sprog.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

	if(lev > DLEV+1)
		return 0;
	if(*ep=='s') {
		switch(ep[-1]) {
		case 'y':
			if(ISVOWEL(ep[-2])||ISUPPER(*word))
				break;	/*says Kennedys*/
		case 'x':
		case 'z':
		case 's':
			return 0;
		case 'h':
			switch(ep[-2]) {
			case 'c':
			case 's':
				return 0;
			}
		}
	}
	return strip(ep,d,a,lev,flag);
}

Bits
an(char* ep, char* d, char* a, int lev, int flag)
{
	USED(d);
	if(!ISUPPER(*word))	/*must be proper name*/
		return 0;
	return trypref(ep,a,lev,flag);
}

Bits
ize(char* ep, char* d, char* a, int lev, int flag)
{
	int temp = ep[-1];
	Bits h;

	USED(a);
	ep[-1] = 'e';
	h = strip(ep,"",d,lev,flag);
	ep[-1] = temp;
	return h;
}

Bits
y_to_e(char* ep, char* d, char* a, int lev, int flag)
{
	Bits h;
	int  temp;

	USED(a);
	switch(ep[-1]) {
	case 'a':
	case 'e':
	case 'i':
		return 0;
	}
	temp = *ep;
	*ep++ = 'e';
	h = strip(ep,"",d,lev,flag);
	ep[-1] = temp;
	return h;
}

Bits
ily(char* ep, char* d, char* a, int lev, int flag)
{
	int temp = ep[0];
	char *cp = ep;

	if(temp==ep[-1]&&temp==ep[-2])		/* sillly */
		return 0;
	if(*--cp=='y' && !ISVOWEL(*--cp))	/* happyly */
		while(cp>word)
			if(ISVOWEL(*--cp))	/* shyness */
				return 0;
	if(ep[-1]=='i')
		return i_to_y(ep,d,a,lev,flag);
	return cstrip(ep,d,a,lev,flag);
}

Bits
bility(char* ep, char* d, char* a, int lev, int flag)
{
	*ep++ = 'l';
	return y_to_e(ep,d,a,lev,flag);
}

Bits
i_to_y(char* ep, char* d, char* a, int lev, int flag)
{
	Bits h;
	int temp;

	if(ISUPPER(*word))
		return 0;
	if((temp=ep[-1])=='i' && !ISVOWEL(ep[-2])) {
		ep[-1] = 'y';
		a = d;
	}
	h = cstrip(ep,"",a,lev,flag);
	ep[-1] = temp;
	return h;
}

Bits
es(char* ep, char* d, char* a, int lev, int flag)
{
	if(lev>DLEV)
		return 0;
	switch(ep[-1]) {
	default:
		return 0;
	case 'i':
		return i_to_y(ep,d,a,lev,flag);
	case 'h':
		switch(ep[-2]) {
		default:
			return 0;
		case 'c':
		case 's':
			break;
		}
	case 's':
	case 'z':
	case 'x':
		return strip(ep,d,a,lev,flag);
	}
}

Bits
subst(char* ep, char* d, char* a, int lev, int flag)
{
	char *u,*t;
	Bits h;

	USED(a);
	if(skipv(skipv(ep-1)) < word)
		return 0;
	for(t=d; *t!='+'; t++)
		continue;
	for(u=ep; *--t!='-';)
		*--u = *t;
	h = strip(ep,"",d,lev,flag);
	while(*++t != '+')
		continue;
	while(*++t)
		*u++ = *t;
	return h;
}

Bits
tion(char* ep, char* d, char* a, int lev, int flag)
{
	switch(ep[-2]) {
	default:
		return trypref(ep,a,lev,flag);
	case 'a':
	case 'e':
	case 'i':
	case 'o':
	case 'u':
		return y_to_e(ep,d,a,lev,flag);
	}
}

/*
 * possible consonant-consonant-e ending
 */
Bits
CCe(char* ep, char* d, char* a, int lev, int flag)
{
	Bits h;

	switch(ep[-1]) {
	case 'l':
		if(ISVOWEL(ep[-2]))
			break;
		switch(ep[-2]) {
		case 'l':
		case 'r':
		case 'w':
			break;
		default:
			return y_to_e(ep,d,a,lev,flag);
		}
		break;
	case 'c':
	case 'g':
		if(*ep == 'a')	/* prevent -able for -eable */
			return 0;
	case 's':
	case 'v':
	case 'z':
		if(ep[-2]==ep[-1])
			break;
		if(ISVOWEL(ep[-2]))
			break;
	case 'u':
		if(h = y_to_e(ep,d,a,lev,flag))
			return h;
		if(!(ep[-2]=='n' && ep[-1]=='g'))
			return 0;
	}
	return VCe(ep,d,a,lev,flag);
}

/*
 * possible consonant-vowel-consonant-e ending
 */
Bits
VCe(char* ep, char* d, char* a, int lev, int flag)
{
	int c;
	Bits h;

	c = ep[-1];
	if(c=='e')
		return 0;
	if(!ISVOWEL(c) && ISVOWEL(ep[-2])) {
		c = *ep;
		*ep++ = 'e';
		h = trypref(ep,d,lev,flag);
		if(!h)
			h = trysuff(ep,lev,flag);
		if(h)
			return h;
		ep--;
		*ep = c;
	}
	return cstrip(ep,d,a,lev,flag);
}

Ptab*
lookuppref(uchar** wp, char* ep)
{
	Ptab *sp;
	uchar *bp,*cp;
	unsigned int initchar = Tolower(**wp);

	if(!ISALPHA(initchar))
		return 0;
	for(sp=preftab[initchar-'a'];sp->s;sp++) {
		bp = *wp;
		for(cp= (uchar*)sp->s;*cp; )
			if(*bp++!=*cp++)
				goto next;
		for(cp=bp;cp<(uchar*)ep;cp++)
			if(ISVOWEL(*cp)) {
				*wp = bp;
				return sp;
			}
	next:;
	}
	return 0;
}

/*	while word is not in dictionary try stripping
 *	prefixes. Fail if no more prefixes.
 */
Bits
trypref(char* ep, char* a, int lev, int flag)
{
	Ptab *tp;
	char *bp, *cp;
	char *pp;
	Bits h;
	char space[20];

	if(lev<DSIZ) {
		deriv[lev].mesg = a;
		deriv[lev].type = *a=='.'? NONE: SUFF;
	}
	if(h = tryword(word,ep,lev,flag)) {
		if(Set(h, flag&~MONO) && (flag&MONO) <= Set(h, MONO))
			return h;
		h = 0;
	}
	bp = word;
	pp = space;
	if(lev<DSIZ) {
		deriv[lev+1].mesg = pp;
		deriv[lev+1].type = 0;
	}
	while(tp=lookuppref((uchar**)&bp,ep)) {
		*pp++ = '+';
		cp = tp->s;
		while(pp<space+sizeof(space) && (*pp = *cp++))
			pp++;
		deriv[lev+1].type += PREF;
		h = tryword(bp,ep,lev+1,flag);
		if(Set(h,NOPREF) ||
		   ((tp->flag&IN) && inun(bp-2,h)==0)) {
			h = 0;
			break;
		}
		if(Set(h,flag&~MONO) && (flag&MONO) <= Set(h, MONO))
			break;
		h = 0;
	}
	if(lev < DSIZ) {
		deriv[lev+1] = emptyderiv;
		deriv[lev+2] = emptyderiv;
	}
	return h;
}

Bits
tryword(char* bp, char* ep, int lev, int flag)
{
	int  j;
	Bits h = 0;
	char duple[3];

	if(ep-bp <= 1)
		return h;
	if(flag&MONO) {
		if(lev<DSIZ) {
			deriv[++lev].mesg = duple;
			deriv[lev].type = SUFF;
		}
		duple[0] = '+';
		duple[1] = *ep;
		duple[2] = 0;
	}
	h = dict(bp, ep);
	if(vflag==0 || h==0)
		return h;
	/*
	 * when derivations are wanted, collect them
	 * for printing
	 */
	j = lev;
	prefcount = suffcount = 0;
	do {
		if(j<DSIZ && deriv[j].type) {
			strcat(affix, deriv[j].mesg);
			if(deriv[j].type == SUFF)
				suffcount++;
			else if(deriv[j].type != NONE)
				prefcount = deriv[j].type/PREF;
		}
	} while(--j > 0);
	return h;
}

int
inun(char* bp, Bits h)
{
	if(*bp == 'u')
		return Set(h, IN) == 0;
	/* *bp == 'i' */
	if(Set(h, IN) == 0)
		return 0;
	switch(bp[2]) {
	case 'r':
		return bp[1] == 'r';
	case 'm':
	case 'p':
		return bp[1] == 'm';
	}
	return bp[1] == 'n';
}

char*
skipv(char *s)
{
	if(s >= word && ISVOWEL(*s))
		s--;
	while(s >= word && !ISVOWEL(*s))
		s--;
	return s;
}

/*
 * crummy way to Britishise
 */
void
ise(void)
{
	Suftab *p;
	int i;

	for(i=0; i<26; i++)
		for(p = suftab[i]; p->suf; p++) {
			p->suf = ztos(p->suf);
			p->d1 = ztos(p->d1);
			p->a1 = ztos(p->a1);
		}
}

char*
ztos(char *as)
{
	char *s, *ds;

	for(s=as; *s; s++)
		if(*s == 'z')
			goto copy;
	return as;

copy:
	ds = strdup(as);
	for(s=ds; *s; s++)
		if(*s == 'z')
			*s = 's';
	return ds;
}

Bits
dict(char* bp, char* ep)
{
	char *cp, *cp1, *w, *wp, *we;
	int n, f;

	w = bp;
	we = ep;
	n = ep-bp;
	if(n <= 1)
		return NOUN;

	f = w[0] & 0x7f;
	f *= 128;
	f += w[1] & 0x7f;
	bp = spacep[f];
	ep = spacep[f+1];

loop:
	if(bp >= ep) {
		if(xflag) 
			fprint(2, "=%.*s\n", utfnlen(w, n), w);
		return 0;
	}
	/*
	 * find the beginning of some word in the middle
	 */
	cp = bp + (ep-bp)/2;

	while(cp > bp && !(*cp & 0x80))
		cp--;
	while(cp > bp && (cp[-1] & 0x80))
		cp--;

	wp = w + 2;	/* skip two letters */
	cp1 = cp + 2;	/* skip affix code */
	for(;;) {
		if(wp >= we) {
			if(*cp1 & 0x80)
				goto found;
			else
				f = 1;
			break;
		}
		if(*cp1 & 0x80) {
			f = -1;
			break;
		}
		f = *cp1++ - *wp++;
		if(f != 0)
			break;
	}

	if(f < 0) {
		while(!(*cp1 & 0x80))
			cp1++;
		bp = cp1;
		goto loop;
	}
	ep = cp;
	goto loop;

found:
	f = ((cp[0] & 0x7) << 8) |
		(cp[1] & 0xff);
	if(xflag) {
		fprint(2, "=%.*s ", utfnlen(w, n), w);
		typeprint(encode[f]);
	}
	return encode[f];
}

void
typeprint(Bits h)
{

	pcomma("");
	if(h & NOUN)
		pcomma("n");
	if(h & PROP_COLLECT)
		pcomma("pc");
	if(h & VERB) {
		if((h & VERB) == VERB)
			pcomma("v");
		else
		if((h & VERB) == V_IRREG)
			pcomma("vi");
		else
		if(h & ED)
			pcomma("ed");
	}
	if(h & ADJ)
		pcomma("a");
	if(h & COMP) {
		if((h & COMP) == ACTOR)
			pcomma("er");
		else
			pcomma("comp");
	}
	if(h & DONT_TOUCH)
		pcomma("d");
	if(h & N_AFFIX)
		pcomma("na");
	if(h & ADV)
		pcomma("adv");
	if(h & ION)
		pcomma("ion");
	if(h & V_AFFIX)
		pcomma("va");
	if(h & MAN)
		pcomma("man");
	if(h & NOPREF)
		pcomma("nopref");
	if(h & MONO)
		pcomma("ms");
	if(h & IN)
		pcomma("in");
	if(h & _Y)
		pcomma("y");
	if(h & STOP)
		pcomma("s");
	fprint(2, "\n");
}

void
pcomma(char *s)
{
	static flag;

	if(*s == 0) {
		flag = 0;
		return;
	}
	if(!flag) {
		fprint(2, "%s", s);
		flag = 1;
	} else
		fprint(2, ",%s", s);
}

/*
 * is the word on of the following
 *	12th	teen
 *	21st	end in 1
 *	23rd	end in 3
 *	77th	default
 * called knowing word[0] is a digit
 */
int
ordinal(void)
{
	char *cp = word;
	static char sp[4];

	while(ISDIGIT(*cp))
		cp++;
	strncpy(sp,cp,3);
	if(ISUPPER(cp[0]) && ISUPPER(cp[1])) {
		sp[0] = Tolower(cp[0]);
		sp[1] = Tolower(cp[1]);
	}
	return 0 == strncmp(sp,
		cp[-2]=='1'? "th":	/* out of bounds if 1 digit */
		*--cp=='1'? "st":	/* harmless */
		*cp=='2'? "nd":
		*cp=='3'? "rd":
		"th", 3);
}

/*
 * read in the dictionary.
 * format is
 * {
 *	short	nencode;
 *	long	encode[nencode];
 *	char	space[*];
 * };
 *
 * the encodings are a table all different
 * affixes.
 * the dictionary proper has 2 bytes
 * that demark and then the rest of the
 * word. the 2 bytes have the following
 *	0x80 0x00	flag
 *	0x78 0x00	count of prefix bytes
 *			common with prev word
 *	0x07 0xff	affix code
 *
 * all ints are big endians in the file.
 */
void
readdict(char *file)
{
	char *s, *is, *lasts, *ls;
	int c, i, sp, p;
	int f;
	long l;

	lasts = 0;
	f = open(file, 0);
	if(f == -1) {
		fprint(2, "cannot open %s\n", file);
		exits("open");
	}
	if(read(f, space, 2) != 2)
		goto bad;
	nencode = ((space[0]&0xff)<<8) | (space[1]&0xff);
	if(read(f, space, 4*nencode) != 4*nencode)
		goto bad;
	s = space;
	for(i=0; i<nencode; i++) {
		l = (long)(s[0] & 0xff) << 24;
		l |= (s[1] & 0xff) << 16;
		l |= (s[2] & 0xff) << 8;
		l |= s[3] & 0xff;
		encode[i] = (Bits)l;
		s += 4;
	}
	l = read(f, space, sizeof(space));
	if(l == sizeof(space))
		goto noroom;
	is = space + (sizeof(space) - l);
	memmove(is, space, l);

	s = space;
	c = *is++ & 0xff;
	sp = -1;
	i = 0;

loop:
	if(s > is)
		goto noroom;
	if(c < 0) {
		close(f);
		while(sp < 128*128)
			spacep[++sp] = s;
		*s = 0x80;		/* fence */
		return;
	}
	p = (c>>3) & 0xf;
	*s++ = c;
	*s++ = *is++ & 0xff;
	if(p <= 0)
		i = (*is++ & 0xff)*128;
	if(p <= 1) {
		if(!(*is & 0x80))
			i = i/128*128 + (*is++ & 0xff);
		if(i <= sp) {
			fprint(2, "the dict isnt sorted or \n");
			fprint(2, "memmove didn't work\n");
			goto bad;
		}
		while(sp < i)
			spacep[++sp] = s-2;
	}
	ls = lasts;
	lasts = s;
	for(p-=2; p>0; p--)
		*s++ = *ls++;
	for(;;) {
		if(is >= space+sizeof(space)) {
			c = -1;
			break;
		}
		c = *is++ & 0xff;
		if(c & 0x80)
			break;
		*s++ = c;
	}
	*s = 0;
	goto loop;

bad:
	fprint(2, "trouble reading %s\n", file);
	exits("read");
noroom:
	fprint(2, "not enough space for dictionary\n");
	exits("space");
}