* Sebastian Hammer, Adam Dickmeiss
*
* $Log: zrpn.c,v $
- * Revision 1.13 1995-09-18 14:17:50 adam
+ * 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
+ * expressons in the search engine.
+ *
+ * Revision 1.35 1995/11/27 09:29:00 adam
+ * Bug fixes regarding conversion to regular expressions.
+ *
+ * Revision 1.34 1995/11/16 17:00:56 adam
+ * Better logging of rpn query.
+ *
+ * Revision 1.33 1995/11/01 13:58:28 quinn
+ * Moving data1 to yaz/retrieval
+ *
+ * Revision 1.32 1995/10/27 14:00:11 adam
+ * Implemented detection of database availability.
+ *
+ * Revision 1.31 1995/10/17 18:02:10 adam
+ * New feature: databases. Implemented as prefix to words in dictionary.
+ *
+ * Revision 1.30 1995/10/16 09:32:38 adam
+ * More work on relational op.
+ *
+ * Revision 1.29 1995/10/13 16:01:49 adam
+ * Work on relations.
+ *
+ * Revision 1.28 1995/10/13 12:26:43 adam
+ * Optimization of truncation.
+ *
+ * Revision 1.27 1995/10/12 17:07:22 adam
+ * Truncation works.
+ *
+ * Revision 1.26 1995/10/12 12:40:54 adam
+ * Bug fixes in rpn_prox.
+ *
+ * Revision 1.25 1995/10/10 13:59:24 adam
+ * Function rset_open changed its wflag parameter to general flags.
+ *
+ * Revision 1.24 1995/10/09 16:18:37 adam
+ * Function dict_lookup_grep got extra client data parameter.
+ *
+ * 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
+ * New member in recRetrieveCtrl: diagnostic.
+ *
+ * Revision 1.14 1995/09/28 12:10:32 adam
+ * Bug fixes. Field prefix used in queries.
+ *
+ * Revision 1.13 1995/09/18 14:17:50 adam
* Minor changes.
*
* Revision 1.12 1995/09/15 14:45:21 adam
#include <stdio.h>
#include <assert.h>
#include <unistd.h>
+#include <ctype.h>
#include "zserver.h"
+#include "attribute.h"
#include <rsisam.h>
#include <rstemp.h>
#include <rsbool.h>
#include <rsrel.h>
-/*
- * attr_print: log attributes
- */
-static void attr_print (Z_AttributesPlusTerm *t)
+int index_word_prefix_map (char *string, oid_value attrSet, int attrUse,
+ char *basename)
{
- int of, i;
- for (of = 0; of < t->num_attributes; of++)
- {
- Z_AttributeElement *element;
- element = t->attributeList[of];
+ attent *attp;
- switch (element->which)
- {
- case Z_AttributeValue_numeric:
- logf (LOG_DEBUG, "attributeType=%d value=%d",
- *element->attributeType,
- *element->value.numeric);
- break;
- case Z_AttributeValue_complex:
- logf (LOG_DEBUG, "attributeType=%d complex",
- *element->attributeType);
- for (i = 0; i<element->value.complex->num_list; i++)
- {
- if (element->value.complex->list[i]->which ==
- Z_StringOrNumeric_string)
- logf (LOG_DEBUG, " string: '%s'",
- element->value.complex->list[i]->u.string);
- else if (element->value.complex->list[i]->which ==
- Z_StringOrNumeric_numeric)
- logf (LOG_DEBUG, " numeric: '%d'",
- *element->value.complex->list[i]->u.numeric);
- }
- break;
- default:
- assert (0);
- }
- }
+ logf (LOG_DEBUG, "oid_value attrSet = %d, attrUse = %d", attrSet, attrUse);
+ attp = att_getentbyatt (attrSet, attrUse);
+ if (!attp)
+ return -1;
+ logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal);
+ return index_word_prefix (string, attp->attset_ordinal,
+ attp->local_attribute, basename);
}
typedef struct {
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);
}
struct trunc_info {
+ int *ptr;
int *indx;
char **heap;
int heapnum;
{
int swap;
- memcpy (ti->swapbuf, ti->heap[i1], ti->keysize);
- memcpy (ti->heap[i1], ti->heap[i2], ti->keysize);
- memcpy (ti->heap[i2], ti->swapbuf, ti->keysize);
-
- swap = ti->indx[i1];
- ti->indx[i1] = ti->indx[i2];
- ti->indx[i2] = swap;
+ swap = ti->ptr[i1];
+ ti->ptr[i1] = ti->ptr[i2];
+ ti->ptr[i2] = swap;
}
static void heap_delete (struct trunc_info *ti)
{
int cur = 1, child = 2;
- assert (ti->heapnum > 0);
- memcpy (ti->heap[1], ti->heap[ti->heapnum], ti->keysize);
- ti->indx[1] = ti->indx[ti->heapnum--];
+ heap_swap (ti, 1, ti->heapnum--);
while (child <= ti->heapnum) {
if (child < ti->heapnum &&
- (*ti->cmp)(ti->heap[child], ti->heap[1+child]) > 0)
+ (*ti->cmp)(ti->heap[ti->ptr[child]],
+ ti->heap[ti->ptr[1+child]]) > 0)
child++;
- if ((*ti->cmp)(ti->heap[cur], ti->heap[child]) > 0)
+ if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
+ ti->heap[ti->ptr[child]]) > 0)
{
heap_swap (ti, cur, child);
cur = child;
int cur, parent;
cur = ++(ti->heapnum);
- memcpy (ti->heap[cur], buf, ti->keysize);
- ti->indx[cur] = indx;
+ memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize);
+ ti->indx[ti->ptr[cur]] = indx;
parent = cur/2;
- while (parent && (*ti->cmp)(ti->heap[parent], ti->heap[cur]) > 0)
+ while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
+ ti->heap[ti->ptr[cur]]) > 0)
{
heap_swap (ti, cur, parent);
cur = parent;
ti->cmp = cmp;
ti->indx = xmalloc (size * sizeof(*ti->indx));
ti->heap = xmalloc (size * sizeof(*ti->heap));
+ ti->ptr = xmalloc (size * sizeof(*ti->ptr));
ti->swapbuf = xmalloc (ti->keysize);
ti->tmpbuf = xmalloc (ti->keysize);
ti->buf = xmalloc (size * ti->keysize);
for (i = size; --i >= 0; )
+ {
+ ti->ptr[i] = i;
ti->heap[i] = ti->buf + ti->keysize * i;
+ }
return ti;
}
static void heap_close (struct trunc_info *ti)
{
+ xfree (ti->ptr);
xfree (ti->indx);
xfree (ti->heap);
xfree (ti->swapbuf);
xfree (ti);
}
-static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int from, int to,
- int merge_chunk)
+static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
+ int merge_chunk)
{
- logf (LOG_DEBUG, "rset_trunc, range=%d-%d", from, to-1);
+ RSET result;
+ RSFD result_rsfd;
+ rset_temp_parms parms;
+
+ parms.key_size = sizeof(struct it_key);
+ result = rset_create (rset_kind_temp, &parms);
+ result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
+
if (to - from > merge_chunk)
{
- return NULL;
+ RSFD *rsfd;
+ RSET *rset;
+ int i, i_add = (to-from)/merge_chunk + 1;
+ struct trunc_info *ti;
+ int rscur = 0;
+ int rsmax = (to-from)/i_add + 1;
+
+ rset = xmalloc (sizeof(*rset) * rsmax);
+ rsfd = xmalloc (sizeof(*rsfd) * rsmax);
+
+ for (i = from; i < to; i += i_add)
+ {
+ if (i_add <= to - i)
+ rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add,
+ merge_chunk);
+ else
+ rset[rscur] = rset_trunc_r (isam, isam_p, i, to,
+ merge_chunk);
+ rscur++;
+ }
+ ti = heap_init (rscur, sizeof(struct it_key), key_compare);
+ for (i = rscur; --i >= 0; )
+ {
+ rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
+ if (rset_read (rset[i], rsfd[i], ti->tmpbuf))
+ heap_insert (ti, ti->tmpbuf, i);
+ else
+ {
+ rset_close (rset[i], rsfd[i]);
+ rset_delete (rset[i]);
+ }
+ }
+ while (ti->heapnum)
+ {
+ int n = ti->indx[ti->ptr[1]];
+
+ rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
+
+ while (1)
+ {
+ if (!rset_read (rset[n], rsfd[n], ti->tmpbuf))
+ {
+ heap_delete (ti);
+ rset_close (rset[n], rsfd[n]);
+ rset_delete (rset[n]);
+ break;
+ }
+ if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
+ {
+ heap_delete (ti);
+ heap_insert (ti, ti->tmpbuf, n);
+ break;
+ }
+ }
+ }
+ xfree (rset);
+ xfree (rsfd);
+ heap_close (ti);
}
else
{
ISPT *ispt;
int i;
struct trunc_info *ti;
- RSET result;
- RSFD rsfd;
- rset_temp_parms parms;
ispt = xmalloc (sizeof(*ispt) * (to-from));
- parms.key_size = sizeof (struct it_key);
- result = rset_create (rset_kind_temp, &parms);
- rsfd = rset_open (result, 1);
ti = heap_init (to-from, sizeof(struct it_key),
key_compare);
ispt[i] = is_position (isam, isam_p[from+i]);
if (is_readkey (ispt[i], ti->tmpbuf))
heap_insert (ti, ti->tmpbuf, i);
+ else
+ is_pt_free (ispt[i]);
}
while (ti->heapnum)
{
- int n = ti->indx[1];
+ int n = ti->indx[ti->ptr[1]];
- rset_write (result, rsfd, ti->heap[1]);
+ rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
+#if 0
+/* section that preserve all keys */
heap_delete (ti);
if (is_readkey (ispt[n], ti->tmpbuf))
heap_insert (ti, ti->tmpbuf, n);
+ else
+ is_pt_free (ispt[n]);
+#else
+/* section that preserve all keys with unique sysnos */
+ while (1)
+ {
+ if (!is_readkey (ispt[n], ti->tmpbuf))
+ {
+ heap_delete (ti);
+ is_pt_free (ispt[n]);
+ break;
+ }
+ if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
+ {
+ heap_delete (ti);
+ heap_insert (ti, ti->tmpbuf, n);
+ break;
+ }
+ }
+#endif
}
- for (i = to-from; --i >= 0; )
- is_pt_free (ispt[i]);
- rset_close (result, rsfd);
heap_close (ti);
xfree (ispt);
- return result;
}
+ rset_close (result, result_rsfd);
+ return result;
}
-static ISAM_P *isam_p_buf = NULL;
-static int isam_p_size = 0;
-static int isam_p_indx;
+static int isam_trunc_cmp (const void *p1, const void *p2)
+{
+ ISAM_P i1 = *(ISAM_P*) p1;
+ ISAM_P i2 = *(ISAM_P*) p2;
+ int d;
+
+ d = is_type (i1) - is_type (i2);
+ if (d)
+ return d;
+ return is_block (i1) - is_block (i2);
+}
-static void add_isam_p (const char *info)
+static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no)
{
- if (isam_p_indx == isam_p_size)
+
+ qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp);
+ return rset_trunc_r (isam, isam_p, 0, no, 100);
+}
+
+struct grep_info {
+ ISAM_P *isam_p_buf;
+ int isam_p_size;
+ int isam_p_indx;
+};
+
+static void add_isam_p (const char *info, struct grep_info *p)
+{
+ if (p->isam_p_indx == p->isam_p_size)
{
ISAM_P *new_isam_p_buf;
- isam_p_size = 2*isam_p_size + 100;
+ p->isam_p_size = 2*p->isam_p_size + 100;
new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
- isam_p_size);
- if (isam_p_buf)
+ p->isam_p_size);
+ if (p->isam_p_buf)
{
- memcpy (new_isam_p_buf, isam_p_buf,
- isam_p_indx * sizeof(*isam_p_buf));
- xfree (isam_p_buf);
+ memcpy (new_isam_p_buf, p->isam_p_buf,
+ p->isam_p_indx * sizeof(*p->isam_p_buf));
+ xfree (p->isam_p_buf);
}
- isam_p_buf = new_isam_p_buf;
+ p->isam_p_buf = new_isam_p_buf;
}
- assert (*info == sizeof(*isam_p_buf));
- memcpy (isam_p_buf + isam_p_indx, info+1, sizeof(*isam_p_buf));
- isam_p_indx++;
+ assert (*info == sizeof(*p->isam_p_buf));
+ memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
+ (p->isam_p_indx)++;
}
-static int grep_handle (Dict_char *name, const char *info)
+static int grep_handle (Dict_char *name, const char *info, void *p)
{
logf (LOG_DEBUG, "dict name: %s", name);
- add_isam_p (info);
+ add_isam_p (info, p);
return 0;
}
+static void gen_regular_rel (char *dst, int val, int islt)
+{
+ int dst_p = 1;
+ int w, d, i;
+ int pos = 0;
+ char numstr[20];
+
+ *dst = '(';
+ sprintf (numstr, "%d", val);
+ for (w = strlen(numstr); --w >= 0; pos++)
+ {
+ d = numstr[w];
+ if (pos > 0)
+ {
+ if (islt)
+ {
+ if (d == '0')
+ continue;
+ d--;
+ }
+ else
+ {
+ if (d == '9')
+ continue;
+ d++;
+ }
+ }
+
+ strcpy (dst + dst_p, numstr);
+ dst_p = strlen(dst) - pos - 1;
+
+ if (islt)
+ {
+ if (d != '0')
+ {
+ dst[dst_p++] = '[';
+ dst[dst_p++] = '0';
+ dst[dst_p++] = '-';
+ dst[dst_p++] = d;
+ dst[dst_p++] = ']';
+ }
+ else
+ dst[dst_p++] = d;
+ }
+ else
+ {
+ if (d != '9')
+ {
+ dst[dst_p++] = '[';
+ dst[dst_p++] = d;
+ dst[dst_p++] = '-';
+ dst[dst_p++] = '9';
+ dst[dst_p++] = ']';
+ }
+ else
+ dst[dst_p++] = d;
+ }
+ for (i = 0; i<pos; i++)
+ {
+ dst[dst_p++] = '[';
+ dst[dst_p++] = '0';
+ dst[dst_p++] = '-';
+ dst[dst_p++] = '9';
+ dst[dst_p++] = ']';
+ }
+ dst[dst_p++] = '|';
+ }
+ dst[dst_p] = '\0';
+ if (islt)
+ {
+ for (i=1; i<pos; i++)
+ strcat (dst, "[0-9]?");
+ }
+ else
+ {
+ for (i = 0; i <= pos; i++)
+ strcat (dst, "[0-9]");
+ strcat (dst, "[0-9]*");
+ }
+ strcat (dst, ")");
+}
+
+static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
+ const char *term_sub,
+ char *term_dict,
+ oid_value attributeSet,
+ struct grep_info *grep_info,
+ int *max_pos)
+{
+ AttrType relation;
+ int relation_value;
+ int term_value;
+ int r;
+
+ attr_init (&relation, zapt, 2);
+ relation_value = attr_find (&relation, NULL);
+ term_value = atoi (term_sub);
+
+ switch (relation_value)
+ {
+ case 1:
+ if (term_value <= 0)
+ return 1;
+ logf (LOG_DEBUG, "Relation <");
+ gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1);
+ break;
+ case 2:
+ if (term_value < 0)
+ return 1;
+ logf (LOG_DEBUG, "Relation <=");
+ gen_regular_rel (term_dict + strlen(term_dict), term_value, 1);
+ break;
+ case 4:
+ if (term_value < 0)
+ term_value = 0;
+ logf (LOG_DEBUG, "Relation >=");
+ gen_regular_rel (term_dict + strlen(term_dict), term_value, 0);
+ break;
+ case 5:
+ if (term_value < 0)
+ term_value = 0;
+ logf (LOG_DEBUG, "Relation >");
+ gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0);
+ break;
+ default:
+ return 0;
+ }
+ logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
+ r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos,
+ 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);
+ return 1;
+}
+
+static void verbatim_char (int ch, int *indx, char *dst)
+{
+ if (!isalnum (ch))
+ dst[(*indx)++] = '\\';
+ dst[(*indx)++] = ch;
+}
+
static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
- const char *term_sub, ISAM_P **isam_ps)
+ const char *term_sub,
+ oid_value attributeSet, struct grep_info *grep_info,
+ int num_bases, char **basenames)
{
char term_dict[2*IT_MAX_WORD+2];
- int i, j;
- const char *info;
+ int i, j, r, base_no;
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);
- switch (truncation_value)
+
+ if (use_value == -1)
+ use_value = 1016;
+
+ for (base_no = 0; base_no < num_bases; base_no++)
{
- case -1: /* not specified */
- case 100: /* do not truncate */
- strcpy (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, ".*");
- dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle);
- break;
- case 2: /* left truncation */
- case 3: /* left&right truncation */
- zi->errCode = 120;
- return -1;
- case 101: /* process # in term */
- for (j = 0, i = 0; term_sub[i] && i < 2; i++)
- term_dict[j++] = term_sub[i];
- for (; term_sub[i]; i++)
- if (term_sub[i] == '#')
+ int max_pos;
+ int prefix_len = index_word_prefix_map (term_dict, curAttributeSet,
+ use_value,
+ basenames[base_no]);
+ if (prefix_len < 0)
+ {
+ zi->errCode = 114;
+ return -1;
+ }
+ if (!relational_term (zi, zapt, term_sub, term_dict,
+ attributeSet, grep_info, &max_pos))
+ {
+ switch (truncation_value)
{
- term_dict[j++] = '.';
- term_dict[j++] = '*';
+ 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);
+ 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);
+ 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)
+ {
+ 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, 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);
+ r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
+ &max_pos, grep_handle);
+ if (r)
+ logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
+ r);
+ break;
}
- else
- term_dict[j++] = term_sub[i];
- term_dict[j] = '\0';
- dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle);
- break;
- case 102: /* regular expression */
- strcpy (term_dict, term_sub);
- dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle);
- break;
+ }
+ if (max_pos <= strlen(basenames[base_no]))
+ {
+ zi->errCode = 109; /* Database unavailable */
+ zi->errString = basenames[base_no];
+ return -1;
+ }
}
- *isam_ps = isam_p_buf;
- logf (LOG_DEBUG, "%d positions", isam_p_indx);
+ logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
return 0;
}
+static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
+ 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;
+ 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,
+ int num_bases, char **basenames)
{
rset_relevance_parms parms;
char termz[IT_MAX_WORD+1];
char term_sub[IT_MAX_WORD+1];
+ struct grep_info grep_info;
char *p0 = termz, *p1 = NULL;
- Z_Term *term = zapt->term;
- size_t sizez, i;
+ RSET result;
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;
}
- sizez = term->u.general->len;
- if (sizez > IT_MAX_WORD)
- sizez = IT_MAX_WORD;
- for (i = 0; i<sizez; i++)
- termz[i] = index_char_cvt (term->u.general->buf[i]);
- termz[i] = '\0';
-
- isam_p_indx = 0; /* global, set by trunc_term - see below */
+ trans_term (zi, zapt, termz);
+ grep_info.isam_p_indx = 0;
+ grep_info.isam_p_size = 0;
+ grep_info.isam_p_buf = NULL;
while (1)
{
if ((p1 = strchr (p0, ' ')))
}
else
strcpy (term_sub, p0);
- if (trunc_term (zi, zapt, term_sub, &parms.isam_positions))
+ if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
+ num_bases, basenames))
return NULL;
if (!p1)
break;
- p0 = p1+1;
+ p0 = p1;
+ while (*++p0 == ' ')
+ ;
}
- parms.no_isam_positions = isam_p_indx;
- if (isam_p_indx > 0)
- return rset_create (rset_kind_relevance, &parms);
+ parms.isam_positions = grep_info.isam_p_buf;
+ parms.no_isam_positions = grep_info.isam_p_indx;
+ if (grep_info.isam_p_indx > 0)
+ result = rset_create (rset_kind_relevance, &parms);
else
- return rset_create (rset_kind_null, NULL);
+ result = rset_create (rset_kind_null, NULL);
+ xfree (grep_info.isam_p_buf);
+ return result;
}
static RSET rpn_search_APT_word (ZServerInfo *zi,
- Z_AttributesPlusTerm *zapt)
+ Z_AttributesPlusTerm *zapt,
+ oid_value attributeSet,
+ int num_bases, char **basenames)
{
- ISAM_P *isam_positions;
rset_isam_parms parms;
-
char termz[IT_MAX_WORD+1];
- Z_Term *term = zapt->term;
- size_t sizez, i;
+ struct grep_info grep_info;
+ RSET result;
- if (term->which != Z_Term_general)
+ if (zapt->term->which != Z_Term_general)
{
zi->errCode = 124;
return NULL;
}
- sizez = term->u.general->len;
- if (sizez > IT_MAX_WORD)
- sizez = IT_MAX_WORD;
- for (i = 0; i<sizez; i++)
- termz[i] = index_char_cvt (term->u.general->buf[i]);
- termz[i] = '\0';
+ trans_term (zi, zapt, termz);
+
+ grep_info.isam_p_indx = 0;
+ grep_info.isam_p_size = 0;
+ grep_info.isam_p_buf = NULL;
- 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, attributeSet, &grep_info,
+ num_bases, basenames))
return NULL;
- if (isam_p_indx < 1)
- return rset_create (rset_kind_null, NULL);
- else if (isam_p_indx == 1)
+ 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 = *isam_positions;
- return rset_create (rset_kind_isam, &parms);
+ parms.pos = *grep_info.isam_p_buf;
+ result = rset_create (rset_kind_isam, &parms);
}
else
- return rset_trunc (zi->wordIsam, isam_positions, 0, isam_p_indx, 400);
+ 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_phrase (ZServerInfo *zi,
- Z_AttributesPlusTerm *zapt)
+static RSET rpn_prox (RSET *rset, int rset_no)
{
- ISAM_P *isam_positions;
- rset_isam_parms parms;
+ int i;
+ RSFD *rsfd;
+ int *more;
+ struct it_key **buf;
+ RSFD rsfd_result;
+ RSET result;
+ rset_temp_parms parms;
+
+ rsfd = xmalloc (sizeof(*rsfd)*rset_no);
+ more = xmalloc (sizeof(*more)*rset_no);
+ buf = xmalloc (sizeof(*buf)*rset_no);
+ for (i = 0; i<rset_no; i++)
+ {
+ buf[i] = xmalloc (sizeof(**buf));
+ rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
+ if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i])))
+ {
+ while (i >= 0)
+ {
+ rset_close (rset[i], rsfd[i]);
+ xfree (buf[i]);
+ --i;
+ }
+ xfree (rsfd);
+ xfree (more);
+ xfree (buf);
+ return rset_create (rset_kind_null, NULL);
+ }
+ }
+ parms.key_size = sizeof (struct it_key);
+ result = rset_create (rset_kind_temp, &parms);
+ rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
+
+ while (*more)
+ {
+ for (i = 1; i<rset_no; i++)
+ {
+ int cmp;
+
+ if (!more[i])
+ {
+ *more = 0;
+ break;
+ }
+ cmp = key_compare (buf[i], buf[i-1]);
+ if (cmp > 1)
+ {
+ more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
+ break;
+ }
+ else if (cmp == 1)
+ {
+ if (buf[i-1]->seqno+1 != buf[i]->seqno)
+ {
+ more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
+ break;
+ }
+ }
+ else
+ {
+ more[i] = rset_read (rset[i], rsfd[i], buf[i]);
+ break;
+ }
+ }
+ if (i == rset_no)
+ {
+ rset_write (result, rsfd_result, buf[0]);
+ more[0] = rset_read (*rset, *rsfd, *buf);
+ }
+ }
+
+ for (i = 0; i<rset_no; i++)
+ {
+ rset_close (rset[i], rsfd[i]);
+ xfree (buf[i]);
+ }
+ rset_close (result, rsfd_result);
+ xfree (buf);
+ xfree (more);
+ xfree (rsfd);
+ return result;
+}
+
+static RSET rpn_search_APT_phrase (ZServerInfo *zi,
+ Z_AttributesPlusTerm *zapt,
+ oid_value attributeSet,
+ int num_bases, char **basenames)
+{
char termz[IT_MAX_WORD+1];
- Z_Term *term = zapt->term;
- size_t sizez, i;
+ char term_sub[IT_MAX_WORD+1];
+ char *p0 = termz, *p1 = NULL;
+ RSET rset[60], result;
+ int i, rset_no = 0;
+ struct grep_info grep_info;
- if (term->which != Z_Term_general)
+ if (zapt->term->which != Z_Term_general)
{
zi->errCode = 124;
return NULL;
}
- sizez = term->u.general->len;
- if (sizez > IT_MAX_WORD)
- sizez = IT_MAX_WORD;
- for (i = 0; i<sizez; i++)
- termz[i] = index_char_cvt (term->u.general->buf[i]);
- termz[i] = '\0';
+ trans_term (zi, zapt, termz);
- isam_p_indx = 0; /* global, set by trunc_term - see below */
- if (trunc_term (zi, zapt, termz, &isam_positions))
- return NULL;
- if (isam_p_indx != 1)
+ grep_info.isam_p_size = 0;
+ grep_info.isam_p_buf = NULL;
+
+ while (1)
+ {
+ if ((p1 = strchr (p0, ' ')))
+ {
+ memcpy (term_sub, p0, p1-p0);
+ term_sub[p1-p0] = '\0';
+ }
+ else
+ strcpy (term_sub, p0);
+
+ grep_info.isam_p_indx = 0;
+ if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
+ num_bases, basenames))
+ return NULL;
+ if (grep_info.isam_p_indx == 0)
+ rset[rset_no] = rset_create (rset_kind_null, NULL);
+ else if (grep_info.isam_p_indx > 1)
+ rset[rset_no] = rset_trunc (zi->wordIsam,
+ grep_info.isam_p_buf,
+ grep_info.isam_p_indx);
+ else
+ {
+ rset_isam_parms parms;
+
+ parms.is = zi->wordIsam;
+ parms.pos = *grep_info.isam_p_buf;
+ rset[rset_no] = rset_create (rset_kind_isam, &parms);
+ }
+ assert (rset[rset_no]);
+ if (++rset_no >= sizeof(rset)/sizeof(*rset))
+ break;
+ if (!p1)
+ break;
+ p0 = p1;
+ while (*++p0 == ' ')
+ ;
+ }
+ xfree (grep_info.isam_p_buf);
+ if (rset_no == 0)
return rset_create (rset_kind_null, NULL);
- parms.is = zi->wordIsam;
- parms.pos = *isam_positions;
- return rset_create (rset_kind_isam, &parms);
+ else if (rset_no == 1)
+ return (rset[0]);
+
+ result = rpn_prox (rset, rset_no);
+ for (i = 0; i<rset_no; i++)
+ rset_delete (rset[i]);
+ return result;
}
-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, RSETF_WRITE|RSETF_SORT_SYSNO);
+
+ 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,
+ int num_bases, char **basenames)
{
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,
+ 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);
- return rpn_search_APT_phrase (zi, zapt);
+ return rpn_search_APT_relevance (zi, zapt, attributeSet,
+ num_bases, basenames);
+ return rpn_search_APT_phrase (zi, zapt, attributeSet,
+ num_bases, basenames);
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,
+ num_bases, basenames);
+ return rpn_search_APT_word (zi, zapt, attributeSet,
+ num_bases, basenames);
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,
+ num_bases, basenames);
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,
+ num_bases, basenames);
case 106: /* document-text */
- return rpn_search_APT_relevance (zi, zapt);
+ return rpn_search_APT_relevance (zi, zapt, attributeSet,
+ num_bases, basenames);
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,
+ num_bases, basenames);
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,
+ int num_bases, char **basenames)
{
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,
+ num_bases, basenames);
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,
+ num_bases, basenames);
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, num_bases, basenames);
}
else if (zs->u.simple->which == Z_Operand_resultSetId)
{
logf (LOG_DEBUG, "rpn_save_set");
*count = 0;
- rfd = rset_open (r, 0);
+ rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO);
while (rset_read (r, rfd, &key))
{
if (key.sysno != psysno)
const char *setname, int *hits)
{
RSET rset;
+ oident *attrset;
+ oid_value attributeSet;
+
+ zlog_rpn (rpn);
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,
+ num_bases, basenames);
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;
}
+struct scan_info {
+ struct scan_entry *list;
+ ODR odr;
+ int before, after;
+ ISAM isam;
+ char prefix[20];
+};
+
+static int scan_handle (Dict_char *name, const char *info, int pos,
+ void *client)
+{
+ int len_prefix, idx;
+ ISAM_P isam_p;
+ RSET rset;
+ struct scan_info *scan_info = client;
+
+ rset_isam_parms parms;
+
+ len_prefix = strlen(scan_info->prefix);
+ if (memcmp (name, scan_info->prefix, len_prefix))
+ return 1;
+ if (pos > 0)
+ idx = scan_info->after - pos + scan_info->before;
+ else
+ idx = - pos - 1;
+ scan_info->list[idx].term = 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 (&isam_p, info+1, sizeof(isam_p));
+ parms.is = scan_info->isam;
+ parms.pos = isam_p;
+#if 1
+ rset = rset_create (rset_kind_isam, &parms);
+ count_set (rset, &scan_info->list[idx].occurrences);
+ rset_delete (rset);
+#else
+ scan_info->list[idx].occurrences = 1;
+#endif
+ logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name);
+ return 0;
+}
+
+
+static int dummy_handle (Dict_char *name, const char *info, void *p)
+{
+ return 0;
+}
+
+int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
+ int num_bases, char **basenames,
+ int *position, int *num_entries, struct scan_entry **list,
+ int *status)
+{
+ int i, j, sizez, max_pos;
+ 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;
+ struct scan_info scan_info;
+
+ logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
+
+ if (num_bases != 1)
+ return 111;
+ scan_info.before = before = pos-1;
+ scan_info.after = after = 1+num-pos;
+ scan_info.odr = zi->odr;
+
+ logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after);
+
+ scan_info.isam = zi->wordIsam;
+ scan_info.list = odr_malloc (zi->odr, (before+after)*
+ sizeof(*scan_info.list));
+ for (j = 0; j<before+after; j++)
+ scan_info.list[j].term = NULL;
+ 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, *basenames);
+
+ dict_lookup_grep (zi->wordDict, termz, 0, NULL, &max_pos,
+ dummy_handle);
+ if (max_pos <= strlen(*basenames))
+ {
+ zi->errString = *basenames;
+ return zi->errCode = 109; /* Database unavailable */
+ }
+ strcpy (scan_info.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_info, scan_handle);
+
+ *status = BEND_SCAN_SUCCESS;
+
+ for (i = 0; i<scan_info.after; i++)
+ if (scan_info.list[scan_info.before+scan_info.after-i-1].term)
+ break;
+ *num_entries -= i;
+ if (i)
+ *status = BEND_SCAN_PARTIAL;
+
+ for (i = 0; i<scan_info.before; i++)
+ if (scan_info.list[i].term)
+ break;
+ if (i)
+ *status = BEND_SCAN_PARTIAL;
+ *position -= i;
+ *num_entries -= i;
+
+ *list = scan_info.list+i; /* list is set to first 'real' entry */
+
+ if (*num_entries == 0) /* signal 'unsupported use-attribute' */
+ zi->errCode = 114; /* if no entries was found */
+ logf (LOG_DEBUG, "position = %d, num_entries = %d",
+ *position, *num_entries);
+ if (zi->errCode)
+ logf (LOG_DEBUG, "scan error: %d", zi->errCode);
+ return 0;
+}
+
+
+