4.3BSD - /usr/src/bin/expr.y

   1: /* Yacc productions for "expr" command: */
   2: 
   3: %token OR AND ADD SUBT MULT DIV REM EQ GT GEQ LT LEQ NEQ
   4: %token A_STRING SUBSTR LENGTH INDEX NOARG MATCH
   5: 
   6: /* operators listed below in increasing precedence: */
   7: %left OR
   8: %left AND
   9: %left EQ LT GT GEQ LEQ NEQ
  10: %left ADD SUBT
  11: %left MULT DIV REM
  12: %left MCH
  13: %left MATCH
  14: %left SUBSTR
  15: %left LENGTH INDEX
  16: %%
  17: 
  18: /* a single `expression' is evaluated and printed: */
  19: 
  20: expression: expr NOARG = {
  21:             printf("%s\n", $1);
  22:             exit((!strcmp($1,"0")||!strcmp($1,"\0"))? 1: 0);
  23:             }
  24:     ;
  25: 
  26: 
  27: expr:   '(' expr ')' = { $$ = $2; }
  28:     | expr OR expr   = { $$ = conj(OR, $1, $3); }
  29:     | expr AND expr   = { $$ = conj(AND, $1, $3); }
  30:     | expr EQ expr   = { $$ = rel(EQ, $1, $3); }
  31:     | expr GT expr   = { $$ = rel(GT, $1, $3); }
  32:     | expr GEQ expr   = { $$ = rel(GEQ, $1, $3); }
  33:     | expr LT expr   = { $$ = rel(LT, $1, $3); }
  34:     | expr LEQ expr   = { $$ = rel(LEQ, $1, $3); }
  35:     | expr NEQ expr   = { $$ = rel(NEQ, $1, $3); }
  36:     | expr ADD expr   = { $$ = arith(ADD, $1, $3); }
  37:     | expr SUBT expr   = { $$ = arith(SUBT, $1, $3); }
  38:     | expr MULT expr   = { $$ = arith(MULT, $1, $3); }
  39:     | expr DIV expr   = { $$ = arith(DIV, $1, $3); }
  40:     | expr REM expr   = { $$ = arith(REM, $1, $3); }
  41:     | expr MCH expr  = { $$ = match($1, $3); }
  42:     | MATCH expr expr = { $$ = match($2, $3); }
  43:     | SUBSTR expr expr expr = { $$ = substr($2, $3, $4); }
  44:     | LENGTH expr       = { $$ = length($2); }
  45:     | INDEX expr expr = { $$ = index($2, $3); }
  46:     | A_STRING
  47:     ;
  48: %%
  49: /*	expression command */
  50: #include <stdio.h>
  51: #define ESIZE   256
  52: #define error(c)    errxx(c)
  53: #define EQL(x,y) !strcmp(x,y)
  54: long atol();
  55: char    **Av;
  56: int Ac;
  57: int Argi;
  58: 
  59: char Mstring[1][128];
  60: char *malloc();
  61: extern int nbra;
  62: 
  63: main(argc, argv) char **argv; {
  64:     Ac = argc;
  65:     Argi = 1;
  66:     Av = argv;
  67:     yyparse();
  68: }
  69: 
  70: char *operators[] = { "|", "&", "+", "-", "*", "/", "%", ":",
  71:     "=", "==", "<", "<=", ">", ">=", "!=",
  72:     "match", "substr", "length", "index", "\0" };
  73: int op[] = { OR, AND, ADD,  SUBT, MULT, DIV, REM, MCH,
  74:     EQ, EQ, LT, LEQ, GT, GEQ, NEQ,
  75:     MATCH, SUBSTR, LENGTH, INDEX };
  76: yylex() {
  77:     register char *p;
  78:     register i;
  79: 
  80:     if(Argi >= Ac) return NOARG;
  81: 
  82:     p = Av[Argi++];
  83: 
  84:     if(*p == '(' || *p == ')')
  85:         return (int)*p;
  86:     for(i = 0; *operators[i]; ++i)
  87:         if(EQL(operators[i], p))
  88:             return op[i];
  89: 
  90:     yylval = p;
  91:     return A_STRING;
  92: }
  93: 
  94: char *rel(op, r1, r2) register char *r1, *r2; {
  95:     register long i;
  96: 
  97:     if(ematch(r1, "-*[0-9]*$") && ematch(r2, "[0-9]*$"))
  98:         i = atol(r1) - atol(r2);
  99:     else
 100:         i = strcmp(r1, r2);
 101:     switch(op) {
 102:     case EQ: i = i==0; break;
 103:     case GT: i = i>0; break;
 104:     case GEQ: i = i>=0; break;
 105:     case LT: i = i<0; break;
 106:     case LEQ: i = i<=0; break;
 107:     case NEQ: i = i!=0; break;
 108:     }
 109:     return i? "1": "0";
 110: }
 111: 
 112: char *arith(op, r1, r2) char *r1, *r2; {
 113:     long i1, i2;
 114:     register char *rv;
 115: 
 116:     if(!(ematch(r1, "[0-9]*$") && ematch(r2, "[0-9]*$")))
 117:         yyerror("non-numeric argument");
 118:     i1 = atol(r1);
 119:     i2 = atol(r2);
 120: 
 121:     switch(op) {
 122:     case ADD: i1 = i1 + i2; break;
 123:     case SUBT: i1 = i1 - i2; break;
 124:     case MULT: i1 = i1 * i2; break;
 125:     case DIV: i1 = i1 / i2; break;
 126:     case REM: i1 = i1 % i2; break;
 127:     }
 128:     rv = malloc(16);
 129:     sprintf(rv, "%D", i1);
 130:     return rv;
 131: }
 132: char *conj(op, r1, r2) char *r1, *r2; {
 133:     register char *rv;
 134: 
 135:     switch(op) {
 136: 
 137:     case OR:
 138:         if(EQL(r1, "0")
 139:         || EQL(r1, ""))
 140:             if(EQL(r2, "0")
 141:             || EQL(r2, ""))
 142:                 rv = "0";
 143:             else
 144:                 rv = r2;
 145:         else
 146:             rv = r1;
 147:         break;
 148:     case AND:
 149:         if(EQL(r1, "0")
 150:         || EQL(r1, ""))
 151:             rv = "0";
 152:         else if(EQL(r2, "0")
 153:         || EQL(r2, ""))
 154:             rv = "0";
 155:         else
 156:             rv = r1;
 157:         break;
 158:     }
 159:     return rv;
 160: }
 161: 
 162: char *substr(v, s, w) char *v, *s, *w; {
 163: register si, wi;
 164: register char *res;
 165: 
 166:     si = atol(s);
 167:     wi = atol(w);
 168:     while(--si) if(*v) ++v;
 169: 
 170:     res = v;
 171: 
 172:     while(wi--) if(*v) ++v;
 173: 
 174:     *v = '\0';
 175:     return res;
 176: }
 177: 
 178: char *length(s) register char *s; {
 179:     register i = 0;
 180:     register char *rv;
 181: 
 182:     while(*s++) ++i;
 183: 
 184:     rv = malloc(8);
 185:     sprintf(rv, "%d", i);
 186:     return rv;
 187: }
 188: 
 189: char *index(s, t) char *s, *t; {
 190:     register i, j;
 191:     register char *rv;
 192: 
 193:     for(i = 0; s[i] ; ++i)
 194:         for(j = 0; t[j] ; ++j)
 195:             if(s[i]==t[j]) {
 196:                 sprintf(rv = malloc(8), "%d", ++i);
 197:                 return rv;
 198:             }
 199:     return "0";
 200: }
 201: 
 202: char *match(s, p)
 203: {
 204:     register char *rv;
 205: 
 206:     sprintf(rv = malloc(8), "%d", ematch(s, p));
 207:     if(nbra) {
 208:         rv = malloc(strlen(Mstring[0])+1);
 209:         strcpy(rv, Mstring[0]);
 210:     }
 211:     return rv;
 212: }
 213: 
 214: #define INIT    register char *sp = instring;
 215: #define GETC()      (*sp++)
 216: #define PEEKC()     (*sp)
 217: #define UNGETC(c)   (--sp)
 218: #define RETURN(c)   return
 219: #define ERROR(c)    errxx(c)
 220: 
 221: 
 222: ematch(s, p)
 223: char *s;
 224: register char *p;
 225: {
 226:     static char expbuf[ESIZE];
 227:     char *compile();
 228:     register num;
 229:     extern char *braslist[], *braelist[], *loc2;
 230: 
 231:     compile(p, expbuf, &expbuf[ESIZE], 0);
 232:     if(nbra > 1)
 233:         yyerror("Too many '\\('s");
 234:     if(advance(s, expbuf)) {
 235:         if(nbra == 1) {
 236:             p = braslist[0];
 237:             num = braelist[0] - p;
 238:             strncpy(Mstring[0], p, num);
 239:             Mstring[0][num] = '\0';
 240:         }
 241:         return(loc2-s);
 242:     }
 243:     return(0);
 244: }
 245: 
 246: errxx(c)
 247: {
 248:     yyerror("RE error");
 249: }
 250: 
 251: #define CBRA    2
 252: #define CCHR    4
 253: #define CDOT    8
 254: #define CCL 12
 255: #define CDOL    20
 256: #define CEOF    22
 257: #define CKET    24
 258: #define CBACK   36
 259: 
 260: #define STAR    01
 261: #define RNGE    03
 262: 
 263: #define NBRA    9
 264: 
 265: #define PLACE(c)    ep[c >> 3] |= bittab[c & 07]
 266: #define ISTHERE(c)  (ep[c >> 3] & bittab[c & 07])
 267: 
 268: char    *braslist[NBRA];
 269: char    *braelist[NBRA];
 270: int nbra;
 271: char *loc1, *loc2, *locs;
 272: int sed;
 273: 
 274: int circf;
 275: int low;
 276: int size;
 277: 
 278: char    bittab[] = {
 279:     1,
 280:     2,
 281:     4,
 282:     8,
 283:     16,
 284:     32,
 285:     64,
 286:     128
 287: };
 288: 
 289: char *
 290: compile(instring, ep, endbuf, seof)
 291: register char *ep;
 292: char *instring, *endbuf;
 293: {
 294:     INIT    /* Dependent declarations and initializations */
 295:     register c;
 296:     register eof = seof;
 297:     char *lastep = instring;
 298:     int cclcnt;
 299:     char bracket[NBRA], *bracketp;
 300:     int closed;
 301:     char neg;
 302:     int lc;
 303:     int i, cflg;
 304: 
 305:     lastep = 0;
 306:     if((c = GETC()) == eof) {
 307:         if(*ep == 0 && !sed)
 308:             ERROR(41);
 309:         RETURN(ep);
 310:     }
 311:     bracketp = bracket;
 312:     circf = closed = nbra = 0;
 313:     if (c == '^')
 314:         circf++;
 315:     else
 316:         UNGETC(c);
 317:     for (;;) {
 318:         if (ep >= endbuf)
 319:             ERROR(50);
 320:         if((c = GETC()) != '*' && ((c != '\\') || (PEEKC() != '{')))
 321:             lastep = ep;
 322:         if (c == eof) {
 323:             *ep++ = CEOF;
 324:             RETURN(ep);
 325:         }
 326:         switch (c) {
 327: 
 328:         case '.':
 329:             *ep++ = CDOT;
 330:             continue;
 331: 
 332:         case '\n':
 333:             ERROR(36);
 334:         case '*':
 335:             if (lastep==0 || *lastep==CBRA || *lastep==CKET)
 336:                 goto defchar;
 337:             *lastep |= STAR;
 338:             continue;
 339: 
 340:         case '$':
 341:             if(PEEKC() != eof)
 342:                 goto defchar;
 343:             *ep++ = CDOL;
 344:             continue;
 345: 
 346:         case '[':
 347:             if(&ep[17] >= endbuf)
 348:                 ERROR(50);
 349: 
 350:             *ep++ = CCL;
 351:             lc = 0;
 352:             for(i = 0; i < 16; i++)
 353:                 ep[i] = 0;
 354: 
 355:             neg = 0;
 356:             if((c = GETC()) == '^') {
 357:                 neg = 1;
 358:                 c = GETC();
 359:             }
 360: 
 361:             do {
 362:                 if(c == '\0' || c == '\n')
 363:                     ERROR(49);
 364:                 if(c == '-' && lc != 0) {
 365:                     if ((c = GETC()) == ']') {
 366:                         PLACE('-');
 367:                         break;
 368:                     }
 369:                     while(lc < c) {
 370:                         PLACE(lc);
 371:                         lc++;
 372:                     }
 373:                 }
 374:                 lc = c;
 375:                 PLACE(c);
 376:             } while((c = GETC()) != ']');
 377:             if(neg) {
 378:                 for(cclcnt = 0; cclcnt < 16; cclcnt++)
 379:                     ep[cclcnt] ^= -1;
 380:                 ep[0] &= 0376;
 381:             }
 382: 
 383:             ep += 16;
 384: 
 385:             continue;
 386: 
 387:         case '\\':
 388:             switch(c = GETC()) {
 389: 
 390:             case '(':
 391:                 if(nbra >= NBRA)
 392:                     ERROR(43);
 393:                 *bracketp++ = nbra;
 394:                 *ep++ = CBRA;
 395:                 *ep++ = nbra++;
 396:                 continue;
 397: 
 398:             case ')':
 399:                 if(bracketp <= bracket)
 400:                     ERROR(42);
 401:                 *ep++ = CKET;
 402:                 *ep++ = *--bracketp;
 403:                 closed++;
 404:                 continue;
 405: 
 406:             case '{':
 407:                 if(lastep == (char *) (0))
 408:                     goto defchar;
 409:                 *lastep |= RNGE;
 410:                 cflg = 0;
 411:             nlim:
 412:                 c = GETC();
 413:                 i = 0;
 414:                 do {
 415:                     if ('0' <= c && c <= '9')
 416:                         i = 10 * i + c - '0';
 417:                     else
 418:                         ERROR(16);
 419:                 } while(((c = GETC()) != '\\') && (c != ','));
 420:                 if (i > 255)
 421:                     ERROR(11);
 422:                 *ep++ = i;
 423:                 if (c == ',') {
 424:                     if(cflg++)
 425:                         ERROR(44);
 426:                     if((c = GETC()) == '\\')
 427:                         *ep++ = 255;
 428:                     else {
 429:                         UNGETC(c);
 430:                         goto nlim; /* get 2'nd number */
 431:                     }
 432:                 }
 433:                 if(GETC() != '}')
 434:                     ERROR(45);
 435:                 if(!cflg)   /* one number */
 436:                     *ep++ = i;
 437:                 else if((ep[-1] & 0377) < (ep[-2] & 0377))
 438:                     ERROR(46);
 439:                 continue;
 440: 
 441:             case '\n':
 442:                 ERROR(36);
 443: 
 444:             case 'n':
 445:                 c = '\n';
 446:                 goto defchar;
 447: 
 448:             default:
 449:                 if(c >= '1' && c <= '9') {
 450:                     if((c -= '1') >= closed)
 451:                         ERROR(25);
 452:                     *ep++ = CBACK;
 453:                     *ep++ = c;
 454:                     continue;
 455:                 }
 456:             }
 457:             /* Drop through to default to use \ to turn off special chars */
 458: 
 459:         defchar:
 460:         default:
 461:             lastep = ep;
 462:             *ep++ = CCHR;
 463:             *ep++ = c;
 464:         }
 465:     }
 466: }
 467: 
 468: step(p1, p2)
 469: register char *p1, *p2;
 470: {
 471:     register c;
 472: 
 473:     if (circf) {
 474:         loc1 = p1;
 475:         return(advance(p1, p2));
 476:     }
 477:     /* fast check for first character */
 478:     if (*p2==CCHR) {
 479:         c = p2[1];
 480:         do {
 481:             if (*p1 != c)
 482:                 continue;
 483:             if (advance(p1, p2)) {
 484:                 loc1 = p1;
 485:                 return(1);
 486:             }
 487:         } while (*p1++);
 488:         return(0);
 489:     }
 490:         /* regular algorithm */
 491:     do {
 492:         if (advance(p1, p2)) {
 493:             loc1 = p1;
 494:             return(1);
 495:         }
 496:     } while (*p1++);
 497:     return(0);
 498: }
 499: 
 500: advance(lp, ep)
 501: register char *lp, *ep;
 502: {
 503:     register char *curlp;
 504:     char c;
 505:     char *bbeg;
 506:     int ct;
 507: 
 508:     for (;;) switch (*ep++) {
 509: 
 510:     case CCHR:
 511:         if (*ep++ == *lp++)
 512:             continue;
 513:         return(0);
 514: 
 515:     case CDOT:
 516:         if (*lp++)
 517:             continue;
 518:         return(0);
 519: 
 520:     case CDOL:
 521:         if (*lp==0)
 522:             continue;
 523:         return(0);
 524: 
 525:     case CEOF:
 526:         loc2 = lp;
 527:         return(1);
 528: 
 529:     case CCL:
 530:         c = *lp++ & 0177;
 531:         if(ISTHERE(c)) {
 532:             ep += 16;
 533:             continue;
 534:         }
 535:         return(0);
 536:     case CBRA:
 537:         braslist[*ep++] = lp;
 538:         continue;
 539: 
 540:     case CKET:
 541:         braelist[*ep++] = lp;
 542:         continue;
 543: 
 544:     case CCHR|RNGE:
 545:         c = *ep++;
 546:         getrnge(ep);
 547:         while(low--)
 548:             if(*lp++ != c)
 549:                 return(0);
 550:         curlp = lp;
 551:         while(size--)
 552:             if(*lp++ != c)
 553:                 break;
 554:         if(size < 0)
 555:             lp++;
 556:         ep += 2;
 557:         goto star;
 558: 
 559:     case CDOT|RNGE:
 560:         getrnge(ep);
 561:         while(low--)
 562:             if(*lp++ == '\0')
 563:                 return(0);
 564:         curlp = lp;
 565:         while(size--)
 566:             if(*lp++ == '\0')
 567:                 break;
 568:         if(size < 0)
 569:             lp++;
 570:         ep += 2;
 571:         goto star;
 572: 
 573:     case CCL|RNGE:
 574:         getrnge(ep + 16);
 575:         while(low--) {
 576:             c = *lp++ & 0177;
 577:             if(!ISTHERE(c))
 578:                 return(0);
 579:         }
 580:         curlp = lp;
 581:         while(size--) {
 582:             c = *lp++ & 0177;
 583:             if(!ISTHERE(c))
 584:                 break;
 585:         }
 586:         if(size < 0)
 587:             lp++;
 588:         ep += 18;       /* 16 + 2 */
 589:         goto star;
 590: 
 591:     case CBACK:
 592:         bbeg = braslist[*ep];
 593:         ct = braelist[*ep++] - bbeg;
 594: 
 595:         if(ecmp(bbeg, lp, ct)) {
 596:             lp += ct;
 597:             continue;
 598:         }
 599:         return(0);
 600: 
 601:     case CBACK|STAR:
 602:         bbeg = braslist[*ep];
 603:         ct = braelist[*ep++] - bbeg;
 604:         curlp = lp;
 605:         while(ecmp(bbeg, lp, ct))
 606:             lp += ct;
 607: 
 608:         while(lp >= curlp) {
 609:             if(advance(lp, ep)) return(1);
 610:             lp -= ct;
 611:         }
 612:         return(0);
 613: 
 614: 
 615:     case CDOT|STAR:
 616:         curlp = lp;
 617:         while (*lp++);
 618:         goto star;
 619: 
 620:     case CCHR|STAR:
 621:         curlp = lp;
 622:         while (*lp++ == *ep);
 623:         ep++;
 624:         goto star;
 625: 
 626:     case CCL|STAR:
 627:         curlp = lp;
 628:         do {
 629:             c = *lp++ & 0177;
 630:         } while(ISTHERE(c));
 631:         ep += 16;
 632:         goto star;
 633: 
 634:     star:
 635:         do {
 636:             if(--lp == locs)
 637:                 break;
 638:             if (advance(lp, ep))
 639:                 return(1);
 640:         } while (lp > curlp);
 641:         return(0);
 642: 
 643:     }
 644: }
 645: 
 646: getrnge(str)
 647: register char *str;
 648: {
 649:     low = *str++ & 0377;
 650:     size = *str == 255 ? 20000 : (*str &0377) - low;
 651: }
 652: 
 653: ecmp(a, b, count)
 654: register char   *a, *b;
 655: register    count;
 656: {
 657:     if(a == b) /* should have been caught in compile() */
 658:         error(51);
 659:     while(count--)
 660:         if(*a++ != *b++)    return(0);
 661:     return(1);
 662: }
 663: 
 664: static char *sccsid = "@(#)expr.y	4.4 (Berkeley) 5/21/84";
 665: yyerror(s)
 666: 
 667: {
 668:     fprintf(stderr, "%s\n", s);
 669:     exit(2);
 670: }

Defined functions

Defined variables

Defined macros