* Sebastian Hammer, Adam Dickmeiss
*
* $Log: zrpn.c,v $
- * Revision 1.96 1999-09-23 10:05:05 adam
+ * Revision 1.101 2000-03-02 14:35:03 adam
+ * Fixed proximity handling.
+ *
+ * Revision 1.100 1999/12/28 15:48:12 adam
+ * Minor Fix.
+ *
+ * Revision 1.99 1999/12/23 09:03:32 adam
+ * Changed behaviour of trunc=105 so that * is regular .* and ! is regular .
+ *
+ * Revision 1.98 1999/11/30 13:48:04 adam
+ * Improved installation. Updated for inclusion of YAZ header files.
+ *
+ * Revision 1.97 1999/10/14 14:33:50 adam
+ * Added truncation 5=106.
+ *
+ * Revision 1.96 1999/09/23 10:05:05 adam
* Implemented structure=105 searching.
*
* Revision 1.95 1999/09/07 07:19:21 adam
#ifdef TERM_COUNT
int *term_no;
#endif
- ISAM_P *isam_p_buf;
+ ISAMS_P *isam_p_buf;
int isam_p_size;
int isam_p_indx;
ZebraHandle zh;
{
if (p->isam_p_indx == p->isam_p_size)
{
- ISAM_P *new_isam_p_buf;
+ ISAMS_P *new_isam_p_buf;
#ifdef TERM_COUNT
int *new_term_no;
#endif
p->isam_p_size = 2*p->isam_p_size + 100;
- new_isam_p_buf = (ISAM_P *) xmalloc (sizeof(*new_isam_p_buf) *
+ new_isam_p_buf = (ISAMS_P *) xmalloc (sizeof(*new_isam_p_buf) *
p->isam_p_size);
if (p->isam_p_buf)
{
return i;
}
-/* term_105: handle term, where trunc=Process # and ! and right trunc */
+/* term_105/106: handle term, where trunc=Process * and ! and right trunc */
static int term_105 (ZebraMaps zebra_maps, int reg_type,
const char **src, char *dst, int space_split,
- char *dst_term)
+ char *dst_term, int right_truncate)
{
const char *s0, *s1;
const char **map;
if (*s0 == '*')
{
dst[i++] = '.';
- dst[i++] = '+';
+ dst[i++] = '*';
dst_term[j++] = *s0++;
}
else if (*s0 == '!')
}
}
}
- dst[i++] = '.';
- dst[i++] = '*';
+ if (right_truncate)
+ {
+ dst[i++] = '.';
+ dst[i++] = '*';
+ }
dst[i] = '\0';
dst_term[j++] = '\0';
case 105: /* process * and ! in term */
term_dict[j++] = '(';
if (!term_105 (zh->zebra_maps, reg_type,
- &termp, term_dict + j, space_split, term_dst))
+ &termp, term_dict + j, space_split, term_dst, 1))
+ return 0;
+ strcat (term_dict, ")");
+ r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info,
+ &max_pos, 0, grep_handle);
+ if (r)
+ logf (LOG_WARN, "dict_lookup_grep err, trunc=*/!: %d", r);
+ break;
+ case 106: /* process * and ! in term */
+ term_dict[j++] = '(';
+ if (!term_105 (zh->zebra_maps, reg_type,
+ &termp, term_dict + j, space_split, term_dst, 0))
return 0;
strcat (term_dict, ")");
r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info,
termz[i] = '\0';
}
-static RSET rpn_proximity (ZebraHandle zh, RSET rset1, RSET rset2,
- int ordered,
- int exclusion, int relation, int distance)
-{
- int i;
- RSFD rsfd1, rsfd2;
- int more1, more2;
- struct it_key buf1, buf2;
- RSFD rsfd_result;
- RSET result;
- rset_temp_parms parms;
- int term_index;
-
- rsfd1 = rset_open (rset1, RSETF_READ);
- more1 = rset_read (rset1, rsfd1, &buf1, &term_index);
-
- rsfd2 = rset_open (rset2, RSETF_READ);
- more2 = rset_read (rset2, rsfd2, &buf2, &term_index);
-
- parms.key_size = sizeof (struct it_key);
- parms.temp_path = res_get (zh->res, "setTmpDir");
- result = rset_create (rset_kind_temp, &parms);
- rsfd_result = rset_open (result, RSETF_WRITE);
-
- logf (LOG_DEBUG, "rpn_proximity excl=%d ord=%d rel=%d dis=%d",
- exclusion, ordered, relation, distance);
- while (more1 && more2)
- {
- int cmp = key_compare_it (&buf1, &buf2);
- if (cmp < -1)
- more1 = rset_read (rset1, rsfd1, &buf1, &term_index);
- else if (cmp > 1)
- more2 = rset_read (rset2, rsfd2, &buf2, &term_index);
- else
- {
- int sysno = buf1.sysno;
- int seqno[500];
- int n = 0;
-
- seqno[n++] = buf1.seqno;
- while ((more1 = rset_read (rset1, rsfd1, &buf1, &term_index)) &&
- sysno == buf1.sysno)
- if (n < 500)
- seqno[n++] = buf1.seqno;
- do
- {
- for (i = 0; i<n; i++)
- {
- int diff = buf2.seqno - seqno[i];
- int excl = exclusion;
- if (!ordered && diff < 0)
- diff = -diff;
- switch (relation)
- {
- case 1: /* < */
- if (diff < distance)
- excl = !excl;
- break;
- case 2: /* <= */
- if (diff <= distance)
- excl = !excl;
- break;
- case 3: /* == */
- if (diff == distance)
- excl = !excl;
- break;
- case 4: /* >= */
- if (diff >= distance)
- excl = !excl;
- break;
- case 5: /* > */
- if (diff > distance)
- excl = !excl;
- break;
- case 6: /* != */
- if (diff != distance)
- excl = !excl;
- break;
- }
- if (excl)
- rset_write (result, rsfd_result, &buf2);
- }
- } while ((more2 = rset_read (rset2, rsfd2, &buf2, &term_index)) &&
- sysno == buf2.sysno);
- }
- }
- rset_close (result, rsfd_result);
- rset_close (rset1, rsfd1);
- rset_close (rset2, rsfd2);
- return result;
-}
-
-static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no)
+static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no,
+ int ordered, int exclusion, int relation, int distance)
{
int i;
RSFD *rsfd;
}
for (i = 0; i<rset_no; i++)
{
+ buf[i] = 0;
+ rsfd[i] = 0;
+ }
+ for (i = 0; i<rset_no; i++)
+ {
buf[i] = (struct it_key *) xmalloc (sizeof(**buf));
rsfd[i] = rset_open (rset[i], RSETF_READ);
if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i], &term_index)))
}
if (i != rset_no)
{
+ /* at least one is empty ... return null set */
rset_null_parms parms;
-
- while (i >= 0)
- {
- rset_close (rset[i], rsfd[i]);
- xfree (buf[i]);
- --i;
- }
+
parms.rset_term = rset_term_create (prox_term, length_prox_term,
flags);
parms.rset_term->nn = 0;
result = rset_create (rset_kind_null, &parms);
}
- else
+ else if (ordered && relation == 3 && exclusion == 0 && distance == 1)
{
+ /* special proximity case = phrase search ... */
rset_temp_parms parms;
RSFD rsfd_result;
more[0] = rset_read (*rset, *rsfd, *buf, &term_index);
}
}
-
- for (i = 0; i<rset_no; i++)
+ rset_close (result, rsfd_result);
+ }
+ else if (rset_no == 2)
+ {
+ /* generic proximity case (two input sets only) ... */
+ rset_temp_parms parms;
+ RSFD rsfd_result;
+
+ logf (LOG_LOG, "generic prox, dist = %d, relation = %d, ordered =%d, exclusion=%d",
+ distance, relation, ordered, exclusion);
+ parms.rset_term = rset_term_create (prox_term, length_prox_term,
+ flags);
+ parms.rset_term->nn = min_nn;
+ parms.key_size = sizeof (struct it_key);
+ parms.temp_path = res_get (zh->res, "setTmpDir");
+ result = rset_create (rset_kind_temp, &parms);
+ rsfd_result = rset_open (result, RSETF_WRITE);
+
+ while (more[0] && more[1])
{
- rset_close (rset[i], rsfd[i]);
- xfree (buf[i]);
+ int cmp = key_compare_it (buf[0], buf[1]);
+ if (cmp < -1)
+ more[0] = rset_read (rset[0], rsfd[0], buf[0], &term_index);
+ else if (cmp > 1)
+ more[1] = rset_read (rset[1], rsfd[1], buf[1], &term_index);
+ else
+ {
+ int sysno = buf[0]->sysno;
+ int seqno[500];
+ int n = 0;
+
+ seqno[n++] = buf[0]->seqno;
+ while ((more[0] = rset_read (rset[0], rsfd[0], buf[0],
+ &term_index)) &&
+ sysno == buf[0]->sysno)
+ if (n < 500)
+ seqno[n++] = buf[0]->seqno;
+ do
+ {
+ for (i = 0; i<n; i++)
+ {
+ int diff = buf[1]->seqno - seqno[i];
+ int excl = exclusion;
+ if (!ordered && diff < 0)
+ diff = -diff;
+ switch (relation)
+ {
+ case 1: /* < */
+ if (diff < distance && diff >= 0)
+ excl = !excl;
+ break;
+ case 2: /* <= */
+ if (diff <= distance && diff >= 0)
+ excl = !excl;
+ break;
+ case 3: /* == */
+ if (diff == distance && diff >= 0)
+ excl = !excl;
+ break;
+ case 4: /* >= */
+ if (diff >= distance && diff >= 0)
+ excl = !excl;
+ break;
+ case 5: /* > */
+ if (diff > distance && diff >= 0)
+ excl = !excl;
+ break;
+ case 6: /* != */
+ if (diff != distance && diff >= 0)
+ excl = !excl;
+ break;
+ }
+ if (excl)
+ {
+ rset_write (result, rsfd_result, buf[1]);
+ break;
+ }
+ }
+ } while ((more[1] = rset_read (rset[1], rsfd[1], buf[1],
+ &term_index)) &&
+ sysno == buf[1]->sysno);
+ }
}
rset_close (result, rsfd_result);
}
+ else
+ {
+ rset_null_parms parms;
+
+ parms.rset_term = rset_term_create (prox_term, length_prox_term,
+ flags);
+ parms.rset_term->nn = 0;
+ result = rset_create (rset_kind_null, &parms);
+ }
+ for (i = 0; i<rset_no; i++)
+ {
+ if (rset[i])
+ rset_close (rset[i], rsfd[i]);
+ xfree (buf[i]);
+ }
xfree (buf);
xfree (more);
xfree (rsfd);
}
else if (rset_no == 1)
return (rset[0]);
- result = rpn_prox (zh, rset, rset_no);
+ result = rpn_prox (zh, rset, rset_no, 1, 0, 3, 1);
for (i = 0; i<rset_no; i++)
rset_delete (rset[i]);
return result;
return NULL;
}
#endif
- r = rpn_proximity (zh, bool_parms.rset_l, bool_parms.rset_r,
- *zop->u.prox->ordered,
- (!zop->u.prox->exclusion ? 0 :
- *zop->u.prox->exclusion),
- *zop->u.prox->relationType,
- *zop->u.prox->distance);
+ else
+ {
+ RSET rsets[2];
+
+ rsets[0] = bool_parms.rset_l;
+ rsets[1] = bool_parms.rset_r;
+
+ r = rpn_prox (zh, rsets, 2,
+ *zop->u.prox->ordered,
+ (!zop->u.prox->exclusion ? 0 :
+ *zop->u.prox->exclusion),
+ *zop->u.prox->relationType,
+ *zop->u.prox->distance);
+ rset_delete (rsets[0]);
+ rset_delete (rsets[1]);
+ }
break;
default:
zh->errCode = 110;
struct scan_info_entry {
char *term;
- ISAM_P isam_p;
+ ISAMS_P isam_p;
};
struct scan_info {
scan_info->list[idx].term = (char *)
odr_malloc (scan_info->odr, strlen(name + len_prefix)+1);
strcpy (scan_info->list[idx].term, name + len_prefix);
- assert (*info == sizeof(ISAM_P));
- memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAM_P));
+ assert (*info == sizeof(ISAMS_P));
+ memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAMS_P));
return 0;
}
projects / idzebra-moved-to-github.git / blobdiff