1 /* vi:set ts=8 sts=4 sw=4 noet: 2 * 3 * VIM - Vi IMproved by Bram Moolenaar 4 * Multibyte extensions partly by Sung-Hoon Baek 5 * 6 * Do ":help uganda" in Vim to read copying and usage conditions. 7 * Do ":help credits" in Vim to see a list of people who contributed. 8 * See README.txt for an overview of the Vim source code. 9 */ 10 /* 11 * mbyte.c: Code specifically for handling multi-byte characters. 12 * 13 * The encoding used in the core is set with 'encoding'. When 'encoding' is 14 * changed, the following four variables are set (for speed). 15 * Currently these types of character encodings are supported: 16 * 17 * "enc_dbcs" When non-zero it tells the type of double byte character 18 * encoding (Chinese, Korean, Japanese, etc.). 19 * The cell width on the display is equal to the number of 20 * bytes. (exception: DBCS_JPNU with first byte 0x8e) 21 * Recognizing the first or second byte is difficult, it 22 * requires checking a byte sequence from the start. 23 * "enc_utf8" When TRUE use Unicode characters in UTF-8 encoding. 24 * The cell width on the display needs to be determined from 25 * the character value. 26 * Recognizing bytes is easy: 0xxx.xxxx is a single-byte 27 * char, 10xx.xxxx is a trailing byte, 11xx.xxxx is a leading 28 * byte of a multi-byte character. 29 * To make things complicated, up to six composing characters 30 * are allowed. These are drawn on top of the first char. 31 * For most editing the sequence of bytes with composing 32 * characters included is considered to be one character. 33 * "enc_unicode" When 2 use 16-bit Unicode characters (or UTF-16). 34 * When 4 use 32-but Unicode characters. 35 * Internally characters are stored in UTF-8 encoding to 36 * avoid NUL bytes. Conversion happens when doing I/O. 37 * "enc_utf8" will also be TRUE. 38 * 39 * "has_mbyte" is set when "enc_dbcs" or "enc_utf8" is non-zero. 40 * 41 * If none of these is TRUE, 8-bit bytes are used for a character. The 42 * encoding isn't currently specified (TODO). 43 * 44 * 'encoding' specifies the encoding used in the core. This is in registers, 45 * text manipulation, buffers, etc. Conversion has to be done when characters 46 * in another encoding are received or send: 47 * 48 * clipboard 49 * ^ 50 * | (2) 51 * V 52 * +---------------+ 53 * (1) | | (3) 54 * keyboard ----->| core |-----> display 55 * | | 56 * +---------------+ 57 * ^ 58 * | (4) 59 * V 60 * file 61 * 62 * (1) Typed characters arrive in the current locale. Conversion is to be 63 * done when 'encoding' is different from 'termencoding'. 64 * (2) Text will be made available with the encoding specified with 65 * 'encoding'. If this is not sufficient, system-specific conversion 66 * might be required. 67 * (3) For the GUI the correct font must be selected, no conversion done. 68 * Otherwise, conversion is to be done when 'encoding' differs from 69 * 'termencoding'. (Different in the GTK+ 2 port -- 'termencoding' 70 * is always used for both input and output and must always be set to 71 * "utf-8". gui_mch_init() does this automatically.) 72 * (4) The encoding of the file is specified with 'fileencoding'. Conversion 73 * is to be done when it's different from 'encoding'. 74 * 75 * The viminfo file is a special case: Only text is converted, not file names. 76 * Vim scripts may contain an ":encoding" command. This has an effect for 77 * some commands, like ":menutrans" 78 */ 79 80 #include "vim.h" 81 82 #ifdef WIN32UNIX 83 # ifndef WIN32_LEAN_AND_MEAN 84 # define WIN32_LEAN_AND_MEAN 85 # endif 86 # if defined(FEAT_GUI) || defined(FEAT_XCLIPBOARD) 87 # include <X11/Xwindows.h> 88 # define WINBYTE wBYTE 89 # else 90 # include <windows.h> 91 # define WINBYTE BYTE 92 # endif 93 # ifdef WIN32 94 # undef WIN32 // Some windows.h define WIN32, we don't want that here. 95 # endif 96 #else 97 # define WINBYTE BYTE 98 #endif 99 100 #if (defined(MSWIN) || defined(WIN32UNIX)) && !defined(__MINGW32__) 101 # include <winnls.h> 102 #endif 103 104 #ifdef FEAT_GUI_X11 105 # include <X11/Intrinsic.h> 106 #endif 107 #ifdef X_LOCALE 108 # include <X11/Xlocale.h> 109 # if !defined(HAVE_MBLEN) && !defined(mblen) 110 # define mblen _Xmblen 111 # endif 112 #endif 113 114 #ifdef HAVE_WCHAR_H 115 # include <wchar.h> 116 #endif 117 118 #if 0 119 // This has been disabled, because several people reported problems with the 120 // wcwidth() and iswprint() library functions, esp. for Hebrew. 121 # ifdef __STDC_ISO_10646__ 122 # define USE_WCHAR_FUNCTIONS 123 # endif 124 #endif 125 126 static int dbcs_char2len(int c); 127 static int dbcs_char2bytes(int c, char_u *buf); 128 static int dbcs_ptr2len(char_u *p); 129 static int dbcs_ptr2len_len(char_u *p, int size); 130 static int utf_ptr2cells_len(char_u *p, int size); 131 static int dbcs_char2cells(int c); 132 static int dbcs_ptr2cells_len(char_u *p, int size); 133 static int dbcs_ptr2char(char_u *p); 134 static int dbcs_head_off(char_u *base, char_u *p); 135 #ifdef FEAT_EVAL 136 static int cw_value(int c); 137 #endif 138 139 /* 140 * Lookup table to quickly get the length in bytes of a UTF-8 character from 141 * the first byte of a UTF-8 string. 142 * Bytes which are illegal when used as the first byte have a 1. 143 * The NUL byte has length 1. 144 */ 145 static char utf8len_tab[256] = 146 { 147 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 148 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 149 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 150 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 151 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 152 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 153 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, 154 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6,1,1, 155 }; 156 157 /* 158 * Like utf8len_tab above, but using a zero for illegal lead bytes. 159 */ 160 static char utf8len_tab_zero[256] = 161 { 162 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 163 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 164 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 165 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 166 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 167 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 168 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, 169 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6,0,0, 170 }; 171 172 173 /* 174 * Canonical encoding names and their properties. 175 * "iso-8859-n" is handled by enc_canonize() directly. 176 */ 177 static struct 178 { char *name; int prop; int codepage;} 179 enc_canon_table[] = 180 { 181 #define IDX_LATIN_1 0 182 {"latin1", ENC_8BIT + ENC_LATIN1, 1252}, 183 #define IDX_ISO_2 1 184 {"iso-8859-2", ENC_8BIT, 0}, 185 #define IDX_ISO_3 2 186 {"iso-8859-3", ENC_8BIT, 0}, 187 #define IDX_ISO_4 3 188 {"iso-8859-4", ENC_8BIT, 0}, 189 #define IDX_ISO_5 4 190 {"iso-8859-5", ENC_8BIT, 0}, 191 #define IDX_ISO_6 5 192 {"iso-8859-6", ENC_8BIT, 0}, 193 #define IDX_ISO_7 6 194 {"iso-8859-7", ENC_8BIT, 0}, 195 #define IDX_ISO_8 7 196 {"iso-8859-8", ENC_8BIT, 0}, 197 #define IDX_ISO_9 8 198 {"iso-8859-9", ENC_8BIT, 0}, 199 #define IDX_ISO_10 9 200 {"iso-8859-10", ENC_8BIT, 0}, 201 #define IDX_ISO_11 10 202 {"iso-8859-11", ENC_8BIT, 0}, 203 #define IDX_ISO_13 11 204 {"iso-8859-13", ENC_8BIT, 0}, 205 #define IDX_ISO_14 12 206 {"iso-8859-14", ENC_8BIT, 0}, 207 #define IDX_ISO_15 13 208 {"iso-8859-15", ENC_8BIT + ENC_LATIN9, 0}, 209 #define IDX_KOI8_R 14 210 {"koi8-r", ENC_8BIT, 0}, 211 #define IDX_KOI8_U 15 212 {"koi8-u", ENC_8BIT, 0}, 213 #define IDX_UTF8 16 214 {"utf-8", ENC_UNICODE, 0}, 215 #define IDX_UCS2 17 216 {"ucs-2", ENC_UNICODE + ENC_ENDIAN_B + ENC_2BYTE, 0}, 217 #define IDX_UCS2LE 18 218 {"ucs-2le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2BYTE, 0}, 219 #define IDX_UTF16 19 220 {"utf-16", ENC_UNICODE + ENC_ENDIAN_B + ENC_2WORD, 0}, 221 #define IDX_UTF16LE 20 222 {"utf-16le", ENC_UNICODE + ENC_ENDIAN_L + ENC_2WORD, 0}, 223 #define IDX_UCS4 21 224 {"ucs-4", ENC_UNICODE + ENC_ENDIAN_B + ENC_4BYTE, 0}, 225 #define IDX_UCS4LE 22 226 {"ucs-4le", ENC_UNICODE + ENC_ENDIAN_L + ENC_4BYTE, 0}, 227 228 // For debugging DBCS encoding on Unix. 229 #define IDX_DEBUG 23 230 {"debug", ENC_DBCS, DBCS_DEBUG}, 231 #define IDX_EUC_JP 24 232 {"euc-jp", ENC_DBCS, DBCS_JPNU}, 233 #define IDX_SJIS 25 234 {"sjis", ENC_DBCS, DBCS_JPN}, 235 #define IDX_EUC_KR 26 236 {"euc-kr", ENC_DBCS, DBCS_KORU}, 237 #define IDX_EUC_CN 27 238 {"euc-cn", ENC_DBCS, DBCS_CHSU}, 239 #define IDX_EUC_TW 28 240 {"euc-tw", ENC_DBCS, DBCS_CHTU}, 241 #define IDX_BIG5 29 242 {"big5", ENC_DBCS, DBCS_CHT}, 243 244 // MS-DOS and MS-Windows codepages are included here, so that they can be 245 // used on Unix too. Most of them are similar to ISO-8859 encodings, but 246 // not exactly the same. 247 #define IDX_CP437 30 248 {"cp437", ENC_8BIT, 437}, // like iso-8859-1 249 #define IDX_CP737 31 250 {"cp737", ENC_8BIT, 737}, // like iso-8859-7 251 #define IDX_CP775 32 252 {"cp775", ENC_8BIT, 775}, // Baltic 253 #define IDX_CP850 33 254 {"cp850", ENC_8BIT, 850}, // like iso-8859-4 255 #define IDX_CP852 34 256 {"cp852", ENC_8BIT, 852}, // like iso-8859-1 257 #define IDX_CP855 35 258 {"cp855", ENC_8BIT, 855}, // like iso-8859-2 259 #define IDX_CP857 36 260 {"cp857", ENC_8BIT, 857}, // like iso-8859-5 261 #define IDX_CP860 37 262 {"cp860", ENC_8BIT, 860}, // like iso-8859-9 263 #define IDX_CP861 38 264 {"cp861", ENC_8BIT, 861}, // like iso-8859-1 265 #define IDX_CP862 39 266 {"cp862", ENC_8BIT, 862}, // like iso-8859-1 267 #define IDX_CP863 40 268 {"cp863", ENC_8BIT, 863}, // like iso-8859-8 269 #define IDX_CP865 41 270 {"cp865", ENC_8BIT, 865}, // like iso-8859-1 271 #define IDX_CP866 42 272 {"cp866", ENC_8BIT, 866}, // like iso-8859-5 273 #define IDX_CP869 43 274 {"cp869", ENC_8BIT, 869}, // like iso-8859-7 275 #define IDX_CP874 44 276 {"cp874", ENC_8BIT, 874}, // Thai 277 #define IDX_CP932 45 278 {"cp932", ENC_DBCS, DBCS_JPN}, 279 #define IDX_CP936 46 280 {"cp936", ENC_DBCS, DBCS_CHS}, 281 #define IDX_CP949 47 282 {"cp949", ENC_DBCS, DBCS_KOR}, 283 #define IDX_CP950 48 284 {"cp950", ENC_DBCS, DBCS_CHT}, 285 #define IDX_CP1250 49 286 {"cp1250", ENC_8BIT, 1250}, // Czech, Polish, etc. 287 #define IDX_CP1251 50 288 {"cp1251", ENC_8BIT, 1251}, // Cyrillic 289 // cp1252 is considered to be equal to latin1 290 #define IDX_CP1253 51 291 {"cp1253", ENC_8BIT, 1253}, // Greek 292 #define IDX_CP1254 52 293 {"cp1254", ENC_8BIT, 1254}, // Turkish 294 #define IDX_CP1255 53 295 {"cp1255", ENC_8BIT, 1255}, // Hebrew 296 #define IDX_CP1256 54 297 {"cp1256", ENC_8BIT, 1256}, // Arabic 298 #define IDX_CP1257 55 299 {"cp1257", ENC_8BIT, 1257}, // Baltic 300 #define IDX_CP1258 56 301 {"cp1258", ENC_8BIT, 1258}, // Vietnamese 302 303 #define IDX_MACROMAN 57 304 {"macroman", ENC_8BIT + ENC_MACROMAN, 0}, // Mac OS 305 #define IDX_DECMCS 58 306 {"dec-mcs", ENC_8BIT, 0}, // DEC MCS 307 #define IDX_HPROMAN8 59 308 {"hp-roman8", ENC_8BIT, 0}, // HP Roman8 309 #define IDX_COUNT 60 310 }; 311 312 /* 313 * Aliases for encoding names. 314 */ 315 static struct 316 { char *name; int canon;} 317 enc_alias_table[] = 318 { 319 {"ansi", IDX_LATIN_1}, 320 {"iso-8859-1", IDX_LATIN_1}, 321 {"latin2", IDX_ISO_2}, 322 {"latin3", IDX_ISO_3}, 323 {"latin4", IDX_ISO_4}, 324 {"cyrillic", IDX_ISO_5}, 325 {"arabic", IDX_ISO_6}, 326 {"greek", IDX_ISO_7}, 327 #ifdef MSWIN 328 {"hebrew", IDX_CP1255}, 329 #else 330 {"hebrew", IDX_ISO_8}, 331 #endif 332 {"latin5", IDX_ISO_9}, 333 {"turkish", IDX_ISO_9}, // ? 334 {"latin6", IDX_ISO_10}, 335 {"nordic", IDX_ISO_10}, // ? 336 {"thai", IDX_ISO_11}, // ? 337 {"latin7", IDX_ISO_13}, 338 {"latin8", IDX_ISO_14}, 339 {"latin9", IDX_ISO_15}, 340 {"utf8", IDX_UTF8}, 341 {"unicode", IDX_UCS2}, 342 {"ucs2", IDX_UCS2}, 343 {"ucs2be", IDX_UCS2}, 344 {"ucs-2be", IDX_UCS2}, 345 {"ucs2le", IDX_UCS2LE}, 346 {"utf16", IDX_UTF16}, 347 {"utf16be", IDX_UTF16}, 348 {"utf-16be", IDX_UTF16}, 349 {"utf16le", IDX_UTF16LE}, 350 {"ucs4", IDX_UCS4}, 351 {"ucs4be", IDX_UCS4}, 352 {"ucs-4be", IDX_UCS4}, 353 {"ucs4le", IDX_UCS4LE}, 354 {"utf32", IDX_UCS4}, 355 {"utf-32", IDX_UCS4}, 356 {"utf32be", IDX_UCS4}, 357 {"utf-32be", IDX_UCS4}, 358 {"utf32le", IDX_UCS4LE}, 359 {"utf-32le", IDX_UCS4LE}, 360 {"932", IDX_CP932}, 361 {"949", IDX_CP949}, 362 {"936", IDX_CP936}, 363 {"gbk", IDX_CP936}, 364 {"950", IDX_CP950}, 365 {"eucjp", IDX_EUC_JP}, 366 {"unix-jis", IDX_EUC_JP}, 367 {"ujis", IDX_EUC_JP}, 368 {"shift-jis", IDX_SJIS}, 369 {"pck", IDX_SJIS}, // Sun: PCK 370 {"euckr", IDX_EUC_KR}, 371 {"5601", IDX_EUC_KR}, // Sun: KS C 5601 372 {"euccn", IDX_EUC_CN}, 373 {"gb2312", IDX_EUC_CN}, 374 {"euctw", IDX_EUC_TW}, 375 #if defined(MSWIN) || defined(WIN32UNIX) || defined(MACOS_X) 376 {"japan", IDX_CP932}, 377 {"korea", IDX_CP949}, 378 {"prc", IDX_CP936}, 379 {"chinese", IDX_CP936}, 380 {"taiwan", IDX_CP950}, 381 {"big5", IDX_CP950}, 382 #else 383 {"japan", IDX_EUC_JP}, 384 {"korea", IDX_EUC_KR}, 385 {"prc", IDX_EUC_CN}, 386 {"chinese", IDX_EUC_CN}, 387 {"taiwan", IDX_EUC_TW}, 388 {"cp950", IDX_BIG5}, 389 {"950", IDX_BIG5}, 390 #endif 391 {"mac", IDX_MACROMAN}, 392 {"mac-roman", IDX_MACROMAN}, 393 {NULL, 0} 394 }; 395 396 #ifndef CP_UTF8 397 # define CP_UTF8 65001 // magic number from winnls.h 398 #endif 399 400 /* 401 * Find encoding "name" in the list of canonical encoding names. 402 * Returns -1 if not found. 403 */ 404 static int 405 enc_canon_search(char_u *name) 406 { 407 int i; 408 409 for (i = 0; i < IDX_COUNT; ++i) 410 if (STRCMP(name, enc_canon_table[i].name) == 0) 411 return i; 412 return -1; 413 } 414 415 416 /* 417 * Find canonical encoding "name" in the list and return its properties. 418 * Returns 0 if not found. 419 */ 420 int 421 enc_canon_props(char_u *name) 422 { 423 int i; 424 425 i = enc_canon_search(name); 426 if (i >= 0) 427 return enc_canon_table[i].prop; 428 #ifdef MSWIN 429 if (name[0] == 'c' && name[1] == 'p' && VIM_ISDIGIT(name[2])) 430 { 431 CPINFO cpinfo; 432 433 // Get info on this codepage to find out what it is. 434 if (GetCPInfo(atoi((char *)name + 2), &cpinfo) != 0) 435 { 436 if (cpinfo.MaxCharSize == 1) // some single-byte encoding 437 return ENC_8BIT; 438 if (cpinfo.MaxCharSize == 2 439 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 440 // must be a DBCS encoding 441 return ENC_DBCS; 442 } 443 return 0; 444 } 445 #endif 446 if (STRNCMP(name, "2byte-", 6) == 0) 447 return ENC_DBCS; 448 if (STRNCMP(name, "8bit-", 5) == 0 || STRNCMP(name, "iso-8859-", 9) == 0) 449 return ENC_8BIT; 450 return 0; 451 } 452 453 /* 454 * Set up for using multi-byte characters. 455 * Called in three cases: 456 * - by main() to initialize (p_enc == NULL) 457 * - by set_init_1() after 'encoding' was set to its default. 458 * - by do_set() when 'encoding' has been set. 459 * p_enc must have been passed through enc_canonize() already. 460 * Sets the "enc_unicode", "enc_utf8", "enc_dbcs" and "has_mbyte" flags. 461 * Fills mb_bytelen_tab[] and returns NULL when there are no problems. 462 * When there is something wrong: Returns an error message and doesn't change 463 * anything. 464 */ 465 char * 466 mb_init(void) 467 { 468 int i; 469 int idx; 470 int n; 471 int enc_dbcs_new = 0; 472 #if defined(USE_ICONV) && !defined(MSWIN) && !defined(WIN32UNIX) \ 473 && !defined(MACOS_CONVERT) 474 # define LEN_FROM_CONV 475 vimconv_T vimconv; 476 char_u *p; 477 #endif 478 479 if (p_enc == NULL) 480 { 481 // Just starting up: set the whole table to one's. 482 for (i = 0; i < 256; ++i) 483 mb_bytelen_tab[i] = 1; 484 input_conv.vc_type = CONV_NONE; 485 input_conv.vc_factor = 1; 486 output_conv.vc_type = CONV_NONE; 487 return NULL; 488 } 489 490 #ifdef MSWIN 491 if (p_enc[0] == 'c' && p_enc[1] == 'p' && VIM_ISDIGIT(p_enc[2])) 492 { 493 CPINFO cpinfo; 494 495 // Get info on this codepage to find out what it is. 496 if (GetCPInfo(atoi((char *)p_enc + 2), &cpinfo) != 0) 497 { 498 if (cpinfo.MaxCharSize == 1) 499 { 500 // some single-byte encoding 501 enc_unicode = 0; 502 enc_utf8 = FALSE; 503 } 504 else if (cpinfo.MaxCharSize == 2 505 && (cpinfo.LeadByte[0] != 0 || cpinfo.LeadByte[1] != 0)) 506 { 507 // must be a DBCS encoding, check below 508 enc_dbcs_new = atoi((char *)p_enc + 2); 509 } 510 else 511 goto codepage_invalid; 512 } 513 else if (GetLastError() == ERROR_INVALID_PARAMETER) 514 { 515 codepage_invalid: 516 return N_("E543: Not a valid codepage"); 517 } 518 } 519 #endif 520 else if (STRNCMP(p_enc, "8bit-", 5) == 0 521 || STRNCMP(p_enc, "iso-8859-", 9) == 0) 522 { 523 // Accept any "8bit-" or "iso-8859-" name. 524 enc_unicode = 0; 525 enc_utf8 = FALSE; 526 } 527 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 528 { 529 #ifdef MSWIN 530 // Windows: accept only valid codepage numbers, check below. 531 if (p_enc[6] != 'c' || p_enc[7] != 'p' 532 || (enc_dbcs_new = atoi((char *)p_enc + 8)) == 0) 533 return e_invarg; 534 #else 535 // Unix: accept any "2byte-" name, assume current locale. 536 enc_dbcs_new = DBCS_2BYTE; 537 #endif 538 } 539 else if ((idx = enc_canon_search(p_enc)) >= 0) 540 { 541 i = enc_canon_table[idx].prop; 542 if (i & ENC_UNICODE) 543 { 544 // Unicode 545 enc_utf8 = TRUE; 546 if (i & (ENC_2BYTE | ENC_2WORD)) 547 enc_unicode = 2; 548 else if (i & ENC_4BYTE) 549 enc_unicode = 4; 550 else 551 enc_unicode = 0; 552 } 553 else if (i & ENC_DBCS) 554 { 555 // 2byte, handle below 556 enc_dbcs_new = enc_canon_table[idx].codepage; 557 } 558 else 559 { 560 // Must be 8-bit. 561 enc_unicode = 0; 562 enc_utf8 = FALSE; 563 } 564 } 565 else // Don't know what encoding this is, reject it. 566 return e_invarg; 567 568 if (enc_dbcs_new != 0) 569 { 570 #ifdef MSWIN 571 // Check if the DBCS code page is OK. 572 if (!IsValidCodePage(enc_dbcs_new)) 573 goto codepage_invalid; 574 #endif 575 enc_unicode = 0; 576 enc_utf8 = FALSE; 577 } 578 enc_dbcs = enc_dbcs_new; 579 has_mbyte = (enc_dbcs != 0 || enc_utf8); 580 581 #if defined(MSWIN) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 582 enc_codepage = encname2codepage(p_enc); 583 enc_latin9 = (STRCMP(p_enc, "iso-8859-15") == 0); 584 #endif 585 586 // Detect an encoding that uses latin1 characters. 587 enc_latin1like = (enc_utf8 || STRCMP(p_enc, "latin1") == 0 588 || STRCMP(p_enc, "iso-8859-15") == 0); 589 590 /* 591 * Set the function pointers. 592 */ 593 if (enc_utf8) 594 { 595 mb_ptr2len = utfc_ptr2len; 596 mb_ptr2len_len = utfc_ptr2len_len; 597 mb_char2len = utf_char2len; 598 mb_char2bytes = utf_char2bytes; 599 mb_ptr2cells = utf_ptr2cells; 600 mb_ptr2cells_len = utf_ptr2cells_len; 601 mb_char2cells = utf_char2cells; 602 mb_off2cells = utf_off2cells; 603 mb_ptr2char = utf_ptr2char; 604 mb_head_off = utf_head_off; 605 } 606 else if (enc_dbcs != 0) 607 { 608 mb_ptr2len = dbcs_ptr2len; 609 mb_ptr2len_len = dbcs_ptr2len_len; 610 mb_char2len = dbcs_char2len; 611 mb_char2bytes = dbcs_char2bytes; 612 mb_ptr2cells = dbcs_ptr2cells; 613 mb_ptr2cells_len = dbcs_ptr2cells_len; 614 mb_char2cells = dbcs_char2cells; 615 mb_off2cells = dbcs_off2cells; 616 mb_ptr2char = dbcs_ptr2char; 617 mb_head_off = dbcs_head_off; 618 } 619 else 620 { 621 mb_ptr2len = latin_ptr2len; 622 mb_ptr2len_len = latin_ptr2len_len; 623 mb_char2len = latin_char2len; 624 mb_char2bytes = latin_char2bytes; 625 mb_ptr2cells = latin_ptr2cells; 626 mb_ptr2cells_len = latin_ptr2cells_len; 627 mb_char2cells = latin_char2cells; 628 mb_off2cells = latin_off2cells; 629 mb_ptr2char = latin_ptr2char; 630 mb_head_off = latin_head_off; 631 } 632 633 /* 634 * Fill the mb_bytelen_tab[] for MB_BYTE2LEN(). 635 */ 636 #ifdef LEN_FROM_CONV 637 // When 'encoding' is different from the current locale mblen() won't 638 // work. Use conversion to "utf-8" instead. 639 vimconv.vc_type = CONV_NONE; 640 if (enc_dbcs) 641 { 642 p = enc_locale(); 643 if (p == NULL || STRCMP(p, p_enc) != 0) 644 { 645 convert_setup(&vimconv, p_enc, (char_u *)"utf-8"); 646 vimconv.vc_fail = TRUE; 647 } 648 vim_free(p); 649 } 650 #endif 651 652 for (i = 0; i < 256; ++i) 653 { 654 // Our own function to reliably check the length of UTF-8 characters, 655 // independent of mblen(). 656 if (enc_utf8) 657 n = utf8len_tab[i]; 658 else if (enc_dbcs == 0) 659 n = 1; 660 else 661 { 662 #if defined(MSWIN) || defined(WIN32UNIX) 663 // enc_dbcs is set by setting 'fileencoding'. It becomes a Windows 664 // CodePage identifier, which we can pass directly in to Windows 665 // API 666 n = IsDBCSLeadByteEx(enc_dbcs, (WINBYTE)i) ? 2 : 1; 667 #else 668 # if defined(__amigaos4__) || defined(__ANDROID__) || \ 669 !(defined(HAVE_MBLEN) || defined(X_LOCALE)) 670 /* 671 * if mblen() is not available, character which MSB is turned on 672 * are treated as leading byte character. (note : This assumption 673 * is not always true.) 674 */ 675 n = (i & 0x80) ? 2 : 1; 676 # else 677 char buf[MB_MAXBYTES + 1]; 678 679 if (i == NUL) // just in case mblen() can't handle "" 680 n = 1; 681 else 682 { 683 buf[0] = i; 684 buf[1] = 0; 685 # ifdef LEN_FROM_CONV 686 if (vimconv.vc_type != CONV_NONE) 687 { 688 /* 689 * string_convert() should fail when converting the first 690 * byte of a double-byte character. 691 */ 692 p = string_convert(&vimconv, (char_u *)buf, NULL); 693 if (p != NULL) 694 { 695 vim_free(p); 696 n = 1; 697 } 698 else 699 n = 2; 700 } 701 else 702 # endif 703 { 704 /* 705 * mblen() should return -1 for invalid (means the leading 706 * multibyte) character. However there are some platforms 707 * where mblen() returns 0 for invalid character. 708 * Therefore, following condition includes 0. 709 */ 710 vim_ignored = mblen(NULL, 0); // First reset the state. 711 if (mblen(buf, (size_t)1) <= 0) 712 n = 2; 713 else 714 n = 1; 715 } 716 } 717 # endif 718 #endif 719 } 720 721 mb_bytelen_tab[i] = n; 722 } 723 724 #ifdef LEN_FROM_CONV 725 convert_setup(&vimconv, NULL, NULL); 726 #endif 727 728 // The cell width depends on the type of multi-byte characters. 729 (void)init_chartab(); 730 731 // When enc_utf8 is set or reset, (de)allocate ScreenLinesUC[] 732 screenalloc(FALSE); 733 734 // When using Unicode, set default for 'fileencodings'. 735 if (enc_utf8 && !option_was_set((char_u *)"fencs")) 736 set_fencs_unicode(); 737 738 #if defined(HAVE_BIND_TEXTDOMAIN_CODESET) && defined(FEAT_GETTEXT) 739 // GNU gettext 0.10.37 supports this feature: set the codeset used for 740 // translated messages independently from the current locale. 741 (void)bind_textdomain_codeset(VIMPACKAGE, 742 enc_utf8 ? "utf-8" : (char *)p_enc); 743 #endif 744 745 #ifdef MSWIN 746 // When changing 'encoding' while starting up, then convert the command 747 // line arguments from the active codepage to 'encoding'. 748 if (starting != 0) 749 fix_arg_enc(); 750 #endif 751 752 // Fire an autocommand to let people do custom font setup. This must be 753 // after Vim has been setup for the new encoding. 754 apply_autocmds(EVENT_ENCODINGCHANGED, NULL, (char_u *)"", FALSE, curbuf); 755 756 #ifdef FEAT_SPELL 757 // Need to reload spell dictionaries 758 spell_reload(); 759 #endif 760 761 return NULL; 762 } 763 764 /* 765 * Return the size of the BOM for the current buffer: 766 * 0 - no BOM 767 * 2 - UCS-2 or UTF-16 BOM 768 * 4 - UCS-4 BOM 769 * 3 - UTF-8 BOM 770 */ 771 int 772 bomb_size(void) 773 { 774 int n = 0; 775 776 if (curbuf->b_p_bomb && !curbuf->b_p_bin) 777 { 778 if (*curbuf->b_p_fenc == NUL) 779 { 780 if (enc_utf8) 781 { 782 if (enc_unicode != 0) 783 n = enc_unicode; 784 else 785 n = 3; 786 } 787 } 788 else if (STRCMP(curbuf->b_p_fenc, "utf-8") == 0) 789 n = 3; 790 else if (STRNCMP(curbuf->b_p_fenc, "ucs-2", 5) == 0 791 || STRNCMP(curbuf->b_p_fenc, "utf-16", 6) == 0) 792 n = 2; 793 else if (STRNCMP(curbuf->b_p_fenc, "ucs-4", 5) == 0) 794 n = 4; 795 } 796 return n; 797 } 798 799 #if defined(FEAT_QUICKFIX) || defined(PROTO) 800 /* 801 * Remove all BOM from "s" by moving remaining text. 802 */ 803 void 804 remove_bom(char_u *s) 805 { 806 if (enc_utf8) 807 { 808 char_u *p = s; 809 810 while ((p = vim_strbyte(p, 0xef)) != NULL) 811 { 812 if (p[1] == 0xbb && p[2] == 0xbf) 813 STRMOVE(p, p + 3); 814 else 815 ++p; 816 } 817 } 818 } 819 #endif 820 821 /* 822 * Get class of pointer: 823 * 0 for blank or NUL 824 * 1 for punctuation 825 * 2 for an (ASCII) word character 826 * >2 for other word characters 827 */ 828 int 829 mb_get_class(char_u *p) 830 { 831 return mb_get_class_buf(p, curbuf); 832 } 833 834 int 835 mb_get_class_buf(char_u *p, buf_T *buf) 836 { 837 if (MB_BYTE2LEN(p[0]) == 1) 838 { 839 if (p[0] == NUL || VIM_ISWHITE(p[0])) 840 return 0; 841 if (vim_iswordc_buf(p[0], buf)) 842 return 2; 843 return 1; 844 } 845 if (enc_dbcs != 0 && p[0] != NUL && p[1] != NUL) 846 return dbcs_class(p[0], p[1]); 847 if (enc_utf8) 848 return utf_class_buf(utf_ptr2char(p), buf); 849 return 0; 850 } 851 852 /* 853 * Get class of a double-byte character. This always returns 3 or bigger. 854 * TODO: Should return 1 for punctuation. 855 */ 856 int 857 dbcs_class(unsigned lead, unsigned trail) 858 { 859 switch (enc_dbcs) 860 { 861 // please add classify routine for your language in here 862 863 case DBCS_JPNU: // ? 864 case DBCS_JPN: 865 { 866 // JIS code classification 867 unsigned char lb = lead; 868 unsigned char tb = trail; 869 870 // convert process code to JIS 871 # if defined(MSWIN) || defined(WIN32UNIX) || defined(MACOS_X) 872 // process code is SJIS 873 if (lb <= 0x9f) 874 lb = (lb - 0x81) * 2 + 0x21; 875 else 876 lb = (lb - 0xc1) * 2 + 0x21; 877 if (tb <= 0x7e) 878 tb -= 0x1f; 879 else if (tb <= 0x9e) 880 tb -= 0x20; 881 else 882 { 883 tb -= 0x7e; 884 lb += 1; 885 } 886 # else 887 /* 888 * XXX: Code page identification can not use with all 889 * system! So, some other encoding information 890 * will be needed. 891 * In japanese: SJIS,EUC,UNICODE,(JIS) 892 * Note that JIS-code system don't use as 893 * process code in most system because it uses 894 * escape sequences(JIS is context depend encoding). 895 */ 896 // assume process code is JAPANESE-EUC 897 lb &= 0x7f; 898 tb &= 0x7f; 899 # endif 900 // exceptions 901 switch (lb << 8 | tb) 902 { 903 case 0x2121: // ZENKAKU space 904 return 0; 905 case 0x2122: // TOU-TEN (Japanese comma) 906 case 0x2123: // KU-TEN (Japanese period) 907 case 0x2124: // ZENKAKU comma 908 case 0x2125: // ZENKAKU period 909 return 1; 910 case 0x213c: // prolongedsound handled as KATAKANA 911 return 13; 912 } 913 // sieved by KU code 914 switch (lb) 915 { 916 case 0x21: 917 case 0x22: 918 // special symbols 919 return 10; 920 case 0x23: 921 // alphanumeric 922 return 11; 923 case 0x24: 924 // hiragana 925 return 12; 926 case 0x25: 927 // katakana 928 return 13; 929 case 0x26: 930 // greek 931 return 14; 932 case 0x27: 933 // russian 934 return 15; 935 case 0x28: 936 // lines 937 return 16; 938 default: 939 // kanji 940 return 17; 941 } 942 } 943 944 case DBCS_KORU: // ? 945 case DBCS_KOR: 946 { 947 // KS code classification 948 unsigned char c1 = lead; 949 unsigned char c2 = trail; 950 951 /* 952 * 20 : Hangul 953 * 21 : Hanja 954 * 22 : Symbols 955 * 23 : Alphanumeric/Roman Letter (Full width) 956 * 24 : Hangul Letter(Alphabet) 957 * 25 : Roman Numeral/Greek Letter 958 * 26 : Box Drawings 959 * 27 : Unit Symbols 960 * 28 : Circled/Parenthesized Letter 961 * 29 : Hiragana/Katakana 962 * 30 : Cyrillic Letter 963 */ 964 965 if (c1 >= 0xB0 && c1 <= 0xC8) 966 // Hangul 967 return 20; 968 #if defined(MSWIN) || defined(WIN32UNIX) 969 else if (c1 <= 0xA0 || c2 <= 0xA0) 970 // Extended Hangul Region : MS UHC(Unified Hangul Code) 971 // c1: 0x81-0xA0 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xFE 972 // c1: 0xA1-0xC6 with c2: 0x41-0x5A, 0x61-0x7A, 0x81-0xA0 973 return 20; 974 #endif 975 976 else if (c1 >= 0xCA && c1 <= 0xFD) 977 // Hanja 978 return 21; 979 else switch (c1) 980 { 981 case 0xA1: 982 case 0xA2: 983 // Symbols 984 return 22; 985 case 0xA3: 986 // Alphanumeric 987 return 23; 988 case 0xA4: 989 // Hangul Letter(Alphabet) 990 return 24; 991 case 0xA5: 992 // Roman Numeral/Greek Letter 993 return 25; 994 case 0xA6: 995 // Box Drawings 996 return 26; 997 case 0xA7: 998 // Unit Symbols 999 return 27; 1000 case 0xA8: 1001 case 0xA9: 1002 if (c2 <= 0xAF) 1003 return 25; // Roman Letter 1004 else if (c2 >= 0xF6) 1005 return 22; // Symbols 1006 else 1007 // Circled/Parenthesized Letter 1008 return 28; 1009 case 0xAA: 1010 case 0xAB: 1011 // Hiragana/Katakana 1012 return 29; 1013 case 0xAC: 1014 // Cyrillic Letter 1015 return 30; 1016 } 1017 } 1018 default: 1019 break; 1020 } 1021 return 3; 1022 } 1023 1024 /* 1025 * mb_char2len() function pointer. 1026 * Return length in bytes of character "c". 1027 * Returns 1 for a single-byte character. 1028 */ 1029 int 1030 latin_char2len(int c UNUSED) 1031 { 1032 return 1; 1033 } 1034 1035 static int 1036 dbcs_char2len( 1037 int c) 1038 { 1039 if (c >= 0x100) 1040 return 2; 1041 return 1; 1042 } 1043 1044 /* 1045 * mb_char2bytes() function pointer. 1046 * Convert a character to its bytes. 1047 * Returns the length in bytes. 1048 */ 1049 int 1050 latin_char2bytes(int c, char_u *buf) 1051 { 1052 buf[0] = c; 1053 return 1; 1054 } 1055 1056 static int 1057 dbcs_char2bytes(int c, char_u *buf) 1058 { 1059 if (c >= 0x100) 1060 { 1061 buf[0] = (unsigned)c >> 8; 1062 buf[1] = c; 1063 // Never use a NUL byte, it causes lots of trouble. It's an invalid 1064 // character anyway. 1065 if (buf[1] == NUL) 1066 buf[1] = '\n'; 1067 return 2; 1068 } 1069 buf[0] = c; 1070 return 1; 1071 } 1072 1073 /* 1074 * mb_ptr2len() function pointer. 1075 * Get byte length of character at "*p" but stop at a NUL. 1076 * For UTF-8 this includes following composing characters. 1077 * Returns 0 when *p is NUL. 1078 */ 1079 int 1080 latin_ptr2len(char_u *p) 1081 { 1082 return MB_BYTE2LEN(*p); 1083 } 1084 1085 static int 1086 dbcs_ptr2len( 1087 char_u *p) 1088 { 1089 int len; 1090 1091 // Check if second byte is not missing. 1092 len = MB_BYTE2LEN(*p); 1093 if (len == 2 && p[1] == NUL) 1094 len = 1; 1095 return len; 1096 } 1097 1098 /* 1099 * mb_ptr2len_len() function pointer. 1100 * Like mb_ptr2len(), but limit to read "size" bytes. 1101 * Returns 0 for an empty string. 1102 * Returns 1 for an illegal char or an incomplete byte sequence. 1103 */ 1104 int 1105 latin_ptr2len_len(char_u *p, int size) 1106 { 1107 if (size < 1 || *p == NUL) 1108 return 0; 1109 return 1; 1110 } 1111 1112 static int 1113 dbcs_ptr2len_len(char_u *p, int size) 1114 { 1115 int len; 1116 1117 if (size < 1 || *p == NUL) 1118 return 0; 1119 if (size == 1) 1120 return 1; 1121 // Check that second byte is not missing. 1122 len = MB_BYTE2LEN(*p); 1123 if (len == 2 && p[1] == NUL) 1124 len = 1; 1125 return len; 1126 } 1127 1128 struct interval 1129 { 1130 long first; 1131 long last; 1132 }; 1133 1134 /* 1135 * Return TRUE if "c" is in "table[size / sizeof(struct interval)]". 1136 */ 1137 static int 1138 intable(struct interval *table, size_t size, int c) 1139 { 1140 int mid, bot, top; 1141 1142 // first quick check for Latin1 etc. characters 1143 if (c < table[0].first) 1144 return FALSE; 1145 1146 // binary search in table 1147 bot = 0; 1148 top = (int)(size / sizeof(struct interval) - 1); 1149 while (top >= bot) 1150 { 1151 mid = (bot + top) / 2; 1152 if (table[mid].last < c) 1153 bot = mid + 1; 1154 else if (table[mid].first > c) 1155 top = mid - 1; 1156 else 1157 return TRUE; 1158 } 1159 return FALSE; 1160 } 1161 1162 // Sorted list of non-overlapping intervals of East Asian Ambiguous 1163 // characters, generated with ../runtime/tools/unicode.vim. 1164 static struct interval ambiguous[] = 1165 { 1166 {0x00a1, 0x00a1}, 1167 {0x00a4, 0x00a4}, 1168 {0x00a7, 0x00a8}, 1169 {0x00aa, 0x00aa}, 1170 {0x00ad, 0x00ae}, 1171 {0x00b0, 0x00b4}, 1172 {0x00b6, 0x00ba}, 1173 {0x00bc, 0x00bf}, 1174 {0x00c6, 0x00c6}, 1175 {0x00d0, 0x00d0}, 1176 {0x00d7, 0x00d8}, 1177 {0x00de, 0x00e1}, 1178 {0x00e6, 0x00e6}, 1179 {0x00e8, 0x00ea}, 1180 {0x00ec, 0x00ed}, 1181 {0x00f0, 0x00f0}, 1182 {0x00f2, 0x00f3}, 1183 {0x00f7, 0x00fa}, 1184 {0x00fc, 0x00fc}, 1185 {0x00fe, 0x00fe}, 1186 {0x0101, 0x0101}, 1187 {0x0111, 0x0111}, 1188 {0x0113, 0x0113}, 1189 {0x011b, 0x011b}, 1190 {0x0126, 0x0127}, 1191 {0x012b, 0x012b}, 1192 {0x0131, 0x0133}, 1193 {0x0138, 0x0138}, 1194 {0x013f, 0x0142}, 1195 {0x0144, 0x0144}, 1196 {0x0148, 0x014b}, 1197 {0x014d, 0x014d}, 1198 {0x0152, 0x0153}, 1199 {0x0166, 0x0167}, 1200 {0x016b, 0x016b}, 1201 {0x01ce, 0x01ce}, 1202 {0x01d0, 0x01d0}, 1203 {0x01d2, 0x01d2}, 1204 {0x01d4, 0x01d4}, 1205 {0x01d6, 0x01d6}, 1206 {0x01d8, 0x01d8}, 1207 {0x01da, 0x01da}, 1208 {0x01dc, 0x01dc}, 1209 {0x0251, 0x0251}, 1210 {0x0261, 0x0261}, 1211 {0x02c4, 0x02c4}, 1212 {0x02c7, 0x02c7}, 1213 {0x02c9, 0x02cb}, 1214 {0x02cd, 0x02cd}, 1215 {0x02d0, 0x02d0}, 1216 {0x02d8, 0x02db}, 1217 {0x02dd, 0x02dd}, 1218 {0x02df, 0x02df}, 1219 {0x0300, 0x036f}, 1220 {0x0391, 0x03a1}, 1221 {0x03a3, 0x03a9}, 1222 {0x03b1, 0x03c1}, 1223 {0x03c3, 0x03c9}, 1224 {0x0401, 0x0401}, 1225 {0x0410, 0x044f}, 1226 {0x0451, 0x0451}, 1227 {0x2010, 0x2010}, 1228 {0x2013, 0x2016}, 1229 {0x2018, 0x2019}, 1230 {0x201c, 0x201d}, 1231 {0x2020, 0x2022}, 1232 {0x2024, 0x2027}, 1233 {0x2030, 0x2030}, 1234 {0x2032, 0x2033}, 1235 {0x2035, 0x2035}, 1236 {0x203b, 0x203b}, 1237 {0x203e, 0x203e}, 1238 {0x2074, 0x2074}, 1239 {0x207f, 0x207f}, 1240 {0x2081, 0x2084}, 1241 {0x20ac, 0x20ac}, 1242 {0x2103, 0x2103}, 1243 {0x2105, 0x2105}, 1244 {0x2109, 0x2109}, 1245 {0x2113, 0x2113}, 1246 {0x2116, 0x2116}, 1247 {0x2121, 0x2122}, 1248 {0x2126, 0x2126}, 1249 {0x212b, 0x212b}, 1250 {0x2153, 0x2154}, 1251 {0x215b, 0x215e}, 1252 {0x2160, 0x216b}, 1253 {0x2170, 0x2179}, 1254 {0x2189, 0x2189}, 1255 {0x2190, 0x2199}, 1256 {0x21b8, 0x21b9}, 1257 {0x21d2, 0x21d2}, 1258 {0x21d4, 0x21d4}, 1259 {0x21e7, 0x21e7}, 1260 {0x2200, 0x2200}, 1261 {0x2202, 0x2203}, 1262 {0x2207, 0x2208}, 1263 {0x220b, 0x220b}, 1264 {0x220f, 0x220f}, 1265 {0x2211, 0x2211}, 1266 {0x2215, 0x2215}, 1267 {0x221a, 0x221a}, 1268 {0x221d, 0x2220}, 1269 {0x2223, 0x2223}, 1270 {0x2225, 0x2225}, 1271 {0x2227, 0x222c}, 1272 {0x222e, 0x222e}, 1273 {0x2234, 0x2237}, 1274 {0x223c, 0x223d}, 1275 {0x2248, 0x2248}, 1276 {0x224c, 0x224c}, 1277 {0x2252, 0x2252}, 1278 {0x2260, 0x2261}, 1279 {0x2264, 0x2267}, 1280 {0x226a, 0x226b}, 1281 {0x226e, 0x226f}, 1282 {0x2282, 0x2283}, 1283 {0x2286, 0x2287}, 1284 {0x2295, 0x2295}, 1285 {0x2299, 0x2299}, 1286 {0x22a5, 0x22a5}, 1287 {0x22bf, 0x22bf}, 1288 {0x2312, 0x2312}, 1289 {0x2460, 0x24e9}, 1290 {0x24eb, 0x254b}, 1291 {0x2550, 0x2573}, 1292 {0x2580, 0x258f}, 1293 {0x2592, 0x2595}, 1294 {0x25a0, 0x25a1}, 1295 {0x25a3, 0x25a9}, 1296 {0x25b2, 0x25b3}, 1297 {0x25b6, 0x25b7}, 1298 {0x25bc, 0x25bd}, 1299 {0x25c0, 0x25c1}, 1300 {0x25c6, 0x25c8}, 1301 {0x25cb, 0x25cb}, 1302 {0x25ce, 0x25d1}, 1303 {0x25e2, 0x25e5}, 1304 {0x25ef, 0x25ef}, 1305 {0x2605, 0x2606}, 1306 {0x2609, 0x2609}, 1307 {0x260e, 0x260f}, 1308 {0x261c, 0x261c}, 1309 {0x261e, 0x261e}, 1310 {0x2640, 0x2640}, 1311 {0x2642, 0x2642}, 1312 {0x2660, 0x2661}, 1313 {0x2663, 0x2665}, 1314 {0x2667, 0x266a}, 1315 {0x266c, 0x266d}, 1316 {0x266f, 0x266f}, 1317 {0x269e, 0x269f}, 1318 {0x26bf, 0x26bf}, 1319 {0x26c6, 0x26cd}, 1320 {0x26cf, 0x26d3}, 1321 {0x26d5, 0x26e1}, 1322 {0x26e3, 0x26e3}, 1323 {0x26e8, 0x26e9}, 1324 {0x26eb, 0x26f1}, 1325 {0x26f4, 0x26f4}, 1326 {0x26f6, 0x26f9}, 1327 {0x26fb, 0x26fc}, 1328 {0x26fe, 0x26ff}, 1329 {0x273d, 0x273d}, 1330 {0x2776, 0x277f}, 1331 {0x2b56, 0x2b59}, 1332 {0x3248, 0x324f}, 1333 {0xe000, 0xf8ff}, 1334 {0xfe00, 0xfe0f}, 1335 {0xfffd, 0xfffd}, 1336 {0x1f100, 0x1f10a}, 1337 {0x1f110, 0x1f12d}, 1338 {0x1f130, 0x1f169}, 1339 {0x1f170, 0x1f18d}, 1340 {0x1f18f, 0x1f190}, 1341 {0x1f19b, 0x1f1ac}, 1342 {0xe0100, 0xe01ef}, 1343 {0xf0000, 0xffffd}, 1344 {0x100000, 0x10fffd} 1345 }; 1346 1347 #if defined(FEAT_TERMINAL) || defined(PROTO) 1348 /* 1349 * utf_char2cells() with different argument type for libvterm. 1350 */ 1351 int 1352 utf_uint2cells(UINT32_T c) 1353 { 1354 if (c >= 0x100 && utf_iscomposing((int)c)) 1355 return 0; 1356 return utf_char2cells((int)c); 1357 } 1358 #endif 1359 1360 /* 1361 * For UTF-8 character "c" return 2 for a double-width character, 1 for others. 1362 * Returns 4 or 6 for an unprintable character. 1363 * Is only correct for characters >= 0x80. 1364 * When p_ambw is "double", return 2 for a character with East Asian Width 1365 * class 'A'(mbiguous). 1366 */ 1367 int 1368 utf_char2cells(int c) 1369 { 1370 // Sorted list of non-overlapping intervals of East Asian double width 1371 // characters, generated with ../runtime/tools/unicode.vim. 1372 static struct interval doublewidth[] = 1373 { 1374 {0x1100, 0x115f}, 1375 {0x231a, 0x231b}, 1376 {0x2329, 0x232a}, 1377 {0x23e9, 0x23ec}, 1378 {0x23f0, 0x23f0}, 1379 {0x23f3, 0x23f3}, 1380 {0x25fd, 0x25fe}, 1381 {0x2614, 0x2615}, 1382 {0x2648, 0x2653}, 1383 {0x267f, 0x267f}, 1384 {0x2693, 0x2693}, 1385 {0x26a1, 0x26a1}, 1386 {0x26aa, 0x26ab}, 1387 {0x26bd, 0x26be}, 1388 {0x26c4, 0x26c5}, 1389 {0x26ce, 0x26ce}, 1390 {0x26d4, 0x26d4}, 1391 {0x26ea, 0x26ea}, 1392 {0x26f2, 0x26f5}, 1393 {0x26fa, 0x26fa}, 1394 {0x26fd, 0x26fd}, 1395 {0x2705, 0x2705}, 1396 {0x270a, 0x270b}, 1397 {0x2728, 0x2728}, 1398 {0x274c, 0x274c}, 1399 {0x274e, 0x274e}, 1400 {0x2753, 0x2755}, 1401 {0x2757, 0x2757}, 1402 {0x2795, 0x2797}, 1403 {0x27b0, 0x27b0}, 1404 {0x27bf, 0x27bf}, 1405 {0x2b1b, 0x2b1c}, 1406 {0x2b50, 0x2b50}, 1407 {0x2b55, 0x2b55}, 1408 {0x2e80, 0x2e99}, 1409 {0x2e9b, 0x2ef3}, 1410 {0x2f00, 0x2fd5}, 1411 {0x2ff0, 0x2ffb}, 1412 {0x3000, 0x303e}, 1413 {0x3041, 0x3096}, 1414 {0x3099, 0x30ff}, 1415 {0x3105, 0x312f}, 1416 {0x3131, 0x318e}, 1417 {0x3190, 0x31ba}, 1418 {0x31c0, 0x31e3}, 1419 {0x31f0, 0x321e}, 1420 {0x3220, 0x3247}, 1421 {0x3250, 0x4dbf}, 1422 {0x4e00, 0xa48c}, 1423 {0xa490, 0xa4c6}, 1424 {0xa960, 0xa97c}, 1425 {0xac00, 0xd7a3}, 1426 {0xf900, 0xfaff}, 1427 {0xfe10, 0xfe19}, 1428 {0xfe30, 0xfe52}, 1429 {0xfe54, 0xfe66}, 1430 {0xfe68, 0xfe6b}, 1431 {0xff01, 0xff60}, 1432 {0xffe0, 0xffe6}, 1433 {0x16fe0, 0x16fe3}, 1434 {0x17000, 0x187f7}, 1435 {0x18800, 0x18af2}, 1436 {0x1b000, 0x1b11e}, 1437 {0x1b150, 0x1b152}, 1438 {0x1b164, 0x1b167}, 1439 {0x1b170, 0x1b2fb}, 1440 {0x1f004, 0x1f004}, 1441 {0x1f0cf, 0x1f0cf}, 1442 {0x1f18e, 0x1f18e}, 1443 {0x1f191, 0x1f19a}, 1444 {0x1f200, 0x1f202}, 1445 {0x1f210, 0x1f23b}, 1446 {0x1f240, 0x1f248}, 1447 {0x1f250, 0x1f251}, 1448 {0x1f260, 0x1f265}, 1449 {0x1f300, 0x1f320}, 1450 {0x1f32d, 0x1f335}, 1451 {0x1f337, 0x1f37c}, 1452 {0x1f37e, 0x1f393}, 1453 {0x1f3a0, 0x1f3ca}, 1454 {0x1f3cf, 0x1f3d3}, 1455 {0x1f3e0, 0x1f3f0}, 1456 {0x1f3f4, 0x1f3f4}, 1457 {0x1f3f8, 0x1f43e}, 1458 {0x1f440, 0x1f440}, 1459 {0x1f442, 0x1f4fc}, 1460 {0x1f4ff, 0x1f53d}, 1461 {0x1f54b, 0x1f54e}, 1462 {0x1f550, 0x1f567}, 1463 {0x1f57a, 0x1f57a}, 1464 {0x1f595, 0x1f596}, 1465 {0x1f5a4, 0x1f5a4}, 1466 {0x1f5fb, 0x1f64f}, 1467 {0x1f680, 0x1f6c5}, 1468 {0x1f6cc, 0x1f6cc}, 1469 {0x1f6d0, 0x1f6d2}, 1470 {0x1f6d5, 0x1f6d5}, 1471 {0x1f6eb, 0x1f6ec}, 1472 {0x1f6f4, 0x1f6fa}, 1473 {0x1f7e0, 0x1f7eb}, 1474 {0x1f90d, 0x1f971}, 1475 {0x1f973, 0x1f976}, 1476 {0x1f97a, 0x1f9a2}, 1477 {0x1f9a5, 0x1f9aa}, 1478 {0x1f9ae, 0x1f9ca}, 1479 {0x1f9cd, 0x1f9ff}, 1480 {0x1fa70, 0x1fa73}, 1481 {0x1fa78, 0x1fa7a}, 1482 {0x1fa80, 0x1fa82}, 1483 {0x1fa90, 0x1fa95}, 1484 {0x20000, 0x2fffd}, 1485 {0x30000, 0x3fffd} 1486 }; 1487 1488 // Sorted list of non-overlapping intervals of Emoji characters that don't 1489 // have ambiguous or double width, 1490 // based on http://unicode.org/emoji/charts/emoji-list.html 1491 static struct interval emoji_wide[] = 1492 { 1493 {0x23ed, 0x23ef}, 1494 {0x23f1, 0x23f2}, 1495 {0x23f8, 0x23fa}, 1496 {0x24c2, 0x24c2}, 1497 {0x261d, 0x261d}, 1498 {0x26c8, 0x26c8}, 1499 {0x26cf, 0x26cf}, 1500 {0x26d1, 0x26d1}, 1501 {0x26d3, 0x26d3}, 1502 {0x26e9, 0x26e9}, 1503 {0x26f0, 0x26f1}, 1504 {0x26f7, 0x26f9}, 1505 {0x270c, 0x270d}, 1506 {0x2934, 0x2935}, 1507 {0x1f170, 0x1f189}, 1508 {0x1f1e6, 0x1f1ff}, 1509 {0x1f321, 0x1f321}, 1510 {0x1f324, 0x1f32c}, 1511 {0x1f336, 0x1f336}, 1512 {0x1f37d, 0x1f37d}, 1513 {0x1f396, 0x1f397}, 1514 {0x1f399, 0x1f39b}, 1515 {0x1f39e, 0x1f39f}, 1516 {0x1f3cb, 0x1f3ce}, 1517 {0x1f3d4, 0x1f3df}, 1518 {0x1f3f3, 0x1f3f5}, 1519 {0x1f3f7, 0x1f3f7}, 1520 {0x1f43f, 0x1f43f}, 1521 {0x1f441, 0x1f441}, 1522 {0x1f4fd, 0x1f4fd}, 1523 {0x1f549, 0x1f54a}, 1524 {0x1f56f, 0x1f570}, 1525 {0x1f573, 0x1f579}, 1526 {0x1f587, 0x1f587}, 1527 {0x1f58a, 0x1f58d}, 1528 {0x1f590, 0x1f590}, 1529 {0x1f5a5, 0x1f5a5}, 1530 {0x1f5a8, 0x1f5a8}, 1531 {0x1f5b1, 0x1f5b2}, 1532 {0x1f5bc, 0x1f5bc}, 1533 {0x1f5c2, 0x1f5c4}, 1534 {0x1f5d1, 0x1f5d3}, 1535 {0x1f5dc, 0x1f5de}, 1536 {0x1f5e1, 0x1f5e1}, 1537 {0x1f5e3, 0x1f5e3}, 1538 {0x1f5e8, 0x1f5e8}, 1539 {0x1f5ef, 0x1f5ef}, 1540 {0x1f5f3, 0x1f5f3}, 1541 {0x1f5fa, 0x1f5fa}, 1542 {0x1f6cb, 0x1f6cf}, 1543 {0x1f6e0, 0x1f6e5}, 1544 {0x1f6e9, 0x1f6e9}, 1545 {0x1f6f0, 0x1f6f0}, 1546 {0x1f6f3, 0x1f6f3} 1547 1548 #ifdef MACOS_X 1549 // Include SF Symbols characters, which should be rendered as 1550 // double-width. All of them are in the Supplementary Private Use 1551 // Area-B range. The exact range was determined by downloading the "SF 1552 // Symbols" app from Apple, and then selecting all symbols, copying 1553 // them out, and inspecting the unicode values of them. 1554 , {0x100000, 0x100d7f} 1555 #endif 1556 }; 1557 1558 if (c >= 0x100) 1559 { 1560 #if defined(FEAT_EVAL) || defined(USE_WCHAR_FUNCTIONS) 1561 int n; 1562 #endif 1563 1564 #ifdef FEAT_EVAL 1565 n = cw_value(c); 1566 if (n != 0) 1567 return n; 1568 #endif 1569 1570 #ifdef USE_WCHAR_FUNCTIONS 1571 /* 1572 * Assume the library function wcwidth() works better than our own 1573 * stuff. It should return 1 for ambiguous width chars! 1574 */ 1575 n = wcwidth(c); 1576 1577 if (n < 0) 1578 return 6; // unprintable, displays <xxxx> 1579 if (n > 1) 1580 return n; 1581 #else 1582 if (!utf_printable(c)) 1583 return 6; // unprintable, displays <xxxx> 1584 if (intable(doublewidth, sizeof(doublewidth), c)) 1585 return 2; 1586 #endif 1587 if (p_emoji && intable(emoji_wide, sizeof(emoji_wide), c)) 1588 return 2; 1589 } 1590 1591 // Characters below 0x100 are influenced by 'isprint' option 1592 else if (c >= 0x80 && !vim_isprintc(c)) 1593 return 4; // unprintable, displays <xx> 1594 1595 if (c >= 0x80 && *p_ambw == 'd' && intable(ambiguous, sizeof(ambiguous), c)) 1596 return 2; 1597 1598 return 1; 1599 } 1600 1601 /* 1602 * mb_ptr2cells() function pointer. 1603 * Return the number of display cells character at "*p" occupies. 1604 * This doesn't take care of unprintable characters, use ptr2cells() for that. 1605 */ 1606 int 1607 latin_ptr2cells(char_u *p UNUSED) 1608 { 1609 return 1; 1610 } 1611 1612 int 1613 utf_ptr2cells( 1614 char_u *p) 1615 { 1616 int c; 1617 1618 // Need to convert to a character number. 1619 if (*p >= 0x80) 1620 { 1621 c = utf_ptr2char(p); 1622 // An illegal byte is displayed as <xx>. 1623 if (utf_ptr2len(p) == 1 || c == NUL) 1624 return 4; 1625 // If the char is ASCII it must be an overlong sequence. 1626 if (c < 0x80) 1627 return char2cells(c); 1628 return utf_char2cells(c); 1629 } 1630 return 1; 1631 } 1632 1633 int 1634 dbcs_ptr2cells(char_u *p) 1635 { 1636 // Number of cells is equal to number of bytes, except for euc-jp when 1637 // the first byte is 0x8e. 1638 if (enc_dbcs == DBCS_JPNU && *p == 0x8e) 1639 return 1; 1640 return MB_BYTE2LEN(*p); 1641 } 1642 1643 /* 1644 * mb_ptr2cells_len() function pointer. 1645 * Like mb_ptr2cells(), but limit string length to "size". 1646 * For an empty string or truncated character returns 1. 1647 */ 1648 int 1649 latin_ptr2cells_len(char_u *p UNUSED, int size UNUSED) 1650 { 1651 return 1; 1652 } 1653 1654 static int 1655 utf_ptr2cells_len(char_u *p, int size) 1656 { 1657 int c; 1658 1659 // Need to convert to a wide character. 1660 if (size > 0 && *p >= 0x80) 1661 { 1662 if (utf_ptr2len_len(p, size) < utf8len_tab[*p]) 1663 return 1; // truncated 1664 c = utf_ptr2char(p); 1665 // An illegal byte is displayed as <xx>. 1666 if (utf_ptr2len(p) == 1 || c == NUL) 1667 return 4; 1668 // If the char is ASCII it must be an overlong sequence. 1669 if (c < 0x80) 1670 return char2cells(c); 1671 return utf_char2cells(c); 1672 } 1673 return 1; 1674 } 1675 1676 static int 1677 dbcs_ptr2cells_len(char_u *p, int size) 1678 { 1679 // Number of cells is equal to number of bytes, except for euc-jp when 1680 // the first byte is 0x8e. 1681 if (size <= 1 || (enc_dbcs == DBCS_JPNU && *p == 0x8e)) 1682 return 1; 1683 return MB_BYTE2LEN(*p); 1684 } 1685 1686 /* 1687 * mb_char2cells() function pointer. 1688 * Return the number of display cells character "c" occupies. 1689 * Only takes care of multi-byte chars, not "^C" and such. 1690 */ 1691 int 1692 latin_char2cells(int c UNUSED) 1693 { 1694 return 1; 1695 } 1696 1697 static int 1698 dbcs_char2cells(int c) 1699 { 1700 // Number of cells is equal to number of bytes, except for euc-jp when 1701 // the first byte is 0x8e. 1702 if (enc_dbcs == DBCS_JPNU && ((unsigned)c >> 8) == 0x8e) 1703 return 1; 1704 // use the first byte 1705 return MB_BYTE2LEN((unsigned)c >> 8); 1706 } 1707 1708 /* 1709 * Return the number of cells occupied by string "p". 1710 * Stop at a NUL character. When "len" >= 0 stop at character "p[len]". 1711 */ 1712 int 1713 mb_string2cells(char_u *p, int len) 1714 { 1715 int i; 1716 int clen = 0; 1717 1718 for (i = 0; (len < 0 || i < len) && p[i] != NUL; i += (*mb_ptr2len)(p + i)) 1719 clen += (*mb_ptr2cells)(p + i); 1720 return clen; 1721 } 1722 1723 /* 1724 * mb_off2cells() function pointer. 1725 * Return number of display cells for char at ScreenLines[off]. 1726 * We make sure that the offset used is less than "max_off". 1727 */ 1728 int 1729 latin_off2cells(unsigned off UNUSED, unsigned max_off UNUSED) 1730 { 1731 return 1; 1732 } 1733 1734 int 1735 dbcs_off2cells(unsigned off, unsigned max_off) 1736 { 1737 // never check beyond end of the line 1738 if (off >= max_off) 1739 return 1; 1740 1741 // Number of cells is equal to number of bytes, except for euc-jp when 1742 // the first byte is 0x8e. 1743 if (enc_dbcs == DBCS_JPNU && ScreenLines[off] == 0x8e) 1744 return 1; 1745 return MB_BYTE2LEN(ScreenLines[off]); 1746 } 1747 1748 int 1749 utf_off2cells(unsigned off, unsigned max_off) 1750 { 1751 return (off + 1 < max_off && ScreenLines[off + 1] == 0) ? 2 : 1; 1752 } 1753 1754 /* 1755 * mb_ptr2char() function pointer. 1756 * Convert a byte sequence into a character. 1757 */ 1758 int 1759 latin_ptr2char(char_u *p) 1760 { 1761 return *p; 1762 } 1763 1764 static int 1765 dbcs_ptr2char(char_u *p) 1766 { 1767 if (MB_BYTE2LEN(*p) > 1 && p[1] != NUL) 1768 return (p[0] << 8) + p[1]; 1769 return *p; 1770 } 1771 1772 /* 1773 * Convert a UTF-8 byte sequence to a character number. 1774 * If the sequence is illegal or truncated by a NUL the first byte is 1775 * returned. 1776 * For an overlong sequence this may return zero. 1777 * Does not include composing characters, of course. 1778 */ 1779 int 1780 utf_ptr2char(char_u *p) 1781 { 1782 int len; 1783 1784 if (p[0] < 0x80) // be quick for ASCII 1785 return p[0]; 1786 1787 len = utf8len_tab_zero[p[0]]; 1788 if (len > 1 && (p[1] & 0xc0) == 0x80) 1789 { 1790 if (len == 2) 1791 return ((p[0] & 0x1f) << 6) + (p[1] & 0x3f); 1792 if ((p[2] & 0xc0) == 0x80) 1793 { 1794 if (len == 3) 1795 return ((p[0] & 0x0f) << 12) + ((p[1] & 0x3f) << 6) 1796 + (p[2] & 0x3f); 1797 if ((p[3] & 0xc0) == 0x80) 1798 { 1799 if (len == 4) 1800 return ((p[0] & 0x07) << 18) + ((p[1] & 0x3f) << 12) 1801 + ((p[2] & 0x3f) << 6) + (p[3] & 0x3f); 1802 if ((p[4] & 0xc0) == 0x80) 1803 { 1804 if (len == 5) 1805 return ((p[0] & 0x03) << 24) + ((p[1] & 0x3f) << 18) 1806 + ((p[2] & 0x3f) << 12) + ((p[3] & 0x3f) << 6) 1807 + (p[4] & 0x3f); 1808 if ((p[5] & 0xc0) == 0x80 && len == 6) 1809 return ((p[0] & 0x01) << 30) + ((p[1] & 0x3f) << 24) 1810 + ((p[2] & 0x3f) << 18) + ((p[3] & 0x3f) << 12) 1811 + ((p[4] & 0x3f) << 6) + (p[5] & 0x3f); 1812 } 1813 } 1814 } 1815 } 1816 // Illegal value, just return the first byte 1817 return p[0]; 1818 } 1819 1820 /* 1821 * Convert a UTF-8 byte sequence to a wide character. 1822 * String is assumed to be terminated by NUL or after "n" bytes, whichever 1823 * comes first. 1824 * The function is safe in the sense that it never accesses memory beyond the 1825 * first "n" bytes of "s". 1826 * 1827 * On success, returns decoded codepoint, advances "s" to the beginning of 1828 * next character and decreases "n" accordingly. 1829 * 1830 * If end of string was reached, returns 0 and, if "n" > 0, advances "s" past 1831 * NUL byte. 1832 * 1833 * If byte sequence is illegal or incomplete, returns -1 and does not advance 1834 * "s". 1835 */ 1836 static int 1837 utf_safe_read_char_adv(char_u **s, size_t *n) 1838 { 1839 int c, k; 1840 1841 if (*n == 0) // end of buffer 1842 return 0; 1843 1844 k = utf8len_tab_zero[**s]; 1845 1846 if (k == 1) 1847 { 1848 // ASCII character or NUL 1849 (*n)--; 1850 return *(*s)++; 1851 } 1852 1853 if ((size_t)k <= *n) 1854 { 1855 // We have a multibyte sequence and it isn't truncated by buffer 1856 // limits so utf_ptr2char() is safe to use. Or the first byte is 1857 // illegal (k=0), and it's also safe to use utf_ptr2char(). 1858 c = utf_ptr2char(*s); 1859 1860 // On failure, utf_ptr2char() returns the first byte, so here we 1861 // check equality with the first byte. The only non-ASCII character 1862 // which equals the first byte of its own UTF-8 representation is 1863 // U+00C3 (UTF-8: 0xC3 0x83), so need to check that special case too. 1864 // It's safe even if n=1, else we would have k=2 > n. 1865 if (c != (int)(**s) || (c == 0xC3 && (*s)[1] == 0x83)) 1866 { 1867 // byte sequence was successfully decoded 1868 *s += k; 1869 *n -= k; 1870 return c; 1871 } 1872 } 1873 1874 // byte sequence is incomplete or illegal 1875 return -1; 1876 } 1877 1878 /* 1879 * Get character at **pp and advance *pp to the next character. 1880 * Note: composing characters are skipped! 1881 */ 1882 int 1883 mb_ptr2char_adv(char_u **pp) 1884 { 1885 int c; 1886 1887 c = (*mb_ptr2char)(*pp); 1888 *pp += (*mb_ptr2len)(*pp); 1889 return c; 1890 } 1891 1892 /* 1893 * Get character at **pp and advance *pp to the next character. 1894 * Note: composing characters are returned as separate characters. 1895 */ 1896 int 1897 mb_cptr2char_adv(char_u **pp) 1898 { 1899 int c; 1900 1901 c = (*mb_ptr2char)(*pp); 1902 if (enc_utf8) 1903 *pp += utf_ptr2len(*pp); 1904 else 1905 *pp += (*mb_ptr2len)(*pp); 1906 return c; 1907 } 1908 1909 #if defined(FEAT_ARABIC) || defined(PROTO) 1910 /* 1911 * Check if the character pointed to by "p2" is a composing character when it 1912 * comes after "p1". For Arabic sometimes "ab" is replaced with "c", which 1913 * behaves like a composing character. 1914 */ 1915 int 1916 utf_composinglike(char_u *p1, char_u *p2) 1917 { 1918 int c2; 1919 1920 c2 = utf_ptr2char(p2); 1921 if (utf_iscomposing(c2)) 1922 return TRUE; 1923 if (!arabic_maycombine(c2)) 1924 return FALSE; 1925 return arabic_combine(utf_ptr2char(p1), c2); 1926 } 1927 #endif 1928 1929 /* 1930 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1931 * composing characters. 1932 */ 1933 int 1934 utfc_ptr2char( 1935 char_u *p, 1936 int *pcc) // return: composing chars, last one is 0 1937 { 1938 int len; 1939 int c; 1940 int cc; 1941 int i = 0; 1942 1943 c = utf_ptr2char(p); 1944 len = utf_ptr2len(p); 1945 1946 // Only accept a composing char when the first char isn't illegal. 1947 if ((len > 1 || *p < 0x80) 1948 && p[len] >= 0x80 1949 && UTF_COMPOSINGLIKE(p, p + len)) 1950 { 1951 cc = utf_ptr2char(p + len); 1952 for (;;) 1953 { 1954 pcc[i++] = cc; 1955 if (i == MAX_MCO) 1956 break; 1957 len += utf_ptr2len(p + len); 1958 if (p[len] < 0x80 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 1959 break; 1960 } 1961 } 1962 1963 if (i < MAX_MCO) // last composing char must be 0 1964 pcc[i] = 0; 1965 1966 return c; 1967 } 1968 1969 /* 1970 * Convert a UTF-8 byte string to a wide character. Also get up to MAX_MCO 1971 * composing characters. Use no more than p[maxlen]. 1972 */ 1973 int 1974 utfc_ptr2char_len( 1975 char_u *p, 1976 int *pcc, // return: composing chars, last one is 0 1977 int maxlen) 1978 { 1979 int len; 1980 int c; 1981 int cc; 1982 int i = 0; 1983 1984 c = utf_ptr2char(p); 1985 len = utf_ptr2len_len(p, maxlen); 1986 // Only accept a composing char when the first char isn't illegal. 1987 if ((len > 1 || *p < 0x80) 1988 && len < maxlen 1989 && p[len] >= 0x80 1990 && UTF_COMPOSINGLIKE(p, p + len)) 1991 { 1992 cc = utf_ptr2char(p + len); 1993 for (;;) 1994 { 1995 pcc[i++] = cc; 1996 if (i == MAX_MCO) 1997 break; 1998 len += utf_ptr2len_len(p + len, maxlen - len); 1999 if (len >= maxlen 2000 || p[len] < 0x80 2001 || !utf_iscomposing(cc = utf_ptr2char(p + len))) 2002 break; 2003 } 2004 } 2005 2006 if (i < MAX_MCO) // last composing char must be 0 2007 pcc[i] = 0; 2008 2009 return c; 2010 } 2011 2012 /* 2013 * Convert the character at screen position "off" to a sequence of bytes. 2014 * Includes the composing characters. 2015 * "buf" must at least have the length MB_MAXBYTES + 1. 2016 * Only to be used when ScreenLinesUC[off] != 0. 2017 * Returns the produced number of bytes. 2018 */ 2019 int 2020 utfc_char2bytes(int off, char_u *buf) 2021 { 2022 int len; 2023 int i; 2024 2025 len = utf_char2bytes(ScreenLinesUC[off], buf); 2026 for (i = 0; i < Screen_mco; ++i) 2027 { 2028 if (ScreenLinesC[i][off] == 0) 2029 break; 2030 len += utf_char2bytes(ScreenLinesC[i][off], buf + len); 2031 } 2032 return len; 2033 } 2034 2035 /* 2036 * Get the length of a UTF-8 byte sequence, not including any following 2037 * composing characters. 2038 * Returns 0 for "". 2039 * Returns 1 for an illegal byte sequence. 2040 */ 2041 int 2042 utf_ptr2len(char_u *p) 2043 { 2044 int len; 2045 int i; 2046 2047 if (*p == NUL) 2048 return 0; 2049 len = utf8len_tab[*p]; 2050 for (i = 1; i < len; ++i) 2051 if ((p[i] & 0xc0) != 0x80) 2052 return 1; 2053 return len; 2054 } 2055 2056 /* 2057 * Return length of UTF-8 character, obtained from the first byte. 2058 * "b" must be between 0 and 255! 2059 * Returns 1 for an invalid first byte value. 2060 */ 2061 int 2062 utf_byte2len(int b) 2063 { 2064 return utf8len_tab[b]; 2065 } 2066 2067 /* 2068 * Get the length of UTF-8 byte sequence "p[size]". Does not include any 2069 * following composing characters. 2070 * Returns 1 for "". 2071 * Returns 1 for an illegal byte sequence (also in incomplete byte seq.). 2072 * Returns number > "size" for an incomplete byte sequence. 2073 * Never returns zero. 2074 */ 2075 int 2076 utf_ptr2len_len(char_u *p, int size) 2077 { 2078 int len; 2079 int i; 2080 int m; 2081 2082 len = utf8len_tab[*p]; 2083 if (len == 1) 2084 return 1; // NUL, ascii or illegal lead byte 2085 if (len > size) 2086 m = size; // incomplete byte sequence. 2087 else 2088 m = len; 2089 for (i = 1; i < m; ++i) 2090 if ((p[i] & 0xc0) != 0x80) 2091 return 1; 2092 return len; 2093 } 2094 2095 /* 2096 * Return the number of bytes the UTF-8 encoding of the character at "p" takes. 2097 * This includes following composing characters. 2098 */ 2099 int 2100 utfc_ptr2len(char_u *p) 2101 { 2102 int len; 2103 int b0 = *p; 2104 #ifdef FEAT_ARABIC 2105 int prevlen; 2106 #endif 2107 2108 if (b0 == NUL) 2109 return 0; 2110 if (b0 < 0x80 && p[1] < 0x80) // be quick for ASCII 2111 return 1; 2112 2113 // Skip over first UTF-8 char, stopping at a NUL byte. 2114 len = utf_ptr2len(p); 2115 2116 // Check for illegal byte. 2117 if (len == 1 && b0 >= 0x80) 2118 return 1; 2119 2120 /* 2121 * Check for composing characters. We can handle only the first six, but 2122 * skip all of them (otherwise the cursor would get stuck). 2123 */ 2124 #ifdef FEAT_ARABIC 2125 prevlen = 0; 2126 #endif 2127 for (;;) 2128 { 2129 if (p[len] < 0x80 || !UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2130 return len; 2131 2132 // Skip over composing char 2133 #ifdef FEAT_ARABIC 2134 prevlen = len; 2135 #endif 2136 len += utf_ptr2len(p + len); 2137 } 2138 } 2139 2140 /* 2141 * Return the number of bytes the UTF-8 encoding of the character at "p[size]" 2142 * takes. This includes following composing characters. 2143 * Returns 0 for an empty string. 2144 * Returns 1 for an illegal char or an incomplete byte sequence. 2145 */ 2146 int 2147 utfc_ptr2len_len(char_u *p, int size) 2148 { 2149 int len; 2150 #ifdef FEAT_ARABIC 2151 int prevlen; 2152 #endif 2153 2154 if (size < 1 || *p == NUL) 2155 return 0; 2156 if (p[0] < 0x80 && (size == 1 || p[1] < 0x80)) // be quick for ASCII 2157 return 1; 2158 2159 // Skip over first UTF-8 char, stopping at a NUL byte. 2160 len = utf_ptr2len_len(p, size); 2161 2162 // Check for illegal byte and incomplete byte sequence. 2163 if ((len == 1 && p[0] >= 0x80) || len > size) 2164 return 1; 2165 2166 /* 2167 * Check for composing characters. We can handle only the first six, but 2168 * skip all of them (otherwise the cursor would get stuck). 2169 */ 2170 #ifdef FEAT_ARABIC 2171 prevlen = 0; 2172 #endif 2173 while (len < size) 2174 { 2175 int len_next_char; 2176 2177 if (p[len] < 0x80) 2178 break; 2179 2180 /* 2181 * Next character length should not go beyond size to ensure that 2182 * UTF_COMPOSINGLIKE(...) does not read beyond size. 2183 */ 2184 len_next_char = utf_ptr2len_len(p + len, size - len); 2185 if (len_next_char > size - len) 2186 break; 2187 2188 if (!UTF_COMPOSINGLIKE(p + prevlen, p + len)) 2189 break; 2190 2191 // Skip over composing char 2192 #ifdef FEAT_ARABIC 2193 prevlen = len; 2194 #endif 2195 len += len_next_char; 2196 } 2197 return len; 2198 } 2199 2200 /* 2201 * Return the number of bytes the UTF-8 encoding of character "c" takes. 2202 * This does not include composing characters. 2203 */ 2204 int 2205 utf_char2len(int c) 2206 { 2207 if (c < 0x80) 2208 return 1; 2209 if (c < 0x800) 2210 return 2; 2211 if (c < 0x10000) 2212 return 3; 2213 if (c < 0x200000) 2214 return 4; 2215 if (c < 0x4000000) 2216 return 5; 2217 return 6; 2218 } 2219 2220 /* 2221 * Convert Unicode character "c" to UTF-8 string in "buf[]". 2222 * Returns the number of bytes. 2223 */ 2224 int 2225 utf_char2bytes(int c, char_u *buf) 2226 { 2227 if (c < 0x80) // 7 bits 2228 { 2229 buf[0] = c; 2230 return 1; 2231 } 2232 if (c < 0x800) // 11 bits 2233 { 2234 buf[0] = 0xc0 + ((unsigned)c >> 6); 2235 buf[1] = 0x80 + (c & 0x3f); 2236 return 2; 2237 } 2238 if (c < 0x10000) // 16 bits 2239 { 2240 buf[0] = 0xe0 + ((unsigned)c >> 12); 2241 buf[1] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2242 buf[2] = 0x80 + (c & 0x3f); 2243 return 3; 2244 } 2245 if (c < 0x200000) // 21 bits 2246 { 2247 buf[0] = 0xf0 + ((unsigned)c >> 18); 2248 buf[1] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2249 buf[2] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2250 buf[3] = 0x80 + (c & 0x3f); 2251 return 4; 2252 } 2253 if (c < 0x4000000) // 26 bits 2254 { 2255 buf[0] = 0xf8 + ((unsigned)c >> 24); 2256 buf[1] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2257 buf[2] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2258 buf[3] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2259 buf[4] = 0x80 + (c & 0x3f); 2260 return 5; 2261 } 2262 // 31 bits 2263 buf[0] = 0xfc + ((unsigned)c >> 30); 2264 buf[1] = 0x80 + (((unsigned)c >> 24) & 0x3f); 2265 buf[2] = 0x80 + (((unsigned)c >> 18) & 0x3f); 2266 buf[3] = 0x80 + (((unsigned)c >> 12) & 0x3f); 2267 buf[4] = 0x80 + (((unsigned)c >> 6) & 0x3f); 2268 buf[5] = 0x80 + (c & 0x3f); 2269 return 6; 2270 } 2271 2272 #if defined(FEAT_TERMINAL) || defined(PROTO) 2273 /* 2274 * utf_iscomposing() with different argument type for libvterm. 2275 */ 2276 int 2277 utf_iscomposing_uint(UINT32_T c) 2278 { 2279 return utf_iscomposing((int)c); 2280 } 2281 #endif 2282 2283 /* 2284 * Return TRUE if "c" is a composing UTF-8 character. This means it will be 2285 * drawn on top of the preceding character. 2286 * Based on code from Markus Kuhn. 2287 */ 2288 int 2289 utf_iscomposing(int c) 2290 { 2291 // Sorted list of non-overlapping intervals. 2292 // Generated by ../runtime/tools/unicode.vim. 2293 static struct interval combining[] = 2294 { 2295 {0x0300, 0x036f}, 2296 {0x0483, 0x0489}, 2297 {0x0591, 0x05bd}, 2298 {0x05bf, 0x05bf}, 2299 {0x05c1, 0x05c2}, 2300 {0x05c4, 0x05c5}, 2301 {0x05c7, 0x05c7}, 2302 {0x0610, 0x061a}, 2303 {0x064b, 0x065f}, 2304 {0x0670, 0x0670}, 2305 {0x06d6, 0x06dc}, 2306 {0x06df, 0x06e4}, 2307 {0x06e7, 0x06e8}, 2308 {0x06ea, 0x06ed}, 2309 {0x0711, 0x0711}, 2310 {0x0730, 0x074a}, 2311 {0x07a6, 0x07b0}, 2312 {0x07eb, 0x07f3}, 2313 {0x07fd, 0x07fd}, 2314 {0x0816, 0x0819}, 2315 {0x081b, 0x0823}, 2316 {0x0825, 0x0827}, 2317 {0x0829, 0x082d}, 2318 {0x0859, 0x085b}, 2319 {0x08d3, 0x08e1}, 2320 {0x08e3, 0x0903}, 2321 {0x093a, 0x093c}, 2322 {0x093e, 0x094f}, 2323 {0x0951, 0x0957}, 2324 {0x0962, 0x0963}, 2325 {0x0981, 0x0983}, 2326 {0x09bc, 0x09bc}, 2327 {0x09be, 0x09c4}, 2328 {0x09c7, 0x09c8}, 2329 {0x09cb, 0x09cd}, 2330 {0x09d7, 0x09d7}, 2331 {0x09e2, 0x09e3}, 2332 {0x09fe, 0x09fe}, 2333 {0x0a01, 0x0a03}, 2334 {0x0a3c, 0x0a3c}, 2335 {0x0a3e, 0x0a42}, 2336 {0x0a47, 0x0a48}, 2337 {0x0a4b, 0x0a4d}, 2338 {0x0a51, 0x0a51}, 2339 {0x0a70, 0x0a71}, 2340 {0x0a75, 0x0a75}, 2341 {0x0a81, 0x0a83}, 2342 {0x0abc, 0x0abc}, 2343 {0x0abe, 0x0ac5}, 2344 {0x0ac7, 0x0ac9}, 2345 {0x0acb, 0x0acd}, 2346 {0x0ae2, 0x0ae3}, 2347 {0x0afa, 0x0aff}, 2348 {0x0b01, 0x0b03}, 2349 {0x0b3c, 0x0b3c}, 2350 {0x0b3e, 0x0b44}, 2351 {0x0b47, 0x0b48}, 2352 {0x0b4b, 0x0b4d}, 2353 {0x0b56, 0x0b57}, 2354 {0x0b62, 0x0b63}, 2355 {0x0b82, 0x0b82}, 2356 {0x0bbe, 0x0bc2}, 2357 {0x0bc6, 0x0bc8}, 2358 {0x0bca, 0x0bcd}, 2359 {0x0bd7, 0x0bd7}, 2360 {0x0c00, 0x0c04}, 2361 {0x0c3e, 0x0c44}, 2362 {0x0c46, 0x0c48}, 2363 {0x0c4a, 0x0c4d}, 2364 {0x0c55, 0x0c56}, 2365 {0x0c62, 0x0c63}, 2366 {0x0c81, 0x0c83}, 2367 {0x0cbc, 0x0cbc}, 2368 {0x0cbe, 0x0cc4}, 2369 {0x0cc6, 0x0cc8}, 2370 {0x0cca, 0x0ccd}, 2371 {0x0cd5, 0x0cd6}, 2372 {0x0ce2, 0x0ce3}, 2373 {0x0d00, 0x0d03}, 2374 {0x0d3b, 0x0d3c}, 2375 {0x0d3e, 0x0d44}, 2376 {0x0d46, 0x0d48}, 2377 {0x0d4a, 0x0d4d}, 2378 {0x0d57, 0x0d57}, 2379 {0x0d62, 0x0d63}, 2380 {0x0d82, 0x0d83}, 2381 {0x0dca, 0x0dca}, 2382 {0x0dcf, 0x0dd4}, 2383 {0x0dd6, 0x0dd6}, 2384 {0x0dd8, 0x0ddf}, 2385 {0x0df2, 0x0df3}, 2386 {0x0e31, 0x0e31}, 2387 {0x0e34, 0x0e3a}, 2388 {0x0e47, 0x0e4e}, 2389 {0x0eb1, 0x0eb1}, 2390 {0x0eb4, 0x0ebc}, 2391 {0x0ec8, 0x0ecd}, 2392 {0x0f18, 0x0f19}, 2393 {0x0f35, 0x0f35}, 2394 {0x0f37, 0x0f37}, 2395 {0x0f39, 0x0f39}, 2396 {0x0f3e, 0x0f3f}, 2397 {0x0f71, 0x0f84}, 2398 {0x0f86, 0x0f87}, 2399 {0x0f8d, 0x0f97}, 2400 {0x0f99, 0x0fbc}, 2401 {0x0fc6, 0x0fc6}, 2402 {0x102b, 0x103e}, 2403 {0x1056, 0x1059}, 2404 {0x105e, 0x1060}, 2405 {0x1062, 0x1064}, 2406 {0x1067, 0x106d}, 2407 {0x1071, 0x1074}, 2408 {0x1082, 0x108d}, 2409 {0x108f, 0x108f}, 2410 {0x109a, 0x109d}, 2411 {0x135d, 0x135f}, 2412 {0x1712, 0x1714}, 2413 {0x1732, 0x1734}, 2414 {0x1752, 0x1753}, 2415 {0x1772, 0x1773}, 2416 {0x17b4, 0x17d3}, 2417 {0x17dd, 0x17dd}, 2418 {0x180b, 0x180d}, 2419 {0x1885, 0x1886}, 2420 {0x18a9, 0x18a9}, 2421 {0x1920, 0x192b}, 2422 {0x1930, 0x193b}, 2423 {0x1a17, 0x1a1b}, 2424 {0x1a55, 0x1a5e}, 2425 {0x1a60, 0x1a7c}, 2426 {0x1a7f, 0x1a7f}, 2427 {0x1ab0, 0x1abe}, 2428 {0x1b00, 0x1b04}, 2429 {0x1b34, 0x1b44}, 2430 {0x1b6b, 0x1b73}, 2431 {0x1b80, 0x1b82}, 2432 {0x1ba1, 0x1bad}, 2433 {0x1be6, 0x1bf3}, 2434 {0x1c24, 0x1c37}, 2435 {0x1cd0, 0x1cd2}, 2436 {0x1cd4, 0x1ce8}, 2437 {0x1ced, 0x1ced}, 2438 {0x1cf4, 0x1cf4}, 2439 {0x1cf7, 0x1cf9}, 2440 {0x1dc0, 0x1df9}, 2441 {0x1dfb, 0x1dff}, 2442 {0x20d0, 0x20f0}, 2443 {0x2cef, 0x2cf1}, 2444 {0x2d7f, 0x2d7f}, 2445 {0x2de0, 0x2dff}, 2446 {0x302a, 0x302f}, 2447 {0x3099, 0x309a}, 2448 {0xa66f, 0xa672}, 2449 {0xa674, 0xa67d}, 2450 {0xa69e, 0xa69f}, 2451 {0xa6f0, 0xa6f1}, 2452 {0xa802, 0xa802}, 2453 {0xa806, 0xa806}, 2454 {0xa80b, 0xa80b}, 2455 {0xa823, 0xa827}, 2456 {0xa880, 0xa881}, 2457 {0xa8b4, 0xa8c5}, 2458 {0xa8e0, 0xa8f1}, 2459 {0xa8ff, 0xa8ff}, 2460 {0xa926, 0xa92d}, 2461 {0xa947, 0xa953}, 2462 {0xa980, 0xa983}, 2463 {0xa9b3, 0xa9c0}, 2464 {0xa9e5, 0xa9e5}, 2465 {0xaa29, 0xaa36}, 2466 {0xaa43, 0xaa43}, 2467 {0xaa4c, 0xaa4d}, 2468 {0xaa7b, 0xaa7d}, 2469 {0xaab0, 0xaab0}, 2470 {0xaab2, 0xaab4}, 2471 {0xaab7, 0xaab8}, 2472 {0xaabe, 0xaabf}, 2473 {0xaac1, 0xaac1}, 2474 {0xaaeb, 0xaaef}, 2475 {0xaaf5, 0xaaf6}, 2476 {0xabe3, 0xabea}, 2477 {0xabec, 0xabed}, 2478 {0xfb1e, 0xfb1e}, 2479 {0xfe00, 0xfe0f}, 2480 {0xfe20, 0xfe2f}, 2481 {0x101fd, 0x101fd}, 2482 {0x102e0, 0x102e0}, 2483 {0x10376, 0x1037a}, 2484 {0x10a01, 0x10a03}, 2485 {0x10a05, 0x10a06}, 2486 {0x10a0c, 0x10a0f}, 2487 {0x10a38, 0x10a3a}, 2488 {0x10a3f, 0x10a3f}, 2489 {0x10ae5, 0x10ae6}, 2490 {0x10d24, 0x10d27}, 2491 {0x10f46, 0x10f50}, 2492 {0x11000, 0x11002}, 2493 {0x11038, 0x11046}, 2494 {0x1107f, 0x11082}, 2495 {0x110b0, 0x110ba}, 2496 {0x11100, 0x11102}, 2497 {0x11127, 0x11134}, 2498 {0x11145, 0x11146}, 2499 {0x11173, 0x11173}, 2500 {0x11180, 0x11182}, 2501 {0x111b3, 0x111c0}, 2502 {0x111c9, 0x111cc}, 2503 {0x1122c, 0x11237}, 2504 {0x1123e, 0x1123e}, 2505 {0x112df, 0x112ea}, 2506 {0x11300, 0x11303}, 2507 {0x1133b, 0x1133c}, 2508 {0x1133e, 0x11344}, 2509 {0x11347, 0x11348}, 2510 {0x1134b, 0x1134d}, 2511 {0x11357, 0x11357}, 2512 {0x11362, 0x11363}, 2513 {0x11366, 0x1136c}, 2514 {0x11370, 0x11374}, 2515 {0x11435, 0x11446}, 2516 {0x1145e, 0x1145e}, 2517 {0x114b0, 0x114c3}, 2518 {0x115af, 0x115b5}, 2519 {0x115b8, 0x115c0}, 2520 {0x115dc, 0x115dd}, 2521 {0x11630, 0x11640}, 2522 {0x116ab, 0x116b7}, 2523 {0x1171d, 0x1172b}, 2524 {0x1182c, 0x1183a}, 2525 {0x119d1, 0x119d7}, 2526 {0x119da, 0x119e0}, 2527 {0x119e4, 0x119e4}, 2528 {0x11a01, 0x11a0a}, 2529 {0x11a33, 0x11a39}, 2530 {0x11a3b, 0x11a3e}, 2531 {0x11a47, 0x11a47}, 2532 {0x11a51, 0x11a5b}, 2533 {0x11a8a, 0x11a99}, 2534 {0x11c2f, 0x11c36}, 2535 {0x11c38, 0x11c3f}, 2536 {0x11c92, 0x11ca7}, 2537 {0x11ca9, 0x11cb6}, 2538 {0x11d31, 0x11d36}, 2539 {0x11d3a, 0x11d3a}, 2540 {0x11d3c, 0x11d3d}, 2541 {0x11d3f, 0x11d45}, 2542 {0x11d47, 0x11d47}, 2543 {0x11d8a, 0x11d8e}, 2544 {0x11d90, 0x11d91}, 2545 {0x11d93, 0x11d97}, 2546 {0x11ef3, 0x11ef6}, 2547 {0x16af0, 0x16af4}, 2548 {0x16b30, 0x16b36}, 2549 {0x16f4f, 0x16f4f}, 2550 {0x16f51, 0x16f87}, 2551 {0x16f8f, 0x16f92}, 2552 {0x1bc9d, 0x1bc9e}, 2553 {0x1d165, 0x1d169}, 2554 {0x1d16d, 0x1d172}, 2555 {0x1d17b, 0x1d182}, 2556 {0x1d185, 0x1d18b}, 2557 {0x1d1aa, 0x1d1ad}, 2558 {0x1d242, 0x1d244}, 2559 {0x1da00, 0x1da36}, 2560 {0x1da3b, 0x1da6c}, 2561 {0x1da75, 0x1da75}, 2562 {0x1da84, 0x1da84}, 2563 {0x1da9b, 0x1da9f}, 2564 {0x1daa1, 0x1daaf}, 2565 {0x1e000, 0x1e006}, 2566 {0x1e008, 0x1e018}, 2567 {0x1e01b, 0x1e021}, 2568 {0x1e023, 0x1e024}, 2569 {0x1e026, 0x1e02a}, 2570 {0x1e130, 0x1e136}, 2571 {0x1e2ec, 0x1e2ef}, 2572 {0x1e8d0, 0x1e8d6}, 2573 {0x1e944, 0x1e94a}, 2574 {0xe0100, 0xe01ef} 2575 }; 2576 2577 return intable(combining, sizeof(combining), c); 2578 } 2579 2580 /* 2581 * Return TRUE for characters that can be displayed in a normal way. 2582 * Only for characters of 0x100 and above! 2583 */ 2584 int 2585 utf_printable(int c) 2586 { 2587 #ifdef USE_WCHAR_FUNCTIONS 2588 /* 2589 * Assume the iswprint() library function works better than our own stuff. 2590 */ 2591 return iswprint(c); 2592 #else 2593 // Sorted list of non-overlapping intervals. 2594 // 0xd800-0xdfff is reserved for UTF-16, actually illegal. 2595 static struct interval nonprint[] = 2596 { 2597 {0x070f, 0x070f}, {0x180b, 0x180e}, {0x200b, 0x200f}, {0x202a, 0x202e}, 2598 {0x206a, 0x206f}, {0xd800, 0xdfff}, {0xfeff, 0xfeff}, {0xfff9, 0xfffb}, 2599 {0xfffe, 0xffff} 2600 }; 2601 2602 return !intable(nonprint, sizeof(nonprint), c); 2603 #endif 2604 } 2605 2606 // Sorted list of non-overlapping intervals of all Emoji characters, 2607 // based on http://unicode.org/emoji/charts/emoji-list.html 2608 // Generated by ../runtime/tools/unicode.vim. 2609 // Excludes 0x00a9 and 0x00ae because they are considered latin1. 2610 static struct interval emoji_all[] = 2611 { 2612 {0x203c, 0x203c}, 2613 {0x2049, 0x2049}, 2614 {0x2122, 0x2122}, 2615 {0x2139, 0x2139}, 2616 {0x2194, 0x2199}, 2617 {0x21a9, 0x21aa}, 2618 {0x231a, 0x231b}, 2619 {0x2328, 0x2328}, 2620 {0x23cf, 0x23cf}, 2621 {0x23e9, 0x23f3}, 2622 {0x23f8, 0x23fa}, 2623 {0x24c2, 0x24c2}, 2624 {0x25aa, 0x25ab}, 2625 {0x25b6, 0x25b6}, 2626 {0x25c0, 0x25c0}, 2627 {0x25fb, 0x25fe}, 2628 {0x2600, 0x2604}, 2629 {0x260e, 0x260e}, 2630 {0x2611, 0x2611}, 2631 {0x2614, 0x2615}, 2632 {0x2618, 0x2618}, 2633 {0x261d, 0x261d}, 2634 {0x2620, 0x2620}, 2635 {0x2622, 0x2623}, 2636 {0x2626, 0x2626}, 2637 {0x262a, 0x262a}, 2638 {0x262e, 0x262f}, 2639 {0x2638, 0x263a}, 2640 {0x2640, 0x2640}, 2641 {0x2642, 0x2642}, 2642 {0x2648, 0x2653}, 2643 {0x265f, 0x2660}, 2644 {0x2663, 0x2663}, 2645 {0x2665, 0x2666}, 2646 {0x2668, 0x2668}, 2647 {0x267b, 0x267b}, 2648 {0x267e, 0x267f}, 2649 {0x2692, 0x2697}, 2650 {0x2699, 0x2699}, 2651 {0x269b, 0x269c}, 2652 {0x26a0, 0x26a1}, 2653 {0x26aa, 0x26ab}, 2654 {0x26b0, 0x26b1}, 2655 {0x26bd, 0x26be}, 2656 {0x26c4, 0x26c5}, 2657 {0x26c8, 0x26c8}, 2658 {0x26ce, 0x26cf}, 2659 {0x26d1, 0x26d1}, 2660 {0x26d3, 0x26d4}, 2661 {0x26e9, 0x26ea}, 2662 {0x26f0, 0x26f5}, 2663 {0x26f7, 0x26fa}, 2664 {0x26fd, 0x26fd}, 2665 {0x2702, 0x2702}, 2666 {0x2705, 0x2705}, 2667 {0x2708, 0x270d}, 2668 {0x270f, 0x270f}, 2669 {0x2712, 0x2712}, 2670 {0x2714, 0x2714}, 2671 {0x2716, 0x2716}, 2672 {0x271d, 0x271d}, 2673 {0x2721, 0x2721}, 2674 {0x2728, 0x2728}, 2675 {0x2733, 0x2734}, 2676 {0x2744, 0x2744}, 2677 {0x2747, 0x2747}, 2678 {0x274c, 0x274c}, 2679 {0x274e, 0x274e}, 2680 {0x2753, 0x2755}, 2681 {0x2757, 0x2757}, 2682 {0x2763, 0x2764}, 2683 {0x2795, 0x2797}, 2684 {0x27a1, 0x27a1}, 2685 {0x27b0, 0x27b0}, 2686 {0x27bf, 0x27bf}, 2687 {0x2934, 0x2935}, 2688 {0x2b05, 0x2b07}, 2689 {0x2b1b, 0x2b1c}, 2690 {0x2b50, 0x2b50}, 2691 {0x2b55, 0x2b55}, 2692 {0x3030, 0x3030}, 2693 {0x303d, 0x303d}, 2694 {0x3297, 0x3297}, 2695 {0x3299, 0x3299}, 2696 {0x1f004, 0x1f004}, 2697 {0x1f0cf, 0x1f0cf}, 2698 {0x1f170, 0x1f189}, 2699 {0x1f18e, 0x1f18e}, 2700 {0x1f191, 0x1f19a}, 2701 {0x1f1e6, 0x1f1ff}, 2702 {0x1f201, 0x1f202}, 2703 {0x1f21a, 0x1f21a}, 2704 {0x1f22f, 0x1f22f}, 2705 {0x1f232, 0x1f23a}, 2706 {0x1f250, 0x1f251}, 2707 {0x1f300, 0x1f321}, 2708 {0x1f324, 0x1f393}, 2709 {0x1f396, 0x1f397}, 2710 {0x1f399, 0x1f39b}, 2711 {0x1f39e, 0x1f3f0}, 2712 {0x1f3f3, 0x1f3f5}, 2713 {0x1f3f7, 0x1f4fd}, 2714 {0x1f4ff, 0x1f53d}, 2715 {0x1f549, 0x1f54e}, 2716 {0x1f550, 0x1f567}, 2717 {0x1f56f, 0x1f570}, 2718 {0x1f573, 0x1f57a}, 2719 {0x1f587, 0x1f587}, 2720 {0x1f58a, 0x1f58d}, 2721 {0x1f590, 0x1f590}, 2722 {0x1f595, 0x1f596}, 2723 {0x1f5a4, 0x1f5a5}, 2724 {0x1f5a8, 0x1f5a8}, 2725 {0x1f5b1, 0x1f5b2}, 2726 {0x1f5bc, 0x1f5bc}, 2727 {0x1f5c2, 0x1f5c4}, 2728 {0x1f5d1, 0x1f5d3}, 2729 {0x1f5dc, 0x1f5de}, 2730 {0x1f5e1, 0x1f5e1}, 2731 {0x1f5e3, 0x1f5e3}, 2732 {0x1f5e8, 0x1f5e8}, 2733 {0x1f5ef, 0x1f5ef}, 2734 {0x1f5f3, 0x1f5f3}, 2735 {0x1f5fa, 0x1f64f}, 2736 {0x1f680, 0x1f6c5}, 2737 {0x1f6cb, 0x1f6d2}, 2738 {0x1f6e0, 0x1f6e5}, 2739 {0x1f6e9, 0x1f6e9}, 2740 {0x1f6eb, 0x1f6ec}, 2741 {0x1f6f0, 0x1f6f0}, 2742 {0x1f6f3, 0x1f6f9}, 2743 {0x1f910, 0x1f93a}, 2744 {0x1f93c, 0x1f93e}, 2745 {0x1f940, 0x1f945}, 2746 {0x1f947, 0x1f970}, 2747 {0x1f973, 0x1f976}, 2748 {0x1f97a, 0x1f97a}, 2749 {0x1f97c, 0x1f9a2}, 2750 {0x1f9b0, 0x1f9b9}, 2751 {0x1f9c0, 0x1f9c2}, 2752 {0x1f9d0, 0x1f9ff} 2753 }; 2754 2755 /* 2756 * Get class of a Unicode character. 2757 * 0: white space 2758 * 1: punctuation 2759 * 2 or bigger: some class of word character. 2760 */ 2761 int 2762 utf_class(int c) 2763 { 2764 return utf_class_buf(c, curbuf); 2765 } 2766 2767 int 2768 utf_class_buf(int c, buf_T *buf) 2769 { 2770 // sorted list of non-overlapping intervals 2771 static struct clinterval 2772 { 2773 unsigned int first; 2774 unsigned int last; 2775 unsigned int class; 2776 } classes[] = 2777 { 2778 {0x037e, 0x037e, 1}, // Greek question mark 2779 {0x0387, 0x0387, 1}, // Greek ano teleia 2780 {0x055a, 0x055f, 1}, // Armenian punctuation 2781 {0x0589, 0x0589, 1}, // Armenian full stop 2782 {0x05be, 0x05be, 1}, 2783 {0x05c0, 0x05c0, 1}, 2784 {0x05c3, 0x05c3, 1}, 2785 {0x05f3, 0x05f4, 1}, 2786 {0x060c, 0x060c, 1}, 2787 {0x061b, 0x061b, 1}, 2788 {0x061f, 0x061f, 1}, 2789 {0x066a, 0x066d, 1}, 2790 {0x06d4, 0x06d4, 1}, 2791 {0x0700, 0x070d, 1}, // Syriac punctuation 2792 {0x0964, 0x0965, 1}, 2793 {0x0970, 0x0970, 1}, 2794 {0x0df4, 0x0df4, 1}, 2795 {0x0e4f, 0x0e4f, 1}, 2796 {0x0e5a, 0x0e5b, 1}, 2797 {0x0f04, 0x0f12, 1}, 2798 {0x0f3a, 0x0f3d, 1}, 2799 {0x0f85, 0x0f85, 1}, 2800 {0x104a, 0x104f, 1}, // Myanmar punctuation 2801 {0x10fb, 0x10fb, 1}, // Georgian punctuation 2802 {0x1361, 0x1368, 1}, // Ethiopic punctuation 2803 {0x166d, 0x166e, 1}, // Canadian Syl. punctuation 2804 {0x1680, 0x1680, 0}, 2805 {0x169b, 0x169c, 1}, 2806 {0x16eb, 0x16ed, 1}, 2807 {0x1735, 0x1736, 1}, 2808 {0x17d4, 0x17dc, 1}, // Khmer punctuation 2809 {0x1800, 0x180a, 1}, // Mongolian punctuation 2810 {0x2000, 0x200b, 0}, // spaces 2811 {0x200c, 0x2027, 1}, // punctuation and symbols 2812 {0x2028, 0x2029, 0}, 2813 {0x202a, 0x202e, 1}, // punctuation and symbols 2814 {0x202f, 0x202f, 0}, 2815 {0x2030, 0x205e, 1}, // punctuation and symbols 2816 {0x205f, 0x205f, 0}, 2817 {0x2060, 0x27ff, 1}, // punctuation and symbols 2818 {0x2070, 0x207f, 0x2070}, // superscript 2819 {0x2080, 0x2094, 0x2080}, // subscript 2820 {0x20a0, 0x27ff, 1}, // all kinds of symbols 2821 {0x2800, 0x28ff, 0x2800}, // braille 2822 {0x2900, 0x2998, 1}, // arrows, brackets, etc. 2823 {0x29d8, 0x29db, 1}, 2824 {0x29fc, 0x29fd, 1}, 2825 {0x2e00, 0x2e7f, 1}, // supplemental punctuation 2826 {0x3000, 0x3000, 0}, // ideographic space 2827 {0x3001, 0x3020, 1}, // ideographic punctuation 2828 {0x3030, 0x3030, 1}, 2829 {0x303d, 0x303d, 1}, 2830 {0x3040, 0x309f, 0x3040}, // Hiragana 2831 {0x30a0, 0x30ff, 0x30a0}, // Katakana 2832 {0x3300, 0x9fff, 0x4e00}, // CJK Ideographs 2833 {0xac00, 0xd7a3, 0xac00}, // Hangul Syllables 2834 {0xf900, 0xfaff, 0x4e00}, // CJK Ideographs 2835 {0xfd3e, 0xfd3f, 1}, 2836 {0xfe30, 0xfe6b, 1}, // punctuation forms 2837 {0xff00, 0xff0f, 1}, // half/fullwidth ASCII 2838 {0xff1a, 0xff20, 1}, // half/fullwidth ASCII 2839 {0xff3b, 0xff40, 1}, // half/fullwidth ASCII 2840 {0xff5b, 0xff65, 1}, // half/fullwidth ASCII 2841 {0x1d000, 0x1d24f, 1}, // Musical notation 2842 {0x1d400, 0x1d7ff, 1}, // Mathematical Alphanumeric Symbols 2843 {0x1f000, 0x1f2ff, 1}, // Game pieces; enclosed characters 2844 {0x1f300, 0x1f9ff, 1}, // Many symbol blocks 2845 {0x20000, 0x2a6df, 0x4e00}, // CJK Ideographs 2846 {0x2a700, 0x2b73f, 0x4e00}, // CJK Ideographs 2847 {0x2b740, 0x2b81f, 0x4e00}, // CJK Ideographs 2848 {0x2f800, 0x2fa1f, 0x4e00}, // CJK Ideographs 2849 }; 2850 2851 int bot = 0; 2852 int top = ARRAY_LENGTH(classes) - 1; 2853 int mid; 2854 2855 // First quick check for Latin1 characters, use 'iskeyword'. 2856 if (c < 0x100) 2857 { 2858 if (c == ' ' || c == '\t' || c == NUL || c == 0xa0) 2859 return 0; // blank 2860 if (vim_iswordc_buf(c, buf)) 2861 return 2; // word character 2862 return 1; // punctuation 2863 } 2864 2865 // emoji 2866 if (intable(emoji_all, sizeof(emoji_all), c)) 2867 return 3; 2868 2869 // binary search in table 2870 while (top >= bot) 2871 { 2872 mid = (bot + top) / 2; 2873 if (classes[mid].last < (unsigned int)c) 2874 bot = mid + 1; 2875 else if (classes[mid].first > (unsigned int)c) 2876 top = mid - 1; 2877 else 2878 return (int)classes[mid].class; 2879 } 2880 2881 // most other characters are "word" characters 2882 return 2; 2883 } 2884 2885 int 2886 utf_ambiguous_width(int c) 2887 { 2888 return c >= 0x80 && (intable(ambiguous, sizeof(ambiguous), c) 2889 || intable(emoji_all, sizeof(emoji_all), c)); 2890 } 2891 2892 /* 2893 * Code for Unicode case-dependent operations. Based on notes in 2894 * http://www.unicode.org/Public/UNIDATA/CaseFolding.txt 2895 * This code uses simple case folding, not full case folding. 2896 * Last updated for Unicode 5.2. 2897 */ 2898 2899 /* 2900 * The following tables are built by ../runtime/tools/unicode.vim. 2901 * They must be in numeric order, because we use binary search. 2902 * An entry such as {0x41,0x5a,1,32} means that Unicode characters in the 2903 * range from 0x41 to 0x5a inclusive, stepping by 1, are changed to 2904 * folded/upper/lower by adding 32. 2905 */ 2906 typedef struct 2907 { 2908 int rangeStart; 2909 int rangeEnd; 2910 int step; 2911 int offset; 2912 } convertStruct; 2913 2914 static convertStruct foldCase[] = 2915 { 2916 {0x41,0x5a,1,32}, 2917 {0xb5,0xb5,-1,775}, 2918 {0xc0,0xd6,1,32}, 2919 {0xd8,0xde,1,32}, 2920 {0x100,0x12e,2,1}, 2921 {0x132,0x136,2,1}, 2922 {0x139,0x147,2,1}, 2923 {0x14a,0x176,2,1}, 2924 {0x178,0x178,-1,-121}, 2925 {0x179,0x17d,2,1}, 2926 {0x17f,0x17f,-1,-268}, 2927 {0x181,0x181,-1,210}, 2928 {0x182,0x184,2,1}, 2929 {0x186,0x186,-1,206}, 2930 {0x187,0x187,-1,1}, 2931 {0x189,0x18a,1,205}, 2932 {0x18b,0x18b,-1,1}, 2933 {0x18e,0x18e,-1,79}, 2934 {0x18f,0x18f,-1,202}, 2935 {0x190,0x190,-1,203}, 2936 {0x191,0x191,-1,1}, 2937 {0x193,0x193,-1,205}, 2938 {0x194,0x194,-1,207}, 2939 {0x196,0x196,-1,211}, 2940 {0x197,0x197,-1,209}, 2941 {0x198,0x198,-1,1}, 2942 {0x19c,0x19c,-1,211}, 2943 {0x19d,0x19d,-1,213}, 2944 {0x19f,0x19f,-1,214}, 2945 {0x1a0,0x1a4,2,1}, 2946 {0x1a6,0x1a6,-1,218}, 2947 {0x1a7,0x1a7,-1,1}, 2948 {0x1a9,0x1a9,-1,218}, 2949 {0x1ac,0x1ac,-1,1}, 2950 {0x1ae,0x1ae,-1,218}, 2951 {0x1af,0x1af,-1,1}, 2952 {0x1b1,0x1b2,1,217}, 2953 {0x1b3,0x1b5,2,1}, 2954 {0x1b7,0x1b7,-1,219}, 2955 {0x1b8,0x1bc,4,1}, 2956 {0x1c4,0x1c4,-1,2}, 2957 {0x1c5,0x1c5,-1,1}, 2958 {0x1c7,0x1c7,-1,2}, 2959 {0x1c8,0x1c8,-1,1}, 2960 {0x1ca,0x1ca,-1,2}, 2961 {0x1cb,0x1db,2,1}, 2962 {0x1de,0x1ee,2,1}, 2963 {0x1f1,0x1f1,-1,2}, 2964 {0x1f2,0x1f4,2,1}, 2965 {0x1f6,0x1f6,-1,-97}, 2966 {0x1f7,0x1f7,-1,-56}, 2967 {0x1f8,0x21e,2,1}, 2968 {0x220,0x220,-1,-130}, 2969 {0x222,0x232,2,1}, 2970 {0x23a,0x23a,-1,10795}, 2971 {0x23b,0x23b,-1,1}, 2972 {0x23d,0x23d,-1,-163}, 2973 {0x23e,0x23e,-1,10792}, 2974 {0x241,0x241,-1,1}, 2975 {0x243,0x243,-1,-195}, 2976 {0x244,0x244,-1,69}, 2977 {0x245,0x245,-1,71}, 2978 {0x246,0x24e,2,1}, 2979 {0x345,0x345,-1,116}, 2980 {0x370,0x372,2,1}, 2981 {0x376,0x376,-1,1}, 2982 {0x37f,0x37f,-1,116}, 2983 {0x386,0x386,-1,38}, 2984 {0x388,0x38a,1,37}, 2985 {0x38c,0x38c,-1,64}, 2986 {0x38e,0x38f,1,63}, 2987 {0x391,0x3a1,1,32}, 2988 {0x3a3,0x3ab,1,32}, 2989 {0x3c2,0x3c2,-1,1}, 2990 {0x3cf,0x3cf,-1,8}, 2991 {0x3d0,0x3d0,-1,-30}, 2992 {0x3d1,0x3d1,-1,-25}, 2993 {0x3d5,0x3d5,-1,-15}, 2994 {0x3d6,0x3d6,-1,-22}, 2995 {0x3d8,0x3ee,2,1}, 2996 {0x3f0,0x3f0,-1,-54}, 2997 {0x3f1,0x3f1,-1,-48}, 2998 {0x3f4,0x3f4,-1,-60}, 2999 {0x3f5,0x3f5,-1,-64}, 3000 {0x3f7,0x3f7,-1,1}, 3001 {0x3f9,0x3f9,-1,-7}, 3002 {0x3fa,0x3fa,-1,1}, 3003 {0x3fd,0x3ff,1,-130}, 3004 {0x400,0x40f,1,80}, 3005 {0x410,0x42f,1,32}, 3006 {0x460,0x480,2,1}, 3007 {0x48a,0x4be,2,1}, 3008 {0x4c0,0x4c0,-1,15}, 3009 {0x4c1,0x4cd,2,1}, 3010 {0x4d0,0x52e,2,1}, 3011 {0x531,0x556,1,48}, 3012 {0x10a0,0x10c5,1,7264}, 3013 {0x10c7,0x10cd,6,7264}, 3014 {0x13f8,0x13fd,1,-8}, 3015 {0x1c80,0x1c80,-1,-6222}, 3016 {0x1c81,0x1c81,-1,-6221}, 3017 {0x1c82,0x1c82,-1,-6212}, 3018 {0x1c83,0x1c84,1,-6210}, 3019 {0x1c85,0x1c85,-1,-6211}, 3020 {0x1c86,0x1c86,-1,-6204}, 3021 {0x1c87,0x1c87,-1,-6180}, 3022 {0x1c88,0x1c88,-1,35267}, 3023 {0x1c90,0x1cba,1,-3008}, 3024 {0x1cbd,0x1cbf,1,-3008}, 3025 {0x1e00,0x1e94,2,1}, 3026 {0x1e9b,0x1e9b,-1,-58}, 3027 {0x1e9e,0x1e9e,-1,-7615}, 3028 {0x1ea0,0x1efe,2,1}, 3029 {0x1f08,0x1f0f,1,-8}, 3030 {0x1f18,0x1f1d,1,-8}, 3031 {0x1f28,0x1f2f,1,-8}, 3032 {0x1f38,0x1f3f,1,-8}, 3033 {0x1f48,0x1f4d,1,-8}, 3034 {0x1f59,0x1f5f,2,-8}, 3035 {0x1f68,0x1f6f,1,-8}, 3036 {0x1f88,0x1f8f,1,-8}, 3037 {0x1f98,0x1f9f,1,-8}, 3038 {0x1fa8,0x1faf,1,-8}, 3039 {0x1fb8,0x1fb9,1,-8}, 3040 {0x1fba,0x1fbb,1,-74}, 3041 {0x1fbc,0x1fbc,-1,-9}, 3042 {0x1fbe,0x1fbe,-1,-7173}, 3043 {0x1fc8,0x1fcb,1,-86}, 3044 {0x1fcc,0x1fcc,-1,-9}, 3045 {0x1fd8,0x1fd9,1,-8}, 3046 {0x1fda,0x1fdb,1,-100}, 3047 {0x1fe8,0x1fe9,1,-8}, 3048 {0x1fea,0x1feb,1,-112}, 3049 {0x1fec,0x1fec,-1,-7}, 3050 {0x1ff8,0x1ff9,1,-128}, 3051 {0x1ffa,0x1ffb,1,-126}, 3052 {0x1ffc,0x1ffc,-1,-9}, 3053 {0x2126,0x2126,-1,-7517}, 3054 {0x212a,0x212a,-1,-8383}, 3055 {0x212b,0x212b,-1,-8262}, 3056 {0x2132,0x2132,-1,28}, 3057 {0x2160,0x216f,1,16}, 3058 {0x2183,0x2183,-1,1}, 3059 {0x24b6,0x24cf,1,26}, 3060 {0x2c00,0x2c2e,1,48}, 3061 {0x2c60,0x2c60,-1,1}, 3062 {0x2c62,0x2c62,-1,-10743}, 3063 {0x2c63,0x2c63,-1,-3814}, 3064 {0x2c64,0x2c64,-1,-10727}, 3065 {0x2c67,0x2c6b,2,1}, 3066 {0x2c6d,0x2c6d,-1,-10780}, 3067 {0x2c6e,0x2c6e,-1,-10749}, 3068 {0x2c6f,0x2c6f,-1,-10783}, 3069 {0x2c70,0x2c70,-1,-10782}, 3070 {0x2c72,0x2c75,3,1}, 3071 {0x2c7e,0x2c7f,1,-10815}, 3072 {0x2c80,0x2ce2,2,1}, 3073 {0x2ceb,0x2ced,2,1}, 3074 {0x2cf2,0xa640,31054,1}, 3075 {0xa642,0xa66c,2,1}, 3076 {0xa680,0xa69a,2,1}, 3077 {0xa722,0xa72e,2,1}, 3078 {0xa732,0xa76e,2,1}, 3079 {0xa779,0xa77b,2,1}, 3080 {0xa77d,0xa77d,-1,-35332}, 3081 {0xa77e,0xa786,2,1}, 3082 {0xa78b,0xa78b,-1,1}, 3083 {0xa78d,0xa78d,-1,-42280}, 3084 {0xa790,0xa792,2,1}, 3085 {0xa796,0xa7a8,2,1}, 3086 {0xa7aa,0xa7aa,-1,-42308}, 3087 {0xa7ab,0xa7ab,-1,-42319}, 3088 {0xa7ac,0xa7ac,-1,-42315}, 3089 {0xa7ad,0xa7ad,-1,-42305}, 3090 {0xa7ae,0xa7ae,-1,-42308}, 3091 {0xa7b0,0xa7b0,-1,-42258}, 3092 {0xa7b1,0xa7b1,-1,-42282}, 3093 {0xa7b2,0xa7b2,-1,-42261}, 3094 {0xa7b3,0xa7b3,-1,928}, 3095 {0xa7b4,0xa7be,2,1}, 3096 {0xa7c2,0xa7c2,-1,1}, 3097 {0xa7c4,0xa7c4,-1,-48}, 3098 {0xa7c5,0xa7c5,-1,-42307}, 3099 {0xa7c6,0xa7c6,-1,-35384}, 3100 {0xab70,0xabbf,1,-38864}, 3101 {0xff21,0xff3a,1,32}, 3102 {0x10400,0x10427,1,40}, 3103 {0x104b0,0x104d3,1,40}, 3104 {0x10c80,0x10cb2,1,64}, 3105 {0x118a0,0x118bf,1,32}, 3106 {0x16e40,0x16e5f,1,32}, 3107 {0x1e900,0x1e921,1,34} 3108 }; 3109 3110 /* 3111 * Generic conversion function for case operations. 3112 * Return the converted equivalent of "a", which is a UCS-4 character. Use 3113 * the given conversion "table". Uses binary search on "table". 3114 */ 3115 static int 3116 utf_convert( 3117 int a, 3118 convertStruct table[], 3119 int tableSize) 3120 { 3121 int start, mid, end; // indices into table 3122 int entries = tableSize / sizeof(convertStruct); 3123 3124 start = 0; 3125 end = entries; 3126 while (start < end) 3127 { 3128 // need to search further 3129 mid = (end + start) / 2; 3130 if (table[mid].rangeEnd < a) 3131 start = mid + 1; 3132 else 3133 end = mid; 3134 } 3135 if (start < entries 3136 && table[start].rangeStart <= a 3137 && a <= table[start].rangeEnd 3138 && (a - table[start].rangeStart) % table[start].step == 0) 3139 return (a + table[start].offset); 3140 else 3141 return a; 3142 } 3143 3144 /* 3145 * Return the folded-case equivalent of "a", which is a UCS-4 character. Uses 3146 * simple case folding. 3147 */ 3148 int 3149 utf_fold(int a) 3150 { 3151 if (a < 0x80) 3152 // be fast for ASCII 3153 return a >= 0x41 && a <= 0x5a ? a + 32 : a; 3154 return utf_convert(a, foldCase, (int)sizeof(foldCase)); 3155 } 3156 3157 static convertStruct toLower[] = 3158 { 3159 {0x41,0x5a,1,32}, 3160 {0xc0,0xd6,1,32}, 3161 {0xd8,0xde,1,32}, 3162 {0x100,0x12e,2,1}, 3163 {0x130,0x130,-1,-199}, 3164 {0x132,0x136,2,1}, 3165 {0x139,0x147,2,1}, 3166 {0x14a,0x176,2,1}, 3167 {0x178,0x178,-1,-121}, 3168 {0x179,0x17d,2,1}, 3169 {0x181,0x181,-1,210}, 3170 {0x182,0x184,2,1}, 3171 {0x186,0x186,-1,206}, 3172 {0x187,0x187,-1,1}, 3173 {0x189,0x18a,1,205}, 3174 {0x18b,0x18b,-1,1}, 3175 {0x18e,0x18e,-1,79}, 3176 {0x18f,0x18f,-1,202}, 3177 {0x190,0x190,-1,203}, 3178 {0x191,0x191,-1,1}, 3179 {0x193,0x193,-1,205}, 3180 {0x194,0x194,-1,207}, 3181 {0x196,0x196,-1,211}, 3182 {0x197,0x197,-1,209}, 3183 {0x198,0x198,-1,1}, 3184 {0x19c,0x19c,-1,211}, 3185 {0x19d,0x19d,-1,213}, 3186 {0x19f,0x19f,-1,214}, 3187 {0x1a0,0x1a4,2,1}, 3188 {0x1a6,0x1a6,-1,218}, 3189 {0x1a7,0x1a7,-1,1}, 3190 {0x1a9,0x1a9,-1,218}, 3191 {0x1ac,0x1ac,-1,1}, 3192 {0x1ae,0x1ae,-1,218}, 3193 {0x1af,0x1af,-1,1}, 3194 {0x1b1,0x1b2,1,217}, 3195 {0x1b3,0x1b5,2,1}, 3196 {0x1b7,0x1b7,-1,219}, 3197 {0x1b8,0x1bc,4,1}, 3198 {0x1c4,0x1c4,-1,2}, 3199 {0x1c5,0x1c5,-1,1}, 3200 {0x1c7,0x1c7,-1,2}, 3201 {0x1c8,0x1c8,-1,1}, 3202 {0x1ca,0x1ca,-1,2}, 3203 {0x1cb,0x1db,2,1}, 3204 {0x1de,0x1ee,2,1}, 3205 {0x1f1,0x1f1,-1,2}, 3206 {0x1f2,0x1f4,2,1}, 3207 {0x1f6,0x1f6,-1,-97}, 3208 {0x1f7,0x1f7,-1,-56}, 3209 {0x1f8,0x21e,2,1}, 3210 {0x220,0x220,-1,-130}, 3211 {0x222,0x232,2,1}, 3212 {0x23a,0x23a,-1,10795}, 3213 {0x23b,0x23b,-1,1}, 3214 {0x23d,0x23d,-1,-163}, 3215 {0x23e,0x23e,-1,10792}, 3216 {0x241,0x241,-1,1}, 3217 {0x243,0x243,-1,-195}, 3218 {0x244,0x244,-1,69}, 3219 {0x245,0x245,-1,71}, 3220 {0x246,0x24e,2,1}, 3221 {0x370,0x372,2,1}, 3222 {0x376,0x376,-1,1}, 3223 {0x37f,0x37f,-1,116}, 3224 {0x386,0x386,-1,38}, 3225 {0x388,0x38a,1,37}, 3226 {0x38c,0x38c,-1,64}, 3227 {0x38e,0x38f,1,63}, 3228 {0x391,0x3a1,1,32}, 3229 {0x3a3,0x3ab,1,32}, 3230 {0x3cf,0x3cf,-1,8}, 3231 {0x3d8,0x3ee,2,1}, 3232 {0x3f4,0x3f4,-1,-60}, 3233 {0x3f7,0x3f7,-1,1}, 3234 {0x3f9,0x3f9,-1,-7}, 3235 {0x3fa,0x3fa,-1,1}, 3236 {0x3fd,0x3ff,1,-130}, 3237 {0x400,0x40f,1,80}, 3238 {0x410,0x42f,1,32}, 3239 {0x460,0x480,2,1}, 3240 {0x48a,0x4be,2,1}, 3241 {0x4c0,0x4c0,-1,15}, 3242 {0x4c1,0x4cd,2,1}, 3243 {0x4d0,0x52e,2,1}, 3244 {0x531,0x556,1,48}, 3245 {0x10a0,0x10c5,1,7264}, 3246 {0x10c7,0x10cd,6,7264}, 3247 {0x13a0,0x13ef,1,38864}, 3248 {0x13f0,0x13f5,1,8}, 3249 {0x1c90,0x1cba,1,-3008}, 3250 {0x1cbd,0x1cbf,1,-3008}, 3251 {0x1e00,0x1e94,2,1}, 3252 {0x1e9e,0x1e9e,-1,-7615}, 3253 {0x1ea0,0x1efe,2,1}, 3254 {0x1f08,0x1f0f,1,-8}, 3255 {0x1f18,0x1f1d,1,-8}, 3256 {0x1f28,0x1f2f,1,-8}, 3257 {0x1f38,0x1f3f,1,-8}, 3258 {0x1f48,0x1f4d,1,-8}, 3259 {0x1f59,0x1f5f,2,-8}, 3260 {0x1f68,0x1f6f,1,-8}, 3261 {0x1f88,0x1f8f,1,-8}, 3262 {0x1f98,0x1f9f,1,-8}, 3263 {0x1fa8,0x1faf,1,-8}, 3264 {0x1fb8,0x1fb9,1,-8}, 3265 {0x1fba,0x1fbb,1,-74}, 3266 {0x1fbc,0x1fbc,-1,-9}, 3267 {0x1fc8,0x1fcb,1,-86}, 3268 {0x1fcc,0x1fcc,-1,-9}, 3269 {0x1fd8,0x1fd9,1,-8}, 3270 {0x1fda,0x1fdb,1,-100}, 3271 {0x1fe8,0x1fe9,1,-8}, 3272 {0x1fea,0x1feb,1,-112}, 3273 {0x1fec,0x1fec,-1,-7}, 3274 {0x1ff8,0x1ff9,1,-128}, 3275 {0x1ffa,0x1ffb,1,-126}, 3276 {0x1ffc,0x1ffc,-1,-9}, 3277 {0x2126,0x2126,-1,-7517}, 3278 {0x212a,0x212a,-1,-8383}, 3279 {0x212b,0x212b,-1,-8262}, 3280 {0x2132,0x2132,-1,28}, 3281 {0x2160,0x216f,1,16}, 3282 {0x2183,0x2183,-1,1}, 3283 {0x24b6,0x24cf,1,26}, 3284 {0x2c00,0x2c2e,1,48}, 3285 {0x2c60,0x2c60,-1,1}, 3286 {0x2c62,0x2c62,-1,-10743}, 3287 {0x2c63,0x2c63,-1,-3814}, 3288 {0x2c64,0x2c64,-1,-10727}, 3289 {0x2c67,0x2c6b,2,1}, 3290 {0x2c6d,0x2c6d,-1,-10780}, 3291 {0x2c6e,0x2c6e,-1,-10749}, 3292 {0x2c6f,0x2c6f,-1,-10783}, 3293 {0x2c70,0x2c70,-1,-10782}, 3294 {0x2c72,0x2c75,3,1}, 3295 {0x2c7e,0x2c7f,1,-10815}, 3296 {0x2c80,0x2ce2,2,1}, 3297 {0x2ceb,0x2ced,2,1}, 3298 {0x2cf2,0xa640,31054,1}, 3299 {0xa642,0xa66c,2,1}, 3300 {0xa680,0xa69a,2,1}, 3301 {0xa722,0xa72e,2,1}, 3302 {0xa732,0xa76e,2,1}, 3303 {0xa779,0xa77b,2,1}, 3304 {0xa77d,0xa77d,-1,-35332}, 3305 {0xa77e,0xa786,2,1}, 3306 {0xa78b,0xa78b,-1,1}, 3307 {0xa78d,0xa78d,-1,-42280}, 3308 {0xa790,0xa792,2,1}, 3309 {0xa796,0xa7a8,2,1}, 3310 {0xa7aa,0xa7aa,-1,-42308}, 3311 {0xa7ab,0xa7ab,-1,-42319}, 3312 {0xa7ac,0xa7ac,-1,-42315}, 3313 {0xa7ad,0xa7ad,-1,-42305}, 3314 {0xa7ae,0xa7ae,-1,-42308}, 3315 {0xa7b0,0xa7b0,-1,-42258}, 3316 {0xa7b1,0xa7b1,-1,-42282}, 3317 {0xa7b2,0xa7b2,-1,-42261}, 3318 {0xa7b3,0xa7b3,-1,928}, 3319 {0xa7b4,0xa7be,2,1}, 3320 {0xa7c2,0xa7c2,-1,1}, 3321 {0xa7c4,0xa7c4,-1,-48}, 3322 {0xa7c5,0xa7c5,-1,-42307}, 3323 {0xa7c6,0xa7c6,-1,-35384}, 3324 {0xff21,0xff3a,1,32}, 3325 {0x10400,0x10427,1,40}, 3326 {0x104b0,0x104d3,1,40}, 3327 {0x10c80,0x10cb2,1,64}, 3328 {0x118a0,0x118bf,1,32}, 3329 {0x16e40,0x16e5f,1,32}, 3330 {0x1e900,0x1e921,1,34} 3331 }; 3332 3333 static convertStruct toUpper[] = 3334 { 3335 {0x61,0x7a,1,-32}, 3336 {0xb5,0xb5,-1,743}, 3337 {0xe0,0xf6,1,-32}, 3338 {0xf8,0xfe,1,-32}, 3339 {0xff,0xff,-1,121}, 3340 {0x101,0x12f,2,-1}, 3341 {0x131,0x131,-1,-232}, 3342 {0x133,0x137,2,-1}, 3343 {0x13a,0x148,2,-1}, 3344 {0x14b,0x177,2,-1}, 3345 {0x17a,0x17e,2,-1}, 3346 {0x17f,0x17f,-1,-300}, 3347 {0x180,0x180,-1,195}, 3348 {0x183,0x185,2,-1}, 3349 {0x188,0x18c,4,-1}, 3350 {0x192,0x192,-1,-1}, 3351 {0x195,0x195,-1,97}, 3352 {0x199,0x199,-1,-1}, 3353 {0x19a,0x19a,-1,163}, 3354 {0x19e,0x19e,-1,130}, 3355 {0x1a1,0x1a5,2,-1}, 3356 {0x1a8,0x1ad,5,-1}, 3357 {0x1b0,0x1b4,4,-1}, 3358 {0x1b6,0x1b9,3,-1}, 3359 {0x1bd,0x1bd,-1,-1}, 3360 {0x1bf,0x1bf,-1,56}, 3361 {0x1c5,0x1c5,-1,-1}, 3362 {0x1c6,0x1c6,-1,-2}, 3363 {0x1c8,0x1c8,-1,-1}, 3364 {0x1c9,0x1c9,-1,-2}, 3365 {0x1cb,0x1cb,-1,-1}, 3366 {0x1cc,0x1cc,-1,-2}, 3367 {0x1ce,0x1dc,2,-1}, 3368 {0x1dd,0x1dd,-1,-79}, 3369 {0x1df,0x1ef,2,-1}, 3370 {0x1f2,0x1f2,-1,-1}, 3371 {0x1f3,0x1f3,-1,-2}, 3372 {0x1f5,0x1f9,4,-1}, 3373 {0x1fb,0x21f,2,-1}, 3374 {0x223,0x233,2,-1}, 3375 {0x23c,0x23c,-1,-1}, 3376 {0x23f,0x240,1,10815}, 3377 {0x242,0x247,5,-1}, 3378 {0x249,0x24f,2,-1}, 3379 {0x250,0x250,-1,10783}, 3380 {0x251,0x251,-1,10780}, 3381 {0x252,0x252,-1,10782}, 3382 {0x253,0x253,-1,-210}, 3383 {0x254,0x254,-1,-206}, 3384 {0x256,0x257,1,-205}, 3385 {0x259,0x259,-1,-202}, 3386 {0x25b,0x25b,-1,-203}, 3387 {0x25c,0x25c,-1,42319}, 3388 {0x260,0x260,-1,-205}, 3389 {0x261,0x261,-1,42315}, 3390 {0x263,0x263,-1,-207}, 3391 {0x265,0x265,-1,42280}, 3392 {0x266,0x266,-1,42308}, 3393 {0x268,0x268,-1,-209}, 3394 {0x269,0x269,-1,-211}, 3395 {0x26a,0x26a,-1,42308}, 3396 {0x26b,0x26b,-1,10743}, 3397 {0x26c,0x26c,-1,42305}, 3398 {0x26f,0x26f,-1,-211}, 3399 {0x271,0x271,-1,10749}, 3400 {0x272,0x272,-1,-213}, 3401 {0x275,0x275,-1,-214}, 3402 {0x27d,0x27d,-1,10727}, 3403 {0x280,0x280,-1,-218}, 3404 {0x282,0x282,-1,42307}, 3405 {0x283,0x283,-1,-218}, 3406 {0x287,0x287,-1,42282}, 3407 {0x288,0x288,-1,-218}, 3408 {0x289,0x289,-1,-69}, 3409 {0x28a,0x28b,1,-217}, 3410 {0x28c,0x28c,-1,-71}, 3411 {0x292,0x292,-1,-219}, 3412 {0x29d,0x29d,-1,42261}, 3413 {0x29e,0x29e,-1,42258}, 3414 {0x345,0x345,-1,84}, 3415 {0x371,0x373,2,-1}, 3416 {0x377,0x377,-1,-1}, 3417 {0x37b,0x37d,1,130}, 3418 {0x3ac,0x3ac,-1,-38}, 3419 {0x3ad,0x3af,1,-37}, 3420 {0x3b1,0x3c1,1,-32}, 3421 {0x3c2,0x3c2,-1,-31}, 3422 {0x3c3,0x3cb,1,-32}, 3423 {0x3cc,0x3cc,-1,-64}, 3424 {0x3cd,0x3ce,1,-63}, 3425 {0x3d0,0x3d0,-1,-62}, 3426 {0x3d1,0x3d1,-1,-57}, 3427 {0x3d5,0x3d5,-1,-47}, 3428 {0x3d6,0x3d6,-1,-54}, 3429 {0x3d7,0x3d7,-1,-8}, 3430 {0x3d9,0x3ef,2,-1}, 3431 {0x3f0,0x3f0,-1,-86}, 3432 {0x3f1,0x3f1,-1,-80}, 3433 {0x3f2,0x3f2,-1,7}, 3434 {0x3f3,0x3f3,-1,-116}, 3435 {0x3f5,0x3f5,-1,-96}, 3436 {0x3f8,0x3fb,3,-1}, 3437 {0x430,0x44f,1,-32}, 3438 {0x450,0x45f,1,-80}, 3439 {0x461,0x481,2,-1}, 3440 {0x48b,0x4bf,2,-1}, 3441 {0x4c2,0x4ce,2,-1}, 3442 {0x4cf,0x4cf,-1,-15}, 3443 {0x4d1,0x52f,2,-1}, 3444 {0x561,0x586,1,-48}, 3445 {0x10d0,0x10fa,1,3008}, 3446 {0x10fd,0x10ff,1,3008}, 3447 {0x13f8,0x13fd,1,-8}, 3448 {0x1c80,0x1c80,-1,-6254}, 3449 {0x1c81,0x1c81,-1,-6253}, 3450 {0x1c82,0x1c82,-1,-6244}, 3451 {0x1c83,0x1c84,1,-6242}, 3452 {0x1c85,0x1c85,-1,-6243}, 3453 {0x1c86,0x1c86,-1,-6236}, 3454 {0x1c87,0x1c87,-1,-6181}, 3455 {0x1c88,0x1c88,-1,35266}, 3456 {0x1d79,0x1d79,-1,35332}, 3457 {0x1d7d,0x1d7d,-1,3814}, 3458 {0x1d8e,0x1d8e,-1,35384}, 3459 {0x1e01,0x1e95,2,-1}, 3460 {0x1e9b,0x1e9b,-1,-59}, 3461 {0x1ea1,0x1eff,2,-1}, 3462 {0x1f00,0x1f07,1,8}, 3463 {0x1f10,0x1f15,1,8}, 3464 {0x1f20,0x1f27,1,8}, 3465 {0x1f30,0x1f37,1,8}, 3466 {0x1f40,0x1f45,1,8}, 3467 {0x1f51,0x1f57,2,8}, 3468 {0x1f60,0x1f67,1,8}, 3469 {0x1f70,0x1f71,1,74}, 3470 {0x1f72,0x1f75,1,86}, 3471 {0x1f76,0x1f77,1,100}, 3472 {0x1f78,0x1f79,1,128}, 3473 {0x1f7a,0x1f7b,1,112}, 3474 {0x1f7c,0x1f7d,1,126}, 3475 {0x1f80,0x1f87,1,8}, 3476 {0x1f90,0x1f97,1,8}, 3477 {0x1fa0,0x1fa7,1,8}, 3478 {0x1fb0,0x1fb1,1,8}, 3479 {0x1fb3,0x1fb3,-1,9}, 3480 {0x1fbe,0x1fbe,-1,-7205}, 3481 {0x1fc3,0x1fc3,-1,9}, 3482 {0x1fd0,0x1fd1,1,8}, 3483 {0x1fe0,0x1fe1,1,8}, 3484 {0x1fe5,0x1fe5,-1,7}, 3485 {0x1ff3,0x1ff3,-1,9}, 3486 {0x214e,0x214e,-1,-28}, 3487 {0x2170,0x217f,1,-16}, 3488 {0x2184,0x2184,-1,-1}, 3489 {0x24d0,0x24e9,1,-26}, 3490 {0x2c30,0x2c5e,1,-48}, 3491 {0x2c61,0x2c61,-1,-1}, 3492 {0x2c65,0x2c65,-1,-10795}, 3493 {0x2c66,0x2c66,-1,-10792}, 3494 {0x2c68,0x2c6c,2,-1}, 3495 {0x2c73,0x2c76,3,-1}, 3496 {0x2c81,0x2ce3,2,-1}, 3497 {0x2cec,0x2cee,2,-1}, 3498 {0x2cf3,0x2cf3,-1,-1}, 3499 {0x2d00,0x2d25,1,-7264}, 3500 {0x2d27,0x2d2d,6,-7264}, 3501 {0xa641,0xa66d,2,-1}, 3502 {0xa681,0xa69b,2,-1}, 3503 {0xa723,0xa72f,2,-1}, 3504 {0xa733,0xa76f,2,-1}, 3505 {0xa77a,0xa77c,2,-1}, 3506 {0xa77f,0xa787,2,-1}, 3507 {0xa78c,0xa791,5,-1}, 3508 {0xa793,0xa793,-1,-1}, 3509 {0xa794,0xa794,-1,48}, 3510 {0xa797,0xa7a9,2,-1}, 3511 {0xa7b5,0xa7bf,2,-1}, 3512 {0xa7c3,0xa7c3,-1,-1}, 3513 {0xab53,0xab53,-1,-928}, 3514 {0xab70,0xabbf,1,-38864}, 3515 {0xff41,0xff5a,1,-32}, 3516 {0x10428,0x1044f,1,-40}, 3517 {0x104d8,0x104fb,1,-40}, 3518 {0x10cc0,0x10cf2,1,-64}, 3519 {0x118c0,0x118df,1,-32}, 3520 {0x16e60,0x16e7f,1,-32}, 3521 {0x1e922,0x1e943,1,-34} 3522 }; 3523 3524 /* 3525 * Return the upper-case equivalent of "a", which is a UCS-4 character. Use 3526 * simple case folding. 3527 */ 3528 int 3529 utf_toupper(int a) 3530 { 3531 // If 'casemap' contains "keepascii" use ASCII style toupper(). 3532 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3533 return TOUPPER_ASC(a); 3534 3535 #if defined(HAVE_TOWUPPER) && defined(__STDC_ISO_10646__) 3536 // If towupper() is available and handles Unicode, use it. 3537 if (!(cmp_flags & CMP_INTERNAL)) 3538 return towupper(a); 3539 #endif 3540 3541 // For characters below 128 use locale sensitive toupper(). 3542 if (a < 128) 3543 return TOUPPER_LOC(a); 3544 3545 // For any other characters use the above mapping table. 3546 return utf_convert(a, toUpper, (int)sizeof(toUpper)); 3547 } 3548 3549 int 3550 utf_islower(int a) 3551 { 3552 // German sharp s is lower case but has no upper case equivalent. 3553 return (utf_toupper(a) != a) || a == 0xdf; 3554 } 3555 3556 /* 3557 * Return the lower-case equivalent of "a", which is a UCS-4 character. Use 3558 * simple case folding. 3559 */ 3560 int 3561 utf_tolower(int a) 3562 { 3563 // If 'casemap' contains "keepascii" use ASCII style tolower(). 3564 if (a < 128 && (cmp_flags & CMP_KEEPASCII)) 3565 return TOLOWER_ASC(a); 3566 3567 #if defined(HAVE_TOWLOWER) && defined(__STDC_ISO_10646__) 3568 // If towlower() is available and handles Unicode, use it. 3569 if (!(cmp_flags & CMP_INTERNAL)) 3570 return towlower(a); 3571 #endif 3572 3573 // For characters below 128 use locale sensitive tolower(). 3574 if (a < 128) 3575 return TOLOWER_LOC(a); 3576 3577 // For any other characters use the above mapping table. 3578 return utf_convert(a, toLower, (int)sizeof(toLower)); 3579 } 3580 3581 int 3582 utf_isupper(int a) 3583 { 3584 return (utf_tolower(a) != a); 3585 } 3586 3587 static int 3588 utf_strnicmp( 3589 char_u *s1, 3590 char_u *s2, 3591 size_t n1, 3592 size_t n2) 3593 { 3594 int c1, c2, cdiff; 3595 char_u buffer[6]; 3596 3597 for (;;) 3598 { 3599 c1 = utf_safe_read_char_adv(&s1, &n1); 3600 c2 = utf_safe_read_char_adv(&s2, &n2); 3601 3602 if (c1 <= 0 || c2 <= 0) 3603 break; 3604 3605 if (c1 == c2) 3606 continue; 3607 3608 cdiff = utf_fold(c1) - utf_fold(c2); 3609 if (cdiff != 0) 3610 return cdiff; 3611 } 3612 3613 // some string ended or has an incomplete/illegal character sequence 3614 3615 if (c1 == 0 || c2 == 0) 3616 { 3617 // some string ended. shorter string is smaller 3618 if (c1 == 0 && c2 == 0) 3619 return 0; 3620 return c1 == 0 ? -1 : 1; 3621 } 3622 3623 // Continue with bytewise comparison to produce some result that 3624 // would make comparison operations involving this function transitive. 3625 // 3626 // If only one string had an error, comparison should be made with 3627 // folded version of the other string. In this case it is enough 3628 // to fold just one character to determine the result of comparison. 3629 3630 if (c1 != -1 && c2 == -1) 3631 { 3632 n1 = utf_char2bytes(utf_fold(c1), buffer); 3633 s1 = buffer; 3634 } 3635 else if (c2 != -1 && c1 == -1) 3636 { 3637 n2 = utf_char2bytes(utf_fold(c2), buffer); 3638 s2 = buffer; 3639 } 3640 3641 while (n1 > 0 && n2 > 0 && *s1 != NUL && *s2 != NUL) 3642 { 3643 cdiff = (int)(*s1) - (int)(*s2); 3644 if (cdiff != 0) 3645 return cdiff; 3646 3647 s1++; 3648 s2++; 3649 n1--; 3650 n2--; 3651 } 3652 3653 if (n1 > 0 && *s1 == NUL) 3654 n1 = 0; 3655 if (n2 > 0 && *s2 == NUL) 3656 n2 = 0; 3657 3658 if (n1 == 0 && n2 == 0) 3659 return 0; 3660 return n1 == 0 ? -1 : 1; 3661 } 3662 3663 /* 3664 * Version of strnicmp() that handles multi-byte characters. 3665 * Needed for Big5, Shift-JIS and UTF-8 encoding. Other DBCS encodings can 3666 * probably use strnicmp(), because there are no ASCII characters in the 3667 * second byte. 3668 * Returns zero if s1 and s2 are equal (ignoring case), the difference between 3669 * two characters otherwise. 3670 */ 3671 int 3672 mb_strnicmp(char_u *s1, char_u *s2, size_t nn) 3673 { 3674 int i, l; 3675 int cdiff; 3676 int n = (int)nn; 3677 3678 if (enc_utf8) 3679 { 3680 return utf_strnicmp(s1, s2, nn, nn); 3681 } 3682 else 3683 { 3684 for (i = 0; i < n; i += l) 3685 { 3686 if (s1[i] == NUL && s2[i] == NUL) // both strings end 3687 return 0; 3688 3689 l = (*mb_ptr2len)(s1 + i); 3690 if (l <= 1) 3691 { 3692 // Single byte: first check normally, then with ignore case. 3693 if (s1[i] != s2[i]) 3694 { 3695 cdiff = MB_TOLOWER(s1[i]) - MB_TOLOWER(s2[i]); 3696 if (cdiff != 0) 3697 return cdiff; 3698 } 3699 } 3700 else 3701 { 3702 // For non-Unicode multi-byte don't ignore case. 3703 if (l > n - i) 3704 l = n - i; 3705 cdiff = STRNCMP(s1 + i, s2 + i, l); 3706 if (cdiff != 0) 3707 return cdiff; 3708 } 3709 } 3710 } 3711 return 0; 3712 } 3713 3714 /* 3715 * "g8": show bytes of the UTF-8 char under the cursor. Doesn't matter what 3716 * 'encoding' has been set to. 3717 */ 3718 void 3719 show_utf8(void) 3720 { 3721 int len; 3722 int rlen = 0; 3723 char_u *line; 3724 int clen; 3725 int i; 3726 3727 // Get the byte length of the char under the cursor, including composing 3728 // characters. 3729 line = ml_get_cursor(); 3730 len = utfc_ptr2len(line); 3731 if (len == 0) 3732 { 3733 msg("NUL"); 3734 return; 3735 } 3736 3737 clen = 0; 3738 for (i = 0; i < len; ++i) 3739 { 3740 if (clen == 0) 3741 { 3742 // start of (composing) character, get its length 3743 if (i > 0) 3744 { 3745 STRCPY(IObuff + rlen, "+ "); 3746 rlen += 2; 3747 } 3748 clen = utf_ptr2len(line + i); 3749 } 3750 sprintf((char *)IObuff + rlen, "%02x ", 3751 (line[i] == NL) ? NUL : line[i]); // NUL is stored as NL 3752 --clen; 3753 rlen += (int)STRLEN(IObuff + rlen); 3754 if (rlen > IOSIZE - 20) 3755 break; 3756 } 3757 3758 msg((char *)IObuff); 3759 } 3760 3761 /* 3762 * mb_head_off() function pointer. 3763 * Return offset from "p" to the first byte of the character it points into. 3764 * If "p" points to the NUL at the end of the string return 0. 3765 * Returns 0 when already at the first byte of a character. 3766 */ 3767 int 3768 latin_head_off(char_u *base UNUSED, char_u *p UNUSED) 3769 { 3770 return 0; 3771 } 3772 3773 static int 3774 dbcs_head_off(char_u *base, char_u *p) 3775 { 3776 char_u *q; 3777 3778 // It can't be a trailing byte when not using DBCS, at the start of the 3779 // string or the previous byte can't start a double-byte. 3780 if (p <= base || MB_BYTE2LEN(p[-1]) == 1 || *p == NUL) 3781 return 0; 3782 3783 // This is slow: need to start at the base and go forward until the 3784 // byte we are looking for. Return 1 when we went past it, 0 otherwise. 3785 q = base; 3786 while (q < p) 3787 q += dbcs_ptr2len(q); 3788 return (q == p) ? 0 : 1; 3789 } 3790 3791 /* 3792 * Special version of dbcs_head_off() that works for ScreenLines[], where 3793 * single-width DBCS_JPNU characters are stored separately. 3794 */ 3795 int 3796 dbcs_screen_head_off(char_u *base, char_u *p) 3797 { 3798 char_u *q; 3799 3800 // It can't be a trailing byte when not using DBCS, at the start of the 3801 // string or the previous byte can't start a double-byte. 3802 // For euc-jp an 0x8e byte in the previous cell always means we have a 3803 // lead byte in the current cell. 3804 if (p <= base 3805 || (enc_dbcs == DBCS_JPNU && p[-1] == 0x8e) 3806 || MB_BYTE2LEN(p[-1]) == 1 3807 || *p == NUL) 3808 return 0; 3809 3810 // This is slow: need to start at the base and go forward until the 3811 // byte we are looking for. Return 1 when we went past it, 0 otherwise. 3812 // For DBCS_JPNU look out for 0x8e, which means the second byte is not 3813 // stored as the next byte. 3814 q = base; 3815 while (q < p) 3816 { 3817 if (enc_dbcs == DBCS_JPNU && *q == 0x8e) 3818 ++q; 3819 else 3820 q += dbcs_ptr2len(q); 3821 } 3822 return (q == p) ? 0 : 1; 3823 } 3824 3825 int 3826 utf_head_off(char_u *base, char_u *p) 3827 { 3828 char_u *q; 3829 char_u *s; 3830 int c; 3831 int len; 3832 #ifdef FEAT_ARABIC 3833 char_u *j; 3834 #endif 3835 3836 if (*p < 0x80) // be quick for ASCII 3837 return 0; 3838 3839 // Skip backwards over trailing bytes: 10xx.xxxx 3840 // Skip backwards again if on a composing char. 3841 for (q = p; ; --q) 3842 { 3843 // Move s to the last byte of this char. 3844 for (s = q; (s[1] & 0xc0) == 0x80; ++s) 3845 ; 3846 // Move q to the first byte of this char. 3847 while (q > base && (*q & 0xc0) == 0x80) 3848 --q; 3849 // Check for illegal sequence. Do allow an illegal byte after where we 3850 // started. 3851 len = utf8len_tab[*q]; 3852 if (len != (int)(s - q + 1) && len != (int)(p - q + 1)) 3853 return 0; 3854 3855 if (q <= base) 3856 break; 3857 3858 c = utf_ptr2char(q); 3859 if (utf_iscomposing(c)) 3860 continue; 3861 3862 #ifdef FEAT_ARABIC 3863 if (arabic_maycombine(c)) 3864 { 3865 // Advance to get a sneak-peak at the next char 3866 j = q; 3867 --j; 3868 // Move j to the first byte of this char. 3869 while (j > base && (*j & 0xc0) == 0x80) 3870 --j; 3871 if (arabic_combine(utf_ptr2char(j), c)) 3872 continue; 3873 } 3874 #endif 3875 break; 3876 } 3877 3878 return (int)(p - q); 3879 } 3880 3881 /* 3882 * Whether space is NOT allowed before/after 'c'. 3883 */ 3884 int 3885 utf_eat_space(int cc) 3886 { 3887 return ((cc >= 0x2000 && cc <= 0x206F) // General punctuations 3888 || (cc >= 0x2e00 && cc <= 0x2e7f) // Supplemental punctuations 3889 || (cc >= 0x3000 && cc <= 0x303f) // CJK symbols and punctuations 3890 || (cc >= 0xff01 && cc <= 0xff0f) // Full width ASCII punctuations 3891 || (cc >= 0xff1a && cc <= 0xff20) // .. 3892 || (cc >= 0xff3b && cc <= 0xff40) // .. 3893 || (cc >= 0xff5b && cc <= 0xff65)); // .. 3894 } 3895 3896 /* 3897 * Whether line break is allowed before "cc". 3898 */ 3899 int 3900 utf_allow_break_before(int cc) 3901 { 3902 static const int BOL_prohibition_punct[] = 3903 { 3904 '!', 3905 '%', 3906 ')', 3907 ',', 3908 ':', 3909 ';', 3910 '>', 3911 '?', 3912 ']', 3913 '}', 3914 0x2019, // ’ right single quotation mark 3915 0x201d, // ” right double quotation mark 3916 0x2020, // † dagger 3917 0x2021, // ‡ double dagger 3918 0x2026, // … horizontal ellipsis 3919 0x2030, // ‰ per mille sign 3920 0x2031, // ‱ per then thousand sign 3921 0x203c, // ‼ double exclamation mark 3922 0x2047, // ⁇ double question mark 3923 0x2048, // ⁈ question exclamation mark 3924 0x2049, // ⁉ exclamation question mark 3925 0x2103, // ℃ degree celsius 3926 0x2109, // ℉ degree fahrenheit 3927 0x3001, // 、 ideographic comma 3928 0x3002, // 。 ideographic full stop 3929 0x3009, // 〉 right angle bracket 3930 0x300b, // 》 right double angle bracket 3931 0x300d, // 」 right corner bracket 3932 0x300f, // 』 right white corner bracket 3933 0x3011, // 】 right black lenticular bracket 3934 0x3015, // 〕 right tortoise shell bracket 3935 0x3017, // 〗 right white lenticular bracket 3936 0x3019, // 〙 right white tortoise shell bracket 3937 0x301b, // 〛 right white square bracket 3938 0xff01, // ！ fullwidth exclamation mark 3939 0xff09, // ） fullwidth right parenthesis 3940 0xff0c, // ， fullwidth comma 3941 0xff0e, // ． fullwidth full stop 3942 0xff1a, // ： fullwidth colon 3943 0xff1b, // ； fullwidth semicolon 3944 0xff1f, // ？ fullwidth question mark 3945 0xff3d, // ］ fullwidth right square bracket 3946 0xff5d, // ｝ fullwidth right curly bracket 3947 }; 3948 3949 int first = 0; 3950 int last = ARRAY_LENGTH(BOL_prohibition_punct) - 1; 3951 int mid = 0; 3952 3953 while (first < last) 3954 { 3955 mid = (first + last)/2; 3956 3957 if (cc == BOL_prohibition_punct[mid]) 3958 return FALSE; 3959 else if (cc > BOL_prohibition_punct[mid]) 3960 first = mid + 1; 3961 else 3962 last = mid - 1; 3963 } 3964 3965 return cc != BOL_prohibition_punct[first]; 3966 } 3967 3968 /* 3969 * Whether line break is allowed after "cc". 3970 */ 3971 static int 3972 utf_allow_break_after(int cc) 3973 { 3974 static const int EOL_prohibition_punct[] = 3975 { 3976 '(', 3977 '<', 3978 '[', 3979 '`', 3980 '{', 3981 //0x2014, // — em dash 3982 0x2018, // ‘ left single quotation mark 3983 0x201c, // “ left double quotation mark 3984 //0x2053, // ～ swung dash 3985 0x3008, // 〈 left angle bracket 3986 0x300a, // 《 left double angle bracket 3987 0x300c, // 「 left corner bracket 3988 0x300e, // 『 left white corner bracket 3989 0x3010, // 【 left black lenticular bracket 3990 0x3014, // 〔 left tortoise shell bracket 3991 0x3016, // 〖 left white lenticular bracket 3992 0x3018, // 〘 left white tortoise shell bracket 3993 0x301a, // 〚 left white square bracket 3994 0xff08, // （ fullwidth left parenthesis 3995 0xff3b, // ［ fullwidth left square bracket 3996 0xff5b, // ｛ fullwidth left curly bracket 3997 }; 3998 3999 int first = 0; 4000 int last = ARRAY_LENGTH(EOL_prohibition_punct) - 1; 4001 int mid = 0; 4002 4003 while (first < last) 4004 { 4005 mid = (first + last)/2; 4006 4007 if (cc == EOL_prohibition_punct[mid]) 4008 return FALSE; 4009 else if (cc > EOL_prohibition_punct[mid]) 4010 first = mid + 1; 4011 else 4012 last = mid - 1; 4013 } 4014 4015 return cc != EOL_prohibition_punct[first]; 4016 } 4017 4018 /* 4019 * Whether line break is allowed between "cc" and "ncc". 4020 */ 4021 int 4022 utf_allow_break(int cc, int ncc) 4023 { 4024 // don't break between two-letter punctuations 4025 if (cc == ncc 4026 && (cc == 0x2014 // em dash 4027 || cc == 0x2026)) // horizontal ellipsis 4028 return FALSE; 4029 4030 return utf_allow_break_after(cc) && utf_allow_break_before(ncc); 4031 } 4032 4033 /* 4034 * Copy a character from "*fp" to "*tp" and advance the pointers. 4035 */ 4036 void 4037 mb_copy_char(char_u **fp, char_u **tp) 4038 { 4039 int l = (*mb_ptr2len)(*fp); 4040 4041 mch_memmove(*tp, *fp, (size_t)l); 4042 *tp += l; 4043 *fp += l; 4044 } 4045 4046 /* 4047 * Return the offset from "p" to the first byte of a character. When "p" is 4048 * at the start of a character 0 is returned, otherwise the offset to the next 4049 * character. Can start anywhere in a stream of bytes. 4050 */ 4051 int 4052 mb_off_next(char_u *base, char_u *p) 4053 { 4054 int i; 4055 int j; 4056 4057 if (enc_utf8) 4058 { 4059 if (*p < 0x80) // be quick for ASCII 4060 return 0; 4061 4062 // Find the next character that isn't 10xx.xxxx 4063 for (i = 0; (p[i] & 0xc0) == 0x80; ++i) 4064 ; 4065 if (i > 0) 4066 { 4067 // Check for illegal sequence. 4068 for (j = 0; p - j > base; ++j) 4069 if ((p[-j] & 0xc0) != 0x80) 4070 break; 4071 if (utf8len_tab[p[-j]] != i + j) 4072 return 0; 4073 } 4074 return i; 4075 } 4076 4077 // Only need to check if we're on a trail byte, it doesn't matter if we 4078 // want the offset to the next or current character. 4079 return (*mb_head_off)(base, p); 4080 } 4081 4082 /* 4083 * Return the offset from "p" to the last byte of the character it points 4084 * into. Can start anywhere in a stream of bytes. 4085 */ 4086 int 4087 mb_tail_off(char_u *base, char_u *p) 4088 { 4089 int i; 4090 int j; 4091 4092 if (*p == NUL) 4093 return 0; 4094 4095 if (enc_utf8) 4096 { 4097 // Find the last character that is 10xx.xxxx 4098 for (i = 0; (p[i + 1] & 0xc0) == 0x80; ++i) 4099 ; 4100 // Check for illegal sequence. 4101 for (j = 0; p - j > base; ++j) 4102 if ((p[-j] & 0xc0) != 0x80) 4103 break; 4104 if (utf8len_tab[p[-j]] != i + j + 1) 4105 return 0; 4106 return i; 4107 } 4108 4109 // It can't be the first byte if a double-byte when not using DBCS, at the 4110 // end of the string or the byte can't start a double-byte. 4111 if (enc_dbcs == 0 || p[1] == NUL || MB_BYTE2LEN(*p) == 1) 4112 return 0; 4113 4114 // Return 1 when on the lead byte, 0 when on the tail byte. 4115 return 1 - dbcs_head_off(base, p); 4116 } 4117 4118 /* 4119 * Find the next illegal byte sequence. 4120 */ 4121 void 4122 utf_find_illegal(void) 4123 { 4124 pos_T pos = curwin->w_cursor; 4125 char_u *p; 4126 int len; 4127 vimconv_T vimconv; 4128 char_u *tofree = NULL; 4129 4130 vimconv.vc_type = CONV_NONE; 4131 if (enc_utf8 && (enc_canon_props(curbuf->b_p_fenc) & ENC_8BIT)) 4132 { 4133 // 'encoding' is "utf-8" but we are editing a 8-bit encoded file, 4134 // possibly a utf-8 file with illegal bytes. Setup for conversion 4135 // from utf-8 to 'fileencoding'. 4136 convert_setup(&vimconv, p_enc, curbuf->b_p_fenc); 4137 } 4138 4139 curwin->w_cursor.coladd = 0; 4140 for (;;) 4141 { 4142 p = ml_get_cursor(); 4143 if (vimconv.vc_type != CONV_NONE) 4144 { 4145 vim_free(tofree); 4146 tofree = string_convert(&vimconv, p, NULL); 4147 if (tofree == NULL) 4148 break; 4149 p = tofree; 4150 } 4151 4152 while (*p != NUL) 4153 { 4154 // Illegal means that there are not enough trail bytes (checked by 4155 // utf_ptr2len()) or too many of them (overlong sequence). 4156 len = utf_ptr2len(p); 4157 if (*p >= 0x80 && (len == 1 4158 || utf_char2len(utf_ptr2char(p)) != len)) 4159 { 4160 if (vimconv.vc_type == CONV_NONE) 4161 curwin->w_cursor.col += (colnr_T)(p - ml_get_cursor()); 4162 else 4163 { 4164 int l; 4165 4166 len = (int)(p - tofree); 4167 for (p = ml_get_cursor(); *p != NUL && len-- > 0; p += l) 4168 { 4169 l = utf_ptr2len(p); 4170 curwin->w_cursor.col += l; 4171 } 4172 } 4173 goto theend; 4174 } 4175 p += len; 4176 } 4177 if (curwin->w_cursor.lnum == curbuf->b_ml.ml_line_count) 4178 break; 4179 ++curwin->w_cursor.lnum; 4180 curwin->w_cursor.col = 0; 4181 } 4182 4183 // didn't find it: don't move and beep 4184 curwin->w_cursor = pos; 4185 beep_flush(); 4186 4187 theend: 4188 vim_free(tofree); 4189 convert_setup(&vimconv, NULL, NULL); 4190 } 4191 4192 #if defined(FEAT_GUI_GTK) || defined(PROTO) 4193 /* 4194 * Return TRUE if string "s" is a valid utf-8 string. 4195 * When "end" is NULL stop at the first NUL. 4196 * When "end" is positive stop there. 4197 */ 4198 int 4199 utf_valid_string(char_u *s, char_u *end) 4200 { 4201 int l; 4202 char_u *p = s; 4203 4204 while (end == NULL ? *p != NUL : p < end) 4205 { 4206 l = utf8len_tab_zero[*p]; 4207 if (l == 0) 4208 return FALSE; // invalid lead byte 4209 if (end != NULL && p + l > end) 4210 return FALSE; // incomplete byte sequence 4211 ++p; 4212 while (--l > 0) 4213 if ((*p++ & 0xc0) != 0x80) 4214 return FALSE; // invalid trail byte 4215 } 4216 return TRUE; 4217 } 4218 #endif 4219 4220 #if defined(FEAT_GUI) || defined(PROTO) 4221 /* 4222 * Special version of mb_tail_off() for use in ScreenLines[]. 4223 */ 4224 int 4225 dbcs_screen_tail_off(char_u *base, char_u *p) 4226 { 4227 // It can't be the first byte if a double-byte when not using DBCS, at the 4228 // end of the string or the byte can't start a double-byte. 4229 // For euc-jp an 0x8e byte always means we have a lead byte in the current 4230 // cell. 4231 if (*p == NUL || p[1] == NUL 4232 || (enc_dbcs == DBCS_JPNU && *p == 0x8e) 4233 || MB_BYTE2LEN(*p) == 1) 4234 return 0; 4235 4236 // Return 1 when on the lead byte, 0 when on the tail byte. 4237 return 1 - dbcs_screen_head_off(base, p); 4238 } 4239 #endif 4240 4241 /* 4242 * If the cursor moves on an trail byte, set the cursor on the lead byte. 4243 * Thus it moves left if necessary. 4244 * Return TRUE when the cursor was adjusted. 4245 */ 4246 void 4247 mb_adjust_cursor(void) 4248 { 4249 mb_adjustpos(curbuf, &curwin->w_cursor); 4250 } 4251 4252 /* 4253 * Adjust position "*lp" to point to the first byte of a multi-byte character. 4254 * If it points to a tail byte it's moved backwards to the head byte. 4255 */ 4256 void 4257 mb_adjustpos(buf_T *buf, pos_T *lp) 4258 { 4259 char_u *p; 4260 4261 if (lp->col > 0 || lp->coladd > 1) 4262 { 4263 p = ml_get_buf(buf, lp->lnum, FALSE); 4264 if (*p == NUL || (int)STRLEN(p) < lp->col) 4265 lp->col = 0; 4266 else 4267 lp->col -= (*mb_head_off)(p, p + lp->col); 4268 // Reset "coladd" when the cursor would be on the right half of a 4269 // double-wide character. 4270 if (lp->coladd == 1 4271 && p[lp->col] != TAB 4272 && vim_isprintc((*mb_ptr2char)(p + lp->col)) 4273 && ptr2cells(p + lp->col) > 1) 4274 lp->coladd = 0; 4275 } 4276 } 4277 4278 /* 4279 * Return a pointer to the character before "*p", if there is one. 4280 */ 4281 char_u * 4282 mb_prevptr( 4283 char_u *line, // start of the string 4284 char_u *p) 4285 { 4286 if (p > line) 4287 MB_PTR_BACK(line, p); 4288 return p; 4289 } 4290 4291 /* 4292 * Return the character length of "str". Each multi-byte character (with 4293 * following composing characters) counts as one. 4294 */ 4295 int 4296 mb_charlen(char_u *str) 4297 { 4298 char_u *p = str; 4299 int count; 4300 4301 if (p == NULL) 4302 return 0; 4303 4304 for (count = 0; *p != NUL; count++) 4305 p += (*mb_ptr2len)(p); 4306 4307 return count; 4308 } 4309 4310 /* 4311 * Like mb_charlen() but for a string with specified length. 4312 */ 4313 int 4314 mb_charlen_len(char_u *str, int len) 4315 { 4316 char_u *p = str; 4317 int count; 4318 4319 for (count = 0; *p != NUL && p < str + len; count++) 4320 p += (*mb_ptr2len)(p); 4321 4322 return count; 4323 } 4324 4325 /* 4326 * Try to un-escape a multi-byte character. 4327 * Used for the "to" and "from" part of a mapping. 4328 * Return the un-escaped string if it is a multi-byte character, and advance 4329 * "pp" to just after the bytes that formed it. 4330 * Return NULL if no multi-byte char was found. 4331 */ 4332 char_u * 4333 mb_unescape(char_u **pp) 4334 { 4335 static char_u buf[6]; 4336 int n; 4337 int m = 0; 4338 char_u *str = *pp; 4339 4340 // Must translate K_SPECIAL KS_SPECIAL KE_FILLER to K_SPECIAL and CSI 4341 // KS_EXTRA KE_CSI to CSI. 4342 // Maximum length of a utf-8 character is 4 bytes. 4343 for (n = 0; str[n] != NUL && m < 4; ++n) 4344 { 4345 if (str[n] == K_SPECIAL 4346 && str[n + 1] == KS_SPECIAL 4347 && str[n + 2] == KE_FILLER) 4348 { 4349 buf[m++] = K_SPECIAL; 4350 n += 2; 4351 } 4352 else if ((str[n] == K_SPECIAL 4353 # ifdef FEAT_GUI 4354 || str[n] == CSI 4355 # endif 4356 ) 4357 && str[n + 1] == KS_EXTRA 4358 && str[n + 2] == (int)KE_CSI) 4359 { 4360 buf[m++] = CSI; 4361 n += 2; 4362 } 4363 else if (str[n] == K_SPECIAL 4364 # ifdef FEAT_GUI 4365 || str[n] == CSI 4366 # endif 4367 ) 4368 break; // a special key can't be a multibyte char 4369 else 4370 buf[m++] = str[n]; 4371 buf[m] = NUL; 4372 4373 // Return a multi-byte character if it's found. An illegal sequence 4374 // will result in a 1 here. 4375 if ((*mb_ptr2len)(buf) > 1) 4376 { 4377 *pp = str + n + 1; 4378 return buf; 4379 } 4380 4381 // Bail out quickly for ASCII. 4382 if (buf[0] < 128) 4383 break; 4384 } 4385 return NULL; 4386 } 4387 4388 /* 4389 * Return TRUE if the character at "row"/"col" on the screen is the left side 4390 * of a double-width character. 4391 * Caller must make sure "row" and "col" are not invalid! 4392 */ 4393 int 4394 mb_lefthalve(int row, int col) 4395 { 4396 return (*mb_off2cells)(LineOffset[row] + col, 4397 LineOffset[row] + screen_Columns) > 1; 4398 } 4399 4400 /* 4401 * Correct a position on the screen, if it's the right half of a double-wide 4402 * char move it to the left half. Returns the corrected column. 4403 */ 4404 int 4405 mb_fix_col(int col, int row) 4406 { 4407 int off; 4408 4409 col = check_col(col); 4410 row = check_row(row); 4411 off = LineOffset[row] + col; 4412 if (has_mbyte && ScreenLines != NULL && col > 0 4413 && ((enc_dbcs 4414 && ScreenLines[off] != NUL 4415 && dbcs_screen_head_off(ScreenLines + LineOffset[row], 4416 ScreenLines + off)) 4417 || (enc_utf8 && ScreenLines[off] == 0 4418 && ScreenLinesUC[off] == 0))) 4419 return col - 1; 4420 return col; 4421 } 4422 4423 static int enc_alias_search(char_u *name); 4424 4425 /* 4426 * Skip the Vim specific head of a 'encoding' name. 4427 */ 4428 char_u * 4429 enc_skip(char_u *p) 4430 { 4431 if (STRNCMP(p, "2byte-", 6) == 0) 4432 return p + 6; 4433 if (STRNCMP(p, "8bit-", 5) == 0) 4434 return p + 5; 4435 return p; 4436 } 4437 4438 /* 4439 * Find the canonical name for encoding "enc". 4440 * When the name isn't recognized, returns "enc" itself, but with all lower 4441 * case characters and '_' replaced with '-'. 4442 * Returns an allocated string. NULL for out-of-memory. 4443 */ 4444 char_u * 4445 enc_canonize(char_u *enc) 4446 { 4447 char_u *r; 4448 char_u *p, *s; 4449 int i; 4450 4451 if (STRCMP(enc, "default") == 0) 4452 { 4453 #ifdef MSWIN 4454 // Use the system encoding, the default is always utf-8. 4455 r = enc_locale(); 4456 #else 4457 // Use the default encoding as it's found by set_init_1(). 4458 r = get_encoding_default(); 4459 #endif 4460 if (r == NULL) 4461 r = (char_u *)ENC_DFLT; 4462 return vim_strsave(r); 4463 } 4464 4465 // copy "enc" to allocated memory, with room for two '-' 4466 r = alloc(STRLEN(enc) + 3); 4467 if (r != NULL) 4468 { 4469 // Make it all lower case and replace '_' with '-'. 4470 p = r; 4471 for (s = enc; *s != NUL; ++s) 4472 { 4473 if (*s == '_') 4474 *p++ = '-'; 4475 else 4476 *p++ = TOLOWER_ASC(*s); 4477 } 4478 *p = NUL; 4479 4480 // Skip "2byte-" and "8bit-". 4481 p = enc_skip(r); 4482 4483 // Change "microsoft-cp" to "cp". Used in some spell files. 4484 if (STRNCMP(p, "microsoft-cp", 12) == 0) 4485 STRMOVE(p, p + 10); 4486 4487 // "iso8859" -> "iso-8859" 4488 if (STRNCMP(p, "iso8859", 7) == 0) 4489 { 4490 STRMOVE(p + 4, p + 3); 4491 p[3] = '-'; 4492 } 4493 4494 // "iso-8859n" -> "iso-8859-n" 4495 if (STRNCMP(p, "iso-8859", 8) == 0 && p[8] != '-') 4496 { 4497 STRMOVE(p + 9, p + 8); 4498 p[8] = '-'; 4499 } 4500 4501 // "latin-N" -> "latinN" 4502 if (STRNCMP(p, "latin-", 6) == 0) 4503 STRMOVE(p + 5, p + 6); 4504 4505 if (enc_canon_search(p) >= 0) 4506 { 4507 // canonical name can be used unmodified 4508 if (p != r) 4509 STRMOVE(r, p); 4510 } 4511 else if ((i = enc_alias_search(p)) >= 0) 4512 { 4513 // alias recognized, get canonical name 4514 vim_free(r); 4515 r = vim_strsave((char_u *)enc_canon_table[i].name); 4516 } 4517 } 4518 return r; 4519 } 4520 4521 /* 4522 * Search for an encoding alias of "name". 4523 * Returns -1 when not found. 4524 */ 4525 static int 4526 enc_alias_search(char_u *name) 4527 { 4528 int i; 4529 4530 for (i = 0; enc_alias_table[i].name != NULL; ++i) 4531 if (STRCMP(name, enc_alias_table[i].name) == 0) 4532 return enc_alias_table[i].canon; 4533 return -1; 4534 } 4535 4536 4537 #ifdef HAVE_LANGINFO_H 4538 # include <langinfo.h> 4539 #endif 4540 4541 #if !defined(FEAT_GUI_MSWIN) || defined(VIMDLL) 4542 /* 4543 * Get the canonicalized encoding from the specified locale string "locale" 4544 * or from the environment variables LC_ALL, LC_CTYPE and LANG. 4545 * Returns an allocated string when successful, NULL when not. 4546 */ 4547 char_u * 4548 enc_locale_env(char *locale) 4549 { 4550 char *s = locale; 4551 char *p; 4552 int i; 4553 char buf[50]; 4554 4555 if (s == NULL || *s == NUL) 4556 if ((s = getenv("LC_ALL")) == NULL || *s == NUL) 4557 if ((s = getenv("LC_CTYPE")) == NULL || *s == NUL) 4558 s = getenv("LANG"); 4559 4560 if (s == NULL || *s == NUL) 4561 return NULL; 4562 4563 // The most generic locale format is: 4564 // language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]] 4565 // If there is a '.' remove the part before it. 4566 // if there is something after the codeset, remove it. 4567 // Make the name lowercase and replace '_' with '-'. 4568 // Exception: "ja_JP.EUC" == "euc-jp", "zh_CN.EUC" = "euc-cn", 4569 // "ko_KR.EUC" == "euc-kr" 4570 if ((p = (char *)vim_strchr((char_u *)s, '.')) != NULL) 4571 { 4572 if (p > s + 2 && STRNICMP(p + 1, "EUC", 3) == 0 4573 && !isalnum((int)p[4]) && p[4] != '-' && p[-3] == '_') 4574 { 4575 // copy "XY.EUC" to "euc-XY" to buf[10] 4576 STRCPY(buf + 10, "euc-"); 4577 buf[14] = p[-2]; 4578 buf[15] = p[-1]; 4579 buf[16] = 0; 4580 s = buf + 10; 4581 } 4582 else 4583 s = p + 1; 4584 } 4585 for (i = 0; i < (int)sizeof(buf) - 1 && s[i] != NUL; ++i) 4586 { 4587 if (s[i] == '_' || s[i] == '-') 4588 buf[i] = '-'; 4589 else if (isalnum((int)s[i])) 4590 buf[i] = TOLOWER_ASC(s[i]); 4591 else 4592 break; 4593 } 4594 buf[i] = NUL; 4595 4596 return enc_canonize((char_u *)buf); 4597 } 4598 #endif 4599 4600 /* 4601 * Get the canonicalized encoding of the current locale. 4602 * Returns an allocated string when successful, NULL when not. 4603 */ 4604 char_u * 4605 enc_locale(void) 4606 { 4607 #ifdef MSWIN 4608 char buf[50]; 4609 long acp = GetACP(); 4610 4611 if (acp == 1200) 4612 STRCPY(buf, "ucs-2le"); 4613 else if (acp == 1252) // cp1252 is used as latin1 4614 STRCPY(buf, "latin1"); 4615 else if (acp == 65001) 4616 STRCPY(buf, "utf-8"); 4617 else 4618 sprintf(buf, "cp%ld", acp); 4619 4620 return enc_canonize((char_u *)buf); 4621 #else 4622 char *s; 4623 4624 # ifdef HAVE_NL_LANGINFO_CODESET 4625 if ((s = nl_langinfo(CODESET)) == NULL || *s == NUL) 4626 # endif 4627 # if defined(HAVE_LOCALE_H) || defined(X_LOCALE) 4628 if ((s = setlocale(LC_CTYPE, NULL)) == NULL || *s == NUL) 4629 # endif 4630 s = NULL; 4631 4632 return enc_locale_env(s); 4633 #endif 4634 } 4635 4636 # if defined(MSWIN) || defined(PROTO) || defined(FEAT_CYGWIN_WIN32_CLIPBOARD) 4637 /* 4638 * Convert an encoding name to an MS-Windows codepage. 4639 * Returns zero if no codepage can be figured out. 4640 */ 4641 int 4642 encname2codepage(char_u *name) 4643 { 4644 int cp; 4645 char_u *p = name; 4646 int idx; 4647 4648 if (STRNCMP(p, "8bit-", 5) == 0) 4649 p += 5; 4650 else if (STRNCMP(p_enc, "2byte-", 6) == 0) 4651 p += 6; 4652 4653 if (p[0] == 'c' && p[1] == 'p') 4654 cp = atoi((char *)p + 2); 4655 else if ((idx = enc_canon_search(p)) >= 0) 4656 cp = enc_canon_table[idx].codepage; 4657 else 4658 return 0; 4659 if (IsValidCodePage(cp)) 4660 return cp; 4661 return 0; 4662 } 4663 # endif 4664 4665 # if defined(USE_ICONV) || defined(PROTO) 4666 4667 /* 4668 * Call iconv_open() with a check if iconv() works properly (there are broken 4669 * versions). 4670 * Returns (void *)-1 if failed. 4671 * (should return iconv_t, but that causes problems with prototypes). 4672 */ 4673 void * 4674 my_iconv_open(char_u *to, char_u *from) 4675 { 4676 iconv_t fd; 4677 #define ICONV_TESTLEN 400 4678 char_u tobuf[ICONV_TESTLEN]; 4679 char *p; 4680 size_t tolen; 4681 static int iconv_ok = -1; 4682 4683 if (iconv_ok == FALSE) 4684 return (void *)-1; // detected a broken iconv() previously 4685 4686 #ifdef DYNAMIC_ICONV 4687 // Check if the iconv.dll can be found. 4688 if (!iconv_enabled(TRUE)) 4689 return (void *)-1; 4690 #endif 4691 4692 fd = iconv_open((char *)enc_skip(to), (char *)enc_skip(from)); 4693 4694 if (fd != (iconv_t)-1 && iconv_ok == -1) 4695 { 4696 /* 4697 * Do a dummy iconv() call to check if it actually works. There is a 4698 * version of iconv() on Linux that is broken. We can't ignore it, 4699 * because it's wide-spread. The symptoms are that after outputting 4700 * the initial shift state the "to" pointer is NULL and conversion 4701 * stops for no apparent reason after about 8160 characters. 4702 */ 4703 p = (char *)tobuf; 4704 tolen = ICONV_TESTLEN; 4705 (void)iconv(fd, NULL, NULL, &p, &tolen); 4706 if (p == NULL) 4707 { 4708 iconv_ok = FALSE; 4709 iconv_close(fd); 4710 fd = (iconv_t)-1; 4711 } 4712 else 4713 iconv_ok = TRUE; 4714 } 4715 4716 return (void *)fd; 4717 } 4718 4719 /* 4720 * Convert the string "str[slen]" with iconv(). 4721 * If "unconvlenp" is not NULL handle the string ending in an incomplete 4722 * sequence and set "*unconvlenp" to the length of it. 4723 * Returns the converted string in allocated memory. NULL for an error. 4724 * If resultlenp is not NULL, sets it to the result length in bytes. 4725 */ 4726 static char_u * 4727 iconv_string( 4728 vimconv_T *vcp, 4729 char_u *str, 4730 int slen, 4731 int *unconvlenp, 4732 int *resultlenp) 4733 { 4734 const char *from; 4735 size_t fromlen; 4736 char *to; 4737 size_t tolen; 4738 size_t len = 0; 4739 size_t done = 0; 4740 char_u *result = NULL; 4741 char_u *p; 4742 int l; 4743 4744 from = (char *)str; 4745 fromlen = slen; 4746 for (;;) 4747 { 4748 if (len == 0 || ICONV_ERRNO == ICONV_E2BIG) 4749 { 4750 // Allocate enough room for most conversions. When re-allocating 4751 // increase the buffer size. 4752 len = len + fromlen * 2 + 40; 4753 p = alloc(len); 4754 if (p != NULL && done > 0) 4755 mch_memmove(p, result, done); 4756 vim_free(result); 4757 result = p; 4758 if (result == NULL) // out of memory 4759 break; 4760 } 4761 4762 to = (char *)result + done; 4763 tolen = len - done - 2; 4764 // Avoid a warning for systems with a wrong iconv() prototype by 4765 // casting the second argument to void *. 4766 if (iconv(vcp->vc_fd, (void *)&from, &fromlen, &to, &tolen) 4767 != (size_t)-1) 4768 { 4769 // Finished, append a NUL. 4770 *to = NUL; 4771 break; 4772 } 4773 4774 // Check both ICONV_EINVAL and EINVAL, because the dynamically loaded 4775 // iconv library may use one of them. 4776 if (!vcp->vc_fail && unconvlenp != NULL 4777 && (ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4778 { 4779 // Handle an incomplete sequence at the end. 4780 *to = NUL; 4781 *unconvlenp = (int)fromlen; 4782 break; 4783 } 4784 4785 // Check both ICONV_EILSEQ and EILSEQ, because the dynamically loaded 4786 // iconv library may use one of them. 4787 else if (!vcp->vc_fail 4788 && (ICONV_ERRNO == ICONV_EILSEQ || ICONV_ERRNO == EILSEQ 4789 || ICONV_ERRNO == ICONV_EINVAL || ICONV_ERRNO == EINVAL)) 4790 { 4791 // Can't convert: insert a '?' and skip a character. This assumes 4792 // conversion from 'encoding' to something else. In other 4793 // situations we don't know what to skip anyway. 4794 *to++ = '?'; 4795 if ((*mb_ptr2cells)((char_u *)from) > 1) 4796 *to++ = '?'; 4797 if (enc_utf8) 4798 l = utfc_ptr2len_len((char_u *)from, (int)fromlen); 4799 else 4800 { 4801 l = (*mb_ptr2len)((char_u *)from); 4802 if (l > (int)fromlen) 4803 l = (int)fromlen; 4804 } 4805 from += l; 4806 fromlen -= l; 4807 } 4808 else if (ICONV_ERRNO != ICONV_E2BIG) 4809 { 4810 // conversion failed 4811 VIM_CLEAR(result); 4812 break; 4813 } 4814 // Not enough room or skipping illegal sequence. 4815 done = to - (char *)result; 4816 } 4817 4818 if (resultlenp != NULL && result != NULL) 4819 *resultlenp = (int)(to - (char *)result); 4820 return result; 4821 } 4822 4823 # if defined(DYNAMIC_ICONV) || defined(PROTO) 4824 /* 4825 * Dynamically load the "iconv.dll" on Win32. 4826 */ 4827 4828 # ifndef DYNAMIC_ICONV // must be generating prototypes 4829 # define HINSTANCE int 4830 # endif 4831 static HINSTANCE hIconvDLL = 0; 4832 static HINSTANCE hMsvcrtDLL = 0; 4833 4834 # ifndef DYNAMIC_ICONV_DLL 4835 # define DYNAMIC_ICONV_DLL "iconv.dll" 4836 # define DYNAMIC_ICONV_DLL_ALT1 "libiconv.dll" 4837 # define DYNAMIC_ICONV_DLL_ALT2 "libiconv2.dll" 4838 # define DYNAMIC_ICONV_DLL_ALT3 "libiconv-2.dll" 4839 # endif 4840 # ifndef DYNAMIC_MSVCRT_DLL 4841 # define DYNAMIC_MSVCRT_DLL "msvcrt.dll" 4842 # endif 4843 4844 /* 4845 * Try opening the iconv.dll and return TRUE if iconv() can be used. 4846 */ 4847 int 4848 iconv_enabled(int verbose) 4849 { 4850 if (hIconvDLL != 0 && hMsvcrtDLL != 0) 4851 return TRUE; 4852 4853 // The iconv DLL file goes under different names, try them all. 4854 // Do the "2" version first, it's newer. 4855 #ifdef DYNAMIC_ICONV_DLL_ALT2 4856 if (hIconvDLL == 0) 4857 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT2); 4858 #endif 4859 #ifdef DYNAMIC_ICONV_DLL_ALT3 4860 if (hIconvDLL == 0) 4861 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT3); 4862 #endif 4863 if (hIconvDLL == 0) 4864 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL); 4865 #ifdef DYNAMIC_ICONV_DLL_ALT1 4866 if (hIconvDLL == 0) 4867 hIconvDLL = vimLoadLib(DYNAMIC_ICONV_DLL_ALT1); 4868 #endif 4869 4870 if (hIconvDLL != 0) 4871 hMsvcrtDLL = vimLoadLib(DYNAMIC_MSVCRT_DLL); 4872 if (hIconvDLL == 0 || hMsvcrtDLL == 0) 4873 { 4874 // Only give the message when 'verbose' is set, otherwise it might be 4875 // done whenever a conversion is attempted. 4876 if (verbose && p_verbose > 0) 4877 { 4878 verbose_enter(); 4879 semsg(_(e_loadlib), 4880 hIconvDLL == 0 ? DYNAMIC_ICONV_DLL : DYNAMIC_MSVCRT_DLL); 4881 verbose_leave(); 4882 } 4883 iconv_end(); 4884 return FALSE; 4885 } 4886 4887 iconv = (void *)GetProcAddress(hIconvDLL, "libiconv"); 4888 iconv_open = (void *)GetProcAddress(hIconvDLL, "libiconv_open"); 4889 iconv_close = (void *)GetProcAddress(hIconvDLL, "libiconv_close"); 4890 iconvctl = (void *)GetProcAddress(hIconvDLL, "libiconvctl"); 4891 iconv_errno = get_dll_import_func(hIconvDLL, "_errno"); 4892 if (iconv_errno == NULL) 4893 iconv_errno = (void *)GetProcAddress(hMsvcrtDLL, "_errno"); 4894 if (iconv == NULL || iconv_open == NULL || iconv_close == NULL 4895 || iconvctl == NULL || iconv_errno == NULL) 4896 { 4897 iconv_end(); 4898 if (verbose && p_verbose > 0) 4899 { 4900 verbose_enter(); 4901 semsg(_(e_loadfunc), "for libiconv"); 4902 verbose_leave(); 4903 } 4904 return FALSE; 4905 } 4906 return TRUE; 4907 } 4908 4909 void 4910 iconv_end(void) 4911 { 4912 // Don't use iconv() when inputting or outputting characters. 4913 if (input_conv.vc_type == CONV_ICONV) 4914 convert_setup(&input_conv, NULL, NULL); 4915 if (output_conv.vc_type == CONV_ICONV) 4916 convert_setup(&output_conv, NULL, NULL); 4917 4918 if (hIconvDLL != 0) 4919 FreeLibrary(hIconvDLL); 4920 if (hMsvcrtDLL != 0) 4921 FreeLibrary(hMsvcrtDLL); 4922 hIconvDLL = 0; 4923 hMsvcrtDLL = 0; 4924 } 4925 # endif // DYNAMIC_ICONV 4926 # endif // USE_ICONV 4927 4928 #if defined(FEAT_EVAL) || defined(PROTO) 4929 /* 4930 * "getimstatus()" function 4931 */ 4932 void 4933 f_getimstatus(typval_T *argvars UNUSED, typval_T *rettv) 4934 { 4935 # if defined(HAVE_INPUT_METHOD) 4936 rettv->vval.v_number = im_get_status(); 4937 # endif 4938 } 4939 #endif 4940 4941 /* 4942 * Setup "vcp" for conversion from "from" to "to". 4943 * The names must have been made canonical with enc_canonize(). 4944 * vcp->vc_type must have been initialized to CONV_NONE. 4945 * Note: cannot be used for conversion from/to ucs-2 and ucs-4 (will use utf-8 4946 * instead). 4947 * Afterwards invoke with "from" and "to" equal to NULL to cleanup. 4948 * Return FAIL when conversion is not supported, OK otherwise. 4949 */ 4950 int 4951 convert_setup(vimconv_T *vcp, char_u *from, char_u *to) 4952 { 4953 return convert_setup_ext(vcp, from, TRUE, to, TRUE); 4954 } 4955 4956 /* 4957 * As convert_setup(), but only when from_unicode_is_utf8 is TRUE will all 4958 * "from" unicode charsets be considered utf-8. Same for "to". 4959 */ 4960 int 4961 convert_setup_ext( 4962 vimconv_T *vcp, 4963 char_u *from, 4964 int from_unicode_is_utf8, 4965 char_u *to, 4966 int to_unicode_is_utf8) 4967 { 4968 int from_prop; 4969 int to_prop; 4970 int from_is_utf8; 4971 int to_is_utf8; 4972 4973 // Reset to no conversion. 4974 #ifdef USE_ICONV 4975 if (vcp->vc_type == CONV_ICONV && vcp->vc_fd != (iconv_t)-1) 4976 iconv_close(vcp->vc_fd); 4977 #endif 4978 vcp->vc_type = CONV_NONE; 4979 vcp->vc_factor = 1; 4980 vcp->vc_fail = FALSE; 4981 4982 // No conversion when one of the names is empty or they are equal. 4983 if (from == NULL || *from == NUL || to == NULL || *to == NUL 4984 || STRCMP(from, to) == 0) 4985 return OK; 4986 4987 from_prop = enc_canon_props(from); 4988 to_prop = enc_canon_props(to); 4989 if (from_unicode_is_utf8) 4990 from_is_utf8 = from_prop & ENC_UNICODE; 4991 else 4992 from_is_utf8 = from_prop == ENC_UNICODE; 4993 if (to_unicode_is_utf8) 4994 to_is_utf8 = to_prop & ENC_UNICODE; 4995 else 4996 to_is_utf8 = to_prop == ENC_UNICODE; 4997 4998 if ((from_prop & ENC_LATIN1) && to_is_utf8) 4999 { 5000 // Internal latin1 -> utf-8 conversion. 5001 vcp->vc_type = CONV_TO_UTF8; 5002 vcp->vc_factor = 2; // up to twice as long 5003 } 5004 else if ((from_prop & ENC_LATIN9) && to_is_utf8) 5005 { 5006 // Internal latin9 -> utf-8 conversion. 5007 vcp->vc_type = CONV_9_TO_UTF8; 5008 vcp->vc_factor = 3; // up to three as long (euro sign) 5009 } 5010 else if (from_is_utf8 && (to_prop & ENC_LATIN1)) 5011 { 5012 // Internal utf-8 -> latin1 conversion. 5013 vcp->vc_type = CONV_TO_LATIN1; 5014 } 5015 else if (from_is_utf8 && (to_prop & ENC_LATIN9)) 5016 { 5017 // Internal utf-8 -> latin9 conversion. 5018 vcp->vc_type = CONV_TO_LATIN9; 5019 } 5020 #ifdef MSWIN 5021 // Win32-specific codepage <-> codepage conversion without iconv. 5022 else if ((from_is_utf8 || encname2codepage(from) > 0) 5023 && (to_is_utf8 || encname2codepage(to) > 0)) 5024 { 5025 vcp->vc_type = CONV_CODEPAGE; 5026 vcp->vc_factor = 2; // up to twice as long 5027 vcp->vc_cpfrom = from_is_utf8 ? 0 : encname2codepage(from); 5028 vcp->vc_cpto = to_is_utf8 ? 0 : encname2codepage(to); 5029 } 5030 #endif 5031 #ifdef MACOS_CONVERT 5032 else if ((from_prop & ENC_MACROMAN) && (to_prop & ENC_LATIN1)) 5033 { 5034 vcp->vc_type = CONV_MAC_LATIN1; 5035 } 5036 else if ((from_prop & ENC_MACROMAN) && to_is_utf8) 5037 { 5038 vcp->vc_type = CONV_MAC_UTF8; 5039 vcp->vc_factor = 2; // up to twice as long 5040 } 5041 else if ((from_prop & ENC_LATIN1) && (to_prop & ENC_MACROMAN)) 5042 { 5043 vcp->vc_type = CONV_LATIN1_MAC; 5044 } 5045 else if (from_is_utf8 && (to_prop & ENC_MACROMAN)) 5046 { 5047 vcp->vc_type = CONV_UTF8_MAC; 5048 } 5049 #endif 5050 #ifdef USE_ICONV 5051 else 5052 { 5053 // Use iconv() for conversion. 5054 vcp->vc_fd = (iconv_t)my_iconv_open( 5055 to_is_utf8 ? (char_u *)"utf-8" : to, 5056 from_is_utf8 ? (char_u *)"utf-8" : from); 5057 if (vcp->vc_fd != (iconv_t)-1) 5058 { 5059 vcp->vc_type = CONV_ICONV; 5060 vcp->vc_factor = 4; // could be longer too... 5061 } 5062 } 5063 #endif 5064 if (vcp->vc_type == CONV_NONE) 5065 return FAIL; 5066 5067 return OK; 5068 } 5069 5070 #if defined(FEAT_GUI) || defined(AMIGA) || defined(MSWIN) \ 5071 || defined(PROTO) 5072 /* 5073 * Do conversion on typed input characters in-place. 5074 * The input and output are not NUL terminated! 5075 * Returns the length after conversion. 5076 */ 5077 int 5078 convert_input(char_u *ptr, int len, int maxlen) 5079 { 5080 return convert_input_safe(ptr, len, maxlen, NULL, NULL); 5081 } 5082 #endif 5083 5084 /* 5085 * Like convert_input(), but when there is an incomplete byte sequence at the 5086 * end return that as an allocated string in "restp" and set "*restlenp" to 5087 * the length. If "restp" is NULL it is not used. 5088 */ 5089 int 5090 convert_input_safe( 5091 char_u *ptr, 5092 int len, 5093 int maxlen, 5094 char_u **restp, 5095 int *restlenp) 5096 { 5097 char_u *d; 5098 int dlen = len; 5099 int unconvertlen = 0; 5100 5101 d = string_convert_ext(&input_conv, ptr, &dlen, 5102 restp == NULL ? NULL : &unconvertlen); 5103 if (d != NULL) 5104 { 5105 if (dlen <= maxlen) 5106 { 5107 if (unconvertlen > 0) 5108 { 5109 // Move the unconverted characters to allocated memory. 5110 *restp = alloc(unconvertlen); 5111 if (*restp != NULL) 5112 mch_memmove(*restp, ptr + len - unconvertlen, unconvertlen); 5113 *restlenp = unconvertlen; 5114 } 5115 mch_memmove(ptr, d, dlen); 5116 } 5117 else 5118 // result is too long, keep the unconverted text (the caller must 5119 // have done something wrong!) 5120 dlen = len; 5121 vim_free(d); 5122 } 5123 return dlen; 5124 } 5125 5126 /* 5127 * Convert text "ptr[*lenp]" according to "vcp". 5128 * Returns the result in allocated memory and sets "*lenp". 5129 * When "lenp" is NULL, use NUL terminated strings. 5130 * Illegal chars are often changed to "?", unless vcp->vc_fail is set. 5131 * When something goes wrong, NULL is returned and "*lenp" is unchanged. 5132 */ 5133 char_u * 5134 string_convert( 5135 vimconv_T *vcp, 5136 char_u *ptr, 5137 int *lenp) 5138 { 5139 return string_convert_ext(vcp, ptr, lenp, NULL); 5140 } 5141 5142 /* 5143 * Like string_convert(), but when "unconvlenp" is not NULL and there are is 5144 * an incomplete sequence at the end it is not converted and "*unconvlenp" is 5145 * set to the number of remaining bytes. 5146 */ 5147 char_u * 5148 string_convert_ext( 5149 vimconv_T *vcp, 5150 char_u *ptr, 5151 int *lenp, 5152 int *unconvlenp) 5153 { 5154 char_u *retval = NULL; 5155 char_u *d; 5156 int len; 5157 int i; 5158 int l; 5159 int c; 5160 5161 if (lenp == NULL) 5162 len = (int)STRLEN(ptr); 5163 else 5164 len = *lenp; 5165 if (len == 0) 5166 return vim_strsave((char_u *)""); 5167 5168 switch (vcp->vc_type) 5169 { 5170 case CONV_TO_UTF8: // latin1 to utf-8 conversion 5171 retval = alloc(len * 2 + 1); 5172 if (retval == NULL) 5173 break; 5174 d = retval; 5175 for (i = 0; i < len; ++i) 5176 { 5177 c = ptr[i]; 5178 if (c < 0x80) 5179 *d++ = c; 5180 else 5181 { 5182 *d++ = 0xc0 + ((unsigned)c >> 6); 5183 *d++ = 0x80 + (c & 0x3f); 5184 } 5185 } 5186 *d = NUL; 5187 if (lenp != NULL) 5188 *lenp = (int)(d - retval); 5189 break; 5190 5191 case CONV_9_TO_UTF8: // latin9 to utf-8 conversion 5192 retval = alloc(len * 3 + 1); 5193 if (retval == NULL) 5194 break; 5195 d = retval; 5196 for (i = 0; i < len; ++i) 5197 { 5198 c = ptr[i]; 5199 switch (c) 5200 { 5201 case 0xa4: c = 0x20ac; break; // euro 5202 case 0xa6: c = 0x0160; break; // S hat 5203 case 0xa8: c = 0x0161; break; // S -hat 5204 case 0xb4: c = 0x017d; break; // Z hat 5205 case 0xb8: c = 0x017e; break; // Z -hat 5206 case 0xbc: c = 0x0152; break; // OE 5207 case 0xbd: c = 0x0153; break; // oe 5208 case 0xbe: c = 0x0178; break; // Y 5209 } 5210 d += utf_char2bytes(c, d); 5211 } 5212 *d = NUL; 5213 if (lenp != NULL) 5214 *lenp = (int)(d - retval); 5215 break; 5216 5217 case CONV_TO_LATIN1: // utf-8 to latin1 conversion 5218 case CONV_TO_LATIN9: // utf-8 to latin9 conversion 5219 retval = alloc(len + 1); 5220 if (retval == NULL) 5221 break; 5222 d = retval; 5223 for (i = 0; i < len; ++i) 5224 { 5225 l = utf_ptr2len_len(ptr + i, len - i); 5226 if (l == 0) 5227 *d++ = NUL; 5228 else if (l == 1) 5229 { 5230 int l_w = utf8len_tab_zero[ptr[i]]; 5231 5232 if (l_w == 0) 5233 { 5234 // Illegal utf-8 byte cannot be converted 5235 vim_free(retval); 5236 return NULL; 5237 } 5238 if (unconvlenp != NULL && l_w > len - i) 5239 { 5240 // Incomplete sequence at the end. 5241 *unconvlenp = len - i; 5242 break; 5243 } 5244 *d++ = ptr[i]; 5245 } 5246 else 5247 { 5248 c = utf_ptr2char(ptr + i); 5249 if (vcp->vc_type == CONV_TO_LATIN9) 5250 switch (c) 5251 { 5252 case 0x20ac: c = 0xa4; break; // euro 5253 case 0x0160: c = 0xa6; break; // S hat 5254 case 0x0161: c = 0xa8; break; // S -hat 5255 case 0x017d: c = 0xb4; break; // Z hat 5256 case 0x017e: c = 0xb8; break; // Z -hat 5257 case 0x0152: c = 0xbc; break; // OE 5258 case 0x0153: c = 0xbd; break; // oe 5259 case 0x0178: c = 0xbe; break; // Y 5260 case 0xa4: 5261 case 0xa6: 5262 case 0xa8: 5263 case 0xb4: 5264 case 0xb8: 5265 case 0xbc: 5266 case 0xbd: 5267 case 0xbe: c = 0x100; break; // not in latin9 5268 } 5269 if (!utf_iscomposing(c)) // skip composing chars 5270 { 5271 if (c < 0x100) 5272 *d++ = c; 5273 else if (vcp->vc_fail) 5274 { 5275 vim_free(retval); 5276 return NULL; 5277 } 5278 else 5279 { 5280 *d++ = 0xbf; 5281 if (utf_char2cells(c) > 1) 5282 *d++ = '?'; 5283 } 5284 } 5285 i += l - 1; 5286 } 5287 } 5288 *d = NUL; 5289 if (lenp != NULL) 5290 *lenp = (int)(d - retval); 5291 break; 5292 5293 # ifdef MACOS_CONVERT 5294 case CONV_MAC_LATIN1: 5295 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 5296 'm', 'l', unconvlenp); 5297 break; 5298 5299 case CONV_LATIN1_MAC: 5300 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 5301 'l', 'm', unconvlenp); 5302 break; 5303 5304 case CONV_MAC_UTF8: 5305 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 5306 'm', 'u', unconvlenp); 5307 break; 5308 5309 case CONV_UTF8_MAC: 5310 retval = mac_string_convert(ptr, len, lenp, vcp->vc_fail, 5311 'u', 'm', unconvlenp); 5312 break; 5313 # endif 5314 5315 # ifdef USE_ICONV 5316 case CONV_ICONV: // conversion with output_conv.vc_fd 5317 retval = iconv_string(vcp, ptr, len, unconvlenp, lenp); 5318 break; 5319 # endif 5320 # ifdef MSWIN 5321 case CONV_CODEPAGE: // codepage -> codepage 5322 { 5323 int retlen; 5324 int tmp_len; 5325 short_u *tmp; 5326 5327 // 1. codepage/UTF-8 -> ucs-2. 5328 if (vcp->vc_cpfrom == 0) 5329 tmp_len = utf8_to_utf16(ptr, len, NULL, NULL); 5330 else 5331 { 5332 tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 5333 unconvlenp ? MB_ERR_INVALID_CHARS : 0, 5334 (char *)ptr, len, 0, 0); 5335 if (tmp_len == 0 5336 && GetLastError() == ERROR_NO_UNICODE_TRANSLATION) 5337 { 5338 if (lenp != NULL) 5339 *lenp = 0; 5340 if (unconvlenp != NULL) 5341 *unconvlenp = len; 5342 retval = alloc(1); 5343 if (retval) 5344 retval[0] = NUL; 5345 return retval; 5346 } 5347 } 5348 tmp = ALLOC_MULT(short_u, tmp_len); 5349 if (tmp == NULL) 5350 break; 5351 if (vcp->vc_cpfrom == 0) 5352 utf8_to_utf16(ptr, len, tmp, unconvlenp); 5353 else 5354 MultiByteToWideChar(vcp->vc_cpfrom, 0, 5355 (char *)ptr, len, tmp, tmp_len); 5356 5357 // 2. ucs-2 -> codepage/UTF-8. 5358 if (vcp->vc_cpto == 0) 5359 retlen = utf16_to_utf8(tmp, tmp_len, NULL); 5360 else 5361 retlen = WideCharToMultiByte(vcp->vc_cpto, 0, 5362 tmp, tmp_len, 0, 0, 0, 0); 5363 retval = alloc(retlen + 1); 5364 if (retval != NULL) 5365 { 5366 if (vcp->vc_cpto == 0) 5367 utf16_to_utf8(tmp, tmp_len, retval); 5368 else 5369 WideCharToMultiByte(vcp->vc_cpto, 0, 5370 tmp, tmp_len, 5371 (char *)retval, retlen, 0, 0); 5372 retval[retlen] = NUL; 5373 if (lenp != NULL) 5374 *lenp = retlen; 5375 } 5376 vim_free(tmp); 5377 break; 5378 } 5379 # endif 5380 } 5381 5382 return retval; 5383 } 5384 5385 #if defined(FEAT_EVAL) || defined(PROTO) 5386 5387 /* 5388 * Table set by setcellwidths(). 5389 */ 5390 typedef struct 5391 { 5392 long first; 5393 long last; 5394 char width; 5395 } cw_interval_T; 5396 5397 static cw_interval_T *cw_table = NULL; 5398 static size_t cw_table_size = 0; 5399 5400 /* 5401 * Return 1 or 2 when "c" is in the cellwidth table. 5402 * Return 0 if not. 5403 */ 5404 static int 5405 cw_value(int c) 5406 { 5407 int mid, bot, top; 5408 5409 if (cw_table == NULL) 5410 return 0; 5411 5412 // first quick check for Latin1 etc. characters 5413 if (c < cw_table[0].first) 5414 return 0; 5415 5416 // binary search in table 5417 bot = 0; 5418 top = (int)cw_table_size - 1; 5419 while (top >= bot) 5420 { 5421 mid = (bot + top) / 2; 5422 if (cw_table[mid].last < c) 5423 bot = mid + 1; 5424 else if (cw_table[mid].first > c) 5425 top = mid - 1; 5426 else 5427 return cw_table[mid].width; 5428 } 5429 return 0; 5430 } 5431 5432 static int 5433 tv_nr_compare(const void *a1, const void *a2) 5434 { 5435 listitem_T *li1 = *(listitem_T **)a1; 5436 listitem_T *li2 = *(listitem_T **)a2; 5437 5438 return li1->li_tv.vval.v_number - li2->li_tv.vval.v_number; 5439 } 5440 5441 void 5442 f_setcellwidths(typval_T *argvars, typval_T *rettv UNUSED) 5443 { 5444 list_T *l; 5445 listitem_T *li; 5446 int item; 5447 int i; 5448 listitem_T **ptrs; 5449 cw_interval_T *table; 5450 5451 if (argvars[0].v_type != VAR_LIST || argvars[0].vval.v_list == NULL) 5452 { 5453 emsg(_(e_listreq)); 5454 return; 5455 } 5456 l = argvars[0].vval.v_list; 5457 if (l->lv_len == 0) 5458 { 5459 // Clearing the table. 5460 vim_free(cw_table); 5461 cw_table = NULL; 5462 cw_table_size = 0; 5463 return; 5464 } 5465 5466 ptrs = ALLOC_MULT(listitem_T *, l->lv_len); 5467 if (ptrs == NULL) 5468 return; 5469 5470 // Check that all entries are a list with three numbers, the range is 5471 // valid and the cell width is valid. 5472 item = 0; 5473 for (li = l->lv_first; li != NULL; li = li->li_next) 5474 { 5475 listitem_T *lili; 5476 varnumber_T n1; 5477 5478 if (li->li_tv.v_type != VAR_LIST || li->li_tv.vval.v_list == NULL) 5479 { 5480 semsg(_(e_list_item_nr_is_not_list), item); 5481 vim_free(ptrs); 5482 return; 5483 } 5484 5485 lili = li->li_tv.vval.v_list->lv_first; 5486 ptrs[item] = lili; 5487 for (i = 0; lili != NULL; lili = lili->li_next, ++i) 5488 { 5489 if (lili->li_tv.v_type != VAR_NUMBER) 5490 break; 5491 if (i == 0) 5492 { 5493 n1 = lili->li_tv.vval.v_number; 5494 if (n1 < 0x100) 5495 { 5496 emsg(_(e_only_values_of_0x100_and_higher_supported)); 5497 vim_free(ptrs); 5498 return; 5499 } 5500 } 5501 else if (i == 1 && lili->li_tv.vval.v_number < n1) 5502 { 5503 semsg(_(e_list_item_nr_range_invalid), item); 5504 vim_free(ptrs); 5505 return; 5506 } 5507 else if (i == 2 && (lili->li_tv.vval.v_number < 1 5508 || lili->li_tv.vval.v_number > 2)) 5509 { 5510 semsg(_(e_list_item_nr_cell_width_invalid), item); 5511 vim_free(ptrs); 5512 return; 5513 } 5514 } 5515 if (i != 3) 5516 { 5517 semsg(_(e_list_item_nr_does_not_contain_3_numbers), item); 5518 vim_free(ptrs); 5519 return; 5520 } 5521 ++item; 5522 } 5523 5524 // Sort the list on the first number. 5525 qsort((void *)ptrs, (size_t)l->lv_len, sizeof(listitem_T *), tv_nr_compare); 5526 5527 table = ALLOC_MULT(cw_interval_T, l->lv_len); 5528 if (table == NULL) 5529 { 5530 vim_free(ptrs); 5531 return; 5532 } 5533 5534 // Store the items in the new table. 5535 item = 0; 5536 for (item = 0; item < l->lv_len; ++item) 5537 { 5538 listitem_T *lili = ptrs[item]; 5539 varnumber_T n1; 5540 5541 n1 = lili->li_tv.vval.v_number; 5542 if (item > 0 && n1 <= table[item - 1].last) 5543 { 5544 semsg(_(e_overlapping_ranges_for_nr), (long)n1); 5545 vim_free(ptrs); 5546 vim_free(table); 5547 return; 5548 } 5549 table[item].first = n1; 5550 lili = lili->li_next; 5551 table[item].last = lili->li_tv.vval.v_number; 5552 lili = lili->li_next; 5553 table[item].width = lili->li_tv.vval.v_number; 5554 } 5555 5556 vim_free(ptrs); 5557 vim_free(cw_table); 5558 cw_table = table; 5559 cw_table_size = l->lv_len; 5560 } 5561 5562 void 5563 f_charclass(typval_T *argvars, typval_T *rettv UNUSED) 5564 { 5565 if (check_for_string_arg(argvars, 0) == FAIL) 5566 return; 5567 rettv->vval.v_number = mb_get_class(argvars[0].vval.v_string); 5568 } 5569 #endif 5570