dbz.c

早期freebsd实现

	register char *sepp;

	DEBUG(("fetch: (%s)\n", key.dptr));
	output.dptr = NULL;
	output.dsize = 0;
	prevp = FRESH;

	/* Key is supposed to be less than DBZMAXKEY */
	keysize = key.dsize;
	if (keysize >= DBZMAXKEY) {
		keysize = DBZMAXKEY;
		DEBUG(("keysize is %d - truncated to %d\n", key.dsize, DBZMAXKEY));
	}

	if (pagf == NULL) {
		DEBUG(("fetch: database not open!\n"));
		return(output);
	} else if (basef == NULL) {	/* basef didn't exist yet */
		basef = latebase();
		if (basef == NULL)
			return(output);
	}

	cmplen = keysize;
	sepp = &conf.fieldsep;
	if (key.dptr[keysize-1] == '\0') {
		cmplen--;
		sepp = &buffer[keysize-1];
	}
	start(&srch, &key, FRESH);
	while ((key_ptr = search(&srch)) != NOTFOUND) {
		DEBUG(("got 0x%lx\n", key_ptr));

		/* fetch the key */
		if (fseek(basef, key_ptr, SEEK_SET) != 0) {
			DEBUG(("fetch: seek failed\n"));
			return(output);
		}
		if (fread((POINTER)buffer, 1, keysize, basef) != keysize) {
			DEBUG(("fetch: read failed\n"));
			return(output);
		}

		/* try it */
		buffer[keysize] = '\0';		/* terminated for DEBUG */
		(void) mapcase(buffer, buffer, keysize);
		DEBUG(("fetch: buffer (%s) looking for (%s) size = %d\n", 
						buffer, key.dptr, keysize));
		if (memcmp((POINTER)key.dptr, (POINTER)buffer, cmplen) == 0 &&
				(*sepp == conf.fieldsep || *sepp == '\0')) {
			/* we found it */
			output.dptr = (char *)&key_ptr;
			output.dsize = SOF;
			DEBUG(("fetch: successful\n"));
			return(output);
		}
	}

	/* we didn't find it */
	DEBUG(("fetch: failed\n"));
	prevp = &srch;			/* remember where we stopped */
	return(output);
}

/*
 - latebase - try to open a base file that wasn't there at the start
 */
static FILE *
latebase()
{
	register FILE *it;

	if (basefname == NULL) {
		DEBUG(("latebase: name foulup\n"));
		return(NULL);
	}
	it = fopen(basefname, "r");
	if (it == NULL) {
		DEBUG(("latebase: still can't open base\n"));
	} else {
		DEBUG(("latebase: late open succeeded\n"));
		free((POINTER)basefname);
		basefname = NULL;
#ifdef _IOFBF
		(void) setvbuf(it, basebuf, _IOFBF, sizeof(basebuf));
#endif
	}
	if (it != NULL)
		CloseOnExec((int)fileno(it), 1);
	return(it);
}

/*
 - dbzstore - store() with case mapping built in
 */
int
dbzstore(key, data)
datum key;
datum data;
{
	char buffer[DBZMAXKEY + 1];
	datum mappedkey;
	register size_t keysize;

	DEBUG(("dbzstore: (%s)\n", key.dptr));

	/* Key is supposed to be less than DBZMAXKEY */
	keysize = key.dsize;
	if (keysize >= DBZMAXKEY) {
		DEBUG(("dbzstore: key size too big (%d)\n", key.dsize));
		return(-1);
	}

	mappedkey.dptr = mapcase(buffer, key.dptr, keysize);
	buffer[keysize] = '\0';	/* just a debug aid */
	mappedkey.dsize = keysize;

	return(store(mappedkey, data));
}

/*
 - store - add an entry to the database
 */
int				/* 0 success, -1 failure */
store(key, data)
datum key;
datum data;
{
	of_t value;

	if (pagf == NULL) {
		DEBUG(("store: database not open!\n"));
		return(-1);
	} else if (basef == NULL) {	/* basef didn't exist yet */
		basef = latebase();
		if (basef == NULL)
			return(-1);
	}
	if (pagronly) {
		DEBUG(("store: database open read-only\n"));
		return(-1);
	}
	if (data.dsize != SOF) {
		DEBUG(("store: value size wrong (%d)\n", data.dsize));
		return(-1);
	}
	if (key.dsize >= DBZMAXKEY) {
		DEBUG(("store: key size too big (%d)\n", key.dsize));
		return(-1);
	}

	/* copy the value in to ensure alignment */
	(void) memcpy((POINTER)&value, (POINTER)data.dptr, SOF);
	DEBUG(("store: (%s, %ld)\n", key.dptr, (long)value));
	if (!okayvalue(value)) {
		DEBUG(("store: reserved bit or overflow in 0x%lx\n", value));
		return(-1);
	}

	/* find the place, exploiting previous search if possible */
	start(&srch, &key, prevp);
	while (search(&srch) != NOTFOUND)
		continue;

	prevp = FRESH;
	conf.used[0]++;
	DEBUG(("store: used count %ld\n", conf.used[0]));
	written = 1;
	return(set(&srch, value));
}

/*
 - dbzincore - control attempts to keep .pag file in core
 */
int				/* old setting */
dbzincore(value)
int value;
{
	register int old = incore;

#ifndef MMAP
	incore = value;
#endif
	return(old);
}

/*
 - dbzwritethrough - write through the pag file in core
 */
int				/* old setting */
dbzwritethrough(value)
int value;
{
	register int old = writethrough;

	writethrough = value;
	return(old);
}

/*
 - dbztagmask - calculate the correct tagmask for the given base file size
 */
long
dbztagmask(size)
register long size;
{
	register long m;
	register long tagmask;
	register int i;

	if (size <= 0)
		return(0L);		/* silly size */

	for (m = 1, i = 0; m < size; i++, m <<= 1)
		continue;

	if (m < (1 << TAGSHIFT))
		return(0L);		/* not worth tagging */

	tagmask = (~(unsigned long)0) >> (i + 1);
	tagmask = tagmask << i;
	return(tagmask);
}

/*
 - getconf - get configuration from .dir file
 */
static int			/* 0 success, -1 failure */
getconf(df, pf, cp)
register FILE *df;		/* NULL means just give me the default */
register FILE *pf;		/* NULL means don't care about .pag */
register struct dbzconfig *cp;
{
	register int c;
	register int i;
	int err = 0;

	c = (df != NULL) ? getc(df) : EOF;
	if (c == EOF) {		/* empty file, no configuration known */
		cp->olddbz = 0;
		if (df != NULL && pf != NULL && getc(pf) != EOF)
			cp->olddbz = 1;
		cp->tsize = DEFSIZE;
		cp->fieldsep = '\t';
		for (i = 0; i < NUSEDS; i++)
			cp->used[i] = 0;
		cp->valuesize = SOF;
		mybytemap(cp->bytemap);
		cp->casemap = DEFCASE;
		cp->tagenb = TAGENB;
		cp->tagmask = TAGMASK;
		cp->tagshift = TAGSHIFT;
		DEBUG(("getconf: defaults (%ld, %c, (0x%lx/0x%lx<<%d))\n",
			cp->tsize, cp->casemap, cp->tagenb, 
			cp->tagmask, cp->tagshift));
		return(0);
	}
	(void) ungetc(c, df);

	/* first line, the vital stuff */
	if (getc(df) != 'd' || getc(df) != 'b' || getc(df) != 'z')
		err = -1;
	if (getno(df, &err) != dbzversion)
		err = -1;
	cp->tsize = getno(df, &err);
	cp->fieldsep = (int)getno(df, &err);
	while ((c = getc(df)) == ' ')
		continue;
	cp->casemap = c;
	cp->tagenb = getno(df, &err);
	cp->tagmask = getno(df, &err);
	cp->tagshift = getno(df, &err);
	cp->valuesize = getno(df, &err);
	if (cp->valuesize != SOF) {
		DEBUG(("getconf: wrong of_t size (%d)\n", cp->valuesize));
		err = -1;
		cp->valuesize = SOF;	/* to protect the loops below */
	}
	for (i = 0; i < cp->valuesize; i++)
		cp->bytemap[i] = getno(df, &err);
	if (getc(df) != '\n')
		err = -1;
#ifdef DBZDEBUG
	DEBUG(("size %ld, sep %d, cmap %c, tags 0x%lx/0x%lx<<%d, ", cp->tsize,
			cp->fieldsep, cp->casemap, cp->tagenb, cp->tagmask,
			cp->tagshift));
	DEBUG(("bytemap (%d)", cp->valuesize));
	for (i = 0; i < cp->valuesize; i++) {
		DEBUG((" %d", cp->bytemap[i]));
	}
	DEBUG(("\n"));
#endif

	/* second line, the usages */
	for (i = 0; i < NUSEDS; i++)
		cp->used[i] = getno(df, &err);
	if (getc(df) != '\n')
		err = -1;
	DEBUG(("used %ld %ld %ld...\n", cp->used[0], cp->used[1], cp->used[2]));

	if (err < 0) {
		DEBUG(("getconf error\n"));
		return(-1);
	}
	return(0);
}

/*
 - getno - get a long
 */
static long
getno(f, ep)
FILE *f;
int *ep;
{
	register char *p;
#	define	MAXN	50
	char getbuf[MAXN];
	register int c;

	while ((c = getc(f)) == ' ')
		continue;
	if (c == EOF || c == '\n') {
		DEBUG(("getno: missing number\n"));
		*ep = -1;
		return(0);
	}
	p = getbuf;
	*p++ = c;
	while ((c = getc(f)) != EOF && c != '\n' && c != ' ')
		if (p < &getbuf[MAXN-1])
			*p++ = c;
	if (c == EOF) {
		DEBUG(("getno: EOF\n"));
		*ep = -1;
	} else
		(void) ungetc(c, f);
	*p = '\0';

	if (strspn(getbuf, "-1234567890") != strlen(getbuf)) {
		DEBUG(("getno: `%s' non-numeric\n", getbuf));
		*ep = -1;
	}
	return(atol(getbuf));
}

/*
 - putconf - write configuration to .dir file
 */
static int			/* 0 success, -1 failure */
putconf(f, cp)
register FILE *f;
register struct dbzconfig *cp;
{
	register int i;
	register int ret = 0;

	if (fseek(f, (of_t)0, SEEK_SET) != 0) {
		DEBUG(("fseek failure in putconf\n"));
		ret = -1;
	}
	(void) fprintf(f, "dbz %d %ld %d %c %ld %ld %d %d", dbzversion, cp->tsize,
				cp->fieldsep, cp->casemap, cp->tagenb,
				cp->tagmask, cp->tagshift, cp->valuesize);
	for (i = 0; i < cp->valuesize; i++)
		(void) fprintf(f, " %d", cp->bytemap[i]);
	(void) fprintf(f, "\n");
	for (i = 0; i < NUSEDS; i++)
		(void) fprintf(f, "%ld%c", cp->used[i], (i < NUSEDS-1) ? ' ' : '\n');

	(void) fflush(f);
	if (ferror(f))
		ret = -1;

	DEBUG(("putconf status %d\n", ret));
	return(ret);
}

/*
 - getcore - try to set up an in-core copy of .pag file
 */
static of_t *			/* pointer to copy, or NULL */
getcore(f)
FILE *f;
{
	register of_t *p;
	register size_t i;
	register size_t nread;
	register char *it;
#ifdef MMAP
	struct stat st;

	if (fstat(fileno(f), &st) == -1) {
		DEBUG(("getcore: fstat failed\n"));
		return(NULL);
	}
	if (((size_t)conf.tsize * SOF) > st.st_size) {
		/* file too small; extend it */
		if (ftruncate((int)fileno(f), conf.tsize * SOF) == -1) {
			DEBUG(("getcore: ftruncate failed\n"));
			return(NULL);
		}
	}
	it = mmap((caddr_t)0, (size_t)conf.tsize * SOF, 
		pagronly ? PROT_READ : PROT_WRITE | PROT_READ, MAP__ARG,
		(int)fileno(f), (off_t)0);
	if (it == (char *)-1) {
		DEBUG(("getcore: mmap failed\n"));
		return(NULL);
	}
#ifdef MC_ADVISE
	/* not present in all versions of mmap() */
	madvise(it, (size_t)conf.tsize * SOF, MADV_RANDOM);
#endif
#else
	it = malloc((size_t)conf.tsize * SOF);
	if (it == NULL) {
		DEBUG(("getcore: malloc failed\n"));
		return(NULL);
	}

	nread = fread((POINTER)it, SOF, (size_t)conf.tsize, f);
	if (ferror(f)) {
		DEBUG(("getcore: read failed\n"));
		free((POINTER)it);
		return(NULL);
	}

 	/* NOSTRICT *//* Possible pointer alignment problem */
	p = (of_t *)it + nread;
	i = (size_t)conf.tsize - nread;
	while (i-- > 0)
		*p++ = VACANT;
#endif
 	/* NOSTRICT *//* Possible pointer alignment problem */
	return((of_t *)it);
}

#ifndef MMAP
/*
 - putcore - try to rewrite an in-core table
 */
static int			/* 0 okay, -1 fail */
putcore(tab, f)
of_t *tab;
FILE *f;
{
	if (fseek(f, (of_t)0, SEEK_SET) != 0) {
		DEBUG(("fseek failure in putcore\n"));
		return(-1);
	}
	(void) fwrite((POINTER)tab, SOF, (size_t)conf.tsize, f);
	(void) fflush(f);
	return((ferror(f)) ? -1 : 0);
}
#endif

/*
 - start - set up to start or restart a search
 */
static void
start(sp, kp, osp)
register struct searcher *sp;
register datum *kp;
register struct searcher *osp;		/* may be FRESH, i.e. NULL */
{
	register long h;

	h = hash(kp->dptr, kp->dsize);
	if (osp != FRESH && osp->hash == h) {
		if (sp != osp)
			*sp = *osp;
		DEBUG(("search restarted\n"));
	} else {
		sp->hash = h;
		sp->tag = MKTAG(h / conf.tsize);
		DEBUG(("tag 0x%lx\n", sp->tag));
		sp->place = h % conf.tsize;
		sp->tabno = 0;
		sp->run = (conf.olddbz) ? conf.tsize : MAXRUN;