Research V7 development
[unix-history] / usr / src / cmd / cpp / cpp.c
CommitLineData
8150d190
KT
1#
2# include "stdio.h"
3/* C command
4/* written by John F. Reiser
5/* July/August 1978
6*/
7
8#define STATIC
9
10#define STDIN 0
11#define STDOUT 1
12#define STDERR 2
13#define READ 0
14#define WRITE 1
15#define SALT '#'
16#ifndef BUFSIZ
17#define BUFSIZ 512
18#endif
19
20char *pbeg,*pbuf,*pend;
21char *outp,*inp;
22char *newp;
23char cinit;
24
25/* some code depends on whether characters are sign or zero extended */
26/* #if '\377' < 0 not used here, old cpp doesn't understand */
27#if pdp11 | vax
28#define COFF 128
29#else
30#define COFF 0
31#endif
32
33# if gcos
34#define ALFSIZ 512 /* alphabet size */
35# else
36#define ALFSIZ 256 /* alphabet size */
37# endif
38char macbit[ALFSIZ+11];
39char toktyp[ALFSIZ];
40#define BLANK 1
41#define IDENT 2
42#define NUMBR 3
43
44/* a superimposed code is used to reduce the number of calls to the
45/* symbol table lookup routine. (if the kth character of an identifier
46/* is 'a' and there are no macro names whose kth character is 'a'
47/* then the identifier cannot be a macro name, hence there is no need
48/* to look in the symbol table.) 'scw1' enables the test based on
49/* single characters and their position in the identifier. 'scw2'
50/* enables the test based on adjacent pairs of characters and their
51/* position in the identifier. scw1 typically costs 1 indexed fetch,
52/* an AND, and a jump per character of identifier, until the identifier
53/* is known as a non-macro name or until the end of the identifier.
54/* scw1 is inexpensive. scw2 typically costs 4 indexed fetches,
55/* an add, an AND, and a jump per character of identifier, but it is also
56/* slightly more effective at reducing symbol table searches.
57/* scw2 usually costs too much because the symbol table search is
58/* usually short; but if symbol table search should become expensive,
59/* the code is here.
60/* using both scw1 and scw2 is of dubious value.
61*/
62#define scw1 1
63#define scw2 0
64
65#if scw2
66char t21[ALFSIZ],t22[ALFSIZ],t23[ALFSIZ+8];
67#endif
68
69#if scw1
70#define b0 1
71#define b1 2
72#define b2 4
73#define b3 8
74#define b4 16
75#define b5 32
76#define b6 64
77#define b7 128
78#endif
79
80#define IB 1
81#define SB 2
82#define NB 4
83#define CB 8
84#define QB 16
85#define WB 32
86char fastab[ALFSIZ];
87char slotab[ALFSIZ];
88char *ptrtab;
89#define isslo (ptrtab==(slotab+COFF))
90#define isid(a) ((fastab+COFF)[a]&IB)
91#define isspc(a) (ptrtab[a]&SB)
92#define isnum(a) ((fastab+COFF)[a]&NB)
93#define iscom(a) ((fastab+COFF)[a]&CB)
94#define isquo(a) ((fastab+COFF)[a]&QB)
95#define iswarn(a) ((fastab+COFF)[a]&WB)
96
97#define eob(a) ((a)>=pend)
98#define bob(a) (pbeg>=(a))
99
100char buffer[8+BUFSIZ+BUFSIZ+8];
101
102# define SBSIZE 12000
103char sbf[SBSIZE];
104char *savch = sbf;
105
106# define DROP 0xFE /* special character not legal ASCII or EBCDIC */
107# define WARN DROP
108# define SAME 0
109# define MAXINC 10
110# define MAXFRE 14 /* max buffers of macro pushback */
111# define MAXFRM 31 /* max number of formals/actuals to a macro */
112
113static char warnc = WARN;
114
115int mactop,fretop;
116char *instack[MAXFRE],*bufstack[MAXFRE],*endbuf[MAXFRE];
117
118int plvl; /* parenthesis level during scan for macro actuals */
119int maclin; /* line number of macro call requiring actuals */
120char *macfil; /* file name of macro call requiring actuals */
121char *macnam; /* name of macro requiring actuals */
122int maclvl; /* # calls since last decrease in nesting level */
123char *macforw; /* pointer which must be exceeded to decrease nesting level */
124int macdam; /* offset to macforw due to buffer shifting */
125
126#if tgp
127int tgpscan; /* flag for dump(); */
128#endif
129
130STATIC int inctop[MAXINC];
131STATIC char *fnames[MAXINC];
132STATIC char *dirnams[MAXINC]; /* actual directory of #include files */
133STATIC int fins[MAXINC];
134STATIC int lineno[MAXINC];
135
136STATIC char *dirs[10]; /* -I and <> directories */
137char *strdex(), *copy(), *subst(), *trmdir();
138struct symtab *stsym();
139STATIC int fin = STDIN;
140STATIC FILE *fout = stdout;
141STATIC int nd = 1;
142STATIC int pflag; /* don't put out lines "# 12 foo.c" */
143STATIC int passcom; /* don't delete comments */
144STATIC int rflag; /* allow macro recursion */
145STATIC int ifno;
146# define NPREDEF 20
147STATIC char *prespc[NPREDEF];
148STATIC char **predef = prespc;
149STATIC char *punspc[NPREDEF];
150STATIC char **prund = punspc;
151STATIC int exfail;
152struct symtab {
153 char *name;
154 char *value;
155} *lastsym, *lookup(), *slookup();
156
157# if gcos
158#include <setjmp.h>
159static jmp_buf env;
160# define main mainpp
161# undef exit
162# define exit(S) longjmp(env, 1)
163# define open(S,D) fileno(fopen(S, "r"))
164# define close(F) fclose(_f[F])
165extern FILE *_f[];
166# define symsiz 500
167# else
168# define symsiz 400
169# endif
170STATIC struct symtab stab[symsiz];
171
172STATIC struct symtab *defloc;
173STATIC struct symtab *udfloc;
174STATIC struct symtab *incloc;
175STATIC struct symtab *ifloc;
176STATIC struct symtab *elsloc;
177STATIC struct symtab *eifloc;
178STATIC struct symtab *ifdloc;
179STATIC struct symtab *ifnloc;
180STATIC struct symtab *ysysloc;
181STATIC struct symtab *varloc;
182STATIC struct symtab *lneloc;
183STATIC struct symtab *ulnloc;
184STATIC struct symtab *uflloc;
185STATIC int trulvl;
186STATIC int flslvl;
187
188sayline() {
189 if (pflag==0) fprintf(fout,"# %d \"%s\"\n", lineno[ifno], fnames[ifno]);
190}
191
192/* data structure guide
193/*
194/* most of the scanning takes place in the buffer:
195/*
196/* (low address) (high address)
197/* pbeg pbuf pend
198/* | <-- BUFSIZ chars --> | <-- BUFSIZ chars --> |
199/* _______________________________________________________________________
200/* |_______________________________________________________________________|
201/* | | |
202/* |<-- waiting -->| |<-- waiting -->
203/* | to be |<-- current -->| to be
204/* | written | token | scanned
205/* | | |
206/* outp inp p
207/*
208/* *outp first char not yet written to output file
209/* *inp first char of current token
210/* *p first char not yet scanned
211/*
212/* macro expansion: write from *outp to *inp (chars waiting to be written),
213/* ignore from *inp to *p (chars of the macro call), place generated
214/* characters in front of *p (in reverse order), update pointers,
215/* resume scanning.
216/*
217/* symbol table pointers point to just beyond the end of macro definitions;
218/* the first preceding character is the number of formal parameters.
219/* the appearance of a formal in the body of a definition is marked by
220/* 2 chars: the char WARN, and a char containing the parameter number.
221/* the first char of a definition is preceded by a zero character.
222/*
223/* when macro expansion attempts to back up over the beginning of the
224/* buffer, some characters preceding *pend are saved in a side buffer,
225/* the address of the side buffer is put on 'instack', and the rest
226/* of the main buffer is moved to the right. the end of the saved buffer
227/* is kept in 'endbuf' since there may be nulls in the saved buffer.
228/*
229/* similar action is taken when an 'include' statement is processed,
230/* except that the main buffer must be completely emptied. the array
231/* element 'inctop[ifno]' records the last side buffer saved when
232/* file 'ifno' was included. these buffers remain dormant while
233/* the file is being read, and are reactivated at end-of-file.
234/*
235/* instack[0 : mactop] holds the addresses of all pending side buffers.
236/* instack[inctop[ifno]+1 : mactop-1] holds the addresses of the side
237/* buffers which are "live"; the side buffers instack[0 : inctop[ifno]]
238/* are dormant, waiting for end-of-file on the current file.
239/*
240/* space for side buffers is obtained from 'savch' and is never returned.
241/* bufstack[0:fretop-1] holds addresses of side buffers which
242/* are available for use.
243*/
244
245dump() {
246/* write part of buffer which lies between outp and inp .
247/* this should be a direct call to 'write', but the system slows to a crawl
248/* if it has to do an unaligned copy. thus we buffer. this silly loop
249/* is 15% of the total time, thus even the 'putc' macro is too slow.
250*/
251 register char *p1,*p2; register FILE *f;
252 if ((p1=outp)==inp || flslvl!=0) return;
253#if tgp
254#define MAXOUT 80
255 if (!tgpscan) {/* scan again to insure <= MAXOUT chars between linefeeds */
256 register char c,*pblank; char savc,stopc,brk;
257 tgpscan=1; brk=stopc=pblank=0; p2=inp; savc= *p2; *p2='\0';
258 while (c= *p1++) {
259 if (c=='\\') c= *p1++;
260 if (stopc==c) stopc=0;
261 else if (c=='"' || c=='\'') stopc=c;
262 if (p1-outp>MAXOUT && pblank!=0) {
263 *pblank++='\n'; inp=pblank; dump(); brk=1; pblank=0;
264 }
265 if (c==' ' && stopc==0) pblank=p1-1;
266 }
267 if (brk) sayline();
268 *p2=savc; inp=p2; p1=outp; tgpscan=0;
269 }
270#endif
271 f=fout;
272# if gcos
273/* filter out "$ program c" card if first line of input */
274/* gmatch is a simple pattern matcher in the GCOS Standard Library */
275{ static int gmfirst = 0;
276 if (!gmfirst) {
277 ++gmfirst;
278 if (gmatch(p1, "^$*program[ \t]*c*"))
279 p1 = strdex(p1, '\n');
280 }
281}
282# endif
283 while (p1<inp) putc(*p1++,f);
284 outp=p1;
285}
286
287char *
288refill(p) register char *p; {
289/* dump buffer. save chars from inp to p. read into buffer at pbuf,
290/* contiguous with p. update pointers, return new p.
291*/
292 register char *np,*op; register int ninbuf;
293 dump(); np=pbuf-(p-inp); op=inp;
294 if (bob(np+1)) {pperror("token too long"); np=pbeg; p=inp+BUFSIZ;}
295 macdam += np-inp; outp=inp=np;
296 while (op<p) *np++= *op++;
297 p=np;
298 for (;;) {
299 if (mactop>inctop[ifno]) {/* retrieve hunk of pushed-back macro text */
300 op=instack[--mactop]; np=pbuf;
301 do {while (*np++= *op++);} while (op<endbuf[mactop]); pend=np-1;
302 /* make buffer space avail for 'include' processing */
303 if (fretop<MAXFRE) bufstack[fretop++]=instack[mactop];
304 return(p);
305 } else {/* get more text from file(s) */
306 maclvl=0;
307 if (0<(ninbuf=read(fin,pbuf,BUFSIZ))) {
308 pend=pbuf+ninbuf; *pend='\0';
309 return(p);
310 }
311 /* end of #include file */
312 if (ifno==0) {/* end of input */
313 if (plvl!=0) {
314 int n=plvl,tlin=lineno[ifno]; char *tfil=fnames[ifno];
315 lineno[ifno]=maclin; fnames[ifno]=macfil;
316 pperror("%s: unterminated macro call",macnam);
317 lineno[ifno]=tlin; fnames[ifno]=tfil;
318 np=p; *np++='\n'; /* shut off unterminated quoted string */
319 while (--n>=0) *np++=')'; /* supply missing parens */
320 pend=np; *np='\0'; if (plvl<0) plvl=0;
321 return(p);
322 }
323 inp=p; dump(); exit(exfail);
324 }
325 close(fin); fin=fins[--ifno]; dirs[0]=dirnams[ifno]; sayline();
326 }
327 }
328}
329
330#define BEG 0
331#define LF 1
332
333char *
334cotoken(p) register char *p; {
335 register int c,i; char quoc;
336 static int state = BEG;
337
338 if (state!=BEG) goto prevlf;
339for (;;) {
340again:
341 while (!isspc(*p++));
342 switch (*(inp=p-1)) {
343 case 0: {
344 if (eob(--p)) {p=refill(p); goto again;}
345 else ++p; /* ignore null byte */
346 } break;
347 case '|': case '&': for (;;) {/* sloscan only */
348 if (*p++== *inp) break;
349 if (eob(--p)) p=refill(p);
350 else break;
351 } break;
352 case '=': case '!': for (;;) {/* sloscan only */
353 if (*p++=='=') break;
354 if (eob(--p)) p=refill(p);
355 else break;
356 } break;
357 case '<': case '>': for (;;) {/* sloscan only */
358 if (*p++=='=' || p[-2]==p[-1]) break;
359 if (eob(--p)) p=refill(p);
360 else break;
361 } break;
362 case '\\': for (;;) {
363 if (*p++=='\n') {++lineno[ifno]; break;}
364 if (eob(--p)) p=refill(p);
365 else {++p; break;}
366 } break;
367 case '/': for (;;) {
368 if (*p++=='*') {/* comment */
369 if (!passcom) {inp=p-2; dump(); ++flslvl;}
370 for (;;) {
371 while (!iscom(*p++));
372 if (p[-1]=='*') for (;;) {
373 if (*p++=='/') goto endcom;
374 if (eob(--p)) {
375 if (!passcom) {inp=p; p=refill(p);}
376 else if ((p-inp)>=BUFSIZ) {/* split long comment */
377 inp=p; p=refill(p); /* last char written is '*' */
378 putc('/',fout); /* terminate first part */
379 /* and fake start of 2nd */
380 outp=inp=p-=3; *p++='/'; *p++='*'; *p++='*';
381 } else p=refill(p);
382 } else break;
383 } else if (p[-1]=='\n') {
384 ++lineno[ifno]; if (!passcom) putc('\n',fout);
385 } else if (eob(--p)) {
386 if (!passcom) {inp=p; p=refill(p);}
387 else if ((p-inp)>=BUFSIZ) {/* split long comment */
388 inp=p; p=refill(p);
389 putc('*',fout); putc('/',fout);
390 outp=inp=p-=2; *p++='/'; *p++='*';
391 } else p=refill(p);
392 } else ++p; /* ignore null byte */
393 }
394 endcom:
395 if (!passcom) {outp=inp=p; --flslvl; goto again;}
396 break;
397 }
398 if (eob(--p)) p=refill(p);
399 else break;
400 } break;
401# if gcos
402 case '`':
403# endif
404 case '"': case '\'': {
405 quoc=p[-1];
406 for (;;) {
407 while (!isquo(*p++));
408 if (p[-1]==quoc) break;
409 if (p[-1]=='\n') {--p; break;} /* bare \n terminates quotation */
410 if (p[-1]=='\\') for (;;) {
411 if (*p++=='\n') {++lineno[ifno]; break;} /* escaped \n ignored */
412 if (eob(--p)) p=refill(p);
413 else {++p; break;}
414 } else if (eob(--p)) p=refill(p);
415 else ++p; /* it was a different quote character */
416 }
417 } break;
418 case '\n': {
419 ++lineno[ifno]; if (isslo) {state=LF; return(p);}
420prevlf:
421 state=BEG;
422 for (;;) {
423 if (*p++=='#') return(p);
424 if (eob(inp= --p)) p=refill(p);
425 else goto again;
426 }
427 } break;
428 case '0': case '1': case '2': case '3': case '4':
429 case '5': case '6': case '7': case '8': case '9':
430 for (;;) {
431 while (isnum(*p++));
432 if (eob(--p)) p=refill(p);
433 else break;
434 } break;
435 case 'A': case 'B': case 'C': case 'D': case 'E':
436 case 'F': case 'G': case 'H': case 'I': case 'J':
437 case 'K': case 'L': case 'M': case 'N': case 'O':
438 case 'P': case 'Q': case 'R': case 'S': case 'T':
439 case 'U': case 'V': case 'W': case 'X': case 'Y':
440 case 'Z': case '_':
441 case 'a': case 'b': case 'c': case 'd': case 'e':
442 case 'f': case 'g': case 'h': case 'i': case 'j':
443 case 'k': case 'l': case 'm': case 'n': case 'o':
444 case 'p': case 'q': case 'r': case 's': case 't':
445 case 'u': case 'v': case 'w': case 'x': case 'y':
446 case 'z':
447#if scw1
448#define tmac1(c,bit) if (!xmac1(c,bit,&)) goto nomac
449#define xmac1(c,bit,op) ((macbit+COFF)[c] op (bit))
450#else
451#define tmac1(c,bit)
452#define xmac1(c,bit,op)
453#endif
454
455#if scw2
456#define tmac2(c0,c1,cpos) if (!xmac2(c0,c1,cpos,&)) goto nomac
457#define xmac2(c0,c1,cpos,op)\
458 ((macbit+COFF)[(t21+COFF)[c0]+(t22+COFF)[c1]] op (t23+COFF+cpos)[c0])
459#else
460#define tmac2(c0,c1,cpos)
461#define xmac2(c0,c1,cpos,op)
462#endif
463
464 if (flslvl) goto nomac;
465 for (;;) {
466 c= p[-1]; tmac1(c,b0);
467 i= *p++; if (!isid(i)) goto endid; tmac1(i,b1); tmac2(c,i,0);
468 c= *p++; if (!isid(c)) goto endid; tmac1(c,b2); tmac2(i,c,1);
469 i= *p++; if (!isid(i)) goto endid; tmac1(i,b3); tmac2(c,i,2);
470 c= *p++; if (!isid(c)) goto endid; tmac1(c,b4); tmac2(i,c,3);
471 i= *p++; if (!isid(i)) goto endid; tmac1(i,b5); tmac2(c,i,4);
472 c= *p++; if (!isid(c)) goto endid; tmac1(c,b6); tmac2(i,c,5);
473 i= *p++; if (!isid(i)) goto endid; tmac1(i,b7); tmac2(c,i,6);
474 tmac2(i,0,7);
475 while (isid(*p++));
476 if (eob(--p)) {refill(p); p=inp+1; continue;}
477 goto lokid;
478 endid:
479 if (eob(--p)) {refill(p); p=inp+1; continue;}
480 tmac2(p[-1],0,-1+(p-inp));
481 lokid:
482 slookup(inp,p,0); if (newp) {p=newp; goto again;}
483 else break;
484 nomac:
485 while (isid(*p++));
486 if (eob(--p)) {p=refill(p); goto nomac;}
487 else break;
488 } break;
489 } /* end of switch */
490
491 if (isslo) return(p);
492} /* end of infinite loop */
493}
494
495char *
496skipbl(p) register char *p; {/* get next non-blank token */
497 do {outp=inp=p; p=cotoken(p);} while ((toktyp+COFF)[*inp]==BLANK);
498 return(p);
499}
500
501char *
502unfill(p) register char *p; {
503/* take <= BUFSIZ chars from right end of buffer and put them on instack .
504/* slide rest of buffer to the right, update pointers, return new p.
505*/
506 register char *np,*op; register int d;
507 if (mactop>=MAXFRE) {
508 pperror("%s: too much pushback",macnam);
509 p=inp=pend; dump(); /* begin flushing pushback */
510 while (mactop>inctop[ifno]) {p=refill(p); p=inp=pend; dump();}
511 }
512 if (fretop>0) np=bufstack[--fretop];
513 else {
514 np=savch; savch+=BUFSIZ;
515 if (savch>=sbf+SBSIZE) {pperror("no space"); exit(exfail);}
516 *savch++='\0';
517 }
518 instack[mactop]=np; op=pend-BUFSIZ; if (op<p) op=p;
519 for (;;) {while (*np++= *op++); if (eob(op)) break;} /* out with old */
520 endbuf[mactop++]=np; /* mark end of saved text */
521 np=pbuf+BUFSIZ; op=pend-BUFSIZ; pend=np; if (op<p) op=p;
522 while (outp<op) *--np= *--op; /* slide over new */
523 if (bob(np)) pperror("token too long");
524 d=np-outp; outp+=d; inp+=d; macdam+=d; return(p+d);
525}
526
527char *
528doincl(p) register char *p; {
529 int filok,inctype;
530 register char *cp; char **dirp,*nfil; char filname[BUFSIZ];
531
532 p=skipbl(p); cp=filname;
533 if (*inp++=='<') {/* special <> syntax */
534 inctype=1;
535 for (;;) {
536 outp=inp=p; p=cotoken(p);
537 if (*inp=='\n') {--p; *cp='\0'; break;}
538 if (*inp=='>') { *cp='\0'; break;}
539# ifdef gimpel
540 if (*inp=='.' && !intss()) *inp='#';
541# endif
542 while (inp<p) *cp++= *inp++;
543 }
544 } else if (inp[-1]=='"') {/* regular "" syntax */
545 inctype=0;
546# ifdef gimpel
547 while (inp<p) {if (*inp=='.' && !intss()) *inp='#'; *cp++= *inp++;}
548# else
549 while (inp<p) *cp++= *inp++;
550# endif
551 if (*--cp=='"') *cp='\0';
552 } else {pperror("bad include syntax",0); inctype=2;}
553 /* flush current file to \n , then write \n */
554 ++flslvl; do {outp=inp=p; p=cotoken(p);} while (*inp!='\n'); --flslvl;
555 inp=p; dump(); if (inctype==2) return(p);
556 /* look for included file */
557 if (ifno+1 >=MAXINC) {
558 pperror("Unreasonable include nesting",0); return(p);
559 }
560 if((nfil=savch)>sbf+SBSIZE-BUFSIZ) {pperror("no space"); exit(exfail);}
561 filok=0;
562 for (dirp=dirs+inctype; *dirp; ++dirp) {
563 if (
564# if gcos
565 strdex(filname, '/')
566# else
567 filname[0]=='/'
568# endif
569 || **dirp=='\0') strcpy(nfil,filname);
570 else {
571 strcpy(nfil,*dirp);
572# if unix || gcos
573 strcat(nfil,"/");
574# endif
575#ifdef ibm
576#ifndef gimpel
577 strcat(nfil,".");
578#endif
579#endif
580 strcat(nfil,filname);
581 }
582 if (0<(fins[ifno+1]=open(nfil,READ))) {
583 filok=1; fin=fins[++ifno]; break;
584 }
585 }
586 if (filok==0) pperror("Can't find include file %s",filname);
587 else {
588 lineno[ifno]=1; fnames[ifno]=cp=nfil; while (*cp++); savch=cp;
589 dirnams[ifno]=dirs[0]=trmdir(copy(nfil));
590 sayline();
591 /* save current contents of buffer */
592 while (!eob(p)) p=unfill(p);
593 inctop[ifno]=mactop;
594 }
595 return(p);
596}
597
598equfrm(a,p1,p2) register char *a,*p1,*p2; {
599 register char c; int flag;
600 c= *p2; *p2='\0';
601 flag=strcmp(a,p1); *p2=c; return(flag==SAME);
602}
603
604char *
605dodef(p) char *p; {/* process '#define' */
606 register char *pin,*psav,*cf;
607 char **pf,**qf; int b,c,params; struct symtab *np;
608 char *oldval,*oldsavch;
609 char *formal[MAXFRM]; /* formal[n] is name of nth formal */
610 char formtxt[BUFSIZ]; /* space for formal names */
611
612 if (savch>sbf+SBSIZE-BUFSIZ) {pperror("too much defining"); return(p);}
613 oldsavch=savch; /* to reclaim space if redefinition */
614 ++flslvl; /* prevent macro expansion during 'define' */
615 p=skipbl(p); pin=inp;
616 if ((toktyp+COFF)[*pin]!=IDENT) {
617 ppwarn("illegal macro name"); while (*inp!='\n') p=skipbl(p); return(p);
618 }
619 np=slookup(pin,p,1);
620 if (oldval=np->value) savch=oldsavch; /* was previously defined */
621 b=1; cf=pin;
622 while (cf<p) {/* update macbit */
623 c= *cf++; xmac1(c,b,|=); b=(b+b)&0xFF;
624 if (cf!=p) xmac2(c,*cf,-1+(cf-pin),|=);
625 else xmac2(c,0,-1+(cf-pin),|=);
626 }
627 params=0; outp=inp=p; p=cotoken(p); pin=inp;
628 if (*pin=='(') {/* with parameters; identify the formals */
629 cf=formtxt; pf=formal;
630 for (;;) {
631 p=skipbl(p); pin=inp;
632 if (*pin=='\n') {
633 --lineno[ifno]; --p; pperror("%s: missing )",np->name); break;
634 }
635 if (*pin==')') break;
636 if (*pin==',') continue;
637 if ((toktyp+COFF)[*pin]!=IDENT) {
638 c= *p; *p='\0'; pperror("bad formal: %s",pin); *p=c;
639 } else if (pf>= &formal[MAXFRM]) {
640 c= *p; *p='\0'; pperror("too many formals: %s",pin); *p=c;
641 } else {
642 *pf++=cf; while (pin<p) *cf++= *pin++; *cf++='\0'; ++params;
643 }
644 }
645 if (params==0) --params; /* #define foo() ... */
646 } else if (*pin=='\n') {--lineno[ifno]; --p;}
647 /* remember beginning of macro body, so that we can
648 /* warn if a redefinition is different from old value.
649 */
650 oldsavch=psav=savch;
651 for (;;) {/* accumulate definition until linefeed */
652 outp=inp=p; p=cotoken(p); pin=inp;
653 if (*pin=='\\' && pin[1]=='\n') continue; /* ignore escaped lf */
654 if (*pin=='\n') break;
655 if (params) {/* mark the appearance of formals in the definiton */
656 if ((toktyp+COFF)[*pin]==IDENT) {
657 for (qf=pf; --qf>=formal; ) {
658 if (equfrm(*qf,pin,p)) {
659 *psav++=qf-formal+1; *psav++=WARN; pin=p; break;
660 }
661 }
662 } else if (*pin=='"' || *pin=='\''
663# if gcos
664 || *pin=='`'
665# endif
666 ) {/* inside quotation marks, too */
667 char quoc= *pin;
668 for (*psav++= *pin++; pin<p && *pin!=quoc; ) {
669 while (pin<p && !isid(*pin)) *psav++= *pin++;
670 cf=pin; while (cf<p && isid(*cf)) ++cf;
671 for (qf=pf; --qf>=formal; ) {
672 if (equfrm(*qf,pin,cf)) {
673 *psav++=qf-formal+1; *psav++=WARN; pin=cf; break;
674 }
675 }
676 while (pin<cf) *psav++= *pin++;
677 }
678 }
679 }
680 while (pin<p) *psav++= *pin++;
681 }
682 *psav++=params; *psav++='\0';
683 if ((cf=oldval)!=NULL) {/* redefinition */
684 --cf; /* skip no. of params, which may be zero */
685 while (*--cf); /* go back to the beginning */
686 if (0!=strcmp(++cf,oldsavch)) {/* redefinition different from old */
687 --lineno[ifno]; ppwarn("%s redefined",np->name); ++lineno[ifno];
688 np->value=psav-1;
689 } else psav=oldsavch; /* identical redef.; reclaim space */
690 } else np->value=psav-1;
691 --flslvl; inp=pin; savch=psav; return(p);
692}
693
694#define fasscan() ptrtab=fastab+COFF
695#define sloscan() ptrtab=slotab+COFF
696
697char *
698control(p) register char *p; {/* find and handle preprocessor control lines */
699 register struct symtab *np;
700for (;;) {
701 fasscan(); p=cotoken(p); if (*inp=='\n') ++inp; dump();
702 sloscan(); p=skipbl(p);
703 *--inp=SALT; outp=inp; ++flslvl; np=slookup(inp,p,0); --flslvl;
704 if (np==defloc) {/* define */
705 if (flslvl==0) {p=dodef(p); continue;}
706 } else if (np==incloc) {/* include */
707 if (flslvl==0) {p=doincl(p); continue;}
708 } else if (np==ifnloc) {/* ifndef */
709 ++flslvl; p=skipbl(p); np=slookup(inp,p,0); --flslvl;
710 if (flslvl==0 && np->value==0) ++trulvl;
711 else ++flslvl;
712 } else if (np==ifdloc) {/* ifdef */
713 ++flslvl; p=skipbl(p); np=slookup(inp,p,0); --flslvl;
714 if (flslvl==0 && np->value!=0) ++trulvl;
715 else ++flslvl;
716 } else if (np==eifloc) {/* endif */
717 if (flslvl) {if (--flslvl==0) sayline();}
718 else if (trulvl) --trulvl;
719 else pperror("If-less endif",0);
720 } else if (np==elsloc) {/* else */
721 if (flslvl) {
722 if (--flslvl!=0) ++flslvl;
723 else {++trulvl; sayline();}
724 }
725 else if (trulvl) {++flslvl; --trulvl;}
726 else pperror("If-less else",0);
727 } else if (np==udfloc) {/* undefine */
728 if (flslvl==0) {
729 ++flslvl; p=skipbl(p); slookup(inp,p,DROP); --flslvl;
730 }
731 } else if (np==ifloc) {/* if */
732#if tgp
733 pperror(" IF not implemented, true assumed", 0);
734 if (flslvl==0) ++trulvl; else ++flslvl;
735#else
736 newp=p;
737 if (flslvl==0 && yyparse()) ++trulvl; else ++flslvl;
738 p=newp;
739#endif
740 } else if (np==lneloc) {/* line */
741 if (flslvl==0 && pflag==0) {
742 outp=inp=p; *--outp='#'; while (*inp!='\n') p=cotoken(p);
743 continue;
744 }
745 } else if (*++inp=='\n') outp=inp; /* allows blank line after # */
746 else pperror("undefined control",0);
747 /* flush to lf */
748 ++flslvl; while (*inp!='\n') {outp=inp=p; p=cotoken(p);} --flslvl;
749}
750}
751
752struct symtab *
753stsym(s) register char *s; {
754 char buf[BUFSIZ]; register char *p;
755
756 /* make definition look exactly like end of #define line */
757 /* copy to avoid running off end of world when param list is at end */
758 p=buf; while (*p++= *s++);
759 p=buf; while (isid(*p++)); /* skip first identifier */
760 if (*--p=='=') {*p++=' '; while (*p++);}
761 else {s=" 1"; while (*p++= *s++);}
762 pend=p; *--p='\n';
763 sloscan(); dodef(buf); return(lastsym);
764}
765
766struct symtab *
767ppsym(s) char *s; {/* kluge */
768 register struct symtab *sp;
769 cinit=SALT; *savch++=SALT; sp=stsym(s); --sp->name; cinit=0; return(sp);
770}
771
772/* VARARGS1 */
773pperror(s,x,y) char *s; {
774 if (fnames[ifno][0]) fprintf(stderr,
775# if gcos
776 "*%c* \"%s\", line ", exfail >= 0 ? 'F' : 'W',
777# else
778 "%s: ",
779# endif
780 fnames[ifno]);
781 fprintf(stderr, "%d: ",lineno[ifno]);
782 fprintf(stderr, s, x, y);
783 fprintf(stderr,"\n");
784 ++exfail;
785}
786
787yyerror(s,a,b) char *s; {
788 pperror(s,a,b);
789}
790
791ppwarn(s,x) char *s; {
792 int fail = exfail;
793 exfail = -1;
794 pperror(s,x);
795 exfail = fail;
796}
797
798struct symtab *
799lookup(namep, enterf)
800char *namep;
801{
802 register char *np, *snp;
803 register int c, i; int around;
804 register struct symtab *sp;
805
806 /* namep had better not be too long (currently, <=8 chars) */
807 np=namep; around=0; i=cinit;
808 while (c= *np++) i += i+c; c=i; /* c=i for register usage on pdp11 */
809 c %= symsiz; if (c<0) c += symsiz;
810 sp = &stab[c];
811 while (snp=sp->name) {
812 np = namep;
813 while (*snp++ == *np) if (*np++ == '\0') {
814 if (enterf==DROP) {sp->name[0]= DROP; sp->value=0;}
815 return(lastsym=sp);
816 }
817 if (--sp < &stab[0])
818 if (around) {pperror("too many defines", 0); exit(exfail);}
819 else {++around; sp = &stab[symsiz-1];}
820 }
821 if (enterf>0) sp->name=namep;
822 return(lastsym=sp);
823}
824
825struct symtab *
826slookup(p1,p2,enterf) register char *p1,*p2; int enterf;{
827 register char *p3; char c2,c3; struct symtab *np;
828 c2= *p2; *p2='\0'; /* mark end of token */
829 if ((p2-p1)>8) p3=p1+8; else p3=p2;
830 c3= *p3; *p3='\0'; /* truncate to 8 chars or less */
831 if (enterf==1) p1=copy(p1);
832 np=lookup(p1,enterf); *p3=c3; *p2=c2;
833 if (np->value!=0 && flslvl==0) newp=subst(p2,np);
834 else newp=0;
835 return(np);
836}
837
838char *
839subst(p,sp) register char *p; struct symtab *sp; {
840 static char match[]="%s: argument mismatch";
841 register char *ca,*vp; int params;
842 char *actual[MAXFRM]; /* actual[n] is text of nth actual */
843 char acttxt[BUFSIZ]; /* space for actuals */
844
845 if (0==(vp=sp->value)) return(p);
846 if ((p-macforw)<=macdam) {
847 if (++maclvl>symsiz && !rflag) {
848 pperror("%s: macro recursion",sp->name); return(p);
849 }
850 } else maclvl=0; /* level decreased */
851 macforw=p; macdam=0; /* new target for decrease in level */
852 macnam=sp->name;
853 dump();
854 if (sp==ulnloc) {
855 vp=acttxt; *vp++='\0';
856 sprintf(vp,"%d",lineno[ifno]); while (*vp++);
857 } else if (sp==uflloc) {
858 vp=acttxt; *vp++='\0';
859 sprintf(vp,"\"%s\"",fnames[ifno]); while (*vp++);
860 }
861 if (0!=(params= *--vp&0xFF)) {/* definition calls for params */
862 register char **pa;
863 ca=acttxt; pa=actual;
864 if (params==0xFF) params=1; /* #define foo() ... */
865 sloscan(); ++flslvl; /* no expansion during search for actuals */
866 plvl= -1;
867 do p=skipbl(p); while (*inp=='\n'); /* skip \n too */
868 if (*inp=='(') {
869 maclin=lineno[ifno]; macfil=fnames[ifno];
870 for (plvl=1; plvl!=0; ) {
871 *ca++='\0';
872 for (;;) {
873 outp=inp=p; p=cotoken(p);
874 if (*inp=='(') ++plvl;
875 if (*inp==')' && --plvl==0) {--params; break;}
876 if (plvl==1 && *inp==',') {--params; break;}
877 while (inp<p) *ca++= *inp++;
878 if (ca> &acttxt[BUFSIZ])
879 pperror("%s: actuals too long",sp->name);
880 }
881 if (pa>= &actual[MAXFRM]) ppwarn(match,sp->name);
882 else *pa++=ca;
883 }
884 }
885 if (params!=0) ppwarn(match,sp->name);
886 while (--params>=0) *pa++=""+1; /* null string for missing actuals */
887 --flslvl; fasscan();
888 }
889 for (;;) {/* push definition onto front of input stack */
890 while (!iswarn(*--vp)) {
891 if (bob(p)) {outp=inp=p; p=unfill(p);}
892 *--p= *vp;
893 }
894 if (*vp==warnc) {/* insert actual param */
895 ca=actual[*--vp-1];
896 while (*--ca) {
897 if (bob(p)) {outp=inp=p; p=unfill(p);}
898 *--p= *ca;
899 }
900 } else break;
901 }
902 outp=inp=p;
903 return(p);
904}
905
906
907
908
909char *
910trmdir(s) register char *s; {
911 register char *p = s;
912 while (*p++); --p; while (p>s && *--p!='/');
913# if unix
914 if (p==s) *p++='.';
915# endif
916 *p='\0';
917 return(s);
918}
919
920STATIC char *
921copy(s) register char *s; {
922 register char *old;
923
924 old = savch; while (*savch++ = *s++);
925 return(old);
926}
927
928char *
929strdex(s,c) char *s,c; {
930 while (*s) if (*s++==c) return(--s);
931 return(0);
932}
933
934yywrap(){ return(1); }
935
936main(argc,argv)
937 char *argv[];
938{
939 register int i,c;
940 register char *p;
941 char *tf,**cp2;
942
943# if gcos
944 if (setjmp(env)) return (exfail);
945# endif
946 p="_ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789";
947 i=0;
948 while (c= *p++) {
949 (fastab+COFF)[c] |= IB|NB|SB; (toktyp+COFF)[c]=IDENT;
950#if scw2
951 /* 53 == 63-10; digits rarely appear in identifiers,
952 /* and can never be the first char of an identifier.
953 /* 11 == 53*53/sizeof(macbit) .
954 */
955 ++i; (t21+COFF)[c]=(53*i)/11; (t22+COFF)[c]=i%11;
956#endif
957 }
958 p="0123456789.";
959 while (c= *p++) {(fastab+COFF)[c] |= NB|SB; (toktyp+COFF)[c]=NUMBR;}
960# if gcos
961 p="\n\"'`/\\";
962# else
963 p="\n\"'/\\";
964# endif
965 while (c= *p++) (fastab+COFF)[c] |= SB;
966# if gcos
967 p="\n\"'`\\";
968# else
969 p="\n\"'\\";
970# endif
971 while (c= *p++) (fastab+COFF)[c] |= QB;
972 p="*\n"; while (c= *p++) (fastab+COFF)[c] |= CB;
973 (fastab+COFF)[warnc] |= WB;
974 (fastab+COFF)['\0'] |= CB|QB|SB|WB;
975 for (i=ALFSIZ; --i>=0; ) slotab[i]=fastab[i]|SB;
976 p=" \t\013\f\r"; /* note no \n; \v not legal for vertical tab? */
977 while (c= *p++) (toktyp+COFF)[c]=BLANK;
978#if scw2
979 for ((t23+COFF)[i=ALFSIZ+7-COFF]=1; --i>=-COFF; )
980 if (((t23+COFF)[i]=(t23+COFF+1)[i]<<1)==0) (t23+COFF)[i]=1;
981#endif
982
983# if unix
984 fnames[ifno=0] = "";
985# endif
986# if ibm
987 fnames[ifno=0] = "";
988# endif
989# if gcos
990 if (inquire(stdin, _TTY)) freopen("*src", "rt", stdin);
991# endif
992# if gimpel || gcos
993 fnames[ifno=0] = (char *)inquire(stdin, _FILENAME);
994 dirnams[0] = dirs[0] = trmdir(copy(fnames[0]));
995# endif
996 for(i=1; i<argc; i++)
997 {
998 switch(argv[i][0])
999 {
1000 case '-':
1001# if gcos
1002 switch(toupper(argv[i][1])) { /* case-independent on GCOS */
1003# else
1004 switch(argv[i][1]) {
1005# endif
1006 case 'P': pflag++;
1007 case 'E': continue;
1008 case 'R': ++rflag; continue;
1009 case 'C': passcom++; continue;
1010 case 'D':
1011 if (predef>prespc+NPREDEF) {
1012 pperror("too many -D options, ignoring %s",argv[i]);
1013 continue;
1014 }
1015 /* ignore plain "-D" (no argument) */
1016 if (*(argv[i]+2)) *predef++ = argv[i]+2;
1017 continue;
1018 case 'U':
1019 if (prund>punspc+NPREDEF) {
1020 pperror("too many -U options, ignoring %s",argv[i]);
1021 continue;
1022 }
1023 *prund++ = argv[i]+2;
1024 continue;
1025 case 'I':
1026 if (nd>8) pperror("excessive -I file (%s) ignored",argv[i]);
1027 else dirs[nd++] = argv[i]+2;
1028 continue;
1029 case '\0': continue;
1030 default:
1031 pperror("unknown flag %s", argv[i]);
1032 continue;
1033 }
1034 default:
1035 if (fin==STDIN) {
1036 if (0>(fin=open(argv[i], READ))) {
1037 pperror("No source file %s",argv[i]); exit(8);
1038 }
1039 fnames[ifno]=copy(argv[i]);
1040 dirs[0]=dirnams[ifno]=trmdir(argv[i]);
1041# ifndef gcos
1042/* too dangerous to have file name in same syntactic position
1043 be input or output file depending on file redirections,
1044 so force output to stdout, willy-nilly
1045 [i don't see what the problem is. jfr]
1046*/
1047 } else if (fout==stdout) {
1048 extern char _sobuf[BUFSIZ];
1049 if (NULL==(fout=fopen(argv[i], "w"))) {
1050 pperror("Can't create %s", argv[i]); exit(8);
1051 } else {fclose(stdout); setbuf(fout,_sobuf);}
1052# endif
1053 } else pperror("extraneous name %s", argv[i]);
1054 }
1055 }
1056
1057 fins[ifno]=fin;
1058 exfail = 0;
1059 /* after user -I files here are the standard include libraries */
1060# if unix
1061 dirs[nd++] = "/usr/include";
1062# endif
1063# if gcos
1064 dirs[nd++] = "cc/include";
1065# endif
1066# if ibm
1067# ifndef gimpel
1068 dirs[nd++] = "BTL$CLIB";
1069# endif
1070# endif
1071# ifdef gimpel
1072 dirs[nd++] = intss() ? "SYS3.C." : "" ;
1073# endif
1074 /* dirs[nd++] = "/compool"; */
1075 dirs[nd++] = 0;
1076 defloc=ppsym("define");
1077 udfloc=ppsym("undef");
1078 incloc=ppsym("include");
1079 elsloc=ppsym("else");
1080 eifloc=ppsym("endif");
1081 ifdloc=ppsym("ifdef");
1082 ifnloc=ppsym("ifndef");
1083 ifloc=ppsym("if");
1084 lneloc=ppsym("line");
1085 for (i=sizeof(macbit)/sizeof(macbit[0]); --i>=0; ) macbit[i]=0;
1086# if unix
1087 ysysloc=stsym("unix");
1088# endif
1089# if gcos
1090 ysysloc=stsym ("gcos");
1091# endif
1092# if ibm
1093 ysysloc=stsym ("ibm");
1094# endif
1095# if pdp11
1096 varloc=stsym("pdp11");
1097# endif
1098# if vax
1099 varloc=stsym("vax");
1100# endif
1101# if interdata
1102 varloc=stsym ("interdata");
1103# endif
1104# if tss
1105 varloc=stsym ("tss");
1106# endif
1107# if os
1108 varloc=stsym ("os");
1109# endif
1110# if mert
1111 varloc=stsym ("mert");
1112# endif
1113 ulnloc=stsym ("__LINE__");
1114 uflloc=stsym ("__FILE__");
1115
1116 tf=fnames[ifno]; fnames[ifno]="command line"; lineno[ifno]=1;
1117 cp2=prespc;
1118 while (cp2<predef) stsym(*cp2++);
1119 cp2=punspc;
1120 while (cp2<prund) {
1121 if (p=strdex(*cp2, '=')) *p++='\0';
1122 lookup(*cp2++, DROP);
1123 }
1124 fnames[ifno]=tf;
1125 pbeg=buffer+8; pbuf=pbeg+BUFSIZ; pend=pbuf+BUFSIZ;
1126
1127 trulvl = 0; flslvl = 0;
1128 lineno[0] = 1; sayline();
1129 outp=inp=pend;
1130 control(pend);
1131 return (exfail);
1132}