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); 135 136 /* 137 * Lookup table to quickly get the length in bytes of a UTF-8 character from 138 * the first byte of a UTF-8 string. 139 * Bytes which are illegal when used as the first byte have a 1. 140 * The NUL byte has length 1. 141 */ 142 static char utf8len_tab[256] = 143 { 144 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 145 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 146 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 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 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, 151 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, 152 }; 153 154 /* 155 * Like utf8len_tab above, but using a zero for illegal lead bytes. 156 */ 157 static char utf8len_tab_zero[256] = 158 { 159 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 160 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 161 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 162 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 163 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, 164 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, 165 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, 166 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, 167 }; 168 169 170 /* 171 * Canonical encoding names and their properties. 172 * "iso-8859-n" is handled by enc_canonize() directly. 173 */ 174 static struct 175 { char *name; int prop; int codepage;} 176 enc_canon_table[] = 177 { 178 #define IDX_LATIN_1 0 179 {"latin1", ENC_8BIT + ENC_LATIN1, 1252}, 180 #define IDX_ISO_2 1 181 {"iso-8859-2", ENC_8BIT, 0}, 182 #define IDX_ISO_3 2 183 {"iso-8859-3", ENC_8BIT, 0}, 184 #define IDX_ISO_4 3 185 {"iso-8859-4", ENC_8BIT, 0}, 186 #define IDX_ISO_5 4 187 {"iso-8859-5", ENC_8BIT, 0}, 188 #define IDX_ISO_6 5 189 {"iso-8859-6", ENC_8BIT, 0}, 190 #define IDX_ISO_7 6 191 {"iso-8859-7", ENC_8BIT, 0}, 192 #define IDX_ISO_8 7 193 {"iso-8859-8", ENC_8BIT, 0}, 194 #define IDX_ISO_9 8 195 {"iso-8859-9", ENC_8BIT, 0}, 196 #define IDX_ISO_10 9 197 {"iso-8859-10", ENC_8BIT, 0}, 198 #define IDX_ISO_11 10 199 {"iso-8859-11", ENC_8BIT, 0}, 200 #define IDX_ISO_13 11 201 {"iso-8859-13", ENC_8BIT, 0}, 202 #define IDX_ISO_14 12 203 {"iso-8859-14", ENC_8BIT, 0}, 204 #define IDX_ISO_15 13 205 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0}, 206 #define IDX_KOI8_R 14 207 {"koi8-r", ENC_8BIT, 0}, 208 #define IDX_KOI8_U 15 209 {"koi8-u", ENC_8BIT, 0}, 210 #define IDX_UTF8 16 211 {"utf-8", ENC_UNICODE, 0}, 212 #define IDX_UCS2 17 213 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0}, 214 #define IDX_UCS2LE 18 215 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0}, 216 #define IDX_UTF16 19 217 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0}, 218 #define IDX_UTF16LE 20 219 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0}, 220 #define IDX_UCS4 21 221 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0}, 222 #define IDX_UCS4LE 22 223 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0}, 224 225 // For debugging DBCS encoding on Unix. 226 #define IDX_DEBUG 23 227 {"debug", ENC_DBCS, DBCS_DEBUG}, 228 #define IDX_EUC_JP 24 229 {"euc-jp", ENC_DBCS, DBCS_JPNU}, 230 #define IDX_SJIS 25 231 {"sjis", ENC_DBCS, DBCS_JPN}, 232 #define IDX_EUC_KR 26 233 {"euc-kr", ENC_DBCS, DBCS_KORU}, 234 #define IDX_EUC_CN 27 235 {"euc-cn", ENC_DBCS, DBCS_CHSU}, 236 #define IDX_EUC_TW 28 237 {"euc-tw", ENC_DBCS, DBCS_CHTU}, 238 #define IDX_BIG5 29 239 {"big5", ENC_DBCS, DBCS_CHT}, 240 241 // MS-DOS and MS-Windows codepages are included here, so that they can be 242 // used on Unix too. Most of them are similar to ISO-8859 encodings, but 243 // not exactly the same. 244 #define IDX_CP437 30 245 {"cp437", ENC_8BIT, 437}, // like iso-8859-1 246 #define IDX_CP737 31 247 {"cp737", ENC_8BIT, 737}, // like iso-8859-7 248 #define IDX_CP775 32 249 {"cp775", ENC_8BIT, 775}, // Baltic 250 #define IDX_CP850 33 251 {"cp850", ENC_8BIT, 850}, // like iso-8859-4 252 #define IDX_CP852 34 253 {"cp852", ENC_8BIT, 852}, // like iso-8859-1 254 #define IDX_CP855 35 255 {"cp855", ENC_8BIT, 855}, // like iso-8859-2 256 #define IDX_CP857 36 257 {"cp857", ENC_8BIT, 857}, // like iso-8859-5 258 #define IDX_CP860 37 259 {"cp860", ENC_8BIT, 860}, // like iso-8859-9 260 #define IDX_CP861 38 261 {"cp861", ENC_8BIT, 861}, // like iso-8859-1 262 #define IDX_CP862 39 263 {"cp862", ENC_8BIT, 862}, // like iso-8859-1 264 #define IDX_CP863 40 265 {"cp863", ENC_8BIT, 863}, // like iso-8859-8 266 #define IDX_CP865 41 267 {"cp865", ENC_8BIT, 865}, // like iso-8859-1 268 #define IDX_CP866 42 269 {"cp866", ENC_8BIT, 866}, // like iso-8859-5 270 #define IDX_CP869 43 271 {"cp869", ENC_8BIT, 869}, // like iso-8859-7 272 #define IDX_CP874 44 273 {"cp874", ENC_8BIT, 874}, // Thai 274 #define IDX_CP932 45 275 {"cp932", ENC_DBCS, DBCS_JPN}, 276 #define IDX_CP936 46 277 {"cp936", ENC_DBCS, DBCS_CHS}, 278 #define IDX_CP949 47 279 {"cp949", ENC_DBCS, DBCS_KOR}, 280 #define IDX_CP950 48 281 {"cp950", ENC_DBCS, DBCS_CHT}, 282 #define IDX_CP1250 49 283 {"cp1250", ENC_8BIT, 1250}, // Czech, Polish, etc. 284 #define IDX_CP1251 50 285 {"cp1251", ENC_8BIT, 1251}, // Cyrillic 286 // cp1252 is considered to be equal to latin1 287 #define IDX_CP1253 51 288 {"cp1253", ENC_8BIT, 1253}, // Greek 289 #define IDX_CP1254 52 290 {"cp1254", ENC_8BIT, 1254}, // Turkish 291 #define IDX_CP1255 53 292 {"cp1255", ENC_8BIT, 1255}, // Hebrew 293 #define IDX_CP1256 54 294 {"cp1256", ENC_8BIT, 1256}, // Arabic 295 #define IDX_CP1257 55 296 {"cp1257", ENC_8BIT, 1257}, // Baltic 297 #define IDX_CP1258 56 298 {"cp1258", ENC_8BIT, 1258}, // Vietnamese 299 300 #define IDX_MACROMAN 57 301 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, // Mac OS 302 #define IDX_DECMCS 58 303 {"dec-mcs", ENC_8BIT, 0}, // DEC MCS 304 #define IDX_HPROMAN8 59 305 {"hp-roman8", ENC_8BIT, 0}, // HP Roman8 306 #define IDX_COUNT 60 307 }; 308 309 /* 310 * Aliases for encoding names. 311 */ 312 static struct 313 { char *name; int canon;} 314 enc_alias_table[] = 315 { 316 {"ansi", IDX_LATIN_1}, 317 {"iso-8859-1", IDX_LATIN_1}, 318 {"latin2", IDX_ISO_2}, 319 {"latin3", IDX_ISO_3}, 320 {"latin4", IDX_ISO_4}, 321 {"cyrillic", IDX_ISO_5}, 322 {"arabic", IDX_ISO_6}, 323 {"greek", IDX_ISO_7}, 324 #ifdef MSWIN 325 {"hebrew", IDX_CP1255}, 326 #else 327 {"hebrew", IDX_ISO_8}, 328 #endif 329 {"latin5", IDX_ISO_9}, 330 {"turkish", IDX_ISO_9}, // ? 331 {"latin6", IDX_ISO_10}, 332 {"nordic", IDX_ISO_10}, // ? 333 {"thai", IDX_ISO_11}, // ? 334 {"latin7", IDX_ISO_13}, 335 {"latin8", IDX_ISO_14}, 336 {"latin9", IDX_ISO_15}, 337 {"utf8", IDX_UTF8}, 338 {"unicode", IDX_UCS2}, 339 {"ucs2", IDX_UCS2}, 340 {"ucs2be", IDX_UCS2}, 341 {"ucs-2be", IDX_UCS2}, 342 {"ucs2le", IDX_UCS2LE}, 343 {"utf16", IDX_UTF16}, 344 {"utf16be", IDX_UTF16}, 345 {"utf-16be", IDX_UTF16}, 346 {"utf16le", IDX_UTF16LE}, 347 {"ucs4", IDX_UCS4}, 348 {"ucs4be", IDX_UCS4}, 349 {"ucs-4be", IDX_UCS4}, 350 {"ucs4le", IDX_UCS4LE}, 351 {"utf32", IDX_UCS4}, 352 {"utf-32", IDX_UCS4}, 353 {"utf32be", IDX_UCS4}, 354 {"utf-32be", IDX_UCS4}, 355 {"utf32le", IDX_UCS4LE}, 356 {"utf-32le", IDX_UCS4LE}, 357 {"932", IDX_CP932}, 358 {"949", IDX_CP949}, 359 {"936", IDX_CP936}, 360 {"gbk", IDX_CP936}, 361 {"950", IDX_CP950}, 362 {"eucjp", IDX_EUC_JP}, 363 {"unix-jis", IDX_EUC_JP}, 364 {"ujis", IDX_EUC_JP}, 365 {"shift-jis", IDX_SJIS}, 366 {"pck", IDX_SJIS}, // Sun: PCK 367 {"euckr", IDX_EUC_KR}, 368 {"5601", IDX_EUC_KR}, // Sun: KS C 5601 369 {"euccn", IDX_EUC_CN}, 370 {"gb2312", IDX_EUC_CN}, 371 {"euctw", IDX_EUC_TW}, 372 #if defined(MSWIN) || defined(WIN32UNIX) || defined(MACOS_X) 373 {"japan", IDX_CP932}, 374 {"korea", IDX_CP949}, 375 {"prc", IDX_CP936}, 376 {"chinese", IDX_CP936}, 377 {"taiwan", IDX_CP950}, 378 {"big5", IDX_CP950}, 379 #else 380 {"japan", IDX_EUC_JP}, 381 {"korea", IDX_EUC_KR}, 382 {"prc", IDX_EUC_CN}, 383 {"chinese", IDX_EUC_CN}, 384 {"taiwan", IDX_EUC_TW}, 385 {"cp950", IDX_BIG5}, 386 {"950", IDX_BIG5}, 387 #endif 388 {"mac", IDX_MACROMAN}, 389 {"mac-roman", IDX_MACROMAN}, 390 {NULL, 0} 391 }; 392 393 #ifndef CP_UTF8 394 # define CP_UTF8 65001 // magic number from winnls.h 395 #endif 396 397 /* 398 * Find encoding "name" in the list of canonical encoding names. 399 * Returns -1 if not found. 400 */ 401 static int 402 enc_canon_search(char_u *name) 403 { 404 int i; 405 406 for (i = 0; i < IDX_COUNT; ++i) 407 if (STRCMP(name, enc_canon_table[i].name) == 0) 408 return i; 409 return -1; 410 } 411 412 413 /* 414 * Find canonical encoding "name" in the list and return its properties. 415 * Returns 0 if not found. 416 */ 417 int 418 enc_canon_props(char_u *name) 419 { 420 int i; 421 422 i = enc_canon_search(name); 423 if (i >= 0) 424 return enc_canon_table[i].prop; 425 #ifdef MSWIN 426 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2])) 427 { 428 CPINFO cpinfo; 429 430 // Get info on this codepage to find out what it is. 431 if (GetCPInfo(atoi((char *)name + 2), &cpinfo) != 0) 432 { 433 if (cpinfo.MaxCharSize == 1) // some single-byte encoding 434 return ENC_8BIT; 435 if (cpinfo.MaxCharSize == 2 436 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 437 // must be a DBCS encoding 438 return ENC_DBCS; 439 } 440 return 0; 441 } 442 #endif 443 if (STRNCMP(name, "2byte-", 6) == 0) 444 return ENC_DBCS; 445 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0) 446 return ENC_8BIT; 447 return 0; 448 } 449 450 /* 451 * Set up for using multi-byte characters. 452 * Called in three cases: 453 * - by main() to initialize (p_enc == NULL) 454 * - by set_init_1() after 'encoding' was set to its default. 455 * - by do_set() when 'encoding' has been set. 456 * p_enc must have been passed through enc_canonize() already. 457 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags. 458 * Fills mb_bytelen_tab[] and returns NULL when there are no problems. 459 * When there is something wrong: Returns an error message and doesn't change 460 * anything. 461 */ 462 char * 463 mb_init(void) 464 { 465 int i; 466 int idx; 467 int n; 468 int enc_dbcs_new = 0; 469 #if defined(USE_ICONV) && !defined(MSWIN) && !defined(WIN32UNIX) \ 470 && !defined(MACOS_CONVERT) 471 # define LEN_FROM_CONV 472 vimconv_T vimconv; 473 char_u *p; 474 #endif 475 476 if (p_enc == NULL) 477 { 478 // Just starting up: set the whole table to one's. 479 for (i = 0; i < 256; ++i) 480 mb_bytelen_tab[i] = 1; 481 input_conv.vc_type = CONV_NONE; 482 input_conv.vc_factor = 1; 483 output_conv.vc_type = CONV_NONE; 484 return NULL; 485 } 486 487 #ifdef MSWIN 488 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2])) 489 { 490 CPINFO cpinfo; 491 492 // Get info on this codepage to find out what it is. 493 if (GetCPInfo(atoi((char *)p_enc + 2), &cpinfo) != 0) 494 { 495 if (cpinfo.MaxCharSize == 1) 496 { 497 // some single-byte encoding 498 enc_unicode = 0; 499 enc_utf8 = FALSE; 500 } 501 else if (cpinfo.MaxCharSize == 2 502 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 503 { 504 // must be a DBCS encoding, check below 505 enc_dbcs_new = atoi((char *)p_enc + 2); 506 } 507 else 508 goto codepage_invalid; 509 } 510 else if (GetLastError() == ERROR_INVALID_PARAMETER) 511 { 512 codepage_invalid: 513 return N_("E543: Not a valid codepage"); 514 } 515 } 516 #endif 517 else if (STRNCMP(p_enc, "8bit-", 5) == 0 518 || STRNCMP(p_enc, "iso-8859-", 9) == 0) 519 { 520 // Accept any "8bit-" or "iso-8859-" name. 521 enc_unicode = 0; 522 enc_utf8 = FALSE; 523 } 524 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 525 { 526 #ifdef MSWIN 527 // Windows: accept only valid codepage numbers, check below. 528 if (p_enc[6] != 'c' || p_enc[7] != 'p' 529 || (enc_dbcs_new = atoi((char *)p_enc + 8)) == 0) 530 return e_invarg; 531 #else 532 // Unix: accept any "2byte-" name, assume current locale. 533 enc_dbcs_new = DBCS_2BYTE; 534 #endif 535 } 536 else if ((idx = enc_canon_search(p_enc)) >= 0) 537 { 538 i = enc_canon_table[idx].prop; 539 if (i & ENC_UNICODE) 540 { 541 // Unicode 542 enc_utf8 = TRUE; 543 if (i & (ENC_2BYTE | ENC_2WORD)) 544 enc_unicode = 2; 545 else if (i & ENC_4BYTE) 546 enc_unicode = 4; 547 else 548 enc_unicode = 0; 549 } 550 else if (i & ENC_DBCS) 551 { 552 // 2byte, handle below 553 enc_dbcs_new = enc_canon_table[idx].codepage; 554 } 555 else 556 { 557 // Must be 8-bit. 558 enc_unicode = 0; 559 enc_utf8 = FALSE; 560 } 561 } 562 else // Don't know what encoding this is, reject it. 563 return e_invarg; 564 565 if (enc_dbcs_new != 0) 566 { 567 #ifdef MSWIN 568 // Check if the DBCS code page is OK. 569 if (!IsValidCodePage(enc_dbcs_new)) 570 goto codepage_invalid; 571 #endif 572 enc_unicode = 0; 573 enc_utf8 = FALSE; 574 } 575 enc_dbcs = enc_dbcs_new; 576 has_mbyte = (enc_dbcs != 0 || enc_utf8); 577 578 #if defined(MSWIN) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 579 enc_codepage = encname2codepage(p_enc); 580 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0); 581 #endif 582 583 // Detect an encoding that uses latin1 characters. 584 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0 585 || STRCMP(p_enc, "iso-8859-15") == 0); 586 587 /* 588 * Set the function pointers. 589 */ 590 if (enc_utf8) 591 { 592 mb_ptr2len = utfc_ptr2len; 593 mb_ptr2len_len = utfc_ptr2len_len; 594 mb_char2len = utf_char2len; 595 mb_char2bytes = utf_char2bytes; 596 mb_ptr2cells = utf_ptr2cells; 597 mb_ptr2cells_len = utf_ptr2cells_len; 598 mb_char2cells = utf_char2cells; 599 mb_off2cells = utf_off2cells; 600 mb_ptr2char = utf_ptr2char; 601 mb_head_off = utf_head_off; 602 } 603 else if (enc_dbcs != 0) 604 { 605 mb_ptr2len = dbcs_ptr2len; 606 mb_ptr2len_len = dbcs_ptr2len_len; 607 mb_char2len = dbcs_char2len; 608 mb_char2bytes = dbcs_char2bytes; 609 mb_ptr2cells = dbcs_ptr2cells; 610 mb_ptr2cells_len = dbcs_ptr2cells_len; 611 mb_char2cells = dbcs_char2cells; 612 mb_off2cells = dbcs_off2cells; 613 mb_ptr2char = dbcs_ptr2char; 614 mb_head_off = dbcs_head_off; 615 } 616 else 617 { 618 mb_ptr2len = latin_ptr2len; 619 mb_ptr2len_len = latin_ptr2len_len; 620 mb_char2len = latin_char2len; 621 mb_char2bytes = latin_char2bytes; 622 mb_ptr2cells = latin_ptr2cells; 623 mb_ptr2cells_len = latin_ptr2cells_len; 624 mb_char2cells = latin_char2cells; 625 mb_off2cells = latin_off2cells; 626 mb_ptr2char = latin_ptr2char; 627 mb_head_off = latin_head_off; 628 } 629 630 /* 631 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN(). 632 */ 633 #ifdef LEN_FROM_CONV 634 // When 'encoding' is different from the current locale mblen() won't 635 // work. Use conversion to "utf-8" instead. 636 vimconv.vc_type = CONV_NONE; 637 if (enc_dbcs) 638 { 639 p = enc_locale(); 640 if (p == NULL || STRCMP(p, p_enc) != 0) 641 { 642 convert_setup(&vimconv, p_enc, (char_u *)"utf-8"); 643 vimconv.vc_fail = TRUE; 644 } 645 vim_free(p); 646 } 647 #endif 648 649 for (i = 0; i < 256; ++i) 650 { 651 // Our own function to reliably check the length of UTF-8 characters, 652 // independent of mblen(). 653 if (enc_utf8) 654 n = utf8len_tab[i]; 655 else if (enc_dbcs == 0) 656 n = 1; 657 else 658 { 659 #if defined(MSWIN) || defined(WIN32UNIX) 660 // enc_dbcs is set by setting 'fileencoding'. It becomes a Windows 661 // CodePage identifier, which we can pass directly in to Windows 662 // API 663 n = IsDBCSLeadByteEx(enc_dbcs, (WINBYTE)i) ? 2 : 1; 664 #else 665 # if defined(__amigaos4__) || defined(__ANDROID__) || \ 666 !(defined(HAVE_MBLEN) || defined(X_LOCALE)) 667 /* 668 * if mblen() is not available, character which MSB is turned on 669 * are treated as leading byte character. (note : This assumption 670 * is not always true.) 671 */ 672 n = (i & 0x80) ? 2 : 1; 673 # else 674 char buf[MB_MAXBYTES + 1]; 675 676 if (i == NUL) // just in case mblen() can't handle "" 677 n = 1; 678 else 679 { 680 buf[0] = i; 681 buf[1] = 0; 682 # ifdef LEN_FROM_CONV 683 if (vimconv.vc_type != CONV_NONE) 684 { 685 /* 686 * string_convert() should fail when converting the first 687 * byte of a double-byte character. 688 */ 689 p = string_convert(&vimconv, (char_u *)buf, NULL); 690 if (p != NULL) 691 { 692 vim_free(p); 693 n = 1; 694 } 695 else 696 n = 2; 697 } 698 else 699 # endif 700 { 701 /* 702 * mblen() should return -1 for invalid (means the leading 703 * multibyte) character. However there are some platforms 704 * where mblen() returns 0 for invalid character. 705 * Therefore, following condition includes 0. 706 */ 707 vim_ignored = mblen(NULL, 0); // First reset the state. 708 if (mblen(buf, (size_t)1) <= 0) 709 n = 2; 710 else 711 n = 1; 712 } 713 } 714 # endif 715 #endif 716 } 717 718 mb_bytelen_tab[i] = n; 719 } 720 721 #ifdef LEN_FROM_CONV 722 convert_setup(&vimconv, NULL, NULL); 723 #endif 724 725 // The cell width depends on the type of multi-byte characters. 726 (void)init_chartab(); 727 728 // When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] 729 screenalloc(FALSE); 730 731 // When using Unicode, set default for 'fileencodings'. 732 if (enc_utf8 && !option_was_set((char_u *)"fencs")) 733 set_string_option_direct((char_u *)"fencs", -1, 734 (char_u *)"ucs-bom,utf-8,default,latin1", OPT_FREE, 0); 735 736 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT) 737 // GNU gettext 0.10.37 supports this feature: set the codeset used for 738 // translated messages independently from the current locale. 739 (void)bind_textdomain_codeset(VIMPACKAGE, 740 enc_utf8 ? "utf-8" : (char *)p_enc); 741 #endif 742 743 #ifdef MSWIN 744 // When changing 'encoding' while starting up, then convert the command 745 // line arguments from the active codepage to 'encoding'. 746 if (starting != 0) 747 fix_arg_enc(); 748 #endif 749 750 // Fire an autocommand to let people do custom font setup. This must be 751 // after Vim has been setup for the new encoding. 752 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); 753 754 #ifdef FEAT_SPELL 755 // Need to reload spell dictionaries 756 spell_reload(); 757 #endif 758 759 return NULL; 760 } 761 762 /* 763 * Return the size of the BOM for the current buffer: 764 * 0 - no BOM 765 * 2 - UCS-2 or UTF-16 BOM 766 * 4 - UCS-4 BOM 767 * 3 - UTF-8 BOM 768 */ 769 int 770 bomb_size(void) 771 { 772 int n = 0; 773 774 if (curbuf->b_p_bomb && !curbuf->b_p_bin) 775 { 776 if (*curbuf->b_p_fenc == NUL) 777 { 778 if (enc_utf8) 779 { 780 if (enc_unicode != 0) 781 n = enc_unicode; 782 else 783 n = 3; 784 } 785 } 786 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) 787 n = 3; 788 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 789 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) 790 n = 2; 791 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) 792 n = 4; 793 } 794 return n; 795 } 796 797 #if defined(FEAT_QUICKFIX) || defined(PROTO) 798 /* 799 * Remove all BOM from "s" by moving remaining text. 800 */ 801 void 802 remove_bom(char_u *s) 803 { 804 if (enc_utf8) 805 { 806 char_u *p = s; 807 808 while ((p = vim_strbyte(p, 0xef)) != NULL) 809 { 810 if (p[1] == 0xbb && p[2] == 0xbf) 811 STRMOVE(p, p + 3); 812 else 813 ++p; 814 } 815 } 816 } 817 #endif 818 819 /* 820 * Get class of pointer: 821 * 0 for blank or NUL 822 * 1 for punctuation 823 * 2 for an (ASCII) word character 824 * >2 for other word characters 825 */ 826 int 827 mb_get_class(char_u *p) 828 { 829 return mb_get_class_buf(p, curbuf); 830 } 831 832 int 833 mb_get_class_buf(char_u *p, buf_T *buf) 834 { 835 if (MB_BYTE2LEN(p[0]) == 1) 836 { 837 if (p[0] == NUL || VIM_ISWHITE(p[0])) 838 return 0; 839 if (vim_iswordc_buf(p[0], buf)) 840 return 2; 841 return 1; 842 } 843 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) 844 return dbcs_class(p[0], p[1]); 845 if (enc_utf8) 846 return utf_class_buf(utf_ptr2char(p), buf); 847 return 0; 848 } 849 850 /* 851 * Get class of a double-byte character. This always returns 3 or bigger. 852 * TODO: Should return 1 for punctuation. 853 */ 854 int 855 dbcs_class(unsigned lead, unsigned trail) 856 { 857 switch (enc_dbcs) 858 { 859 // please add classify routine for your language in here 860 861 case DBCS_JPNU: // ? 862 case DBCS_JPN: 863 { 864 // JIS code classification 865 unsigned char lb = lead; 866 unsigned char tb = trail; 867 868 // convert process code to JIS 869 # if defined(MSWIN) || defined(WIN32UNIX) || defined(MACOS_X) 870 // process code is SJIS 871 if (lb <= 0x9f) 872 lb = (lb - 0x81) * 2 + 0x21; 873 else 874 lb = (lb - 0xc1) * 2 + 0x21; 875 if (tb <= 0x7e) 876 tb -= 0x1f; 877 else if (tb <= 0x9e) 878 tb -= 0x20; 879 else 880 { 881 tb -= 0x7e; 882 lb += 1; 883 } 884 # else 885 /* 886 * XXX: Code page identification can not use with all 887 * system! So, some other encoding information 888 * will be needed. 889 * In japanese: SJIS,EUC,UNICODE,(JIS) 890 * Note that JIS-code system don't use as 891 * process code in most system because it uses 892 * escape sequences(JIS is context depend encoding). 893 */ 894 // assume process code is JAPANESE-EUC 895 lb &= 0x7f; 896 tb &= 0x7f; 897 # endif 898 // exceptions 899 switch (lb << 8 | tb) 900 { 901 case 0x2121: // ZENKAKU space 902 return 0; 903 case 0x2122: // TOU-TEN (Japanese comma) 904 case 0x2123: // KU-TEN (Japanese period) 905 case 0x2124: // ZENKAKU comma 906 case 0x2125: // ZENKAKU period 907 return 1; 908 case 0x213c: // prolongedsound handled as KATAKANA 909 return 13; 910 } 911 // sieved by KU code 912 switch (lb) 913 { 914 case 0x21: 915 case 0x22: 916 // special symbols 917 return 10; 918 case 0x23: 919 // alphanumeric 920 return 11; 921 case 0x24: 922 // hiragana 923 return 12; 924 case 0x25: 925 // katakana 926 return 13; 927 case 0x26: 928 // greek 929 return 14; 930 case 0x27: 931 // russian 932 return 15; 933 case 0x28: 934 // lines 935 return 16; 936 default: 937 // kanji 938 return 17; 939 } 940 } 941 942 case DBCS_KORU: // ? 943 case DBCS_KOR: 944 { 945 // KS code classification 946 unsigned char c1 = lead; 947 unsigned char c2 = trail; 948 949 /* 950 * 20 : Hangul 951 * 21 : Hanja 952 * 22 : Symbols 953 * 23 : Alphanumeric/Roman Letter (Full width) 954 * 24 : Hangul Letter(Alphabet) 955 * 25 : Roman Numeral/Greek Letter 956 * 26 : Box Drawings 957 * 27 : Unit Symbols 958 * 28 : Circled/Parenthesized Letter 959 * 29 : Hiragana/Katakana 960 * 30 : Cyrillic Letter 961 */ 962 963 if (c1 >= 0xB0 && c1 <= 0xC8) 964 // Hangul 965 return 20; 966 #if defined(MSWIN) || defined(WIN32UNIX) 967 else if (c1 <= 0xA0 || c2 <= 0xA0) 968 // Extended Hangul Region : MS UHC(Unified Hangul Code) 969 // c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE 970 // c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 971 return 20; 972 #endif 973 974 else if (c1 >= 0xCA && c1 <= 0xFD) 975 // Hanja 976 return 21; 977 else switch (c1) 978 { 979 case 0xA1: 980 case 0xA2: 981 // Symbols 982 return 22; 983 case 0xA3: 984 // Alphanumeric 985 return 23; 986 case 0xA4: 987 // Hangul Letter(Alphabet) 988 return 24; 989 case 0xA5: 990 // Roman Numeral/Greek Letter 991 return 25; 992 case 0xA6: 993 // Box Drawings 994 return 26; 995 case 0xA7: 996 // Unit Symbols 997 return 27; 998 case 0xA8: 999 case 0xA9: 1000 if (c2 <= 0xAF) 1001 return 25; // Roman Letter 1002 else if (c2 >= 0xF6) 1003 return 22; // Symbols 1004 else 1005 // Circled/Parenthesized Letter 1006 return 28; 1007 case 0xAA: 1008 case 0xAB: 1009 // Hiragana/Katakana 1010 return 29; 1011 case 0xAC: 1012 // Cyrillic Letter 1013 return 30; 1014 } 1015 } 1016 default: 1017 break; 1018 } 1019 return 3; 1020 } 1021 1022 /* 1023 * mb_char2len() function pointer. 1024 * Return length in bytes of character "c". 1025 * Returns 1 for a single-byte character. 1026 */ 1027 int 1028 latin_char2len(int c UNUSED) 1029 { 1030 return 1; 1031 } 1032 1033 static int 1034 dbcs_char2len( 1035 int c) 1036 { 1037 if (c >= 0x100) 1038 return 2; 1039 return 1; 1040 } 1041 1042 /* 1043 * mb_char2bytes() function pointer. 1044 * Convert a character to its bytes. 1045 * Returns the length in bytes. 1046 */ 1047 int 1048 latin_char2bytes(int c, char_u *buf) 1049 { 1050 buf[0] = c; 1051 return 1; 1052 } 1053 1054 static int 1055 dbcs_char2bytes(int c, char_u *buf) 1056 { 1057 if (c >= 0x100) 1058 { 1059 buf[0] = (unsigned)c >> 8; 1060 buf[1] = c; 1061 // Never use a NUL byte, it causes lots of trouble. It's an invalid 1062 // character anyway. 1063 if (buf[1] == NUL) 1064 buf[1] = '\n'; 1065 return 2; 1066 } 1067 buf[0] = c; 1068 return 1; 1069 } 1070 1071 /* 1072 * mb_ptr2len() function pointer. 1073 * Get byte length of character at "*p" but stop at a NUL. 1074 * For UTF-8 this includes following composing characters. 1075 * Returns 0 when *p is NUL. 1076 */ 1077 int 1078 latin_ptr2len(char_u *p) 1079 { 1080 return MB_BYTE2LEN(*p); 1081 } 1082 1083 static int 1084 dbcs_ptr2len( 1085 char_u *p) 1086 { 1087 int len; 1088 1089 // Check if second byte is not missing. 1090 len = MB_BYTE2LEN(*p); 1091 if (len == 2 && p[1] == NUL) 1092 len = 1; 1093 return len; 1094 } 1095 1096 /* 1097 * mb_ptr2len_len() function pointer. 1098 * Like mb_ptr2len(), but limit to read "size" bytes. 1099 * Returns 0 for an empty string. 1100 * Returns 1 for an illegal char or an incomplete byte sequence. 1101 */ 1102 int 1103 latin_ptr2len_len(char_u *p, int size) 1104 { 1105 if (size < 1 || *p == NUL) 1106 return 0; 1107 return 1; 1108 } 1109 1110 static int 1111 dbcs_ptr2len_len(char_u *p, int size) 1112 { 1113 int len; 1114 1115 if (size < 1 || *p == NUL) 1116 return 0; 1117 if (size == 1) 1118 return 1; 1119 // Check that second byte is not missing. 1120 len = MB_BYTE2LEN(*p); 1121 if (len == 2 && p[1] == NUL) 1122 len = 1; 1123 return len; 1124 } 1125 1126 struct interval 1127 { 1128 long first; 1129 long last; 1130 }; 1131 1132 /* 1133 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]". 1134 */ 1135 static int 1136 intable(struct interval *table, size_t size, int c) 1137 { 1138 int mid, bot, top; 1139 1140 // first quick check for Latin1 etc. characters 1141 if (c < table[0].first) 1142 return FALSE; 1143 1144 // binary search in table 1145 bot = 0; 1146 top = (int)(size / sizeof(struct interval) - 1); 1147 while (top >= bot) 1148 { 1149 mid = (bot + top) / 2; 1150 if (table[mid].last < c) 1151 bot = mid + 1; 1152 else if (table[mid].first > c) 1153 top = mid - 1; 1154 else 1155 return TRUE; 1156 } 1157 return FALSE; 1158 } 1159 1160 // Sorted list of non-overlapping intervals of East Asian Ambiguous 1161 // characters, generated with ../runtime/tools/unicode.vim. 1162 static struct interval ambiguous[] = 1163 { 1164 {0x00a1, 0x00a1}, 1165 {0x00a4, 0x00a4}, 1166 {0x00a7, 0x00a8}, 1167 {0x00aa, 0x00aa}, 1168 {0x00ad, 0x00ae}, 1169 {0x00b0, 0x00b4}, 1170 {0x00b6, 0x00ba}, 1171 {0x00bc, 0x00bf}, 1172 {0x00c6, 0x00c6}, 1173 {0x00d0, 0x00d0}, 1174 {0x00d7, 0x00d8}, 1175 {0x00de, 0x00e1}, 1176 {0x00e6, 0x00e6}, 1177 {0x00e8, 0x00ea}, 1178 {0x00ec, 0x00ed}, 1179 {0x00f0, 0x00f0}, 1180 {0x00f2, 0x00f3}, 1181 {0x00f7, 0x00fa}, 1182 {0x00fc, 0x00fc}, 1183 {0x00fe, 0x00fe}, 1184 {0x0101, 0x0101}, 1185 {0x0111, 0x0111}, 1186 {0x0113, 0x0113}, 1187 {0x011b, 0x011b}, 1188 {0x0126, 0x0127}, 1189 {0x012b, 0x012b}, 1190 {0x0131, 0x0133}, 1191 {0x0138, 0x0138}, 1192 {0x013f, 0x0142}, 1193 {0x0144, 0x0144}, 1194 {0x0148, 0x014b}, 1195 {0x014d, 0x014d}, 1196 {0x0152, 0x0153}, 1197 {0x0166, 0x0167}, 1198 {0x016b, 0x016b}, 1199 {0x01ce, 0x01ce}, 1200 {0x01d0, 0x01d0}, 1201 {0x01d2, 0x01d2}, 1202 {0x01d4, 0x01d4}, 1203 {0x01d6, 0x01d6}, 1204 {0x01d8, 0x01d8}, 1205 {0x01da, 0x01da}, 1206 {0x01dc, 0x01dc}, 1207 {0x0251, 0x0251}, 1208 {0x0261, 0x0261}, 1209 {0x02c4, 0x02c4}, 1210 {0x02c7, 0x02c7}, 1211 {0x02c9, 0x02cb}, 1212 {0x02cd, 0x02cd}, 1213 {0x02d0, 0x02d0}, 1214 {0x02d8, 0x02db}, 1215 {0x02dd, 0x02dd}, 1216 {0x02df, 0x02df}, 1217 {0x0300, 0x036f}, 1218 {0x0391, 0x03a1}, 1219 {0x03a3, 0x03a9}, 1220 {0x03b1, 0x03c1}, 1221 {0x03c3, 0x03c9}, 1222 {0x0401, 0x0401}, 1223 {0x0410, 0x044f}, 1224 {0x0451, 0x0451}, 1225 {0x2010, 0x2010}, 1226 {0x2013, 0x2016}, 1227 {0x2018, 0x2019}, 1228 {0x201c, 0x201d}, 1229 {0x2020, 0x2022}, 1230 {0x2024, 0x2027}, 1231 {0x2030, 0x2030}, 1232 {0x2032, 0x2033}, 1233 {0x2035, 0x2035}, 1234 {0x203b, 0x203b}, 1235 {0x203e, 0x203e}, 1236 {0x2074, 0x2074}, 1237 {0x207f, 0x207f}, 1238 {0x2081, 0x2084}, 1239 {0x20ac, 0x20ac}, 1240 {0x2103, 0x2103}, 1241 {0x2105, 0x2105}, 1242 {0x2109, 0x2109}, 1243 {0x2113, 0x2113}, 1244 {0x2116, 0x2116}, 1245 {0x2121, 0x2122}, 1246 {0x2126, 0x2126}, 1247 {0x212b, 0x212b}, 1248 {0x2153, 0x2154}, 1249 {0x215b, 0x215e}, 1250 {0x2160, 0x216b}, 1251 {0x2170, 0x2179}, 1252 {0x2189, 0x2189}, 1253 {0x2190, 0x2199}, 1254 {0x21b8, 0x21b9}, 1255 {0x21d2, 0x21d2}, 1256 {0x21d4, 0x21d4}, 1257 {0x21e7, 0x21e7}, 1258 {0x2200, 0x2200}, 1259 {0x2202, 0x2203}, 1260 {0x2207, 0x2208}, 1261 {0x220b, 0x220b}, 1262 {0x220f, 0x220f}, 1263 {0x2211, 0x2211}, 1264 {0x2215, 0x2215}, 1265 {0x221a, 0x221a}, 1266 {0x221d, 0x2220}, 1267 {0x2223, 0x2223}, 1268 {0x2225, 0x2225}, 1269 {0x2227, 0x222c}, 1270 {0x222e, 0x222e}, 1271 {0x2234, 0x2237}, 1272 {0x223c, 0x223d}, 1273 {0x2248, 0x2248}, 1274 {0x224c, 0x224c}, 1275 {0x2252, 0x2252}, 1276 {0x2260, 0x2261}, 1277 {0x2264, 0x2267}, 1278 {0x226a, 0x226b}, 1279 {0x226e, 0x226f}, 1280 {0x2282, 0x2283}, 1281 {0x2286, 0x2287}, 1282 {0x2295, 0x2295}, 1283 {0x2299, 0x2299}, 1284 {0x22a5, 0x22a5}, 1285 {0x22bf, 0x22bf}, 1286 {0x2312, 0x2312}, 1287 {0x2460, 0x24e9}, 1288 {0x24eb, 0x254b}, 1289 {0x2550, 0x2573}, 1290 {0x2580, 0x258f}, 1291 {0x2592, 0x2595}, 1292 {0x25a0, 0x25a1}, 1293 {0x25a3, 0x25a9}, 1294 {0x25b2, 0x25b3}, 1295 {0x25b6, 0x25b7}, 1296 {0x25bc, 0x25bd}, 1297 {0x25c0, 0x25c1}, 1298 {0x25c6, 0x25c8}, 1299 {0x25cb, 0x25cb}, 1300 {0x25ce, 0x25d1}, 1301 {0x25e2, 0x25e5}, 1302 {0x25ef, 0x25ef}, 1303 {0x2605, 0x2606}, 1304 {0x2609, 0x2609}, 1305 {0x260e, 0x260f}, 1306 {0x261c, 0x261c}, 1307 {0x261e, 0x261e}, 1308 {0x2640, 0x2640}, 1309 {0x2642, 0x2642}, 1310 {0x2660, 0x2661}, 1311 {0x2663, 0x2665}, 1312 {0x2667, 0x266a}, 1313 {0x266c, 0x266d}, 1314 {0x266f, 0x266f}, 1315 {0x269e, 0x269f}, 1316 {0x26bf, 0x26bf}, 1317 {0x26c6, 0x26cd}, 1318 {0x26cf, 0x26d3}, 1319 {0x26d5, 0x26e1}, 1320 {0x26e3, 0x26e3}, 1321 {0x26e8, 0x26e9}, 1322 {0x26eb, 0x26f1}, 1323 {0x26f4, 0x26f4}, 1324 {0x26f6, 0x26f9}, 1325 {0x26fb, 0x26fc}, 1326 {0x26fe, 0x26ff}, 1327 {0x273d, 0x273d}, 1328 {0x2776, 0x277f}, 1329 {0x2b56, 0x2b59}, 1330 {0x3248, 0x324f}, 1331 {0xe000, 0xf8ff}, 1332 {0xfe00, 0xfe0f}, 1333 {0xfffd, 0xfffd}, 1334 {0x1f100, 0x1f10a}, 1335 {0x1f110, 0x1f12d}, 1336 {0x1f130, 0x1f169}, 1337 {0x1f170, 0x1f18d}, 1338 {0x1f18f, 0x1f190}, 1339 {0x1f19b, 0x1f1ac}, 1340 {0xe0100, 0xe01ef}, 1341 {0xf0000, 0xffffd}, 1342 {0x100000, 0x10fffd} 1343 }; 1344 1345 #if defined(FEAT_TERMINAL) || defined(PROTO) 1346 /* 1347 * utf_char2cells() with different argument type for libvterm. 1348 */ 1349 int 1350 utf_uint2cells(UINT32_T c) 1351 { 1352 if (c >= 0x100 && utf_iscomposing((int)c)) 1353 return 0; 1354 return utf_char2cells((int)c); 1355 } 1356 #endif 1357 1358 /* 1359 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1360 * Returns 4 or 6 for an unprintable character. 1361 * Is only correct for characters >= 0x80. 1362 * When p_ambw is "double", return 2 for a character with East Asian Width 1363 * class 'A'(mbiguous). 1364 */ 1365 int 1366 utf_char2cells(int c) 1367 { 1368 // Sorted list of non-overlapping intervals of East Asian double width 1369 // characters, generated with ../runtime/tools/unicode.vim. 1370 static struct interval doublewidth[] = 1371 { 1372 {0x1100, 0x115f}, 1373 {0x231a, 0x231b}, 1374 {0x2329, 0x232a}, 1375 {0x23e9, 0x23ec}, 1376 {0x23f0, 0x23f0}, 1377 {0x23f3, 0x23f3}, 1378 {0x25fd, 0x25fe}, 1379 {0x2614, 0x2615}, 1380 {0x2648, 0x2653}, 1381 {0x267f, 0x267f}, 1382 {0x2693, 0x2693}, 1383 {0x26a1, 0x26a1}, 1384 {0x26aa, 0x26ab}, 1385 {0x26bd, 0x26be}, 1386 {0x26c4, 0x26c5}, 1387 {0x26ce, 0x26ce}, 1388 {0x26d4, 0x26d4}, 1389 {0x26ea, 0x26ea}, 1390 {0x26f2, 0x26f3}, 1391 {0x26f5, 0x26f5}, 1392 {0x26fa, 0x26fa}, 1393 {0x26fd, 0x26fd}, 1394 {0x2705, 0x2705}, 1395 {0x270a, 0x270b}, 1396 {0x2728, 0x2728}, 1397 {0x274c, 0x274c}, 1398 {0x274e, 0x274e}, 1399 {0x2753, 0x2755}, 1400 {0x2757, 0x2757}, 1401 {0x2795, 0x2797}, 1402 {0x27b0, 0x27b0}, 1403 {0x27bf, 0x27bf}, 1404 {0x2b1b, 0x2b1c}, 1405 {0x2b50, 0x2b50}, 1406 {0x2b55, 0x2b55}, 1407 {0x2e80, 0x2e99}, 1408 {0x2e9b, 0x2ef3}, 1409 {0x2f00, 0x2fd5}, 1410 {0x2ff0, 0x2ffb}, 1411 {0x3000, 0x303e}, 1412 {0x3041, 0x3096}, 1413 {0x3099, 0x30ff}, 1414 {0x3105, 0x312f}, 1415 {0x3131, 0x318e}, 1416 {0x3190, 0x31ba}, 1417 {0x31c0, 0x31e3}, 1418 {0x31f0, 0x321e}, 1419 {0x3220, 0x3247}, 1420 {0x3250, 0x4dbf}, 1421 {0x4e00, 0xa48c}, 1422 {0xa490, 0xa4c6}, 1423 {0xa960, 0xa97c}, 1424 {0xac00, 0xd7a3}, 1425 {0xf900, 0xfaff}, 1426 {0xfe10, 0xfe19}, 1427 {0xfe30, 0xfe52}, 1428 {0xfe54, 0xfe66}, 1429 {0xfe68, 0xfe6b}, 1430 {0xff01, 0xff60}, 1431 {0xffe0, 0xffe6}, 1432 {0x16fe0, 0x16fe3}, 1433 {0x17000, 0x187f7}, 1434 {0x18800, 0x18af2}, 1435 {0x1b000, 0x1b11e}, 1436 {0x1b150, 0x1b152}, 1437 {0x1b164, 0x1b167}, 1438 {0x1b170, 0x1b2fb}, 1439 {0x1f004, 0x1f004}, 1440 {0x1f0cf, 0x1f0cf}, 1441 {0x1f18e, 0x1f18e}, 1442 {0x1f191, 0x1f19a}, 1443 {0x1f200, 0x1f202}, 1444 {0x1f210, 0x1f23b}, 1445 {0x1f240, 0x1f248}, 1446 {0x1f250, 0x1f251}, 1447 {0x1f260, 0x1f265}, 1448 {0x1f300, 0x1f320}, 1449 {0x1f32d, 0x1f335}, 1450 {0x1f337, 0x1f37c}, 1451 {0x1f37e, 0x1f393}, 1452 {0x1f3a0, 0x1f3ca}, 1453 {0x1f3cf, 0x1f3d3}, 1454 {0x1f3e0, 0x1f3f0}, 1455 {0x1f3f4, 0x1f3f4}, 1456 {0x1f3f8, 0x1f43e}, 1457 {0x1f440, 0x1f440}, 1458 {0x1f442, 0x1f4fc}, 1459 {0x1f4ff, 0x1f53d}, 1460 {0x1f54b, 0x1f54e}, 1461 {0x1f550, 0x1f567}, 1462 {0x1f57a, 0x1f57a}, 1463 {0x1f595, 0x1f596}, 1464 {0x1f5a4, 0x1f5a4}, 1465 {0x1f5fb, 0x1f64f}, 1466 {0x1f680, 0x1f6c5}, 1467 {0x1f6cc, 0x1f6cc}, 1468 {0x1f6d0, 0x1f6d2}, 1469 {0x1f6d5, 0x1f6d5}, 1470 {0x1f6eb, 0x1f6ec}, 1471 {0x1f6f4, 0x1f6fa}, 1472 {0x1f7e0, 0x1f7eb}, 1473 {0x1f90d, 0x1f971}, 1474 {0x1f973, 0x1f976}, 1475 {0x1f97a, 0x1f9a2}, 1476 {0x1f9a5, 0x1f9aa}, 1477 {0x1f9ae, 0x1f9ca}, 1478 {0x1f9cd, 0x1f9ff}, 1479 {0x1fa70, 0x1fa73}, 1480 {0x1fa78, 0x1fa7a}, 1481 {0x1fa80, 0x1fa82}, 1482 {0x1fa90, 0x1fa95}, 1483 {0x20000, 0x2fffd}, 1484 {0x30000, 0x3fffd} 1485 }; 1486 1487 // Sorted list of non-overlapping intervals of Emoji characters that don't 1488 // have ambiguous or double width, 1489 // based on http://unicode.org/emoji/charts/emoji-list.html 1490 static struct interval emoji_width[] = 1491 { 1492 {0x1f1e6, 0x1f1ff}, 1493 {0x1f321, 0x1f321}, 1494 {0x1f324, 0x1f32c}, 1495 {0x1f336, 0x1f336}, 1496 {0x1f37d, 0x1f37d}, 1497 {0x1f396, 0x1f397}, 1498 {0x1f399, 0x1f39b}, 1499 {0x1f39e, 0x1f39f}, 1500 {0x1f3cb, 0x1f3ce}, 1501 {0x1f3d4, 0x1f3df}, 1502 {0x1f3f3, 0x1f3f5}, 1503 {0x1f3f7, 0x1f3f7}, 1504 {0x1f43f, 0x1f43f}, 1505 {0x1f441, 0x1f441}, 1506 {0x1f4fd, 0x1f4fd}, 1507 {0x1f549, 0x1f54a}, 1508 {0x1f56f, 0x1f570}, 1509 {0x1f573, 0x1f579}, 1510 {0x1f587, 0x1f587}, 1511 {0x1f58a, 0x1f58d}, 1512 {0x1f590, 0x1f590}, 1513 {0x1f5a5, 0x1f5a5}, 1514 {0x1f5a8, 0x1f5a8}, 1515 {0x1f5b1, 0x1f5b2}, 1516 {0x1f5bc, 0x1f5bc}, 1517 {0x1f5c2, 0x1f5c4}, 1518 {0x1f5d1, 0x1f5d3}, 1519 {0x1f5dc, 0x1f5de}, 1520 {0x1f5e1, 0x1f5e1}, 1521 {0x1f5e3, 0x1f5e3}, 1522 {0x1f5e8, 0x1f5e8}, 1523 {0x1f5ef, 0x1f5ef}, 1524 {0x1f5f3, 0x1f5f3}, 1525 {0x1f5fa, 0x1f5fa}, 1526 {0x1f6cb, 0x1f6cf}, 1527 {0x1f6e0, 0x1f6e5}, 1528 {0x1f6e9, 0x1f6e9}, 1529 {0x1f6f0, 0x1f6f0}, 1530 {0x1f6f3, 0x1f6f3} 1531 }; 1532 1533 if (c >= 0x100) 1534 { 1535 #ifdef USE_WCHAR_FUNCTIONS 1536 /* 1537 * Assume the library function wcwidth() works better than our own 1538 * stuff. It should return 1 for ambiguous width chars! 1539 */ 1540 int n = wcwidth(c); 1541 1542 if (n < 0) 1543 return 6; // unprintable, displays <xxxx> 1544 if (n > 1) 1545 return n; 1546 #else 1547 if (!utf_printable(c)) 1548 return 6; // unprintable, displays <xxxx> 1549 if (intable(doublewidth, sizeof(doublewidth), c)) 1550 return 2; 1551 #endif 1552 if (p_emoji && intable(emoji_width, sizeof(emoji_width), c)) 1553 return 2; 1554 } 1555 1556 // Characters below 0x100 are influenced by 'isprint' option 1557 else if (c >= 0x80 && !vim_isprintc(c)) 1558 return 4; // unprintable, displays <xx> 1559 1560 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c)) 1561 return 2; 1562 1563 return 1; 1564 } 1565 1566 /* 1567 * mb_ptr2cells() function pointer. 1568 * Return the number of display cells character at "*p" occupies. 1569 * This doesn't take care of unprintable characters, use ptr2cells() for that. 1570 */ 1571 int 1572 latin_ptr2cells(char_u *p UNUSED) 1573 { 1574 return 1; 1575 } 1576 1577 int 1578 utf_ptr2cells( 1579 char_u *p) 1580 { 1581 int c; 1582 1583 // Need to convert to a wide character. 1584 if (*p >= 0x80) 1585 { 1586 c = utf_ptr2char(p); 1587 // An illegal byte is displayed as <xx>. 1588 if (utf_ptr2len(p) == 1 || c == NUL) 1589 return 4; 1590 // If the char is ASCII it must be an overlong sequence. 1591 if (c < 0x80) 1592 return char2cells(c); 1593 return utf_char2cells(c); 1594 } 1595 return 1; 1596 } 1597 1598 int 1599 dbcs_ptr2cells(char_u *p) 1600 { 1601 // Number of cells is equal to number of bytes, except for euc-jp when 1602 // the first byte is 0x8e. 1603 if (enc_dbcs == DBCS_JPNU && *p == 0x8e) 1604 return 1; 1605 return MB_BYTE2LEN(*p); 1606 } 1607 1608 /* 1609 * mb_ptr2cells_len() function pointer. 1610 * Like mb_ptr2cells(), but limit string length to "size". 1611 * For an empty string or truncated character returns 1. 1612 */ 1613 int 1614 latin_ptr2cells_len(char_u *p UNUSED, int size UNUSED) 1615 { 1616 return 1; 1617 } 1618 1619 static int 1620 utf_ptr2cells_len(char_u *p, int size) 1621 { 1622 int c; 1623 1624 // Need to convert to a wide character. 1625 if (size > 0 && *p >= 0x80) 1626 { 1627 if (utf_ptr2len_len(p, size) < utf8len_tab[*p]) 1628 return 1; // truncated 1629 c = utf_ptr2char(p); 1630 // An illegal byte is displayed as <xx>. 1631 if (utf_ptr2len(p) == 1 || c == NUL) 1632 return 4; 1633 // If the char is ASCII it must be an overlong sequence. 1634 if (c < 0x80) 1635 return char2cells(c); 1636 return utf_char2cells(c); 1637 } 1638 return 1; 1639 } 1640 1641 static int 1642 dbcs_ptr2cells_len(char_u *p, int size) 1643 { 1644 // Number of cells is equal to number of bytes, except for euc-jp when 1645 // the first byte is 0x8e. 1646 if (size <= 1 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)) 1647 return 1; 1648 return MB_BYTE2LEN(*p); 1649 } 1650 1651 /* 1652 * mb_char2cells() function pointer. 1653 * Return the number of display cells character "c" occupies. 1654 * Only takes care of multi-byte chars, not "^C" and such. 1655 */ 1656 int 1657 latin_char2cells(int c UNUSED) 1658 { 1659 return 1; 1660 } 1661 1662 static int 1663 dbcs_char2cells(int c) 1664 { 1665 // Number of cells is equal to number of bytes, except for euc-jp when 1666 // the first byte is 0x8e. 1667 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e) 1668 return 1; 1669 // use the first byte 1670 return MB_BYTE2LEN((unsigned)c >> 8); 1671 } 1672 1673 /* 1674 * Return the number of cells occupied by string "p". 1675 * Stop at a NUL character. When "len" >= 0 stop at character "p[len]". 1676 */ 1677 int 1678 mb_string2cells(char_u *p, int len) 1679 { 1680 int i; 1681 int clen = 0; 1682 1683 for (i = 0; (len < 0 || i < len) && p[i] != NUL; i += (*mb_ptr2len)(p + i)) 1684 clen += (*mb_ptr2cells)(p + i); 1685 return clen; 1686 } 1687 1688 /* 1689 * mb_off2cells() function pointer. 1690 * Return number of display cells for char at ScreenLines[off]. 1691 * We make sure that the offset used is less than "max_off". 1692 */ 1693 int 1694 latin_off2cells(unsigned off UNUSED, unsigned max_off UNUSED) 1695 { 1696 return 1; 1697 } 1698 1699 int 1700 dbcs_off2cells(unsigned off, unsigned max_off) 1701 { 1702 // never check beyond end of the line 1703 if (off >= max_off) 1704 return 1; 1705 1706 // Number of cells is equal to number of bytes, except for euc-jp when 1707 // the first byte is 0x8e. 1708 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e) 1709 return 1; 1710 return MB_BYTE2LEN(ScreenLines[off]); 1711 } 1712 1713 int 1714 utf_off2cells(unsigned off, unsigned max_off) 1715 { 1716 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1; 1717 } 1718 1719 /* 1720 * mb_ptr2char() function pointer. 1721 * Convert a byte sequence into a character. 1722 */ 1723 int 1724 latin_ptr2char(char_u *p) 1725 { 1726 return *p; 1727 } 1728 1729 static int 1730 dbcs_ptr2char(char_u *p) 1731 { 1732 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL) 1733 return (p[0] << 8) + p[1]; 1734 return *p; 1735 } 1736 1737 /* 1738 * Convert a UTF-8 byte sequence to a wide character. 1739 * If the sequence is illegal or truncated by a NUL the first byte is 1740 * returned. 1741 * For an overlong sequence this may return zero. 1742 * Does not include composing characters, of course. 1743 */ 1744 int 1745 utf_ptr2char(char_u *p) 1746 { 1747 int len; 1748 1749 if (p[0] < 0x80) // be quick for ASCII 1750 return p[0]; 1751 1752 len = utf8len_tab_zero[p[0]]; 1753 if (len > 1 && (p[1] & 0xc0) == 0x80) 1754 { 1755 if (len == 2) 1756 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); 1757 if ((p[2] & 0xc0) == 0x80) 1758 { 1759 if (len == 3) 1760 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) 1761 + (p[2] & 0x3f); 1762 if ((p[3] & 0xc0) == 0x80) 1763 { 1764 if (len == 4) 1765 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) 1766 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); 1767 if ((p[4] & 0xc0) == 0x80) 1768 { 1769 if (len == 5) 1770 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) 1771 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) 1772 + (p[4] & 0x3f); 1773 if ((p[5] & 0xc0) == 0x80 && len == 6) 1774 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) 1775 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) 1776 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); 1777 } 1778 } 1779 } 1780 } 1781 // Illegal value, just return the first byte 1782 return p[0]; 1783 } 1784 1785 /* 1786 * Convert a UTF-8 byte sequence to a wide character. 1787 * String is assumed to be terminated by NUL or after "n" bytes, whichever 1788 * comes first. 1789 * The function is safe in the sense that it never accesses memory beyond the 1790 * first "n" bytes of "s". 1791 * 1792 * On success, returns decoded codepoint, advances "s" to the beginning of 1793 * next character and decreases "n" accordingly. 1794 * 1795 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past 1796 * NUL byte. 1797 * 1798 * If byte sequence is illegal or incomplete, returns -1 and does not advance 1799 * "s". 1800 */ 1801 static int 1802 utf_safe_read_char_adv(char_u **s, size_t *n) 1803 { 1804 int c, k; 1805 1806 if (*n == 0) // end of buffer 1807 return 0; 1808 1809 k = utf8len_tab_zero[**s]; 1810 1811 if (k == 1) 1812 { 1813 // ASCII character or NUL 1814 (*n)--; 1815 return *(*s)++; 1816 } 1817 1818 if ((size_t)k <= *n) 1819 { 1820 // We have a multibyte sequence and it isn't truncated by buffer 1821 // limits so utf_ptr2char() is safe to use. Or the first byte is 1822 // illegal (k=0), and it's also safe to use utf_ptr2char(). 1823 c = utf_ptr2char(*s); 1824 1825 // On failure, utf_ptr2char() returns the first byte, so here we 1826 // check equality with the first byte. The only non-ASCII character 1827 // which equals the first byte of its own UTF-8 representation is 1828 // U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too. 1829 // It's safe even if n=1, else we would have k=2 > n. 1830 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83)) 1831 { 1832 // byte sequence was successfully decoded 1833 *s += k; 1834 *n -= k; 1835 return c; 1836 } 1837 } 1838 1839 // byte sequence is incomplete or illegal 1840 return -1; 1841 } 1842 1843 /* 1844 * Get character at **pp and advance *pp to the next character. 1845 * Note: composing characters are skipped! 1846 */ 1847 int 1848 mb_ptr2char_adv(char_u **pp) 1849 { 1850 int c; 1851 1852 c = (*mb_ptr2char)(*pp); 1853 *pp += (*mb_ptr2len)(*pp); 1854 return c; 1855 } 1856 1857 /* 1858 * Get character at **pp and advance *pp to the next character. 1859 * Note: composing characters are returned as separate characters. 1860 */ 1861 int 1862 mb_cptr2char_adv(char_u **pp) 1863 { 1864 int c; 1865 1866 c = (*mb_ptr2char)(*pp); 1867 if (enc_utf8) 1868 *pp += utf_ptr2len(*pp); 1869 else 1870 *pp += (*mb_ptr2len)(*pp); 1871 return c; 1872 } 1873 1874 #if defined(FEAT_ARABIC) || defined(PROTO) 1875 /* 1876 * Check if the character pointed to by "p2" is a composing character when it 1877 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which 1878 * behaves like a composing character. 1879 */ 1880 int 1881 utf_composinglike(char_u *p1, char_u *p2) 1882 { 1883 int c2; 1884 1885 c2 = utf_ptr2char(p2); 1886 if (utf_iscomposing(c2)) 1887 return TRUE; 1888 if (!arabic_maycombine(c2)) 1889 return FALSE; 1890 return arabic_combine(utf_ptr2char(p1), c2); 1891 } 1892 #endif 1893 1894 /* 1895 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1896 * composing characters. 1897 */ 1898 int 1899 utfc_ptr2char( 1900 char_u *p, 1901 int *pcc) // return: composing chars, last one is 0 1902 { 1903 int len; 1904 int c; 1905 int cc; 1906 int i = 0; 1907 1908 c = utf_ptr2char(p); 1909 len = utf_ptr2len(p); 1910 1911 // Only accept a composing char when the first char isn't illegal. 1912 if ((len > 1 || *p < 0x80) 1913 && p[len] >= 0x80 1914 && UTF_COMPOSINGLIKE(p, p + len)) 1915 { 1916 cc = utf_ptr2char(p + len); 1917 for (;;) 1918 { 1919 pcc[i++] = cc; 1920 if (i == MAX_MCO) 1921 break; 1922 len += utf_ptr2len(p + len); 1923 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1924 break; 1925 } 1926 } 1927 1928 if (i < MAX_MCO) // last composing char must be 0 1929 pcc[i] = 0; 1930 1931 return c; 1932 } 1933 1934 /* 1935 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1936 * composing characters. Use no more than p[maxlen]. 1937 */ 1938 int 1939 utfc_ptr2char_len( 1940 char_u *p, 1941 int *pcc, // return: composing chars, last one is 0 1942 int maxlen) 1943 { 1944 int len; 1945 int c; 1946 int cc; 1947 int i = 0; 1948 1949 c = utf_ptr2char(p); 1950 len = utf_ptr2len_len(p, maxlen); 1951 // Only accept a composing char when the first char isn't illegal. 1952 if ((len > 1 || *p < 0x80) 1953 && len < maxlen 1954 && p[len] >= 0x80 1955 && UTF_COMPOSINGLIKE(p, p + len)) 1956 { 1957 cc = utf_ptr2char(p + len); 1958 for (;;) 1959 { 1960 pcc[i++] = cc; 1961 if (i == MAX_MCO) 1962 break; 1963 len += utf_ptr2len_len(p + len, maxlen - len); 1964 if (len >= maxlen 1965 || p[len] < 0x80 1966 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1967 break; 1968 } 1969 } 1970 1971 if (i < MAX_MCO) // last composing char must be 0 1972 pcc[i] = 0; 1973 1974 return c; 1975 } 1976 1977 /* 1978 * Convert the character at screen position "off" to a sequence of bytes. 1979 * Includes the composing characters. 1980 * "buf" must at least have the length MB_MAXBYTES + 1. 1981 * Only to be used when ScreenLinesUC[off] != 0. 1982 * Returns the produced number of bytes. 1983 */ 1984 int 1985 utfc_char2bytes(int off, char_u *buf) 1986 { 1987 int len; 1988 int i; 1989 1990 len = utf_char2bytes(ScreenLinesUC[off], buf); 1991 for (i = 0; i < Screen_mco; ++i) 1992 { 1993 if (ScreenLinesC[i][off] == 0) 1994 break; 1995 len += utf_char2bytes(ScreenLinesC[i][off], buf + len); 1996 } 1997 return len; 1998 } 1999 2000 /* 2001 * Get the length of a UTF-8 byte sequence, not including any following 2002 * composing characters. 2003 * Returns 0 for "". 2004 * Returns 1 for an illegal byte sequence. 2005 */ 2006 int 2007 utf_ptr2len(char_u *p) 2008 { 2009 int len; 2010 int i; 2011 2012 if (*p == NUL) 2013 return 0; 2014 len = utf8len_tab[*p]; 2015 for (i = 1; i < len; ++i) 2016 if ((p[i] & 0xc0) != 0x80) 2017 return 1; 2018 return len; 2019 } 2020 2021 /* 2022 * Return length of UTF-8 character, obtained from the first byte. 2023 * "b" must be between 0 and 255! 2024 * Returns 1 for an invalid first byte value. 2025 */ 2026 int 2027 utf_byte2len(int b) 2028 { 2029 return utf8len_tab[b]; 2030 } 2031 2032 /* 2033 * Get the length of UTF-8 byte sequence "p[size]". Does not include any 2034 * following composing characters. 2035 * Returns 1 for "". 2036 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.). 2037 * Returns number > "size" for an incomplete byte sequence. 2038 * Never returns zero. 2039 */ 2040 int 2041 utf_ptr2len_len(char_u *p, int size) 2042 { 2043 int len; 2044 int i; 2045 int m; 2046 2047 len = utf8len_tab[*p]; 2048 if (len == 1) 2049 return 1; // NUL, ascii or illegal lead byte 2050 if (len > size) 2051 m = size; // incomplete byte sequence. 2052 else 2053 m = len; 2054 for (i = 1; i < m; ++i) 2055 if ((p[i] & 0xc0) != 0x80) 2056 return 1; 2057 return len; 2058 } 2059 2060 /* 2061 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. 2062 * This includes following composing characters. 2063 */ 2064 int 2065 utfc_ptr2len(char_u *p) 2066 { 2067 int len; 2068 int b0 = *p; 2069 #ifdef FEAT_ARABIC 2070 int prevlen; 2071 #endif 2072 2073 if (b0 == NUL) 2074 return 0; 2075 if (b0 < 0x80 && p[1] < 0x80) // be quick for ASCII 2076 return 1; 2077 2078 // Skip over first UTF-8 char, stopping at a NUL byte. 2079 len = utf_ptr2len(p); 2080 2081 // Check for illegal byte. 2082 if (len == 1 && b0 >= 0x80) 2083 return 1; 2084 2085 /* 2086 * Check for composing characters. We can handle only the first six, but 2087 * skip all of them (otherwise the cursor would get stuck). 2088 */ 2089 #ifdef FEAT_ARABIC 2090 prevlen = 0; 2091 #endif 2092 for (;;) 2093 { 2094 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2095 return len; 2096 2097 // Skip over composing char 2098 #ifdef FEAT_ARABIC 2099 prevlen = len; 2100 #endif 2101 len += utf_ptr2len(p + len); 2102 } 2103 } 2104 2105 /* 2106 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" 2107 * takes. This includes following composing characters. 2108 * Returns 0 for an empty string. 2109 * Returns 1 for an illegal char or an incomplete byte sequence. 2110 */ 2111 int 2112 utfc_ptr2len_len(char_u *p, int size) 2113 { 2114 int len; 2115 #ifdef FEAT_ARABIC 2116 int prevlen; 2117 #endif 2118 2119 if (size < 1 || *p == NUL) 2120 return 0; 2121 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) // be quick for ASCII 2122 return 1; 2123 2124 // Skip over first UTF-8 char, stopping at a NUL byte. 2125 len = utf_ptr2len_len(p, size); 2126 2127 // Check for illegal byte and incomplete byte sequence. 2128 if ((len == 1 && p[0] >= 0x80) || len > size) 2129 return 1; 2130 2131 /* 2132 * Check for composing characters. We can handle only the first six, but 2133 * skip all of them (otherwise the cursor would get stuck). 2134 */ 2135 #ifdef FEAT_ARABIC 2136 prevlen = 0; 2137 #endif 2138 while (len < size) 2139 { 2140 int len_next_char; 2141 2142 if (p[len] < 0x80) 2143 break; 2144 2145 /* 2146 * Next character length should not go beyond size to ensure that 2147 * UTF_COMPOSINGLIKE(...) does not read beyond size. 2148 */ 2149 len_next_char = utf_ptr2len_len(p + len, size - len); 2150 if (len_next_char > size - len) 2151 break; 2152 2153 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2154 break; 2155 2156 // Skip over composing char 2157 #ifdef FEAT_ARABIC 2158 prevlen = len; 2159 #endif 2160 len += len_next_char; 2161 } 2162 return len; 2163 } 2164 2165 /* 2166 * Return the number of bytes the UTF-8 encoding of character "c" takes. 2167 * This does not include composing characters. 2168 */ 2169 int 2170 utf_char2len(int c) 2171 { 2172 if (c < 0x80) 2173 return 1; 2174 if (c < 0x800) 2175 return 2; 2176 if (c < 0x10000) 2177 return 3; 2178 if (c < 0x200000) 2179 return 4; 2180 if (c < 0x4000000) 2181 return 5; 2182 return 6; 2183 } 2184 2185 /* 2186 * Convert Unicode character "c" to UTF-8 string in "buf[]". 2187 * Returns the number of bytes. 2188 */ 2189 int 2190 utf_char2bytes(int c, char_u *buf) 2191 { 2192 if (c < 0x80) // 7 bits 2193 { 2194 buf[0] = c; 2195 return 1; 2196 } 2197 if (c < 0x800) // 11 bits 2198 { 2199 buf[0] = 0xc0 + ((unsigned)c >> 6); 2200 buf[1] = 0x80 + (c & 0x3f); 2201 return 2; 2202 } 2203 if (c < 0x10000) // 16 bits 2204 { 2205 buf[0] = 0xe0 + ((unsigned)c >> 12); 2206 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2207 buf[2] = 0x80 + (c & 0x3f); 2208 return 3; 2209 } 2210 if (c < 0x200000) // 21 bits 2211 { 2212 buf[0] = 0xf0 + ((unsigned)c >> 18); 2213 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2214 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2215 buf[3] = 0x80 + (c & 0x3f); 2216 return 4; 2217 } 2218 if (c < 0x4000000) // 26 bits 2219 { 2220 buf[0] = 0xf8 + ((unsigned)c >> 24); 2221 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2222 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2223 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2224 buf[4] = 0x80 + (c & 0x3f); 2225 return 5; 2226 } 2227 // 31 bits 2228 buf[0] = 0xfc + ((unsigned)c >> 30); 2229 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f); 2230 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2231 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2232 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2233 buf[5] = 0x80 + (c & 0x3f); 2234 return 6; 2235 } 2236 2237 #if defined(FEAT_TERMINAL) || defined(PROTO) 2238 /* 2239 * utf_iscomposing() with different argument type for libvterm. 2240 */ 2241 int 2242 utf_iscomposing_uint(UINT32_T c) 2243 { 2244 return utf_iscomposing((int)c); 2245 } 2246 #endif 2247 2248 /* 2249 * Return TRUE if "c" is a composing UTF-8 character. This means it will be 2250 * drawn on top of the preceding character. 2251 * Based on code from Markus Kuhn. 2252 */ 2253 int 2254 utf_iscomposing(int c) 2255 { 2256 // Sorted list of non-overlapping intervals. 2257 // Generated by ../runtime/tools/unicode.vim. 2258 static struct interval combining[] = 2259 { 2260 {0x0300, 0x036f}, 2261 {0x0483, 0x0489}, 2262 {0x0591, 0x05bd}, 2263 {0x05bf, 0x05bf}, 2264 {0x05c1, 0x05c2}, 2265 {0x05c4, 0x05c5}, 2266 {0x05c7, 0x05c7}, 2267 {0x0610, 0x061a}, 2268 {0x064b, 0x065f}, 2269 {0x0670, 0x0670}, 2270 {0x06d6, 0x06dc}, 2271 {0x06df, 0x06e4}, 2272 {0x06e7, 0x06e8}, 2273 {0x06ea, 0x06ed}, 2274 {0x0711, 0x0711}, 2275 {0x0730, 0x074a}, 2276 {0x07a6, 0x07b0}, 2277 {0x07eb, 0x07f3}, 2278 {0x07fd, 0x07fd}, 2279 {0x0816, 0x0819}, 2280 {0x081b, 0x0823}, 2281 {0x0825, 0x0827}, 2282 {0x0829, 0x082d}, 2283 {0x0859, 0x085b}, 2284 {0x08d3, 0x08e1}, 2285 {0x08e3, 0x0903}, 2286 {0x093a, 0x093c}, 2287 {0x093e, 0x094f}, 2288 {0x0951, 0x0957}, 2289 {0x0962, 0x0963}, 2290 {0x0981, 0x0983}, 2291 {0x09bc, 0x09bc}, 2292 {0x09be, 0x09c4}, 2293 {0x09c7, 0x09c8}, 2294 {0x09cb, 0x09cd}, 2295 {0x09d7, 0x09d7}, 2296 {0x09e2, 0x09e3}, 2297 {0x09fe, 0x09fe}, 2298 {0x0a01, 0x0a03}, 2299 {0x0a3c, 0x0a3c}, 2300 {0x0a3e, 0x0a42}, 2301 {0x0a47, 0x0a48}, 2302 {0x0a4b, 0x0a4d}, 2303 {0x0a51, 0x0a51}, 2304 {0x0a70, 0x0a71}, 2305 {0x0a75, 0x0a75}, 2306 {0x0a81, 0x0a83}, 2307 {0x0abc, 0x0abc}, 2308 {0x0abe, 0x0ac5}, 2309 {0x0ac7, 0x0ac9}, 2310 {0x0acb, 0x0acd}, 2311 {0x0ae2, 0x0ae3}, 2312 {0x0afa, 0x0aff}, 2313 {0x0b01, 0x0b03}, 2314 {0x0b3c, 0x0b3c}, 2315 {0x0b3e, 0x0b44}, 2316 {0x0b47, 0x0b48}, 2317 {0x0b4b, 0x0b4d}, 2318 {0x0b56, 0x0b57}, 2319 {0x0b62, 0x0b63}, 2320 {0x0b82, 0x0b82}, 2321 {0x0bbe, 0x0bc2}, 2322 {0x0bc6, 0x0bc8}, 2323 {0x0bca, 0x0bcd}, 2324 {0x0bd7, 0x0bd7}, 2325 {0x0c00, 0x0c04}, 2326 {0x0c3e, 0x0c44}, 2327 {0x0c46, 0x0c48}, 2328 {0x0c4a, 0x0c4d}, 2329 {0x0c55, 0x0c56}, 2330 {0x0c62, 0x0c63}, 2331 {0x0c81, 0x0c83}, 2332 {0x0cbc, 0x0cbc}, 2333 {0x0cbe, 0x0cc4}, 2334 {0x0cc6, 0x0cc8}, 2335 {0x0cca, 0x0ccd}, 2336 {0x0cd5, 0x0cd6}, 2337 {0x0ce2, 0x0ce3}, 2338 {0x0d00, 0x0d03}, 2339 {0x0d3b, 0x0d3c}, 2340 {0x0d3e, 0x0d44}, 2341 {0x0d46, 0x0d48}, 2342 {0x0d4a, 0x0d4d}, 2343 {0x0d57, 0x0d57}, 2344 {0x0d62, 0x0d63}, 2345 {0x0d82, 0x0d83}, 2346 {0x0dca, 0x0dca}, 2347 {0x0dcf, 0x0dd4}, 2348 {0x0dd6, 0x0dd6}, 2349 {0x0dd8, 0x0ddf}, 2350 {0x0df2, 0x0df3}, 2351 {0x0e31, 0x0e31}, 2352 {0x0e34, 0x0e3a}, 2353 {0x0e47, 0x0e4e}, 2354 {0x0eb1, 0x0eb1}, 2355 {0x0eb4, 0x0ebc}, 2356 {0x0ec8, 0x0ecd}, 2357 {0x0f18, 0x0f19}, 2358 {0x0f35, 0x0f35}, 2359 {0x0f37, 0x0f37}, 2360 {0x0f39, 0x0f39}, 2361 {0x0f3e, 0x0f3f}, 2362 {0x0f71, 0x0f84}, 2363 {0x0f86, 0x0f87}, 2364 {0x0f8d, 0x0f97}, 2365 {0x0f99, 0x0fbc}, 2366 {0x0fc6, 0x0fc6}, 2367 {0x102b, 0x103e}, 2368 {0x1056, 0x1059}, 2369 {0x105e, 0x1060}, 2370 {0x1062, 0x1064}, 2371 {0x1067, 0x106d}, 2372 {0x1071, 0x1074}, 2373 {0x1082, 0x108d}, 2374 {0x108f, 0x108f}, 2375 {0x109a, 0x109d}, 2376 {0x135d, 0x135f}, 2377 {0x1712, 0x1714}, 2378 {0x1732, 0x1734}, 2379 {0x1752, 0x1753}, 2380 {0x1772, 0x1773}, 2381 {0x17b4, 0x17d3}, 2382 {0x17dd, 0x17dd}, 2383 {0x180b, 0x180d}, 2384 {0x1885, 0x1886}, 2385 {0x18a9, 0x18a9}, 2386 {0x1920, 0x192b}, 2387 {0x1930, 0x193b}, 2388 {0x1a17, 0x1a1b}, 2389 {0x1a55, 0x1a5e}, 2390 {0x1a60, 0x1a7c}, 2391 {0x1a7f, 0x1a7f}, 2392 {0x1ab0, 0x1abe}, 2393 {0x1b00, 0x1b04}, 2394 {0x1b34, 0x1b44}, 2395 {0x1b6b, 0x1b73}, 2396 {0x1b80, 0x1b82}, 2397 {0x1ba1, 0x1bad}, 2398 {0x1be6, 0x1bf3}, 2399 {0x1c24, 0x1c37}, 2400 {0x1cd0, 0x1cd2}, 2401 {0x1cd4, 0x1ce8}, 2402 {0x1ced, 0x1ced}, 2403 {0x1cf4, 0x1cf4}, 2404 {0x1cf7, 0x1cf9}, 2405 {0x1dc0, 0x1df9}, 2406 {0x1dfb, 0x1dff}, 2407 {0x20d0, 0x20f0}, 2408 {0x2cef, 0x2cf1}, 2409 {0x2d7f, 0x2d7f}, 2410 {0x2de0, 0x2dff}, 2411 {0x302a, 0x302f}, 2412 {0x3099, 0x309a}, 2413 {0xa66f, 0xa672}, 2414 {0xa674, 0xa67d}, 2415 {0xa69e, 0xa69f}, 2416 {0xa6f0, 0xa6f1}, 2417 {0xa802, 0xa802}, 2418 {0xa806, 0xa806}, 2419 {0xa80b, 0xa80b}, 2420 {0xa823, 0xa827}, 2421 {0xa880, 0xa881}, 2422 {0xa8b4, 0xa8c5}, 2423 {0xa8e0, 0xa8f1}, 2424 {0xa8ff, 0xa8ff}, 2425 {0xa926, 0xa92d}, 2426 {0xa947, 0xa953}, 2427 {0xa980, 0xa983}, 2428 {0xa9b3, 0xa9c0}, 2429 {0xa9e5, 0xa9e5}, 2430 {0xaa29, 0xaa36}, 2431 {0xaa43, 0xaa43}, 2432 {0xaa4c, 0xaa4d}, 2433 {0xaa7b, 0xaa7d}, 2434 {0xaab0, 0xaab0}, 2435 {0xaab2, 0xaab4}, 2436 {0xaab7, 0xaab8}, 2437 {0xaabe, 0xaabf}, 2438 {0xaac1, 0xaac1}, 2439 {0xaaeb, 0xaaef}, 2440 {0xaaf5, 0xaaf6}, 2441 {0xabe3, 0xabea}, 2442 {0xabec, 0xabed}, 2443 {0xfb1e, 0xfb1e}, 2444 {0xfe00, 0xfe0f}, 2445 {0xfe20, 0xfe2f}, 2446 {0x101fd, 0x101fd}, 2447 {0x102e0, 0x102e0}, 2448 {0x10376, 0x1037a}, 2449 {0x10a01, 0x10a03}, 2450 {0x10a05, 0x10a06}, 2451 {0x10a0c, 0x10a0f}, 2452 {0x10a38, 0x10a3a}, 2453 {0x10a3f, 0x10a3f}, 2454 {0x10ae5, 0x10ae6}, 2455 {0x10d24, 0x10d27}, 2456 {0x10f46, 0x10f50}, 2457 {0x11000, 0x11002}, 2458 {0x11038, 0x11046}, 2459 {0x1107f, 0x11082}, 2460 {0x110b0, 0x110ba}, 2461 {0x11100, 0x11102}, 2462 {0x11127, 0x11134}, 2463 {0x11145, 0x11146}, 2464 {0x11173, 0x11173}, 2465 {0x11180, 0x11182}, 2466 {0x111b3, 0x111c0}, 2467 {0x111c9, 0x111cc}, 2468 {0x1122c, 0x11237}, 2469 {0x1123e, 0x1123e}, 2470 {0x112df, 0x112ea}, 2471 {0x11300, 0x11303}, 2472 {0x1133b, 0x1133c}, 2473 {0x1133e, 0x11344}, 2474 {0x11347, 0x11348}, 2475 {0x1134b, 0x1134d}, 2476 {0x11357, 0x11357}, 2477 {0x11362, 0x11363}, 2478 {0x11366, 0x1136c}, 2479 {0x11370, 0x11374}, 2480 {0x11435, 0x11446}, 2481 {0x1145e, 0x1145e}, 2482 {0x114b0, 0x114c3}, 2483 {0x115af, 0x115b5}, 2484 {0x115b8, 0x115c0}, 2485 {0x115dc, 0x115dd}, 2486 {0x11630, 0x11640}, 2487 {0x116ab, 0x116b7}, 2488 {0x1171d, 0x1172b}, 2489 {0x1182c, 0x1183a}, 2490 {0x119d1, 0x119d7}, 2491 {0x119da, 0x119e0}, 2492 {0x119e4, 0x119e4}, 2493 {0x11a01, 0x11a0a}, 2494 {0x11a33, 0x11a39}, 2495 {0x11a3b, 0x11a3e}, 2496 {0x11a47, 0x11a47}, 2497 {0x11a51, 0x11a5b}, 2498 {0x11a8a, 0x11a99}, 2499 {0x11c2f, 0x11c36}, 2500 {0x11c38, 0x11c3f}, 2501 {0x11c92, 0x11ca7}, 2502 {0x11ca9, 0x11cb6}, 2503 {0x11d31, 0x11d36}, 2504 {0x11d3a, 0x11d3a}, 2505 {0x11d3c, 0x11d3d}, 2506 {0x11d3f, 0x11d45}, 2507 {0x11d47, 0x11d47}, 2508 {0x11d8a, 0x11d8e}, 2509 {0x11d90, 0x11d91}, 2510 {0x11d93, 0x11d97}, 2511 {0x11ef3, 0x11ef6}, 2512 {0x16af0, 0x16af4}, 2513 {0x16b30, 0x16b36}, 2514 {0x16f4f, 0x16f4f}, 2515 {0x16f51, 0x16f87}, 2516 {0x16f8f, 0x16f92}, 2517 {0x1bc9d, 0x1bc9e}, 2518 {0x1d165, 0x1d169}, 2519 {0x1d16d, 0x1d172}, 2520 {0x1d17b, 0x1d182}, 2521 {0x1d185, 0x1d18b}, 2522 {0x1d1aa, 0x1d1ad}, 2523 {0x1d242, 0x1d244}, 2524 {0x1da00, 0x1da36}, 2525 {0x1da3b, 0x1da6c}, 2526 {0x1da75, 0x1da75}, 2527 {0x1da84, 0x1da84}, 2528 {0x1da9b, 0x1da9f}, 2529 {0x1daa1, 0x1daaf}, 2530 {0x1e000, 0x1e006}, 2531 {0x1e008, 0x1e018}, 2532 {0x1e01b, 0x1e021}, 2533 {0x1e023, 0x1e024}, 2534 {0x1e026, 0x1e02a}, 2535 {0x1e130, 0x1e136}, 2536 {0x1e2ec, 0x1e2ef}, 2537 {0x1e8d0, 0x1e8d6}, 2538 {0x1e944, 0x1e94a}, 2539 {0xe0100, 0xe01ef} 2540 }; 2541 2542 return intable(combining, sizeof(combining), c); 2543 } 2544 2545 /* 2546 * Return TRUE for characters that can be displayed in a normal way. 2547 * Only for characters of 0x100 and above! 2548 */ 2549 int 2550 utf_printable(int c) 2551 { 2552 #ifdef USE_WCHAR_FUNCTIONS 2553 /* 2554 * Assume the iswprint() library function works better than our own stuff. 2555 */ 2556 return iswprint(c); 2557 #else 2558 // Sorted list of non-overlapping intervals. 2559 // 0xd800-0xdfff is reserved for UTF-16, actually illegal. 2560 static struct interval nonprint[] = 2561 { 2562 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2563 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2564 {0xfffe, 0xffff} 2565 }; 2566 2567 return !intable(nonprint, sizeof(nonprint), c); 2568 #endif 2569 } 2570 2571 // Sorted list of non-overlapping intervals of all Emoji characters, 2572 // based on http://unicode.org/emoji/charts/emoji-list.html 2573 static struct interval emoji_all[] = 2574 { 2575 {0x203c, 0x203c}, 2576 {0x2049, 0x2049}, 2577 {0x2122, 0x2122}, 2578 {0x2139, 0x2139}, 2579 {0x2194, 0x2199}, 2580 {0x21a9, 0x21aa}, 2581 {0x231a, 0x231b}, 2582 {0x2328, 0x2328}, 2583 {0x23cf, 0x23cf}, 2584 {0x23e9, 0x23f3}, 2585 {0x23f8, 0x23fa}, 2586 {0x24c2, 0x24c2}, 2587 {0x25aa, 0x25ab}, 2588 {0x25b6, 0x25b6}, 2589 {0x25c0, 0x25c0}, 2590 {0x25fb, 0x25fe}, 2591 {0x2600, 0x2604}, 2592 {0x260e, 0x260e}, 2593 {0x2611, 0x2611}, 2594 {0x2614, 0x2615}, 2595 {0x2618, 0x2618}, 2596 {0x261d, 0x261d}, 2597 {0x2620, 0x2620}, 2598 {0x2622, 0x2623}, 2599 {0x2626, 0x2626}, 2600 {0x262a, 0x262a}, 2601 {0x262e, 0x262f}, 2602 {0x2638, 0x263a}, 2603 {0x2640, 0x2640}, 2604 {0x2642, 0x2642}, 2605 {0x2648, 0x2653}, 2606 {0x265f, 0x2660}, 2607 {0x2663, 0x2663}, 2608 {0x2665, 0x2666}, 2609 {0x2668, 0x2668}, 2610 {0x267b, 0x267b}, 2611 {0x267e, 0x267f}, 2612 {0x2692, 0x2697}, 2613 {0x2699, 0x2699}, 2614 {0x269b, 0x269c}, 2615 {0x26a0, 0x26a1}, 2616 {0x26aa, 0x26ab}, 2617 {0x26b0, 0x26b1}, 2618 {0x26bd, 0x26be}, 2619 {0x26c4, 0x26c5}, 2620 {0x26c8, 0x26c8}, 2621 {0x26ce, 0x26cf}, 2622 {0x26d1, 0x26d1}, 2623 {0x26d3, 0x26d4}, 2624 {0x26e9, 0x26ea}, 2625 {0x26f0, 0x26f5}, 2626 {0x26f7, 0x26fa}, 2627 {0x26fd, 0x26fd}, 2628 {0x2702, 0x2702}, 2629 {0x2705, 0x2705}, 2630 {0x2708, 0x270d}, 2631 {0x270f, 0x270f}, 2632 {0x2712, 0x2712}, 2633 {0x2714, 0x2714}, 2634 {0x2716, 0x2716}, 2635 {0x271d, 0x271d}, 2636 {0x2721, 0x2721}, 2637 {0x2728, 0x2728}, 2638 {0x2733, 0x2734}, 2639 {0x2744, 0x2744}, 2640 {0x2747, 0x2747}, 2641 {0x274c, 0x274c}, 2642 {0x274e, 0x274e}, 2643 {0x2753, 0x2755}, 2644 {0x2757, 0x2757}, 2645 {0x2763, 0x2764}, 2646 {0x2795, 0x2797}, 2647 {0x27a1, 0x27a1}, 2648 {0x27b0, 0x27b0}, 2649 {0x27bf, 0x27bf}, 2650 {0x2934, 0x2935}, 2651 {0x2b05, 0x2b07}, 2652 {0x2b1b, 0x2b1c}, 2653 {0x2b50, 0x2b50}, 2654 {0x2b55, 0x2b55}, 2655 {0x3030, 0x3030}, 2656 {0x303d, 0x303d}, 2657 {0x3297, 0x3297}, 2658 {0x3299, 0x3299}, 2659 {0x1f004, 0x1f004}, 2660 {0x1f0cf, 0x1f0cf}, 2661 {0x1f170, 0x1f171}, 2662 {0x1f17e, 0x1f17f}, 2663 {0x1f18e, 0x1f18e}, 2664 {0x1f191, 0x1f19a}, 2665 {0x1f1e6, 0x1f1ff}, 2666 {0x1f201, 0x1f202}, 2667 {0x1f21a, 0x1f21a}, 2668 {0x1f22f, 0x1f22f}, 2669 {0x1f232, 0x1f23a}, 2670 {0x1f250, 0x1f251}, 2671 {0x1f300, 0x1f321}, 2672 {0x1f324, 0x1f393}, 2673 {0x1f396, 0x1f397}, 2674 {0x1f399, 0x1f39b}, 2675 {0x1f39e, 0x1f3f0}, 2676 {0x1f3f3, 0x1f3f5}, 2677 {0x1f3f7, 0x1f4fd}, 2678 {0x1f4ff, 0x1f53d}, 2679 {0x1f549, 0x1f54e}, 2680 {0x1f550, 0x1f567}, 2681 {0x1f56f, 0x1f570}, 2682 {0x1f573, 0x1f57a}, 2683 {0x1f587, 0x1f587}, 2684 {0x1f58a, 0x1f58d}, 2685 {0x1f590, 0x1f590}, 2686 {0x1f595, 0x1f596}, 2687 {0x1f5a4, 0x1f5a5}, 2688 {0x1f5a8, 0x1f5a8}, 2689 {0x1f5b1, 0x1f5b2}, 2690 {0x1f5bc, 0x1f5bc}, 2691 {0x1f5c2, 0x1f5c4}, 2692 {0x1f5d1, 0x1f5d3}, 2693 {0x1f5dc, 0x1f5de}, 2694 {0x1f5e1, 0x1f5e1}, 2695 {0x1f5e3, 0x1f5e3}, 2696 {0x1f5e8, 0x1f5e8}, 2697 {0x1f5ef, 0x1f5ef}, 2698 {0x1f5f3, 0x1f5f3}, 2699 {0x1f5fa, 0x1f64f}, 2700 {0x1f680, 0x1f6c5}, 2701 {0x1f6cb, 0x1f6d2}, 2702 {0x1f6e0, 0x1f6e5}, 2703 {0x1f6e9, 0x1f6e9}, 2704 {0x1f6eb, 0x1f6ec}, 2705 {0x1f6f0, 0x1f6f0}, 2706 {0x1f6f3, 0x1f6f9}, 2707 {0x1f910, 0x1f93a}, 2708 {0x1f93c, 0x1f93e}, 2709 {0x1f940, 0x1f945}, 2710 {0x1f947, 0x1f970}, 2711 {0x1f973, 0x1f976}, 2712 {0x1f97a, 0x1f97a}, 2713 {0x1f97c, 0x1f9a2}, 2714 {0x1f9b0, 0x1f9b9}, 2715 {0x1f9c0, 0x1f9c2}, 2716 {0x1f9d0, 0x1f9ff} 2717 }; 2718 2719 /* 2720 * Get class of a Unicode character. 2721 * 0: white space 2722 * 1: punctuation 2723 * 2 or bigger: some class of word character. 2724 */ 2725 int 2726 utf_class(int c) 2727 { 2728 return utf_class_buf(c, curbuf); 2729 } 2730 2731 int 2732 utf_class_buf(int c, buf_T *buf) 2733 { 2734 // sorted list of non-overlapping intervals 2735 static struct clinterval 2736 { 2737 unsigned int first; 2738 unsigned int last; 2739 unsigned int class; 2740 } classes[] = 2741 { 2742 {0x037e, 0x037e, 1}, // Greek question mark 2743 {0x0387, 0x0387, 1}, // Greek ano teleia 2744 {0x055a, 0x055f, 1}, // Armenian punctuation 2745 {0x0589, 0x0589, 1}, // Armenian full stop 2746 {0x05be, 0x05be, 1}, 2747 {0x05c0, 0x05c0, 1}, 2748 {0x05c3, 0x05c3, 1}, 2749 {0x05f3, 0x05f4, 1}, 2750 {0x060c, 0x060c, 1}, 2751 {0x061b, 0x061b, 1}, 2752 {0x061f, 0x061f, 1}, 2753 {0x066a, 0x066d, 1}, 2754 {0x06d4, 0x06d4, 1}, 2755 {0x0700, 0x070d, 1}, // Syriac punctuation 2756 {0x0964, 0x0965, 1}, 2757 {0x0970, 0x0970, 1}, 2758 {0x0df4, 0x0df4, 1}, 2759 {0x0e4f, 0x0e4f, 1}, 2760 {0x0e5a, 0x0e5b, 1}, 2761 {0x0f04, 0x0f12, 1}, 2762 {0x0f3a, 0x0f3d, 1}, 2763 {0x0f85, 0x0f85, 1}, 2764 {0x104a, 0x104f, 1}, // Myanmar punctuation 2765 {0x10fb, 0x10fb, 1}, // Georgian punctuation 2766 {0x1361, 0x1368, 1}, // Ethiopic punctuation 2767 {0x166d, 0x166e, 1}, // Canadian Syl. punctuation 2768 {0x1680, 0x1680, 0}, 2769 {0x169b, 0x169c, 1}, 2770 {0x16eb, 0x16ed, 1}, 2771 {0x1735, 0x1736, 1}, 2772 {0x17d4, 0x17dc, 1}, // Khmer punctuation 2773 {0x1800, 0x180a, 1}, // Mongolian punctuation 2774 {0x2000, 0x200b, 0}, // spaces 2775 {0x200c, 0x2027, 1}, // punctuation and symbols 2776 {0x2028, 0x2029, 0}, 2777 {0x202a, 0x202e, 1}, // punctuation and symbols 2778 {0x202f, 0x202f, 0}, 2779 {0x2030, 0x205e, 1}, // punctuation and symbols 2780 {0x205f, 0x205f, 0}, 2781 {0x2060, 0x27ff, 1}, // punctuation and symbols 2782 {0x2070, 0x207f, 0x2070}, // superscript 2783 {0x2080, 0x2094, 0x2080}, // subscript 2784 {0x20a0, 0x27ff, 1}, // all kinds of symbols 2785 {0x2800, 0x28ff, 0x2800}, // braille 2786 {0x2900, 0x2998, 1}, // arrows, brackets, etc. 2787 {0x29d8, 0x29db, 1}, 2788 {0x29fc, 0x29fd, 1}, 2789 {0x2e00, 0x2e7f, 1}, // supplemental punctuation 2790 {0x3000, 0x3000, 0}, // ideographic space 2791 {0x3001, 0x3020, 1}, // ideographic punctuation 2792 {0x3030, 0x3030, 1}, 2793 {0x303d, 0x303d, 1}, 2794 {0x3040, 0x309f, 0x3040}, // Hiragana 2795 {0x30a0, 0x30ff, 0x30a0}, // Katakana 2796 {0x3300, 0x9fff, 0x4e00}, // CJK Ideographs 2797 {0xac00, 0xd7a3, 0xac00}, // Hangul Syllables 2798 {0xf900, 0xfaff, 0x4e00}, // CJK Ideographs 2799 {0xfd3e, 0xfd3f, 1}, 2800 {0xfe30, 0xfe6b, 1}, // punctuation forms 2801 {0xff00, 0xff0f, 1}, // half/fullwidth ASCII 2802 {0xff1a, 0xff20, 1}, // half/fullwidth ASCII 2803 {0xff3b, 0xff40, 1}, // half/fullwidth ASCII 2804 {0xff5b, 0xff65, 1}, // half/fullwidth ASCII 2805 {0x1d000, 0x1d24f, 1}, // Musical notation 2806 {0x1d400, 0x1d7ff, 1}, // Mathematical Alphanumeric Symbols 2807 {0x1f000, 0x1f2ff, 1}, // Game pieces; enclosed characters 2808 {0x1f300, 0x1f9ff, 1}, // Many symbol blocks 2809 {0x20000, 0x2a6df, 0x4e00}, // CJK Ideographs 2810 {0x2a700, 0x2b73f, 0x4e00}, // CJK Ideographs 2811 {0x2b740, 0x2b81f, 0x4e00}, // CJK Ideographs 2812 {0x2f800, 0x2fa1f, 0x4e00}, // CJK Ideographs 2813 }; 2814 2815 int bot = 0; 2816 int top = sizeof(classes) / sizeof(struct clinterval) - 1; 2817 int mid; 2818 2819 // First quick check for Latin1 characters, use 'iskeyword'. 2820 if (c < 0x100) 2821 { 2822 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2823 return 0; // blank 2824 if (vim_iswordc_buf(c, buf)) 2825 return 2; // word character 2826 return 1; // punctuation 2827 } 2828 2829 // binary search in table 2830 while (top >= bot) 2831 { 2832 mid = (bot + top) / 2; 2833 if (classes[mid].last < (unsigned int)c) 2834 bot = mid + 1; 2835 else if (classes[mid].first > (unsigned int)c) 2836 top = mid - 1; 2837 else 2838 return (int)classes[mid].class; 2839 } 2840 2841 // emoji 2842 if (intable(emoji_all, sizeof(emoji_all), c)) 2843 return 3; 2844 2845 // most other characters are "word" characters 2846 return 2; 2847 } 2848 2849 int 2850 utf_ambiguous_width(int c) 2851 { 2852 return c >= 0x80 && (intable(ambiguous, sizeof(ambiguous), c) 2853 || intable(emoji_all, sizeof(emoji_all), c)); 2854 } 2855 2856 /* 2857 * Code for Unicode case-dependent operations. Based on notes in 2858 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2859 * This code uses simple case folding, not full case folding. 2860 * Last updated for Unicode 5.2. 2861 */ 2862 2863 /* 2864 * The following tables are built by ../runtime/tools/unicode.vim. 2865 * They must be in numeric order, because we use binary search. 2866 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2867 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2868 * folded/upper/lower by adding 32. 2869 */ 2870 typedef struct 2871 { 2872 int rangeStart; 2873 int rangeEnd; 2874 int step; 2875 int offset; 2876 } convertStruct; 2877 2878 static convertStruct foldCase[] = 2879 { 2880 {0x41,0x5a,1,32}, 2881 {0xb5,0xb5,-1,775}, 2882 {0xc0,0xd6,1,32}, 2883 {0xd8,0xde,1,32}, 2884 {0x100,0x12e,2,1}, 2885 {0x132,0x136,2,1}, 2886 {0x139,0x147,2,1}, 2887 {0x14a,0x176,2,1}, 2888 {0x178,0x178,-1,-121}, 2889 {0x179,0x17d,2,1}, 2890 {0x17f,0x17f,-1,-268}, 2891 {0x181,0x181,-1,210}, 2892 {0x182,0x184,2,1}, 2893 {0x186,0x186,-1,206}, 2894 {0x187,0x187,-1,1}, 2895 {0x189,0x18a,1,205}, 2896 {0x18b,0x18b,-1,1}, 2897 {0x18e,0x18e,-1,79}, 2898 {0x18f,0x18f,-1,202}, 2899 {0x190,0x190,-1,203}, 2900 {0x191,0x191,-1,1}, 2901 {0x193,0x193,-1,205}, 2902 {0x194,0x194,-1,207}, 2903 {0x196,0x196,-1,211}, 2904 {0x197,0x197,-1,209}, 2905 {0x198,0x198,-1,1}, 2906 {0x19c,0x19c,-1,211}, 2907 {0x19d,0x19d,-1,213}, 2908 {0x19f,0x19f,-1,214}, 2909 {0x1a0,0x1a4,2,1}, 2910 {0x1a6,0x1a6,-1,218}, 2911 {0x1a7,0x1a7,-1,1}, 2912 {0x1a9,0x1a9,-1,218}, 2913 {0x1ac,0x1ac,-1,1}, 2914 {0x1ae,0x1ae,-1,218}, 2915 {0x1af,0x1af,-1,1}, 2916 {0x1b1,0x1b2,1,217}, 2917 {0x1b3,0x1b5,2,1}, 2918 {0x1b7,0x1b7,-1,219}, 2919 {0x1b8,0x1bc,4,1}, 2920 {0x1c4,0x1c4,-1,2}, 2921 {0x1c5,0x1c5,-1,1}, 2922 {0x1c7,0x1c7,-1,2}, 2923 {0x1c8,0x1c8,-1,1}, 2924 {0x1ca,0x1ca,-1,2}, 2925 {0x1cb,0x1db,2,1}, 2926 {0x1de,0x1ee,2,1}, 2927 {0x1f1,0x1f1,-1,2}, 2928 {0x1f2,0x1f4,2,1}, 2929 {0x1f6,0x1f6,-1,-97}, 2930 {0x1f7,0x1f7,-1,-56}, 2931 {0x1f8,0x21e,2,1}, 2932 {0x220,0x220,-1,-130}, 2933 {0x222,0x232,2,1}, 2934 {0x23a,0x23a,-1,10795}, 2935 {0x23b,0x23b,-1,1}, 2936 {0x23d,0x23d,-1,-163}, 2937 {0x23e,0x23e,-1,10792}, 2938 {0x241,0x241,-1,1}, 2939 {0x243,0x243,-1,-195}, 2940 {0x244,0x244,-1,69}, 2941 {0x245,0x245,-1,71}, 2942 {0x246,0x24e,2,1}, 2943 {0x345,0x345,-1,116}, 2944 {0x370,0x372,2,1}, 2945 {0x376,0x376,-1,1}, 2946 {0x37f,0x37f,-1,116}, 2947 {0x386,0x386,-1,38}, 2948 {0x388,0x38a,1,37}, 2949 {0x38c,0x38c,-1,64}, 2950 {0x38e,0x38f,1,63}, 2951 {0x391,0x3a1,1,32}, 2952 {0x3a3,0x3ab,1,32}, 2953 {0x3c2,0x3c2,-1,1}, 2954 {0x3cf,0x3cf,-1,8}, 2955 {0x3d0,0x3d0,-1,-30}, 2956 {0x3d1,0x3d1,-1,-25}, 2957 {0x3d5,0x3d5,-1,-15}, 2958 {0x3d6,0x3d6,-1,-22}, 2959 {0x3d8,0x3ee,2,1}, 2960 {0x3f0,0x3f0,-1,-54}, 2961 {0x3f1,0x3f1,-1,-48}, 2962 {0x3f4,0x3f4,-1,-60}, 2963 {0x3f5,0x3f5,-1,-64}, 2964 {0x3f7,0x3f7,-1,1}, 2965 {0x3f9,0x3f9,-1,-7}, 2966 {0x3fa,0x3fa,-1,1}, 2967 {0x3fd,0x3ff,1,-130}, 2968 {0x400,0x40f,1,80}, 2969 {0x410,0x42f,1,32}, 2970 {0x460,0x480,2,1}, 2971 {0x48a,0x4be,2,1}, 2972 {0x4c0,0x4c0,-1,15}, 2973 {0x4c1,0x4cd,2,1}, 2974 {0x4d0,0x52e,2,1}, 2975 {0x531,0x556,1,48}, 2976 {0x10a0,0x10c5,1,7264}, 2977 {0x10c7,0x10cd,6,7264}, 2978 {0x13f8,0x13fd,1,-8}, 2979 {0x1c80,0x1c80,-1,-6222}, 2980 {0x1c81,0x1c81,-1,-6221}, 2981 {0x1c82,0x1c82,-1,-6212}, 2982 {0x1c83,0x1c84,1,-6210}, 2983 {0x1c85,0x1c85,-1,-6211}, 2984 {0x1c86,0x1c86,-1,-6204}, 2985 {0x1c87,0x1c87,-1,-6180}, 2986 {0x1c88,0x1c88,-1,35267}, 2987 {0x1c90,0x1cba,1,-3008}, 2988 {0x1cbd,0x1cbf,1,-3008}, 2989 {0x1e00,0x1e94,2,1}, 2990 {0x1e9b,0x1e9b,-1,-58}, 2991 {0x1e9e,0x1e9e,-1,-7615}, 2992 {0x1ea0,0x1efe,2,1}, 2993 {0x1f08,0x1f0f,1,-8}, 2994 {0x1f18,0x1f1d,1,-8}, 2995 {0x1f28,0x1f2f,1,-8}, 2996 {0x1f38,0x1f3f,1,-8}, 2997 {0x1f48,0x1f4d,1,-8}, 2998 {0x1f59,0x1f5f,2,-8}, 2999 {0x1f68,0x1f6f,1,-8}, 3000 {0x1f88,0x1f8f,1,-8}, 3001 {0x1f98,0x1f9f,1,-8}, 3002 {0x1fa8,0x1faf,1,-8}, 3003 {0x1fb8,0x1fb9,1,-8}, 3004 {0x1fba,0x1fbb,1,-74}, 3005 {0x1fbc,0x1fbc,-1,-9}, 3006 {0x1fbe,0x1fbe,-1,-7173}, 3007 {0x1fc8,0x1fcb,1,-86}, 3008 {0x1fcc,0x1fcc,-1,-9}, 3009 {0x1fd8,0x1fd9,1,-8}, 3010 {0x1fda,0x1fdb,1,-100}, 3011 {0x1fe8,0x1fe9,1,-8}, 3012 {0x1fea,0x1feb,1,-112}, 3013 {0x1fec,0x1fec,-1,-7}, 3014 {0x1ff8,0x1ff9,1,-128}, 3015 {0x1ffa,0x1ffb,1,-126}, 3016 {0x1ffc,0x1ffc,-1,-9}, 3017 {0x2126,0x2126,-1,-7517}, 3018 {0x212a,0x212a,-1,-8383}, 3019 {0x212b,0x212b,-1,-8262}, 3020 {0x2132,0x2132,-1,28}, 3021 {0x2160,0x216f,1,16}, 3022 {0x2183,0x2183,-1,1}, 3023 {0x24b6,0x24cf,1,26}, 3024 {0x2c00,0x2c2e,1,48}, 3025 {0x2c60,0x2c60,-1,1}, 3026 {0x2c62,0x2c62,-1,-10743}, 3027 {0x2c63,0x2c63,-1,-3814}, 3028 {0x2c64,0x2c64,-1,-10727}, 3029 {0x2c67,0x2c6b,2,1}, 3030 {0x2c6d,0x2c6d,-1,-10780}, 3031 {0x2c6e,0x2c6e,-1,-10749}, 3032 {0x2c6f,0x2c6f,-1,-10783}, 3033 {0x2c70,0x2c70,-1,-10782}, 3034 {0x2c72,0x2c75,3,1}, 3035 {0x2c7e,0x2c7f,1,-10815}, 3036 {0x2c80,0x2ce2,2,1}, 3037 {0x2ceb,0x2ced,2,1}, 3038 {0x2cf2,0xa640,31054,1}, 3039 {0xa642,0xa66c,2,1}, 3040 {0xa680,0xa69a,2,1}, 3041 {0xa722,0xa72e,2,1}, 3042 {0xa732,0xa76e,2,1}, 3043 {0xa779,0xa77b,2,1}, 3044 {0xa77d,0xa77d,-1,-35332}, 3045 {0xa77e,0xa786,2,1}, 3046 {0xa78b,0xa78b,-1,1}, 3047 {0xa78d,0xa78d,-1,-42280}, 3048 {0xa790,0xa792,2,1}, 3049 {0xa796,0xa7a8,2,1}, 3050 {0xa7aa,0xa7aa,-1,-42308}, 3051 {0xa7ab,0xa7ab,-1,-42319}, 3052 {0xa7ac,0xa7ac,-1,-42315}, 3053 {0xa7ad,0xa7ad,-1,-42305}, 3054 {0xa7ae,0xa7ae,-1,-42308}, 3055 {0xa7b0,0xa7b0,-1,-42258}, 3056 {0xa7b1,0xa7b1,-1,-42282}, 3057 {0xa7b2,0xa7b2,-1,-42261}, 3058 {0xa7b3,0xa7b3,-1,928}, 3059 {0xa7b4,0xa7be,2,1}, 3060 {0xa7c2,0xa7c2,-1,1}, 3061 {0xa7c4,0xa7c4,-1,-48}, 3062 {0xa7c5,0xa7c5,-1,-42307}, 3063 {0xa7c6,0xa7c6,-1,-35384}, 3064 {0xab70,0xabbf,1,-38864}, 3065 {0xff21,0xff3a,1,32}, 3066 {0x10400,0x10427,1,40}, 3067 {0x104b0,0x104d3,1,40}, 3068 {0x10c80,0x10cb2,1,64}, 3069 {0x118a0,0x118bf,1,32}, 3070 {0x16e40,0x16e5f,1,32}, 3071 {0x1e900,0x1e921,1,34} 3072 }; 3073 3074 /* 3075 * Generic conversion function for case operations. 3076 * Return the converted equivalent of "a", which is a UCS-4 character. Use 3077 * the given conversion "table". Uses binary search on "table". 3078 */ 3079 static int 3080 utf_convert( 3081 int a, 3082 convertStruct table[], 3083 int tableSize) 3084 { 3085 int start, mid, end; // indices into table 3086 int entries = tableSize / sizeof(convertStruct); 3087 3088 start = 0; 3089 end = entries; 3090 while (start < end) 3091 { 3092 // need to search further 3093 mid = (end + start) / 2; 3094 if (table[mid].rangeEnd < a) 3095 start = mid + 1; 3096 else 3097 end = mid; 3098 } 3099 if (start < entries 3100 && table[start].rangeStart <= a 3101 && a <= table[start].rangeEnd 3102 && (a - table[start].rangeStart) % table[start].step == 0) 3103 return (a + table[start].offset); 3104 else 3105 return a; 3106 } 3107 3108 /* 3109 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 3110 * simple case folding. 3111 */ 3112 int 3113 utf_fold(int a) 3114 { 3115 if (a < 0x80) 3116 // be fast for ASCII 3117 return a >= 0x41 && a <= 0x5a ? a + 32 : a; 3118 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 3119 } 3120 3121 static convertStruct toLower[] = 3122 { 3123 {0x41,0x5a,1,32}, 3124 {0xc0,0xd6,1,32}, 3125 {0xd8,0xde,1,32}, 3126 {0x100,0x12e,2,1}, 3127 {0x130,0x130,-1,-199}, 3128 {0x132,0x136,2,1}, 3129 {0x139,0x147,2,1}, 3130 {0x14a,0x176,2,1}, 3131 {0x178,0x178,-1,-121}, 3132 {0x179,0x17d,2,1}, 3133 {0x181,0x181,-1,210}, 3134 {0x182,0x184,2,1}, 3135 {0x186,0x186,-1,206}, 3136 {0x187,0x187,-1,1}, 3137 {0x189,0x18a,1,205}, 3138 {0x18b,0x18b,-1,1}, 3139 {0x18e,0x18e,-1,79}, 3140 {0x18f,0x18f,-1,202}, 3141 {0x190,0x190,-1,203}, 3142 {0x191,0x191,-1,1}, 3143 {0x193,0x193,-1,205}, 3144 {0x194,0x194,-1,207}, 3145 {0x196,0x196,-1,211}, 3146 {0x197,0x197,-1,209}, 3147 {0x198,0x198,-1,1}, 3148 {0x19c,0x19c,-1,211}, 3149 {0x19d,0x19d,-1,213}, 3150 {0x19f,0x19f,-1,214}, 3151 {0x1a0,0x1a4,2,1}, 3152 {0x1a6,0x1a6,-1,218}, 3153 {0x1a7,0x1a7,-1,1}, 3154 {0x1a9,0x1a9,-1,218}, 3155 {0x1ac,0x1ac,-1,1}, 3156 {0x1ae,0x1ae,-1,218}, 3157 {0x1af,0x1af,-1,1}, 3158 {0x1b1,0x1b2,1,217}, 3159 {0x1b3,0x1b5,2,1}, 3160 {0x1b7,0x1b7,-1,219}, 3161 {0x1b8,0x1bc,4,1}, 3162 {0x1c4,0x1c4,-1,2}, 3163 {0x1c5,0x1c5,-1,1}, 3164 {0x1c7,0x1c7,-1,2}, 3165 {0x1c8,0x1c8,-1,1}, 3166 {0x1ca,0x1ca,-1,2}, 3167 {0x1cb,0x1db,2,1}, 3168 {0x1de,0x1ee,2,1}, 3169 {0x1f1,0x1f1,-1,2}, 3170 {0x1f2,0x1f4,2,1}, 3171 {0x1f6,0x1f6,-1,-97}, 3172 {0x1f7,0x1f7,-1,-56}, 3173 {0x1f8,0x21e,2,1}, 3174 {0x220,0x220,-1,-130}, 3175 {0x222,0x232,2,1}, 3176 {0x23a,0x23a,-1,10795}, 3177 {0x23b,0x23b,-1,1}, 3178 {0x23d,0x23d,-1,-163}, 3179 {0x23e,0x23e,-1,10792}, 3180 {0x241,0x241,-1,1}, 3181 {0x243,0x243,-1,-195}, 3182 {0x244,0x244,-1,69}, 3183 {0x245,0x245,-1,71}, 3184 {0x246,0x24e,2,1}, 3185 {0x370,0x372,2,1}, 3186 {0x376,0x376,-1,1}, 3187 {0x37f,0x37f,-1,116}, 3188 {0x386,0x386,-1,38}, 3189 {0x388,0x38a,1,37}, 3190 {0x38c,0x38c,-1,64}, 3191 {0x38e,0x38f,1,63}, 3192 {0x391,0x3a1,1,32}, 3193 {0x3a3,0x3ab,1,32}, 3194 {0x3cf,0x3cf,-1,8}, 3195 {0x3d8,0x3ee,2,1}, 3196 {0x3f4,0x3f4,-1,-60}, 3197 {0x3f7,0x3f7,-1,1}, 3198 {0x3f9,0x3f9,-1,-7}, 3199 {0x3fa,0x3fa,-1,1}, 3200 {0x3fd,0x3ff,1,-130}, 3201 {0x400,0x40f,1,80}, 3202 {0x410,0x42f,1,32}, 3203 {0x460,0x480,2,1}, 3204 {0x48a,0x4be,2,1}, 3205 {0x4c0,0x4c0,-1,15}, 3206 {0x4c1,0x4cd,2,1}, 3207 {0x4d0,0x52e,2,1}, 3208 {0x531,0x556,1,48}, 3209 {0x10a0,0x10c5,1,7264}, 3210 {0x10c7,0x10cd,6,7264}, 3211 {0x13a0,0x13ef,1,38864}, 3212 {0x13f0,0x13f5,1,8}, 3213 {0x1c90,0x1cba,1,-3008}, 3214 {0x1cbd,0x1cbf,1,-3008}, 3215 {0x1e00,0x1e94,2,1}, 3216 {0x1e9e,0x1e9e,-1,-7615}, 3217 {0x1ea0,0x1efe,2,1}, 3218 {0x1f08,0x1f0f,1,-8}, 3219 {0x1f18,0x1f1d,1,-8}, 3220 {0x1f28,0x1f2f,1,-8}, 3221 {0x1f38,0x1f3f,1,-8}, 3222 {0x1f48,0x1f4d,1,-8}, 3223 {0x1f59,0x1f5f,2,-8}, 3224 {0x1f68,0x1f6f,1,-8}, 3225 {0x1f88,0x1f8f,1,-8}, 3226 {0x1f98,0x1f9f,1,-8}, 3227 {0x1fa8,0x1faf,1,-8}, 3228 {0x1fb8,0x1fb9,1,-8}, 3229 {0x1fba,0x1fbb,1,-74}, 3230 {0x1fbc,0x1fbc,-1,-9}, 3231 {0x1fc8,0x1fcb,1,-86}, 3232 {0x1fcc,0x1fcc,-1,-9}, 3233 {0x1fd8,0x1fd9,1,-8}, 3234 {0x1fda,0x1fdb,1,-100}, 3235 {0x1fe8,0x1fe9,1,-8}, 3236 {0x1fea,0x1feb,1,-112}, 3237 {0x1fec,0x1fec,-1,-7}, 3238 {0x1ff8,0x1ff9,1,-128}, 3239 {0x1ffa,0x1ffb,1,-126}, 3240 {0x1ffc,0x1ffc,-1,-9}, 3241 {0x2126,0x2126,-1,-7517}, 3242 {0x212a,0x212a,-1,-8383}, 3243 {0x212b,0x212b,-1,-8262}, 3244 {0x2132,0x2132,-1,28}, 3245 {0x2160,0x216f,1,16}, 3246 {0x2183,0x2183,-1,1}, 3247 {0x24b6,0x24cf,1,26}, 3248 {0x2c00,0x2c2e,1,48}, 3249 {0x2c60,0x2c60,-1,1}, 3250 {0x2c62,0x2c62,-1,-10743}, 3251 {0x2c63,0x2c63,-1,-3814}, 3252 {0x2c64,0x2c64,-1,-10727}, 3253 {0x2c67,0x2c6b,2,1}, 3254 {0x2c6d,0x2c6d,-1,-10780}, 3255 {0x2c6e,0x2c6e,-1,-10749}, 3256 {0x2c6f,0x2c6f,-1,-10783}, 3257 {0x2c70,0x2c70,-1,-10782}, 3258 {0x2c72,0x2c75,3,1}, 3259 {0x2c7e,0x2c7f,1,-10815}, 3260 {0x2c80,0x2ce2,2,1}, 3261 {0x2ceb,0x2ced,2,1}, 3262 {0x2cf2,0xa640,31054,1}, 3263 {0xa642,0xa66c,2,1}, 3264 {0xa680,0xa69a,2,1}, 3265 {0xa722,0xa72e,2,1}, 3266 {0xa732,0xa76e,2,1}, 3267 {0xa779,0xa77b,2,1}, 3268 {0xa77d,0xa77d,-1,-35332}, 3269 {0xa77e,0xa786,2,1}, 3270 {0xa78b,0xa78b,-1,1}, 3271 {0xa78d,0xa78d,-1,-42280}, 3272 {0xa790,0xa792,2,1}, 3273 {0xa796,0xa7a8,2,1}, 3274 {0xa7aa,0xa7aa,-1,-42308}, 3275 {0xa7ab,0xa7ab,-1,-42319}, 3276 {0xa7ac,0xa7ac,-1,-42315}, 3277 {0xa7ad,0xa7ad,-1,-42305}, 3278 {0xa7ae,0xa7ae,-1,-42308}, 3279 {0xa7b0,0xa7b0,-1,-42258}, 3280 {0xa7b1,0xa7b1,-1,-42282}, 3281 {0xa7b2,0xa7b2,-1,-42261}, 3282 {0xa7b3,0xa7b3,-1,928}, 3283 {0xa7b4,0xa7be,2,1}, 3284 {0xa7c2,0xa7c2,-1,1}, 3285 {0xa7c4,0xa7c4,-1,-48}, 3286 {0xa7c5,0xa7c5,-1,-42307}, 3287 {0xa7c6,0xa7c6,-1,-35384}, 3288 {0xff21,0xff3a,1,32}, 3289 {0x10400,0x10427,1,40}, 3290 {0x104b0,0x104d3,1,40}, 3291 {0x10c80,0x10cb2,1,64}, 3292 {0x118a0,0x118bf,1,32}, 3293 {0x16e40,0x16e5f,1,32}, 3294 {0x1e900,0x1e921,1,34} 3295 }; 3296 3297 static convertStruct toUpper[] = 3298 { 3299 {0x61,0x7a,1,-32}, 3300 {0xb5,0xb5,-1,743}, 3301 {0xe0,0xf6,1,-32}, 3302 {0xf8,0xfe,1,-32}, 3303 {0xff,0xff,-1,121}, 3304 {0x101,0x12f,2,-1}, 3305 {0x131,0x131,-1,-232}, 3306 {0x133,0x137,2,-1}, 3307 {0x13a,0x148,2,-1}, 3308 {0x14b,0x177,2,-1}, 3309 {0x17a,0x17e,2,-1}, 3310 {0x17f,0x17f,-1,-300}, 3311 {0x180,0x180,-1,195}, 3312 {0x183,0x185,2,-1}, 3313 {0x188,0x18c,4,-1}, 3314 {0x192,0x192,-1,-1}, 3315 {0x195,0x195,-1,97}, 3316 {0x199,0x199,-1,-1}, 3317 {0x19a,0x19a,-1,163}, 3318 {0x19e,0x19e,-1,130}, 3319 {0x1a1,0x1a5,2,-1}, 3320 {0x1a8,0x1ad,5,-1}, 3321 {0x1b0,0x1b4,4,-1}, 3322 {0x1b6,0x1b9,3,-1}, 3323 {0x1bd,0x1bd,-1,-1}, 3324 {0x1bf,0x1bf,-1,56}, 3325 {0x1c5,0x1c5,-1,-1}, 3326 {0x1c6,0x1c6,-1,-2}, 3327 {0x1c8,0x1c8,-1,-1}, 3328 {0x1c9,0x1c9,-1,-2}, 3329 {0x1cb,0x1cb,-1,-1}, 3330 {0x1cc,0x1cc,-1,-2}, 3331 {0x1ce,0x1dc,2,-1}, 3332 {0x1dd,0x1dd,-1,-79}, 3333 {0x1df,0x1ef,2,-1}, 3334 {0x1f2,0x1f2,-1,-1}, 3335 {0x1f3,0x1f3,-1,-2}, 3336 {0x1f5,0x1f9,4,-1}, 3337 {0x1fb,0x21f,2,-1}, 3338 {0x223,0x233,2,-1}, 3339 {0x23c,0x23c,-1,-1}, 3340 {0x23f,0x240,1,10815}, 3341 {0x242,0x247,5,-1}, 3342 {0x249,0x24f,2,-1}, 3343 {0x250,0x250,-1,10783}, 3344 {0x251,0x251,-1,10780}, 3345 {0x252,0x252,-1,10782}, 3346 {0x253,0x253,-1,-210}, 3347 {0x254,0x254,-1,-206}, 3348 {0x256,0x257,1,-205}, 3349 {0x259,0x259,-1,-202}, 3350 {0x25b,0x25b,-1,-203}, 3351 {0x25c,0x25c,-1,42319}, 3352 {0x260,0x260,-1,-205}, 3353 {0x261,0x261,-1,42315}, 3354 {0x263,0x263,-1,-207}, 3355 {0x265,0x265,-1,42280}, 3356 {0x266,0x266,-1,42308}, 3357 {0x268,0x268,-1,-209}, 3358 {0x269,0x269,-1,-211}, 3359 {0x26a,0x26a,-1,42308}, 3360 {0x26b,0x26b,-1,10743}, 3361 {0x26c,0x26c,-1,42305}, 3362 {0x26f,0x26f,-1,-211}, 3363 {0x271,0x271,-1,10749}, 3364 {0x272,0x272,-1,-213}, 3365 {0x275,0x275,-1,-214}, 3366 {0x27d,0x27d,-1,10727}, 3367 {0x280,0x280,-1,-218}, 3368 {0x282,0x282,-1,42307}, 3369 {0x283,0x283,-1,-218}, 3370 {0x287,0x287,-1,42282}, 3371 {0x288,0x288,-1,-218}, 3372 {0x289,0x289,-1,-69}, 3373 {0x28a,0x28b,1,-217}, 3374 {0x28c,0x28c,-1,-71}, 3375 {0x292,0x292,-1,-219}, 3376 {0x29d,0x29d,-1,42261}, 3377 {0x29e,0x29e,-1,42258}, 3378 {0x345,0x345,-1,84}, 3379 {0x371,0x373,2,-1}, 3380 {0x377,0x377,-1,-1}, 3381 {0x37b,0x37d,1,130}, 3382 {0x3ac,0x3ac,-1,-38}, 3383 {0x3ad,0x3af,1,-37}, 3384 {0x3b1,0x3c1,1,-32}, 3385 {0x3c2,0x3c2,-1,-31}, 3386 {0x3c3,0x3cb,1,-32}, 3387 {0x3cc,0x3cc,-1,-64}, 3388 {0x3cd,0x3ce,1,-63}, 3389 {0x3d0,0x3d0,-1,-62}, 3390 {0x3d1,0x3d1,-1,-57}, 3391 {0x3d5,0x3d5,-1,-47}, 3392 {0x3d6,0x3d6,-1,-54}, 3393 {0x3d7,0x3d7,-1,-8}, 3394 {0x3d9,0x3ef,2,-1}, 3395 {0x3f0,0x3f0,-1,-86}, 3396 {0x3f1,0x3f1,-1,-80}, 3397 {0x3f2,0x3f2,-1,7}, 3398 {0x3f3,0x3f3,-1,-116}, 3399 {0x3f5,0x3f5,-1,-96}, 3400 {0x3f8,0x3fb,3,-1}, 3401 {0x430,0x44f,1,-32}, 3402 {0x450,0x45f,1,-80}, 3403 {0x461,0x481,2,-1}, 3404 {0x48b,0x4bf,2,-1}, 3405 {0x4c2,0x4ce,2,-1}, 3406 {0x4cf,0x4cf,-1,-15}, 3407 {0x4d1,0x52f,2,-1}, 3408 {0x561,0x586,1,-48}, 3409 {0x10d0,0x10fa,1,3008}, 3410 {0x10fd,0x10ff,1,3008}, 3411 {0x13f8,0x13fd,1,-8}, 3412 {0x1c80,0x1c80,-1,-6254}, 3413 {0x1c81,0x1c81,-1,-6253}, 3414 {0x1c82,0x1c82,-1,-6244}, 3415 {0x1c83,0x1c84,1,-6242}, 3416 {0x1c85,0x1c85,-1,-6243}, 3417 {0x1c86,0x1c86,-1,-6236}, 3418 {0x1c87,0x1c87,-1,-6181}, 3419 {0x1c88,0x1c88,-1,35266}, 3420 {0x1d79,0x1d79,-1,35332}, 3421 {0x1d7d,0x1d7d,-1,3814}, 3422 {0x1d8e,0x1d8e,-1,35384}, 3423 {0x1e01,0x1e95,2,-1}, 3424 {0x1e9b,0x1e9b,-1,-59}, 3425 {0x1ea1,0x1eff,2,-1}, 3426 {0x1f00,0x1f07,1,8}, 3427 {0x1f10,0x1f15,1,8}, 3428 {0x1f20,0x1f27,1,8}, 3429 {0x1f30,0x1f37,1,8}, 3430 {0x1f40,0x1f45,1,8}, 3431 {0x1f51,0x1f57,2,8}, 3432 {0x1f60,0x1f67,1,8}, 3433 {0x1f70,0x1f71,1,74}, 3434 {0x1f72,0x1f75,1,86}, 3435 {0x1f76,0x1f77,1,100}, 3436 {0x1f78,0x1f79,1,128}, 3437 {0x1f7a,0x1f7b,1,112}, 3438 {0x1f7c,0x1f7d,1,126}, 3439 {0x1f80,0x1f87,1,8}, 3440 {0x1f90,0x1f97,1,8}, 3441 {0x1fa0,0x1fa7,1,8}, 3442 {0x1fb0,0x1fb1,1,8}, 3443 {0x1fb3,0x1fb3,-1,9}, 3444 {0x1fbe,0x1fbe,-1,-7205}, 3445 {0x1fc3,0x1fc3,-1,9}, 3446 {0x1fd0,0x1fd1,1,8}, 3447 {0x1fe0,0x1fe1,1,8}, 3448 {0x1fe5,0x1fe5,-1,7}, 3449 {0x1ff3,0x1ff3,-1,9}, 3450 {0x214e,0x214e,-1,-28}, 3451 {0x2170,0x217f,1,-16}, 3452 {0x2184,0x2184,-1,-1}, 3453 {0x24d0,0x24e9,1,-26}, 3454 {0x2c30,0x2c5e,1,-48}, 3455 {0x2c61,0x2c61,-1,-1}, 3456 {0x2c65,0x2c65,-1,-10795}, 3457 {0x2c66,0x2c66,-1,-10792}, 3458 {0x2c68,0x2c6c,2,-1}, 3459 {0x2c73,0x2c76,3,-1}, 3460 {0x2c81,0x2ce3,2,-1}, 3461 {0x2cec,0x2cee,2,-1}, 3462 {0x2cf3,0x2cf3,-1,-1}, 3463 {0x2d00,0x2d25,1,-7264}, 3464 {0x2d27,0x2d2d,6,-7264}, 3465 {0xa641,0xa66d,2,-1}, 3466 {0xa681,0xa69b,2,-1}, 3467 {0xa723,0xa72f,2,-1}, 3468 {0xa733,0xa76f,2,-1}, 3469 {0xa77a,0xa77c,2,-1}, 3470 {0xa77f,0xa787,2,-1}, 3471 {0xa78c,0xa791,5,-1}, 3472 {0xa793,0xa793,-1,-1}, 3473 {0xa794,0xa794,-1,48}, 3474 {0xa797,0xa7a9,2,-1}, 3475 {0xa7b5,0xa7bf,2,-1}, 3476 {0xa7c3,0xa7c3,-1,-1}, 3477 {0xab53,0xab53,-1,-928}, 3478 {0xab70,0xabbf,1,-38864}, 3479 {0xff41,0xff5a,1,-32}, 3480 {0x10428,0x1044f,1,-40}, 3481 {0x104d8,0x104fb,1,-40}, 3482 {0x10cc0,0x10cf2,1,-64}, 3483 {0x118c0,0x118df,1,-32}, 3484 {0x16e60,0x16e7f,1,-32}, 3485 {0x1e922,0x1e943,1,-34} 3486 }; 3487 3488 /* 3489 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3490 * simple case folding. 3491 */ 3492 int 3493 utf_toupper(int a) 3494 { 3495 // If 'casemap' contains "keepascii" use ASCII style toupper(). 3496 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3497 return TOUPPER_ASC(a); 3498 3499 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3500 // If towupper() is available and handles Unicode, use it. 3501 if (!(cmp_flags & CMP_INTERNAL)) 3502 return towupper(a); 3503 #endif 3504 3505 // For characters below 128 use locale sensitive toupper(). 3506 if (a < 128) 3507 return TOUPPER_LOC(a); 3508 3509 // For any other characters use the above mapping table. 3510 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3511 } 3512 3513 int 3514 utf_islower(int a) 3515 { 3516 // German sharp s is lower case but has no upper case equivalent. 3517 return (utf_toupper(a) != a) || a == 0xdf; 3518 } 3519 3520 /* 3521 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3522 * simple case folding. 3523 */ 3524 int 3525 utf_tolower(int a) 3526 { 3527 // If 'casemap' contains "keepascii" use ASCII style tolower(). 3528 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3529 return TOLOWER_ASC(a); 3530 3531 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3532 // If towlower() is available and handles Unicode, use it. 3533 if (!(cmp_flags & CMP_INTERNAL)) 3534 return towlower(a); 3535 #endif 3536 3537 // For characters below 128 use locale sensitive tolower(). 3538 if (a < 128) 3539 return TOLOWER_LOC(a); 3540 3541 // For any other characters use the above mapping table. 3542 return utf_convert(a, toLower, (int)sizeof(toLower)); 3543 } 3544 3545 int 3546 utf_isupper(int a) 3547 { 3548 return (utf_tolower(a) != a); 3549 } 3550 3551 static int 3552 utf_strnicmp( 3553 char_u *s1, 3554 char_u *s2, 3555 size_t n1, 3556 size_t n2) 3557 { 3558 int c1, c2, cdiff; 3559 char_u buffer[6]; 3560 3561 for (;;) 3562 { 3563 c1 = utf_safe_read_char_adv(&s1, &n1); 3564 c2 = utf_safe_read_char_adv(&s2, &n2); 3565 3566 if (c1 <= 0 || c2 <= 0) 3567 break; 3568 3569 if (c1 == c2) 3570 continue; 3571 3572 cdiff = utf_fold(c1) - utf_fold(c2); 3573 if (cdiff != 0) 3574 return cdiff; 3575 } 3576 3577 // some string ended or has an incomplete/illegal character sequence 3578 3579 if (c1 == 0 || c2 == 0) 3580 { 3581 // some string ended. shorter string is smaller 3582 if (c1 == 0 && c2 == 0) 3583 return 0; 3584 return c1 == 0 ? -1 : 1; 3585 } 3586 3587 // Continue with bytewise comparison to produce some result that 3588 // would make comparison operations involving this function transitive. 3589 // 3590 // If only one string had an error, comparison should be made with 3591 // folded version of the other string. In this case it is enough 3592 // to fold just one character to determine the result of comparison. 3593 3594 if (c1 != -1 && c2 == -1) 3595 { 3596 n1 = utf_char2bytes(utf_fold(c1), buffer); 3597 s1 = buffer; 3598 } 3599 else if (c2 != -1 && c1 == -1) 3600 { 3601 n2 = utf_char2bytes(utf_fold(c2), buffer); 3602 s2 = buffer; 3603 } 3604 3605 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3606 { 3607 cdiff = (int)(*s1) - (int)(*s2); 3608 if (cdiff != 0) 3609 return cdiff; 3610 3611 s1++; 3612 s2++; 3613 n1--; 3614 n2--; 3615 } 3616 3617 if (n1 > 0 && *s1 == NUL) 3618 n1 = 0; 3619 if (n2 > 0 && *s2 == NUL) 3620 n2 = 0; 3621 3622 if (n1 == 0 && n2 == 0) 3623 return 0; 3624 return n1 == 0 ? -1 : 1; 3625 } 3626 3627 /* 3628 * Version of strnicmp() that handles multi-byte characters. 3629 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3630 * probably use strnicmp(), because there are no ASCII characters in the 3631 * second byte. 3632 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3633 * two characters otherwise. 3634 */ 3635 int 3636 mb_strnicmp(char_u *s1, char_u *s2, size_t nn) 3637 { 3638 int i, l; 3639 int cdiff; 3640 int n = (int)nn; 3641 3642 if (enc_utf8) 3643 { 3644 return utf_strnicmp(s1, s2, nn, nn); 3645 } 3646 else 3647 { 3648 for (i = 0; i < n; i += l) 3649 { 3650 if (s1[i] == NUL && s2[i] == NUL) // both strings end 3651 return 0; 3652 3653 l = (*mb_ptr2len)(s1 + i); 3654 if (l <= 1) 3655 { 3656 // Single byte: first check normally, then with ignore case. 3657 if (s1[i] != s2[i]) 3658 { 3659 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3660 if (cdiff != 0) 3661 return cdiff; 3662 } 3663 } 3664 else 3665 { 3666 // For non-Unicode multi-byte don't ignore case. 3667 if (l > n - i) 3668 l = n - i; 3669 cdiff = STRNCMP(s1 + i, s2 + i, l); 3670 if (cdiff != 0) 3671 return cdiff; 3672 } 3673 } 3674 } 3675 return 0; 3676 } 3677 3678 /* 3679 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3680 * 'encoding' has been set to. 3681 */ 3682 void 3683 show_utf8(void) 3684 { 3685 int len; 3686 int rlen = 0; 3687 char_u *line; 3688 int clen; 3689 int i; 3690 3691 // Get the byte length of the char under the cursor, including composing 3692 // characters. 3693 line = ml_get_cursor(); 3694 len = utfc_ptr2len(line); 3695 if (len == 0) 3696 { 3697 msg("NUL"); 3698 return; 3699 } 3700 3701 clen = 0; 3702 for (i = 0; i < len; ++i) 3703 { 3704 if (clen == 0) 3705 { 3706 // start of (composing) character, get its length 3707 if (i > 0) 3708 { 3709 STRCPY(IObuff + rlen, "+ "); 3710 rlen += 2; 3711 } 3712 clen = utf_ptr2len(line + i); 3713 } 3714 sprintf((char *)IObuff + rlen, "%02x ", 3715 (line[i] == NL) ? NUL : line[i]); // NUL is stored as NL 3716 --clen; 3717 rlen += (int)STRLEN(IObuff + rlen); 3718 if (rlen > IOSIZE - 20) 3719 break; 3720 } 3721 3722 msg((char *)IObuff); 3723 } 3724 3725 /* 3726 * mb_head_off() function pointer. 3727 * Return offset from "p" to the first byte of the character it points into. 3728 * If "p" points to the NUL at the end of the string return 0. 3729 * Returns 0 when already at the first byte of a character. 3730 */ 3731 int 3732 latin_head_off(char_u *base UNUSED, char_u *p UNUSED) 3733 { 3734 return 0; 3735 } 3736 3737 static int 3738 dbcs_head_off(char_u *base, char_u *p) 3739 { 3740 char_u *q; 3741 3742 // It can't be a trailing byte when not using DBCS, at the start of the 3743 // string or the previous byte can't start a double-byte. 3744 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3745 return 0; 3746 3747 // This is slow: need to start at the base and go forward until the 3748 // byte we are looking for. Return 1 when we went past it, 0 otherwise. 3749 q = base; 3750 while (q < p) 3751 q += dbcs_ptr2len(q); 3752 return (q == p) ? 0 : 1; 3753 } 3754 3755 /* 3756 * Special version of dbcs_head_off() that works for ScreenLines[], where 3757 * single-width DBCS_JPNU characters are stored separately. 3758 */ 3759 int 3760 dbcs_screen_head_off(char_u *base, char_u *p) 3761 { 3762 char_u *q; 3763 3764 // It can't be a trailing byte when not using DBCS, at the start of the 3765 // string or the previous byte can't start a double-byte. 3766 // For euc-jp an 0x8e byte in the previous cell always means we have a 3767 // lead byte in the current cell. 3768 if (p <= base 3769 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3770 || MB_BYTE2LEN(p[-1]) == 1 3771 || *p == NUL) 3772 return 0; 3773 3774 // This is slow: need to start at the base and go forward until the 3775 // byte we are looking for. Return 1 when we went past it, 0 otherwise. 3776 // For DBCS_JPNU look out for 0x8e, which means the second byte is not 3777 // stored as the next byte. 3778 q = base; 3779 while (q < p) 3780 { 3781 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3782 ++q; 3783 else 3784 q += dbcs_ptr2len(q); 3785 } 3786 return (q == p) ? 0 : 1; 3787 } 3788 3789 int 3790 utf_head_off(char_u *base, char_u *p) 3791 { 3792 char_u *q; 3793 char_u *s; 3794 int c; 3795 int len; 3796 #ifdef FEAT_ARABIC 3797 char_u *j; 3798 #endif 3799 3800 if (*p < 0x80) // be quick for ASCII 3801 return 0; 3802 3803 // Skip backwards over trailing bytes: 10xx.xxxx 3804 // Skip backwards again if on a composing char. 3805 for (q = p; ; --q) 3806 { 3807 // Move s to the last byte of this char. 3808 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3809 ; 3810 // Move q to the first byte of this char. 3811 while (q > base && (*q & 0xc0) == 0x80) 3812 --q; 3813 // Check for illegal sequence. Do allow an illegal byte after where we 3814 // started. 3815 len = utf8len_tab[*q]; 3816 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3817 return 0; 3818 3819 if (q <= base) 3820 break; 3821 3822 c = utf_ptr2char(q); 3823 if (utf_iscomposing(c)) 3824 continue; 3825 3826 #ifdef FEAT_ARABIC 3827 if (arabic_maycombine(c)) 3828 { 3829 // Advance to get a sneak-peak at the next char 3830 j = q; 3831 --j; 3832 // Move j to the first byte of this char. 3833 while (j > base && (*j & 0xc0) == 0x80) 3834 --j; 3835 if (arabic_combine(utf_ptr2char(j), c)) 3836 continue; 3837 } 3838 #endif 3839 break; 3840 } 3841 3842 return (int)(p - q); 3843 } 3844 3845 /* 3846 * Whether space is NOT allowed before/after 'c'. 3847 */ 3848 int 3849 utf_eat_space(int cc) 3850 { 3851 return ((cc >= 0x2000 && cc <= 0x206F) // General punctuations 3852 || (cc >= 0x2e00 && cc <= 0x2e7f) // Supplemental punctuations 3853 || (cc >= 0x3000 && cc <= 0x303f) // CJK symbols and punctuations 3854 || (cc >= 0xff01 && cc <= 0xff0f) // Full width ASCII punctuations 3855 || (cc >= 0xff1a && cc <= 0xff20) // .. 3856 || (cc >= 0xff3b && cc <= 0xff40) // .. 3857 || (cc >= 0xff5b && cc <= 0xff65)); // .. 3858 } 3859 3860 /* 3861 * Whether line break is allowed before "cc". 3862 */ 3863 int 3864 utf_allow_break_before(int cc) 3865 { 3866 static const int BOL_prohibition_punct[] = 3867 { 3868 '!', 3869 '%', 3870 ')', 3871 ',', 3872 ':', 3873 ';', 3874 '>', 3875 '?', 3876 ']', 3877 '}', 3878 0x2019, // ’ right single quotation mark 3879 0x201d, // ” right double quotation mark 3880 0x2020, // † dagger 3881 0x2021, // ‡ double dagger 3882 0x2026, // … horizontal ellipsis 3883 0x2030, // ‰ per mille sign 3884 0x2031, // ‱ per then thousand sign 3885 0x203c, // ‼ double exclamation mark 3886 0x2047, // ⁇ double question mark 3887 0x2048, // ⁈ question exclamation mark 3888 0x2049, // ⁉ exclamation question mark 3889 0x2103, // ℃ degree celsius 3890 0x2109, // ℉ degree fahrenheit 3891 0x3001, // 、 ideographic comma 3892 0x3002, // 。 ideographic full stop 3893 0x3009, // 〉 right angle bracket 3894 0x300b, // 》 right double angle bracket 3895 0x300d, // 」 right corner bracket 3896 0x300f, // 』 right white corner bracket 3897 0x3011, // 】 right black lenticular bracket 3898 0x3015, // 〕 right tortoise shell bracket 3899 0x3017, // 〗 right white lenticular bracket 3900 0x3019, // 〙 right white tortoise shell bracket 3901 0x301b, // 〛 right white square bracket 3902 0xff01, // ！ fullwidth exclamation mark 3903 0xff09, // ） fullwidth right parenthesis 3904 0xff0c, // ， fullwidth comma 3905 0xff0e, // ． fullwidth full stop 3906 0xff1a, // ： fullwidth colon 3907 0xff1b, // ； fullwidth semicolon 3908 0xff1f, // ？ fullwidth question mark 3909 0xff3d, // ］ fullwidth right square bracket 3910 0xff5d, // ｝ fullwidth right curly bracket 3911 }; 3912 3913 int first = 0; 3914 int last = sizeof(BOL_prohibition_punct)/sizeof(int) - 1; 3915 int mid = 0; 3916 3917 while (first < last) 3918 { 3919 mid = (first + last)/2; 3920 3921 if (cc == BOL_prohibition_punct[mid]) 3922 return FALSE; 3923 else if (cc > BOL_prohibition_punct[mid]) 3924 first = mid + 1; 3925 else 3926 last = mid - 1; 3927 } 3928 3929 return cc != BOL_prohibition_punct[first]; 3930 } 3931 3932 /* 3933 * Whether line break is allowed after "cc". 3934 */ 3935 static int 3936 utf_allow_break_after(int cc) 3937 { 3938 static const int EOL_prohibition_punct[] = 3939 { 3940 '(', 3941 '<', 3942 '[', 3943 '`', 3944 '{', 3945 //0x2014, // — em dash 3946 0x2018, // ‘ left single quotation mark 3947 0x201c, // “ left double quotation mark 3948 //0x2053, // ～ swung dash 3949 0x3008, // 〈 left angle bracket 3950 0x300a, // 《 left double angle bracket 3951 0x300c, // 「 left corner bracket 3952 0x300e, // 『 left white corner bracket 3953 0x3010, // 【 left black lenticular bracket 3954 0x3014, // 〔 left tortoise shell bracket 3955 0x3016, // 〖 left white lenticular bracket 3956 0x3018, // 〘 left white tortoise shell bracket 3957 0x301a, // 〚 left white square bracket 3958 0xff08, // （ fullwidth left parenthesis 3959 0xff3b, // ［ fullwidth left square bracket 3960 0xff5b, // ｛ fullwidth left curly bracket 3961 }; 3962 3963 int first = 0; 3964 int last = sizeof(EOL_prohibition_punct)/sizeof(int) - 1; 3965 int mid = 0; 3966 3967 while (first < last) 3968 { 3969 mid = (first + last)/2; 3970 3971 if (cc == EOL_prohibition_punct[mid]) 3972 return FALSE; 3973 else if (cc > EOL_prohibition_punct[mid]) 3974 first = mid + 1; 3975 else 3976 last = mid - 1; 3977 } 3978 3979 return cc != EOL_prohibition_punct[first]; 3980 } 3981 3982 /* 3983 * Whether line break is allowed between "cc" and "ncc". 3984 */ 3985 int 3986 utf_allow_break(int cc, int ncc) 3987 { 3988 // don't break between two-letter punctuations 3989 if (cc == ncc 3990 && (cc == 0x2014 // em dash 3991 || cc == 0x2026)) // horizontal ellipsis 3992 return FALSE; 3993 3994 return utf_allow_break_after(cc) && utf_allow_break_before(ncc); 3995 } 3996 3997 /* 3998 * Copy a character from "*fp" to "*tp" and advance the pointers. 3999 */ 4000 void 4001 mb_copy_char(char_u **fp, char_u **tp) 4002 { 4003 int l = (*mb_ptr2len)(*fp); 4004 4005 mch_memmove(*tp, *fp, (size_t)l); 4006 *tp += l; 4007 *fp += l; 4008 } 4009 4010 /* 4011 * Return the offset from "p" to the first byte of a character. When "p" is 4012 * at the start of a character 0 is returned, otherwise the offset to the next 4013 * character. Can start anywhere in a stream of bytes. 4014 */ 4015 int 4016 mb_off_next(char_u *base, char_u *p) 4017 { 4018 int i; 4019 int j; 4020 4021 if (enc_utf8) 4022 { 4023 if (*p < 0x80) // be quick for ASCII 4024 return 0; 4025 4026 // Find the next character that isn't 10xx.xxxx 4027 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 4028 ; 4029 if (i > 0) 4030 { 4031 // Check for illegal sequence. 4032 for (j = 0; p - j > base; ++j) 4033 if ((p[-j] & 0xc0) != 0x80) 4034 break; 4035 if (utf8len_tab[p[-j]] != i + j) 4036 return 0; 4037 } 4038 return i; 4039 } 4040 4041 // Only need to check if we're on a trail byte, it doesn't matter if we 4042 // want the offset to the next or current character. 4043 return (*mb_head_off)(base, p); 4044 } 4045 4046 /* 4047 * Return the offset from "p" to the last byte of the character it points 4048 * into. Can start anywhere in a stream of bytes. 4049 */ 4050 int 4051 mb_tail_off(char_u *base, char_u *p) 4052 { 4053 int i; 4054 int j; 4055 4056 if (*p == NUL) 4057 return 0; 4058 4059 if (enc_utf8) 4060 { 4061 // Find the last character that is 10xx.xxxx 4062 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 4063 ; 4064 // Check for illegal sequence. 4065 for (j = 0; p - j > base; ++j) 4066 if ((p[-j] & 0xc0) != 0x80) 4067 break; 4068 if (utf8len_tab[p[-j]] != i + j + 1) 4069 return 0; 4070 return i; 4071 } 4072 4073 // It can't be the first byte if a double-byte when not using DBCS, at the 4074 // end of the string or the byte can't start a double-byte. 4075 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 4076 return 0; 4077 4078 // Return 1 when on the lead byte, 0 when on the tail byte. 4079 return 1 - dbcs_head_off(base, p); 4080 } 4081 4082 /* 4083 * Find the next illegal byte sequence. 4084 */ 4085 void 4086 utf_find_illegal(void) 4087 { 4088 pos_T pos = curwin->w_cursor; 4089 char_u *p; 4090 int len; 4091 vimconv_T vimconv; 4092 char_u *tofree = NULL; 4093 4094 vimconv.vc_type = CONV_NONE; 4095 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 4096 { 4097 // 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 4098 // possibly a utf-8 file with illegal bytes. Setup for conversion 4099 // from utf-8 to 'fileencoding'. 4100 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 4101 } 4102 4103 curwin->w_cursor.coladd = 0; 4104 for (;;) 4105 { 4106 p = ml_get_cursor(); 4107 if (vimconv.vc_type != CONV_NONE) 4108 { 4109 vim_free(tofree); 4110 tofree = string_convert(&vimconv, p, NULL); 4111 if (tofree == NULL) 4112 break; 4113 p = tofree; 4114 } 4115 4116 while (*p != NUL) 4117 { 4118 // Illegal means that there are not enough trail bytes (checked by 4119 // utf_ptr2len()) or too many of them (overlong sequence). 4120 len = utf_ptr2len(p); 4121 if (*p >= 0x80 && (len == 1 4122 || utf_char2len(utf_ptr2char(p)) != len)) 4123 { 4124 if (vimconv.vc_type == CONV_NONE) 4125 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 4126 else 4127 { 4128 int l; 4129 4130 len = (int)(p - tofree); 4131 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 4132 { 4133 l = utf_ptr2len(p); 4134 curwin->w_cursor.col += l; 4135 } 4136 } 4137 goto theend; 4138 } 4139 p += len; 4140 } 4141 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 4142 break; 4143 ++curwin->w_cursor.lnum; 4144 curwin->w_cursor.col = 0; 4145 } 4146 4147 // didn't find it: don't move and beep 4148 curwin->w_cursor = pos; 4149 beep_flush(); 4150 4151 theend: 4152 vim_free(tofree); 4153 convert_setup(&vimconv, NULL, NULL); 4154 } 4155 4156 #if defined(FEAT_GUI_GTK) || defined(PROTO) 4157 /* 4158 * Return TRUE if string "s" is a valid utf-8 string. 4159 * When "end" is NULL stop at the first NUL. 4160 * When "end" is positive stop there. 4161 */ 4162 int 4163 utf_valid_string(char_u *s, char_u *end) 4164 { 4165 int l; 4166 char_u *p = s; 4167 4168 while (end == NULL ? *p != NUL : p < end) 4169 { 4170 l = utf8len_tab_zero[*p]; 4171 if (l == 0) 4172 return FALSE; // invalid lead byte 4173 if (end != NULL && p + l > end) 4174 return FALSE; // incomplete byte sequence 4175 ++p; 4176 while (--l > 0) 4177 if ((*p++ & 0xc0) != 0x80) 4178 return FALSE; // invalid trail byte 4179 } 4180 return TRUE; 4181 } 4182 #endif 4183 4184 #if defined(FEAT_GUI) || defined(PROTO) 4185 /* 4186 * Special version of mb_tail_off() for use in ScreenLines[]. 4187 */ 4188 int 4189 dbcs_screen_tail_off(char_u *base, char_u *p) 4190 { 4191 // It can't be the first byte if a double-byte when not using DBCS, at the 4192 // end of the string or the byte can't start a double-byte. 4193 // For euc-jp an 0x8e byte always means we have a lead byte in the current 4194 // cell. 4195 if (*p == NUL || p[1] == NUL 4196 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 4197 || MB_BYTE2LEN(*p) == 1) 4198 return 0; 4199 4200 // Return 1 when on the lead byte, 0 when on the tail byte. 4201 return 1 - dbcs_screen_head_off(base, p); 4202 } 4203 #endif 4204 4205 /* 4206 * If the cursor moves on an trail byte, set the cursor on the lead byte. 4207 * Thus it moves left if necessary. 4208 * Return TRUE when the cursor was adjusted. 4209 */ 4210 void 4211 mb_adjust_cursor(void) 4212 { 4213 mb_adjustpos(curbuf, &curwin->w_cursor); 4214 } 4215 4216 /* 4217 * Adjust position "*lp" to point to the first byte of a multi-byte character. 4218 * If it points to a tail byte it's moved backwards to the head byte. 4219 */ 4220 void 4221 mb_adjustpos(buf_T *buf, pos_T *lp) 4222 { 4223 char_u *p; 4224 4225 if (lp->col > 0 || lp->coladd > 1) 4226 { 4227 p = ml_get_buf(buf, lp->lnum, FALSE); 4228 if (*p == NUL || (int)STRLEN(p) < lp->col) 4229 lp->col = 0; 4230 else 4231 lp->col -= (*mb_head_off)(p, p + lp->col); 4232 // Reset "coladd" when the cursor would be on the right half of a 4233 // double-wide character. 4234 if (lp->coladd == 1 4235 && p[lp->col] != TAB 4236 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 4237 && ptr2cells(p + lp->col) > 1) 4238 lp->coladd = 0; 4239 } 4240 } 4241 4242 /* 4243 * Return a pointer to the character before "*p", if there is one. 4244 */ 4245 char_u * 4246 mb_prevptr( 4247 char_u *line, // start of the string 4248 char_u *p) 4249 { 4250 if (p > line) 4251 MB_PTR_BACK(line, p); 4252 return p; 4253 } 4254 4255 /* 4256 * Return the character length of "str". Each multi-byte character (with 4257 * following composing characters) counts as one. 4258 */ 4259 int 4260 mb_charlen(char_u *str) 4261 { 4262 char_u *p = str; 4263 int count; 4264 4265 if (p == NULL) 4266 return 0; 4267 4268 for (count = 0; *p != NUL; count++) 4269 p += (*mb_ptr2len)(p); 4270 4271 return count; 4272 } 4273 4274 #if defined(FEAT_SPELL) || defined(PROTO) 4275 /* 4276 * Like mb_charlen() but for a string with specified length. 4277 */ 4278 int 4279 mb_charlen_len(char_u *str, int len) 4280 { 4281 char_u *p = str; 4282 int count; 4283 4284 for (count = 0; *p != NUL && p < str + len; count++) 4285 p += (*mb_ptr2len)(p); 4286 4287 return count; 4288 } 4289 #endif 4290 4291 /* 4292 * Try to un-escape a multi-byte character. 4293 * Used for the "to" and "from" part of a mapping. 4294 * Return the un-escaped string if it is a multi-byte character, and advance 4295 * "pp" to just after the bytes that formed it. 4296 * Return NULL if no multi-byte char was found. 4297 */ 4298 char_u * 4299 mb_unescape(char_u **pp) 4300 { 4301 static char_u buf[6]; 4302 int n; 4303 int m = 0; 4304 char_u *str = *pp; 4305 4306 // Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 4307 // KS_EXTRA KE_CSI to CSI. 4308 // Maximum length of a utf-8 character is 4 bytes. 4309 for (n = 0; str[n] != NUL && m < 4; ++n) 4310 { 4311 if (str[n] == K_SPECIAL 4312 && str[n + 1] == KS_SPECIAL 4313 && str[n + 2] == KE_FILLER) 4314 { 4315 buf[m++] = K_SPECIAL; 4316 n += 2; 4317 } 4318 else if ((str[n] == K_SPECIAL 4319 # ifdef FEAT_GUI 4320 || str[n] == CSI 4321 # endif 4322 ) 4323 && str[n + 1] == KS_EXTRA 4324 && str[n + 2] == (int)KE_CSI) 4325 { 4326 buf[m++] = CSI; 4327 n += 2; 4328 } 4329 else if (str[n] == K_SPECIAL 4330 # ifdef FEAT_GUI 4331 || str[n] == CSI 4332 # endif 4333 ) 4334 break; // a special key can't be a multibyte char 4335 else 4336 buf[m++] = str[n]; 4337 buf[m] = NUL; 4338 4339 // Return a multi-byte character if it's found. An illegal sequence 4340 // will result in a 1 here. 4341 if ((*mb_ptr2len)(buf) > 1) 4342 { 4343 *pp = str + n + 1; 4344 return buf; 4345 } 4346 4347 // Bail out quickly for ASCII. 4348 if (buf[0] < 128) 4349 break; 4350 } 4351 return NULL; 4352 } 4353 4354 /* 4355 * Return TRUE if the character at "row"/"col" on the screen is the left side 4356 * of a double-width character. 4357 * Caller must make sure "row" and "col" are not invalid! 4358 */ 4359 int 4360 mb_lefthalve(int row, int col) 4361 { 4362 return (*mb_off2cells)(LineOffset[row] + col, 4363 LineOffset[row] + screen_Columns) > 1; 4364 } 4365 4366 /* 4367 * Correct a position on the screen, if it's the right half of a double-wide 4368 * char move it to the left half. Returns the corrected column. 4369 */ 4370 int 4371 mb_fix_col(int col, int row) 4372 { 4373 int off; 4374 4375 col = check_col(col); 4376 row = check_row(row); 4377 off = LineOffset[row] + col; 4378 if (has_mbyte && ScreenLines != NULL && col > 0 4379 && ((enc_dbcs 4380 && ScreenLines[off] != NUL 4381 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 4382 ScreenLines + off)) 4383 || (enc_utf8 && ScreenLines[off] == 0 4384 && ScreenLinesUC[off] == 0))) 4385 return col - 1; 4386 return col; 4387 } 4388 4389 static int enc_alias_search(char_u *name); 4390 4391 /* 4392 * Skip the Vim specific head of a 'encoding' name. 4393 */ 4394 char_u * 4395 enc_skip(char_u *p) 4396 { 4397 if (STRNCMP(p, "2byte-", 6) == 0) 4398 return p + 6; 4399 if (STRNCMP(p, "8bit-", 5) == 0) 4400 return p + 5; 4401 return p; 4402 } 4403 4404 /* 4405 * Find the canonical name for encoding "enc". 4406 * When the name isn't recognized, returns "enc" itself, but with all lower 4407 * case characters and '_' replaced with '-'. 4408 * Returns an allocated string. NULL for out-of-memory. 4409 */ 4410 char_u * 4411 enc_canonize(char_u *enc) 4412 { 4413 char_u *r; 4414 char_u *p, *s; 4415 int i; 4416 4417 if (STRCMP(enc, "default") == 0) 4418 { 4419 // Use the default encoding as it's found by set_init_1(). 4420 r = get_encoding_default(); 4421 if (r == NULL) 4422 r = (char_u *)"latin1"; 4423 return vim_strsave(r); 4424 } 4425 4426 // copy "enc" to allocated memory, with room for two '-' 4427 r = alloc(STRLEN(enc) + 3); 4428 if (r != NULL) 4429 { 4430 // Make it all lower case and replace '_' with '-'. 4431 p = r; 4432 for (s = enc; *s != NUL; ++s) 4433 { 4434 if (*s == '_') 4435 *p++ = '-'; 4436 else 4437 *p++ = TOLOWER_ASC(*s); 4438 } 4439 *p = NUL; 4440 4441 // Skip "2byte-" and "8bit-". 4442 p = enc_skip(r); 4443 4444 // Change "microsoft-cp" to "cp". Used in some spell files. 4445 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4446 STRMOVE(p, p + 10); 4447 4448 // "iso8859" -> "iso-8859" 4449 if (STRNCMP(p, "iso8859", 7) == 0) 4450 { 4451 STRMOVE(p + 4, p + 3); 4452 p[3] = '-'; 4453 } 4454 4455 // "iso-8859n" -> "iso-8859-n" 4456 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4457 { 4458 STRMOVE(p + 9, p + 8); 4459 p[8] = '-'; 4460 } 4461 4462 // "latin-N" -> "latinN" 4463 if (STRNCMP(p, "latin-", 6) == 0) 4464 STRMOVE(p + 5, p + 6); 4465 4466 if (enc_canon_search(p) >= 0) 4467 { 4468 // canonical name can be used unmodified 4469 if (p != r) 4470 STRMOVE(r, p); 4471 } 4472 else if ((i = enc_alias_search(p)) >= 0) 4473 { 4474 // alias recognized, get canonical name 4475 vim_free(r); 4476 r = vim_strsave((char_u *)enc_canon_table[i].name); 4477 } 4478 } 4479 return r; 4480 } 4481 4482 /* 4483 * Search for an encoding alias of "name". 4484 * Returns -1 when not found. 4485 */ 4486 static int 4487 enc_alias_search(char_u *name) 4488 { 4489 int i; 4490 4491 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4492 if (STRCMP(name, enc_alias_table[i].name) == 0) 4493 return enc_alias_table[i].canon; 4494 return -1; 4495 } 4496 4497 4498 #ifdef HAVE_LANGINFO_H 4499 # include <langinfo.h> 4500 #endif 4501 4502 #if !defined(FEAT_GUI_MSWIN) || defined(VIMDLL) 4503 /* 4504 * Get the canonicalized encoding from the specified locale string "locale" 4505 * or from the environment variables LC_ALL, LC_CTYPE and LANG. 4506 * Returns an allocated string when successful, NULL when not. 4507 */ 4508 char_u * 4509 enc_locale_env(char *locale) 4510 { 4511 char *s = locale; 4512 char *p; 4513 int i; 4514 char buf[50]; 4515 4516 if (s == NULL || *s == NUL) 4517 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4518 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4519 s = getenv("LANG"); 4520 4521 if (s == NULL || *s == NUL) 4522 return NULL; 4523 4524 // The most generic locale format is: 4525 // language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4526 // If there is a '.' remove the part before it. 4527 // if there is something after the codeset, remove it. 4528 // Make the name lowercase and replace '_' with '-'. 4529 // Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4530 // "ko_KR.EUC" == "euc-kr" 4531 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4532 { 4533 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4534 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4535 { 4536 // copy "XY.EUC" to "euc-XY" to buf[10] 4537 STRCPY(buf + 10, "euc-"); 4538 buf[14] = p[-2]; 4539 buf[15] = p[-1]; 4540 buf[16] = 0; 4541 s = buf + 10; 4542 } 4543 else 4544 s = p + 1; 4545 } 4546 for (i = 0; i < (int)sizeof(buf) - 1 && s[i] != NUL; ++i) 4547 { 4548 if (s[i] == '_' || s[i] == '-') 4549 buf[i] = '-'; 4550 else if (isalnum((int)s[i])) 4551 buf[i] = TOLOWER_ASC(s[i]); 4552 else 4553 break; 4554 } 4555 buf[i] = NUL; 4556 4557 return enc_canonize((char_u *)buf); 4558 } 4559 #endif 4560 4561 /* 4562 * Get the canonicalized encoding of the current locale. 4563 * Returns an allocated string when successful, NULL when not. 4564 */ 4565 char_u * 4566 enc_locale(void) 4567 { 4568 #ifdef MSWIN 4569 char buf[50]; 4570 long acp = GetACP(); 4571 4572 if (acp == 1200) 4573 STRCPY(buf, "ucs-2le"); 4574 else if (acp == 1252) // cp1252 is used as latin1 4575 STRCPY(buf, "latin1"); 4576 else if (acp == 65001) 4577 STRCPY(buf, "utf-8"); 4578 else 4579 sprintf(buf, "cp%ld", acp); 4580 4581 return enc_canonize((char_u *)buf); 4582 #else 4583 char *s; 4584 4585 # ifdef HAVE_NL_LANGINFO_CODESET 4586 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4587 # endif 4588 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4589 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4590 # endif 4591 s = NULL; 4592 4593 return enc_locale_env(s); 4594 #endif 4595 } 4596 4597 # if defined(MSWIN) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4598 /* 4599 * Convert an encoding name to an MS-Windows codepage. 4600 * Returns zero if no codepage can be figured out. 4601 */ 4602 int 4603 encname2codepage(char_u *name) 4604 { 4605 int cp; 4606 char_u *p = name; 4607 int idx; 4608 4609 if (STRNCMP(p, "8bit-", 5) == 0) 4610 p += 5; 4611 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4612 p += 6; 4613 4614 if (p[0] == 'c' && p[1] == 'p') 4615 cp = atoi((char *)p + 2); 4616 else if ((idx = enc_canon_search(p)) >= 0) 4617 cp = enc_canon_table[idx].codepage; 4618 else 4619 return 0; 4620 if (IsValidCodePage(cp)) 4621 return cp; 4622 return 0; 4623 } 4624 # endif 4625 4626 # if defined(USE_ICONV) || defined(PROTO) 4627 4628 /* 4629 * Call iconv_open() with a check if iconv() works properly (there are broken 4630 * versions). 4631 * Returns (void *)-1 if failed. 4632 * (should return iconv_t, but that causes problems with prototypes). 4633 */ 4634 void * 4635 my_iconv_open(char_u *to, char_u *from) 4636 { 4637 iconv_t fd; 4638 #define ICONV_TESTLEN 400 4639 char_u tobuf[ICONV_TESTLEN]; 4640 char *p; 4641 size_t tolen; 4642 static int iconv_ok = -1; 4643 4644 if (iconv_ok == FALSE) 4645 return (void *)-1; // detected a broken iconv() previously 4646 4647 #ifdef DYNAMIC_ICONV 4648 // Check if the iconv.dll can be found. 4649 if (!iconv_enabled(TRUE)) 4650 return (void *)-1; 4651 #endif 4652 4653 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4654 4655 if (fd != (iconv_t)-1 && iconv_ok == -1) 4656 { 4657 /* 4658 * Do a dummy iconv() call to check if it actually works. There is a 4659 * version of iconv() on Linux that is broken. We can't ignore it, 4660 * because it's wide-spread. The symptoms are that after outputting 4661 * the initial shift state the "to" pointer is NULL and conversion 4662 * stops for no apparent reason after about 8160 characters. 4663 */ 4664 p = (char *)tobuf; 4665 tolen = ICONV_TESTLEN; 4666 (void)iconv(fd, NULL, NULL, &p, &tolen); 4667 if (p == NULL) 4668 { 4669 iconv_ok = FALSE; 4670 iconv_close(fd); 4671 fd = (iconv_t)-1; 4672 } 4673 else 4674 iconv_ok = TRUE; 4675 } 4676 4677 return (void *)fd; 4678 } 4679 4680 /* 4681 * Convert the string "str[slen]" with iconv(). 4682 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4683 * sequence and set "*unconvlenp" to the length of it. 4684 * Returns the converted string in allocated memory. NULL for an error. 4685 * If resultlenp is not NULL, sets it to the result length in bytes. 4686 */ 4687 static char_u * 4688 iconv_string( 4689 vimconv_T *vcp, 4690 char_u *str, 4691 int slen, 4692 int *unconvlenp, 4693 int *resultlenp) 4694 { 4695 const char *from; 4696 size_t fromlen; 4697 char *to; 4698 size_t tolen; 4699 size_t len = 0; 4700 size_t done = 0; 4701 char_u *result = NULL; 4702 char_u *p; 4703 int l; 4704 4705 from = (char *)str; 4706 fromlen = slen; 4707 for (;;) 4708 { 4709 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4710 { 4711 // Allocate enough room for most conversions. When re-allocating 4712 // increase the buffer size. 4713 len = len + fromlen * 2 + 40; 4714 p = alloc(len); 4715 if (p != NULL && done > 0) 4716 mch_memmove(p, result, done); 4717 vim_free(result); 4718 result = p; 4719 if (result == NULL) // out of memory 4720 break; 4721 } 4722 4723 to = (char *)result + done; 4724 tolen = len - done - 2; 4725 // Avoid a warning for systems with a wrong iconv() prototype by 4726 // casting the second argument to void *. 4727 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4728 != (size_t)-1) 4729 { 4730 // Finished, append a NUL. 4731 *to = NUL; 4732 break; 4733 } 4734 4735 // Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4736 // iconv library may use one of them. 4737 if (!vcp->vc_fail && unconvlenp != NULL 4738 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4739 { 4740 // Handle an incomplete sequence at the end. 4741 *to = NUL; 4742 *unconvlenp = (int)fromlen; 4743 break; 4744 } 4745 4746 // Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4747 // iconv library may use one of them. 4748 else if (!vcp->vc_fail 4749 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4750 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4751 { 4752 // Can't convert: insert a '?' and skip a character. This assumes 4753 // conversion from 'encoding' to something else. In other 4754 // situations we don't know what to skip anyway. 4755 *to++ = '?'; 4756 if ((*mb_ptr2cells)((char_u *)from) > 1) 4757 *to++ = '?'; 4758 if (enc_utf8) 4759 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4760 else 4761 { 4762 l = (*mb_ptr2len)((char_u *)from); 4763 if (l > (int)fromlen) 4764 l = (int)fromlen; 4765 } 4766 from += l; 4767 fromlen -= l; 4768 } 4769 else if (ICONV_ERRNO != ICONV_E2BIG) 4770 { 4771 // conversion failed 4772 VIM_CLEAR(result); 4773 break; 4774 } 4775 // Not enough room or skipping illegal sequence. 4776 done = to - (char *)result; 4777 } 4778 4779 if (resultlenp != NULL && result != NULL) 4780 *resultlenp = (int)(to - (char *)result); 4781 return result; 4782 } 4783 4784 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4785 /* 4786 * Dynamically load the "iconv.dll" on Win32. 4787 */ 4788 4789 # ifndef DYNAMIC_ICONV // must be generating prototypes 4790 # define HINSTANCE int 4791 # endif 4792 static HINSTANCE hIconvDLL = 0; 4793 static HINSTANCE hMsvcrtDLL = 0; 4794 4795 # ifndef DYNAMIC_ICONV_DLL 4796 # define DYNAMIC_ICONV_DLL "iconv.dll" 4797 # define DYNAMIC_ICONV_DLL_ALT1 "libiconv.dll" 4798 # define DYNAMIC_ICONV_DLL_ALT2 "libiconv2.dll" 4799 # define DYNAMIC_ICONV_DLL_ALT3 "libiconv-2.dll" 4800 # endif 4801 # ifndef DYNAMIC_MSVCRT_DLL 4802 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4803 # endif 4804 4805 /* 4806 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4807 */ 4808 int 4809 iconv_enabled(int verbose) 4810 { 4811 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4812 return TRUE; 4813 4814 // The iconv DLL file goes under different names, try them all. 4815 // Do the "2" version first, it's newer. 4816 #ifdef DYNAMIC_ICONV_DLL_ALT2 4817 if (hIconvDLL == 0) 4818 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT2); 4819 #endif 4820 #ifdef DYNAMIC_ICONV_DLL_ALT3 4821 if (hIconvDLL == 0) 4822 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT3); 4823 #endif 4824 if (hIconvDLL == 0) 4825 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4826 #ifdef DYNAMIC_ICONV_DLL_ALT1 4827 if (hIconvDLL == 0) 4828 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT1); 4829 #endif 4830 4831 if (hIconvDLL != 0) 4832 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4833 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4834 { 4835 // Only give the message when 'verbose' is set, otherwise it might be 4836 // done whenever a conversion is attempted. 4837 if (verbose && p_verbose > 0) 4838 { 4839 verbose_enter(); 4840 semsg(_(e_loadlib), 4841 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4842 verbose_leave(); 4843 } 4844 iconv_end(); 4845 return FALSE; 4846 } 4847 4848 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4849 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4850 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4851 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4852 iconv_errno = get_dll_import_func(hIconvDLL, "_errno"); 4853 if (iconv_errno == NULL) 4854 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4855 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4856 || iconvctl == NULL || iconv_errno == NULL) 4857 { 4858 iconv_end(); 4859 if (verbose && p_verbose > 0) 4860 { 4861 verbose_enter(); 4862 semsg(_(e_loadfunc), "for libiconv"); 4863 verbose_leave(); 4864 } 4865 return FALSE; 4866 } 4867 return TRUE; 4868 } 4869 4870 void 4871 iconv_end(void) 4872 { 4873 // Don't use iconv() when inputting or outputting characters. 4874 if (input_conv.vc_type == CONV_ICONV) 4875 convert_setup(&input_conv, NULL, NULL); 4876 if (output_conv.vc_type == CONV_ICONV) 4877 convert_setup(&output_conv, NULL, NULL); 4878 4879 if (hIconvDLL != 0) 4880 FreeLibrary(hIconvDLL); 4881 if (hMsvcrtDLL != 0) 4882 FreeLibrary(hMsvcrtDLL); 4883 hIconvDLL = 0; 4884 hMsvcrtDLL = 0; 4885 } 4886 # endif // DYNAMIC_ICONV 4887 # endif // USE_ICONV 4888 4889 #if defined(FEAT_EVAL) || defined(PROTO) 4890 /* 4891 * "getimstatus()" function 4892 */ 4893 void 4894 f_getimstatus(typval_T *argvars UNUSED, typval_T *rettv) 4895 { 4896 # if defined(HAVE_INPUT_METHOD) 4897 rettv->vval.v_number = im_get_status(); 4898 # endif 4899 } 4900 #endif 4901 4902 /* 4903 * Setup "vcp" for conversion from "from" to "to". 4904 * The names must have been made canonical with enc_canonize(). 4905 * vcp->vc_type must have been initialized to CONV_NONE. 4906 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 4907 * instead). 4908 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 4909 * Return FAIL when conversion is not supported, OK otherwise. 4910 */ 4911 int 4912 convert_setup(vimconv_T *vcp, char_u *from, char_u *to) 4913 { 4914 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 4915 } 4916 4917 /* 4918 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 4919 * "from" unicode charsets be considered utf-8. Same for "to". 4920 */ 4921 int 4922 convert_setup_ext( 4923 vimconv_T *vcp, 4924 char_u *from, 4925 int from_unicode_is_utf8, 4926 char_u *to, 4927 int to_unicode_is_utf8) 4928 { 4929 int from_prop; 4930 int to_prop; 4931 int from_is_utf8; 4932 int to_is_utf8; 4933 4934 // Reset to no conversion. 4935 #ifdef USE_ICONV 4936 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 4937 iconv_close(vcp->vc_fd); 4938 #endif 4939 vcp->vc_type = CONV_NONE; 4940 vcp->vc_factor = 1; 4941 vcp->vc_fail = FALSE; 4942 4943 // No conversion when one of the names is empty or they are equal. 4944 if (from == NULL || *from == NUL || to == NULL || *to == NUL 4945 || STRCMP(from, to) == 0) 4946 return OK; 4947 4948 from_prop = enc_canon_props(from); 4949 to_prop = enc_canon_props(to); 4950 if (from_unicode_is_utf8) 4951 from_is_utf8 = from_prop & ENC_UNICODE; 4952 else 4953 from_is_utf8 = from_prop == ENC_UNICODE; 4954 if (to_unicode_is_utf8) 4955 to_is_utf8 = to_prop & ENC_UNICODE; 4956 else 4957 to_is_utf8 = to_prop == ENC_UNICODE; 4958 4959 if ((from_prop & ENC_LATIN1) && to_is_utf8) 4960 { 4961 // Internal latin1 -> utf-8 conversion. 4962 vcp->vc_type = CONV_TO_UTF8; 4963 vcp->vc_factor = 2; // up to twice as long 4964 } 4965 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 4966 { 4967 // Internal latin9 -> utf-8 conversion. 4968 vcp->vc_type = CONV_9_TO_UTF8; 4969 vcp->vc_factor = 3; // up to three as long (euro sign) 4970 } 4971 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 4972 { 4973 // Internal utf-8 -> latin1 conversion. 4974 vcp->vc_type = CONV_TO_LATIN1; 4975 } 4976 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 4977 { 4978 // Internal utf-8 -> latin9 conversion. 4979 vcp->vc_type = CONV_TO_LATIN9; 4980 } 4981 #ifdef MSWIN 4982 // Win32-specific codepage <-> codepage conversion without iconv. 4983 else if ((from_is_utf8 || encname2codepage(from) > 0) 4984 && (to_is_utf8 || encname2codepage(to) > 0)) 4985 { 4986 vcp->vc_type = CONV_CODEPAGE; 4987 vcp->vc_factor = 2; // up to twice as long 4988 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 4989 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 4990 } 4991 #endif 4992 #ifdef MACOS_CONVERT 4993 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 4994 { 4995 vcp->vc_type = CONV_MAC_LATIN1; 4996 } 4997 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 4998 { 4999 vcp->vc_type = CONV_MAC_UTF8; 5000 vcp->vc_factor = 2; // up to twice as long 5001 } 5002 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 5003 { 5004 vcp->vc_type = CONV_LATIN1_MAC; 5005 } 5006 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 5007 { 5008 vcp->vc_type = CONV_UTF8_MAC; 5009 } 5010 #endif 5011 #ifdef USE_ICONV 5012 else 5013 { 5014 // Use iconv() for conversion. 5015 vcp->vc_fd = (iconv_t)my_iconv_open( 5016 to_is_utf8 ? (char_u *)"utf-8" : to, 5017 from_is_utf8 ? (char_u *)"utf-8" : from); 5018 if (vcp->vc_fd != (iconv_t)-1) 5019 { 5020 vcp->vc_type = CONV_ICONV; 5021 vcp->vc_factor = 4; // could be longer too... 5022 } 5023 } 5024 #endif 5025 if (vcp->vc_type == CONV_NONE) 5026 return FAIL; 5027 5028 return OK; 5029 } 5030 5031 #if defined(FEAT_GUI) || defined(AMIGA) || defined(MSWIN) \ 5032 || defined(PROTO) 5033 /* 5034 * Do conversion on typed input characters in-place. 5035 * The input and output are not NUL terminated! 5036 * Returns the length after conversion. 5037 */ 5038 int 5039 convert_input(char_u *ptr, int len, int maxlen) 5040 { 5041 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 5042 } 5043 #endif 5044 5045 /* 5046 * Like convert_input(), but when there is an incomplete byte sequence at the 5047 * end return that as an allocated string in "restp" and set "*restlenp" to 5048 * the length. If "restp" is NULL it is not used. 5049 */ 5050 int 5051 convert_input_safe( 5052 char_u *ptr, 5053 int len, 5054 int maxlen, 5055 char_u **restp, 5056 int *restlenp) 5057 { 5058 char_u *d; 5059 int dlen = len; 5060 int unconvertlen = 0; 5061 5062 d = string_convert_ext(&input_conv, ptr, &dlen, 5063 restp == NULL ? NULL : &unconvertlen); 5064 if (d != NULL) 5065 { 5066 if (dlen <= maxlen) 5067 { 5068 if (unconvertlen > 0) 5069 { 5070 // Move the unconverted characters to allocated memory. 5071 *restp = alloc(unconvertlen); 5072 if (*restp != NULL) 5073 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 5074 *restlenp = unconvertlen; 5075 } 5076 mch_memmove(ptr, d, dlen); 5077 } 5078 else 5079 // result is too long, keep the unconverted text (the caller must 5080 // have done something wrong!) 5081 dlen = len; 5082 vim_free(d); 5083 } 5084 return dlen; 5085 } 5086 5087 /* 5088 * Convert text "ptr[*lenp]" according to "vcp". 5089 * Returns the result in allocated memory and sets "*lenp". 5090 * When "lenp" is NULL, use NUL terminated strings. 5091 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 5092 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 5093 */ 5094 char_u * 5095 string_convert( 5096 vimconv_T *vcp, 5097 char_u *ptr, 5098 int *lenp) 5099 { 5100 return string_convert_ext(vcp, ptr, lenp, NULL); 5101 } 5102 5103 /* 5104 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 5105 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 5106 * set to the number of remaining bytes. 5107 */ 5108 char_u * 5109 string_convert_ext( 5110 vimconv_T *vcp, 5111 char_u *ptr, 5112 int *lenp, 5113 int *unconvlenp) 5114 { 5115 char_u *retval = NULL; 5116 char_u *d; 5117 int len; 5118 int i; 5119 int l; 5120 int c; 5121 5122 if (lenp == NULL) 5123 len = (int)STRLEN(ptr); 5124 else 5125 len = *lenp; 5126 if (len == 0) 5127 return vim_strsave((char_u *)""); 5128 5129 switch (vcp->vc_type) 5130 { 5131 case CONV_TO_UTF8: // latin1 to utf-8 conversion 5132 retval = alloc(len * 2 + 1); 5133 if (retval == NULL) 5134 break; 5135 d = retval; 5136 for (i = 0; i < len; ++i) 5137 { 5138 c = ptr[i]; 5139 if (c < 0x80) 5140 *d++ = c; 5141 else 5142 { 5143 *d++ = 0xc0 + ((unsigned)c >> 6); 5144 *d++ = 0x80 + (c & 0x3f); 5145 } 5146 } 5147 *d = NUL; 5148 if (lenp != NULL) 5149 *lenp = (int)(d - retval); 5150 break; 5151 5152 case CONV_9_TO_UTF8: // latin9 to utf-8 conversion 5153 retval = alloc(len * 3 + 1); 5154 if (retval == NULL) 5155 break; 5156 d = retval; 5157 for (i = 0; i < len; ++i) 5158 { 5159 c = ptr[i]; 5160 switch (c) 5161 { 5162 case 0xa4: c = 0x20ac; break; // euro 5163 case 0xa6: c = 0x0160; break; // S hat 5164 case 0xa8: c = 0x0161; break; // S -hat 5165 case 0xb4: c = 0x017d; break; // Z hat 5166 case 0xb8: c = 0x017e; break; // Z -hat 5167 case 0xbc: c = 0x0152; break; // OE 5168 case 0xbd: c = 0x0153; break; // oe 5169 case 0xbe: c = 0x0178; break; // Y 5170 } 5171 d += utf_char2bytes(c, d); 5172 } 5173 *d = NUL; 5174 if (lenp != NULL) 5175 *lenp = (int)(d - retval); 5176 break; 5177 5178 case CONV_TO_LATIN1: // utf-8 to latin1 conversion 5179 case CONV_TO_LATIN9: // utf-8 to latin9 conversion 5180 retval = alloc(len + 1); 5181 if (retval == NULL) 5182 break; 5183 d = retval; 5184 for (i = 0; i < len; ++i) 5185 { 5186 l = utf_ptr2len_len(ptr + i, len - i); 5187 if (l == 0) 5188 *d++ = NUL; 5189 else if (l == 1) 5190 { 5191 int l_w = utf8len_tab_zero[ptr[i]]; 5192 5193 if (l_w == 0) 5194 { 5195 // Illegal utf-8 byte cannot be converted 5196 vim_free(retval); 5197 return NULL; 5198 } 5199 if (unconvlenp != NULL && l_w > len - i) 5200 { 5201 // Incomplete sequence at the end. 5202 *unconvlenp = len - i; 5203 break; 5204 } 5205 *d++ = ptr[i]; 5206 } 5207 else 5208 { 5209 c = utf_ptr2char(ptr + i); 5210 if (vcp->vc_type == CONV_TO_LATIN9) 5211 switch (c) 5212 { 5213 case 0x20ac: c = 0xa4; break; // euro 5214 case 0x0160: c = 0xa6; break; // S hat 5215 case 0x0161: c = 0xa8; break; // S -hat 5216 case 0x017d: c = 0xb4; break; // Z hat 5217 case 0x017e: c = 0xb8; break; // Z -hat 5218 case 0x0152: c = 0xbc; break; // OE 5219 case 0x0153: c = 0xbd; break; // oe 5220 case 0x0178: c = 0xbe; break; // Y 5221 case 0xa4: 5222 case 0xa6: 5223 case 0xa8: 5224 case 0xb4: 5225 case 0xb8: 5226 case 0xbc: 5227 case 0xbd: 5228 case 0xbe: c = 0x100; break; // not in latin9 5229 } 5230 if (!utf_iscomposing(c)) // skip composing chars 5231 { 5232 if (c < 0x100) 5233 *d++ = c; 5234 else if (vcp->vc_fail) 5235 { 5236 vim_free(retval); 5237 return NULL; 5238 } 5239 else 5240 { 5241 *d++ = 0xbf; 5242 if (utf_char2cells(c) > 1) 5243 *d++ = '?'; 5244 } 5245 } 5246 i += l - 1; 5247 } 5248 } 5249 *d = NUL; 5250 if (lenp != NULL) 5251 *lenp = (int)(d - retval); 5252 break; 5253 5254 # ifdef MACOS_CONVERT 5255 case CONV_MAC_LATIN1: 5256 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 5257 'm', 'l', unconvlenp); 5258 break; 5259 5260 case CONV_LATIN1_MAC: 5261 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 5262 'l', 'm', unconvlenp); 5263 break; 5264 5265 case CONV_MAC_UTF8: 5266 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 5267 'm', 'u', unconvlenp); 5268 break; 5269 5270 case CONV_UTF8_MAC: 5271 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 5272 'u', 'm', unconvlenp); 5273 break; 5274 # endif 5275 5276 # ifdef USE_ICONV 5277 case CONV_ICONV: // conversion with output_conv.vc_fd 5278 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 5279 break; 5280 # endif 5281 # ifdef MSWIN 5282 case CONV_CODEPAGE: // codepage -> codepage 5283 { 5284 int retlen; 5285 int tmp_len; 5286 short_u *tmp; 5287 5288 // 1. codepage/UTF-8 -> ucs-2. 5289 if (vcp->vc_cpfrom == 0) 5290 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 5291 else 5292 { 5293 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 5294 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 5295 (char *)ptr, len, 0, 0); 5296 if (tmp_len == 0 5297 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 5298 { 5299 if (lenp != NULL) 5300 *lenp = 0; 5301 if (unconvlenp != NULL) 5302 *unconvlenp = len; 5303 retval = alloc(1); 5304 if (retval) 5305 retval[0] = NUL; 5306 return retval; 5307 } 5308 } 5309 tmp = ALLOC_MULT(short_u, tmp_len); 5310 if (tmp == NULL) 5311 break; 5312 if (vcp->vc_cpfrom == 0) 5313 utf8_to_utf16(ptr, len, tmp, unconvlenp); 5314 else 5315 MultiByteToWideChar(vcp->vc_cpfrom, 0, 5316 (char *)ptr, len, tmp, tmp_len); 5317 5318 // 2. ucs-2 -> codepage/UTF-8. 5319 if (vcp->vc_cpto == 0) 5320 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 5321 else 5322 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 5323 tmp, tmp_len, 0, 0, 0, 0); 5324 retval = alloc(retlen + 1); 5325 if (retval != NULL) 5326 { 5327 if (vcp->vc_cpto == 0) 5328 utf16_to_utf8(tmp, tmp_len, retval); 5329 else 5330 WideCharToMultiByte(vcp->vc_cpto, 0, 5331 tmp, tmp_len, 5332 (char *)retval, retlen, 0, 0); 5333 retval[retlen] = NUL; 5334 if (lenp != NULL) 5335 *lenp = retlen; 5336 } 5337 vim_free(tmp); 5338 break; 5339 } 5340 # endif 5341 } 5342 5343 return retval; 5344 } 5345