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 <ctype.h>
#define	EXTERN
#include "a.h"
#include "y.tab.h"

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

	thechar = '8';
	thestring = "386";
	memset(debug, 0, sizeof(debug));
	cinit();
	outfile = 0;
	include[ninclude++] = ".";
	ARGBEGIN {
	default:
		c = ARGC();
		if(c >= 0 || c < sizeof(debug))
			debug[c] = 1;
		break;

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

	case 'D':
		p = ARGF();
		if(p)
			Dlist[nDlist++] = p;
		break;

	case 'I':
		p = ARGF();
		setinclude(p);
		break;
	} ARGEND
	if(*argv == 0) {
		print("usage: %ca [-options] file.s\n", thechar);
		errorexit();
	}
	if(argc > 1 && systemtype(Windows)){
		print("can't assemble multiple files on windows\n");
		errorexit();
	}
	if(argc > 1 && !systemtype(Windows)) {
		nproc = 1;
		if(p = getenv("NPROC"))
			nproc = atol(p);	/* */
		c = 0;
		nout = 0;
		for(;;) {
			while(nout < nproc && argc > 0) {
				i = myfork();
				if(i < 0) {
					i = mywait(&status);
					if(i < 0)
						errorexit();
					if(status)
						c++;
					nout--;
					continue;
				}
				if(i == 0) {
					print("%s:\n", *argv);
					if(assemble(*argv))
						errorexit();
					exits(0);
				}
				nout++;
				argc--;
				argv++;
			}
			i = mywait(&status);
			if(i < 0) {
				if(c)
					errorexit();
				exits(0);
			}
			if(status)
				c++;
			nout--;
		}
	}
	if(assemble(argv[0]))
		errorexit();
	exits(0);
}

int
assemble(char *file)
{
	char ofile[100], incfile[20], *p;
	int i, of;

	strcpy(ofile, file);
	p = utfrrune(ofile, pathchar());
	if(p) {
		include[0] = ofile;
		*p++ = 0;
	} else
		p = ofile;
	if(outfile == 0) {
		outfile = p;
		if(outfile){
			p = utfrrune(outfile, '.');
			if(p)
				if(p[1] == 's' && p[2] == 0)
					p[0] = 0;
			p = utfrune(outfile, 0);
			p[0] = '.';
			p[1] = thechar;
			p[2] = 0;
		} else
			outfile = "/dev/null";
	}
	p = getenv("INCLUDE");
	if(p) {
		setinclude(p);
	} else {
		if(systemtype(Plan9)) {
			sprint(incfile,"/%s/include", thestring);
			setinclude(strdup(incfile));
		}
	}

	of = mycreat(outfile, 0664);
	if(of < 0) {
		yyerror("%ca: cannot create %s", thechar, outfile);
		errorexit();
	}
	Binit(&obuf, of, OWRITE);

	pass = 1;
	pinit(file);
	for(i=0; i<nDlist; i++)
		dodefine(Dlist[i]);
	yyparse();
	if(nerrors) {
		cclean();
		return nerrors;
	}

	pass = 2;
	outhist();
	pinit(file);
	for(i=0; i<nDlist; i++)
		dodefine(Dlist[i]);
	yyparse();
	cclean();
	return nerrors;
}

struct
{
	char	*name;
	ushort	type;
	ushort	value;
} itab[] =
{
	"SP",		LSP,	D_AUTO,
	"SB",		LSB,	D_EXTERN,
	"FP",		LFP,	D_PARAM,
	"PC",		LPC,	D_BRANCH,

	"AL",		LBREG,	D_AL,
	"CL",		LBREG,	D_CL,
	"DL",		LBREG,	D_DL,
	"BL",		LBREG,	D_BL,
	"AH",		LBREG,	D_AH,
	"CH",		LBREG,	D_CH,
	"DH",		LBREG,	D_DH,
	"BH",		LBREG,	D_BH,

	"AX",		LLREG,	D_AX,
	"CX",		LLREG,	D_CX,
	"DX",		LLREG,	D_DX,
	"BX",		LLREG,	D_BX,
/*	"SP",		LLREG,	D_SP,	*/
	"BP",		LLREG,	D_BP,
	"SI",		LLREG,	D_SI,
	"DI",		LLREG,	D_DI,

	"F0",		LFREG,	D_F0+0,
	"F1",		LFREG,	D_F0+1,
	"F2",		LFREG,	D_F0+2,
	"F3",		LFREG,	D_F0+3,
	"F4",		LFREG,	D_F0+4,
	"F5",		LFREG,	D_F0+5,
	"F6",		LFREG,	D_F0+6,
	"F7",		LFREG,	D_F0+7,

	"CS",		LSREG,	D_CS,
	"SS",		LSREG,	D_SS,
	"DS",		LSREG,	D_DS,
	"ES",		LSREG,	D_ES,
	"FS",		LSREG,	D_FS,
	"GS",		LSREG,	D_GS,

	"GDTR",		LBREG,	D_GDTR,
	"IDTR",		LBREG,	D_IDTR,
	"LDTR",		LBREG,	D_LDTR,
	"MSW",		LBREG,	D_MSW,
	"TASK",		LBREG,	D_TASK,

	"CR0",		LBREG,	D_CR+0,
	"CR1",		LBREG,	D_CR+1,
	"CR2",		LBREG,	D_CR+2,
	"CR3",		LBREG,	D_CR+3,
	"CR4",		LBREG,	D_CR+4,
	"CR5",		LBREG,	D_CR+5,
	"CR6",		LBREG,	D_CR+6,
	"CR7",		LBREG,	D_CR+7,

	"DR0",		LBREG,	D_DR+0,
	"DR1",		LBREG,	D_DR+1,
	"DR2",		LBREG,	D_DR+2,
	"DR3",		LBREG,	D_DR+3,
	"DR4",		LBREG,	D_DR+4,
	"DR5",		LBREG,	D_DR+5,
	"DR6",		LBREG,	D_DR+6,
	"DR7",		LBREG,	D_DR+7,

	"TR0",		LBREG,	D_TR+0,
	"TR1",		LBREG,	D_TR+1,
	"TR2",		LBREG,	D_TR+2,
	"TR3",		LBREG,	D_TR+3,
	"TR4",		LBREG,	D_TR+4,
	"TR5",		LBREG,	D_TR+5,
	"TR6",		LBREG,	D_TR+6,
	"TR7",		LBREG,	D_TR+7,

	"AAA",		LTYPE0,	AAAA,
	"AAD",		LTYPE0,	AAAD,
	"AAM",		LTYPE0,	AAAM,
	"AAS",		LTYPE0,	AAAS,
	"ADCB",		LTYPE3,	AADCB,
	"ADCL",		LTYPE3,	AADCL,
	"ADCW",		LTYPE3,	AADCW,
	"ADDB",		LTYPE3,	AADDB,
	"ADDL",		LTYPE3,	AADDL,
	"ADDW",		LTYPE3,	AADDW,
	"ADJSP",	LTYPE2,	AADJSP,
	"ANDB",		LTYPE3,	AANDB,
	"ANDL",		LTYPE3,	AANDL,
	"ANDW",		LTYPE3,	AANDW,
	"ARPL",		LTYPE3,	AARPL,
	"BOUNDL",	LTYPE3,	ABOUNDL,
	"BOUNDW",	LTYPE3,	ABOUNDW,
	"BSFL",		LTYPE3,	ABSFL,
	"BSFW",		LTYPE3,	ABSFW,
	"BSRL",		LTYPE3,	ABSRL,
	"BSRW",		LTYPE3,	ABSRW,
	"BTCL",		LTYPE3,	ABTCL,
	"BTCW",		LTYPE3,	ABTCW,
	"BTL",		LTYPE3,	ABTL,
	"BTRL",		LTYPE3,	ABTRL,
	"BTRW",		LTYPE3,	ABTRW,
	"BTSL",		LTYPE3,	ABTSL,
	"BTSW",		LTYPE3,	ABTSW,
	"BTW",		LTYPE3,	ABTW,
	"BYTE",		LTYPE2,	ABYTE,
	"CALL",		LTYPEC,	ACALL,
	"CLC",		LTYPE0,	ACLC,
	"CLD",		LTYPE0,	ACLD,
	"CLI",		LTYPE0,	ACLI,
	"CLTS",		LTYPE0,	ACLTS,
	"CMC",		LTYPE0,	ACMC,
	"CMPB",		LTYPE4,	ACMPB,
	"CMPL",		LTYPE4,	ACMPL,
	"CMPW",		LTYPE4,	ACMPW,
	"CMPSB",	LTYPE0,	ACMPSB,
	"CMPSL",	LTYPE0,	ACMPSL,
	"CMPSW",	LTYPE0,	ACMPSW,
	"DAA",		LTYPE0,	ADAA,
	"DAS",		LTYPE0,	ADAS,
	"DATA",		LTYPED,	ADATA,
	"DECB",		LTYPE1,	ADECB,
	"DECL",		LTYPE1,	ADECL,
	"DECW",		LTYPE1,	ADECW,
	"DIVB",		LTYPE2,	ADIVB,
	"DIVL",		LTYPE2,	ADIVL,
	"DIVW",		LTYPE2,	ADIVW,
	"END",		LTYPE0,	AEND,
	"ENTER",	LTYPE2,	AENTER,
	"GLOBL",	LTYPET,	AGLOBL,
	"HLT",		LTYPE0,	AHLT,
	"IDIVB",	LTYPE2,	AIDIVB,
	"IDIVL",	LTYPE2,	AIDIVL,
	"IDIVW",	LTYPE2,	AIDIVW,
	"IMULB",	LTYPE2,	AIMULB,
	"IMULL",	LTYPE2,	AIMULL,
	"IMULW",	LTYPE2,	AIMULW,
	"INB",		LTYPE0,	AINB,
	"INL",		LTYPE0,	AINL,
	"INW",		LTYPE0,	AINW,
	"INCB",		LTYPE1,	AINCB,
	"INCL",		LTYPE1,	AINCL,
	"INCW",		LTYPE1,	AINCW,
	"INSB",		LTYPE0,	AINSB,
	"INSL",		LTYPE0,	AINSL,
	"INSW",		LTYPE0,	AINSW,
	"INT",		LTYPE2,	AINT,
	"INTO",		LTYPE0,	AINTO,
	"IRETL",	LTYPE0,	AIRETL,
	"IRETW",	LTYPE0,	AIRETW,

	"JOS",		LTYPER,	AJOS,
	"JO",		LTYPER,	AJOS,	/* alternate */
	"JOC",		LTYPER,	AJOC,
	"JNO",		LTYPER,	AJOC,	/* alternate */
	"JCS",		LTYPER,	AJCS,
	"JB",		LTYPER,	AJCS,	/* alternate */
	"JC",		LTYPER,	AJCS,	/* alternate */
	"JNAE",		LTYPER,	AJCS,	/* alternate */
	"JLO",		LTYPER,	AJCS,	/* alternate */
	"JCC",		LTYPER,	AJCC,
	"JAE",		LTYPER,	AJCC,	/* alternate */
	"JNB",		LTYPER,	AJCC,	/* alternate */
	"JNC",		LTYPER,	AJCC,	/* alternate */
	"JHS",		LTYPER,	AJCC,	/* alternate */
	"JEQ",		LTYPER,	AJEQ,
	"JE",		LTYPER,	AJEQ,	/* alternate */
	"JZ",		LTYPER,	AJEQ,	/* alternate */
	"JNE",		LTYPER,	AJNE,
	"JNZ",		LTYPER,	AJNE,	/* alternate */
	"JLS",		LTYPER,	AJLS,
	"JBE",		LTYPER,	AJLS,	/* alternate */
	"JNA",		LTYPER,	AJLS,	/* alternate */
	"JHI",		LTYPER,	AJHI,
	"JA",		LTYPER,	AJHI,	/* alternate */
	"JNBE",		LTYPER,	AJHI,	/* alternate */
	"JMI",		LTYPER,	AJMI,
	"JS",		LTYPER,	AJMI,	/* alternate */
	"JPL",		LTYPER,	AJPL,
	"JNS",		LTYPER,	AJPL,	/* alternate */
	"JPS",		LTYPER,	AJPS,
	"JP",		LTYPER,	AJPS,	/* alternate */
	"JPE",		LTYPER,	AJPS,	/* alternate */
	"JPC",		LTYPER,	AJPC,
	"JNP",		LTYPER,	AJPC,	/* alternate */
	"JPO",		LTYPER,	AJPC,	/* alternate */
	"JLT",		LTYPER,	AJLT,
	"JL",		LTYPER,	AJLT,	/* alternate */
	"JNGE",		LTYPER,	AJLT,	/* alternate */
	"JGE",		LTYPER,	AJGE,
	"JNL",		LTYPER,	AJGE,	/* alternate */
	"JLE",		LTYPER,	AJLE,
	"JNG",		LTYPER,	AJLE,	/* alternate */
	"JGT",		LTYPER,	AJGT,
	"JG",		LTYPER,	AJGT,	/* alternate */
	"JNLE",		LTYPER,	AJGT,	/* alternate */

	"JCXZ",		LTYPER,	AJCXZ,
	"JMP",		LTYPEC,	AJMP,
	"LAHF",		LTYPE0,	ALAHF,
	"LARL",		LTYPE3,	ALARL,
	"LARW",		LTYPE3,	ALARW,
	"LEAL",		LTYPE3,	ALEAL,
	"LEAW",		LTYPE3,	ALEAW,
	"LEAVEL",	LTYPE0,	ALEAVEL,
	"LEAVEW",	LTYPE0,	ALEAVEW,
	"LOCK",		LTYPE0,	ALOCK,
	"LODSB",	LTYPE0,	ALODSB,
	"LODSL",	LTYPE0,	ALODSL,
	"LODSW",	LTYPE0,	ALODSW,
	"LONG",		LTYPE2,	ALONG,
	"LOOP",		LTYPER,	ALOOP,
	"LOOPEQ",	LTYPER,	ALOOPEQ,
	"LOOPNE",	LTYPER,	ALOOPNE,
	"LSLL",		LTYPE3,	ALSLL,
	"LSLW",		LTYPE3,	ALSLW,
	"MOVB",		LTYPE3,	AMOVB,
	"MOVL",		LTYPEM,	AMOVL,
	"MOVW",		LTYPEM,	AMOVW,
	"MOVBLSX",	LTYPE3, AMOVBLSX,
	"MOVBLZX",	LTYPE3, AMOVBLZX,
	"MOVBWSX",	LTYPE3, AMOVBWSX,
	"MOVBWZX",	LTYPE3, AMOVBWZX,
	"MOVWLSX",	LTYPE3, AMOVWLSX,
	"MOVWLZX",	LTYPE3, AMOVWLZX,
	"MOVSB",	LTYPE0,	AMOVSB,
	"MOVSL",	LTYPE0,	AMOVSL,
	"MOVSW",	LTYPE0,	AMOVSW,
	"MULB",		LTYPE2,	AMULB,
	"MULL",		LTYPE2,	AMULL,
	"MULW",		LTYPE2,	AMULW,
	"NEGB",		LTYPE1,	ANEGB,
	"NEGL",		LTYPE1,	ANEGL,
	"NEGW",		LTYPE1,	ANEGW,
	"NOP",		LTYPEN,	ANOP,
	"NOTB",		LTYPE1,	ANOTB,
	"NOTL",		LTYPE1,	ANOTL,
	"NOTW",		LTYPE1,	ANOTW,
	"ORB",		LTYPE3,	AORB,
	"ORL",		LTYPE3,	AORL,
	"ORW",		LTYPE3,	AORW,
	"OUTB",		LTYPE0,	AOUTB,
	"OUTL",		LTYPE0,	AOUTL,
	"OUTW",		LTYPE0,	AOUTW,
	"OUTSB",	LTYPE0,	AOUTSB,
	"OUTSL",	LTYPE0,	AOUTSL,
	"OUTSW",	LTYPE0,	AOUTSW,
	"POPAL",	LTYPE0,	APOPAL,
	"POPAW",	LTYPE0,	APOPAW,
	"POPFL",	LTYPE0,	APOPFL,
	"POPFW",	LTYPE0,	APOPFW,
	"POPL",		LTYPE1,	APOPL,
	"POPW",		LTYPE1,	APOPW,
	"PUSHAL",	LTYPE0,	APUSHAL,
	"PUSHAW",	LTYPE0,	APUSHAW,
	"PUSHFL",	LTYPE0,	APUSHFL,
	"PUSHFW",	LTYPE0,	APUSHFW,
	"PUSHL",	LTYPE2,	APUSHL,
	"PUSHW",	LTYPE2,	APUSHW,
	"RCLB",		LTYPE3,	ARCLB,
	"RCLL",		LTYPE3,	ARCLL,
	"RCLW",		LTYPE3,	ARCLW,
	"RCRB",		LTYPE3,	ARCRB,
	"RCRL",		LTYPE3,	ARCRL,
	"RCRW",		LTYPE3,	ARCRW,
	"REP",		LTYPE0,	AREP,
	"REPN",		LTYPE0,	AREPN,
	"RET",		LTYPE0,	ARET,
	"ROLB",		LTYPE3,	AROLB,
	"ROLL",		LTYPE3,	AROLL,
	"ROLW",		LTYPE3,	AROLW,
	"RORB",		LTYPE3,	ARORB,
	"RORL",		LTYPE3,	ARORL,
	"RORW",		LTYPE3,	ARORW,
	"SAHF",		LTYPE0,	ASAHF,
	"SALB",		LTYPE3,	ASALB,
	"SALL",		LTYPE3,	ASALL,
	"SALW",		LTYPE3,	ASALW,
	"SARB",		LTYPE3,	ASARB,
	"SARL",		LTYPE3,	ASARL,
	"SARW",		LTYPE3,	ASARW,
	"SBBB",		LTYPE3,	ASBBB,
	"SBBL",		LTYPE3,	ASBBL,
	"SBBW",		LTYPE3,	ASBBW,
	"SCASB",	LTYPE0,	ASCASB,
	"SCASL",	LTYPE0,	ASCASL,
	"SCASW",	LTYPE0,	ASCASW,
	"SETCC",	LTYPE1,	ASETCC,
	"SETCS",	LTYPE1,	ASETCS,
	"SETEQ",	LTYPE1,	ASETEQ,
	"SETGE",	LTYPE1,	ASETGE,