lex.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

#include	"cc.h"
#include	"y.tab.h"

#ifndef	CPP
#define	CPP	"/bin/cpp"
#endif

/*
 * known debug flags
 *	-a		acid declaration output
 *	-A		!B
 *	-B		non ANSI
 *	-d		print declarations
 *	-D name		define
 *	-F		format specification check
 *	-i		print initialization
 *	-I path		include
 *	-l		generate little-endian code
 *	-L		print every NAME symbol
 *	-M		constant multiplication
 *	-m		print add/sub/mul trees
 *	-n		print acid to file (%.c=%.acid) (with -a or -aa)
 *	-o file		output file
 *	-p		use standard cpp ANSI preprocessor (not on windows)
 *	-r		print registerization
 *	-s		print structure offsets (with -a or -aa)
 *	-S		print assembly
 *	-t		print type trees
 *	-V		enable void* conversion warnings
 *	-v		verbose printing
 *	-w		print warnings
 *	-X		abort on error
 */

void
main(int argc, char *argv[])
{
	char *defs[50], *p;
	int nproc, nout, status, i, c, ndef;

	memset(debug, 0, sizeof(debug));
	tinit();
	cinit();
	ginit();
	arginit();

	profileflg = 1;	/* #pragma can turn it off */
	tufield = simplet((1L<<tfield->etype) | BUNSIGNED);
	ndef = 0;
	outfile = 0;
	include[ninclude++] = ".";
	ARGBEGIN {
	default:
		c = ARGC();
		if(c >= 0 && c < sizeof(debug))
			debug[c]++;
		break;

	case 'l':			/* for little-endian mips */
		if(thechar != 'v'){
			print("can only use -l with vc");
			errorexit();
		}
		thechar = '0';
		thestring = "spim";
		break;

	case 'o':
		outfile = ARGF();
		break;

	case 'D':
		p = ARGF();
		if(p) {
			defs[ndef++] = p;
			dodefine(p);
		}
		break;

	case 'I':
		p = ARGF();
		setinclude(p);
		break;
	} ARGEND
	if(argc < 1 && outfile == 0) {
		print("usage: %cc [-options] files\n", thechar);
		errorexit();
	}
	if(argc > 1 && systemtype(Windows)){
		print("can't compile multiple files on windows\n");
		errorexit();
	}
	if(argc > 1 && !systemtype(Windows)) {
		nproc = 1;
		/*
		 * if we're writing acid to standard output, don't compile
		 * concurrently, to avoid interleaving output.
		 */
		if(((!debug['a'] && !debug['Z']) || debug['n']) &&
		    (p = getenv("NPROC")) != nil)
			nproc = atol(p);	/* */
		c = 0;
		nout = 0;
		for(;;) {
			while(nout < nproc && argc > 0) {
				i = myfork();
				if(i < 0) {
					i = mywait(&status);
					if(i < 0) {
						print("cannot create a process\n");
						errorexit();
					}
					if(status)
						c++;
					nout--;
					continue;
				}
				if(i == 0) {
					fprint(2, "%s:\n", *argv);
					if (compile(*argv, defs, ndef))
						errorexit();
					exits(0);
				}
				nout++;
				argc--;
				argv++;
			}
			i = mywait(&status);
			if(i < 0) {
				if(c)
					errorexit();
				exits(0);
			}
			if(status)
				c++;
			nout--;
		}
	}

	if(argc == 0)
		c = compile("stdin", defs, ndef);
	else
		c = compile(argv[0], defs, ndef);

	if(c)
		errorexit();
	exits(0);
}

int
compile(char *file, char **defs, int ndef)
{
	char ofile[400], incfile[20];
	char *p, *av[100], opt[256];
	int i, c, fd[2];
	static int first = 1;

	strcpy(ofile, file);
	p = utfrrune(ofile, pathchar());
	if(p) {
		*p++ = 0;
		if(!debug['.'])
			include[0] = strdup(ofile);
	} else
		p = ofile;

	if(outfile == 0) {
		outfile = p;
		if(outfile) {
			if(p = utfrrune(outfile, '.'))
				if(p[1] == 'c' && p[2] == 0)
					p[0] = 0;
			p = utfrune(outfile, 0);
			if(debug['a'] && debug['n'])
				strcat(p, ".acid");
			else if(debug['Z'] && debug['n'])
				strcat(p, "_pickle.c");
			else {
				p[0] = '.';
				p[1] = thechar;
				p[2] = 0;
			}
		} else
			outfile = "/dev/null";
	}

	if(p = getenv("INCLUDE")) {
		setinclude(p);
	} else {
		if(systemtype(Plan9)) {
			sprint(incfile, "/%s/include", thestring);
			setinclude(strdup(incfile));
			setinclude("/sys/include");
		}
	}
	if (first)
		Binit(&diagbuf, 1, OWRITE);
	/*
	 * if we're writing acid to standard output, don't keep scratching
	 * outbuf.
	 */
	if((debug['a'] || debug['Z']) && !debug['n']) {
		if (first) {
			outfile = 0;
			Binit(&outbuf, dup(1, -1), OWRITE);
			dup(2, 1);
		}
	} else {
		c = mycreat(outfile, 0664);
		if(c < 0) {
			diag(Z, "cannot open %s - %r", outfile);
			outfile = 0;
			errorexit();
		}
		Binit(&outbuf, c, OWRITE);
	}
	newio();
	first = 0;

	/* Use an ANSI preprocessor */
	if(debug['p']) {
		if(systemtype(Windows)) {
			diag(Z, "-p option not supported on windows");
			errorexit();
		}
		if(myaccess(file) < 0) {
			diag(Z, "%s does not exist", file);
			errorexit();
		}
		if(mypipe(fd) < 0) {
			diag(Z, "pipe failed");
			errorexit();
		}
		switch(myfork()) {
		case -1:
			diag(Z, "fork failed");
			errorexit();
		case 0:
			close(fd[0]);
			mydup(fd[1], 1);
			close(fd[1]);
			av[0] = CPP;
			i = 1;
			if(debug['+']) {
				sprint(opt, "-+");
				av[i++] = strdup(opt);
			}
			for(c = 0; c < ndef; c++) {
				sprint(opt, "-D%s", defs[c]);
				av[i++] = strdup(opt);
			}
			for(c = 0; c < ninclude; c++) {
				sprint(opt, "-I%s", include[c]);
				av[i++] = strdup(opt);
			}
			if(strcmp(file, "stdin") != 0)
				av[i++] = file;
			av[i] = 0;
			if(debug['p'] > 1) {
				for(c = 0; c < i; c++)
					fprint(2, "%s ", av[c]);
				fprint(2, "\n");
			}
			myexec(av[0], av);
			fprint(2, "can't exec C preprocessor %s: %r\n", CPP);
			errorexit();
		default:
			close(fd[1]);
			newfile(file, fd[0]);
			break;
		}
	} else {
		if(strcmp(file, "stdin") == 0)
			newfile(file, 0);
		else
			newfile(file, -1);
	}
	yyparse();
	if(!debug['a'] && !debug['Z'])
		gclean();
	return nerrors;
}

void
errorexit(void)
{
	if(outfile)
		remove(outfile);
	exits("error");
}

void
pushio(void)
{
	Io *i;

	i = iostack;
	if(i == I) {
		yyerror("botch in pushio");
		errorexit();
	}
	i->p = fi.p;
	i->c = fi.c;
}

void
newio(void)
{
	Io *i;
	static int pushdepth = 0;

	i = iofree;
	if(i == I) {
		pushdepth++;
		if(pushdepth > 1000) {
			yyerror("macro/io expansion too deep");
			errorexit();
		}
		i = alloc(sizeof(*i));
	} else
		iofree = i->link;
	i->c = 0;
	i->f = -1;
	ionext = i;
}

void
newfile(char *s, int f)
{
	Io *i;

	if(debug['e'])
		print("%L: %s\n", lineno, s);

	i = ionext;
	i->link = iostack;
	iostack = i;
	i->f = f;
	if(f < 0)
		i->f = open(s, 0);
	if(i->f < 0) {
		yyerror("%cc: %r: %s", thechar, s);
		errorexit();
	}
	fi.c = 0;
	linehist(s, 0);
}

Sym*
slookup(char *s)
{

	strcpy(symb, s);
	return lookup();
}

Sym*
lookup(void)
{
	Sym *s;
	ulong h;
	char *p;
	int c, n;

	h = 0;
	for(p=symb; *p;) {
		h = h * 3;
		h += *p++;
	}
	n = (p - symb) + 1;
	if((long)h < 0)
		h = ~h;
	h %= NHASH;
	c = symb[0];
	for(s = hash[h]; s != S; s = s->link) {
		if(s->name[0] != c)
			continue;
		if(strcmp(s->name, symb) == 0)
			return s;
	}
	s = alloc(sizeof(*s));
	s->name = alloc(n);
	memmove(s->name, symb, n);

	strcpy(s->name, symb);
	s->link = hash[h];
	hash[h] = s;
	syminit(s);

	return s;
}

void
syminit(Sym *s)
{
	s->lexical = LNAME;
	s->block = 0;
	s->offset = 0;
	s->type = T;
	s->suetag = T;
	s->class = CXXX;
	s->aused = 0;
	s->sig = SIGNONE;
}

#define	EOF	(-1)
#define	IGN	(-2)
#define	ESC	(1<<20)
#define	GETC()	((--fi.c < 0)? filbuf(): (*fi.p++ & 0xff))

enum
{
	Numdec		= 1<<0,
	Numlong		= 1<<1,
	Numuns		= 1<<2,
	Numvlong	= 1<<3,
	Numflt		= 1<<4,
};

long
yylex(void)
{
	vlong vv;
	long c, c1, t;
	char *cp;
	Rune rune;
	Sym *s;

	if(peekc != IGN) {
		c = peekc;
		peekc = IGN;
		goto l1;
	}
l0:
	c = GETC();

l1:
	if(c >= Runeself) {
		/*
		 * extension --
		 *	all multibyte runes are alpha
		 */
		cp = symb;
		goto talph;
	}
	if(isspace(c)) {
		if(c == '\n')
			lineno++;
		goto l0;
	}
	if(isalpha(c)) {
		cp = symb;
		if(c != 'L')
			goto talph;
		*cp++ = c;
		c = GETC();
		if(c == '\'') {
			/* L'x' */
			c = escchar('\'', 1, 0);
			if(c == EOF)
				c = '\'';
			c1 = escchar('\'', 1, 0);
			if(c1 != EOF) {
				yyerror("missing '");
				peekc = c1;
			}
			yylval.vval = convvtox(c, TUSHORT);
			return LUCONST;
		}
		if(c == '"') {
			goto caselq;
		}
		goto talph;
	}
	if(isdigit(c))
		goto tnum;
	switch(c)
	{

	case EOF:
		peekc = EOF;
		return -1;

	case '_':
		cp = symb;
		goto talph;

	case '#':
		domacro();
		goto l0;

	case '.':
		c1 = GETC();
		if(isdigit(c1)) {
			cp = symb;
			*cp++ = c;
			c = c1;
			c1 = 0;
			goto casedot;
		}
		break;

	case '"':
		strcpy(symb, "\"<string>\"");
		cp = alloc(0);
		c1 = 0;

		/* "..." */
		for(;;) {
			c = escchar('"', 0, 1);
			if(c == EOF)
				break;
			if(c & ESC) {
				cp = allocn(cp, c1, 1);
				cp[c1++] = c;
			} else {
				rune = c;
				c = runelen(rune);
				cp = allocn(cp, c1, c);
				runetochar(cp+c1, &rune);
				c1 += c;
			}
		}
		yylval.sval.l = c1;
		do {
			cp = allocn(cp, c1, 1);
			cp[c1++] = 0;
		} while(c1 & MAXALIGN);
		yylval.sval.s = cp;
		return LSTRING;

	caselq:
		/* L"..." */
		strcpy(symb, "\"L<string>\"");
		cp = alloc(0);
		c1 = 0;
		for(;;) {
			c = escchar('"', 1, 0);
			if(c == EOF)
				break;
			cp = allocn(cp, c1, sizeof(ushort));
			*(ushort*)(cp + c1) = c;
			c1 += sizeof(ushort);
		}
		yylval.sval.l = c1;
		do {
			cp = allocn(cp, c1, sizeof(ushort));
			*(ushort*)(cp + c1) = 0;
			c1 += sizeof(ushort);
		} while(c1 & MAXALIGN);
		yylval.sval.s = cp;
		return LLSTRING;

	case '\'':
		/* '.' */
		c = escchar('\'', 0, 0);
		if(c == EOF)
			c = '\'';
		c1 = escchar('\'', 0, 0);
		if(c1 != EOF) {
			yyerror("missing '");
			peekc = c1;
		}
		vv = c;
		yylval.vval = convvtox(vv, TUCHAR);
		if(yylval.vval != vv)
			yyerror("overflow in character constant: 0x%lx", c);
		else
		if(c & 0x80){
			nearln = lineno;
			warn(Z, "sign-extended character constant");
		}
		yylval.vval = convvtox(vv, TCHAR);
		return LCONST;

	case '/':
		c1 = GETC();
		if(c1 == '*') {
			for(;;) {
				c = getr();
				while(c == '*') {
					c = getr();
					if(c == '/')
						goto l0;
				}
				if(c == EOF) {
					yyerror("eof in comment");
					errorexit();
				}
			}
		}
		if(c1 == '/') {
			for(;;) {
				c = getr();
				if(c == '\n')
					goto l0;
				if(c == EOF) {
					yyerror("eof in comment");
					errorexit();
				}
			}
		}
		if(c1 == '=')
			return LDVE;
		break;

	case '*':
		c1 = GETC();
		if(c1 == '=')
			return LMLE;
		break;

	case '%':
		c1 = GETC();
		if(c1 == '=')
			return LMDE;
		break;

	case '+':
		c1 = GETC();
		if(c1 == '+')
			return LPP;
		if(c1 == '=')
			return LPE;
		break;

	case '-':
		c1 = GETC();
		if(c1 == '-')
			return LMM;
		if(c1 == '=')
			return LME;
		if(c1 == '>')
			return LMG;
		break;

	case '>':
		c1 = GETC();
		if(c1 == '>') {
			c = LRSH;
			c1 = GETC();
			if(c1 == '=')
				return LRSHE;
			break;
		}
		if(c1 == '=')
			return LGE;
		break;

	case '<':
		c1 = GETC();
		if(c1 == '<') {
			c = LLSH;
			c1 = GETC();
			if(c1 == '=')
				return LLSHE;
			break;
		}
		if(c1 == '=')
			return LLE;
		break;

	case '=':
		c1 = GETC();
		if(c1 == '=')
			return LEQ;
		break;

	case '!':
		c1 = GETC();
		if(c1 == '=')
			return LNE;
		break;

	case '&':
		c1 = GETC();
		if(c1 == '&')
			return LANDAND;
		if(c1 == '=')
			return LANDE;
		break;

	case '|':
		c1 = GETC();
		if(c1 == '|')
			return LOROR;
		if(c1 == '=')
			return LORE;
		break;

	case '^':
		c1 = GETC();
		if(c1 == '=')
			return LXORE;
		break;

	default:
		return c;
	}
	peekc = c1;
	return c;

talph:
	/*
	 * cp is set to symb and some
	 * prefix has been stored
	 */
	for(;;) {
		if(c >= Runeself) {
			for(c1=0;;) {
				cp[c1++] = c;
				if(fullrune(cp, c1))
					break;
				c = GETC();
			}
			cp += c1;
			c = GETC();
			continue;
		}
		if(!isalnum(c) && c != '_')
			break;
		*cp++ = c;
		c = GETC();
	}
	*cp = 0;
	if(debug['L'])
		print("%L: %s\n", lineno, symb);
	peekc = c;
	s = lookup();
	if(s->macro) {
		newio();
		cp = ionext->b;
		macexpand(s, cp);
		pushio();
		ionext->link = iostack;
		iostack = ionext;
		fi.p = cp;
		fi.c = strlen(cp);
		if(peekc != IGN) {
			cp[fi.c++] = peekc;
			cp[fi.c] = 0;
			peekc = IGN;
		}
		goto l0;
	}
	yylval.sym = s;
	if(s->class == CTYPEDEF || s->class == CTYPESTR)
		return LTYPE;
	return s->lexical;

tnum:
	c1 = 0;
	cp = symb;
	if(c != '0') {
		c1 |= Numdec;
		for(;;) {
			*cp++ = c;
			c = GETC();
			if(isdigit(c))
				continue;
			goto dc;
		}
	}
	*cp++ = c;
	c = GETC();
	if(c == 'x' || c == 'X')
		for(;;) {
			*cp++ = c;
			c = GETC();
			if(isdigit(c))
				continue;
			if(c >= 'a' && c <= 'f')
				continue;
			if(c >= 'A' && c <= 'F')