0x270d}, 1507 {0x2934, 0x2935}, 1508 {0x1f170, 0x1f189}, 1509 {0x1f1e6, 0x1f1ff}, 1510 {0x1f321, 0x1f321}, 1511 {0x1f324, 0x1f32c}, 1512 {0x1f336, 0x1f336}, 1513 {0x1f37d, 0x1f37d}, 1514 {0x1f396, 0x1f397}, 1515 {0x1f399, 0x1f39b}, 1516 {0x1f39e, 0x1f39f}, 1517 {0x1f3cb, 0x1f3ce}, 1518 {0x1f3d4, 0x1f3df}, 1519 {0x1f3f3, 0x1f3f5}, 1520 {0x1f3f7, 0x1f3f7}, 1521 {0x1f43f, 0x1f43f}, 1522 {0x1f441, 0x1f441}, 1523 {0x1f4fd, 0x1f4fd}, 1524 {0x1f549, 0x1f54a}, 1525 {0x1f56f, 0x1f570}, 1526 {0x1f573, 0x1f579}, 1527 {0x1f587, 0x1f587}, 1528 {0x1f58a, 0x1f58d}, 1529 {0x1f590, 0x1f590}, 1530 {0x1f5a5, 0x1f5a5}, 1531 {0x1f5a8, 0x1f5a8}, 1532 {0x1f5b1, 0x1f5b2}, 1533 {0x1f5bc, 0x1f5bc}, 1534 {0x1f5c2, 0x1f5c4}, 1535 {0x1f5d1, 0x1f5d3}, 1536 {0x1f5dc, 0x1f5de}, 1537 {0x1f5e1, 0x1f5e1}, 1538 {0x1f5e3, 0x1f5e3}, 1539 {0x1f5e8, 0x1f5e8}, 1540 {0x1f5ef, 0x1f5ef}, 1541 {0x1f5f3, 0x1f5f3}, 1542 {0x1f5fa, 0x1f5fa}, 1543 {0x1f6cb, 0x1f6cf}, 1544 {0x1f6e0, 0x1f6e5}, 1545 {0x1f6e9, 0x1f6e9}, 1546 {0x1f6f0, 0x1f6f0}, 1547 {0x1f6f3, 0x1f6f3} 1548 1549 #ifdef MACOS_X 1550 // Include SF Symbols characters, which should be rendered as 1551 // double-width. All of them are in the Supplementary Private Use 1552 // Area-B range. The exact range was determined by downloading the "SF 1553 // Symbols" app from Apple, and then selecting all symbols, copying 1554 // them out, and inspecting the unicode values of them. 1555 , {0x100000, 0x100d7f} 1556 #endif 1557 }; 1558 1559 if (c >= 0x100) 1560 { 1561 #if defined(FEAT_EVAL) || defined(USE_WCHAR_FUNCTIONS) 1562 int n; 1563 #endif 1564 1565 #ifdef FEAT_EVAL 1566 n = cw_value(c); 1567 if (n != 0) 1568 return n; 1569 #endif 1570 1571 #ifdef USE_WCHAR_FUNCTIONS 1572 /* 1573 * Assume the library function wcwidth() works better than our own 1574 * stuff. It should return 1 for ambiguous width chars! 1575 */ 1576 n = wcwidth(c); 1577 1578 if (n < 0) 1579 return 6; // unprintable, displays <xxxx> 1580 if (n > 1) 1581 return n; 1582 #else 1583 if (!utf_printable(c)) 1584 return 6; // unprintable, displays <xxxx> 1585 if (intable(doublewidth, sizeof(doublewidth), c)) 1586 return 2; 1587 #endif 1588 if (p_emoji && intable(emoji_wide, sizeof(emoji_wide), c)) 1589 return 2; 1590 } 1591 1592 // Characters below 0x100 are influenced by 'isprint' option 1593 else if (c >= 0x80 && !vim_isprintc(c)) 1594 return 4; // unprintable, displays <xx> 1595 1596 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c)) 1597 return 2; 1598 1599 return 1; 1600 } 1601 1602 /* 1603 * mb_ptr2cells() function pointer. 1604 * Return the number of display cells character at "*p" occupies. 1605 * This doesn't take care of unprintable characters, use ptr2cells() for that. 1606 */ 1607 int 1608 latin_ptr2cells(char_u *p UNUSED) 1609 { 1610 return 1; 1611 } 1612 1613 int 1614 utf_ptr2cells( 1615 char_u *p) 1616 { 1617 int c; 1618 1619 // Need to convert to a character number. 1620 if (*p >= 0x80) 1621 { 1622 c = utf_ptr2char(p); 1623 // An illegal byte is displayed as <xx>. 1624 if (utf_ptr2len(p) == 1 || c == NUL) 1625 return 4; 1626 // If the char is ASCII it must be an overlong sequence. 1627 if (c < 0x80) 1628 return char2cells(c); 1629 return utf_char2cells(c); 1630 } 1631 return 1; 1632 } 1633 1634 int 1635 dbcs_ptr2cells(char_u *p) 1636 { 1637 // Number of cells is equal to number of bytes, except for euc-jp when 1638 // the first byte is 0x8e. 1639 if (enc_dbcs == DBCS_JPNU && *p == 0x8e) 1640 return 1; 1641 return MB_BYTE2LEN(*p); 1642 } 1643 1644 /* 1645 * mb_ptr2cells_len() function pointer. 1646 * Like mb_ptr2cells(), but limit string length to "size". 1647 * For an empty string or truncated character returns 1. 1648 */ 1649 int 1650 latin_ptr2cells_len(char_u *p UNUSED, int size UNUSED) 1651 { 1652 return 1; 1653 } 1654 1655 static int 1656 utf_ptr2cells_len(char_u *p, int size) 1657 { 1658 int c; 1659 1660 // Need to convert to a wide character. 1661 if (size > 0 && *p >= 0x80) 1662 { 1663 if (utf_ptr2len_len(p, size) < utf8len_tab[*p]) 1664 return 1; // truncated 1665 c = utf_ptr2char(p); 1666 // An illegal byte is displayed as <xx>. 1667 if (utf_ptr2len(p) == 1 || c == NUL) 1668 return 4; 1669 // If the char is ASCII it must be an overlong sequence. 1670 if (c < 0x80) 1671 return char2cells(c); 1672 return utf_char2cells(c); 1673 } 1674 return 1; 1675 } 1676 1677 static int 1678 dbcs_ptr2cells_len(char_u *p, int size) 1679 { 1680 // Number of cells is equal to number of bytes, except for euc-jp when 1681 // the first byte is 0x8e. 1682 if (size <= 1 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)) 1683 return 1; 1684 return MB_BYTE2LEN(*p); 1685 } 1686 1687 /* 1688 * mb_char2cells() function pointer. 1689 * Return the number of display cells character "c" occupies. 1690 * Only takes care of multi-byte chars, not "^C" and such. 1691 */ 1692 int 1693 latin_char2cells(int c UNUSED) 1694 { 1695 return 1; 1696 } 1697 1698 static int 1699 dbcs_char2cells(int c) 1700 { 1701 // Number of cells is equal to number of bytes, except for euc-jp when 1702 // the first byte is 0x8e. 1703 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e) 1704 return 1; 1705 // use the first byte 1706 return MB_BYTE2LEN((unsigned)c >> 8); 1707 } 1708 1709 /* 1710 * Return the number of cells occupied by string "p". 1711 * Stop at a NUL character. When "len" >= 0 stop at character "p[len]". 1712 */ 1713 int 1714 mb_string2cells(char_u *p, int len) 1715 { 1716 int i; 1717 int clen = 0; 1718 1719 for (i = 0; (len < 0 || i < len) && p[i] != NUL; i += (*mb_ptr2len)(p + i)) 1720 clen += (*mb_ptr2cells)(p + i); 1721 return clen; 1722 } 1723 1724 /* 1725 * mb_off2cells() function pointer. 1726 * Return number of display cells for char at ScreenLines[off]. 1727 * We make sure that the offset used is less than "max_off". 1728 */ 1729 int 1730 latin_off2cells(unsigned off UNUSED, unsigned max_off UNUSED) 1731 { 1732 return 1; 1733 } 1734 1735 int 1736 dbcs_off2cells(unsigned off, unsigned max_off) 1737 { 1738 // never check beyond end of the line 1739 if (off >= max_off) 1740 return 1; 1741 1742 // Number of cells is equal to number of bytes, except for euc-jp when 1743 // the first byte is 0x8e. 1744 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e) 1745 return 1; 1746 return MB_BYTE2LEN(ScreenLines[off]); 1747 } 1748 1749 int 1750 utf_off2cells(unsigned off, unsigned max_off) 1751 { 1752 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1; 1753 } 1754 1755 /* 1756 * mb_ptr2char() function pointer. 1757 * Convert a byte sequence into a character. 1758 */ 1759 int 1760 latin_ptr2char(char_u *p) 1761 { 1762 return *p; 1763 } 1764 1765 static int 1766 dbcs_ptr2char(char_u *p) 1767 { 1768 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL) 1769 return (p[0] << 8) + p[1]; 1770 return *p; 1771 } 1772 1773 /* 1774 * Convert a UTF-8 byte sequence to a character number. 1775 * If the sequence is illegal or truncated by a NUL the first byte is 1776 * returned. 1777 * For an overlong sequence this may return zero. 1778 * Does not include composing characters, of course. 1779 */ 1780 int 1781 utf_ptr2char(char_u *p) 1782 { 1783 int len; 1784 1785 if (p[0] < 0x80) // be quick for ASCII 1786 return p[0]; 1787 1788 len = utf8len_tab_zero[p[0]]; 1789 if (len > 1 && (p[1] & 0xc0) == 0x80) 1790 { 1791 if (len == 2) 1792 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); 1793 if ((p[2] & 0xc0) == 0x80) 1794 { 1795 if (len == 3) 1796 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) 1797 + (p[2] & 0x3f); 1798 if ((p[3] & 0xc0) == 0x80) 1799 { 1800 if (len == 4) 1801 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) 1802 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); 1803 if ((p[4] & 0xc0) == 0x80) 1804 { 1805 if (len == 5) 1806 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) 1807 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) 1808 + (p[4] & 0x3f); 1809 if ((p[5] & 0xc0) == 0x80 && len == 6) 1810 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) 1811 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) 1812 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); 1813 } 1814 } 1815 } 1816 } 1817 // Illegal value, just return the first byte 1818 return p[0]; 1819 } 1820 1821 /* 1822 * Convert a UTF-8 byte sequence to a wide character. 1823 * String is assumed to be terminated by NUL or after "n" bytes, whichever 1824 * comes first. 1825 * The function is safe in the sense that it never accesses memory beyond the 1826 * first "n" bytes of "s". 1827 * 1828 * On success, returns decoded codepoint, advances "s" to the beginning of 1829 * next character and decreases "n" accordingly. 1830 * 1831 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past 1832 * NUL byte. 1833 * 1834 * If byte sequence is illegal or incomplete, returns -1 and does not advance 1835 * "s". 1836 */ 1837 static int 1838 utf_safe_read_char_adv(char_u **s, size_t *n) 1839 { 1840 int c, k; 1841 1842 if (*n == 0) // end of buffer 1843 return 0; 1844 1845 k = utf8len_tab_zero[**s]; 1846 1847 if (k == 1) 1848 { 1849 // ASCII character or NUL 1850 (*n)--; 1851 return *(*s)++; 1852 } 1853 1854 if ((size_t)k <= *n) 1855 { 1856 // We have a multibyte sequence and it isn't truncated by buffer 1857 // limits so utf_ptr2char() is safe to use. Or the first byte is 1858 // illegal (k=0), and it's also safe to use utf_ptr2char(). 1859 c = utf_ptr2char(*s); 1860 1861 // On failure, utf_ptr2char() returns the first byte, so here we 1862 // check equality with the first byte. The only non-ASCII character 1863 // which equals the first byte of its own UTF-8 representation is 1864 // U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too. 1865 // It's safe even if n=1, else we would have k=2 > n. 1866 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83)) 1867 { 1868 // byte sequence was successfully decoded 1869 *s += k; 1870 *n -= k; 1871 return c; 1872 } 1873 } 1874 1875 // byte sequence is incomplete or illegal 1876 return -1; 1877 } 1878 1879 /* 1880 * Get character at **pp and advance *pp to the next character. 1881 * Note: composing characters are skipped! 1882 */ 1883 int 1884 mb_ptr2char_adv(char_u **pp) 1885 { 1886 int c; 1887 1888 c = (*mb_ptr2char)(*pp); 1889 *pp += (*mb_ptr2len)(*pp); 1890 return c; 1891 } 1892 1893 /* 1894 * Get character at **pp and advance *pp to the next character. 1895 * Note: composing characters are returned as separate characters. 1896 */ 1897 int 1898 mb_cptr2char_adv(char_u **pp) 1899 { 1900 int c; 1901 1902 c = (*mb_ptr2char)(*pp); 1903 if (enc_utf8) 1904 *pp += utf_ptr2len(*pp); 1905 else 1906 *pp += (*mb_ptr2len)(*pp); 1907 return c; 1908 } 1909 1910 #if defined(FEAT_ARABIC) || defined(PROTO) 1911 /* 1912 * Check if the character pointed to by "p2" is a composing character when it 1913 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which 1914 * behaves like a composing character. 1915 */ 1916 int 1917 utf_composinglike(char_u *p1, char_u *p2) 1918 { 1919 int c2; 1920 1921 c2 = utf_ptr2char(p2); 1922 if (utf_iscomposing(c2)) 1923 return TRUE; 1924 if (!arabic_maycombine(c2)) 1925 return FALSE; 1926 return arabic_combine(utf_ptr2char(p1), c2); 1927 } 1928 #endif 1929 1930 /* 1931 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1932 * composing characters. 1933 */ 1934 int 1935 utfc_ptr2char( 1936 char_u *p, 1937 int *pcc) // return: composing chars, last one is 0 1938 { 1939 int len; 1940 int c; 1941 int cc; 1942 int i = 0; 1943 1944 c = utf_ptr2char(p); 1945 len = utf_ptr2len(p); 1946 1947 // Only accept a composing char when the first char isn't illegal. 1948 if ((len > 1 || *p < 0x80) 1949 && p[len] >= 0x80 1950 && UTF_COMPOSINGLIKE(p, p + len)) 1951 { 1952 cc = utf_ptr2char(p + len); 1953 for (;;) 1954 { 1955 pcc[i++] = cc; 1956 if (i == MAX_MCO) 1957 break; 1958 len += utf_ptr2len(p + len); 1959 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1960 break; 1961 } 1962 } 1963 1964 if (i < MAX_MCO) // last composing char must be 0 1965 pcc[i] = 0; 1966 1967 return c; 1968 } 1969 1970 /* 1971 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1972 * composing characters. Use no more than p[maxlen]. 1973 */ 1974 int 1975 utfc_ptr2char_len( 1976 char_u *p, 1977 int *pcc, // return: composing chars, last one is 0 1978 int maxlen) 1979 { 1980 int len; 1981 int c; 1982 int cc; 1983 int i = 0; 1984 1985 c = utf_ptr2char(p); 1986 len = utf_ptr2len_len(p, maxlen); 1987 // Only accept a composing char when the first char isn't illegal. 1988 if ((len > 1 || *p < 0x80) 1989 && len < maxlen 1990 && p[len] >= 0x80 1991 && UTF_COMPOSINGLIKE(p, p + len)) 1992 { 1993 cc = utf_ptr2char(p + len); 1994 for (;;) 1995 { 1996 pcc[i++] = cc; 1997 if (i == MAX_MCO) 1998 break; 1999 len += utf_ptr2len_len(p + len, maxlen - len); 2000 if (len >= maxlen 2001 || p[len] < 0x80 2002 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 2003 break; 2004 } 2005 } 2006 2007 if (i < MAX_MCO) // last composing char must be 0 2008 pcc[i] = 0; 2009 2010 return c; 2011 } 2012 2013 /* 2014 * Convert the character at screen position "off" to a sequence of bytes. 2015 * Includes the composing characters. 2016 * "buf" must at least have the length MB_MAXBYTES + 1. 2017 * Only to be used when ScreenLinesUC[off] != 0. 2018 * Returns the produced number of bytes. 2019 */ 2020 int 2021 utfc_char2bytes(int off, char_u *buf) 2022 { 2023 int len; 2024 int i; 2025 2026 len = utf_char2bytes(ScreenLinesUC[off], buf); 2027 for (i = 0; i < Screen_mco; ++i) 2028 { 2029 if (ScreenLinesC[i][off] == 0) 2030 break; 2031 len += utf_char2bytes(ScreenLinesC[i][off], buf + len); 2032 } 2033 return len; 2034 } 2035 2036 /* 2037 * Get the length of a UTF-8 byte sequence, not including any following 2038 * composing characters. 2039 * Returns 0 for "". 2040 * Returns 1 for an illegal byte sequence. 2041 */ 2042 int 2043 utf_ptr2len(char_u *p) 2044 { 2045 int len; 2046 int i; 2047 2048 if (*p == NUL) 2049 return 0; 2050 len = utf8len_tab[*p]; 2051 for (i = 1; i < len; ++i) 2052 if ((p[i] & 0xc0) != 0x80) 2053 return 1; 2054 return len; 2055 } 2056 2057 /* 2058 * Return length of UTF-8 character, obtained from the first byte. 2059 * "b" must be between 0 and 255! 2060 * Returns 1 for an invalid first byte value. 2061 */ 2062 int 2063 utf_byte2len(int b) 2064 { 2065 return utf8len_tab[b]; 2066 } 2067 2068 /* 2069 * Get the length of UTF-8 byte sequence "p[size]". Does not include any 2070 * following composing characters. 2071 * Returns 1 for "". 2072 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.). 2073 * Returns number > "size" for an incomplete byte sequence. 2074 * Never returns zero. 2075 */ 2076 int 2077 utf_ptr2len_len(char_u *p, int size) 2078 { 2079 int len; 2080 int i; 2081 int m; 2082 2083 len = utf8len_tab[*p]; 2084 if (len == 1) 2085 return 1; // NUL, ascii or illegal lead byte 2086 if (len > size) 2087 m = size; // incomplete byte sequence. 2088 else 2089 m = len; 2090 for (i = 1; i < m; ++i) 2091 if ((p[i] & 0xc0) != 0x80) 2092 return 1; 2093 return len; 2094 } 2095 2096 /* 2097 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. 2098 * This includes following composing characters. 2099 */ 2100 int 2101 utfc_ptr2len(char_u *p) 2102 { 2103 int len; 2104 int b0 = *p; 2105 #ifdef FEAT_ARABIC 2106 int prevlen; 2107 #endif 2108 2109 if (b0 == NUL) 2110 return 0; 2111 if (b0 < 0x80 && p[1] < 0x80) // be quick for ASCII 2112 return 1; 2113 2114 // Skip over first UTF-8 char, stopping at a NUL byte. 2115 len = utf_ptr2len(p); 2116 2117 // Check for illegal byte. 2118 if (len == 1 && b0 >= 0x80) 2119 return 1; 2120 2121 /* 2122 * Check for composing characters. We can handle only the first six, but 2123 * skip all of them (otherwise the cursor would get stuck). 2124 */ 2125 #ifdef FEAT_ARABIC 2126 prevlen = 0; 2127 #endif 2128 for (;;) 2129 { 2130 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2131 return len; 2132 2133 // Skip over composing char 2134 #ifdef FEAT_ARABIC 2135 prevlen = len; 2136 #endif 2137 len += utf_ptr2len(p + len); 2138 } 2139 } 2140 2141 /* 2142 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" 2143 * takes. This includes following composing characters. 2144 * Returns 0 for an empty string. 2145 * Returns 1 for an illegal char or an incomplete byte sequence. 2146 */ 2147 int 2148 utfc_ptr2len_len(char_u *p, int size) 2149 { 2150 int len; 2151 #ifdef FEAT_ARABIC 2152 int prevlen; 2153 #endif 2154 2155 if (size < 1 || *p == NUL) 2156 return 0; 2157 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) // be quick for ASCII 2158 return 1; 2159 2160 // Skip over first UTF-8 char, stopping at a NUL byte. 2161 len = utf_ptr2len_len(p, size); 2162 2163 // Check for illegal byte and incomplete byte sequence. 2164 if ((len == 1 && p[0] >= 0x80) || len > size) 2165 return 1; 2166 2167 /* 2168 * Check for composing characters. We can handle only the first six, but 2169 * skip all of them (otherwise the cursor would get stuck). 2170 */ 2171 #ifdef FEAT_ARABIC 2172 prevlen = 0; 2173 #endif 2174 while (len < size) 2175 { 2176 int len_next_char; 2177 2178 if (p[len] < 0x80) 2179 break; 2180 2181 /* 2182 * Next character length should not go beyond size to ensure that 2183 * UTF_COMPOSINGLIKE(...) does not read beyond size. 2184 */ 2185 len_next_char = utf_ptr2len_len(p + len, size - len); 2186 if (len_next_char > size - len) 2187 break; 2188 2189 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2190 break; 2191 2192 // Skip over composing char 2193 #ifdef FEAT_ARABIC 2194 prevlen = len; 2195 #endif 2196 len += len_next_char; 2197 } 2198 return len; 2199 } 2200 2201 /* 2202 * Return the number of bytes the UTF-8 encoding of character "c" takes. 2203 * This does not include composing characters. 2204 */ 2205 int 2206 utf_char2len(int c) 2207 { 2208 if (c < 0x80) 2209 return 1; 2210 if (c < 0x800) 2211 return 2; 2212 if (c < 0x10000) 2213 return 3; 2214 if (c < 0x200000) 2215 return 4; 2216 if (c < 0x4000000) 2217 return 5; 2218 return 6; 2219 } 2220 2221 /* 2222 * Convert Unicode character "c" to UTF-8 string in "buf[]". 2223 * Returns the number of bytes. 2224 */ 2225 int 2226 utf_char2bytes(int c, char_u *buf) 2227 { 2228 if (c < 0x80) // 7 bits 2229 { 2230 buf[0] = c; 2231 return 1; 2232 } 2233 if (c < 0x800) // 11 bits 2234 { 2235 buf[0] = 0xc0 + ((unsigned)c >> 6); 2236 buf[1] = 0x80 + (c & 0x3f); 2237 return 2; 2238 } 2239 if (c < 0x10000) // 16 bits 2240 { 2241 buf[0] = 0xe0 + ((unsigned)c >> 12); 2242 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2243 buf[2] = 0x80 + (c & 0x3f); 2244 return 3; 2245 } 2246 if (c < 0x200000) // 21 bits 2247 { 2248 buf[0] = 0xf0 + ((unsigned)c >> 18); 2249 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2250 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2251 buf[3] = 0x80 + (c & 0x3f); 2252 return 4; 2253 } 2254 if (c < 0x4000000) // 26 bits 2255 { 2256 buf[0] = 0xf8 + ((unsigned)c >> 24); 2257 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2258 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2259 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2260 buf[4] = 0x80 + (c & 0x3f); 2261 return 5; 2262 } 2263 // 31 bits 2264 buf[0] = 0xfc + ((unsigned)c >> 30); 2265 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f); 2266 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2267 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2268 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2269 buf[5] = 0x80 + (c & 0x3f); 2270 return 6; 2271 } 2272 2273 #if defined(FEAT_TERMINAL) || defined(PROTO) 2274 /* 2275 * utf_iscomposing() with different argument type for libvterm. 2276 */ 2277 int 2278 utf_iscomposing_uint(UINT32_T c) 2279 { 2280 return utf_iscomposing((int)c); 2281 } 2282 #endif 2283 2284 /* 2285 * Return TRUE if "c" is a composing UTF-8 character. This means it will be 2286 * drawn on top of the preceding character. 2287 * Based on code from Markus Kuhn. 2288 */ 2289 int 2290 utf_iscomposing(int c) 2291 { 2292 // Sorted list of non-overlapping intervals. 2293 // Generated by ../runtime/tools/unicode.vim. 2294 static struct interval combining[] = 2295 { 2296 {0x0300, 0x036f}, 2297 {0x0483, 0x0489}, 2298 {0x0591, 0x05bd}, 2299 {0x05bf, 0x05bf}, 2300 {0x05c1, 0x05c2}, 2301 {0x05c4, 0x05c5}, 2302 {0x05c7, 0x05c7}, 2303 {0x0610, 0x061a}, 2304 {0x064b, 0x065f}, 2305 {0x0670, 0x0670}, 2306 {0x06d6, 0x06dc}, 2307 {0x06df, 0x06e4}, 2308 {0x06e7, 0x06e8}, 2309 {0x06ea, 0x06ed}, 2310 {0x0711, 0x0711}, 2311 {0x0730, 0x074a}, 2312 {0x07a6, 0x07b0}, 2313 {0x07eb, 0x07f3}, 2314 {0x07fd, 0x07fd}, 2315 {0x0816, 0x0819}, 2316 {0x081b, 0x0823}, 2317 {0x0825, 0x0827}, 2318 {0x0829, 0x082d}, 2319 {0x0859, 0x085b}, 2320 {0x08d3, 0x08e1}, 2321 {0x08e3, 0x0903}, 2322 {0x093a, 0x093c}, 2323 {0x093e, 0x094f}, 2324 {0x0951, 0x0957}, 2325 {0x0962, 0x0963}, 2326 {0x0981, 0x0983}, 2327 {0x09bc, 0x09bc}, 2328 {0x09be, 0x09c4}, 2329 {0x09c7, 0x09c8}, 2330 {0x09cb, 0x09cd}, 2331 {0x09d7, 0x09d7}, 2332 {0x09e2, 0x09e3}, 2333 {0x09fe, 0x09fe}, 2334 {0x0a01, 0x0a03}, 2335 {0x0a3c, 0x0a3c}, 2336 {0x0a3e, 0x0a42}, 2337 {0x0a47, 0x0a48}, 2338 {0x0a4b, 0x0a4d}, 2339 {0x0a51, 0x0a51}, 2340 {0x0a70, 0x0a71}, 2341 {0x0a75, 0x0a75}, 2342 {0x0a81, 0x0a83}, 2343 {0x0abc, 0x0abc}, 2344 {0x0abe, 0x0ac5}, 2345 {0x0ac7, 0x0ac9}, 2346 {0x0acb, 0x0acd}, 2347 {0x0ae2, 0x0ae3}, 2348 {0x0afa, 0x0aff}, 2349 {0x0b01, 0x0b03}, 2350 {0x0b3c, 0x0b3c}, 2351 {0x0b3e, 0x0b44}, 2352 {0x0b47, 0x0b48}, 2353 {0x0b4b, 0x0b4d}, 2354 {0x0b56, 0x0b57}, 2355 {0x0b62, 0x0b63}, 2356 {0x0b82, 0x0b82}, 2357 {0x0bbe, 0x0bc2}, 2358 {0x0bc6, 0x0bc8}, 2359 {0x0bca, 0x0bcd}, 2360 {0x0bd7, 0x0bd7}, 2361 {0x0c00, 0x0c04}, 2362 {0x0c3e, 0x0c44}, 2363 {0x0c46, 0x0c48}, 2364 {0x0c4a, 0x0c4d}, 2365 {0x0c55, 0x0c56}, 2366 {0x0c62, 0x0c63}, 2367 {0x0c81, 0x0c83}, 2368 {0x0cbc, 0x0cbc}, 2369 {0x0cbe, 0x0cc4}, 2370 {0x0cc6, 0x0cc8}, 2371 {0x0cca, 0x0ccd}, 2372 {0x0cd5, 0x0cd6}, 2373 {0x0ce2, 0x0ce3}, 2374 {0x0d00, 0x0d03}, 2375 {0x0d3b, 0x0d3c}, 2376 {0x0d3e, 0x0d44}, 2377 {0x0d46, 0x0d48}, 2378 {0x0d4a, 0x0d4d}, 2379 {0x0d57, 0x0d57}, 2380 {0x0d62, 0x0d63}, 2381 {0x0d82, 0x0d83}, 2382 {0x0dca, 0x0dca}, 2383 {0x0dcf, 0x0dd4}, 2384 {0x0dd6, 0x0dd6}, 2385 {0x0dd8, 0x0ddf}, 2386 {0x0df2, 0x0df3}, 2387 {0x0e31, 0x0e31}, 2388 {0x0e34, 0x0e3a}, 2389 {0x0e47, 0x0e4e}, 2390 {0x0eb1, 0x0eb1}, 2391 {0x0eb4, 0x0ebc}, 2392 {0x0ec8, 0x0ecd}, 2393 {0x0f18, 0x0f19}, 2394 {0x0f35, 0x0f35}, 2395 {0x0f37, 0x0f37}, 2396 {0x0f39, 0x0f39}, 2397 {0x0f3e, 0x0f3f}, 2398 {0x0f71, 0x0f84}, 2399 {0x0f86, 0x0f87}, 2400 {0x0f8d, 0x0f97}, 2401 {0x0f99, 0x0fbc}, 2402 {0x0fc6, 0x0fc6}, 2403 {0x102b, 0x103e}, 2404 {0x1056, 0x1059}, 2405 {0x105e, 0x1060}, 2406 {0x1062, 0x1064}, 2407 {0x1067, 0x106d}, 2408 {0x1071, 0x1074}, 2409 {0x1082, 0x108d}, 2410 {0x108f, 0x108f}, 2411 {0x109a, 0x109d}, 2412 {0x135d, 0x135f}, 2413 {0x1712, 0x1714}, 2414 {0x1732, 0x1734}, 2415 {0x1752, 0x1753}, 2416 {0x1772, 0x1773}, 2417 {0x17b4, 0x17d3}, 2418 {0x17dd, 0x17dd}, 2419 {0x180b, 0x180d}, 2420 {0x1885, 0x1886}, 2421 {0x18a9, 0x18a9}, 2422 {0x1920, 0x192b}, 2423 {0x1930, 0x193b}, 2424 {0x1a17, 0x1a1b}, 2425 {0x1a55, 0x1a5e}, 2426 {0x1a60, 0x1a7c}, 2427 {0x1a7f, 0x1a7f}, 2428 {0x1ab0, 0x1abe}, 2429 {0x1b00, 0x1b04}, 2430 {0x1b34, 0x1b44}, 2431 {0x1b6b, 0x1b73}, 2432 {0x1b80, 0x1b82}, 2433 {0x1ba1, 0x1bad}, 2434 {0x1be6, 0x1bf3}, 2435 {0x1c24, 0x1c37}, 2436 {0x1cd0, 0x1cd2}, 2437 {0x1cd4, 0x1ce8}, 2438 {0x1ced, 0x1ced}, 2439 {0x1cf4, 0x1cf4}, 2440 {0x1cf7, 0x1cf9}, 2441 {0x1dc0, 0x1df9}, 2442 {0x1dfb, 0x1dff}, 2443 {0x20d0, 0x20f0}, 2444 {0x2cef, 0x2cf1}, 2445 {0x2d7f, 0x2d7f}, 2446 {0x2de0, 0x2dff}, 2447 {0x302a, 0x302f}, 2448 {0x3099, 0x309a}, 2449 {0xa66f, 0xa672}, 2450 {0xa674, 0xa67d}, 2451 {0xa69e, 0xa69f}, 2452 {0xa6f0, 0xa6f1}, 2453 {0xa802, 0xa802}, 2454 {0xa806, 0xa806}, 2455 {0xa80b, 0xa80b}, 2456 {0xa823, 0xa827}, 2457 {0xa880, 0xa881}, 2458 {0xa8b4, 0xa8c5}, 2459 {0xa8e0, 0xa8f1}, 2460 {0xa8ff, 0xa8ff}, 2461 {0xa926, 0xa92d}, 2462 {0xa947, 0xa953}, 2463 {0xa980, 0xa983}, 2464 {0xa9b3, 0xa9c0}, 2465 {0xa9e5, 0xa9e5}, 2466 {0xaa29, 0xaa36}, 2467 {0xaa43, 0xaa43}, 2468 {0xaa4c, 0xaa4d}, 2469 {0xaa7b, 0xaa7d}, 2470 {0xaab0, 0xaab0}, 2471 {0xaab2, 0xaab4}, 2472 {0xaab7, 0xaab8}, 2473 {0xaabe, 0xaabf}, 2474 {0xaac1, 0xaac1}, 2475 {0xaaeb, 0xaaef}, 2476 {0xaaf5, 0xaaf6}, 2477 {0xabe3, 0xabea}, 2478 {0xabec, 0xabed}, 2479 {0xfb1e, 0xfb1e}, 2480 {0xfe00, 0xfe0f}, 2481 {0xfe20, 0xfe2f}, 2482 {0x101fd, 0x101fd}, 2483 {0x102e0, 0x102e0}, 2484 {0x10376, 0x1037a}, 2485 {0x10a01, 0x10a03}, 2486 {0x10a05, 0x10a06}, 2487 {0x10a0c, 0x10a0f}, 2488 {0x10a38, 0x10a3a}, 2489 {0x10a3f, 0x10a3f}, 2490 {0x10ae5, 0x10ae6}, 2491 {0x10d24, 0x10d27}, 2492 {0x10f46, 0x10f50}, 2493 {0x11000, 0x11002}, 2494 {0x11038, 0x11046}, 2495 {0x1107f, 0x11082}, 2496 {0x110b0, 0x110ba}, 2497 {0x11100, 0x11102}, 2498 {0x11127, 0x11134}, 2499 {0x11145, 0x11146}, 2500 {0x11173, 0x11173}, 2501 {0x11180, 0x11182}, 2502 {0x111b3, 0x111c0}, 2503 {0x111c9, 0x111cc}, 2504 {0x1122c, 0x11237}, 2505 {0x1123e, 0x1123e}, 2506 {0x112df, 0x112ea}, 2507 {0x11300, 0x11303}, 2508 {0x1133b, 0x1133c}, 2509 {0x1133e, 0x11344}, 2510 {0x11347, 0x11348}, 2511 {0x1134b, 0x1134d}, 2512 {0x11357, 0x11357}, 2513 {0x11362, 0x11363}, 2514 {0x11366, 0x1136c}, 2515 {0x11370, 0x11374}, 2516 {0x11435, 0x11446}, 2517 {0x1145e, 0x1145e}, 2518 {0x114b0, 0x114c3}, 2519 {0x115af, 0x115b5}, 2520 {0x115b8, 0x115c0}, 2521 {0x115dc, 0x115dd}, 2522 {0x11630, 0x11640}, 2523 {0x116ab, 0x116b7}, 2524 {0x1171d, 0x1172b}, 2525 {0x1182c, 0x1183a}, 2526 {0x119d1, 0x119d7}, 2527 {0x119da, 0x119e0}, 2528 {0x119e4, 0x119e4}, 2529 {0x11a01, 0x11a0a}, 2530 {0x11a33, 0x11a39}, 2531 {0x11a3b, 0x11a3e}, 2532 {0x11a47, 0x11a47}, 2533 {0x11a51, 0x11a5b}, 2534 {0x11a8a, 0x11a99}, 2535 {0x11c2f, 0x11c36}, 2536 {0x11c38, 0x11c3f}, 2537 {0x11c92, 0x11ca7}, 2538 {0x11ca9, 0x11cb6}, 2539 {0x11d31, 0x11d36}, 2540 {0x11d3a, 0x11d3a}, 2541 {0x11d3c, 0x11d3d}, 2542 {0x11d3f, 0x11d45}, 2543 {0x11d47, 0x11d47}, 2544 {0x11d8a, 0x11d8e}, 2545 {0x11d90, 0x11d91}, 2546 {0x11d93, 0x11d97}, 2547 {0x11ef3, 0x11ef6}, 2548 {0x16af0, 0x16af4}, 2549 {0x16b30, 0x16b36}, 2550 {0x16f4f, 0x16f4f}, 2551 {0x16f51, 0x16f87}, 2552 {0x16f8f, 0x16f92}, 2553 {0x1bc9d, 0x1bc9e}, 2554 {0x1d165, 0x1d169}, 2555 {0x1d16d, 0x1d172}, 2556 {0x1d17b, 0x1d182}, 2557 {0x1d185, 0x1d18b}, 2558 {0x1d1aa, 0x1d1ad}, 2559 {0x1d242, 0x1d244}, 2560 {0x1da00, 0x1da36}, 2561 {0x1da3b, 0x1da6c}, 2562 {0x1da75, 0x1da75}, 2563 {0x1da84, 0x1da84}, 2564 {0x1da9b, 0x1da9f}, 2565 {0x1daa1, 0x1daaf}, 2566 {0x1e000, 0x1e006}, 2567 {0x1e008, 0x1e018}, 2568 {0x1e01b, 0x1e021}, 2569 {0x1e023, 0x1e024}, 2570 {0x1e026, 0x1e02a}, 2571 {0x1e130, 0x1e136}, 2572 {0x1e2ec, 0x1e2ef}, 2573 {0x1e8d0, 0x1e8d6}, 2574 {0x1e944, 0x1e94a}, 2575 {0xe0100, 0xe01ef} 2576 }; 2577 2578 return intable(combining, sizeof(combining), c); 2579 } 2580 2581 /* 2582 * Return TRUE for characters that can be displayed in a normal way. 2583 * Only for characters of 0x100 and above! 2584 */ 2585 int 2586 utf_printable(int c) 2587 { 2588 #ifdef USE_WCHAR_FUNCTIONS 2589 /* 2590 * Assume the iswprint() library function works better than our own stuff. 2591 */ 2592 return iswprint(c); 2593 #else 2594 // Sorted list of non-overlapping intervals. 2595 // 0xd800-0xdfff is reserved for UTF-16, actually illegal. 2596 static struct interval nonprint[] = 2597 { 2598 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2599 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2600 {0xfffe, 0xffff} 2601 }; 2602 2603 return !intable(nonprint, sizeof(nonprint), c); 2604 #endif 2605 } 2606 2607 // Sorted list of non-overlapping intervals of all Emoji characters, 2608 // based on http://unicode.org/emoji/charts/emoji-list.html 2609 // Generated by ../runtime/tools/unicode.vim. 2610 // Excludes 0x00a9 and 0x00ae because they are considered latin1. 2611 static struct interval emoji_all[] = 2612 { 2613 {0x203c, 0x203c}, 2614 {0x2049, 0x2049}, 2615 {0x2122, 0x2122}, 2616 {0x2139, 0x2139}, 2617 {0x2194, 0x2199}, 2618 {0x21a9, 0x21aa}, 2619 {0x231a, 0x231b}, 2620 {0x2328, 0x2328}, 2621 {0x23cf, 0x23cf}, 2622 {0x23e9, 0x23f3}, 2623 {0x23f8, 0x23fa}, 2624 {0x24c2, 0x24c2}, 2625 {0x25aa, 0x25ab}, 2626 {0x25b6, 0x25b6}, 2627 {0x25c0, 0x25c0}, 2628 {0x25fb, 0x25fe}, 2629 {0x2600, 0x2604}, 2630 {0x260e, 0x260e}, 2631 {0x2611, 0x2611}, 2632 {0x2614, 0x2615}, 2633 {0x2618, 0x2618}, 2634 {0x261d, 0x261d}, 2635 {0x2620, 0x2620}, 2636 {0x2622, 0x2623}, 2637 {0x2626, 0x2626}, 2638 {0x262a, 0x262a}, 2639 {0x262e, 0x262f}, 2640 {0x2638, 0x263a}, 2641 {0x2640, 0x2640}, 2642 {0x2642, 0x2642}, 2643 {0x2648, 0x2653}, 2644 {0x265f, 0x2660}, 2645 {0x2663, 0x2663}, 2646 {0x2665, 0x2666}, 2647 {0x2668, 0x2668}, 2648 {0x267b, 0x267b}, 2649 {0x267e, 0x267f}, 2650 {0x2692, 0x2697}, 2651 {0x2699, 0x2699}, 2652 {0x269b, 0x269c}, 2653 {0x26a0, 0x26a1}, 2654 {0x26aa, 0x26ab}, 2655 {0x26b0, 0x26b1}, 2656 {0x26bd, 0x26be}, 2657 {0x26c4, 0x26c5}, 2658 {0x26c8, 0x26c8}, 2659 {0x26ce, 0x26cf}, 2660 {0x26d1, 0x26d1}, 2661 {0x26d3, 0x26d4}, 2662 {0x26e9, 0x26ea}, 2663 {0x26f0, 0x26f5}, 2664 {0x26f7, 0x26fa}, 2665 {0x26fd, 0x26fd}, 2666 {0x2702, 0x2702}, 2667 {0x2705, 0x2705}, 2668 {0x2708, 0x270d}, 2669 {0x270f, 0x270f}, 2670 {0x2712, 0x2712}, 2671 {0x2714, 0x2714}, 2672 {0x2716, 0x2716}, 2673 {0x271d, 0x271d}, 2674 {0x2721, 0x2721}, 2675 {0x2728, 0x2728}, 2676 {0x2733, 0x2734}, 2677 {0x2744, 0x2744}, 2678 {0x2747, 0x2747}, 2679 {0x274c, 0x274c}, 2680 {0x274e, 0x274e}, 2681 {0x2753, 0x2755}, 2682 {0x2757, 0x2757}, 2683 {0x2763, 0x2764}, 2684 {0x2795, 0x2797}, 2685 {0x27a1, 0x27a1}, 2686 {0x27b0, 0x27b0}, 2687 {0x27bf, 0x27bf}, 2688 {0x2934, 0x2935}, 2689 {0x2b05, 0x2b07}, 2690 {0x2b1b, 0x2b1c}, 2691 {0x2b50, 0x2b50}, 2692 {0x2b55, 0x2b55}, 2693 {0x3030, 0x3030}, 2694 {0x303d, 0x303d}, 2695 {0x3297, 0x3297}, 2696 {0x3299, 0x3299}, 2697 {0x1f004, 0x1f004}, 2698 {0x1f0cf, 0x1f0cf}, 2699 {0x1f170, 0x1f189}, 2700 {0x1f18e, 0x1f18e}, 2701 {0x1f191, 0x1f19a}, 2702 {0x1f1e6, 0x1f1ff}, 2703 {0x1f201, 0x1f202}, 2704 {0x1f21a, 0x1f21a}, 2705 {0x1f22f, 0x1f22f}, 2706 {0x1f232, 0x1f23a}, 2707 {0x1f250, 0x1f251}, 2708 {0x1f300, 0x1f321}, 2709 {0x1f324, 0x1f393}, 2710 {0x1f396, 0x1f397}, 2711 {0x1f399, 0x1f39b}, 2712 {0x1f39e, 0x1f3f0}, 2713 {0x1f3f3, 0x1f3f5}, 2714 {0x1f3f7, 0x1f4fd}, 2715 {0x1f4ff, 0x1f53d}, 2716 {0x1f549, 0x1f54e}, 2717 {0x1f550, 0x1f567}, 2718 {0x1f56f, 0x1f570}, 2719 {0x1f573, 0x1f57a}, 2720 {0x1f587, 0x1f587}, 2721 {0x1f58a, 0x1f58d}, 2722 {0x1f590, 0x1f590}, 2723 {0x1f595, 0x1f596}, 2724 {0x1f5a4, 0x1f5a5}, 2725 {0x1f5a8, 0x1f5a8}, 2726 {0x1f5b1, 0x1f5b2}, 2727 {0x1f5bc, 0x1f5bc}, 2728 {0x1f5c2, 0x1f5c4}, 2729 {0x1f5d1, 0x1f5d3}, 2730 {0x1f5dc, 0x1f5de}, 2731 {0x1f5e1, 0x1f5e1}, 2732 {0x1f5e3, 0x1f5e3}, 2733 {0x1f5e8, 0x1f5e8}, 2734 {0x1f5ef, 0x1f5ef}, 2735 {0x1f5f3, 0x1f5f3}, 2736 {0x1f5fa, 0x1f64f}, 2737 {0x1f680, 0x1f6c5}, 2738 {0x1f6cb, 0x1f6d2}, 2739 {0x1f6e0, 0x1f6e5}, 2740 {0x1f6e9, 0x1f6e9}, 2741 {0x1f6eb, 0x1f6ec}, 2742 {0x1f6f0, 0x1f6f0}, 2743 {0x1f6f3, 0x1f6f9}, 2744 {0x1f910, 0x1f93a}, 2745 {0x1f93c, 0x1f93e}, 2746 {0x1f940, 0x1f945}, 2747 {0x1f947, 0x1f970}, 2748 {0x1f973, 0x1f976}, 2749 {0x1f97a, 0x1f97a}, 2750 {0x1f97c, 0x1f9a2}, 2751 {0x1f9b0, 0x1f9b9}, 2752 {0x1f9c0, 0x1f9c2}, 2753 {0x1f9d0, 0x1f9ff} 2754 }; 2755 2756 /* 2757 * Get class of a Unicode character. 2758 * 0: white space 2759 * 1: punctuation 2760 * 2 or bigger: some class of word character. 2761 */ 2762 int 2763 utf_class(int c) 2764 { 2765 return utf_class_buf(c, curbuf); 2766 } 2767 2768 int 2769 utf_class_buf(int c, buf_T *buf) 2770 { 2771 // sorted list of non-overlapping intervals 2772 static struct clinterval 2773 { 2774 unsigned int first; 2775 unsigned int last; 2776 unsigned int class; 2777 } classes[] = 2778 { 2779 {0x037e, 0x037e, 1}, // Greek question mark 2780 {0x0387, 0x0387, 1}, // Greek ano teleia 2781 {0x055a, 0x055f, 1}, // Armenian punctuation 2782 {0x0589, 0x0589, 1}, // Armenian full stop 2783 {0x05be, 0x05be, 1}, 2784 {0x05c0, 0x05c0, 1}, 2785 {0x05c3, 0x05c3, 1}, 2786 {0x05f3, 0x05f4, 1}, 2787 {0x060c, 0x060c, 1}, 2788 {0x061b, 0x061b, 1}, 2789 {0x061f, 0x061f, 1}, 2790 {0x066a, 0x066d, 1}, 2791 {0x06d4, 0x06d4, 1}, 2792 {0x0700, 0x070d, 1}, // Syriac punctuation 2793 {0x0964, 0x0965, 1}, 2794 {0x0970, 0x0970, 1}, 2795 {0x0df4, 0x0df4, 1}, 2796 {0x0e4f, 0x0e4f, 1}, 2797 {0x0e5a, 0x0e5b, 1}, 2798 {0x0f04, 0x0f12, 1}, 2799 {0x0f3a, 0x0f3d, 1}, 2800 {0x0f85, 0x0f85, 1}, 2801 {0x104a, 0x104f, 1}, // Myanmar punctuation 2802 {0x10fb, 0x10fb, 1}, // Georgian punctuation 2803 {0x1361, 0x1368, 1}, // Ethiopic punctuation 2804 {0x166d, 0x166e, 1}, // Canadian Syl. punctuation 2805 {0x1680, 0x1680, 0}, 2806 {0x169b, 0x169c, 1}, 2807 {0x16eb, 0x16ed, 1}, 2808 {0x1735, 0x1736, 1}, 2809 {0x17d4, 0x17dc, 1}, // Khmer punctuation 2810 {0x1800, 0x180a, 1}, // Mongolian punctuation 2811 {0x2000, 0x200b, 0}, // spaces 2812 {0x200c, 0x2027, 1}, // punctuation and symbols 2813 {0x2028, 0x2029, 0}, 2814 {0x202a, 0x202e, 1}, // punctuation and symbols 2815 {0x202f, 0x202f, 0}, 2816 {0x2030, 0x205e, 1}, // punctuation and symbols 2817 {0x205f, 0x205f, 0}, 2818 {0x2060, 0x27ff, 1}, // punctuation and symbols 2819 {0x2070, 0x207f, 0x2070}, // superscript 2820 {0x2080, 0x2094, 0x2080}, // subscript 2821 {0x20a0, 0x27ff, 1}, // all kinds of symbols 2822 {0x2800, 0x28ff, 0x2800}, // braille 2823 {0x2900, 0x2998, 1}, // arrows, brackets, etc. 2824 {0x29d8, 0x29db, 1}, 2825 {0x29fc, 0x29fd, 1}, 2826 {0x2e00, 0x2e7f, 1}, // supplemental punctuation 2827 {0x3000, 0x3000, 0}, // ideographic space 2828 {0x3001, 0x3020, 1}, // ideographic punctuation 2829 {0x3030, 0x3030, 1}, 2830 {0x303d, 0x303d, 1}, 2831 {0x3040, 0x309f, 0x3040}, // Hiragana 2832 {0x30a0, 0x30ff, 0x30a0}, // Katakana 2833 {0x3300, 0x9fff, 0x4e00}, // CJK Ideographs 2834 {0xac00, 0xd7a3, 0xac00}, // Hangul Syllables 2835 {0xf900, 0xfaff, 0x4e00}, // CJK Ideographs 2836 {0xfd3e, 0xfd3f, 1}, 2837 {0xfe30, 0xfe6b, 1}, // punctuation forms 2838 {0xff00, 0xff0f, 1}, // half/fullwidth ASCII 2839 {0xff1a, 0xff20, 1}, // half/fullwidth ASCII 2840 {0xff3b, 0xff40, 1}, // half/fullwidth ASCII 2841 {0xff5b, 0xff65, 1}, // half/fullwidth ASCII 2842 {0x1d000, 0x1d24f, 1}, // Musical notation 2843 {0x1d400, 0x1d7ff, 1}, // Mathematical Alphanumeric Symbols 2844 {0x1f000, 0x1f2ff, 1}, // Game pieces; enclosed characters 2845 {0x1f300, 0x1f9ff, 1}, // Many symbol blocks 2846 {0x20000, 0x2a6df, 0x4e00}, // CJK Ideographs 2847 {0x2a700, 0x2b73f, 0x4e00}, // CJK Ideographs 2848 {0x2b740, 0x2b81f, 0x4e00}, // CJK Ideographs 2849 {0x2f800, 0x2fa1f, 0x4e00}, // CJK Ideographs 2850 }; 2851 2852 int bot = 0; 2853 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2854 int mid; 2855 2856 // First quick check for Latin1 characters, use 'iskeyword'. 2857 if (c < 0x100) 2858 { 2859 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2860 return 0; // blank 2861 if (vim_iswordc_buf(c, buf)) 2862 return 2; // word character 2863 return 1; // punctuation 2864 } 2865 2866 // emoji 2867 if (intable(emoji_all, sizeof(emoji_all), c)) 2868 return 3; 2869 2870 // binary search in table 2871 while (top >= bot) 2872 { 2873 mid = (bot + top) / 2; 2874 if (classes[mid].last < (unsigned int)c) 2875 bot = mid + 1; 2876 else if (classes[mid].first > (unsigned int)c) 2877 top = mid - 1; 2878 else 2879 return (int)classes[mid].class; 2880 } 2881 2882 // most other characters are "word" characters 2883 return 2; 2884 } 2885 2886 int 2887 utf_ambiguous_width(int c) 2888 { 2889 return c >= 0x80 && (intable(ambiguous, sizeof(ambiguous), c) 2890 || intable(emoji_all, sizeof(emoji_all), c)); 2891 } 2892 2893 /* 2894 * Code for Unicode case-dependent operations. Based on notes in 2895 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2896 * This code uses simple case folding, not full case folding. 2897 * Last updated for Unicode 5.2. 2898 */ 2899 2900 /* 2901 * The following tables are built by ../runtime/tools/unicode.vim. 2902 * They must be in numeric order, because we use binary search. 2903 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2904 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2905 * folded/upper/lower by adding 32. 2906 */ 2907 typedef struct 2908 { 2909 int rangeStart; 2910 int rangeEnd; 2911 int step; 2912 int offset; 2913 } convertStruct; 2914 2915 static convertStruct foldCase[] = 2916 { 2917 {0x41,0x5a,1,32}, 2918 {0xb5,0xb5,-1,775}, 2919 {0xc0,0xd6,1,32}, 2920 {0xd8,0xde,1,32}, 2921 {0x100,0x12e,2,1}, 2922 {0x132,0x136,2,1}, 2923 {0x139,0x147,2,1}, 2924 {0x14a,0x176,2,1}, 2925 {0x178,0x178,-1,-121}, 2926 {0x179,0x17d,2,1}, 2927 {0x17f,0x17f,-1,-268}, 2928 {0x181,0x181,-1,210}, 2929 {0x182,0x184,2,1}, 2930 {0x186,0x186,-1,206}, 2931 {0x187,0x187,-1,1}, 2932 {0x189,0x18a,1,205}, 2933 {0x18b,0x18b,-1,1}, 2934 {0x18e,0x18e,-1,79}, 2935 {0x18f,0x18f,-1,202}, 2936 {0x190,0x190,-1,203}, 2937 {0x191,0x191,-1,1}, 2938 {0x193,0x193,-1,205}, 2939 {0x194,0x194,-1,207}, 2940 {0x196,0x196,-1,211}, 2941 {0x197,0x197,-1,209}, 2942 {0x198,0x198,-1,1}, 2943 {0x19c,0x19c,-1,211}, 2944 {0x19d,0x19d,-1,213}, 2945 {0x19f,0x19f,-1,214}, 2946 {0x1a0,0x1a4,2,1}, 2947 {0x1a6,0x1a6,-1,218}, 2948 {0x1a7,0x1a7,-1,1}, 2949 {0x1a9,0x1a9,-1,218}, 2950 {0x1ac,0x1ac,-1,1}, 2951 {0x1ae,0x1ae,-1,218}, 2952 {0x1af,0x1af,-1,1}, 2953 {0x1b1,0x1b2,1,217}, 2954 {0x1b3,0x1b5,2,1}, 2955 {0x1b7,0x1b7,-1,219}, 2956 {0x1b8,0x1bc,4,1}, 2957 {0x1c4,0x1c4,-1,2}, 2958 {0x1c5,0x1c5,-1,1}, 2959 {0x1c7,0x1c7,-1,2}, 2960 {0x1c8,0x1c8,-1,1}, 2961 {0x1ca,0x1ca,-1,2}, 2962 {0x1cb,0x1db,2,1}, 2963 {0x1de,0x1ee,2,1}, 2964 {0x1f1,0x1f1,-1,2}, 2965 {0x1f2,0x1f4,2,1}, 2966 {0x1f6,0x1f6,-1,-97}, 2967 {0x1f7,0x1f7,-1,-56}, 2968 {0x1f8,0x21e,2,1}, 2969 {0x220,0x220,-1,-130}, 2970 {0x222,0x232,2,1}, 2971 {0x23a,0x23a,-1,10795}, 2972 {0x23b,0x23b,-1,1}, 2973 {0x23d,0x23d,-1,-163}, 2974 {0x23e,0x23e,-1,10792}, 2975 {0x241,0x241,-1,1}, 2976 {0x243,0x243,-1,-195}, 2977 {0x244,0x244,-1,69}, 2978 {0x245,0x245,-1,71}, 2979 {0x246,0x24e,2,1}, 2980 {0x345,0x345,-1,116}, 2981 {0x370,0x372,2,1}, 2982 {0x376,0x376,-1,1}, 2983 {0x37f,0x37f,-1,116}, 2984 {0x386,0x386,-1,38}, 2985 {0x388,0x38a,1,37}, 2986 {0x38c,0x38c,-1,64}, 2987 {0x38e,0x38f,1,63}, 2988 {0x391,0x3a1,1,32}, 2989 {0x3a3,0x3ab,1,32}, 2990 {0x3c2,0x3c2,-1,1}, 2991 {0x3cf,0x3cf,-1,8}, 2992 {0x3d0,0x3d0,-1,-30}, 2993 {0x3d1,0x3d1,-1,-25}, 2994 {0x3d5,0x3d5,-1,-15}, 2995 {0x3d6,0x3d6,-1,-22}, 2996 {0x3d8,0x3ee,2,1}, 2997 {0x3f0,0x3f0,-1,-54}, 2998 {0x3f1,0x3f1,-1,-48}, 2999 {0x3f4,0x3f4,-1,-60}, 3000 {0x3f5,0x3f5,-1,-64}, 3001 {0x3f7,0x3f7,-1,1}, 3002 {0x3f9,0x3f9,-1,-7}, 3003 {0x3fa,0x3fa,-1,1}, 3004 {0x3fd,0x3ff,1,-130}, 3005 {0x400,0x40f,1,80}, 3006 {0x410,0x42f,1,32}, 3007 {0x460,0x480,2,1}, 3008 {0x48a,0x4be,2,1}, 3009 {0x4c0,0x4c0,-1,15}, 3010 {0x4c1,0x4cd,2,1}, 3011 {0x4d0,0x52e,2,1}, 3012 {0x531,0x556,1,48}, 3013 {0x10a0,0x10c5,1,7264}, 3014 {0x10c7,0x10cd,6,7264}, 3015 {0x13f8,0x13fd,1,-8}, 3016 {0x1c80,0x1c80,-1,-6222}, 3017 {0x1c81,0x1c81,-1,-6221}, 3018 {0x1c82,0x1c82,-1,-6212}, 3019 {0x1c83,0x1c84,1,-6210}, 3020 {0x1c85,0x1c85,-1,-6211}, 3021 {0x1c86,0x1c86,-1,-6204}, 3022 {0x1c87,0x1c87,-1,-6180}, 3023 {0x1c88,0x1c88,-1,35267}, 3024 {0x1c90,0x1cba,1,-3008}, 3025 {0x1cbd,0x1cbf,1,-3008}, 3026 {0x1e00,0x1e94,2,1}, 3027 {0x1e9b,0x1e9b,-1,-58}, 3028 {0x1e9e,0x1e9e,-1,-7615}, 3029 {0x1ea0,0x1efe,2,1}, 3030 {0x1f08,0x1f0f,1,-8}, 3031 {0x1f18,0x1f1d,1,-8}, 3032 {0x1f28,0x1f2f,1,-8}, 3033 {0x1f38,0x1f3f,1,-8}, 3034 {0x1f48,0x1f4d,1,-8}, 3035 {0x1f59,0x1f5f,2,-8}, 3036 {0x1f68,0x1f6f,1,-8}, 3037 {0x1f88,0x1f8f,1,-8}, 3038 {0x1f98,0x1f9f,1,-8}, 3039 {0x1fa8,0x1faf,1,-8}, 3040 {0x1fb8,0x1fb9,1,-8}, 3041 {0x1fba,0x1fbb,1,-74}, 3042 {0x1fbc,0x1fbc,-1,-9}, 3043 {0x1fbe,0x1fbe,-1,-7173}, 3044 {0x1fc8,0x1fcb,1,-86}, 3045 {0x1fcc,0x1fcc,-1,-9}, 3046 {0x1fd8,0x1fd9,1,-8}, 3047 {0x1fda,0x1fdb,1,-100}, 3048 {0x1fe8,0x1fe9,1,-8}, 3049 {0x1fea,0x1feb,1,-112}, 3050 {0x1fec,0x1fec,-1,-7}, 3051 {0x1ff8,0x1ff9,1,-128}, 3052 {0x1ffa,0x1ffb,1,-126}, 3053 {0x1ffc,0x1ffc,-1,-9}, 3054 {0x2126,0x2126,-1,-7517}, 3055 {0x212a,0x212a,-1,-8383}, 3056 {0x212b,0x212b,-1,-8262}, 3057 {0x2132,0x2132,-1,28}, 3058 {0x2160,0x216f,1,16}, 3059 {0x2183,0x2183,-1,1}, 3060 {0x24b6,0x24cf,1,26}, 3061 {0x2c00,0x2c2e,1,48}, 3062 {0x2c60,0x2c60,-1,1}, 3063 {0x2c62,0x2c62,-1,-10743}, 3064 {0x2c63,0x2c63,-1,-3814}, 3065 {0x2c64,0x2c64,-1,-10727}, 3066 {0x2c67,0x2c6b,2,1}, 3067 {0x2c6d,0x2c6d,-1,-10780}, 3068 {0x2c6e,0x2c6e,-1,-10749}, 3069 {0x2c6f,0x2c6f,-1,-10783}, 3070 {0x2c70,0x2c70,-1,-10782}, 3071 {0x2c72,0x2c75,3,1}, 3072 {0x2c7e,0x2c7f,1,-10815}, 3073 {0x2c80,0x2ce2,2,1}, 3074 {0x2ceb,0x2ced,2,1}, 3075 {0x2cf2,0xa640,31054,1}, 3076 {0xa642,0xa66c,2,1}, 3077 {0xa680,0xa69a,2,1}, 3078 {0xa722,0xa72e,2,1}, 3079 {0xa732,0xa76e,2,1}, 3080 {0xa779,0xa77b,2,1}, 3081 {0xa77d,0xa77d,-1,-35332}, 3082 {0xa77e,0xa786,2,1}, 3083 {0xa78b,0xa78b,-1,1}, 3084 {0xa78d,0xa78d,-1,-42280}, 3085 {0xa790,0xa792,2,1}, 3086 {0xa796,0xa7a8,2,1}, 3087 {0xa7aa,0xa7aa,-1,-42308}, 3088 {0xa7ab,0xa7ab,-1,-42319}, 3089 {0xa7ac,0xa7ac,-1,-42315}, 3090 {0xa7ad,0xa7ad,-1,-42305}, 3091 {0xa7ae,0xa7ae,-1,-42308}, 3092 {0xa7b0,0xa7b0,-1,-42258}, 3093 {0xa7b1,0xa7b1,-1,-42282}, 3094 {0xa7b2,0xa7b2,-1,-42261}, 3095 {0xa7b3,0xa7b3,-1,928}, 3096 {0xa7b4,0xa7be,2,1}, 3097 {0xa7c2,0xa7c2,-1,1}, 3098 {0xa7c4,0xa7c4,-1,-48}, 3099 {0xa7c5,0xa7c5,-1,-42307}, 3100 {0xa7c6,0xa7c6,-1,-35384}, 3101 {0xab70,0xabbf,1,-38864}, 3102 {0xff21,0xff3a,1,32}, 3103 {0x10400,0x10427,1,40}, 3104 {0x104b0,0x104d3,1,40}, 3105 {0x10c80,0x10cb2,1,64}, 3106 {0x118a0,0x118bf,1,32}, 3107 {0x16e40,0x16e5f,1,32}, 3108 {0x1e900,0x1e921,1,34} 3109 }; 3110 3111 /* 3112 * Generic conversion function for case operations. 3113 * Return the converted equivalent of "a", which is a UCS-4 character. Use 3114 * the given conversion "table". Uses binary search on "table". 3115 */ 3116 static int 3117 utf_convert( 3118 int a, 3119 convertStruct table[], 3120 int tableSize) 3121 { 3122 int start, mid, end; // indices into table 3123 int entries = tableSize / sizeof(convertStruct); 3124 3125 start = 0; 3126 end = entries; 3127 while (start < end) 3128 { 3129 // need to search further 3130 mid = (end + start) / 2; 3131 if (table[mid].rangeEnd < a) 3132 start = mid + 1; 3133 else 3134 end = mid; 3135 } 3136 if (start < entries 3137 && table[start].rangeStart <= a 3138 && a <= table[start].rangeEnd 3139 && (a - table[start].rangeStart) % table[start].step == 0) 3140 return (a + table[start].offset); 3141 else 3142 return a; 3143 } 3144 3145 /* 3146 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 3147 * simple case folding. 3148 */ 3149 int 3150 utf_fold(int a) 3151 { 3152 if (a < 0x80) 3153 // be fast for ASCII 3154 return a >= 0x41 && a <= 0x5a ? a + 32 : a; 3155 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 3156 } 3157 3158 static convertStruct toLower[] = 3159 { 3160 {0x41,0x5a,1,32}, 3161 {0xc0,0xd6,1,32}, 3162 {0xd8,0xde,1,32}, 3163 {0x100,0x12e,2,1}, 3164 {0x130,0x130,-1,-199}, 3165 {0x132,0x136,2,1}, 3166 {0x139,0x147,2,1}, 3167 {0x14a,0x176,2,1}, 3168 {0x178,0x178,-1,-121}, 3169 {0x179,0x17d,2,1}, 3170 {0x181,0x181,-1,210}, 3171 {0x182,0x184,2,1}, 3172 {0x186,0x186,-1,206}, 3173 {0x187,0x187,-1,1}, 3174 {0x189,0x18a,1,205}, 3175 {0x18b,0x18b,-1,1}, 3176 {0x18e,0x18e,-1,79}, 3177 {0x18f,0x18f,-1,202}, 3178 {0x190,0x190,-1,203}, 3179 {0x191,0x191,-1,1}, 3180 {0x193,0x193,-1,205}, 3181 {0x194,0x194,-1,207}, 3182 {0x196,0x196,-1,211}, 3183 {0x197,0x197,-1,209}, 3184 {0x198,0x198,-1,1}, 3185 {0x19c,0x19c,-1,211}, 3186 {0x19d,0x19d,-1,213}, 3187 {0x19f,0x19f,-1,214}, 3188 {0x1a0,0x1a4,2,1}, 3189 {0x1a6,0x1a6,-1,218}, 3190 {0x1a7,0x1a7,-1,1}, 3191 {0x1a9,0x1a9,-1,218}, 3192 {0x1ac,0x1ac,-1,1}, 3193 {0x1ae,0x1ae,-1,218}, 3194 {0x1af,0x1af,-1,1}, 3195 {0x1b1,0x1b2,1,217}, 3196 {0x1b3,0x1b5,2,1}, 3197 {0x1b7,0x1b7,-1,219}, 3198 {0x1b8,0x1bc,4,1}, 3199 {0x1c4,0x1c4,-1,2}, 3200 {0x1c5,0x1c5,-1,1}, 3201 {0x1c7,0x1c7,-1,2}, 3202 {0x1c8,0x1c8,-1,1}, 3203 {0x1ca,0x1ca,-1,2}, 3204 {0x1cb,0x1db,2,1}, 3205 {0x1de,0x1ee,2,1}, 3206 {0x1f1,0x1f1,-1,2}, 3207 {0x1f2,0x1f4,2,1}, 3208 {0x1f6,0x1f6,-1,-97}, 3209 {0x1f7,0x1f7,-1,-56}, 3210 {0x1f8,0x21e,2,1}, 3211 {0x220,0x220,-1,-130}, 3212 {0x222,0x232,2,1}, 3213 {0x23a,0x23a,-1,10795}, 3214 {0x23b,0x23b,-1,1}, 3215 {0x23d,0x23d,-1,-163}, 3216 {0x23e,0x23e,-1,10792}, 3217 {0x241,0x241,-1,1}, 3218 {0x243,0x243,-1,-195}, 3219 {0x244,0x244,-1,69}, 3220 {0x245,0x245,-1,71}, 3221 {0x246,0x24e,2,1}, 3222 {0x370,0x372,2,1}, 3223 {0x376,0x376,-1,1}, 3224 {0x37f,0x37f,-1,116}, 3225 {0x386,0x386,-1,38}, 3226 {0x388,0x38a,1,37}, 3227 {0x38c,0x38c,-1,64}, 3228 {0x38e,0x38f,1,63}, 3229 {0x391,0x3a1,1,32}, 3230 {0x3a3,0x3ab,1,32}, 3231 {0x3cf,0x3cf,-1,8}, 3232 {0x3d8,0x3ee,2,1}, 3233 {0x3f4,0x3f4,-1,-60}, 3234 {0x3f7,0x3f7,-1,1}, 3235 {0x3f9,0x3f9,-1,-7}, 3236 {0x3fa,0x3fa,-1,1}, 3237 {0x3fd,0x3ff,1,-130}, 3238 {0x400,0x40f,1,80}, 3239 {0x410,0x42f,1,32}, 3240 {0x460,0x480,2,1}, 3241 {0x48a,0x4be,2,1}, 3242 {0x4c0,0x4c0,-1,15}, 3243 {0x4c1,0x4cd,2,1}, 3244 {0x4d0,0x52e,2,1}, 3245 {0x531,0x556,1,48}, 3246 {0x10a0,0x10c5,1,7264}, 3247 {0x10c7,0x10cd,6,7264}, 3248 {0x13a0,0x13ef,1,38864}, 3249 {0x13f0,0x13f5,1,8}, 3250 {0x1c90,0x1cba,1,-3008}, 3251 {0x1cbd,0x1cbf,1,-3008}, 3252 {0x1e00,0x1e94,2,1}, 3253 {0x1e9e,0x1e9e,-1,-7615}, 3254 {0x1ea0,0x1efe,2,1}, 3255 {0x1f08,0x1f0f,1,-8}, 3256 {0x1f18,0x1f1d,1,-8}, 3257 {0x1f28,0x1f2f,1,-8}, 3258 {0x1f38,0x1f3f,1,-8}, 3259 {0x1f48,0x1f4d,1,-8}, 3260 {0x1f59,0x1f5f,2,-8}, 3261 {0x1f68,0x1f6f,1,-8}, 3262 {0x1f88,0x1f8f,1,-8}, 3263 {0x1f98,0x1f9f,1,-8}, 3264 {0x1fa8,0x1faf,1,-8}, 3265 {0x1fb8,0x1fb9,1,-8}, 3266 {0x1fba,0x1fbb,1,-74}, 3267 {0x1fbc,0x1fbc,-1,-9}, 3268 {0x1fc8,0x1fcb,1,-86}, 3269 {0x1fcc,0x1fcc,-1,-9}, 3270 {0x1fd8,0x1fd9,1,-8}, 3271 {0x1fda,0x1fdb,1,-100}, 3272 {0x1fe8,0x1fe9,1,-8}, 3273 {0x1fea,0x1feb,1,-112}, 3274 {0x1fec,0x1fec,-1,-7}, 3275 {0x1ff8,0x1ff9,1,-128}, 3276 {0x1ffa,0x1ffb,1,-126}, 3277 {0x1ffc,0x1ffc,-1,-9}, 3278 {0x2126,0x2126,-1,-7517}, 3279 {0x212a,0x212a,-1,-8383}, 3280 {0x212b,0x212b,-1,-8262}, 3281 {0x2132,0x2132,-1,28}, 3282 {0x2160,0x216f,1,16}, 3283 {0x2183,0x2183,-1,1}, 3284 {0x24b6,0x24cf,1,26}, 3285 {0x2c00,0x2c2e,1,48}, 3286 {0x2c60,0x2c60,-1,1}, 3287 {0x2c62,0x2c62,-1,-10743}, 3288 {0x2c63,0x2c63,-1,-3814}, 3289 {0x2c64,0x2c64,-1,-10727}, 3290 {0x2c67,0x2c6b,2,1}, 3291 {0x2c6d,0x2c6d,-1,-10780}, 3292 {0x2c6e,0x2c6e,-1,-10749}, 3293 {0x2c6f,0x2c6f,-1,-10783}, 3294 {0x2c70,0x2c70,-1,-10782}, 3295 {0x2c72,0x2c75,3,1}, 3296 {0x2c7e,0x2c7f,1,-10815}, 3297 {0x2c80,0x2ce2,2,1}, 3298 {0x2ceb,0x2ced,2,1}, 3299 {0x2cf2,0xa640,31054,1}, 3300 {0xa642,0xa66c,2,1}, 3301 {0xa680,0xa69a,2,1}, 3302 {0xa722,0xa72e,2,1}, 3303 {0xa732,0xa76e,2,1}, 3304 {0xa779,0xa77b,2,1}, 3305 {0xa77d,0xa77d,-1,-35332}, 3306 {0xa77e,0xa786,2,1}, 3307 {0xa78b,0xa78b,-1,1}, 3308 {0xa78d,0xa78d,-1,-42280}, 3309 {0xa790,0xa792,2,1}, 3310 {0xa796,0xa7a8,2,1}, 3311 {0xa7aa,0xa7aa,-1,-42308}, 3312 {0xa7ab,0xa7ab,-1,-42319}, 3313 {0xa7ac,0xa7ac,-1,-42315}, 3314 {0xa7ad,0xa7ad,-1,-42305}, 3315 {0xa7ae,0xa7ae,-1,-42308}, 3316 {0xa7b0,0xa7b0,-1,-42258}, 3317 {0xa7b1,0xa7b1,-1,-42282}, 3318 {0xa7b2,0xa7b2,-1,-42261}, 3319 {0xa7b3,0xa7b3,-1,928}, 3320 {0xa7b4,0xa7be,2,1}, 3321 {0xa7c2,0xa7c2,-1,1}, 3322 {0xa7c4,0xa7c4,-1,-48}, 3323 {0xa7c5,0xa7c5,-1,-42307}, 3324 {0xa7c6,0xa7c6,-1,-35384}, 3325 {0xff21,0xff3a,1,32}, 3326 {0x10400,0x10427,1,40}, 3327 {0x104b0,0x104d3,1,40}, 3328 {0x10c80,0x10cb2,1,64}, 3329 {0x118a0,0x118bf,1,32}, 3330 {0x16e40,0x16e5f,1,32}, 3331 {0x1e900,0x1e921,1,34} 3332 }; 3333 3334 static convertStruct toUpper[] = 3335 { 3336 {0x61,0x7a,1,-32}, 3337 {0xb5,0xb5,-1,743}, 3338 {0xe0,0xf6,1,-32}, 3339 {0xf8,0xfe,1,-32}, 3340 {0xff,0xff,-1,121}, 3341 {0x101,0x12f,2,-1}, 3342 {0x131,0x131,-1,-232}, 3343 {0x133,0x137,2,-1}, 3344 {0x13a,0x148,2,-1}, 3345 {0x14b,0x177,2,-1}, 3346 {0x17a,0x17e,2,-1}, 3347 {0x17f,0x17f,-1,-300}, 3348 {0x180,0x180,-1,195}, 3349 {0x183,0x185,2,-1}, 3350 {0x188,0x18c,4,-1}, 3351 {0x192,0x192,-1,-1}, 3352 {0x195,0x195,-1,97}, 3353 {0x199,0x199,-1,-1}, 3354 {0x19a,0x19a,-1,163}, 3355 {0x19e,0x19e,-1,130}, 3356 {0x1a1,0x1a5,2,-1}, 3357 {0x1a8,0x1ad,5,-1}, 3358 {0x1b0,0x1b4,4,-1}, 3359 {0x1b6,0x1b9,3,-1}, 3360 {0x1bd,0x1bd,-1,-1}, 3361 {0x1bf,0x1bf,-1,56}, 3362 {0x1c5,0x1c5,-1,-1}, 3363 {0x1c6,0x1c6,-1,-2}, 3364 {0x1c8,0x1c8,-1,-1}, 3365 {0x1c9,0x1c9,-1,-2}, 3366 {0x1cb,0x1cb,-1,-1}, 3367 {0x1cc,0x1cc,-1,-2}, 3368 {0x1ce,0x1dc,2,-1}, 3369 {0x1dd,0x1dd,-1,-79}, 3370 {0x1df,0x1ef,2,-1}, 3371 {0x1f2,0x1f2,-1,-1}, 3372 {0x1f3,0x1f3,-1,-2}, 3373 {0x1f5,0x1f9,4,-1}, 3374 {0x1fb,0x21f,2,-1}, 3375 {0x223,0x233,2,-1}, 3376 {0x23c,0x23c,-1,-1}, 3377 {0x23f,0x240,1,10815}, 3378 {0x242,0x247,5,-1}, 3379 {0x249,0x24f,2,-1}, 3380 {0x250,0x250,-1,10783}, 3381 {0x251,0x251,-1,10780}, 3382 {0x252,0x252,-1,10782}, 3383 {0x253,0x253,-1,-210}, 3384 {0x254,0x254,-1,-206}, 3385 {0x256,0x257,1,-205}, 3386 {0x259,0x259,-1,-202}, 3387 {0x25b,0x25b,-1,-203}, 3388 {0x25c,0x25c,-1,42319}, 3389 {0x260,0x260,-1,-205}, 3390 {0x261,0x261,-1,42315}, 3391 {0x263,0x263,-1,-207}, 3392 {0x265,0x265,-1,42280}, 3393 {0x266,0x266,-1,42308}, 3394 {0x268,0x268,-1,-209}, 3395 {0x269,0x269,-1,-211}, 3396 {0x26a,0x26a,-1,42308}, 3397 {0x26b,0x26b,-1,10743}, 3398 {0x26c,0x26c,-1,42305}, 3399 {0x26f,0x26f,-1,-211}, 3400 {0x271,0x271,-1,10749}, 3401 {0x272,0x272,-1,-213}, 3402 {0x275,0x275,-1,-214}, 3403 {0x27d,0x27d,-1,10727}, 3404 {0x280,0x280,-1,-218}, 3405 {0x282,0x282,-1,42307}, 3406 {0x283,0x283,-1,-218}, 3407 {0x287,0x287,-1,42282}, 3408 {0x288,0x288,-1,-218}, 3409 {0x289,0x289,-1,-69}, 3410 {0x28a,0x28b,1,-217}, 3411 {0x28c,0x28c,-1,-71}, 3412 {0x292,0x292,-1,-219}, 3413 {0x29d,0x29d,-1,42261}, 3414 {0x29e,0x29e,-1,42258}, 3415 {0x345,0x345,-1,84}, 3416 {0x371,0x373,2,-1}, 3417 {0x377,0x377,-1,-1}, 3418 {0x37b,0x37d,1,130}, 3419 {0x3ac,0x3ac,-1,-38}, 3420 {0x3ad,0x3af,1,-37}, 3421 {0x3b1,0x3c1,1,-32}, 3422 {0x3c2,0x3c2,-1,-31}, 3423 {0x3c3,0x3cb,1,-32}, 3424 {0x3cc,0x3cc,-1,-64}, 3425 {0x3cd,0x3ce,1,-63}, 3426 {0x3d0,0x3d0,-1,-62}, 3427 {0x3d1,0x3d1,-1,-57}, 3428 {0x3d5,0x3d5,-1,-47}, 3429 {0x3d6,0x3d6,-1,-54}, 3430 {0x3d7,0x3d7,-1,-8}, 3431 {0x3d9,0x3ef,2,-1}, 3432 {0x3f0,0x3f0,-1,-86}, 3433 {0x3f1,0x3f1,-1,-80}, 3434 {0x3f2,0x3f2,-1,7}, 3435 {0x3f3,0x3f3,-1,-116}, 3436 {0x3f5,0x3f5,-1,-96}, 3437 {0x3f8,0x3fb,3,-1}, 3438 {0x430,0x44f,1,-32}, 3439 {0x450,0x45f,1,-80}, 3440 {0x461,0x481,2,-1}, 3441 {0x48b,0x4bf,2,-1}, 3442 {0x4c2,0x4ce,2,-1}, 3443 {0x4cf,0x4cf,-1,-15}, 3444 {0x4d1,0x52f,2,-1}, 3445 {0x561,0x586,1,-48}, 3446 {0x10d0,0x10fa,1,3008}, 3447 {0x10fd,0x10ff,1,3008}, 3448 {0x13f8,0x13fd,1,-8}, 3449 {0x1c80,0x1c80,-1,-6254}, 3450 {0x1c81,0x1c81,-1,-6253}, 3451 {0x1c82,0x1c82,-1,-6244}, 3452 {0x1c83,0x1c84,1,-6242}, 3453 {0x1c85,0x1c85,-1,-6243}, 3454 {0x1c86,0x1c86,-1,-6236}, 3455 {0x1c87,0x1c87,-1,-6181}, 3456 {0x1c88,0x1c88,-1,35266}, 3457 {0x1d79,0x1d79,-1,35332}, 3458 {0x1d7d,0x1d7d,-1,3814}, 3459 {0x1d8e,0x1d8e,-1,35384}, 3460 {0x1e01,0x1e95,2,-1}, 3461 {0x1e9b,0x1e9b,-1,-59}, 3462 {0x1ea1,0x1eff,2,-1}, 3463 {0x1f00,0x1f07,1,8}, 3464 {0x1f10,0x1f15,1,8}, 3465 {0x1f20,0x1f27,1,8}, 3466 {0x1f30,0x1f37,1,8}, 3467 {0x1f40,0x1f45,1,8}, 3468 {0x1f51,0x1f57,2,8}, 3469 {0x1f60,0x1f67,1,8}, 3470 {0x1f70,0x1f71,1,74}, 3471 {0x1f72,0x1f75,1,86}, 3472 {0x1f76,0x1f77,1,100}, 3473 {0x1f78,0x1f79,1,128}, 3474 {0x1f7a,0x1f7b,1,112}, 3475 {0x1f7c,0x1f7d,1,126}, 3476 {0x1f80,0x1f87,1,8}, 3477 {0x1f90,0x1f97,1,8}, 3478 {0x1fa0,0x1fa7,1,8}, 3479 {0x1fb0,0x1fb1,1,8}, 3480 {0x1fb3,0x1fb3,-1,9}, 3481 {0x1fbe,0x1fbe,-1,-7205}, 3482 {0x1fc3,0x1fc3,-1,9}, 3483 {0x1fd0,0x1fd1,1,8}, 3484 {0x1fe0,0x1fe1,1,8}, 3485 {0x1fe5,0x1fe5,-1,7}, 3486 {0x1ff3,0x1ff3,-1,9}, 3487 {0x214e,0x214e,-1,-28}, 3488 {0x2170,0x217f,1,-16}, 3489 {0x2184,0x2184,-1,-1}, 3490 {0x24d0,0x24e9,1,-26}, 3491 {0x2c30,0x2c5e,1,-48}, 3492 {0x2c61,0x2c61,-1,-1}, 3493 {0x2c65,0x2c65,-1,-10795}, 3494 {0x2c66,0x2c66,-1,-10792}, 3495 {0x2c68,0x2c6c,2,-1}, 3496 {0x2c73,0x2c76,3,-1}, 3497 {0x2c81,0x2ce3,2,-1}, 3498 {0x2cec,0x2cee,2,-1}, 3499 {0x2cf3,0x2cf3,-1,-1}, 3500 {0x2d00,0x2d25,1,-7264}, 3501 {0x2d27,0x2d2d,6,-7264}, 3502 {0xa641,0xa66d,2,-1}, 3503 {0xa681,0xa69b,2,-1}, 3504 {0xa723,0xa72f,2,-1}, 3505 {0xa733,0xa76f,2,-1}, 3506 {0xa77a,0xa77c,2,-1}, 3507 {0xa77f,0xa787,2,-1}, 3508 {0xa78c,0xa791,5,-1}, 3509 {0xa793,0xa793,-1,-1}, 3510 {0xa794,0xa794,-1,48}, 3511 {0xa797,0xa7a9,2,-1}, 3512 {0xa7b5,0xa7bf,2,-1}, 3513 {0xa7c3,0xa7c3,-1,-1}, 3514 {0xab53,0xab53,-1,-928}, 3515 {0xab70,0xabbf,1,-38864}, 3516 {0xff41,0xff5a,1,-32}, 3517 {0x10428,0x1044f,1,-40}, 3518 {0x104d8,0x104fb,1,-40}, 3519 {0x10cc0,0x10cf2,1,-64}, 3520 {0x118c0,0x118df,1,-32}, 3521 {0x16e60,0x16e7f,1,-32}, 3522 {0x1e922,0x1e943,1,-34} 3523 }; 3524 3525 /* 3526 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3527 * simple case folding. 3528 */ 3529 int 3530 utf_toupper(int a) 3531 { 3532 // If 'casemap' contains "keepascii" use ASCII style toupper(). 3533 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3534 return TOUPPER_ASC(a); 3535 3536 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3537 // If towupper() is available and handles Unicode, use it. 3538 if (!(cmp_flags & CMP_INTERNAL)) 3539 return towupper(a); 3540 #endif 3541 3542 // For characters below 128 use locale sensitive toupper(). 3543 if (a < 128) 3544 return TOUPPER_LOC(a); 3545 3546 // For any other characters use the above mapping table. 3547 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3548 } 3549 3550 int 3551 utf_islower(int a) 3552 { 3553 // German sharp s is lower case but has no upper case equivalent. 3554 return (utf_toupper(a) != a) || a == 0xdf; 3555 } 3556 3557 /* 3558 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3559 * simple case folding. 3560 */ 3561 int 3562 utf_tolower(int a) 3563 { 3564 // If 'casemap' contains "keepascii" use ASCII style tolower(). 3565 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3566 return TOLOWER_ASC(a); 3567 3568 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3569 // If towlower() is available and handles Unicode, use it. 3570 if (!(cmp_flags & CMP_INTERNAL)) 3571 return towlower(a); 3572 #endif 3573 3574 // For characters below 128 use locale sensitive tolower(). 3575 if (a < 128) 3576 return TOLOWER_LOC(a); 3577 3578 // For any other characters use the above mapping table. 3579 return utf_convert(a, toLower, (int)sizeof(toLower)); 3580 } 3581 3582 int 3583 utf_isupper(int a) 3584 { 3585 return (utf_tolower(a) != a); 3586 } 3587 3588 static int 3589 utf_strnicmp( 3590 char_u *s1, 3591 char_u *s2, 3592 size_t n1, 3593 size_t n2) 3594 { 3595 int c1, c2, cdiff; 3596 char_u buffer[6]; 3597 3598 for (;;) 3599 { 3600 c1 = utf_safe_read_char_adv(&s1, &n1); 3601 c2 = utf_safe_read_char_adv(&s2, &n2); 3602 3603 if (c1 <= 0 || c2 <= 0) 3604 break; 3605 3606 if (c1 == c2) 3607 continue; 3608 3609 cdiff = utf_fold(c1) - utf_fold(c2); 3610 if (cdiff != 0) 3611 return cdiff; 3612 } 3613 3614 // some string ended or has an incomplete/illegal character sequence 3615 3616 if (c1 == 0 || c2 == 0) 3617 { 3618 // some string ended. shorter string is smaller 3619 if (c1 == 0 && c2 == 0) 3620 return 0; 3621 return c1 == 0 ? -1 : 1; 3622 } 3623 3624 // Continue with bytewise comparison to produce some result that 3625 // would make comparison operations involving this function transitive. 3626 // 3627 // If only one string had an error, comparison should be made with 3628 // folded version of the other string. In this case it is enough 3629 // to fold just one character to determine the result of comparison. 3630 3631 if (c1 != -1 && c2 == -1) 3632 { 3633 n1 = utf_char2bytes(utf_fold(c1), buffer); 3634 s1 = buffer; 3635 } 3636 else if (c2 != -1 && c1 == -1) 3637 { 3638 n2 = utf_char2bytes(utf_fold(c2), buffer); 3639 s2 = buffer; 3640 } 3641 3642 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3643 { 3644 cdiff = (int)(*s1) - (int)(*s2); 3645 if (cdiff != 0) 3646 return cdiff; 3647 3648 s1++; 3649 s2++; 3650 n1--; 3651 n2--; 3652 } 3653 3654 if (n1 > 0 && *s1 == NUL) 3655 n1 = 0; 3656 if (n2 > 0 && *s2 == NUL) 3657 n2 = 0; 3658 3659 if (n1 == 0 && n2 == 0) 3660 return 0; 3661 return n1 == 0 ? -1 : 1; 3662 } 3663 3664 /* 3665 * Version of strnicmp() that handles multi-byte characters. 3666 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3667 * probably use strnicmp(), because there are no ASCII characters in the 3668 * second byte. 3669 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3670 * two characters otherwise. 3671 */ 3672 int 3673 mb_strnicmp(char_u *s1, char_u *s2, size_t nn) 3674 { 3675 int i, l; 3676 int cdiff; 3677 int n = (int)nn; 3678 3679 if (enc_utf8) 3680 { 3681 return utf_strnicmp(s1, s2, nn, nn); 3682 } 3683 else 3684 { 3685 for (i = 0; i < n; i += l) 3686 { 3687 if (s1[i] == NUL && s2[i] == NUL) // both strings end 3688 return 0; 3689 3690 l = (*mb_ptr2len)(s1 + i); 3691 if (l <= 1) 3692 { 3693 // Single byte: first check normally, then with ignore case. 3694 if (s1[i] != s2[i]) 3695 { 3696 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3697 if (cdiff != 0) 3698 return cdiff; 3699 } 3700 } 3701 else 3702 { 3703 // For non-Unicode multi-byte don't ignore case. 3704 if (l > n - i) 3705 l = n - i; 3706 cdiff = STRNCMP(s1 + i, s2 + i, l); 3707 if (cdiff != 0) 3708 return cdiff; 3709 } 3710 } 3711 } 3712 return 0; 3713 } 3714 3715 /* 3716 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3717 * 'encoding' has been set to. 3718 */ 3719 void 3720 show_utf8(void) 3721 { 3722 int len; 3723 int rlen = 0; 3724 char_u *line; 3725 int clen; 3726 int i; 3727 3728 // Get the byte length of the char under the cursor, including composing 3729 // characters. 3730 line = ml_get_cursor(); 3731 len = utfc_ptr2len(line); 3732 if (len == 0) 3733 { 3734 msg("NUL"); 3735 return; 3736 } 3737 3738 clen = 0; 3739 for (i = 0; i < len; ++i) 3740 { 3741 if (clen == 0) 3742 { 3743 // start of (composing) character, get its length 3744 if (i > 0) 3745 { 3746 STRCPY(IObuff + rlen, "+ "); 3747 rlen += 2; 3748 } 3749 clen = utf_ptr2len(line + i); 3750 } 3751 sprintf((char *)IObuff + rlen, "%02x ", 3752 (line[i] == NL) ? NUL : line[i]); // NUL is stored as NL 3753 --clen; 3754 rlen += (int)STRLEN(IObuff + rlen); 3755 if (rlen > IOSIZE - 20) 3756 break; 3757 } 3758 3759 msg((char *)IObuff); 3760 } 3761 3762 /* 3763 * mb_head_off() function pointer. 3764 * Return offset from "p" to the first byte of the character it points into. 3765 * If "p" points to the NUL at the end of the string return 0. 3766 * Returns 0 when already at the first byte of a character. 3767 */ 3768 int 3769 latin_head_off(char_u *base UNUSED, char_u *p UNUSED) 3770 { 3771 return 0; 3772 } 3773 3774 static int 3775 dbcs_head_off(char_u *base, char_u *p) 3776 { 3777 char_u *q; 3778 3779 // It can't be a trailing byte when not using DBCS, at the start of the 3780 // string or the previous byte can't start a double-byte. 3781 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3782 return 0; 3783 3784 // This is slow: need to start at the base and go forward until the 3785 // byte we are looking for. Return 1 when we went past it, 0 otherwise. 3786 q = base; 3787 while (q < p) 3788 q += dbcs_ptr2len(q); 3789 return (q == p) ? 0 : 1; 3790 } 3791 3792 /* 3793 * Special version of dbcs_head_off() that works for ScreenLines[], where 3794 * single-width DBCS_JPNU characters are stored separately. 3795 */ 3796 int 3797 dbcs_screen_head_off(char_u *base, char_u *p) 3798 { 3799 char_u *q; 3800 3801 // It can't be a trailing byte when not using DBCS, at the start of the 3802 // string or the previous byte can't start a double-byte. 3803 // For euc-jp an 0x8e byte in the previous cell always means we have a 3804 // lead byte in the current cell. 3805 if (p <= base 3806 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3807 || MB_BYTE2LEN(p[-1]) == 1 3808 || *p == NUL) 3809 return 0; 3810 3811 // This is slow: need to start at the base and go forward until the 3812 // byte we are looking for. Return 1 when we went past it, 0 otherwise. 3813 // For DBCS_JPNU look out for 0x8e, which means the second byte is not 3814 // stored as the next byte. 3815 q = base; 3816 while (q < p) 3817 { 3818 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3819 ++q; 3820 else 3821 q += dbcs_ptr2len(q); 3822 } 3823 return (q == p) ? 0 : 1; 3824 } 3825 3826 int 3827 utf_head_off(char_u *base, char_u *p) 3828 { 3829 char_u *q; 3830 char_u *s; 3831 int c; 3832 int len; 3833 #ifdef FEAT_ARABIC 3834 char_u *j; 3835 #endif 3836 3837 if (*p < 0x80) // be quick for ASCII 3838 return 0; 3839 3840 // Skip backwards over trailing bytes: 10xx.xxxx 3841 // Skip backwards again if on a composing char. 3842 for (q = p; ; --q) 3843 { 3844 // Move s to the last byte of this char. 3845 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3846 ; 3847 // Move q to the first byte of this char. 3848 while (q > base && (*q & 0xc0) == 0x80) 3849 --q; 3850 // Check for illegal sequence. Do allow an illegal byte after where we 3851 // started. 3852 len = utf8len_tab[*q]; 3853 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3854 return 0; 3855 3856 if (q <= base) 3857 break; 3858 3859 c = utf_ptr2char(q); 3860 if (utf_iscomposing(c)) 3861 continue; 3862 3863 #ifdef FEAT_ARABIC 3864 if (arabic_maycombine(c)) 3865 { 3866 // Advance to get a sneak-peak at the next char 3867 j = q; 3868 --j; 3869 // Move j to the first byte of this char. 3870 while (j > base && (*j & 0xc0) == 0x80) 3871 --j; 3872 if (arabic_combine(utf_ptr2char(j), c)) 3873 continue; 3874 } 3875 #endif 3876 break; 3877 } 3878 3879 return (int)(p - q); 3880 } 3881 3882 /* 3883 * Whether space is NOT allowed before/after 'c'. 3884 */ 3885 int 3886 utf_eat_space(int cc) 3887 { 3888 return ((cc >= 0x2000 && cc <= 0x206F) // General punctuations 3889 || (cc >= 0x2e00 && cc <= 0x2e7f) // Supplemental punctuations 3890 || (cc >= 0x3000 && cc <= 0x303f) // CJK symbols and punctuations 3891 || (cc >= 0xff01 && cc <= 0xff0f) // Full width ASCII punctuations 3892 || (cc >= 0xff1a && cc <= 0xff20) // .. 3893 || (cc >= 0xff3b && cc <= 0xff40) // .. 3894 || (cc >= 0xff5b && cc <= 0xff65)); // .. 3895 } 3896 3897 /* 3898 * Whether line break is allowed before "cc". 3899 */ 3900 int 3901 utf_allow_break_before(int cc) 3902 { 3903 static const int BOL_prohibition_punct[] = 3904 { 3905 '!', 3906 '%', 3907 ')', 3908 ',', 3909 ':', 3910 ';', 3911 '>', 3912 '?', 3913 ']', 3914 '}', 3915 0x2019, // ’ right single quotation mark 3916 0x201d, // ” right double quotation mark 3917 0x2020, // † dagger 3918 0x2021, // ‡ double dagger 3919 0x2026, // … horizontal ellipsis 3920 0x2030, // ‰ per mille sign 3921 0x2031, // ‱ per then thousand sign 3922 0x203c, // ‼ double exclamation mark 3923 0x2047, // ⁇ double question mark 3924 0x2048, // ⁈ question exclamation mark 3925 0x2049, // ⁉ exclamation question mark 3926 0x2103, // ℃ degree celsius 3927 0x2109, // ℉ degree fahrenheit 3928 0x3001, // 、 ideographic comma 3929 0x3002, // 。 ideographic full stop 3930 0x3009, // 〉 right angle bracket 3931 0x300b, // 》 right double angle bracket 3932 0x300d, // 」 right corner bracket 3933 0x300f, // 』 right white corner bracket 3934 0x3011, // 】 right black lenticular bracket 3935 0x3015, // 〕 right tortoise shell bracket 3936 0x3017, // 〗 right white lenticular bracket 3937 0x3019, // 〙 right white tortoise shell bracket 3938 0x301b, // 〛 right white square bracket 3939 0xff01, // ！ fullwidth exclamation mark 3940 0xff09, // ） fullwidth right parenthesis 3941 0xff0c, // ， fullwidth comma 3942 0xff0e, // ． fullwidth full stop 3943 0xff1a, // ： fullwidth colon 3944 0xff1b, // ； fullwidth semicolon 3945 0xff1f, // ？ fullwidth question mark 3946 0xff3d, // ］ fullwidth right square bracket 3947 0xff5d, // ｝ fullwidth right curly bracket 3948 }; 3949 3950 int first = 0; 3951 int last = sizeof(BOL_prohibition_punct)/sizeof(int) - 1; 3952 int mid = 0; 3953 3954 while (first < last) 3955 { 3956 mid = (first + last)/2; 3957 3958 if (cc == BOL_prohibition_punct[mid]) 3959 return FALSE; 3960 else if (cc > BOL_prohibition_punct[mid]) 3961 first = mid + 1; 3962 else 3963 last = mid - 1; 3964 } 3965 3966 return cc != BOL_prohibition_punct[first]; 3967 } 3968 3969 /* 3970 * Whether line break is allowed after "cc". 3971 */ 3972 static int 3973 utf_allow_break_after(int cc) 3974 { 3975 static const int EOL_prohibition_punct[] = 3976 { 3977 '(', 3978 '<', 3979 '[', 3980 '`', 3981 '{', 3982 //0x2014, // — em dash 3983 0x2018, // ‘ left single quotation mark 3984 0x201c, // “ left double quotation mark 3985 //0x2053, // ～ swung dash 3986 0x3008, // 〈 left angle bracket 3987 0x300a, // 《 left double angle bracket 3988 0x300c, // 「 left corner bracket 3989 0x300e, // 『 left white corner bracket 3990 0x3010, // 【 left black lenticular bracket 3991 0x3014, // 〔 left tortoise shell bracket 3992 0x3016, // 〖 left white lenticular bracket 3993 0x3018, // 〘 left white tortoise shell bracket 3994 0x301a, // 〚 left white square bracket 3995 0xff08, // （ fullwidth left parenthesis 3996 0xff3b, // ［ fullwidth left square bracket 3997 0xff5b, // ｛ fullwidth left curly bracket 3998 }; 3999 4000 int first = 0; 4001 int last = sizeof(EOL_prohibition_punct)/sizeof(int) - 1; 4002 int mid = 0; 4003 4004 while (first < last) 4005 { 4006 mid = (first + last)/2; 4007 4008 if (cc == EOL_prohibition_punct[mid]) 4009 return FALSE; 4010 else if (cc > EOL_prohibition_punct[mid]) 4011 first = mid + 1; 4012 else 4013 last = mid - 1; 4014 } 4015 4016 return cc != EOL_prohibition_punct[first]; 4017 } 4018 4019 /* 4020 * Whether line break is allowed between "cc" and "ncc". 4021 */ 4022 int 4023 utf_allow_break(int cc, int ncc) 4024 { 4025 // don't break between two-letter punctuations 4026 if (cc == ncc 4027 && (cc == 0x2014 // em dash 4028 || cc == 0x2026)) // horizontal ellipsis 4029 return FALSE; 4030 4031 return utf_allow_break_after(cc) && utf_allow_break_before(ncc); 4032 } 4033 4034 /* 4035 * Copy a character from "*fp" to "*tp" and advance the pointers. 4036 */ 4037 void 4038 mb_copy_char(char_u **fp, char_u **tp) 4039 { 4040 int l = (*mb_ptr2len)(*fp); 4041 4042 mch_memmove(*tp, *fp, (size_t)l); 4043 *tp += l; 4044 *fp += l; 4045 } 4046 4047 /* 4048 * Return the offset from "p" to the first byte of a character. When "p" is 4049 * at the start of a character 0 is returned, otherwise the offset to the next 4050 * character. Can start anywhere in a stream of bytes. 4051 */ 4052 int 4053 mb_off_next(char_u *base, char_u *p) 4054 { 4055 int i; 4056 int j; 4057 4058 if (enc_utf8) 4059 { 4060 if (*p < 0x80) // be quick for ASCII 4061 return 0; 4062 4063 // Find the next character that isn't 10xx.xxxx 4064 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 4065 ; 4066 if (i > 0) 4067 { 4068 // Check for illegal sequence. 4069 for (j = 0; p - j > base; ++j) 4070 if ((p[-j] & 0xc0) != 0x80) 4071 break; 4072 if (utf8len_tab[p[-j]] != i + j) 4073 return 0; 4074 } 4075 return i; 4076 } 4077 4078 // Only need to check if we're on a trail byte, it doesn't matter if we 4079 // want the offset to the next or current character. 4080 return (*mb_head_off)(base, p); 4081 } 4082 4083 /* 4084 * Return the offset from "p" to the last byte of the character it points 4085 * into. Can start anywhere in a stream of bytes. 4086 */ 4087 int 4088 mb_tail_off(char_u *base, char_u *p) 4089 { 4090 int i; 4091 int j; 4092 4093 if (*p == NUL) 4094 return 0; 4095 4096 if (enc_utf8) 4097 { 4098 // Find the last character that is 10xx.xxxx 4099 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 4100 ; 4101 // Check for illegal sequence. 4102 for (j = 0; p - j > base; ++j) 4103 if ((p[-j] & 0xc0) != 0x80) 4104 break; 4105 if (utf8len_tab[p[-j]] != i + j + 1) 4106 return 0; 4107 return i; 4108 } 4109 4110 // It can't be the first byte if a double-byte when not using DBCS, at the 4111 // end of the string or the byte can't start a double-byte. 4112 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 4113 return 0; 4114 4115 // Return 1 when on the lead byte, 0 when on the tail byte. 4116 return 1 - dbcs_head_off(base, p); 4117 } 4118 4119 /* 4120 * Find the next illegal byte sequence. 4121 */ 4122 void 4123 utf_find_illegal(void) 4124 { 4125 pos_T pos = curwin->w_cursor; 4126 char_u *p; 4127 int len; 4128 vimconv_T vimconv; 4129 char_u *tofree = NULL; 4130 4131 vimconv.vc_type = CONV_NONE; 4132 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 4133 { 4134 // 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 4135 // possibly a utf-8 file with illegal bytes. Setup for conversion 4136 // from utf-8 to 'fileencoding'. 4137 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 4138 } 4139 4140 curwin->w_cursor.coladd = 0; 4141 for (;;) 4142 { 4143 p = ml_get_cursor(); 4144 if (vimconv.vc_type != CONV_NONE) 4145 { 4146 vim_free(tofree); 4147 tofree = string_convert(&vimconv, p, NULL); 4148 if (tofree == NULL) 4149 break; 4150 p = tofree; 4151 } 4152 4153 while (*p != NUL) 4154 { 4155 // Illegal means that there are not enough trail bytes (checked by 4156 // utf_ptr2len()) or too many of them (overlong sequence). 4157 len = utf_ptr2len(p); 4158 if (*p >= 0x80 && (len == 1 4159 || utf_char2len(utf_ptr2char(p)) != len)) 4160 { 4161 if (vimconv.vc_type == CONV_NONE) 4162 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 4163 else 4164 { 4165 int l; 4166 4167 len = (int)(p - tofree); 4168 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 4169 { 4170 l = utf_ptr2len(p); 4171 curwin->w_cursor.col += l; 4172 } 4173 } 4174 goto theend; 4175 } 4176 p += len; 4177 } 4178 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 4179 break; 4180 ++curwin->w_cursor.lnum; 4181 curwin->w_cursor.col = 0; 4182 } 4183 4184 // didn't find it: don't move and beep 4185 curwin->w_cursor = pos; 4186 beep_flush(); 4187 4188 theend: 4189 vim_free(tofree); 4190 convert_setup(&vimconv, NULL, NULL); 4191 } 4192 4193 #if defined(FEAT_GUI_GTK) || defined(PROTO) 4194 /* 4195 * Return TRUE if string "s" is a valid utf-8 string. 4196 * When "end" is NULL stop at the first NUL. 4197 * When "end" is positive stop there. 4198 */ 4199 int 4200 utf_valid_string(char_u *s, char_u *end) 4201 { 4202 int l; 4203 char_u *p = s; 4204 4205 while (end == NULL ? *p != NUL : p < end) 4206 { 4207 l = utf8len_tab_zero[*p]; 4208 if (l == 0) 4209 return FALSE; // invalid lead byte 4210 if (end != NULL && p + l > end) 4211 return FALSE; // incomplete byte sequence 4212 ++p; 4213 while (--l > 0) 4214 if ((*p++ & 0xc0) != 0x80) 4215 return FALSE; // invalid trail byte 4216 } 4217 return TRUE; 4218 } 4219 #endif 4220 4221 #if defined(FEAT_GUI) || defined(PROTO) 4222 /* 4223 * Special version of mb_tail_off() for use in ScreenLines[]. 4224 */ 4225 int 4226 dbcs_screen_tail_off(char_u *base, char_u *p) 4227 { 4228 // It can't be the first byte if a double-byte when not using DBCS, at the 4229 // end of the string or the byte can't start a double-byte. 4230 // For euc-jp an 0x8e byte always means we have a lead byte in the current 4231 // cell. 4232 if (*p == NUL || p[1] == NUL 4233 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 4234 || MB_BYTE2LEN(*p) == 1) 4235 return 0; 4236 4237 // Return 1 when on the lead byte, 0 when on the tail byte. 4238 return 1 - dbcs_screen_head_off(base, p); 4239 } 4240 #endif 4241 4242 /* 4243 * If the cursor moves on an trail byte, set the cursor on the lead byte. 4244 * Thus it moves left if necessary. 4245 * Return TRUE when the cursor was adjusted. 4246 */ 4247 void 4248 mb_adjust_cursor(void) 4249 { 4250 mb_adjustpos(curbuf, &curwin->w_cursor); 4251 } 4252 4253 /* 4254 * Adjust position "*lp" to point to the first byte of a multi-byte character. 4255 * If it points to a tail byte it's moved backwards to the head byte. 4256 */ 4257 void 4258 mb_adjustpos(buf_T *buf, pos_T *lp) 4259 { 4260 char_u *p; 4261 4262 if (lp->col > 0 || lp->coladd > 1) 4263 { 4264 p = ml_get_buf(buf, lp->lnum, FALSE); 4265 if (*p == NUL || (int)STRLEN(p) < lp->col) 4266 lp->col = 0; 4267 else 4268 lp->col -= (*mb_head_off)(p, p + lp->col); 4269 // Reset "coladd" when the cursor would be on the right half of a 4270 // double-wide character. 4271 if (lp->coladd == 1 4272 && p[lp->col] != TAB 4273 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 4274 && ptr2cells(p + lp->col) > 1) 4275 lp->coladd = 0; 4276 } 4277 } 4278 4279 /* 4280 * Return a pointer to the character before "*p", if there is one. 4281 */ 4282 char_u * 4283 mb_prevptr( 4284 char_u *line, // start of the string 4285 char_u *p) 4286 { 4287 if (p > line) 4288 MB_PTR_BACK(line, p); 4289 return p; 4290 } 4291 4292 /* 4293 * Return the character length of "str". Each multi-byte character (with 4294 * following composing characters) counts as one. 4295 */ 4296 int 4297 mb_charlen(char_u *str) 4298 { 4299 char_u *p = str; 4300 int count; 4301 4302 if (p == NULL) 4303 return 0; 4304 4305 for (count = 0; *p != NUL; count++) 4306 p += (*mb_ptr2len)(p); 4307 4308 return count; 4309 } 4310 4311 #if (defined(FEAT_SPELL) || defined(FEAT_EVAL)) || defined(PROTO) 4312 /* 4313 * Like mb_charlen() but for a string with specified length. 4314 */ 4315 int 4316 mb_charlen_len(char_u *str, int len) 4317 { 4318 char_u *p = str; 4319 int count; 4320 4321 for (count = 0; *p != NUL && p < str + len; count++) 4322 p += (*mb_ptr2len)(p); 4323 4324 return count; 4325 } 4326 #endif 4327 4328 /* 4329 * Try to un-escape a multi-byte character. 4330 * Used for the "to" and "from" part of a mapping. 4331 * Return the un-escaped string if it is a multi-byte character, and advance 4332 * "pp" to just after the bytes that formed it. 4333 * Return NULL if no multi-byte char was found. 4334 */ 4335 char_u * 4336 mb_unescape(char_u **pp) 4337 { 4338 static char_u buf[6]; 4339 int n; 4340 int m = 0; 4341 char_u *str = *pp; 4342 4343 // Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 4344 // KS_EXTRA KE_CSI to CSI. 4345 // Maximum length of a utf-8 character is 4 bytes. 4346 for (n = 0; str[n] != NUL && m < 4; ++n) 4347 { 4348 if (str[n] == K_SPECIAL 4349 && str[n + 1] == KS_SPECIAL 4350 && str[n + 2] == KE_FILLER) 4351 { 4352 buf[m++] = K_SPECIAL; 4353 n += 2; 4354 } 4355 else if ((str[n] == K_SPECIAL 4356 # ifdef FEAT_GUI 4357 || str[n] == CSI 4358 # endif 4359 ) 4360 && str[n + 1] == KS_EXTRA 4361 && str[n + 2] == (int)KE_CSI) 4362 { 4363 buf[m++] = CSI; 4364 n += 2; 4365 } 4366 else if (str[n] == K_SPECIAL 4367 # ifdef FEAT_GUI 4368 || str[n] == CSI 4369 # endif 4370 ) 4371 break; // a special key can't be a multibyte char 4372 else 4373 buf[m++] = str[n]; 4374 buf[m] = NUL; 4375 4376 // Return a multi-byte character if it's found. An illegal sequence 4377 // will result in a 1 here. 4378 if ((*mb_ptr2len)(buf) > 1) 4379 { 4380 *pp = str + n + 1; 4381 return buf; 4382 } 4383 4384 // Bail out quickly for ASCII. 4385 if (buf[0] < 128) 4386 break; 4387 } 4388 return NULL; 4389 } 4390 4391 /* 4392 * Return TRUE if the character at "row"/"col" on the screen is the left side 4393 * of a double-width character. 4394 * Caller must make sure "row" and "col" are not invalid! 4395 */ 4396 int 4397 mb_lefthalve(int row, int col) 4398 { 4399 return (*mb_off2cells)(LineOffset[row] + col, 4400 LineOffset[row] + screen_Columns) > 1; 4401 } 4402 4403 /* 4404 * Correct a position on the screen, if it's the right half of a double-wide 4405 * char move it to the left half. Returns the corrected column. 4406 */ 4407 int 4408 mb_fix_col(int col, int row) 4409 { 4410 int off; 4411 4412 col = check_col(col); 4413 row = check_row(row); 4414 off = LineOffset[row] + col; 4415 if (has_mbyte && ScreenLines != NULL && col > 0 4416 && ((enc_dbcs 4417 && ScreenLines[off] != NUL 4418 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 4419 ScreenLines + off)) 4420 || (enc_utf8 && ScreenLines[off] == 0 4421 && ScreenLinesUC[off] == 0))) 4422 return col - 1; 4423 return col; 4424 } 4425 4426 static int enc_alias_search(char_u *name); 4427 4428 /* 4429 * Skip the Vim specific head of a 'encoding' name. 4430 */ 4431 char_u * 4432 enc_skip(char_u *p) 4433 { 4434 if (STRNCMP(p, "2byte-", 6) == 0) 4435 return p + 6; 4436 if (STRNCMP(p, "8bit-", 5) == 0) 4437 return p + 5; 4438 return p; 4439 } 4440 4441 /* 4442 * Find the canonical name for encoding "enc". 4443 * When the name isn't recognized, returns "enc" itself, but with all lower 4444 * case characters and '_' replaced with '-'. 4445 * Returns an allocated string. NULL for out-of-memory. 4446 */ 4447 char_u * 4448 enc_canonize(char_u *enc) 4449 { 4450 char_u *r; 4451 char_u *p, *s; 4452 int i; 4453 4454 if (STRCMP(enc, "default") == 0) 4455 { 4456 // Use the default encoding as it's found by set_init_1(). 4457 r = get_encoding_default(); 4458 if (r == NULL) 4459 r = (char_u *)"latin1"; 4460 return vim_strsave(r); 4461 } 4462 4463 // copy "enc" to allocated memory, with room for two '-' 4464 r = alloc(STRLEN(enc) + 3); 4465 if (r != NULL) 4466 { 4467 // Make it all lower case and replace '_' with '-'. 4468 p = r; 4469 for (s = enc; *s != NUL; ++s) 4470 { 4471 if (*s == '_') 4472 *p++ = '-'; 4473 else 4474 *p++ = TOLOWER_ASC(*s); 4475 } 4476 *p = NUL; 4477 4478 // Skip "2byte-" and "8bit-". 4479 p = enc_skip(r); 4480 4481 // Change "microsoft-cp" to "cp". Used in some spell files. 4482 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4483 STRMOVE(p, p + 10); 4484 4485 // "iso8859" -> "iso-8859" 4486 if (STRNCMP(p, "iso8859", 7) == 0) 4487 { 4488 STRMOVE(p + 4, p + 3); 4489 p[3] = '-'; 4490 } 4491 4492 // "iso-8859n" -> "iso-8859-n" 4493 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4494 { 4495 STRMOVE(p + 9, p + 8); 4496 p[8] = '-'; 4497 } 4498 4499 // "latin-N" -> "latinN" 4500 if (STRNCMP(p, "latin-", 6) == 0) 4501 STRMOVE(p + 5, p + 6); 4502 4503 if (enc_canon_search(p) >= 0) 4504 { 4505 // canonical name can be used unmodified 4506 if (p != r) 4507 STRMOVE(r, p); 4508 } 4509 else if ((i = enc_alias_search(p)) >= 0) 4510 { 4511 // alias recognized, get canonical name 4512 vim_free(r); 4513 r = vim_strsave((char_u *)enc_canon_table[i].name); 4514 } 4515 } 4516 return r; 4517 } 4518 4519 /* 4520 * Search for an encoding alias of "name". 4521 * Returns -1 when not found. 4522 */ 4523 static int 4524 enc_alias_search(char_u *name) 4525 { 4526 int i; 4527 4528 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4529 if (STRCMP(name, enc_alias_table[i].name) == 0) 4530 return enc_alias_table[i].canon; 4531 return -1; 4532 } 4533 4534 4535 #ifdef HAVE_LANGINFO_H 4536 # include <langinfo.h> 4537 #endif 4538 4539 #if !defined(FEAT_GUI_MSWIN) || defined(VIMDLL) 4540 /* 4541 * Get the canonicalized encoding from the specified locale string "locale" 4542 * or from the environment variables LC_ALL, LC_CTYPE and LANG. 4543 * Returns an allocated string when successful, NULL when not. 4544 */ 4545 char_u * 4546 enc_locale_env(char *locale) 4547 { 4548 char *s = locale; 4549 char *p; 4550 int i; 4551 char buf[50]; 4552 4553 if (s == NULL || *s == NUL) 4554 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4555 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4556 s = getenv("LANG"); 4557 4558 if (s == NULL || *s == NUL) 4559 return NULL; 4560 4561 // The most generic locale format is: 4562 // language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4563 // If there is a '.' remove the part before it. 4564 // if there is something after the codeset, remove it. 4565 // Make the name lowercase and replace '_' with '-'. 4566 // Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4567 // "ko_KR.EUC" == "euc-kr" 4568 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4569 { 4570 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4571 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4572 { 4573 // copy "XY.EUC" to "euc-XY" to buf[10] 4574 STRCPY(buf + 10, "euc-"); 4575 buf[14] = p[-2]; 4576 buf[15] = p[-1]; 4577 buf[16] = 0; 4578 s = buf + 10; 4579 } 4580 else 4581 s = p + 1; 4582 } 4583 for (i = 0; i < (int)sizeof(buf) - 1 && s[i] != NUL; ++i) 4584 { 4585 if (s[i] == '_' || s[i] == '-') 4586 buf[i] = '-'; 4587 else if (isalnum((int)s[i])) 4588 buf[i] = TOLOWER_ASC(s[i]); 4589 else 4590 break; 4591 } 4592 buf[i] = NUL; 4593 4594 return enc_canonize((char_u *)buf); 4595 } 4596 #endif 4597 4598 /* 4599 * Get the canonicalized encoding of the current locale. 4600 * Returns an allocated string when successful, NULL when not. 4601 */ 4602 char_u * 4603 enc_locale(void) 4604 { 4605 #ifdef MSWIN 4606 char buf[50]; 4607 long acp = GetACP(); 4608 4609 if (acp == 1200) 4610 STRCPY(buf, "ucs-2le"); 4611 else if (acp == 1252) // cp1252 is used as latin1 4612 STRCPY(buf, "latin1"); 4613 else if (acp == 65001) 4614 STRCPY(buf, "utf-8"); 4615 else 4616 sprintf(buf, "cp%ld", acp); 4617 4618 return enc_canonize((char_u *)buf); 4619 #else 4620 char *s; 4621 4622 # ifdef HAVE_NL_LANGINFO_CODESET 4623 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4624 # endif 4625 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4626 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4627 # endif 4628 s = NULL; 4629 4630 return enc_locale_env(s); 4631 #endif 4632 } 4633 4634 # if defined(MSWIN) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4635 /* 4636 * Convert an encoding name to an MS-Windows codepage. 4637 * Returns zero if no codepage can be figured out. 4638 */ 4639 int 4640 encname2codepage(char_u *name) 4641 { 4642 int cp; 4643 char_u *p = name; 4644 int idx; 4645 4646 if (STRNCMP(p, "8bit-", 5) == 0) 4647 p += 5; 4648 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4649 p += 6; 4650 4651 if (p[0] == 'c' && p[1] == 'p') 4652 cp = atoi((char *)p + 2); 4653 else if ((idx = enc_canon_search(p)) >= 0) 4654 cp = enc_canon_table[idx].codepage; 4655 else 4656 return 0; 4657 if (IsValidCodePage(cp)) 4658 return cp; 4659 return 0; 4660 } 4661 # endif 4662 4663 # if defined(USE_ICONV) || defined(PROTO) 4664 4665 /* 4666 * Call iconv_open() with a check if iconv() works properly (there are broken 4667 * versions). 4668 * Returns (void *)-1 if failed. 4669 * (should return iconv_t, but that causes problems with prototypes). 4670 */ 4671 void * 4672 my_iconv_open(char_u *to, char_u *from) 4673 { 4674 iconv_t fd; 4675 #define ICONV_TESTLEN 400 4676 char_u tobuf[ICONV_TESTLEN]; 4677 char *p; 4678 size_t tolen; 4679 static int iconv_ok = -1; 4680 4681 if (iconv_ok == FALSE) 4682 return (void *)-1; // detected a broken iconv() previously 4683 4684 #ifdef DYNAMIC_ICONV 4685 // Check if the iconv.dll can be found. 4686 if (!iconv_enabled(TRUE)) 4687 return (void *)-1; 4688 #endif 4689 4690 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4691 4692 if (fd != (iconv_t)-1 && iconv_ok == -1) 4693 { 4694 /* 4695 * Do a dummy iconv() call to check if it actually works. There is a 4696 * version of iconv() on Linux that is broken. We can't ignore it, 4697 * because it's wide-spread. The symptoms are that after outputting 4698 * the initial shift state the "to" pointer is NULL and conversion 4699 * stops for no apparent reason after about 8160 characters. 4700 */ 4701 p = (char *)tobuf; 4702 tolen = ICONV_TESTLEN; 4703 (void)iconv(fd, NULL, NULL, &p, &tolen); 4704 if (p == NULL) 4705 { 4706 iconv_ok = FALSE; 4707 iconv_close(fd); 4708 fd = (iconv_t)-1; 4709 } 4710 else 4711 iconv_ok = TRUE; 4712 } 4713 4714 return (void *)fd; 4715 } 4716 4717 /* 4718 * Convert the string "str[slen]" with iconv(). 4719 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4720 * sequence and set "*unconvlenp" to the length of it. 4721 * Returns the converted string in allocated memory. NULL for an error. 4722 * If resultlenp is not NULL, sets it to the result length in bytes. 4723 */ 4724 static char_u * 4725 iconv_string( 4726 vimconv_T *vcp, 4727 char_u *str, 4728 int slen, 4729 int *unconvlenp, 4730 int *resultlenp) 4731 { 4732 const char *from; 4733 size_t fromlen; 4734 char *to; 4735 size_t tolen; 4736 size_t len = 0; 4737 size_t done = 0; 4738 char_u *result = NULL; 4739 char_u *p; 4740 int l; 4741 4742 from = (char *)str; 4743 fromlen = slen; 4744 for (;;) 4745 { 4746 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4747 { 4748 // Allocate enough room for most conversions. When re-allocating 4749 // increase the buffer size. 4750 len = len + fromlen * 2 + 40; 4751 p = alloc(len); 4752 if (p != NULL && done > 0) 4753 mch_memmove(p, result, done); 4754 vim_free(result); 4755 result = p; 4756 if (result == NULL) // out of memory 4757 break; 4758 } 4759 4760 to = (char *)result + done; 4761 tolen = len - done - 2; 4762 // Avoid a warning for systems with a wrong iconv() prototype by 4763 // casting the second argument to void *. 4764 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4765 != (size_t)-1) 4766 { 4767 // Finished, append a NUL. 4768 *to = NUL; 4769 break; 4770 } 4771 4772 // Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4773 // iconv library may use one of them. 4774 if (!vcp->vc_fail && unconvlenp != NULL 4775 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4776 { 4777 // Handle an incomplete sequence at the end. 4778 *to = NUL; 4779 *unconvlenp = (int)fromlen; 4780 break; 4781 } 4782 4783 // Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4784 // iconv library may use one of them. 4785 else if (!vcp->vc_fail 4786 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4787 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4788 { 4789 // Can't convert: insert a '?' and skip a character. This assumes 4790 // conversion from 'encoding' to something else. In other 4791 // situations we don't know what to skip anyway. 4792 *to++ = '?'; 4793 if ((*mb_ptr2cells)((char_u *)from) > 1) 4794 *to++ = '?'; 4795 if (enc_utf8) 4796 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4797 else 4798 { 4799 l = (*mb_ptr2len)((char_u *)from); 4800 if (l > (int)fromlen) 4801 l = (int)fromlen; 4802 } 4803 from += l; 4804 fromlen -= l; 4805 } 4806 else if (ICONV_ERRNO != ICONV_E2BIG) 4807 { 4808 // conversion failed 4809 VIM_CLEAR(result); 4810 break; 4811 } 4812 // Not enough room or skipping illegal sequence. 4813 done = to - (char *)result; 4814 } 4815 4816 if (resultlenp != NULL && result != NULL) 4817 *resultlenp = (int)(to - (char *)result); 4818 return result; 4819 } 4820 4821 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4822 /* 4823 * Dynamically load the "iconv.dll" on Win32. 4824 */ 4825 4826 # ifndef DYNAMIC_ICONV // must be generating prototypes 4827 # define HINSTANCE int 4828 # endif 4829 static HINSTANCE hIconvDLL = 0; 4830 static HINSTANCE hMsvcrtDLL = 0; 4831 4832 # ifndef DYNAMIC_ICONV_DLL 4833 # define DYNAMIC_ICONV_DLL "iconv.dll" 4834 # define DYNAMIC_ICONV_DLL_ALT1 "libiconv.dll" 4835 # define DYNAMIC_ICONV_DLL_ALT2 "libiconv2.dll" 4836 # define DYNAMIC_ICONV_DLL_ALT3 "libiconv-2.dll" 4837 # endif 4838 # ifndef DYNAMIC_MSVCRT_DLL 4839 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4840 # endif 4841 4842 /* 4843 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4844 */ 4845 int 4846 iconv_enabled(int verbose) 4847 { 4848 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4849 return TRUE; 4850 4851 // The iconv DLL file goes under different names, try them all. 4852 // Do the "2" version first, it's newer. 4853 #ifdef DYNAMIC_ICONV_DLL_ALT2 4854 if (hIconvDLL == 0) 4855 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT2); 4856 #endif 4857 #ifdef DYNAMIC_ICONV_DLL_ALT3 4858 if (hIconvDLL == 0) 4859 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT3); 4860 #endif 4861 if (hIconvDLL == 0) 4862 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4863 #ifdef DYNAMIC_ICONV_DLL_ALT1 4864 if (hIconvDLL == 0) 4865 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT1); 4866 #endif 4867 4868 if (hIconvDLL != 0) 4869 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4870 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4871 { 4872 // Only give the message when 'verbose' is set, otherwise it might be 4873 // done whenever a conversion is attempted. 4874 if (verbose && p_verbose > 0) 4875 { 4876 verbose_enter(); 4877 semsg(_(e_loadlib), 4878 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4879 verbose_leave(); 4880 } 4881 iconv_end(); 4882 return FALSE; 4883 } 4884 4885 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4886 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4887 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4888 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4889 iconv_errno = get_dll_import_func(hIconvDLL, "_errno"); 4890 if (iconv_errno == NULL) 4891 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4892 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4893 || iconvctl == NULL || iconv_errno == NULL) 4894 { 4895 iconv_end(); 4896 if (verbose && p_verbose > 0) 4897 { 4898 verbose_enter(); 4899 semsg(_(e_loadfunc), "for libiconv"); 4900 verbose_leave(); 4901 } 4902 return FALSE; 4903 } 4904 return TRUE; 4905 } 4906 4907 void 4908 iconv_end(void) 4909 { 4910 // Don't use iconv() when inputting or outputting characters. 4911 if (input_conv.vc_type == CONV_ICONV) 4912 convert_setup(&input_conv, NULL, NULL); 4913 if (output_conv.vc_type == CONV_ICONV) 4914 convert_setup(&output_conv, NULL, NULL); 4915 4916 if (hIconvDLL != 0) 4917 FreeLibrary(hIconvDLL); 4918 if (hMsvcrtDLL != 0) 4919 FreeLibrary(hMsvcrtDLL); 4920 hIconvDLL = 0; 4921 hMsvcrtDLL = 0; 4922 } 4923 # endif // DYNAMIC_ICONV 4924 # endif // USE_ICONV 4925 4926 #if defined(FEAT_EVAL) || defined(PROTO) 4927 /* 4928 * "getimstatus()" function 4929 */ 4930 void 4931 f_getimstatus(typval_T *argvars UNUSED, typval_T *rettv) 4932 { 4933 # if defined(HAVE_INPUT_METHOD) 4934 rettv->vval.v_number = im_get_status(); 4935 # endif 4936 } 4937 #endif 4938 4939 /* 4940 * Setup "vcp" for conversion from "from" to "to". 4941 * The names must have been made canonical with enc_canonize(). 4942 * vcp->vc_type must have been initialized to CONV_NONE. 4943 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 4944 * instead). 4945 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 4946 * Return FAIL when conversion is not supported, OK otherwise. 4947 */ 4948 int 4949 convert_setup(vimconv_T *vcp, char_u *from, char_u *to) 4950 { 4951 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 4952 } 4953 4954 /* 4955 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 4956 * "from" unicode charsets be considered utf-8. Same for "to". 4957 */ 4958 int 4959 convert_setup_ext( 4960 vimconv_T *vcp, 4961 char_u *from, 4962 int from_unicode_is_utf8, 4963 char_u *to, 4964 int to_unicode_is_utf8) 4965 { 4966 int from_prop; 4967 int to_prop; 4968 int from_is_utf8; 4969 int to_is_utf8; 4970 4971 // Reset to no conversion. 4972 #ifdef USE_ICONV 4973 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 4974 iconv_close(vcp->vc_fd); 4975 #endif 4976 vcp->vc_type = CONV_NONE; 4977 vcp->vc_factor = 1; 4978 vcp->vc_fail = FALSE; 4979 4980 // No conversion when one of the names is empty or they are equal. 4981 if (from == NULL || *from == NUL || to == NULL || *to == NUL 4982 || STRCMP(from, to) == 0) 4983 return OK; 4984 4985 from_prop = enc_canon_props(from); 4986 to_prop = enc_canon_props(to); 4987 if (from_unicode_is_utf8) 4988 from_is_utf8 = from_prop & ENC_UNICODE; 4989 else 4990 from_is_utf8 = from_prop == ENC_UNICODE; 4991 if (to_unicode_is_utf8) 4992 to_is_utf8 = to_prop & ENC_UNICODE; 4993 else 4994 to_is_utf8 = to_prop == ENC_UNICODE; 4995 4996 if ((from_prop & ENC_LATIN1) && to_is_utf8) 4997 { 4998 // Internal latin1 -> utf-8 conversion. 4999 vcp->vc_type = CONV_TO_UTF8; 5000 vcp->vc_factor = 2; // up to twice as long 5001 } 5002 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 5003 { 5004 // Internal latin9 -> utf-8 conversion. 5005 vcp->vc_type = CONV_9_TO_UTF8; 5006 vcp->vc_factor = 3; // up to three as long (euro sign) 5007 } 5008 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 5009 { 5010 // Internal utf-8 -> latin1 conversion. 5011 vcp->vc_type = CONV_TO_LATIN1; 5012 } 5013 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 5014 { 5015 // Internal utf-8 -> latin9 conversion. 5016 vcp->vc_type = CONV_TO_LATIN9; 5017 } 5018 #ifdef MSWIN 5019 // Win32-specific codepage <-> codepage conversion without iconv. 5020 else if ((from_is_utf8 || encname2codepage(from) > 0) 5021 && (to_is_utf8 || encname2codepage(to) > 0)) 5022 { 5023 vcp->vc_type = CONV_CODEPAGE; 5024 vcp->vc_factor = 2; // up to twice as long 5025 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 5026 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 5027 } 5028 #endif 5029 #ifdef MACOS_CONVERT 5030 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 5031 { 5032 vcp->vc_type = CONV_MAC_LATIN1; 5033 } 5034 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 5035 { 5036 vcp->vc_type = CONV_MAC_UTF8; 5037 vcp->vc_factor = 2; // up to twice as long 5038 } 5039 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 5040 { 5041 vcp->vc_type = CONV_LATIN1_MAC; 5042 } 5043 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 5044 { 5045 vcp->vc_type = CONV_UTF8_MAC; 5046 } 5047 #endif 5048 #ifdef USE_ICONV 5049 else 5050 { 5051 // Use iconv() for conversion. 5052 vcp->vc_fd = (iconv_t)my_iconv_open( 5053 to_is_utf8 ? (char_u *)"utf-8" : to, 5054 from_is_utf8 ? (char_u *)"utf-8" : from); 5055 if (vcp->vc_fd != (iconv_t)-1) 5056 { 5057 vcp->vc_type = CONV_ICONV; 5058 vcp->vc_factor = 4; // could be longer too... 5059 } 5060 } 5061 #endif 5062 if (vcp->vc_type == CONV_NONE) 5063 return FAIL; 5064 5065 return OK; 5066 } 5067 5068 #if defined(FEAT_GUI) || defined(AMIGA) || defined(MSWIN) \ 5069 || defined(PROTO) 5070 /* 5071 * Do conversion on typed input characters in-place. 5072 * The input and output are not NUL terminated! 5073 * Returns the length after conversion. 5074 */ 5075 int 5076 convert_input(char_u *ptr, int len, int maxlen) 5077 { 5078 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 5079 } 5080 #endif 5081 5082 /* 5083 * Like convert_input(), but when there is an incomplete byte sequence at the 5084 * end return that as an allocated string in "restp" and set "*restlenp" to 5085 * the length. If "restp" is NULL it is not used. 5086 */ 5087 int 5088 convert_input_safe( 5089 char_u *ptr, 5090 int len, 5091 int maxlen, 5092 char_u **restp, 5093 int *restlenp) 5094 { 5095 char_u *d; 5096 int dlen = len; 5097 int unconvertlen = 0; 5098 5099 d = string_convert_ext(&input_conv, ptr, &dlen, 5100 restp == NULL ? NULL : &unconvertlen); 5101 if (d != NULL) 5102 { 5103 if (dlen <= maxlen) 5104 { 5105 if (unconvertlen > 0) 5106 { 5107 // Move the unconverted characters to allocated memory. 5108 *restp = alloc(unconvertlen); 5109 if (*restp != NULL) 5110 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 5111 *restlenp = unconvertlen; 5112 } 5113 mch_memmove(ptr, d, dlen); 5114 } 5115 else 5116 // result is too long, keep the unconverted text (the caller must 5117 // have done something wrong!) 5118 dlen = len; 5119 vim_free(d); 5120 } 5121 return dlen; 5122 } 5123 5124 /* 5125 * Convert text "ptr[*lenp]" according to "vcp". 5126 * Returns the result in allocated memory and sets "*lenp". 5127 * When "lenp" is NULL, use NUL terminated strings. 5128 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 5129 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 5130 */ 5131 char_u * 5132 string_convert( 5133 vimconv_T *vcp, 5134 char_u *ptr, 5135 int *lenp) 5136 { 5137 return string_convert_ext(vcp, ptr, lenp, NULL); 5138 } 5139 5140 /* 5141 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 5142 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 5143 * set to the number of remaining bytes. 5144 */ 5145 char_u * 5146 string_convert_ext( 5147 vimconv_T *vcp, 5148 char_u *ptr, 5149 int *lenp, 5150 int *unconvlenp) 5151 { 5152 char_u *retval = NULL; 5153 char_u *d; 5154 int len; 5155 int i; 5156 int l; 5157 int c; 5158 5159 if (lenp == NULL) 5160 len = (int)STRLEN(ptr); 5161 else 5162 len = *lenp; 5163 if (len == 0) 5164 return vim_strsave((char_u *)""); 5165 5166 switch (vcp->vc_type) 5167 { 5168 case CONV_TO_UTF8: // latin1 to utf-8 conversion 5169 retval = alloc(len * 2 + 1); 5170 if (retval == NULL) 5171 break; 5172 d = retval; 5173 for (i = 0; i < len; ++i) 5174 { 5175 c = ptr[i]; 5176 if (c < 0x80) 5177 *d++ = c; 5178 else 5179 { 5180 *d++ = 0xc0 + ((unsigned)c >> 6); 5181 *d++ = 0x80 + (c & 0x3f); 5182 } 5183 } 5184 *d = NUL; 5185 if (lenp != NULL) 5186 *lenp = (int)(d - retval); 5187 break; 5188 5189 case CONV_9_TO_UTF8: // latin9 to utf-8 conversion 5190 retval = alloc(len * 3 + 1); 5191 if (retval == NULL) 5192 break; 5193 d = retval; 5194 for (i = 0; i < len; ++i) 5195 { 5196 c = ptr[i]; 5197 switch (c) 5198 { 5199 case 0xa4: c = 0x20ac; break; // euro 5200 case 0xa6: c = 0x0160; break; // S hat 5201 case 0xa8: c = 0x0161; break; // S -hat 5202 case 0xb4: c = 0x017d; break; // Z hat 5203 case 0xb8: c = 0x017e; break; // Z -hat 5204 case 0xbc: c = 0x0152; break; // OE 5205 case 0xbd: c = 0x0153; break; // oe 5206 case 0xbe: c = 0x0178; break; // Y 5207 } 5208 d += utf_char2bytes(c, d); 5209 } 5210 *d = NUL; 5211 if (lenp != NULL) 5212 *lenp = (int)(d - retval); 5213 break; 5214 5215 case CONV_TO_LATIN1: // utf-8 to latin1 conversion 5216 case CONV_TO_LATIN9: // utf-8 to latin9 conversion 5217 retval = alloc(len + 1); 5218 if (retval == NULL) 5219 break; 5220 d = retval; 5221 for (i = 0; i < len; ++i) 5222 { 5223 l = utf_ptr2len_len(ptr + i, len - i); 5224 if (l == 0) 5225 *d++ = NUL; 5226 else if (l == 1) 5227 { 5228 int l_w = utf8len_tab_zero[ptr[i]]; 5229 5230 if (l_w == 0) 5231 { 5232 // Illegal utf-8 byte cannot be converted 5233 vim_free(retval); 5234 return NULL; 5235 } 5236 if (unconvlenp != NULL && l_w > len - i) 5237 { 5238 // Incomplete sequence at the end. 5239 *unconvlenp = len - i; 5240 break; 5241 } 5242 *d++ = ptr[i]; 5243 } 5244 else 5245 { 5246 c = utf_ptr2char(ptr + i); 5247 if (vcp->vc_type == CONV_TO_LATIN9) 5248 switch (c) 5249 { 5250 case 0x20ac: c = 0xa4; break; // euro 5251 case 0x0160: c = 0xa6; break; // S hat 5252 case 0x0161: c = 0xa8; break; // S -hat 5253 case 0x017d: c = 0xb4; break; // Z hat 5254 case 0x017e: c = 0xb8; break; // Z -hat 5255 case 0x0152: c = 0xbc; break; // OE 5256 case 0x0153: c = 0xbd; break; // oe 5257 case 0x0178: c = 0xbe; break; // Y 5258 case 0xa4: 5259 case 0xa6: 5260 case 0xa8: 5261 case 0xb4: 5262 case 0xb8: 5263 case 0xbc: 5264 case 0xbd: 5265 case 0xbe: c = 0x100; break; // not in latin9 5266 } 5267 if (!utf_iscomposing(c)) // skip composing chars 5268 { 5269 if (c < 0x100) 5270 *d++ = c; 5271 else if (vcp->vc_fail) 5272 { 5273 vim_free(retval); 5274 return NULL; 5275 } 5276 else 5277 { 5278 *d++ = 0xbf; 5279 if (utf_char2cells(c) > 1) 5280 *d++ = '?'; 5281 } 5282 } 5283 i += l - 1; 5284 } 5285 } 5286 *d = NUL; 5287 if (lenp != NULL) 5288 *lenp = (int)(d - retval); 5289 break; 5290 5291 # ifdef MACOS_CONVERT 5292 case CONV_MAC_LATIN1: 5293 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 5294 'm', 'l', unconvlenp); 5295 break; 5296 5297 case CONV_LATIN1_MAC: 5298 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 5299 'l', 'm', unconvlenp); 5300 break; 5301 5302 case CONV_MAC_UTF8: 5303 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 5304 'm', 'u', unconvlenp); 5305 break; 5306 5307 case CONV_UTF8_MAC: 5308 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 5309 'u', 'm', unconvlenp); 5310 break; 5311 # endif 5312 5313 # ifdef USE_ICONV 5314 case CONV_ICONV: // conversion with output_conv.vc_fd 5315 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 5316 break; 5317 # endif 5318 # ifdef MSWIN 5319 case CONV_CODEPAGE: // codepage -> codepage 5320 { 5321 int retlen; 5322 int tmp_len; 5323 short_u *tmp; 5324 5325 // 1. codepage/UTF-8 -> ucs-2. 5326 if (vcp->vc_cpfrom == 0) 5327 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 5328 else 5329 { 5330 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 5331 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 5332 (char *)ptr, len, 0, 0); 5333 if (tmp_len == 0 5334 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 5335 { 5336 if (lenp != NULL) 5337 *lenp = 0; 5338 if (unconvlenp != NULL) 5339 *unconvlenp = len; 5340 retval = alloc(1); 5341 if (retval) 5342 retval[0] = NUL; 5343 return retval; 5344 } 5345 } 5346 tmp = ALLOC_MULT(short_u, tmp_len); 5347 if (tmp == NULL) 5348 break; 5349 if (vcp->vc_cpfrom == 0) 5350 utf8_to_utf16(ptr, len, tmp, unconvlenp); 5351 else 5352 MultiByteToWideChar(vcp->vc_cpfrom, 0, 5353 (char *)ptr, len, tmp, tmp_len); 5354 5355 // 2. ucs-2 -> codepage/UTF-8. 5356 if (vcp->vc_cpto == 0) 5357 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 5358 else 5359 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 5360 tmp, tmp_len, 0, 0, 0, 0); 5361 retval = alloc(retlen + 1); 5362 if (retval != NULL) 5363 { 5364 if (vcp->vc_cpto == 0) 5365 utf16_to_utf8(tmp, tmp_len, retval); 5366 else 5367 WideCharToMultiByte(vcp->vc_cpto, 0, 5368 tmp, tmp_len, 5369 (char *)retval, retlen, 0, 0); 5370 retval[retlen] = NUL; 5371 if (lenp != NULL) 5372 *lenp = retlen; 5373 } 5374 vim_free(tmp); 5375 break; 5376 } 5377 # endif 5378 } 5379 5380 return retval; 5381 } 5382 5383 #if defined(FEAT_EVAL) || defined(PROTO) 5384 5385 /* 5386 * Table set by setcellwidths(). 5387 */ 5388 typedef struct 5389 { 5390 long first; 5391 long last; 5392 char width; 5393 } cw_interval_T; 5394 5395 static cw_interval_T *cw_table = NULL; 5396 static size_t cw_table_size = 0; 5397 5398 /* 5399 * Return 1 or 2 when "c" is in the cellwidth table. 5400 * Return 0 if not. 5401 */ 5402 static int 5403 cw_value(int c) 5404 { 5405 int mid, bot, top; 5406 5407 if (cw_table == NULL) 5408 return 0; 5409 5410 // first quick check for Latin1 etc. characters 5411 if (c < cw_table[0].first) 5412 return 0; 5413 5414 // binary search in table 5415 bot = 0; 5416 top = (int)cw_table_size - 1; 5417 while (top >= bot) 5418 { 5419 mid = (bot + top) / 2; 5420 if (cw_table[mid].last < c) 5421 bot = mid + 1; 5422 else if (cw_table[mid].first > c) 5423 top = mid - 1; 5424 else 5425 return cw_table[mid].width; 5426 } 5427 return 0; 5428 } 5429 5430 static int 5431 tv_nr_compare(const void *a1, const void *a2) 5432 { 5433 listitem_T *li1 = *(listitem_T **)a1; 5434 listitem_T *li2 = *(listitem_T **)a2; 5435 5436 return li1->li_tv.vval.v_number - li2->li_tv.vval.v_number; 5437 } 5438 5439 void 5440 f_setcellwidths(typval_T *argvars, typval_T *rettv UNUSED) 5441 { 5442 list_T *l; 5443 listitem_T *li; 5444 int item; 5445 int i; 5446 listitem_T **ptrs; 5447 cw_interval_T *table; 5448 5449 if (argvars[0].v_type != VAR_LIST || argvars[0].vval.v_list == NULL) 5450 { 5451 emsg(_(e_listreq)); 5452 return; 5453 } 5454 l = argvars[0].vval.v_list; 5455 if (l->lv_len == 0) 5456 { 5457 // Clearing the table. 5458 vim_free(cw_table); 5459 cw_table = NULL; 5460 cw_table_size = 0; 5461 return; 5462 } 5463 5464 ptrs = ALLOC_MULT(listitem_T *, l->lv_len); 5465 if (ptrs == NULL) 5466 return; 5467 5468 // Check that all entries are a list with three numbers, the range is 5469 // valid and the cell width is valid. 5470 item = 0; 5471 for (li = l->lv_first; li != NULL; li = li->li_next) 5472 { 5473 listitem_T *lili; 5474 varnumber_T n1; 5475 5476 if (li->li_tv.v_type != VAR_LIST || li->li_tv.vval.v_list == NULL) 5477 { 5478 semsg(_(e_list_item_nr_is_not_list), item); 5479 vim_free(ptrs); 5480 return; 5481 } 5482 5483 lili = li->li_tv.vval.v_list->lv_first; 5484 ptrs[item] = lili; 5485 for (i = 0; lili != NULL; lili = lili->li_next, ++i) 5486 { 5487 if (lili->li_tv.v_type != VAR_NUMBER) 5488 break; 5489 if (i == 0) 5490 { 5491 n1 = lili->li_tv.vval.v_number; 5492 if (n1 < 0x100) 5493 { 5494 emsg(_(e_only_values_of_0x100_and_higher_supported)); 5495 vim_free(ptrs); 5496 return; 5497 } 5498 } 5499 else if (i == 1 && lili->li_tv.vval.v_number < n1) 5500 { 5501 semsg(_(e_list_item_nr_range_invalid), item); 5502 vim_free(ptrs); 5503 return; 5504 } 5505 else if (i == 2 && (lili->li_tv.vval.v_number < 1 5506 || lili->li_tv.vval.v_number > 2)) 5507 { 5508 semsg(_(e_list_item_nr_cell_width_invalid), item); 5509 vim_free(ptrs); 5510 return; 5511 } 5512 } 5513 if (i != 3) 5514 { 5515 semsg(_(e_list_item_nr_does_not_contain_3_numbers), item); 5516 vim_free(ptrs); 5517 return; 5518 } 5519 ++item; 5520 } 5521 5522 // Sort the list on the first number. 5523 qsort((void *)ptrs, (size_t)l->lv_len, sizeof(listitem_T *), tv_nr_compare); 5524 5525 table = ALLOC_MULT(cw_interval_T, l->lv_len); 5526 if (table == NULL) 5527 { 5528 vim_free(ptrs); 5529 return; 5530 } 5531 5532 // Store the items in the new table. 5533 item = 0; 5534 for (item = 0; item < l->lv_len; ++item) 5535 { 5536 listitem_T *lili = ptrs[item]; 5537 varnumber_T n1; 5538 5539 n1 = lili->li_tv.vval.v_number; 5540 if (item > 0 && n1 <= table[item - 1].last) 5541 { 5542 semsg(_(e_overlapping_ranges_for_nr), (long)n1); 5543 vim_free(ptrs); 5544 vim_free(table); 5545 return; 5546 } 5547 table[item].first = n1; 5548 lili = lili->li_next; 5549 table[item].last = lili->li_tv.vval.v_number; 5550 lili = lili->li_next; 5551 table[item].width = lili->li_tv.vval.v_number; 5552 } 5553 5554 vim_free(ptrs); 5555 vim_free(cw_table); 5556 cw_table = table; 5557 cw_table_size = l->lv_len; 5558 } 5559 5560 void 5561 f_charclass(typval_T *argvars, typval_T *rettv UNUSED) 5562 { 5563 if (check_for_string_arg(argvars, 0) == FAIL) 5564 return; 5565 rettv->vval.v_number = mb_get_class(argvars[0].vval.v_string); 5566 } 5567 #endif 5568