cpp.c

unix v7是最后一个广泛发布的研究型UNIX版本

#
# include "stdio.h"
/* C command
/* written by John F. Reiser
/* July/August 1978
*/

#define STATIC

#define STDIN 0
#define STDOUT 1
#define STDERR 2
#define READ 0
#define WRITE 1
#define SALT '#'
#ifndef BUFSIZ
#define BUFSIZ 512
#endif

char *pbeg,*pbuf,*pend;
char *outp,*inp;
char *newp;
char cinit;

/* some code depends on whether characters are sign or zero extended */
/*	#if '\377' < 0		not used here, old cpp doesn't understand */
#if pdp11 | vax
#define COFF 128
#else
#define COFF 0
#endif

# if gcos
#define ALFSIZ 512	/* alphabet size */
# else
#define ALFSIZ 256	/* alphabet size */
# endif
char macbit[ALFSIZ+11];
char toktyp[ALFSIZ];
#define BLANK 1
#define IDENT 2
#define NUMBR 3

/* a superimposed code is used to reduce the number of calls to the
/* symbol table lookup routine.  (if the kth character of an identifier
/* is 'a' and there are no macro names whose kth character is 'a'
/* then the identifier cannot be a macro name, hence there is no need
/* to look in the symbol table.)  'scw1' enables the test based on
/* single characters and their position in the identifier.  'scw2'
/* enables the test based on adjacent pairs of characters and their
/* position in the identifier.  scw1 typically costs 1 indexed fetch,
/* an AND, and a jump per character of identifier, until the identifier
/* is known as a non-macro name or until the end of the identifier.
/* scw1 is inexpensive.  scw2 typically costs 4 indexed fetches,
/* an add, an AND, and a jump per character of identifier, but it is also
/* slightly more effective at reducing symbol table searches.
/* scw2 usually costs too much because the symbol table search is
/* usually short; but if symbol table search should become expensive,
/* the code is here.
/* using both scw1 and scw2 is of dubious value.
*/
#define scw1 1
#define scw2 0

#if scw2
char t21[ALFSIZ],t22[ALFSIZ],t23[ALFSIZ+8];
#endif

#if scw1
#define b0 1
#define b1 2
#define b2 4
#define b3 8
#define b4 16
#define b5 32
#define b6 64
#define b7 128
#endif

#define IB 1
#define SB 2
#define NB 4
#define CB 8
#define QB 16
#define WB 32
char fastab[ALFSIZ];
char slotab[ALFSIZ];
char *ptrtab;
#define isslo (ptrtab==(slotab+COFF))
#define isid(a)  ((fastab+COFF)[a]&IB)
#define isspc(a) (ptrtab[a]&SB)
#define isnum(a) ((fastab+COFF)[a]&NB)
#define iscom(a) ((fastab+COFF)[a]&CB)
#define isquo(a) ((fastab+COFF)[a]&QB)
#define iswarn(a) ((fastab+COFF)[a]&WB)

#define eob(a) ((a)>=pend)
#define bob(a) (pbeg>=(a))

char buffer[8+BUFSIZ+BUFSIZ+8];

# define SBSIZE 12000
char	sbf[SBSIZE];
char	*savch	= sbf;

# define DROP 0xFE	/* special character not legal ASCII or EBCDIC */
# define WARN DROP
# define SAME 0
# define MAXINC 10
# define MAXFRE 14	/* max buffers of macro pushback */
# define MAXFRM 31	/* max number of formals/actuals to a macro */

static char warnc = WARN;

int mactop,fretop;
char *instack[MAXFRE],*bufstack[MAXFRE],*endbuf[MAXFRE];

int plvl;	/* parenthesis level during scan for macro actuals */
int maclin;	/* line number of macro call requiring actuals */
char *macfil;	/* file name of macro call requiring actuals */
char *macnam;	/* name of macro requiring actuals */
int maclvl;	/* # calls since last decrease in nesting level */
char *macforw;	/* pointer which must be exceeded to decrease nesting level */
int macdam;	/* offset to macforw due to buffer shifting */

#if tgp
int tgpscan;	/* flag for dump(); */
#endif

STATIC	int	inctop[MAXINC];
STATIC	char	*fnames[MAXINC];
STATIC	char	*dirnams[MAXINC];	/* actual directory of #include files */
STATIC	int	fins[MAXINC];
STATIC	int	lineno[MAXINC];

STATIC	char	*dirs[10];	/* -I and <> directories */
char *strdex(), *copy(), *subst(), *trmdir();
struct symtab *stsym();
STATIC	int	fin	= STDIN;
STATIC	FILE	*fout	= stdout;
STATIC	int	nd	= 1;
STATIC	int	pflag;	/* don't put out lines "# 12 foo.c" */
STATIC	int	passcom;	/* don't delete comments */
STATIC	int rflag;	/* allow macro recursion */
STATIC	int	ifno;
# define NPREDEF 20
STATIC	char *prespc[NPREDEF];
STATIC	char **predef = prespc;
STATIC	char *punspc[NPREDEF];
STATIC	char **prund = punspc;
STATIC	int	exfail;
struct symtab {
	char	*name;
	char	*value;
} *lastsym, *lookup(), *slookup();

# if gcos
#include <setjmp.h>
static jmp_buf env;
# define main	mainpp
# undef exit
# define exit(S)	longjmp(env, 1)
# define open(S,D)	fileno(fopen(S, "r"))
# define close(F)	fclose(_f[F])
extern FILE *_f[];
# define symsiz 500
# else
# define symsiz 400
# endif
STATIC	struct symtab stab[symsiz];

STATIC	struct symtab *defloc;
STATIC	struct symtab *udfloc;
STATIC	struct symtab *incloc;
STATIC	struct symtab *ifloc;
STATIC	struct symtab *elsloc;
STATIC	struct symtab *eifloc;
STATIC	struct symtab *ifdloc;
STATIC	struct symtab *ifnloc;
STATIC	struct symtab *ysysloc;
STATIC	struct symtab *varloc;
STATIC	struct symtab *lneloc;
STATIC	struct symtab *ulnloc;
STATIC	struct symtab *uflloc;
STATIC	int	trulvl;
STATIC	int	flslvl;

sayline() {
	if (pflag==0) fprintf(fout,"# %d \"%s\"\n", lineno[ifno], fnames[ifno]);
}

/* data structure guide
/*
/* most of the scanning takes place in the buffer:
/*
/*  (low address)                                             (high address)
/*  pbeg                           pbuf                                 pend
/*  |      <-- BUFSIZ chars -->      |         <-- BUFSIZ chars -->        |
/*  _______________________________________________________________________
/* |_______________________________________________________________________|
/*          |               |               |
/*          |<-- waiting -->|               |<-- waiting -->
/*          |    to be      |<-- current -->|    to be
/*          |    written    |    token      |    scanned
/*          |               |               |
/*          outp            inp             p
/*
/*  *outp   first char not yet written to output file
/*  *inp    first char of current token
/*  *p      first char not yet scanned
/*
/* macro expansion: write from *outp to *inp (chars waiting to be written),
/* ignore from *inp to *p (chars of the macro call), place generated
/* characters in front of *p (in reverse order), update pointers,
/* resume scanning.
/*
/* symbol table pointers point to just beyond the end of macro definitions;
/* the first preceding character is the number of formal parameters.
/* the appearance of a formal in the body of a definition is marked by
/* 2 chars: the char WARN, and a char containing the parameter number.
/* the first char of a definition is preceded by a zero character.
/*
/* when macro expansion attempts to back up over the beginning of the
/* buffer, some characters preceding *pend are saved in a side buffer,
/* the address of the side buffer is put on 'instack', and the rest
/* of the main buffer is moved to the right.  the end of the saved buffer
/* is kept in 'endbuf' since there may be nulls in the saved buffer.
/*
/* similar action is taken when an 'include' statement is processed,
/* except that the main buffer must be completely emptied.  the array
/* element 'inctop[ifno]' records the last side buffer saved when
/* file 'ifno' was included.  these buffers remain dormant while
/* the file is being read, and are reactivated at end-of-file.
/*
/* instack[0 : mactop] holds the addresses of all pending side buffers.
/* instack[inctop[ifno]+1 : mactop-1] holds the addresses of the side
/* buffers which are "live"; the side buffers instack[0 : inctop[ifno]]
/* are dormant, waiting for end-of-file on the current file.
/*
/* space for side buffers is obtained from 'savch' and is never returned.
/* bufstack[0:fretop-1] holds addresses of side buffers which
/* are available for use.
*/

dump() {
/* write part of buffer which lies between  outp  and  inp .
/* this should be a direct call to 'write', but the system slows to a crawl
/* if it has to do an unaligned copy.  thus we buffer.  this silly loop
/* is 15% of the total time, thus even the 'putc' macro is too slow.
*/
	register char *p1,*p2; register FILE *f;
	if ((p1=outp)==inp || flslvl!=0) return;
#if tgp
#define MAXOUT 80
	if (!tgpscan) {/* scan again to insure <= MAXOUT chars between linefeeds */
		register char c,*pblank; char savc,stopc,brk;
		tgpscan=1; brk=stopc=pblank=0; p2=inp; savc= *p2; *p2='\0';
		while (c= *p1++) {
			if (c=='\\') c= *p1++;
			if (stopc==c) stopc=0;
			else if (c=='"' || c=='\'') stopc=c;
			if (p1-outp>MAXOUT && pblank!=0) {
				*pblank++='\n'; inp=pblank; dump(); brk=1; pblank=0;
			}
			if (c==' ' && stopc==0) pblank=p1-1;
		}
		if (brk) sayline();
		*p2=savc; inp=p2; p1=outp; tgpscan=0;
	}
#endif
	f=fout;
# if gcos
/* filter out "$ program c" card if first line of input */
/* gmatch is a simple pattern matcher in the GCOS Standard Library */
{	static int gmfirst = 0;
	if (!gmfirst) {
		++gmfirst;
		if (gmatch(p1, "^$*program[ \t]*c*"))
			p1 = strdex(p1, '\n');
	}
}
# endif
	while (p1<inp) putc(*p1++,f);
	outp=p1;
}

char *
refill(p) register char *p; {
/* dump buffer.  save chars from inp to p.  read into buffer at pbuf,
/* contiguous with p.  update pointers, return new p.
*/
	register char *np,*op; register int ninbuf;
	dump(); np=pbuf-(p-inp); op=inp;
	if (bob(np+1)) {pperror("token too long"); np=pbeg; p=inp+BUFSIZ;}
	macdam += np-inp; outp=inp=np;
	while (op<p) *np++= *op++;
	p=np;
	for (;;) {
		if (mactop>inctop[ifno]) {/* retrieve hunk of pushed-back macro text */
			op=instack[--mactop]; np=pbuf;
			do {while (*np++= *op++);} while (op<endbuf[mactop]); pend=np-1;
			/* make buffer space avail for 'include' processing */
			if (fretop<MAXFRE) bufstack[fretop++]=instack[mactop];
			return(p);
		} else {/* get more text from file(s) */
			maclvl=0;
			if (0<(ninbuf=read(fin,pbuf,BUFSIZ))) {
				pend=pbuf+ninbuf; *pend='\0';
				return(p);
			}
			/* end of #include file */
			if (ifno==0) {/* end of input */
				if (plvl!=0) {
					int n=plvl,tlin=lineno[ifno]; char *tfil=fnames[ifno];
					lineno[ifno]=maclin; fnames[ifno]=macfil;
					pperror("%s: unterminated macro call",macnam);
					lineno[ifno]=tlin; fnames[ifno]=tfil;
					np=p; *np++='\n';	/* shut off unterminated quoted string */
					while (--n>=0) *np++=')';	/* supply missing parens */
					pend=np; *np='\0'; if (plvl<0) plvl=0;
					return(p);
				}
				inp=p; dump(); exit(exfail);
			}
			close(fin); fin=fins[--ifno]; dirs[0]=dirnams[ifno]; sayline();
		}
	}
}

#define BEG 0
#define LF 1

char *
cotoken(p) register char *p; {
	register int c,i; char quoc;
	static int state = BEG;

	if (state!=BEG) goto prevlf;
for (;;) {
again:
	while (!isspc(*p++));
	switch (*(inp=p-1)) {
	case 0: {
		if (eob(--p)) {p=refill(p); goto again;}
		else ++p; /* ignore null byte */
	} break;
	case '|': case '&': for (;;) {/* sloscan only */
		if (*p++== *inp) break;
		if (eob(--p)) p=refill(p);
		else break;
	} break;
	case '=': case '!': for (;;) {/* sloscan only */
		if (*p++=='=') break;
		if (eob(--p)) p=refill(p);
		else break;
	} break;
	case '<': case '>': for (;;) {/* sloscan only */
		if (*p++=='=' || p[-2]==p[-1]) break;
		if (eob(--p)) p=refill(p);
		else break;
	} break;
	case '\\': for (;;) {
		if (*p++=='\n') {++lineno[ifno]; break;}
		if (eob(--p)) p=refill(p);
		else {++p; break;}
	} break;
	case '/': for (;;) {
		if (*p++=='*') {/* comment */
			if (!passcom) {inp=p-2; dump(); ++flslvl;}
			for (;;) {
				while (!iscom(*p++));
				if (p[-1]=='*') for (;;) {
					if (*p++=='/') goto endcom;
					if (eob(--p)) {
						if (!passcom) {inp=p; p=refill(p);}
						else if ((p-inp)>=BUFSIZ) {/* split long comment */
							inp=p; p=refill(p);	/* last char written is '*' */
							putc('/',fout);	/* terminate first part */