/*
- * Copyright (C) 1994-1996, Index Data I/S
+ * Copyright (C) 1994-1997, Index Data I/S
* All rights reserved.
* Sebastian Hammer, Adam Dickmeiss
*
* $Log: zrpn.c,v $
- * Revision 1.48 1996-05-28 15:15:01 adam
+ * Revision 1.64 1997-09-18 08:59:20 adam
+ * Extra generic handle for the character mapping routines.
+ *
+ * Revision 1.63 1997/09/17 12:19:18 adam
+ * Zebra version corresponds to YAZ version 1.4.
+ * Changed Zebra server so that it doesn't depend on global common_resource.
+ *
+ * Revision 1.62 1997/09/05 15:30:09 adam
+ * Changed prototype for chr_map_input - added const.
+ * Added support for C++, headers uses extern "C" for public definitions.
+ *
+ * Revision 1.61 1997/02/10 10:21:14 adam
+ * Bug fix: in search terms character (^) wasn't observed.
+ *
+ * Revision 1.60 1997/01/31 11:10:34 adam
+ * Bug fix: Leading and trailing white space weren't removed in scan tokens.
+ *
+ * Revision 1.59 1997/01/17 11:31:46 adam
+ * Bug fix: complete phrase search didn't work.
+ *
+ * Revision 1.58 1996/12/23 15:30:45 adam
+ * Work on truncation.
+ * Bug fix: result sets weren't deleted after server shut down.
+ *
+ * Revision 1.57 1996/11/11 13:38:02 adam
+ * Added proximity support in search.
+ *
+ * Revision 1.56 1996/11/08 11:10:32 adam
+ * Buffers used during file match got bigger.
+ * Compressed ISAM support everywhere.
+ * Bug fixes regarding masking characters in queries.
+ * Redesigned Regexp-2 queries.
+ *
+ * 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
#include <ctype.h>
#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_attributes->local, 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;
-};
+#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 heap_swap (struct trunc_info *ti, int i1, int i2)
+static void add_isam_p (const char *info, struct grep_info *p)
{
- int swap;
-
- swap = ti->ptr[i1];
- ti->ptr[i1] = ti->ptr[i2];
- ti->ptr[i2] = swap;
-}
+ 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) *
+ p->isam_p_size);
+ if (p->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);
+ }
+ p->isam_p_buf = new_isam_p_buf;
-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)
+#ifdef TERM_COUNT
+ new_term_no = xmalloc (sizeof(*new_term_no) *
+ p->isam_p_size);
+ if (p->term_no)
{
- heap_swap (ti, cur, child);
- cur = child;
- child = 2*cur;
+ memcpy (new_term_no, p->isam_p_buf,
+ p->isam_p_indx * sizeof(*p->term_no));
+ xfree (p->term_no);
}
- else
- break;
+ 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 void heap_insert (struct trunc_info *ti, const char *buf, int indx)
+static int grep_handle (char *name, const char *info, void *p)
{
- 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;
- }
+ add_isam_p (info, p);
+ return 0;
}
-static
-struct trunc_info *heap_init (int size, int key_size,
- int (*cmp)(const void *p1, const void *p2))
+static int term_pre (const char **src, const char *ct1, const char *ct2)
{
- struct trunc_info *ti = xmalloc (sizeof(*ti));
- int i;
+ const char *s1, *s0 = *src;
+ const char **map;
- ++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; )
+ /* skip white space */
+ while (*s0)
{
- ti->ptr[i] = i;
- ti->heap[i] = ti->buf + ti->keysize * i;
+ if (ct1 && strchr (ct1, *s0))
+ break;
+ if (ct2 && strchr (ct2, *s0))
+ break;
+ s1 = s0;
+ map = map_chrs_input (0, &s1, strlen(s1));
+ if (**map != *CHR_SPACE)
+ break;
+ s0 = s1;
}
- 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);
+ *src = s0;
+ return *s0;
}
-static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
- int merge_chunk)
+static int term_100 (const char **src, char *dst, int space_split)
{
- 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);
+ const char *s0, *s1;
+ const char **map;
+ int i = 0;
- if (to - from > merge_chunk)
+ if (!term_pre (src, NULL, NULL))
+ return 0;
+ s0 = *src;
+ while (*s0)
{
- 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; )
+ s1 = s0;
+ map = map_chrs_input (0, &s0, strlen(s0));
+ if (space_split && **map == *CHR_SPACE)
+ break;
+ while (s1 < s0)
{
- 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]);
- }
+ if (!isalnum (*s1))
+ dst[i++] = '\\';
+ dst[i++] = *s1++;
}
- 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;
+ dst[i] = '\0';
+ *src = s0;
+ return i;
+}
- ispt = xmalloc (sizeof(*ispt) * (to-from));
+static int term_101 (const char **src, char *dst, int space_split)
+{
+ const char *s0, *s1;
+ const char **map;
+ int i = 0;
- ti = heap_init (to-from, sizeof(struct it_key),
- key_compare);
- for (i = to-from; --i >= 0; )
+ if (!term_pre (src, "#", "#"))
+ return 0;
+ s0 = *src;
+ while (*s0)
+ {
+ if (*s0 == '#')
{
- 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]);
+ dst[i++] = '.';
+ dst[i++] = '*';
+ s0++;
}
- while (ti->heapnum)
+ else
{
- 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)
+ s1 = s0;
+ map = map_chrs_input (0, &s0, strlen(s0));
+ if (space_split && **map == *CHR_SPACE)
+ break;
+ while (s1 < s0)
{
- 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;
- }
+ if (!isalnum (*s1))
+ dst[i++] = '\\';
+ dst[i++] = *s1++;
}
-#endif
}
- heap_close (ti);
- xfree (ispt);
}
- rset_close (result, result_rsfd);
- return result;
+ dst[i] = '\0';
+ *src = s0;
+ return i;
}
-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)
+static int term_103 (const char **src, char *dst, int *errors, int space_split)
{
+ int i = 0;
+ const char *s0, *s1;
+ const char **map;
- 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)
+ if (!term_pre (src, "^\\()[].*+?|", "("))
+ return 0;
+ s0 = *src;
+ if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] &&
+ isdigit (s0[1]))
{
- ISAM_P *new_isam_p_buf;
-
- p->isam_p_size = 2*p->isam_p_size + 100;
- new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
- p->isam_p_size);
- if (p->isam_p_buf)
+ *errors = s0[1] - '0';
+ s0 += 3;
+ if (*errors > 3)
+ *errors = 3;
+ }
+ while (*s0)
+ {
+ if (strchr ("^\\()[].*+?|-", *s0))
+ dst[i++] = *s0++;
+ else
{
- memcpy (new_isam_p_buf, p->isam_p_buf,
- p->isam_p_indx * sizeof(*p->isam_p_buf));
- xfree (p->isam_p_buf);
+ s1 = s0;
+ map = map_chrs_input (0, &s0, strlen(s0));
+ if (**map == *CHR_SPACE)
+ break;
+ while (s1 < s0)
+ {
+ if (!isalnum (*s1))
+ dst[i++] = '\\';
+ dst[i++] = *s1++;
+ }
}
- p->isam_p_buf = new_isam_p_buf;
}
- 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)++;
+ dst[i] = '\0';
+ *src = s0;
+ return i;
}
-static int grep_handle (char *name, const char *info, void *p)
+static int term_102 (const char **src, char *dst, int space_split)
{
- logf (LOG_DEBUG, "dict name: %s", name);
- add_isam_p (info, p);
- return 0;
+ return term_103 (src, dst, NULL, space_split);
}
+/* 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++)
{
}
static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
- const char *term_sub,
+ const char **term_sub,
char *term_dict,
oid_value attributeSet,
struct grep_info *grep_info,
attr_init (&relation, zapt, 2);
relation_value = attr_find (&relation, NULL);
- term_value = atoi (term_sub);
switch (relation_value)
{
case 1:
+ if (!term_100 (term_sub, term_dict, 1))
+ return 0;
+ term_value = atoi (term_dict);
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_100 (term_sub, term_dict, 1))
+ return 0;
+ term_value = atoi (term_dict);
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_100 (term_sub, term_dict, 1))
+ return 0;
+ term_value = atoi (term_dict);
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_100 (term_sub, term_dict, 1))
+ return 0;
+ term_value = atoi (term_dict);
if (term_value < 0)
term_value = 0;
logf (LOG_DEBUG, "Relation >");
return 0;
}
logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
- r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos,
+ 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);
return 1;
}
-static void verbatim_char (int ch, int *indx, char *dst)
-{
- if (!isalnum (ch))
- dst[(*indx)++] = '\\';
- dst[(*indx)++] = ch;
-}
-
static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
- const char *term_sub, int regType,
+ const char **term_sub, int regType,
oid_value attributeSet, struct grep_info *grep_info,
- int num_bases, char **basenames)
+ int num_bases, char **basenames, int space_split)
{
char term_dict[2*IT_MAX_WORD+2];
- int i, j, r, base_no;
+ int j, r, base_no;
AttrType truncation;
int truncation_value;
AttrType use;
int use_value;
oid_value curAttributeSet = attributeSet;
+ const char *termp;
attr_init (&use, zapt, 1);
use_value = attr_find (&use, &curAttributeSet);
- logf (LOG_DEBUG, "use value %d", use_value);
+ logf (LOG_DEBUG, "field_term, use value %d", use_value);
attr_init (&truncation, zapt, 5);
truncation_value = attr_find (&truncation, NULL);
logf (LOG_DEBUG, "truncation value %d", truncation_value);
for (base_no = 0; base_no < num_bases; base_no++)
{
- attent *attp;
+ attent attp;
data1_local_attribute *local_attr;
int max_pos, prefix_len = 0;
- attp = att_getentbyatt (curAttributeSet, use_value);
- if (!attp)
+ termp = *term_sub;
+ if (!att_getentbyatt (zi, &attp, curAttributeSet, use_value))
{
logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
curAttributeSet, use_value);
zi->errString = basenames[base_no];
return -1;
}
- for (local_attr = attp->local_attributes; local_attr;
+ for (local_attr = attp.local_attributes; local_attr;
local_attr = local_attr->next)
{
int ord;
- ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal,
+ ord = zebTargetInfo_lookupSU (zi->zti, attp.attset_ordinal,
local_attr->local);
if (ord < 0)
continue;
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;
- }
+ 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++] = ')';
+ term_dict[prefix_len++] = 1;
term_dict[prefix_len++] = regType;
term_dict[prefix_len] = '\0';
- if (!relational_term (zi, zapt, term_sub, term_dict,
+ if (!relational_term (zi, zapt, &termp, 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,
+ term_dict[j++] = '(';
+ if (!term_100 (&termp, term_dict + j, space_split))
+ return 0;
+ strcat (term_dict, ")");
+ 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 */
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,
+ if (!term_100 (&termp, term_dict + j, space_split))
+ return 0;
+ strcat (term_dict, ".*)");
+ dict_lookup_grep (zi->dict, term_dict, 0, grep_info,
&max_pos, 0, grep_handle);
break;
case 2: /* left truncation */
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,
+ if (!term_101 (&termp, term_dict + j, space_split))
+ return 0;
+ strcat (term_dict, ")");
+ 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);
+ 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,
+ case 102: /* Regexp-1 */
+ term_dict[j++] = '(';
+ if (!term_102 (&termp, term_dict + j, space_split))
+ return 0;
+ strcat (term_dict, ")");
+ logf (LOG_DEBUG, "Regexp-1 tolerance=%d", r);
+ 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->wordDict, term_dict, r, grep_info,
- &max_pos, j, grep_handle);
+ case 103: /* Regexp-1 */
+ r = 1;
+ term_dict[j++] = '(';
+ if (!term_103 (&termp, term_dict + j, &r, space_split))
+ return 0;
+ strcat (term_dict, ")");
+ logf (LOG_DEBUG, "Regexp-2 tolerance=%d", r);
+ r = dict_lookup_grep (zi->dict, term_dict, r, grep_info,
+ &max_pos, 2, grep_handle);
if (r)
logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d",
r);
}
}
}
+ *term_sub = termp;
logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
- return 0;
+ return 1;
}
static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
char *termz)
{
- size_t i, sizez;
- int sep = 0;
+ size_t sizez;
Z_Term *term = zapt->term;
sizez = term->u.general->len;
if (sizez > IT_MAX_WORD-1)
sizez = IT_MAX_WORD-1;
- for (i = 0; i < sizez; i++)
+ 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;
+ const char **map;
+ const char *cp = (const char *) term->u.general->buf;
+ const char *cp_end = cp + term->u.general->len;
+ const char *src;
+ int i = 0;
+ const char *space_map = NULL;
+ int len;
+
+ while ((len = (cp_end - cp)) > 0)
{
- if (!isalnum(term->u.general->buf[i]))
- sep = 1;
+ map = map_chrs_input (0, &cp, len);
+ if (**map == *CHR_SPACE)
+ space_map = *map;
else
{
- if (sep)
- *termz++ = ' ';
- *termz++ = index_char_cvt (term->u.general->buf[i]);
- sep = 0;
+ if (i && space_map)
+ for (src = space_map; *src; src++)
+ termz[i++] = *src;
+ space_map = NULL;
+ for (src = *map; *src; src++)
+ termz[i++] = *src;
}
}
- if (sep)
- *termz++ = ' ';
- *termz = '\0';
+ termz[i] = '\0';
}
static RSET rpn_search_APT_relevance (ZServerInfo *zi,
{
rset_relevance_parms parms;
char termz[IT_MAX_WORD+1];
- char term_sub[IT_MAX_WORD+1];
+ const char *termp = termz;
struct grep_info grep_info;
- char *p0 = termz, *p1 = NULL;
RSET result;
+ int term_index = 0;
+ int r;
parms.key_size = sizeof(struct it_key);
- parms.max_rec = 100;
- parms.cmp = key_compare;
- parms.is = zi->wordIsam;
+ parms.max_rec = 1000;
+ parms.cmp = key_compare_it;
+ parms.is = zi->isam;
+ parms.isc = zi->isamc;
+ parms.no_terms = 0;
if (zapt->term->which != Z_Term_general)
{
}
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, ' ')))
- {
- memcpy (term_sub, p0, p1-p0);
- term_sub[p1-p0] = '\0';
- }
- else
- strcpy (term_sub, p0);
- if (field_term (zi, zapt, term_sub, 'w', attributeSet, &grep_info,
- num_bases, basenames))
- return NULL;
- if (!p1)
+ r = field_term (zi, zapt, &termp, 'w', attributeSet, &grep_info,
+ num_bases, basenames, 1);
+ if (r <= 0)
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;
}
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;
+ const char *termp = termz;
+ int r;
if (zapt->term->which != Z_Term_general)
{
}
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 (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);
+ r = field_term (zi, zapt, &termp, 'p', attributeSet, &grep_info,
+ num_bases, basenames, 0);
+ 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;
}
-static RSET rpn_search_APT_word (ZServerInfo *zi,
- Z_AttributesPlusTerm *zapt,
- oid_value attributeSet,
- int num_bases, char **basenames)
+
+static RSET rpn_proximity (ZServerInfo *zi, RSET rset1, RSET rset2,
+ int ordered,
+ int exclusion, int relation, int distance)
{
- rset_isam_parms parms;
- char termz[IT_MAX_WORD+1];
- struct grep_info grep_info;
+ int i;
+ RSFD rsfd1, rsfd2;
+ int more1, more2;
+ struct it_key buf1, buf2;
+ RSFD rsfd_result;
RSET result;
+ rset_temp_parms parms;
+
+ rsfd1 = rset_open (rset1, RSETF_READ|RSETF_SORT_SYSNO);
+ more1 = rset_read (rset1, rsfd1, &buf1);
+
+ rsfd2 = rset_open (rset2, RSETF_READ|RSETF_SORT_SYSNO);
+ more2 = rset_read (rset2, rsfd2, &buf2);
- if (zapt->term->which != Z_Term_general)
- {
- zi->errCode = 124;
- return NULL;
- }
- trans_term (zi, zapt, 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, 'w', 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.key_size = sizeof (struct it_key);
+ parms.temp_path = res_get (zi->res, "setTmpDir");
+ result = rset_create (rset_kind_temp, &parms);
+ rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
+
+ logf (LOG_DEBUG, "rpn_proximity excl=%d ord=%d rel=%d dis=%d",
+ exclusion, ordered, relation, distance);
+ while (more1 && more2)
{
- parms.is = zi->wordIsam;
- parms.pos = *grep_info.isam_p_buf;
- result = rset_create (rset_kind_isam, &parms);
+ int cmp = key_compare_it (&buf1, &buf2);
+ if (cmp < -1)
+ more1 = rset_read (rset1, rsfd1, &buf1);
+ else if (cmp > 1)
+ more2 = rset_read (rset2, rsfd2, &buf2);
+ else
+ {
+ int sysno = buf1.sysno;
+ int seqno[500];
+ int n = 0;
+
+ seqno[n++] = buf1.seqno;
+ while ((more1 = rset_read (rset1, rsfd1, &buf1)) &&
+ sysno == buf1.sysno)
+ if (n < 500)
+ seqno[n++] = buf1.seqno;
+ do
+ {
+ for (i = 0; i<n; i++)
+ {
+ int diff = buf2.seqno - seqno[i];
+ int excl = exclusion;
+ if (!ordered && diff < 0)
+ diff = -diff;
+ switch (relation)
+ {
+ case 1: /* < */
+ if (diff < distance)
+ excl = !excl;
+ break;
+ case 2: /* <= */
+ if (diff <= distance)
+ excl = !excl;
+ break;
+ case 3: /* == */
+ if (diff == distance)
+ excl = !excl;
+ break;
+ case 4: /* >= */
+ if (diff >= distance)
+ excl = !excl;
+ break;
+ case 5: /* > */
+ if (diff > distance)
+ excl = !excl;
+ break;
+ case 6: /* != */
+ if (diff != distance)
+ excl = !excl;
+ break;
+ }
+ if (excl)
+ rset_write (result, rsfd_result, &buf2);
+ }
+ } while ((more2 = rset_read (rset2, rsfd2, &buf2)) &&
+ sysno == buf2.sysno);
+ }
}
- else
- result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf,
- grep_info.isam_p_indx);
- xfree (grep_info.isam_p_buf);
+ rset_close (result, rsfd_result);
+ rset_close (rset1, rsfd1);
+ rset_close (rset2, rsfd2);
return result;
}
-static RSET rpn_prox (RSET *rset, int rset_no)
+static RSET rpn_prox (ZServerInfo *zi, RSET *rset, int rset_no)
{
int i;
RSFD *rsfd;
}
}
parms.key_size = sizeof (struct it_key);
+ parms.temp_path = res_get (zi->res, "setTmpDir");
result = rset_create (rset_kind_temp, &parms);
rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
*more = 0;
break;
}
- cmp = key_compare (buf[i], buf[i-1]);
+ cmp = key_compare_it (buf[i], buf[i-1]);
if (cmp > 1)
{
more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
int num_bases, char **basenames)
{
char termz[IT_MAX_WORD+1];
- char term_sub[IT_MAX_WORD+1];
- char *p0 = termz, *p1 = NULL;
+ const char *termp = termz;
RSET rset[60], result;
- int i, rset_no = 0;
+ int i, r, rset_no = 0;
struct grep_info grep_info;
if (zapt->term->which != Z_Term_general)
}
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, ' ')))
- {
- memcpy (term_sub, p0, p1-p0);
- term_sub[p1-p0] = '\0';
- }
- else
- strcpy (term_sub, p0);
-
grep_info.isam_p_indx = 0;
- 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);
- }
+ r = field_term (zi, zapt, &termp, 'w', attributeSet, &grep_info,
+ num_bases, basenames, 1);
+ if (r < 1)
+ break;
+ 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);
else if (rset_no == 1)
return (rset[0]);
-
- result = rpn_prox (rset, rset_no);
+ result = rpn_prox (zi, rset, rset_no);
for (i = 0; i<rset_no; i++)
rset_delete (rset[i]);
return result;
return NULL;
}
parms.key_size = sizeof (struct it_key);
+ parms.temp_path = res_get (zi->res, "setTmpDir");
result = rset_create (rset_kind_temp, &parms);
rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
if (completeness_value == 2 || completeness_value == 3)
return rpn_search_APT_cphrase (zi, zapt, attributeSet,
num_bases, basenames);
- return rpn_search_APT_word (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;
}
RSET r = NULL;
if (zs->which == Z_RPNStructure_complex)
{
+ Z_Operator *zop = zs->u.complex->roperator;
rset_bool_parms bool_parms;
int soft = 0;
+
bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
attributeSet,
if (rset_is_ranked(bool_parms.rset_r))
soft = 1;
bool_parms.key_size = sizeof(struct it_key);
- bool_parms.cmp = key_compare;
+ bool_parms.cmp = key_compare_it;
- switch (zs->u.complex->roperator->which)
+ switch (zop->which)
{
case Z_Operator_and:
r = rset_create (soft ? rset_kind_sand:rset_kind_and, &bool_parms);
case Z_Operator_and_not:
r = rset_create (soft ? rset_kind_snot:rset_kind_not, &bool_parms);
break;
+ case Z_Operator_prox:
+ if (zop->u.prox->which != Z_ProxCode_known)
+ {
+ zi->errCode = 132;
+ return NULL;
+ }
+ if (*zop->u.prox->proximityUnitCode != Z_ProxUnit_word)
+ {
+ static char val[16];
+ zi->errCode = 132;
+ zi->errString = val;
+ sprintf (val, "%d", *zop->u.prox->proximityUnitCode);
+ return NULL;
+ }
+ r = rpn_proximity (zi, bool_parms.rset_l, bool_parms.rset_r,
+ *zop->u.prox->ordered,
+ (!zop->u.prox->exclusion ? 0 :
+ *zop->u.prox->exclusion),
+ *zop->u.prox->relationType,
+ *zop->u.prox->distance);
+ break;
default:
- assert (0);
+ zi->errCode = 110;
+ return NULL;
}
}
else if (zs->which == Z_RPNStructure_simple)
}
else
{
- assert (0);
+ zi->errCode = 3;
+ return NULL;
}
}
else
{
- assert (0);
+ zi->errCode = 3;
+ return NULL;
}
return r;
}
-void count_set_save (RSET *r, int *count)
+void count_set_save (ZServerInfo *zi, RSET *r, int *count)
{
int psysno = 0;
int kno = 0;
RSFD rfd, wfd;
RSET w;
rset_temp_parms parms;
-
+ int maxResultSetSize = atoi (res_get_def (zi->res,
+ "maxResultSetSize", "400"));
logf (LOG_DEBUG, "count_set_save");
*count = 0;
parms.key_size = sizeof(struct it_key);
+ parms.temp_path = res_get (zi->res, "setTmpDir");
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);
{
if (key.sysno != psysno)
{
- rset_write (w, wfd, &key);
- psysno = key.sysno;
+ if (*count < maxResultSetSize)
+ rset_write (w, wfd, &key);
(*count)++;
+ psysno = key.sysno;
}
kno++;
}
oident *attrset;
oid_value attributeSet;
+ dict_grep_cmap (zi->dict, 0, map_chrs_input);
zlog_rpn (rpn);
zi->errCode = 0;
if (!rset)
return zi->errCode;
if (rset_is_volatile(rset))
- count_set_save(&rset,hits);
+ count_set_save(zi, &rset,hits);
else
count_set (rset, hits);
resultSetAdd (zi, setname, 1, rset);
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);
}
+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,
logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
+ 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 (attributeset == VAL_NONE)
- attributeset = VAL_BIB1;
-
if (use_value == -1)
use_value = 1016;
for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++)
{
- attent *attp;
+ attent attp;
data1_local_attribute *local_attr;
- attp = att_getentbyatt (attributeset, use_value);
- if (!attp)
+ if (!att_getentbyatt (zi, &attp, attributeset, use_value))
{
logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
attributeset, use_value);
zi->errString = basenames[base_no];
return zi->errCode = 109; /* Database unavailable */
}
- for (local_attr = attp->local_attributes; local_attr && ord_no < 32;
+ for (local_attr = attp.local_attributes; local_attr && ord_no < 32;
local_attr = local_attr->next)
{
int ord;
- ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal,
+ ord = zebTargetInfo_lookupSU (zi->zti, attp.attset_ordinal,
local_attr->local);
if (ord > 0)
ords[ord_no++] = ord;
(completeness_value==2 || completeness_value==3) ? 'p': 'w';
termz[prefix_len] = 0;
strcpy (scan_info->prefix, termz);
- trans_term (zi, zapt, termz+prefix_len);
- dict_scan (zi->wordDict, termz, &before_tmp, &after_tmp, scan_info,
+
+ trans_scan_term (zi, zapt, termz+prefix_len);
+
+ dict_scan (zi->dict, termz, &before_tmp, &after_tmp, scan_info,
scan_handle);
}
glist = odr_malloc (zi->odr, (before+after)*sizeof(*glist));
const char *mterm = NULL;
const char *tst;
RSET rset;
- rset_isam_parms parms;
for (j = 0; j < ord_no; j++)
{
}
if (j0 == -1)
break;
- glist[i+before].term = odr_malloc (zi->odr, strlen(mterm)+1);
- strcpy (glist[i+before].term, mterm);
-
- parms.is = zi->wordIsam;
- parms.pos = scan_info_array[j0].list[ptr[j0]].isam_p;
- rset = rset_create (rset_kind_isam, &parms);
+ 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++)
(tst=scan_info_array[j].list[ptr[j]].term) &&
!strcmp (tst, mterm))
{
- rset_isam_parms parms;
rset_bool_parms bool_parms;
RSET rset2;
- parms.is = zi->wordIsam;
- parms.pos = scan_info_array[j].list[ptr[j]].isam_p;
- rset2 = rset_create (rset_kind_isam, &parms);
+ 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.cmp = key_compare_it;
bool_parms.rset_l = rset;
bool_parms.rset_r = rset2;
const char *mterm = NULL;
const char *tst;
RSET rset;
- rset_isam_parms parms;
for (j = 0; j <ord_no; j++)
{
}
if (j0 == -1)
break;
- glist[before-1-i].term = odr_malloc (zi->odr, strlen(mterm)+1);
- strcpy (glist[before-1-i].term, mterm);
- parms.is = zi->wordIsam;
- parms.pos = scan_info_array[j0].list[before-1-ptr[j0]].isam_p;
- rset = rset_create (rset_kind_isam, &parms);
+ 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]++;
(tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
!strcmp (tst, mterm))
{
- rset_isam_parms parms;
rset_bool_parms bool_parms;
RSET rset2;
- parms.is = zi->wordIsam;
- parms.pos = scan_info_array[j].list[before-1-ptr[j]].isam_p;
- rset2 = rset_create (rset_kind_isam, &parms);
+ 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.cmp = key_compare_it;
bool_parms.rset_l = rset;
bool_parms.rset_r = rset2;
projects / idzebra-moved-to-github.git / blobdiff