-/* $Id: isamb.c,v 1.43 2004-06-03 15:05:05 heikki Exp $
+/* $Id: isamb.c,v 1.47.2.4 2005-01-16 23:13:30 adam Exp $
Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
Index Data Aps
*/
#include <string.h>
+#include <stdlib.h>
#include <yaz/xmalloc.h>
#include <yaz/log.h>
#include <isamb.h>
#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;
struct ISAMB_block **block;
};
+
#if ISAMB_PTR_CODEC
static void encode_ptr (char **dst, unsigned 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;
while (1)
{
char *dst_item = 0;
- char *dst_0 = dst;
char *lookahead_next;
int d = -1;
if (d > 0)
{
if (dst > maxp)
- {
- dst = dst_0;
lookahead_item = 0;
- }
else
{
lookahead_next = lookahead_item;
pp->no_blocks = 0;
pp->skipped_numbers=0;
pp->returned_numbers=0;
- for (i=0;i<ISAMB_MAX_LEVEL;i++)
- pp->skipped_nodes[i] = pp->accessed_nodes[i]=0;
- while (1)
+ for (i = 0; i<ISAMB_MAX_LEVEL; i++)
+ pp->skipped_nodes[i] = pp->accessed_nodes[i] = 0;
+ pp->block[0] = 0;
+ if (pos)
{
- struct ISAMB_block *p = open_block (isamb, pos);
- char *src = p->bytes + p->offset;
- pp->block[pp->level] = p;
-
- pp->total_size += p->size;
- pp->no_blocks++;
- if (p->leaf)
- break;
-
-
- decode_ptr (&src, &pos);
- p->offset = src - p->bytes;
- pp->level++;
- pp->accessed_nodes[pp->level]++;
+ while (1)
+ {
+ struct ISAMB_block *p = open_block (isamb, pos);
+ char *src = p->bytes + p->offset;
+ pp->block[pp->level] = p;
+
+ pp->total_size += p->size;
+ pp->no_blocks++;
+ if (p->leaf)
+ break;
+
+ decode_ptr (&src, &pos);
+ p->offset = src - p->bytes;
+ pp->level++;
+ pp->accessed_nodes[pp->level]++;
+ }
}
pp->block[pp->level+1] = 0;
- pp->maxlevel=pp->level;
+ pp->maxlevel = pp->level;
if (level)
*level = pp->level;
return pp;
void isamb_dump (ISAMB b, ISAMB_P pos, void (*pr)(const char *str))
{
- return isamb_dump_r(b, pos, pr, 0);
+ isamb_dump_r(b, pos, pr, 0);
}
#if 0
char *src;
assert(pp);
assert(buf);
- if (p->offset == p->size) {
+ if (!p || p->offset == p->size) {
#if ISAMB_DEBUG
logf(LOG_DEBUG,"isamb_pp_read_on_leaf returning 0 on node %d",p->pos);
#endif
pp->level, p->pos, p->offset, p->size);
#endif
assert(pp->level >= 0);
+ if (!p)
+ return 0;
assert(p->offset <= p->size);
if (pp->level==0)
{
assert(pp->level>0);
close_block(pp->isamb, pp->block[pp->level]);
pp->block[pp->level]=0;
- assert(pp->level>0);
(pp->level)--;
p=pp->block[pp->level];
#if ISAMB_DEBUG
/* parent node, but that gets messy. Presumably the cost is */
/* pretty low anyway */
struct ISAMB_block *p = pp->block[pp->level];
- char *src=p->bytes + p->offset;
+ char *src = p->bytes + p->offset;
int item_len;
int cmp;
int 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);
+ int skips = 0;
+ if (p)
+ 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);
+ 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) {
+ while(p->offset < p->size)
+ {
decode_ptr(&src,&item_len);
cmp=(*pp->isamb->method->compare_item)(untilbuf,src);
src+=item_len;
{
#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);
+ "at level %d node %d ofs=%d sz=%d",
+ pp->level, p->pos, p->offset, p->size);
#endif
return pos;
} /* found one */
pos=isamb_pp_forward_unode(pp,pos,untilbuf);
++(pp->level);
assert(pos);
- p=open_block(pp->isamb, pos);
- pp->block[pp->level]=p;
+ p = open_block(pp->isamb, pos);
+ pp->block[pp->level] = p;
++(pp->accessed_nodes[pp->maxlevel-pp->level]);
++(pp->no_blocks);
#if ISAMB_DEBUG
#endif
if (p->leaf)
return;
- assert (p->offset==0 );
- src=p->bytes + p->offset;
+ assert (p->offset==0);
+ src = p->bytes + p->offset;
decode_ptr(&src, &pos);
- p->offset=src-(char*)p->bytes;
- /*
- if (untilbuf)
- pos=isamb_pp_forward_unode(pp,pos,untilbuf);
- */
+ 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 "
{
#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);
+ if (p)
+ {
+ 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",
+ if (p)
+ 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);
+ if (p)
+ 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{
+ 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);
+ if(p)
+ 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));
+ } while (isamb_pp_find_next_leaf(pp));
return 0; /* could not find at all */
}
else { /* no untilbuf, a straight read */
* 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);
+ if (p)
+ 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",
+ if (p)
+ 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);
p->pos, p->offset);
#endif
assert(!p->leaf);
- assert(p->offset <= p->size);
+ assert(p->offset <= p->size);
/* skip the child we have handled */
if (p->offset != p->size)
{
{
return 1;
}
+
+static void isamb_pp_leaf_pos( ISAMB_PP pp,
+ int *current, int *total, void *dummybuf )
+{
+ struct ISAMB_block *p = pp->block[pp->level];
+ char *src=p->bytes;
+ char *end=p->bytes+p->size;
+ char *cur=p->bytes+p->offset;
+ char *dst;
+ assert(p->offset <= p->size);
+ assert(cur <= end);
+ assert(p->leaf);
+ *current=0;
+ *total=0;
+
+ while(src < end) {
+ dst=dummybuf;
+ (*pp->isamb->method->code_item)
+ (ISAMC_DECODE, p->decodeClientData,&dst, &src);
+ assert(dst<(char*) dummybuf+100); /*FIXME */
+ (*total)++;
+ if (src<=cur)
+ (*current)++;
+ }
+ logf(LOG_DEBUG, "isamb_pp_leaf_pos: cur=%d tot=%d ofs=%d sz=%d lev=%d",
+ *current, *total, p->offset, p->size, pp->level);
+ assert(src==end);
+}
+
+static void isamb_pp_upper_pos( ISAMB_PP pp, int *current, int *total,
+ int size, int level )
+{ /* estimates total/current occurrences from here up, excl leaf */
+ struct ISAMB_block *p = pp->block[level];
+ char *src=p->bytes;
+ char *end=p->bytes+p->size;
+ char *cur=p->bytes+p->offset;
+ int item_size;
+ int child;
+ assert(level>=0);
+ assert(!p->leaf);
+ logf(LOG_DEBUG,"isamb_pp_upper_pos at beginning l=%d "
+ "cur=%d tot=%d ofs=%d sz=%d pos=%d",
+ level, *current, *total, p->offset, p->size, p->pos);
+ 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, *total, level-1);
+} /* upper_pos */
+
+void isamb_pp_pos( ISAMB_PP pp, int *current, int *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, *total, pp->level-1);
+ /*
+ logf(LOG_DEBUG,"isamb_pp_pos: C=%d T=%d =%6.2f%%",
+ *current, *total, 100.0*(*current)/(*total));
+ */
+}