module/lua/llex.c

   1 /*
   2 ** $Id: llex.c,v 2.63.1.3 2015/02/09 17:56:34 roberto Exp $
   3 ** Lexical Analyzer
   4 ** See Copyright Notice in lua.h
   5 */
   6
   7 #define llex_c
   8 #define LUA_CORE
   9
  10 #include <sys/lua/lua.h>
  11
  12 #include "lctype.h"
  13 #include "ldo.h"
  14 #include "llex.h"
  15 #include "lobject.h"
  16 #include "lparser.h"
  17 #include "lstate.h"
  18 #include "lstring.h"
  19 #include "ltable.h"
  20 #include "lzio.h"
  21
  22
  23
  24 #define next(ls) (ls->current = zgetc(ls->z))
  25
  26
  27
  28 #define currIsNewline(ls)       (ls->current == '\n' || ls->current == '\r')
  29
  30
  31 /* ORDER RESERVED */
  32 static const char *const luaX_tokens [] = {
  33     "and", "break", "do", "else", "elseif",
  34     "end", "false", "for", "function", "goto", "if",
  35     "in", "local", "nil", "not", "or", "repeat",
  36     "return", "then", "true", "until", "while",
  37     "..", "...", "==", ">=", "<=", "~=", "::", "<eof>",
  38     "<number>", "<name>", "<string>"
  39 };
  40
  41
  42 #define save_and_next(ls) (save(ls, ls->current), next(ls))
  43
  44
  45 static l_noret lexerror (LexState *ls, const char *msg, int token);
  46
  47
  48 static void save (LexState *ls, int c) {
  49   Mbuffer *b = ls->buff;
  50   if (luaZ_bufflen(b) + 1 > luaZ_sizebuffer(b)) {
  51     size_t newsize;
  52     if (luaZ_sizebuffer(b) >= MAX_SIZET/2)
  53       lexerror(ls, "lexical element too long", 0);
  54     newsize = luaZ_sizebuffer(b) * 2;
  55     luaZ_resizebuffer(ls->L, b, newsize);
  56   }
  57   b->buffer[luaZ_bufflen(b)++] = cast(char, c);
  58 }
  59
  60
  61 void luaX_init (lua_State *L) {
  62   int i;
  63   for (i=0; i<NUM_RESERVED; i++) {
  64     TString *ts = luaS_new(L, luaX_tokens[i]);
  65     luaS_fix(ts);  /* reserved words are never collected */
  66     ts->tsv.extra = cast_byte(i+1);  /* reserved word */
  67   }
  68 }
  69
  70
  71 const char *luaX_token2str (LexState *ls, int token) {
  72   if (token < FIRST_RESERVED) {  /* single-byte symbols? */
  73     lua_assert(token == cast(unsigned char, token));
  74     return (lisprint(token)) ? luaO_pushfstring(ls->L, LUA_QL("%c"), token) :
  75                               luaO_pushfstring(ls->L, "char(%d)", token);
  76   }
  77   else {
  78     const char *s = luaX_tokens[token - FIRST_RESERVED];
  79     if (token < TK_EOS)  /* fixed format (symbols and reserved words)? */
  80       return luaO_pushfstring(ls->L, LUA_QS, s);
  81     else  /* names, strings, and numerals */
  82       return s;
  83   }
  84 }
  85
  86
  87 static const char *txtToken (LexState *ls, int token) {
  88   switch (token) {
  89     case TK_NAME:
  90     case TK_STRING:
  91     case TK_NUMBER:
  92       save(ls, '\0');
  93       return luaO_pushfstring(ls->L, LUA_QS, luaZ_buffer(ls->buff));
  94     default:
  95       return luaX_token2str(ls, token);
  96   }
  97 }
  98
  99
 100 static l_noret lexerror (LexState *ls, const char *msg, int token) {
 101   char buff[LUA_IDSIZE];
 102   luaO_chunkid(buff, getstr(ls->source), LUA_IDSIZE);
 103   msg = luaO_pushfstring(ls->L, "%s:%d: %s", buff, ls->linenumber, msg);
 104   if (token)
 105     luaO_pushfstring(ls->L, "%s near %s", msg, txtToken(ls, token));
 106   luaD_throw(ls->L, LUA_ERRSYNTAX);
 107 }
 108
 109
 110 l_noret luaX_syntaxerror (LexState *ls, const char *msg) {
 111   lexerror(ls, msg, ls->t.token);
 112 }
 113
 114
 115 /*
 116 ** creates a new string and anchors it in function's table so that
 117 ** it will not be collected until the end of the function's compilation
 118 ** (by that time it should be anchored in function's prototype)
 119 */
 120 TString *luaX_newstring (LexState *ls, const char *str, size_t l) {
 121   lua_State *L = ls->L;
 122   TValue *o;  /* entry for `str' */
 123   TString *ts = luaS_newlstr(L, str, l);  /* create new string */
 124   setsvalue2s(L, L->top++, ts);  /* temporarily anchor it in stack */
 125   o = luaH_set(L, ls->fs->h, L->top - 1);
 126   if (ttisnil(o)) {  /* not in use yet? (see 'addK') */
 127     /* boolean value does not need GC barrier;
 128        table has no metatable, so it does not need to invalidate cache */
 129     setbvalue(o, 1);  /* t[string] = true */
 130     luaC_checkGC(L);
 131   }
 132   else {  /* string already present */
 133     ts = rawtsvalue(keyfromval(o));  /* re-use value previously stored */
 134   }
 135   L->top--;  /* remove string from stack */
 136   return ts;
 137 }
 138
 139
 140 /*
 141 ** increment line number and skips newline sequence (any of
 142 ** \n, \r, \n\r, or \r\n)
 143 */
 144 static void inclinenumber (LexState *ls) {
 145   int old = ls->current;
 146   lua_assert(currIsNewline(ls));
 147   next(ls);  /* skip `\n' or `\r' */
 148   if (currIsNewline(ls) && ls->current != old)
 149     next(ls);  /* skip `\n\r' or `\r\n' */
 150   if (++ls->linenumber >= MAX_INT)
 151     lexerror(ls, "chunk has too many lines", 0);
 152 }
 153
 154
 155 void luaX_setinput (lua_State *L, LexState *ls, ZIO *z, TString *source,
 156                     int firstchar) {
 157   ls->decpoint = '.';
 158   ls->L = L;
 159   ls->current = firstchar;
 160   ls->lookahead.token = TK_EOS;  /* no look-ahead token */
 161   ls->z = z;
 162   ls->fs = NULL;
 163   ls->linenumber = 1;
 164   ls->lastline = 1;
 165   ls->source = source;
 166   ls->envn = luaS_new(L, LUA_ENV);  /* create env name */
 167   luaS_fix(ls->envn);  /* never collect this name */
 168   luaZ_resizebuffer(ls->L, ls->buff, LUA_MINBUFFER);  /* initialize buffer */
 169 }
 170
 171
 172
 173 /*
 174 ** =======================================================
 175 ** LEXICAL ANALYZER
 176 ** =======================================================
 177 */
 178
 179
 180
 181 static int check_next (LexState *ls, const char *set) {
 182   if (ls->current == '\0' || !strchr(set, ls->current))
 183     return 0;
 184   save_and_next(ls);
 185   return 1;
 186 }
 187
 188
 189 /*
 190 ** change all characters 'from' in buffer to 'to'
 191 */
 192 static void buffreplace (LexState *ls, char from, char to) {
 193   size_t n = luaZ_bufflen(ls->buff);
 194   char *p = luaZ_buffer(ls->buff);
 195   while (n--)
 196     if (p[n] == from) p[n] = to;
 197 }
 198
 199
 200 #if !defined(getlocaledecpoint)
 201 #define getlocaledecpoint()     (localeconv()->decimal_point[0])
 202 #endif
 203
 204
 205 #define buff2d(b,e)     luaO_str2d(luaZ_buffer(b), luaZ_bufflen(b) - 1, e)
 206
 207 /*
 208 ** in case of format error, try to change decimal point separator to
 209 ** the one defined in the current locale and check again
 210 */
 211 static void trydecpoint (LexState *ls, SemInfo *seminfo) {
 212   char old = ls->decpoint;
 213   ls->decpoint = getlocaledecpoint();
 214   buffreplace(ls, old, ls->decpoint);  /* try new decimal separator */
 215   if (!buff2d(ls->buff, &seminfo->r)) {
 216     /* format error with correct decimal point: no more options */
 217     buffreplace(ls, ls->decpoint, '.');  /* undo change (for error message) */
 218     lexerror(ls, "malformed number", TK_NUMBER);
 219   }
 220 }
 221
 222
 223 /* LUA_NUMBER */
 224 /*
 225 ** this function is quite liberal in what it accepts, as 'luaO_str2d'
 226 ** will reject ill-formed numerals.
 227 */
 228 static void read_numeral (LexState *ls, SemInfo *seminfo) {
 229   const char *expo = "Ee";
 230   int first = ls->current;
 231   lua_assert(lisdigit(ls->current));
 232   save_and_next(ls);
 233   if (first == '0' && check_next(ls, "Xx"))  /* hexadecimal? */
 234     expo = "Pp";
 235   for (;;) {
 236     if (check_next(ls, expo))  /* exponent part? */
 237       (void) check_next(ls, "+-");  /* optional exponent sign */
 238     if (lisxdigit(ls->current) || ls->current == '.')
 239       save_and_next(ls);
 240     else  break;
 241   }
 242   save(ls, '\0');
 243   buffreplace(ls, '.', ls->decpoint);  /* follow locale for decimal point */
 244   if (!buff2d(ls->buff, &seminfo->r))  /* format error? */
 245     trydecpoint(ls, seminfo); /* try to update decimal point separator */
 246 }
 247
 248
 249 /*
 250 ** skip a sequence '[=*[' or ']=*]' and return its number of '='s or
 251 ** -1 if sequence is malformed
 252 */
 253 static int skip_sep (LexState *ls) {
 254   int count = 0;
 255   int s = ls->current;
 256   lua_assert(s == '[' || s == ']');
 257   save_and_next(ls);
 258   while (ls->current == '=') {
 259     save_and_next(ls);
 260     count++;
 261   }
 262   return (ls->current == s) ? count : (-count) - 1;
 263 }
 264
 265
 266 static void read_long_string (LexState *ls, SemInfo *seminfo, int sep) {
 267   save_and_next(ls);  /* skip 2nd `[' */
 268   if (currIsNewline(ls))  /* string starts with a newline? */
 269     inclinenumber(ls);  /* skip it */
 270   for (;;) {
 271     switch (ls->current) {
 272       case EOZ:
 273         lexerror(ls, (seminfo) ? "unfinished long string" :
 274                                  "unfinished long comment", TK_EOS);
 275         break;  /* to avoid warnings */
 276       case ']': {
 277         if (skip_sep(ls) == sep) {
 278           save_and_next(ls);  /* skip 2nd `]' */
 279           goto endloop;
 280         }
 281         break;
 282       }
 283       case '\n': case '\r': {
 284         save(ls, '\n');
 285         inclinenumber(ls);
 286         if (!seminfo) luaZ_resetbuffer(ls->buff);  /* avoid wasting space */
 287         break;
 288       }
 289       default: {
 290         if (seminfo) save_and_next(ls);
 291         else next(ls);
 292       }
 293     }
 294   } endloop:
 295   if (seminfo)
 296     seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + (2 + sep),
 297                                      luaZ_bufflen(ls->buff) - 2*(2 + sep));
 298 }
 299
 300
 301 static void escerror (LexState *ls, int *c, int n, const char *msg) {
 302   int i;
 303   luaZ_resetbuffer(ls->buff);  /* prepare error message */
 304   save(ls, '\\');
 305   for (i = 0; i < n && c[i] != EOZ; i++)
 306     save(ls, c[i]);
 307   lexerror(ls, msg, TK_STRING);
 308 }
 309
 310
 311 static int readhexaesc (LexState *ls) {
 312   int c[3], i;  /* keep input for error message */
 313   int r = 0;  /* result accumulator */
 314   c[0] = 'x';  /* for error message */
 315   for (i = 1; i < 3; i++) {  /* read two hexadecimal digits */
 316     c[i] = next(ls);
 317     if (!lisxdigit(c[i]))
 318       escerror(ls, c, i + 1, "hexadecimal digit expected");
 319     r = (r << 4) + luaO_hexavalue(c[i]);
 320   }
 321   return r;
 322 }
 323
 324
 325 static int readdecesc (LexState *ls) {
 326   int c[3], i;
 327   int r = 0;  /* result accumulator */
 328   for (i = 0; i < 3 && lisdigit(ls->current); i++) {  /* read up to 3 digits */
 329     c[i] = ls->current;
 330     r = 10*r + c[i] - '0';
 331     next(ls);
 332   }
 333   if (r > UCHAR_MAX)
 334     escerror(ls, c, i, "decimal escape too large");
 335   return r;
 336 }
 337
 338
 339 static void read_string (LexState *ls, int del, SemInfo *seminfo) {
 340   save_and_next(ls);  /* keep delimiter (for error messages) */
 341   while (ls->current != del) {
 342     switch (ls->current) {
 343       case EOZ:
 344         lexerror(ls, "unfinished string", TK_EOS);
 345         break;  /* to avoid warnings */
 346       case '\n':
 347       case '\r':
 348         lexerror(ls, "unfinished string", TK_STRING);
 349         break;  /* to avoid warnings */
 350       case '\\': {  /* escape sequences */
 351         int c;  /* final character to be saved */
 352         next(ls);  /* do not save the `\' */
 353         switch (ls->current) {
 354           case 'a': c = '\a'; goto read_save;
 355           case 'b': c = '\b'; goto read_save;
 356           case 'f': c = '\f'; goto read_save;
 357           case 'n': c = '\n'; goto read_save;
 358           case 'r': c = '\r'; goto read_save;
 359           case 't': c = '\t'; goto read_save;
 360           case 'v': c = '\v'; goto read_save;
 361           case 'x': c = readhexaesc(ls); goto read_save;
 362           case '\n': case '\r':
 363             inclinenumber(ls); c = '\n'; goto only_save;
 364           case '\\': case '\"': case '\'':
 365             c = ls->current; goto read_save;
 366           case EOZ: goto no_save;  /* will raise an error next loop */
 367           case 'z': {  /* zap following span of spaces */
 368             next(ls);  /* skip the 'z' */
 369             while (lisspace(ls->current)) {
 370               if (currIsNewline(ls)) inclinenumber(ls);
 371               else next(ls);
 372             }
 373             goto no_save;
 374           }
 375           default: {
 376             if (!lisdigit(ls->current))
 377               escerror(ls, &ls->current, 1, "invalid escape sequence");
 378             /* digital escape \ddd */
 379             c = readdecesc(ls);
 380             goto only_save;
 381           }
 382         }
 383        read_save: next(ls);  /* read next character */
 384        only_save: save(ls, c);  /* save 'c' */
 385        no_save: break;
 386       }
 387       default:
 388         save_and_next(ls);
 389     }
 390   }
 391   save_and_next(ls);  /* skip delimiter */
 392   seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + 1,
 393                                    luaZ_bufflen(ls->buff) - 2);
 394 }
 395
 396
 397 static int llex (LexState *ls, SemInfo *seminfo) {
 398   luaZ_resetbuffer(ls->buff);
 399   for (;;) {
 400     switch (ls->current) {
 401       case '\n': case '\r': {  /* line breaks */
 402         inclinenumber(ls);
 403         break;
 404       }
 405       case ' ': case '\f': case '\t': case '\v': {  /* spaces */
 406         next(ls);
 407         break;
 408       }
 409       case '-': {  /* '-' or '--' (comment) */
 410         next(ls);
 411         if (ls->current != '-') return '-';
 412         /* else is a comment */
 413         next(ls);
 414         if (ls->current == '[') {  /* long comment? */
 415           int sep = skip_sep(ls);
 416           luaZ_resetbuffer(ls->buff);  /* `skip_sep' may dirty the buffer */
 417           if (sep >= 0) {
 418             read_long_string(ls, NULL, sep);  /* skip long comment */
 419             luaZ_resetbuffer(ls->buff);  /* previous call may dirty the buff. */
 420             break;
 421           }
 422         }
 423         /* else short comment */
 424         while (!currIsNewline(ls) && ls->current != EOZ)
 425           next(ls);  /* skip until end of line (or end of file) */
 426         break;
 427       }
 428       case '[': {  /* long string or simply '[' */
 429         int sep = skip_sep(ls);
 430         if (sep >= 0) {
 431           read_long_string(ls, seminfo, sep);
 432           return TK_STRING;
 433         } else if (sep == -1) {
 434                 return '[';
 435         } else {
 436                 lexerror(ls, "invalid long string delimiter", TK_STRING);
 437                 break;
 438         }
 439       }
 440       case '=': {
 441         next(ls);
 442         if (ls->current != '=') return '=';
 443         else { next(ls); return TK_EQ; }
 444       }
 445       case '<': {
 446         next(ls);
 447         if (ls->current != '=') return '<';
 448         else { next(ls); return TK_LE; }
 449       }
 450       case '>': {
 451         next(ls);
 452         if (ls->current != '=') return '>';
 453         else { next(ls); return TK_GE; }
 454       }
 455       case '~': {
 456         next(ls);
 457         if (ls->current != '=') return '~';
 458         else { next(ls); return TK_NE; }
 459       }
 460       case ':': {
 461         next(ls);
 462         if (ls->current != ':') return ':';
 463         else { next(ls); return TK_DBCOLON; }
 464       }
 465       case '"': case '\'': {  /* short literal strings */
 466         read_string(ls, ls->current, seminfo);
 467         return TK_STRING;
 468       }
 469       case '.': {  /* '.', '..', '...', or number */
 470         save_and_next(ls);
 471         if (check_next(ls, ".")) {
 472           if (check_next(ls, "."))
 473             return TK_DOTS;   /* '...' */
 474           else return TK_CONCAT;   /* '..' */
 475         }
 476         else if (!lisdigit(ls->current)) return '.';
 477         /* else go through */
 478       }
 479         zfs_fallthrough;
 480       case '0': case '1': case '2': case '3': case '4':
 481       case '5': case '6': case '7': case '8': case '9': {
 482         read_numeral(ls, seminfo);
 483         return TK_NUMBER;
 484       }
 485       case EOZ: {
 486         return TK_EOS;
 487       }
 488       default: {
 489         if (lislalpha(ls->current)) {  /* identifier or reserved word? */
 490           TString *ts;
 491           do {
 492             save_and_next(ls);
 493           } while (lislalnum(ls->current));
 494           ts = luaX_newstring(ls, luaZ_buffer(ls->buff),
 495                                   luaZ_bufflen(ls->buff));
 496           seminfo->ts = ts;
 497           if (isreserved(ts))  /* reserved word? */
 498             return ts->tsv.extra - 1 + FIRST_RESERVED;
 499           else {
 500             return TK_NAME;
 501           }
 502         }
 503         else {  /* single-char tokens (+ - / ...) */
 504           int c = ls->current;
 505           next(ls);
 506           return c;
 507         }
 508       }
 509     }
 510   }
 511 }
 512
 513
 514 void luaX_next (LexState *ls) {
 515   ls->lastline = ls->linenumber;
 516   if (ls->lookahead.token != TK_EOS) {  /* is there a look-ahead token? */
 517     ls->t = ls->lookahead;  /* use this one */
 518     ls->lookahead.token = TK_EOS;  /* and discharge it */
 519   }
 520   else
 521     ls->t.token = llex(ls, &ls->t.seminfo);  /* read next token */
 522 }
 523
 524
 525 int luaX_lookahead (LexState *ls) {
 526   lua_assert(ls->lookahead.token == TK_EOS);
 527   ls->lookahead.token = llex(ls, &ls->lookahead.seminfo);
 528   return ls->lookahead.token;
 529 }