-/* $Id: isamb.c,v 1.67 2005-01-15 18:43:05 adam Exp $
+/* $Id: isamb.c,v 1.77 2005-04-15 10:47:49 adam Exp $
Copyright (C) 1995-2005
- Index Data Aps
+ Index Data ApS
This file is part of the Zebra server.
02111-1307, USA.
*/
+#include <stdlib.h>
#include <string.h>
#include <yaz/log.h>
#include <yaz/xmalloc.h>
int block_offset;
};
-/* if 1, interior nodes items are encoded; 0 if not encoded */
+/* if 1, upper nodes items are encoded; 0 if not encoded */
#define INT_ENCODE 1
/* maximum size of encoded buffer */
#define ISAMB_PTR_CODEC 1
struct ISAMB_cache_entry {
- ISAMB_P pos;
+ ISAM_P pos;
unsigned char *buf;
int dirty;
int hits;
};
struct ISAMB_block {
- ISAMB_P pos;
+ ISAM_P pos;
int cat;
int size;
int leaf;
struct ISAMB_PP_s {
ISAMB isamb;
- ISAMB_P pos;
+ ISAM_P pos;
int level;
int maxlevel; /* total depth */
zint total_size;
while (pos > 127)
{
- *bp++ = 128 | (pos & 127);
+ *bp++ = (unsigned char) (128 | (pos & 127));
pos = pos >> 7;
}
- *bp++ = pos;
+ *bp++ = (unsigned char) pos;
*dst = (char *) bp;
}
#else
#endif
ISAMB isamb_open(BFiles bfs, const char *name, int writeflag, ISAMC_M *method,
- int cache)
+ int cache)
{
ISAMB isamb = xmalloc(sizeof(*isamb));
int i, b_size = ISAMB_MIN_SIZE;
isamb->bfs = bfs;
isamb->method = (ISAMC_M *) xmalloc(sizeof(*method));
- memcpy (isamb->method, method, sizeof(*method));
+ memcpy(isamb->method, method, sizeof(*method));
isamb->no_cat = CAT_NO;
isamb->log_io = 0;
isamb->log_freelist = 0;
isamb->cache = cache;
isamb->skipped_numbers = 0;
isamb->returned_numbers = 0;
- for (i = 0;i<ISAMB_MAX_LEVEL;i++)
- isamb->skipped_nodes[i]= isamb->accessed_nodes[i]=0;
+ for (i = 0; i<ISAMB_MAX_LEVEL; i++)
+ isamb->skipped_nodes[i] = isamb->accessed_nodes[i] = 0;
assert(cache == 0);
isamb->file = xmalloc(sizeof(*isamb->file) * isamb->no_cat);
decode_ptr(&src, &isamb->file[i].head.first_block);
decode_ptr(&src, &isamb->file[i].head.last_block);
decode_ptr(&src, &zint_tmp);
- isamb->file[i].head.block_size = zint_tmp;
+ isamb->file[i].head.block_size = (int) zint_tmp;
decode_ptr(&src, &zint_tmp);
- isamb->file[i].head.block_max = zint_tmp;
+ isamb->file[i].head.block_max = (int) zint_tmp;
decode_ptr(&src, &isamb->file[i].head.free_list);
}
assert (isamb->file[i].head.block_size >= isamb->file[i].head.block_offset);
}
}
-static int get_block (ISAMB b, ISAMC_P pos, char *userbuf, int wr)
+static int cache_block (ISAMB b, ISAM_P pos, char *userbuf, int wr)
{
int cat = (int) (pos&CAT_MASK);
int off = (int) (((pos/CAT_MAX) &
*ce_last = 0; /* remove the last entry from list */
if (ce_this->dirty)
{
- yaz_log(b->log_io, "bf_write: get_block");
+ yaz_log(b->log_io, "bf_write: cache_block");
bf_write(b->file[cat].bf, ce_this->pos, 0, 0, ce_this->buf);
}
xfree(ce_this->buf);
b->file[cat].cache_entries = ce_this;
ce_this->buf = xmalloc(ISAMB_CACHE_ENTRY_SIZE);
ce_this->pos = norm;
- yaz_log(b->log_io, "bf_read: get_block");
+ yaz_log(b->log_io, "bf_read: cache_block");
if (!bf_read(b->file[cat].bf, norm, 0, 0, ce_this->buf))
memset (ce_this->buf, 0, ISAMB_CACHE_ENTRY_SIZE);
if (wr)
void isamb_close (ISAMB isamb)
{
int i;
- for (i = 0;isamb->accessed_nodes[i];i++)
+ for (i = 0; isamb->accessed_nodes[i]; i++)
yaz_log(YLOG_DEBUG, "isamb_close level leaf-%d: "ZINT_FORMAT" read, "
ZINT_FORMAT" skipped",
i, isamb->accessed_nodes[i], isamb->skipped_nodes[i]);
* Reserve 5 bytes for large block sizes. 1 for small ones .. Number
of items. We can thus have at most 2^40 nodes.
*/
-static struct ISAMB_block *open_block(ISAMB b, ISAMC_P pos)
+static struct ISAMB_block *open_block(ISAMB b, ISAM_P pos)
{
int cat = (int) (pos&CAT_MASK);
const char *src;
p->buf = xmalloc(b->file[cat].head.block_size);
p->cbuf = 0;
- if (!get_block (b, pos, p->buf, 0))
+ if (!cache_block (b, pos, p->buf, 0))
{
yaz_log(b->log_io, "bf_read: open_block");
if (!bf_read(b->file[cat].bf, pos/CAT_MAX, 0, 0, p->buf))
{
p->pos = b->file[cat].head.free_list;
assert((p->pos & CAT_MASK) == cat);
- if (!get_block (b, p->pos, p->buf, 0))
+ if (!cache_block (b, p->pos, p->buf, 0))
{
yaz_log(b->log_io, "bf_read: new_block");
if (!bf_read(b->file[cat].bf, p->pos/CAT_MAX, 0, 0, p->buf))
char *startp = p->bytes;
const char *src = startp;
char *endp = p->bytes + p->size;
- ISAMB_P pos;
+ ISAM_P pos;
void *c1 = (*b->method->codec.start)();
decode_ptr(&src, &pos);
p->pos, p->cat, p->pos/CAT_MAX);
memcpy (p->buf, &b->file[p->cat].head.free_list, sizeof(zint));
b->file[p->cat].head.free_list = p->pos;
- if (!get_block (b, p->pos, p->buf, 1))
+ if (!cache_block (b, p->pos, p->buf, 1))
{
yaz_log(b->log_io, "bf_write: close_block (deleted)");
bf_write(b->file[p->cat].bf, p->pos/CAT_MAX, 0, 0, p->buf);
p->buf[2] = size >> 8;
encode_ptr(&dst, p->no_items);
check_block(b, p);
- if (!get_block (b, p->pos, p->buf, 1))
+ if (!cache_block (b, p->pos, p->buf, 1))
{
yaz_log(b->log_io, "bf_write: close_block");
bf_write(b->file[p->cat].bf, p->pos/CAT_MAX, 0, 0, p->buf);
char *startp = p->bytes;
const char *src = startp;
char *endp = p->bytes + p->size;
- ISAMB_P pos;
+ ISAM_P pos;
struct ISAMB_block *sub_p1 = 0, *sub_p2 = 0;
char sub_item[DST_ITEM_MAX];
int sub_size;
/* there was a split - must insert pointer in this one */
char dst_buf[DST_BUF_SIZE];
char *dst = dst_buf;
-
+#if INT_ENCODE
+ const char *sub_item_ptr = sub_item;
+#endif
assert (sub_size < 80 && sub_size > 1);
memcpy (dst, startp, src - startp);
dst += src - startp;
#if INT_ENCODE
- const char *sub_item_ptr = sub_item;
(*b->method->codec.reset)(c1);
(*b->method->codec.encode)(c1, &dst, &sub_item_ptr);
#else
/* must split _this_ block as well .. */
struct ISAMB_block *sub_p3;
#if INT_ENCODE
+ char file_item_buf[DST_ITEM_MAX];
+ char *file_item = file_item_buf;
#else
zint split_size_tmp;
#endif
while (src <= half)
{
#if INT_ENCODE
- char file_item_buf[DST_ITEM_MAX];
- char *file_item = file_item_buf;
+ file_item = file_item_buf;
(*b->method->codec.reset)(c1);
(*b->method->codec.decode)(c1, &file_item, &src);
#else
memcpy (p->bytes, dst_buf, p_new_size);
#if INT_ENCODE
- char file_item_buf[DST_ITEM_MAX];
- char *file_item = file_item_buf;
+ file_item = file_item_buf;
(*b->method->codec.reset)(c1);
(*b->method->codec.decode)(c1, &file_item, &src);
*split_size = file_item - file_item_buf;
int cut_item_size = 0;
int no_items = 0; /* number of items (total) */
int no_items_1 = 0; /* number of items (first half) */
+ int inserted_dst_bytes = 0;
if (p && p->size)
{
{
const char *dst_item = 0; /* resulting item to be inserted */
char *lookahead_next;
+ char *dst_0 = dst;
int d = -1;
if (lookahead_item)
else
dst_item = file_item_buf;
-
if (!*lookahead_mode && d == 0)
{
/* it's a deletion and they match so there is nothing to be
{
/* we must move the lookahead pointer */
- if (dst > maxp)
+ inserted_dst_bytes += (dst - dst_0);
+ if (inserted_dst_bytes >= quater)
/* no more room. Mark lookahead as "gone".. */
lookahead_item = 0;
else
}
}
}
- maxp = dst_buf + b->file[b->no_cat-1].head.block_max + quater;
+
/* this loop runs when we are "appending" to a leaf page. That is
either it's empty (new) or all file items have been read in
previous loop */
+
+ maxp = dst_buf + b->file[b->no_cat-1].head.block_max + quater;
while (lookahead_item)
{
char *dst_item;
const char *src = lookahead_item;
char *dst_0 = dst;
- /* compare lookahead with max item */
+ /* if we have a lookahead item, we stop if we exceed the value of it */
if (max_item &&
(*b->method->compare_item)(max_item, lookahead_item) <= 0)
{
/* first half */
p->size = half1 - dst_buf;
+ assert(p->size <= b->file[p->cat].head.block_max);
memcpy (p->bytes, dst_buf, half1 - dst_buf);
p->no_items = no_items_1;
memcpy (first_dst, half2, dst - half2);
(*sp2)->size = (first_dst - (*sp2)->bytes) + (dst - half2);
+ assert((*sp2)->size <= b->file[p->cat].head.block_max);
(*sp2)->no_items = no_items - no_items_1;
(*sp2)->dirty = 1;
p->dirty = 1;
sub_size, max_item);
}
-int isamb_unlink (ISAMB b, ISAMC_P pos)
+int isamb_unlink (ISAMB b, ISAM_P pos)
{
struct ISAMB_block *p1;
{
zint sub_p;
const char *src = p1->bytes + p1->offset;
-
+#if INT_ENCODE
+ void *c1 = (*b->method->codec.start)();
+#endif
decode_ptr(&src, &sub_p);
isamb_unlink(b, sub_p);
- void *c1 = (*b->method->codec.start)();
while (src != p1->bytes + p1->size)
{
decode_ptr(&src, &sub_p);
isamb_unlink(b, sub_p);
}
+#if INT_ENCODE
(*b->method->codec.stop)(c1);
+#endif
}
close_block(b, p1);
return 0;
}
-ISAMB_P isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I *stream)
+void isamb_merge(ISAMB b, ISAM_P *pos, ISAMC_I *stream)
{
char item_buf[DST_ITEM_MAX];
char *item_ptr;
more =
(*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
}
- return 1;
+ *pos = 1;
+ return;
}
item_ptr = item_buf;
more = (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
char sub_item[DST_ITEM_MAX];
int sub_size;
- if (pos)
- p = open_block(b, pos);
+ if (*pos)
+ p = open_block(b, *pos);
more = insert_sub (b, &p, item_buf, &i_mode, stream, &sp,
sub_item, &sub_size, 0);
if (sp)
{ /* increase level of tree by one */
struct ISAMB_block *p2 = new_int (b, p->cat);
char *dst = p2->bytes + p2->size;
+#if INT_ENCODE
void *c1 = (*b->method->codec.start)();
-
+ const char *sub_item_ptr = sub_item;
+#endif
+
encode_ptr(&dst, p->pos);
assert (sub_size < 80 && sub_size > 1);
#if INT_ENCODE
- const char *sub_item_ptr = sub_item;
(*b->method->codec.reset)(c1);
(*b->method->codec.encode)(c1, &dst, &sub_item_ptr);
#else
p2->size = dst - p2->bytes;
p2->no_items = p->no_items + sp->no_items;
- pos = p2->pos; /* return new super page */
+ *pos = p2->pos; /* return new super page */
close_block(b, sp);
close_block(b, p2);
+#if INT_ENCODE
(*b->method->codec.stop)(c1);
+#endif
}
else
{
- pos = p->pos; /* return current one (again) */
+ *pos = p->pos; /* return current one (again) */
}
if (p->no_items == 0)
must_delete = 1;
}
if (must_delete)
{
- isamb_unlink(b, pos);
- return 0;
+ isamb_unlink(b, *pos);
+ *pos = 0;
}
- return pos;
}
-ISAMB_PP isamb_pp_open_x(ISAMB isamb, ISAMB_P pos, int *level, int scope)
+ISAMB_PP isamb_pp_open_x(ISAMB isamb, ISAM_P pos, int *level, int scope)
{
ISAMB_PP pp = xmalloc(sizeof(*pp));
int i;
pp->skipped_numbers = 0;
pp->returned_numbers = 0;
pp->scope = scope;
- for (i = 0;i<ISAMB_MAX_LEVEL;i++)
- pp->skipped_nodes[i] = pp->accessed_nodes[i]=0;
+ for (i = 0; i<ISAMB_MAX_LEVEL; i++)
+ pp->skipped_nodes[i] = pp->accessed_nodes[i] = 0;
while (1)
{
struct ISAMB_block *p = open_block(isamb, pos);
return pp;
}
-ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos, int scope)
+ISAMB_PP isamb_pp_open (ISAMB isamb, ISAM_P pos, int scope)
{
return isamb_pp_open_x(isamb, pos, 0, scope);
}
-void isamb_pp_close_x(ISAMB_PP pp, int *size, int *blocks)
+void isamb_pp_close_x(ISAMB_PP pp, zint *size, zint *blocks)
{
int i;
if (!pp)
yaz_log(YLOG_DEBUG, "isamb_pp_close lev=%d returned "ZINT_FORMAT" values, "
"skipped "ZINT_FORMAT,
pp->maxlevel, pp->skipped_numbers, pp->returned_numbers);
- for (i = pp->maxlevel;i>=0;i--)
+ for (i = pp->maxlevel; i>=0; i--)
if (pp->skipped_nodes[i] || pp->accessed_nodes[i])
yaz_log(YLOG_DEBUG, "isamb_pp_close level leaf-%d: "
ZINT_FORMAT" read, "ZINT_FORMAT" skipped", i,
pp->accessed_nodes[i], pp->skipped_nodes[i]);
pp->isamb->skipped_numbers += pp->skipped_numbers;
pp->isamb->returned_numbers += pp->returned_numbers;
- for (i = pp->maxlevel;i>=0;i--)
+ for (i = pp->maxlevel; i>=0; i--)
{
pp->isamb->accessed_nodes[i] += pp->accessed_nodes[i];
pp->isamb->skipped_nodes[i] += pp->skipped_nodes[i];
}
/* simple recursive dumper .. */
-static void isamb_dump_r (ISAMB b, ISAMB_P pos, void (*pr)(const char *str),
+static void isamb_dump_r (ISAMB b, ISAM_P pos, void (*pr)(const char *str),
int level)
{
char buf[1024];
else
{
const char *src = p->bytes + p->offset;
- ISAMB_P sub;
+ ISAM_P sub;
decode_ptr(&src, &sub);
p->offset = src - (char*) p->bytes;
}
}
-void isamb_dump(ISAMB b, ISAMB_P pos, void (*pr)(const char *str))
+void isamb_dump(ISAMB b, ISAM_P pos, void (*pr)(const char *str))
{
isamb_dump_r(b, pos, pr, 0);
}
}
} /* forward_on_leaf */
-static int isamb_pp_climb_level(ISAMB_PP pp, ISAMB_P *pos)
+static int isamb_pp_climb_level(ISAMB_PP pp, ISAM_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;
- ISAMB b = pp->isamb;
#if ISAMB_DEBUG
yaz_log(YLOG_DEBUG, "isamb_pp_climb_level starting "
"at level %d node %d ofs=%d sz=%d",
}
assert(pp->level>0);
close_block(pp->isamb, pp->block[pp->level]);
- pp->block[pp->level]=0;
+ pp->block[pp->level] = 0;
(pp->level)--;
p = pp->block[pp->level];
#if ISAMB_DEBUG
}
else
{
+#if INT_ENCODE
+ char file_item_buf[DST_ITEM_MAX];
+ char *file_item = file_item_buf;
+ ISAMB b = pp->isamb;
+ void *c1 = (*b->method->codec.start)();
+#else
+ zint item_len;
+#endif
/* skip the child we just came from */
#if ISAMB_DEBUG
yaz_log(YLOG_DEBUG, "isam_pp_climb_level: skipping lev=%d ofs=%d sz=%d",
assert (p->offset < p->size);
src = p->bytes + p->offset;
#if INT_ENCODE
- char file_item_buf[DST_ITEM_MAX];
- char *file_item = file_item_buf;
- void *c1 = (*b->method->codec.start)();
(*b->method->codec.decode)(c1, &file_item, &src);
(*b->method->codec.stop)(c1);
#else
- zint item_len;
decode_item_len(&src, &item_len);
src += item_len;
#endif
} /* forward_unode */
-static void isamb_pp_descend_to_leaf(ISAMB_PP pp, ISAMB_P pos,
+static void isamb_pp_descend_to_leaf(ISAMB_PP pp, ISAM_P pos,
const void *untilbuf)
{ /* climbs down the tree, from pos, to the leftmost leaf */
struct ISAMB_block *p = pp->block[pp->level];
static int isamb_pp_find_next_leaf(ISAMB_PP pp)
{ /* finds the next leaf by climbing up and down */
- ISAMB_P pos;
+ ISAM_P pos;
if (!isamb_pp_climb_level(pp, &pos))
return 0;
isamb_pp_descend_to_leaf(pp, pos, 0);
static int isamb_pp_climb_desc(ISAMB_PP pp, const void *untilbuf)
{ /* climbs up and descends to a leaf where values >= *untilbuf are found */
- ISAMB_P pos;
+ ISAM_P pos;
#if ISAMB_DEBUG
struct ISAMB_block *p = pp->block[pp->level];
yaz_log(YLOG_DEBUG, "isamb_pp_climb_desc starting "
if (isamb_pp_forward_on_leaf(pp, buf, untilbuf))
{
#if ISAMB_DEBUG
- yaz_log(YLOG_DEBUG, "isamb_pp_forward (f) returning (C) "
- "at level %d node %d ofs=%d sz=%d",
- pp->level, p->pos, p->offset, p->size);
+ yaz_log(YLOG_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;
}
assert(current);
assert(p->leaf);
- *total = pp->block[0]->no_items;
+ *total = (double) (pp->block[0]->no_items);
*current = (double) pp->returned_numbers;
#if ISAMB_DEBUG
yaz_log(YLOG_LOG, "isamb_pp_pos returning: cur= %0.1f tot=%0.1f rn="
again:
while (p->offset == p->size)
{
- ISAMB_P pos;
+ ISAM_P pos;
#if INT_ENCODE
const char *src_0;
void *c1;
projects / idzebra-moved-to-github.git / blobdiff