1 /* $Id: logic.c,v 1.56 2007-07-25 11:41:32 adam Exp $
2 Copyright (c) 2006-2007, Index Data.
4 This file is part of Pazpar2.
6 Pazpar2 is free software; you can redistribute it and/or modify it under
7 the terms of the GNU General Public License as published by the Free
8 Software Foundation; either version 2, or (at your option) any later
11 Pazpar2 is distributed in the hope that it will be useful, but WITHOUT ANY
12 WARRANTY; without even the implied warranty of MERCHANTABILITY or
13 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
16 You should have received a copy of the GNU General Public License
17 along with Pazpar2; see the file LICENSE. If not, write to the
18 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
23 \brief high-level logic; mostly user sessions and settings
31 #include <sys/socket.h>
37 #include <yaz/marcdisp.h>
38 #include <yaz/comstack.h>
39 #include <yaz/tcpip.h>
40 #include <yaz/proto.h>
41 #include <yaz/readconf.h>
42 #include <yaz/pquery.h>
43 #include <yaz/otherinfo.h>
44 #include <yaz/yaz-util.h>
46 #include <yaz/query-charset.h>
47 #include <yaz/querytowrbuf.h>
48 #include <yaz/oid_db.h>
49 #include <yaz/snprintf.h>
57 #include <yaz/timing.h>
60 #include <netinet/in.h>
65 #include "termlists.h"
67 #include "relevance.h"
72 #include "normalize7bit.h"
74 #define TERMLIST_HIGH_SCORE 25
78 // Note: Some things in this structure will eventually move to configuration
79 struct parameters global_parameters =
101 // Recursively traverse query structure to extract terms.
102 void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
114 pull_terms(nmem, n->u.p[0], termlist, num);
115 pull_terms(nmem, n->u.p[1], termlist, num);
118 nmem_strsplit(nmem, " ", n->u.t.term, &words, &numwords);
119 for (i = 0; i < numwords; i++)
120 termlist[(*num)++] = words[i];
129 static void add_facet(struct session *s, const char *type, const char *value)
135 for (i = 0; i < s->num_termlists; i++)
136 if (!strcmp(s->termlists[i].name, type))
138 if (i == s->num_termlists)
140 if (i == SESSION_MAX_TERMLISTS)
142 yaz_log(YLOG_FATAL, "Too many termlists");
146 s->termlists[i].name = nmem_strdup(s->nmem, type);
147 s->termlists[i].termlist
148 = termlist_create(s->nmem, s->expected_maxrecs,
149 TERMLIST_HIGH_SCORE);
150 s->num_termlists = i + 1;
152 termlist_insert(s->termlists[i].termlist, value);
155 xmlDoc *record_to_xml(struct session_database *sdb, Z_External *rec)
157 struct database *db = sdb->database;
159 const Odr_oid *oid = rec->direct_reference;
161 /* convert response record to XML somehow */
162 if (rec->which == Z_External_octet && oid
163 && !oid_oidcmp(oid, yaz_oid_recsyn_xml))
166 rdoc = xmlParseMemory((char*) rec->u.octet_aligned->buf,
167 rec->u.octet_aligned->len);
170 yaz_log(YLOG_FATAL, "Non-wellformed XML received from %s",
175 else if (rec->which == Z_External_OPAC)
179 yaz_log(YLOG_WARN, "MARC decoding not configured");
184 /* OPAC gets converted to XML too */
185 WRBUF wrbuf_opac = wrbuf_alloc();
186 /* MARCXML inside the OPAC XML. Charset is in effect because we
187 use the yaz_marc handle */
188 yaz_marc_xml(sdb->yaz_marc, YAZ_MARC_MARCXML);
189 yaz_opac_decode_wrbuf(sdb->yaz_marc, rec->u.opac, wrbuf_opac);
191 rdoc = xmlParseMemory((char*) wrbuf_buf(wrbuf_opac),
192 wrbuf_len(wrbuf_opac));
194 yaz_log(YLOG_WARN, "Unable to parse OPAC XML");
195 wrbuf_destroy(wrbuf_opac);
198 else if (oid && yaz_oid_is_iso2709(oid))
200 /* ISO2709 gets converted to MARCXML */
203 yaz_log(YLOG_WARN, "MARC decoding not configured");
212 if (rec->which != Z_External_octet)
214 yaz_log(YLOG_WARN, "Unexpected external branch, probably BER %s",
218 buf = (char*) rec->u.octet_aligned->buf;
219 len = rec->u.octet_aligned->len;
220 if (yaz_marc_read_iso2709(sdb->yaz_marc, buf, len) < 0)
222 yaz_log(YLOG_WARN, "Failed to decode MARC %s", db->url);
226 yaz_marc_write_using_libxml2(sdb->yaz_marc, 1);
227 if (yaz_marc_write_xml(sdb->yaz_marc, &res,
228 "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
230 yaz_log(YLOG_WARN, "Failed to encode as XML %s",
234 rdoc = xmlNewDoc((xmlChar *) "1.0");
235 xmlDocSetRootElement(rdoc, res);
240 char oid_name_buf[OID_STR_MAX];
241 const char *oid_name = yaz_oid_to_string_buf(oid, 0, oid_name_buf);
243 "Unable to handle record of type %s from %s",
248 if (global_parameters.dump_records)
250 FILE *lf = yaz_log_file();
253 yaz_log(YLOG_LOG, "Un-normalized record from %s", db->url);
254 #if LIBXML_VERSION >= 20600
255 xmlDocFormatDump(lf, rdoc, 1);
257 xmlDocDump(lf, rdoc);
265 xmlDoc *normalize_record(struct session_database *sdb, Z_External *rec)
267 struct database_retrievalmap *m;
268 xmlDoc *rdoc = record_to_xml(sdb, rec);
271 for (m = sdb->map; m; m = m->next){
276 new = xsltApplyStylesheet(m->stylesheet, rdoc, 0);
277 root= xmlDocGetRootElement(new);
278 if (!new || !root || !(root->children))
280 yaz_log(YLOG_WARN, "XSLT transformation failed from %s",
291 if (global_parameters.dump_records)
293 FILE *lf = yaz_log_file();
297 yaz_log(YLOG_LOG, "Normalized record from %s",
299 #if LIBXML_VERSION >= 20600
300 xmlDocFormatDump(lf, rdoc, 1);
302 xmlDocDump(lf, rdoc);
311 // Retrieve first defined value for 'name' for given database.
312 // Will be extended to take into account user associated with session
313 char *session_setting_oneval(struct session_database *db, int offset)
315 if (!db->settings[offset])
317 return db->settings[offset]->value;
322 // Initialize YAZ Map structures for MARC-based targets
323 static int prepare_yazmarc(struct session_database *sdb)
329 yaz_log(YLOG_WARN, "No settings for %s", sdb->database->url);
332 if ((s = session_setting_oneval(sdb, PZ_NATIVESYNTAX))
333 && !strncmp(s, "iso2709", 7))
335 char *encoding = "marc-8s", *e;
338 // See if a native encoding is specified
339 if ((e = strchr(s, ';')))
342 sdb->yaz_marc = yaz_marc_create();
343 yaz_marc_subfield_str(sdb->yaz_marc, "\t");
345 cm = yaz_iconv_open("utf-8", encoding);
349 "Unable to map from %s to UTF-8 for target %s",
350 encoding, sdb->database->url);
353 yaz_marc_iconv(sdb->yaz_marc, cm);
358 // Prepare XSLT stylesheets for record normalization
359 // Structures are allocated on the session_wide nmem to avoid having
360 // to recompute this for every search. This would lead
361 // to leaking if a single session was to repeatedly change the PZ_XSLT
362 // setting. However, this is not a realistic use scenario.
363 static int prepare_map(struct session *se, struct session_database *sdb)
369 yaz_log(YLOG_WARN, "No settings on %s", sdb->database->url);
372 if ((s = session_setting_oneval(sdb, PZ_XSLT)))
375 struct database_retrievalmap **m = &sdb->map;
377 char auto_stylesheet[256];
379 if (!strcmp(s, "auto"))
381 char *request_syntax = session_setting_oneval(sdb,
386 yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
387 "%s.xsl", request_syntax);
388 for (cp = auto_stylesheet; *cp; cp++)
390 /* deliberately only consider ASCII */
391 if (*cp > 32 && *cp < 127)
398 yaz_log(YLOG_WARN, "No pz:requestsyntax for auto stylesheet");
401 nmem_strsplit(se->session_nmem, ",", s, &stylesheets, &num);
402 for (i = 0; i < num; i++)
404 (*m) = nmem_malloc(se->session_nmem, sizeof(**m));
406 if (!((*m)->stylesheet = conf_load_stylesheet(stylesheets[i])))
408 yaz_log(YLOG_FATAL|YLOG_ERRNO, "Unable to load stylesheet: %s",
416 yaz_log(YLOG_WARN, "No Normalization stylesheet for target %s",
421 // This analyzes settings and recomputes any supporting data structures
423 static int prepare_session_database(struct session *se,
424 struct session_database *sdb)
429 "No settings associated with %s", sdb->database->url);
432 if (sdb->settings[PZ_NATIVESYNTAX] && !sdb->yaz_marc)
434 if (prepare_yazmarc(sdb) < 0)
437 if (sdb->settings[PZ_XSLT] && !sdb->map)
439 if (prepare_map(se, sdb) < 0)
445 // called if watch should be removed because http_channel is to be destroyed
446 static void session_watch_cancel(void *data, struct http_channel *c)
448 struct session_watchentry *ent = data;
455 // set watch. Returns 0=OK, -1 if watch is already set
456 int session_set_watch(struct session *s, int what,
457 session_watchfun fun, void *data,
458 struct http_channel *chan)
460 if (s->watchlist[what].fun)
462 s->watchlist[what].fun = fun;
463 s->watchlist[what].data = data;
464 s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
465 session_watch_cancel);
469 void session_alert_watch(struct session *s, int what)
471 if (!s->watchlist[what].fun)
473 http_remove_observer(s->watchlist[what].obs);
474 (*s->watchlist[what].fun)(s->watchlist[what].data);
475 s->watchlist[what].fun = 0;
476 s->watchlist[what].data = 0;
477 s->watchlist[what].obs = 0;
480 //callback for grep_databases
481 static void select_targets_callback(void *context, struct session_database *db)
483 struct session *se = (struct session*) context;
484 struct client *cl = client_create();
485 client_set_database(cl, db);
486 client_set_session(cl, se);
489 // Associates a set of clients with a session;
490 // Note: Session-databases represent databases with per-session
492 int select_targets(struct session *se, struct database_criterion *crit)
495 client_destroy(se->clients);
497 return session_grep_databases(se, crit, select_targets_callback);
500 int session_active_clients(struct session *s)
505 for (c = s->clients; c; c = client_next_in_session(c))
506 if (client_is_active(c))
512 // parses crit1=val1,crit2=val2|val3,...
513 static struct database_criterion *parse_filter(NMEM m, const char *buf)
515 struct database_criterion *res = 0;
522 nmem_strsplit(m, ",", buf, &values, &num);
523 for (i = 0; i < num; i++)
528 struct database_criterion *new = nmem_malloc(m, sizeof(*new));
529 char *eq = strchr(values[i], '=');
532 yaz_log(YLOG_WARN, "Missing equal-sign in filter");
536 new->name = values[i];
537 nmem_strsplit(m, "|", eq, &subvalues, &subnum);
539 for (subi = 0; subi < subnum; subi++)
541 struct database_criterion_value *newv
542 = nmem_malloc(m, sizeof(*newv));
543 newv->value = subvalues[subi];
544 newv->next = new->values;
553 enum pazpar2_error_code search(struct session *se,
554 char *query, char *filter,
555 const char **addinfo)
557 int live_channels = 0;
561 struct database_criterion *criteria;
563 yaz_log(YLOG_DEBUG, "Search");
566 nmem_reset(se->nmem);
567 criteria = parse_filter(se->nmem, filter);
569 live_channels = select_targets(se, criteria);
572 int maxrecs = live_channels * global_parameters.toget;
573 se->num_termlists = 0;
574 se->reclist = reclist_create(se->nmem, maxrecs);
575 // This will be initialized in send_search()
576 se->total_records = se->total_hits = se->total_merged = 0;
577 se->expected_maxrecs = maxrecs;
580 return PAZPAR2_NO_TARGETS;
584 for (cl = se->clients; cl; cl = client_next_in_session(cl))
586 if (prepare_session_database(se, client_get_database(cl)) < 0)
588 *addinfo = client_get_database(cl)->database->url;
589 return PAZPAR2_CONFIG_TARGET;
591 // Parse query for target
592 if (client_parse_query(cl, query) < 0)
597 client_prep_connection(cl);
601 // If no queries could be mapped, we signal an error
605 return PAZPAR2_MALFORMED_PARAMETER_VALUE;
607 return PAZPAR2_NO_ERROR;
610 // Creates a new session_database object for a database
611 static void session_init_databases_fun(void *context, struct database *db)
613 struct session *se = (struct session *) context;
614 struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
615 int num = settings_num();
622 if (global_parameters.server && global_parameters.server->icu_chn)
623 new->pct = pp2_charset_create(global_parameters.server->icu_chn);
625 new->pct = pp2_charset_create(0);
627 new->pct = pp2_charset_create(0);
632 = nmem_malloc(se->session_nmem, sizeof(struct settings *) * num);
633 memset(new->settings, 0, sizeof(struct settings*) * num);
637 for (i = 0; i < num; i++)
638 new->settings[i] = db->settings[i];
640 new->next = se->databases;
644 // Doesn't free memory associated with sdb -- nmem takes care of that
645 static void session_database_destroy(struct session_database *sdb)
647 struct database_retrievalmap *m;
649 for (m = sdb->map; m; m = m->next)
650 xsltFreeStylesheet(m->stylesheet);
652 yaz_marc_destroy(sdb->yaz_marc);
654 pp2_charset_destroy(sdb->pct);
657 // Initialize session_database list -- this represents this session's view
658 // of the database list -- subject to modification by the settings ws command
659 void session_init_databases(struct session *se)
662 predef_grep_databases(se, 0, session_init_databases_fun);
665 // Probably session_init_databases_fun should be refactored instead of
667 static struct session_database *load_session_database(struct session *se,
670 struct database *db = find_database(id, 0);
672 session_init_databases_fun((void*) se, db);
673 // New sdb is head of se->databases list
674 return se->databases;
677 // Find an existing session database. If not found, load it
678 static struct session_database *find_session_database(struct session *se,
681 struct session_database *sdb;
683 for (sdb = se->databases; sdb; sdb = sdb->next)
684 if (!strcmp(sdb->database->url, id))
686 return load_session_database(se, id);
689 // Apply a session override to a database
690 void session_apply_setting(struct session *se, char *dbname, char *setting,
693 struct session_database *sdb = find_session_database(se, dbname);
694 struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
695 int offset = settings_offset_cprefix(setting);
699 yaz_log(YLOG_WARN, "Unknown setting %s", setting);
702 // Jakub: This breaks the filter setting.
703 /*if (offset == PZ_ID)
705 yaz_log(YLOG_WARN, "No need to set pz:id setting. Ignoring");
709 new->target = dbname;
712 new->next = sdb->settings[offset];
713 sdb->settings[offset] = new;
715 // Force later recompute of settings-driven data structures
716 // (happens when a search starts and client connections are prepared)
719 case PZ_NATIVESYNTAX:
722 yaz_marc_destroy(sdb->yaz_marc);
729 struct database_retrievalmap *m;
730 // We don't worry about the map structure -- it's in nmem
731 for (m = sdb->map; m; m = m->next)
732 xsltFreeStylesheet(m->stylesheet);
739 void destroy_session(struct session *s)
741 struct session_database *sdb;
744 client_destroy(s->clients);
745 for (sdb = s->databases; sdb; sdb = sdb->next)
746 session_database_destroy(sdb);
747 nmem_destroy(s->nmem);
748 wrbuf_destroy(s->wrbuf);
751 struct session *new_session(NMEM nmem)
754 struct session *session = nmem_malloc(nmem, sizeof(*session));
756 yaz_log(YLOG_DEBUG, "New Pazpar2 session");
758 session->relevance = 0;
759 session->total_hits = 0;
760 session->total_records = 0;
761 session->num_termlists = 0;
762 session->reclist = 0;
763 session->requestid = -1;
764 session->clients = 0;
765 session->expected_maxrecs = 0;
766 session->session_nmem = nmem;
767 session->nmem = nmem_create();
768 session->wrbuf = wrbuf_alloc();
769 session->databases = 0;
770 for (i = 0; i <= SESSION_WATCH_MAX; i++)
772 session->watchlist[i].data = 0;
773 session->watchlist[i].fun = 0;
779 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
781 static struct hitsbytarget res[1000]; // FIXME MM
785 for (cl = se->clients; cl; cl = client_next_in_session(cl))
787 char *name = session_setting_oneval(client_get_database(cl), PZ_NAME);
789 res[*count].id = client_get_database(cl)->database->url;
790 res[*count].name = *name ? name : "Unknown";
791 res[*count].hits = client_get_hits(cl);
792 res[*count].records = client_get_num_records(cl);
793 res[*count].diagnostic = client_get_diagnostic(cl);
794 res[*count].state = client_get_state_str(cl);
795 res[*count].connected = client_get_connection(cl) ? 1 : 0;
802 struct termlist_score **termlist(struct session *s, const char *name, int *num)
806 for (i = 0; i < s->num_termlists; i++)
807 if (!strcmp((const char *) s->termlists[i].name, name))
808 return termlist_highscore(s->termlists[i].termlist, num);
812 #ifdef MISSING_HEADERS
813 void report_nmem_stats(void)
815 size_t in_use, is_free;
817 nmem_get_memory_in_use(&in_use);
818 nmem_get_memory_free(&is_free);
820 yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
821 (long) in_use, (long) is_free);
825 struct record_cluster *show_single(struct session *s, const char *id)
827 struct record_cluster *r;
829 reclist_rewind(s->reclist);
830 while ((r = reclist_read_record(s->reclist)))
831 if (!strcmp(r->recid, id))
836 struct record_cluster **show(struct session *s, struct reclist_sortparms *sp,
837 int start, int *num, int *total, int *sumhits,
840 struct record_cluster **recs = nmem_malloc(nmem_show, *num
841 * sizeof(struct record_cluster *));
842 struct reclist_sortparms *spp;
845 yaz_timing_t t = yaz_timing_create();
857 for (spp = sp; spp; spp = spp->next)
858 if (spp->type == Metadata_sortkey_relevance)
860 relevance_prepare_read(s->relevance, s->reclist);
863 reclist_sort(s->reclist, sp);
865 *total = s->reclist->num_records;
866 *sumhits = s->total_hits;
868 for (i = 0; i < start; i++)
869 if (!reclist_read_record(s->reclist))
876 for (i = 0; i < *num; i++)
878 struct record_cluster *r = reclist_read_record(s->reclist);
889 yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
890 yaz_timing_get_real(t), yaz_timing_get_user(t),
891 yaz_timing_get_sys(t));
892 yaz_timing_destroy(&t);
897 void statistics(struct session *se, struct statistics *stat)
902 memset(stat, 0, sizeof(*stat));
903 for (cl = se->clients; cl; cl = client_next_in_session(cl))
905 if (!client_get_connection(cl))
906 stat->num_no_connection++;
907 switch (client_get_state(cl))
909 case Client_Connecting: stat->num_connecting++; break;
910 case Client_Initializing: stat->num_initializing++; break;
911 case Client_Searching: stat->num_searching++; break;
912 case Client_Presenting: stat->num_presenting++; break;
913 case Client_Idle: stat->num_idle++; break;
914 case Client_Failed: stat->num_failed++; break;
915 case Client_Error: stat->num_error++; break;
920 stat->num_hits = se->total_hits;
921 stat->num_records = se->total_records;
923 stat->num_clients = count;
926 void start_http_listener(void)
929 struct conf_server *ser = global_parameters.server;
931 if (*global_parameters.listener_override)
932 strcpy(hp, global_parameters.listener_override);
935 strcpy(hp, ser->host ? ser->host : "");
940 sprintf(hp + strlen(hp), "%d", ser->port);
946 void start_proxy(void)
949 struct conf_server *ser = global_parameters.server;
951 if (*global_parameters.proxy_override)
952 strcpy(hp, global_parameters.proxy_override);
953 else if (ser->proxy_host || ser->proxy_port)
955 strcpy(hp, ser->proxy_host ? ser->proxy_host : "");
960 sprintf(hp + strlen(hp), "%d", ser->proxy_port);
966 http_set_proxyaddr(hp, ser->myurl ? ser->myurl : "");
970 // Master list of connections we're handling events to
971 static IOCHAN channel_list = 0;
972 void pazpar2_add_channel(IOCHAN chan)
974 chan->next = channel_list;
978 void pazpar2_event_loop()
980 event_loop(&channel_list);
983 static struct record_metadata *record_metadata_init(
984 NMEM nmem, char *value, enum conf_metadata_type type)
986 struct record_metadata *rec_md = record_metadata_create(nmem);
987 if (type == Metadata_type_generic)
990 p = normalize7bit_generic(p, " ,/.:([");
992 rec_md->data.text = nmem_strdup(nmem, p);
994 else if (type == Metadata_type_year)
997 if (extract7bit_years((char *) value, &first, &last) < 0)
999 rec_md->data.number.min = first;
1000 rec_md->data.number.max = last;
1007 struct record *ingest_record(struct client *cl, Z_External *rec,
1010 xmlDoc *xdoc = normalize_record(client_get_database(cl), rec);
1012 struct record *record;
1013 struct record_cluster *cluster;
1014 struct session *se = client_get_session(cl);
1015 xmlChar *mergekey, *mergekey_norm;
1018 struct conf_service *service = global_parameters.server->service;
1023 root = xmlDocGetRootElement(xdoc);
1024 if (!(mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
1026 yaz_log(YLOG_WARN, "No mergekey found in record");
1031 record = record_create(se->nmem,
1032 service->num_metadata, service->num_sortkeys, cl,
1035 mergekey_norm = (xmlChar *) nmem_strdup(se->nmem, (char*) mergekey);
1037 normalize7bit_mergekey((char *) mergekey_norm, 0);
1039 cluster = reclist_insert(se->reclist,
1040 global_parameters.server->service,
1041 record, (char *) mergekey_norm,
1043 if (global_parameters.dump_records)
1044 yaz_log(YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1045 client_get_database(cl)->database->url, record_no);
1048 /* no room for record */
1052 relevance_newrec(se->relevance, cluster);
1055 // now parsing XML record and adding data to cluster or record metadata
1056 for (n = root->children; n; n = n->next)
1064 if (n->type != XML_ELEMENT_NODE)
1066 if (!strcmp((const char *) n->name, "metadata"))
1068 struct conf_metadata *ser_md = 0;
1069 struct conf_sortkey *ser_sk = 0;
1070 struct record_metadata **wheretoput = 0;
1071 struct record_metadata *rec_md = 0;
1072 int md_field_id = -1;
1073 int sk_field_id = -1;
1075 type = xmlGetProp(n, (xmlChar *) "type");
1076 value = xmlNodeListGetString(xdoc, n->children, 1);
1078 if (!type || !value)
1082 = conf_service_metadata_field_id(service, (const char *) type);
1083 if (md_field_id < 0)
1086 "Ignoring unknown metadata element: %s", type);
1090 ser_md = &service->metadata[md_field_id];
1092 if (ser_md->sortkey_offset >= 0){
1093 sk_field_id = ser_md->sortkey_offset;
1094 ser_sk = &service->sortkeys[sk_field_id];
1097 // non-merged metadata
1098 rec_md = record_metadata_init(se->nmem, (char *) value,
1102 yaz_log(YLOG_WARN, "bad metadata data '%s' for element '%s'",
1106 rec_md->next = record->metadata[md_field_id];
1107 record->metadata[md_field_id] = rec_md;
1110 rec_md = record_metadata_init(se->nmem, (char *) value,
1112 wheretoput = &cluster->metadata[md_field_id];
1114 // and polulate with data:
1115 // assign cluster or record based on merge action
1116 if (ser_md->merge == Metadata_merge_unique)
1118 struct record_metadata *mnode;
1119 for (mnode = *wheretoput; mnode; mnode = mnode->next)
1120 if (!strcmp((const char *) mnode->data.text,
1125 rec_md->next = *wheretoput;
1126 *wheretoput = rec_md;
1129 else if (ser_md->merge == Metadata_merge_longest)
1132 || strlen(rec_md->data.text)
1133 > strlen((*wheretoput)->data.text))
1135 *wheretoput = rec_md;
1138 char *s = nmem_strdup(se->nmem, rec_md->data.text);
1139 if (!cluster->sortkeys[sk_field_id])
1140 cluster->sortkeys[sk_field_id] =
1141 nmem_malloc(se->nmem,
1142 sizeof(union data_types));
1143 normalize7bit_mergekey(s,
1144 (ser_sk->type == Metadata_sortkey_skiparticle));
1145 cluster->sortkeys[sk_field_id]->text = s;
1149 else if (ser_md->merge == Metadata_merge_all)
1151 rec_md->next = *wheretoput;
1152 *wheretoput = rec_md;
1154 else if (ser_md->merge == Metadata_merge_range)
1158 *wheretoput = rec_md;
1160 cluster->sortkeys[sk_field_id]
1165 int this_min = rec_md->data.number.min;
1166 int this_max = rec_md->data.number.max;
1167 if (this_min < (*wheretoput)->data.number.min)
1168 (*wheretoput)->data.number.min = this_min;
1169 if (this_max > (*wheretoput)->data.number.max)
1170 (*wheretoput)->data.number.max = this_max;
1175 union data_types *sdata
1176 = cluster->sortkeys[sk_field_id];
1177 yaz_log(YLOG_LOG, "SK range: %d-%d",
1178 sdata->number.min, sdata->number.max);
1184 // ranking of _all_ fields enabled ...
1186 relevance_countwords(se->relevance, cluster,
1187 (char *) value, ser_md->rank);
1189 // construct facets ...
1190 if (ser_md->termlist)
1192 if (ser_md->type == Metadata_type_year)
1195 sprintf(year, "%d", rec_md->data.number.max);
1196 add_facet(se, (char *) type, year);
1197 if (rec_md->data.number.max != rec_md->data.number.min)
1199 sprintf(year, "%d", rec_md->data.number.min);
1200 add_facet(se, (char *) type, year);
1204 add_facet(se, (char *) type, (char *) value);
1214 "Unexpected element %s in internal record", n->name);
1223 relevance_donerecord(se->relevance, cluster);
1224 se->total_records++;
1234 * indent-tabs-mode: nil
1236 * vim: shiftwidth=4 tabstop=8 expandtab