src/lex.c

   1 char *rcs_lex = "$Id: lex.c,v 3.5 1997/06/16 16:50:22 roberto Exp $";
   2
   3
   4 #include <ctype.h>
   5 #include <string.h>
   6
   7 #include "auxlib.h"
   8 #include "luamem.h"
   9 #include "tree.h"
  10 #include "table.h"
  11 #include "lex.h"
  12 #include "inout.h"
  13 #include "luadebug.h"
  14 #include "parser.h"
  15
  16 #define MINBUFF 250
  17
  18 static int current;  /* look ahead character */
  19 static ZIO *lex_z;
  20
  21
  22 #define next() (current = zgetc(lex_z))
  23 #define save(x) (yytext[tokensize++] = (x))
  24 #define save_and_next()  (save(current), next())
  25
  26
  27 #define MAX_IFS 5
  28
  29 /* "ifstate" keeps the state of each nested $if the lexical is dealing with. */
  30
  31 static struct {
  32   int elsepart;  /* true if its in the $else part */
  33   int condition;  /* true if $if condition is true */
  34   int skip;  /* true if part must be skiped */
  35 } ifstate[MAX_IFS];
  36
  37 static int iflevel;  /* level of nested $if's */
  38
  39
  40 void lua_setinput (ZIO *z)
  41 {
  42   current = '\n';
  43   lua_linenumber = 0;
  44   iflevel = 0;
  45   ifstate[0].skip = 0;
  46   ifstate[0].elsepart = 1;  /* to avoid a free $else */
  47   lex_z = z;
  48 }
  49
  50
  51 static void luaI_auxsyntaxerror (char *s)
  52 {
  53   luaL_verror("%s;\n> at line %d in file %s",
  54                s, lua_linenumber, lua_parsedfile);
  55 }
  56
  57 static void luaI_auxsynterrbf (char *s, char *token)
  58 {
  59   if (token[0] == 0)
  60     token = "<eof>";
  61   luaL_verror("%s;\n> last token read: \"%s\" at line %d in file %s",
  62            s, token, lua_linenumber, lua_parsedfile);
  63 }
  64
  65 void luaI_syntaxerror (char *s)
  66 {
  67   luaI_auxsynterrbf(s, luaI_buffer(1));
  68 }
  69
  70
  71 static struct
  72   {
  73     char *name;
  74     int token;
  75   } reserved [] = {
  76       {"and", AND},
  77       {"do", DO},
  78       {"else", ELSE},
  79       {"elseif", ELSEIF},
  80       {"end", END},
  81       {"function", FUNCTION},
  82       {"if", IF},
  83       {"local", LOCAL},
  84       {"nil", NIL},
  85       {"not", NOT},
  86       {"or", OR},
  87       {"repeat", REPEAT},
  88       {"return", RETURN},
  89       {"then", THEN},
  90       {"until", UNTIL},
  91       {"while", WHILE} };
  92
  93
  94 #define RESERVEDSIZE (sizeof(reserved)/sizeof(reserved[0]))
  95
  96
  97 void luaI_addReserved (void)
  98 {
  99   int i;
 100   for (i=0; i<RESERVEDSIZE; i++)
 101   {
 102     TaggedString *ts = lua_createstring(reserved[i].name);
 103     ts->marked = reserved[i].token;  /* reserved word  (always > 255) */
 104   }
 105 }
 106
 107
 108 /*
 109 ** Pragma handling
 110 */
 111
 112 #define PRAGMASIZE      20
 113
 114 static void skipspace (void)
 115 {
 116   while (current == ' ' || current == '\t') next();
 117 }
 118
 119
 120 static int checkcond (char *buff)
 121 {
 122   static char *opts[] = {"nil", "1"};
 123   int i = luaI_findstring(buff, opts);
 124   if (i >= 0) return i;
 125   else if (isalpha((unsigned char)buff[0]) || buff[0] == '_')
 126     return luaI_globaldefined(buff);
 127   else {
 128     luaI_auxsynterrbf("invalid $if condition", buff);
 129     return 0;  /* to avoid warnings */
 130   }
 131 }
 132
 133
 134 static void readname (char *buff)
 135 {
 136   int i = 0;
 137   skipspace();
 138   while (isalnum((unsigned char)current) || current == '_') {
 139     if (i >= PRAGMASIZE) {
 140       buff[PRAGMASIZE] = 0;
 141       luaI_auxsynterrbf("pragma too long", buff);
 142     }
 143     buff[i++] = current;
 144     next();
 145   }
 146   buff[i] = 0;
 147 }
 148
 149
 150 static void inclinenumber (void);
 151
 152
 153 static void ifskip (void)
 154 {
 155   while (ifstate[iflevel].skip) {
 156     if (current == '\n')
 157       inclinenumber();
 158     else if (current == EOZ)
 159       luaI_auxsyntaxerror("input ends inside a $if");
 160     else next();
 161   }
 162 }
 163
 164
 165 static void inclinenumber (void)
 166 {
 167   static char *pragmas [] =
 168     {"debug", "nodebug", "endinput", "end", "ifnot", "if", "else", NULL};
 169   next();  /* skip '\n' */
 170   ++lua_linenumber;
 171   if (current == '$') {  /* is a pragma? */
 172     char buff[PRAGMASIZE+1];
 173     int ifnot = 0;
 174     int skip = ifstate[iflevel].skip;
 175     next();  /* skip $ */
 176     readname(buff);
 177     switch (luaI_findstring(buff, pragmas)) {
 178       case 0:  /* debug */
 179         if (!skip) lua_debug = 1;
 180         break;
 181       case 1:  /* nodebug */
 182         if (!skip) lua_debug = 0;
 183         break;
 184       case 2:  /* endinput */
 185         if (!skip) {
 186           current = EOZ;
 187           iflevel = 0;  /* to allow $endinput inside a $if */
 188         }
 189         break;
 190       case 3:  /* end */
 191         if (iflevel-- == 0)
 192           luaI_auxsyntaxerror("unmatched $endif");
 193         break;
 194       case 4:  /* ifnot */
 195         ifnot = 1;
 196         /* go through */
 197       case 5:  /* if */
 198         if (iflevel == MAX_IFS-1)
 199           luaI_auxsyntaxerror("too many nested `$ifs'");
 200         readname(buff);
 201         iflevel++;
 202         ifstate[iflevel].elsepart = 0;
 203         ifstate[iflevel].condition = checkcond(buff) ? !ifnot : ifnot;
 204         ifstate[iflevel].skip = skip || !ifstate[iflevel].condition;
 205         break;
 206       case 6:  /* else */
 207         if (ifstate[iflevel].elsepart)
 208           luaI_auxsyntaxerror("unmatched $else");
 209         ifstate[iflevel].elsepart = 1;
 210         ifstate[iflevel].skip =
 211                     ifstate[iflevel-1].skip || ifstate[iflevel].condition;
 212         break;
 213       default:
 214         luaI_auxsynterrbf("invalid pragma", buff);
 215     }
 216     skipspace();
 217     if (current == '\n')  /* pragma must end with a '\n' ... */
 218       inclinenumber();
 219     else if (current != EOZ)  /* or eof */
 220       luaI_auxsyntaxerror("invalid pragma format");
 221     ifskip();
 222   }
 223 }
 224
 225 static int read_long_string (char *yytext, int buffsize)
 226 {
 227   int cont = 0;
 228   int tokensize = 2;  /* '[[' already stored */
 229   while (1)
 230   {
 231     if (buffsize-tokensize <= 2) /* may read more than 1 char in one cicle */
 232       yytext = luaI_buffer(buffsize *= 2);
 233     switch (current)
 234     {
 235       case EOZ:
 236         save(0);
 237         return WRONGTOKEN;
 238       case '[':
 239         save_and_next();
 240         if (current == '[')
 241         {
 242           cont++;
 243           save_and_next();
 244         }
 245         continue;
 246       case ']':
 247         save_and_next();
 248         if (current == ']')
 249         {
 250           if (cont == 0) goto endloop;
 251           cont--;
 252           save_and_next();
 253         }
 254         continue;
 255       case '\n':
 256         save('\n');
 257         inclinenumber();
 258         continue;
 259       default:
 260         save_and_next();
 261     }
 262   } endloop:
 263   save_and_next();  /* pass the second ']' */
 264   yytext[tokensize-2] = 0;  /* erases ']]' */
 265   luaY_lval.vWord = luaI_findconstantbyname(yytext+2);
 266   yytext[tokensize-2] = ']';  /* restores ']]' */
 267   save(0);
 268   return STRING;
 269 }
 270
 271 int luaY_lex (void)
 272 {
 273   static int linelasttoken = 0;
 274   double a;
 275   int buffsize = MINBUFF;
 276   char *yytext = luaI_buffer(buffsize);
 277   yytext[1] = yytext[2] = yytext[3] = 0;
 278   if (lua_debug)
 279     luaI_codedebugline(linelasttoken);
 280   linelasttoken = lua_linenumber;
 281   while (1)
 282   {
 283     int tokensize = 0;
 284     switch (current)
 285     {
 286       case '\n':
 287         inclinenumber();
 288         linelasttoken = lua_linenumber;
 289         continue;
 290
 291       case ' ': case '\t': case '\r':  /* CR: to avoid problems with DOS */
 292         next();
 293         continue;
 294
 295       case '-':
 296         save_and_next();
 297         if (current != '-') return '-';
 298         do { next(); } while (current != '\n' && current != EOZ);
 299         continue;
 300
 301       case '[':
 302         save_and_next();
 303         if (current != '[') return '[';
 304         else
 305         {
 306           save_and_next();  /* pass the second '[' */
 307           return read_long_string(yytext, buffsize);
 308         }
 309
 310       case '=':
 311         save_and_next();
 312         if (current != '=') return '=';
 313         else { save_and_next(); return EQ; }
 314
 315       case '<':
 316         save_and_next();
 317         if (current != '=') return '<';
 318         else { save_and_next(); return LE; }
 319
 320       case '>':
 321         save_and_next();
 322         if (current != '=') return '>';
 323         else { save_and_next(); return GE; }
 324
 325       case '~':
 326         save_and_next();
 327         if (current != '=') return '~';
 328         else { save_and_next(); return NE; }
 329
 330       case '"':
 331       case '\'':
 332       {
 333         int del = current;
 334         save_and_next();
 335         while (current != del)
 336         {
 337           if (buffsize-tokensize <= 2) /* may read more than 1 char in one cicle */
 338             yytext = luaI_buffer(buffsize *= 2);
 339           switch (current)
 340           {
 341             case EOZ:
 342             case '\n':
 343               save(0);
 344               return WRONGTOKEN;
 345             case '\\':
 346               next();  /* do not save the '\' */
 347               switch (current)
 348               {
 349                 case 'n': save('\n'); next(); break;
 350                 case 't': save('\t'); next(); break;
 351                 case 'r': save('\r'); next(); break;
 352                 case '\n': save('\n'); inclinenumber(); break;
 353                 default : save_and_next(); break;
 354               }
 355               break;
 356             default:
 357               save_and_next();
 358           }
 359         }
 360         next();  /* skip delimiter */
 361         save(0);
 362         luaY_lval.vWord = luaI_findconstantbyname(yytext+1);
 363         tokensize--;
 364         save(del); save(0);  /* restore delimiter */
 365         return STRING;
 366       }
 367
 368       case 'a': case 'b': case 'c': case 'd': case 'e':
 369       case 'f': case 'g': case 'h': case 'i': case 'j':
 370       case 'k': case 'l': case 'm': case 'n': case 'o':
 371       case 'p': case 'q': case 'r': case 's': case 't':
 372       case 'u': case 'v': case 'w': case 'x': case 'y':
 373       case 'z':
 374       case 'A': case 'B': case 'C': case 'D': case 'E':
 375       case 'F': case 'G': case 'H': case 'I': case 'J':
 376       case 'K': case 'L': case 'M': case 'N': case 'O':
 377       case 'P': case 'Q': case 'R': case 'S': case 'T':
 378       case 'U': case 'V': case 'W': case 'X': case 'Y':
 379       case 'Z':
 380       case '_':
 381       {
 382         TaggedString *ts;
 383         do {
 384           save_and_next();
 385         } while (isalnum((unsigned char)current) || current == '_');
 386         save(0);
 387         ts = lua_createstring(yytext);
 388         if (ts->marked > 2)
 389           return ts->marked;  /* reserved word */
 390         luaY_lval.pTStr = ts;
 391         ts->marked = 2;  /* avoid GC */
 392         return NAME;
 393       }
 394
 395       case '.':
 396         save_and_next();
 397         if (current == '.')
 398         {
 399           save_and_next();
 400           if (current == '.')
 401           {
 402             save_and_next();
 403             return DOTS;   /* ... */
 404           }
 405           else return CONC;   /* .. */
 406         }
 407         else if (!isdigit((unsigned char)current)) return '.';
 408         /* current is a digit: goes through to number */
 409         a=0.0;
 410         goto fraction;
 411
 412       case '0': case '1': case '2': case '3': case '4':
 413       case '5': case '6': case '7': case '8': case '9':
 414         a=0.0;
 415         do {
 416           a=10.0*a+(current-'0');
 417           save_and_next();
 418         } while (isdigit((unsigned char)current));
 419         if (current == '.') {
 420           save_and_next();
 421           if (current == '.')
 422             luaI_syntaxerror(
 423               "ambiguous syntax (decimal point x string concatenation)");
 424         }
 425       fraction:
 426         { double da=0.1;
 427           while (isdigit((unsigned char)current))
 428           {
 429             a+=(current-'0')*da;
 430             da/=10.0;
 431             save_and_next();
 432           }
 433           if (current == 'e' || current == 'E')
 434           {
 435             int e=0;
 436             int neg;
 437             double ea;
 438             save_and_next();
 439             neg=(current=='-');
 440             if (current == '+' || current == '-') save_and_next();
 441             if (!isdigit((unsigned char)current)) {
 442               save(0); return WRONGTOKEN; }
 443             do {
 444               e=10.0*e+(current-'0');
 445               save_and_next();
 446             } while (isdigit((unsigned char)current));
 447             for (ea=neg?0.1:10.0; e>0; e>>=1)
 448             {
 449               if (e & 1) a*=ea;
 450               ea*=ea;
 451             }
 452           }
 453           luaY_lval.vFloat = a;
 454           save(0);
 455           return NUMBER;
 456         }
 457
 458       case EOZ:
 459         save(0);
 460         if (iflevel > 0)
 461           luaI_syntaxerror("missing $endif");
 462         return 0;
 463
 464       default:
 465         save_and_next();
 466         return yytext[0];
 467     }
 468   }
 469 }
 470