1 /* $Id: zsets.c,v 1.119 2007-01-17 15:35:48 adam Exp $
2 Copyright (C) 1995-2007
5 This file is part of the Zebra server.
7 Zebra is free software; you can redistribute it and/or modify it under
8 the terms of the GNU General Public License as published by the Free
9 Software Foundation; either version 2, or (at your option) any later
12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
17 You should have received a copy of the GNU General Public License
18 along with this program; if not, write to the Free Software
19 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
34 #include <yaz/diagbib1.h>
37 #define ZSET_SORT_MAX_LEVEL 10
39 struct zebra_set_term_entry {
50 NMEM rset_nmem; /* for creating the rsets in */
55 Z_SortKeySpecList *sortSpec;
56 struct zset_sort_info *sort_info;
57 struct zebra_set_term_entry *term_entries;
59 struct zebra_set *next;
61 int estimated_hit_count;
63 zint cache_position; /* last position */
64 RSFD cache_rfd; /* rfd (NULL if not existing) */
65 zint cache_psysno; /* sysno for last position */
66 zint approx_limit; /* limit before we do approx */
69 struct zset_sort_entry {
74 struct zset_sort_info {
77 struct zset_sort_entry *all_entries;
78 struct zset_sort_entry **entries;
81 static int log_level_set=0;
82 static int log_level_sort=0;
83 static int log_level_searchhits=0;
84 static int log_level_searchterms=0;
85 static int log_level_resultsets=0;
87 static void loglevels(void)
91 log_level_sort = yaz_log_module_level("sorting");
92 log_level_searchhits = yaz_log_module_level("searchhits");
93 log_level_searchterms = yaz_log_module_level("searchterms");
94 log_level_resultsets = yaz_log_module_level("resultsets");
99 static ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem,
100 Z_RPNQuery *rpn, ZebraSet sset)
104 Z_SortKeySpecList *sort_sequence;
106 ZEBRA_RES res = ZEBRA_OK;
108 sort_sequence = (Z_SortKeySpecList *)
109 nmem_malloc(nmem, sizeof(*sort_sequence));
110 sort_sequence->num_specs = 10; /* FIXME - Hard-coded number */
111 sort_sequence->specs = (Z_SortKeySpec **)
112 nmem_malloc(nmem, sort_sequence->num_specs *
113 sizeof(*sort_sequence->specs));
114 for (i = 0; i<sort_sequence->num_specs; i++)
115 sort_sequence->specs[i] = 0;
117 attrset = oid_getentbyoid (rpn->attributeSetId);
119 rpn_get_top_approx_limit(zh, rpn->RPNStructure, &sset->approx_limit);
121 res = rpn_search_top(zh, rpn->RPNStructure, attrset->value,
124 sset->num_bases, sset->basenames,
131 for (i = 0; sort_sequence->specs[i]; i++)
133 sort_sequence->num_specs = i;
134 rset->hits_limit = sset->approx_limit;
137 res = resultSetRank (zh, sset, rset, rset_nmem);
141 res = resultSetSortSingle (zh, nmem, sset, rset,
142 sort_sequence, &sort_status);
149 ZEBRA_RES resultSetAddRPN(ZebraHandle zh, NMEM m, Z_RPNQuery *rpn,
150 int num_bases, char **basenames,
152 zint *hits, int *estimated_hit_count)
159 *estimated_hit_count = 0;
161 zebraSet = resultSetAdd(zh, setname, 1);
164 zebraSet->locked = 1;
167 zebraSet->rset_nmem = nmem_create();
169 zebraSet->num_bases = num_bases;
170 zebraSet->basenames =
171 nmem_malloc (zebraSet->nmem, num_bases * sizeof(*zebraSet->basenames));
172 for (i = 0; i<num_bases; i++)
173 zebraSet->basenames[i] = nmem_strdup(zebraSet->nmem, basenames[i]);
175 res = resultSetSearch(zh, zebraSet->nmem, zebraSet->rset_nmem,
177 *hits = zebraSet->hits;
178 if (zebraSet->estimated_hit_count)
179 *estimated_hit_count = 1;
183 zebraSet->locked = 0;
189 void resultSetAddTerm(ZebraHandle zh, ZebraSet s, int reg_type,
190 const char *db, const char *index_name,
193 assert(zh); /* compiler shut up */
195 s->nmem = nmem_create ();
196 if (!s->term_entries)
199 s->term_entries_max = 1000;
201 nmem_malloc (s->nmem, s->term_entries_max *
202 sizeof(*s->term_entries));
203 for (i = 0; i < s->term_entries_max; i++)
204 s->term_entries[i].term = 0;
206 if (s->hits < s->term_entries_max)
208 s->term_entries[s->hits].reg_type = reg_type;
209 s->term_entries[s->hits].db = nmem_strdup (s->nmem, db);
210 s->term_entries[s->hits].index_name = nmem_strdup(s->nmem, index_name);
211 s->term_entries[s->hits].term = nmem_strdup(s->nmem, term);
216 ZebraSet resultSetAdd(ZebraHandle zh, const char *name, int ov)
221 for (s = zh->sets; s; s = s->next)
222 if (!strcmp (s->name, name))
229 yaz_log(log_level_resultsets, "updating result set %s", name);
230 if (!ov || s->locked)
235 rset_close(s->cache_rfd);
236 rset_delete (s->rset);
239 nmem_destroy (s->rset_nmem);
241 nmem_destroy (s->nmem);
245 const char *sort_max_str = zebra_get_resource(zh, "sortmax", "1000");
247 yaz_log(log_level_resultsets, "adding result set %s", name);
248 s = (ZebraSet) xmalloc (sizeof(*s));
251 s->name = (char *) xmalloc (strlen(name)+1);
252 strcpy (s->name, name);
254 s->sort_info = (struct zset_sort_info *)
255 xmalloc (sizeof(*s->sort_info));
256 s->sort_info->max_entries = atoi(sort_max_str);
257 if (s->sort_info->max_entries < 2)
258 s->sort_info->max_entries = 2;
260 s->sort_info->entries = (struct zset_sort_entry **)
261 xmalloc (sizeof(*s->sort_info->entries) *
262 s->sort_info->max_entries);
263 s->sort_info->all_entries = (struct zset_sort_entry *)
264 xmalloc (sizeof(*s->sort_info->all_entries) *
265 s->sort_info->max_entries);
266 for (i = 0; i < s->sort_info->max_entries; i++)
267 s->sort_info->entries[i] = s->sort_info->all_entries + i;
277 s->cache_position = 0;
279 s->approx_limit = zh->approx_limit;
280 s->estimated_hit_count = 0;
284 ZebraSet resultSetGet(ZebraHandle zh, const char *name)
288 for (s = zh->sets; s; s = s->next)
289 if (!strcmp (s->name, name))
291 if (!s->term_entries && !s->rset && s->rpn)
293 NMEM nmem = nmem_create ();
294 yaz_log(log_level_resultsets, "research %s", name);
296 s->rset_nmem=nmem_create();
297 resultSetSearch(zh, nmem, s->rset_nmem, s->rpn, s);
298 if (s->rset && s->sortSpec)
301 yaz_log(log_level_resultsets, "resort %s", name);
302 resultSetSortSingle (zh, nmem, s, s->rset, s->sortSpec,
312 void resultSetInvalidate (ZebraHandle zh)
314 ZebraSet s = zh->sets;
316 yaz_log(log_level_resultsets, "invalidating result sets");
317 for (; s; s = s->next)
322 rset_close(s->cache_rfd);
323 rset_delete (s->rset);
327 s->cache_position = 0;
329 nmem_destroy(s->rset_nmem);
334 void resultSetDestroy(ZebraHandle zh, int num, char **names,int *statuses)
336 ZebraSet * ss = &zh->sets;
340 for (i = 0; i<num; i++)
341 statuses[i] = Z_DeleteStatus_resultSetDidNotExist;
348 for (i = 0; i<num; i++)
349 if (!strcmp (s->name, names[i]))
352 statuses[i] = Z_DeleteStatus_success;
361 xfree (s->sort_info->all_entries);
362 xfree (s->sort_info->entries);
363 xfree (s->sort_info);
366 nmem_destroy (s->nmem);
370 rset_close(s->cache_rfd);
371 rset_delete (s->rset);
374 nmem_destroy(s->rset_nmem);
383 ZebraMetaRecord *zebra_meta_records_create_range(ZebraHandle zh,
388 zint *pos = pos_small;
392 if (num > 10000 || num <= 0)
396 pos = xmalloc(sizeof(*pos) * num);
398 for (i = 0; i<num; i++)
401 mr = zebra_meta_records_create(zh, name, num, pos);
408 ZebraMetaRecord *zebra_meta_records_create(ZebraHandle zh, const char *name,
409 int num, zint *positions)
412 ZebraMetaRecord *sr = 0;
415 struct zset_sort_info *sort_info;
416 size_t sysno_mem_index = 0;
418 if (zh->m_staticrank)
423 if (!(sset = resultSetGet (zh, name)))
425 if (!(rset = sset->rset))
427 if (!sset->term_entries)
429 sr = (ZebraMetaRecord *) xmalloc (sizeof(*sr) * num);
430 for (i = 0; i<num; i++)
437 if (positions[i] <= sset->term_entries_max)
439 sr[i].term = sset->term_entries[positions[i]-1].term;
440 sr[i].db = sset->term_entries[positions[i]-1].db;
446 sr = (ZebraMetaRecord *) xmalloc (sizeof(*sr) * num);
447 for (i = 0; i<num; i++)
454 sort_info = sset->sort_info;
459 for (i = 0; i<num; i++)
461 position = positions[i];
462 if (position > 0 && position <= sort_info->num_entries)
464 yaz_log(log_level_sort, "got pos=" ZINT_FORMAT
465 " (sorted)", position);
466 sr[i].sysno = sort_info->entries[position-1]->sysno;
467 sr[i].score = sort_info->entries[position-1]->score;
471 /* did we really get all entries using sort ? */
472 for (i = 0; i<num; i++)
477 if (i < num) /* nope, get the rest, unsorted - sorry */
486 position = sort_info->num_entries;
487 while (num_i < num && positions[num_i] <= position)
490 if (sset->cache_rfd &&
491 num_i < num && positions[num_i] > sset->cache_position)
493 position = sset->cache_position;
494 rfd = sset->cache_rfd;
495 psysno = sset->cache_psysno;
500 rset_close(sset->cache_rfd);
501 rfd = rset_open (rset, RSETF_READ);
503 while (num_i < num && rset_read (rfd, &key, 0))
505 zint this_sys = key.mem[sysno_mem_index];
506 if (this_sys != psysno)
511 /* determine we alreay have this in our set */
512 for (i = sort_info->num_entries; --i >= 0; )
513 if (psysno == sort_info->entries[i]->sysno)
519 assert (num_i < num);
520 if (position == positions[num_i])
522 sr[num_i].sysno = psysno;
523 yaz_log(log_level_sort, "got pos=" ZINT_FORMAT " (unsorted)", position);
524 sr[num_i].score = -1;
529 sset->cache_position = position;
530 sset->cache_psysno = psysno;
531 sset->cache_rfd = rfd;
537 void zebra_meta_records_destroy (ZebraHandle zh, ZebraMetaRecord *records,
540 assert(zh); /* compiler shut up about unused arg */
551 void resultSetInsertSort(ZebraHandle zh, ZebraSet sset,
552 struct sortKeyInfo *criteria, int num_criteria,
554 char *cmp_buf[], char *tmp_cmp_buf[])
556 struct zset_sort_entry *new_entry = NULL;
557 struct zset_sort_info *sort_info = sset->sort_info;
560 zebra_sort_sysno(zh->reg->sort_index, sysno);
561 for (i = 0; i<num_criteria; i++)
563 char *this_entry_buf = tmp_cmp_buf[i];
564 memset(this_entry_buf, '\0', SORT_IDX_ENTRYSIZE);
565 if (criteria[i].ord != -1)
567 zebra_sort_type(zh->reg->sort_index, criteria[i].ord);
568 zebra_sort_read(zh->reg->sort_index, this_entry_buf);
571 i = sort_info->num_entries;
575 for (j = 0; j<num_criteria; j++)
577 char *this_entry_buf = tmp_cmp_buf[j];
578 char *other_entry_buf =
579 cmp_buf[j] + i * SORT_IDX_ENTRYSIZE;
580 if (criteria[j].numerical)
582 char this_entry_org[1024];
583 char other_entry_org[1024];
585 int index_type = criteria[j].index_type;
586 zebra_term_untrans(zh, index_type, this_entry_org,
588 zebra_term_untrans(zh, index_type, other_entry_org,
590 diff = atof(this_entry_org) - atof(other_entry_org);
601 rel = memcmp(this_entry_buf, other_entry_buf,
609 if (criteria[j].relation == 'A')
614 else if (criteria[j].relation == 'D')
621 j = sort_info->max_entries;
625 if (sort_info->num_entries == j)
628 j = (sort_info->num_entries)++;
629 new_entry = sort_info->entries[j];
633 for (k = 0; k<num_criteria; k++)
635 char *j_buf = cmp_buf[k] + j * SORT_IDX_ENTRYSIZE;
636 char *j_1_buf = cmp_buf[k] + (j-1) * SORT_IDX_ENTRYSIZE;
637 memcpy(j_buf, j_1_buf, SORT_IDX_ENTRYSIZE);
639 sort_info->entries[j] = sort_info->entries[j-1];
642 sort_info->entries[i] = new_entry;
644 for (i = 0; i<num_criteria; i++)
646 char *new_entry_buf = cmp_buf[i] + j * SORT_IDX_ENTRYSIZE;
647 char *this_entry_buf = tmp_cmp_buf[i];
648 memcpy(new_entry_buf, this_entry_buf, SORT_IDX_ENTRYSIZE);
650 new_entry->sysno = sysno;
651 new_entry->score = -1;
654 void resultSetInsertRank(ZebraHandle zh, struct zset_sort_info *sort_info,
655 zint sysno, int score, int relation)
657 struct zset_sort_entry *new_entry = NULL;
659 assert(zh); /* compiler shut up about unused arg */
661 i = sort_info->num_entries;
666 rel = score - sort_info->entries[i]->score;
673 else if (relation == 'A')
680 j = sort_info->max_entries;
684 if (sort_info->num_entries == j)
687 j = (sort_info->num_entries)++;
689 new_entry = sort_info->entries[j];
692 sort_info->entries[j] = sort_info->entries[j-1];
695 sort_info->entries[i] = new_entry;
697 new_entry->sysno = sysno;
698 new_entry->score = score;
701 static Z_RPNQuery *copy_RPNQuery(Z_RPNQuery *src, NMEM nmem)
704 ODR encode = odr_createmem(ODR_ENCODE);
705 ODR decode = odr_createmem(ODR_DECODE);
707 if (z_RPNQuery(encode, &src, 0, 0))
710 char *buf = odr_getbuf(encode, &len, 0);
714 odr_setbuf(decode, buf, len, 0);
715 z_RPNQuery(decode, &dst, 0, 0);
718 nmem_transfer(nmem, decode->mem);
724 static Z_SortKeySpecList *copy_SortKeySpecList(Z_SortKeySpecList *src, NMEM nmem)
726 Z_SortKeySpecList *dst = 0;
727 ODR encode = odr_createmem(ODR_ENCODE);
728 ODR decode = odr_createmem(ODR_DECODE);
730 if (z_SortKeySpecList(encode, &src, 0, 0))
733 char *buf = odr_getbuf(encode, &len, 0);
737 odr_setbuf(decode, buf, len, 0);
738 z_SortKeySpecList(decode, &dst, 0, 0);
741 nmem_transfer(nmem, decode->mem);
747 ZebraSet resultSetClone(ZebraHandle zh, const char *setname,
753 nset = resultSetAdd(zh, setname, 1);
757 nset->nmem = nmem_create();
759 nset->num_bases = rset->num_bases;
761 nmem_malloc (nset->nmem, nset->num_bases * sizeof(*rset->basenames));
762 for (i = 0; i<rset->num_bases; i++)
763 nset->basenames[i] = nmem_strdup(nset->nmem, rset->basenames[i]);
766 nset->rset = rset_dup(rset->rset);
768 nset->rpn = copy_RPNQuery(rset->rpn, nset->nmem);
772 ZEBRA_RES resultSetSort(ZebraHandle zh, NMEM nmem,
773 int num_input_setnames, const char **input_setnames,
774 const char *output_setname,
775 Z_SortKeySpecList *sort_sequence, int *sort_status)
780 if (num_input_setnames == 0)
782 zebra_setError(zh, YAZ_BIB1_NO_RESULT_SET_NAME_SUPPLIED_ON_SORT, 0);
785 if (num_input_setnames > 1)
787 zebra_setError(zh, YAZ_BIB1_SORT_TOO_MANY_INPUT_RESULTS, 0);
792 yaz_log(log_level_sort, "result set sort input=%s output=%s",
793 *input_setnames, output_setname);
794 sset = resultSetGet (zh, input_setnames[0]);
797 zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
801 if (!(rset = sset->rset))
803 zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
807 if (strcmp (output_setname, input_setnames[0]))
808 sset = resultSetClone(zh, output_setname, sset);
809 sset->sortSpec = copy_SortKeySpecList(sort_sequence, sset->nmem);
810 return resultSetSortSingle (zh, nmem, sset, rset, sort_sequence,
814 ZEBRA_RES resultSetSortSingle(ZebraHandle zh, NMEM nmem,
815 ZebraSet sset, RSET rset,
816 Z_SortKeySpecList *sort_sequence,
824 struct sortKeyInfo sort_criteria[ZSET_SORT_MAX_LEVEL];
825 char *cmp_buf[ZSET_SORT_MAX_LEVEL];
826 char *tmp_cmp_buf[ZSET_SORT_MAX_LEVEL];
832 size_t sysno_mem_index = 0;
834 if (zh->m_staticrank)
837 assert(nmem); /* compiler shut up about unused param */
838 sset->sort_info->num_entries = 0;
840 rset_getterms(rset, 0, 0, &n);
841 terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n);
842 rset_getterms(rset, terms, n, &numTerms);
845 num_criteria = sort_sequence->num_specs;
846 if (num_criteria > ZSET_SORT_MAX_LEVEL)
847 num_criteria = ZSET_SORT_MAX_LEVEL;
848 for (i = 0; i < num_criteria; i++)
850 Z_SortKeySpec *sks = sort_sequence->specs[i];
854 sort_criteria[i].ord = -1;
855 sort_criteria[i].numerical = 0;
857 if (sks->which == Z_SortKeySpec_missingValueData)
859 zebra_setError(zh, YAZ_BIB1_UNSUPP_MISSING_DATA_ACTION, 0);
862 if (*sks->sortRelation == Z_SortKeySpec_ascending)
863 sort_criteria[i].relation = 'A';
864 else if (*sks->sortRelation == Z_SortKeySpec_descending)
865 sort_criteria[i].relation = 'D';
868 zebra_setError(zh, YAZ_BIB1_ILLEGAL_SORT_RELATION, 0);
871 if (sks->sortElement->which == Z_SortElement_databaseSpecific)
873 zebra_setError(zh, YAZ_BIB1_DATABASE_SPECIFIC_SORT_UNSUPP, 0);
876 else if (sks->sortElement->which != Z_SortElement_generic)
878 zebra_setError(zh, YAZ_BIB1_SORT_ILLEGAL_SORT, 0);
881 sk = sks->sortElement->u.generic;
884 case Z_SortKey_sortField:
885 yaz_log(log_level_sort, "key %d is of type sortField",
887 sort_criteria[i].numerical = 0;
888 sort_criteria[i].ord =
889 zebraExplain_lookup_attr_str(zh->reg->zei,
890 zinfo_index_category_sort,
891 -1, sk->u.sortField);
892 if (sks->which != Z_SortKeySpec_null
893 && sort_criteria[i].ord == -1)
896 YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE, 0);
900 case Z_SortKey_elementSpec:
901 yaz_log(log_level_sort, "key %d is of type elementSpec",
903 zebra_setError(zh, YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE, 0);
905 case Z_SortKey_sortAttributes:
906 yaz_log(log_level_sort, "key %d is of type sortAttributes", i+1);
907 res = zebra_sort_get_ord(zh, sk->u.sortAttributes,
909 &sort_criteria[i].ord,
910 &sort_criteria[i].numerical);
911 if (sks->which != Z_SortKeySpec_null && res != ZEBRA_OK)
915 if (zebraExplain_lookup_ord(zh->reg->zei, sort_criteria[i].ord,
916 &sort_criteria[i].index_type,
919 zebra_setError(zh, YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE, 0);
923 /* allocate space for each cmpare buf + one extra for tmp comparison */
924 for (i = 0; i<num_criteria; i++)
926 cmp_buf[i] = xmalloc(sset->sort_info->max_entries
927 * SORT_IDX_ENTRYSIZE);
928 tmp_cmp_buf[i] = xmalloc(SORT_IDX_ENTRYSIZE);
930 rfd = rset_open (rset, RSETF_READ);
931 while (rset_read (rfd, &key, &termid))
933 zint this_sys = key.mem[sysno_mem_index];
934 if (log_level_searchhits)
935 key_logdump_txt(log_level_searchhits, &key, termid->name);
937 if (this_sys != psysno)
939 if ((sset->hits & 255) == 0 && zh->break_handler_func)
941 if (zh->break_handler_func(zh->break_handler_data))
943 rset_set_hits_limit(rset, 0);
949 resultSetInsertSort(zh, sset,
950 sort_criteria, num_criteria, psysno, cmp_buf,
956 for (i = 0; i<num_criteria; i++)
959 xfree(tmp_cmp_buf[i]);
962 yaz_log(log_level_sort, ZINT_FORMAT " keys, " ZINT_FORMAT " sysnos, sort",
964 for (i = 0; i < numTerms; i++)
965 yaz_log(log_level_sort, "term=\"%s\" type=%s count=" ZINT_FORMAT,
966 terms[i]->name, terms[i]->flags, terms[i]->rset->hits_count);
967 *sort_status = Z_SortResponse_success;
971 RSET resultSetRef(ZebraHandle zh, const char *resultSetId)
975 if ((s = resultSetGet (zh, resultSetId)))
980 ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet,
981 RSET rset, NMEM nmem)
990 ZebraRankClass rank_class;
991 struct zset_sort_info *sort_info;
992 const char *rank_handler_name = res_get_def(zh->res, "rank", "rank-1");
993 size_t sysno_mem_index = 0;
995 if (zh->m_staticrank)
1000 sort_info = zebraSet->sort_info;
1001 sort_info->num_entries = 0;
1003 zebraSet->estimated_hit_count = 0;
1004 rset_getterms(rset, 0, 0, &n);
1005 terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n);
1006 rset_getterms(rset, terms, n, &numTerms);
1008 rank_class = zebraRankLookup(zh, rank_handler_name);
1011 yaz_log(YLOG_WARN, "No such rank handler: %s", rank_handler_name);
1012 zebra_setError(zh, YAZ_BIB1_UNSUPP_SEARCH, "Cannot find rank handler");
1017 RSFD rfd = rset_open(rset, RSETF_READ);
1018 struct rank_control *rc = rank_class->control;
1021 void *handle = (*rc->begin) (zh->reg, rank_class->class_handle, rset,
1022 nmem, terms, numTerms);
1023 zint psysno = 0; /* previous doc id / sys no */
1024 zint pstaticrank = 0; /* previous static rank */
1026 while (rset_read(rfd, &key, &termid))
1028 zint this_sys = key.mem[sysno_mem_index];
1030 zint seqno = key.mem[key.len-1];
1032 if (log_level_searchhits)
1033 key_logdump_txt(log_level_searchhits, &key, termid->name);
1034 if (this_sys != psysno)
1035 { /* new record .. */
1036 if (!(rfd->counted_items & 255) && zh->break_handler_func)
1038 if (zh->break_handler_func(zh->break_handler_data))
1040 yaz_log(YLOG_LOG, "Aborted search");
1044 if (rfd->counted_items > rset->hits_limit)
1047 { /* only if we did have a previous record */
1048 score = (*rc->calc) (handle, psysno, pstaticrank,
1050 /* insert the hit. A=Ascending */
1051 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
1056 zebraSet->estimated_hit_count = 1;
1057 rset_set_hits_limit(rset, 0);
1061 if (zh->m_staticrank)
1062 pstaticrank = key.mem[0];
1064 (*rc->add) (handle, CAST_ZINT_TO_INT(seqno), termid);
1068 { /* we had - at least - one record */
1069 score = (*rc->calc)(handle, psysno, pstaticrank, &stop_flag);
1070 /* insert the hit. A=Ascending */
1071 resultSetInsertRank(zh, sort_info, psysno, score, 'A');
1074 (*rc->end) (zh->reg, handle);
1077 zebraSet->hits = rset->hits_count;
1079 yaz_log(log_level_searchterms, ZINT_FORMAT " keys, "
1080 ZINT_FORMAT " sysnos, rank", kno, zebraSet->hits);
1081 for (i = 0; i < numTerms; i++)
1083 yaz_log(log_level_searchterms, "term=\"%s\" type=%s count="
1085 terms[i]->name, terms[i]->flags, terms[i]->rset->hits_count);
1090 ZebraRankClass zebraRankLookup(ZebraHandle zh, const char *name)
1092 ZebraRankClass p = zh->reg->rank_classes;
1093 while (p && strcmp (p->control->name, name))
1095 if (p && !p->init_flag)
1097 if (p->control->create)
1098 p->class_handle = (*p->control->create)(zh);
1104 void zebraRankInstall(struct zebra_register *reg, struct rank_control *ctrl)
1106 ZebraRankClass p = (ZebraRankClass) xmalloc (sizeof(*p));
1107 p->control = (struct rank_control *) xmalloc (sizeof(*p->control));
1108 memcpy (p->control, ctrl, sizeof(*p->control));
1109 p->control->name = xstrdup (ctrl->name);
1111 p->next = reg->rank_classes;
1112 reg->rank_classes = p;
1115 void zebraRankDestroy(struct zebra_register *reg)
1117 ZebraRankClass p = reg->rank_classes;
1120 ZebraRankClass p_next = p->next;
1121 if (p->init_flag && p->control->destroy)
1122 (*p->control->destroy)(reg, p->class_handle);
1123 xfree(p->control->name);
1128 reg->rank_classes = NULL;
1131 static int trav_rset_for_termids(RSET rset, TERMID *termid_array,
1132 zint *hits_array, int *approx_array)
1136 for (i = 0; i<rset->no_children; i++)
1137 no += trav_rset_for_termids(rset->children[i],
1138 (termid_array ? termid_array + no : 0),
1139 (hits_array ? hits_array + no : 0),
1140 (approx_array ? approx_array + no : 0));
1144 termid_array[no] = rset->term;
1146 hits_array[no] = rset->hits_count;
1148 approx_array[no] = rset->hits_approx;
1150 yaz_log(YLOG_LOG, "rset=%p term=%s limit=" ZINT_FORMAT
1151 " count=" ZINT_FORMAT,
1152 rset, rset->term->name, rset->hits_limit, rset->hits_count);
1159 ZEBRA_RES zebra_result_set_term_no(ZebraHandle zh, const char *setname,
1162 ZebraSet sset = resultSetGet(zh, setname);
1166 *num_terms = trav_rset_for_termids(sset->rset, 0, 0, 0);
1172 ZEBRA_RES zebra_result_set_term_info(ZebraHandle zh, const char *setname,
1173 int no, zint *count, int *approx,
1174 char *termbuf, size_t *termlen,
1175 const char **term_ref_id)
1177 ZebraSet sset = resultSetGet(zh, setname);
1180 int num_terms = trav_rset_for_termids(sset->rset, 0, 0, 0);
1181 if (no >= 0 && no < num_terms)
1183 TERMID *term_array = xmalloc(num_terms * sizeof(*term_array));
1184 zint *hits_array = xmalloc(num_terms * sizeof(*hits_array));
1185 int *approx_array = xmalloc(num_terms * sizeof(*approx_array));
1187 trav_rset_for_termids(sset->rset, term_array,
1188 hits_array, approx_array);
1191 *count = hits_array[no];
1193 *approx = approx_array[no];
1196 char *inbuf = term_array[no]->name;
1197 size_t inleft = strlen(inbuf);
1198 size_t outleft = *termlen - 1;
1200 if (zh->iconv_from_utf8 != 0)
1202 char *outbuf = termbuf;
1205 ret = yaz_iconv(zh->iconv_from_utf8, &inbuf, &inleft,
1207 if (ret == (size_t)(-1))
1210 *termlen = outbuf - termbuf;
1214 if (inleft > outleft)
1217 memcpy(termbuf, inbuf, *termlen);
1219 termbuf[*termlen] = '\0';
1222 *term_ref_id = term_array[no]->ref_id;
1226 xfree(approx_array);
1233 ZEBRA_RES zebra_snippets_hit_vector(ZebraHandle zh, const char *setname,
1234 zint sysno, zebra_snippets *snippets)
1236 ZebraSet sset = resultSetGet(zh, setname);
1237 yaz_log(YLOG_DEBUG, "zebra_get_hit_vector setname=%s zysno=" ZINT_FORMAT,
1243 struct rset_key_control *kc = zebra_key_control_create(zh);
1244 NMEM nmem = nmem_create();
1246 RSET rsets[2], rset_comb;
1247 RSET rset_temp = rset_create_temp(nmem, kc, kc->scope,
1248 res_get (zh->res, "setTmpDir"),0 );
1251 RSFD rsfd = rset_open(rset_temp, RSETF_WRITE);
1258 rset_write (rsfd, &key);
1261 rsets[0] = rset_temp;
1262 rsets[1] = rset_dup(sset->rset);
1264 rset_comb = rset_create_and(nmem, kc, kc->scope, 2, rsets);
1266 rsfd = rset_open(rset_comb, RSETF_READ);
1268 while (rset_read(rsfd, &key, &termid))
1272 struct ord_list *ol;
1273 for (ol = termid->ol; ol; ol = ol->next)
1275 zebra_snippets_append(snippets, key.mem[key.len-1],
1276 ol->ord, termid->name);
1282 rset_delete(rset_comb);
1292 * indent-tabs-mode: nil
1294 * vim: shiftwidth=4 tabstop=8 expandtab