* Sebastian Hammer, Adam Dickmeiss
*
* $Log: zrpn.c,v $
- * Revision 1.36 1995-12-06 12:41:27 adam
+ * Revision 1.44 1996-05-14 06:16:44 adam
+ * Compact use/set bytes used in search service.
+ *
+ * Revision 1.43 1996/05/09 09:54:43 adam
+ * Server supports maps from one logical attributes to a list of physical
+ * attributes.
+ * The extraction process doesn't make space consuming 'any' keys.
+ *
+ * Revision 1.42 1996/05/09 07:28:56 quinn
+ * Work towards phrases and multiple registers
+ *
+ * Revision 1.41 1996/03/20 09:36:43 adam
+ * Function dict_lookup_grep got extra parameter, init_pos, which marks
+ * from which position in pattern approximate pattern matching should occur.
+ * Approximate pattern matching is used in relevance=re-2.
+ *
+ * Revision 1.40 1996/02/02 13:44:44 adam
+ * The public dictionary functions simply use char instead of Dict_char
+ * to represent search strings. Dict_char is used internally only.
+ *
+ * Revision 1.39 1996/01/03 16:22:13 quinn
+ * operator->roperator
+ *
+ * Revision 1.38 1995/12/11 09:12:55 adam
+ * The rec_get function returns NULL if record doesn't exist - will
+ * happen in the server if the result set records have been deleted since
+ * the creation of the set (i.e. the search).
+ * The server saves a result temporarily if it is 'volatile', i.e. the
+ * set is register dependent.
+ *
+ * Revision 1.37 1995/12/06 15:05:28 adam
+ * More verbose in count_set.
+ *
+ * Revision 1.36 1995/12/06 12:41:27 adam
* New command 'stat' for the index program.
* Filenames can be read from stdin by specifying '-'.
* Bug fix/enhancement of the transformation from terms to regular
return -1;
logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal);
return index_word_prefix (string, attp->attset_ordinal,
- attp->local_attribute, basename);
+ attp->local_attributes->local, basename);
}
typedef struct {
(p->isam_p_indx)++;
}
-static int grep_handle (Dict_char *name, const char *info, void *p)
+static int grep_handle (char *name, const char *info, void *p)
{
logf (LOG_DEBUG, "dict name: %s", name);
add_isam_p (info, p);
}
logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos,
- grep_handle);
+ 0, grep_handle);
if (r)
logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
dst[(*indx)++] = ch;
}
-static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
+static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
const char *term_sub,
oid_value attributeSet, struct grep_info *grep_info,
int num_bases, char **basenames)
for (base_no = 0; base_no < num_bases; base_no++)
{
+#if 1
+ attent *attp;
+ data1_local_attribute *local_attr;
+ int max_pos, prefix_len = 0;
+
+ attp = att_getentbyatt (curAttributeSet, use_value);
+ if (!attp)
+ {
+ zi->errCode = 114;
+ return -1;
+ }
+ if (zebTargetInfo_curDatabase (zi->zti, basenames[base_no]))
+ {
+ zi->errCode = 109; /* Database unavailable */
+ zi->errString = basenames[base_no];
+ }
+ for (local_attr = attp->local_attributes; local_attr;
+ local_attr = local_attr->next)
+ {
+ int ord;
+
+ ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal,
+ local_attr->local);
+ if (ord < 0)
+ continue;
+ if (prefix_len)
+ term_dict[prefix_len++] = '|';
+ else
+ term_dict[prefix_len++] = '(';
+ if ((ord >= 'A' && ord <= 'Z') || (ord >= 'a' && ord <= 'z'))
+ term_dict[prefix_len++] = ord;
+ else
+ {
+ term_dict[prefix_len++] = '\\';
+ term_dict[prefix_len++] = ord;
+ }
+ }
+ if (!prefix_len)
+ {
+ zi->errCode = 114;
+ return -1;
+ }
+ term_dict[prefix_len++] = ')';
+ term_dict[prefix_len] = '\0';
+ if (!relational_term (zi, zapt, term_sub, term_dict,
+ attributeSet, grep_info, &max_pos))
+ {
+ const char *cp;
+
+ j = prefix_len;
+ switch (truncation_value)
+ {
+ case -1: /* not specified */
+ case 100: /* do not truncate */
+ term_dict[j++] = '(';
+ for (i = 0; term_sub[i]; i++)
+ verbatim_char (term_sub[i], &j, term_dict);
+ strcpy (term_dict+j, ")");
+ r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
+ &max_pos, 0, grep_handle);
+ if (r)
+ logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
+ break;
+ case 1: /* right truncation */
+ term_dict[j++] = '(';
+ for (i = 0; term_sub[i]; i++)
+ verbatim_char (term_sub[i], &j, term_dict);
+ strcpy (term_dict+j, ".*)");
+ dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
+ &max_pos, 0, grep_handle);
+ break;
+ case 2: /* left truncation */
+ case 3: /* left&right truncation */
+ zi->errCode = 120;
+ return -1;
+ case 101: /* process # in term */
+ term_dict[j++] = '(';
+ for (i=0; term_sub[i]; i++)
+ if (term_sub[i] == '#' && i > 2)
+ {
+ term_dict[j++] = '.';
+ term_dict[j++] = '*';
+ }
+ else
+ verbatim_char (term_sub[i], &j, term_dict);
+ strcpy (term_dict+j, ")");
+ r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
+ &max_pos, 0, grep_handle);
+ if (r)
+ logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
+ r);
+ break;
+ case 102: /* regular expression */
+ sprintf (term_dict + j, "(%s)", term_sub);
+ r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
+ &max_pos, 0, grep_handle);
+ if (r)
+ logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
+ r);
+ break;
+ case 103: /* regular expression with error correction */
+ cp = term_sub;
+ r = 0;
+ if (*cp == '*' && cp[1] && cp[2])
+ {
+ r = atoi (cp+1);
+ cp += 2;
+ }
+ sprintf (term_dict + j, "(%s)", cp);
+ r = dict_lookup_grep (zi->wordDict, term_dict, r, grep_info,
+ &max_pos, j, grep_handle);
+ if (r)
+ logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d",
+ r);
+ break;
+ }
+ }
+#else
int max_pos;
+#if 1
+ attent *attp;
+ data1_local_attribute *local_attr;
+ int prefix_len;
+
+ attp = att_getentbyatt (curAttributeSet, use_value);
+ if (!attp)
+ {
+ zi->errCode = 114;
+ return -1;
+ }
+ for (local_attr = attp->local_attributes; local_attr;
+ local_attr = local_attr->next)
+ {
+ prefix_len = index_word_prefix (term_dict, attp->attset_ordinal,
+ local_attr->local,
+ basenames[base_no]);
+
+ if (!relational_term (zi, zapt, term_sub, term_dict,
+ attributeSet, grep_info, &max_pos))
+ {
+ const char *cp;
+
+ j = prefix_len;
+ switch (truncation_value)
+ {
+ case -1: /* not specified */
+ case 100: /* do not truncate */
+ term_dict[j++] = '(';
+ for (i = 0; term_sub[i]; i++)
+ verbatim_char (term_sub[i], &j, term_dict);
+ strcpy (term_dict+j, ")");
+ r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
+ &max_pos, 0, grep_handle);
+ if (r)
+ logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
+ break;
+ case 1: /* right truncation */
+ term_dict[j++] = '(';
+ for (i = 0; term_sub[i]; i++)
+ verbatim_char (term_sub[i], &j, term_dict);
+ strcpy (term_dict+j, ".*)");
+ dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
+ &max_pos, 0, grep_handle);
+ break;
+ case 2: /* left truncation */
+ case 3: /* left&right truncation */
+ zi->errCode = 120;
+ return -1;
+ case 101: /* process # in term */
+ term_dict[j++] = '(';
+ for (i=0; term_sub[i]; i++)
+ if (term_sub[i] == '#' && i > 2)
+ {
+ term_dict[j++] = '.';
+ term_dict[j++] = '*';
+ }
+ else
+ verbatim_char (term_sub[i], &j, term_dict);
+ strcpy (term_dict+j, ")");
+ r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
+ &max_pos, 0, grep_handle);
+ if (r)
+ logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
+ r);
+ break;
+ case 102: /* regular expression */
+ sprintf (term_dict + j, "(%s)", term_sub);
+ r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
+ &max_pos, 0, grep_handle);
+ if (r)
+ logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
+ r);
+ break;
+ case 103: /* regular expression with error correction */
+ cp = term_sub;
+ r = 0;
+ if (*cp == '*' && cp[1] && cp[2])
+ {
+ r = atoi (cp+1);
+ cp += 2;
+ }
+ sprintf (term_dict + j, "(%s)", cp);
+ r = dict_lookup_grep (zi->wordDict, term_dict, r, grep_info,
+ &max_pos, j, grep_handle);
+ if (r)
+ logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d",
+ r);
+ break;
+ }
+ }
+ if (max_pos <= strlen(basenames[base_no]))
+ {
+ zi->errCode = 109; /* Database unavailable */
+ zi->errString = basenames[base_no];
+ return -1;
+ }
+ }
+#else
int prefix_len = index_word_prefix_map (term_dict, curAttributeSet,
use_value,
basenames[base_no]);
if (!relational_term (zi, zapt, term_sub, term_dict,
attributeSet, grep_info, &max_pos))
{
+ const char *cp;
+
+ j = prefix_len;
switch (truncation_value)
{
case -1: /* not specified */
case 100: /* do not truncate */
- j = strlen(term_dict);
term_dict[j++] = '(';
for (i = 0; term_sub[i]; i++)
verbatim_char (term_sub[i], &j, term_dict);
strcpy (term_dict+j, ")");
r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
- &max_pos, grep_handle);
+ &max_pos, 0, grep_handle);
if (r)
logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
break;
case 1: /* right truncation */
- j = strlen(term_dict);
term_dict[j++] = '(';
for (i = 0; term_sub[i]; i++)
verbatim_char (term_sub[i], &j, term_dict);
strcpy (term_dict+j, ".*)");
dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
- &max_pos, grep_handle);
+ &max_pos, 0, grep_handle);
break;
case 2: /* left truncation */
case 3: /* left&right truncation */
zi->errCode = 120;
return -1;
case 101: /* process # in term */
- j = strlen(term_dict);
term_dict[j++] = '(';
for (i=0; term_sub[i]; i++)
if (term_sub[i] == '#' && i > 2)
verbatim_char (term_sub[i], &j, term_dict);
strcpy (term_dict+j, ")");
r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
- &max_pos, grep_handle);
+ &max_pos, 0, grep_handle);
if (r)
logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
r);
break;
case 102: /* regular expression */
- sprintf (term_dict + strlen(term_dict), "(%s)", term_sub);
+ sprintf (term_dict + j, "(%s)", term_sub);
r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
- &max_pos, grep_handle);
+ &max_pos, 0, grep_handle);
if (r)
logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
r);
break;
+ case 103: /* regular expression with error correction */
+ cp = term_sub;
+ r = 0;
+ if (*cp == '*' && cp[1] && cp[2])
+ {
+ r = atoi (cp+1);
+ cp += 2;
+ }
+ sprintf (term_dict + j, "(%s)", cp);
+ r = dict_lookup_grep (zi->wordDict, term_dict, r, grep_info,
+ &max_pos, j, grep_handle);
+ if (r)
+ logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d",
+ r);
+ break;
}
}
if (max_pos <= strlen(basenames[base_no]))
zi->errString = basenames[base_no];
return -1;
}
+#endif
+#endif
}
logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
return 0;
}
static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
- char *termz)
+ int regType, char *termz)
{
size_t i, sizez;
Z_Term *term = zapt->term;
sizez = term->u.general->len;
- if (sizez > IT_MAX_WORD)
- sizez = IT_MAX_WORD;
+ if (sizez > IT_MAX_WORD-1)
+ sizez = IT_MAX_WORD-1;
+ termz[0] = regType;
for (i = 0; i < sizez; i++)
- termz[i] = index_char_cvt (term->u.general->buf[i]);
- termz[i] = '\0';
+ termz[i+1] = index_char_cvt (term->u.general->buf[i]);
+ termz[i+1] = '\0';
}
static RSET rpn_search_APT_relevance (ZServerInfo *zi,
zi->errCode = 124;
return NULL;
}
- trans_term (zi, zapt, termz);
+ trans_term (zi, zapt, 'w', termz);
+
grep_info.isam_p_indx = 0;
grep_info.isam_p_size = 0;
grep_info.isam_p_buf = NULL;
}
else
strcpy (term_sub, p0);
- if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
+ if (field_term (zi, zapt, term_sub, attributeSet, &grep_info,
num_bases, basenames))
return NULL;
if (!p1)
return result;
}
+static RSET rpn_search_APT_cphrase (ZServerInfo *zi,
+ Z_AttributesPlusTerm *zapt,
+ oid_value attributeSet,
+ int num_bases, char **basenames)
+{
+ rset_isam_parms parms;
+ char termz[IT_MAX_WORD+1];
+ struct grep_info grep_info;
+ RSET result;
+
+ if (zapt->term->which != Z_Term_general)
+ {
+ zi->errCode = 124;
+ return NULL;
+ }
+ trans_term (zi, zapt, 'p', termz);
+
+ grep_info.isam_p_indx = 0;
+ grep_info.isam_p_size = 0;
+ grep_info.isam_p_buf = NULL;
+
+ if (field_term (zi, zapt, termz, attributeSet, &grep_info,
+ num_bases, basenames))
+ return NULL;
+ if (grep_info.isam_p_indx < 1)
+ result = rset_create (rset_kind_null, NULL);
+ else if (grep_info.isam_p_indx == 1)
+ {
+ parms.is = zi->wordIsam;
+ parms.pos = *grep_info.isam_p_buf;
+ result = rset_create (rset_kind_isam, &parms);
+ }
+ else
+ result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf,
+ grep_info.isam_p_indx);
+ xfree (grep_info.isam_p_buf);
+ return result;
+}
static RSET rpn_search_APT_word (ZServerInfo *zi,
Z_AttributesPlusTerm *zapt,
oid_value attributeSet,
zi->errCode = 124;
return NULL;
}
- trans_term (zi, zapt, termz);
+ trans_term (zi, zapt, 'w', termz);
grep_info.isam_p_indx = 0;
grep_info.isam_p_size = 0;
grep_info.isam_p_buf = NULL;
- if (trunc_term (zi, zapt, termz, attributeSet, &grep_info,
+ if (field_term (zi, zapt, termz, attributeSet, &grep_info,
num_bases, basenames))
return NULL;
if (grep_info.isam_p_indx < 1)
zi->errCode = 124;
return NULL;
}
- trans_term (zi, zapt, termz);
+ trans_term (zi, zapt, 'w', termz);
grep_info.isam_p_size = 0;
grep_info.isam_p_buf = NULL;
strcpy (term_sub, p0);
grep_info.isam_p_indx = 0;
- if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
+ if (field_term (zi, zapt, term_sub, attributeSet, &grep_info,
num_bases, basenames))
return NULL;
if (grep_info.isam_p_indx == 0)
result = rset_create (rset_kind_temp, &parms);
rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
- trans_term (zi, zapt, termz);
- key.sysno = atoi (termz);
+ trans_term (zi, zapt, 'w', termz);
+
+ key.sysno = atoi (termz+1);
if (key.sysno <= 0)
key.sysno = 1;
rset_write (result, rsfd, &key);
{
AttrType relation;
AttrType structure;
- int relation_value, structure_value;
+ AttrType completeness;
+ int relation_value, structure_value, completeness_value;
attr_init (&relation, zapt, 2);
attr_init (&structure, zapt, 4);
+ attr_init (&completeness, zapt, 6);
relation_value = attr_find (&relation, NULL);
structure_value = attr_find (&structure, NULL);
+ completeness_value = attr_find (&completeness, NULL);
switch (structure_value)
{
case -1:
if (relation_value == 102) /* relevance relation */
return rpn_search_APT_relevance (zi, zapt, attributeSet,
num_bases, basenames);
+ if (completeness_value == 2 || completeness_value == 3)
+ return rpn_search_APT_cphrase (zi, zapt, attributeSet,
+ num_bases, basenames);
return rpn_search_APT_phrase (zi, zapt, attributeSet,
num_bases, basenames);
case 1: /* phrase */
if (relation_value == 102) /* relevance relation */
return rpn_search_APT_relevance (zi, zapt, attributeSet,
num_bases, basenames);
+ if (completeness_value == 2 || completeness_value == 3)
+ return rpn_search_APT_cphrase (zi, zapt, attributeSet,
+ num_bases, basenames);
return rpn_search_APT_phrase (zi, zapt, attributeSet,
num_bases, basenames);
break;
if (zs->which == Z_RPNStructure_complex)
{
rset_bool_parms bool_parms;
+ int soft = 0;
bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
attributeSet,
num_bases, basenames);
if (bool_parms.rset_l == NULL)
return NULL;
+ if (rset_is_ranked(bool_parms.rset_l))
+ soft = 1;
bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2,
attributeSet,
num_bases, basenames);
rset_delete (bool_parms.rset_l);
return NULL;
}
+ if (rset_is_ranked(bool_parms.rset_r))
+ soft = 1;
bool_parms.key_size = sizeof(struct it_key);
bool_parms.cmp = key_compare;
- switch (zs->u.complex->operator->which)
+ switch (zs->u.complex->roperator->which)
{
case Z_Operator_and:
- r = rset_create (rset_kind_and, &bool_parms);
+ r = rset_create (soft ? rset_kind_sand:rset_kind_and, &bool_parms);
break;
case Z_Operator_or:
- r = rset_create (rset_kind_or, &bool_parms);
+ r = rset_create (soft ? rset_kind_sor:rset_kind_or, &bool_parms);
break;
case Z_Operator_and_not:
- r = rset_create (rset_kind_not, &bool_parms);
+ r = rset_create (soft ? rset_kind_snot:rset_kind_not, &bool_parms);
break;
default:
assert (0);
return r;
}
+void count_set_save (RSET *r, int *count)
+{
+ int psysno = 0;
+ int kno = 0;
+ struct it_key key;
+ RSFD rfd, wfd;
+ RSET w;
+ rset_temp_parms parms;
+
+ logf (LOG_DEBUG, "count_set_save");
+ *count = 0;
+ parms.key_size = sizeof(struct it_key);
+ w = rset_create (rset_kind_temp, &parms);
+ wfd = rset_open (w, RSETF_WRITE|RSETF_SORT_SYSNO);
+ rfd = rset_open (*r, RSETF_READ|RSETF_SORT_SYSNO);
+ while (rset_read (*r, rfd, &key))
+ {
+ if (key.sysno != psysno)
+ {
+ rset_write (w, wfd, &key);
+ psysno = key.sysno;
+ (*count)++;
+ }
+ kno++;
+ }
+ rset_close (*r, rfd);
+ rset_delete (*r);
+ rset_close (w, wfd);
+ *r = w;
+ logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count);
+}
+
static void count_set (RSET r, int *count)
{
int psysno = 0;
+ int kno = 0;
struct it_key key;
RSFD rfd;
- logf (LOG_DEBUG, "rpn_save_set");
+ logf (LOG_DEBUG, "count_set");
*count = 0;
rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO);
while (rset_read (r, rfd, &key))
psysno = key.sysno;
(*count)++;
}
+ kno++;
}
rset_close (r, rfd);
- logf (LOG_DEBUG, "%d distinct sysnos", *count);
+ logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count);
}
int rpn_search (ZServerInfo *zi,
num_bases, basenames);
if (!rset)
return zi->errCode;
- count_set (rset, hits);
+ if (rset_is_volatile(rset))
+ count_set_save(&rset,hits);
+ else
+ count_set (rset, hits);
resultSetAdd (zi, setname, 1, rset);
if (zi->errCode)
logf (LOG_DEBUG, "search error: %d", zi->errCode);
char prefix[20];
};
-static int scan_handle (Dict_char *name, const char *info, int pos,
- void *client)
+static int scan_handle (char *name, const char *info, int pos, void *client)
{
int len_prefix, idx;
ISAM_P isam_p;
}
-static int dummy_handle (Dict_char *name, const char *info, void *p)
+static int dummy_handle (char *name, const char *info, void *p)
{
return 0;
}
use_value = 1016;
i = index_word_prefix (termz, 1, use_value, *basenames);
- dict_lookup_grep (zi->wordDict, termz, 0, NULL, &max_pos,
+ dict_lookup_grep (zi->wordDict, termz, 0, NULL, &max_pos, 0,
dummy_handle);
if (max_pos <= strlen(*basenames))
{