-/* $Id: isamb.c,v 1.31 2004-06-01 13:46:41 adam Exp $
+/* $Id: isamb.c,v 1.52 2004-08-06 12:28:23 adam Exp $
Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
Index Data Aps
#endif
struct ISAMB_head {
- int first_block;
- int last_block;
+ zint first_block;
+ zint last_block;
int block_size;
int block_max;
- int free_list;
+ zint free_list;
};
#define ISAMB_DATA_OFFSET 3
#define CAT_NO 4
/* ISAMB_PTR_CODEC=1 var, =0 fixed */
-#define ISAMB_PTR_CODEC 1
+#define ISAMB_PTR_CODEC 0
struct ISAMB_cache_entry {
ISAMB_P pos;
int deleted;
int offset;
char *bytes;
+ char *cbuf;
unsigned char *buf;
void *decodeClientData;
int log_rw;
struct ISAMB_block **block;
};
+
#if ISAMB_PTR_CODEC
-static void encode_ptr (char **dst, unsigned pos)
+static void encode_ptr (char **dst, zint pos)
{
unsigned char *bp = (unsigned char*) *dst;
*dst = (char *) bp;
}
#else
-static void encode_ptr (char **dst, unsigned pos)
+static void encode_ptr (char **dst, zint pos)
{
memcpy(*dst, &pos, sizeof(pos));
(*dst) += sizeof(pos);
#endif
#if ISAMB_PTR_CODEC
-static void decode_ptr (char **src1, int *pos)
+static void decode_ptr (const char **src1, zint *pos)
{
- unsigned char **src = (unsigned char **) src1;
- unsigned d = 0;
+ const unsigned char **src = (const unsigned char **) src1;
+ zint d = 0;
unsigned char c;
unsigned r = 0;
while (((c = *(*src)++) & 128))
{
- d += ((c & 127) << r);
+ d += ((zint) (c & 127) << r);
r += 7;
}
- d += (c << r);
+ d += ((zint) c << r);
*pos = d;
}
#else
-static void decode_ptr (char **src, int *pos)
+static void decode_ptr (const char **src, zint *pos)
{
memcpy (pos, *src, sizeof(*pos));
(*src) += sizeof(*pos);
assert(isamb->file[i].head.block_size == b_size);
b_size = b_size * 4;
}
+#if ISAMB_DEBUG
+ logf(LOG_WARN, "isamb debug enabled. Things will be slower than usual");
+#endif
return isamb;
}
static int get_block (ISAMB b, ISAMC_P pos, char *userbuf, int wr)
{
- int cat = pos&CAT_MASK;
- int off = ((pos/CAT_MAX) &
+ int cat = (int) (pos&CAT_MASK);
+ int off = (int) (((pos/CAT_MAX) &
(ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size - 1))
- * b->file[cat].head.block_size;
- int norm = pos / (CAT_MASK*ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size);
+ * b->file[cat].head.block_size);
+ zint norm = pos / (CAT_MASK*ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size);
int no = 0;
struct ISAMB_cache_entry **ce, *ce_this = 0, **ce_last = 0;
xfree (isamb);
}
-
-struct ISAMB_block *open_block (ISAMB b, ISAMC_P pos)
+static struct ISAMB_block *open_block (ISAMB b, ISAMC_P pos)
{
- int cat = pos&CAT_MASK;
+ int cat = (int) (pos&CAT_MASK);
struct ISAMB_block *p;
if (!pos)
return 0;
p = xmalloc (sizeof(*p));
p->pos = pos;
- p->cat = pos & CAT_MASK;
+ p->cat = (int) (pos & CAT_MASK);
p->buf = xmalloc (b->file[cat].head.block_size);
+ p->cbuf = 0;
if (!get_block (b, pos, p->buf, 0))
{
p->size = (p->buf[1] + 256 * p->buf[2]) - ISAMB_DATA_OFFSET;
if (p->size < 0)
{
- yaz_log (LOG_FATAL, "Bad block size %d in pos=%d\n", p->size, pos);
+ yaz_log (LOG_FATAL, "Bad block size %d in pos=" ZINT_FORMAT "\n",
+ p->size, pos);
}
assert (p->size >= 0);
p->offset = 0;
p->dirty = 0;
p->deleted = 0;
- p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
- yaz_log (LOG_DEBUG, "isamb_open_block: Opened block %d ofs=%d",pos, p->offset);
+ p->decodeClientData = (*b->method->codec.start)();
+ yaz_log (LOG_DEBUG, "isamb_open_block: Opened block " ZINT_FORMAT " ofs=%d",pos, p->offset);
return p;
}
if (!b->file[cat].head.free_list)
{
- int block_no;
+ zint block_no;
block_no = b->file[cat].head.last_block++;
p->pos = block_no * CAT_MAX + cat;
}
abort ();
}
}
- yaz_log (b->log_freelist, "got block %d from freelist %d:%d", p->pos,
+ yaz_log (b->log_freelist, "got block " ZINT_FORMAT " from freelist %d:" ZINT_FORMAT, p->pos,
cat, p->pos/CAT_MAX);
memcpy (&b->file[cat].head.free_list, p->buf, sizeof(int));
}
p->dirty = 1;
p->deleted = 0;
p->offset = 0;
- p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
+ p->decodeClientData = (*b->method->codec.start)();
return p;
}
{
/* sanity check */
char *startp = p->bytes;
- char *src = startp;
+ const char *src = startp;
char *endp = p->bytes + p->size;
- int pos;
+ ISAMB_P pos;
decode_ptr (&src, &pos);
assert ((pos&CAT_MASK) == p->cat);
while (src != endp)
{
- int item_len;
+ zint item_len;
decode_ptr (&src, &item_len);
- assert (item_len > 0 && item_len < 30);
+ assert (item_len > 0 && item_len < 80);
src += item_len;
decode_ptr (&src, &pos);
assert ((pos&CAT_MASK) == p->cat);
return;
if (p->deleted)
{
- yaz_log (b->log_freelist, "release block %d from freelist %d:%d",
+ yaz_log (b->log_freelist, "release block " ZINT_FORMAT " from freelist %d:" ZINT_FORMAT,
p->pos, p->cat, p->pos/CAT_MAX);
memcpy (p->buf, &b->file[p->cat].head.free_list, sizeof(int));
b->file[p->cat].head.free_list = p->pos;
bf_write (b->file[p->cat].bf, p->pos/CAT_MAX, 0, 0, p->buf);
}
}
- (*b->method->code_stop)(ISAMC_DECODE, p->decodeClientData);
+ (*b->method->codec.stop)(p->decodeClientData);
xfree (p->buf);
xfree (p);
}
ISAMC_I *stream,
struct ISAMB_block **sp,
void *sub_item, int *sub_size,
- void *max_item);
+ const void *max_item);
int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
int *mode,
ISAMC_I *stream, struct ISAMB_block **sp,
- void *split_item, int *split_size, void *last_max_item)
+ void *split_item, int *split_size, const void *last_max_item)
{
char *startp = p->bytes;
- char *src = startp;
+ const char *src = startp;
char *endp = p->bytes + p->size;
- int pos;
+ ISAMB_P pos;
struct ISAMB_block *sub_p1 = 0, *sub_p2 = 0;
char sub_item[DST_ITEM_MAX];
int sub_size;
- int more;
+ int more = 0;
*sp = 0;
decode_ptr (&src, &pos);
while (src != endp)
{
- int item_len;
+ zint item_len;
int d;
- char *src0 = src;
+ const char *src0 = src;
decode_ptr (&src, &item_len);
d = (*b->method->compare_item)(src, lookahead_item);
if (d > 0)
char dst_buf[DST_BUF_SIZE];
char *dst = dst_buf;
- assert (sub_size < 30 && sub_size > 1);
+ assert (sub_size < 80 && sub_size > 1);
memcpy (dst, startp, src - startp);
}
else
{
+ zint split_size_tmp;
int p_new_size;
- char *half;
+ const char *half;
src = dst_buf;
endp = dst;
decode_ptr (&src, &pos);
while (src <= half)
{
- decode_ptr (&src, split_size);
+ decode_ptr (&src, &split_size_tmp);
+ *split_size = (int) split_size_tmp;
+
src += *split_size;
decode_ptr (&src, &pos);
}
p_new_size = src - dst_buf;
memcpy (p->bytes, dst_buf, p_new_size);
- decode_ptr (&src, split_size);
+ decode_ptr (&src, &split_size_tmp);
+ *split_size = (int) split_size_tmp;
memcpy (split_item, src, *split_size);
src += *split_size;
return more;
}
-
int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
int *lookahead_mode, ISAMC_I *stream,
struct ISAMB_block **sp2,
void *sub_item, int *sub_size,
- void *max_item)
+ const void *max_item)
{
struct ISAMB_block *p = *sp1;
- char *src = 0, *endp = 0;
+ char *endp = 0;
+ const char *src = 0;
char dst_buf[DST_BUF_SIZE], *dst = dst_buf;
int new_size;
- void *c1 = (*b->method->code_start)(ISAMC_DECODE);
- void *c2 = (*b->method->code_start)(ISAMC_ENCODE);
+ void *c1 = (*b->method->codec.start)();
+ void *c2 = (*b->method->codec.start)();
int more = 1;
int quater = b->file[b->no_cat-1].head.block_max / CAT_MAX;
char *cut = dst_buf + quater * 2;
src = p->bytes;
endp = p->bytes + p->size;
- (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
+ (*b->method->codec.decode)(c1, &file_item, &src);
while (1)
{
- char *dst_item = 0;
+ const char *dst_item = 0;
char *dst_0 = dst;
char *lookahead_next;
int d = -1;
}
else if (!half1 && dst > cut)
{
- char *dst_item_0 = dst_item;
+ const char *dst_item_0 = dst_item;
half1 = dst; /* candidate for splitting */
- (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
+ (*b->method->codec.encode)(c2, &dst, &dst_item);
cut_item_size = dst_item - dst_item_0;
+ assert(cut_item_size > 0);
memcpy (cut_item_buf, dst_item_0, cut_item_size);
half2 = dst;
}
else
- (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
+ (*b->method->codec.encode)(c2, &dst, &dst_item);
if (d > 0)
{
if (dst > maxp)
if (src == endp)
break;
file_item = file_item_buf;
- (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
+ (*b->method->codec.decode)(c1, &file_item, &src);
}
else
{
if (src == endp)
break;
file_item = file_item_buf;
- (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
+ (*b->method->codec.decode)(c1, &file_item, &src);
}
}
}
maxp = dst_buf + b->file[b->no_cat-1].head.block_max + quater;
while (lookahead_item)
{
- char *dst_item = lookahead_item;
+ char *dst_item;
+ const char *src = lookahead_item;
char *dst_0 = dst;
if (max_item &&
}
else if (!half1 && dst > cut)
{
- char *dst_item_0 = dst_item;
+ const char *src_0 = src;
half1 = dst; /* candidate for splitting */
- (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
+ (*b->method->codec.encode)(c2, &dst, &src);
- cut_item_size = dst_item - dst_item_0;
- memcpy (cut_item_buf, dst_item_0, cut_item_size);
+ cut_item_size = src - src_0;
+ assert(cut_item_size > 0);
+ memcpy (cut_item_buf, src_0, cut_item_size);
half2 = dst;
}
else
- (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
+ (*b->method->codec.encode)(c2, &dst, &src);
if (dst > maxp)
{
if (new_size > b->file[p->cat].head.block_max)
{
char *first_dst;
- char *cut_item = cut_item_buf;
+ const char *cut_item = cut_item_buf;
assert (half1);
assert (half2);
- /* first half */
+ assert(cut_item_size > 0);
+
+ /* first half */
p->size = half1 - dst_buf;
memcpy (p->bytes, dst_buf, half1 - dst_buf);
/* second half */
*sp2 = new_leaf (b, p->cat);
- (*b->method->code_reset)(c2);
+ (*b->method->codec.reset)(c2);
first_dst = (*sp2)->bytes;
- (*b->method->code_item)(ISAMC_ENCODE, c2, &first_dst, &cut_item);
+ (*b->method->codec.encode)(c2, &first_dst, &cut_item);
memcpy (first_dst, half2, dst - half2);
memcpy (p->bytes, dst_buf, dst - dst_buf);
p->size = new_size;
}
- (*b->method->code_stop)(ISAMC_DECODE, c1);
- (*b->method->code_stop)(ISAMC_ENCODE, c2);
+ (*b->method->codec.stop)(c1);
+ (*b->method->codec.stop)(c2);
*sp1 = p;
return more;
}
ISAMC_I *stream,
struct ISAMB_block **sp,
void *sub_item, int *sub_size,
- void *max_item)
+ const void *max_item)
{
if (!*p || (*p)->leaf)
return insert_leaf (b, p, new_item, mode, stream, sp, sub_item,
p1->deleted = 1;
if (!p1->leaf)
{
- int sub_p;
- int item_len;
- char *src = p1->bytes + p1->offset;
+ zint sub_p;
+ zint item_len;
+ const char *src = p1->bytes + p1->offset;
decode_ptr(&src, &sub_p);
isamb_unlink(b, sub_p);
return 0;
}
-int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I *stream)
+ISAMB_P isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I *stream)
{
char item_buf[DST_ITEM_MAX];
char *item_ptr;
char *dst = p2->bytes + p2->size;
encode_ptr (&dst, p->pos);
- assert (sub_size < 20);
+ assert (sub_size < 40);
encode_ptr (&dst, sub_size);
memcpy (dst, sub_item, sub_size);
dst += sub_size;
while (1)
{
struct ISAMB_block *p = open_block (isamb, pos);
- char *src = p->bytes + p->offset;
+ const char *src = p->bytes + p->offset;
pp->block[pp->level] = p;
pp->total_size += p->size;
isamb_pp_close_x (pp, 0, 0);
}
+/* simple recursive dumper .. */
+static void isamb_dump_r (ISAMB b, ISAMB_P pos, void (*pr)(const char *str),
+ int level)
+{
+ char buf[1024];
+ char prefix_str[1024];
+ if (pos)
+ {
+ struct ISAMB_block *p = open_block (b, pos);
+ sprintf(prefix_str, "%*s " ZINT_FORMAT " cat=%d size=%d max=%d", level*2, "",
+ pos, p->cat, p->size, b->file[p->cat].head.block_max);
+ (*pr)(prefix_str);
+ sprintf(prefix_str, "%*s " ZINT_FORMAT, level*2, "", pos);
+ if (p->leaf)
+ {
+ while (p->offset < p->size)
+ {
+ const char *src = p->bytes + p->offset;
+ char *dst = buf;
+ (*b->method->codec.decode)(p->decodeClientData, &dst, &src);
+ (*b->method->log_item)(LOG_DEBUG, buf, prefix_str);
+ p->offset = src - (char*) p->bytes;
+ }
+ assert(p->offset == p->size);
+ }
+ else
+ {
+ const char *src = p->bytes + p->offset;
+ ISAMB_P sub;
+ zint item_len;
+
+ decode_ptr (&src, &sub);
+ p->offset = src - (char*) p->bytes;
+
+ isamb_dump_r(b, sub, pr, level+1);
+
+ while (p->offset < p->size)
+ {
+ decode_ptr (&src, &item_len);
+ (*b->method->log_item)(LOG_DEBUG, src, prefix_str);
+ src += item_len;
+ decode_ptr (&src, &sub);
+
+ p->offset = src - (char*) p->bytes;
+
+ isamb_dump_r(b, sub, pr, level+1);
+ }
+ }
+ close_block(b,p);
+ }
+}
+
+void isamb_dump (ISAMB b, ISAMB_P pos, void (*pr)(const char *str))
+{
+ isamb_dump_r(b, pos, pr, 0);
+}
#if 0
/* Old isamb_pp_read that Adam wrote, kept as a reference in case we need to
debug the more complex pp_read that also forwards. May be deleted near end
of 2004, if it has not shown to be useful */
+
int isamb_pp_read (ISAMB_PP pp, void *buf)
{
char *dst = buf;
assert (p->offset < p->size);
assert (p->leaf);
src = p->bytes + p->offset;
- (*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
+ (*pp->isamb->method->codec.code_item)(ISAMC_DECODE, p->decodeClientData,
&dst, &src);
p->offset = src - (char*) p->bytes;
- key_logdump_txt(LOG_DEBUG,buf, "isamb_pp_read returning 1");
+ /* key_logdump_txt(LOG_DEBUG,buf, "isamb_pp_read returning 1"); */
return 1;
}
#else
int isamb_pp_read (ISAMB_PP pp, void *buf)
{
- return isamb_pp_forward(pp,buf,0);
+ return isamb_pp_forward(pp, buf, 0);
}
#endif
+#define NEW_FORWARD 1
+
+#if NEW_FORWARD == 1
+
+static int isamb_pp_on_right_node(ISAMB_PP pp, int level, const void *untilbuf)
+{ /* looks one node higher to see if we should be on this node at all */
+ /* useful in backing off quickly, and in avoiding tail descends */
+ /* call with pp->level to begin with */
+ struct ISAMB_block *p;
+ int cmp;
+ const char *src;
+ zint item_len;
+ assert(level>=0);
+ if ( level == 0) {
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_on_right returning true for root");
+#endif
+ return 1; /* we can never skip the root node */
+ }
+ level--;
+ p=pp->block[level];
+ assert(p->offset <= p->size);
+ if (p->offset < p->size )
+ {
+ assert(p->offset>0);
+ src=p->bytes + p->offset;
+ decode_ptr(&src, &item_len);
+#if ISAMB_DEBUG
+ (*pp->isamb->method->codec.log_item)(LOG_DEBUG,untilbuf,"on_leaf: until");
+ (*pp->isamb->method->codec.log_item)(LOG_DEBUG,src,"on_leaf: value");
+#endif
+ cmp=(*pp->isamb->method->compare_item)(untilbuf,src);
+ if (cmp<2) {
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_on_right returning true "
+ "cmp=%d lev=%d ofs=%d",cmp,level,p->offset);
+#endif
+ return 1;
+ }
+ else {
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_on_right returning false "
+ "cmp=%d lev=%d ofs=%d",cmp,level,p->offset);
+#endif
+ return 0;
+ }
+ }
+ else {
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_on_right at tail, looking higher "
+ "lev=%d",level);
+#endif
+ return isamb_pp_on_right_node(pp, level, untilbuf);
+ }
+} /* isamb_pp_on_right_node */
+
+static int isamb_pp_read_on_leaf(ISAMB_PP pp, void *buf)
+{ /* reads the next item on the current leaf, returns 0 if end of leaf*/
+ struct ISAMB_block *p = pp->block[pp->level];
+ char *dst;
+ const char *src;
+ assert(pp);
+ assert(buf);
+ if (p->offset == p->size) {
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_read_on_leaf returning 0 on node %d",p->pos);
+#endif
+ return 0; /* at end of leaf */
+ }
+ src=p->bytes + p->offset;
+ dst=buf;
+ (*pp->isamb->method->codec.decode)(p->decodeClientData,&dst, &src);
+ p->offset = src - (char*) p->bytes;
+ /*
+#if ISAMB_DEBUG
+ (*pp->isamb->method->codec.log_item)(LOG_DEBUG, buf, "read_on_leaf returning 1");
+#endif
+*/
+ return 1;
+} /* read_on_leaf */
+
+static int isamb_pp_forward_on_leaf(ISAMB_PP pp, void *buf, const void *untilbuf)
+{ /* forwards on the current leaf, returns 0 if not found */
+ int cmp;
+ int skips=0;
+ while (1){
+ if (!isamb_pp_read_on_leaf(pp,buf))
+ return 0;
+ /* FIXME - this is an extra function call, inline the read? */
+ cmp=(*pp->isamb->method->compare_item)(untilbuf,buf);
+ if (cmp <2){ /* found a good one */
+#if ISAMB_DEBUG
+ if (skips)
+ logf(LOG_DEBUG, "isam_pp_fwd_on_leaf skipped %d items",skips);
+#endif
+ pp->returned_numbers++;
+ return 1;
+ }
+ if (!skips)
+ if (!isamb_pp_on_right_node(pp, pp->level, untilbuf))
+ return 0; /* never mind the rest of this leaf */
+ pp->skipped_numbers++;
+ skips++;
+ }
+} /* forward_on_leaf */
+
+static int isamb_pp_climb_level(ISAMB_PP pp, ISAMB_P *pos)
+{ /* climbs higher in the tree, until finds a level with data left */
+ /* returns the node to (consider to) descend to in *pos) */
+ struct ISAMB_block *p = pp->block[pp->level];
+ const char *src;
+ zint item_len;
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_climb_level starting "
+ "at level %d node %d ofs=%d sz=%d",
+ pp->level, p->pos, p->offset, p->size);
+#endif
+ assert(pp->level >= 0);
+ assert(p->offset <= p->size);
+ if (pp->level==0)
+ {
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_climb_level returning 0 at root");
+#endif
+ return 0;
+ }
+ assert(pp->level>0);
+ close_block(pp->isamb, pp->block[pp->level]);
+ pp->block[pp->level]=0;
+ (pp->level)--;
+ p=pp->block[pp->level];
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_climb_level climbed to level %d node %d ofs=%d",
+ pp->level, p->pos, p->offset);
+#endif
+ assert(!p->leaf);
+ assert(p->offset <= p->size);
+ if (p->offset == p->size ) {
+ /* we came from the last pointer, climb on */
+ if (!isamb_pp_climb_level(pp,pos))
+ return 0;
+ p=pp->block[pp->level];
+ }
+ else
+ {
+ /* skip the child we just came from */
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isam_pp_climb_level: skipping lev=%d ofs=%d sz=%d",
+ pp->level, p->offset, p->size);
+#endif
+ assert (p->offset < p->size );
+ src=p->bytes + p->offset;
+ decode_ptr(&src, &item_len);
+ src += item_len;
+ decode_ptr(&src, pos);
+ p->offset=src - (char *)p->bytes;
+
+ }
+ return 1;
+} /* climb_level */
+
+
+static zint isamb_pp_forward_unode(ISAMB_PP pp, zint pos, const void *untilbuf)
+{ /* scans a upper node until it finds a child <= untilbuf */
+ /* pp points to the key value, as always. pos is the child read from */
+ /* the buffer */
+ /* if all values are too small, returns the last child in the node */
+ /* FIXME - this can be detected, and avoided by looking at the */
+ /* parent node, but that gets messy. Presumably the cost is */
+ /* pretty low anyway */
+ struct ISAMB_block *p = pp->block[pp->level];
+ const char *src=p->bytes + p->offset;
+ zint item_len;
+ int cmp;
+ zint nxtpos;
+#if ISAMB_DEBUG
+ int skips=0;
+ logf(LOG_DEBUG,"isamb_pp_forward_unode starting "
+ "at level %d node %d ofs=%di sz=%d",
+ pp->level, p->pos, p->offset, p->size);
+#endif
+ assert(!p->leaf);
+ assert(p->offset <= p->size);
+ if (p->offset == p->size) {
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_forward_unode returning at end "
+ "at level %d node %d ofs=%di sz=%d",
+ pp->level, p->pos, p->offset, p->size);
+#endif
+ return pos; /* already at the end of it */
+ }
+ while(p->offset < p->size) {
+ decode_ptr(&src,&item_len);
+ cmp=(*pp->isamb->method->compare_item)(untilbuf,src);
+ src+=item_len;
+ decode_ptr(&src,&nxtpos);
+ if (cmp<2)
+ {
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_forward_unode returning a hit "
+ "at level %d node %d ofs=%d sz=%d",
+ pp->level, p->pos, p->offset, p->size);
+#endif
+ return pos;
+ } /* found one */
+ pos=nxtpos;
+ p->offset=src-(char*)p->bytes;
+ (pp->skipped_nodes[pp->maxlevel - pp->level -1])++;
+#if ISAMB_DEBUG
+ skips++;
+#endif
+ }
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_forward_unode returning at tail "
+ "at level %d node %d ofs=%d sz=%d skips=%d",
+ pp->level, p->pos, p->offset, p->size, skips);
+#endif
+ return pos; /* that's the last one in the line */
+
+} /* forward_unode */
+
+static void isamb_pp_descend_to_leaf(ISAMB_PP pp, ISAMB_P pos, const void *untilbuf)
+{ /* climbs down the tree, from pos, to the leftmost leaf */
+ struct ISAMB_block *p = pp->block[pp->level];
+ const char *src;
+ assert(!p->leaf);
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_descend_to_leaf "
+ "starting at lev %d node %d ofs=%d lf=%d u=%p",
+ pp->level, p->pos, p->offset, p->leaf, untilbuf);
+#endif
+ if (untilbuf)
+ pos=isamb_pp_forward_unode(pp,pos,untilbuf);
+ ++(pp->level);
+ assert(pos);
+ p=open_block(pp->isamb, pos);
+ pp->block[pp->level]=p;
+ ++(pp->accessed_nodes[pp->maxlevel-pp->level]);
+ ++(pp->no_blocks);
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_descend_to_leaf "
+ "got lev %d node %d lf=%d",
+ pp->level, p->pos, p->leaf);
+#endif
+ if (p->leaf)
+ return;
+ assert (p->offset==0 );
+ src=p->bytes + p->offset;
+ decode_ptr(&src, &pos);
+ p->offset=src-(char*)p->bytes;
+ isamb_pp_descend_to_leaf(pp,pos,untilbuf);
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_descend_to_leaf "
+ "returning at lev %d node %d ofs=%d lf=%d",
+ pp->level, p->pos, p->offset, p->leaf);
+#endif
+} /* descend_to_leaf */
+
+static int isamb_pp_find_next_leaf(ISAMB_PP pp)
+{ /* finds the next leaf by climbing up and down */
+ ISAMB_P pos;
+ if (!isamb_pp_climb_level(pp,&pos))
+ return 0;
+ isamb_pp_descend_to_leaf(pp, pos,0);
+ return 1;
+}
+
+static int isamb_pp_climb_desc(ISAMB_PP pp, void *buf, const void *untilbuf)
+{ /* climbs up and descends to a leaf where values >= *untilbuf are found */
+ ISAMB_P pos;
+#if ISAMB_DEBUG
+ struct ISAMB_block *p = pp->block[pp->level];
+ logf(LOG_DEBUG,"isamb_pp_climb_desc starting "
+ "at level %d node %d ofs=%d sz=%d",
+ pp->level, p->pos, p->offset, p->size);
+#endif
+ if (!isamb_pp_climb_level(pp,&pos))
+ return 0;
+ /* see if it would pay to climb one higher */
+ if (!isamb_pp_on_right_node(pp, pp->level, untilbuf))
+ if (!isamb_pp_climb_level(pp,&pos))
+ return 0;
+ isamb_pp_descend_to_leaf(pp, pos,untilbuf);
+#if ISAMB_DEBUG
+ p = pp->block[pp->level];
+ logf(LOG_DEBUG,"isamb_pp_climb_desc done "
+ "at level %d node %d ofs=%d sz=%d",
+ pp->level, p->pos, p->offset, p->size);
+#endif
+ return 1;
+} /* climb_desc */
+
+int isamb_pp_forward (ISAMB_PP pp, void *buf, const void *untilbuf)
+{
+#if ISAMB_DEBUG
+ struct ISAMB_block *p = pp->block[pp->level];
+ assert(p->leaf);
+ logf(LOG_DEBUG,"isamb_pp_forward starting "
+ "at level %d node %d ofs=%d sz=%d u=%p",
+ pp->level, p->pos, p->offset, p->size,untilbuf);
+#endif
+ if (untilbuf) {
+ if (isamb_pp_forward_on_leaf( pp, buf, untilbuf)) {
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_forward (f) returning (A) "
+ "at level %d node %d ofs=%d sz=%d",
+ pp->level, p->pos, p->offset, p->size);
+#endif
+ return 1;
+ }
+ if (! isamb_pp_climb_desc( pp, buf, untilbuf)) {
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_forward (f) returning notfound (B) "
+ "at level %d node %d ofs=%d sz=%d",
+ pp->level, p->pos, p->offset, p->size);
+#endif
+ return 0; /* could not find a leaf */
+ }
+ do{
+ if (isamb_pp_forward_on_leaf( pp, buf, untilbuf)) {
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_forward (f) returning (C) "
+ "at level %d node %d ofs=%d sz=%d",
+ pp->level, p->pos, p->offset, p->size);
+#endif
+ return 1;
+ }
+ }while ( isamb_pp_find_next_leaf(pp));
+ return 0; /* could not find at all */
+ }
+ else { /* no untilbuf, a straight read */
+ /* FIXME - this should be moved
+ * directly into the pp_read */
+ /* keeping here now, to keep same
+ * interface as the old fwd */
+ if (isamb_pp_read_on_leaf( pp, buf)) {
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_forward (read) returning (D) "
+ "at level %d node %d ofs=%d sz=%d",
+ pp->level, p->pos, p->offset, p->size);
+#endif
+ return 1;
+ }
+ if (isamb_pp_find_next_leaf(pp)) {
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_forward (read) returning (E) "
+ "at level %d node %d ofs=%d sz=%d",
+ pp->level, p->pos, p->offset, p->size);
+#endif
+ return isamb_pp_read_on_leaf(pp, buf);
+ }
+ else
+ return 0;
+ }
+} /* isam_pp_forward (new version) */
+
+#elif NEW_FORWARD == 0
+
int isamb_pp_forward (ISAMB_PP pp, void *buf, const void *untilbuf)
{
/* pseudocode:
* if cmp
* return 1
*/
- /*
- * The upper nodes consist of a sequence of nodenumbers and keys
- * When opening a block, the first node number is read in, and
- * offset points to the first key, which is the upper limit of keys
- * in the node just read.
- */
+ /*
+ * The upper nodes consist of a sequence of nodenumbers and keys
+ * When opening a block, the first node number is read in, and
+ * offset points to the first key, which is the upper limit of keys
+ * in the node just read.
+ */
char *dst = buf;
- char *src;
+ const char *src;
struct ISAMB_block *p = pp->block[pp->level];
int cmp;
int item_len;
int pos;
int nxtpos;
+ int descending=0; /* used to prevent a border condition error */
if (!p)
return 0;
#if ISAMB_DEBUG
logf(LOG_DEBUG,"isamb_pp_forward starting [%p] p=%d",pp,p->pos);
- (*pp->isamb->method->log_item)(LOG_DEBUG, untilbuf, "until");
- (*pp->isamb->method->log_item)(LOG_DEBUG, buf, "buf");
+ (*pp->isamb->method->codec.log_item)(LOG_DEBUG, untilbuf, "until");
+ (*pp->isamb->method->codec.log_item)(LOG_DEBUG, buf, "buf");
#endif
while (1)
{
- while ( p->offset == p->size)
+ while ( (p->offset == p->size) && !descending )
{ /* end of this block - climb higher */
+ assert (p->offset <= p->size);
#if ISAMB_DEBUG
logf(LOG_DEBUG,"isamb_pp_forward climbing from l=%d",
pp->level);
p->pos, p->offset);
#endif
assert(!p->leaf);
+ assert(p->offset <= p->size);
/* skip the child we have handled */
if (p->offset != p->size)
{
src = p->bytes + p->offset;
decode_ptr(&src, &item_len);
#if ISAMB_DEBUG
- (*pp->isamb->method->log_item)(LOG_DEBUG, src,
+ (*pp->isamb->method->codec.log_item)(LOG_DEBUG, src,
" isamb_pp_forward "
"climb skipping old key");
#endif
src += item_len;
decode_ptr(&src,&pos);
p->offset = src - (char*) p->bytes;
- break; /* even if this puts us at the end of the block, we need to */
- /* descend to the last pos. UGLY coding, clean up some */
- /* day */
+ break; /* even if this puts us at the end of the block, we
+ need to descend to the last pos. UGLY coding,
+ clean up some day */
}
}
if (!p->leaf)
{
src = p->bytes + p->offset;
if (p->offset == p->size)
- cmp=-2 ; /* descend to the last node, as we have no value to cmp */
+ cmp=-2 ; /* descend to the last node, as we have
+ no value to cmp */
else
{
decode_ptr(&src, &item_len);
#if ISAMB_DEBUG
- logf(LOG_DEBUG,"isamb_pp_forward (B) on a high node. ofs=%d sz=%d nxtpos=%d ",
+ logf(LOG_DEBUG,"isamb_pp_forward (B) on a high node. "
+ "ofs=%d sz=%d nxtpos=%d ",
p->offset,p->size,pos);
- (*pp->isamb->method->log_item)(LOG_DEBUG, src, "");
+ (*pp->isamb->method->codec.log_item)(LOG_DEBUG, src, "");
#endif
if (untilbuf)
cmp=(*pp->isamb->method->compare_item)(untilbuf,src);
logf(LOG_DEBUG,"isambb_pp_forward descending l=%d p=%d ",
pp->level, pos);
#endif
+ descending=1; /* prevent climbing for a while */
++(pp->level);
p = open_block(pp->isamb,pos);
pp->block[pp->level] = p ;
(pp->skipped_nodes[pp->maxlevel - pp->level -1])++;
#if ISAMB_DEBUG
logf(LOG_DEBUG,
- "isamb_pp_forward: skipping block on level %d, noting on %d (%d)",
+ "isamb_pp_forward: skipping block on level %d, noting "
+ "on %d (%d)",
pp->level, pp->maxlevel - pp->level-1 ,
pp->skipped_nodes[pp->maxlevel - pp->level-1 ]);
#endif
} /* not on a leaf */
else
{ /* on a leaf */
- src = p->bytes + p->offset;
- dst=buf;
- (*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
- &dst, &src);
- p->offset = src - (char*) p->bytes;
- if (untilbuf)
- cmp=(*pp->isamb->method->compare_item)(untilbuf,buf);
- else
- cmp=-2;
+ if (p->offset == p->size) {
+ descending = 0;
+ }
+ else
+ {
+ assert (p->offset < p->size);
+ src = p->bytes + p->offset;
+ dst=buf;
+ (*pp->isamb->method->codec.decode)(p->decodeClientData,
+ &dst, &src);
+ p->offset = src - (char*) p->bytes;
+ if (untilbuf)
+ cmp=(*pp->isamb->method->compare_item)(untilbuf,buf);
+ else
+ cmp=-2;
#if ISAMB_DEBUG
- logf(LOG_DEBUG,"isamb_pp_forward on a leaf. cmp=%d",
- cmp);
- (*pp->isamb->method->log_item)(LOG_DEBUG, buf, "");
+ logf(LOG_DEBUG,"isamb_pp_forward on a leaf. cmp=%d",
+ cmp);
+ (*pp->isamb->method->codec.log_item)(LOG_DEBUG, buf, "");
#endif
- if (cmp <2)
- {
-#if ISAMB_DEBUG
- if (untilbuf)
- {
- (*pp->isamb->method->log_item)(LOG_DEBUG, buf,
- "isamb_pp_forward returning 1");
- }
- else
+ if (cmp <2)
{
- (*pp->isamb->method->log_item)(LOG_DEBUG, buf,
- "isamb_pp_read returning 1 (fwd)");
- }
+#if ISAMB_DEBUG
+ if (untilbuf)
+ {
+ (*pp->isamb->method->codec.log_item)(
+ LOG_DEBUG, buf, "isamb_pp_forward returning 1");
+ }
+ else
+ {
+ (*pp->isamb->method->codec.log_item)(
+ LOG_DEBUG, buf, "isamb_pp_read returning 1 (fwd)");
+ }
#endif
- pp->returned_numbers++;
- return 1;
- }
- else
- pp->skipped_numbers++;
+ pp->returned_numbers++;
+ return 1;
+ }
+ else
+ pp->skipped_numbers++;
+ }
} /* leaf */
} /* main loop */
}
+#elif NEW_FORWARD == 2
+
+int isamb_pp_forward (ISAMB_PP pp, void *buf, const void *untilb)
+{
+ char *dst = buf;
+ const char *src;
+ struct ISAMB_block *p = pp->block[pp->level];
+ if (!p)
+ return 0;
+
+again:
+ while (p->offset == p->size)
+ {
+ int pos, item_len;
+ while (p->offset == p->size)
+ {
+ if (pp->level == 0)
+ return 0;
+ close_block (pp->isamb, pp->block[pp->level]);
+ pp->block[pp->level] = 0;
+ (pp->level)--;
+ p = pp->block[pp->level];
+ assert (!p->leaf);
+ }
+
+ assert(!p->leaf);
+ src = p->bytes + p->offset;
+
+ decode_ptr (&src, &item_len);
+ src += item_len;
+ decode_ptr (&src, &pos);
+
+ p->offset = src - (char*) p->bytes;
+
+ src = p->bytes + p->offset;
+
+ while(1)
+ {
+ if (!untilb || p->offset == p->size)
+ break;
+ assert(p->offset < p->size);
+ decode_ptr (&src, &item_len);
+ if ((*pp->isamb->method->compare_item)(untilb, src) <= 1)
+ break;
+ src += item_len;
+ decode_ptr (&src, &pos);
+ p->offset = src - (char*) p->bytes;
+ }
+
+ pp->level++;
+
+ while (1)
+ {
+ pp->block[pp->level] = p = open_block (pp->isamb, pos);
+
+ pp->total_size += p->size;
+ pp->no_blocks++;
+
+ if (p->leaf)
+ {
+ break;
+ }
+
+ src = p->bytes + p->offset;
+ while(1)
+ {
+ decode_ptr (&src, &pos);
+ p->offset = src - (char*) p->bytes;
+
+ if (!untilb || p->offset == p->size)
+ break;
+ assert(p->offset < p->size);
+ decode_ptr (&src, &item_len);
+ if ((*pp->isamb->method->compare_item)(untilb, src) <= 1)
+ break;
+ src += item_len;
+ }
+ pp->level++;
+ }
+ }
+ assert (p->offset < p->size);
+ assert (p->leaf);
+ while(1)
+ {
+ char *dst0 = dst;
+ src = p->bytes + p->offset;
+ (*pp->isamb->method->codec.decode)(p->decodeClientData, &dst, &src);
+ p->offset = src - (char*) p->bytes;
+ if (!untilb || (*pp->isamb->method->compare_item)(untilb, dst0) <= 1)
+ break;
+ dst = dst0;
+ if (p->offset == p->size) goto again;
+ }
+ /* key_logdump_txt(LOG_DEBUG,buf, "isamb_pp_read returning 1"); */
+ return 1;
+}
+
+#endif
int isamb_pp_num (ISAMB_PP pp)
{
return 1;
}
+
+static void isamb_pp_leaf_pos( ISAMB_PP pp,
+ double *current, double *total,
+ void *dummybuf )
+{
+ struct ISAMB_block *p = pp->block[pp->level];
+ const char *src=p->bytes;
+ char *end=p->bytes+p->size;
+ char *cur=p->bytes+p->offset;
+ char *dst;
+ void *decodeClientData;
+ assert(p->offset <= p->size);
+ assert(cur <= end);
+ assert(p->leaf);
+ *current=0;
+ *total=0;
+
+ decodeClientData = (pp->isamb->method->codec.start)();
+
+ while(src < end) {
+ dst=dummybuf;
+ (*pp->isamb->method->codec.decode)(decodeClientData,&dst, &src);
+ assert(dst<(char*) dummybuf+100); /*FIXME */
+ (*total)++;
+ if (src<=cur)
+ (*current)++;
+ }
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG, "isamb_pp_leaf_pos: cur= %0.1f tot=%0.1f "
+ " ofs=%d sz=%d lev=%d",
+ *current, *total, p->offset, p->size, pp->level);
+#endif
+ assert(src==end);
+ (pp->isamb->method->codec.stop)(decodeClientData);
+}
+
+static void isamb_pp_upper_pos( ISAMB_PP pp, double *current, double *total,
+ zint size, int level )
+{ /* estimates total/current occurrences from here up, excl leaf */
+ struct ISAMB_block *p = pp->block[level];
+ const char *src=p->bytes;
+ char *end=p->bytes+p->size;
+ char *cur=p->bytes+p->offset;
+ zint item_size;
+ ISAMB_P child;
+
+ assert(level>=0);
+ assert(!p->leaf);
+
+#if ISAMB_DEBUG
+ logf(LOG_DEBUG,"isamb_pp_upper_pos at beginning l=%d "
+ "cur="ZINT_FORMAT" tot="ZINT_FORMAT
+ " ofs=%d sz=%d pos=" ZINT_FORMAT,
+ level, *current, *total, p->offset, p->size, p->pos);
+#endif
+ assert (p->offset <= p->size);
+ decode_ptr (&src, &child ); /* first child */
+ while(src < end) {
+ if (src!=cur) {
+ *total += size;
+ if (src < cur)
+ *current +=size;
+ }
+ decode_ptr (&src, &item_size );
+ assert(src+item_size<=end);
+ src += item_size;
+ decode_ptr (&src, &child );
+ }
+ if (level>0)
+ isamb_pp_upper_pos(pp, current, total, (zint) *total, level-1);
+} /* upper_pos */
+
+void isamb_pp_pos( ISAMB_PP pp, double *current, double *total )
+{ /* return an estimate of the current position and of the total number of */
+ /* occureences in the isam tree, based on the current leaf */
+ struct ISAMB_block *p = pp->block[pp->level];
+ char dummy[100]; /* 100 bytes/entry must be enough */
+ assert(total);
+ assert(current);
+ assert(p->leaf);
+ isamb_pp_leaf_pos(pp,current, total, dummy);
+ if (pp->level>0)
+ isamb_pp_upper_pos(pp, current, total, (zint) *total, pp->level-1);
+}
projects / idzebra-moved-to-github.git / blobdiff