lex.c

早期freebsd实现

	 * if the :g modifier is followed by a newline, then error right away!
	 * -strike
	 */

	int     gmodflag = 0, amodflag = 0;

	do {
	    *np++ = c, c = getC(DOEXCL);
	    if (c == 'g' || c == 'a') {
		if (c == 'g')
		    gmodflag++;
		else
		    amodflag++;
		*np++ = c; c = getC(DOEXCL);
	    }
	    if ((c == 'g' && !gmodflag) || (c == 'a' && !amodflag)) {
		if (c == 'g')
		    gmodflag++;
		else
		    amodflag++;
		*np++ = c; c = getC(DOEXCL);
	    }
	    *np++ = c;
	    /* scan s// [eichin:19910926.0512EST] */
	    if (c == 's') {
		int delimcnt = 2;
		int delim = getC(0);
		*np++ = delim;
		
		if (!delim || letter(delim)
		    || Isdigit(delim) || any(" \t\n", delim)) {
		    seterror(ERR_BADSUBST);
		    break;
		}	
		while ((c = getC(0)) != (-1)) {
		    *np++ = c;
		    if(c == delim) delimcnt--;
		    if(!delimcnt) break;
		}
		if(delimcnt) {
		    seterror(ERR_BADSUBST);
		    break;
		}
		c = 's';
	    }
	    if (!any("htrqxes", c)) {
		if ((amodflag || gmodflag) && c == '\n')
		    stderror(ERR_VARSYN);	/* strike */
		seterror(ERR_VARMOD, c);
		*np = 0;
		addla(name);
		return;
	    }
	}
	while ((c = getC(DOEXCL)) == ':');
	ungetD(c);
    }
    else
	ungetD(c);
    if (sc == '{') {
	c = getC(DOEXCL);
	if (c != '}') {
	    ungetD(c);
	    seterror(ERR_MISSING, '}');
	    *np = 0;
	    addla(name);
	    return;
	}
	*np++ = c;
    }
    *np = 0;
    addla(name);
    return;
}

void
addla(cp)
    Char   *cp;
{
    Char    buf[BUFSIZ];

    if (Strlen(cp) + (lap ? Strlen(lap) : 0) >=
	(sizeof(labuf) - 4) / sizeof(Char)) {
	seterror(ERR_EXPOVFL);
	return;
    }
    if (lap)
	(void) Strcpy(buf, lap);
    (void) Strcpy(labuf, cp);
    if (lap)
	(void) Strcat(labuf, buf);
    lap = labuf;
}

static Char lhsb[32];
static Char slhs[32];
static Char rhsb[64];
static int quesarg;

static void
getexcl(sc)
    int    sc;
{
    register struct wordent *hp, *ip;
    int     left, right, dol;
    register int c;

    if (sc == 0) {
	sc = getC(0);
	if (sc != '{') {
	    ungetC(sc);
	    sc = 0;
	}
    }
    quesarg = -1;
    lastev = eventno;
    hp = gethent(sc);
    if (hp == 0)
	return;
    hadhist = 1;
    dol = 0;
    if (hp == alhistp)
	for (ip = hp->next->next; ip != alhistt; ip = ip->next)
	    dol++;
    else
	for (ip = hp->next->next; ip != hp->prev; ip = ip->next)
	    dol++;
    left = 0, right = dol;
    if (sc == HISTSUB) {
	ungetC('s'), unreadc(HISTSUB), c = ':';
	goto subst;
    }
    c = getC(0);
    if (!any(":^$*-%", c))
	goto subst;
    left = right = -1;
    if (c == ':') {
	c = getC(0);
	unreadc(c);
	if (letter(c) || c == '&') {
	    c = ':';
	    left = 0, right = dol;
	    goto subst;
	}
    }
    else
	ungetC(c);
    if (!getsel(&left, &right, dol))
	return;
    c = getC(0);
    if (c == '*')
	ungetC(c), c = '-';
    if (c == '-') {
	if (!getsel(&left, &right, dol))
	    return;
	c = getC(0);
    }
subst:
    exclc = right - left + 1;
    while (--left >= 0)
	hp = hp->next;
    if (sc == HISTSUB || c == ':') {
	do {
	    hp = getsub(hp);
	    c = getC(0);
	} while (c == ':');
    }
    unreadc(c);
    if (sc == '{') {
	c = getC(0);
	if (c != '}')
	    seterror(ERR_BADBANG);
    }
    exclnxt = hp;
}

static struct wordent *
getsub(en)
    struct wordent *en;
{
    register Char *cp;
    int     delim;
    register int c;
    int     sc;
    bool global;
    Char    orhsb[sizeof(rhsb) / sizeof(Char)];

    do {
	exclnxt = 0;
	global = 0;
	sc = c = getC(0);
	if (c == 'g' || c == 'a') {
	    global |= (c == 'g') ? 1 : 2;
	    sc = c = getC(0);
	}
	if (((c =='g') && !(global & 1)) || ((c == 'a') && !(global & 2))) {
	    global |= (c == 'g') ? 1 : 2;
	    sc = c = getC(0);
	}

	switch (c) {
	case 'p':
	    justpr++;
	    return (en);

	case 'x':
	case 'q':
	    global |= 1;

	    /* fall into ... */

	case 'h':
	case 'r':
	case 't':
	case 'e':
	    break;

	case '&':
	    if (slhs[0] == 0) {
		seterror(ERR_NOSUBST);
		return (en);
	    }
	    (void) Strcpy(lhsb, slhs);
	    break;

#ifdef notdef
	case '~':
	    if (lhsb[0] == 0)
		goto badlhs;
	    break;
#endif

	case 's':
	    delim = getC(0);
	    if (letter(delim) || Isdigit(delim) || any(" \t\n", delim)) {
		unreadc(delim);
		lhsb[0] = 0;
		seterror(ERR_BADSUBST);
		return (en);
	    }
	    cp = lhsb;
	    for (;;) {
		c = getC(0);
		if (c == '\n') {
		    unreadc(c);
		    break;
		}
		if (c == delim)
		    break;
		if (cp > &lhsb[sizeof(lhsb) / sizeof(Char) - 2]) {
		    lhsb[0] = 0;
		    seterror(ERR_BADSUBST);
		    return (en);
		}
		if (c == '\\') {
		    c = getC(0);
		    if (c != delim && c != '\\')
			*cp++ = '\\';
		}
		*cp++ = c;
	    }
	    if (cp != lhsb)
		*cp++ = 0;
	    else if (lhsb[0] == 0) {
		seterror(ERR_LHS);
		return (en);
	    }
	    cp = rhsb;
	    (void) Strcpy(orhsb, cp);
	    for (;;) {
		c = getC(0);
		if (c == '\n') {
		    unreadc(c);
		    break;
		}
		if (c == delim)
		    break;
#ifdef notdef
		if (c == '~') {
		    if (&cp[Strlen(orhsb)] > &rhsb[sizeof(rhsb) /
						   sizeof(Char) - 2])
			goto toorhs;
		    (void) Strcpy(cp, orhsb);
		    cp = Strend(cp);
		    continue;
		}
#endif
		if (cp > &rhsb[sizeof(rhsb) / sizeof(Char) - 2]) {
		    seterror(ERR_RHSLONG);
		    return (en);
		}
		if (c == '\\') {
		    c = getC(0);
		    if (c != delim /* && c != '~' */ )
			*cp++ = '\\';
		}
		*cp++ = c;
	    }
	    *cp++ = 0;
	    break;

	default:
	    if (c == '\n')
		unreadc(c);
	    seterror(ERR_BADBANGMOD, c);
	    return (en);
	}
	(void) Strcpy(slhs, lhsb);
	if (exclc)
	    en = dosub(sc, en, global);
    }
    while ((c = getC(0)) == ':');
    unreadc(c);
    return (en);
}

static struct wordent *
dosub(sc, en, global)
    int     sc;
    struct wordent *en;
    bool global;
{
    struct wordent lexi;
    bool    didsub = 0, didone = 0;
    struct wordent *hp = &lexi;
    register struct wordent *wdp;
    register int i = exclc;

    wdp = hp;
    while (--i >= 0) {
	register struct wordent *new = 
		(struct wordent *) xcalloc(1, sizeof *wdp);

	new->word = 0;
	new->prev = wdp;
	new->next = hp;
	wdp->next = new;
	wdp = new;
	en = en->next;
	if (en->word) {
	    Char *tword, *otword;

	    if ((global & 1) || didsub == 0) {
		tword = subword(en->word, sc, &didone);
		if (didone)
		    didsub = 1;
		if (global & 2) {
		    while (didone && tword != STRNULL) {
			otword = tword;
			tword = subword(otword, sc, &didone);
			if (Strcmp(tword, otword) == 0) {
			    xfree((ptr_t) otword);
			    break;
			}
			else
			    xfree((ptr_t) otword);
		    }
		}
	    }
	    else
		tword = Strsave(en->word);
	    wdp->word = tword;
	}
    }
    if (didsub == 0)
	seterror(ERR_MODFAIL);
    hp->prev = wdp;
    return (&enthist(-1000, &lexi, 0)->Hlex);
}

static Char *
subword(cp, type, adid)
    Char   *cp;
    int     type;
    bool   *adid;
{
    Char    wbuf[BUFSIZ];
    register Char *wp, *mp, *np;
    register int i;

    *adid = 0;
    switch (type) {

    case 'r':
    case 'e':
    case 'h':
    case 't':
    case 'q':
    case 'x':
	wp = domod(cp, type);
	if (wp == 0)
	    return (Strsave(cp));
	*adid = 1;
	return (wp);

    default:
	wp = wbuf;
	i = BUFSIZ - 4;
	for (mp = cp; *mp; mp++)
	    if (matchs(mp, lhsb)) {
		for (np = cp; np < mp;)
		    *wp++ = *np++, --i;
		for (np = rhsb; *np; np++)
		    switch (*np) {

		    case '\\':
			if (np[1] == '&')
			    np++;
			/* fall into ... */

		    default:
			if (--i < 0) {
			    seterror(ERR_SUBOVFL);
			    return (STRNULL);
			}
			*wp++ = *np;
			continue;

		    case '&':
			i -= Strlen(lhsb);
			if (i < 0) {
			    seterror(ERR_SUBOVFL);
			    return (STRNULL);
			}
			*wp = 0;
			(void) Strcat(wp, lhsb);
			wp = Strend(wp);
			continue;
		    }
		mp += Strlen(lhsb);
		i -= Strlen(mp);
		if (i < 0) {
		    seterror(ERR_SUBOVFL);
		    return (STRNULL);
		}
		*wp = 0;
		(void) Strcat(wp, mp);
		*adid = 1;
		return (Strsave(wbuf));
	    }
	return (Strsave(cp));
    }
}

Char   *
domod(cp, type)
    Char   *cp;
    int     type;
{
    register Char *wp, *xp;
    register int c;

    switch (type) {

    case 'x':
    case 'q':
	wp = Strsave(cp);
	for (xp = wp; (c = *xp) != '\0'; xp++)
	    if ((c != ' ' && c != '\t') || type == 'q')
		*xp |= QUOTE;
	return (wp);

    case 'h':
    case 't':
	if (!any(short2str(cp), '/'))
	    return (type == 't' ? Strsave(cp) : 0);
	wp = Strend(cp);
	while (*--wp != '/')
	    continue;
	if (type == 'h')
	    xp = Strsave(cp), xp[wp - cp] = 0;
	else
	    xp = Strsave(wp + 1);
	return (xp);

    case 'e':
    case 'r':
	wp = Strend(cp);
	for (wp--; wp >= cp && *wp != '/'; wp--)
	    if (*wp == '.') {
		if (type == 'e')
		    xp = Strsave(wp + 1);
		else
		    xp = Strsave(cp), xp[wp - cp] = 0;
		return (xp);
	    }
	return (Strsave(type == 'e' ? STRNULL : cp));
    default:
	break;
    }
    return (0);
}

static int
matchs(str, pat)
    register Char *str, *pat;
{
    while (*str && *pat && *str == *pat)
	str++, pat++;
    return (*pat == 0);
}

static int
getsel(al, ar, dol)
    register int *al, *ar;
    int     dol;
{
    register int c = getC(0);
    register int i;
    bool    first = *al < 0;

    switch (c) {

    case '%':
	if (quesarg == -1) {
	    seterror(ERR_BADBANGARG);
	    return (0);
	}
	if (*al < 0)
	    *al = quesarg;
	*ar = quesarg;
	break;

    case '-':
	if (*al < 0) {
	    *al = 0;
	    *ar = dol - 1;
	    unreadc(c);
	}
	return (1);

    case '^':
	if (*al < 0)
	    *al = 1;
	*ar = 1;
	break;

    case '$':
	if (*al < 0)
	    *al = dol;
	*ar = dol;
	break;

    case '*':
	if (*al < 0)
	    *al = 1;
	*ar = dol;