sed.c

这是一个同样来自贝尔实验室的和UNIX有着渊源的操作系统, 其简洁的设计和实现易于我们学习和理解

				c = '\n';	/* fake an end-of-line */
				prog.curr = 0;
			}
		} else 
			c = -1;
	} else if ((c = Bgetrune(prog.bp)) < 0)
			Bterm(prog.bp);
	return c;
}

void
address(Addr *ap)
{
	int c;
	long	lno;

	if((c = *cp++) == '$')
		ap->type = A_DOL;
	else if(c == '/') {
		seof = c;
		if (ap->rp = compile())
			ap->type = A_RE;
		else
			ap->type = A_LAST;
	}
	else if (c >= '0' && c <= '9') {
		lno = c-'0';
		while ((c = *cp) >= '0' && c <= '9')
			lno = lno*10 + *cp++-'0';
		if(!lno)
			quit("line number 0 is illegal",0);
		ap->type = A_LINE;
		ap->line = lno;
	}
	else {
		cp--;
		ap->type = A_NONE;
	}
}

cmp(char *a, char *b)		/* compare characters */
{
	while(*a == *b++)
		if (*a == '\0')
			return(0);
		else a++;
	return(1);
}
rcmp(Rune *a, Rune *b)		/* compare runes */
{
	while(*a == *b++)
		if (*a == '\0')
			return(0);
		else a++;
	return(1);
}

char *
text(char *p)		/* extract character string */
{
	Rune	r;

	while(*cp == '\t' || *cp == ' ')
			cp++;
	while (*cp) {
		if ((r = *cp++) == '\\')
			if ((r = *cp++) == 0)
				break;;
		if (r == '\n')
			while (*cp == '\t' || *cp == ' ')
					cp++;
		p += runetochar(p, &r);
	}
	*p++ = '\0';
	return p;
}

Rune *
stext(Rune *p, Rune *end)		/* extract rune string */
{
	while(*cp == '\t' || *cp == ' ')
		cp++;
	while (*cp) {
		if (*cp == '\\')
			if (*++cp == 0)
				break;
		if (p >= end-1)
			quit(TMMES, (char *) linebuf);
		if ((*p++ = *cp++) == '\n')
			while(*cp == '\t' || *cp == ' ')
					cp++;
	}
	*p++ = 0;
	return p;
}


Label *
search (Label *ptr)
{
	Label	*rp;

	for (rp = ltab; rp < ptr; rp++)
		if(rcmp(rp->asc, ptr->asc) == 0)
			return(rp);
	return(0);
}

void
dechain(void)
{
	Label	*lptr;
	SedCom	*rptr, *trptr;

	for(lptr = ltab; lptr < lab; lptr++) {

		if(lptr->address == 0)
			quit("Undefined label: %S", (char *) lptr->asc);

		if(lptr->chain) {
			rptr = lptr->chain;
			while(trptr = rptr->lb1) {
				rptr->lb1 = lptr->address;
				rptr = trptr;
			}
			rptr->lb1 = lptr->address;
		}
	}
}

int
ycomp(SedCom *r)
{
	int 	i;
	Rune	*rp;
	Rune	c, *tsp, highc;
	Rune	*sp;

	highc = 0;
	for(tsp = cp; *tsp != seof; tsp++) {
		if(*tsp == '\\')
			tsp++;
		if(*tsp == '\n' || *tsp == '\0')
			return(0);
		if (*tsp > highc) highc = *tsp;
	}
	tsp++;
	if ((rp = r->text = (Rune *) malloc(sizeof(Rune)*(highc+2))) == 0)
		quit("Out of memory", 0);
	*rp++ = highc;				/* save upper bound */
	for (i = 0; i <= highc; i++)
		rp[i] = i;
	sp = cp;
	while((c = *sp++) != seof) {
		if(c == '\\' && *sp == 'n') {
			sp++;
			c = '\n';
		}
		if((rp[c] = *tsp++) == '\\' && *tsp == 'n') {
			rp[c] = '\n';
			tsp++;
		}
		if(rp[c] == seof || rp[c] == '\0') {
			free(r->re1);
			r->re1 = 0;
			return(0);
		}
	}
	if(*tsp != seof) {
		free(r->re1);
		r->re1 = 0;
		return(0);
	}
	cp = tsp+1;
	return(1);
}

void
execute(void)
{
	SedCom	*ipc;

	while (spend = gline(linebuf)){
		for(ipc = pspace; ipc->command; ) {
			if (!executable(ipc)) {
				ipc++;
				continue;
			}
			command(ipc);

			if(delflag)
				break;
			if(jflag) {
				jflag = 0;
				if((ipc = ipc->lb1) == 0)
					break;
			} else
				ipc++;

		}
		if(!nflag && !delflag)
			putline(&fout, linebuf, spend-linebuf);
		if(aptr > abuf) {
			arout();
		}
		delflag = 0;
	}
}
	/* determine if a statement should be applied to an input line */
int
executable(SedCom *ipc)
{
	if (ipc->active) {	/* Addr1 satisfied - accept until Addr2 */
		if (ipc->active == 1)		/* Second line */
			ipc->active = 2;
		switch(ipc->ad2.type) {
			case A_NONE:	/* No second addr; use first */
				ipc->active = 0;
				break;
			case A_DOL:	/* Accept everything */
				return !ipc->negfl;
			case A_LINE:	/* Line at end of range? */
				if (lnum <= ipc->ad2.line) {
					if (ipc->ad2.line == lnum)
						ipc->active = 0;
					return !ipc->negfl;
				}
				ipc->active = 0;	/* out of range */
				return ipc->negfl;
			case A_RE:	/* Check for matching R.E. */
				if (match(ipc->ad2.rp, linebuf))
					ipc->active = 0;
				return !ipc->negfl;
			default:		/* internal error */
				quit("Internal error", 0);
		}
	}
	switch (ipc->ad1.type) {	/* Check first address */
		case A_NONE:			/* Everything matches */
			return !ipc->negfl;
		case A_DOL:			/* Only last line */
			if (dolflag)
				return !ipc->negfl;
			break;
		case A_LINE:			/* Check line number */
			if (ipc->ad1.line == lnum) {
				ipc->active = 1;	/* In range */
				return !ipc->negfl;
			}
			break;
		case A_RE:			/* Check R.E. */
			if (match(ipc->ad1.rp, linebuf)) {
				ipc->active = 1;	/* In range */
				return !ipc->negfl;
			}
			break;
		default:
			quit("Internal error", 0);
	}
	return ipc->negfl;
}
match(Reprog *pattern, Rune *buf)
{
	if (!pattern)
		return 0;
	subexp[0].rsp = buf; 
	subexp[0].ep = 0;
	if (rregexec(pattern, linebuf, subexp, MAXSUB)) {
		loc1 = subexp[0].rsp;
		loc2 = subexp[0].rep;
		return 1;
	}
	loc1 = loc2 = 0;
	return 0;
}
substitute(SedCom *ipc)
{
	int len;

	if(!match(ipc->re1, linebuf))
		return 0;

	/*
	 * we have at least one match.  some patterns, e.g. '$' or '^', can
	 * produce zero-length matches, so during a global substitute we
	 * must bump to the character after a zero-length match to keep from looping.
	 */
	sflag = 1;
	if(ipc->gfl == 0)		/* single substitution */
		dosub(ipc->rhs);
	else
	do{				/* global substitution */
		len = loc2-loc1;	/* length of match */
		dosub(ipc->rhs);	/* dosub moves loc2 */
		if(*loc2 == 0)		/* end of string */
			break;
		if(len == 0)		/* zero-length R.E. match */
			loc2++;		/* bump over zero-length match */
		if(*loc2 == 0)		/* end of string */
			break;
	} while(match(ipc->re1, loc2));
	return 1;
}

void
dosub(Rune *rhsbuf)
{
	Rune *lp, *sp;
	Rune *rp;
	int c, n;

	lp = linebuf;
	sp = genbuf;
	rp = rhsbuf;
	while (lp < loc1)
		*sp++ = *lp++;
	while(c = *rp++) {
		if (c == '&') {
			sp = place(sp, loc1, loc2);
			continue;
		}
		if (c == 0xFFFF && (c = *rp++) >= '1' && c < MAXSUB+'0') {
			n = c-'0';
			if (subexp[n].rsp && subexp[n].rep) {
				sp = place(sp, subexp[n].rsp, subexp[n].rep);
				continue;
			}
			else {
				fprint(2, "sed: Invalid back reference \\%d\n",n);
				errexit();
			}
		}
		*sp++ = c;
		if (sp >= &genbuf[LBSIZE])
			fprint(2, "sed: Output line too long.\n");
	}
	lp = loc2;
	loc2 = sp - genbuf + linebuf;
	while (*sp++ = *lp++)
		if (sp >= &genbuf[LBSIZE])
			fprint(2, "sed: Output line too long.\n");
	lp = linebuf;
	sp = genbuf;
	while (*lp++ = *sp++)
		;
	spend = lp-1;
}

Rune *
place(Rune *sp, Rune *l1, Rune *l2)
{
	while (l1 < l2) {
		*sp++ = *l1++;
		if (sp >= &genbuf[LBSIZE])
			fprint(2, "sed: Output line too long.\n");
	}
	return(sp);
}

char *
trans(int c)
{
	static char buf[] = "\\x0000";
	static char hex[] = "0123456789abcdef";

	switch(c) {
		case '\b':
			return "\\b";
		case '\n':
			return "\\n";
		case '\r':
			return "\\r";
		case '\t':
			return "\\t";
		case '\\':
			return "\\\\";
	}
	buf[2] = hex[(c>>12)&0xF];
	buf[3] = hex[(c>>8)&0xF];
	buf[4] = hex[(c>>4)&0xF];
	buf[5] = hex[c&0xF];
	return buf;
}

void
command(SedCom *ipc)
{
	int	i, c;
	Rune	*p1, *p2;
	char	*ucp;
	Rune	*rp;
	Rune	*execp;

	switch(ipc->command) {

		case ACOM:
			*aptr++ = ipc;
			if(aptr >= abuf+MAXADDS) {
				quit("sed: Too many appends after line %ld\n",
					(char *) lnum);
			}
			*aptr = 0;
			break;
		case CCOM:
			delflag = 1;
			if(ipc->active == 1) {
				for(rp = ipc->text; *rp; rp++)
					Bputrune(&fout, *rp);
				Bputc(&fout, '\n');
			}
			break;
		case DCOM:
			delflag++;
			break;
		case CDCOM:
			p1 = p2 = linebuf;
			while(*p1 != '\n') {
				if(*p1++ == 0) {
					delflag++;
					return;
				}
			}
			p1++;
			while(*p2++ = *p1++)
				;
			spend = p2-1;
			jflag++;
			break;
		case EQCOM:
			Bprint(&fout, "%ld\n", lnum);
			break;
		case GCOM:
			p1 = linebuf;
			p2 = holdsp;
			while(*p1++ = *p2++)
				;
			spend = p1-1;
			break;
		case CGCOM:
			*spend++ = '\n';
			p1 = spend;
			p2 = holdsp;
			while(*p1++ = *p2++)
				if(p1 >= lbend)
					break;
			spend = p1-1;
			break;
		case HCOM:
			p1 = holdsp;
			p2 = linebuf;
			while(*p1++ = *p2++);
			hspend = p1-1;
			break;
		case CHCOM:
			*hspend++ = '\n';
			p1 = hspend;
			p2 = linebuf;
			while(*p1++ = *p2++)
				if(p1 >= hend)
					break;
			hspend = p1-1;
			break;
		case ICOM:
			for(rp = ipc->text; *rp; rp++)
				Bputrune(&fout, *rp);
			Bputc(&fout, '\n');
			break;
		case BCOM:
			jflag = 1;
			break;
		case LCOM:
			c = 0;
			for (i = 0, rp = linebuf; *rp; rp++) {
				c = *rp;
				if(c >= 0x20 && c < 0x7F && c != '\\') {
					Bputc(&fout, c);
					if(i++ > 71) {
						Bprint(&fout, "\\\n");
						i = 0;
					}
				} else {
					for (ucp = trans(*rp); *ucp; ucp++){
						c = *ucp;
						Bputc(&fout, c);
						if(i++ > 71) {
							Bprint(&fout, "\\\n");
							i = 0;
						}
					}
				}
			}
			if(c == ' ')
				Bprint(&fout, "\\n");
			Bputc(&fout, '\n');
			break;
		case NCOM:
			if(!nflag)
				putline(&fout, linebuf, spend-linebuf);

			if(aptr > abuf)
				arout();
			if((execp = gline(linebuf)) == 0) {
				delflag = 1;
				break;
			}
			spend = execp;
			break;
		case CNCOM:
			if(aptr > abuf)
				arout();
			*spend++ = '\n';
			if((execp = gline(spend)) == 0) {
				delflag = 1;
				break;
			}
			spend = execp;
			break;
		case PCOM:
			putline(&fout, linebuf, spend-linebuf);
			break;
		case CPCOM:
	cpcom:
			for(rp = linebuf; *rp && *rp != '\n'; rp++)
				Bputc(&fout, *rp);
			Bputc(&fout, '\n');
			break;
		case QCOM:
			if(!nflag)
				putline(&fout, linebuf, spend-linebuf);
			if(aptr > abuf)
				arout();
			exits(0);
		case RCOM:
			*aptr++ = ipc;
			if(aptr >= &abuf[MAXADDS])
				quit("sed: Too many reads after line %ld\n",
					(char *) lnum);
			*aptr = 0;
			break;
		case SCOM:
			i = substitute(ipc);
			if(i && ipc->pfl)
				if(ipc->pfl == 1)
					putline(&fout, linebuf, spend-linebuf);
				else
					goto cpcom;
			if(i && ipc->fcode)
				goto wcom;
			break;

		case TCOM:
			if(sflag == 0)	break;
			sflag = 0;
			jflag = 1;
			break;

		wcom:
		case WCOM:
			putline(ipc->fcode,linebuf, spend-linebuf);
			break;
		case XCOM:
			p1 = linebuf;
			p2 = genbuf;
			while(*p2++ = *p1++);
			p1 = holdsp;
			p2 = linebuf;
			while(*p2++ = *p1++);
			spend = p2 - 1;
			p1 = genbuf;
			p2 = holdsp;
			while(*p2++ = *p1++);
			hspend = p2 - 1;
			break;
		case YCOM:
			p1 = linebuf;
			p2 = ipc->text;
			for (i = *p2++;	*p1; p1++){
				if (*p1 <= i) *p1 = p2[*p1];
			}
			break;
	}

}

void
putline(Biobuf *bp, Rune *buf, int n)
{
	while (n--)
		Bputrune(bp, *buf++);
	Bputc(bp, '\n');
}
ecmp(Rune *a, Rune *b, int count)
{
	while(count--)
		if(*a++ != *b++)	return(0);
	return(1);
}

void
arout(void)
{
	Rune	*p1;
	Biobuf	*fi;
	int	c;
	char	*s;
	char	buf[128];

	for (aptr = abuf; *aptr; aptr++) {
		if((*aptr)->command == ACOM) {
			for(p1 = (*aptr)->text; *p1; p1++ )
				Bputrune(&fout, *p1);
			Bputc(&fout, '\n');
		} else {
			for(s = buf, p1= (*aptr)->text; *p1; p1++)
					s += runetochar(s, p1);
			*s = '\0';
			if((fi = Bopen(buf, OREAD)) == 0)
				continue;
			while((c = Bgetc(fi)) >= 0)
				Bputc(&fout, c);
			Bterm(fi);
		}
	}
	aptr = abuf;
	*aptr = 0;
}

void
errexit(void)
{
	exits("error");
}

void
quit (char *msg, char *arg)
{
	fprint(2, "sed: ");
	fprint(2, msg, arg);
	fprint(2, "\n");
	errexit();
}

Rune *
gline(Rune *addr)
{
	long	c;
	Rune *p;

	static long peekc = 0;

	if (f == 0 && opendata() < 0)
		return 0;
	sflag = 0;
	lnum++;
/*	Bflush(&fout);********* dumped 4/30/92 - bobf****/
	do {
		p = addr;
		for (c = (peekc ? peekc : Bgetrune(f)); c >= 0; c = Bgetrune(f)) {
			if (c == '\n') {
				if ((peekc = Bgetrune(f)) < 0) {
					if (fhead == 0)
						dolflag = 1;
				}
				*p = '\0';
				return p;
			}
			if (c && p < lbend)
				*p++ = c;
		}
		/* return partial final line, adding implicit newline */
		if(p != addr) {
			*p = '\0';
			peekc = -1;
			if (fhead == 0)
				dolflag = 1;
			return p;
		}
		peekc = 0;
		Bterm(f);
	} while (opendata() > 0);	/* Switch to next stream */
	f = 0;
	return 0;
}

	/* Data file input section - the intent is to transparently
	 *	catenate all data input streams.
	 */
void
enroll(char *filename)		/* Add a file to the input file cache */
{
	FileCache *fp;

	if ((fp = (FileCache *) malloc(sizeof (FileCache))) == 0)
		quit("Out of memory", 0);
	if (ftail == 0)
		fhead = fp;
	else
		ftail->next = fp;
	ftail = fp;
	fp->next = 0;
	fp->name = filename;	/* 0 => stdin */
}

int
opendata(void)
{
	if (fhead == 0)
		return -1;
	if (fhead->name) {
		if ((f = Bopen(fhead->name, OREAD)) == 0)
			quit("Can't open %s", fhead->name);
	} else {
		Binit(&stdin, 0, OREAD);
		f = &stdin;
	}
	fhead = fhead->next;
	return 1;
}