lex.c

unix v7是最后一个广泛发布的研究型UNIX版本

#include "defs"
#include "tokdefs"

# define BLANK	' '
# define MYQUOTE (2)
# define SEOF 0

/* card types */

# define STEOF 1
# define STINITIAL 2
# define STCONTINUE 3

/* lex states */

#define NEWSTMT	1
#define FIRSTTOKEN	2
#define OTHERTOKEN	3
#define RETEOS	4


LOCAL int stkey;
ftnint yystno;
LOCAL long int stno;
LOCAL long int nxtstno;
LOCAL int parlev;
LOCAL int expcom;
LOCAL int expeql;
LOCAL char *nextch;
LOCAL char *lastch;
LOCAL char *nextcd 	= NULL;
LOCAL char *endcd;
LOCAL int prevlin;
LOCAL int thislin;
LOCAL int code;
LOCAL int lexstate	= NEWSTMT;
LOCAL char s[1390];
LOCAL char *send	= s+20*66;
LOCAL int nincl	= 0;

struct inclfile
	{
	struct inclfile *inclnext;
	FILEP inclfp;
	char *inclname;
	int incllno;
	char *incllinp;
	int incllen;
	int inclcode;
	ftnint inclstno;
	} ;

LOCAL struct inclfile *inclp	=  NULL;
LOCAL struct keylist { char *keyname; int keyval; } ;
LOCAL struct punctlist { char punchar; int punval; };
LOCAL struct fmtlist { char fmtchar; int fmtval; };
LOCAL struct dotlist { char *dotname; int dotval; };
LOCAL struct keylist *keystart[26], *keyend[26];




inilex(name)
char *name;
{
nincl = 0;
inclp = NULL;
doinclude(name);
lexstate = NEWSTMT;
return(NO);
}



/* throw away the rest of the current line */
flline()
{
lexstate = RETEOS;
}



char *lexline(n)
ftnint *n;
{
*n = (lastch - nextch) + 1;
return(nextch);
}





doinclude(name)
char *name;
{
FILEP fp;
struct inclfile *t;

if(inclp)
	{
	inclp->incllno = thislin;
	inclp->inclcode = code;
	inclp->inclstno = nxtstno;
	if(nextcd)
		inclp->incllinp = copyn(inclp->incllen = endcd-nextcd , nextcd);
	else
		inclp->incllinp = 0;
	}
nextcd = NULL;

if(++nincl >= MAXINCLUDE)
	fatal("includes nested too deep");
if(name[0] == '\0')
	fp = stdin;
else
	fp = fopen(name, "r");
if( fp )
	{
	t = inclp;
	inclp = ALLOC(inclfile);
	inclp->inclnext = t;
	prevlin = thislin = 0;
	infname = inclp->inclname = name;
	infile = inclp->inclfp = fp;
	}
else
	{
	fprintf(diagfile, "Cannot open file %s", name);
	done(1);
	}
}




LOCAL popinclude()
{
struct inclfile *t;
register char *p;
register int k;

if(infile != stdin)
	clf(&infile);
free(infname);

--nincl;
t = inclp->inclnext;
free(inclp);
inclp = t;
if(inclp == NULL)
	return(NO);

infile = inclp->inclfp;
infname = inclp->inclname;
prevlin = thislin = inclp->incllno;
code = inclp->inclcode;
stno = nxtstno = inclp->inclstno;
if(inclp->incllinp)
	{
	endcd = nextcd = s;
	k = inclp->incllen;
	p = inclp->incllinp;
	while(--k >= 0)
		*endcd++ = *p++;
	free(inclp->incllinp);
	}
else
	nextcd = NULL;
return(YES);
}




yylex()
{
static int  tokno;

	switch(lexstate)
	{
case NEWSTMT :	/* need a new statement */
	if(getcds() == STEOF)
		return(SEOF);
	crunch();
	tokno = 0;
	lexstate = FIRSTTOKEN;
	yystno = stno;
	stno = nxtstno;
	toklen = 0;
	return(SLABEL);

first:
case FIRSTTOKEN :	/* first step on a statement */
	analyz();
	lexstate = OTHERTOKEN;
	tokno = 1;
	return(stkey);

case OTHERTOKEN :	/* return next token */
	if(nextch > lastch)
		goto reteos;
	++tokno;
	if((stkey==SLOGIF || stkey==SELSEIF) && parlev==0 && tokno>3) goto first;
	if(stkey==SASSIGN && tokno==3 && nextch<lastch &&
		nextch[0]=='t' && nextch[1]=='o')
			{
			nextch+=2;
			return(STO);
			}
	return(gettok());

reteos:
case RETEOS:
	lexstate = NEWSTMT;
	return(SEOS);
	}
fatal1("impossible lexstate %d", lexstate);
/* NOTREACHED */
}

LOCAL getcds()
{
register char *p, *q;

top:
	if(nextcd == NULL)
		{
		code = getcd( nextcd = s );
		stno = nxtstno;
		prevlin = thislin;
		}
	if(code == STEOF)
		if( popinclude() )
			goto top;
		else
			return(STEOF);

	if(code == STCONTINUE)
		{
		lineno = thislin;
		err("illegal continuation card ignored");
		nextcd = NULL;
		goto top;
		}

	if(nextcd > s)
		{
		q = nextcd;
		p = s;
		while(q < endcd)
			*p++ = *q++;
		endcd = p;
		}
	for(nextcd = endcd ;
		nextcd+66<=send && (code = getcd(nextcd))==STCONTINUE ;
		nextcd = endcd )
			;
	nextch = s;
	lastch = nextcd - 1;
	if(nextcd >= send)
		nextcd = NULL;
	lineno = prevlin;
	prevlin = thislin;
	return(STINITIAL);
}

LOCAL getcd(b)
register char *b;
{
register int c;
register char *p, *bend;
int speclin;
static char a[6];
static char *aend	= a+6;

top:
	endcd = b;
	bend = b+66;
	speclin = NO;

	if( (c = getc(infile)) == '&')
		{
		a[0] = BLANK;
		a[5] = 'x';
		speclin = YES;
		bend = send;
		}
	else if(c=='c' || c=='C' || c=='*')
		{
		while( (c = getc(infile)) != '\n')
			if(c == EOF)
				return(STEOF);
		++thislin;
		goto top;
		}

	else if(c != EOF)
		{
		/* a tab in columns 1-6 skips to column 7 */
		ungetc(c, infile);
		for(p=a; p<aend && (c=getc(infile)) != '\n' && c!=EOF; )
			if(c == '\t')
				{
				while(p < aend)
					*p++ = BLANK;
				speclin = YES;
				bend = send;
				}
			else
				*p++ = c;
		}
	if(c == EOF)
		return(STEOF);
	if(c == '\n')
		{
		while(p < aend)
			*p++ = BLANK;
		if( ! speclin )
			while(endcd < bend)
				*endcd++ = BLANK;
		}
	else	{	/* read body of line */
		while( endcd<bend && (c=getc(infile)) != '\n' && c!=EOF )
			*endcd++ = c;
		if(c == EOF)
			return(STEOF);
		if(c != '\n')
			{
			while( (c=getc(infile)) != '\n')
				if(c == EOF)
					return(STEOF);
			}

		if( ! speclin )
			while(endcd < bend)
				*endcd++ = BLANK;
		}
	++thislin;
	if( !isspace(a[5]) && a[5]!='0')
		return(STCONTINUE);
	for(p=a; p<aend; ++p)
		if( !isspace(*p) ) goto initline;
	for(p = b ; p<endcd ; ++p)
		if( !isspace(*p) ) goto initline;
	goto top;

initline:
	nxtstno = 0;
	for(p = a ; p<a+5 ; ++p)
		if( !isspace(*p) )
			if(isdigit(*p))
				nxtstno = 10*nxtstno + (*p - '0');
			else	{
				lineno = thislin;
				err("nondigit in statement number field");
				nxtstno = 0;
				break;
				}
	return(STINITIAL);
}

LOCAL crunch()
{
register char *i, *j, *j0, *j1, *prvstr;
int ten, nh, quote;

/* i is the next input character to be looked at
j is the next output character */
parlev = 0;
expcom = 0;	/* exposed ','s */
expeql = 0;	/* exposed equal signs */
j = s;
prvstr = s;
for(i=s ; i<=lastch ; ++i)
	{
	if(isspace(*i) )
		continue;
	if(*i=='\'' ||  *i=='"')
		{
		quote = *i;
		*j = MYQUOTE; /* special marker */
		for(;;)
			{
			if(++i > lastch)
				{
				err("unbalanced quotes; closing quote supplied");
				break;
				}
			if(*i == quote)
				if(i<lastch && i[1]==quote) ++i;
				else break;
			else if(*i=='\\' && i<lastch)
				switch(*++i)
					{
					case 't':
						*i = '\t'; break;
					case 'b':
						*i = '\b'; break;
					case 'n':
						*i = '\n'; break;
					case 'f':
						*i = '\f'; break;
					case '0':
						*i = '\0'; break;
					default:
						break;
					}
			*++j = *i;
			}
		j[1] = MYQUOTE;
		j += 2;
		prvstr = j;
		}
	else if( (*i=='h' || *i=='H')  && j>prvstr)	/* test for Hollerith strings */
		{
		if( ! isdigit(j[-1])) goto copychar;
		nh = j[-1] - '0';
		ten = 10;
		j1 = prvstr - 1;
		if (j1<j-5) j1=j-5;
		for(j0=j-2 ; j0>j1; -- j0)
			{
			if( ! isdigit(*j0 ) ) break;
			nh += ten * (*j0-'0');
			ten*=10;
			}
		if(j0 <= j1) goto copychar;
/* a hollerith must be preceded by a punctuation mark.
   '*' is possible only as repetition factor in a data statement
   not, in particular, in character*2h
*/

		if( !(*j0=='*'&&s[0]=='d') && *j0!='/' && *j0!='(' &&
			*j0!=',' && *j0!='=' && *j0!='.')
				goto copychar;
		if(i+nh > lastch)
			{
			err1("%dH too big", nh);
			nh = lastch - i;
			}
		j0[1] = MYQUOTE; /* special marker */
		j = j0 + 1;
		while(nh-- > 0)
			{
			if(*++i == '\\')
				switch(*++i)
					{
					case 't':
						*i = '\t'; break;
					case 'b':