compile.c

早期freebsd实现

	*repp = xmalloc(sizeof(regex_t));
	if (p && (eval = regcomp(*repp, re, 0)) != 0)
		err(COMPILE, "RE error: %s", strregerror(eval, *repp));
	if (maxnsub < (*repp)->re_nsub)
		maxnsub = (*repp)->re_nsub;
	return (p);
}

/*
 * Compile the substitution string of a regular expression and set res to
 * point to a saved copy of it.  Nsub is the number of parenthesized regular
 * expressions.
 */
static char *
compile_subst(p, s)
	char *p;
	struct s_subst *s;
{
	static char lbuf[_POSIX2_LINE_MAX + 1];
	int asize, ref, size;
	char c, *text, *op, *sp;

	c = *p++;			/* Terminator character */
	if (c == '\0')
		return (NULL);

	s->maxbref = 0;
	s->linenum = linenum;
	asize = 2 * _POSIX2_LINE_MAX + 1;
	text = xmalloc(asize);
	size = 0;
	do {
		op = sp = text + size;
		for (; *p; p++) {
			if (*p == '\\') {
				p++;
				if (strchr("123456789", *p) != NULL) {
					*sp++ = '\\';
					ref = *p - '0';
					if (s->re != NULL &&
					    ref > s->re->re_nsub)
						err(COMPILE,
"\\%c not defined in the RE", *p);
					if (s->maxbref < ref)
						s->maxbref = ref;
				} else if (*p == '&' || *p == '\\')
					*sp++ = '\\';
			} else if (*p == c) {
				p++;
				*sp++ = '\0';
				size += sp - op;
				s->new = xrealloc(text, size);
				return (p);
			} else if (*p == '\n') {
				err(COMPILE,
"unescaped newline inside substitute pattern");
				/* NOTREACHED */
			}
			*sp++ = *p;
		}
		size += sp - op;
		if (asize - size < _POSIX2_LINE_MAX + 1) {
			asize *= 2;
			text = xmalloc(asize);
		}
	} while (cu_fgets(p = lbuf, sizeof(lbuf)));
	err(COMPILE, "unterminated substitute in regular expression");
	/* NOTREACHED */
}

/*
 * Compile the flags of the s command
 */
static char *
compile_flags(p, s)
	char *p;
	struct s_subst *s;
{
	int gn;			/* True if we have seen g or n */
	char wfile[_POSIX2_LINE_MAX + 1], *q;

	s->n = 1;				/* Default */
	s->p = 0;
	s->wfile = NULL;
	s->wfd = -1;
	for (gn = 0;;) {
		EATSPACE();			/* EXTENSION */
		switch (*p) {
		case 'g':
			if (gn)
				err(COMPILE,
"more than one number or 'g' in substitute flags");
			gn = 1;
			s->n = 0;
			break;
		case '\0':
		case '\n':
		case ';':
			return (p);
		case 'p':
			s->p = 1;
			break;
		case '1': case '2': case '3':
		case '4': case '5': case '6':
		case '7': case '8': case '9':
			if (gn)
				err(COMPILE,
"more than one number or 'g' in substitute flags");
			gn = 1;
			/* XXX Check for overflow */
			s->n = (int)strtol(p, &p, 10);
			break;
		case 'w':
			p++;
#ifdef HISTORIC_PRACTICE
			if (*p != ' ') {
				err(WARNING, "space missing before w wfile");
				return (p);
			}
#endif
			EATSPACE();
			q = wfile;
			while (*p) {
				if (*p == '\n')
					break;
				*q++ = *p++;
			}
			*q = '\0';
			if (q == wfile)
				err(COMPILE, "no wfile specified");
			s->wfile = strdup(wfile);
			if (!aflag && (s->wfd = open(wfile,
			    O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
			    DEFFILEMODE)) == -1)
				err(FATAL, "%s: %s\n", wfile, strerror(errno));
			return (p);
		default:
			err(COMPILE,
			    "bad flag in substitute command: '%c'", *p);
			break;
		}
		p++;
	}
}

/*
 * Compile a translation set of strings into a lookup table.
 */
static char *
compile_tr(p, transtab)
	char *p;
	char **transtab;
{
	int i;
	char *lt, *op, *np;
	char old[_POSIX2_LINE_MAX + 1];
	char new[_POSIX2_LINE_MAX + 1];

	if (*p == '\0' || *p == '\\')
		err(COMPILE,
"transform pattern can not be delimited by newline or backslash");
	p = compile_delimited(p, old);
	if (p == NULL) {
		err(COMPILE, "unterminated transform source string");
		return (NULL);
	}
	p = compile_delimited(--p, new);
	if (p == NULL) {
		err(COMPILE, "unterminated transform target string");
		return (NULL);
	}
	EATSPACE();
	if (strlen(new) != strlen(old)) {
		err(COMPILE, "transform strings are not the same length");
		return (NULL);
	}
	/* We assume characters are 8 bits */
	lt = xmalloc(UCHAR_MAX);
	for (i = 0; i <= UCHAR_MAX; i++)
		lt[i] = (char)i;
	for (op = old, np = new; *op; op++, np++)
		lt[(u_char)*op] = *np;
	*transtab = lt;
	return (p);
}

/*
 * Compile the text following an a or i command.
 */
static char *
compile_text()
{
	int asize, size;
	char *text, *p, *op, *s;
	char lbuf[_POSIX2_LINE_MAX + 1];

	asize = 2 * _POSIX2_LINE_MAX + 1;
	text = xmalloc(asize);
	size = 0;
	while (cu_fgets(lbuf, sizeof(lbuf))) {
		op = s = text + size;
		p = lbuf;
		EATSPACE();
		for (; *p; p++) {
			if (*p == '\\')
				p++;
			*s++ = *p;
		}
		size += s - op;
		if (p[-2] != '\\') {
			*s = '\0';
			break;
		}
		if (asize - size < _POSIX2_LINE_MAX + 1) {
			asize *= 2;
			text = xmalloc(asize);
		}
	}
	return (xrealloc(text, size + 1));
}

/*
 * Get an address and return a pointer to the first character after
 * it.  Fill the structure pointed to according to the address.
 */
static char *
compile_addr(p, a)
	char *p;
	struct s_addr *a;
{
	char *end;

	switch (*p) {
	case '\\':				/* Context address */
		++p;
		/* FALLTHROUGH */
	case '/':				/* Context address */
		p = compile_re(p, &a->u.r);
		if (p == NULL)
			err(COMPILE, "unterminated regular expression");
		a->type = AT_RE;
		return (p);

	case '$':				/* Last line */
		a->type = AT_LAST;
		return (p + 1);
						/* Line number */
	case '0': case '1': case '2': case '3': case '4': 
	case '5': case '6': case '7': case '8': case '9':
		a->type = AT_LINE;
		a->u.l = strtol(p, &end, 10);
		return (end);
	default:
		err(COMPILE, "expected context address");
		return (NULL);
	}
}

/*
 * duptoeol --
 *	Return a copy of all the characters up to \n or \0.
 */
static char *
duptoeol(s, ctype)
	register char *s;
	char *ctype;
{
	size_t len;
	int ws;
	char *start;

	ws = 0;
	for (start = s; *s != '\0' && *s != '\n'; ++s)
		ws = isspace(*s);
	*s = '\0';
	if (ws)
		err(WARNING, "whitespace after %s", ctype);
	len = s - start + 1;
	return (memmove(xmalloc(len), start, len));
}

/*
 * Convert goto label names to addresses, and count a and r commands, in
 * the given subset of the script.  Free the memory used by labels in b
 * and t commands (but not by :).
 *
 * TODO: Remove } nodes
 */
static void
fixuplabel(cp, end)
	struct s_command *cp, *end;
{

	for (; cp != end; cp = cp->next)
		switch (cp->code) {
		case 'a':
		case 'r':
			appendnum++;
			break;
		case 'b':
		case 't':
			/* Resolve branch target. */
			if (cp->t == NULL) {
				cp->u.c = NULL;
				break;
			}
			if ((cp->u.c = findlabel(cp->t)) == NULL)
				err(COMPILE2, "undefined label '%s'", cp->t);
			free(cp->t);
			break;
		case '{':
			/* Do interior commands. */
			fixuplabel(cp->u.c, cp->next);
			break;
		}
}

/*
 * Associate the given command label for later lookup.
 */
static void
enterlabel(cp)
	struct s_command *cp;
{
	register struct labhash **lhp, *lh;
	register u_char *p;
	register u_int h, c;

	for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
		h = (h << 5) + h + c;
	lhp = &labels[h & LHMASK];
	for (lh = *lhp; lh != NULL; lh = lh->lh_next)
		if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
			err(COMPILE2, "duplicate label '%s'", cp->t);
	lh = xmalloc(sizeof *lh);
	lh->lh_next = *lhp;
	lh->lh_hash = h;
	lh->lh_cmd = cp;
	lh->lh_ref = 0;
	*lhp = lh;
}

/*
 * Find the label contained in the command l in the command linked
 * list cp.  L is excluded from the search.  Return NULL if not found.
 */
static struct s_command *
findlabel(name)
	char *name;
{
	register struct labhash *lh;
	register u_char *p;
	register u_int h, c;

	for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
		h = (h << 5) + h + c;
	for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
		if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
			lh->lh_ref = 1;
			return (lh->lh_cmd);
		}
	}
	return (NULL);
}

/* 
 * Warn about any unused labels.  As a side effect, release the label hash
 * table space.
 */
static void
uselabel()
{
	register struct labhash *lh, *next;
	register int i;

	for (i = 0; i < LHSZ; i++) {
		for (lh = labels[i]; lh != NULL; lh = next) {
			next = lh->lh_next;
			if (!lh->lh_ref)
				err(WARNING, "unused label '%s'",
				    lh->lh_cmd->t);
			free(lh);
		}
	}
}