-/* $Id: zsets.c,v 1.100 2006-05-03 09:31:26 marc Exp $
- Copyright (C) 1995-2005
+/* $Id: zsets.c,v 1.122 2007-08-21 11:06:47 adam Exp $
+ Copyright (C) 1995-2007
Index Data ApS
This file is part of the Zebra server.
for more details.
You should have received a copy of the GNU General Public License
-along with Zebra; see the file LICENSE.zebra. If not, write to the
-Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
-02111-1307, USA.
+along with this program; if not, write to the Free Software
+Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
+
*/
#include <yaz/diagbib1.h>
#include <rset.h>
-#define SORT_IDX_ENTRYSIZE 64
-#define ZSET_SORT_MAX_LEVEL 3
+#define ZSET_SORT_MAX_LEVEL 10
struct zebra_set_term_entry {
int reg_type;
char *db;
- int set;
- int use;
+ char *index_name;
char *term;
};
int term_entries_max;
struct zebra_set *next;
int locked;
+ int estimated_hit_count;
zint cache_position; /* last position */
RSFD cache_rfd; /* rfd (NULL if not existing) */
struct zset_sort_entry {
zint sysno;
int score;
- char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
};
struct zset_sort_info {
static int log_level_searchterms=0;
static int log_level_resultsets=0;
-static void loglevels()
+static void loglevels(void)
{
if (log_level_set)
return;
log_level_set = 1;
}
-ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem,
- Z_RPNQuery *rpn, ZebraSet sset)
+
+static ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem,
+ Z_RPNQuery *rpn, ZebraSet sset)
{
RSET rset = 0;
- oident *attrset;
Z_SortKeySpecList *sort_sequence;
int sort_status, i;
ZEBRA_RES res = ZEBRA_OK;
- zh->hits = 0;
-
sort_sequence = (Z_SortKeySpecList *)
nmem_malloc(nmem, sizeof(*sort_sequence));
sort_sequence->num_specs = 10; /* FIXME - Hard-coded number */
for (i = 0; i<sort_sequence->num_specs; i++)
sort_sequence->specs[i] = 0;
- attrset = oid_getentbyoid (rpn->attributeSetId);
- res = rpn_search_top(zh, rpn->RPNStructure, attrset->value,
+ rpn_get_top_approx_limit(zh, rpn->RPNStructure, &sset->approx_limit);
+
+ res = rpn_search_top(zh, rpn->RPNStructure, rpn->attributeSetId,
nmem, rset_nmem,
sort_sequence,
sset->num_bases, sset->basenames,
}
-ZEBRA_RES resultSetAddRPN (ZebraHandle zh, NMEM m, Z_RPNQuery *rpn,
- int num_bases, char **basenames,
- const char *setname)
+ZEBRA_RES resultSetAddRPN(ZebraHandle zh, NMEM m, Z_RPNQuery *rpn,
+ int num_bases, char **basenames,
+ const char *setname,
+ zint *hits, int *estimated_hit_count)
{
ZebraSet zebraSet;
int i;
ZEBRA_RES res;
- zh->hits = 0;
+ *hits = 0;
+ *estimated_hit_count = 0;
zebraSet = resultSetAdd(zh, setname, 1);
if (!zebraSet)
res = resultSetSearch(zh, zebraSet->nmem, zebraSet->rset_nmem,
rpn, zebraSet);
- zh->hits = zebraSet->hits;
+ *hits = zebraSet->hits;
+ if (zebraSet->estimated_hit_count)
+ *estimated_hit_count = 1;
+
if (zebraSet->rset)
zebraSet->rpn = rpn;
zebraSet->locked = 0;
return res;
}
-void resultSetAddTerm (ZebraHandle zh, ZebraSet s, int reg_type,
- const char *db, int set,
- int use, const char *term)
+void resultSetAddTerm(ZebraHandle zh, ZebraSet s, int reg_type,
+ const char *db, const char *index_name,
+ const char *term)
{
assert(zh); /* compiler shut up */
if (!s->nmem)
{
s->term_entries[s->hits].reg_type = reg_type;
s->term_entries[s->hits].db = nmem_strdup (s->nmem, db);
- s->term_entries[s->hits].set = set;
- s->term_entries[s->hits].use = use;
- s->term_entries[s->hits].term = nmem_strdup (s->nmem, term);
+ s->term_entries[s->hits].index_name = nmem_strdup(s->nmem, index_name);
+ s->term_entries[s->hits].term = nmem_strdup(s->nmem, term);
}
(s->hits)++;
}
s->cache_position = 0;
s->cache_rfd = 0;
s->approx_limit = zh->approx_limit;
+ s->estimated_hit_count = 0;
return s;
}
struct sortKeyInfo {
int relation;
- int attrUse;
+ int ord;
int numerical;
+ int index_type;
};
-void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
- struct sortKeyInfo *criteria, int num_criteria,
- zint sysno)
+void resultSetInsertSort(ZebraHandle zh, ZebraSet sset,
+ struct sortKeyInfo *criteria, int num_criteria,
+ zint sysno,
+ char *cmp_buf[], char *tmp_cmp_buf[])
{
- struct zset_sort_entry this_entry;
struct zset_sort_entry *new_entry = NULL;
struct zset_sort_info *sort_info = sset->sort_info;
int i, j;
- sortIdx_sysno (zh->reg->sortIdx, sysno);
+ zebra_sort_sysno(zh->reg->sort_index, sysno);
for (i = 0; i<num_criteria; i++)
{
- sortIdx_type (zh->reg->sortIdx, criteria[i].attrUse);
- sortIdx_read (zh->reg->sortIdx, this_entry.buf[i]);
+ char *this_entry_buf = tmp_cmp_buf[i];
+ memset(this_entry_buf, '\0', SORT_IDX_ENTRYSIZE);
+ if (criteria[i].ord != -1)
+ {
+ zebra_sort_type(zh->reg->sort_index, criteria[i].ord);
+ zebra_sort_read(zh->reg->sort_index, this_entry_buf);
+ }
}
i = sort_info->num_entries;
while (--i >= 0)
int rel = 0;
for (j = 0; j<num_criteria; j++)
{
+ char *this_entry_buf = tmp_cmp_buf[j];
+ char *other_entry_buf =
+ cmp_buf[j] + i * SORT_IDX_ENTRYSIZE;
if (criteria[j].numerical)
{
- double diff = atof(this_entry.buf[j]) -
- atof(sort_info->entries[i]->buf[j]);
- rel = 0;
+ char this_entry_org[1024];
+ char other_entry_org[1024];
+ double diff;
+ int index_type = criteria[j].index_type;
+ zebra_term_untrans(zh, index_type, this_entry_org,
+ this_entry_buf);
+ zebra_term_untrans(zh, index_type, other_entry_org,
+ other_entry_buf);
+ diff = atof(this_entry_org) - atof(other_entry_org);
+
if (diff > 0.0)
rel = 1;
else if (diff < 0.0)
rel = -1;
+ else
+ rel = 0;
}
else
{
- rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
- SORT_IDX_ENTRYSIZE);
+ rel = memcmp(this_entry_buf, other_entry_buf,
+ SORT_IDX_ENTRYSIZE);
}
if (rel)
break;
new_entry = sort_info->entries[j];
while (j != i)
{
+ int k;
+ for (k = 0; k<num_criteria; k++)
+ {
+ char *j_buf = cmp_buf[k] + j * SORT_IDX_ENTRYSIZE;
+ char *j_1_buf = cmp_buf[k] + (j-1) * SORT_IDX_ENTRYSIZE;
+ memcpy(j_buf, j_1_buf, SORT_IDX_ENTRYSIZE);
+ }
sort_info->entries[j] = sort_info->entries[j-1];
--j;
}
sort_info->entries[i] = new_entry;
assert (new_entry);
for (i = 0; i<num_criteria; i++)
- memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
+ {
+ char *new_entry_buf = cmp_buf[i] + j * SORT_IDX_ENTRYSIZE;
+ char *this_entry_buf = tmp_cmp_buf[i];
+ memcpy(new_entry_buf, this_entry_buf, SORT_IDX_ENTRYSIZE);
+ }
new_entry->sysno = sysno;
new_entry->score = -1;
}
-void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
- zint sysno, int score, int relation)
+void resultSetInsertRank(ZebraHandle zh, struct zset_sort_info *sort_info,
+ zint sysno, int score, int relation)
{
struct zset_sort_entry *new_entry = NULL;
int i, j;
zint kno = 0;
zint psysno = 0;
struct it_key key;
- struct sortKeyInfo sort_criteria[3];
+ struct sortKeyInfo sort_criteria[ZSET_SORT_MAX_LEVEL];
+ char *cmp_buf[ZSET_SORT_MAX_LEVEL];
+ char *tmp_cmp_buf[ZSET_SORT_MAX_LEVEL];
int num_criteria;
RSFD rfd;
TERMID termid;
if (zh->m_staticrank)
sysno_mem_index = 1;
-
assert(nmem); /* compiler shut up about unused param */
sset->sort_info->num_entries = 0;
sset->hits = 0;
num_criteria = sort_sequence->num_specs;
- if (num_criteria > 3)
- num_criteria = 3;
+ if (num_criteria > ZSET_SORT_MAX_LEVEL)
+ num_criteria = ZSET_SORT_MAX_LEVEL;
for (i = 0; i < num_criteria; i++)
{
Z_SortKeySpec *sks = sort_sequence->specs[i];
Z_SortKey *sk;
+ ZEBRA_RES res;
+ sort_criteria[i].ord = -1;
+ sort_criteria[i].numerical = 0;
+
+ if (sks->which == Z_SortKeySpec_missingValueData)
+ {
+ zebra_setError(zh, YAZ_BIB1_UNSUPP_MISSING_DATA_ACTION, 0);
+ return ZEBRA_FAIL;
+ }
if (*sks->sortRelation == Z_SortKeySpec_ascending)
sort_criteria[i].relation = 'A';
else if (*sks->sortRelation == Z_SortKeySpec_descending)
case Z_SortKey_sortField:
yaz_log(log_level_sort, "key %d is of type sortField",
i+1);
- zebra_setError(zh, YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE, 0);
- return ZEBRA_FAIL;
+ sort_criteria[i].numerical = 0;
+ sort_criteria[i].ord =
+ zebraExplain_lookup_attr_str(zh->reg->zei,
+ zinfo_index_category_sort,
+ -1, sk->u.sortField);
+ if (sks->which != Z_SortKeySpec_null
+ && sort_criteria[i].ord == -1)
+ {
+ zebra_setError(zh,
+ YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE, 0);
+ return ZEBRA_FAIL;
+ }
+ break;
case Z_SortKey_elementSpec:
yaz_log(log_level_sort, "key %d is of type elementSpec",
i+1);
return ZEBRA_FAIL;
case Z_SortKey_sortAttributes:
yaz_log(log_level_sort, "key %d is of type sortAttributes", i+1);
- sort_criteria[i].attrUse =
- zebra_maps_sort (zh->reg->zebra_maps,
- sk->u.sortAttributes,
- &sort_criteria[i].numerical);
- yaz_log(log_level_sort, "use value = %d", sort_criteria[i].attrUse);
- if (sort_criteria[i].attrUse == -1)
- {
- zebra_setError(
- zh, YAZ_BIB1_USE_ATTRIBUTE_REQUIRED_BUT_NOT_SUPPLIED, 0);
- return ZEBRA_FAIL;
- }
- if (sortIdx_type (zh->reg->sortIdx, sort_criteria[i].attrUse))
- {
- zebra_setError(
- zh, YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE, 0);
+ res = zebra_sort_get_ord(zh, sk->u.sortAttributes,
+
+ &sort_criteria[i].ord,
+ &sort_criteria[i].numerical);
+ if (sks->which != Z_SortKeySpec_null && res != ZEBRA_OK)
return ZEBRA_FAIL;
- }
break;
}
+ if (zebraExplain_lookup_ord(zh->reg->zei, sort_criteria[i].ord,
+ &sort_criteria[i].index_type,
+ 0, 0))
+ {
+ zebra_setError(zh, YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE, 0);
+ return ZEBRA_FAIL;
+ }
+ }
+ /* allocate space for each cmpare buf + one extra for tmp comparison */
+ for (i = 0; i<num_criteria; i++)
+ {
+ cmp_buf[i] = xmalloc(sset->sort_info->max_entries
+ * SORT_IDX_ENTRYSIZE);
+ tmp_cmp_buf[i] = xmalloc(SORT_IDX_ENTRYSIZE);
}
rfd = rset_open (rset, RSETF_READ);
while (rset_read (rfd, &key, &termid))
kno++;
if (this_sys != psysno)
{
+ if ((sset->hits & 255) == 0 && zh->break_handler_func)
+ {
+ if (zh->break_handler_func(zh->break_handler_data))
+ {
+ rset_set_hits_limit(rset, 0);
+ break;
+ }
+ }
(sset->hits)++;
psysno = this_sys;
- resultSetInsertSort (zh, sset,
- sort_criteria, num_criteria, psysno);
+ resultSetInsertSort(zh, sset,
+ sort_criteria, num_criteria, psysno, cmp_buf,
+ tmp_cmp_buf);
}
}
rset_close (rfd);
+
+ for (i = 0; i<num_criteria; i++)
+ {
+ xfree(cmp_buf[i]);
+ xfree(tmp_cmp_buf[i]);
+ }
+
yaz_log(log_level_sort, ZINT_FORMAT " keys, " ZINT_FORMAT " sysnos, sort",
kno, sset->hits);
for (i = 0; i < numTerms; i++)
sort_info = zebraSet->sort_info;
sort_info->num_entries = 0;
zebraSet->hits = 0;
+ zebraSet->estimated_hit_count = 0;
rset_getterms(rset, 0, 0, &n);
terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n);
rset_getterms(rset, terms, n, &numTerms);
-
rank_class = zebraRankLookup(zh, rank_handler_name);
if (!rank_class)
{
{
RSFD rfd = rset_open(rset, RSETF_READ);
struct rank_control *rc = rank_class->control;
- double score;
+ int score;
zint count = 0;
-
- void *handle =
- (*rc->begin) (zh->reg, rank_class->class_handle, rset, nmem,
- terms, numTerms);
+ void *handle = (*rc->begin) (zh->reg, rank_class->class_handle, rset,
+ nmem, terms, numTerms);
zint psysno = 0; /* previous doc id / sys no */
zint pstaticrank = 0; /* previous static rank */
int stop_flag = 0;
key_logdump_txt(log_level_searchhits, &key, termid->name);
if (this_sys != psysno)
{ /* new record .. */
+ if (!(rfd->counted_items & 255) && zh->break_handler_func)
+ {
+ if (zh->break_handler_func(zh->break_handler_data))
+ {
+ yaz_log(YLOG_LOG, "Aborted search");
+ stop_flag = 1;
+ }
+ }
if (rfd->counted_items > rset->hits_limit)
- break;
+ stop_flag = 1;
if (psysno)
{ /* only if we did have a previous record */
score = (*rc->calc) (handle, psysno, pstaticrank,
/* insert the hit. A=Ascending */
resultSetInsertRank (zh, sort_info, psysno, score, 'A');
count++;
- if (stop_flag)
- break;
}
+ if (stop_flag)
+ {
+ zebraSet->estimated_hit_count = 1;
+ rset_set_hits_limit(rset, 0);
+ break;
+ }
psysno = this_sys;
if (zh->m_staticrank)
pstaticrank = key.mem[0];
if (ret == (size_t)(-1))
*termlen = 0;
else
+ {
+ yaz_iconv(zh->iconv_from_utf8, 0, 0,
+ &outbuf, &outleft);
*termlen = outbuf - termbuf;
+ }
}
else
{
NMEM nmem = nmem_create();
struct it_key key;
RSET rsets[2], rset_comb;
- RSET rset_temp = rstemp_create(nmem, kc, kc->scope,
- res_get (zh->res, "setTmpDir"),0 );
+ RSET rset_temp = rset_create_temp(nmem, kc, kc->scope,
+ res_get (zh->res, "setTmpDir"),0 );
TERMID termid;
RSFD rsfd = rset_open(rset_temp, RSETF_WRITE);
rsets[0] = rset_temp;
rsets[1] = rset_dup(sset->rset);
- rset_comb = rsmulti_and_create(nmem, kc, kc->scope, 2, rsets);
+ rset_comb = rset_create_and(nmem, kc, kc->scope, 2, rsets);
rsfd = rset_open(rset_comb, RSETF_READ);
struct ord_list *ol;
for (ol = termid->ol; ol; ol = ol->next)
{
- zebra_snippets_append(snippets, key.mem[key.len-1],
+ zebra_snippets_append(snippets, key.mem[key.len-1], 0,
ol->ord, termid->name);
}
}
return ZEBRA_OK;
}
+/*
+ * Local variables:
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ * vim: shiftwidth=4 tabstop=8 expandtab
+ */
+
projects / idzebra-moved-to-github.git / blobdiff