external/mit/lua/dist/src/llex.c

   1 /*      $NetBSD: llex.c,v 1.1.1.2 2012/03/15 00:08:13 alnsn Exp $       */
   2
   3 /*
   4 ** $Id: llex.c,v 1.1.1.2 2012/03/15 00:08:13 alnsn Exp $
   5 ** Lexical Analyzer
   6 ** See Copyright Notice in lua.h
   7 */
   8
   9
  10 #include <ctype.h>
  11 #include <locale.h>
  12 #include <string.h>
  13
  14 #define llex_c
  15 #define LUA_CORE
  16
  17 #include "lua.h"
  18
  19 #include "ldo.h"
  20 #include "llex.h"
  21 #include "lobject.h"
  22 #include "lparser.h"
  23 #include "lstate.h"
  24 #include "lstring.h"
  25 #include "ltable.h"
  26 #include "lzio.h"
  27
  28
  29
  30 #define next(ls) (ls->current = zgetc(ls->z))
  31
  32
  33
  34
  35 #define currIsNewline(ls)       (ls->current == '\n' || ls->current == '\r')
  36
  37
  38 /* ORDER RESERVED */
  39 const char *const luaX_tokens [] = {
  40     "and", "break", "do", "else", "elseif",
  41     "end", "false", "for", "function", "if",
  42     "in", "local", "nil", "not", "or", "repeat",
  43     "return", "then", "true", "until", "while",
  44     "..", "...", "==", ">=", "<=", "~=",
  45     "<number>", "<name>", "<string>", "<eof>",
  46     NULL
  47 };
  48
  49
  50 #define save_and_next(ls) (save(ls, ls->current), next(ls))
  51
  52
  53 static void save (LexState *ls, int c) {
  54   Mbuffer *b = ls->buff;
  55   if (b->n + 1 > b->buffsize) {
  56     size_t newsize;
  57     if (b->buffsize >= MAX_SIZET/2)
  58       luaX_lexerror(ls, "lexical element too long", 0);
  59     newsize = b->buffsize * 2;
  60     luaZ_resizebuffer(ls->L, b, newsize);
  61   }
  62   b->buffer[b->n++] = cast(char, c);
  63 }
  64
  65
  66 void luaX_init (lua_State *L) {
  67   int i;
  68   for (i=0; i<NUM_RESERVED; i++) {
  69     TString *ts = luaS_new(L, luaX_tokens[i]);
  70     luaS_fix(ts);  /* reserved words are never collected */
  71     lua_assert(strlen(luaX_tokens[i])+1 <= TOKEN_LEN);
  72     ts->tsv.reserved = cast_byte(i+1);  /* reserved word */
  73   }
  74 }
  75
  76
  77 #define MAXSRC          80
  78
  79
  80 const char *luaX_token2str (LexState *ls, int token) {
  81   if (token < FIRST_RESERVED) {
  82     lua_assert(token == cast(unsigned char, token));
  83     return (iscntrl(token)) ? luaO_pushfstring(ls->L, "char(%d)", token) :
  84                               luaO_pushfstring(ls->L, "%c", token);
  85   }
  86   else
  87     return luaX_tokens[token-FIRST_RESERVED];
  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 luaZ_buffer(ls->buff);
  98     default:
  99       return luaX_token2str(ls, token);
 100   }
 101 }
 102
 103
 104 void luaX_lexerror (LexState *ls, const char *msg, int token) {
 105   char buff[MAXSRC];
 106   luaO_chunkid(buff, getstr(ls->source), MAXSRC);
 107   msg = luaO_pushfstring(ls->L, "%s:%d: %s", buff, ls->linenumber, msg);
 108   if (token)
 109     luaO_pushfstring(ls->L, "%s near " LUA_QS, msg, txtToken(ls, token));
 110   luaD_throw(ls->L, LUA_ERRSYNTAX);
 111 }
 112
 113
 114 void luaX_syntaxerror (LexState *ls, const char *msg) {
 115   luaX_lexerror(ls, msg, ls->t.token);
 116 }
 117
 118
 119 TString *luaX_newstring (LexState *ls, const char *str, size_t l) {
 120   lua_State *L = ls->L;
 121   TString *ts = luaS_newlstr(L, str, l);
 122   TValue *o = luaH_setstr(L, ls->fs->h, ts);  /* entry for `str' */
 123   if (ttisnil(o)) {
 124     setbvalue(o, 1);  /* make sure `str' will not be collected */
 125     luaC_checkGC(L);
 126   }
 127   return ts;
 128 }
 129
 130
 131 static void inclinenumber (LexState *ls) {
 132   int old = ls->current;
 133   lua_assert(currIsNewline(ls));
 134   next(ls);  /* skip `\n' or `\r' */
 135   if (currIsNewline(ls) && ls->current != old)
 136     next(ls);  /* skip `\n\r' or `\r\n' */
 137   if (++ls->linenumber >= MAX_INT)
 138     luaX_syntaxerror(ls, "chunk has too many lines");
 139 }
 140
 141
 142 void luaX_setinput (lua_State *L, LexState *ls, ZIO *z, TString *source) {
 143   ls->decpoint = '.';
 144   ls->L = L;
 145   ls->lookahead.token = TK_EOS;  /* no look-ahead token */
 146   ls->z = z;
 147   ls->fs = NULL;
 148   ls->linenumber = 1;
 149   ls->lastline = 1;
 150   ls->source = source;
 151   luaZ_resizebuffer(ls->L, ls->buff, LUA_MINBUFFER);  /* initialize buffer */
 152   next(ls);  /* read first char */
 153 }
 154
 155
 156
 157 /*
 158 ** =======================================================
 159 ** LEXICAL ANALYZER
 160 ** =======================================================
 161 */
 162
 163
 164
 165 static int check_next (LexState *ls, const char *set) {
 166   if (!strchr(set, ls->current))
 167     return 0;
 168   save_and_next(ls);
 169   return 1;
 170 }
 171
 172
 173 static void buffreplace (LexState *ls, char from, char to) {
 174   size_t n = luaZ_bufflen(ls->buff);
 175   char *p = luaZ_buffer(ls->buff);
 176   while (n--)
 177     if (p[n] == from) p[n] = to;
 178 }
 179
 180
 181 static void trydecpoint (LexState *ls, SemInfo *seminfo) {
 182   /* format error: try to update decimal point separator */
 183   struct lconv *cv = localeconv();
 184   char old = ls->decpoint;
 185   ls->decpoint = (cv ? cv->decimal_point[0] : '.');
 186   buffreplace(ls, old, ls->decpoint);  /* try updated decimal separator */
 187   if (!luaO_str2d(luaZ_buffer(ls->buff), &seminfo->r)) {
 188     /* format error with correct decimal point: no more options */
 189     buffreplace(ls, ls->decpoint, '.');  /* undo change (for error message) */
 190     luaX_lexerror(ls, "malformed number", TK_NUMBER);
 191   }
 192 }
 193
 194
 195 /* LUA_NUMBER */
 196 static void read_numeral (LexState *ls, SemInfo *seminfo) {
 197   lua_assert(isdigit(ls->current));
 198   do {
 199     save_and_next(ls);
 200   } while (isdigit(ls->current) || ls->current == '.');
 201   if (check_next(ls, "Ee"))  /* `E'? */
 202     check_next(ls, "+-");  /* optional exponent sign */
 203   while (isalnum(ls->current) || ls->current == '_')
 204     save_and_next(ls);
 205   save(ls, '\0');
 206   buffreplace(ls, '.', ls->decpoint);  /* follow locale for decimal point */
 207   if (!luaO_str2d(luaZ_buffer(ls->buff), &seminfo->r))  /* format error? */
 208     trydecpoint(ls, seminfo); /* try to update decimal point separator */
 209 }
 210
 211
 212 static int skip_sep (LexState *ls) {
 213   int count = 0;
 214   int s = ls->current;
 215   lua_assert(s == '[' || s == ']');
 216   save_and_next(ls);
 217   while (ls->current == '=') {
 218     save_and_next(ls);
 219     count++;
 220   }
 221   return (ls->current == s) ? count : (-count) - 1;
 222 }
 223
 224
 225 static void read_long_string (LexState *ls, SemInfo *seminfo, int sep) {
 226   int cont = 0;
 227   (void)(cont);  /* avoid warnings when `cont' is not used */
 228   save_and_next(ls);  /* skip 2nd `[' */
 229   if (currIsNewline(ls))  /* string starts with a newline? */
 230     inclinenumber(ls);  /* skip it */
 231   for (;;) {
 232     switch (ls->current) {
 233       case EOZ:
 234         luaX_lexerror(ls, (seminfo) ? "unfinished long string" :
 235                                    "unfinished long comment", TK_EOS);
 236         break;  /* to avoid warnings */
 237 #if defined(LUA_COMPAT_LSTR)
 238       case '[': {
 239         if (skip_sep(ls) == sep) {
 240           save_and_next(ls);  /* skip 2nd `[' */
 241           cont++;
 242 #if LUA_COMPAT_LSTR == 1
 243           if (sep == 0)
 244             luaX_lexerror(ls, "nesting of [[...]] is deprecated", '[');
 245 #endif
 246         }
 247         break;
 248       }
 249 #endif
 250       case ']': {
 251         if (skip_sep(ls) == sep) {
 252           save_and_next(ls);  /* skip 2nd `]' */
 253 #if defined(LUA_COMPAT_LSTR) && LUA_COMPAT_LSTR == 2
 254           cont--;
 255           if (sep == 0 && cont >= 0) break;
 256 #endif
 257           goto endloop;
 258         }
 259         break;
 260       }
 261       case '\n':
 262       case '\r': {
 263         save(ls, '\n');
 264         inclinenumber(ls);
 265         if (!seminfo) luaZ_resetbuffer(ls->buff);  /* avoid wasting space */
 266         break;
 267       }
 268       default: {
 269         if (seminfo) save_and_next(ls);
 270         else next(ls);
 271       }
 272     }
 273   } endloop:
 274   if (seminfo)
 275     seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + (2 + sep),
 276                                      luaZ_bufflen(ls->buff) - 2*(2 + sep));
 277 }
 278
 279
 280 static void read_string (LexState *ls, int del, SemInfo *seminfo) {
 281   save_and_next(ls);
 282   while (ls->current != del) {
 283     switch (ls->current) {
 284       case EOZ:
 285         luaX_lexerror(ls, "unfinished string", TK_EOS);
 286         continue;  /* to avoid warnings */
 287       case '\n':
 288       case '\r':
 289         luaX_lexerror(ls, "unfinished string", TK_STRING);
 290         continue;  /* to avoid warnings */
 291       case '\\': {
 292         int c;
 293         next(ls);  /* do not save the `\' */
 294         switch (ls->current) {
 295           case 'a': c = '\a'; break;
 296           case 'b': c = '\b'; break;
 297           case 'f': c = '\f'; break;
 298           case 'n': c = '\n'; break;
 299           case 'r': c = '\r'; break;
 300           case 't': c = '\t'; break;
 301           case 'v': c = '\v'; break;
 302           case '\n':  /* go through */
 303           case '\r': save(ls, '\n'); inclinenumber(ls); continue;
 304           case EOZ: continue;  /* will raise an error next loop */
 305           default: {
 306             if (!isdigit(ls->current))
 307               save_and_next(ls);  /* handles \\, \", \', and \? */
 308             else {  /* \xxx */
 309               int i = 0;
 310               c = 0;
 311               do {
 312                 c = 10*c + (ls->current-'0');
 313                 next(ls);
 314               } while (++i<3 && isdigit(ls->current));
 315               if (c > UCHAR_MAX)
 316                 luaX_lexerror(ls, "escape sequence too large", TK_STRING);
 317               save(ls, c);
 318             }
 319             continue;
 320           }
 321         }
 322         save(ls, c);
 323         next(ls);
 324         continue;
 325       }
 326       default:
 327         save_and_next(ls);
 328     }
 329   }
 330   save_and_next(ls);  /* skip delimiter */
 331   seminfo->ts = luaX_newstring(ls, luaZ_buffer(ls->buff) + 1,
 332                                    luaZ_bufflen(ls->buff) - 2);
 333 }
 334
 335
 336 static int llex (LexState *ls, SemInfo *seminfo) {
 337   luaZ_resetbuffer(ls->buff);
 338   for (;;) {
 339     switch (ls->current) {
 340       case '\n':
 341       case '\r': {
 342         inclinenumber(ls);
 343         continue;
 344       }
 345       case '-': {
 346         next(ls);
 347         if (ls->current != '-') return '-';
 348         /* else is a comment */
 349         next(ls);
 350         if (ls->current == '[') {
 351           int sep = skip_sep(ls);
 352           luaZ_resetbuffer(ls->buff);  /* `skip_sep' may dirty the buffer */
 353           if (sep >= 0) {
 354             read_long_string(ls, NULL, sep);  /* long comment */
 355             luaZ_resetbuffer(ls->buff);
 356             continue;
 357           }
 358         }
 359         /* else short comment */
 360         while (!currIsNewline(ls) && ls->current != EOZ)
 361           next(ls);
 362         continue;
 363       }
 364       case '[': {
 365         int sep = skip_sep(ls);
 366         if (sep >= 0) {
 367           read_long_string(ls, seminfo, sep);
 368           return TK_STRING;
 369         }
 370         else if (sep == -1) return '[';
 371         else luaX_lexerror(ls, "invalid long string delimiter", TK_STRING);
 372       }
 373       case '=': {
 374         next(ls);
 375         if (ls->current != '=') return '=';
 376         else { next(ls); return TK_EQ; }
 377       }
 378       case '<': {
 379         next(ls);
 380         if (ls->current != '=') return '<';
 381         else { next(ls); return TK_LE; }
 382       }
 383       case '>': {
 384         next(ls);
 385         if (ls->current != '=') return '>';
 386         else { next(ls); return TK_GE; }
 387       }
 388       case '~': {
 389         next(ls);
 390         if (ls->current != '=') return '~';
 391         else { next(ls); return TK_NE; }
 392       }
 393       case '"':
 394       case '\'': {
 395         read_string(ls, ls->current, seminfo);
 396         return TK_STRING;
 397       }
 398       case '.': {
 399         save_and_next(ls);
 400         if (check_next(ls, ".")) {
 401           if (check_next(ls, "."))
 402             return TK_DOTS;   /* ... */
 403           else return TK_CONCAT;   /* .. */
 404         }
 405         else if (!isdigit(ls->current)) return '.';
 406         else {
 407           read_numeral(ls, seminfo);
 408           return TK_NUMBER;
 409         }
 410       }
 411       case EOZ: {
 412         return TK_EOS;
 413       }
 414       default: {
 415         if (isspace(ls->current)) {
 416           lua_assert(!currIsNewline(ls));
 417           next(ls);
 418           continue;
 419         }
 420         else if (isdigit(ls->current)) {
 421           read_numeral(ls, seminfo);
 422           return TK_NUMBER;
 423         }
 424         else if (isalpha(ls->current) || ls->current == '_') {
 425           /* identifier or reserved word */
 426           TString *ts;
 427           do {
 428             save_and_next(ls);
 429           } while (isalnum(ls->current) || ls->current == '_');
 430           ts = luaX_newstring(ls, luaZ_buffer(ls->buff),
 431                                   luaZ_bufflen(ls->buff));
 432           if (ts->tsv.reserved > 0)  /* reserved word? */
 433             return ts->tsv.reserved - 1 + FIRST_RESERVED;
 434           else {
 435             seminfo->ts = ts;
 436             return TK_NAME;
 437           }
 438         }
 439         else {
 440           int c = ls->current;
 441           next(ls);
 442           return c;  /* single-char tokens (+ - / ...) */
 443         }
 444       }
 445     }
 446   }
 447 }
 448
 449
 450 void luaX_next (LexState *ls) {
 451   ls->lastline = ls->linenumber;
 452   if (ls->lookahead.token != TK_EOS) {  /* is there a look-ahead token? */
 453     ls->t = ls->lookahead;  /* use this one */
 454     ls->lookahead.token = TK_EOS;  /* and discharge it */
 455   }
 456   else
 457     ls->t.token = llex(ls, &ls->t.seminfo);  /* read next token */
 458 }
 459
 460
 461 void luaX_lookahead (LexState *ls) {
 462   lua_assert(ls->lookahead.token == TK_EOS);
 463   ls->lookahead.token = llex(ls, &ls->lookahead.seminfo);
 464 }
 465