* Sebastian Hammer, Adam Dickmeiss
*
* $Log: zrpn.c,v $
- * Revision 1.16 1995-10-02 16:24:40 adam
+ * Revision 1.23 1995-10-06 16:33:37 adam
+ * Use attribute mappings.
+ *
+ * Revision 1.22 1995/10/06 15:07:39 adam
+ * Structure 'local-number' handled.
+ *
+ * Revision 1.21 1995/10/06 13:52:06 adam
+ * Bug fixes. Handler may abort further scanning.
+ *
+ * Revision 1.20 1995/10/06 11:06:33 adam
+ * Scan entries include 'occurrences' now.
+ *
+ * Revision 1.19 1995/10/06 10:43:56 adam
+ * Scan added. 'occurrences' in scan entries not set yet.
+ *
+ * Revision 1.18 1995/10/04 16:57:20 adam
+ * Key input and merge sort in one pass.
+ *
+ * Revision 1.17 1995/10/04 12:55:17 adam
+ * Bug fix in ranked search. Use=Any keys inserted.
+ *
+ * Revision 1.16 1995/10/02 16:24:40 adam
* Use attribute actually used in search requests.
*
* Revision 1.15 1995/10/02 15:18:52 adam
#include <unistd.h>
#include "zserver.h"
+#include <attribute.h>
#include <rsisam.h>
#include <rstemp.h>
#include <rsbool.h>
#include <rsrel.h>
+int index_word_prefix_map (char *string, oid_value attrSet, int attrUse)
+{
+ attent *attp;
+
+ logf (LOG_DEBUG, "oid_value attrSet = %d", attrSet);
+ logf (LOG_DEBUG, "int attrUse = %d", attrUse);
+ attp = att_getentbyatt (attrSet, attrUse);
+ assert (attp);
+ logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal);
+ return index_word_prefix (string, attp->attset_ordinal,
+ attp->local_attribute);
+}
+
/*
* attr_print: log attributes
*/
Z_AttributesPlusTerm *zapt;
} AttrType;
-static int attr_find (AttrType *src)
+static int attr_find (AttrType *src, oid_value *attributeSetP)
{
while (src->major < src->zapt->num_attributes)
{
Z_AttributeElement *element;
- element = src->zapt->attributeList[src->major];
+ element = src->zapt->attributeList[src->major];
if (src->type == *element->attributeType)
{
switch (element->which)
{
case Z_AttributeValue_numeric:
++(src->major);
+ if (element->attributeSet && attributeSetP)
+ {
+ oident *attrset;
+
+ attrset = oid_getentbyoid (element->attributeSet);
+ *attributeSetP = attrset->value;
+ }
return *element->value.numeric;
break;
case Z_AttributeValue_complex:
Z_StringOrNumeric_numeric)
break;
++(src->minor);
+ if (element->attributeSet && attributeSetP)
+ {
+ oident *attrset;
+
+ attrset = oid_getentbyoid (element->attributeSet);
+ *attributeSetP = attrset->value;
+ }
return *element->value.complex->list[src->minor-1]->u.numeric;
default:
assert (0);
}
static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
- const char *term_sub, ISAM_P **isam_ps)
+ const char *term_sub, ISAM_P **isam_ps,
+ oid_value attributeSet)
{
char term_dict[2*IT_MAX_WORD+2];
int i, j;
const char *info;
AttrType truncation;
int truncation_value;
+ AttrType use;
+ int use_value;
+ oid_value curAttributeSet = attributeSet;
+ attr_init (&use, zapt, 1);
+ use_value = attr_find (&use, &curAttributeSet);
+ logf (LOG_DEBUG, "use value %d", use_value);
attr_init (&truncation, zapt, 5);
- truncation_value = attr_find (&truncation);
+ truncation_value = attr_find (&truncation, NULL);
logf (LOG_DEBUG, "truncation value %d", truncation_value);
+
+ if (use_value == -1)
+ use_value = 1016;
+ i = index_word_prefix_map (term_dict, curAttributeSet, use_value);
+
switch (truncation_value)
{
case -1: /* not specified */
case 100: /* do not truncate */
- strcpy (term_dict, term_sub);
+ strcat (term_dict, term_sub);
logf (LOG_DEBUG, "dict_lookup: %s", term_dict);
if ((info = dict_lookup (zi->wordDict, term_dict)))
add_isam_p (info);
break;
case 1: /* right truncation */
- strcpy (term_dict, term_sub);
+ strcat (term_dict, term_sub);
strcat (term_dict, ".*");
dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle);
break;
zi->errCode = 120;
return -1;
case 101: /* process # in term */
- for (j = 0, i = 0; term_sub[i] && i < 2; i++)
+ for (j = strlen(term_dict), i = 0; term_sub[i] && i < 2; i++)
term_dict[j++] = term_sub[i];
for (; term_sub[i]; i++)
if (term_sub[i] == '#')
dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle);
break;
case 102: /* regular expression */
- strcpy (term_dict, term_sub);
+ strcat (term_dict, term_sub);
dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle);
break;
}
return 0;
}
-static void field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
+static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
char *termz)
{
- size_t i, j, sizez;
- AttrType use;
- int use_value;
+ size_t i, sizez;
Z_Term *term = zapt->term;
- attr_init (&use, zapt, 1);
- use_value = attr_find (&use);
- if (use_value == -1)
- use_value = 1016;
-
- i = index_word_prefix (termz, 1, use_value);
- sizez = i + term->u.general->len;
+ sizez = term->u.general->len;
if (sizez > IT_MAX_WORD)
sizez = IT_MAX_WORD;
- for (j = 0; i < sizez; i++, j++)
- termz[i] = index_char_cvt (term->u.general->buf[j]);
+ for (i = 0; i < sizez; i++)
+ termz[i] = index_char_cvt (term->u.general->buf[i]);
termz[i] = '\0';
}
-
static RSET rpn_search_APT_relevance (ZServerInfo *zi,
- Z_AttributesPlusTerm *zapt)
+ Z_AttributesPlusTerm *zapt,
+ oid_value attributeSet)
{
rset_relevance_parms parms;
char termz[IT_MAX_WORD+1];
char term_sub[IT_MAX_WORD+1];
char *p0 = termz, *p1 = NULL;
- Z_Term *term = zapt->term;
parms.key_size = sizeof(struct it_key);
parms.max_rec = 100;
parms.cmp = key_compare;
parms.is = zi->wordIsam;
- if (term->which != Z_Term_general)
+ if (zapt->term->which != Z_Term_general)
{
zi->errCode = 124;
return NULL;
}
- field_term (zi, zapt, termz);
+ trans_term (zi, zapt, termz);
isam_p_indx = 0; /* global, set by trunc_term - see below */
while (1)
{
}
else
strcpy (term_sub, p0);
- if (trunc_term (zi, zapt, term_sub, &parms.isam_positions))
+ if (trunc_term (zi, zapt, term_sub, &parms.isam_positions,
+ attributeSet))
return NULL;
if (!p1)
break;
}
static RSET rpn_search_APT_word (ZServerInfo *zi,
- Z_AttributesPlusTerm *zapt)
+ Z_AttributesPlusTerm *zapt,
+ oid_value attributeSet)
{
ISAM_P *isam_positions;
rset_isam_parms parms;
-
char termz[IT_MAX_WORD+1];
- Z_Term *term = zapt->term;
- if (term->which != Z_Term_general)
+ if (zapt->term->which != Z_Term_general)
{
zi->errCode = 124;
return NULL;
}
- field_term (zi, zapt, termz);
+ trans_term (zi, zapt, termz);
isam_p_indx = 0; /* global, set by trunc_term - see below */
- if (trunc_term (zi, zapt, termz, &isam_positions))
+ if (trunc_term (zi, zapt, termz, &isam_positions, attributeSet))
return NULL;
if (isam_p_indx < 1)
return rset_create (rset_kind_null, NULL);
}
static RSET rpn_search_APT_phrase (ZServerInfo *zi,
- Z_AttributesPlusTerm *zapt)
+ Z_AttributesPlusTerm *zapt,
+ oid_value attributeSet)
{
ISAM_P *isam_positions;
rset_isam_parms parms;
-
char termz[IT_MAX_WORD+1];
- Z_Term *term = zapt->term;
- if (term->which != Z_Term_general)
+ if (zapt->term->which != Z_Term_general)
{
zi->errCode = 124;
return NULL;
}
- field_term (zi, zapt, termz);
+ trans_term (zi, zapt, termz);
isam_p_indx = 0; /* global, set by trunc_term - see below */
- if (trunc_term (zi, zapt, termz, &isam_positions))
+ if (trunc_term (zi, zapt, termz, &isam_positions, attributeSet))
return NULL;
if (isam_p_indx != 1)
return rset_create (rset_kind_null, NULL);
return rset_create (rset_kind_isam, &parms);
}
-static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt)
+static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
+ oid_value attributeSet)
+{
+ RSET result;
+ RSFD rsfd;
+ struct it_key key;
+ rset_temp_parms parms;
+ char termz[IT_MAX_WORD+1];
+
+ if (zapt->term->which != Z_Term_general)
+ {
+ zi->errCode = 124;
+ return NULL;
+ }
+ parms.key_size = sizeof (struct it_key);
+ result = rset_create (rset_kind_temp, &parms);
+ rsfd = rset_open (result, 1);
+
+ trans_term (zi, zapt, termz);
+ key.sysno = atoi (termz);
+ if (key.sysno <= 0)
+ key.sysno = 1;
+ rset_write (result, rsfd, &key);
+ rset_close (result, rsfd);
+ return result;
+}
+
+
+static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
+ oid_value attributeSet)
{
AttrType relation;
AttrType structure;
attr_init (&relation, zapt, 2);
attr_init (&structure, zapt, 4);
- relation_value = attr_find (&relation);
- structure_value = attr_find (&structure);
+ relation_value = attr_find (&relation, NULL);
+ structure_value = attr_find (&structure, NULL);
switch (structure_value)
{
case -1:
if (relation_value == 102) /* relevance relation */
- return rpn_search_APT_relevance (zi, zapt);
- return rpn_search_APT_word (zi, zapt);
+ return rpn_search_APT_relevance (zi, zapt, attributeSet);
+ return rpn_search_APT_word (zi, zapt, attributeSet);
case 1: /* phrase */
if (relation_value == 102) /* relevance relation */
- return rpn_search_APT_relevance (zi, zapt);
- return rpn_search_APT_phrase (zi, zapt);
+ return rpn_search_APT_relevance (zi, zapt, attributeSet);
+ return rpn_search_APT_phrase (zi, zapt, attributeSet);
break;
case 2: /* word */
if (relation_value == 102) /* relevance relation */
- return rpn_search_APT_relevance (zi, zapt);
- return rpn_search_APT_word (zi, zapt);
+ return rpn_search_APT_relevance (zi, zapt, attributeSet);
+ return rpn_search_APT_word (zi, zapt, attributeSet);
case 3: /* key */
break;
case 4: /* year */
case 5: /* date - normalized */
break;
case 6: /* word list */
- return rpn_search_APT_relevance (zi, zapt);
+ return rpn_search_APT_relevance (zi, zapt, attributeSet);
case 100: /* date - un-normalized */
break;
case 101: /* name - normalized */
case 104: /* urx */
break;
case 105: /* free-form-text */
- return rpn_search_APT_relevance (zi, zapt);
+ return rpn_search_APT_relevance (zi, zapt, attributeSet);
case 106: /* document-text */
- return rpn_search_APT_relevance (zi, zapt);
+ return rpn_search_APT_relevance (zi, zapt, attributeSet);
case 107: /* local-number */
- break;
+ return rpn_search_APT_local (zi, zapt, attributeSet);
case 108: /* string */
- return rpn_search_APT_word (zi, zapt);
+ return rpn_search_APT_word (zi, zapt, attributeSet);
case 109: /* numeric string */
break;
}
return s->rset;
}
-static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs)
+static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs,
+ oid_value attributeSet)
{
RSET r = NULL;
if (zs->which == Z_RPNStructure_complex)
{
rset_bool_parms bool_parms;
- bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1);
+ bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
+ attributeSet);
if (bool_parms.rset_l == NULL)
return NULL;
- bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2);
+ bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2,
+ attributeSet);
if (bool_parms.rset_r == NULL)
{
rset_delete (bool_parms.rset_l);
if (zs->u.simple->which == Z_Operand_APT)
{
logf (LOG_DEBUG, "rpn_search_APT");
- r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm);
+ r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm,
+ attributeSet);
}
else if (zs->u.simple->which == Z_Operand_resultSetId)
{
const char *setname, int *hits)
{
RSET rset;
+ oident *attrset;
+ oid_value attributeSet;
zi->errCode = 0;
zi->errString = NULL;
- rset = rpn_search_structure (zi, rpn->RPNStructure);
+
+ attrset = oid_getentbyoid (rpn->attributeSetId);
+ attributeSet = attrset->value;
+
+ rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet);
if (!rset)
return zi->errCode;
count_set (rset, hits);
resultSetAdd (zi, setname, 1, rset);
+ if (zi->errCode)
+ logf (LOG_DEBUG, "search error: %d", zi->errCode);
return zi->errCode;
}
+static struct scan_entry *scan_list;
+static ODR scan_odr;
+static int scan_before, scan_after;
+static ISAM scan_isam;
+static char scan_prefix[20];
+
+static int scan_handle (Dict_char *name, const char *info, int pos)
+{
+ int len_prefix, idx;
+ ISAM_P isam_p;
+ RSET rset;
+
+ rset_isam_parms parms;
+
+ len_prefix = strlen(scan_prefix);
+ if (memcmp (name, scan_prefix, len_prefix))
+ return 1;
+ if (pos > 0)
+ idx = scan_after - pos + scan_before;
+ else
+ idx = - pos - 1;
+ scan_list[idx].term = odr_malloc (scan_odr, strlen(name + len_prefix)+1);
+ strcpy (scan_list[idx].term, name + len_prefix);
+ assert (*info == sizeof(isam_p));
+ memcpy (&isam_p, info+1, sizeof(isam_p));
+ parms.is = scan_isam;
+ parms.pos = isam_p;
+#if 1
+ rset = rset_create (rset_kind_isam, &parms);
+ count_set (rset, &scan_list[idx].occurrences);
+ rset_delete (rset);
+#else
+ scan_list[idx].occurrences = 1;
+#endif
+ logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name);
+ return 0;
+}
+
+int rpn_scan (ZServerInfo *zi, ODR odr, Z_AttributesPlusTerm *zapt,
+ int *position, int *num_entries, struct scan_entry **list)
+{
+ int i, j, sizez;
+ int pos = *position;
+ int num = *num_entries;
+ int before;
+ int after;
+ char termz[IT_MAX_WORD+20];
+ AttrType use;
+ int use_value;
+ Z_Term *term = zapt->term;
+
+ logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
+ scan_before = before = pos-1;
+ scan_after = after = 1+num-pos;
+ scan_odr = odr;
+
+ logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after);
+
+ scan_isam = zi->wordIsam;
+ scan_list = *list = odr_malloc (odr, (before+after)*sizeof(**list));
+ for (j = 0; j<before+after; j++)
+ scan_list[j].term = "------";
+ attr_init (&use, zapt, 1);
+ use_value = attr_find (&use, NULL);
+ logf (LOG_DEBUG, "use value %d", use_value);
+
+ if (use_value == -1)
+ use_value = 1016;
+ i = index_word_prefix (termz, 1, use_value);
+ strcpy (scan_prefix, termz);
+ sizez = term->u.general->len;
+ if (sizez > IT_MAX_WORD)
+ sizez = IT_MAX_WORD;
+ for (j = 0; j<sizez; j++)
+ termz[j+i] = index_char_cvt (term->u.general->buf[j]);
+ termz[j+i] = '\0';
+
+ dict_scan (zi->wordDict, termz, &before, &after, scan_handle);
+
+ if (zi->errCode)
+ logf (LOG_DEBUG, "scan error: %d", zi->errCode);
+ return 0;
+}
+
projects / idzebra-moved-to-github.git / blobdiff