1 #ifndef lint 2 static char sccsid[] = "@(#)cpp.c 1.8 05/16/84"; 3 #endif lint 4 5 #ifdef FLEXNAMES 6 #define NCPS 128 7 #else 8 #define NCPS 8 9 #endif 10 11 # include "stdio.h" 12 # include "ctype.h" 13 /* C command 14 /* written by John F. Reiser 15 /* July/August 1978 16 */ 17 18 #define STATIC 19 20 #define STDIN 0 21 #define STDOUT 1 22 #define STDERR 2 23 #define READ 0 24 #define WRITE 1 25 #define SALT '#' 26 #ifndef BUFSIZ 27 #define BUFSIZ 512 28 #endif 29 30 char *pbeg,*pbuf,*pend; 31 char *outp,*inp; 32 char *newp; 33 char cinit; 34 35 /* some code depends on whether characters are sign or zero extended */ 36 /* #if '\377' < 0 not used here, old cpp doesn't understand */ 37 #if pdp11 | vax | mc68000 38 #define COFF 128 39 #else 40 #define COFF 0 41 #endif 42 43 # if gcos 44 #define ALFSIZ 512 /* alphabet size */ 45 # else 46 #define ALFSIZ 256 /* alphabet size */ 47 # endif 48 char macbit[ALFSIZ+11]; 49 char toktyp[ALFSIZ]; 50 #define BLANK 1 51 #define IDENT 2 52 #define NUMBR 3 53 54 /* a superimposed code is used to reduce the number of calls to the 55 /* symbol table lookup routine. (if the kth character of an identifier 56 /* is 'a' and there are no macro names whose kth character is 'a' 57 /* then the identifier cannot be a macro name, hence there is no need 58 /* to look in the symbol table.) 'scw1' enables the test based on 59 /* single characters and their position in the identifier. 'scw2' 60 /* enables the test based on adjacent pairs of characters and their 61 /* position in the identifier. scw1 typically costs 1 indexed fetch, 62 /* an AND, and a jump per character of identifier, until the identifier 63 /* is known as a non-macro name or until the end of the identifier. 64 /* scw1 is inexpensive. scw2 typically costs 4 indexed fetches, 65 /* an add, an AND, and a jump per character of identifier, but it is also 66 /* slightly more effective at reducing symbol table searches. 67 /* scw2 usually costs too much because the symbol table search is 68 /* usually short; but if symbol table search should become expensive, 69 /* the code is here. 70 /* using both scw1 and scw2 is of dubious value. 71 */ 72 #define scw1 1 73 #define scw2 0 74 75 #if scw2 76 char t21[ALFSIZ],t22[ALFSIZ],t23[ALFSIZ+NCPS]; 77 #endif 78 79 #if scw1 80 #define b0 1 81 #define b1 2 82 #define b2 4 83 #define b3 8 84 #define b4 16 85 #define b5 32 86 #define b6 64 87 #define b7 128 88 #endif 89 90 #define IB 1 91 #define SB 2 92 #define NB 4 93 #define CB 8 94 #define QB 16 95 #define WB 32 96 char fastab[ALFSIZ]; 97 char slotab[ALFSIZ]; 98 char *ptrtab; 99 #define isslo (ptrtab==(slotab+COFF)) 100 #define isid(a) ((fastab+COFF)[a]&IB) 101 #define isspc(a) (ptrtab[a]&SB) 102 #define isnum(a) ((fastab+COFF)[a]&NB) 103 #define iscom(a) ((fastab+COFF)[a]&CB) 104 #define isquo(a) ((fastab+COFF)[a]&QB) 105 #define iswarn(a) ((fastab+COFF)[a]&WB) 106 107 #define eob(a) ((a)>=pend) 108 #define bob(a) (pbeg>=(a)) 109 110 # define cputc(a,b) if(!flslvl) putc(a,b) 111 112 char buffer[NCPS+BUFSIZ+BUFSIZ+NCPS]; 113 114 # define SBSIZE 60000 /* std = 12000, wnj aug 1979 */ 115 char sbf[SBSIZE]; 116 char *savch = sbf; 117 118 # define DROP 0xFE /* special character not legal ASCII or EBCDIC */ 119 # define WARN DROP 120 # define SAME 0 121 # define MAXINC 10 122 # define MAXFRE 14 /* max buffers of macro pushback */ 123 # define MAXFRM 31 /* max number of formals/actuals to a macro */ 124 125 static char warnc = WARN; 126 127 int mactop,fretop; 128 char *instack[MAXFRE],*bufstack[MAXFRE],*endbuf[MAXFRE]; 129 130 int plvl; /* parenthesis level during scan for macro actuals */ 131 int maclin; /* line number of macro call requiring actuals */ 132 char *macfil; /* file name of macro call requiring actuals */ 133 char *macnam; /* name of macro requiring actuals */ 134 int maclvl; /* # calls since last decrease in nesting level */ 135 char *macforw; /* pointer which must be exceeded to decrease nesting level */ 136 int macdam; /* offset to macforw due to buffer shifting */ 137 138 #if tgp 139 int tgpscan; /* flag for dump(); */ 140 #endif 141 142 STATIC int inctop[MAXINC]; 143 STATIC char *fnames[MAXINC]; 144 STATIC char *dirnams[MAXINC]; /* actual directory of #include files */ 145 STATIC int fins[MAXINC]; 146 STATIC int lineno[MAXINC]; 147 148 STATIC char *dirs[10]; /* -I and <> directories */ 149 char *strdex(), *copy(), *subst(), *trmdir(); 150 struct symtab *stsym(); 151 STATIC int fin = STDIN; 152 STATIC FILE *fout = stdout; 153 STATIC int nd = 1; 154 STATIC int pflag; /* don't put out lines "# 12 foo.c" */ 155 int passcom; /* don't delete comments */ 156 STATIC int rflag; /* allow macro recursion */ 157 STATIC int mflag; /* generate makefile dependencies */ 158 STATIC char *infile; /* name of .o file to build dependencies from */ 159 STATIC FILE *mout; /* file to place dependencies on */ 160 #define START 1 161 #define CONT 2 162 #define BACK 3 163 STATIC int ifno; 164 # define NPREDEF 20 165 STATIC char *prespc[NPREDEF]; 166 STATIC char **predef = prespc; 167 STATIC char *punspc[NPREDEF]; 168 STATIC char **prund = punspc; 169 STATIC int exfail; 170 struct symtab { 171 char *name; 172 char *value; 173 } *lastsym, *lookup(), *slookup(); 174 175 # if gcos 176 #include <setjmp.h> 177 static jmp_buf env; 178 # define main mainpp 179 # undef exit 180 # define exit(S) longjmp(env, 1) 181 # define open(S,D) fileno(fopen(S, "r")) 182 # define close(F) fclose(_f[F]) 183 extern FILE *_f[]; 184 # define symsiz 500 185 # else 186 # define symsiz 2000 /* std = 500, wnj aug 1979 */ 187 # endif 188 STATIC struct symtab stab[symsiz]; 189 190 STATIC struct symtab *defloc; 191 STATIC struct symtab *udfloc; 192 STATIC struct symtab *incloc; 193 STATIC struct symtab *ifloc; 194 STATIC struct symtab *elsloc; 195 STATIC struct symtab *eifloc; 196 STATIC struct symtab *ifdloc; 197 STATIC struct symtab *ifnloc; 198 STATIC struct symtab *ysysloc; 199 STATIC struct symtab *varloc; 200 STATIC struct symtab *lneloc; 201 STATIC struct symtab *ulnloc; 202 STATIC struct symtab *uflloc; 203 STATIC int trulvl; 204 STATIC int flslvl; 205 206 sayline(where) 207 int where; 208 { 209 if (mflag && where==START) fprintf(mout, "%s: %s\n", infile, fnames[ifno]); 210 if (pflag==0) fprintf(fout,"# %d \"%s\"\n", lineno[ifno], fnames[ifno]); 211 } 212 213 /* data structure guide 214 /* 215 /* most of the scanning takes place in the buffer: 216 /* 217 /* (low address) (high address) 218 /* pbeg pbuf pend 219 /* | <-- BUFSIZ chars --> | <-- BUFSIZ chars --> | 220 /* _______________________________________________________________________ 221 /* |_______________________________________________________________________| 222 /* | | | 223 /* |<-- waiting -->| |<-- waiting --> 224 /* | to be |<-- current -->| to be 225 /* | written | token | scanned 226 /* | | | 227 /* outp inp p 228 /* 229 /* *outp first char not yet written to output file 230 /* *inp first char of current token 231 /* *p first char not yet scanned 232 /* 233 /* macro expansion: write from *outp to *inp (chars waiting to be written), 234 /* ignore from *inp to *p (chars of the macro call), place generated 235 /* characters in front of *p (in reverse order), update pointers, 236 /* resume scanning. 237 /* 238 /* symbol table pointers point to just beyond the end of macro definitions; 239 /* the first preceding character is the number of formal parameters. 240 /* the appearance of a formal in the body of a definition is marked by 241 /* 2 chars: the char WARN, and a char containing the parameter number. 242 /* the first char of a definition is preceded by a zero character. 243 /* 244 /* when macro expansion attempts to back up over the beginning of the 245 /* buffer, some characters preceding *pend are saved in a side buffer, 246 /* the address of the side buffer is put on 'instack', and the rest 247 /* of the main buffer is moved to the right. the end of the saved buffer 248 /* is kept in 'endbuf' since there may be nulls in the saved buffer. 249 /* 250 /* similar action is taken when an 'include' statement is processed, 251 /* except that the main buffer must be completely emptied. the array 252 /* element 'inctop[ifno]' records the last side buffer saved when 253 /* file 'ifno' was included. these buffers remain dormant while 254 /* the file is being read, and are reactivated at end-of-file. 255 /* 256 /* instack[0 : mactop] holds the addresses of all pending side buffers. 257 /* instack[inctop[ifno]+1 : mactop-1] holds the addresses of the side 258 /* buffers which are "live"; the side buffers instack[0 : inctop[ifno]] 259 /* are dormant, waiting for end-of-file on the current file. 260 /* 261 /* space for side buffers is obtained from 'savch' and is never returned. 262 /* bufstack[0:fretop-1] holds addresses of side buffers which 263 /* are available for use. 264 */ 265 266 dump() { 267 /* write part of buffer which lies between outp and inp . 268 /* this should be a direct call to 'write', but the system slows to a crawl 269 /* if it has to do an unaligned copy. thus we buffer. this silly loop 270 /* is 15% of the total time, thus even the 'putc' macro is too slow. 271 */ 272 register char *p1,*p2; register FILE *f; 273 if ((p1=outp)==inp || flslvl!=0) return; 274 #if tgp 275 #define MAXOUT 80 276 if (!tgpscan) {/* scan again to insure <= MAXOUT chars between linefeeds */ 277 register char c,*pblank; char savc,stopc,brk; 278 tgpscan=1; brk=stopc=pblank=0; p2=inp; savc= *p2; *p2='\0'; 279 while (c= *p1++) { 280 if (c=='\\') c= *p1++; 281 if (stopc==c) stopc=0; 282 else if (c=='"' || c=='\'') stopc=c; 283 if (p1-outp>MAXOUT && pblank!=0) { 284 *pblank++='\n'; inp=pblank; dump(); brk=1; pblank=0; 285 } 286 if (c==' ' && stopc==0) pblank=p1-1; 287 } 288 if (brk) sayline(CONT); 289 *p2=savc; inp=p2; p1=outp; tgpscan=0; 290 } 291 #endif 292 f=fout; 293 # if gcos 294 /* filter out "$ program c" card if first line of input */ 295 /* gmatch is a simple pattern matcher in the GCOS Standard Library */ 296 { static int gmfirst = 0; 297 if (!gmfirst) { 298 ++gmfirst; 299 if (gmatch(p1, "^$*program[ \t]*c*")) 300 p1 = strdex(p1, '\n'); 301 } 302 } 303 # endif 304 while (p1<inp) putc(*p1++,f); 305 outp=p1; 306 } 307 308 char * 309 refill(p) register char *p; { 310 /* dump buffer. save chars from inp to p. read into buffer at pbuf, 311 /* contiguous with p. update pointers, return new p. 312 */ 313 register char *np,*op; register int ninbuf; 314 dump(); np=pbuf-(p-inp); op=inp; 315 if (bob(np+1)) {pperror("token too long"); np=pbeg; p=inp+BUFSIZ;} 316 macdam += np-inp; outp=inp=np; 317 while (op<p) *np++= *op++; 318 p=np; 319 for (;;) { 320 if (mactop>inctop[ifno]) {/* retrieve hunk of pushed-back macro text */ 321 op=instack[--mactop]; np=pbuf; 322 do {while (*np++= *op++);} while (op<endbuf[mactop]); pend=np-1; 323 /* make buffer space avail for 'include' processing */ 324 if (fretop<MAXFRE) bufstack[fretop++]=instack[mactop]; 325 return(p); 326 } else {/* get more text from file(s) */ 327 maclvl=0; 328 if (0<(ninbuf=read(fin,pbuf,BUFSIZ))) { 329 pend=pbuf+ninbuf; *pend='\0'; 330 return(p); 331 } 332 /* end of #include file */ 333 if (ifno==0) {/* end of input */ 334 if (plvl!=0) { 335 int n=plvl,tlin=lineno[ifno]; char *tfil=fnames[ifno]; 336 lineno[ifno]=maclin; fnames[ifno]=macfil; 337 pperror("%s: unterminated macro call",macnam); 338 lineno[ifno]=tlin; fnames[ifno]=tfil; 339 np=p; *np++='\n'; /* shut off unterminated quoted string */ 340 while (--n>=0) *np++=')'; /* supply missing parens */ 341 pend=np; *np='\0'; if (plvl<0) plvl=0; 342 return(p); 343 } 344 if (trulvl || flslvl) 345 pperror("missing endif"); 346 inp=p; dump(); exit(exfail); 347 } 348 close(fin); fin=fins[--ifno]; dirs[0]=dirnams[ifno]; sayline(BACK); 349 } 350 } 351 } 352 353 #define BEG 0 354 #define LF 1 355 356 char * 357 cotoken(p) register char *p; { 358 register int c,i; char quoc; 359 static int state = BEG; 360 361 if (state!=BEG) goto prevlf; 362 for (;;) { 363 again: 364 while (!isspc(*p++)); 365 switch (*(inp=p-1)) { 366 case 0: { 367 if (eob(--p)) {p=refill(p); goto again;} 368 else ++p; /* ignore null byte */ 369 } break; 370 case '|': case '&': for (;;) {/* sloscan only */ 371 if (*p++== *inp) break; 372 if (eob(--p)) p=refill(p); 373 else break; 374 } break; 375 case '=': case '!': for (;;) {/* sloscan only */ 376 if (*p++=='=') break; 377 if (eob(--p)) p=refill(p); 378 else break; 379 } break; 380 case '<': case '>': for (;;) {/* sloscan only */ 381 if (*p++=='=' || p[-2]==p[-1]) break; 382 if (eob(--p)) p=refill(p); 383 else break; 384 } break; 385 case '\\': for (;;) { 386 if (*p++=='\n') {++lineno[ifno]; break;} 387 if (eob(--p)) p=refill(p); 388 else {++p; break;} 389 } break; 390 case '/': for (;;) { 391 if (*p++=='*') {/* comment */ 392 if (!passcom) {inp=p-2; dump(); ++flslvl;} 393 for (;;) { 394 while (!iscom(*p++)); 395 if (p[-1]=='*') for (;;) { 396 if (*p++=='/') goto endcom; 397 if (eob(--p)) { 398 if (!passcom) {inp=p; p=refill(p);} 399 else if ((p-inp)>=BUFSIZ) {/* split long comment */ 400 inp=p; p=refill(p); /* last char written is '*' */ 401 cputc('/',fout); /* terminate first part */ 402 /* and fake start of 2nd */ 403 outp=inp=p-=3; *p++='/'; *p++='*'; *p++='*'; 404 } else p=refill(p); 405 } else break; 406 } else if (p[-1]=='\n') { 407 ++lineno[ifno]; if (!passcom) putc('\n',fout); 408 } else if (eob(--p)) { 409 if (!passcom) {inp=p; p=refill(p);} 410 else if ((p-inp)>=BUFSIZ) {/* split long comment */ 411 inp=p; p=refill(p); 412 cputc('*',fout); cputc('/',fout); 413 outp=inp=p-=2; *p++='/'; *p++='*'; 414 } else p=refill(p); 415 } else ++p; /* ignore null byte */ 416 } 417 endcom: 418 if (!passcom) {outp=inp=p; --flslvl; goto again;} 419 break; 420 } 421 if (eob(--p)) p=refill(p); 422 else break; 423 } break; 424 # if gcos 425 case '`': 426 # endif 427 case '"': case '\'': { 428 quoc=p[-1]; 429 for (;;) { 430 while (!isquo(*p++)); 431 if (p[-1]==quoc) break; 432 if (p[-1]=='\n') {--p; break;} /* bare \n terminates quotation */ 433 if (p[-1]=='\\') for (;;) { 434 if (*p++=='\n') {++lineno[ifno]; break;} /* escaped \n ignored */ 435 if (eob(--p)) p=refill(p); 436 else {++p; break;} 437 } else if (eob(--p)) p=refill(p); 438 else ++p; /* it was a different quote character */ 439 } 440 } break; 441 case '\n': { 442 ++lineno[ifno]; if (isslo) {state=LF; return(p);} 443 prevlf: 444 state=BEG; 445 for (;;) { 446 if (*p++=='#') return(p); 447 if (eob(inp= --p)) p=refill(p); 448 else goto again; 449 } 450 } break; 451 case '0': case '1': case '2': case '3': case '4': 452 case '5': case '6': case '7': case '8': case '9': 453 for (;;) { 454 while (isnum(*p++)); 455 if (eob(--p)) p=refill(p); 456 else break; 457 } break; 458 case 'A': case 'B': case 'C': case 'D': case 'E': 459 case 'F': case 'G': case 'H': case 'I': case 'J': 460 case 'K': case 'L': case 'M': case 'N': case 'O': 461 case 'P': case 'Q': case 'R': case 'S': case 'T': 462 case 'U': case 'V': case 'W': case 'X': case 'Y': 463 case 'Z': case '_': 464 case 'a': case 'b': case 'c': case 'd': case 'e': 465 case 'f': case 'g': case 'h': case 'i': case 'j': 466 case 'k': case 'l': case 'm': case 'n': case 'o': 467 case 'p': case 'q': case 'r': case 's': case 't': 468 case 'u': case 'v': case 'w': case 'x': case 'y': 469 case 'z': 470 #if scw1 471 #define tmac1(c,bit) if (!xmac1(c,bit,&)) goto nomac 472 #define xmac1(c,bit,op) ((macbit+COFF)[c] op (bit)) 473 #else 474 #define tmac1(c,bit) 475 #define xmac1(c,bit,op) 476 #endif 477 478 #if scw2 479 #define tmac2(c0,c1,cpos) if (!xmac2(c0,c1,cpos,&)) goto nomac 480 #define xmac2(c0,c1,cpos,op)\ 481 ((macbit+COFF)[(t21+COFF)[c0]+(t22+COFF)[c1]] op (t23+COFF+cpos)[c0]) 482 #else 483 #define tmac2(c0,c1,cpos) 484 #define xmac2(c0,c1,cpos,op) 485 #endif 486 487 if (flslvl) goto nomac; 488 for (;;) { 489 c= p[-1]; tmac1(c,b0); 490 i= *p++; if (!isid(i)) goto endid; tmac1(i,b1); tmac2(c,i,0); 491 c= *p++; if (!isid(c)) goto endid; tmac1(c,b2); tmac2(i,c,1); 492 i= *p++; if (!isid(i)) goto endid; tmac1(i,b3); tmac2(c,i,2); 493 c= *p++; if (!isid(c)) goto endid; tmac1(c,b4); tmac2(i,c,3); 494 i= *p++; if (!isid(i)) goto endid; tmac1(i,b5); tmac2(c,i,4); 495 c= *p++; if (!isid(c)) goto endid; tmac1(c,b6); tmac2(i,c,5); 496 i= *p++; if (!isid(i)) goto endid; tmac1(i,b7); tmac2(c,i,6); 497 tmac2(i,0,7); 498 while (isid(*p++)); 499 if (eob(--p)) {refill(p); p=inp+1; continue;} 500 goto lokid; 501 endid: 502 if (eob(--p)) {refill(p); p=inp+1; continue;} 503 tmac2(p[-1],0,-1+(p-inp)); 504 lokid: 505 slookup(inp,p,0); if (newp) {p=newp; goto again;} 506 else break; 507 nomac: 508 while (isid(*p++)); 509 if (eob(--p)) {p=refill(p); goto nomac;} 510 else break; 511 } break; 512 } /* end of switch */ 513 514 if (isslo) return(p); 515 } /* end of infinite loop */ 516 } 517 518 char * 519 skipbl(p) register char *p; {/* get next non-blank token */ 520 do {outp=inp=p; p=cotoken(p);} while ((toktyp+COFF)[*inp]==BLANK); 521 return(p); 522 } 523 524 char * 525 unfill(p) register char *p; { 526 /* take <= BUFSIZ chars from right end of buffer and put them on instack . 527 /* slide rest of buffer to the right, update pointers, return new p. 528 */ 529 register char *np,*op; register int d; 530 if (mactop>=MAXFRE) { 531 pperror("%s: too much pushback",macnam); 532 p=inp=pend; dump(); /* begin flushing pushback */ 533 while (mactop>inctop[ifno]) {p=refill(p); p=inp=pend; dump();} 534 } 535 if (fretop>0) np=bufstack[--fretop]; 536 else { 537 np=savch; savch+=BUFSIZ; 538 if (savch>=sbf+SBSIZE) {pperror("no space"); exit(exfail);} 539 *savch++='\0'; 540 } 541 instack[mactop]=np; op=pend-BUFSIZ; if (op<p) op=p; 542 for (;;) {while (*np++= *op++); if (eob(op)) break;} /* out with old */ 543 endbuf[mactop++]=np; /* mark end of saved text */ 544 np=pbuf+BUFSIZ; op=pend-BUFSIZ; pend=np; if (op<p) op=p; 545 while (outp<op) *--np= *--op; /* slide over new */ 546 if (bob(np)) pperror("token too long"); 547 d=np-outp; outp+=d; inp+=d; macdam+=d; return(p+d); 548 } 549 550 char * 551 doincl(p) register char *p; { 552 int filok,inctype; 553 register char *cp; char **dirp,*nfil; char filname[BUFSIZ]; 554 555 p=skipbl(p); cp=filname; 556 if (*inp++=='<') {/* special <> syntax */ 557 inctype=1; 558 ++flslvl; /* prevent macro expansion */ 559 for (;;) { 560 outp=inp=p; p=cotoken(p); 561 if (*inp=='\n') {--p; *cp='\0'; break;} 562 if (*inp=='>') { *cp='\0'; break;} 563 # ifdef gimpel 564 if (*inp=='.' && !intss()) *inp='#'; 565 # endif 566 while (inp<p) *cp++= *inp++; 567 } 568 --flslvl; /* reenable macro expansion */ 569 } else if (inp[-1]=='"') {/* regular "" syntax */ 570 inctype=0; 571 # ifdef gimpel 572 while (inp<p) {if (*inp=='.' && !intss()) *inp='#'; *cp++= *inp++;} 573 # else 574 while (inp<p) *cp++= *inp++; 575 # endif 576 if (*--cp=='"') *cp='\0'; 577 } else {pperror("bad include syntax",0); inctype=2;} 578 /* flush current file to \n , then write \n */ 579 ++flslvl; do {outp=inp=p; p=cotoken(p);} while (*inp!='\n'); --flslvl; 580 inp=p; dump(); if (inctype==2) return(p); 581 /* look for included file */ 582 if (ifno+1 >=MAXINC) { 583 pperror("Unreasonable include nesting",0); return(p); 584 } 585 if((nfil=savch)>sbf+SBSIZE-BUFSIZ) {pperror("no space"); exit(exfail);} 586 filok=0; 587 for (dirp=dirs+inctype; *dirp; ++dirp) { 588 if ( 589 # if gcos 590 strdex(filname, '/') 591 # else 592 filname[0]=='/' 593 # endif 594 || **dirp=='\0') strcpy(nfil,filname); 595 else { 596 strcpy(nfil,*dirp); 597 # if unix || gcos 598 strcat(nfil,"/"); 599 # endif 600 #ifdef ibm 601 #ifndef gimpel 602 strcat(nfil,"."); 603 #endif 604 #endif 605 strcat(nfil,filname); 606 } 607 if (0<(fins[ifno+1]=open(nfil,READ))) { 608 filok=1; fin=fins[++ifno]; break; 609 } 610 } 611 if (filok==0) pperror("Can't find include file %s",filname); 612 else { 613 lineno[ifno]=1; fnames[ifno]=cp=nfil; while (*cp++); savch=cp; 614 dirnams[ifno]=dirs[0]=trmdir(copy(nfil)); 615 sayline(START); 616 /* save current contents of buffer */ 617 while (!eob(p)) p=unfill(p); 618 inctop[ifno]=mactop; 619 } 620 return(p); 621 } 622 623 equfrm(a,p1,p2) register char *a,*p1,*p2; { 624 register char c; int flag; 625 c= *p2; *p2='\0'; 626 flag=strcmp(a,p1); *p2=c; return(flag==SAME); 627 } 628 629 char * 630 dodef(p) char *p; {/* process '#define' */ 631 register char *pin,*psav,*cf; 632 char **pf,**qf; int b,c,params; struct symtab *np; 633 char *oldval,*oldsavch; 634 char *formal[MAXFRM]; /* formal[n] is name of nth formal */ 635 char formtxt[BUFSIZ]; /* space for formal names */ 636 637 if (savch>sbf+SBSIZE-BUFSIZ) {pperror("too much defining"); return(p);} 638 oldsavch=savch; /* to reclaim space if redefinition */ 639 ++flslvl; /* prevent macro expansion during 'define' */ 640 p=skipbl(p); pin=inp; 641 if ((toktyp+COFF)[*pin]!=IDENT) { 642 ppwarn("illegal macro name"); while (*inp!='\n') p=skipbl(p); return(p); 643 } 644 np=slookup(pin,p,1); 645 if (oldval=np->value) savch=oldsavch; /* was previously defined */ 646 b=1; cf=pin; 647 while (cf<p) {/* update macbit */ 648 c= *cf++; xmac1(c,b,|=); b=(b+b)&0xFF; 649 if (cf!=p) xmac2(c,*cf,-1+(cf-pin),|=); 650 else xmac2(c,0,-1+(cf-pin),|=); 651 } 652 params=0; outp=inp=p; p=cotoken(p); pin=inp; 653 if (*pin=='(') {/* with parameters; identify the formals */ 654 cf=formtxt; pf=formal; 655 for (;;) { 656 p=skipbl(p); pin=inp; 657 if (*pin=='\n') { 658 --lineno[ifno]; --p; pperror("%s: missing )",np->name); break; 659 } 660 if (*pin==')') break; 661 if (*pin==',') continue; 662 if ((toktyp+COFF)[*pin]!=IDENT) { 663 c= *p; *p='\0'; pperror("bad formal: %s",pin); *p=c; 664 } else if (pf>= &formal[MAXFRM]) { 665 c= *p; *p='\0'; pperror("too many formals: %s",pin); *p=c; 666 } else { 667 *pf++=cf; while (pin<p) *cf++= *pin++; *cf++='\0'; ++params; 668 } 669 } 670 if (params==0) --params; /* #define foo() ... */ 671 } else if (*pin=='\n') {--lineno[ifno]; --p;} 672 /* remember beginning of macro body, so that we can 673 /* warn if a redefinition is different from old value. 674 */ 675 oldsavch=psav=savch; 676 for (;;) {/* accumulate definition until linefeed */ 677 outp=inp=p; p=cotoken(p); pin=inp; 678 if (*pin=='\\' && pin[1]=='\n') {putc('\n',fout); continue;} /* ignore escaped lf */ 679 if (*pin=='\n') break; 680 if (params) {/* mark the appearance of formals in the definiton */ 681 if ((toktyp+COFF)[*pin]==IDENT) { 682 for (qf=pf; --qf>=formal; ) { 683 if (equfrm(*qf,pin,p)) { 684 *psav++=qf-formal+1; *psav++=WARN; pin=p; break; 685 } 686 } 687 } else if (*pin=='"' || *pin=='\'' 688 # if gcos 689 || *pin=='`' 690 # endif 691 ) {/* inside quotation marks, too */ 692 char quoc= *pin; 693 for (*psav++= *pin++; pin<p && *pin!=quoc; ) { 694 while (pin<p && !isid(*pin)) *psav++= *pin++; 695 cf=pin; while (cf<p && isid(*cf)) ++cf; 696 for (qf=pf; --qf>=formal; ) { 697 if (equfrm(*qf,pin,cf)) { 698 *psav++=qf-formal+1; *psav++=WARN; pin=cf; break; 699 } 700 } 701 while (pin<cf) *psav++= *pin++; 702 } 703 } 704 } 705 while (pin<p) *psav++= *pin++; 706 } 707 *psav++=params; *psav++='\0'; 708 if ((cf=oldval)!=NULL) {/* redefinition */ 709 --cf; /* skip no. of params, which may be zero */ 710 while (*--cf); /* go back to the beginning */ 711 if (0!=strcmp(++cf,oldsavch)) {/* redefinition different from old */ 712 --lineno[ifno]; ppwarn("%s redefined",np->name); ++lineno[ifno]; 713 np->value=psav-1; 714 } else psav=oldsavch; /* identical redef.; reclaim space */ 715 } else np->value=psav-1; 716 --flslvl; inp=pin; savch=psav; return(p); 717 } 718 719 #define fasscan() ptrtab=fastab+COFF 720 #define sloscan() ptrtab=slotab+COFF 721 722 char * 723 control(p) register char *p; {/* find and handle preprocessor control lines */ 724 register struct symtab *np; 725 for (;;) { 726 fasscan(); p=cotoken(p); if (*inp=='\n') ++inp; dump(); 727 sloscan(); p=skipbl(p); 728 *--inp=SALT; outp=inp; ++flslvl; np=slookup(inp,p,0); --flslvl; 729 if (np==defloc) {/* define */ 730 if (flslvl==0) {p=dodef(p); continue;} 731 } else if (np==incloc) {/* include */ 732 if (flslvl==0) {p=doincl(p); continue;} 733 } else if (np==ifnloc) {/* ifndef */ 734 ++flslvl; p=skipbl(p); np=slookup(inp,p,0); --flslvl; 735 if (flslvl==0 && np->value==0) ++trulvl; 736 else ++flslvl; 737 } else if (np==ifdloc) {/* ifdef */ 738 ++flslvl; p=skipbl(p); np=slookup(inp,p,0); --flslvl; 739 if (flslvl==0 && np->value!=0) ++trulvl; 740 else ++flslvl; 741 } else if (np==eifloc) {/* endif */ 742 if (flslvl) {if (--flslvl==0) sayline(CONT);} 743 else if (trulvl) --trulvl; 744 else pperror("If-less endif",0); 745 } else if (np==elsloc) {/* else */ 746 if (flslvl) { 747 if (--flslvl!=0) ++flslvl; 748 else {++trulvl; sayline(CONT);} 749 } 750 else if (trulvl) {++flslvl; --trulvl;} 751 else pperror("If-less else",0); 752 } else if (np==udfloc) {/* undefine */ 753 if (flslvl==0) { 754 ++flslvl; p=skipbl(p); slookup(inp,p,DROP); --flslvl; 755 } 756 } else if (np==ifloc) {/* if */ 757 #if tgp 758 pperror(" IF not implemented, true assumed", 0); 759 if (flslvl==0) ++trulvl; else ++flslvl; 760 #else 761 newp=p; 762 if (flslvl==0 && yyparse()) ++trulvl; else ++flslvl; 763 p=newp; 764 #endif 765 } else if (np==lneloc) {/* line */ 766 if (flslvl==0 && pflag==0) { 767 char *cp, *cp2, *savestring(); 768 outp=inp=p; *--outp='#'; while (*inp!='\n') p=cotoken(p); 769 cp = outp + 1; 770 while (isspace(*cp) && cp < inp) 771 cp++; 772 while (isdigit(*cp) && cp < inp) 773 cp++; 774 while (*cp != '"' && cp < inp) 775 cp++; 776 if (cp < inp) { 777 cp++; 778 cp2 = cp; 779 while (*cp2 != '"' && cp2 < inp) 780 cp2++; 781 fnames[ifno] = savestring(cp, cp2); 782 } 783 continue; 784 } 785 } else if (*++inp=='\n') outp=inp; /* allows blank line after # */ 786 else pperror("undefined control",0); 787 /* flush to lf */ 788 ++flslvl; while (*inp!='\n') {outp=inp=p; p=cotoken(p);} --flslvl; 789 } 790 } 791 792 char * 793 savestring(start, finish) 794 register char *start, *finish; 795 { 796 char *retbuf; 797 register char *cp; 798 799 retbuf = (char *) calloc(finish - start + 1, sizeof (char)); 800 cp = retbuf; 801 while (start < finish) 802 *cp++ = *start++; 803 *cp = 0; 804 return(retbuf); 805 } 806 807 struct symtab * 808 stsym(s) register char *s; { 809 char buf[BUFSIZ]; register char *p; 810 811 /* make definition look exactly like end of #define line */ 812 /* copy to avoid running off end of world when param list is at end */ 813 p=buf; while (*p++= *s++); 814 p=buf; while (isid(*p++)); /* skip first identifier */ 815 if (*--p=='=') {*p++=' '; while (*p++);} 816 else {s=" 1"; while (*p++= *s++);} 817 pend=p; *--p='\n'; 818 sloscan(); dodef(buf); return(lastsym); 819 } 820 821 struct symtab * 822 ppsym(s) char *s; {/* kluge */ 823 register struct symtab *sp; 824 cinit=SALT; *savch++=SALT; sp=stsym(s); --sp->name; cinit=0; return(sp); 825 } 826 827 /* VARARGS1 */ 828 pperror(s,x,y) char *s; { 829 if (fnames[ifno][0]) fprintf(stderr, 830 # if gcos 831 "*%c* \"%s\", line ", exfail >= 0 ? 'F' : 'W', 832 # else 833 "%s: ", 834 # endif 835 fnames[ifno]); 836 fprintf(stderr, "%d: ",lineno[ifno]); 837 fprintf(stderr, s, x, y); 838 fprintf(stderr,"\n"); 839 ++exfail; 840 } 841 842 yyerror(s,a,b) char *s; { 843 pperror(s,a,b); 844 } 845 846 ppwarn(s,x) char *s; { 847 int fail = exfail; 848 exfail = -1; 849 pperror(s,x); 850 exfail = fail; 851 } 852 853 struct symtab * 854 lookup(namep, enterf) 855 char *namep; 856 { 857 register char *np, *snp; 858 register int c, i; int around; 859 register struct symtab *sp; 860 861 /* namep had better not be too long (currently, <=NCPS chars) */ 862 np=namep; around=0; i=cinit; 863 while (c= *np++) i += i+c; c=i; /* c=i for register usage on pdp11 */ 864 c %= symsiz; if (c<0) c += symsiz; 865 sp = &stab[c]; 866 while (snp=sp->name) { 867 np = namep; 868 while (*snp++ == *np) if (*np++ == '\0') { 869 if (enterf==DROP) {sp->name[0]= DROP; sp->value=0;} 870 return(lastsym=sp); 871 } 872 if (--sp < &stab[0]) 873 if (around) {pperror("too many defines", 0); exit(exfail);} 874 else {++around; sp = &stab[symsiz-1];} 875 } 876 if (enterf==1) sp->name=namep; 877 return(lastsym=sp); 878 } 879 880 struct symtab * 881 slookup(p1,p2,enterf) register char *p1,*p2; int enterf;{ 882 register char *p3; char c2,c3; struct symtab *np; 883 c2= *p2; *p2='\0'; /* mark end of token */ 884 if ((p2-p1)>NCPS) p3=p1+NCPS; else p3=p2; 885 c3= *p3; *p3='\0'; /* truncate to NCPS chars or less */ 886 if (enterf==1) p1=copy(p1); 887 np=lookup(p1,enterf); *p3=c3; *p2=c2; 888 if (np->value!=0 && flslvl==0) newp=subst(p2,np); 889 else newp=0; 890 return(np); 891 } 892 893 char * 894 subst(p,sp) register char *p; struct symtab *sp; { 895 static char match[]="%s: argument mismatch"; 896 register char *ca,*vp; int params; 897 char *actual[MAXFRM]; /* actual[n] is text of nth actual */ 898 char actused[MAXFRM]; /* for newline processing in actuals */ 899 char acttxt[BUFSIZ]; /* space for actuals */ 900 int nlines = 0; 901 902 if (0==(vp=sp->value)) return(p); 903 if ((p-macforw)<=macdam) { 904 if (++maclvl>symsiz && !rflag) { 905 pperror("%s: macro recursion",sp->name); return(p); 906 } 907 } else maclvl=0; /* level decreased */ 908 macforw=p; macdam=0; /* new target for decrease in level */ 909 macnam=sp->name; 910 dump(); 911 if (sp==ulnloc) { 912 vp=acttxt; *vp++='\0'; 913 sprintf(vp,"%d",lineno[ifno]); while (*vp++); 914 } else if (sp==uflloc) { 915 vp=acttxt; *vp++='\0'; 916 sprintf(vp,"\"%s\"",fnames[ifno]); while (*vp++); 917 } 918 if (0!=(params= *--vp&0xFF)) {/* definition calls for params */ 919 register char **pa; 920 ca=acttxt; pa=actual; 921 if (params==0xFF) params=1; /* #define foo() ... */ 922 sloscan(); ++flslvl; /* no expansion during search for actuals */ 923 plvl= -1; 924 do p=skipbl(p); while (*inp=='\n'); /* skip \n too */ 925 if (*inp=='(') { 926 maclin=lineno[ifno]; macfil=fnames[ifno]; 927 for (plvl=1; plvl!=0; ) { 928 *ca++='\0'; 929 for (;;) { 930 outp=inp=p; p=cotoken(p); 931 if (*inp=='(') ++plvl; 932 if (*inp==')' && --plvl==0) {--params; break;} 933 if (plvl==1 && *inp==',') {--params; break;} 934 while (inp<p) *ca++= *inp++; 935 if (ca> &acttxt[BUFSIZ]) 936 pperror("%s: actuals too long",sp->name); 937 } 938 if (pa>= &actual[MAXFRM]) ppwarn(match,sp->name); 939 else { actused[pa-actual]=0; *pa++=ca; } 940 } 941 nlines = lineno[ifno] - maclin; 942 lineno[ifno] = maclin; /* don't count newlines here */ 943 } 944 if (params!=0) ppwarn(match,sp->name); 945 while (--params>=0) *pa++=""+1; /* null string for missing actuals */ 946 --flslvl; fasscan(); 947 } 948 for (;;) {/* push definition onto front of input stack */ 949 while (!iswarn(*--vp)) { 950 if (bob(p)) {outp=inp=p; p=unfill(p);} 951 *--p= *vp; 952 } 953 if (*vp==warnc) {/* insert actual param */ 954 ca=actual[*--vp-1]; 955 while (*--ca) { 956 if (bob(p)) {outp=inp=p; p=unfill(p);} 957 /* Actuals with newlines confuse line numbering */ 958 if (*ca == '\n' && actused[*vp-1]) 959 if (*(ca-1) == '\\') ca--; 960 else *--p = ' '; 961 else { *--p= *ca; if (*ca == '\n') nlines--; } 962 } 963 actused[*vp-1] = 1; 964 } else { 965 if (nlines > 0 ) 966 while (nlines-- > 0) 967 *--p = '\n'; 968 break; 969 } 970 } 971 outp=inp=p; 972 return(p); 973 } 974 975 976 977 978 char * 979 trmdir(s) register char *s; { 980 register char *p = s; 981 while (*p++); --p; while (p>s && *--p!='/'); 982 # if unix 983 if (p==s) *p++='.'; 984 # endif 985 *p='\0'; 986 return(s); 987 } 988 989 STATIC char * 990 copy(s) register char *s; { 991 register char *old; 992 993 old = savch; while (*savch++ = *s++); 994 return(old); 995 } 996 997 char * 998 strdex(s,c) char *s,c; { 999 while (*s) if (*s++==c) return(--s); 1000 return(0); 1001 } 1002 1003 yywrap(){ return(1); } 1004 1005 main(argc,argv) 1006 char *argv[]; 1007 { 1008 register int i,c; 1009 register char *p; 1010 char *tf,**cp2; 1011 1012 # if gcos 1013 if (setjmp(env)) return (exfail); 1014 # endif 1015 p="_$ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789"; 1016 i=0; 1017 while (c= *p++) { 1018 (fastab+COFF)[c] |= IB|NB|SB; (toktyp+COFF)[c]=IDENT; 1019 #if scw2 1020 /* 53 == 63-10; digits rarely appear in identifiers, 1021 /* and can never be the first char of an identifier. 1022 /* 11 == 53*53/sizeof(macbit) . 1023 */ 1024 ++i; (t21+COFF)[c]=(53*i)/11; (t22+COFF)[c]=i%11; 1025 #endif 1026 } 1027 p="0123456789."; 1028 while (c= *p++) {(fastab+COFF)[c] |= NB|SB; (toktyp+COFF)[c]=NUMBR;} 1029 # if gcos 1030 p="\n\"'`/\\"; 1031 # else 1032 p="\n\"'/\\"; 1033 # endif 1034 while (c= *p++) (fastab+COFF)[c] |= SB; 1035 # if gcos 1036 p="\n\"'`\\"; 1037 # else 1038 p="\n\"'\\"; 1039 # endif 1040 while (c= *p++) (fastab+COFF)[c] |= QB; 1041 p="*\n"; while (c= *p++) (fastab+COFF)[c] |= CB; 1042 (fastab+COFF)[warnc] |= WB; 1043 (fastab+COFF)['\0'] |= CB|QB|SB|WB; 1044 for (i=ALFSIZ; --i>=0; ) slotab[i]=fastab[i]|SB; 1045 p=" \t\013\f\r"; /* note no \n; \v not legal for vertical tab? */ 1046 while (c= *p++) (toktyp+COFF)[c]=BLANK; 1047 #if scw2 1048 for ((t23+COFF)[i=ALFSIZ+7-COFF]=1; --i>=-COFF; ) 1049 if (((t23+COFF)[i]=(t23+COFF+1)[i]<<1)==0) (t23+COFF)[i]=1; 1050 #endif 1051 1052 # if unix 1053 fnames[ifno=0] = ""; dirnams[0]=dirs[0]="."; 1054 # endif 1055 # if ibm 1056 fnames[ifno=0] = ""; 1057 # endif 1058 # if gcos 1059 if (inquire(stdin, _TTY)) freopen("*src", "rt", stdin); 1060 # endif 1061 # if gimpel || gcos 1062 fnames[ifno=0] = (char *)inquire(stdin, _FILENAME); 1063 dirnams[0] = dirs[0] = trmdir(copy(fnames[0])); 1064 # endif 1065 for(i=1; i<argc; i++) 1066 { 1067 switch(argv[i][0]) 1068 { 1069 case '-': 1070 # if gcos 1071 switch(toupper(argv[i][1])) { /* case-independent on GCOS */ 1072 # else 1073 switch(argv[i][1]) { 1074 # endif 1075 case 'M': mflag++; 1076 case 'P': pflag++; 1077 case 'E': continue; 1078 case 'R': ++rflag; continue; 1079 case 'C': passcom++; continue; 1080 case 'D': 1081 if (predef>prespc+NPREDEF) { 1082 pperror("too many -D options, ignoring %s",argv[i]); 1083 continue; 1084 } 1085 /* ignore plain "-D" (no argument) */ 1086 if (*(argv[i]+2)) *predef++ = argv[i]+2; 1087 continue; 1088 case 'U': 1089 if (prund>punspc+NPREDEF) { 1090 pperror("too many -U options, ignoring %s",argv[i]); 1091 continue; 1092 } 1093 *prund++ = argv[i]+2; 1094 continue; 1095 case 'I': 1096 if (nd>8) pperror("excessive -I file (%s) ignored",argv[i]); 1097 else dirs[nd++] = argv[i]+2; 1098 continue; 1099 case '\0': continue; 1100 default: 1101 pperror("unknown flag %s", argv[i]); 1102 continue; 1103 } 1104 default: 1105 if (fin==STDIN) { 1106 if (0>(fin=open(argv[i], READ))) { 1107 pperror("No source file %s",argv[i]); exit(8); 1108 } 1109 fnames[ifno]=copy(argv[i]); 1110 infile=copy(argv[i]); 1111 dirs[0]=dirnams[ifno]=trmdir(argv[i]); 1112 # ifndef gcos 1113 /* too dangerous to have file name in same syntactic position 1114 be input or output file depending on file redirections, 1115 so force output to stdout, willy-nilly 1116 [i don't see what the problem is. jfr] 1117 */ 1118 } else if (fout==stdout) { 1119 if (NULL==(fout=fopen(argv[i], "w"))) { 1120 pperror("Can't create %s", argv[i]); exit(8); 1121 } else fclose(stdout); 1122 # endif 1123 } else pperror("extraneous name %s", argv[i]); 1124 } 1125 } 1126 1127 if (mflag) { 1128 if (infile==(char *)0) { 1129 fprintf(stderr, 1130 "no input file specified with -M flag\n"); 1131 exit(8); 1132 } 1133 tf=(char *)rindex(infile, '.'); 1134 if (tf==0) { 1135 fprintf(stderr, "missing component name on %s\n", 1136 infile); 1137 exit(8); 1138 } 1139 tf[1]='o'; 1140 tf=(char *)rindex(infile, '/'); 1141 if (tf!=(char *)0) 1142 infile = tf + 1; 1143 mout=fout; 1144 if (NULL==(fout=fopen("/dev/null", "w"))) { 1145 pperror("Can't open /dev/null"); 1146 exit(8); 1147 } 1148 } 1149 fins[ifno]=fin; 1150 exfail = 0; 1151 /* after user -I files here are the standard include libraries */ 1152 # if unix 1153 dirs[nd++] = "/usr/include"; 1154 # endif 1155 # if gcos 1156 dirs[nd++] = "cc/include"; 1157 # endif 1158 # if ibm 1159 # ifndef gimpel 1160 dirs[nd++] = "BTL$CLIB"; 1161 # endif 1162 # endif 1163 # ifdef gimpel 1164 dirs[nd++] = intss() ? "SYS3.C." : "" ; 1165 # endif 1166 /* dirs[nd++] = "/compool"; */ 1167 dirs[nd++] = 0; 1168 defloc=ppsym("define"); 1169 udfloc=ppsym("undef"); 1170 incloc=ppsym("include"); 1171 elsloc=ppsym("else"); 1172 eifloc=ppsym("endif"); 1173 ifdloc=ppsym("ifdef"); 1174 ifnloc=ppsym("ifndef"); 1175 ifloc=ppsym("if"); 1176 lneloc=ppsym("line"); 1177 for (i=sizeof(macbit)/sizeof(macbit[0]); --i>=0; ) macbit[i]=0; 1178 # if unix 1179 ysysloc=stsym("unix"); 1180 # endif 1181 # if gcos 1182 ysysloc=stsym ("gcos"); 1183 # endif 1184 # if ibm 1185 ysysloc=stsym ("ibm"); 1186 # endif 1187 # if pdp11 1188 varloc=stsym("pdp11"); 1189 # endif 1190 # if vax 1191 varloc=stsym("vax"); 1192 # endif 1193 # if interdata 1194 varloc=stsym ("interdata"); 1195 # endif 1196 # if tss 1197 varloc=stsym ("tss"); 1198 # endif 1199 # if os 1200 varloc=stsym ("os"); 1201 # endif 1202 # if mert 1203 varloc=stsym ("mert"); 1204 # endif 1205 # if mc68000 1206 varloc=stsym("mc68000"); 1207 # endif 1208 # if sun 1209 varloc=stsym("sun"); 1210 # endif 1211 ulnloc=stsym ("__LINE__"); 1212 uflloc=stsym ("__FILE__"); 1213 1214 tf=fnames[ifno]; fnames[ifno]="command line"; lineno[ifno]=1; 1215 cp2=prespc; 1216 while (cp2<predef) stsym(*cp2++); 1217 cp2=punspc; 1218 while (cp2<prund) { 1219 if (p=strdex(*cp2, '=')) *p++='\0'; 1220 lookup(*cp2++, DROP); 1221 } 1222 fnames[ifno]=tf; 1223 pbeg=buffer+NCPS; pbuf=pbeg+BUFSIZ; pend=pbuf+BUFSIZ; 1224 1225 trulvl = 0; flslvl = 0; 1226 lineno[0] = 1; sayline(START); 1227 outp=inp=pend; 1228 control(pend); 1229 return (exfail); 1230 } 1231