| 1 | /* |
| 2 | ** $Id: llex.c,v 2.63 2013/03/16 21:10:18 roberto Exp $ |
| 3 | ** Lexical Analyzer |
| 4 | ** See Copyright Notice in lua.h |
| 5 | */ |
| 6 | |
| 7 | |
| 8 | #include <locale.h> |
| 9 | #include <string.h> |
| 10 | |
| 11 | #define llex_c |
| 12 | #define LUA_CORE |
| 13 | |
| 14 | #include "lua.h" |
| 15 | |
| 16 | #include "lctype.h" |
| 17 | #include "ldo.h" |
| 18 | #include "llex.h" |
| 19 | #include "lobject.h" |
| 20 | #include "lparser.h" |
| 21 | #include "lstate.h" |
| 22 | #include "lstring.h" |
| 23 | #include "ltable.h" |
| 24 | #include "lzio.h" |
| 25 | |
| 26 | |
| 27 | |
| 28 | #define next(ls) (ls->current = zgetc(ls->z)) |
| 29 | |
| 30 | |
| 31 | |
| 32 | #define currIsNewline(ls) (ls->current == '\n' || ls->current == '\r') |
| 33 | |
| 34 | |
| 35 | /* ORDER RESERVED */ |
| 36 | static const char *const luaX_tokens [] = { |
| 37 | "and", "break", "do", "else", "elseif", |
| 38 | "end", "false", "for", "function", "goto", "if", |
| 39 | "in", "local", "nil", "not", "or", "repeat", |
| 40 | "return", "then", "true", "until", "while", |
| 41 | "..", "...", "==", ">=", "<=", "~=", "::", "<eof>", |
| 42 | "<number>", "<name>", "<string>" |
| 43 | }; |
| 44 | |
| 45 | |
| 46 | #define save_and_next(ls) (save(ls, ls->current), next(ls)) |
| 47 | |
| 48 | |
| 49 | static l_noret lexerror (LexState *ls, const char *msg, int token); |
| 50 | |
| 51 | |
| 52 | static void save (LexState *ls, int c) { |
| 53 | Mbuffer *b = ls->buff; |
| 54 | if (luaZ_bufflen(b) + 1 > luaZ_sizebuffer(b)) { |
| 55 | size_t newsize; |
| 56 | if (luaZ_sizebuffer(b) >= MAX_SIZET/2) |
| 57 | lexerror(ls, "lexical element too long", 0); |
| 58 | newsize = luaZ_sizebuffer(b) * 2; |
| 59 | luaZ_resizebuffer(ls->L, b, newsize); |
| 60 | } |
| 61 | b->buffer[luaZ_bufflen(b)++] = cast(char, c); |
| 62 | } |
| 63 | |
| 64 | |
| 65 | void luaX_init (lua_State *L) { |
| 66 | int i; |
| 67 | for (i=0; i<NUM_RESERVED; i++) { |
| 68 | TString *ts = luaS_new(L, luaX_tokens[i]); |
| 69 | luaS_fix(ts); /* reserved words are never collected */ |
| 70 | ts->tsv.extra = cast_byte(i+1); /* reserved word */ |
| 71 | } |
| 72 | } |
| 73 | |
| 74 | |
| 75 | const char *luaX_token2str (LexState *ls, int token) { |
| 76 | if (token < FIRST_RESERVED) { /* single-byte symbols? */ |
| 77 | lua_assert(token == cast(unsigned char, token)); |
| 78 | return (lisprint(token)) ? luaO_pushfstring(ls->L, LUA_QL("%c"), token) : |
| 79 | luaO_pushfstring(ls->L, "char(%d)", token); |
| 80 | } |
| 81 | else { |
| 82 | const char *s = luaX_tokens[token - FIRST_RESERVED]; |
| 83 | if (token < TK_EOS) /* fixed format (symbols and reserved words)? */ |
| 84 | return luaO_pushfstring(ls->L, LUA_QS, s); |
| 85 | else /* names, strings, and numerals */ |
| 86 | return s; |
| 87 | } |
| 88 | } |
| 89 | |
| 90 | |
| 91 | static const char *txtToken (LexState *ls, int token) { |
| 92 | switch (token) { |
| 93 | case TK_NAME: |
| 94 | case TK_STRING: |
| 95 | case TK_NUMBER: |
| 96 | save(ls, '\0'); |
| 97 | return luaO_pushfstring(ls->L, LUA_QS, luaZ_buffer(ls->buff)); |
| 98 | default: |
| 99 | return luaX_token2str(ls, token); |
| 100 | } |
| 101 | } |
| 102 | |
| 103 | |
| 104 | static l_noret lexerror (LexState *ls, const char *msg, int token) { |
| 105 | char buff[LUA_IDSIZE]; |
| 106 | luaO_chunkid(buff, getstr(ls->source), LUA_IDSIZE); |
| 107 | msg = luaO_pushfstring(ls->L, "%s:%d: %s", buff, ls->linenumber, msg); |
| 108 | if (token) |
| 109 | luaO_pushfstring(ls->L, "%s near %s", msg, txtToken(ls, token)); |
| 110 | luaD_throw(ls->L, LUA_ERRSYNTAX); |
| 111 | } |
| 112 | |
| 113 | |
| 114 | l_noret luaX_syntaxerror (LexState *ls, const char *msg) { |
| 115 | lexerror(ls, msg, ls->t.token); |
| 116 | } |
| 117 | |
| 118 | |
| 119 | /* |
| 120 | ** creates a new string and anchors it in function's table so that |
| 121 | ** it will not be collected until the end of the function's compilation |
| 122 | ** (by that time it should be anchored in function's prototype) |
| 123 | */ |
| 124 | TString *luaX_newstring (LexState *ls, const char *str, size_t l) { |
| 125 | lua_State *L = ls->L; |
| 126 | TValue *o; /* entry for `str' */ |
| 127 | TString *ts = luaS_newlstr(L, str, l); /* create new string */ |
| 128 | setsvalue2s(L, L->top++, ts); /* temporarily anchor it in stack */ |
| 129 | o = luaH_set(L, ls->fs->h, L->top - 1); |
| 130 | if (ttisnil(o)) { /* not in use yet? (see 'addK') */ |
| 131 | /* boolean value does not need GC barrier; |
| 132 | table has no metatable, so it does not need to invalidate cache */ |
| 133 | setbvalue(o, 1); /* t[string] = true */ |
| 134 | luaC_checkGC(L); |
| 135 | } |
| 136 | L->top--; /* remove string from stack */ |
| 137 | return ts; |
| 138 | } |
| 139 | |
| 140 | |
| 141 | /* |
| 142 | ** increment line number and skips newline sequence (any of |
| 143 | ** \n, \r, \n\r, or \r\n) |
| 144 | */ |
| 145 | static void inclinenumber (LexState *ls) { |
| 146 | int old = ls->current; |
| 147 | lua_assert(currIsNewline(ls)); |
| 148 | next(ls); /* skip `\n' or `\r' */ |
| 149 | if (currIsNewline(ls) && ls->current != old) |
| 150 | next(ls); /* skip `\n\r' or `\r\n' */ |
| 151 | if (++ls->linenumber >= MAX_INT) |
| 152 | luaX_syntaxerror(ls, "chunk has too many lines"); |
| 153 | } |
| 154 | |
| 155 | |
| 156 | void luaX_setinput (lua_State *L, LexState *ls, ZIO *z, TString *source, |
| 157 | int firstchar) { |
| 158 | ls->decpoint = '.'; |
| 159 | ls->L = L; |
| 160 | ls->current = firstchar; |
| 161 | ls->lookahead.token = TK_EOS; /* no look-ahead token */ |
| 162 | ls->z = z; |
| 163 | ls->fs = NULL; |
| 164 | ls->linenumber = 1; |
| 165 | ls->lastline = 1; |
| 166 | ls->source = source; |
| 167 | ls->envn = luaS_new(L, LUA_ENV); /* create env name */ |
| 168 | luaS_fix(ls->envn); /* never collect this name */ |
| 169 | luaZ_resizebuffer(ls->L, ls->buff, LUA_MINBUFFER); /* initialize buffer */ |
| 170 | } |
| 171 | |
| 172 | |
| 173 | |
| 174 | /* |
| 175 | ** ======================================================= |
| 176 | ** LEXICAL ANALYZER |
| 177 | ** ======================================================= |
| 178 | */ |
| 179 | |
| 180 | |
| 181 | |
| 182 | static int check_next (LexState *ls, const char *set) { |
| 183 | if (ls->current == '\0' || !strchr(set, ls->current)) |
| 184 | return 0; |
| 185 | save_and_next(ls); |
| 186 | return 1; |
| 187 | } |
| 188 | |
| 189 | |
| 190 | /* |
| 191 | ** change all characters 'from' in buffer to 'to' |
| 192 | */ |
| 193 | static void buffreplace (LexState *ls, char from, char to) { |
| 194 | size_t n = luaZ_bufflen(ls->buff); |
| 195 | char *p = luaZ_buffer(ls->buff); |
| 196 | while (n--) |
| 197 | if (p[n] == from) p[n] = to; |
| 198 | } |
| 199 | |
| 200 | |
| 201 | #if !defined(getlocaledecpoint) |
| 202 | #define getlocaledecpoint() (localeconv()->decimal_point[0]) |
| 203 | #endif |
| 204 | |
| 205 | |
| 206 | #define buff2d(b,e) luaO_str2d(luaZ_buffer(b), luaZ_bufflen(b) - 1, e) |
| 207 | |
| 208 | /* |
| 209 | ** in case of format error, try to change decimal point separator to |
| 210 | ** the one defined in the current locale and check again |
| 211 | */ |
| 212 | static void trydecpoint (LexState *ls, SemInfo *seminfo) { |
| 213 | char old = ls->decpoint; |
| 214 | ls->decpoint = getlocaledecpoint(); |
| 215 | buffreplace(ls, old, ls->decpoint); /* try new decimal separator */ |
| 216 | if (!buff2d(ls->buff, &seminfo->r)) { |
| 217 | /* format error with correct decimal point: no more options */ |
| 218 | buffreplace(ls, ls->decpoint, '.'); /* undo change (for error message) */ |
| 219 | lexerror(ls, "malformed number", TK_NUMBER); |
| 220 | } |
| 221 | } |
| 222 | |
| 223 | |
| 224 | /* LUA_NUMBER */ |
| 225 | /* |
| 226 | ** this function is quite liberal in what it accepts, as 'luaO_str2d' |
| 227 | ** will reject ill-formed numerals. |
| 228 | */ |
| 229 | static void read_numeral (LexState *ls, SemInfo *seminfo) { |
| 230 | const char *expo = "Ee"; |
| 231 | int first = ls->current; |
| 232 | lua_assert(lisdigit(ls->current)); |
| 233 | save_and_next(ls); |
| 234 | if (first == '0' && check_next(ls, "Xx")) /* hexadecimal? */ |
| 235 | expo = "Pp"; |
| 236 | for (;;) { |
| 237 | if (check_next(ls, expo)) /* exponent part? */ |
| 238 | check_next(ls, "+-"); /* optional exponent sign */ |
| 239 | if (lisxdigit(ls->current) || ls->current == '.') |
| 240 | save_and_next(ls); |
| 241 | else break; |
| 242 | } |
| 243 | save(ls, '\0'); |
| 244 | buffreplace(ls, '.', ls->decpoint); /* follow locale for decimal point */ |
| 245 | if (!buff2d(ls->buff, &seminfo->r)) /* format error? */ |
| 246 | trydecpoint(ls, seminfo); /* try to update decimal point separator */ |
| 247 | } |
| 248 | |
| 249 | |
| 250 | /* |
| 251 | ** skip a sequence '[=*[' or ']=*]' and return its number of '='s or |
| 252 | ** -1 if sequence is malformed |
| 253 | */ |
| 254 | static int skip_sep (LexState *ls) { |
| 255 | int count = 0; |
| 256 | int s = ls->current; |
| 257 | lua_assert(s == '[' || s == ']'); |
| 258 | save_and_next(ls); |
| 259 | while (ls->current == '=') { |
| 260 | save_and_next(ls); |
| 261 | count++; |
| 262 | } |
| 263 | return (ls->current == s) ? count : (-count) - 1; |
| 264 | } |
| 265 | |
| 266 | |
| 267 | static void read_long_string (LexState *ls, SemInfo *seminfo, int sep) { |
| 268 | save_and_next(ls); /* skip 2nd `[' */ |
| 269 | if (currIsNewline(ls)) /* string starts with a newline? */ |
| 270 | inclinenumber(ls); /* skip it */ |
| 271 | for (;;) { |
| 272 | switch (ls->current) { |
| 273 | case EOZ: |
| 274 | lexerror(ls, (seminfo) ? "unfinished long string" : |
| 275 | "unfinished long comment", TK_EOS); |
| 276 | break; /* to avoid warnings */ |
| 277 | case ']': { |
| 278 | if (skip_sep(ls) == sep) { |
| 279 | save_and_next(ls); /* skip 2nd `]' */ |
| 280 | goto endloop; |
| 281 | } |
| 282 | break; |
| 283 | } |
| 284 | case '\n': case '\r': { |
| 285 | save(ls, '\n'); |
| 286 | inclinenumber(ls); |
| 287 | if (!seminfo) luaZ_resetbuffer(ls->buff); /* avoid wasting space */ |
| 288 | break; |
| 289 | } |
| 290 | default: { |
| 291 | if (seminfo) save_and_next(ls); |
| 292 | else next(ls); |
| 293 | } |
| 294 | } |
| 295 | } endloop: |
| 296 | if (seminfo) |
| 297 | seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + (2 + sep), |
| 298 | luaZ_bufflen(ls->buff) - 2*(2 + sep)); |
| 299 | } |
| 300 | |
| 301 | |
| 302 | static void escerror (LexState *ls, int *c, int n, const char *msg) { |
| 303 | int i; |
| 304 | luaZ_resetbuffer(ls->buff); /* prepare error message */ |
| 305 | save(ls, '\\'); |
| 306 | for (i = 0; i < n && c[i] != EOZ; i++) |
| 307 | save(ls, c[i]); |
| 308 | lexerror(ls, msg, TK_STRING); |
| 309 | } |
| 310 | |
| 311 | |
| 312 | static int readhexaesc (LexState *ls) { |
| 313 | int c[3], i; /* keep input for error message */ |
| 314 | int r = 0; /* result accumulator */ |
| 315 | c[0] = 'x'; /* for error message */ |
| 316 | for (i = 1; i < 3; i++) { /* read two hexadecimal digits */ |
| 317 | c[i] = next(ls); |
| 318 | if (!lisxdigit(c[i])) |
| 319 | escerror(ls, c, i + 1, "hexadecimal digit expected"); |
| 320 | r = (r << 4) + luaO_hexavalue(c[i]); |
| 321 | } |
| 322 | return r; |
| 323 | } |
| 324 | |
| 325 | |
| 326 | static int readdecesc (LexState *ls) { |
| 327 | int c[3], i; |
| 328 | int r = 0; /* result accumulator */ |
| 329 | for (i = 0; i < 3 && lisdigit(ls->current); i++) { /* read up to 3 digits */ |
| 330 | c[i] = ls->current; |
| 331 | r = 10*r + c[i] - '0'; |
| 332 | next(ls); |
| 333 | } |
| 334 | if (r > UCHAR_MAX) |
| 335 | escerror(ls, c, i, "decimal escape too large"); |
| 336 | return r; |
| 337 | } |
| 338 | |
| 339 | |
| 340 | static void read_string (LexState *ls, int del, SemInfo *seminfo) { |
| 341 | save_and_next(ls); /* keep delimiter (for error messages) */ |
| 342 | while (ls->current != del) { |
| 343 | switch (ls->current) { |
| 344 | case EOZ: |
| 345 | lexerror(ls, "unfinished string", TK_EOS); |
| 346 | break; /* to avoid warnings */ |
| 347 | case '\n': |
| 348 | case '\r': |
| 349 | lexerror(ls, "unfinished string", TK_STRING); |
| 350 | break; /* to avoid warnings */ |
| 351 | case '\\': { /* escape sequences */ |
| 352 | int c; /* final character to be saved */ |
| 353 | next(ls); /* do not save the `\' */ |
| 354 | switch (ls->current) { |
| 355 | case 'a': c = '\a'; goto read_save; |
| 356 | case 'b': c = '\b'; goto read_save; |
| 357 | case 'f': c = '\f'; goto read_save; |
| 358 | case 'n': c = '\n'; goto read_save; |
| 359 | case 'r': c = '\r'; goto read_save; |
| 360 | case 't': c = '\t'; goto read_save; |
| 361 | case 'v': c = '\v'; goto read_save; |
| 362 | case 'x': c = readhexaesc(ls); goto read_save; |
| 363 | case '\n': case '\r': |
| 364 | inclinenumber(ls); c = '\n'; goto only_save; |
| 365 | case '\\': case '\"': case '\'': |
| 366 | c = ls->current; goto read_save; |
| 367 | case EOZ: goto no_save; /* will raise an error next loop */ |
| 368 | case 'z': { /* zap following span of spaces */ |
| 369 | next(ls); /* skip the 'z' */ |
| 370 | while (lisspace(ls->current)) { |
| 371 | if (currIsNewline(ls)) inclinenumber(ls); |
| 372 | else next(ls); |
| 373 | } |
| 374 | goto no_save; |
| 375 | } |
| 376 | default: { |
| 377 | if (!lisdigit(ls->current)) |
| 378 | escerror(ls, &ls->current, 1, "invalid escape sequence"); |
| 379 | /* digital escape \ddd */ |
| 380 | c = readdecesc(ls); |
| 381 | goto only_save; |
| 382 | } |
| 383 | } |
| 384 | read_save: next(ls); /* read next character */ |
| 385 | only_save: save(ls, c); /* save 'c' */ |
| 386 | no_save: break; |
| 387 | } |
| 388 | default: |
| 389 | save_and_next(ls); |
| 390 | } |
| 391 | } |
| 392 | save_and_next(ls); /* skip delimiter */ |
| 393 | seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + 1, |
| 394 | luaZ_bufflen(ls->buff) - 2); |
| 395 | } |
| 396 | |
| 397 | |
| 398 | static int llex (LexState *ls, SemInfo *seminfo) { |
| 399 | luaZ_resetbuffer(ls->buff); |
| 400 | for (;;) { |
| 401 | switch (ls->current) { |
| 402 | case '\n': case '\r': { /* line breaks */ |
| 403 | inclinenumber(ls); |
| 404 | break; |
| 405 | } |
| 406 | case ' ': case '\f': case '\t': case '\v': { /* spaces */ |
| 407 | next(ls); |
| 408 | break; |
| 409 | } |
| 410 | case '-': { /* '-' or '--' (comment) */ |
| 411 | next(ls); |
| 412 | if (ls->current != '-') return '-'; |
| 413 | /* else is a comment */ |
| 414 | next(ls); |
| 415 | if (ls->current == '[') { /* long comment? */ |
| 416 | int sep = skip_sep(ls); |
| 417 | luaZ_resetbuffer(ls->buff); /* `skip_sep' may dirty the buffer */ |
| 418 | if (sep >= 0) { |
| 419 | read_long_string(ls, NULL, sep); /* skip long comment */ |
| 420 | luaZ_resetbuffer(ls->buff); /* previous call may dirty the buff. */ |
| 421 | break; |
| 422 | } |
| 423 | } |
| 424 | /* else short comment */ |
| 425 | while (!currIsNewline(ls) && ls->current != EOZ) |
| 426 | next(ls); /* skip until end of line (or end of file) */ |
| 427 | break; |
| 428 | } |
| 429 | case '[': { /* long string or simply '[' */ |
| 430 | int sep = skip_sep(ls); |
| 431 | if (sep >= 0) { |
| 432 | read_long_string(ls, seminfo, sep); |
| 433 | return TK_STRING; |
| 434 | } |
| 435 | else if (sep == -1) return '['; |
| 436 | else lexerror(ls, "invalid long string delimiter", TK_STRING); |
| 437 | } |
| 438 | case '=': { |
| 439 | next(ls); |
| 440 | if (ls->current != '=') return '='; |
| 441 | else { next(ls); return TK_EQ; } |
| 442 | } |
| 443 | case '<': { |
| 444 | next(ls); |
| 445 | if (ls->current != '=') return '<'; |
| 446 | else { next(ls); return TK_LE; } |
| 447 | } |
| 448 | case '>': { |
| 449 | next(ls); |
| 450 | if (ls->current != '=') return '>'; |
| 451 | else { next(ls); return TK_GE; } |
| 452 | } |
| 453 | case '~': { |
| 454 | next(ls); |
| 455 | if (ls->current != '=') return '~'; |
| 456 | else { next(ls); return TK_NE; } |
| 457 | } |
| 458 | case ':': { |
| 459 | next(ls); |
| 460 | if (ls->current != ':') return ':'; |
| 461 | else { next(ls); return TK_DBCOLON; } |
| 462 | } |
| 463 | case '"': case '\'': { /* short literal strings */ |
| 464 | read_string(ls, ls->current, seminfo); |
| 465 | return TK_STRING; |
| 466 | } |
| 467 | case '.': { /* '.', '..', '...', or number */ |
| 468 | save_and_next(ls); |
| 469 | if (check_next(ls, ".")) { |
| 470 | if (check_next(ls, ".")) |
| 471 | return TK_DOTS; /* '...' */ |
| 472 | else return TK_CONCAT; /* '..' */ |
| 473 | } |
| 474 | else if (!lisdigit(ls->current)) return '.'; |
| 475 | /* else go through */ |
| 476 | } |
| 477 | case '0': case '1': case '2': case '3': case '4': |
| 478 | case '5': case '6': case '7': case '8': case '9': { |
| 479 | read_numeral(ls, seminfo); |
| 480 | return TK_NUMBER; |
| 481 | } |
| 482 | case EOZ: { |
| 483 | return TK_EOS; |
| 484 | } |
| 485 | default: { |
| 486 | if (lislalpha(ls->current)) { /* identifier or reserved word? */ |
| 487 | TString *ts; |
| 488 | do { |
| 489 | save_and_next(ls); |
| 490 | } while (lislalnum(ls->current)); |
| 491 | ts = luaX_newstring(ls, luaZ_buffer(ls->buff), |
| 492 | luaZ_bufflen(ls->buff)); |
| 493 | seminfo->ts = ts; |
| 494 | if (isreserved(ts)) /* reserved word? */ |
| 495 | return ts->tsv.extra - 1 + FIRST_RESERVED; |
| 496 | else { |
| 497 | return TK_NAME; |
| 498 | } |
| 499 | } |
| 500 | else { /* single-char tokens (+ - / ...) */ |
| 501 | int c = ls->current; |
| 502 | next(ls); |
| 503 | return c; |
| 504 | } |
| 505 | } |
| 506 | } |
| 507 | } |
| 508 | } |
| 509 | |
| 510 | |
| 511 | void luaX_next (LexState *ls) { |
| 512 | ls->lastline = ls->linenumber; |
| 513 | if (ls->lookahead.token != TK_EOS) { /* is there a look-ahead token? */ |
| 514 | ls->t = ls->lookahead; /* use this one */ |
| 515 | ls->lookahead.token = TK_EOS; /* and discharge it */ |
| 516 | } |
| 517 | else |
| 518 | ls->t.token = llex(ls, &ls->t.seminfo); /* read next token */ |
| 519 | } |
| 520 | |
| 521 | |
| 522 | int luaX_lookahead (LexState *ls) { |
| 523 | lua_assert(ls->lookahead.token == TK_EOS); |
| 524 | ls->lookahead.token = llex(ls, &ls->lookahead.seminfo); |
| 525 | return ls->lookahead.token; |
| 526 | } |
| 527 | |