lburg.c

lcc,一个可变目标c语言编译器的源码

#include <assert.h>
#include <ctype.h>
#include <stdarg.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <time.h>
#include "lburg.h"

static char rcsid[] = "$Id: lburg.c,v 2.10 2002/03/08 18:45:21 drh Exp $";
static char *prefix = "";
static int Tflag = 0;
static int ntnumber = 0;
static Nonterm start = 0;
static Term terms;
static Nonterm nts;
static Rule rules;
static int nrules;
static struct block {
	struct block *link;
} *memlist;			/* list of allocated blocks */

static char *stringf(char *fmt, ...);
static void print(char *fmt, ...);
static void ckreach(Nonterm p);
static void emitclosure(Nonterm nts);
static void emitcost(Tree t, char *v);
static void emitdefs(Nonterm nts, int ntnumber);
static void emitheader(void);
static void emitkids(Rule rules, int nrules);
static void emitnts(Rule rules, int nrules);
static void emitrecalc(char *pre, Term root, Term kid);
static void emitrecord(char *pre, Rule r, char *c, int cost);
static void emitrule(Nonterm nts);
static void emitlabel(Term terms, Nonterm start, int ntnumber);
static void emitstring(Rule rules);
static void emitstruct(Nonterm nts, int ntnumber);
static void emittest(Tree t, char *v, char *suffix);

int main(int argc, char *argv[]) {
	int c, i;
	Nonterm p;
	
	for (i = 1; i < argc; i++)
		if (strcmp(argv[i], "-T") == 0)
			Tflag = 1;
		else if (strncmp(argv[i], "-p", 2) == 0 && argv[i][2])
			prefix = &argv[i][2];
		else if (strncmp(argv[i], "-p", 2) == 0 && i + 1 < argc)
			prefix = argv[++i];
		else if (*argv[i] == '-' && argv[i][1]) {
			yyerror("usage: %s [-T | -p prefix]... [ [ input ] output ] \n",
				argv[0]);
			exit(1);
		} else if (infp == NULL) {
			if (strcmp(argv[i], "-") == 0)
				infp = stdin;
			else if ((infp = fopen(argv[i], "r")) == NULL) {
				yyerror("%s: can't read `%s'\n", argv[0], argv[i]);
				exit(1);
			}
		} else if (outfp == NULL) {
			if (strcmp(argv[i], "-") == 0)
				outfp = stdout;
			if ((outfp = fopen(argv[i], "w")) == NULL) {
				yyerror("%s: can't write `%s'\n", argv[0], argv[i]);
				exit(1);
			}
		}
	if (infp == NULL)
		infp = stdin;
	if (outfp == NULL)
		outfp = stdout;
	yyparse();
	if (start)
		ckreach(start);
	for (p = nts; p; p = p->link) {
		if (p->rules == NULL)
			yyerror("undefined nonterminal `%s'\n", p->name);
		if (!p->reached)
			yyerror("can't reach nonterminal `%s'\n", p->name);
	}
	emitheader();
	emitdefs(nts, ntnumber);
	emitstruct(nts, ntnumber);
	emitnts(rules, nrules);
	emitstring(rules);
	emitrule(nts);
	emitclosure(nts);
	if (start)
		emitlabel(terms, start, ntnumber);
	emitkids(rules, nrules);
	if (!feof(infp))
		while ((c = getc(infp)) != EOF)
			putc(c, outfp);
	while (memlist) {	/* for purify */
		struct block *q = memlist->link;
		free(memlist);
		memlist = q;
	}
	return errcnt > 0;
}

/* alloc - allocate nbytes or issue fatal error */
void *alloc(int nbytes) {
	struct block *p = calloc(1, sizeof *p + nbytes);

	if (p == NULL) {
		yyerror("out of memory\n");
		exit(1);
	}
	p->link = memlist;
	memlist = p;
	return p + 1;
}

/* stringf - format and save a string */
static char *stringf(char *fmt, ...) {
	va_list ap;
	char buf[512];

	va_start(ap, fmt);
	vsprintf(buf, fmt, ap);
	va_end(ap);
	return strcpy(alloc(strlen(buf) + 1), buf);
}	

struct entry {
	union {
		char *name;
		struct term t;
		struct nonterm nt;
	} sym;
	struct entry *link;
} *table[211];
#define HASHSIZE (sizeof table/sizeof table[0])

/* hash - return hash number for str */
static unsigned hash(char *str) {
	unsigned h = 0;

	while (*str)
		h = (h<<1) + *str++;
	return h;
}

/* lookup - lookup symbol name */
static void *lookup(char *name) {
	struct entry *p = table[hash(name)%HASHSIZE];

	for ( ; p; p = p->link)
		if (strcmp(name, p->sym.name) == 0)
			return &p->sym;
	return 0;
}

/* install - install symbol name */
static void *install(char *name) {
	struct entry *p = alloc(sizeof *p);
	int i = hash(name)%HASHSIZE;

	p->sym.name = name;
	p->link = table[i];
	table[i] = p;
	return &p->sym;
}

/* nonterm - create a new terminal id, if necessary */
Nonterm nonterm(char *id) {
	Nonterm p = lookup(id), *q = &nts;

	if (p && p->kind == NONTERM)
		return p;
	if (p && p->kind == TERM)
		yyerror("`%s' is a terminal\n", id);
	p = install(id);
	p->kind = NONTERM;
	p->number = ++ntnumber;
	if (p->number == 1)
		start = p;
	while (*q && (*q)->number < p->number)
		q = &(*q)->link;
	assert(*q == 0 || (*q)->number != p->number);
	p->link = *q;
	*q = p;
	return p;
}

/* term - create a new terminal id with external symbol number esn */
Term term(char *id, int esn) {
	Term p = lookup(id), *q = &terms;

	if (p)
		yyerror("redefinition of terminal `%s'\n", id);
	else
		p = install(id);
	p->kind = TERM;
	p->esn = esn;
	p->arity = -1;
	while (*q && (*q)->esn < p->esn)
		q = &(*q)->link;
	if (*q && (*q)->esn == p->esn)
		yyerror("duplicate external symbol number `%s=%d'\n",
			p->name, p->esn);
	p->link = *q;
	*q = p;
	return p;
}

/* tree - create & initialize a tree node with the given fields */
Tree tree(char *id, Tree left, Tree right) {
	Tree t = alloc(sizeof *t);
	Term p = lookup(id);
	int arity = 0;

	if (left && right)
		arity = 2;
	else if (left)
		arity = 1;
	if (p == NULL && arity > 0) {
		yyerror("undefined terminal `%s'\n", id);
		p = term(id, -1);
	} else if (p == NULL && arity == 0)
		p = (Term)nonterm(id);
	else if (p && p->kind == NONTERM && arity > 0) {
		yyerror("`%s' is a nonterminal\n", id);
		p = term(id, -1);
	}
	if (p->kind == TERM && p->arity == -1)
		p->arity = arity;
	if (p->kind == TERM && arity != p->arity)
		yyerror("inconsistent arity for terminal `%s'\n", id);
	t->op = p;
	t->nterms = p->kind == TERM;
	if ((t->left = left) != NULL)
		t->nterms += left->nterms;
	if ((t->right = right) != NULL)
		t->nterms += right->nterms;
	return t;
}

/* rule - create & initialize a rule with the given fields */
Rule rule(char *id, Tree pattern, char *template, char *code) {
	Rule r = alloc(sizeof *r), *q;
	Term p = pattern->op;
	char *end;

	r->lhs = nonterm(id);
	r->packed = ++r->lhs->lhscount;
	for (q = &r->lhs->rules; *q; q = &(*q)->decode)
		;
	*q = r;
	r->pattern = pattern;
	r->ern = ++nrules;
	r->template = template;
	r->code = code;
	r->cost = strtol(code, &end, 10);
	if (*end) {
		r->cost = -1;
		r->code = stringf("(%s)", code);
	}
	if (p->kind == TERM) {
		for (q = &p->rules; *q; q = &(*q)->next)
			;
		*q = r;
	} else if (pattern->left == NULL && pattern->right == NULL) {
		Nonterm p = pattern->op;
		r->chain = p->chain;
	        p->chain = r;
		if (r->cost == -1)
			yyerror("illegal nonconstant cost `%s'\n", code);
	}
	for (q = &rules; *q; q = &(*q)->link)
		;
	r->link = *q;
	*q = r;
	return r;
}

/* print - formatted output */
static void print(char *fmt, ...) {
	va_list ap;

	va_start(ap, fmt);
	for ( ; *fmt; fmt++)
		if (*fmt == '%')
			switch (*++fmt) {
			case 'd': fprintf(outfp, "%d", va_arg(ap, int)); break;
			case 's': fputs(va_arg(ap, char *), outfp); break;
			case 'P': fprintf(outfp, "%s_", prefix); break;
			case 'T': {
				Tree t = va_arg(ap, Tree);
				print("%S", t->op);
				if (t->left && t->right)
					print("(%T,%T)", t->left, t->right);
				else if (t->left)
					print("(%T)", t->left);
				break;
				}
			case 'R': {
				Rule r = va_arg(ap, Rule);
				print("%S: %T", r->lhs, r->pattern);
				break;
				}
			case 'S': {
				Term t = va_arg(ap, Term);
				fputs(t->name, outfp);
				break;
				}
			case '1': case '2': case '3': case '4': case '5': {
				int n = *fmt - '0';
				while (n-- > 0)
					putc('\t', outfp);
				break;
				}
			default: putc(*fmt, outfp); break;			
			}
		else
			putc(*fmt, outfp);
	va_end(ap);
}

/* reach - mark all nonterminals in tree t as reachable */
static void reach(Tree t) {
	Nonterm p = t->op;

	if (p->kind == NONTERM)
		if (!p->reached)
			ckreach(p);
	if (t->left)
		reach(t->left);
	if (t->right)
		reach(t->right);
}

/* ckreach - mark all nonterminals reachable from p */
static void ckreach(Nonterm p) {
	Rule r;