1 /* $Id: logic.c,v 1.59 2007-08-13 12:12:30 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));
195 yaz_log(YLOG_WARN, "Unable to parse OPAC XML");
196 /* Was used to debug bug #1348 */
198 FILE *f = fopen("/tmp/opac.xml.txt", "wb");
201 fwrite(wrbuf_buf(wrbuf_opac), 1, wrbuf_len(wrbuf_opac), f);
206 wrbuf_destroy(wrbuf_opac);
209 else if (oid && yaz_oid_is_iso2709(oid))
211 /* ISO2709 gets converted to MARCXML */
214 yaz_log(YLOG_WARN, "MARC decoding not configured");
223 if (rec->which != Z_External_octet)
225 yaz_log(YLOG_WARN, "Unexpected external branch, probably BER %s",
229 buf = (char*) rec->u.octet_aligned->buf;
230 len = rec->u.octet_aligned->len;
231 if (yaz_marc_read_iso2709(sdb->yaz_marc, buf, len) < 0)
233 yaz_log(YLOG_WARN, "Failed to decode MARC %s", db->url);
237 if (yaz_marc_write_xml(sdb->yaz_marc, &res,
238 "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
240 yaz_log(YLOG_WARN, "Failed to encode as XML %s",
244 rdoc = xmlNewDoc((xmlChar *) "1.0");
245 xmlDocSetRootElement(rdoc, res);
250 char oid_name_buf[OID_STR_MAX];
251 const char *oid_name = yaz_oid_to_string_buf(oid, 0, oid_name_buf);
253 "Unable to handle record of type %s from %s",
258 if (global_parameters.dump_records)
260 FILE *lf = yaz_log_file();
263 yaz_log(YLOG_LOG, "Un-normalized record from %s", db->url);
264 #if LIBXML_VERSION >= 20600
265 xmlDocFormatDump(lf, rdoc, 1);
267 xmlDocDump(lf, rdoc);
275 #define MAX_XSLT_ARGS 16
277 // Add static values from session database settings if applicable
278 static void insert_settings_parameters(struct session_database *sdb,
279 struct session *se, char **parms)
281 struct conf_service *service = global_parameters.server->service;
286 for (i = 0; i < service->num_metadata; i++)
288 struct conf_metadata *md = &service->metadata[i];
291 if (md->setting == Metadata_setting_parameter &&
292 (setting = settings_offset(md->name)) > 0)
294 char *val = session_setting_oneval(sdb, setting);
295 if (val && nparms < MAX_XSLT_ARGS)
298 int len = strlen(val);
299 buf = nmem_malloc(se->nmem, len + 3);
301 strcpy(buf + 1, val);
304 parms[offset++] = md->name;
305 parms[offset++] = buf;
313 // Add static values from session database settings if applicable
314 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc)
316 struct conf_service *service = global_parameters.server->service;
319 for (i = 0; i < service->num_metadata; i++)
321 struct conf_metadata *md = &service->metadata[i];
324 if (md->setting == Metadata_setting_postproc &&
325 (offset = settings_offset(md->name)) > 0)
327 char *val = session_setting_oneval(sdb, offset);
330 xmlNode *r = xmlDocGetRootElement(doc);
331 xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
333 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
339 xmlDoc *normalize_record(struct session_database *sdb, struct session *se,
342 struct database_retrievalmap *m;
343 xmlDoc *rdoc = record_to_xml(sdb, rec);
346 for (m = sdb->map; m; m = m->next)
352 char *parms[MAX_XSLT_ARGS*2+1];
354 insert_settings_parameters(sdb, se, parms);
356 new = xsltApplyStylesheet(m->stylesheet, rdoc, (const char **) parms);
357 root= xmlDocGetRootElement(new);
358 if (!new || !root || !(root->children))
360 yaz_log(YLOG_WARN, "XSLT transformation failed from %s",
372 insert_settings_values(sdb, rdoc);
374 if (global_parameters.dump_records)
376 FILE *lf = yaz_log_file();
380 yaz_log(YLOG_LOG, "Normalized record from %s",
382 #if LIBXML_VERSION >= 20600
383 xmlDocFormatDump(lf, rdoc, 1);
385 xmlDocDump(lf, rdoc);
394 // Retrieve first defined value for 'name' for given database.
395 // Will be extended to take into account user associated with session
396 char *session_setting_oneval(struct session_database *db, int offset)
398 if (!db->settings[offset])
400 return db->settings[offset]->value;
405 // Initialize YAZ Map structures for MARC-based targets
406 static int prepare_yazmarc(struct session_database *sdb)
412 yaz_log(YLOG_WARN, "No settings for %s", sdb->database->url);
415 if ((s = session_setting_oneval(sdb, PZ_NATIVESYNTAX))
416 && !strncmp(s, "iso2709", 7))
418 char *encoding = "marc-8s", *e;
421 // See if a native encoding is specified
422 if ((e = strchr(s, ';')))
425 sdb->yaz_marc = yaz_marc_create();
426 yaz_marc_subfield_str(sdb->yaz_marc, "\t");
428 cm = yaz_iconv_open("utf-8", encoding);
432 "Unable to map from %s to UTF-8 for target %s",
433 encoding, sdb->database->url);
436 yaz_marc_iconv(sdb->yaz_marc, cm);
441 // Prepare XSLT stylesheets for record normalization
442 // Structures are allocated on the session_wide nmem to avoid having
443 // to recompute this for every search. This would lead
444 // to leaking if a single session was to repeatedly change the PZ_XSLT
445 // setting. However, this is not a realistic use scenario.
446 static int prepare_map(struct session *se, struct session_database *sdb)
452 yaz_log(YLOG_WARN, "No settings on %s", sdb->database->url);
455 if ((s = session_setting_oneval(sdb, PZ_XSLT)))
458 struct database_retrievalmap **m = &sdb->map;
460 char auto_stylesheet[256];
462 if (!strcmp(s, "auto"))
464 char *request_syntax = session_setting_oneval(sdb,
469 yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
470 "%s.xsl", request_syntax);
471 for (cp = auto_stylesheet; *cp; cp++)
473 /* deliberately only consider ASCII */
474 if (*cp > 32 && *cp < 127)
481 yaz_log(YLOG_WARN, "No pz:requestsyntax for auto stylesheet");
484 nmem_strsplit(se->session_nmem, ",", s, &stylesheets, &num);
485 for (i = 0; i < num; i++)
487 (*m) = nmem_malloc(se->session_nmem, sizeof(**m));
489 if (!((*m)->stylesheet = conf_load_stylesheet(stylesheets[i])))
491 yaz_log(YLOG_FATAL|YLOG_ERRNO, "Unable to load stylesheet: %s",
499 yaz_log(YLOG_WARN, "No Normalization stylesheet for target %s",
504 // This analyzes settings and recomputes any supporting data structures
506 static int prepare_session_database(struct session *se,
507 struct session_database *sdb)
512 "No settings associated with %s", sdb->database->url);
515 if (sdb->settings[PZ_NATIVESYNTAX] && !sdb->yaz_marc)
517 if (prepare_yazmarc(sdb) < 0)
520 if (sdb->settings[PZ_XSLT] && !sdb->map)
522 if (prepare_map(se, sdb) < 0)
528 // called if watch should be removed because http_channel is to be destroyed
529 static void session_watch_cancel(void *data, struct http_channel *c)
531 struct session_watchentry *ent = data;
538 // set watch. Returns 0=OK, -1 if watch is already set
539 int session_set_watch(struct session *s, int what,
540 session_watchfun fun, void *data,
541 struct http_channel *chan)
543 if (s->watchlist[what].fun)
545 s->watchlist[what].fun = fun;
546 s->watchlist[what].data = data;
547 s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
548 session_watch_cancel);
552 void session_alert_watch(struct session *s, int what)
554 if (!s->watchlist[what].fun)
556 http_remove_observer(s->watchlist[what].obs);
557 (*s->watchlist[what].fun)(s->watchlist[what].data);
558 s->watchlist[what].fun = 0;
559 s->watchlist[what].data = 0;
560 s->watchlist[what].obs = 0;
563 //callback for grep_databases
564 static void select_targets_callback(void *context, struct session_database *db)
566 struct session *se = (struct session*) context;
567 struct client *cl = client_create();
568 client_set_database(cl, db);
569 client_set_session(cl, se);
572 // Associates a set of clients with a session;
573 // Note: Session-databases represent databases with per-session
575 int select_targets(struct session *se, struct database_criterion *crit)
578 client_destroy(se->clients);
580 return session_grep_databases(se, crit, select_targets_callback);
583 int session_active_clients(struct session *s)
588 for (c = s->clients; c; c = client_next_in_session(c))
589 if (client_is_active(c))
595 // parses crit1=val1,crit2=val2|val3,...
596 static struct database_criterion *parse_filter(NMEM m, const char *buf)
598 struct database_criterion *res = 0;
605 nmem_strsplit(m, ",", buf, &values, &num);
606 for (i = 0; i < num; i++)
611 struct database_criterion *new = nmem_malloc(m, sizeof(*new));
612 char *eq = strchr(values[i], '=');
615 yaz_log(YLOG_WARN, "Missing equal-sign in filter");
619 new->name = values[i];
620 nmem_strsplit(m, "|", eq, &subvalues, &subnum);
622 for (subi = 0; subi < subnum; subi++)
624 struct database_criterion_value *newv
625 = nmem_malloc(m, sizeof(*newv));
626 newv->value = subvalues[subi];
627 newv->next = new->values;
636 enum pazpar2_error_code search(struct session *se,
637 char *query, char *filter,
638 const char **addinfo)
640 int live_channels = 0;
644 struct database_criterion *criteria;
646 yaz_log(YLOG_DEBUG, "Search");
649 nmem_reset(se->nmem);
650 criteria = parse_filter(se->nmem, filter);
652 live_channels = select_targets(se, criteria);
655 int maxrecs = live_channels * global_parameters.toget;
656 se->num_termlists = 0;
657 se->reclist = reclist_create(se->nmem, maxrecs);
658 // This will be initialized in send_search()
659 se->total_records = se->total_hits = se->total_merged = 0;
660 se->expected_maxrecs = maxrecs;
663 return PAZPAR2_NO_TARGETS;
667 for (cl = se->clients; cl; cl = client_next_in_session(cl))
669 if (prepare_session_database(se, client_get_database(cl)) < 0)
671 *addinfo = client_get_database(cl)->database->url;
672 return PAZPAR2_CONFIG_TARGET;
674 // Parse query for target
675 if (client_parse_query(cl, query) < 0)
680 client_prep_connection(cl);
684 // If no queries could be mapped, we signal an error
688 return PAZPAR2_MALFORMED_PARAMETER_VALUE;
690 return PAZPAR2_NO_ERROR;
693 // Creates a new session_database object for a database
694 static void session_init_databases_fun(void *context, struct database *db)
696 struct session *se = (struct session *) context;
697 struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
698 int num = settings_num();
705 if (global_parameters.server && global_parameters.server->icu_chn)
706 new->pct = pp2_charset_create(global_parameters.server->icu_chn);
708 new->pct = pp2_charset_create(0);
710 new->pct = pp2_charset_create(0);
715 = nmem_malloc(se->session_nmem, sizeof(struct settings *) * num);
716 memset(new->settings, 0, sizeof(struct settings*) * num);
720 for (i = 0; i < num; i++)
721 new->settings[i] = db->settings[i];
723 new->next = se->databases;
727 // Doesn't free memory associated with sdb -- nmem takes care of that
728 static void session_database_destroy(struct session_database *sdb)
730 struct database_retrievalmap *m;
732 for (m = sdb->map; m; m = m->next)
733 xsltFreeStylesheet(m->stylesheet);
735 yaz_marc_destroy(sdb->yaz_marc);
737 pp2_charset_destroy(sdb->pct);
740 // Initialize session_database list -- this represents this session's view
741 // of the database list -- subject to modification by the settings ws command
742 void session_init_databases(struct session *se)
745 predef_grep_databases(se, 0, session_init_databases_fun);
748 // Probably session_init_databases_fun should be refactored instead of
750 static struct session_database *load_session_database(struct session *se,
753 struct database *db = find_database(id, 0);
755 session_init_databases_fun((void*) se, db);
756 // New sdb is head of se->databases list
757 return se->databases;
760 // Find an existing session database. If not found, load it
761 static struct session_database *find_session_database(struct session *se,
764 struct session_database *sdb;
766 for (sdb = se->databases; sdb; sdb = sdb->next)
767 if (!strcmp(sdb->database->url, id))
769 return load_session_database(se, id);
772 // Apply a session override to a database
773 void session_apply_setting(struct session *se, char *dbname, char *setting,
776 struct session_database *sdb = find_session_database(se, dbname);
777 struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
778 int offset = settings_offset_cprefix(setting);
782 yaz_log(YLOG_WARN, "Unknown setting %s", setting);
785 // Jakub: This breaks the filter setting.
786 /*if (offset == PZ_ID)
788 yaz_log(YLOG_WARN, "No need to set pz:id setting. Ignoring");
792 new->target = dbname;
795 new->next = sdb->settings[offset];
796 sdb->settings[offset] = new;
798 // Force later recompute of settings-driven data structures
799 // (happens when a search starts and client connections are prepared)
802 case PZ_NATIVESYNTAX:
805 yaz_marc_destroy(sdb->yaz_marc);
812 struct database_retrievalmap *m;
813 // We don't worry about the map structure -- it's in nmem
814 for (m = sdb->map; m; m = m->next)
815 xsltFreeStylesheet(m->stylesheet);
822 void destroy_session(struct session *s)
824 struct session_database *sdb;
827 client_destroy(s->clients);
828 for (sdb = s->databases; sdb; sdb = sdb->next)
829 session_database_destroy(sdb);
830 nmem_destroy(s->nmem);
831 wrbuf_destroy(s->wrbuf);
834 struct session *new_session(NMEM nmem)
837 struct session *session = nmem_malloc(nmem, sizeof(*session));
839 yaz_log(YLOG_DEBUG, "New Pazpar2 session");
841 session->relevance = 0;
842 session->total_hits = 0;
843 session->total_records = 0;
844 session->num_termlists = 0;
845 session->reclist = 0;
846 session->requestid = -1;
847 session->clients = 0;
848 session->expected_maxrecs = 0;
849 session->session_nmem = nmem;
850 session->nmem = nmem_create();
851 session->wrbuf = wrbuf_alloc();
852 session->databases = 0;
853 for (i = 0; i <= SESSION_WATCH_MAX; i++)
855 session->watchlist[i].data = 0;
856 session->watchlist[i].fun = 0;
862 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
864 static struct hitsbytarget res[1000]; // FIXME MM
868 for (cl = se->clients; cl; cl = client_next_in_session(cl))
870 char *name = session_setting_oneval(client_get_database(cl), PZ_NAME);
872 res[*count].id = client_get_database(cl)->database->url;
873 res[*count].name = *name ? name : "Unknown";
874 res[*count].hits = client_get_hits(cl);
875 res[*count].records = client_get_num_records(cl);
876 res[*count].diagnostic = client_get_diagnostic(cl);
877 res[*count].state = client_get_state_str(cl);
878 res[*count].connected = client_get_connection(cl) ? 1 : 0;
885 struct termlist_score **termlist(struct session *s, const char *name, int *num)
889 for (i = 0; i < s->num_termlists; i++)
890 if (!strcmp((const char *) s->termlists[i].name, name))
891 return termlist_highscore(s->termlists[i].termlist, num);
895 #ifdef MISSING_HEADERS
896 void report_nmem_stats(void)
898 size_t in_use, is_free;
900 nmem_get_memory_in_use(&in_use);
901 nmem_get_memory_free(&is_free);
903 yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
904 (long) in_use, (long) is_free);
908 struct record_cluster *show_single(struct session *s, const char *id)
910 struct record_cluster *r;
912 reclist_rewind(s->reclist);
913 while ((r = reclist_read_record(s->reclist)))
914 if (!strcmp(r->recid, id))
919 struct record_cluster **show(struct session *s, struct reclist_sortparms *sp,
920 int start, int *num, int *total, int *sumhits,
923 struct record_cluster **recs = nmem_malloc(nmem_show, *num
924 * sizeof(struct record_cluster *));
925 struct reclist_sortparms *spp;
928 yaz_timing_t t = yaz_timing_create();
940 for (spp = sp; spp; spp = spp->next)
941 if (spp->type == Metadata_sortkey_relevance)
943 relevance_prepare_read(s->relevance, s->reclist);
946 reclist_sort(s->reclist, sp);
948 *total = s->reclist->num_records;
949 *sumhits = s->total_hits;
951 for (i = 0; i < start; i++)
952 if (!reclist_read_record(s->reclist))
959 for (i = 0; i < *num; i++)
961 struct record_cluster *r = reclist_read_record(s->reclist);
972 yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
973 yaz_timing_get_real(t), yaz_timing_get_user(t),
974 yaz_timing_get_sys(t));
975 yaz_timing_destroy(&t);
980 void statistics(struct session *se, struct statistics *stat)
985 memset(stat, 0, sizeof(*stat));
986 for (cl = se->clients; cl; cl = client_next_in_session(cl))
988 if (!client_get_connection(cl))
989 stat->num_no_connection++;
990 switch (client_get_state(cl))
992 case Client_Connecting: stat->num_connecting++; break;
993 case Client_Initializing: stat->num_initializing++; break;
994 case Client_Searching: stat->num_searching++; break;
995 case Client_Presenting: stat->num_presenting++; break;
996 case Client_Idle: stat->num_idle++; break;
997 case Client_Failed: stat->num_failed++; break;
998 case Client_Error: stat->num_error++; break;
1003 stat->num_hits = se->total_hits;
1004 stat->num_records = se->total_records;
1006 stat->num_clients = count;
1009 void start_http_listener(void)
1012 struct conf_server *ser = global_parameters.server;
1014 if (*global_parameters.listener_override)
1015 strcpy(hp, global_parameters.listener_override);
1018 strcpy(hp, ser->host ? ser->host : "");
1023 sprintf(hp + strlen(hp), "%d", ser->port);
1029 void start_proxy(void)
1032 struct conf_server *ser = global_parameters.server;
1034 if (*global_parameters.proxy_override)
1035 strcpy(hp, global_parameters.proxy_override);
1036 else if (ser->proxy_host || ser->proxy_port)
1038 strcpy(hp, ser->proxy_host ? ser->proxy_host : "");
1039 if (ser->proxy_port)
1043 sprintf(hp + strlen(hp), "%d", ser->proxy_port);
1049 http_set_proxyaddr(hp, ser->myurl ? ser->myurl : "");
1053 // Master list of connections we're handling events to
1054 static IOCHAN channel_list = 0;
1055 void pazpar2_add_channel(IOCHAN chan)
1057 chan->next = channel_list;
1058 channel_list = chan;
1061 void pazpar2_event_loop()
1063 event_loop(&channel_list);
1066 static struct record_metadata *record_metadata_init(
1067 NMEM nmem, char *value, enum conf_metadata_type type)
1069 struct record_metadata *rec_md = record_metadata_create(nmem);
1070 if (type == Metadata_type_generic)
1073 p = normalize7bit_generic(p, " ,/.:([");
1075 rec_md->data.text = nmem_strdup(nmem, p);
1077 else if (type == Metadata_type_year)
1080 if (extract7bit_years((char *) value, &first, &last) < 0)
1082 rec_md->data.number.min = first;
1083 rec_md->data.number.max = last;
1090 struct record *ingest_record(struct client *cl, Z_External *rec,
1093 xmlDoc *xdoc = normalize_record(client_get_database(cl),
1094 client_get_session(cl), rec);
1096 struct record *record;
1097 struct record_cluster *cluster;
1098 struct session *se = client_get_session(cl);
1099 xmlChar *mergekey, *mergekey_norm;
1102 struct conf_service *service = global_parameters.server->service;
1107 root = xmlDocGetRootElement(xdoc);
1108 if (!(mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
1110 yaz_log(YLOG_WARN, "No mergekey found in record");
1115 record = record_create(se->nmem,
1116 service->num_metadata, service->num_sortkeys, cl,
1119 mergekey_norm = (xmlChar *) nmem_strdup(se->nmem, (char*) mergekey);
1121 normalize7bit_mergekey((char *) mergekey_norm, 0);
1123 cluster = reclist_insert(se->reclist,
1124 global_parameters.server->service,
1125 record, (char *) mergekey_norm,
1127 if (global_parameters.dump_records)
1128 yaz_log(YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1129 client_get_database(cl)->database->url, record_no);
1132 /* no room for record */
1136 relevance_newrec(se->relevance, cluster);
1139 // now parsing XML record and adding data to cluster or record metadata
1140 for (n = root->children; n; n = n->next)
1148 if (n->type != XML_ELEMENT_NODE)
1150 if (!strcmp((const char *) n->name, "metadata"))
1152 struct conf_metadata *ser_md = 0;
1153 struct conf_sortkey *ser_sk = 0;
1154 struct record_metadata **wheretoput = 0;
1155 struct record_metadata *rec_md = 0;
1156 int md_field_id = -1;
1157 int sk_field_id = -1;
1159 type = xmlGetProp(n, (xmlChar *) "type");
1160 value = xmlNodeListGetString(xdoc, n->children, 1);
1162 if (!type || !value)
1166 = conf_service_metadata_field_id(service, (const char *) type);
1167 if (md_field_id < 0)
1170 "Ignoring unknown metadata element: %s", type);
1174 ser_md = &service->metadata[md_field_id];
1176 if (ser_md->sortkey_offset >= 0){
1177 sk_field_id = ser_md->sortkey_offset;
1178 ser_sk = &service->sortkeys[sk_field_id];
1181 // non-merged metadata
1182 rec_md = record_metadata_init(se->nmem, (char *) value,
1186 yaz_log(YLOG_WARN, "bad metadata data '%s' for element '%s'",
1190 rec_md->next = record->metadata[md_field_id];
1191 record->metadata[md_field_id] = rec_md;
1194 rec_md = record_metadata_init(se->nmem, (char *) value,
1196 wheretoput = &cluster->metadata[md_field_id];
1198 // and polulate with data:
1199 // assign cluster or record based on merge action
1200 if (ser_md->merge == Metadata_merge_unique)
1202 struct record_metadata *mnode;
1203 for (mnode = *wheretoput; mnode; mnode = mnode->next)
1204 if (!strcmp((const char *) mnode->data.text,
1209 rec_md->next = *wheretoput;
1210 *wheretoput = rec_md;
1213 else if (ser_md->merge == Metadata_merge_longest)
1216 || strlen(rec_md->data.text)
1217 > strlen((*wheretoput)->data.text))
1219 *wheretoput = rec_md;
1222 char *s = nmem_strdup(se->nmem, rec_md->data.text);
1223 if (!cluster->sortkeys[sk_field_id])
1224 cluster->sortkeys[sk_field_id] =
1225 nmem_malloc(se->nmem,
1226 sizeof(union data_types));
1227 normalize7bit_mergekey(s,
1228 (ser_sk->type == Metadata_sortkey_skiparticle));
1229 cluster->sortkeys[sk_field_id]->text = s;
1233 else if (ser_md->merge == Metadata_merge_all)
1235 rec_md->next = *wheretoput;
1236 *wheretoput = rec_md;
1238 else if (ser_md->merge == Metadata_merge_range)
1242 *wheretoput = rec_md;
1244 cluster->sortkeys[sk_field_id]
1249 int this_min = rec_md->data.number.min;
1250 int this_max = rec_md->data.number.max;
1251 if (this_min < (*wheretoput)->data.number.min)
1252 (*wheretoput)->data.number.min = this_min;
1253 if (this_max > (*wheretoput)->data.number.max)
1254 (*wheretoput)->data.number.max = this_max;
1259 union data_types *sdata
1260 = cluster->sortkeys[sk_field_id];
1261 yaz_log(YLOG_LOG, "SK range: %d-%d",
1262 sdata->number.min, sdata->number.max);
1268 // ranking of _all_ fields enabled ...
1270 relevance_countwords(se->relevance, cluster,
1271 (char *) value, ser_md->rank);
1273 // construct facets ...
1274 if (ser_md->termlist)
1276 if (ser_md->type == Metadata_type_year)
1279 sprintf(year, "%d", rec_md->data.number.max);
1280 add_facet(se, (char *) type, year);
1281 if (rec_md->data.number.max != rec_md->data.number.min)
1283 sprintf(year, "%d", rec_md->data.number.min);
1284 add_facet(se, (char *) type, year);
1288 add_facet(se, (char *) type, (char *) value);
1298 "Unexpected element %s in internal record", n->name);
1307 relevance_donerecord(se->relevance, cluster);
1308 se->total_records++;
1318 * indent-tabs-mode: nil
1320 * vim: shiftwidth=4 tabstop=8 expandtab