/*
- * Copyright (C) 1994-1995, Index Data I/S
+ * Copyright (C) 1994-1996, Index Data I/S
* All rights reserved.
* Sebastian Hammer, Adam Dickmeiss
*
* $Log: zrpn.c,v $
- * Revision 1.36 1995-12-06 12:41:27 adam
+ * Revision 1.55 1996-11-04 14:07:44 adam
+ * Moved truncation code to trunc.c.
+ *
+ * Revision 1.54 1996/10/29 14:09:52 adam
+ * Use of cisam system - enabled if setting isamc is 1.
+ *
+ * Revision 1.53 1996/06/26 09:21:43 adam
+ * Bug fix: local attribute set wasn't obeyed in scan.
+ *
+ * Revision 1.52 1996/06/17 14:26:20 adam
+ * Function gen_regular_rel changed to handle negative numbers.
+ *
+ * Revision 1.51 1996/06/11 10:54:15 quinn
+ * Relevance work
+ *
+ * Revision 1.50 1996/06/07 08:51:53 adam
+ * Bug fix: Character mapping was broken (introducued by last revision).
+ *
+ * Revision 1.49 1996/06/04 10:18:11 adam
+ * Search/scan uses character mapping module.
+ *
+ * Revision 1.48 1996/05/28 15:15:01 adam
+ * Bug fix: Didn't handle unknown database correctly.
+ *
+ * Revision 1.47 1996/05/15 18:36:28 adam
+ * Function trans_term transforms unsearchable characters to blanks.
+ *
+ * Revision 1.46 1996/05/15 11:57:56 adam
+ * Fixed bug introduced by set/field mapping in search operations.
+ *
+ * Revision 1.45 1996/05/14 11:34:00 adam
+ * Scan support in multiple registers/databases.
+ *
+ * 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
#include "zserver.h"
#include "attribute.h"
-#include <rsisam.h>
+#include <charmap.h>
#include <rstemp.h>
#include <rsnull.h>
#include <rsbool.h>
#include <rsrel.h>
-int index_word_prefix_map (char *string, oid_value attrSet, int attrUse,
- char *basename)
-{
- attent *attp;
-
- 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 {
int type;
int major;
src->minor = 0;
}
-struct trunc_info {
- int *ptr;
- int *indx;
- char **heap;
- int heapnum;
- int (*cmp)(const void *p1, const void *p2);
- int keysize;
- char *swapbuf;
- char *tmpbuf;
- char *buf;
-};
-
-static void heap_swap (struct trunc_info *ti, int i1, int i2)
-{
- int 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;
-
- heap_swap (ti, 1, ti->heapnum--);
- while (child <= ti->heapnum) {
- if (child < ti->heapnum &&
- (*ti->cmp)(ti->heap[ti->ptr[child]],
- ti->heap[ti->ptr[1+child]]) > 0)
- child++;
- if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
- ti->heap[ti->ptr[child]]) > 0)
- {
- heap_swap (ti, cur, child);
- cur = child;
- child = 2*cur;
- }
- else
- break;
- }
-}
-
-static void heap_insert (struct trunc_info *ti, const char *buf, int indx)
-{
- int cur, parent;
-
- cur = ++(ti->heapnum);
- memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize);
- ti->indx[ti->ptr[cur]] = indx;
- parent = cur/2;
- while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
- ti->heap[ti->ptr[cur]]) > 0)
- {
- heap_swap (ti, cur, parent);
- cur = parent;
- parent = cur/2;
- }
-}
-
-static
-struct trunc_info *heap_init (int size, int key_size,
- int (*cmp)(const void *p1, const void *p2))
-{
- struct trunc_info *ti = xmalloc (sizeof(*ti));
- int i;
-
- ++size;
- ti->heapnum = 0;
- ti->keysize = key_size;
- 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->tmpbuf);
- xfree (ti);
-}
-
-static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
- int merge_chunk)
-{
- 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)
- {
- 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;
-
- ispt = xmalloc (sizeof(*ispt) * (to-from));
-
- ti = heap_init (to-from, sizeof(struct it_key),
- key_compare);
- for (i = to-from; --i >= 0; )
- {
- 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[ti->ptr[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
- }
- heap_close (ti);
- xfree (ispt);
- }
- rset_close (result, result_rsfd);
- return result;
-}
-
-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 RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no)
-{
-
- 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;
-};
+#define TERM_COUNT
+
+struct grep_info {
+#ifdef TERM_COUNT
+ int *term_no;
+#endif
+ 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;
+#ifdef TERM_COUNT
+ int *new_term_no;
+#endif
p->isam_p_size = 2*p->isam_p_size + 100;
new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
xfree (p->isam_p_buf);
}
p->isam_p_buf = new_isam_p_buf;
+
+#ifdef TERM_COUNT
+ new_term_no = xmalloc (sizeof(*new_term_no) *
+ p->isam_p_size);
+ if (p->term_no)
+ {
+ memcpy (new_term_no, p->isam_p_buf,
+ p->isam_p_indx * sizeof(*p->term_no));
+ xfree (p->term_no);
+ }
+ p->term_no = new_term_no;
+#endif
}
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, 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);
return 0;
}
+/* gen_regular_rel - generate regular expression from relation
+ * val: border value (inclusive)
+ * islt: 1 if <=; 0 if >=.
+ */
static void gen_regular_rel (char *dst, int val, int islt)
{
- int dst_p = 1;
+ int dst_p;
int w, d, i;
int pos = 0;
char numstr[20];
- *dst = '(';
+ logf (LOG_DEBUG, "gen_regular_rel. val=%d, islt=%d", val, islt);
+ if (val >= 0)
+ {
+ if (islt)
+ strcpy (dst, "(-[0-9]+|");
+ else
+ strcpy (dst, "(");
+ }
+ else
+ {
+ if (!islt)
+ {
+ strcpy (dst, "([0-9]+|-");
+ dst_p = strlen (dst);
+ islt = 1;
+ }
+ else
+ {
+ strcpy (dst, "(-");
+ islt = 0;
+ }
+ val = -val;
+ }
+ dst_p = strlen (dst);
sprintf (numstr, "%d", val);
for (w = strlen(numstr); --w >= 0; pos++)
{
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);
+ r = dict_lookup_grep (zi->dict, term_dict, 0, grep_info, max_pos,
+ 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,
- const char *term_sub,
+static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
+ const char *term_sub, int regType,
oid_value attributeSet, struct grep_info *grep_info,
int num_bases, char **basenames)
{
for (base_no = 0; base_no < num_bases; base_no++)
{
- int max_pos;
- int prefix_len = index_word_prefix_map (term_dict, curAttributeSet,
- use_value,
- basenames[base_no]);
- if (prefix_len < 0)
+ attent *attp;
+ data1_local_attribute *local_attr;
+ int max_pos, prefix_len = 0;
+
+ attp = att_getentbyatt (curAttributeSet, use_value);
+ if (!attp)
{
+ logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
+ curAttributeSet, use_value);
zi->errCode = 114;
return -1;
}
+ if (zebTargetInfo_curDatabase (zi->zti, basenames[base_no]))
+ {
+ zi->errCode = 109; /* Database unavailable */
+ zi->errString = basenames[base_no];
+ return -1;
+ }
+ 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++] = '(';
+ term_dict[prefix_len++] = 1;
+ term_dict[prefix_len++] = ord;
+ }
+ if (!prefix_len)
+ {
+ zi->errCode = 114;
+ return -1;
+ }
+ term_dict[prefix_len++] = ')';
+ term_dict[prefix_len++] = 1;
+ term_dict[prefix_len++] = regType;
+ 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 */
- 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);
+ r = dict_lookup_grep (zi->dict, 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 */
- 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);
+ dict_lookup_grep (zi->dict, 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 */
- j = strlen(term_dict);
term_dict[j++] = '(';
for (i=0; term_sub[i]; i++)
if (term_sub[i] == '#' && i > 2)
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);
+ r = dict_lookup_grep (zi->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 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);
+ sprintf (term_dict + j, "(%s)", term_sub);
+ r = dict_lookup_grep (zi->dict, 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->dict, 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;
- }
}
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;
+ size_t 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]);
+ if (sizez > IT_MAX_WORD-1)
+ sizez = IT_MAX_WORD-1;
+ memcpy (termz, term->u.general->buf, sizez);
+ termz[sizez] = '\0';
+}
+
+static void trans_scan_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
+ char *termz)
+{
+ Z_Term *term = zapt->term;
+ char **map;
+ char *cp = (char*) term->u.general->buf;
+ const char *cp_end = cp + term->u.general->len;
+ const char *src;
+ int i = 0;
+ int prev_space = 0;
+ int len;
+
+ while ((len = (cp_end - cp)) > 0)
+ {
+ map = map_chrs_input (&cp, len);
+ if (**map == *CHR_SPACE)
+ {
+ if (prev_space)
+ continue;
+ prev_space = 1;
+ }
+ else
+ prev_space = 0;
+ for (src = *map; *src; src++)
+ termz[i++] = *src;
+ }
termz[i] = '\0';
}
char termz[IT_MAX_WORD+1];
char term_sub[IT_MAX_WORD+1];
struct grep_info grep_info;
- char *p0 = termz, *p1 = NULL;
+ char *p0 = termz;
RSET result;
+ int term_index = 0;
parms.key_size = sizeof(struct it_key);
parms.max_rec = 100;
parms.cmp = key_compare;
- parms.is = zi->wordIsam;
+ parms.is = zi->isam;
+ parms.no_terms = 0;
if (zapt->term->which != Z_Term_general)
{
return NULL;
}
trans_term (zi, zapt, termz);
+
+#ifdef TERM_COUNT
+ grep_info.term_no = 0;
+#endif
grep_info.isam_p_indx = 0;
grep_info.isam_p_size = 0;
grep_info.isam_p_buf = NULL;
while (1)
{
- if ((p1 = strchr (p0, ' ')))
+ char **map;
+ char *p2, *p1;
+
+ p1 = p0;
+ while (*(p0 = p1))
{
- memcpy (term_sub, p0, p1-p0);
- term_sub[p1-p0] = '\0';
+ map = map_chrs_input (&p1, strlen(p1));
+ if (**map != *CHR_SPACE)
+ break;
}
- else
- strcpy (term_sub, p0);
- if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
+ if (!*p0)
+ break;
+
+ p1 = p0;
+ while (*(p2 = p1))
+ {
+ map = map_chrs_input (&p1, strlen(p1));
+ if (**map == *CHR_SPACE)
+ break;
+ }
+ if (p2 == p0)
+ break;
+ memcpy (term_sub, p0, p2-p0);
+ term_sub[p2-p0] = '\0';
+ p0 = p2;
+ if (field_term (zi, zapt, term_sub, 'w', attributeSet, &grep_info,
num_bases, basenames))
return NULL;
- if (!p1)
- break;
- p0 = p1;
- while (*++p0 == ' ')
- ;
+#ifdef TERM_COUNT
+ for (; term_index < grep_info.isam_p_indx; term_index++)
+ grep_info.term_no[term_index] = parms.no_terms;
+ parms.no_terms++;
+#endif
}
+ parms.term_no = grep_info.term_no;
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
result = rset_create (rset_kind_null, NULL);
+#ifdef TERM_COUNT
+ xfree(grep_info.term_no);
+#endif
xfree (grep_info.isam_p_buf);
return result;
}
-static RSET rpn_search_APT_word (ZServerInfo *zi,
- Z_AttributesPlusTerm *zapt,
- oid_value attributeSet,
- int num_bases, char **basenames)
+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;
}
trans_term (zi, zapt, termz);
+#ifdef TERM_COUNT
+ grep_info.term_no = 0;
+#endif
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, 'p', 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);
+ result = rset_trunc (zi, grep_info.isam_p_buf, grep_info.isam_p_indx);
+#ifdef TERM_COUNT
+ xfree(grep_info.term_no);
+#endif
xfree (grep_info.isam_p_buf);
return result;
}
{
char termz[IT_MAX_WORD+1];
char term_sub[IT_MAX_WORD+1];
- char *p0 = termz, *p1 = NULL;
+ char *p0 = termz;
RSET rset[60], result;
int i, rset_no = 0;
struct grep_info grep_info;
}
trans_term (zi, zapt, termz);
+#ifdef TERM_COUNT
+ grep_info.term_no = 0;
+#endif
grep_info.isam_p_size = 0;
grep_info.isam_p_buf = NULL;
while (1)
{
- if ((p1 = strchr (p0, ' ')))
+ char **map;
+ char *p2, *p1;
+
+ p1 = p0;
+ while (*(p0 = p1))
{
- memcpy (term_sub, p0, p1-p0);
- term_sub[p1-p0] = '\0';
+ map = map_chrs_input (&p1, strlen(p1));
+ if (**map != *CHR_SPACE)
+ break;
}
- else
- strcpy (term_sub, p0);
+ if (!*p0)
+ break;
+
+ p1 = p0;
+ while (*(p2 = p1))
+ {
+ map = map_chrs_input (&p1, strlen(p1));
+ if (**map == *CHR_SPACE)
+ break;
+ }
+ if (p2 == p0)
+ break;
+
+ memcpy (term_sub, p0, p2-p0);
+ term_sub[p2-p0] = '\0';
+ p0 = p2;
grep_info.isam_p_indx = 0;
- if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
+ if (field_term (zi, zapt, term_sub, 'w', 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);
- }
+ rset[rset_no] = rset_trunc (zi, grep_info.isam_p_buf,
+ grep_info.isam_p_indx);
assert (rset[rset_no]);
if (++rset_no >= sizeof(rset)/sizeof(*rset))
break;
- if (!p1)
- break;
- p0 = p1;
- while (*++p0 == ' ')
- ;
}
+#ifdef TERM_COUNT
+ xfree(grep_info.term_no);
+#endif
xfree (grep_info.isam_p_buf);
if (rset_no == 0)
return rset_create (rset_kind_null, NULL);
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;
{
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 (relation_value == 102) /* relevance relation */
return rpn_search_APT_relevance (zi, zapt, attributeSet,
num_bases, basenames);
- return rpn_search_APT_word (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 3: /* key */
break;
case 4: /* year */
case 107: /* local-number */
return rpn_search_APT_local (zi, zapt, attributeSet);
case 108: /* string */
- return rpn_search_APT_word (zi, zapt, attributeSet,
- num_bases, basenames);
+ return rpn_search_APT_phrase (zi, zapt, attributeSet,
+ num_bases, basenames);
case 109: /* numeric string */
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))
+ {
+ logf (LOG_DEBUG, "sysno=%-7d seqno=%d", key.sysno, key.seqno);
+ 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,
oident *attrset;
oid_value attributeSet;
+ dict_grep_cmap (zi->dict, map_chrs_input);
zlog_rpn (rpn);
zi->errCode = 0;
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);
return zi->errCode;
}
+struct scan_info_entry {
+ char *term;
+ ISAM_P isam_p;
+};
+
struct scan_info {
- struct scan_entry *list;
+ struct scan_info_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)
+static int scan_handle (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;
idx = scan_info->after - pos + scan_info->before;
else
idx = - pos - 1;
+ logf (LOG_DEBUG, "%-3d %s", idx, name+len_prefix);
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);
+ assert (*info == sizeof(ISAM_P));
+ memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAM_P));
return 0;
}
-static int dummy_handle (Dict_char *name, const char *info, void *p)
-{
- return 0;
+static void scan_term_untrans (ODR odr, char **dstp, const char *src)
+{
+ char *dst = odr_malloc (odr, strlen(src)*2+1);
+ *dstp = dst;
+
+ while (*src)
+ {
+ const char *cp = map_chrs_output (&src);
+ while (*cp)
+ *dst++ = *cp++;
+ }
+ *dst = '\0';
}
int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
+ oid_value attributeset,
int num_bases, char **basenames,
int *position, int *num_entries, struct scan_entry **list,
int *status)
{
- int i, j, sizez, max_pos;
+ int i;
int pos = *position;
int num = *num_entries;
int before;
int after;
+ int base_no;
char termz[IT_MAX_WORD+20];
AttrType use;
int use_value;
- Z_Term *term = zapt->term;
- struct scan_info scan_info;
+ AttrType completeness;
+ int completeness_value;
+ struct scan_info *scan_info_array;
+ struct scan_entry *glist;
+ int ords[32], ord_no = 0;
+ int ptr[32];
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;
+ if (attributeset == VAL_NONE)
+ attributeset = VAL_BIB1;
+
attr_init (&use, zapt, 1);
- use_value = attr_find (&use, NULL);
+ use_value = attr_find (&use, &attributeset);
logf (LOG_DEBUG, "use value %d", use_value);
+ attr_init (&completeness, zapt, 6);
+ completeness_value = attr_find (&completeness, NULL);
+ logf (LOG_DEBUG, "completeness value %d", completeness_value);
+
if (use_value == -1)
use_value = 1016;
- i = index_word_prefix (termz, 1, use_value, *basenames);
+ for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++)
+ {
+ attent *attp;
+ data1_local_attribute *local_attr;
+
+ attp = att_getentbyatt (attributeset, use_value);
+ if (!attp)
+ {
+ logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
+ attributeset, use_value);
+ return zi->errCode = 114;
+ }
+ if (zebTargetInfo_curDatabase (zi->zti, basenames[base_no]))
+ {
+ zi->errString = basenames[base_no];
+ return zi->errCode = 109; /* Database unavailable */
+ }
+ for (local_attr = attp->local_attributes; local_attr && ord_no < 32;
+ local_attr = local_attr->next)
+ {
+ int ord;
- dict_lookup_grep (zi->wordDict, termz, 0, NULL, &max_pos,
- dummy_handle);
- if (max_pos <= strlen(*basenames))
+ ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal,
+ local_attr->local);
+ if (ord > 0)
+ ords[ord_no++] = ord;
+ }
+ }
+ if (ord_no == 0)
+ return zi->errCode = 113;
+ before = pos-1;
+ after = 1+num-pos;
+ scan_info_array = odr_malloc (zi->odr, ord_no * sizeof(*scan_info_array));
+ for (i = 0; i < ord_no; i++)
{
- zi->errString = *basenames;
- return zi->errCode = 109; /* Database unavailable */
+ int j, prefix_len = 0;
+ int before_tmp = before, after_tmp = after;
+ struct scan_info *scan_info = scan_info_array + i;
+
+ scan_info->before = before;
+ scan_info->after = after;
+ scan_info->odr = zi->odr;
+
+ 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;
+ termz[prefix_len++] = ords[i];
+ termz[prefix_len++] =
+ (completeness_value==2 || completeness_value==3) ? 'p': 'w';
+ termz[prefix_len] = 0;
+ strcpy (scan_info->prefix, termz);
+
+ trans_scan_term (zi, zapt, termz+prefix_len);
+
+ dict_scan (zi->dict, termz, &before_tmp, &after_tmp, scan_info,
+ scan_handle);
}
- 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';
+ glist = odr_malloc (zi->odr, (before+after)*sizeof(*glist));
+ for (i = 0; i < ord_no; i++)
+ ptr[i] = before;
- 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)
+ for (i = 0; i<after; i++)
+ {
+ int j, j0 = -1;
+ const char *mterm = NULL;
+ const char *tst;
+ RSET rset;
+
+ for (j = 0; j < ord_no; j++)
+ {
+ if (ptr[j] < before+after &&
+ (tst=scan_info_array[j].list[ptr[j]].term) &&
+ (!mterm || strcmp (tst, mterm) < 0))
+ {
+ j0 = j;
+ mterm = tst;
+ }
+ }
+ if (j0 == -1)
break;
- *num_entries -= i;
- if (i)
+ scan_term_untrans (zi->odr, &glist[i+before].term, mterm);
+ rset = rset_trunc (zi, &scan_info_array[j0].list[ptr[j0]].isam_p, 1);
+
+ ptr[j0]++;
+ for (j = j0+1; j<ord_no; j++)
+ {
+ if (ptr[j] < before+after &&
+ (tst=scan_info_array[j].list[ptr[j]].term) &&
+ !strcmp (tst, mterm))
+ {
+ rset_bool_parms bool_parms;
+ RSET rset2;
+
+ rset2 =
+ rset_trunc (zi, &scan_info_array[j].list[ptr[j]].isam_p, 1);
+
+ bool_parms.key_size = sizeof(struct it_key);
+ bool_parms.cmp = key_compare;
+ bool_parms.rset_l = rset;
+ bool_parms.rset_r = rset2;
+
+ rset = rset_create (rset_kind_or, &bool_parms);
+
+ ptr[j]++;
+ }
+ }
+ count_set (rset, &glist[i+before].occurrences);
+ rset_delete (rset);
+ }
+ if (i < after)
+ {
+ *num_entries -= (after-i);
*status = BEND_SCAN_PARTIAL;
+ }
- for (i = 0; i<scan_info.before; i++)
- if (scan_info.list[i].term)
+ for (i = 0; i<ord_no; i++)
+ ptr[i] = 0;
+
+ for (i = 0; i<before; i++)
+ {
+ int j, j0 = -1;
+ const char *mterm = NULL;
+ const char *tst;
+ RSET rset;
+
+ for (j = 0; j <ord_no; j++)
+ {
+ if (ptr[j] < before &&
+ (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
+ (!mterm || strcmp (tst, mterm) > 0))
+ {
+ j0 = j;
+ mterm = tst;
+ }
+ }
+ if (j0 == -1)
break;
- if (i)
- *status = BEND_SCAN_PARTIAL;
- *position -= i;
- *num_entries -= i;
- *list = scan_info.list+i; /* list is set to first 'real' entry */
+ scan_term_untrans (zi->odr, &glist[before-1-i].term, mterm);
+
+ rset = rset_trunc
+ (zi, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1);
+
+ ptr[j0]++;
- if (*num_entries == 0) /* signal 'unsupported use-attribute' */
- zi->errCode = 114; /* if no entries was found */
+ for (j = j0+1; j<ord_no; j++)
+ {
+ if (ptr[j] < before &&
+ (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
+ !strcmp (tst, mterm))
+ {
+ rset_bool_parms bool_parms;
+ RSET rset2;
+
+ rset2 = rset_trunc (zi,
+ &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1);
+
+ bool_parms.key_size = sizeof(struct it_key);
+ bool_parms.cmp = key_compare;
+ bool_parms.rset_l = rset;
+ bool_parms.rset_r = rset2;
+
+ rset = rset_create (rset_kind_or, &bool_parms);
+
+ ptr[j]++;
+ }
+ }
+ count_set (rset, &glist[before-1-i].occurrences);
+ rset_delete (rset);
+ }
+ i = before-i;
+ if (i)
+ {
+ *status = BEND_SCAN_PARTIAL;
+ *position -= i;
+ *num_entries -= i;
+ }
+ *list = glist + i; /* list is set to first 'real' entry */
+
logf (LOG_DEBUG, "position = %d, num_entries = %d",
*position, *num_entries);
if (zi->errCode)
logf (LOG_DEBUG, "scan error: %d", zi->errCode);
- return 0;
+ return zi->errCode;
}
-
-